Description
A marketplace is a location where goods and services are exchanged. The traditional market square is a city square where traders set up stalls and buyers browse the merchandise.
ABSTRACT
Title of Dissertation: LABOR ADJUSTMENT IN AN EVOLVING
MARKETPLACE
Manisha G. Singh, Doctor of Philosophy, 2005
Dissertation directed by: Professor John C. Haltiwanger
Department of Economics
This thesis is about the process of employment adjustment. It studies adjustment costs
and their impact on employment and labor demand. It starts by describing key
characteristics of India's labor market; documents legal, economic, and social framework;
investigates the impact and finds magnitudes of worker adjustment costs. Then, it
estimates econometrically parameters of labor adjustment based on ASI industry data
from 1973 to 1997. Specifically effects of the job security regime in India on
employment are estimated while accounting for the concurrent impact of product markets’
liberalization. It establishes that adjustment costs are substantial, adjustment is slow, and
competition mitigates only some of the adverse impact of adjustment costs. Third, it
investigates retrenchment programs across countries. This one-time or episode form of
adjustment is a preferred mechanism in the presence of excess labor. Multi-dimensional
program designs are found to be more successful. These analyses can be better conducted
using a panel dataset based on the unit level data of the ASI.
LABOR ADJUSTMENT
IN AN EVOLVING MARKETPLACE
by
Manisha G. Singh
Dissertation Submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of requirements for the degree of
Doctor of Philosophy
2005
Advisory Committee:
Professor John C. Haltiwanger, Chair
Professor Judy Hellerstein
Professor Howard Leathers
Professor Michael Pries
Professor John Shea
© Copyright by
Manisha G. Singh
2005
ii
ACKNOWLEDGEMENTS
The research is based on Annual Survey of Industries (ASI) data provided by the Central
Statistical Organization (CSO), Ministry of Planning and Programme Implementation,
Government of India. I thank the officials of the CSO for their time and valuable insights
about the data. I also thank the officials of the Ministry of Labour, Government of India
and the V.V. Giri National Labour Institute, Ministry of Labour, Government of India for
sharing their knowledge and experience with this subject. I appreciate the help provided
by the library staff of the Ministry of Labour in locating reference materials and enabling
their use.
I thank the Department of Economics and the Graduate School of the University of
Maryland, College Park for enabling my research amid difficult circumstances. I am
especially grateful to Professor J ohn Haltiwanger whose guidance and support has been
immense.
Part of this research is sponsored by the World Bank, Washington D.C.. I thank Dr.
Martin Rama for this support and discussions during the initial period of this research.
Many thanks are due to my father for helping me to persist with this endeavor, to my
husband for being a steadfast supporter, and to my friends for their numerous cheerful
contributions.
iii
TABLE OF CONTENTS
Acknowledgements………………………………………………………………… ii
Table of Contents……………………………………………………………………iii
List of Tables………………………………………………………………………...iv
I. Introduction and Overview ………………………………………….. 1
1.1 The Indian Economy …………………………………………………... 4
1.2 The Worker Separation Regime ………………………………………. 7
II. Labor Adjustment: Framework and Evaluation
2.1 Introduction…………………………………………………………… 10
2.2 Framework – Labor Adjustment
2.2.1 Legal Framework………………………………………………..14
2.2.2 Economic Framework – Labor Institutions……………………..23
2.2.3 Social Framework – Support Institutions……………………….35
2.3 Impact of Labor Adjustment
2.3.1 Legal Impact – Enforcement…………………………………….37
2.3.2 Economic Impact………………………………………………...39
2.3.3 Social Impact…………………………………………………….46
2.4 Concluding Remarks…………………………………………………….47
III. Labor Adjustment – An Empirical Evaluation
3.1 The Issues and Literature ………………………………………………. 49
3.1.1 Nature of Adjustment Costs……………………………………...50
3.1.2 Alternative Structure of Adjustment Costs………………………51
3.1.3 Evidence on Adjustment Costs………………………………......55
3.1.4 Objectives..………………………………………………………57
3.2 Data and Background
3.2.1 Dataset…………………………………………………………...58
3.2.2 Data Description………………………………………………....62
3.3 The Model and Regression Equations ………………………………… 64
3.3.1 Symmetric Quadratic Convex Adjustment Costs………………. 65
3.3.2 The Regression Equations ……………..……………………….68
3.3.3 GLS and System GMM Estimators …………………………….74
3.4 Results and Findings …………………………………………………… 76
3.4.1 Changes in the speed of adjustment – GLS ……………………..78
3.4.2 Changes in the speed of adjustment – System GMM Results.....81
3.5 Concluding Remarks……………………………………………………..83
IV. Public Sector Retrenchment Programs
4.1 Introduction …………………………………………………………… 85
4.2 Conceptual Framework and Measurement Issues ……………………. 88
4.3 Data Collection ………………………………………………………... 95
iv
TABLE OF CONTENTS (concld.)
4.4 Survey and Analysis of Cross country Evidence
4.4.1 Basic Facts ……………………………………………………… 97
4.4.2 The Relationship Between Rehires, New Hires and Other Program
Characteristics …………………………………………………… 101
4.4.3 Summary Financial Indicators ………………………………….. 103
4.4.4 Highlights of Individual Programs ……………………………… 107
4.4.5 Aggregate Factors ……………………………………………... 117
4.4.6 Missing Pieces -- Measurement and Characterization of Adjustment
Costs ………………………………………………………………….. 119
4.5 Concluding Remarks …………………………………………………. 121
V. Conclusions………………………………………………………………... 124
Tables …..………………………………………………………………………… 128
References ………………………………………………………………………… 171
v
LIST OF TABLES
2.1A Employment Aggregates……………………………………………….128
2.1B Growth Rates – Employment and GDP……………………………….. 129
2.2 Employment in Organised Sector………………………………………130
2.3 Disputes Data…………………………………………………………...131
2.4A Enforcement – Industrial Tribunals and Labour Courts – Dispute
Resolution………………………………………………………………132
2.4B Enforcement – Other Acts……………………………………………...133
2.4C Enforcement – Standing Orders (Employment) Act…………………... 134
2.5 Workers’ Adjustment Assistance – National Renewal Fund (NRF) ..…135
2.6 Adjustment Costs – Labor Restructuring……………………………… 136
2.7 Industrial Disputes Act 1947 with Amendments ………………………138
3.1 Summary Statistics…………………………………………………….. 142
3.2 India: J ob flows ………………………………………………………. 143
3.3 J ob Flows: Comparisons ……………………………………………… 144
3.4 GLS Estimates: Quadratic Convex Adjustment Costs …………………145
3.5 GLS Estimates: Quadratic Convex Adjustment Costs – Parsimonious ..149
3.6 Estimates: Quadratic Convex Adjustment Costs – Parsimonius (System-
GMM Estimator ………………………………………………………. 150
4.1 Scope of Retrenchment: Number of Workers ………………………….151
4.2 Scope of Retrenchment: Financial Costs …………………………….. 152
4.3 Distribution of Employment Reduction Method ……………………... 153
4.4 Distribution of Compensation and Transition Assistance ……………. 154
4.5 Distribution of Targeting …………………………………………….. 155
4.6 Relationship Between Rehiring and Program Characteristics ……….. 156
4.7 Relationship Between New Hires and Program Characteristics ……... 157
4.8 Summary Statistics for Financial Indicators ………………………….. 158
4.9 Relationship Between Net Losses and Program Characteristics …….. 159
4.10 Relationship Between Financial Break-even Period (BEP) Program
Characteristics ………………………………………………………… 160
4.11 Relationship Between Payback Period (PBP) and Program
Characteristics…………………………………………………………. 161
4.12 Key Characteristics: Selected Retrenchment Programs ……………… 162
4.13 Conditions before and during programs: deviations from country-specific
means …………………………………………………………………. 163
4A.1 Summary Statistics for all Countries ………………………………… 164
Chapter1: Introduction and Overview
This thesis is about the process of employment adjustment. The specific issue is
jobs and job security. Concern about both have led policymakers in many countries
to legislate regulation preventing job loss and so protect means of livelihood and
economic welfare. Regulation for job security and severance payments may protect
labor but slows labor adjustment. Yet, at times market forces necessitate adjustment
larger than that permitted by regulatory frameworks. In many countries this
adjustment takes the form of retrenchment programs, specially in the public sector
where de facto job security is perpetual. This one-time adjustment accounts for
substantial (latent) redundancy or labor hoarding. However, accumulating excess
labor and limiting adjustment is costly for firms and may have significant adverse
effects. Both continual adjustment in the presence of job security regulations and
episodal adjustment via retrenchment programs are studied. The application for the
former is India with description of the regulatory and economic framework and
direct measurement of adjustment costs; followed by econometric estimation of the
impact of adjustment costs on employment adjustment. Several public sector
retrenchment programs are examined over 37 countries. Some of the key issues are
discussed below.
1
Economists argue that job security motivated regulation is restrictive. Some find
that it is counter-productive because it slows reallocation and subsequent job
creation. Others find that it increases employment. Bentolila and Bertola (1990)
simulations using aggregate European data find that separation costs impact the
separation margin more than the hiring margin. Net employment may increase in
the long run despite mandated severance payments. Hopenhayn and Rogerson
(1993) simulations using parameter estimates from establishment-level US data find
that a firing tax would reduce labor turnover, employment, and average labor
productivity in the long run. Blanchard (1997) summarized that “firing restrictions
lead to a more sclerotic labor market, a market with lower turnover and lower
productivity; but it is not clear at all that they lead to high unemployment.” The
diverse results follow partly from the data - aggregate versus establishment - and
partly from the differing specifications modeling adjustment costs.
In general, the standard symmetric convex costs model is the most widely used
with aggregate and industry data whereas asymmetric piecewise linear and lumpy
adjustment cost models are being used with microeconomic data. A survey of the
various models is presented by Hamermesh and Pfann (1996). Gauging the impact
of adjustment costs is best accomplished with establishment level panel data
2
wherein any discontinuities in the labor demand decision rule and the impact of
idiosyncratic firm level shocks can be accounted for. This is also important in case
of mandated severance payments and job security regulations that directly affect
only the separation margin, i.e., these have asymmetric effects. With industry data,
using these models is inappropriate since the discontinuities, heterogeneity, and
asymmetries would be largely aggregated away. Moreover, using the standard
convex costs model with industry data retains value in that it accounts for net
adjustment costs that are also substantial.
The idea here is to unravel the effects of adjustment costs, specifically that of
mandated severance payments and job security measures, on the process of
employment adjustment. In examining this issue, most studies focus on labor
market institutions. Here, the attention is also on product markets. If substantial
changes occur in the product markets, as in industry deregulation and/or
economy-wide reform, the effects of job security policy and mandated severance
payments may be masked by the impact of these changes. Or it may be that these
costs have no adverse impact on employment. Given that only 28 million
employees of total employment of over 400 million are in the organised sector
which is subject to the job security regulations, any impact may be of marginal
3
value to start with. However, it is the organised sector, specially manufacturing
industries that may have the potential for higher employment generation.
Understanding whether adjustment costs may be dampening job creation is hence
important. Although use of industry data as opposed to establishment data limits
analysis, investigating the issue of job security while accounting for the concurrent
impact of product markets’ liberalization.should shed some more light on how it
works.
A job security regime has been in place since 1976. Gradual deregulation
continued from late seventies to 1991 when a liberalization program was initiated.
Apart from job security and mandated severance payments, other institutional
(unions, wage-setting) and technological adjustment costs (transaction costs,
shopfloor rearrangements) operate in some measure in India. However, the single
most visible change in the past two decades has been liberalization in product
markets. Below are brief descriptions of the Indian economy with product market
changes, and the job security regime.
1.1 The Indian Economy
India is a large democracy with GDP in year 2000 (year 1970) of about $380
4
($94) billion at 1993-94 prices but a low per capita income of $375 ($175). With
low growth of per capita incomes a principal development objective was poverty
reduction through employment generation. Successive five-year development plans
underscored the need for progressive reduction of unemployment especially in the
rural areas and in cottage industry. This emphasis was driven also by the
consideration to encourage wide-spread participation in the development process.
Adoption of big bang development theories led to a large government presence in
the production sector. The Indian economy became a mixed economy with a
substantial public sector (21 percent in 1973–74 to 32 percent in 1997-98) that
controlled almost all of the core sector, namely, mining, oil, power, transport and a
dominant part of heavy manufacturing, for example, steel and capital goods, and
financial services. A large private sector operated in consumer goods and shared the
financial sector. Almost all firms were subject to quantity controls through
extensive licensing. Foreign competition was limited with phases of ownership
dilution of subsidiaries of foreign firms. Protection through prohibitive tariffs and
binding quotas was granted on the basis of the infant industry argument. An
elaborate and complicated system of taxes, subsidies, and transfers grew from years
of such policy practices. Consequently special interest groups emerged which
sought to maintain the special privileges.
5
Economic performance was disappointing with annual growth rates averaging
about 3 percent during 1960s and 1970s. This focused attention on the causes of
lethargy in the Indian economy leading to several research committees during the
seventies. From late-seventies, India began gradual deregulation allowing firms to
rationalize costs and expand in related products. From 1983, firms were allowed to
expand production and enter closely allied product lines, a process called
broad-banding. From 1989, selected industries were delicensed, that is, entry
restrictions diluted substantially. Such selective mechanisms continued till
July-August 1991. A full-scale adjustment and liberalization program was launched
assisted partly by the IMF and the World Bank. The 1991 program aimed at both
stabilization and structural adjustment. It attempted to correct the fiscal and balance
of payments deficits created largely during the latter half of 1980s, in part due to
large foreign short term borrowing, and to build a legal and policy framework to
allow restructuring. In 1991, delicensing was extended to all (except a short
negative list) industries. Foreign trade and investment regimes were liberalized.
Growth responded, rising to about 5 percent per year in 1980s and around 6 percent
per year in 1990s. However, per capita growth rates remain sub-5 percent. With
about a quarter of the population below poverty line, growth acceleration with job
6
creation remains a key economic objective.
There is a shift of emphasis from socialist, poverty-alleviation,
import-competing development strategy to market-oriented, poverty removal &
improving standard of living, and globally competing growth strategy. Promoting
productive employment growth alongwith allowing greater mobility in industrial
employment and developing safety nets for retrenched/displaced labor are being
considered. This is expected to lead to improved reallocation, improved labor
market functioning and growing human capital. In the attempt to develop safety
nets is an implicit acknowledgment of the need to allow greater labor adjustment.
There is also a concurrent need to assess the costs and benefits of the job security
policy in order to enable modifications, if deemed desirable.
1.2 The Worker Separation Regime
The socialist and nationalist accent to policy peaked during mid-1970s. During
this time the labor law was strengthened and forms the current regime that first
came into force March 1976. It requires industrial establishments to seek written
permission of the government for any separations. Prior to this, severance pay was
fifteen days pay per year of service. Since then, severance is subject to negotiation
7
between employers and workers as separations must be induced or bought. To this
extent, the new regime is that of mandated severance payments.
As per the Industrial Disputes Act (IDA) 1947, retrenchment of workers with
more than 240 days of service requires one month notice in writing stating reasons
for retrenchment and 15 days’ average pay per each year of service as
compensation. The 1976 Amendment to IDA-1947 required “lay off, retrenchment
and closure illegal except with the written permission of the government.” Initially
it covered establishments employing more than 300 workers on an average day in
the 12 months’ prior to the amendment. Another amendment in 1982 extended
coverage to establishments with 100 or more workers. The penalty for retrenchment
or closure without permission includes a fine and/or a prison sentence.
According to Mathur (1989), the number of closures diminished during 1980s,
lock-outs increased, and unions put up effective opposition to “casualization” of
workers. However, adjustment came via increasing recourse to induced voluntary
quits, since this required no government permission. This is costlier and one month
pay for each year of service is common. Voluntary retrenchment programs,
enabling shedding hoarded and excess labor, document the severance pay formulae
8
in chapter four.
The continual and episodal forms of adjustment are dealt with on two levels.
First, chapter 2 starts by describing key characteristics of India’s labor market and
employment situation; documents legal, economic, and social framework; and
investigates the impact and magnitudes of employer and worker adjustment costs.
The labor market institutions - labor law, unions, wage-setting, labor
administration, support (such as pension) institutions - are discussed. Given this
framework, the impact of labor adjustment on workers, employers and the economy
is evaluated. Chapter 3 examines econometrically parameters of labor adjustment
based on ASI industry data from 1973 to 1997. Second, chapter 4 also begins with
description of 41 public sector retrenchment programs over 37 countries and then
provides empirical insights based on these 41 observations. Chapter 5 concludes.
Study limitations and suggestions for further research are also discussed. The
emphasis is on using establishment or unit-level or firm-level panel datasets. The
industry data based estimates are revealing but the magnitudes may well vary
substantially at the unit level.
9
Chapter 2: Labor Adjustment - Framework and Evaluation
2.1 Introduction
India is a labor abundant country. In over 50 years since independence in 1947,
agricultural employment remains dominant at more than 50 percent (see table
2.1A). Further, despite labor abundance, employment has not kept pace with GDP
growth (see table 2.1B). While GDP growth picked up from about 3 percent per
annum up to 1970s to about 5.8 percent in 1980s and 1990s, employment growth
has been at most 2.4 percent per annum. Even this rate was achieved during an
investment boom in 1992-97 after economic reforms in 1991. As per Planning
Commission (2001) the employment elasticity to GDP for all sectors has declined
from 0.53 (during 1977-78 to 1983) to 0.41 (1983 to 1993-94) and further to 0.15
(1993-94 to 1999-2000).
Organised sector employment offering quality employment, that is, a living
wage with medical and retirement benefits, decent working conditions, and some
employment-tenure security grew at just over 2 percent per year in most years until
1983 (see table 2.2). Since then, this growth has dwindled and since 1998 has been
negative. Only one year, 1996, saw spectacular (given other years’ performance)
employment growth of 5.6 percent. This was in the private sector during the
10
investment boom. For the key contributing sector of manufacturing, employment
elasticity has fallen from 0.67 (1977-78 to 1983) to 0.33 (1983 to 1993-94) and
further to 0.26 (1993-94 to 1999-2000). Although services’ sector employment
elasticity are still higher than manufacturing, these are also decreasing, for instance,
the best performing financial sector elasticities are1.00 to 0.92 to 0.73 for the
corresponding periods.
Though the services sector would continue to absorb persons, these persons are
overwhelmingly likely to be the educated and the skilled. The avenues for the
semi-skilled and the unskilled, with at most secondary education, are to learn on the
job in manufacturing and grow. Planning Commission (2001) reports that over 80
percent of the urban population and around 90 percent of the rural population
possess no skill as per the National Sample Survey Organization (NSSO) data for
1993-94. Given declining employment elasticity in manufacturing, where are these
persons to go? Quite a few find relief in rural wage employment programs wherein
over 4 million person-years of employment per year has been generated during the
1990s. To juxtapose this with regular employment growth, if manufacturing
employment elasticity were to bounce back to 0.67 and GDP growth is 6.5 percent
per year, then using manufacturing employment of 48 million in 1999-2000 as base,
11
increase of just over 2 million manufacturing jobs is obtained. However, these jobs
lift persons out of poverty and enable their growth, hence their importance.
The situation becomes more urgent given that both the public sector and
agriculture are shedding labor. Agriculture shedding labor is the usual consequence
of development and of more attractive opportunities emerging elsewhere. Public
sector labor shedding has its genesis in over-employment. The excess labor cannot
be sustained due to fiscal deficits (government) and increasing competition
post-reforms (public sector enterprises). See chapter 4 for more details on public
sector retrenchment.
The surplus labor phenomenon extends beyond the public sector to the private
sector. The genesis here lies in protected product markets and even more protected
labor markets. The extent of redundancy in the organised sector in 1990-91 is
estimated to be a conservative 16 percent (PRAGYA 1999). With the organised
sector employment being about 27 million in 1990-91 (table 2), redundancy works
out to over 3 million workers. With normal growth, industrial restructuring induced
reallocation of workers may be accommodated within the sector from loss-making
activities to value-adding activities. This has an added effect of using
industry-specific skills of displaced workers. However, with accumulated surplus
12
labor of this magnitude, any redeployment within the sector/industry would likely
be swamped. Since the government is itself in re-sizing mode and industrial
enterprises in technological upgradation and catch-up (with international standards)
phase, opportunities in other industries for displaced workers would be scarce.
Enterprises also face increasing competition, external and domestic, following
deregulation and liberalization of (i) entry with the abolition of licensing in 1991 in
most sectors, and (ii) of foreign trade and investment. Indian enterprises have gone
through years of cost-cutting that among other things (soft interest rate regime,
lower oil prices) helped these to become profitable during 1990s and early 2000s
despite hardening competition.
This chapter investigates issues in labor adjustment in India, beginning with the
framework - legal, economic, and social - within which this adjustment takes place.
The legal framework discusses the key labor law - the Industrial Disputes Act in
1947 - along with other germane labor laws. The economic framework describes
the myriad labor market institutions such as wage-setting, unions, and labor
administration that influence employment adjustment. It also outlines formal
institutions that aid workers during unemployment spells and during retirement; the
former are important for continual adjustment and the latter for voluntary
13
retrenchment (frequently amounting to voluntary early retirement) based
adjustment. These influence worker adjustment costs and indirectly employer
adjustment costs.
2.2 Framework - Labor Adjustment
2.2.1 Legal Framework - Labor Laws
The state’s intervention in the labor market in India dates back to 1920. Starting
then three laws were enacted, namely, the Trade Disputes Act in 1920, the
Industrial Employment Standing Orders Act in 1946, and the Industrial Disputes
Act in 1947. Several amendments have modified and expanded the scope of this
last key law. These are listed in a chronological sequence in table 2.7 and are
discussed below.
The Trade Disputes Act 1920 established courts of enquiry and conciliation
boards but provided for no machinery for settling of industrial disputes. It forbade
strikes in public utility services without a month’s notice in writing. No norms
dealing with involuntary separation or other industrial disputes were set.
Subsequently, in 1929 this act was amended and provisions for state intervention in
peacefully settling industrial disputes were made through the conciliation apparatus.
14
During World War II, as emergency measures, the Defense of India Rules
empowered the government to appoint industrial tribunals and enforce the tribunal
awards. In 1946 the Industrial Employment Standing Orders Act provided for
devising, framing, and certifying of standing orders covering service conditions;
and making these known to the workers. Throughout these changes, the British
government in India followed a policy of “laizzez faire” with interventions being
selective.
The government of independent India enacted the Industrial Disputes Act 1947.
It embodies essential principles of the Defense of India Rules regarding industrial
disputes and certain principles of the Trade Disputes Act 1929 about investigation
and settlement of industrial disputes. It sought to achieve social justice on the basis
of collective bargaining, conciliation, arbitration, or finally compulsory
adjudication. The instruments for these processes were to be works committees,
joint management councils, recognized unions; regional and national labor
commissioners; regional and national industrial tribunals, and labor courts. Strikes
and lockouts were prohibited while proceedings remained pending in the tribunals.
Over the years up to 1976, several amendments added new requirements, such as
maintenance of muster rolls by the employer, compensation for layoff and formula
15
for retrenchment based on the length of service.
In 1976, the Industrial Disputes Act was made stringent in that any
retrenchment required prior permission from the appropriate government. This was
further amended and restricted in 1982 by broadening the applicability and by
bringing closures within the ambit of prior permission also. Thus, changing the use
of labor in response to demand conditions was no longer the domain of a producer
but firmly in the realm of an industrial dispute.
Other acts germane to labor adjustment are the Contract Labor (Regulation &
Abolition) Act 1970, the Trade Unions Act 1926, and the Sick Industrial
Companies Act.
The Industrial Disputes Act (IDA) 1947 and Key Amendments
In the first few years, the IDA 1947 specified and established the legal
mechanisms defining industrial disputes, to address them, and to resolve them. In
1953, several textile mills were reportedly planning closing down of one shift. In
most mills, worker rotation was not the norm, hence workers were assigned to one
shift and remained in it. Further, textile mills were the largest employer in the
organised sector, largely setting the terms and practices of labor use and labor
16
relations. Closing of one shift meant that all workers in that shift would be rendered
unemployed. In response to an imminent public disaster, the government amended
the law.
Newly added was chapter VA that specified compensation in case of lay-off at
50 percent of wages and dearness allowance (inflation linked) for 45 days during
any 12 months. The eligibility was that the worker’s name was borne on the
establishment muster rolls and service was continuous for at least one year. Also
added was the requirement of muster rolls to be made available for entry by the
employer since entry recording the worker’s availability for work at the
establishment was mandatory for lay-off compensation. However, the law also
provided for exceptions to the general provision for lay-off compensation: (i)
employer offered alternate employment in the same or any other establishment, (ii)
worker did not present himself for work at the appointed time, and (iii) the lay-off
was due to strike or slow-down of production in another part of establishment.
Regarding retrenchment, employers were required to give one month’s written
notice stating reasons for retrenchment. Mandated was compensation at fifteen
days’ average pay for each year of continuous service completed or any residual
17
part in excess of six months. Retrenchment compensation was to follow a simple
yardstick of length of service on the grounds of a humane public policy and the
consideration that involuntary employment may result in general economic
insecurity. This was not applicable to closures. Employers responded by resorting
to closures to address economic redundancy.
In 1956, to check indiscriminate resort to layoff and retrenchment via closures,
an IDA 1947 amendment provided that termination of employment due to closure
or transfer be considered as retrenchment. Thus, compensation was legally required
for closures also. Subsequent to retrenchment, the employer was to offer an
opportunity for re-employment of retrenched workers if further employment
proposed. Layoff compensation was to extend beyond 45 days to all days laid off,
unless an express agreement between the employee and the employer limits it to the
statutory 45 days.
Further, the recovery of dues to workers by the government may be treated in
the same manner as arrears of land revenue or as public demand by the government
on an application to it by the worker/s. This included compensation for layoff and
retrenchment as also money due under settlements and awards.
18
Occurrence of redundancy due to market conditions continued and employers
resorted to closures as also layoff and retrenchment. In an effort to save jobs,
another IDA 1947 amendment in 1972 required employers to give 60 days’ notice
to state governments prior to closure. The idea was to enable the government to
take remedial steps to prevent job loss. Frequently, the state response was
nationalization of the unit and continued operation of the loss-making unit. The
employer was bought out but the resources remain locked in an unproductive
activity. Employment, albeit in a loss-making but state-owned unit, delayed
transition to a productive job. An implicit social insurance from the government
(soft budget constraint for the unit) reduced worker adjustment cost and incentive to
move while uncertainty about future jobs with little training avenues also reduced
potential benefit from moving.
In 1976, chapter VB was added to IDA 1947. The key provision “lay off,
retrenchment and closure illegal except with the written prior permission of the
government” covered establishments employing more than 300 workers on an
average day in the 12 months’ prior to the amendment. In 1982, another
amendment to IDA 1947 extended coverage to establishments with 100 or more
workers. The penalty for retrenchment or closure without permission included a
19
prison sentence and a fine. This extreme form of job security - prior permission
required for any layoff - altered the form and magnitude of adjustment in the labor
markets as discussed in the next section on impact. The IDA 1947 specified that
retrenchment means the termination by employer of the service of a worker for any
reason other than disciplinary action; excludes termination upon voluntary
retirement, superannuation, non-renewal of a contract, continued ill-health, and
cases where the appointment letter stipulates discharge from service without notice
or reasons.
Employers’ resorted to increasing the workforce mainly by using contractual
workers as opposed to hiring permanently in order to limit adjustment costs during
a downturn. This was opposed by workers and unions. Subsequently, judicial
rulings deemed non-renewal of a contract as retrenchment and liable for
retrenchment compensation. With most avenues for adjustment blocked, the
employers used Voluntary Retirement Schemes (VRS) and its versions wherein
workers were induced to exit by offer of compensation. The formula was along the
lines of that specified under the law. However, depending upon the financial health
of the firm and its prospects, the days of average pay per year of service mostly
varied from 30 to 45 days.
20
Employers facing closure of the unit also sought to force the issue using
lockouts. Though illegal, this form of retrenchment bailed out the employer through
government intervention (nationalization). However, the workers were left hapless
and the resources remained locked in. The land of most of the older textile mills
could have been sold and dues to creditors and workers settled. However, owing to
suspected criminality, this was banned in the 1970s and has only been allowed in
May 2005 following a Supreme Court ruling to that effect. As discussed in the
economic impact sub-section, this offers relief to workers some of whom have been
waiting for over 10 years.
Other Labor Use related Acts
The Contract Labour (Regulation and Abolition) Act 1970 was intending to
regulate the practice of contract labor to prevent exploitation. Supreme Court
judgements served to establish, however, that where outsourcing services are
performed on the premises, the act makes the employer liable to giving permanent
status to such workers who perform these services. The Contract Labor Act 1970 is
proposed to be liberalized - allowing for outsourcing of peripheral activities without
attendant risk of required absorption as permanent employees, thus retaining
21
flexibility for the employer while allowing the outsourcing agency and staff to
specialize more and generate greater efficiency gains and reduced costs.
The Trade Unions Act 1926 allows any seven workers in an enterprise to form
a trade union and register it. This has led to multiple unions in one unit. The
employers are empowered to choose the union that will represent the workers,
giving them the power to use the multiplicity to their advantage by selecting a
weaker union. Further, half of the office bearers of the union may be outsiders
further eroding representativeness of the union.
The above discussed laws provide a high degree of protection to workers. The
Sick Industrial Companies Act (SICA) affords protection to entrepreneurs. Under
this act, sick or loss making companies would be referred to the Board for Industrial
and Financial Reconstruction (BIFR). The BIFR would recommend either
rehabilitation through assistance to save jobs or closure for unviable units. In
practice, closures happened as an extreme measure after years of rehabilitation
failed to produce results. This also allowed crooked managements time to divert
funds from the unit. Ultimately, the units would be rendered unrevivable and the
jobs would be lost. The workers’ adjustment to productive jobs and hence more
22
stable employment is needlessly prolonged. SICA’s adverse effects are alleviated to
some extent by allowing creditors, namely, banks to take control of debtor firms’
assets if firms should default on amortization payments for an extended period.
Settlement of outstanding dues accords priority to worker dues.
2.2.2 Economic Framework - Labor Institutions
Producers can either adjust wages or employment. Adjusting wages would clear
the labor market and obviate the need for massive wage employment support
programs. However, efficiency wages prevail in almost every segment of the labor
market. Further, wage-setting in India being largely exogenous to employers, their
primary response is in adjusting employment.
Wage-setting
Wages are subject to several regulations - minimum wages, acts relating to
payment of wages and payment of bonus (a statutory minimum of 8.33% to a
maximum of 20% of annual wages, thereby raising labor cost by an additional
month). The government’s overarching role dilutes the role of bilateral
employer-employees negotiations or collective bargaining. Another direct influence
is through establishment of tripartite wage boards for some industries whose
primary role is working out a wage structure. Increments to wages (through
23
dearness allowance) in the organised sector are indexed to inflation.
Further, informally, public sector being the dominant employer in the organised
sector (of 28 million employed in the organised sector, about 16 million are in the
public sector), it influences wage movements in the private sector too. Although
central public sector undertakings (PSUs) have some autonomy and collective
bargaining is allowed, actual agreements have to be consistent with the detailed
guidelines issued by the Department of Public Enterprises (DPE, the nodal
coordinating interface between the PSUs and the government). The state PSUs have
independent guidelines but emulate the central PSUs. In practice, both the wage
boards and the DPE use as reference points the Central Pay Commissions’ (set up
periodically) recommendations for salaries of central government employees. This
ripple wage-setting effect emanating from a central government institution extends
to the private sector too.
In the informal or unorganised sector, the central and state governments
intervene by fixing minimum wages for an eight-hour work day applicable to
certain occupations where workers are considered vulnerable. Minimum wage
varies across states, industries and occupations. Awareness of minimum wage
24
appears to be low and multiplicity makes it confusing. Further, since the
unorganised sector is dispersed and workers not unionised, the minimum wage may
not be/is frequently not binding.
Through direct and indirect influences of the government over wage-setting, the
market condition that emerges is that individual employers have little flexibility in
adjusting wages. Thus, employers’ response in labor adjustment is toward
employment adjustment and less to wage adjustment. Impeding workforce
flexibility are regulations restricting separation. Thus, the wage bill, normally a
variable component of costs, has become a fixed cost.
Unions and Contractors
Unions are a pervasive presence in the organised sector. Most units have an
internal union or affiliation to an external union.Union membership increased 150
percent from 1966 to 1979 and stagnates thereafter. In early 1980s, the unions
operating in the Bombay textile mills struck work for 20 months asking for higher
wages (real wage growth was almost zero) and better working conditions. The mill
owners refused to negotiate and ultimately the workers returned without much
change in their conditions.
25
Originating from unions formed pre-independence are three main federations.
The Indian National Trade Union Congress (INTUC) is affiliated to the Congress
(I) political party (that is credited with securing India’s freedom from colonial era
and currently governing); the All India Trade Union Congress (AITUC) is a wing
of the Communist Party of India (CPI) and the Congress of Indian Trade Unions
(CITU) that of the Communist Party of India (Marxist) or CPM; and the Bhartiya
Mazdoor Sangh (BMS) is allied to the the Bhartiya Janata political Party (BJP) that
was in power until 2003. Guided by the self-interest of currently employed
unionised workers, the unions have used their affiliations to block labor law reform.
They demand a safety net for unemployment and redeployment first.
In the unorganised sector employing more than 350 million workers, the
contractor acts as an interface between workers and employers. Now, contractors
are required to be registered with the Labor Commissioner’s office. For workers
displaced from the organised sector, the contractor can be a source of support since
jobs and number of days for which jobs are secured, all flow via a contractor.
Labor Administration - Labor Commissioners
26
The national and regional labor commissioners are the first point of interface of
government with workers and with employers on labor issues. They are entrusted
with the implementation of labor laws. They also provide the first conciliation
platform. During employment distress, illegal separation, and unpaid dues the labor
commissioner can provide redressal. The use of their offices depends upon the
education and awareness level of labor.
Retirement and Disability Funds
Both disability and retirement support are linked to employment. During
unemployment spells, there would be no coverage. Disability, partial or total,
through accidents on the job is covered by the Workmen’s Compensation Act,
1923. Incapacitation due to sickness or due to injury as also maternity leave is
covered by the Employees’ State Insurance Act 1948.
Safety net for retirement exists in several forms. Gratuity is payable since 1972
to workers if in continuous service for 5 years. From May 1994 the coverage
extends to supervisors and managers.
The chief mechanism for retirement support is a contingency fund called the
Employees Provident Fund (EPF). Law requires the institution of mandatory
27
provident funds for employees in factories and establishments. Since its inception
in 1952, the coverage has steadily expanded. Even so, it covers less than 7 percent
of factories and establishments. The purpose is to make provisions for retirement
needs of the employee as well as those of the dependents in case of early death.
Contributions required from employees and employers are fixed by the government.
Effective May 1, 1997, the rate is 12 percent of wages with a lower rate of 10
percent for five industries. The central government, on the recommendation of the
board of trustees of the Employees Provident Fund Organisation (EPFO), declares
the rate of interest to be credited annually to the accounts of provident fund
members. The EPF of private enterprises may be self-managed or be managed by
the EPFO on payment of management fees.
Final withdrawals from the EPF are allowed at retirement or prematurely at the
time of resignation. In effect, the EPF operates only for the employed. This
sufficed in a protected market where employment tenure mostly lasted until
retirement. Partial withdrawals are allowed for illness, invalidation, etcetera; and
for discharging social responsibilities such as marriages of dependents, higher
education of children, construction or purchase of a house/dwelling.
The Employees Pension (EP) scheme was introduced effective November
28
1995. Members of an earlier family pension scheme as well as of EPF were allowed
to join this scheme. Upon its introduction, the earlier pension scheme ceases to
operate and the benefits and entitlements of members remain protected and are
continued under the EP scheme. Eligibility is ten years of service with benefits
commencing as early as 50 years of age and vesting fully at 58 years of age. No
additional contributions from either the employee or the employer are required.
The pension fund is set up and a 8.33 percent of wage share of the employer’s
contribution to the employee’s EPF is diverted to it. Further, the government
contributes 1.16% of members’ wage. Withdrawal of own contributions is
permitted.
The scheme provides for a monthly pension to the employee or dependents as
the case may be for (i) superannuation at 58 years of age, retirement, permanent
total disability, (ii) death during service or after superannuation, retirement,
disability, and (iii) children and orphan pension. The employees have an option to
accept an admissible pension or reduced pension with return of capital.
A third scheme in this family is the Employees Deposit Linked Insurance
Scheme, 1976. It provides that in the event of death of a member, the person
entitled to receive EPF accumulations would be paid an additional amount
29
calculated on the basis of the last 12 months’ balance in the employee’s EPF
account.
The organised sector schemes cover only a minuscule section of the
unorganised sector. The informal segments therefore have no formal safety net
supporting them. A Social Security Group Insurance Scheme operated by the
state-owned Life Insurance Corporation (LIC) offers some insurance to unorganised
sector workers in the event of death, permanent total disability, or loss of limbs.
Half the contribution is paid from a social security fund. The central government
operates five welfare funds funded from a cess on producers in the corresponding
sector. Since 2001, the central government operates a pilot scheme for 20,000
agricultural laborers offering life-cum-accident insurance, money-back and the
superannuation benefits.
Transition, Training and Search Support
No unemployment insurance system exists for either organised or unorganised
sector workers. Discussions inform that a pilot unemployment scheme is being
drawn up by the Employees’ State Insurance Corporation. To deal with massive
redundancy and to accommodate the IDA 1947, VRS compensation packages are
30
offered. As stated earlier, the formula to compute VRS is along the lines of that
specified under the law. However, depending upon the financial health of the firm
and its prospects, the days of average pay per year of service mostly vary from 30
to 45 days.
Vocational training at the national level is vested with the Directorate General
of Employment and Training (DGET), Ministry of Labour, the nodal department
for formulating policies, laying down standards, conducting trade testing and
certification, and directly running several training institutions. At the state level, the
state government departments perform these functions for vocational training.
The key institution for imparting vocational training is the Indian Training
Institute (ITI). It has 4274 centers, 1654 in the public sector and the rest 2620 in
the private sector. The trainee capacity is 628,000 seats. According to Planning
Commission (2002), there exist about 19 million dropouts after class VIII and
another 2.2 million after class X or about 21.2 million children looking for
vocational training every year! The available formal training capacity is only 2.3
million or a gap of 18.9 million needs to be filled. To fill the gap, the ITI system
would require modernization and expansion. Supplementary programs such as the
nodal agencies-EACs under NRF or CRR, and leveraging of existing capacity as
31
envisaged under CII Skills Development Initiative would also be required. These
are discussed below.
The National Renewal Fund (NRF) was established in February 1992 and
remained formally in operation from 1992-93 until 2000-01. Its objectives were to
provide funds (i) for retrenchment and/or closure compensation, where necessary,
to affected workers both in public and private sectors, (ii) for retraining and
redeployment of displaced workers, and (iii) for employment generation schemes as
part of safety net development for both organised and unorganised sectors. The
NRF received funds essentially from budgetary allocations though it was open to
contributions from financial institutions, insurance companies, state governments,
and industrial undertakings. It was administered by an Empowered Authority in the
Department of Industrial Development, Ministry of Industry (now Ministry of
Commerce and Industry) with representatives from several ministries, one of
employers, and two external experts of industrial relations. It considered schemes
recommended by either central and state governments or Board of Industrial and
Financial Reconstruction (BIFR). The BIFR is the authority that determines closure
or rehabilitation of companies/units declared sick (loss-making for at least three
consecutive years). The scheme was thus open to part of the private sector.
32
To implement retraining and redeployment schemes, the NRF worked with
Employee Resource Centers (ERC). An ERC documented details about retrenched
workers, provided counseling, and disseminated information regarding retraining
and redeployment possibilities. Any PSU availing of NRF funds was directed to
establish an ERC. More than 30 ERCs were set up during NRF operation. The
retraining was provided by the Employee Assistance Centers (EAC) that were
operated by designated nodal agencies at locations where substantial concentration
of displaced workers was expected. Other than organizing training, an EAC
surveyed affected workers, assessed wage and self-employment opportunities,
provided counseling and escort services to workers, and provided management
information services for NRF. Some of the EACs were Associated Chambers of
Commerce and Industry (ASSOCHAM) at Kanpur (restructuring textiles firms’
hub), Confederation of Indian Industry (CII) at Bombay (also textiles), Gandhi
Labor Institute at Ahmedabad (textiles), Small Industries Service Institutes at
Indore (textiles), and National Small Industries Corporation at Calcutta.
To implement employment generation schemes, Area Generation Councils
(ARC) were established at Ahmedabad, Indore, and Kanpur.
33
The successor to NRF is the Counseling, Retraining, Redeployment (CRR)
scheme from 2001-02 and aimed at assisting the rationalised employees of central
public sector enterprises (CPSEs). It includes a sensitization program. Selected
nodal agencies counsel VRS workers, impart training and reorientation, develop
curriculum/materials, prepare feasibility reports, market surveys, post-training
follow up interface with credit institutions, support in self-employment, regular
liaison with CPSEs, convene meeting of the coordination committee. The CPSEs
help by clearing all dues of the rationalised workers before release. Further, they
provide valuable input in identifying workers’ retraining needs. The cost per 40 day
module is Rupees 7900/? of which Rupees 2400/? is paid at the rate of Rupees
60/? per day to the trainee.
The Confederation of Indian Industry (CII), a major industry association is
leading a “CII-City & Guilds Skills Development Initiative” on a pilot basis to
train and place workers - new entrants and lateral movers. It has drawn up a fairly
long list of skills available. City & Guilds is reportedly the world’s largest
vocational training and awarding body. It is expected to provide technical
assistance as well as certification reflecting international standards. CII intends to
leverage spare capacity in the existing training apparatus, e.g., using ITIs outside
regular hours and during weekends. The cost is targeted at Rupees 5000/? including
34
training and certification to benchmarked global standards approved by CII. In this
program, there is no free-riding since from the very outset the costs are being
shared among all potential employers, with an industry association playing the role
of the coordinator. It helps in integrating the labor market across regions (and
languages and sub-nationalities that exist in the Indian sub-continent, for example,
a Bihari worker from the east may be employable in southern Kerala without a
hefty deduction to a labor contractor) and internationally too.
To effectively deal with rural unemployment, the National Rural Employment
Guarantee (NREG) scheme is being proposed, for about 200 districts, and to be
extended to all 604 districts of India. Its formulation is controversial, the
contentious issues being coverage, eligibility, minimum number of days of wage
employment to be specified, and the wage level. Also under consideration is an
urban equivalent of NREG with a fixed per person expense of Rs. 10,000. Of this
part is earmarked for training and rest for subsequent on-the-job training.
2.2.3 Social Framework - Support Institutions
Social networking among family, friends, and relatives offers critical support in
times of crises. From better placed members, it affords access to credit and provides
35
information on job possibilities as well. In India, social networks are characterized
by workplace, place of residence as also by class, caste, region of origin, and
religion. Traditional segments based on caste and gender are deeply entrenched in
rural areas. Surveys of urban labor markets have found some evidence of
“systematic correspondence” between inherited (caste, gender, urban/rural, father’s
education) and acquired (own education) characteristics on the one hand and labor
status on the other hand. Occupational rigidities perpetuated by caste are eroding
but the speed is debatable since little data is available (caste data not recorded in
census since 1931.)
Shopkeepers or neighborhood-grocers provide credit for consumption that helps
ward off distress. The longer a worker stays at one place, higher is the trust quotient
and consequent access to consumption credit.
In rural areas, the local government or the Panchayati Raj institution with its
head the “sarpanch” play a role in securing employment. The Panchayat is
instrumental in provision of amenities such as drinking water, sanitation, electricity,
roads, and documents such as the ration card (provides access to subsidized food).
In times of displacement, the Panchayat could mobilize support and thus influences
social security. In urban areas, to a limited extent, this role is played by the unions
36
in the organised sector and the contractor in the unorganised sector.
2.3 Impact of Labor Adjustment
2.3.1 Legal Impact - Enforcement
The IDA 1947 provisions slow employment adjustment. Prior permission is
rarely granted. Then, employers can adjust the workforce by seeking recourse to
courts or by unilateral action that would be disputed by workers. Examining
disputes’ data should provide insights, specially in the absence of direct employer
adjustment costs.
The number of disputes, strikes and lockouts, and associated loss of mandays
have varied over the years, increasing and decreasing (see table 2.3). The more
striking fact is that the proportion of lockouts in total disputes has risen from about
15 percent until 1983 to more than 20 percent in 1980s, more than 30 percent in
1990s, and beyond to 40 percent in 2000s. Moreover, the intensity of lockouts as
reflected in the share of mandays lost due to lockouts has also risen dramatically
from about 20-30 percent in 1970s to about 60 percent in most of 1990s and 2000s.
The major reason stated by employers are wages, indiscipline, and others with
retrenchment cited less than 5 percent. However, the timing of the change offers a
37
clue. The IDA 1947 current regime essentially dates back to 1982 when the prior
permission requirement for retrenchment was extended to all units with 100 or more
workers. The pointer to (latent) redundancy or excess labor as the cause of
increasing lockouts proportion is quite indicative though clear evidence from the
data is not there. Direct loss from cost of mandays lost rose from under Rupees 900
million in 1982 to Rupees 3900 million in 2002, almost the same in real terms. The
indirect loss may be in terms of falling employment elasticity.
A survey study by Ruddar Datt (2000) called “Lockouts in India” covered 42
cases in the state of West Bengal and found that employers mis-stated reasons to be
indiscipline and others instead of retrenchment when indeed the lockout was aimed
at downsizing. Putting both this and the above evidences together, the increasing
incidence and intensity of lockouts could be attributed as a response to redundancy.
With retrenchment related disputes increasing, dispute resolution via the labor
courts and tribunals become important. Table 2.4A shows that the percent share of
disputes and applications disposed off by courts and tribunals has been variable
over the years but the declines are more so after 1993. Is this indifferent
enforcement limited to the IDA 1947? Table 2.4B shows that inspections of
premises under the Factories Act 1948 and the Shops and Establishment Act have
38
also been declining since late-1980s/early-1990s. Convictions under the Factories
Act have declined substantially. Prosecutions under the Shops and Establishment
Act have also reduced. Similarly, certification under the Standing Orders Act have
varied but declined more during late 1990s/early 2000s. In this environment of
weakening enforcement, a measure that has helped to expedite law enforcement and
dispute resolution is the use of lok adalats (people’s courts), these helped reduced
case pendency. Delayed enforcement or delayed processing of requirements or
protracted dispute resolution under different labor laws is causing costs to build up
- time delays, legal costs, managerial costs, production loss, missed opportunities.
2.3.2 Economic Impact
Determination of adjustment costs and their impact is gauged through
evaluation of pre- and post-restructuring outcomes: earnings, employment, and
other worker outcomes. Results are compiled here based on several survey studies
on worker outcomes commissioned by the Ministry of Labour, Government of
India. Simple measures of adjustment costs are computed. Data on productivity,
output, and employment in downsized units and the costs borne by the system
(labor administration, judicial costs) are not available. These are evaluated
qualitatively. For a complete evaluation, the entire spectrum needs to be covered.
39
Nonetheless, several insights are obtained from the analysis here.
Characterizing Downsizing and VRS
Common patterns from the Ministry of Labour Survey Reports
Among reasons for undertaking downsizing and VRS program, the employers
cited demand deficiency, increase in (price) competition, modernizing existing
facilities, and expanding business while rationalising costs. Competitive pressure
to improve efficiency causes churn of firms wherein some firms remain weak and
need to be pushed out, specially those units operating below economic scale.
Released workers from weaker units usually have a high probability of being
absorbed in the stronger units due to some usable specificity of skills. Other
resources released - scrap, machinery and land - get better utilized in expanding
productive activities. This churn is normal among different units within an
industry. However, with accumulated (latent) redundancy or excess labor within a
sector, the absorption capacity within an adjusting industry is impaired. To account
for large-scale (latent) redundancy, redeployment has to be across sectors. With
(latent) redundancy pervasive across sectors, redeployment is a challenge. As
shown in table 2.6, post-VRS most workers are self-employed and in the informal
or unorganised sector. Of those employed, most are wage earners in casual jobs.
Excess labor has accumulated due to protective product and labor markets. In the
40
survey studies available, table 2.6 shows that it varies from 30 percent to 100
percent (closure candidates). Recall that in 1990-91, the extent of (latent)
redundancy in the organised sector is 16 percent.
Reasons for workers accepting VRS included poor financial health of the
firm, a perception of fear psychosis fed by the management, age and other factors.
Most VRS schemes are aimed at workers above 40 years of age. Nevertheless,
table 2.6 shows that there is a significant (in some cases, dominant) proportion of
workers above 50 years of age. This has implications for both re-training and
redeployment. However, most workers are also educated to the secondary level
affording them higher employability.
From all survey reports it emerges that the VRS money was used in
non-productive activities such as repaying debts, children’s marriages, and
house-building. While all are legitimate uses of the lump sum, these generate no
income. A major reason for mounting adjustment costs is the failure of management
to provide adequate financial counseling in the use of VRS money. In the TISS
(2001) study, although 54 percent of the MNC sample had invested the VRS money
in a return generating activity, only 14 percent of the NTC sample had done so.
41
Then, redeployment even if in low-paying unorganised sector assumes greater
urgency. Evidence indicates that most displaced workers (95%) are either
self-employed or employed in the informal sector (increasing casualization) or not
working. Only 5 percent could secure formal wage employment. In adjusting to
post-VRS reduced earnings, expenditure was reduced (45% MNC, 21%NTC),
including on health and children’s education. A significant number (more than 10
percent) of NTC mill workers had children drop out of school.
Other data on VRS and downsizing are not available.
Counseling, Training and Search Assistance
Common patterns from the Ministry of Labour Survey Reports
In the cases studied, table 2.6 shows that some workers were provided
counseling and retraining. Among those who could not avail of this benefit, the
major reason was “no information” about avenues of training under NRF. TISS
(2001) reported that the advanced training institute of DGET failed to attract a
single worker from the NTC mills in Bombay. Where the workers could avail of
training, in Indore, the incentive was the stipend of Rs. 2400/? over 40 days. The
training itself is found to be unrelated to the activity taken up post-VRS and
post-training. Indore textile mill workers displaced due to VRS/lockouts/closures
42
are primarily relocated in the informal sector. The conditions in the informal sector
are thus important for the success of restructuring and reform.
From secondary data
On a macro level, the NRF assisted 107 thousand workers by 1997-98 (table
2.5) with over Rs. 18 billion expenditure. About 90 percent of this was utilized for
VRS compensation of covered workers as per PRAGYA (1999) report. The average
compensation works out to Rs. 142,000/? per worker. Experience so far indicates
that NRF, ERCs, EACs, and ARCs were not effective in aiding redeployment. Of
107 thousand workers assisted from the NRF, only about 5 thousand workers were
deployed (table 2.5) by 1997-98. Reorienting existing mechanisms is required to
address the retraining, skill enhancement, and job search needs of displaced
workers. Labor reallocation in India is impeded by the absence of appropriate labor
institutions aiding worker transition to another job or activity.
Social (Income) Security
Along with training and search assistance, income security during the transition
to a productive activity through an unemployment benefit for a specified period
help. An instance is the Textile Workers’ Rehabilitation Welfare Fund (TWRFS)
scheme of financial assistance: 75 percent of wage is given in the first year, 50
43
percent in the second year, and finally only 25 percent in the third year. Such a
benefit must be fully funded either through contributions or taxes. However,
unemployment spell assistance for organised sector does not exist. The Employees
Provident Fund Organisation (EPFO) allows partial withdrawals for several events
(house-building, dependents’ marriages, children’s higher education) but not for
unemployment spells. The key to smoother transition in both formal and informal
sectors is segregation between employment security and income/social security,
with employment security becoming employment mobility.
Adjustment Costs - Worker and Overall Economy
Using the data from the survey studies, we find that worker adjustment costs are
substantial. Comparing pre-separation earnings to post-separation earnings the loss
of earnings in constant 2002 prices is calculated to vary from 53 percent to 77
percent (see table 2.6). One case had a loss of only 6 percent because the private
mill had been brought to closure point and the wages were close to the informal
sector wages. The huge loss in earnings points to dismal alternate earnings. This is
compounded by the absence of investing VRS money in income generating assets.
Most of the workers are above 50 years of age, so the loss of limited productive
years may be acceptable.
44
However, the loss in earnings compelled expenditure cuts including in health
and children’s education expenses. A significant proportion of workers in one case
had children drop out of school and join the labor market. We use average labor
cost per manday in the ASI-census sector as a proxy for organised sector earnings,
and in the ASI-sample sector as a proxy for unorganised sector earnings in the year
of adjustment, that is, 1997. We calculate that the loss in earnings of children
owing to dropping out of school is a large 46 percent (see table 2.6). This
generational effect where income loss is perpetuating down generations may not be
widespread. However, the huge self-earnings loss of workers may have such
effects.
Apart from legal restructuring using VRS compensation to induce worker
separation, two other mechanisms of adjustment have emerged - illegal
restructuring resorting to lockouts and declared restructuring subject to formal
industrial dispute resolution. The proportion of lockouts in industrial disputes (table
2.3) is increasing rapidly. [Two out of eight cases captured the practice (table 2.6)
of employers using lockouts as a de facto means of downsizing. It gets more
credence from and the Datt (2000) finding that firms mis-state indiscipline as cause
45
of the dispute when in truth it is redundancy and retrenchment.] The third method is
the legal but cumbersome one of declaring the cause of dispute to be retrenchment.
Employer adjustment costs by resorting to lockout and subsequent bailout via
BIFR under the SICA law are the least among the three options. Employee
adjustment costs are the least in disputed restructuring. Where unions are strong,
this outcome may dominate. In both these cases the government labor machinery
incurs costs. Also, the assets/resources remain locked in. In the VRS option, both
employer and workers incur costs, government incurs none, and the resources are
freed up for reallocation. The mechanisms of adjustment are changing to
outsourcing instead of expanding operations, inducing voluntary worker exit with
enhanced severance payments instead of simple separations, and lock-outs instead
of closure/exit. Thus adjustment seems to have been possible albeit at a higher cost.
2.3.3 Social Impact
Displacement from a wage/salary job with fair working conditions to variable
low earnings in casual employment or self-employment amidst worse working
conditions would impact economic well-being and perceptions of self-worth,
standing in the family and society. It may engender negative or positive feelings
46
toward VRS and restructuring. From the case studies, such feelings did emerge but
were addressed by social networks, mainly, family. However, in one case of a
lockout, 57 workers committed suicides. Not only is this an economic and social
waste, it influences the future of democratic governments, and thus can undermine
the support to reform itself in India.
2.4 Concluding Remarks
To sum up, in India there exists a great concern about the adverse impact of job
insecurity on workers. This underlies the IDA 1947 provisions and guides many of
the other policies discussed above that have been set in place regarding the labor
market. In other countries, notably Western European, this concern has also led to
high mandated severance payments but not quite so extreme a version of requiring
prior permission for any layoff or retrenchment.
Adjustment costs of restructuring are substantial for both workers and
employers. Post-separation, the workers’ loss of earnings varies from 53 percent
to 77 percent. A generational effect also exists - due to parental job loss and
consequent children dropping out of school, the loss in earnings of children is a
47
large 46 percent. Employer adjustment costs are high in that costlier means of
adjustment are adopted to enable restructuring.
Transition support structure seems geared to one-time adjustment as embodied
in one-time (sometimes repeated) VRS schemes. There exists an ongoing
restructuring process in the economy and hence continual dynamic support to
displaced workers (e.g. market-driven retraining and placement) is necessary. Apart
from training/search support the provision of a fully funded unemployment
insurance is also a pressing requirement.
Though the future benefits in terms of productivity, output and employment
growth cannot be determined without panel data on restructuring firms, qualitative
conclusions may be drawn. The tradeoff between slow adjustment-lower
cost-lower incremental labor absorption and higher adjustment-higher cost-higher
incremental labor absorption (that is, employment elasticity) exists. It can be
partially ameliorated with efficient functioning of transition support institutions.
48
Chapter 3: Labor Adjustment - an Empirical Evaluation
3.1 The Issues and Literature
This chapter focuses on the econometric estimation of the size of adjustment
costs and their impact on employment in India. The effect of exogenous events,
namely, (i) the changes in the job security and mandated severance payments’
regime in 1976, (ii) the expanded coverage of job security regime in 1982 as well as
the deregulation (“broadbanding”) process begun about 1982, and (iii) the
liberalization and economic reforms’ program in 1991 on adjustment costs and
employment is gauged over-time. Further, these effects’ variation by average
employment size is also obtained. The discussion centers on labor. Though
employment refers to all employees, here the emphasis is on workers. In the
literature discussion, workers and employees may be treated as synonymous.
Workers are treated as a homogenous input post-training and hours are
fixed.Capital and investment demand are ignored (see data limitations below).
Substantial evidence (Hamermesh (1993), chapter 7) indicates that estimates of
employment dynamics are not biased if the adjustment of capital is ignored.
(Anecdotal evidence in India indicates that after job security measures are enforced,
some substitution of capital for labor may have taken place. The paper abstracts
49
from this aspect.) Below is a discussion of the nature and types of adjustment costs.
The sections following describe the data; present the model and regression
equations; and the results.
3.1.1 Nature of adjustment costs
Net costs of adjusting labor demand are those of changing the level (or number)
of workers in any productive unit. These include costs due to disruptions in
production (new equipment, shopfloor rearrangements) and other similar costs not
linked to the identity of the workers. Gross costs of adjustment are related to the
flows of workers, i.e., to the identity of workers. These include hiring costs (search
costs - advertising, screening, interviewing), training costs, separation costs
(severance payments, unemployment benefits), and overhead personnel department
costs dealing with recruitments and separations.
Net costs are internal and implicit in nature. Lost output is not measured.
Training of new workers by senior workers may reduce average efficiency during
the transition period of adjustment. Costs unrelated to production and external in
nature, for example, severance payments, may be measurable. Hamermesh and
Pfann (1996) survey presents the following “extremely tentative conclusions” - (i)
external costs of adjusting labor demand are large, amounting to as much as one
50
year of payroll cost for the average worker; (ii) the average adjustment cost rises
very rapidly with the skill of the worker; and (iii) costs of hiring exceed costs of
separations.
Adjustment costs may arise from the economic environment (factor markets,
technology, product markets) the firm faces. However, these may also arise directly
or indirectly from government policies such as mandated severance payments and
mandated advance notice of layoffs.
3.1.2 Alternative Structures of Adjustment Costs
Several forms have been proposed in the literature, under assumptions about the
expectations formation process and that given the expectations-process, the only
reason for slow adjustment is the cost associated with changing labor demand.
The most used structure of adjustment costs in the literature is that of symmetric
convex adjustment costs, usually restricted to be quadratic. Convexity is imposed
on marginal adjustment cost and implies that it is increasing in labor demanded.
Continuity and differentiability ensure that changes in market conditions, however
small, will lead the firm to adjust labor demand continually. Symmetry around zero
ensures that the marginal cost of increasing labor is the same as that of an
51
equal-sized decrease. Sargent (1978) formulated a symmetric quadratic adjustment
cost function and resultant linear labor demand based on it with the assumption of
rational expectations. The quadratic approximation and symmetry imply that even
in the presence of unit-specific shocks, the linearity of labor demand function
allows its aggregation across units. Thus, this form may be applied to aggregated
data with no other consideration and yet retaining its theoretical basis.
The symmetric quadratic form is a convenient tool. However, data from several
countries point to differences in magnitude and persistence between positive input
adjustments and negative ones. A non-linear functional form that has the quadratic
model nested within is proposed by Pfann and Verspagen (1989). It has a parameter
that captures asymmetry, which if negative implies that marginal cost associated
with a negative adjustment or separation exceeds that of a positive adjustment and
vice-versa. If the parameter is zero, it implies that there is no asymmetry and the
symmetric convex costs model applies. Though non-linear, the model can be used
with aggregate data with the assumptions of no idiosyncratic shocks and a fixed
number of firms. Then, a Taylor-series expansion of the firm’s adjustment function
around zero net employment change and subsequent aggregation of these across
firms is identical to the Taylor-series expansion of the aggregate (of firms)
52
adjustment cost function. The quadratic convex cost structure mostly abstracts from
attrition.
Nickell (1978) discussed piecewise linear adjustment costs that are more
relevant for asymmetric gross costs or per worker costs such as severance payments
and interviewing. Gross adjustment costs such as severance payments are distinct
and differ in magnitude from recruitment costs such as interviewing. Further, these
are state dependent, that is, severance payments are incurred only for separations
and hiring costs are incurred only for hiring. Both separations and hiring may take
place simultaneously. In Nickell’s model, at any point, the firm is either expanding
or contracting in response to a corresponding product demand shock over the
business cycle. Attrition is assumed to be zero. His model analyzed regulatory
adjustment costs in the context of the business cycle assuming perfect foresight.
Demand variations over the business cycle are therefore known. Labor use then
varies in terms of employment or hours, though the number of shifts is fixed at one.
According to the paper, empirically, differences between the convex costs model
and the linear costs model “would tend to be washed out” at the aggregate level but
may be picked up at the industry level. These costs engender a discontinuity in the
optimal decision rule. In the presence of idiosyncratic shocks and firm
53
heterogeneity, this form requires unit level data. Application to aggregated data
amounts to assuming away firm heterogeneity. Using industry data captures
inter-industry variation but still loses firm heterogeneity.
In most recent studies of adjustment costs, there is an emphasis on delineating
net costs from gross costs. This stems from prior evidence (Hamermesh and Pfann
(1996)) that the two are different in magnitude and impact, and partly from
evidence on employment flows. Job flows - jobs created and destroyed, by
continuing firms and by entrants and exiters - underlying the net result, that is,
change in employment stock behave quite differently to net change. Davis and
Haltiwanger (1998) document job flows for most developed countries and several
developing countries. The salient features across economies remain that gross job
flows - job creation and job destruction - co-exist in almost every time period and
sector, are substantial, persistent, asymmetric and concentrated at relatively few
plants. The key factor is the existence of heterogeneity among firms as reflected in
the predominance of idiosyncratic factors as opposed to aggregate and sectoral
factors. Gross adjustment costs are related to gross worker flows. Worker flows
exceed job flows to the extent of any attrition. Net costs are related to net
employment change.
54
Models of adjustment costs described above deal mostly with net costs as
opposed to per worker costs. The symmetric convex costs based model is based on
net employment change. The asymmetric convex costs model recognizes that the
costs associated with a net negative or a net positive change may be different and
hence also examines net costs though asymmetric. The piecewise linear costs model
explicitly accounts for gross costs - recruitment costs versus separation costs, each
per worker. The symmetric convex costs model, despite its limitations, is the most
widely used specification with aggregate macroeconomic data (Hamermesh and
Pfann (1996)) and is suitable for disaggregated industry data.
3.1.3 Evidence on adjustment costs
The evidence on adjustment costs, both for labor demand and investment
demand is discussed at length by Hamermesh and Pfann (1996). Here, the emphasis
is on mandated severance payments related papers. Two major approaches emerge.
One uses simulations based on statistics or estimates of parameters, and the other is
the more direct parametric estimation.
Bentolila and Bertola (1990) model a forward looking monopolist firm subject
55
to separation costs and demand uncertainty. Their simulations using aggregate
European data find that separation costs impact the separation margin more than the
hiring margin. Net employment may increase in the long run despite mandated
severance payments, if the shocks faced by the firms are highly persistent and
worker attrition is high. With more persistent shocks, the need to alter labor demand
is lower. With higher worker attrition, the need for separations is reduced. It is
unclear what happens if the shocks are more variable and/or attrition is lower.
Hopenhayn and Rogerson (1993) model a competitive firm also subject to a firing
tax and idiosyncratic productivity shocks in a general equilibrium framework. They
perform simulations using parameter estimates from establishment-level US data
and find that a firing tax would reduce labor turnover, employment, and average
labor productivity in the long run. The opposing results could be due, in part, to
assumptions about market structure (monopolist versus perfect competition) and
inclusion of the entry-exit process in the latter study.
Attempts to explain these contrary conclusions include accounting for other
labor institutions that might be simultaneously operating in conjunction with a tax
on labor adjustment. Bertola and Rogerson (1996) analyze the process of
wage-setting. They argue that relative-wage compression in Europe may be part of
56
the puzzle. It is conducive to higher employer initiated job turnover. Both
institutions together can explain similar job flows in US and Europe but higher
unemployment rates in Europe. Cabrales and Hopenhayn (1997) allow firms to
choose between two coexisting types of contracts - permanent and temporary. Their
calibrations using Spanish data find that with the inclusion of temporary contracts,
there is a substantial increase in reallocation but not in aggregate productivity.
Inferring of adjustment costs within labor demand estimation is attempted by
Fallon and Lucas (1991) based on the symmetric quadratic convex cost
specification and using disaggregated industry data from India. Job security is
instituted via requirements of prior government approval for separation and
severance payments for any separations since March 1976. Fallon and Lucas use a
difference approach by introducing a dummy variable for regime change to gauge
the impact of the severance requirements. They find a significant one-time
reduction in labor demand (intercept) but no significant fall in the speed of
adjustment.
3.1.4 Objectives
This paper attempts to unravel the effects of adjustment costs on employment
57
while accounting for (i) the changes in the job security and mandated severance
payments’ regime in 1976, (ii) the expanded coverage of job security regime in
1982 as well as the deregulation (“broadbanding”) process begun about 1982, and
(iii) the liberalization and economic reforms’ program in 1991, and (iv) average
employment size. The hypotheses are that average employment and speed of
adjustment (a) decrease after job security provisions are more restricted and (b)
increase post-economic reform in product markets despite no further change in
labor law restrictions. Labor demand is systematically related to the degree of
competition; and, the speed of adjustment is also affected by the degree of
competition. The effects are greater for larger firms. Note that after 1982, with
changes in both job security regime and product markets, the net effect can go
either way.
3.2 Data
3.2.1 Dataset
The study dataset comprises the Annual Survey of Industries (ASI) series from
1973 through 1997. The unit of observation is a factory in case of manufacturing
industries. It covers registered factories/plants, i.e., units with 10 or more workers
in plants using electric power or 20 or more workers in plants using no electric
58
power (and as registered under sections 2m(1) and 2m(11) of the Factory Act,
1948.) It consists of (i) a census of firms with over 50 employees with power (99
without power) and (ii) a sample of the remaining registered plants or factories with
10-50 employees with power (20-99 employees without power.) The reference
period for annual coverage of premises is the preceding 12 months. The cut-off
point is a registered factory working on any day of the preceding twelve months,
and in any part of which a manufacturing process is being carried or ordinarily so
carried on. The information generated from the ASI data includes number of
workers and number of employees besides the economic profile of the plants.
Though the data is collected at plant level, it is publicly available, up to 1997-98,
only at the 3-digit industry level of disaggregation.
Study Universe
ASI is classified using the national industrial classification (NIC), that changes
from NIC1970 to NIC1987. NIC1970 is applicable to years 1973 through 1988.
NIC1987 applies to years 1989 though 1997. Developing a consistent set of series
across the 25 years led to merger of several industry groups. These are as follows.
? ASI classification merges some two industry groups of NIC1970 into one group of
NIC1987, for example, NIC70-228 and NIC70-229 merge into NIC87-229. To
generate a comparable series, data of industries NIC70-228 and NIC70-229 for
years under NIC1970 classification (1973-1988) are merged and a 25-year
consistent series is obtained for such industry groups.
? ASI classification splits a NIC1970 industry group into two NIC1987 groups, for
example, NIC70-253 splits into NIC87-253 and NIC87-256. For these cases, a
59
25-year consistent series is obtained by merging the split NIC1987 industry groups
into one series corresponding to the NIC1970 group.
Some industry groups had missing observations (the entire set of 30 variables)
due to ASI data suppression policy wherein data for any industry group with two or
less number of factories is not reported in that year. Such industries (11 in number)
have been entirely dropped. Also omitted are several new groups included in NIC87
for which no earlier data are available. The truncated dataset is 84% (73%) of the
universe (ASI manufacturing) in terms of output for initial-year 1973-74 (end-year
1997-98). The figure in terms of employment (as measured by number of workers)
is 85% (78%).
Deflation to constant price data
Each industry group data is deflated using the wholesale price index (WPI)
corresponding to that industry. The WPI is the most broad-based deflator available.
The CPI is available for industrial workers but not according to industry groups.
The GDP deflator can be computed from available national accounts data at the
2-digit level of disaggregation rather than the 3-digit level possible with the WPI.
The WPI series has two bases, 1970/71 and 1982/83. The 1970/71 WPI series is
spliced to 1982/83 WPI series at the industry group level and the series so obtained
60
is used to deflate the 25-year NIC1987-classification industry data series.
Data Limitations
The dataset is rich in that it includes labor, capital, other inputs, output, value
added variables. However, it is available at present at a disaggregated industry
level instead of at unit level. Any model based on such a dataset assumes away or
ignores idiosyncratic shocks at the factory level. As Davis and Haltiwanger (1998)
and Roberts and Tybout (1996) point out, in both developed and developing
countries, idiosyncratic shocks are important. Significant variation in economic
outcomes may be driven by these shocks. Further, there is a large possibility that
the structure of adjustment costs may be lost in industry averages that can be
examined with this dataset. Second, the annual frequency may be more than the
frequency of decision-making of factory managers, i.e., the economic period may
be less than an year (quarterly or semi-annual). Nonetheless, industry level data
capture substantial variation, and any variation at an annual level is only likely to
understate patterns evident at higher frequency data. Thus, examining this dataset
should be quite instructive. Also, the capital series is a financial measure.
Appropriate economic capital stock or corresponding investment series exist at
macroeconomic level but are not yet available by disaggregated industry. The
61
model adopted hence omits capital but includes output.
3.2.2 Data Description
Summary statistics
Summary statistics of key variables across industry and year are provided in
Table 3.1. Value of output is rescaled from lakhs (hundred thousand) to thousands
to better represent the output values at the lower end of the range. In all variables,
the range and standard deviation are quite large. To the extent that estimation is in
log-linear form (as is the norm for labor demand estimation), the large range
doesn’t complicate the results.
Job flows
Employment numbers may reveal only part of the story. Following the
methodology developed by Davis and Haltiwanger (1992), computations of
manufacturing job flows during 1973/74 - 1997/98 show that job flows in India at
3-digit industry level are large (see Tables 3.2 and 3.3). The 1973-1997 average job
creation is 5.5 percent of employment (US four-digit data based manufacturing
number is 2.5 percent of employment), job destruction is 3.8 percent (US: 3.6
percent), and net employment rate is 1.7 percent (US: -1.1 percent). Further, job
reallocation rate footnote is 9.4 percent and excess job reallocation rate is 6.6
percent. Thus, 70 percent of job reallocation is excess, that is, 70 percent of the
62
turnover is over and above the rate required to accommodate the net change in
employment. Yet, somewhat paradoxically, substantial labor adjustment costs exist
in India and increased from 1976. This co-existence of high labor turnover and high
labor adjustment costs seemingly point to the limited relevance of adjustment costs
for reallocation.
A closer look, however, reveals more. Table 3.2 shows period averages for:
? pre-1976 (1973-75) or pre- labor law restrictions
? 1976-82, immediate period following labor law changes with no change in the
product markets’ regime
? 1983-90, period of gradual deregulation of product markets
? 1991-97, period post-July/August1991 liberalization package.
? 1976-97, complete period post-labor law changes
Excess job reallocation fell from 7.9 percent during 1973-75 (pre-resstrictions)
to 5 percent during 1976-82 (immediately after 1976 job security measures) but
rose to 7.8 percent during 1983-90 (gradual deregulation) falling to 6.4 percent
during 1991-97 (post liberalization program). Similarly, after separation restrictions
and induced higher severance requirements are imposed in 1975 (1975/76 - March
1976), both job creation and destruction rates fall in the 1976-82 period. Net
employment is about the same. During 1983-90 job destruction rate recovers and
surpasses the pre-restrictions level. Unlike job destruction and excess job
reallocation, job creation rate does not bounce back and remains depressed. Over
63
1991-97, job creation rate is higher at 5.8 percent but remains quite below the rate
prior to job security restrictions of 7.4 percent. Overall, in 1976-97, job destruction
rate recovers but job creation rate falls and remains lower, with net employment
rate halving to the previous period’s levels. So, there seems to be a dampening
effect of adjustment costs on employment if not on job turnover.
Part of the story may lie in product markets reform - gradual deregulation in
1980s and a liberalization package in 1990s creating more productive capacity in
response to suppressed demand. With higher demand, job creation will rise and job
destruction may fall. However, in that case, excess job reallocation whould remain
about the same since the flows are accommodating net employment creation. A rise
in excess job reallocation points to variation in both flows - creation and
destruction. So, the story in India appears to be more than the expanding demand
after liberalization. Though the evidence is very limited, it nonetheless points to a
shift in labor restructuring activity despite the continuance of a strong job security
and high adjustment cost regime.
3.3 The Model and Regression Equations
Estimations are based on the symmetric quadratic convex cost form. This is a
64
widely used specification with aggregate macroeconomic data (Hamermesh and
Pfann (1996)) and is used here with the disaggregated industry data. It is suitable
for studying the effects of adjustment costs on employment in India while
accounting for (i) the changes in the job security and mandated severance
payments’ regime in 1976, (ii) the expanded coverage of job security regime in
1982 as well as the deregulation (“broadbanding”) process begun about 1982, and
(iii) the liberalization and economic reforms’ program in 1991, and (iv) average
employment size. Estimates of the reduced form labor demand function, that is,
coefficients of wage, output, mandays, speed of adjustment, and, changes in
average employment and speed of adjustment due to regime changes and
employment size are obtained. The estimating equations are briefly derived below.
3.3.1 Symmetric Quadratic Convex Adjustment Costs
The underlying theory is based on (Sargent (1978)) the standard value
maximization by a firm given production function technology and labor-adjustment
cost function, and ignoring capital. The firm maximizes the present value of the
stream of expected future profits
65
max
?Lt?
V
t
? E
t ?
i?0
?
R
t
i
_
t?i
? E
t ?
i?0
?
R
t
i
?Q
t?i
? w
t?i
L
t?i
? AC
t?i
?
Q
t?i
? F?L
t?i
?? ?_
o
? _
o,t?i
?L
t?i
? 0.5_
1
L
t?i
2
AC
t?i
? 0.5c??L
t?i
?
2
where R ? 1 is the discount factor, F is the production function with _ parameters,
_
o,t?i
being a random parameter having a zero mean and positive variance, w is the
wage, and AC is the adjustment cost, c is adjustment cost parameter. R is assumed
to be constant. The Euler equation is
RE
t
L
t?i?1
? ?
_
1
c
? R ? 1?L
t?i
? L
t?i?1
? c
?1
?w
t?i
? _
o
? _
o,t?i
?, i ? 0, 1, ..
and the solution or decision rule for labor demand is
L
t
? _L
t?1
? _c
?1
E
t ?
i?0
?
R
i
?w
t?i
? _
o
? _
o,t?i
?
0 ? _ ? 1
Decision-makers in the firm are assumed to be risk-neutral and have rational
expectations, based on the information available at time t about the path of shocks.
Simplifying assumptions about the process generating shocks to forcing variables
are made. A first order autoregressive process is used as a reasonable
approximation to the unknown correct form for these shocks. Then, optimal
66
forecasting implies replacing expected values with their lagged values. Let the
autocorrelation parameters be _
w
and _
q
. Then the path of labor demand relating
current period employment to its lagged value and a vector of forcing variables,
wages and productivity shocks can be described as
L
t
? _L
t?1
? _c
?1
??1 ? R_
w
?
?1
w
t
? ?1 ? R?
?1
_
o
? ?1 ? R_
q
?
?1
_
o,,t
?
Sargent (1978) fits his model to data that are deviations from means and trends
(partly accounts for omitting capital). This maximization is equivalent to
minimizing the present value of the stream of expected costs (Hamermesh and
Pfann (1996)). Based on the standard simplifying assumption that deviations from
optimal profits are quadratic both in adjustment costs and in deviations of the actual
demand for labor from the optimal path, this yields
min
?Lt?
E
t ?
i?0
?
C
t?i
? E
t ?
i?0
?
R
t
i
?0.5?_L
t?i
?
? L
t?i
?
2
? 0.5c??L
t?i
?
2
?
where L
t?i
?
constitutes the optimal path in the static optimization without adjustment
costs, i.e., c ? 0. The above assumption implies that the firm is a price-taker in all
its markets. The estimating version is
67
L
t
? ?1 ? _?L
t
?
? _L
it?1
0 ? _ ? 1
_ is a non-linear function of R, c, and _. Given the assumptions, linearity of this
equation allows for aggregation even if firms face different shocks. Then,
adjustment costs are implied by _, the closer it is to one, the higher are the
adjustment costs and lower is the speed of adjustment.
3.3.2 The Regression Equations
In this paper, recall that output is included though capital and investment
demand are ignored (since appropriate series are not yet available and relying on
evidence indicating that the estimates of employment dynamics are not biased if
adjustment of capital is ignored). Thus, the forcing variables included in L
t
?
are the
real wage and output. Further, since the model uses employment and not
labor-hours the variable mandays per worker is also included as a regressor. In
India, manday is a fixed number of hours and labor-use varies with number of
mandays and number of workers (employment). It is assumed that the first
dimension of labor adjustment is number of mandays and second that of number of
workers. Hence, employment and labor (workers) demand does not affect lagged
mandays. The symmetric quadratic convex cost model is readily amenable to
68
aggregation across units as discussed above. Thus, an industry labor demand
function is the basis of the log-linear regression equation. Then, with first-order
autoregressive processes generating shocks to forcing variables in L
it
?
and their
expected values replaced by their lagged values, the estimating version for industry
i ?i ? 1, ..., N; N ? 136?in period t ?t ? 1, ..., T; T ? 25?is
lnL
it
? ?1 ? _?lnL
it
?
? _lnL
it?1
? u
it
, i ? 1, ..., N; t ? 1, ..., T
lnL
it
? ?1 ? _??_
0
? ?_lnw
?1
? _lnQ
?1
? _lnH
?1
?? ? _lnL
it?1
? u
it
lnL
it
? ?1 ? _??_
0
? ?_
?
lnL
it
?
?? ? _lnL
it?1
? u
it
0 ? _ ? 1
u
it
? _
l
u
it?1
? ?
it
?
it
~ i.i.d. ?0, _
?
2
?
Rational expectations based on autoregressive processes imply bona-fide use of the
lagged values of forcing variables and hence serially uncorrelated disturbance in the
labor demand equation. However, the error u
it
here allows for some serial
correlation with i.i.d. errors. L is demand for workers, Q is output, w is wage, H is
hours, _ is the measure of adjustment costs. Note that the autoregressive parameters
of the forcing variables are embedded within their coefficients.
The estimation procedure is a difference-in-difference approach based on size
(average employment size) and period dummies. Also, as a measure of evolving
competition (protected and licensed production yielding to increasing deregulation
69
from 1982 and substantial delicensing in 1991), price-cost margins are generated
and interacted with lagged employment and size dummies.
The primary differences are due to size dummies. The secondary differences in
these size-differences are due to the regime changes captured by the period
dummies. To the above equation are added these size dummies. Average
employment size is the number of workers in an industry divided by the number of
factories in that industry. This is used to generate two size dummies (i) for size 100
and above, S1 takes the value one and above, and (ii) for size 300 and above, S3
takes a value one and above. S3 captures industries that come within the ambit of
the 1976 job security restrictions (that is, all units with employment 300 workers
and above). S1 represents industries that are further included in job security
restrcitions in 1982 (that is, all units with employment 100 workers and above). The
size dummies are cumulative, hence their estimated coefficients indicate marginal
effects. Being based on average employment size rather than actual employment
size may make these size dummies more restrictive than the usual classification of
large, medium, and small units of interest with unit-level panel data. _
S1
and _
S3
are
the increments in effect of adjustment costs due to size differences. Then, the
estimating equation is
70
lnL
it
? ?1 ? _??_
0
? ?_
?
lnL
it
?
?? ? _lnL
it?1
lnL
it
? ?1 ? _
0
? _
S1
S1 ? _
S3
S3??
?_
0
? _
S1
S1 ? _
S3
S3 ? ?_
?
lnL
it
?
??
? _
0
lnL
it?1
? _
S1
S1lnL
it?1
? _
S3
S3lnL
it?1
S1 ? 1 if size ? 100, zero else
S3 ? 1 if size ? 300, zero else
The coefficient for wage is expected to be negative and that of output to be
positive. The coefficient for mandays may be positive or negative depending upon
whether workers and mandays are complements or substitutes. _
0
is expected to be
positive, that is, some adjustment costs exist prior to the job security restrictions
and that speed of adjustment is slow. _
S1
and _
S2
would be positive if larger
industries find it harder to adjust employment. Similarly, average employment
coefficient _
0
is expected to be positive and, _
S1
and _
S2
also positive as larger
firms are expected to have higher average employment.
There are three period dummies, (i) P1 is the period dummy that takes value
one for all periods from 1976 onward representing job security regime from 1976;
(ii) P2 is the period dummy that takes value one for all periods from 1983 onward
reflecting the expanded coverage of job security regulations in 1982 (in the tables,
it is called Year82 dummy to refer to 1982 regime changes as well as the
deregulation or “broadbanding” process begun about 1982), and (iii) P3 is the
71
period dummy that takes value one for all periods from 1991 onward reflecting the
economic reforms and liberalization in 1991. Again, the period dummies are
cumulative, hence their estimated coefficients indicate marginal effects. These
period dummies are interacted with the size dummies. Below is a version of the
estimating equation presented with size dummies and only one period dummy (to
avoid notational clutter).
lnL
it
? ?1 ? _
0
? _
S1
S1 ? _
S3
S3 ? _
P1
P1 ? _
S1.P1
S1.P1 ? _
S3.P1
S3.P1??
?_
0
? _
S1
S1 ? _
S3
S3 ? _
P1
P1 ? _
S1.P1
S1.P1 ? _
S3.P1
S3.P1 ? ?_
?
lnL
it
?
??
? _
0
lnL
it?1
? _
S1
S1lnL
it?1
? _
S3
S3lnL
it?1
? _
P1
P1lnL
it?1
? _
S1.P1
S1.P1lnL
it?1
? _
S3.P1
S3.P1lnL
it?1
S1 ? 1 if size ? 100, zero else
S3 ? 1 if size ? 300, zero else
P1 ? 1 if year ? 1976, zero else
Post-1976: Due to job security restrictions and consequent higher severance
payments, adjustment costs in general will rise and an increase in the adjustment
cost parameter _ or a positive marginal effect _
P1
is expected. Further, _
S3.P1
(size
300 and post-1976 restrictions) should be positive since the restrictions apply
particularly to size 300 and above. The intercept terms or average employment
parameters _
P1
and _
S3.P1
may be negative indicating a one-time downward shift of
the labor demand function for all industries and specially for those with size 300
and above. _
S1.P1
(size 100 and post-1976 restrictions) may be positive or
72
insignificant and _
S1.P1
may be negative or insignificant. So, the marginal effects
for industries with size 100 and above may be negligible.
Post-1982: _
S1.P2
(size 100 and post-1982) is expected to be positive on account
of job security coverage expanding to units with employment 100 and above but
negative on account of deregulation. _
S1.P2
is expected to be negative due to
expanded job security coverage and positive otherwise. The net effect may be
positive or negative. Similarly, other coefficients may be negative or positive.
Post-1991: If the impact of liberalization in product markets filtered to labor
markets, a reduction in adjustment costs, i.e., negative _
P3
, _
S1.P3
, and _
S3.P3
are
expected, even though there is no change in the job security regime. _
P3
, _
S1.P3
, and
_
S3.P3
may similarly be positive.
Since gradual deregulation in product markets (protected and licensed
production yielding to increasing deregulation from 1982 and substantial
delicensing in 1991) continued over several years, a competition measure every
period to capture the evolutionary change is more appropriate. Herfindahl indexes
by industry in India could not be found. Thus, as an inverse measure of gradually
evolving competition, price-cost margins (PCM, G
it?1
) are also generated and
interacted with lagged employment and size dummies. A truncated version of the
estimating equation without period dummies is shown below.
73
lnL
it
? ?1 ? _
0
? _
S1
S1 ? _
S3
S3 ? _
g
lnG
it?1
? _
S1.g
S1lnG
it?1
? _
S3.g
S3lnG
it?1
??
?_
0
? _
S1
S1 ? _
S3
S3 ? _
g
lnG
it?1
? _
S1.g
S1lnG
it?1
? _
S3.g
S3lnG
it?1
? ?_
?
lnL
it
?
??
? _
0
lnL
it?1
? _
S1
S1lnL
it?1
? _
S3
S3lnL
it?1
? _
g
lnG
it?1
lnL
it?1
? _
S1.g
S1 ? G
it?1
lnL
it?1
? _
S3.g
S3 ? G
it?1
lnL
it?1
S1 ? 1 if size ? 100, zero else
S3 ? 1 if size ? 300, zero else
The premise here is that a higher degree of competition makes the need to adjust
more urgent and may thus blunt the impact of adjustment costs and increase the
speed of adjustment. So, the change in _ due to competition (PCM) is negative
(positive) or that _
g
is expected to be positive. The PCM coefficient, _
g
, will be
positive or negative depending upon whether employment increases or decreases
with greater price-cost margin. Higher PCM being associated with increasingly
imperfect competition and excess capacity, it is expected to be negative. The
interactions of PCM and size dummies would reflect these effects coupled with
labor hoarding effects.
3.3.3 GLS and System-GMM Estimators
The estimates are obtained using GLS with errors robust to heteroskedasticity
and autocorrelation. In the presence of any disturbance autocorrelation, the lagged
variables (lagged dependent variable, lagged output; usually wage also, not in India
application, see chapter two) are endogenous and need to be instrumented. In the
74
absence of external instruments, lagged first differences can serve as instruments
for levels equation. However, with substantial sluggishness in employment, the
correlation of these instruments with regressors is weak. Shea R-squared (Shea
(1997)) is about 0.01 only indicating poor instrument relevance.
A more comprehensive set of instruments is obtained by using the system GMM
estimator. Efforts to account for group effects in dynamic panel data estimations
with autoregressive panel series produced the first-differenced GMM estimator.
The strategy is first-differencing the equation and using lagged levels as
instruments for the pre-determined and endogenous variables in first differences.
Though sound, the technique performed poorly in that for panels where the series
are highly autoregressive, this estimator is found to have large finite sample bias
and poor precision. An alternative is found in the system-GMM estimator (Blundell
and Bond (1998)), where equations in first differences use lagged levels as
instruments as before and additionally levels equations are instrumented using
lagged first differences. The system GMM estimator adds moment conditions based
on initial conditions restrictions (jointly stationary means of the regressand and the
regressors, to ensure that the respective first differences are uncorrelated with group
effects). When the additional moment conditions are valid, the system GMM
75
estimator is found to reduce bias and improve precision (Blundell, Bond and
Windmeijer 2000). System-GMM allows for some serial correlation in the residual
disturbance, with the lags specified for instruments increasing by the order of serial
correlation. The number of moment conditions reduces correspondingly.
To exploit the rich instrumentation possible with the system-GMM technique,
these GMM estimates are obtained. However, owing to the use of average
employment size dummies and period dummies, only a parsimonious specification
is feasible. The interactions of size and period dummies with the forcing variables
are excluded. This parsimonious specification is estimated with both GLS and
GMM to enable appropriate comparisons. In so doing, the first three years’
observations are lost. These are 1973-1975, the years prior to the job security
restrictions’ regime change in 1976. The impact of these 1976 restrictions alone is
thus not directly available with the GMM instruments.
3.4 Results and Findings
The estimates of model I (symmetric quadratic convex adjustment costs) based
equation are presented in Table 3.4. As noted in the data description above, an year,
say 1973, in this dataset refers to the period April73-March74. The equation is
76
estimated for three periods - 1973-1975, the period before the job security
restrictions become mandatory; 1973-1982, the period after restrictions but before
the job security provisions’ coverage is extended and gradual deregulation become
operational; and, 1973-1997, the full period for which data are available.
Size dummies are called Size100 and Size300. These dummies are interacted
with wages, mandays, output, and lagged workers to assess the impact of regime
changes on the relevant regression coefficients. Thus, there are three panels in the
base regression for 1973-75: (i) effects for all employment sizes, (ii) marginal
effects for size 100 and above, and (iii) further marginal effects for 300 and above.
Period dummies are called Year76, Year82, and Year91. All these dummies are
interacted with the entire first set of three panels including wages, mandays, output,
lagged workers, and their interactions with size dummies. Year dummies from 1979
onward when a regime (new government) change and before the gradual
deregulation process is operational and industry dummies at 2-digit level are also
included.
The wage coefficient is statistically significant and negative as expected. The
estimate for output is also significant and positive. Mandays (proxy variable for
77
hours) coefficient is significant and positive. The constant term representing
average or autonomous employment is significant and positive. The coefficient for
lagged workers that captures the effect of adjustment costs (extent of sluggishness
in adjustment and opposite of the speed of adjustment) is significant and positive.
Note that, this coefficient is high and larger in absolute terms than that of wage,
mandays, or output. The evidence for 1973-1975 indicates the existence of
adjustment costs even before the job security restrictions came into effect. Size
effects are significant. Average employment is substantially higher for size300
industries (but negative for size100 industries). The lagged employment coefficient
for size300 industries is negative implying that larger size in India may be
associated with lower adjustment costs or higher speed of adjustment.
Most industry and year dummies are insignificant. The results for wage,
mandays, output, and lagged workers remain unchanged in the different model
specifications. Following is the discussion on size and period dummies and their
interactions. It focuses on average employment and the speed of adjustment.
3.4.1 Changes in the speed of adjustment - GLS
The impact of the labor restrictions through the period dummy “Year76” is very
significant for average employment and quite high (the intercept or the coefficient
78
for Year76). Recall that in the log-linear labor demand equation, this coefficient is
log of average employment. Calculations indicate a fall in average employment of
about 28 percent (reduces to about 25 percent when the full sample period is taken).
The speed of adjustment is the opposite of the coefficient on lagged employment.
The coefficient of lagged workers interacted with Year76 is a significant 0.032 or 3
percent (significance declines in the full sample period). Further, the size300 and
lagged workers interaction is very significant. It is 0.377 indicating a large 38
percent (36 percent in full sample period) increase in sluggishness or a
corresponding decrease in the speed of adjustment over and above the 3 percent
found for all sizes. Post job security restrictions in 1976, (i) there is a one-time
fall (about 28 percent) in labor demand and (ii) a fall in the speed of
adjustment (3 percent) and additionally so (38 percent) for large industries.
Fallon and Lucas (1991) had found a significant negative coefficient on the period
dummy (here, Year76) but no significant change in the speed of adjustment. Note
that the interactions of Year76 with wages, mandays, and output are all highly
significant and opposite to the signs without interaction. This is consistent with the
theoretical model - when _ increases, the coefficients magnitude decreases - and
lends considerable support to a significant increase in _ and hence fall in ?1 ? _?or
speed of adjustment.
79
Using the full sample period 1973-1997 and all dummies, results are somewhat
unexpected for Year82. The coefficient of lagged employment, size300 and Year82
dummy interaction is significant, about 0.15. This increase in sluggishness is
expected for size100 industries (insignificant here) since job security regime
coverage extended to units with employment 100 and above (from 300 and above in
1976). Nonetheless, this does indicate increasing adjustment costs for larger
industries with increasing coverage. Further, there is a small decrease in
sluggishness (-0.026) in lagged employment and Year82 interaction; perhaps the
impact of deregulation on industries with size below 100 or those not specifically
covered by job security regime. Unlike the case of Year76 and Year82 period
dummies, the only significant result for Year91 is the large increase in average
employment for size300 or large industries. Until 1997, there is no significant effect
of lagged employment-Year91 interactions. So, there is some evidence of increase
in employment but no corresponding decrease in sluggishness.
The coefficients of price-cost margin and its interaction with lagged workers are
insignificant. However, the size dummy interactions are very significant. For
size100, the results are as expected - PCM decreases average employment and
80
increases the sluggishness in adjustment, i.e., a positive coefficient of lagged
employment-PCM-size100. Paradoxically, the results are opposite for size300 -
PCM-size300 dummy has a positive coefficient (increases employment) and
PCM-size300-lagged employment has a negative coefficient or there is a decrease
in sluggishness.
Competition thus has differing impacts. For medium sized (size100) firms, it
increases average employment and decreases sluggishness with opposing
results for larger (siz300) firms (lower average employment and higher
sluggishness). Effects of pre-existing labor hoarding (requiring shedding of
excess labor lowering average employment and net employment change) may
be responsible.
3.4.2 Changes in the speed of adjustment - System GMM Results
The results for system-GMM estimations are shown in table 3.6 and the
corresponding GLS results in table 3.5. Before presenting the estimates, note that
the over-identifying conditions fail to be rejected using the Hansen test or are tested
as valid. Though the absence of autocorrelation of the first order in first differenced
residuals is rejected (as expected, if the level equations have i.i.d. errors), no-AR(2)
fails to be rejected or that no serial correlation in the levels’ equation disturbance is
valid. The regressions are thus well specified. Second, both tables 3.5 and 3.6 use
81
the parsimonious specification. Even so, all relevant variables cannot be
instrumented given sample size. With restricted lags for the instruments, along with
base regressors there are two distinct sets of variables that are instrumented. One set
includes additionally Year82 period dummy with its interactions. The second set
includes additionally Year91 dummy and its interactions.
Recall that there are no observations for the pre-1976 period. Results pertain to
1976-1982 and 1976-1997 periods. Hence, a direct before-after comparison is not
possible. For both periods and with either set of instruments for the 1976-1997
period, GMM-lagged employment coefficient is very significant and high, the
magnitude being slightly smaller than corresponding significant GLS estimates.
GMM-size effects are obtained for size100 industries whereas in corresponding
GLS these are obtained for size300 industries. Both indicate higher average
employment and lower sluggishness for larger industries. The magnitude of GMM
estimate is many times over those of GLS.
Additional results are obtained for the full sample with Year82 and Year91
period dummies and the second set of instruments. GMM-estimates for
size100-Year82 show a large fall in average employment and for
82
size100-Year82-lagged employment show a large increase in sluggishness. This
is expected since the coverage of job security restrictions expanded to size100
industries in 1982. However, this result is not obtained using the first set of
instruments, the estimates are insignificant. With corresponding GLS, these results
are obtained for size300 industries as also in the full specification GLS and in both,
the magnitudes are smaller. Further, there is a decrease in sluggishness or the
coefficient of lagged employment-Year82 interaction is negative for small firms.
This is not obtained with parsimonious GLS but is found in full GLS model.
Year91-lagged employment GMM estimates are significant but opposing with
the two sets of instruments. No significant results for Year91 and its interactions are
obtained in corresponding GLS. Similarly, no significant results for PCM and its
interaction with lagged employment are obtained in either GMM or corresponding
GLS. Note that parsimony led to exclusion of PCM-size dummies’ interactions
which produced results in the full GLS model.
3.5 Concluding remarks
Overall, there is evidence that after job security restrictions in 1976, (i) there is
a one-time fall (about 28 percent) in labor demand and (ii) a fall in the speed of
adjustment (3 percent) and additionally so (38 percent) for large industries. Weaker
83
evidence indicates increasing adjustment costs for larger industries with increasing
coverage and lower adjustment costs for smaller ones excluded from coverage.
GMM-estimates post-Year82 for size100 industries show a large fall in average
employment and a large increase in sluggishness; and a decrease in sluggishness for
smaller firms. This is expected since the coverage of job security restrictions
expanded to size100 industries in 1982. The liberalization package in 1991
generates a one-time boost to labor demand for large firms but no clear reversal in
sluggishness of labor adjustment.
For medium sized (size100) firms, competition increases average employment
and decreases sluggishness with opposing results for larger (size300) firms (lower
average employment and higher sluggishness). Effects of pre-existing labor
hoarding (requiring shedding of excess labor lowering average employment and net
employment change) may be responsible.
Industry data used with GLS and system-GMM estimation techniques provides
interesting results. Using the system-GMM estimators helps to instrument for
endogeneity in the labor demand equation. GMM estimates strengthen the GLS
results and provide additional results.
84
1
We take a broad view of public sector employment and associated retrenchment.
Public sector employment includes civil service workers, military personnel and
employees of public sector enterprises.
2
Important exceptions are Svejnar and Terrell (1991) which provides a very useful
comparison across six countries for retrenchment in the transport sector, and Lindauer
and Nunberg (1994) which provides a book-length discussion of civil service reform.
Our study benefitted substantially from the insights in both of these publications.
Chapter 4: Public Sector Retrenchment Programs
4.1. Introduction
The retrenchment of public sector employment is an increasingly pervasive
phenomenon.
1
Many advanced countries, developing economies, and transition
economies have recently faced the issue of downsizing of public sector employment.
The reasons underlying the downsizing vary considerably across countries. For some it
is a general move towards a more market economy, for others it is a reduction in the role
of the military or an attempt to reduce a bloated bureaucracy, for others the retrenchment
is sparked by a fiscal crisis necessitating a severe cutback in government spending, and
finally for some it is a combination of some or all of these. Given the pervasiveness of
the phenomenon, there is no shortage of studies of individual countries or episodes.
While these studies are quite useful, it is difficult to draw common lessons from these
country specific studies and there is relatively little analysis comparing the experiences
across countries.
2
In this paper, we compile and analyze information on the recent
retrenchment experiences across a large number of countries and retrenchment episodes.
Comparing and contrasting the experiences across countries is a formidable task.
85
3
Details regarding the collection of information is provided in section 3. A complete
listing of individual country summaries is available upon request.
4
The converse is not true i.e. the absence of rehiring is not synonymous with success.
Also, conceivably a restructuring could involve separating employees en masse and
subsequently rehiring those with necessary specific skills or rehiring at temporary terms
or both. No program in our sample intended such purposive rehiring.
Beyond the usual problems of cross country comparisons, there are no central databases
with the requisite information compiled that permits such analysis. Accordingly, we
compiled and collected our own information from a myriad of sources but primarily from
internal World Bank documents and interviews of staff members at the World Bank with
operational information about retrenchment experiences in individual countries.
3
Our
analysis is based upon 37 countries and 41 programs. Most of the programs for which we
could collect detailed information commenced in the early 1990s and many of the
programs are ongoing (which makes analysis of the ultimate success of the programs
difficult). The gross number of workers separated in all the programs we consider
exceeds 5 million, with a somewhat smaller net employment reduction. The discrepancy
reflects the fact that some programs exhibited significant rehiring of the separated
workers who departed and/or new hires by the public sector. In the analysis that follows,
we look closely at the characteristics of programs that involve significant rehiring and
new hires. While the latter may be part of a coherent plan to restructure the public sector
workforce, the former is clearly not an indicator of success.
4
Evaluating the costs and the benefits of the individual programs in a more
comprehensive fashion is inherently difficult. In principle, the requisite information
includes the compensation packages offered, the relative productivities and wages of the
86
displaced workers in the public and private sectors, as well as the adjustment costs borne
directly by the affected workers and the entire economy. Measurement of much of this is
well beyond the scope of available information, particularly the nature of the adjustment
costs. Nevertheless, we can document considerable details about the magnitude and
forms of compensation used in retrenchment programs, the wage bill savings, and
characterize some of the other aspects that are relevant for evaluation. To summarize the
available information on costs and benefits, we calculate a simple financial break-even
indicator measured as the number of years required for the present value of financial
costs to equal the present value of financial gains. We also examine refinements of this
indicator that attempt to incorporate the productivity gains generated from the
retrenchment. In turn, we relate these summary financial indicators to other program
characteristics.
The chapter proceeds as follows. Section 2 provides a brief presentation of the
underlying conceptual framework for characterizing the private and social costs and
benefits of a public sector retrenchment program. The analysis in this section is
deliberately simple and borrows heavily from the existing literature. Our objective here
is primarily to provide guidance for the type of information that is required to compare
and analyze alternative programs and to discuss the interaction between the conceptual
and measurement issues that must be confronted. Section 3 outlines the methodology
used to collect information for the individual countries and to compile the information in
a systematic fashion. Section 4 presents the analysis of the information collected as well
as some detailed discussion of individual countries to illustrate the patterns that emerge.
Section 5 contains concluding remarks.
87
5
For example, many of the conceptual issues discussed in this section are discussed more
formally and with greater attention to the range of relevant issues in Diwan (1993a,
1993b) and Rama (1997).
6
While we do not characterize all relevant sources of heterogeneity explicitly, differences
across workers in the following present values may reflect differences in ability,
experience, skills, horizons, discount rates and mobility costs.
4.2. Conceptual Framework and Measurement Issues
To provide some guidance for the type of factors relevant for comparing and
analyzing retrenchment programs across countries, it is useful to sketch out a simple
conceptual framework. This framework helps us organize our analysis and discussion of
the information we have collected. While the discussion in this section borrows heavily
from the existing literature,
5
our characterization of the relevant issues emphasizes some
points that are neglected in the literature. Further, since our ultimate objective is to
quantify the relevant measures using data across countries, our focus in this discussion is
on the interaction between the conceptual and measurement issues that must be
confronted.
Consider the relevant incentives (private and social) for the retrenchment decision
for the marginal worker. Let:
6
P
private
= Present discounted value of the worker's productivity in private sector
P
public
= Present discounted value of the worker's productivity in public sector
C
individual
= Present discounted value of adjustment costs born by the individual in
relocating from public to private sector (e.g., job search costs, relocation costs, time spent
unemployed).
88
7
The discussion here and the subsequent analysis primarily focuses on the reallocation of
workers from the public to the private sector taking into account the relevant
productivities, wages and costs of labor market adjustment. Public sector downsizing
also likely involves the reallocation of capital. Even though our focus is on the
reallocation of workers, the implications of capital reallocation and the interaction with
the worker reallocation deserves further attention.
8
Explicit modeling of the distinction between the present discounted value of earnings
and adjustment costs would require incorporating the fact that transitions from a public
sector job to a private sector job may involve several transitions and accordingly several
spells of unemployment. As emphasized by Hall (1995), a potentially important
explanation of persistence in unemployment rate dynamics is that separations tend to
beget further separations.
C
social
= Present discounted value of adjustment costs borne by society for the individual
to relocate from public to private sector (i.e., adjustment costs borne by individual plus
spillover effects e.g., congestion effects).
W
public
= Present discounted value of earnings of the worker in public sector.
W
private
= Present discounted value of earnings of the worker in private sector.
The principle of social optimality considered here is that all resources should be
allocated to their highest valued use (net of relocation costs given the existing allocation
of resources). Thus, it is socially optimal to relocate the marginal worker to the private
sector if:
7
P
private
- C
social
> P
public
(1)
An individual worker will choose to stay on a public sector job as long as:
8
W
public
> W
private
- C
individual
(2)
If workers are paid more than their marginal product in the public sector (i.e., W
public
>
P
public
), then it is easy to imagine that both (1) and (2) hold simultaneously -- that is, it is
89
9
It may be that individual workers and the government discount the future differently
because of differential access to capital markets. Such differential access to capital
markets is, in principle, consistent with the specification and associated discussion
considered here but explicit consideration of the role of capital markets deserves more
attention in this context. For example, unemployment benefits and other forms of worker
safety net assistance are often justified as a form of social insurance in the face of
imperfections in the capital market. Since worker safety net enhancements are often part
of a retrenchment program, explicit consideration of the role of differential capital market
access is quite relevant.
in society’s interest for the marginal worker to relocate but it is not privately optimal.
9
A number of additional factors are potentially important in evaluating the
optimality of a retrenchment program. One factor is the expenditures incurred to pay
public workers wages or alternatively the compensation packages offered to induce
workers to relocate voluntarily. If taxes are not distortionary and there is no distortionary
rent seeking behavior, these expenditures should be viewed as transfers and thus not
impact efficiency. Under these strong assumptions, these terms should not appear in
equation (1). However, in the presence of distortionary taxes (including the inflation tax)
and distortions from rent seeking behavior, such transfers yield efficiency losses. In
addition, an excessive wage bill and/or fiscal crisis may imply that other government
services are adversely affected by retaining redundant workers. To incorporate these
effects in a modified version of (1), suppose that there is a distortions-based loss function
L(.) which is an increasing function of the transfers to workers. Taking into account
these distortionary losses in (1) would yield that retrenchment is optimal if:
P
private
- C
social
- L(COMP) > P
public
- L(W
public
) (3)
where COMP reflects the present discounted value of the compensation or other
90
10
A related issue is whether the nature of the distortions from the transfers depends not
only the net present value of the transfers but also the timing. For example, rents in the
form of artificially high public sector wages may have dynamic distortionary effects
beyond those associated with a one-time transfer.
11
For further discussion of the congestion externalities among job seekers that might be
important see, e.g., Blanchard and Diamond (1989,1990). For further discussion of the
idea that employment and earnings losses by workers in one sector may generate
spillover effects to other sectors through final good demand spillover effects, see, e.g.,
Cooper and Haltiwanger (1996).
assistance programs that accompany the retrenchment.
10
One problem faced by policymakers (and shared in our analysis) is measuring the
components above in (1), (2) and (3). Many of the components are very difficult to
measure and/or difficult to observe in the presence of worker heterogeneity (e.g., outside
opportunities and individual productivity). Among the most difficult to assess are the
adjustment costs. Factors influencing adjustment costs include the degree of labor
market flexibility (i.e., barriers to adjustment in terms of hiring and firing costs), the
worker safety net, informal vs. formal sector development, and the method, scope and
speed of retrenchment. Social adjustment costs will be greater than the private
adjustment costs to the extent there are spillover or externality effects of worker
displacement. The spillover/externality effects include the impact of final goods demand
spillover effects from reduced consumption expenditures by affected workers, congestion
externalities among job searchers, and social disruption (e.g., national strikes) induced
by the affected workers.
11
The method, scope and speed of retrenchment potentially
influence these spillover effects. In addition, the concentration of the retrenchment in a
local community may imply important local spillover effects even if there are not
economy-wide spillover effects.
91
12
The nature of the measurement problems in evaluating pre and post retrenchment
productivity will vary depending on the type of public sector employment (i.e., civil
service, military, or public sector enterprises producing goods and services).
As noted, the nature and magnitude of these adjustment costs are very difficult to
measure and as will become clear we have relatively little quantifiable information on
adjustment costs that we can use in our comparison across countries. The most direct
evidence available are from recent studies of privatization in transition economies (see,
e.g., the studies in Commander and Coricelli (1995)) for which the massive restructuring
implies that adjustment costs take center stage. Even in these latter studies, the
information on the nature of the adjustment process is often quite limited. In general, and
particularly in these transition economies, characterizing these adjustment costs is at the
heart of the debate regarding gradualism versus a big push in the efforts to reduce the role
of the public sector in the economy.
Another difficult measurement/conceptual problem is evaluating the pre and post
retrenchment productivity of workers in the public sector. While there may be
widespread agreement that there are redundant public sector workers, quantifying this is
extremely difficult.
12
Further, some of the need for retrenchment may reflect more of a
need of restructuring than simply downsizing. For example, low morale and productivity
of public sector employees may reflect poor organizational structure and/or wage
compression. Thus, a program may involve organizational structure and wage structure
changes that appear to be financially expensive but are well-motivated by such factors.
This raises concerns about the use of simple financial indicators that do not adequately
incorporate such factors to evaluate the need and success of a program. In terms of (3),
this discussion implies that the public sector productivity term needs to be broadly
92
interpreted. That is, retrenching the marginal worker will yield the direct loss of the
worker’s output (if any) but may also yield productivity gains from the reorganization
that accompanies downsizing.
Beyond the above problems of measurement and assessment, it is often politically
necessary to implement the retrenchment scheme via a voluntary program. This
necessitates the provision of some incentive payment in form of severance pay, pensions,
etc. (denoted INCENTIVE below) such that for an individual worker:
INCENTIVE > W
public
- (W
private
- C
individual
) (4)
In principle, with complete information about worker heterogeneity, it is possible
to design an optimal incentive scheme for each worker (see Diwan 1993a, 1993b and
Rama (1997) for a formal analysis). One well-recognized problem that immediately
emerges is that if a simple, common incentive package is offered to all workers,
heterogeneity across workers implies that only those with lowest rent will depart. It is
important to emphasize that some aspects of the selection process may be unfavorable,
while others may be advantageous. For example, in environments with wage
compression, a common package offered to a wide class of workers implies that the most
skilled and capable workers will leave. Alternatively, individuals with better outside
opportunities and/or low adjustment costs are more likely to leave -- other things equal,
this voluntary aspect of selection will be favorable.
The implication of this discussion is that an optimal incentive scheme must be
individually tailored to reflect worker heterogeneity in rents and adjustment costs but that
this is difficult to implement in practice in the face of imperfect information. This
tension is at the center of the debate about the pros and cons of voluntary and involuntary
93
13
For a discussion on optimal schemes using either different types of severance contracts
or alternatively auctions, see Levy and Mclean (1996).
retrenchment programs. As noted above, voluntary programs have the advantage that
they yield some favorable voluntary selection of workers with good outside opportunities
and/or low adjustment costs without requiring the policy maker to have complete
information. However, workers with good outside opportunities are likely the most
productive public sector workers so that the voluntary selection may be adverse. The
problem with adverse selection is apt to be especially severe in environments with wage
compression
13
. Involuntary programs potentially permit specific targeting of groups of
workers on observable characteristics but don’t take advantage of favorable voluntary
selection on unobservable worker characteristics. Further, involuntary schemes may
also yield high adjustment costs. For example, mass involuntary layoffs may involve
substantial private and social adjustment costs and in this case the nature of the safety net
takes on particular importance. Finally, the political will (ability) to undertake and
sustain schemes with an involuntary component may be lacking.
Putting these pieces together suggests that our comparison and analysis of
alternative programs should consider the following factors. First, we must consider the
factors leading to retrenchment (fiscal crisis, overstaffing, morale problems given wage
compression, etc.) since they may provide insights about the relationship between wages
and productivity in the private and public sectors. Second, several factors influence the
adjustment costs including the scope and speed of retrenchment, the mechanism used
(involuntary/voluntary), whether the scheme involves targeting (e.g., skill or age biased),
and the nature of labor market flexibility and the safety net. Many of these same factors
94
14
Type of the World Bank documents included: Staff Appraisal Report (SAR),
Memorandum and President's Recommendation (MOP), President's Report (PR),
Supervision memorandum, Project Completion Report (PCR), Implementation
Completion Report (ICR), Project Performance Audit Report (PPAR), Operations
Evaluation Studis, Country Economic Memorandum (CEM), Sector and other economic
reports.
are important (e.g., the method used and the nature of targeting) in terms of the need for
individual tailoring of plans given worker heterogeneity. Finally, the magnitudes of the
financial costs (e.g. W
public
and INCENTIVE) are relevant to characterize the magnitude
of the transfers.
4.3 Data Collection
The survey of public sector retrenchment programs across developing and
transition countries was carried out on two dimensions. The first involved collection of
internal World Bank documents relating to the macro-economic and public sector
adjustment in these countries. A preliminary portrait of the issues relating to each
country was drawn using myriad types of World Bank Documents
14
. These were, in
turn, supplemented by a variety of external sources. The second step was interviewing
World Bank officials associated with the retrenchment programs to obtain more direct
information and assessment. The interviews typically yielded further acquisition of
relevant documents. The countries, the retrenchment programs and the associated
officials were selected using the World Bank electronic management information system
(MIS) and the adjustment lending database called ALCID. See appendix Table 4A.1 for
a list of surveyed programs/countries. The projects and loans listed on the World Bank
95
15
The selection of programs was mostly restricted to this list. In general, World Bank
staff mobility with movements away from early programs reduced usefulness of such
interviews.
16
The detailed country summaries total more than 200 pages including extensive
references to documents, papers, and other sources beyond those cited in the paper.
MIS go back to mid-eighties.
15
The interviews were conducted for more than eight
weeks during the summer of 1995.
The interview transcripts, documents, reports and articles relating to each
program were synthesized and summarized using a uniform outline. Information
presented relates to four broad aspects - characterizing retrenchment, the nature of labor
turnover and institutions, cost-benefit analyses in financial and economic terms, and
monitoring and evaluation of the program. A complete listing of the individual country
summaries are available upon request.
16
As will become apparent, the information collected and assimilated ranges from
quantitative information about the scope, speed, and financial costs and benefits to
qualitative information about the factors precipitating the retrenchment as well as
characteristics of the program such as the methods used and the productivity gains
observed in the public sector. The quality and completeness of the information collected
varies substantially across programs. This blend of quantitative and qualitative
information yields that the analysis that follows is primarily descriptive -- an attempt at
summarizing the basic facts and drawing out observable patterns. Not surprisingly, a
host of institutional and idiosyncratic factors appear to be important as one learns the
details of each individual program. These institutional and idiosyncratic factors serve as
an additional important caution for interpreting the analysis of basic patterns that follows
96
17
As of February 1997, the World Bank allowed lending for severance payments provided
it entails no organization closure.
in sections 4.4.1-4.4.5.
4.4 Survey and Analysis of Cross country Evidence
4.4.1 Basic Facts
We begin the analysis of the information we have collected by characterizing
some basic facts about the programs for which we gathered information. Tables 4.1-4.2
present summary statistics about the programs we surveyed (summary information on
each program is provided in Table 4A.1 in the appendix). As noted in the introduction,
our analysis is based on 41 programs in 37 countries. The World Bank has not
historically provided direct assistance for public sector employment retrenchment
programs
17
. However, under the umbrella of a comprehensive assistance package, an
agreement with a country often stipulates that the country use domestic counterpart funds
for specific purposes. Under such umbrella agreements, public sector retrenchment may
be part of the overall package. In our survey, about 65 percent of the programs had this
“umbrella” connection to World Bank financing.
Table 4.1 makes clear that total separations are greater than the total employment
reduction reflecting rehiring and new hires. Most programs did not experience rehiring
of the same workers that had departed but a nontrivial fraction (20 percent) experienced
significant rehiring. New hires are also relatively rare ( about 13 percent of programs).
Some programs represented only good intentions with no workers actually separated.
One of the programs included in our survey is Hungary with a massive public sector
97
18
The total cost is driven in part by enormous increases in pension and safety net
expenditures in Poland. The increases for Poland reported in Table 4A.1 are, from all
accounts, real but they do skew totals and averages somewhat. However, much of the
subsequent analysis is not sensitive to outliers on this dimension. That is, much of the
analysis is based on the percent of programs that exhibit various characteristics.
retrenchment program. This program is a clear outlier on a number of dimensions
(discussed below) but we felt it was important to include large programs among the
transition economies as points of contrast. The method of employment reduction is
categorized into three basic categories: involuntary (hard) refers to layoffs, involuntary
(soft) refers to employment reductions generated by strict enforcement of rules such as
mandatory retirement and the removal of ghost workers, and voluntary refers to programs
in which employment reductions were achieved through workers voluntarily quitting
(e.g., early retirements). As is evident from the table, involuntary-hard reductions
dominate total employment reductions. However, this is primarily driven by the massive
involuntary reductions in Eastern Europe. In contrast, the use of voluntary and
involuntary-soft measures are the prevailing patterns in Latin America, Asia, and Africa.
Table 4.2 indicates that the total financial costs of the program are large
(exceeding 12 billion dollars) with the average program having total cost of 400 million
dollars.
18
The financial costs took a variety of forms including severance payments,
increased pension liabilities, and enhancements to the worker safety net (which we use a
shorthand term of the various worker assistance programs including unemployment
benefits, job search assistance, training, and relocation assistance designed to aid in the
process of worker reallocation). As is clear from the minimum and the maximum, the
scope and mix of the packages varied considerably across countries. As the patterns
98
across continents show, the transition economies in Eastern Europe relied heavily on the
worker safety net, while countries in the Southern Hemisphere and Asia relied much
more heavily on direct compensation (severance pay and enhanced pensions, i.e.,
payments made directly to workers) to workers. The latter pattern is particularly striking
for severance pay (Asia-92 percent, Latin America-74 percent, and Africa-51 percent of
total costs). In section 4.4.6, we return to this observation in the context of discussing the
likely differences in the nature of labor market adjustment costs across countries.
A wide variety of circumstances led to the retrenchment programs across
countries. The most prominent factors leading to retrenchment were fiscal crises and as
part of a general effort to reduce the size of government in the economy. However, in
some cases, the compelling factors appeared to be more structural problems with the type
and mix of government workers. Wage compression among public sector workers
leading to morale and staffing problems was a common complaint. Overstaffing
(including the problem of “ghost workers” -- workers on the payroll who did not exist but
someone was collecting the paycheck) was another common complaint. Finally, the
downsizing of the military played a prominent role as well. Not surprisingly, many
programs had multiple factors leading to the retrenchment.
The differences in the factors generating retrenchment are closely linked to the
mix of packages used across countries and continents. For some countries, the
retrenchment episode is viewed as a one-time event correcting a perceived relatively
narrow problem (e.g., ghost workers, productivity/morale problems due to wage
compression) in a particular public sector agency or enterprise. These one-time event
cases typically used some form of direct compensation (i.e., severance and enhanced
99
19
The inclusion of Hungary in our 41 programs has a clear influence on employment-
weighted results. For the most part, our results are depicted on a unweighted basis so that
outliers such as Hungary do not play a disproportionate role.
pension) and accordingly were more likely to use voluntary methods of retrenchment. In
contrast, in some countries, the public sector retrenchment is part of a fundamental
change in the role of the public sector in the economy (e.g., the transition economies).
In the latter cases, there is typically much greater attention to institutional changes in the
programs (e.g., unemployment benefits, relocation assistance, training assistance) that act
as the safety net for worker reallocation.
Much of the detailed information we collected involves the method used for
employment reduction, the nature of compensation provided and the degree of targeting.
Summary information on these characteristics is provided in Tables 4.3-4.5. Many
programs (but less than half) used both involuntary and voluntary reduction methods.
Relatively few programs used all three methods. While voluntary programs appear to be
the most popular in terms of the percentage of programs with a voluntary component,
most employment reductions were actually achieved via involuntary methods
19
. The
removal of ghost workers played a relatively minor supporting role in the employment
reductions but was quite prevalent amongst the African countries in our survey.
Table 4.4 indicates that the primary form of direct compensation to workers was
some form of severance payment. Relatively few programs offered no direct
compensation. Most programs involved some enhancement of the overall safety net
intended to assist unemployed workers and workers attempting to relocate. An important
component of the safety net enhancement was often the stipulation of some form of
training assistance.
100
As discussed in section 2, the underlying theory suggests that designing a
program that targets specific individuals or groups of individuals is likely to be important
to avoid adverse selection and to promote favorable selection. We attempt to summarize
the many different approaches to targeting used in practice by classifying programs into
three admittedly crude groups: skill-biased, age-biased and neutral. Skill-biased
programs are those that restricted the program along detailed occupational or skill
groupings. For example, the restructuring of the workforce in the tax administration in
Peru was based on new and continuing workers passing a written test. Age-biased
programs are primarily those that focused on voluntary retirement (and associated
pension enhancements) as the means for retrenchment. Neutral programs are those that
offered a simple, common package to a wide class of workers with little or no attempt to
target specific groups. Some programs are classified as both skill-biased and age-biased
if they used a mix of packages that induced selection on both types of criteria.
Table 4.5 indicates that targeting on the basis of skills or age (in the latter case
primarily through early retirement programs) is important but not pervasive. Almost 20
percent of the programs used no targeting of workers. As we will see in the following
sections, the nature of targeting is closely related to a variety of measures of “success” of
the program.
4.4.2. The Relationship Between Rehires, New Hires and Other Program
Characteristics
Based upon the discussion in section 2, tailoring a program to individual
characteristics using some form of targeting is likely to be important to avoid losing the
most productive and or essential workers. The absence of targeting may yield excessive
101
20
In many programs, workers are rehired by the same organization/unit. In some cases,
they are rehired by a different branch of the public sector.
losses of workers in key areas and/or with key skills and thus necessitate either rehiring
some of the same workers
20
who departed (a sure sign of poor targeting) or new hiring.
Using the information we have collected, we have examined the simple bivariate
relationships between the likelihood of rehires and new hires and other program
characteristics (including the nature of targeting).
Table 4.6 examines the link between rehiring and program characteristics. The
most striking results involve the role of targeting and the type of compensation offered.
Programs with rehiring are much less likely to use skill or age targeting. For example,
only 25 percent of programs with rehiring also used skill targeting while more than 60
percent of programs without rehiring used skill targeting. The flip side of this is that
more than 60 percent of the programs with rehiring did no targeting while less than 10
percent of programs without rehiring did no targeting. These findings are consistent with
the view that programs that fail to target yield severe adverse selection problems with the
most critical workers departing and the subsequent need to rehire these same workers.
Some other interesting patterns emerge that are large in magnitude but not
statistically significant at conventional levels given the relatively small sample size (i.e.,
41 observations on individual programs). Specifically, programs with rehiring were also
less likely to have a safety net component and in particular some type of training
component. The potential inference here is that the pressure to rehire may be greater if
efforts are not made to assist workers in the transition to the private sector.
Table 4.7 examines the analogous relationships between new hires and program
102
21
As noted in the introduction, rehiring of the same workers that have been separated
could be part of a coherent plan of restructuring. For example, it might be difficult
politically to target specific workers for retrenchment. Given constraints on the ability to
target workers, the optimal strategy might be to induce separations en masse and then
rehire specific workers. We found no evidence for such purposive rehiring.
characteristics. The largest and most significant results again involve the nature of
targeting and the type of compensation assistance. Programs with new hires are all age-
biased with associated pension enhancements and do not include safety net components.
It is not clear, of course, that evidence of new hiring should be interpreted or treated in a
like manner with rehiring. Rehires of the same workers who departed is much more
likely an indicator of problems with the program but new hires may involve some
intended restructuring of the workforce.
21
The results on age targeting and pension
enhancements suggest that for some programs the intent was to replace older, incumbent
workers with new workers.
4.4.3 Summary Financial Indicators
One summary financial indicator that we consider is a measure of the simple,
financial break-even period reflecting the number of years that a program requires before
breaking even on financial costs and benefits. The break-even measure is calculated as
follows. Financial costs are the present discounted value of the compensation package or
safety net expenses incurred. This present discounted value is in principle calculated
over an infinite horizon. In practice, for many of the cases, the financial costs are front-
loaded via severance payments so this present discounted value is easy to calculate.
However, in cases in which the costs are in the form of continuing safety net or pension
liabilities, we generated appropriate present discounted value measures of the financial
103
22
For some programs, the evaluation of the program by World Bank staff had already
yielded such calculations.
costs.
22
Financial benefits are measured as the presented discounted value of the wage
bill savings from the retrenched workers. The annual wage bill savings are assumed to
be constant over time and equal to the wage bill savings relevant at the time of the
implementation of the program. These assumptions permit direct calculation of the
break-even period as the number of years for the present discounted value of wage bill
savings to equal the overall presented discounted value of financial costs.
For purposes of comparability of measures across countries, a common (10
percent) discount rate is used. Under these assumptions, the break-even period is a
scale-free financial indicator that permits comparing and contrasting the financial costs
and benefits across programs. Information was sufficient to calculate the break-even
measure for about 40 percent of programs.
As seen from Table 4.8, across the programs we surveyed, a non-trivial
fraction (22 percent) yielded net losses essentially implying an infinite break-even
period. The presence of net losses reflects in part the impact of rehires or new hires or
rising compensation for retained workers. Accordingly, some programs (e.g., Peru
SUNAT) yielded no wage bill savings given the rehires, new hires and/or increase in
compensation. For the programs for which we could calculate a finite break-even period,
the average break-even period is 2.27 years with the median 1.82 years, the maximum 10
years and the minimum 0 years. Programs with immediate break-even periods (zero
years) are those with involuntary reductions without direct compensation or other
assistance programs
104
How should this simple financial indicator be interpreted? If nothing else, it
provides an indicator of the financial viability of the programs and Table 4.8 indicates
that many of the programs yield relatively rapid financial payoffs. However, as discussed
in section 2, we would ideally like to quantify the present discounted value of all costs
and benefits from retrenching public sector workers and evaluate programs accordingly.
Where does this simple financial indicator fit into such a calculation? To address this
question, it is useful to return to equation (3). For purposes of discussion, suppose we
treat the loss function in (3) as a simple linear function such that a dollar of expenditures
on public sector programs yields a dollar in distortionary losses. Then, equation (3) can
be written as:
W
public
- COMP + P
private
- C
social
- P
public
> 0 (5)
Thus, we see that the break-even period measure is essentially providing an evaluation of
the program based upon only the first two terms of equation (5). A comprehensive
evaluation, even in these crude terms, requires calculating all of the terms in equation (5).
Unfortunately, data limitations imply that we cannot measure all of the
components of (5) and thus cannot literally evaluate programs on this basis. However,
for some programs we can go a few steps further which is instructive. In particular, for
some programs we can estimate the wages that retrenched workers received in the private
sector. Assuming that workers are paid their marginal product in the private sector,
ignoring adjustment costs and assuming that productivity of the redundant (retrenched)
workers is zero in the public sector, permits a crude implementation of the terms in (5)
using this additional information. In particular, for programs that we had this
information, we calculated a modified break-even measure (denoted for labeling purposes
105
23
Sufficient information was available to calculate the payback period indicator for 20
percent of the programs.
the “economic” payback period) using this additional information. Literally, the
“economic” payback period is the number of years it requires for the benefits from the
reduced public sector wage bill to be such that present discounted value of costs equals
the presented discounted value of benefits. For this calculation, the costs and benefits for
all components other than the public sector wage bill are calculated over an infinite
horizon. In addition to measuring benefits based upon retrenched workers’ earnings in
the private sector, for programs with measurable increases in productivity in the public
sector (e.g., PERU SUNAT) we included such effects in the benefits. As seen from Table
4.8, the average calculated economic payback period is 2.1 years and the median is 0.8
years across the programs for which we could generate the measure.
23
Part of the motivation for calculating comparable summary financial and related
indicators across programs is to permit relating these financial indicators to other
program characteristics. Table 4.9 characterizes the relationship between programs with
net financial losses and other program characteristics. Programs with a net financial loss
are much more likely to include a voluntary component than programs without financial
losses. Interestingly, programs with net financial losses are more likely to use targeting,
particularly age targeting through pension enhancements. This result indicates that
targeting may be expensive, at least measured in simple, financial terms. Table 4.10
provides a related look at the relationship between the magnitude of the break-even
period and other program characteristics. Given that these are at best crudely calculated,
we divided programs into two groups -- those in the lower and upper tail of the
106
distribution relative to the median break-even period (which is 1.82 years). The results
from Tables 4.10 reinforce those from Table 4.9. Programs with relatively high break-
even periods are more likely to involve a voluntary component, more likely to use
targeting, and more likely to use direct compensation (either through severance payments
or pension enhancements).
Interestingly, some of these patterns change substantially when we examine the
payback period distribution. As seen in Table 4.11, once we incorporated even crude
information about productivity gains into the calculation, we no longer find that targeting
is especially expensive in terms of implying a high payback period. Not surprisingly, we
still find that programs with high calculated payback periods are more likely to involve
direct compensation and in particular severance pay.
To sum up, simple calculations of financial break-even periods indicate that many
of the programs have relatively rapid financial payoffs. Alternatively, a non-trivial
fraction are financial losers. However, caution must be used in interpreting such
financial indicators as measures of relative success of programs. There are many relevant
components of the private and social costs and benefits not captured by these measures.
Even the limited attempt we made to incorporate relevant economic costs and benefits
indicates that the inferences based upon narrowly defined financial indicators may be
quite misleading.
4.4.4 Highlights of Individual Programs
We now turn to summarizing the experiences for a selected set of individual
programs. The objective here is threefold. First, consideration of individual programs
provides a more comprehensive view of the scope and heterogeneity across programs.
107
24
In general, productivity measurement in service sector, specially public sector is
difficult. Most work uses quantitative or technical measures. See Griliches (1992).
Second, the discussion of individual programs permits illustration of some of the key
patterns detected in the discussion of basic patterns. Third, our summary tables and
related analyses admittedly leave much out. Part of the reason for this is lack of data
suitable for quantifying effects in a summary fashion. For example, quantifying the
nature of productivity gains associated with retrenchment as well as the adjustment costs
is difficult both in principle and in practice given available data
24
. However, in some
cases, qualitative information about these issues are available which sheds some
additional light.
Programs in six countries (Peru, Argentina, Uganda, Ghana, India, and Hungary)
are highlighted. Key characteristics are presented in Table 4.12. In the discussion, the
ordering of the country/program cases reflects a rough attempt to select cases that
highlight the role of the following issues: (i) targeting; (ii) productivity gains; and (iii)
worker safety net and labor market adjustment. This ordering is only approximate since
all of these issues are present for every program. In the summary statistics in Table 12,
bold entries in the table highlight the relevant key issue that motivates our inclusion of
this program for special discussion.
PERU: Two programs were initiated as part of a broader exercise of fiscal
austerity and adjustment. Among the problems were the under-funded pension system,
the posting of “ghost” employees by the regions to enhance employment based federal
transfers, and erosion in public salaries making it difficult to hire and retain qualified
workers.
108
25
The primary source of information on these two programs is a World Bank (principal
external donor for both projects) document: World Bank, 1994. Peru Public Expenditure
Review. Report No. 13190-PE
26
Further, federal government staff reductions were offset by increases in employment by
regional governments with most of it re-hiring of erstwhile federal staff.
With the support of external donors, Peru initiated two labor adjustment
programs in 1991, one for the Peru civil service and the other for the Peru tax
administration authority called SUNAT. These were respectively completed in 1993 and
1992.
25
The civil service program used all three involuntary and voluntary employment
reduction methods - separating about 250,000 workers over three years. Induced
departures employed both lump-sum severance and pension enhancements. Little
evidence on skill targeting was found. Targeting by age was implicit in the use of
pension enhancements to induce voluntary separation. Poor targeting aggravated the
shortage of human resources with many of the most qualified staff leaving. The poor
targeting and accompanying shortages probably led to the significant rehiring of the
separated workers.
26
In total, 163,000 workers of the originally retrenched workers were
rehired. Severance packages of about $1000 were provided to less than half (112000) of
the workers separated. This limited the direct financial losses associated with the
significant rehiring. Our simple measure of the break-even period for this program was
2.6 years, more than the median of 1.82 years for all the programs studied. However, this
measure does not reflect the loss of productivity associated with the shuffling of the same
workers in and out of the same positions.
Interestingly, the other program in Peru appears to be a model of good targeting.
The SUNAT (tax authority) program also used a mix of voluntary and involuntary
109
27
Deflating wages using the CPI instead of the exchange rate returns an increase of
similar magnitude. The instantaneous increase would drop from 20-fold to 18.6 times.
28
This discussion neglects the welfare costs (e.g., reduction in hours and output) due to
increased enforcement of tax collections.
reduction methods. Voluntary separations came with an enhanced pension. Involuntary
workers were selected on the basis of a written test. Thus, targeting was worker specific
and objectively determined. Two-thirds of the workforce (2034 workers) was separated.
Subsequently, SUNAT hired 1309 new workers, again based on a written test. Since
productivity and competence were objectively established and necessary skills acquired
afresh, little basis remained for rehiring (skilled but severance-induced separated)
workers. Rehiring was barred for ten years and none was found. The average salary for
affected workers increased from US$50 per month to US$1000 per month
27
. Tax
collections more than doubled and so did SUNAT’s revenues (2 percent of tax
collections.) These were insufficient to cover the salary increase and the scheme incurred
a net financial loss of US$47 million in present value terms. However, the entire
increase/improvement in tax collections can be interpreted as substantial gain for the
government (and in principle the economy
28
). Incorporating the tax collection gains into
our calculation of the payback period yields a payback period of only 0.002 years. This
highlights the importance of evaluating program performance along multiple dimensions,
not just organization-level financial costs and benefits but the broader impact of the
program. It is worth noting that neither of these two programs emphasized job search,
relocation, training or other forms of assistance to reduce the adjustment costs for the
retrenched workers.
110
29
See Robbins (1996) for a case study on public sector retrenchment in Argentina since
1990. This paper examines issues of adverse selection and post-retrenchment worker
status using micro data.
30
World Bank, 1995. Argentina Public Sector Reform Project Completion Report. Report
No. 14781.
31
In the source reports (Reports no. 14781, 14809 and the World Development Report
1994,) we found no evidence of rehiring. However, some anecdotal evidence indicates
occurrence of rehiring.
ARGENTINA: With external assistance from the World Bank, Argentina
undertook programs for the federal administration and the railroads (as part of a broader
program for public enterprises) during the years 1990-92 and 1991-94 respectively
29
.
Given chronic public sector deficits and endemic inflation, extensive public sector reform
was considered essential. Wages accounted for 70 percent of federal expenditure
excluding interest payments and transfers. Employee redundancy level was estimated at
50 percent
30
. Thus, staff restructuring was an important component of overall reform.
The federal administration program retrenched 400,000 workers. No explicit
targeting by worker was found. Involuntary separations implied some implicit targeting
in the selection of redundant workers. Some evidence indicates an age bias. Also,
targeting by function was attempted in that the federal government’s role was restricted
to the provision of security, social services and economic management. Rehiring was
barred by law and none was found
31
, despite the large scale of retrenchment and the
absence of targeting. The railroads’ program separated about 73,000 workers using
multiple methods. It used both voluntary and involuntary reduction methods. An
important feature of Argentinean programs is the heterogeneity in compensation
amounts. In federal administration, average severance payment per worker was US$
111
32
World Bank, 1995. Argentina Public Sector Reform Project Completion Report. Report
No. 14781.
33
World Bank, 1995. Public Enterprise Adjustment Loan Audit Report. Report No.
14809.
3000 amounting to a one-time cost of US$ 425 million
32
. In railways, the average
severance per worker was US$12000 or a one shot payment of US$ 360
33
million in
railways . Annual wage bill saving was expected to be US$1000 million in the federal
service and US$238 million in the railroads. Calculated break-even periods were 0.4
years and 1.6 years respectively in the federal administration and the railroads, with the
difference reflecting the higher compensation per worker in railroads.
While difficult to measure, some evidence suggests that the restructuring in both
railways and federal administration was characterized by substantial productivity gains.
Railways experienced an increase in freight miles per worker and passenger miles per
worker of about six to seven times. In federal administration, a 50 percent cut in
processing time is claimed.
UGANDA: High fiscal deficits combined with inadequate pay levels and wage
compression led to a civil service labor restructuring program in 1992 - 94.
Concomitantly, the military reduced size in line with peace-time national priorities and
conducted a program in 1992 - 95. Both were partly supported by external financing.
The civil service program employed all three reduction methods restricting the package
offer to a fraction of the workers separated. Separations were targeted by skill and tenure.
Some critical jobs and personnel were “ring-fenced” that is, protected from retrenchment.
Such workers did not have the option of obtaining enhanced benefits associated with
112
34
Republic of Uganda, 1994. Management of Change, Context, Vision, Objectives,
Strategy and Plan: Civil Service Reform.
35
Although an evaluation of the civil service staff performance was not available at the
time of our data collection, measurement indicators to evaluate the performance
improvements were planned by the EDI of the World Bank.
36
World Bank, 1995. From Swords to Ploughshares: The Uganda Demobilization and
Reintegration of Ex-combatants in Uganda. Confidential Mimeo.
voluntary separation. The gross number of separations was approximately 150,000. The
average compensation per worker was US$320
34
or about twice the annual GDP per
capita. However this total expense of US$ 16 million constituted a measured net
financial loss since there was no saving on the wage bill. The wage bill increased due to
salary revisions for the remaining civil service staff and was intended to improve service
performance. Interviews with program advisors revealed a strong motivation for the
salary hike, viz., to achieve parity between the civil service average wage and the living
wage in Uganda so as to motivate the civil service staff and generate productivity
increases.
35
The military program was a post-war retrenchment exercise. It used both
voluntary and involuntary reduction methods to reduce strength by 44000 employees.
Skill targeting was based on soldier performance and military requirements. Average
compensation was US$955, provided partly in kind and distributed over several months.
The calculated break-even period was 2.7 years. In Uganda, some measurement of post-
retrenchment earnings for veterans was available. Actual post-separation earnings found
for the military programs
36
were up to 71 percent of GDP per capita in Uganda.
Including these earnings of a total of US$ 5.4 million per year in Uganda’s case yields a
113
calculated payback period of 1.2 years. The rarely observed actual data on post-
separation earnings is interesting, particularly given that the earnings are lower than the
average. While this evidence is too idiosyncratic to draw general inferences, it does
highlight the difficulty in generating the appropriate cost-benefit calculations. Using
average earnings in the private sector as a proxy (even when this is available ) is
obviously inappropriate in this case.
An interesting feature of the military program was the safety net nature of the
compensation that enabled veterans to reintegrate into civilian society. Assistance in kind
was tailored to the intended post-separation occupation, for instance farmers were
provided land and farm implements. Further the assistance was known to be temporary
(six months), creating an incentive for the veteran to hasten adjustment and thereby
limiting program costs.
GHANA: Motivated by a very large public sector wage bill as a fraction of total
government expenditures, the Government of Ghana initiated a program for the civil
service in 1987 with partial-financing from the World Bank. The program concluded in
1992. It was driven by voluntary departures induced by additional severance and
complemented by soft involuntary measures - removing ghost workers and enforcing
retirement age. Voluntary departure, with compensation, was conditional on
employment being non-critical to the performance of the unit. Necessary skills were thus
sought to be protected. 73,000 workers were separated. More than three-fourths of the
total workers separated received the package. Average compensation was US$700. The
formula of two months’ base pay for each year of uninterrupted service exceeded the
legally required four months’ base pay in most cases. Anecdotal evidence indicated
114
37
Alderman, H., S. Canagarajah and S. Younger (1994).
some rehiring in subvented organizations (attached to ministries but not covered under
budgetary process.) The calculated break-even period was 1.8 years. Despite a
generous package offer, the BEP was relatively low.
Regarding assistance to workers to making a transition to the private sector,
extensive training through Program of Action to Mitigate the Social Costs of Adjustment
(PAMSCAD) was planned. However assistance for job search & placement and for
retraining was inactive. Courses in entrepreneurial development were offered. For
potential farmers inputs including land were provided. In rural areas retrenchees were
also eligible to participate in Food-for-Work schemes. A sample survey of the retrenched
workers
37
found that about 10 percent of the retrenchees quit the labor force. Of the rest
97 percent were re-employed by the second year, about 20 percent in formal sector wage
jobs and the rest in self-employment or informal sector jobs.
INDIA: Fiscal and external payments’ deficits led India to initiate a stabilization
and structural adjustment program in 1991. A program to support public sector labor
restructuring was required. This retrenchment support program was directed at public
enterprises declared to be sick (with several years’ of accumulated losses). A voluntary
retrenchment program in the sick public sector textile firms separated about 70,000
workers over more than two years (1993-94). The average cost per worker was about
US$17000. The formula used was 30 days’ wages for each year worked compared with
the legally required 15 days’ wages for each year of permanent service. The scheme
incurred a net loss of US$276 million in present value terms given the exorbitant
115
38
The high compensation reflects in part the effects of a rigid anti-retrenchment and anti-
closure law in India. For a discussion of this, see Basu, Fields and Debgupta, (1996).
39
PA Consulting Group, 1993. Social Safety Net Adjustment Credit Technical Assistance -
Final Report. Submitted to the World Bank.
40
See, Commander et. al. (1995).
compensation
38
. Given the incentives, the voluntarily departing workers may have had
long tenures increasing the severance amount granted. No explicit targeting mechanism
by tenure or age was found. On skills, the maximum skill level among the retrenchees
was that of a supervisor.
Results from a sample survey
39
indicated that all retrenched workers remained in
the labor force and 80 percent were re-employed. Of the total surveyed, 32 percent were
in wage jobs, one-fourth of them in the same industry, that is, textiles. The rest (48
percent) were self-employed.
HUNGARY: Reform and stabilization programs aimed at a transition to a
competitive economic system led to mass dismissals in state enterprises during 1990-
1992. About 1.7 million workers constituting 8.7 percent of the country’s labor force
were separated. This is extremely large compared with other programs’ scope - Ghana
(0.2 percent of labor force,) India (0.01 percent.) The annual wage saving amounting to
US$298 million but was exceeded by the increase in the annual safety net expenditures of
US$858 million. The exercise was clearly financially costly but reflects a fundamental
change in the structure of the economy. A comprehensive set of institutions designed to
act as a safety net for worker reallocation is very common in western economies and
Hungary is obviously attempting to follow that model.
Surveys
40
indicate that retrenched workers remained in the labor force. Most of
116
the re-employed were in the private sector, in trade and service industries. The average
private manufacturing wage was found to be about 70 percent of state firm wage
(providing some indication of the size of the rent in public sector firms.) The nature of
the labor market adjustment appeared to change as the transition process proceeded and
accelerated. Early on, many of the retrenched workers left voluntarily and left the labor
force (e.g., via voluntary retirement). Others left voluntarily and transited directly to
other jobs. However, as involuntary layoffs increased, an increased fraction of the
retrenched workers entered unemployment. Moreover, the available evidence suggests
that unemployed workers had an increasingly difficult time finding jobs. Early in the
transition in February 1991, the outflow rate from unemployment implied a steady-state
duration of unemployment of 7 months. By November 1992, the implied steady-state
duration was 50 months.
4.4.5 Aggregate Factors
Another means of evaluating the programs is to look at relevant macroeconomic
indicators prior to and during the program including indicators of fiscal posture.
Examining such macro indicators provides a rough independent check of the factors that
lead to a retrenchment episode. Further, evaluating macro indicators prior to and during
the program provides another means of evaluating the impact of the program. In many
cases, the actual retrenchment program is too small by itself to have important macro
effects. An exception are the programs in Eastern Europe (e.g., Hungary) in which there
was massive public sector downsizing. However, even in cases where the specific
program we analyzed is too small by itself to have macro effects, the program may have
been part of a larger effort to restructure and downsize the public sector. Thus, the
117
41
Several sources were used for this purpose including the Penn World Tables, the cross-
country labor market database being generated by Rama (1995), the World Bank World
Tables, and the International Finance Statistics.
42
The data are not complete for all variables and all countries over this entire horizon.
The unemployment rate series is probably the worst in this regard. Considering
deviations from country-specific means mitigates the problems with limited data
availability. However, appropriate caution should be used in considering these results
(particularly for unemployment) with the additional usual caveats that there is likely
considerable measurement error in the aggregates available by country.
results that follow can be interpreted in this light.
For each of the programs we surveyed, we acquired data on the unemployment
rate, the real GDP growth rate, the ratio of the current budget deficit to GDP, the ratio of
domestic debt to GDP, the ratio of Foreign Debt/GDP and the ratio of government
spending to GDP.
41
To remove the influence of idiosyncratic effects across countries, all
variables are characterized in terms of deviations from country-specific means. Country-
specific means for each variable are calculated from data for 1980-92.
42
Table 4.13
presents the patterns of these variables for the five years prior to the initiation of the
retrenchment program in each country and during the program (e.g., for a specific
country if the retrenchment program began in 1990 and ended in 1992 the prior year
calculations reflect data for 1980 to 1989 while the calculations during the program
reflect data averaged over 1990 to 1992).
Unemployment rises in the years preceding a program and GDP growth rates fall
in the years leading up to a program. Interestingly, the unemployment rate appears to fall
somewhat during programs while GDP growth remains below average. These patterns
are roughly consistent with the view that retrenchment programs are often precipitated by
economic crises. The evidence on fiscal indicators is somewhat mixed at first glance.
118
43
The discussion that follows on the formal and informal labor market is inadequate on a
number of dimensions and does not reflect the rich body of research that has been
conducted on the role of formal, informal and rural sectors for labor market dynamics in
developing countries (e.g., see Mazumdar (1989) for an overview of this research). Our
point here is simply to highlight the potential importance of these considerations for
cross-country differences in the nature of labor market adjustment.
The government share of GDP rises steadily prior to the onset of a program and falls
during the program. Somewhat surprisingly, there is not an accompanying pattern for the
deficit to GDP ratio and the debt to GDP ratio. For both of these domestic fiscal
indicators, there is an improvement in the fiscal posture preceding the program. However,
this appears to in part reflect the rising foreign debt to GDP ratio that precedes a program
and continues during a program. Putting these pieces together is consistent with the view
that these retrenchment programs are often part of a more general austerity program that
are supported in part by foreign assistance.
4.4.6 Missing Pieces -- Measurement and Characterization of Adjustment Costs
The discussion in the above subsections provides a range of information about the
experiences of retrenchment programs. In terms of relevant factors for evaluation, one
of the biggest missing pieces is associated with the conceptual and measurement
problems in quantifying private and social adjustment costs. While quantification is
beyond the scope of readily available information, there are a few qualitative
observations that can be made regarding the cross-country variation in labor market
adjustment costs. Specifically, qualitative evidence suggests that the considering the
impact of the formal and informal sectors is important.
43
In countries for which the
informal sector plays a large role (e.g., Ghana), many of the retrenched public sector
workers found employment relatively quickly, albeit mostly not in the formal wage
119
sector. In contrast, in the transition economies, retrenched workers from public sector
enterprises are facing increasing difficulty leaving unemployment because of difficulties
in finding employment in the formal wage sector.
Stated very roughly, these observations suggest the following potentially
important difference across countries in terms of the relevant labor market adjustment.
One set of countries seem to yield relatively quick adjustment for retrenched public
sector workers with the informal sector absorbing retrenched workers. Other countries
seem to yield long adjustment with retrenched workers from public sector enterprises
experiencing long spells of unemployment while seeking jobs in the formal private
sector. These differences are further reflected in the mix of policies we observe across
countries. For example, in Africa and to some extent Latin America, many of the
programs we surveyed involved direct compensation and relatively little changes in the
formal worker safety net. In contrast, in transition economies, the resources for
retrenched workers have primarily been devoted to the worker safety net.
This contrast is likely overstated in a number of respects. First, employment in
the informal sector may involve substantial underemployment. Second, workers reported
to be officially unemployed are arguably often engaged in some unreported informal
sector work activity. Nevertheless, even this rough characterization highlights the
potentially important role of differences across countries in the structure of economies
and in turn the differences in the nature of labor market adjustment costs.
What information would we have liked to have to quantify adjustment costs? To
be fair, this question is difficult to answer in advanced economies as well as for
developed and transition economies. Nevertheless, from advanced economies (and some
120
44
See, e.g., OECD (1993) and OECD (1996).
transition economies) we have learned a great deal about labor market flexibility and
adjustment by examining the flows of workers and the flows of jobs (see, e.g., Davis,
Haltiwanger and Schuh (1996)). For example, some advanced economies exhibit high
inflows and outflow rates to unemployment (e.g., the United States and Canada) while
others exhibit low inflows and outflow rates from unemployment (e.g., France and the
Italy).
44
The variation in the inflow and outflow rates to unemployment translate into
striking differences in the importance of long-term unemployment across countries. In
the United States the percent of long term unemployed (less than 12 months) in the early
1990s was about 6 percent while the equivalent figure in Italy was 71 percent. Even this
limited and indirect evidence yields two immediate inferences for the analysis of public
sector retrenchment. First, the low outflow rates from unemployment and accompanying
importance of long term unemployment in some countries implies that the magnitude of
the adjustment costs are potentially very large. Second, the variation across countries is
substantial implying that evaluation of a specific program in a specific country will
depend critically on the nature of the labor market adjustment in the country.
4.5 Concluding Remarks.
This survey and analysis of cross country experiences with retrenchment reflects
the gathering and processing of a wealth of information about individual programs.
Details about the nature of the compensation packages, the employment reduction
methods used, the nature of targeting, the worker safety net components of the program,
the financial costs and benefits and the nature of adjustment (including the presence of
121
significant rehiring of the same workers) have been documented and examined. One of
the primary objectives and thus contributions of this paper is to gather all of this
information into one place in an easily accessible manner.
A number of interesting findings emerge from analysis of basic facts. Theory
suggests that individually tailoring programs to account for worker heterogeneity and to
avoid adverse selection problems is very important. Consistent with this view, we found
that programs that used targeting on the basis of skills and age, used multiple methods of
employment reduction, and used a combination of compensation packages that included
enhancements of the safety net for assisting the reallocation of worker were much less
likely to exhibit problems with rehiring. We also found that programs with this multi-
dimensional approach with targeting tended to be financially expensive. However, both
quantitative and qualitative information suggest that there is a potentially large payoff in
productivity gains and in reducing adjustment costs.
Our analysis of financial indicators indicates that there is considerable
heterogeneity in the financial viability across the programs. Many of the programs have
a rapid financial payoff in the sense that the wage bill savings from retrenchment quickly
cover the financial costs of the programs from worker compensation and assistance.
Alternatively, a non-trivial fraction of the programs are clear financial losers. However,
we emphasize that while simple financial indicators are of obvious interest, they are
inappropriate for evaluating the relative success of programs because they omit many of
the potentially relevant private and social costs and benefits.
Unfortunately, many of the relevant private and social costs and benefits are
difficult to quantify given data (and conceptual) limitations. Especially difficult to
122
quantify are the individual and economy-wide (social) adjustment costs that are incurred
as part of restructuring programs. The data required to assess these adjustment costs are
generally not available and there are accompanying difficult conceptual issues. This
problem has both micro and macro dimensions. First, on the micro side, little effort has
been provided to collect data systematically on the post-separation experiences of
retrenched workers. Second, on the macro side, characterizing and understanding the
nature of these adjustment costs requires an understanding of the myriad of factors (e.g.,
institutions) that affect the labor market adjustment processes within individual countries.
The large observed differences across even advanced economies on this dimension
provides prima facie evidence in support of the need for this type of information. Third,
in countries with massive public sector retrenchment episodes, the labor market
adjustment process is endogenous to the retrenchment program itself. Given the
importance of the issues, more resources should be devoted to developing the data
necessary to evaluate these adjustment costs.
123
Chapter 5: Conclusions
The direct measurement of adjustment costs establishes the existence of
substantial costs for workers. Weaker but disturbing evidence points to generational
effects where job prospects of displaced workers’ children are adversely affected.
The economy while growing seems unable to absorb displaced workers or
otherwise provide income security and a cushion against accompanying effects
such as on children education. Transition supporting institutions exist in some
measure but their coverage is restricted and where these exist, effectiveness is low.
Employers are adopting complex means to alter workforce in the backdrop of
restrictive law, increasing competition and changing market conditions. These are
legal restructuring using Voluntary Retirement Schemes (VRS), illegal
restructuring using lockouts, and disputed restucturing with resolution sought
through labor courts and tribunals. Though labor adjustment is taking place, the
methods are costly. Given cost-cutting pressures from increasing competition,
costly adjustment reduces the gains from such adjustment.
The econometric analysis establishes that adjustment is slow. There is evidence
that after job security restrictions in 1976, (i) there is a one-time fall (about 28
percent) in labor demand and (ii) a fall in the speed of adjustment (3 percent) and
124
additionally so (38 percent) for large industries. The extended coverage of the job
security regime in 1982 led to these effects becoming larger for covered industries
and slightly smaller for uncovered industries based on average employment size.
These effects are not reversed by the economic reform package of 1991. No
significant increase in speed of adjustment is evident post-1991 though there may
have been a one-time boost to labor demand. Though more adjustment and more
VRS programs are taking place post-1991, the speed of adjustment has not really
increased.
Deregulation begun prior to 1991 and liberalization from 1991 have increased
the extent and intensity of competition. Interacting competition measures with
adjustment variables sheds better light on the effects. Competition increases
average employment and reduces the sluggishness of labor adjustment or increases
the speed of adjustment for medium sized industries (size 100 and above by average
employment size) but the opposite holds for larger industries (size 300 and above
by average employment size). For the latter reuslts, pre-existing labor hoarding
may be responsible.
The survey and analysis of cross country experiences with retrenchment reflects
the gathering and processing of a wealth of information about individual programs.
125
Details about the nature of the compensation packages, the employment reduction
methods used, the nature of targeting, the worker safety net components of the
program, the financial costs and benefits and the nature of adjustment (including
the presence of significant rehiring of the same workers) have been documented
and examined. Both quantitative and qualitative information suggest that there is a
potentially large payoff in productivity gains and in reducing adjustment costs from
programs with a multi-dimensional approach. The analysis of financial indicators
indicates that there is considerable heterogeneity in the financial viability across the
programs - many of the programs have a rapid financial payoff but a non-trivial
fraction of the programs are clear financial losers. However, while simple financial
indicators are of obvious interest, these are inappropriate for evaluating the relative
success of programs because they omit many of the potentially relevant private and
social costs and benefits. Unfortunately, many of the relevant private and social
costs and benefits are difficult to quantify given data (and conceptual) limitations.
Should the job security regime be liberalized and consequent mandated
separation costs be reduced? The debate on this issue has been pending resolution
since the initiation of reforms in 1991 in India. Both the economic and political
costs are immense. The evidence of substantial adjustment costs and sluggish
126
employment adjustment here, though modest, urges a qualified yes.
Several issues remain to be examined. The assumption that ignoring capital
doesn’t bias estimates of labor based on studies of the U.S. economy may be less
true for the Indian economy. Within the existing framework here, data on trade
unions and on external trade should be incorporated. As of now, the trade union
data based on voluntary but intermittent reporting is at best indicative. Data on
external trade can be used once the industrial and trade classifications are mapped.
Finally, heterogeneity arising from idiosyncratic shocks is lost due to the use of
industry level data. The industry data based estimates are revealing but the
magnitudes may well vary substantially at the unit level. Some of the ambiguity
emerging here may be resolved.
127
Table 2.1A: Employment Aggregates
Year
Population Labour
force
Employ-
ment
Agriculture Industry Services Unemploy-
ment rate*
million million million million million million percent
1971 548 - 180 130 20 30 -
1981# 685 - 223 148 - - -
1983 718 309 303 207 43 52 1.9
1987/88 790 333 324 208 52 64 2.8
1991 816 - 306 186 - - -
1993/94 895 382 374 242 58 77 2.0
1999/2000 1004 406 397 238 69 90 2.2
2001 1025 - 403 235 - - -
Source: Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Notes:
# Excludes the state of Assam
* Using usual principal and subsidiary status (upss)
Indian Census data for 1971, 1981, 1991, 2001; Rest- NSSO quinquennial household surveys
Table 2.1B: Growth Rates - Employment and GDP
Period
Reference GDP Employ-
ment
Five-year
plan
Annual
average
Annual
average
1951-56 I 3.6 0.39
1956-61 II 4.2 0.85
1961-66 III 2.8 2.03
1967-69 Annual plans 3.9 2.21
1969-74 IV 3.3 1.99
1974-79 V 4.8 1.84
1980-85 VI 5.7 1.73
1985-90 VII 5.8 1.89
1990-92 Annual plans 3.4 1.50
1992-97 VIII 6.5 2.44
1997-2002 IX 5.8 1.07
Source: Planning Commission, Government of India, Five-Year Plan Documents
128
Table 2.2: Employment in Organised Sector
Year ending
as of
March31
Organised
employment -
public sector
Organised
employment -
private sector
Organised
employment -
total
Growth -
public
sector
Growth -
private
sector
Growth -
total
organised
million million million percent percent percent
1975 12.87 6.80 19.67 - - -
1976 13.36 6.84 20.21 3.8 0.6 2.7
1977 13.88 6.87 20.74 3.8 0.4 2.7
1978 14.44 7.04 21.48 4.1 2.5 3.6
1979 15.05 7.21 22.25 4.2 2.3 3.6
1980 15.08 7.24 22.32 0.2 0.4 0.3
1981 15.48 7.40 22.88 2.7 2.2 2.5
1982 15.88 7.61 23.48 2.5 2.9 2.6
1983 16.46 7.55 24.01 3.7 -0.7 2.2
1984 16.86 7.43 24.29 2.4 -1.6 1.2
1985 17.27 7.31 24.58 2.4 -1.6 1.2
1986 17.68 7.38 25.06 2.4 0.9 1.9
1987 18.03 7.36 25.39 1.9 -0.2 1.3
1988 18.32 7.39 25.71 1.6 0.4 1.3
1989 18.51 7.48 25.98 1.0 1.2 1.1
1990 18.77 7.58 26.35 1.4 1.4 1.4
1991 19.06 7.68 26.73 1.5 1.2 1.4
1992 19.21 7.85 27.06 0.8 2.2 1.2
1993 19.33 7.85 27.18 0.6 0.1 0.4
1994 19.45 7.93 27.38 0.6 1.0 0.7
1995 19.47 8.06 27.53 0.1 1.6 0.5
1996
19.43 8.51 27.94
-0.2 5.6 1.5
1997 19.56 8.69 28.25 0.7 2.0 1.1
1998 19.42 8.75 28.17 -0.7 0.7 -0.3
1999 19.42 8.70 28.11 0.0 -0.6 -0.2
2000 19.31 8.65 27.96 -0.5 -0.6 -0.5
2001 19.14 8.65 27.79 -0.9 0.1 -0.6
2002 18.77 8.43 27.21 -1.9 -2.5 -2.1
2003 18.58 8.42 27.00 -1.0 -0.2 -0.8
Sources:
Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Economic Survey, Ministry of Finance, Government of India
129
Table 2.3: Disputes data
Year Industrial Disputes Strikes Lockouts Strikes-% change Lockouts-% change Lockouts-% share Loss (Rs. Mill)
No. Mandays
lost
No. Mandays
lost
No. Mandays
lost
No.-%
change
Mandays
lost - %
chg
No.-%
change
Mandays
lost - %
chg.
Of no.
disputes
of lost
mandays
Mandays lost*Avg.
Labor Cost/
manday
1970 2889 20563381 2598 14749067 291 5814314
- - - - 10.1 28.3 78.3
1971 2752 16545636 2478 11802589 274 4743047
-4.6 -20.0 -5.8 -18.4 10.0 28.7 68.2
1972 3243 20543916 2857 13748262 386 6795654
15.3 16.5 40.9 43.3 11.9 33.1 -
1973 3370 20626253 2958 13862211 412 6764042
3.5 0.8 6.7 -0.5 12.2 32.8 -
1974 2938 40262417 2510 33643524 428 6618893
-15.1 142.7 3.9 -2.1 14.6 16.4 116.6
1975 1943 21900931 1644 16706369 299 5194562
-34.5 -50.3 -30.1 -21.5 15.4 23.7 113.7
1976 1459 12745735 1241 2798975 218 9946760
-24.5 -83.2 -27.1 91.5 14.9 78.0 239.8
1977 3117 25320072 2691 13410141 426 11909931
116.8 379.1 95.4 19.7 13.7 47.0 286.6
1978 3187 28340199 2762 15423344 425 12916855
2.6 15.0 -0.2 8.5 13.3 45.6 332.0
1979 3048 43853518 2708 35803816 340 8049702
-2.0 132.1 -20.0 -37.7 11.2 18.4 225.6
1980 2856 21925026 2501 12017919 355 9907107
-7.6 -66.4 4.4 23.1 12.4 45.2 310.5
1981 2589 36583564 2245 21208130 344 15375434
-10.2 76.5 -3.1 55.2 13.3 42.0 538.6
1982 2483 74614764 2029 52112874 454 22501890
-9.6 145.7 32.0 46.3 18.3 30.2 878.7
1983 2488 46858464 1993 24921446 495 21937018
-1.8 -52.2 9.0 -2.5 19.9 46.8 -
1984 2094 56025240 1689 39956861 405 16068379
-15.3 60.3 -18.2 -26.8 19.3 28.7 -
1985 1755 29239466 1355 11486803 400 17752663
-19.8 -71.3 -1.2 10.5 22.8 60.7 954.6
1986 1892 32748228 1458 18823648 434 13924580
7.6 63.9 8.5 -21.6 22.9 42.5 837.6
1987 1799 35358372 1348 14026081 451 21332291
-7.5 -25.5 3.9 53.2 25.1 60.3 1404.1
1988 1745 33946925 1304 12529895 441 21417030
-3.3 -10.7 -2.2 0.4 25.3 63.1 1635.2
1989 1786 32663377 1397 10695112 389 21968265
7.1 -14.6 -11.8 2.6 21.8 67.3 -
1990 1825 24086170 1459 10639687 366 13446483
4.4 -0.5 -5.9 -38.8 20.1 55.8 1307.3
1991 1810 26428092 1278 12428333 532 13999759
-12.4 16.8 45.4 4.1 29.4 53.0 1527.2
1992 1714 31258744 1011 15132101 703 16126643
-20.9 21.8 32.1 15.2 41.0 51.6 2019.5
1993 1393 20300653 914 5614515 479 14686138
-9.6 -62.9 -31.9 -8.9 34.4 72.3 2027.9
1994 1201 20983082 808 6651054 393 14332028
-11.6 18.5 -18.0 -2.4 32.7 68.3 2092.6
1995 1066 16289569 732 5719961 334 10569608
-9.4 -14.0 -15.0 -26.3 31.3 64.9 1757.0
1996 1166 20284803 763 7817869 403 12466934
4.2 36.7 20.7 18.0 34.6 61.5 2382.9
1997 1305 16971389 793 6295365 512 10676024
3.9 -19.5 27.0 -14.4 39.2 62.9 2040.6
1998 1097 22061984 665 9349108 432 12712876
-16.1 48.5 -15.6 19.1 39.4 57.6 2679.4
1999 927 26786856 540 10625171 387 16161685
-18.8 13.6 -10.4 27.1 41.7 60.3 4176.7
2000 771 28763121 426 11958694 345 16804427
-21.1 12.6 -10.9 4.0 44.7 58.4 4036.6
2001 674 23766809 372 5562765 302 18204044
-12.7 -53.5 -12.5 8.3 44.8 76.6 4282.5
2002 579 26585919 295 9664537 284 16921382
-20.7 73.7 -6.0 -7.0 49.1 63.6 3907.7
Source: Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Notes: Data pertains to work stoppages where at least 10 workers are involved dirsectly or indirectly.
130
Table 2.4A: Enforcement - Industrial Tribunals and Labor Courts - Dispute Resolution
Year Central Government Industrial Tribunals and Labour Courts
Industrial
disputes
pending
beginning
of year
Industrial
disputes
received
during the
year
Industrial
disputes
disposed
during the
year
Industrial
disputes
pending
end of year
Applica-
tions
pending as
on January
1
Applica-
tions
received
during the
year
Applica-
tions
disposed
during the
year
Applica-
tions
pending
end of year
Percent
disputes
disposed
Percent
applica-
tions
disposed
1981 - - - - - - - - - -
1982 1093 558 337 1314 - - - -
20.4 -
1983 1298 643 602 1206 - - - -
31.0 -
1984 1206 834 652 1173 3359 1844 2129 2796
32.0 40.9
1985 1173 603 591 1185 2796 2842 2151 3485
33.3 38.2
1986 1185 909 456 1648 3485 2390 1720 4116
21.8 29.3
1987 1651 1191 509 2333 4131 4469 2042 6564
17.9 23.7
1988
- - - - - - - - - -
1989 2810 1450 803 3457 9142 3943 3891 9224
18.8 29.7
1990 3505 1504 747 4262 9514 1556 3339 7731
14.9 30.2
1991 4266 1440 1005 4701 7731 2535 2143 8123
17.6 20.9
1992 4702 616 403 4915 8165 1783 955 899
7.6 9.6
1993 5093 989 440 5642 9829 816 2432 8213
7.2 22.8
1994 5807 1211 649 6369 8399 1519 1814 8104
9.2 18.3
1995 6369 1310 1275 6404 8104 1842 3934 6012
16.6 39.6
1996 6409 1042 1149 5928 6033 1350 2510 4849
15.4 34.0
1997 6310 1460 977 6793 4776 1114 1990 3900
12.6 33.8
1998 7127 1280 724 7683 4337 1363 1002 4698
8.6 17.6
1999 7576 1622 770 8428 4203 234 499 3939
8.4 11.2
2000 8398 2035 722 9711 3328 1310 835 3803
6.9 18.0
2001 - - - - - - - - - -
2002 - - - - - - - - - -
Source: Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Notes: Data pertains to work stoppages where at least 10 workers are involved dirsectly or indirectly.
These statistics are collected on a voluntary basis from the primary units. Due to their voluntary nature,
the quality of data is not always upto the mark.
131
Table 2.4B: Enforcement - Other Acts
Year Factories Act 1948 Shops and Establishments Act
Factories
on register
Factories
Inspected
Percent
Inspec-
tions
Factories
Number
of Convic-
tions
Convictions
as % of
inspections
Shops
and
Establ
no.
Shops and
Establ
Inspected
Percent
Inspections
Shops &
Establ
Number of
prosecu-
tions
Cases
disposed
off by
courts
Amount of
fines
realized
(Rs.'000)
Prosecu-
tions as %
of
inspections
1970 84426 62200 73.67 8678 13.95 2320218 2050696 88.38 111703 78450 2167063 5.45
1971 88089 63018 71.54 9567 15.18 2404510 2026412 84.28 105358 89121 2735810 5.20
1972 93053 67379 72.41 9320 13.83 2529173 3135065 123.96 105697 82904 2285066 3.37
1973 - - - - - 2371370 1958748 82.60 96945 94410 2827370 4.95
1974 104680 75547 72.17 9946 13.17 2468137 1946783 78.88 117664 106774 3096046 6.04
1975 112237 79720 71.03 8696 10.91 2584000 - - - - - -
1976 110427 79813 72.28 12139 15.21 2691000 1933294 71.84 129784 111604 4062020 6.71
1977 125164 89088 71.18 14338 16.09 3268000 2116729 64.77 177654 119648 4280568 8.39
1978 114071 80054 70.18 13167 16.45 3315000 1991363 60.07 179951 111088 4225743 9.04
1979 122931 79258 64.47 14920 18.82 3614000 1978296 54.74 220631 152634 6217451 11.15
1980 141028 85336 60.51 14814 17.36 3749000 1941097 51.78 221413 151988 5489004 11.41
1981 149893 95804 63.91 15554 16.24 3839000 2101856 54.75 255963 163370 6120948 12.18
1982 128341 85215 66.40 25865 30.35 3358218 2061775 61.39 208497 142192 6082111 10.11
1983 141434 92938 65.71 28049 30.18 3358218 - - - - - -
1984 171966 113054 65.74 26911 23.80 3392275 - - - - - -
1985 178696 107255 60.02 20224 18.86 3004721 - - - - - -
1986 165637 104435 63.05 19400 18.58 3321566 2086835 62.83 167456 157185 8974302 8.02
1987 185096 111660 60.33 26214 23.48 3428856 2173180 63.38 183798 177969 10779696 8.46
1988 128660 64771 50.34 7715 11.91 3401593 2340044 68.79 188991 156014 11147169 8.08
1989 155925 80900 51.88 18566 22.95 3941589 2431273 61.68 213898 198622 13258789 8.80
1990 122070 78969 71.51 6020 7.62 3293342 2172367 65.96 132596 108545 8879175 6.10
1991 89030 67342 75.64 7846 11.65 3899937 2436067 62.46 141092 110012 11105661 5.79
1992 138907 103157 74.26 13650 13.23 3656245 2055961 56.23 129613 101788 11866481 6.30
1993 - - - - - 3813607 2020591 52.98 142098 119434 14883.73 7.03
1994 84431 63084 74.72 2889 4.58 3767703 1905291 50.57 125716 107902 15467.27 6.60
1995 111742 61216 54.78 3675 6.00 3973776 2059766 51.83 127500 98690 15182.86 6.19
1996 147310 87564 59.44 7723 8.82 5157303 2057470 39.89 154176 125306 17324.97 7.49
1997 86053 58620 68.12 6380 10.88 5472569 2238485 40.90 128179 107441 19317.54 5.73
1998 258440 141930 54.92 8672 6.11 5541409 2136833 38.56 130483 106731 27785.79 6.11
1999 246252 115006 46.70 5378 4.68 5800916 2160467 37.24 115745 106961 17701.02 5.36
2000 88702 50832 57.31 3290 6.47 5536095 2127202 38.42 115092 105387 21547.45 5.41
2001 - - - - - 6023103 2080560 34.54 105943 990939 24199.06 5.09
See table 2.4A for sources and notes. Some excessively high percentage changes may be based on provisional data and may be disregarded.
132
Table 2.4C: Enforcement - Standing Orders (Employment) Act
Year ------------------------------------------------------- Standing Orders (Employment) Act 1946 -------------------------------------------------------------
Number of
establish-
ments
covered
Number of
employees
covered
No. establ
with
certified
orders beg.
of year
No. empl
covered by
certified
orders beg.
of year
Applica-
tions for
certification:
B/F as on
January 1
Applica-
tions for
certification
received
during-year
Applica-
tions for
certification
disposed
during-year
Applica-
tions for
certification
pending end
of year
No. establ
with
certified
orders end
of year
No. empl
covered by
certified
orders end
of year
Percent
application
s disposed
1970 17289 4103107 10936 3024605 1003 261 236 1082 11011 3209336
18.7
1971 18050 4270037 9283 3017326 1028 229 200 1057 9273 2953962
15.9
1972 - - - - - - - - - - -
1973 18745 4384403 10654 3447583 1145 291 230 1206 10702 3423249
16.0
1974 18011 3652803 9529 2711364 921 880 886 922 10399 2817040
49.2
1975 18573 3919894 8869 2626824 923 252 272 909 9124 2711932
23.1
1976 19162 4213827 10507 2703510 895 459 389 965 10884 3165424
28.7
1977 16189 3499811 8453 2295211 751 296 214 833 8584 2311033
20.4
1978 16196 3586615 8495 2303386 833 876 211 1498 8704 2351696
12.3
1979 17871 3308485 8769 2478435 653 283 245 681 8952 2536809
26.2
1980 25455 4231811 10617 3376040 961 171 162 970 11053 3943004
14.3
1981 18265 4309188 11076 3515010 370 237 192 1015 11258 3563515
31.6
1982 17509 3572919 8896 2857817 1015 240 175 1080 9078 2926281
13.9
1983 25650 5423826 14838 4329623 1735 447 677 1505 15161 4431945
31.0
1984 29434 5456601 15323 4681486 1751 594 546 1799 15909 4904807
23.3
1985 - - - - - - - - - - -
1986 - - - - - - - - - - -
1987 40776 5695305 26019 4606395 1829 556 529 1856 26563 4712175
22.2
1988 43992 6107072 26426 4729950 1852 726 658 1920 27089 4789207
25.5
1989 41987 5939797 23763 4763322 1867 628 658 1837 24421 4810413
26.4
1990 43305 5967538 24038 4683810 1833 1771 1509 2095 24646 4762349
41.9
1991 47100 6342735 27082 4966685 2161 738 725 2174 27803 4847477
25.0
1992 42638 5591083 23776 3378160 1334 581 566 1349 24342 4239426
29.6
1993 28410 3976780 13091 3040555 1934 597 455 2076 13291 2854244
18.0
1994 35417 4184413 20625 3074903 1321 469 385 1405 21010 3095465
21.5
1995 30069 3354608 11285 2925828 1957 470 385 2042 11670 2891996
15.9
1996 42566 4999714 11961 2989455 2117 552 562 2107 12503 3049822
21.1
1997 46024 5352919 12823 3201230 2101 530 461 2170 13284 3162004
17.5
1998 42199 5095292 12831 3011426 2070 483 471 2082 13280 3094684
18.4
1999 46115 4808860 12106 2567342 1690 328 354 1664 12460 2613654
17.5
2000 47681 5263181 13019 2822547 1740 444 489 1695 13437 2883562
22.4
2001 68850 6765485 14848 3385791 2068 447 368 2147 15289 3435837
14.6
See table 2.4A for sources and notes
133
Table 2.5: Workers' Adjustment Assistance - National Renewal Fund (NRF)
Year
Budgeted
(Rupees
million)
Released
(Rupees
million)
Total no.
workers
covered
Total no.
workers
covered -
cumulative
No. workers
counseled
via EACs*
No. workers
retrained
via EACs*
No. workers
redeployed
via EACs*
No. workers
covered by
EACs* -
cumulative
1992/93 8297 5667 38626 38626 - - - -
1993/94 10200 4781 31613 70239 - - - -
1994/95 7000 2508 3979 74218 7500 1500 234 9234
1995/96 - 2096 8946 83164 11493 5644 704 27075
1996/97 - 1906 15264 98428 28862 18927 3898 78762
1997/98 2350 - 8733 107161 - - - -
1998/99 - - - - - - - -
1999/2000 - - - - - - - -
2000/01 - - - - - - - -
Sources:
Indian Labour Yearbooks, 1994, 1995, 1996, 1997, 1998, Labour Bureau, Ministry of Labour, Government of India
PRAGYA (1999) "Managing the Negative Impacts of Retrenchments and Layoffs", Ministry of Labour, Government of India
(PRAGYA: 1994/95 data for EACs)
Notes:
* Employee Assistance Centres (EACs): 5 sanctioned in 1994/95 - Bombay, Indore, Ahmedabad, Kanpur, Calcutta
134
Table 2.6: Adjustment Costs - Labor Restructuring
Pragya 1999 TISS Study 2001, Bombay
40 cases** MNC Mill NTC Mills
Continuing-VRS Continuing-VRS Sick -VRS
Workers
No. employed pre-VRS/lockout/dispute 2000+ on average - -
No. displaced in VRS/lockout/dispute - 2000 20000
No. employed post VRS/lockout/dispute
Sample size 190 355 251
VRS - average amount (in Rupees '000) - 500-700 60-250
Worker Dues* held back na na na
Amount - avg./max. Rupees per worker na na na
Time-Years held back na na na
Age: Percent>50 years 24.2 59.5 42.0
Education: Percent>=secondary (8+ yrs) - 72.6 24.8
Assistance
Counseling (Percent of sample) - 21.4 4.7
Re-training (Percent of sample) - 0.0 2.0
Re-deployment help (% of sample) - 0.0 2.0
Post-VRS employment (% of sample) 92.2 26.2 68.6
Of which percent self-employed 57.3 89.0 70.3
Looking for employment (% sample) - 21.7 17.3
Post-VRS avg. earnings (2002 Rs./month)# - 4423 1041
pre-VRS avg. earnings (2002 Rs./month) - - 4486
Percent loss - avg. earnings - - 76.8
Impact on health - reduced expenses (% sample) - 15.7 23.5
Impact on children's education - Neutral Adverse
Percent cases - children drop-out - 1.2 13.8
Percent loss-avg. earnings -children ## - 46.0 46.0
Sources: Reports prepared for the Ministry of Labour, Governmant of India
PRAGYA (1999) "Managing the Negative Impacts of Retrenchments & Layoffs"
Tata Institute of Social Sciences (TISS) (2001), "Life After Voluntary Retirement - Examining
the Human Face of the Structural Adjustment Programme"
Institute for Human Development (IHD) (2004), "Rehabilitation of Cotton Textile Mill
Workers after the Closure of the Mills in Indore"
Notes: na Not Applicable; MNC Multinational company; NTC (STC) National (State) Textile Corp.
* Worker dues may be wages, provident fund, gratuity, and retrenchment or closure compensation
** Covers three industries - textiles, engineering, pharmaceuticals
# Inflation adjusted: Pre- & post-VRS earnings in 1991 rupees; deflation using CPI-industrial workers
## Loss in earnings by comparing average labor cost per manday in ASI-census (proxy for organised
sector) & ASI-sample (proxy-unorganised) in year of adjustment, i.e., 1997 (2002 figure = 45.4%)
135
Table 2.6: Adjustment Costs - Labor Restructuring (contd.)
IHD Study 2004, Indore
NTC - A NTC - B NTC - C
Sick -VRS Sick -VRS Sick -VRS
Workers
No. employed pre-VRS/lockout/dispute 3814 3000 1766
No. displaced in VRS/lockout/dispute 35%, 100% 38%, 100% 100%
No. employed post VRS/lockout/dispute None None None
Sample size 130 82 60
VRS - average amount (in Rupees '000) 125, 200-300 125, 200-300 200-300
Worker Dues* held back na na na
Amount - avg./max. Rupees per worker na na na
Time-Years held back na na na
Age: Percent>50 years 19.2 22.2 60.0
Education: Percent>=secondary (8+ yrs) ------------------------- 52.8% -------------------
Assistance
Counseling (Percent of sample) - - -
Re-training (Percent of sample) Yes Yes Yes
Re-deployment help (% of sample) Yes Yes Yes
Post-VRS employment (% of sample) 51.5 42.0 36.7
Of which percent self-employed 17.9 58.8 27.3
Looking for employment (% sample) 23.9 5.6 14.3
Post-VRS avg. earnings (2002 Rs./month)# 1500 1500 1500
pre-VRS avg. earnings (2002 Rs./month) 3190 3190 4253
Percent loss - avg. earnings 53.0 53.0 64.7
Impact on health - reduced expenses (% sample) --------------------------- 5.7% --------------------
Impact on children's education ------- Adverse, no data reported -----------
Percent cases - children drop-out - - -
Percent loss-avg. earnings -children ## - - -
Sources and Notes: see first page of table 2.6
136
Table 2.6: Adjustment Costs - Labor Restructuring (concld.)
Private - D Private - E STC - F
Lockout Lockout Sick -VRS
Workers
No. employed pre-VRS/lockout/dispute 1100 6000 1600
No. displaced in VRS/lockout/dispute 100% 100% 100%
No. employed post VRS/lockout/dispute 135 reporting None None
Sample size 35 122 59
VRS - average amount (in Rupees '000) None None 6-150
Worker Dues* held back Yes Yes na
Amount - avg./max. Rupees per worker - 90000 max na
Time-Years held back 18 years 12 years na
Age: Percent>50 years 74.3 59.0 30.0
Education: Percent>=secondary (8+ yrs) --------------------- 52.8% -----------
Assistance
Counseling (Percent of sample) None None -
Re-training (Percent of sample) None None Yes
Re-deployment help (% of sample) None None Yes
Post-VRS employment (% of sample) 62.9 82.8 58.3
Of which percent self-employed 22.7 21.8 28.6
Looking for employment (% sample) 25.0 14.3 7.7
Post-VRS avg. earnings (2002 Rs./month)# 1500 1500 1500
pre-VRS avg. earnings (2002 Rs./month) 5293 1595 3190
Percent loss - avg. earnings 71.7 6.0 53.0
Impact on health - reduced expenses (% sample) ----------------------- 5.7% ---------------
Impact on children's education ----- Adverse, no data reported ----
Percent cases - children drop-out - - -
Percent loss-avg. earnings -children ## - - -
Sources and Notes: see first page of table 2.6
137
Table 2.7: Industrial Disputes Act 1947 with Amendments
Date Section Coverage Provision or Institutional Change Flexibility
1920 Trade
Disputes' Act
India Established Courts of Enquiry, Conciliation Boards;
Forbade strikes in public utility services without notice;
but, no machinery for settling of "industrial disputes"
+
1929 Trade
Disputes' Act
India State intervention in settling industrial disputes - provided
conciliation machinery for peaceful settlement
-
Policy of "laissez faire" continued, selective intervention
at best
WW-
II
Defense of
India Rules
(temp)
India Power to appoint industrial tribunals & enforce tribunal
awards
-
1946 Industrial
Employ-ment
(Standing
Order) Act
India Provision for devising, framing, and certifying of
standing orders covering service conditions, and making
these known to workers
-
1947 Chap 1-VIII:
Sections 1-40
India Embodies essential principles of Rule 81A, Defense of
India Rules and certain principles of Trade Disputes Act
1929 about investigation and settlement of industrial
disputes. Seeks to achieve social justice on the basis of
collective bargaining, conciliation, arbitration, or finally
compulsory adjudication. Dismissals, and a decision to
grant or decline permission to retrench are industrial
disputes. Retrenchment means the termination by
employer of the service of a worker for any reason other
than disciplinary action; excludes termination upon
voluntary retirement, superannuation, non-renewal of
a contract, continued ill-health, and cases where the
appointment letter stipulates discharge from service
without notice or reasons.
-
1947 3 India Works committee comprising representatives of
employers and workmen to be constituted in every
industrial establishment employing 100 or more
workers. Industry means any business, trade,
undertaking, manufacture… Member-workers to be
selected in consultation with their trade union. Committee
tasked with promoting measures for securing &
preserving good relations.
-
1947 3A,3B Gujarat Additional: Joint management council instead of works
committee, for every industrial establishment employing
500 or more workers. Duties include promoting
productivity growth and training workers to understand
management responsibilities.
+
1947 3 Maharashtra Additional: Recognized union, if any, shall appoint its
nominees to represent workers
-
138
Table 2.7: Industrial Disputes Act 1947 with Amendments (contd.)
Date Section Coverage Provision or Institutional Change Flexibility
1947 3 Rajasthan Additional: State shall appoint Registrar of Unions for the
whole state and assistant registrar for any local area.
-
1947 7, 7A, 7B India Labor courts, regional and national industrial
tribunals constituted; strikes and lockouts prohibited
pending proceedings in tribunals
-
1953 25A-25J
(Chapter VA)
added
India Influenced by decisions of judiciary, bipartite and
tripartite agreements, and actual implementation of IDA
provisions, substantial modifications made.
-
1953 25A India New addition: Exemption from liability to pay lay-off
compensation or maintaining muster rolls: industrial
establishments with less than 50 workers per working day
in the preceeding month or those of a seasonal nature
+
1953 25B India ?? New addition: Eligibility is that worker's name is
borne on the establishment muster rolls and service is
continuous of at least one year.
-
1953 25C India Newly added: Compensation specified in case of lay-
off (50% of wages and dearness allowance) - 45 days
maximum during any 12 months. Not applicable to
closures. Wages as per section S.2(iv)(d) of the Payment
of Wages Act. Eligibility is that worker's name is borne
on the establishment muster rolls and service is
continuous of at least one year.
-
Immediate provocation for tightening law was imminent
downsizing in (cotton?) textiles mills - closing one shift -
possibly rendering thousands of workers unemployed
with no income.
1953 25D India First time added: Muster rolls entry recording worker's
availability for work at the establishment mandatory for
lay-off compensation, muster rolls to be made available
for entry by the employer.
-
1953 25E India New addition: Provides for exceptions to the general
provision for lay-off compensation: (i) employer offers
alternate employment in the same or any other
estalishment, (ii) worker does not present himself for
work at the appointed time, (iii) lay-off due to strike or
slow-down of production in another part of
establishment.
+
139
Table 2.7: Industrial Disputes Act 1947 with Amendments (contd.)
Date Section Coverage Provision or Institutional Change Flexibility
1953 25F India New addition: Retrenchment compensation to follow a
simple yardstick of length of service; on grounds of
humane public policy and consideration that involuntary
employment may result in general economic insecurity.
Compensation formula: 15 days average pay for every
completed year of continuous service or any part
thereof in excess of six months.
-
1953 25J India Employee rights under a contract cannot be derogated by
reason of any provision in the Act
-
1956 25C India Amended: Layoff compensation to extend beyond 45
days to all days laid off.
-
1956 25FF India Amended and added: To check indiscriminate resort to
layoff & retrenchment via closures, amendment provided
that termination of employment due to closure or
transfer be considered as retrenchment.
-
Immediate cause was a Supreme Court of India ruling
(Hariparshad Shivshanker Shukla vs. A.D. Diwakar, per
S.K. Das J) that denied compensation to a worker
unemployed due to a closure/transfer.
1956 25H India Amended?: Employer to offer an opportunity for re-
employment of retrenched workers if further employment
proposed.
-
1956 25I India Deleted, covered under section 33C: Recovery of dues to
workers may be recovered by the government in the same
manner as arrears of land revenue or as public demand by
govt. on an application to the govt. by the worker/s.
-
1956 33C India Added: recovery of money dues to workers under chap
VA (lay-off and rerenchment), also money due under
settlements or awards
-
1957 25FFF India New addition: If closure due to specified "unavoidable
circumstances" beyond the employer's control, then
retrenchment compensation is reduced to a maximum
of 3 months' average pay of the (eligible) worker.
Unavoidable circumstances refers to financial difficulties
only, or expiry of lease or license on or after April 1,
1967. For construction work, compensation applies only
beyond two years of construction project, as per section
25F.
+
140
Table 2.7: Industrial Disputes Act 1947 with Amendments (concld.)
Date Section Coverage Provision or Institutional Change Flexibility
1964 25B India Amended: Defined "continuous service" for "a period" as
all uninterrupted service for that period, including periods
of interruptions due to sickness, authorised leave,
accident, illegal strike, lockout or cessation of work not
due to the worker; deemed to be a year (six months) if
worker actually worked for 240 (120) days and for
miners 190 (95) days.
-
1964 25FFF India Amended: "completed year of continuous service"
replaced by "completed year of service" for eligibility for
compensation [see 1956, section 25B, and 1957, 25FFF]
-
1964 25FFF India Amended: unavoidable circumstances for closure include
accumulation of undisposed stocks [see 1957, section
25FFF]
+
1964 25H India Amended: Workers to be considered for re-employment
if they are citizens of India.
-
1964 25J India Amended: Provisions of chap. VA prevail over those of
other acts including Standing Orders that are inconsistent
with it. If the inconsistent provision more beneficial to
worker, it'll override chap VA
-
1965 25C India Amended: Compensation specified in case of lay-off -
1971 25FFA West Bengal Employer to give 60 days' notice to state government
(West Bengal) for closure - to enable govt. to take
measures, if required
-
1972 25FFA India Employer to give 60 days' notice to state government for
closure
-
1976 25K-255,
chap VB
India Industrial establishments (as per section 25L) employing
300 or more workers: require "prior permission" to
lay-off or retrench workers, and for closure
-
1982 25K-255??,
(Chapter VB
added)
India Industrial establishments (as per section 25L) requiring
"prior permission" to lay-off or retrench workers, and for
closure: coverage expanded to 100 or more workers
-
Sources:
2005, The Industrial Disputes Act, 1947 with amendments, BARE ACT with short comments,
Professional Book Publishers, New Delhi
2001, G.B. Pai, Labour Law in India, Butterworths India., New Delhi
Discussions with experts: Labour administration officials, Jurists, Lawyers
141
Table 3.1
Summary Statistics
India: Industry (3-digit) Data, 1973-1997
Variable Obs Unit Mean Std. Dev. Min Max
Factories 3400 Number 645 1067 1 10702
Workers 3400 Number 36921 71490 31 783025
Mandays 3400 Thousand 10695 22214 10 250334
Value of output 3400 Rupees
thousand
6772286 14500000 1059 170000000
Wage 3400 Rupees 8706 5240 575 58180
Unit severance pay 3400 Rupees 24292 14140 1090 60867
Unit hiring cost 3400 Rupees 735 362 123 1722
Marginal expected
adjustment cost -
historical memory 3264 Rupees 9416 6667 123 34073
Marginal expected
adjustment cost - five
year memory 2720 Rupees 11548 7759 153 49038
Market share 3400 Percent 14 19 0.002 100
Price Cost Margin 3398 Percent 33 12 1.000 96
Source: Author's dataset (truncated version of Annual Survey of Industries, 1973-1997)
Note: Rupee values in constant 1982/83 terms.
142
Table 3.2
India: Job flows
(Percent of Employment*)
(3-digit Industry Level)
Year
Job creation
rate
Job
destruction
rate
Net
employment
rate
Job
reallocation
rate
Excess job
reallocation
rate
1973
- - - - -
1974 7.4 4.4 3.0 11.8 8.8
1975 7.4 3.5 4.0 10.9 6.9
1976 6.4 2.9 3.5 9.4 5.9
1977 6.4 0.9 5.5 7.3 1.8
1978 4.5 2.4 2.2 6.9 4.8
1979 7.4 1.7 5.7 9.2 3.5
1980 4.7 3.2 1.5 7.8 6.4
1981 4.6 3.6 0.9 8.2 7.3
1982 5.8 2.7 3.1 8.4 5.3
1983 3.2 6.7 -3.6 9.9 6.3
1984 4.4 6.1 -1.7 10.6 8.8
1985 3.5 7.8 -4.3 11.3 7.0
1986 2.9 3.4 -0.5 6.3 5.8
1987 7.9 3.6 4.3 11.5 7.2
1988 4.0 4.4 -0.4 8.5 8.1
1989 8.0 5.5 2.5 13.5 11.0
1990 4.1 4.4 -0.3 8.5 8.2
1991 4.2 3.9 0.3 8.1 7.8
1992 7.2 2.2 5.1 9.4 4.3
1993 3.7 4.0 -0.3 7.7 7.4
1994 5.8 1.8 4.0 7.6 3.6
1995 10.3 1.6 8.7 12.0 3.3
1996 4.0 5.2 -1.2 9.2 8.0
1997 5.1 6.0 -0.8 11.1 10.3
Average(73-97) 5.5 3.8 1.7 9.4 6.6
Variance(73-97) 3.4 2.9 9.4 3.3 4.9
Correlation with net 0.9 -0.8 1.0 0.1 -0.6
Avg(73-75) 7.4 3.9 3.5 11.3 7.9
Var(73-75) 0.0 0.4 0.5 0.4 1.8
Avg(76-82) 5.7 2.5 3.2 8.2 5.0
Var(76-82) 1.3 0.8 3.4 0.8 3.4
Avg(83-90) 4.7 5.3 -0.5 10.0 7.8
Var(83-90) 4.1 2.4 8.2 4.9 2.7
Avg(91-97) 5.8 3.5 2.3 9.3 6.4
Var(91-97) 5.5 2.9 14.0 2.9 7.1
Avg(76-97) 5.4 3.8 1.5 9.2 6.5
Var(76-97) 3.5 3.3 10.4 3.3 5.3
* Employment is averaged over current and previous year.
Source: Author's dataset (truncated version of Annual Survey of Industries, 1973-1997)
143
Table 3.3
Job Flows: Comparisons
JC JD JR NT
Industry - level
US-4I 2.48 3.63 6.11 -1.15
India - 3I 5.5 3.8 9.4 1.7
Establishment - level
Canada 10.9 11.1 22 -0.2
USA 8.8 10.2 19 -1.4
Denmark 16 13.8 29.8 2.2
France 13.9 13.2 27.1 0.7
Norway 7.1 8.4 15.5 -1.3
U.K. 10.2 11.5 21.7 -1.3
Chile 12.9 13.9 26.8 -1
Colombia 12.5 12.2 24.7 0.3
Estonia 9.7 12.9 22.6 -3.2
Morocco 18.6 12.1 30.7 6.5
Sources:
Davis and Haltiwanger (1998)
Author's dataset - India
144
Table 3.4
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1975 1973-1982 1973-1997
With 76-97, 82-97, & 91-97 period dum
With 76-
82
dummy
No ind.
year
dum
Ind.
dum
only
Industry
& year
dummies
Constant 0.777 *** 0.741 *** 0.841 *** 0.942 *** 0.896 ***
Wage -0.102 *** -0.096 *** -0.101 *** -0.106 *** -0.101 ***
Mandays 0.488 *** 0.218 *** 0.202 *** 0.211 *** 0.207 ***
Output 0.064 *** 0.057 *** 0.052 *** 0.048 *** 0.048 ***
Lagged employment 0.960 *** 0.959 *** 0.959 *** 0.962 *** 0.961 ***
Time 0.115 *** -0.004 ** -0.002 ** -0.002 0.001
Average Emplt. Size(>=100) dummy -0.661 *** 0.555 0.352 0.446 0.541
Wage*Size100 0.051 ** 0.027 0.031 0.017 0.009
Mandays*Size100 -0.381 *** 0.010 0.020 0.020 0.031
Output*Size100 -0.031 ** -0.066 ** -0.050 -0.043 -0.043
Lagged employment*Size100 0.011 0.019 0.015 0.008 0.007
Average Emplt. Size(>=300) dummy 6.846 *** 2.761 * 2.497 1.885 1.880
Wage*Size300 -0.650 *** -0.344 ** -0.321 * -0.269 -0.272
Mandays*Size300 1.069 *** 0.205 0.180 0.152 0.150
Output*Size300 0.547 *** 0.379 *** 0.377 *** 0.381 *** 0.380 ***
Lagged employment*Size300 -0.692 *** -0.447 *** -0.441 *** -0.438 *** -0.435 ***
Year76 dummy - -0.533 *** -0.634 *** -0.618 *** -0.563 ***
Wage*Year76 - 0.077 *** 0.083 *** 0.083 *** 0.074 ***
Mandays*Year76 - -0.162 *** -0.155 *** -0.150 *** -0.139 ***
Output*Year76 - -0.045 *** -0.042 *** -0.041 ** -0.039 **
Lagged employment*Year76 - 0.032 ** 0.034 * 0.030 0.031
Average Emplt. Size(>=100)*Year76 - -0.552 -0.219 -0.307 -0.349
Wage*Size100*Year76 - -0.016 -0.012 -0.006 -0.002
Mandays*Size100*Year76 - -0.005 -0.004 -0.018 -0.025
Output*Size100*Year76 - 0.056 0.028 0.025 0.026
Lagged employment*Size100*Year76 - -0.014 -0.007 -0.002 -0.003
Average Emplt. Size(>=300)*Year76 - -3.564 * -3.177 -2.706 -2.822
Wage*Size300*Year76 - 0.455 ** 0.397 * 0.356 0.372 *
Mandays*Size300*Year76 - -0.373 -0.310 -0.267 -0.281
Output*Size300*Year76 - -0.351 *** -0.327 *** -0.329 *** -0.332 ***
Lagged employment*Size300*Year76 - 0.377 *** 0.362 *** 0.360 *** 0.361 ***
Year82 dummy - - 0.000 0.043 dropped
Wage*Year82 - - -0.005 -0.010 -0.011
Mandays*Year82 - - -0.007 -0.004 -0.006
Output*Year82 - - 0.018 0.020 * 0.018
Lagged employment*Year82 - - -0.025 * -0.027 ** -0.026 *
Average Emplt. Size(>=100)*Year82 - - 0.436 0.367 0.201
Wage*Size100*Year82 - - -0.001 0.002 0.014
Mandays*Size100*Year82 - - 0.073 0.045 0.020
Output*Size100*Year82 - - -0.022 -0.016 -0.013
Lagged employment*Size100*Year82 - - 0.001 -0.008 -0.010
Average Emplt. Size(>=300)*Year82 - - -0.807 -0.928 -0.201
Wage*Size300*Year82 - - 0.108 0.128 0.058
Mandays*Size300*Year82 - - 0.441 0.462 0.556 *
Output*Size300*Year82 - - -0.094 * -0.107 * -0.095 *
Lagged employment*Size300*Year82 - - 0.149 ** 0.164 *** 0.151 **
145
Table 3.4
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1975 1973-1982 1973-1997
With 76-97, 82-97, & 91-97 period dum
With 76-
82
dummy
No ind.
year
dum
Ind.
dum
only
Industry
& year
dummies
Year91 dummy - - 0.146 0.224 dropped
Wage*Year91 - - -0.008 -0.014 -0.011
Mandays*Year91 - - 0.050 0.072 0.065
Output*Year91 - - -0.006 -0.004 0.000
Lagged employment*Year91 - - 0.011 0.008 0.004
Average Emplt. Size(>=100)*Year91 - - -0.519 -0.483 -0.304
Wage*Size100*Year91 - - -0.027 -0.038 -0.053
Mandays*Size100*Year91 - - -0.262 * -0.264 * -0.229
Output*Size100*Year91 - - 0.055 0.056 0.054
Lagged employment*Size100*Year91 - - -0.037 -0.032 -0.029
Average Emplt. Size(>=300)*Year91 - - 4.466 ** 4.548 ** 3.633
Wage*Size300*Year91 - - -0.437 -0.455 -0.339
Mandays*Size300*Year91 - - 0.117 0.001 0.012
Output*Size300*Year91 - - 0.082 0.087 0.059
Lagged employment*Size300*Year91 - - -0.132 -0.144 -0.118
Price Cost Margin - - - - -
Lagged emplt*PCM - - - - -
Price Cost Margin*Size100 - - - - -
Lagged emplt*PCM*Size100 - - - - -
Price Cost Margin*Size300 - - - - -
Lagged emplt*PCM*Size300 - - - - -
Pr > chi2 0.000 0.000 0.000 0.000 0.000
No. of observations 272 1224 3264 3264 3264
Notes:
(1) * significant at 10% level, ** significant at 5% level, *** significant at 1% level
(2) Errors robust to heteroscedasticity and autocorrelation
(3) All variables in logs, except time; wage, mandays, output, average size, PCM lagged one period
(4) Industry dummies at 2-digit level
(5) Year dummies from 1979 onward, to capture gradual liberalization process, begun in 1979
146
Table 3.4 (contd.)
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1997
m NO interaction of market variables & period dummies
Market
variables
No dummies
Market variables
(Ind dum)
Market
variables (Ind &
year dum)
Constant 0.776 *** 1.068 *** 1.069 ***
Wage -0.110 *** -0.114 *** -0.110 ***
Mandays 0.200 *** 0.213 *** 0.206 ***
Output 0.058 *** 0.054 *** 0.053 ***
Lagged employment 0.955 *** 0.938 *** 0.933 ***
Time -0.002 ** -0.002 -0.001
Average Emplt. Size(>=100) dummy 2.594 *** 2.382 ** 2.543 **
Wage*Size100 0.044 0.031 0.022
Mandays*Size100 -0.033 -0.049 -0.036
Output*Size100 -0.065 -0.056 -0.056
Lagged employment*Size100 -0.204 ** -0.187 ** -0.195 **
Average Emplt. Size(>=300) dummy -1.489 -2.438 -2.765
Wage*Size300 -0.327 * -0.266 -0.267
Mandays*Size300 0.232 0.200 0.199
Output*Size300 0.389 *** 0.397 *** 0.397 ***
Lagged employment*Size300 -0.075 -0.055 -0.023
Year76 dummy -0.643 *** -0.621 *** -0.548 ***
Wage*Year76 0.082 *** 0.080 *** 0.070 ***
Mandays*Year76 -0.149 *** -0.150 *** -0.134 ***
Output*Year76 -0.041 *** -0.038 *** -0.035 **
Lagged employment*Year76 0.034 * 0.029 0.028
Average Emplt. Size(>=100)*Year76 -0.002 -0.026 -0.093
Wage*Size100*Year76 -0.029 -0.028 -0.021
Mandays*Size100*Year76 0.002 0.010 0.000
Output*Size100*Year76 0.030 0.028 0.027
Lagged employment*Size100*Year76 -0.015 -0.010 -0.009
Average Emplt. Size(>=300)*Year76 -3.423 -2.994 -3.021
Wage*Size300*Year76 0.421 * 0.389 * 0.396 *
Mandays*Size300*Year76 -0.290 -0.271 -0.272
Output*Size300*Year76 -0.306 *** -0.312 *** -0.311 ***
Lagged employment*Size300*Year76 0.340 *** 0.337 *** 0.333 ***
Year82 dummy 0.005 0.043
Wage*Year82 -0.006 -0.012 -0.013
Mandays*Year82 -0.015 -0.015 -0.020
Output*Year82 0.019 0.022 * 0.020 *
Lagged employment*Year82 -0.027 ** -0.030 ** -0.028 **
Average Emplt. Size(>=100)*Year82 0.374 0.352 0.190
Wage*Size100*Year82 0.011 0.013 0.023
Mandays*Size100*Year82 0.101 0.079 0.051
Output*Size100*Year82 -0.023 -0.018 -0.016
Lagged employment*Size100*Year82 0.002 -0.008 -0.009
Average Emplt. Size(>=300)*Year82 -0.394 -0.349 0.286
Wage*Size300*Year82 0.071 0.077 0.015
Mandays*Size300*Year82 0.484 0.537 0.597 *
Output*Size300*Year82 -0.104 * -0.116 ** -0.109 *
Lagged employment*Size300*Year82 0.159 ** 0.175 *** 0.164 ***
147
Table 3.4 (contd.)
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1997
m NO interaction of market variables & period dummies
Market
variables
No dummies
Market variables
(Ind dum)
Market
variables (Ind &
year dum)
Year91 dummy 0.152 0.223
Wage*Year91 -0.008 -0.014 -0.011
Mandays*Year91 0.048 0.064 0.056
Output*Year91 -0.007 -0.005 -0.001
Lagged employment*Year91 0.011 0.008 0.004
Average Emplt. Size(>=100)*Year91 -0.847 -0.916 -0.706
Wage*Size100*Year91 -0.021 -0.016 -0.032
Mandays*Size100*Year91 -0.358 ** -0.383 ** -0.339 *
Output*Size100*Year91 0.051 0.044 0.042
Lagged employment*Size100*Year91 -0.011 -0.003 -0.001
Average Emplt. Size(>=300)*Year91 4.613 ** 4.662 ** 3.633
Wage*Size300*Year91 -0.464 -0.469 -0.326
Mandays*Size300*Year91 -0.002 -0.066 0.032
Output*Size300*Year91 0.085 0.088 0.051
Lagged employment*Size300*Year91 -0.144 -0.156 -0.121
Price Cost Margin 0.034 -0.012 -0.024
Lagged emplt*PCM -0.001 0.005 0.006
Price Cost Margin*Size100 -0.647 *** -0.577 ** -0.593 ***
Lagged emplt*PCM*Size100 0.064 *** 0.057 ** 0.059 ***
Price Cost Margin*Size300 1.145 *** 1.213 *** 1.301 ***
Lagged emplt*PCM*Size300 -0.106 *** -0.113 *** -0.121 ***
Pr > chi2 0.000 0.000 0.000
No. of observations 3264 3264 3264
Notes:
(1) * significant at 10% level, ** significant at 5% level, *** significant at 1% level
(2) Errors robust to heteroscedasticity and autocorrelation
(3) All variables in logs, except time; wage, mandays, output, average size, PCM lagged one period
(4) Industry dummies at 2-digit level
(5) Year dummies from 1979 onward, to capture gradual liberalization process, begun in 1979
148
Table 3.5
GLS Estimates: Quadratic Convex Adjustment Costs - Parsimonious
Employment 1976-1982 1976-1997
82-97 & 91-97 period dummies
With 76-82
dummy
NO Market
Variables
Market
Variables
Constant 0.276 *** 0.274 *** 0.171
Wage -0.026 *** -0.030 *** -0.032 ***
Mandays 0.073 *** 0.074 *** 0.071 ***
Output 0.013 *** 0.016 *** 0.018 ***
Lagged employment 0.989 *** 0.989 *** 0.994 ***
Average Emplt. Size(>=100) dummy 0.077 ** 0.199 * 0.189
Lagged employment*Size100 -0.006 -0.018 -0.017
Average Emplt. Size(>=300) dummy 0.635 ** 0.507 * 0.528 *
Lagged employment*Size300 -0.051 ** -0.039 * -0.041 *
Year82 dummy - 0.045 0.036
Lagged employment*Year82 - -0.008 -0.009
Average Emplt. Size(>=100)*Year82 - 0.100 0.102
Lagged employment*Size100*Year82 - -0.009 -0.009
Average Emplt. Size(>=300)*Year82 - -0.926 ** -0.942 **
Lagged employment*Size300*Year82 - 0.074 ** 0.075 **
Year91 dummy - dropped 0.015
Lagged employment*Year91 - 0.002 0.002
Average Emplt. Size(>=100)*Year91 - -0.084 -0.091
Lagged employment*Size100*Year91 - 0.010 0.011
Average Emplt. Size(>=300)*Year91 - 0.578 0.586
Lagged employment*Size300*Year91 - -0.050 -0.051
Price Cost Margin - - 0.031
Lagged emplt*PCM - - -0.002
Pr > Chi2 0.000 0.000 0.000
Observations 952 2992 2992
Notes:
(1) Significant at 1% level***, at 5% level**, at 10%level*
(2) Errors robust to heteroscedasticity and autocorrelation, all variables in logs
(3) Year dummies from 1979 onward, to capture gradual liberalization process begun in 1979,
except with period dummy 1991-97, where year dummies until 1990 included
149
Table 3.6
Estimates: Quadratic Convex Adjustment Costs - Parsimonious (System GMM Estimators)
Employment 1976-1982 1976-1997 1976-1997: Market Var.
82-97 & 91-97 period dum 82-97 & 91-97 period dum
With
76-82
dum
Instru-
ment
set 1
Instru-
ment
set 2
Instru-
ment
set 1
Instru-
ment
set 2
Constant -1.165 1.087 0.927 2.093 *** 2.085
Wage 0.142 -0.251 -0.240 -0.448 *** -0.115
Mandays 0.166 0.400 * 0.383 * 0.288 *** 0.283 *
Output 0.040 0.199 0.138 0.333 *** 0.083
Lagged employment 0.960 *** 0.880 *** 0.974 *** 0.780 *** 0.838 *
Average Emplt. Size(>=100) dummy 0.699 5.426 * 4.476 * 5.735 *** 4.859 *
Lagged employment*Size100 -0.075 -0.558 * -0.458 * -0.593 ** -0.502 *
Average Emplt. Size(>=300) dummy -1.525 0.163 2.384 -0.196 2.812
Lagged employment*Size300 0.159 0.083 -0.130 0.124 -0.159
Year82 dummy - 0.703 0.845 * 0.742 0.862 **
Lagged employment*Year82 - 0.008 -0.253 ** -0.003 ** -0.268 ***
Average Emplt. Size(>=100)*Year82 - 6.294 -10.633 ** 4.847 -8.964 **
Lagged employment*Size100*Year82 - -0.612 1.091 ** -0.467 0.928 **
Average Emplt. Size(>=300)*Year82 - -6.840 2.169 -5.011 0.498
Lagged employment*Size300*Year82 - 0.642 -0.329 0.470 ** -0.169
Year91 dummy - dropped dropped dropped dropped
Lagged employment*Year91 - -0.093 ** 0.159 * -0.085 0.170 **
Average Emplt. Size(>=100)*Year91 - -10.441 7.268 -9.134 4.464
Lagged employment*Size100*Year91 - 1.049 -0.741 0.921 -0.461
Average Emplt. Size(>=300)*Year91 - 12.862 2.620 10.149 ** 4.317
Lagged employment*Size300*Year91 - -1.238 -0.137 -0.988 -0.319
Price Cost Margin - - - -0.021 -0.657
Lagged emplt*PCM - - - -0.010 0.061
Hansen Test: Pr>Chi2 0.216 0.285 0.450 0.214 0.586
Arellano-Bond Test: No AR(1) 0.001 0.000 0.000 0.000 0.000
Arellano-Bond Test: No AR(2) 0.807 0.170 0.271 0.098 0.279
Observations 1224 3264 3264 3264 3264
No. of industry groups 136 136 136 136 136
Notes: (1) Significant at 1% level***, at 5% level**, at 10%level*
(2) Errors robust to heteroscedasticity, all variables in logs
(3) Year dummies from 1979 onward, to capture gradual liberalization process, begun in 1979,
except with period dummy 1991-97, where year dummies until 1990 included
(4) Using lagged instruments leads to losing first three years' observations, i.e., 1973-1975; variables
instrumented are the first panel - lagged wage, mandays, ouput, employment- and market variables
where included; with industry groups no. 136 and multiple lagged instruments, the no. of
variables instrumented for is constrained
(5a) Instrument set 1 also instruments for period dummy 1982 and its interactions
(5b) Instrument set 2 also instruments for period dummy 1991 and its interactions
(6) Lags restricted to account for some serial correlation AR(1) in residual disturbance
(7) No. of observations reports all years used including those used only for lagged instruments
150
Table 4.1
Scope of Retrenchment: Number of Workers
New Hires Rehires Separations By Instrument Employment
Vol-soft Invol-soft Invol-hard Reduction Continent
41002 282307 5033875 577188 752809 1178394 4710566 TOTAL - All cases
1743 15900 909705 97878 140896 104035 892062 Africa
19000 3000 255311 172711 11000 0 233311 Asia
0 0 2853559 105000 132000 883259 2853559 Europe
20259 263407 1015300 201599 468913 191100 731634 Latin America
0 0 37500 4800 1800 0 32900 MEDIAN - All cases
0 0 10061 3696 6861 1922 10061 Africa
0 0 23425 14373 0 0 21925 Asia
0 0 172959 17500 0 42000 172959 Europe
0 0 56409 4599 0 0 26000 Latin America
1025 7058 125847 18037 27882 39280 117764 AVERAGE - All cases
116 1060 60647 7529 14090 10404 59471 Africa
2375 375 31914 28785 2200 0 29164 Asia
0 0 407651 26250 33000 176652 407651 Europe
2026 26341 101530 22400 58614 19110 73163 Latin America
0 0 247 0 0 0 0 MINIMUM - All cases
0 0 247 0 0 0 247 Africa
0 0 6500 0 0 0 6500 Asia
0 0 35000 0 0 0 35000 Europe
0 0 2034 0 0 0 0 Latin America
19000 163059 1661000 112000 424095 547300 1661000 MAXIMUM - All cases
1743 7500 547200 59810 75000 57000 541200 Africa
19000 3000 69466 69466 7000 0 69466 Asia
0 0 1661000 70000 132000 547300 1661000 Europe
18100 163059 424095 112000 424095 100000 405995 Latin America
Source: Author's calculations
151
Table 4.2
Scope of Retrenchment: Financial Costs
FINANCIAL BENEFITS (In $ million) FINANCIAL COSTS (In $ million)
Wage bill Total Cost per Safety Enhanced Severance Total Cost
savings-annual worker ($) Net Pension Payments Continent
2834 4206 na 3268 6098.29 2707.89 12074.10 TOTAL - All cases
96 673 na 224.86 20.00 255.17 500.03 Africa
-153 179 na 48.40 70.73 1280.73 1399.86 Asia
1446 1846 na 2995 5588.36 0.00 8583.02 Europe
1445 1508 na 0.00 419.20 1171.99 1591.19 Latin America
10 17 1085 0.00 0.00 12.50 42.04 MEDIAN - All cases
8 10 1476 0.00 0.00 12.50 20.00 Africa
9 9 1040 24.20 0.00 24.87 50.00 Asia
723 923 616 858.51 25.00 0.00 454.26 Europe
222 222 4735 0.00 0.00 112.00 280.00 Latin America
123 168 4142 192.23 243.93 87.35 402.47 AVERAGE - All cases
11 61 3630 32.12 2.00 19.63 38.46 Africa
-22 26 3651 24.20 14.15 182.96 199.98 Asia
723 923 3817 998.22 1397.09 0.00 2145.76 Europe
289 302 5695 0.00 69.87 167.43 265.20 Latin America
-157 -157 0 0.00 0.00 0.00 0.00 MINIMUM - All cases
0 0 320 0.00 0.00 0.00 2.12 Africa
-157 -157 0 0.00 0.00 0.00 0.00 Asia
298 298 24 6.00 0.00 0.00 6.00 Europe
-15 -15 0 0.00 0.00 0.00 2.30 Latin America
1148 1548 17108 2130 5538.36 1140.00 7668.51 MAXIMUM - All cases
30 542 13166 199.80 20.00 80.31 199.80 Africa
83 350 17108 48.40 70.73 1140.00 1188.40 Asia
1148 1548 14012 2130 5538.36 0.00 7668.51 Europe
1000 1063 16000 0.00 418.30 425.00 530.30 Latin America
Source: Author's calculations
Note: Percentages computed from these tables (e.g. severance as percent of total cost) may
not match percentages given in table 5 that are based on a count of programs.
152
Table 4.3
Distribution of Employment Reduction Method
Type: Percent of Programs: Percent of
Workers:
Involuntary (hard -- layoffs) 41.5 47.0
Involuntary (soft -- enforcement of rules, removal
of ghost workers, etc.)
65.0 30.0
Involuntary (removal of ghost workers) 22.5 3.3
Voluntary 77.5 23.0
Used Both Voluntary and Involuntary (either) 42.5 19.2
Used All Methods 17.5 13.4
Source: Author's calculations.
Note: Percent of workers may differ from those computable from Table 1 on account of
missing values that are subsumed in totals in Table 1. Here, missing values are excluded.
153
Table 4. 4
Distribution of Compensation and Transition Assistance
Type: Percent of programs:
Severance Payment 68.3
Pension Enhancement 29.3
No Direct Compensation 14.6
Safety Net 63.4
Safety Net (Training) 53.7
Unknown 12.2
Source: Author's calculations
154
Table 4.5
Distribution of Targeting
Type: Percent of programs:
Skill-biased 53.7
Age-biased 51.2
Neutral 19.5
Unknown 2.4
Source: Author's calculations
155
Table 4.6
Relationship Between Rehiring and Program Characteristics
Program
Characteristic:
Percent of Programs with Rehiring Percent of Other
Programs
Employment Reduction
Method:
Involuntary (hard) 50.0 39.4
Involuntary (soft) 62.5 65.6
Voluntary 62.5 81.3
Used All Methods 12.5 18.8
Nature of Targeting:
Skill-biased 25.0* 60.6
Age-biased 12.5** 60.6
Neutral 62.5** 9.1
Compensation --
Assistance:
Severance Payment 75.0 66.7
Pension Enhancement 37.5 27.3
No Direct Compensation 12.5 15.2
Safety Net 50.0 66.7
Safety Net (Training) 37.5 57.6
Source: Author's calculations. Table entries provide the percent of programs with indicated
column characteristic that also have indicated row characteristic. * indicates that column
difference in row is statistically significant at 10% level; ** indicates that column difference is
statistically significant at 5% level.
156
Table 4.7
Relationship Between New Hires and Program Characteristics
Program
Characteristic:
Percent of Programs with New Hires Percent of Programs
without New Hires
Employment Reduction
Method:
Involuntary (hard) 0.0** 47.2
Involuntary (soft) 80.0 62.9
Voluntary 80.0 77.1
Used All Methods 0.0 20.0
Nature of Targeting:
Skill-biased 40.0 55.6
Age-biased 100** 44.4
Neutral 0.0 22.2
Compensation --
Assistance:
Severance Payment 60.0 69.4
Pension Enhancement 60.0 25.0
No Direct Compensation 0.0 16.7
Safety Net 0.0** 72.2
Safety Net (Training) 0.0** 61.1
Source: Author's calculations. Table entries provide the percent of programs with indicated column
characteristic that also have indicated row characteristic. * indicates that column difference in row
is statistically significant at 10% level; ** indicates that column difference is statistically significant
at 5% level.
157
Table 4.8
Summary Statistics for Financial Indicators
_____________________________________________________________________
Average Median Min Max % programs
With Net Loss
Estimated Financial
Break-even Period 2.3 1.8 0 10 22
Estimated Payback
Period 2.1 0.8 0 10 -
_____________________________________________________________________
Source: Author’s Calculations
158
Table 4.9
Relationship Between Net Losses and Program Characteristics
Program
Characteristic:
Percent of Programs with Net
Losses
Percent of Other Programs
Employment Reduction Method:
Involuntary (hard) 44.4 40.6
Involuntary (soft) 50.0 68.8
Voluntary 87.5 75.0
Used All Methods 25.0 15.6
Nature of Targeting:
Skill-biased 66.7 50.0
Age-biased 88.9** 40.6
Neutral 0.00* 25.0
Compensation --
Assistance:
Severance Payment 66.7 68.8
Pension Enhancement 55.6** 21.9
No Direct Compensation 22.2 12.5
Safety Net 55.6 65.6
Safety Net (Training) 55.6 53.1
Source: Author’s Calculations. Table entries provide the percent of programs with indicated column characteristic
that also have indicated row characteristic. * indicates that column difference in row is statistically significant at
10% level; ** indicates that column difference is statistically significant at 5% level.
159
Table 4.10
Relationship Between Financial Break-even Period (BEP) Program Characteristics
Program
Characteristic:
Percent of Programs with below
Median BEP
Percent of Programs with Above
Median BEP
Employment Reduction Method:
Involuntary (hard) 50.0 33.3
Involuntary (soft) 66.7 70.6
Voluntary 33.3** 94.1
Used All Methods 0.0 23.5
Nature of Targeting:
Skill-biased 0.0** 66.7
Age-biased 16.7** 66.7
Neutral 66.7** 5.6
Compensation --
Assistance:
Severance Payment 66.7 83.3
Pension Enhancement 16.7 38.9
No Direct Compensation 33.3 11.1
Safety Net 66.7 50.0
Safety Net (Training) 50.0 50.0
Source/Notes: Author’s Calculations Table entries provide the percent of programs with indicated column
characteristic that also have indicated row characteristic. * indicates that column difference in row is statistically
significant at 10% level; ** indicates that column difference is statistically significant at 5% level.
160
Table 4.11
Relationship Between Payback Period (PBP) and Program Characteristics
Program
Characteristic:
Percent of Programs with below
Median PBP
Percent of Programs with above
Median PBP
Employment Reduction Method:
Involuntary (hard) 60.0 18.2
Involuntary (soft) 80.0 81.8
Voluntary 20.0** 100
Used All Methods 0.0 18.2
Nature of Targeting:
Skill-biased 20.0 54.5
Age-biased 40.0 36.4
Neutral 40.0 27.3
Compensation --
Assistance:
Severance Payment 60.0** 100.0
Pension Enhancement 40.0 18.2
No Direct Compensation 40.0** 0.0
Safety Net 60.0 45.5
Safety Net (Training) 60.0 36.4
Source/Notes: Author’s Calculations. Table entries provide the percent of programs with indicated column
characteristic that also have indicated row characteristic. * indicates that column difference in row is statistically
significant at 10% level; ** indicates that column difference is statistically significant at 5% level.
161
Table 4.12
Key Characteristics: Selected Retrenchment Programs
Uganda Peru Argentina
Military Civil Service SUNAT Civil Service India Hungary Ghana Railroads Federal govt.
Targeting
Yes Yes Yes No No Yes Yes No No Skill bias
No Yes Yes No Yes Yes Yes Yes Yes Age bias
- - - Yes - - - - - Neutral
Reduction Method
Yes Yes No Yes No Yes No No No Involuntary - hard
Yes Yes Yes Yes No na Yes Yes Yes Involuntary - soft
Yes Yes Yes Yes Yes na Yes Yes No Voluntary
No No No Yes No No No No No Rehires
No No Yes No No No No No Yes New Hires
Financial Indicators
2.7 Net Loss Net Loss 2.6 Net Loss Net loss 1.82 1.56 0.41 BEP
1.2 na 0.0023 na na na 1.66 na na PBP
Productivity Gains
No na Yes na na na No Yes Yes Organizational
(Monetary) (Quantitative) (Quantitative)
Yes na na na na na Yes na na Worker
(Monetary) (Monetary)
Yes Yes No No Yes Yes Yes No No Safety Net Provision
(most workers (most workers
re-employed) re-employed)
Source: Author's calculations
162
Table 4.13
Conditions before and during programs: deviations from country-specific means
Measure: Five Years
before Program
Four Years before
Program
Three Years before
Program
Two Years before
Program
One Year before
Program
During Program
Unemployment rate -0.01 -0.26 -0.05 1.09 0.87 -0.18
Real GDP Growth
Rate**
-0.05 0.82 -0.70 -3.25 -0.41 -2.08
Deficit/GDP
**
0.012 -0.005 -0.008 -0.010 -0.012 -0.034
Domestic
Debt/GDP**
-0.036 -0.049 -0.041 -0.036 -0.063 -0.129
Foreign Debt/GDP -0.050 -0.058 0.026 0.055 0.098 0.149
Government
Spending/GDP**
0.020 0.021 0.021 0.014 0.001 0.007
Source/Notes: Tabulations based upon a variety of sources including the Penn World Tables, World Bank World Tables, the International Finance
Statistics, and the Rama (1995) Cross-Country Labor Market Database. See text for methodology. The deficit numbers are equal to expenditures less
revenues and ignore any grants and/or loans. * indicates that the F-test for the null hypothesis that all of the coefficients in the row are equal is rejected
at the 10 percent level. ** indicates that the F-test is rejected at the 5 percent level.
163
Table 4A.1: Summary Statistics for all Countries
Employment (In Numbers) Continent
By Instrument Employment
Vol-soft Invol-soft Invol-hard Reduction Country/Case NO.
Africa
3300 5721 1040 10061 Benin Civil Service 1
- - - 1585 Burkina Faso Joint Railway Line 2
3696 - 2804 6500 Cameroon Civil Service 3
247 - 0 247 Cape Verde Public Enterprises 4
725 2275 0 1100 Central African Republic Civil Service 5
0 8000 - 8000 Congo Civil Service 6
0 - - 541200 Ethiopia Military 7
59810 14000 0 73810 Ghana Civil Service 8
5800 25000 - 30800 Kenya Civil Service 9
- - - - Malawi Civil Service 10
- - - 1900 Mauritania Public Enterprises 11
0 0 57000 49500 Namibia Military 12
4300 1800 0 4357 Senegal Civil Service 13
6000 1100 20852 27452 Sierra Leone Civil Service 14
5000 75000 11339 91339 Uganda Civil Service 15
9000 8000 11000 44211 Uganda Military 16
Asia
21250 4000 0 22250 Bangladesh Jute Sector Public Enterprises 17
- - - 50000 Cambodia Civil Service 18
0 7000 - 7000 China Shenyang Region Reform 19
69466 0 0 69466 India Public Enterprises 20
- - - 21600 Lao PDR Civil Service 21
7495 0 0 7495 Pakistan Public Enterprises 22
6500 - 0 6500 Pakistan Sindh Region Reform 23
68000 0 0 49000 Sri Lanka Civil Service 24
Europe
0 0 253000 253000 Albania Public Enterprises 25
- - - 1661000 Hungary Public Enterprises 26
- 132000 40959 172959 Kazakhstan Public Enterprises 27
70000 0 42000 112000 Macedonia Public Enterprises 28
0 0 547300 547300 Poland Public Enterprises 29
- - - 72300 Russian Federation Coal Sector 30
35000 - 0 35000 Turkey Public Enterprises 31
Latin America
30000 42818 0 72818 Argentina Public Enterprises 32
0 424095 0 405995 Argentina Federal Administration 33
4599 0 0 4251 Bolivia Mining - public corporation 34
0 0 100000 0 Brazil Civil Service 35
0 0 91100 91100 Chile Civil Service and parastatal organizations 36
12000 0 0 12000 Colombia Tourism and Transport Ministry 37
40000 0 0 40000 Ecuador Civil Service 38
3000 2000 0 4150 Mexico SOCEFI - Ministry of Trade & Industry 39
112000 - 0 100595 Peru Civil Service 40
- - 0 725 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
Note: In several cases, the number of workers separated by instrument do not add up to total
separations owing to partial availability of information.
164
Table 4A.1: Summary Statistics for all Countries (contd.)
Employment (In Numbers) Continent
New Hires Rehires Separations Employment
Reduction Country/Case NO.
Africa
0 0 10061 10061 Benin Civil Service 1
0 0 1585 1585 Burkina Faso Joint Railway Line 2
0 0 6500 6500 Cameroon Civil Service 3
0 0 247 247 Cape Verde Public Enterprises 4
0 1900 3000 1100 Central African Republic Civil Service 5
0 0 8000 8000 Congo Civil Service 6
0 6000 547200 541200 Ethiopia Military 7
0 0 73810 73810 Ghana Civil Service 8
0 0 30800 30800 Kenya Civil Service 9
- - - - Malawi Civil Service 10
0 0 1900 1900 Mauritania Public Enterprises 11
0 7500 57000 49500 Namibia Military 12
1743 0 6100 4357 Senegal Civil Service 13
0 500 27952 27452 Sierra Leone Civil Service 14
0 0 91339 91339 Uganda Civil Service 15
0 0 44211 44211 Uganda Military 16
Asia
0 3000 25250 22250 Bangladesh Jute Sector Public Enterprises 17
0 0 50000 50000 Cambodia Civil Service 18
0 0 7000 7000 China Shenyang Region Reform 19
0 0 69466 69466 India Public Enterprises 20
0 0 21600 21600 Lao PDR Civil Service 21
0 0 7495 7495 Pakistan Public Enterprises 22
0 0 6500 6500 Pakistan Sindh Region Reform 23
19000 0 68000 49000 Sri Lanka Civil Service 24
Europe
0 0 253000 253000 Albania Public Enterprises 25
0 0 1661000 1661000 Hungary Public Enterprises 26
0 0 172959 172959 Kazakhstan Public Enterprises 27
0 0 112000 112000 Macedonia Public Enterprises 28
0 0 547300 547300 Poland Public Enterprises 29
0 0 72300 72300 Russian Federation Coal Sector 30
0 0 35000 35000 Turkey Public Enterprises 31
Latin America
0 0 72818 72818 Argentina Public Enterprises 32
18100 0 424095 405995 Argentina Federal Administration 33
0 348 4599 4251 Bolivia Mining - public corporation 34
0 100000 100000 0 Brazil Civil Service 35
0 0 91100 91100 Chile Civil Service and parastatal organizations 36
0 0 12000 12000 Colombia Tourism and Transport Ministry 37
0 0 40000 40000 Ecuador Civil Service 38
850 0 5000 4150 Mexico SOCEFI - Ministry of Trade & Industry 39
0 163059 263654 100595 Peru Civil Service 40
1309 0 2034 725 Peru SUNAT - Tax collecting Authority 41
Source: Author's Calculations
165
Table 4A.1: Summary Statistics for all Countries (contd.)
FINANCIAL COSTS (In $ million) Continent
Cost per Safety Enhanced Severance Total Cost
(a) worker (USD) Net Pension Payments Country/Case NO.
Africa
6424 0.00 0.00 21.20 21.20 Benin Civil Service 1
- - - - - Burkina Faso Joint Railway Line 2
1997 - 0.00 7.38 7.38 Cameroon Civil Service 3
10260 0.06 0.00 2.47 2.53 Cape Verde Public Enterprises 4
- - - - - Central African Republic Civil Service 5
- - - - - Congo Civil Service 6
(d) 365 199.80 0.00 0.00 199.80 Ethiopia Military 7
700 - 0.00 41.87 41.87 Ghana Civil Service 8
3448 - 20.00 0.00 20.00 Kenya Civil Service 9
- 0.00 0.00 20.00 20.00 Malawi Civil Service 10
(b) 4910 - 0.00 9.33 9.33 Mauritania Public Enterprises 11
(d) 658 25.00 - 12.50 37.50 Namibia Military 12
(b) 13166 0.00 - 80.31 80.31 Senegal Civil Service 13
353 0.00 0.00 2.12 2.12 Sierra Leone Civil Service 14
(c) 320 - - 15.79 15.79 Uganda Civil Service 15
(b) 955 - 0.00 42.20 42.20 Uganda Military 16
Asia
2621 - 0.00 55.70 55.70 Bangladesh Jute Sector Public Enterprises 17
(b) 1000 - 0.00 50.00 50.00 Cambodia Civil Service 18
0 - 0.00 0.00 0.00 China Shenyang Region Reform 19
17108 48.40 - 1140.00 1188.40 India Public Enterprises 20
(b) 470 - - 10.16 10.16 Lao PDR Civil Service 21
3318 - 0.00 24.87 24.87 Pakistan Public Enterprises 22
- - - - - Pakistan Sindh Region Reform 23
1040 0.00 70.73 0.00 70.73 Sri Lanka Civil Service 24
Europe
(d) 24 6.00 0.00 0.00 6.00 Albania Public Enterprises 25
(d) 517 858.51 0.00 0.00 858.51 Hungary Public Enterprises 26
- - - - - Kazakhstan Public Enterprises 27
714 - 50.00 0.00 50.00 Macedonia Public Enterprises 28
(d) 14012 2130.2 5538.36 0.00 7668.51 Poland Public Enterprises 29
- - - - - Russian Federation Coal Sector 30
- - - - - Turkey Public Enterprises 31
Latin America
12000 0.00 0.00 360.00 360.00 Argentina Public Enterprises 32
(d) 1002 0.00 0.00 425.00 425.00 Argentina Federal Administration 33
16000 0.00 0.00 73.59 73.59 Bolivia Mining - public corporation 34
- - - - - Brazil Civil Service 35
0 - - 0.00 - Chile Civil Service and parastatal organizations 36
- - - - - Colombia Tourism and Transport Ministry 37
5000 - 0.00 200.00 200.00 Ecuador Civil Service 38
- - - - - Mexico SOCEFI - Ministry of Trade & Industry 39
4735 0.00 418.30 112.00 530.30 Peru Civil Service 40
(b) 1131 0.00 0.90 1.40 2.30 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
(a): Amount of severance per (voluntarily retrenched) worker.
(b): Average severance amount per (all retrenched) worker.
(c): Avg. sev. using all workers (excluding ghost workers, no.=42000)
166
Table 4A.1: Summary Statistics for all Countries (contd.)
FINANCIAL BENEFITS (In $ million) Continent
Other Wages Wages Wages Wage bill Total
New hires Rehires Separatees savings-annual Country/Case NO.
Africa
- - - - 0 0 Benin Civil Service 1
- - - - - - Burkina Faso Joint Railway Line 2
- - - - - - Cameroon Civil Service 3
- - - - - - Cape Verde Public Enterprises 4
- - - - 8 8 Central African Republic Civil Service 5
- - - - - - Congo Civil Service 6
542 - - - - 542 Ethiopia Military 7
- - - 24 24 24 Ghana Civil Service 8
- - - 6 6 6 Kenya Civil Service 9
- - - 10 10 10 Malawi Civil Service 10
- - - - - - Mauritania Public Enterprises 11
- - - - - 34 Namibia Military 12
- - - - 30 30 Senegal Civil Service 13
- - - 1 1 1 Sierra Leone Civil Service 14
- - - - 0 0 Uganda Civil Service 15
- 0 0 17 17 17 Uganda Military 16
Asia
- 0 - - 18 18 Bangladesh Jute Sector Public Enterprises 17
- 144 - 18 -126 -126 Cambodia Civil Service 18
- 0 0 3 3 3 China Shenyang Region Reform 19
- 0 0 83 83 83 India Public Enterprises 20
- - - - 9 9 Lao PDR Civil Service 21
332 - - 18 18 350 Pakistan Public Enterprises 22
- - - - - - Pakistan Sindh Region Reform 23
- - - - -157 -157 Sri Lanka Civil Service 24
Europe
- - - - - - Albania Public Enterprises 25
- - - - 298 298 Hungary Public Enterprises 26
- - - - - - Kazakhstan Public Enterprises 27
- - - - - - Macedonia Public Enterprises 28
400 - - 1148 1148 1548 Poland Public Enterprises 29
- - - - - - Russian Federation Coal Sector 30
- - - - - - Turkey Public Enterprises 31
Latin America
0 - - 237 237 237 Argentina Public Enterprises 32
63 - - - 1000 1063 Argentina Federal Administration 33
0 - - - - - Bolivia Mining - public corporation 34
0 - - - 0 0 Brazil Civil Service 35
- - - - - - Chile Civil Service and parastatal organizations 36
- - - - - - Colombia Tourism and Transport Ministry 37
- - - - - - Ecuador Civil Service 38
- - - - - - Mexico SOCEFI - Ministry of Trade & Industry 39
- - - - 222 222 Peru Civil Service 40
- 16 0 1 -15 -15 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
167
Table 4A.1: Summary Statistics for all Countries (contd.)
PERFORMANCE INDICATORS Continent
Net econ. Payback Net fin. Break-even
gain(loss) period(In yrs) gain(loss) period(In yrs) Country/Case NO.
Africa
- - -21 - Benin Civil Service 1
- - - - Burkina Faso Joint Railway Line 2
- - - - Cameroon Civil Service 3
- - - - Cape Verde Public Enterprises 4
- - - - Central African Republic Civil Service 5
- - - - Congo Civil Service 6
* 365 0.30 * 342 0.40 Ethiopia Military 7
- 1.66 - 1.82 Ghana Civil Service 8
- 3.02 - 3.60 Kenya Civil Service 9
- - - 2.00 Malawi Civil Service 10
- - - - Mauritania Public Enterprises 11
** 115 0.40 ** 48 1.10 Namibia Military 12
- - -409 - Senegal Civil Service 13
- - - 1.87 Sierra Leone Civil Service 14
- - -16 - Uganda Civil Service 15
- 1.20 - 2.70 Uganda Military 16
Asia
- - - 3.40 Bangladesh Jute Sector Public Enterprises 17
- - -1436 - Cambodia Civil Service 18
93 0.00 29 - China Shenyang Region Reform 19
- - -276 - India Public Enterprises 20
- - - 1.10 Lao PDR Civil Service 21
- - # 501 1.44 Pakistan Public Enterprises 22
- - - 0.00 Pakistan Sindh Region Reform 23
- - -1802 - Sri Lanka Civil Service 24
Europe
- - - - Albania Public Enterprises 25
- - -561 - Hungary Public Enterprises 26
- - - - Kazakhstan Public Enterprises 27
- - - - Macedonia Public Enterprises 28
- - -6121 - Poland Public Enterprises 29
- - - - Russian Federation Coal Sector 30
- - - - Turkey Public Enterprises 31
Latin America
- - - 1.56 Argentina Public Enterprises 32
- - - 0.41 Argentina Federal Administration 33
- ## 10.00 - ## 10.00 Bolivia Mining - public corporation 34
- - - - Brazil Civil Service 35
- - - - Chile Civil Service and parastatal organizations 36
- - - - Colombia Tourism and Transport Ministry 37
- - - - Ecuador Civil Service 38
- - - - Mexico SOCEFI - Ministry of Trade & Industry 39
- - - 2.60 Peru Civil Service 40
- 0.002 -47 - Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
* Including arms' imports savings.
** For three year program duration, defence expenditure savings
reversed thereafter.
# Including privatization proceeds.
168
Table 4A.1: Summary Statistics for all Countries (contd.)
ECON. BENEFITS (In $ mill) ECON..COSTS (In $ mill) Continent
Non-finl (productivity rise Total Non-financial Total
Worker Organization (production loss) Country/Case NO.
Africa
- - - - 21 Benin Civil Service 1
- - - - - Burkina Faso Joint Railway Line 2
- - - - 7 Cameroon Civil Service 3
- - - - 3 Cape Verde Public Enterprises 4
- - 8 - - Central African Republic Civil Service 5
- - - - - Congo Civil Service 6
23 - 565 0 200 Ethiopia Military 7
2 - 26 - 42 Ghana Civil Service 8
1 - 7 - 20 Kenya Civil Service 9
- - 10 - 20 Malawi Civil Service 10
- - 0 - 9 Mauritania Public Enterprises 11
67 0 101 - 38 Namibia Military 12
- - 30 - 80 Senegal Civil Service 13
- - 1 - 2 Sierra Leone Civil Service 14
- - 0 - 16 Uganda Civil Service 15
5 - 23 - 42 Uganda Military 16
Asia
- - 18 - 56 Bangladesh Jute Sector Public Enterprises 17
- - -126 - 50 Cambodia Civil Service 18
- 12 15 0 0 China Shenyang Region Reform 19
- - 83 - 1188 India Public Enterprises 20
- - 9 - 10 Lao PDR Civil Service 21
- - 18 - 25 Pakistan Public Enterprises 22
- - - - - Pakistan Sindh Region Reform 23
- - -157 - 71 Sri Lanka Civil Service 24
Europe
- - 0 - 6 Albania Public Enterprises 25
- - 298 - 859 Hungary Public Enterprises 26
- - - - - Kazakhstan Public Enterprises 27
- - - - 50 Macedonia Public Enterprises 28
- - 1148 - 7669 Poland Public Enterprises 29
- - - - - Russian Federation Coal Sector 30
- - - - - Turkey Public Enterprises 31
Latin America
- quantitative ris 237 - 360 Argentina Public Enterprises 32
- quantitative ris 1000 - 425 Argentina Federal Administration 33
- - - - 74 Bolivia Mining - public corporation 34
0 0 0 - 0 Brazil Civil Service 35
- - - - - Chile Civil Service and parastatal organizations 36
- - - - - Colombia Tourism and Transport Ministry 37
- - - - 200 Ecuador Civil Service 38
- - - - - Mexico SOCEFI - Ministry of Trade & Industry 39
- - 222 - 530 Peru Civil Service 40
- 947 933 0 2 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
169
Table 4A.1: Summary Statistics for all Countries (concld.)
Time Period Continent
To From Country/Case NO.
Africa
1994 1991 Benin Civil Service 1
1994 (Ongoing) Burkina Faso Joint Railway Line 2
1994 1989 Cameroon Civil Service 3
1997 1992 Cape Verde Public Enterprises 4
1990 1987 Central African Republic Civil Service 5
1995(Ongoing) Congo Civil Service 6
1995 1991 Ethiopia Military 7
1992 1987 Ghana Civil Service 8
1997 1994 Kenya Civil Service 9
1994 (Ongoing) Malawi Civil Service 10
1994 1990 Mauritania Public Enterprises 11
1994 1991 Namibia Military 12
1991 1989 Senegal Civil Service 13
1997 1992 Sierra Leone Civil Service 14
1994 1992 Uganda Civil Service 15
1995 1992 Uganda Military 16
Asia
1995 1994 Bangladesh Jute Sector Public Enterprises 17
1995/6 (Proposed) Cambodia Civil Service 18
2002 1994 China Shenyang Region Reform 19
1994 1993 India Public Enterprises 20
1993 1989 Lao PDR Civil Service 21
1993 1991 Pakistan Public Enterprises 22
1997 1993 Pakistan Sindh Region Reform 23
1992 1991 Sri Lanka Civil Service 24
Europe
1992 1992 Albania Public Enterprises 25
1992 1990 Hungary Public Enterprises 26
1993 1993 Kazakhstan Public Enterprises 27
1993 1988 Macedonia Public Enterprises 28
1993 1990 Poland Public Enterprises 29
1996 (proposed) Russian Federation Coal Sector 30
1994 1993 Turkey Public Enterprises 31
Latin America
1994 1991 Argentina Public Enterprises 32
1992 1990 Argentina Federal Administration 33
1994 1991 Bolivia Mining - public corporation 34
1991 1990 Brazil Civil Service 35
1977 1973 Chile Civil Service and parastatal organizations 36
1992 1990 Colombia Tourism and Transport Ministry 37
1994 1992 Ecuador Civil Service 38
1992 1989 Mexico SOCEFI - Ministry of Trade & Industry 39
1993 1991 Peru Civil Service 40
1992 1991 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
170
References
Alderman, H., S. Canagarajah and S. Younger, (1994), ”A Comparison of
Ghanian Civil Servants’ Earnings Before and After Retrenchment.” World Bank
Mimeo.
BARE ACT (2005). The Industrial Disputes Act 1947, with amendments,
Professional Book Publishers, New Delhi
Basu, Kaushik, Gary Fields, and Shub Debgupta, (1996), ”Retrenchment, Labor
Laws, and Government Policy: An Analysis with Special Reference to India,” paper
presented at World Bank Conference on Public Sector Retrenchment and Efficient
Compensation Schemes.
Bentolila, Samuel and Bertola, Giuseppe (1990). Firing Costs and Labour
Demand: How Bad is Eurosclerosis?. Review of Economic Studies 57, 381-402.
Bertola, Giuseppe, (1990), ”Job Security, Employment and Wages,” European
Economic Review, 34, 851-886.
Bertola, Giuseppe and Richard Rogerson (1996). Institutions and Labor
Reallocation. NBER Working Paper Series. Working Paper 5828, November 1996.
Blanchard, Olivier J. (1997). Labor Market Flexibility and Aggregate
Employment Volatility: A Comment. Comment on paper by Cabrales and
Hopenhayn (1997). Carnegie-Rochester Conference Series on Public Policy 46,
189-228.
171
Blanchard, Olivier, Simon Commander and Fabrizio Coricelli, (1995),
”Unemployment and Restructuring in Eastern Europe,” in Unemployment,
Restructuring and the Labor Market in Easter Europe and Russia, (Commander and
Coricelli, eds), Washington, D.C.: EDI World Bank.
Blanchard, Olivier and Peter Diamond, (1989), ”The Beveridge Curve,”
Brookings Papers on Economic Activity, 1, 1-60.
Blanchard, Olivier and Peter Diamond, (1990), ”The Cyclical Behavior of
Gross Flows of Workers in the United States,” Brookings Papers on Economic
Activity, 2, 85-155.
Blundell, Richard and Stephen Bond (1998). Initial Conditions and Moment
Restrictions in Dynamic Panel Data Models. Journal of Econometrics 87,115-143.
Blundell, Richard, Stephen Bond, and Frank Windmeijer (2000). Estimation in
Dynamic Panel Data Models: Improving on the Performance of the Standard GMM
Estimator. Advances in Econometrics: Nonstationary Panels, Panel Cointegration
and Dynamic Panels, Volume 15, 53-91.
Caballero, Ricardo J., Eduardo M.R.A.Engel and John Haltiwanger (1997).
Aggregate Employment Dynamics: Building from Microeconomic Evidence.
American Economic Review, 87(1), March 1997.
Caballero, Ricardo J. and Eduardo M.R.A.Engel (1999). Explaining Investment
Dynamics in U.S. Manufacturing: A Generalized (S,s) Approach. Econmetrica
67(4), July 1999.
172
Cabrales, Antonio and Hugo A. Hopenhayn (1997). Labor Market Flexibility
and Aggregate Employment Volatility. Carnegie-Rochester Conference Series on
Public Policy 46, 189-228.
Commander, Simon and Farbrizo Coricelli, eds. (1995), Unemployment,
Restructuring and the Labor Market in Easter Europe and Russia, Washington,
D.C.: EDI World Bank.
Commander, Simon, Janos Kollo, Cecilia Ugaz, and Balacs Vilagi, (1995),
”Hungary” in Unemployment, Restructuring and the Labor Market in Easter Europe
and Russia, (Commander and Coricelli, eds), Washington, D.C.: EDI World Bank.
Cooper, Russell and John Haltiwanger, (1996), ”Evidence on Macroeconomic
Complementarities,” Review of Economics and Statistics, 78, 78-93.
Davis, Steven J. and John Haltiwanger (1992). Gross Job Creation, Gross Job
Destruction, and Employment Reallocation.Quarterly Journal of Economics,
107(3), 819-63.
Davis, Steven J., John Haltiwanger, and Scott Schuh, (1996), Job Creation and
Destruction, Cambridge: MIT Press.
Davis, Steven J. and John Haltiwanger (1998). Gross Job Flows. Handbook of
Labor Economics, edited by Orley Ashenfelter and David Card (1998).
Datt, Ruddar (2000). Lockouts in India - A Case Study of West Bengal, Indian
Institute for Human Development, New Delhi.
173
Datta, R.C., (TISS) (2001). Life After Voluntary Retirement - Examining the
Human Face of the Structural Adjustment Programme, Tata Institute of Social
Sciences, Mumbai.
Debroy, Bibek and P.D. Kaushik, editors (2005). Reforming the Labour Market,
Academic Foundation, New Delhi.
Deshpande, L.K., Alakh Sharma, Anup Karan, Sandip Sarkar (2004).
Liberalisation and Labour, Institute for Human Development, New Delhi
Domowitz, Ian, R. Glenn Hubbard and Bruce Petersen (1988). Market Structure
and Cyclical Fluctuations. The Review of Economics and Statistics. Volume 70
(1), February 1988, 55-66.
Diwan, Ishac, (1993a), ”Public Sector Retrenchment and Efficient Severance
Schemes.” The World Bank, mimeo.
Diwan, Ishac, (1993b), ”Public Sector Retrenchment and Severance Pay: Nine
Propositions,” The World Bank, mimeo.
Fallon, Peter R. and Lucas, Robert E.B. (1991). The Impact of Changes in Job
Security Regulations in India and Zimbabwe. The World Bank Economic Review,
5(3).
Griliches, Zvi (editor), (1992), Output Measurement in the Service Sectors.
NBER Studies in Income and Wealth, Volume 56. University of Chicago Press.
174
Hall, Robert E., (1995), ”Lost Jobs,” Brookings Papers on Economic Activity,
1, 221-256.
Hamermesh, Daniel (1992). A General Model of Dynamic Labor Demand.
Review of Economic Statistics, 74(4), November 1992.
Hamermesh, Daniel (1995). Labor Demand and the Source of Adjustment
Costs. The Economic Journal, 105 (May), 620-634.
Hamermesh, Daniel and Gerard A. Pfann (1996). Adjustment Costs in Factor
Demand. Journal of Economic Literature, Vol. XXXIV (September), 1264-1292.
Haltiwanger, John and Manisha Singh (1999). Cross-Country Evidence on
Public Sector Retrenchment. The World Bank Economic Review, Vol.13 No.1,
May 1999.
Hopenhayn, Hugo A. and Richard Rogerson (1993). Job Turnover. Journal of
Political Economy, 101(5).
Indian Labour Yearbooks, Labor Bureau, Ministry of Labor, Shimla, various
issues.
Institute for Human Development (IHD) (2004). Rehabilitation of Cotton
Textile Mill Workers after the Closure of the Mills in Indore, New Delhi.
Lindauer, David and Barbara Nunberg (editors), (1994), Rehabilitating
Government: Pay and Employment Reform in Africa, Washington, D.C.: World
Bank.
175
Lazear, Edward, (1990), ”Job Security Provisions and Employment,” Quarterly
Journal of Economics, 105, 699-726.
Levy, Anat and Richard Mclean, (1996), ”Optimal and Sub-Optimal
Retrenchment Schemes:: An Analytical Framework,” paper presented at World
Bank Conference on Public Sector Retrenchment and Efficient Compensation
Schemes.
Mazumdar, Dipak, (1989), ”Microeconomic Issues of Labor Markets in
Developing Countries: Analysis and Policy Implications,” EDI Seminar Paper No.
40, World Bank.
Mills, Bradford, David E. Sahn, Edward E. Walden, and Stephen D. Younger,
(1993), ”Public Finance and Employment: An Analysis of Public Sector
Retrenchment Programs in Ghana and Guinea,” mimeo.
National Labour Commission (2002). Report of the National Commission on
Labour: 2002, New Delhi
Nickell, Stephen (1978). Fixed Costs, Employment and Labour Demand over
the Cycle. Economica, 45(180), November 1978, 329-45.
OECD, (1993), Employment Outlook, July, Paris.
OECD, (1996), Job Creation and Loss: Analysis, Policy and Data Development,
Paris.
176
PA Consulting Group, 1993. Social Safety Net Adjustment Credit Technical
Assistance - Final Report. Submitted to the World Bank.
Papola, T.S. (1994). Structural Adjustment, Labor Market Flexibility and
Employment. Indian Journal of Labor Economics, 37(1), 1994.
Pai, G.B. (2001). Labour Law in India, Butterworths India., New Delhi.
Pfann, Gerard A. and Franz C. Palm (1993). Asymmetric Adjustment Costs in
Non-linear labor Demand Models for the Netherlands and U.K. Manufacturing
Sectors. The Review of Economic Studies, 60(2), 397-412.
Planning Commission (2001). Report of the Task Force on Employment
Opportunities, New Delhi.
Planning Commission (2002). Report of the Steering Committee on Labour and
Employment for the 10th Five Year Plan(2002-07), New Delhi.
PRAGYA (1999). Managing the Negative Impacts of Retrenchments &
Layoffs, Sushant Lok, Gurgaon.
Rama, Martin, (1994), ”Flexibility in Sri Lanka’s Labor Market,” Policy
Research Working Paper 1262, The World Bank, March.
Rama, Martin, (1995), ”Labor Market Cross-Country Database – Glossary,”
mimeo.
Rama, Martin, (1997), ”Efficient Public Sector Downsizing,” mimeo.
177
Rama, Martin and Donna Macisaac (1996), ”Activity, Earnings and Welfare
after Retrenchment: Central Bank Employees in Ecuador,” paper presented at
World Bank Conference on Public Sector Retrenchment and Efficient
Compensation Schemes.
Republic of Uganda, 1994. Management of Change, Context, Vision,
Objectives, Strategy and Plan: Civil Service Reform.
Robbins, Donald, (1996), ”Public Sector Retrenchment: A Case Study of
Argentina,” paper presented at World Bank Conference on Public Sector
Retrenchment and Efficient Compensation Schemes.
Roberts, Mark J. and Tybout, James R. (1996). Industrial Evolution in
Developing Countries. A World Bank Book. Oxford University Press.
Shea, John (1997). Instrument Relevance in Multivariate Linear Modela: A
Simple Measure. The Review of Economics and Statistics, Vol. 79, No.2, 348-352.
Svejnar, Jan and Katherine Terrell, (1991), ”Reducing Labor Redundancy in
State-Owned Enterprises,” PRE Working Paper No. 792, World Bank, October.
World Bank, 1994. Peru Public Expenditure Review. Report No. 13190-PE.
World Bank, 1994. Shenyang Industrial Reform Project Appraisal Report.
Report No. 12415-CHA.
178
World Bank, 1995. Argentina Public Sector Reform Project Completion Report.
Report No. 14781.
World Bank, 1995. Public Enterprise Adjustment Loan Audit Report. Report
No. 14809.
World Bank, 1995. From Swords to Ploughshares: The Uganda Demobilization
and Reintegration of Ex-combatants in Uganda. Confidential Mimeo.
179
doc_185905729.pdf
A marketplace is a location where goods and services are exchanged. The traditional market square is a city square where traders set up stalls and buyers browse the merchandise.
ABSTRACT
Title of Dissertation: LABOR ADJUSTMENT IN AN EVOLVING
MARKETPLACE
Manisha G. Singh, Doctor of Philosophy, 2005
Dissertation directed by: Professor John C. Haltiwanger
Department of Economics
This thesis is about the process of employment adjustment. It studies adjustment costs
and their impact on employment and labor demand. It starts by describing key
characteristics of India's labor market; documents legal, economic, and social framework;
investigates the impact and finds magnitudes of worker adjustment costs. Then, it
estimates econometrically parameters of labor adjustment based on ASI industry data
from 1973 to 1997. Specifically effects of the job security regime in India on
employment are estimated while accounting for the concurrent impact of product markets’
liberalization. It establishes that adjustment costs are substantial, adjustment is slow, and
competition mitigates only some of the adverse impact of adjustment costs. Third, it
investigates retrenchment programs across countries. This one-time or episode form of
adjustment is a preferred mechanism in the presence of excess labor. Multi-dimensional
program designs are found to be more successful. These analyses can be better conducted
using a panel dataset based on the unit level data of the ASI.
LABOR ADJUSTMENT
IN AN EVOLVING MARKETPLACE
by
Manisha G. Singh
Dissertation Submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of requirements for the degree of
Doctor of Philosophy
2005
Advisory Committee:
Professor John C. Haltiwanger, Chair
Professor Judy Hellerstein
Professor Howard Leathers
Professor Michael Pries
Professor John Shea
© Copyright by
Manisha G. Singh
2005
ii
ACKNOWLEDGEMENTS
The research is based on Annual Survey of Industries (ASI) data provided by the Central
Statistical Organization (CSO), Ministry of Planning and Programme Implementation,
Government of India. I thank the officials of the CSO for their time and valuable insights
about the data. I also thank the officials of the Ministry of Labour, Government of India
and the V.V. Giri National Labour Institute, Ministry of Labour, Government of India for
sharing their knowledge and experience with this subject. I appreciate the help provided
by the library staff of the Ministry of Labour in locating reference materials and enabling
their use.
I thank the Department of Economics and the Graduate School of the University of
Maryland, College Park for enabling my research amid difficult circumstances. I am
especially grateful to Professor J ohn Haltiwanger whose guidance and support has been
immense.
Part of this research is sponsored by the World Bank, Washington D.C.. I thank Dr.
Martin Rama for this support and discussions during the initial period of this research.
Many thanks are due to my father for helping me to persist with this endeavor, to my
husband for being a steadfast supporter, and to my friends for their numerous cheerful
contributions.
iii
TABLE OF CONTENTS
Acknowledgements………………………………………………………………… ii
Table of Contents……………………………………………………………………iii
List of Tables………………………………………………………………………...iv
I. Introduction and Overview ………………………………………….. 1
1.1 The Indian Economy …………………………………………………... 4
1.2 The Worker Separation Regime ………………………………………. 7
II. Labor Adjustment: Framework and Evaluation
2.1 Introduction…………………………………………………………… 10
2.2 Framework – Labor Adjustment
2.2.1 Legal Framework………………………………………………..14
2.2.2 Economic Framework – Labor Institutions……………………..23
2.2.3 Social Framework – Support Institutions……………………….35
2.3 Impact of Labor Adjustment
2.3.1 Legal Impact – Enforcement…………………………………….37
2.3.2 Economic Impact………………………………………………...39
2.3.3 Social Impact…………………………………………………….46
2.4 Concluding Remarks…………………………………………………….47
III. Labor Adjustment – An Empirical Evaluation
3.1 The Issues and Literature ………………………………………………. 49
3.1.1 Nature of Adjustment Costs……………………………………...50
3.1.2 Alternative Structure of Adjustment Costs………………………51
3.1.3 Evidence on Adjustment Costs………………………………......55
3.1.4 Objectives..………………………………………………………57
3.2 Data and Background
3.2.1 Dataset…………………………………………………………...58
3.2.2 Data Description………………………………………………....62
3.3 The Model and Regression Equations ………………………………… 64
3.3.1 Symmetric Quadratic Convex Adjustment Costs………………. 65
3.3.2 The Regression Equations ……………..……………………….68
3.3.3 GLS and System GMM Estimators …………………………….74
3.4 Results and Findings …………………………………………………… 76
3.4.1 Changes in the speed of adjustment – GLS ……………………..78
3.4.2 Changes in the speed of adjustment – System GMM Results.....81
3.5 Concluding Remarks……………………………………………………..83
IV. Public Sector Retrenchment Programs
4.1 Introduction …………………………………………………………… 85
4.2 Conceptual Framework and Measurement Issues ……………………. 88
4.3 Data Collection ………………………………………………………... 95
iv
TABLE OF CONTENTS (concld.)
4.4 Survey and Analysis of Cross country Evidence
4.4.1 Basic Facts ……………………………………………………… 97
4.4.2 The Relationship Between Rehires, New Hires and Other Program
Characteristics …………………………………………………… 101
4.4.3 Summary Financial Indicators ………………………………….. 103
4.4.4 Highlights of Individual Programs ……………………………… 107
4.4.5 Aggregate Factors ……………………………………………... 117
4.4.6 Missing Pieces -- Measurement and Characterization of Adjustment
Costs ………………………………………………………………….. 119
4.5 Concluding Remarks …………………………………………………. 121
V. Conclusions………………………………………………………………... 124
Tables …..………………………………………………………………………… 128
References ………………………………………………………………………… 171
v
LIST OF TABLES
2.1A Employment Aggregates……………………………………………….128
2.1B Growth Rates – Employment and GDP……………………………….. 129
2.2 Employment in Organised Sector………………………………………130
2.3 Disputes Data…………………………………………………………...131
2.4A Enforcement – Industrial Tribunals and Labour Courts – Dispute
Resolution………………………………………………………………132
2.4B Enforcement – Other Acts……………………………………………...133
2.4C Enforcement – Standing Orders (Employment) Act…………………... 134
2.5 Workers’ Adjustment Assistance – National Renewal Fund (NRF) ..…135
2.6 Adjustment Costs – Labor Restructuring……………………………… 136
2.7 Industrial Disputes Act 1947 with Amendments ………………………138
3.1 Summary Statistics…………………………………………………….. 142
3.2 India: J ob flows ………………………………………………………. 143
3.3 J ob Flows: Comparisons ……………………………………………… 144
3.4 GLS Estimates: Quadratic Convex Adjustment Costs …………………145
3.5 GLS Estimates: Quadratic Convex Adjustment Costs – Parsimonious ..149
3.6 Estimates: Quadratic Convex Adjustment Costs – Parsimonius (System-
GMM Estimator ………………………………………………………. 150
4.1 Scope of Retrenchment: Number of Workers ………………………….151
4.2 Scope of Retrenchment: Financial Costs …………………………….. 152
4.3 Distribution of Employment Reduction Method ……………………... 153
4.4 Distribution of Compensation and Transition Assistance ……………. 154
4.5 Distribution of Targeting …………………………………………….. 155
4.6 Relationship Between Rehiring and Program Characteristics ……….. 156
4.7 Relationship Between New Hires and Program Characteristics ……... 157
4.8 Summary Statistics for Financial Indicators ………………………….. 158
4.9 Relationship Between Net Losses and Program Characteristics …….. 159
4.10 Relationship Between Financial Break-even Period (BEP) Program
Characteristics ………………………………………………………… 160
4.11 Relationship Between Payback Period (PBP) and Program
Characteristics…………………………………………………………. 161
4.12 Key Characteristics: Selected Retrenchment Programs ……………… 162
4.13 Conditions before and during programs: deviations from country-specific
means …………………………………………………………………. 163
4A.1 Summary Statistics for all Countries ………………………………… 164
Chapter1: Introduction and Overview
This thesis is about the process of employment adjustment. The specific issue is
jobs and job security. Concern about both have led policymakers in many countries
to legislate regulation preventing job loss and so protect means of livelihood and
economic welfare. Regulation for job security and severance payments may protect
labor but slows labor adjustment. Yet, at times market forces necessitate adjustment
larger than that permitted by regulatory frameworks. In many countries this
adjustment takes the form of retrenchment programs, specially in the public sector
where de facto job security is perpetual. This one-time adjustment accounts for
substantial (latent) redundancy or labor hoarding. However, accumulating excess
labor and limiting adjustment is costly for firms and may have significant adverse
effects. Both continual adjustment in the presence of job security regulations and
episodal adjustment via retrenchment programs are studied. The application for the
former is India with description of the regulatory and economic framework and
direct measurement of adjustment costs; followed by econometric estimation of the
impact of adjustment costs on employment adjustment. Several public sector
retrenchment programs are examined over 37 countries. Some of the key issues are
discussed below.
1
Economists argue that job security motivated regulation is restrictive. Some find
that it is counter-productive because it slows reallocation and subsequent job
creation. Others find that it increases employment. Bentolila and Bertola (1990)
simulations using aggregate European data find that separation costs impact the
separation margin more than the hiring margin. Net employment may increase in
the long run despite mandated severance payments. Hopenhayn and Rogerson
(1993) simulations using parameter estimates from establishment-level US data find
that a firing tax would reduce labor turnover, employment, and average labor
productivity in the long run. Blanchard (1997) summarized that “firing restrictions
lead to a more sclerotic labor market, a market with lower turnover and lower
productivity; but it is not clear at all that they lead to high unemployment.” The
diverse results follow partly from the data - aggregate versus establishment - and
partly from the differing specifications modeling adjustment costs.
In general, the standard symmetric convex costs model is the most widely used
with aggregate and industry data whereas asymmetric piecewise linear and lumpy
adjustment cost models are being used with microeconomic data. A survey of the
various models is presented by Hamermesh and Pfann (1996). Gauging the impact
of adjustment costs is best accomplished with establishment level panel data
2
wherein any discontinuities in the labor demand decision rule and the impact of
idiosyncratic firm level shocks can be accounted for. This is also important in case
of mandated severance payments and job security regulations that directly affect
only the separation margin, i.e., these have asymmetric effects. With industry data,
using these models is inappropriate since the discontinuities, heterogeneity, and
asymmetries would be largely aggregated away. Moreover, using the standard
convex costs model with industry data retains value in that it accounts for net
adjustment costs that are also substantial.
The idea here is to unravel the effects of adjustment costs, specifically that of
mandated severance payments and job security measures, on the process of
employment adjustment. In examining this issue, most studies focus on labor
market institutions. Here, the attention is also on product markets. If substantial
changes occur in the product markets, as in industry deregulation and/or
economy-wide reform, the effects of job security policy and mandated severance
payments may be masked by the impact of these changes. Or it may be that these
costs have no adverse impact on employment. Given that only 28 million
employees of total employment of over 400 million are in the organised sector
which is subject to the job security regulations, any impact may be of marginal
3
value to start with. However, it is the organised sector, specially manufacturing
industries that may have the potential for higher employment generation.
Understanding whether adjustment costs may be dampening job creation is hence
important. Although use of industry data as opposed to establishment data limits
analysis, investigating the issue of job security while accounting for the concurrent
impact of product markets’ liberalization.should shed some more light on how it
works.
A job security regime has been in place since 1976. Gradual deregulation
continued from late seventies to 1991 when a liberalization program was initiated.
Apart from job security and mandated severance payments, other institutional
(unions, wage-setting) and technological adjustment costs (transaction costs,
shopfloor rearrangements) operate in some measure in India. However, the single
most visible change in the past two decades has been liberalization in product
markets. Below are brief descriptions of the Indian economy with product market
changes, and the job security regime.
1.1 The Indian Economy
India is a large democracy with GDP in year 2000 (year 1970) of about $380
4
($94) billion at 1993-94 prices but a low per capita income of $375 ($175). With
low growth of per capita incomes a principal development objective was poverty
reduction through employment generation. Successive five-year development plans
underscored the need for progressive reduction of unemployment especially in the
rural areas and in cottage industry. This emphasis was driven also by the
consideration to encourage wide-spread participation in the development process.
Adoption of big bang development theories led to a large government presence in
the production sector. The Indian economy became a mixed economy with a
substantial public sector (21 percent in 1973–74 to 32 percent in 1997-98) that
controlled almost all of the core sector, namely, mining, oil, power, transport and a
dominant part of heavy manufacturing, for example, steel and capital goods, and
financial services. A large private sector operated in consumer goods and shared the
financial sector. Almost all firms were subject to quantity controls through
extensive licensing. Foreign competition was limited with phases of ownership
dilution of subsidiaries of foreign firms. Protection through prohibitive tariffs and
binding quotas was granted on the basis of the infant industry argument. An
elaborate and complicated system of taxes, subsidies, and transfers grew from years
of such policy practices. Consequently special interest groups emerged which
sought to maintain the special privileges.
5
Economic performance was disappointing with annual growth rates averaging
about 3 percent during 1960s and 1970s. This focused attention on the causes of
lethargy in the Indian economy leading to several research committees during the
seventies. From late-seventies, India began gradual deregulation allowing firms to
rationalize costs and expand in related products. From 1983, firms were allowed to
expand production and enter closely allied product lines, a process called
broad-banding. From 1989, selected industries were delicensed, that is, entry
restrictions diluted substantially. Such selective mechanisms continued till
July-August 1991. A full-scale adjustment and liberalization program was launched
assisted partly by the IMF and the World Bank. The 1991 program aimed at both
stabilization and structural adjustment. It attempted to correct the fiscal and balance
of payments deficits created largely during the latter half of 1980s, in part due to
large foreign short term borrowing, and to build a legal and policy framework to
allow restructuring. In 1991, delicensing was extended to all (except a short
negative list) industries. Foreign trade and investment regimes were liberalized.
Growth responded, rising to about 5 percent per year in 1980s and around 6 percent
per year in 1990s. However, per capita growth rates remain sub-5 percent. With
about a quarter of the population below poverty line, growth acceleration with job
6
creation remains a key economic objective.
There is a shift of emphasis from socialist, poverty-alleviation,
import-competing development strategy to market-oriented, poverty removal &
improving standard of living, and globally competing growth strategy. Promoting
productive employment growth alongwith allowing greater mobility in industrial
employment and developing safety nets for retrenched/displaced labor are being
considered. This is expected to lead to improved reallocation, improved labor
market functioning and growing human capital. In the attempt to develop safety
nets is an implicit acknowledgment of the need to allow greater labor adjustment.
There is also a concurrent need to assess the costs and benefits of the job security
policy in order to enable modifications, if deemed desirable.
1.2 The Worker Separation Regime
The socialist and nationalist accent to policy peaked during mid-1970s. During
this time the labor law was strengthened and forms the current regime that first
came into force March 1976. It requires industrial establishments to seek written
permission of the government for any separations. Prior to this, severance pay was
fifteen days pay per year of service. Since then, severance is subject to negotiation
7
between employers and workers as separations must be induced or bought. To this
extent, the new regime is that of mandated severance payments.
As per the Industrial Disputes Act (IDA) 1947, retrenchment of workers with
more than 240 days of service requires one month notice in writing stating reasons
for retrenchment and 15 days’ average pay per each year of service as
compensation. The 1976 Amendment to IDA-1947 required “lay off, retrenchment
and closure illegal except with the written permission of the government.” Initially
it covered establishments employing more than 300 workers on an average day in
the 12 months’ prior to the amendment. Another amendment in 1982 extended
coverage to establishments with 100 or more workers. The penalty for retrenchment
or closure without permission includes a fine and/or a prison sentence.
According to Mathur (1989), the number of closures diminished during 1980s,
lock-outs increased, and unions put up effective opposition to “casualization” of
workers. However, adjustment came via increasing recourse to induced voluntary
quits, since this required no government permission. This is costlier and one month
pay for each year of service is common. Voluntary retrenchment programs,
enabling shedding hoarded and excess labor, document the severance pay formulae
8
in chapter four.
The continual and episodal forms of adjustment are dealt with on two levels.
First, chapter 2 starts by describing key characteristics of India’s labor market and
employment situation; documents legal, economic, and social framework; and
investigates the impact and magnitudes of employer and worker adjustment costs.
The labor market institutions - labor law, unions, wage-setting, labor
administration, support (such as pension) institutions - are discussed. Given this
framework, the impact of labor adjustment on workers, employers and the economy
is evaluated. Chapter 3 examines econometrically parameters of labor adjustment
based on ASI industry data from 1973 to 1997. Second, chapter 4 also begins with
description of 41 public sector retrenchment programs over 37 countries and then
provides empirical insights based on these 41 observations. Chapter 5 concludes.
Study limitations and suggestions for further research are also discussed. The
emphasis is on using establishment or unit-level or firm-level panel datasets. The
industry data based estimates are revealing but the magnitudes may well vary
substantially at the unit level.
9
Chapter 2: Labor Adjustment - Framework and Evaluation
2.1 Introduction
India is a labor abundant country. In over 50 years since independence in 1947,
agricultural employment remains dominant at more than 50 percent (see table
2.1A). Further, despite labor abundance, employment has not kept pace with GDP
growth (see table 2.1B). While GDP growth picked up from about 3 percent per
annum up to 1970s to about 5.8 percent in 1980s and 1990s, employment growth
has been at most 2.4 percent per annum. Even this rate was achieved during an
investment boom in 1992-97 after economic reforms in 1991. As per Planning
Commission (2001) the employment elasticity to GDP for all sectors has declined
from 0.53 (during 1977-78 to 1983) to 0.41 (1983 to 1993-94) and further to 0.15
(1993-94 to 1999-2000).
Organised sector employment offering quality employment, that is, a living
wage with medical and retirement benefits, decent working conditions, and some
employment-tenure security grew at just over 2 percent per year in most years until
1983 (see table 2.2). Since then, this growth has dwindled and since 1998 has been
negative. Only one year, 1996, saw spectacular (given other years’ performance)
employment growth of 5.6 percent. This was in the private sector during the
10
investment boom. For the key contributing sector of manufacturing, employment
elasticity has fallen from 0.67 (1977-78 to 1983) to 0.33 (1983 to 1993-94) and
further to 0.26 (1993-94 to 1999-2000). Although services’ sector employment
elasticity are still higher than manufacturing, these are also decreasing, for instance,
the best performing financial sector elasticities are1.00 to 0.92 to 0.73 for the
corresponding periods.
Though the services sector would continue to absorb persons, these persons are
overwhelmingly likely to be the educated and the skilled. The avenues for the
semi-skilled and the unskilled, with at most secondary education, are to learn on the
job in manufacturing and grow. Planning Commission (2001) reports that over 80
percent of the urban population and around 90 percent of the rural population
possess no skill as per the National Sample Survey Organization (NSSO) data for
1993-94. Given declining employment elasticity in manufacturing, where are these
persons to go? Quite a few find relief in rural wage employment programs wherein
over 4 million person-years of employment per year has been generated during the
1990s. To juxtapose this with regular employment growth, if manufacturing
employment elasticity were to bounce back to 0.67 and GDP growth is 6.5 percent
per year, then using manufacturing employment of 48 million in 1999-2000 as base,
11
increase of just over 2 million manufacturing jobs is obtained. However, these jobs
lift persons out of poverty and enable their growth, hence their importance.
The situation becomes more urgent given that both the public sector and
agriculture are shedding labor. Agriculture shedding labor is the usual consequence
of development and of more attractive opportunities emerging elsewhere. Public
sector labor shedding has its genesis in over-employment. The excess labor cannot
be sustained due to fiscal deficits (government) and increasing competition
post-reforms (public sector enterprises). See chapter 4 for more details on public
sector retrenchment.
The surplus labor phenomenon extends beyond the public sector to the private
sector. The genesis here lies in protected product markets and even more protected
labor markets. The extent of redundancy in the organised sector in 1990-91 is
estimated to be a conservative 16 percent (PRAGYA 1999). With the organised
sector employment being about 27 million in 1990-91 (table 2), redundancy works
out to over 3 million workers. With normal growth, industrial restructuring induced
reallocation of workers may be accommodated within the sector from loss-making
activities to value-adding activities. This has an added effect of using
industry-specific skills of displaced workers. However, with accumulated surplus
12
labor of this magnitude, any redeployment within the sector/industry would likely
be swamped. Since the government is itself in re-sizing mode and industrial
enterprises in technological upgradation and catch-up (with international standards)
phase, opportunities in other industries for displaced workers would be scarce.
Enterprises also face increasing competition, external and domestic, following
deregulation and liberalization of (i) entry with the abolition of licensing in 1991 in
most sectors, and (ii) of foreign trade and investment. Indian enterprises have gone
through years of cost-cutting that among other things (soft interest rate regime,
lower oil prices) helped these to become profitable during 1990s and early 2000s
despite hardening competition.
This chapter investigates issues in labor adjustment in India, beginning with the
framework - legal, economic, and social - within which this adjustment takes place.
The legal framework discusses the key labor law - the Industrial Disputes Act in
1947 - along with other germane labor laws. The economic framework describes
the myriad labor market institutions such as wage-setting, unions, and labor
administration that influence employment adjustment. It also outlines formal
institutions that aid workers during unemployment spells and during retirement; the
former are important for continual adjustment and the latter for voluntary
13
retrenchment (frequently amounting to voluntary early retirement) based
adjustment. These influence worker adjustment costs and indirectly employer
adjustment costs.
2.2 Framework - Labor Adjustment
2.2.1 Legal Framework - Labor Laws
The state’s intervention in the labor market in India dates back to 1920. Starting
then three laws were enacted, namely, the Trade Disputes Act in 1920, the
Industrial Employment Standing Orders Act in 1946, and the Industrial Disputes
Act in 1947. Several amendments have modified and expanded the scope of this
last key law. These are listed in a chronological sequence in table 2.7 and are
discussed below.
The Trade Disputes Act 1920 established courts of enquiry and conciliation
boards but provided for no machinery for settling of industrial disputes. It forbade
strikes in public utility services without a month’s notice in writing. No norms
dealing with involuntary separation or other industrial disputes were set.
Subsequently, in 1929 this act was amended and provisions for state intervention in
peacefully settling industrial disputes were made through the conciliation apparatus.
14
During World War II, as emergency measures, the Defense of India Rules
empowered the government to appoint industrial tribunals and enforce the tribunal
awards. In 1946 the Industrial Employment Standing Orders Act provided for
devising, framing, and certifying of standing orders covering service conditions;
and making these known to the workers. Throughout these changes, the British
government in India followed a policy of “laizzez faire” with interventions being
selective.
The government of independent India enacted the Industrial Disputes Act 1947.
It embodies essential principles of the Defense of India Rules regarding industrial
disputes and certain principles of the Trade Disputes Act 1929 about investigation
and settlement of industrial disputes. It sought to achieve social justice on the basis
of collective bargaining, conciliation, arbitration, or finally compulsory
adjudication. The instruments for these processes were to be works committees,
joint management councils, recognized unions; regional and national labor
commissioners; regional and national industrial tribunals, and labor courts. Strikes
and lockouts were prohibited while proceedings remained pending in the tribunals.
Over the years up to 1976, several amendments added new requirements, such as
maintenance of muster rolls by the employer, compensation for layoff and formula
15
for retrenchment based on the length of service.
In 1976, the Industrial Disputes Act was made stringent in that any
retrenchment required prior permission from the appropriate government. This was
further amended and restricted in 1982 by broadening the applicability and by
bringing closures within the ambit of prior permission also. Thus, changing the use
of labor in response to demand conditions was no longer the domain of a producer
but firmly in the realm of an industrial dispute.
Other acts germane to labor adjustment are the Contract Labor (Regulation &
Abolition) Act 1970, the Trade Unions Act 1926, and the Sick Industrial
Companies Act.
The Industrial Disputes Act (IDA) 1947 and Key Amendments
In the first few years, the IDA 1947 specified and established the legal
mechanisms defining industrial disputes, to address them, and to resolve them. In
1953, several textile mills were reportedly planning closing down of one shift. In
most mills, worker rotation was not the norm, hence workers were assigned to one
shift and remained in it. Further, textile mills were the largest employer in the
organised sector, largely setting the terms and practices of labor use and labor
16
relations. Closing of one shift meant that all workers in that shift would be rendered
unemployed. In response to an imminent public disaster, the government amended
the law.
Newly added was chapter VA that specified compensation in case of lay-off at
50 percent of wages and dearness allowance (inflation linked) for 45 days during
any 12 months. The eligibility was that the worker’s name was borne on the
establishment muster rolls and service was continuous for at least one year. Also
added was the requirement of muster rolls to be made available for entry by the
employer since entry recording the worker’s availability for work at the
establishment was mandatory for lay-off compensation. However, the law also
provided for exceptions to the general provision for lay-off compensation: (i)
employer offered alternate employment in the same or any other establishment, (ii)
worker did not present himself for work at the appointed time, and (iii) the lay-off
was due to strike or slow-down of production in another part of establishment.
Regarding retrenchment, employers were required to give one month’s written
notice stating reasons for retrenchment. Mandated was compensation at fifteen
days’ average pay for each year of continuous service completed or any residual
17
part in excess of six months. Retrenchment compensation was to follow a simple
yardstick of length of service on the grounds of a humane public policy and the
consideration that involuntary employment may result in general economic
insecurity. This was not applicable to closures. Employers responded by resorting
to closures to address economic redundancy.
In 1956, to check indiscriminate resort to layoff and retrenchment via closures,
an IDA 1947 amendment provided that termination of employment due to closure
or transfer be considered as retrenchment. Thus, compensation was legally required
for closures also. Subsequent to retrenchment, the employer was to offer an
opportunity for re-employment of retrenched workers if further employment
proposed. Layoff compensation was to extend beyond 45 days to all days laid off,
unless an express agreement between the employee and the employer limits it to the
statutory 45 days.
Further, the recovery of dues to workers by the government may be treated in
the same manner as arrears of land revenue or as public demand by the government
on an application to it by the worker/s. This included compensation for layoff and
retrenchment as also money due under settlements and awards.
18
Occurrence of redundancy due to market conditions continued and employers
resorted to closures as also layoff and retrenchment. In an effort to save jobs,
another IDA 1947 amendment in 1972 required employers to give 60 days’ notice
to state governments prior to closure. The idea was to enable the government to
take remedial steps to prevent job loss. Frequently, the state response was
nationalization of the unit and continued operation of the loss-making unit. The
employer was bought out but the resources remain locked in an unproductive
activity. Employment, albeit in a loss-making but state-owned unit, delayed
transition to a productive job. An implicit social insurance from the government
(soft budget constraint for the unit) reduced worker adjustment cost and incentive to
move while uncertainty about future jobs with little training avenues also reduced
potential benefit from moving.
In 1976, chapter VB was added to IDA 1947. The key provision “lay off,
retrenchment and closure illegal except with the written prior permission of the
government” covered establishments employing more than 300 workers on an
average day in the 12 months’ prior to the amendment. In 1982, another
amendment to IDA 1947 extended coverage to establishments with 100 or more
workers. The penalty for retrenchment or closure without permission included a
19
prison sentence and a fine. This extreme form of job security - prior permission
required for any layoff - altered the form and magnitude of adjustment in the labor
markets as discussed in the next section on impact. The IDA 1947 specified that
retrenchment means the termination by employer of the service of a worker for any
reason other than disciplinary action; excludes termination upon voluntary
retirement, superannuation, non-renewal of a contract, continued ill-health, and
cases where the appointment letter stipulates discharge from service without notice
or reasons.
Employers’ resorted to increasing the workforce mainly by using contractual
workers as opposed to hiring permanently in order to limit adjustment costs during
a downturn. This was opposed by workers and unions. Subsequently, judicial
rulings deemed non-renewal of a contract as retrenchment and liable for
retrenchment compensation. With most avenues for adjustment blocked, the
employers used Voluntary Retirement Schemes (VRS) and its versions wherein
workers were induced to exit by offer of compensation. The formula was along the
lines of that specified under the law. However, depending upon the financial health
of the firm and its prospects, the days of average pay per year of service mostly
varied from 30 to 45 days.
20
Employers facing closure of the unit also sought to force the issue using
lockouts. Though illegal, this form of retrenchment bailed out the employer through
government intervention (nationalization). However, the workers were left hapless
and the resources remained locked in. The land of most of the older textile mills
could have been sold and dues to creditors and workers settled. However, owing to
suspected criminality, this was banned in the 1970s and has only been allowed in
May 2005 following a Supreme Court ruling to that effect. As discussed in the
economic impact sub-section, this offers relief to workers some of whom have been
waiting for over 10 years.
Other Labor Use related Acts
The Contract Labour (Regulation and Abolition) Act 1970 was intending to
regulate the practice of contract labor to prevent exploitation. Supreme Court
judgements served to establish, however, that where outsourcing services are
performed on the premises, the act makes the employer liable to giving permanent
status to such workers who perform these services. The Contract Labor Act 1970 is
proposed to be liberalized - allowing for outsourcing of peripheral activities without
attendant risk of required absorption as permanent employees, thus retaining
21
flexibility for the employer while allowing the outsourcing agency and staff to
specialize more and generate greater efficiency gains and reduced costs.
The Trade Unions Act 1926 allows any seven workers in an enterprise to form
a trade union and register it. This has led to multiple unions in one unit. The
employers are empowered to choose the union that will represent the workers,
giving them the power to use the multiplicity to their advantage by selecting a
weaker union. Further, half of the office bearers of the union may be outsiders
further eroding representativeness of the union.
The above discussed laws provide a high degree of protection to workers. The
Sick Industrial Companies Act (SICA) affords protection to entrepreneurs. Under
this act, sick or loss making companies would be referred to the Board for Industrial
and Financial Reconstruction (BIFR). The BIFR would recommend either
rehabilitation through assistance to save jobs or closure for unviable units. In
practice, closures happened as an extreme measure after years of rehabilitation
failed to produce results. This also allowed crooked managements time to divert
funds from the unit. Ultimately, the units would be rendered unrevivable and the
jobs would be lost. The workers’ adjustment to productive jobs and hence more
22
stable employment is needlessly prolonged. SICA’s adverse effects are alleviated to
some extent by allowing creditors, namely, banks to take control of debtor firms’
assets if firms should default on amortization payments for an extended period.
Settlement of outstanding dues accords priority to worker dues.
2.2.2 Economic Framework - Labor Institutions
Producers can either adjust wages or employment. Adjusting wages would clear
the labor market and obviate the need for massive wage employment support
programs. However, efficiency wages prevail in almost every segment of the labor
market. Further, wage-setting in India being largely exogenous to employers, their
primary response is in adjusting employment.
Wage-setting
Wages are subject to several regulations - minimum wages, acts relating to
payment of wages and payment of bonus (a statutory minimum of 8.33% to a
maximum of 20% of annual wages, thereby raising labor cost by an additional
month). The government’s overarching role dilutes the role of bilateral
employer-employees negotiations or collective bargaining. Another direct influence
is through establishment of tripartite wage boards for some industries whose
primary role is working out a wage structure. Increments to wages (through
23
dearness allowance) in the organised sector are indexed to inflation.
Further, informally, public sector being the dominant employer in the organised
sector (of 28 million employed in the organised sector, about 16 million are in the
public sector), it influences wage movements in the private sector too. Although
central public sector undertakings (PSUs) have some autonomy and collective
bargaining is allowed, actual agreements have to be consistent with the detailed
guidelines issued by the Department of Public Enterprises (DPE, the nodal
coordinating interface between the PSUs and the government). The state PSUs have
independent guidelines but emulate the central PSUs. In practice, both the wage
boards and the DPE use as reference points the Central Pay Commissions’ (set up
periodically) recommendations for salaries of central government employees. This
ripple wage-setting effect emanating from a central government institution extends
to the private sector too.
In the informal or unorganised sector, the central and state governments
intervene by fixing minimum wages for an eight-hour work day applicable to
certain occupations where workers are considered vulnerable. Minimum wage
varies across states, industries and occupations. Awareness of minimum wage
24
appears to be low and multiplicity makes it confusing. Further, since the
unorganised sector is dispersed and workers not unionised, the minimum wage may
not be/is frequently not binding.
Through direct and indirect influences of the government over wage-setting, the
market condition that emerges is that individual employers have little flexibility in
adjusting wages. Thus, employers’ response in labor adjustment is toward
employment adjustment and less to wage adjustment. Impeding workforce
flexibility are regulations restricting separation. Thus, the wage bill, normally a
variable component of costs, has become a fixed cost.
Unions and Contractors
Unions are a pervasive presence in the organised sector. Most units have an
internal union or affiliation to an external union.Union membership increased 150
percent from 1966 to 1979 and stagnates thereafter. In early 1980s, the unions
operating in the Bombay textile mills struck work for 20 months asking for higher
wages (real wage growth was almost zero) and better working conditions. The mill
owners refused to negotiate and ultimately the workers returned without much
change in their conditions.
25
Originating from unions formed pre-independence are three main federations.
The Indian National Trade Union Congress (INTUC) is affiliated to the Congress
(I) political party (that is credited with securing India’s freedom from colonial era
and currently governing); the All India Trade Union Congress (AITUC) is a wing
of the Communist Party of India (CPI) and the Congress of Indian Trade Unions
(CITU) that of the Communist Party of India (Marxist) or CPM; and the Bhartiya
Mazdoor Sangh (BMS) is allied to the the Bhartiya Janata political Party (BJP) that
was in power until 2003. Guided by the self-interest of currently employed
unionised workers, the unions have used their affiliations to block labor law reform.
They demand a safety net for unemployment and redeployment first.
In the unorganised sector employing more than 350 million workers, the
contractor acts as an interface between workers and employers. Now, contractors
are required to be registered with the Labor Commissioner’s office. For workers
displaced from the organised sector, the contractor can be a source of support since
jobs and number of days for which jobs are secured, all flow via a contractor.
Labor Administration - Labor Commissioners
26
The national and regional labor commissioners are the first point of interface of
government with workers and with employers on labor issues. They are entrusted
with the implementation of labor laws. They also provide the first conciliation
platform. During employment distress, illegal separation, and unpaid dues the labor
commissioner can provide redressal. The use of their offices depends upon the
education and awareness level of labor.
Retirement and Disability Funds
Both disability and retirement support are linked to employment. During
unemployment spells, there would be no coverage. Disability, partial or total,
through accidents on the job is covered by the Workmen’s Compensation Act,
1923. Incapacitation due to sickness or due to injury as also maternity leave is
covered by the Employees’ State Insurance Act 1948.
Safety net for retirement exists in several forms. Gratuity is payable since 1972
to workers if in continuous service for 5 years. From May 1994 the coverage
extends to supervisors and managers.
The chief mechanism for retirement support is a contingency fund called the
Employees Provident Fund (EPF). Law requires the institution of mandatory
27
provident funds for employees in factories and establishments. Since its inception
in 1952, the coverage has steadily expanded. Even so, it covers less than 7 percent
of factories and establishments. The purpose is to make provisions for retirement
needs of the employee as well as those of the dependents in case of early death.
Contributions required from employees and employers are fixed by the government.
Effective May 1, 1997, the rate is 12 percent of wages with a lower rate of 10
percent for five industries. The central government, on the recommendation of the
board of trustees of the Employees Provident Fund Organisation (EPFO), declares
the rate of interest to be credited annually to the accounts of provident fund
members. The EPF of private enterprises may be self-managed or be managed by
the EPFO on payment of management fees.
Final withdrawals from the EPF are allowed at retirement or prematurely at the
time of resignation. In effect, the EPF operates only for the employed. This
sufficed in a protected market where employment tenure mostly lasted until
retirement. Partial withdrawals are allowed for illness, invalidation, etcetera; and
for discharging social responsibilities such as marriages of dependents, higher
education of children, construction or purchase of a house/dwelling.
The Employees Pension (EP) scheme was introduced effective November
28
1995. Members of an earlier family pension scheme as well as of EPF were allowed
to join this scheme. Upon its introduction, the earlier pension scheme ceases to
operate and the benefits and entitlements of members remain protected and are
continued under the EP scheme. Eligibility is ten years of service with benefits
commencing as early as 50 years of age and vesting fully at 58 years of age. No
additional contributions from either the employee or the employer are required.
The pension fund is set up and a 8.33 percent of wage share of the employer’s
contribution to the employee’s EPF is diverted to it. Further, the government
contributes 1.16% of members’ wage. Withdrawal of own contributions is
permitted.
The scheme provides for a monthly pension to the employee or dependents as
the case may be for (i) superannuation at 58 years of age, retirement, permanent
total disability, (ii) death during service or after superannuation, retirement,
disability, and (iii) children and orphan pension. The employees have an option to
accept an admissible pension or reduced pension with return of capital.
A third scheme in this family is the Employees Deposit Linked Insurance
Scheme, 1976. It provides that in the event of death of a member, the person
entitled to receive EPF accumulations would be paid an additional amount
29
calculated on the basis of the last 12 months’ balance in the employee’s EPF
account.
The organised sector schemes cover only a minuscule section of the
unorganised sector. The informal segments therefore have no formal safety net
supporting them. A Social Security Group Insurance Scheme operated by the
state-owned Life Insurance Corporation (LIC) offers some insurance to unorganised
sector workers in the event of death, permanent total disability, or loss of limbs.
Half the contribution is paid from a social security fund. The central government
operates five welfare funds funded from a cess on producers in the corresponding
sector. Since 2001, the central government operates a pilot scheme for 20,000
agricultural laborers offering life-cum-accident insurance, money-back and the
superannuation benefits.
Transition, Training and Search Support
No unemployment insurance system exists for either organised or unorganised
sector workers. Discussions inform that a pilot unemployment scheme is being
drawn up by the Employees’ State Insurance Corporation. To deal with massive
redundancy and to accommodate the IDA 1947, VRS compensation packages are
30
offered. As stated earlier, the formula to compute VRS is along the lines of that
specified under the law. However, depending upon the financial health of the firm
and its prospects, the days of average pay per year of service mostly vary from 30
to 45 days.
Vocational training at the national level is vested with the Directorate General
of Employment and Training (DGET), Ministry of Labour, the nodal department
for formulating policies, laying down standards, conducting trade testing and
certification, and directly running several training institutions. At the state level, the
state government departments perform these functions for vocational training.
The key institution for imparting vocational training is the Indian Training
Institute (ITI). It has 4274 centers, 1654 in the public sector and the rest 2620 in
the private sector. The trainee capacity is 628,000 seats. According to Planning
Commission (2002), there exist about 19 million dropouts after class VIII and
another 2.2 million after class X or about 21.2 million children looking for
vocational training every year! The available formal training capacity is only 2.3
million or a gap of 18.9 million needs to be filled. To fill the gap, the ITI system
would require modernization and expansion. Supplementary programs such as the
nodal agencies-EACs under NRF or CRR, and leveraging of existing capacity as
31
envisaged under CII Skills Development Initiative would also be required. These
are discussed below.
The National Renewal Fund (NRF) was established in February 1992 and
remained formally in operation from 1992-93 until 2000-01. Its objectives were to
provide funds (i) for retrenchment and/or closure compensation, where necessary,
to affected workers both in public and private sectors, (ii) for retraining and
redeployment of displaced workers, and (iii) for employment generation schemes as
part of safety net development for both organised and unorganised sectors. The
NRF received funds essentially from budgetary allocations though it was open to
contributions from financial institutions, insurance companies, state governments,
and industrial undertakings. It was administered by an Empowered Authority in the
Department of Industrial Development, Ministry of Industry (now Ministry of
Commerce and Industry) with representatives from several ministries, one of
employers, and two external experts of industrial relations. It considered schemes
recommended by either central and state governments or Board of Industrial and
Financial Reconstruction (BIFR). The BIFR is the authority that determines closure
or rehabilitation of companies/units declared sick (loss-making for at least three
consecutive years). The scheme was thus open to part of the private sector.
32
To implement retraining and redeployment schemes, the NRF worked with
Employee Resource Centers (ERC). An ERC documented details about retrenched
workers, provided counseling, and disseminated information regarding retraining
and redeployment possibilities. Any PSU availing of NRF funds was directed to
establish an ERC. More than 30 ERCs were set up during NRF operation. The
retraining was provided by the Employee Assistance Centers (EAC) that were
operated by designated nodal agencies at locations where substantial concentration
of displaced workers was expected. Other than organizing training, an EAC
surveyed affected workers, assessed wage and self-employment opportunities,
provided counseling and escort services to workers, and provided management
information services for NRF. Some of the EACs were Associated Chambers of
Commerce and Industry (ASSOCHAM) at Kanpur (restructuring textiles firms’
hub), Confederation of Indian Industry (CII) at Bombay (also textiles), Gandhi
Labor Institute at Ahmedabad (textiles), Small Industries Service Institutes at
Indore (textiles), and National Small Industries Corporation at Calcutta.
To implement employment generation schemes, Area Generation Councils
(ARC) were established at Ahmedabad, Indore, and Kanpur.
33
The successor to NRF is the Counseling, Retraining, Redeployment (CRR)
scheme from 2001-02 and aimed at assisting the rationalised employees of central
public sector enterprises (CPSEs). It includes a sensitization program. Selected
nodal agencies counsel VRS workers, impart training and reorientation, develop
curriculum/materials, prepare feasibility reports, market surveys, post-training
follow up interface with credit institutions, support in self-employment, regular
liaison with CPSEs, convene meeting of the coordination committee. The CPSEs
help by clearing all dues of the rationalised workers before release. Further, they
provide valuable input in identifying workers’ retraining needs. The cost per 40 day
module is Rupees 7900/? of which Rupees 2400/? is paid at the rate of Rupees
60/? per day to the trainee.
The Confederation of Indian Industry (CII), a major industry association is
leading a “CII-City & Guilds Skills Development Initiative” on a pilot basis to
train and place workers - new entrants and lateral movers. It has drawn up a fairly
long list of skills available. City & Guilds is reportedly the world’s largest
vocational training and awarding body. It is expected to provide technical
assistance as well as certification reflecting international standards. CII intends to
leverage spare capacity in the existing training apparatus, e.g., using ITIs outside
regular hours and during weekends. The cost is targeted at Rupees 5000/? including
34
training and certification to benchmarked global standards approved by CII. In this
program, there is no free-riding since from the very outset the costs are being
shared among all potential employers, with an industry association playing the role
of the coordinator. It helps in integrating the labor market across regions (and
languages and sub-nationalities that exist in the Indian sub-continent, for example,
a Bihari worker from the east may be employable in southern Kerala without a
hefty deduction to a labor contractor) and internationally too.
To effectively deal with rural unemployment, the National Rural Employment
Guarantee (NREG) scheme is being proposed, for about 200 districts, and to be
extended to all 604 districts of India. Its formulation is controversial, the
contentious issues being coverage, eligibility, minimum number of days of wage
employment to be specified, and the wage level. Also under consideration is an
urban equivalent of NREG with a fixed per person expense of Rs. 10,000. Of this
part is earmarked for training and rest for subsequent on-the-job training.
2.2.3 Social Framework - Support Institutions
Social networking among family, friends, and relatives offers critical support in
times of crises. From better placed members, it affords access to credit and provides
35
information on job possibilities as well. In India, social networks are characterized
by workplace, place of residence as also by class, caste, region of origin, and
religion. Traditional segments based on caste and gender are deeply entrenched in
rural areas. Surveys of urban labor markets have found some evidence of
“systematic correspondence” between inherited (caste, gender, urban/rural, father’s
education) and acquired (own education) characteristics on the one hand and labor
status on the other hand. Occupational rigidities perpetuated by caste are eroding
but the speed is debatable since little data is available (caste data not recorded in
census since 1931.)
Shopkeepers or neighborhood-grocers provide credit for consumption that helps
ward off distress. The longer a worker stays at one place, higher is the trust quotient
and consequent access to consumption credit.
In rural areas, the local government or the Panchayati Raj institution with its
head the “sarpanch” play a role in securing employment. The Panchayat is
instrumental in provision of amenities such as drinking water, sanitation, electricity,
roads, and documents such as the ration card (provides access to subsidized food).
In times of displacement, the Panchayat could mobilize support and thus influences
social security. In urban areas, to a limited extent, this role is played by the unions
36
in the organised sector and the contractor in the unorganised sector.
2.3 Impact of Labor Adjustment
2.3.1 Legal Impact - Enforcement
The IDA 1947 provisions slow employment adjustment. Prior permission is
rarely granted. Then, employers can adjust the workforce by seeking recourse to
courts or by unilateral action that would be disputed by workers. Examining
disputes’ data should provide insights, specially in the absence of direct employer
adjustment costs.
The number of disputes, strikes and lockouts, and associated loss of mandays
have varied over the years, increasing and decreasing (see table 2.3). The more
striking fact is that the proportion of lockouts in total disputes has risen from about
15 percent until 1983 to more than 20 percent in 1980s, more than 30 percent in
1990s, and beyond to 40 percent in 2000s. Moreover, the intensity of lockouts as
reflected in the share of mandays lost due to lockouts has also risen dramatically
from about 20-30 percent in 1970s to about 60 percent in most of 1990s and 2000s.
The major reason stated by employers are wages, indiscipline, and others with
retrenchment cited less than 5 percent. However, the timing of the change offers a
37
clue. The IDA 1947 current regime essentially dates back to 1982 when the prior
permission requirement for retrenchment was extended to all units with 100 or more
workers. The pointer to (latent) redundancy or excess labor as the cause of
increasing lockouts proportion is quite indicative though clear evidence from the
data is not there. Direct loss from cost of mandays lost rose from under Rupees 900
million in 1982 to Rupees 3900 million in 2002, almost the same in real terms. The
indirect loss may be in terms of falling employment elasticity.
A survey study by Ruddar Datt (2000) called “Lockouts in India” covered 42
cases in the state of West Bengal and found that employers mis-stated reasons to be
indiscipline and others instead of retrenchment when indeed the lockout was aimed
at downsizing. Putting both this and the above evidences together, the increasing
incidence and intensity of lockouts could be attributed as a response to redundancy.
With retrenchment related disputes increasing, dispute resolution via the labor
courts and tribunals become important. Table 2.4A shows that the percent share of
disputes and applications disposed off by courts and tribunals has been variable
over the years but the declines are more so after 1993. Is this indifferent
enforcement limited to the IDA 1947? Table 2.4B shows that inspections of
premises under the Factories Act 1948 and the Shops and Establishment Act have
38
also been declining since late-1980s/early-1990s. Convictions under the Factories
Act have declined substantially. Prosecutions under the Shops and Establishment
Act have also reduced. Similarly, certification under the Standing Orders Act have
varied but declined more during late 1990s/early 2000s. In this environment of
weakening enforcement, a measure that has helped to expedite law enforcement and
dispute resolution is the use of lok adalats (people’s courts), these helped reduced
case pendency. Delayed enforcement or delayed processing of requirements or
protracted dispute resolution under different labor laws is causing costs to build up
- time delays, legal costs, managerial costs, production loss, missed opportunities.
2.3.2 Economic Impact
Determination of adjustment costs and their impact is gauged through
evaluation of pre- and post-restructuring outcomes: earnings, employment, and
other worker outcomes. Results are compiled here based on several survey studies
on worker outcomes commissioned by the Ministry of Labour, Government of
India. Simple measures of adjustment costs are computed. Data on productivity,
output, and employment in downsized units and the costs borne by the system
(labor administration, judicial costs) are not available. These are evaluated
qualitatively. For a complete evaluation, the entire spectrum needs to be covered.
39
Nonetheless, several insights are obtained from the analysis here.
Characterizing Downsizing and VRS
Common patterns from the Ministry of Labour Survey Reports
Among reasons for undertaking downsizing and VRS program, the employers
cited demand deficiency, increase in (price) competition, modernizing existing
facilities, and expanding business while rationalising costs. Competitive pressure
to improve efficiency causes churn of firms wherein some firms remain weak and
need to be pushed out, specially those units operating below economic scale.
Released workers from weaker units usually have a high probability of being
absorbed in the stronger units due to some usable specificity of skills. Other
resources released - scrap, machinery and land - get better utilized in expanding
productive activities. This churn is normal among different units within an
industry. However, with accumulated (latent) redundancy or excess labor within a
sector, the absorption capacity within an adjusting industry is impaired. To account
for large-scale (latent) redundancy, redeployment has to be across sectors. With
(latent) redundancy pervasive across sectors, redeployment is a challenge. As
shown in table 2.6, post-VRS most workers are self-employed and in the informal
or unorganised sector. Of those employed, most are wage earners in casual jobs.
Excess labor has accumulated due to protective product and labor markets. In the
40
survey studies available, table 2.6 shows that it varies from 30 percent to 100
percent (closure candidates). Recall that in 1990-91, the extent of (latent)
redundancy in the organised sector is 16 percent.
Reasons for workers accepting VRS included poor financial health of the
firm, a perception of fear psychosis fed by the management, age and other factors.
Most VRS schemes are aimed at workers above 40 years of age. Nevertheless,
table 2.6 shows that there is a significant (in some cases, dominant) proportion of
workers above 50 years of age. This has implications for both re-training and
redeployment. However, most workers are also educated to the secondary level
affording them higher employability.
From all survey reports it emerges that the VRS money was used in
non-productive activities such as repaying debts, children’s marriages, and
house-building. While all are legitimate uses of the lump sum, these generate no
income. A major reason for mounting adjustment costs is the failure of management
to provide adequate financial counseling in the use of VRS money. In the TISS
(2001) study, although 54 percent of the MNC sample had invested the VRS money
in a return generating activity, only 14 percent of the NTC sample had done so.
41
Then, redeployment even if in low-paying unorganised sector assumes greater
urgency. Evidence indicates that most displaced workers (95%) are either
self-employed or employed in the informal sector (increasing casualization) or not
working. Only 5 percent could secure formal wage employment. In adjusting to
post-VRS reduced earnings, expenditure was reduced (45% MNC, 21%NTC),
including on health and children’s education. A significant number (more than 10
percent) of NTC mill workers had children drop out of school.
Other data on VRS and downsizing are not available.
Counseling, Training and Search Assistance
Common patterns from the Ministry of Labour Survey Reports
In the cases studied, table 2.6 shows that some workers were provided
counseling and retraining. Among those who could not avail of this benefit, the
major reason was “no information” about avenues of training under NRF. TISS
(2001) reported that the advanced training institute of DGET failed to attract a
single worker from the NTC mills in Bombay. Where the workers could avail of
training, in Indore, the incentive was the stipend of Rs. 2400/? over 40 days. The
training itself is found to be unrelated to the activity taken up post-VRS and
post-training. Indore textile mill workers displaced due to VRS/lockouts/closures
42
are primarily relocated in the informal sector. The conditions in the informal sector
are thus important for the success of restructuring and reform.
From secondary data
On a macro level, the NRF assisted 107 thousand workers by 1997-98 (table
2.5) with over Rs. 18 billion expenditure. About 90 percent of this was utilized for
VRS compensation of covered workers as per PRAGYA (1999) report. The average
compensation works out to Rs. 142,000/? per worker. Experience so far indicates
that NRF, ERCs, EACs, and ARCs were not effective in aiding redeployment. Of
107 thousand workers assisted from the NRF, only about 5 thousand workers were
deployed (table 2.5) by 1997-98. Reorienting existing mechanisms is required to
address the retraining, skill enhancement, and job search needs of displaced
workers. Labor reallocation in India is impeded by the absence of appropriate labor
institutions aiding worker transition to another job or activity.
Social (Income) Security
Along with training and search assistance, income security during the transition
to a productive activity through an unemployment benefit for a specified period
help. An instance is the Textile Workers’ Rehabilitation Welfare Fund (TWRFS)
scheme of financial assistance: 75 percent of wage is given in the first year, 50
43
percent in the second year, and finally only 25 percent in the third year. Such a
benefit must be fully funded either through contributions or taxes. However,
unemployment spell assistance for organised sector does not exist. The Employees
Provident Fund Organisation (EPFO) allows partial withdrawals for several events
(house-building, dependents’ marriages, children’s higher education) but not for
unemployment spells. The key to smoother transition in both formal and informal
sectors is segregation between employment security and income/social security,
with employment security becoming employment mobility.
Adjustment Costs - Worker and Overall Economy
Using the data from the survey studies, we find that worker adjustment costs are
substantial. Comparing pre-separation earnings to post-separation earnings the loss
of earnings in constant 2002 prices is calculated to vary from 53 percent to 77
percent (see table 2.6). One case had a loss of only 6 percent because the private
mill had been brought to closure point and the wages were close to the informal
sector wages. The huge loss in earnings points to dismal alternate earnings. This is
compounded by the absence of investing VRS money in income generating assets.
Most of the workers are above 50 years of age, so the loss of limited productive
years may be acceptable.
44
However, the loss in earnings compelled expenditure cuts including in health
and children’s education expenses. A significant proportion of workers in one case
had children drop out of school and join the labor market. We use average labor
cost per manday in the ASI-census sector as a proxy for organised sector earnings,
and in the ASI-sample sector as a proxy for unorganised sector earnings in the year
of adjustment, that is, 1997. We calculate that the loss in earnings of children
owing to dropping out of school is a large 46 percent (see table 2.6). This
generational effect where income loss is perpetuating down generations may not be
widespread. However, the huge self-earnings loss of workers may have such
effects.
Apart from legal restructuring using VRS compensation to induce worker
separation, two other mechanisms of adjustment have emerged - illegal
restructuring resorting to lockouts and declared restructuring subject to formal
industrial dispute resolution. The proportion of lockouts in industrial disputes (table
2.3) is increasing rapidly. [Two out of eight cases captured the practice (table 2.6)
of employers using lockouts as a de facto means of downsizing. It gets more
credence from and the Datt (2000) finding that firms mis-state indiscipline as cause
45
of the dispute when in truth it is redundancy and retrenchment.] The third method is
the legal but cumbersome one of declaring the cause of dispute to be retrenchment.
Employer adjustment costs by resorting to lockout and subsequent bailout via
BIFR under the SICA law are the least among the three options. Employee
adjustment costs are the least in disputed restructuring. Where unions are strong,
this outcome may dominate. In both these cases the government labor machinery
incurs costs. Also, the assets/resources remain locked in. In the VRS option, both
employer and workers incur costs, government incurs none, and the resources are
freed up for reallocation. The mechanisms of adjustment are changing to
outsourcing instead of expanding operations, inducing voluntary worker exit with
enhanced severance payments instead of simple separations, and lock-outs instead
of closure/exit. Thus adjustment seems to have been possible albeit at a higher cost.
2.3.3 Social Impact
Displacement from a wage/salary job with fair working conditions to variable
low earnings in casual employment or self-employment amidst worse working
conditions would impact economic well-being and perceptions of self-worth,
standing in the family and society. It may engender negative or positive feelings
46
toward VRS and restructuring. From the case studies, such feelings did emerge but
were addressed by social networks, mainly, family. However, in one case of a
lockout, 57 workers committed suicides. Not only is this an economic and social
waste, it influences the future of democratic governments, and thus can undermine
the support to reform itself in India.
2.4 Concluding Remarks
To sum up, in India there exists a great concern about the adverse impact of job
insecurity on workers. This underlies the IDA 1947 provisions and guides many of
the other policies discussed above that have been set in place regarding the labor
market. In other countries, notably Western European, this concern has also led to
high mandated severance payments but not quite so extreme a version of requiring
prior permission for any layoff or retrenchment.
Adjustment costs of restructuring are substantial for both workers and
employers. Post-separation, the workers’ loss of earnings varies from 53 percent
to 77 percent. A generational effect also exists - due to parental job loss and
consequent children dropping out of school, the loss in earnings of children is a
47
large 46 percent. Employer adjustment costs are high in that costlier means of
adjustment are adopted to enable restructuring.
Transition support structure seems geared to one-time adjustment as embodied
in one-time (sometimes repeated) VRS schemes. There exists an ongoing
restructuring process in the economy and hence continual dynamic support to
displaced workers (e.g. market-driven retraining and placement) is necessary. Apart
from training/search support the provision of a fully funded unemployment
insurance is also a pressing requirement.
Though the future benefits in terms of productivity, output and employment
growth cannot be determined without panel data on restructuring firms, qualitative
conclusions may be drawn. The tradeoff between slow adjustment-lower
cost-lower incremental labor absorption and higher adjustment-higher cost-higher
incremental labor absorption (that is, employment elasticity) exists. It can be
partially ameliorated with efficient functioning of transition support institutions.
48
Chapter 3: Labor Adjustment - an Empirical Evaluation
3.1 The Issues and Literature
This chapter focuses on the econometric estimation of the size of adjustment
costs and their impact on employment in India. The effect of exogenous events,
namely, (i) the changes in the job security and mandated severance payments’
regime in 1976, (ii) the expanded coverage of job security regime in 1982 as well as
the deregulation (“broadbanding”) process begun about 1982, and (iii) the
liberalization and economic reforms’ program in 1991 on adjustment costs and
employment is gauged over-time. Further, these effects’ variation by average
employment size is also obtained. The discussion centers on labor. Though
employment refers to all employees, here the emphasis is on workers. In the
literature discussion, workers and employees may be treated as synonymous.
Workers are treated as a homogenous input post-training and hours are
fixed.Capital and investment demand are ignored (see data limitations below).
Substantial evidence (Hamermesh (1993), chapter 7) indicates that estimates of
employment dynamics are not biased if the adjustment of capital is ignored.
(Anecdotal evidence in India indicates that after job security measures are enforced,
some substitution of capital for labor may have taken place. The paper abstracts
49
from this aspect.) Below is a discussion of the nature and types of adjustment costs.
The sections following describe the data; present the model and regression
equations; and the results.
3.1.1 Nature of adjustment costs
Net costs of adjusting labor demand are those of changing the level (or number)
of workers in any productive unit. These include costs due to disruptions in
production (new equipment, shopfloor rearrangements) and other similar costs not
linked to the identity of the workers. Gross costs of adjustment are related to the
flows of workers, i.e., to the identity of workers. These include hiring costs (search
costs - advertising, screening, interviewing), training costs, separation costs
(severance payments, unemployment benefits), and overhead personnel department
costs dealing with recruitments and separations.
Net costs are internal and implicit in nature. Lost output is not measured.
Training of new workers by senior workers may reduce average efficiency during
the transition period of adjustment. Costs unrelated to production and external in
nature, for example, severance payments, may be measurable. Hamermesh and
Pfann (1996) survey presents the following “extremely tentative conclusions” - (i)
external costs of adjusting labor demand are large, amounting to as much as one
50
year of payroll cost for the average worker; (ii) the average adjustment cost rises
very rapidly with the skill of the worker; and (iii) costs of hiring exceed costs of
separations.
Adjustment costs may arise from the economic environment (factor markets,
technology, product markets) the firm faces. However, these may also arise directly
or indirectly from government policies such as mandated severance payments and
mandated advance notice of layoffs.
3.1.2 Alternative Structures of Adjustment Costs
Several forms have been proposed in the literature, under assumptions about the
expectations formation process and that given the expectations-process, the only
reason for slow adjustment is the cost associated with changing labor demand.
The most used structure of adjustment costs in the literature is that of symmetric
convex adjustment costs, usually restricted to be quadratic. Convexity is imposed
on marginal adjustment cost and implies that it is increasing in labor demanded.
Continuity and differentiability ensure that changes in market conditions, however
small, will lead the firm to adjust labor demand continually. Symmetry around zero
ensures that the marginal cost of increasing labor is the same as that of an
51
equal-sized decrease. Sargent (1978) formulated a symmetric quadratic adjustment
cost function and resultant linear labor demand based on it with the assumption of
rational expectations. The quadratic approximation and symmetry imply that even
in the presence of unit-specific shocks, the linearity of labor demand function
allows its aggregation across units. Thus, this form may be applied to aggregated
data with no other consideration and yet retaining its theoretical basis.
The symmetric quadratic form is a convenient tool. However, data from several
countries point to differences in magnitude and persistence between positive input
adjustments and negative ones. A non-linear functional form that has the quadratic
model nested within is proposed by Pfann and Verspagen (1989). It has a parameter
that captures asymmetry, which if negative implies that marginal cost associated
with a negative adjustment or separation exceeds that of a positive adjustment and
vice-versa. If the parameter is zero, it implies that there is no asymmetry and the
symmetric convex costs model applies. Though non-linear, the model can be used
with aggregate data with the assumptions of no idiosyncratic shocks and a fixed
number of firms. Then, a Taylor-series expansion of the firm’s adjustment function
around zero net employment change and subsequent aggregation of these across
firms is identical to the Taylor-series expansion of the aggregate (of firms)
52
adjustment cost function. The quadratic convex cost structure mostly abstracts from
attrition.
Nickell (1978) discussed piecewise linear adjustment costs that are more
relevant for asymmetric gross costs or per worker costs such as severance payments
and interviewing. Gross adjustment costs such as severance payments are distinct
and differ in magnitude from recruitment costs such as interviewing. Further, these
are state dependent, that is, severance payments are incurred only for separations
and hiring costs are incurred only for hiring. Both separations and hiring may take
place simultaneously. In Nickell’s model, at any point, the firm is either expanding
or contracting in response to a corresponding product demand shock over the
business cycle. Attrition is assumed to be zero. His model analyzed regulatory
adjustment costs in the context of the business cycle assuming perfect foresight.
Demand variations over the business cycle are therefore known. Labor use then
varies in terms of employment or hours, though the number of shifts is fixed at one.
According to the paper, empirically, differences between the convex costs model
and the linear costs model “would tend to be washed out” at the aggregate level but
may be picked up at the industry level. These costs engender a discontinuity in the
optimal decision rule. In the presence of idiosyncratic shocks and firm
53
heterogeneity, this form requires unit level data. Application to aggregated data
amounts to assuming away firm heterogeneity. Using industry data captures
inter-industry variation but still loses firm heterogeneity.
In most recent studies of adjustment costs, there is an emphasis on delineating
net costs from gross costs. This stems from prior evidence (Hamermesh and Pfann
(1996)) that the two are different in magnitude and impact, and partly from
evidence on employment flows. Job flows - jobs created and destroyed, by
continuing firms and by entrants and exiters - underlying the net result, that is,
change in employment stock behave quite differently to net change. Davis and
Haltiwanger (1998) document job flows for most developed countries and several
developing countries. The salient features across economies remain that gross job
flows - job creation and job destruction - co-exist in almost every time period and
sector, are substantial, persistent, asymmetric and concentrated at relatively few
plants. The key factor is the existence of heterogeneity among firms as reflected in
the predominance of idiosyncratic factors as opposed to aggregate and sectoral
factors. Gross adjustment costs are related to gross worker flows. Worker flows
exceed job flows to the extent of any attrition. Net costs are related to net
employment change.
54
Models of adjustment costs described above deal mostly with net costs as
opposed to per worker costs. The symmetric convex costs based model is based on
net employment change. The asymmetric convex costs model recognizes that the
costs associated with a net negative or a net positive change may be different and
hence also examines net costs though asymmetric. The piecewise linear costs model
explicitly accounts for gross costs - recruitment costs versus separation costs, each
per worker. The symmetric convex costs model, despite its limitations, is the most
widely used specification with aggregate macroeconomic data (Hamermesh and
Pfann (1996)) and is suitable for disaggregated industry data.
3.1.3 Evidence on adjustment costs
The evidence on adjustment costs, both for labor demand and investment
demand is discussed at length by Hamermesh and Pfann (1996). Here, the emphasis
is on mandated severance payments related papers. Two major approaches emerge.
One uses simulations based on statistics or estimates of parameters, and the other is
the more direct parametric estimation.
Bentolila and Bertola (1990) model a forward looking monopolist firm subject
55
to separation costs and demand uncertainty. Their simulations using aggregate
European data find that separation costs impact the separation margin more than the
hiring margin. Net employment may increase in the long run despite mandated
severance payments, if the shocks faced by the firms are highly persistent and
worker attrition is high. With more persistent shocks, the need to alter labor demand
is lower. With higher worker attrition, the need for separations is reduced. It is
unclear what happens if the shocks are more variable and/or attrition is lower.
Hopenhayn and Rogerson (1993) model a competitive firm also subject to a firing
tax and idiosyncratic productivity shocks in a general equilibrium framework. They
perform simulations using parameter estimates from establishment-level US data
and find that a firing tax would reduce labor turnover, employment, and average
labor productivity in the long run. The opposing results could be due, in part, to
assumptions about market structure (monopolist versus perfect competition) and
inclusion of the entry-exit process in the latter study.
Attempts to explain these contrary conclusions include accounting for other
labor institutions that might be simultaneously operating in conjunction with a tax
on labor adjustment. Bertola and Rogerson (1996) analyze the process of
wage-setting. They argue that relative-wage compression in Europe may be part of
56
the puzzle. It is conducive to higher employer initiated job turnover. Both
institutions together can explain similar job flows in US and Europe but higher
unemployment rates in Europe. Cabrales and Hopenhayn (1997) allow firms to
choose between two coexisting types of contracts - permanent and temporary. Their
calibrations using Spanish data find that with the inclusion of temporary contracts,
there is a substantial increase in reallocation but not in aggregate productivity.
Inferring of adjustment costs within labor demand estimation is attempted by
Fallon and Lucas (1991) based on the symmetric quadratic convex cost
specification and using disaggregated industry data from India. Job security is
instituted via requirements of prior government approval for separation and
severance payments for any separations since March 1976. Fallon and Lucas use a
difference approach by introducing a dummy variable for regime change to gauge
the impact of the severance requirements. They find a significant one-time
reduction in labor demand (intercept) but no significant fall in the speed of
adjustment.
3.1.4 Objectives
This paper attempts to unravel the effects of adjustment costs on employment
57
while accounting for (i) the changes in the job security and mandated severance
payments’ regime in 1976, (ii) the expanded coverage of job security regime in
1982 as well as the deregulation (“broadbanding”) process begun about 1982, and
(iii) the liberalization and economic reforms’ program in 1991, and (iv) average
employment size. The hypotheses are that average employment and speed of
adjustment (a) decrease after job security provisions are more restricted and (b)
increase post-economic reform in product markets despite no further change in
labor law restrictions. Labor demand is systematically related to the degree of
competition; and, the speed of adjustment is also affected by the degree of
competition. The effects are greater for larger firms. Note that after 1982, with
changes in both job security regime and product markets, the net effect can go
either way.
3.2 Data
3.2.1 Dataset
The study dataset comprises the Annual Survey of Industries (ASI) series from
1973 through 1997. The unit of observation is a factory in case of manufacturing
industries. It covers registered factories/plants, i.e., units with 10 or more workers
in plants using electric power or 20 or more workers in plants using no electric
58
power (and as registered under sections 2m(1) and 2m(11) of the Factory Act,
1948.) It consists of (i) a census of firms with over 50 employees with power (99
without power) and (ii) a sample of the remaining registered plants or factories with
10-50 employees with power (20-99 employees without power.) The reference
period for annual coverage of premises is the preceding 12 months. The cut-off
point is a registered factory working on any day of the preceding twelve months,
and in any part of which a manufacturing process is being carried or ordinarily so
carried on. The information generated from the ASI data includes number of
workers and number of employees besides the economic profile of the plants.
Though the data is collected at plant level, it is publicly available, up to 1997-98,
only at the 3-digit industry level of disaggregation.
Study Universe
ASI is classified using the national industrial classification (NIC), that changes
from NIC1970 to NIC1987. NIC1970 is applicable to years 1973 through 1988.
NIC1987 applies to years 1989 though 1997. Developing a consistent set of series
across the 25 years led to merger of several industry groups. These are as follows.
? ASI classification merges some two industry groups of NIC1970 into one group of
NIC1987, for example, NIC70-228 and NIC70-229 merge into NIC87-229. To
generate a comparable series, data of industries NIC70-228 and NIC70-229 for
years under NIC1970 classification (1973-1988) are merged and a 25-year
consistent series is obtained for such industry groups.
? ASI classification splits a NIC1970 industry group into two NIC1987 groups, for
example, NIC70-253 splits into NIC87-253 and NIC87-256. For these cases, a
59
25-year consistent series is obtained by merging the split NIC1987 industry groups
into one series corresponding to the NIC1970 group.
Some industry groups had missing observations (the entire set of 30 variables)
due to ASI data suppression policy wherein data for any industry group with two or
less number of factories is not reported in that year. Such industries (11 in number)
have been entirely dropped. Also omitted are several new groups included in NIC87
for which no earlier data are available. The truncated dataset is 84% (73%) of the
universe (ASI manufacturing) in terms of output for initial-year 1973-74 (end-year
1997-98). The figure in terms of employment (as measured by number of workers)
is 85% (78%).
Deflation to constant price data
Each industry group data is deflated using the wholesale price index (WPI)
corresponding to that industry. The WPI is the most broad-based deflator available.
The CPI is available for industrial workers but not according to industry groups.
The GDP deflator can be computed from available national accounts data at the
2-digit level of disaggregation rather than the 3-digit level possible with the WPI.
The WPI series has two bases, 1970/71 and 1982/83. The 1970/71 WPI series is
spliced to 1982/83 WPI series at the industry group level and the series so obtained
60
is used to deflate the 25-year NIC1987-classification industry data series.
Data Limitations
The dataset is rich in that it includes labor, capital, other inputs, output, value
added variables. However, it is available at present at a disaggregated industry
level instead of at unit level. Any model based on such a dataset assumes away or
ignores idiosyncratic shocks at the factory level. As Davis and Haltiwanger (1998)
and Roberts and Tybout (1996) point out, in both developed and developing
countries, idiosyncratic shocks are important. Significant variation in economic
outcomes may be driven by these shocks. Further, there is a large possibility that
the structure of adjustment costs may be lost in industry averages that can be
examined with this dataset. Second, the annual frequency may be more than the
frequency of decision-making of factory managers, i.e., the economic period may
be less than an year (quarterly or semi-annual). Nonetheless, industry level data
capture substantial variation, and any variation at an annual level is only likely to
understate patterns evident at higher frequency data. Thus, examining this dataset
should be quite instructive. Also, the capital series is a financial measure.
Appropriate economic capital stock or corresponding investment series exist at
macroeconomic level but are not yet available by disaggregated industry. The
61
model adopted hence omits capital but includes output.
3.2.2 Data Description
Summary statistics
Summary statistics of key variables across industry and year are provided in
Table 3.1. Value of output is rescaled from lakhs (hundred thousand) to thousands
to better represent the output values at the lower end of the range. In all variables,
the range and standard deviation are quite large. To the extent that estimation is in
log-linear form (as is the norm for labor demand estimation), the large range
doesn’t complicate the results.
Job flows
Employment numbers may reveal only part of the story. Following the
methodology developed by Davis and Haltiwanger (1992), computations of
manufacturing job flows during 1973/74 - 1997/98 show that job flows in India at
3-digit industry level are large (see Tables 3.2 and 3.3). The 1973-1997 average job
creation is 5.5 percent of employment (US four-digit data based manufacturing
number is 2.5 percent of employment), job destruction is 3.8 percent (US: 3.6
percent), and net employment rate is 1.7 percent (US: -1.1 percent). Further, job
reallocation rate footnote is 9.4 percent and excess job reallocation rate is 6.6
percent. Thus, 70 percent of job reallocation is excess, that is, 70 percent of the
62
turnover is over and above the rate required to accommodate the net change in
employment. Yet, somewhat paradoxically, substantial labor adjustment costs exist
in India and increased from 1976. This co-existence of high labor turnover and high
labor adjustment costs seemingly point to the limited relevance of adjustment costs
for reallocation.
A closer look, however, reveals more. Table 3.2 shows period averages for:
? pre-1976 (1973-75) or pre- labor law restrictions
? 1976-82, immediate period following labor law changes with no change in the
product markets’ regime
? 1983-90, period of gradual deregulation of product markets
? 1991-97, period post-July/August1991 liberalization package.
? 1976-97, complete period post-labor law changes
Excess job reallocation fell from 7.9 percent during 1973-75 (pre-resstrictions)
to 5 percent during 1976-82 (immediately after 1976 job security measures) but
rose to 7.8 percent during 1983-90 (gradual deregulation) falling to 6.4 percent
during 1991-97 (post liberalization program). Similarly, after separation restrictions
and induced higher severance requirements are imposed in 1975 (1975/76 - March
1976), both job creation and destruction rates fall in the 1976-82 period. Net
employment is about the same. During 1983-90 job destruction rate recovers and
surpasses the pre-restrictions level. Unlike job destruction and excess job
reallocation, job creation rate does not bounce back and remains depressed. Over
63
1991-97, job creation rate is higher at 5.8 percent but remains quite below the rate
prior to job security restrictions of 7.4 percent. Overall, in 1976-97, job destruction
rate recovers but job creation rate falls and remains lower, with net employment
rate halving to the previous period’s levels. So, there seems to be a dampening
effect of adjustment costs on employment if not on job turnover.
Part of the story may lie in product markets reform - gradual deregulation in
1980s and a liberalization package in 1990s creating more productive capacity in
response to suppressed demand. With higher demand, job creation will rise and job
destruction may fall. However, in that case, excess job reallocation whould remain
about the same since the flows are accommodating net employment creation. A rise
in excess job reallocation points to variation in both flows - creation and
destruction. So, the story in India appears to be more than the expanding demand
after liberalization. Though the evidence is very limited, it nonetheless points to a
shift in labor restructuring activity despite the continuance of a strong job security
and high adjustment cost regime.
3.3 The Model and Regression Equations
Estimations are based on the symmetric quadratic convex cost form. This is a
64
widely used specification with aggregate macroeconomic data (Hamermesh and
Pfann (1996)) and is used here with the disaggregated industry data. It is suitable
for studying the effects of adjustment costs on employment in India while
accounting for (i) the changes in the job security and mandated severance
payments’ regime in 1976, (ii) the expanded coverage of job security regime in
1982 as well as the deregulation (“broadbanding”) process begun about 1982, and
(iii) the liberalization and economic reforms’ program in 1991, and (iv) average
employment size. Estimates of the reduced form labor demand function, that is,
coefficients of wage, output, mandays, speed of adjustment, and, changes in
average employment and speed of adjustment due to regime changes and
employment size are obtained. The estimating equations are briefly derived below.
3.3.1 Symmetric Quadratic Convex Adjustment Costs
The underlying theory is based on (Sargent (1978)) the standard value
maximization by a firm given production function technology and labor-adjustment
cost function, and ignoring capital. The firm maximizes the present value of the
stream of expected future profits
65
max
?Lt?
V
t
? E
t ?
i?0
?
R
t
i
_
t?i
? E
t ?
i?0
?
R
t
i
?Q
t?i
? w
t?i
L
t?i
? AC
t?i
?
Q
t?i
? F?L
t?i
?? ?_
o
? _
o,t?i
?L
t?i
? 0.5_
1
L
t?i
2
AC
t?i
? 0.5c??L
t?i
?
2
where R ? 1 is the discount factor, F is the production function with _ parameters,
_
o,t?i
being a random parameter having a zero mean and positive variance, w is the
wage, and AC is the adjustment cost, c is adjustment cost parameter. R is assumed
to be constant. The Euler equation is
RE
t
L
t?i?1
? ?
_
1
c
? R ? 1?L
t?i
? L
t?i?1
? c
?1
?w
t?i
? _
o
? _
o,t?i
?, i ? 0, 1, ..
and the solution or decision rule for labor demand is
L
t
? _L
t?1
? _c
?1
E
t ?
i?0
?
R
i
?w
t?i
? _
o
? _
o,t?i
?
0 ? _ ? 1
Decision-makers in the firm are assumed to be risk-neutral and have rational
expectations, based on the information available at time t about the path of shocks.
Simplifying assumptions about the process generating shocks to forcing variables
are made. A first order autoregressive process is used as a reasonable
approximation to the unknown correct form for these shocks. Then, optimal
66
forecasting implies replacing expected values with their lagged values. Let the
autocorrelation parameters be _
w
and _
q
. Then the path of labor demand relating
current period employment to its lagged value and a vector of forcing variables,
wages and productivity shocks can be described as
L
t
? _L
t?1
? _c
?1
??1 ? R_
w
?
?1
w
t
? ?1 ? R?
?1
_
o
? ?1 ? R_
q
?
?1
_
o,,t
?
Sargent (1978) fits his model to data that are deviations from means and trends
(partly accounts for omitting capital). This maximization is equivalent to
minimizing the present value of the stream of expected costs (Hamermesh and
Pfann (1996)). Based on the standard simplifying assumption that deviations from
optimal profits are quadratic both in adjustment costs and in deviations of the actual
demand for labor from the optimal path, this yields
min
?Lt?
E
t ?
i?0
?
C
t?i
? E
t ?
i?0
?
R
t
i
?0.5?_L
t?i
?
? L
t?i
?
2
? 0.5c??L
t?i
?
2
?
where L
t?i
?
constitutes the optimal path in the static optimization without adjustment
costs, i.e., c ? 0. The above assumption implies that the firm is a price-taker in all
its markets. The estimating version is
67
L
t
? ?1 ? _?L
t
?
? _L
it?1
0 ? _ ? 1
_ is a non-linear function of R, c, and _. Given the assumptions, linearity of this
equation allows for aggregation even if firms face different shocks. Then,
adjustment costs are implied by _, the closer it is to one, the higher are the
adjustment costs and lower is the speed of adjustment.
3.3.2 The Regression Equations
In this paper, recall that output is included though capital and investment
demand are ignored (since appropriate series are not yet available and relying on
evidence indicating that the estimates of employment dynamics are not biased if
adjustment of capital is ignored). Thus, the forcing variables included in L
t
?
are the
real wage and output. Further, since the model uses employment and not
labor-hours the variable mandays per worker is also included as a regressor. In
India, manday is a fixed number of hours and labor-use varies with number of
mandays and number of workers (employment). It is assumed that the first
dimension of labor adjustment is number of mandays and second that of number of
workers. Hence, employment and labor (workers) demand does not affect lagged
mandays. The symmetric quadratic convex cost model is readily amenable to
68
aggregation across units as discussed above. Thus, an industry labor demand
function is the basis of the log-linear regression equation. Then, with first-order
autoregressive processes generating shocks to forcing variables in L
it
?
and their
expected values replaced by their lagged values, the estimating version for industry
i ?i ? 1, ..., N; N ? 136?in period t ?t ? 1, ..., T; T ? 25?is
lnL
it
? ?1 ? _?lnL
it
?
? _lnL
it?1
? u
it
, i ? 1, ..., N; t ? 1, ..., T
lnL
it
? ?1 ? _??_
0
? ?_lnw
?1
? _lnQ
?1
? _lnH
?1
?? ? _lnL
it?1
? u
it
lnL
it
? ?1 ? _??_
0
? ?_
?
lnL
it
?
?? ? _lnL
it?1
? u
it
0 ? _ ? 1
u
it
? _
l
u
it?1
? ?
it
?
it
~ i.i.d. ?0, _
?
2
?
Rational expectations based on autoregressive processes imply bona-fide use of the
lagged values of forcing variables and hence serially uncorrelated disturbance in the
labor demand equation. However, the error u
it
here allows for some serial
correlation with i.i.d. errors. L is demand for workers, Q is output, w is wage, H is
hours, _ is the measure of adjustment costs. Note that the autoregressive parameters
of the forcing variables are embedded within their coefficients.
The estimation procedure is a difference-in-difference approach based on size
(average employment size) and period dummies. Also, as a measure of evolving
competition (protected and licensed production yielding to increasing deregulation
69
from 1982 and substantial delicensing in 1991), price-cost margins are generated
and interacted with lagged employment and size dummies.
The primary differences are due to size dummies. The secondary differences in
these size-differences are due to the regime changes captured by the period
dummies. To the above equation are added these size dummies. Average
employment size is the number of workers in an industry divided by the number of
factories in that industry. This is used to generate two size dummies (i) for size 100
and above, S1 takes the value one and above, and (ii) for size 300 and above, S3
takes a value one and above. S3 captures industries that come within the ambit of
the 1976 job security restrictions (that is, all units with employment 300 workers
and above). S1 represents industries that are further included in job security
restrcitions in 1982 (that is, all units with employment 100 workers and above). The
size dummies are cumulative, hence their estimated coefficients indicate marginal
effects. Being based on average employment size rather than actual employment
size may make these size dummies more restrictive than the usual classification of
large, medium, and small units of interest with unit-level panel data. _
S1
and _
S3
are
the increments in effect of adjustment costs due to size differences. Then, the
estimating equation is
70
lnL
it
? ?1 ? _??_
0
? ?_
?
lnL
it
?
?? ? _lnL
it?1
lnL
it
? ?1 ? _
0
? _
S1
S1 ? _
S3
S3??
?_
0
? _
S1
S1 ? _
S3
S3 ? ?_
?
lnL
it
?
??
? _
0
lnL
it?1
? _
S1
S1lnL
it?1
? _
S3
S3lnL
it?1
S1 ? 1 if size ? 100, zero else
S3 ? 1 if size ? 300, zero else
The coefficient for wage is expected to be negative and that of output to be
positive. The coefficient for mandays may be positive or negative depending upon
whether workers and mandays are complements or substitutes. _
0
is expected to be
positive, that is, some adjustment costs exist prior to the job security restrictions
and that speed of adjustment is slow. _
S1
and _
S2
would be positive if larger
industries find it harder to adjust employment. Similarly, average employment
coefficient _
0
is expected to be positive and, _
S1
and _
S2
also positive as larger
firms are expected to have higher average employment.
There are three period dummies, (i) P1 is the period dummy that takes value
one for all periods from 1976 onward representing job security regime from 1976;
(ii) P2 is the period dummy that takes value one for all periods from 1983 onward
reflecting the expanded coverage of job security regulations in 1982 (in the tables,
it is called Year82 dummy to refer to 1982 regime changes as well as the
deregulation or “broadbanding” process begun about 1982), and (iii) P3 is the
71
period dummy that takes value one for all periods from 1991 onward reflecting the
economic reforms and liberalization in 1991. Again, the period dummies are
cumulative, hence their estimated coefficients indicate marginal effects. These
period dummies are interacted with the size dummies. Below is a version of the
estimating equation presented with size dummies and only one period dummy (to
avoid notational clutter).
lnL
it
? ?1 ? _
0
? _
S1
S1 ? _
S3
S3 ? _
P1
P1 ? _
S1.P1
S1.P1 ? _
S3.P1
S3.P1??
?_
0
? _
S1
S1 ? _
S3
S3 ? _
P1
P1 ? _
S1.P1
S1.P1 ? _
S3.P1
S3.P1 ? ?_
?
lnL
it
?
??
? _
0
lnL
it?1
? _
S1
S1lnL
it?1
? _
S3
S3lnL
it?1
? _
P1
P1lnL
it?1
? _
S1.P1
S1.P1lnL
it?1
? _
S3.P1
S3.P1lnL
it?1
S1 ? 1 if size ? 100, zero else
S3 ? 1 if size ? 300, zero else
P1 ? 1 if year ? 1976, zero else
Post-1976: Due to job security restrictions and consequent higher severance
payments, adjustment costs in general will rise and an increase in the adjustment
cost parameter _ or a positive marginal effect _
P1
is expected. Further, _
S3.P1
(size
300 and post-1976 restrictions) should be positive since the restrictions apply
particularly to size 300 and above. The intercept terms or average employment
parameters _
P1
and _
S3.P1
may be negative indicating a one-time downward shift of
the labor demand function for all industries and specially for those with size 300
and above. _
S1.P1
(size 100 and post-1976 restrictions) may be positive or
72
insignificant and _
S1.P1
may be negative or insignificant. So, the marginal effects
for industries with size 100 and above may be negligible.
Post-1982: _
S1.P2
(size 100 and post-1982) is expected to be positive on account
of job security coverage expanding to units with employment 100 and above but
negative on account of deregulation. _
S1.P2
is expected to be negative due to
expanded job security coverage and positive otherwise. The net effect may be
positive or negative. Similarly, other coefficients may be negative or positive.
Post-1991: If the impact of liberalization in product markets filtered to labor
markets, a reduction in adjustment costs, i.e., negative _
P3
, _
S1.P3
, and _
S3.P3
are
expected, even though there is no change in the job security regime. _
P3
, _
S1.P3
, and
_
S3.P3
may similarly be positive.
Since gradual deregulation in product markets (protected and licensed
production yielding to increasing deregulation from 1982 and substantial
delicensing in 1991) continued over several years, a competition measure every
period to capture the evolutionary change is more appropriate. Herfindahl indexes
by industry in India could not be found. Thus, as an inverse measure of gradually
evolving competition, price-cost margins (PCM, G
it?1
) are also generated and
interacted with lagged employment and size dummies. A truncated version of the
estimating equation without period dummies is shown below.
73
lnL
it
? ?1 ? _
0
? _
S1
S1 ? _
S3
S3 ? _
g
lnG
it?1
? _
S1.g
S1lnG
it?1
? _
S3.g
S3lnG
it?1
??
?_
0
? _
S1
S1 ? _
S3
S3 ? _
g
lnG
it?1
? _
S1.g
S1lnG
it?1
? _
S3.g
S3lnG
it?1
? ?_
?
lnL
it
?
??
? _
0
lnL
it?1
? _
S1
S1lnL
it?1
? _
S3
S3lnL
it?1
? _
g
lnG
it?1
lnL
it?1
? _
S1.g
S1 ? G
it?1
lnL
it?1
? _
S3.g
S3 ? G
it?1
lnL
it?1
S1 ? 1 if size ? 100, zero else
S3 ? 1 if size ? 300, zero else
The premise here is that a higher degree of competition makes the need to adjust
more urgent and may thus blunt the impact of adjustment costs and increase the
speed of adjustment. So, the change in _ due to competition (PCM) is negative
(positive) or that _
g
is expected to be positive. The PCM coefficient, _
g
, will be
positive or negative depending upon whether employment increases or decreases
with greater price-cost margin. Higher PCM being associated with increasingly
imperfect competition and excess capacity, it is expected to be negative. The
interactions of PCM and size dummies would reflect these effects coupled with
labor hoarding effects.
3.3.3 GLS and System-GMM Estimators
The estimates are obtained using GLS with errors robust to heteroskedasticity
and autocorrelation. In the presence of any disturbance autocorrelation, the lagged
variables (lagged dependent variable, lagged output; usually wage also, not in India
application, see chapter two) are endogenous and need to be instrumented. In the
74
absence of external instruments, lagged first differences can serve as instruments
for levels equation. However, with substantial sluggishness in employment, the
correlation of these instruments with regressors is weak. Shea R-squared (Shea
(1997)) is about 0.01 only indicating poor instrument relevance.
A more comprehensive set of instruments is obtained by using the system GMM
estimator. Efforts to account for group effects in dynamic panel data estimations
with autoregressive panel series produced the first-differenced GMM estimator.
The strategy is first-differencing the equation and using lagged levels as
instruments for the pre-determined and endogenous variables in first differences.
Though sound, the technique performed poorly in that for panels where the series
are highly autoregressive, this estimator is found to have large finite sample bias
and poor precision. An alternative is found in the system-GMM estimator (Blundell
and Bond (1998)), where equations in first differences use lagged levels as
instruments as before and additionally levels equations are instrumented using
lagged first differences. The system GMM estimator adds moment conditions based
on initial conditions restrictions (jointly stationary means of the regressand and the
regressors, to ensure that the respective first differences are uncorrelated with group
effects). When the additional moment conditions are valid, the system GMM
75
estimator is found to reduce bias and improve precision (Blundell, Bond and
Windmeijer 2000). System-GMM allows for some serial correlation in the residual
disturbance, with the lags specified for instruments increasing by the order of serial
correlation. The number of moment conditions reduces correspondingly.
To exploit the rich instrumentation possible with the system-GMM technique,
these GMM estimates are obtained. However, owing to the use of average
employment size dummies and period dummies, only a parsimonious specification
is feasible. The interactions of size and period dummies with the forcing variables
are excluded. This parsimonious specification is estimated with both GLS and
GMM to enable appropriate comparisons. In so doing, the first three years’
observations are lost. These are 1973-1975, the years prior to the job security
restrictions’ regime change in 1976. The impact of these 1976 restrictions alone is
thus not directly available with the GMM instruments.
3.4 Results and Findings
The estimates of model I (symmetric quadratic convex adjustment costs) based
equation are presented in Table 3.4. As noted in the data description above, an year,
say 1973, in this dataset refers to the period April73-March74. The equation is
76
estimated for three periods - 1973-1975, the period before the job security
restrictions become mandatory; 1973-1982, the period after restrictions but before
the job security provisions’ coverage is extended and gradual deregulation become
operational; and, 1973-1997, the full period for which data are available.
Size dummies are called Size100 and Size300. These dummies are interacted
with wages, mandays, output, and lagged workers to assess the impact of regime
changes on the relevant regression coefficients. Thus, there are three panels in the
base regression for 1973-75: (i) effects for all employment sizes, (ii) marginal
effects for size 100 and above, and (iii) further marginal effects for 300 and above.
Period dummies are called Year76, Year82, and Year91. All these dummies are
interacted with the entire first set of three panels including wages, mandays, output,
lagged workers, and their interactions with size dummies. Year dummies from 1979
onward when a regime (new government) change and before the gradual
deregulation process is operational and industry dummies at 2-digit level are also
included.
The wage coefficient is statistically significant and negative as expected. The
estimate for output is also significant and positive. Mandays (proxy variable for
77
hours) coefficient is significant and positive. The constant term representing
average or autonomous employment is significant and positive. The coefficient for
lagged workers that captures the effect of adjustment costs (extent of sluggishness
in adjustment and opposite of the speed of adjustment) is significant and positive.
Note that, this coefficient is high and larger in absolute terms than that of wage,
mandays, or output. The evidence for 1973-1975 indicates the existence of
adjustment costs even before the job security restrictions came into effect. Size
effects are significant. Average employment is substantially higher for size300
industries (but negative for size100 industries). The lagged employment coefficient
for size300 industries is negative implying that larger size in India may be
associated with lower adjustment costs or higher speed of adjustment.
Most industry and year dummies are insignificant. The results for wage,
mandays, output, and lagged workers remain unchanged in the different model
specifications. Following is the discussion on size and period dummies and their
interactions. It focuses on average employment and the speed of adjustment.
3.4.1 Changes in the speed of adjustment - GLS
The impact of the labor restrictions through the period dummy “Year76” is very
significant for average employment and quite high (the intercept or the coefficient
78
for Year76). Recall that in the log-linear labor demand equation, this coefficient is
log of average employment. Calculations indicate a fall in average employment of
about 28 percent (reduces to about 25 percent when the full sample period is taken).
The speed of adjustment is the opposite of the coefficient on lagged employment.
The coefficient of lagged workers interacted with Year76 is a significant 0.032 or 3
percent (significance declines in the full sample period). Further, the size300 and
lagged workers interaction is very significant. It is 0.377 indicating a large 38
percent (36 percent in full sample period) increase in sluggishness or a
corresponding decrease in the speed of adjustment over and above the 3 percent
found for all sizes. Post job security restrictions in 1976, (i) there is a one-time
fall (about 28 percent) in labor demand and (ii) a fall in the speed of
adjustment (3 percent) and additionally so (38 percent) for large industries.
Fallon and Lucas (1991) had found a significant negative coefficient on the period
dummy (here, Year76) but no significant change in the speed of adjustment. Note
that the interactions of Year76 with wages, mandays, and output are all highly
significant and opposite to the signs without interaction. This is consistent with the
theoretical model - when _ increases, the coefficients magnitude decreases - and
lends considerable support to a significant increase in _ and hence fall in ?1 ? _?or
speed of adjustment.
79
Using the full sample period 1973-1997 and all dummies, results are somewhat
unexpected for Year82. The coefficient of lagged employment, size300 and Year82
dummy interaction is significant, about 0.15. This increase in sluggishness is
expected for size100 industries (insignificant here) since job security regime
coverage extended to units with employment 100 and above (from 300 and above in
1976). Nonetheless, this does indicate increasing adjustment costs for larger
industries with increasing coverage. Further, there is a small decrease in
sluggishness (-0.026) in lagged employment and Year82 interaction; perhaps the
impact of deregulation on industries with size below 100 or those not specifically
covered by job security regime. Unlike the case of Year76 and Year82 period
dummies, the only significant result for Year91 is the large increase in average
employment for size300 or large industries. Until 1997, there is no significant effect
of lagged employment-Year91 interactions. So, there is some evidence of increase
in employment but no corresponding decrease in sluggishness.
The coefficients of price-cost margin and its interaction with lagged workers are
insignificant. However, the size dummy interactions are very significant. For
size100, the results are as expected - PCM decreases average employment and
80
increases the sluggishness in adjustment, i.e., a positive coefficient of lagged
employment-PCM-size100. Paradoxically, the results are opposite for size300 -
PCM-size300 dummy has a positive coefficient (increases employment) and
PCM-size300-lagged employment has a negative coefficient or there is a decrease
in sluggishness.
Competition thus has differing impacts. For medium sized (size100) firms, it
increases average employment and decreases sluggishness with opposing
results for larger (siz300) firms (lower average employment and higher
sluggishness). Effects of pre-existing labor hoarding (requiring shedding of
excess labor lowering average employment and net employment change) may
be responsible.
3.4.2 Changes in the speed of adjustment - System GMM Results
The results for system-GMM estimations are shown in table 3.6 and the
corresponding GLS results in table 3.5. Before presenting the estimates, note that
the over-identifying conditions fail to be rejected using the Hansen test or are tested
as valid. Though the absence of autocorrelation of the first order in first differenced
residuals is rejected (as expected, if the level equations have i.i.d. errors), no-AR(2)
fails to be rejected or that no serial correlation in the levels’ equation disturbance is
valid. The regressions are thus well specified. Second, both tables 3.5 and 3.6 use
81
the parsimonious specification. Even so, all relevant variables cannot be
instrumented given sample size. With restricted lags for the instruments, along with
base regressors there are two distinct sets of variables that are instrumented. One set
includes additionally Year82 period dummy with its interactions. The second set
includes additionally Year91 dummy and its interactions.
Recall that there are no observations for the pre-1976 period. Results pertain to
1976-1982 and 1976-1997 periods. Hence, a direct before-after comparison is not
possible. For both periods and with either set of instruments for the 1976-1997
period, GMM-lagged employment coefficient is very significant and high, the
magnitude being slightly smaller than corresponding significant GLS estimates.
GMM-size effects are obtained for size100 industries whereas in corresponding
GLS these are obtained for size300 industries. Both indicate higher average
employment and lower sluggishness for larger industries. The magnitude of GMM
estimate is many times over those of GLS.
Additional results are obtained for the full sample with Year82 and Year91
period dummies and the second set of instruments. GMM-estimates for
size100-Year82 show a large fall in average employment and for
82
size100-Year82-lagged employment show a large increase in sluggishness. This
is expected since the coverage of job security restrictions expanded to size100
industries in 1982. However, this result is not obtained using the first set of
instruments, the estimates are insignificant. With corresponding GLS, these results
are obtained for size300 industries as also in the full specification GLS and in both,
the magnitudes are smaller. Further, there is a decrease in sluggishness or the
coefficient of lagged employment-Year82 interaction is negative for small firms.
This is not obtained with parsimonious GLS but is found in full GLS model.
Year91-lagged employment GMM estimates are significant but opposing with
the two sets of instruments. No significant results for Year91 and its interactions are
obtained in corresponding GLS. Similarly, no significant results for PCM and its
interaction with lagged employment are obtained in either GMM or corresponding
GLS. Note that parsimony led to exclusion of PCM-size dummies’ interactions
which produced results in the full GLS model.
3.5 Concluding remarks
Overall, there is evidence that after job security restrictions in 1976, (i) there is
a one-time fall (about 28 percent) in labor demand and (ii) a fall in the speed of
adjustment (3 percent) and additionally so (38 percent) for large industries. Weaker
83
evidence indicates increasing adjustment costs for larger industries with increasing
coverage and lower adjustment costs for smaller ones excluded from coverage.
GMM-estimates post-Year82 for size100 industries show a large fall in average
employment and a large increase in sluggishness; and a decrease in sluggishness for
smaller firms. This is expected since the coverage of job security restrictions
expanded to size100 industries in 1982. The liberalization package in 1991
generates a one-time boost to labor demand for large firms but no clear reversal in
sluggishness of labor adjustment.
For medium sized (size100) firms, competition increases average employment
and decreases sluggishness with opposing results for larger (size300) firms (lower
average employment and higher sluggishness). Effects of pre-existing labor
hoarding (requiring shedding of excess labor lowering average employment and net
employment change) may be responsible.
Industry data used with GLS and system-GMM estimation techniques provides
interesting results. Using the system-GMM estimators helps to instrument for
endogeneity in the labor demand equation. GMM estimates strengthen the GLS
results and provide additional results.
84
1
We take a broad view of public sector employment and associated retrenchment.
Public sector employment includes civil service workers, military personnel and
employees of public sector enterprises.
2
Important exceptions are Svejnar and Terrell (1991) which provides a very useful
comparison across six countries for retrenchment in the transport sector, and Lindauer
and Nunberg (1994) which provides a book-length discussion of civil service reform.
Our study benefitted substantially from the insights in both of these publications.
Chapter 4: Public Sector Retrenchment Programs
4.1. Introduction
The retrenchment of public sector employment is an increasingly pervasive
phenomenon.
1
Many advanced countries, developing economies, and transition
economies have recently faced the issue of downsizing of public sector employment.
The reasons underlying the downsizing vary considerably across countries. For some it
is a general move towards a more market economy, for others it is a reduction in the role
of the military or an attempt to reduce a bloated bureaucracy, for others the retrenchment
is sparked by a fiscal crisis necessitating a severe cutback in government spending, and
finally for some it is a combination of some or all of these. Given the pervasiveness of
the phenomenon, there is no shortage of studies of individual countries or episodes.
While these studies are quite useful, it is difficult to draw common lessons from these
country specific studies and there is relatively little analysis comparing the experiences
across countries.
2
In this paper, we compile and analyze information on the recent
retrenchment experiences across a large number of countries and retrenchment episodes.
Comparing and contrasting the experiences across countries is a formidable task.
85
3
Details regarding the collection of information is provided in section 3. A complete
listing of individual country summaries is available upon request.
4
The converse is not true i.e. the absence of rehiring is not synonymous with success.
Also, conceivably a restructuring could involve separating employees en masse and
subsequently rehiring those with necessary specific skills or rehiring at temporary terms
or both. No program in our sample intended such purposive rehiring.
Beyond the usual problems of cross country comparisons, there are no central databases
with the requisite information compiled that permits such analysis. Accordingly, we
compiled and collected our own information from a myriad of sources but primarily from
internal World Bank documents and interviews of staff members at the World Bank with
operational information about retrenchment experiences in individual countries.
3
Our
analysis is based upon 37 countries and 41 programs. Most of the programs for which we
could collect detailed information commenced in the early 1990s and many of the
programs are ongoing (which makes analysis of the ultimate success of the programs
difficult). The gross number of workers separated in all the programs we consider
exceeds 5 million, with a somewhat smaller net employment reduction. The discrepancy
reflects the fact that some programs exhibited significant rehiring of the separated
workers who departed and/or new hires by the public sector. In the analysis that follows,
we look closely at the characteristics of programs that involve significant rehiring and
new hires. While the latter may be part of a coherent plan to restructure the public sector
workforce, the former is clearly not an indicator of success.
4
Evaluating the costs and the benefits of the individual programs in a more
comprehensive fashion is inherently difficult. In principle, the requisite information
includes the compensation packages offered, the relative productivities and wages of the
86
displaced workers in the public and private sectors, as well as the adjustment costs borne
directly by the affected workers and the entire economy. Measurement of much of this is
well beyond the scope of available information, particularly the nature of the adjustment
costs. Nevertheless, we can document considerable details about the magnitude and
forms of compensation used in retrenchment programs, the wage bill savings, and
characterize some of the other aspects that are relevant for evaluation. To summarize the
available information on costs and benefits, we calculate a simple financial break-even
indicator measured as the number of years required for the present value of financial
costs to equal the present value of financial gains. We also examine refinements of this
indicator that attempt to incorporate the productivity gains generated from the
retrenchment. In turn, we relate these summary financial indicators to other program
characteristics.
The chapter proceeds as follows. Section 2 provides a brief presentation of the
underlying conceptual framework for characterizing the private and social costs and
benefits of a public sector retrenchment program. The analysis in this section is
deliberately simple and borrows heavily from the existing literature. Our objective here
is primarily to provide guidance for the type of information that is required to compare
and analyze alternative programs and to discuss the interaction between the conceptual
and measurement issues that must be confronted. Section 3 outlines the methodology
used to collect information for the individual countries and to compile the information in
a systematic fashion. Section 4 presents the analysis of the information collected as well
as some detailed discussion of individual countries to illustrate the patterns that emerge.
Section 5 contains concluding remarks.
87
5
For example, many of the conceptual issues discussed in this section are discussed more
formally and with greater attention to the range of relevant issues in Diwan (1993a,
1993b) and Rama (1997).
6
While we do not characterize all relevant sources of heterogeneity explicitly, differences
across workers in the following present values may reflect differences in ability,
experience, skills, horizons, discount rates and mobility costs.
4.2. Conceptual Framework and Measurement Issues
To provide some guidance for the type of factors relevant for comparing and
analyzing retrenchment programs across countries, it is useful to sketch out a simple
conceptual framework. This framework helps us organize our analysis and discussion of
the information we have collected. While the discussion in this section borrows heavily
from the existing literature,
5
our characterization of the relevant issues emphasizes some
points that are neglected in the literature. Further, since our ultimate objective is to
quantify the relevant measures using data across countries, our focus in this discussion is
on the interaction between the conceptual and measurement issues that must be
confronted.
Consider the relevant incentives (private and social) for the retrenchment decision
for the marginal worker. Let:
6
P
private
= Present discounted value of the worker's productivity in private sector
P
public
= Present discounted value of the worker's productivity in public sector
C
individual
= Present discounted value of adjustment costs born by the individual in
relocating from public to private sector (e.g., job search costs, relocation costs, time spent
unemployed).
88
7
The discussion here and the subsequent analysis primarily focuses on the reallocation of
workers from the public to the private sector taking into account the relevant
productivities, wages and costs of labor market adjustment. Public sector downsizing
also likely involves the reallocation of capital. Even though our focus is on the
reallocation of workers, the implications of capital reallocation and the interaction with
the worker reallocation deserves further attention.
8
Explicit modeling of the distinction between the present discounted value of earnings
and adjustment costs would require incorporating the fact that transitions from a public
sector job to a private sector job may involve several transitions and accordingly several
spells of unemployment. As emphasized by Hall (1995), a potentially important
explanation of persistence in unemployment rate dynamics is that separations tend to
beget further separations.
C
social
= Present discounted value of adjustment costs borne by society for the individual
to relocate from public to private sector (i.e., adjustment costs borne by individual plus
spillover effects e.g., congestion effects).
W
public
= Present discounted value of earnings of the worker in public sector.
W
private
= Present discounted value of earnings of the worker in private sector.
The principle of social optimality considered here is that all resources should be
allocated to their highest valued use (net of relocation costs given the existing allocation
of resources). Thus, it is socially optimal to relocate the marginal worker to the private
sector if:
7
P
private
- C
social
> P
public
(1)
An individual worker will choose to stay on a public sector job as long as:
8
W
public
> W
private
- C
individual
(2)
If workers are paid more than their marginal product in the public sector (i.e., W
public
>
P
public
), then it is easy to imagine that both (1) and (2) hold simultaneously -- that is, it is
89
9
It may be that individual workers and the government discount the future differently
because of differential access to capital markets. Such differential access to capital
markets is, in principle, consistent with the specification and associated discussion
considered here but explicit consideration of the role of capital markets deserves more
attention in this context. For example, unemployment benefits and other forms of worker
safety net assistance are often justified as a form of social insurance in the face of
imperfections in the capital market. Since worker safety net enhancements are often part
of a retrenchment program, explicit consideration of the role of differential capital market
access is quite relevant.
in society’s interest for the marginal worker to relocate but it is not privately optimal.
9
A number of additional factors are potentially important in evaluating the
optimality of a retrenchment program. One factor is the expenditures incurred to pay
public workers wages or alternatively the compensation packages offered to induce
workers to relocate voluntarily. If taxes are not distortionary and there is no distortionary
rent seeking behavior, these expenditures should be viewed as transfers and thus not
impact efficiency. Under these strong assumptions, these terms should not appear in
equation (1). However, in the presence of distortionary taxes (including the inflation tax)
and distortions from rent seeking behavior, such transfers yield efficiency losses. In
addition, an excessive wage bill and/or fiscal crisis may imply that other government
services are adversely affected by retaining redundant workers. To incorporate these
effects in a modified version of (1), suppose that there is a distortions-based loss function
L(.) which is an increasing function of the transfers to workers. Taking into account
these distortionary losses in (1) would yield that retrenchment is optimal if:
P
private
- C
social
- L(COMP) > P
public
- L(W
public
) (3)
where COMP reflects the present discounted value of the compensation or other
90
10
A related issue is whether the nature of the distortions from the transfers depends not
only the net present value of the transfers but also the timing. For example, rents in the
form of artificially high public sector wages may have dynamic distortionary effects
beyond those associated with a one-time transfer.
11
For further discussion of the congestion externalities among job seekers that might be
important see, e.g., Blanchard and Diamond (1989,1990). For further discussion of the
idea that employment and earnings losses by workers in one sector may generate
spillover effects to other sectors through final good demand spillover effects, see, e.g.,
Cooper and Haltiwanger (1996).
assistance programs that accompany the retrenchment.
10
One problem faced by policymakers (and shared in our analysis) is measuring the
components above in (1), (2) and (3). Many of the components are very difficult to
measure and/or difficult to observe in the presence of worker heterogeneity (e.g., outside
opportunities and individual productivity). Among the most difficult to assess are the
adjustment costs. Factors influencing adjustment costs include the degree of labor
market flexibility (i.e., barriers to adjustment in terms of hiring and firing costs), the
worker safety net, informal vs. formal sector development, and the method, scope and
speed of retrenchment. Social adjustment costs will be greater than the private
adjustment costs to the extent there are spillover or externality effects of worker
displacement. The spillover/externality effects include the impact of final goods demand
spillover effects from reduced consumption expenditures by affected workers, congestion
externalities among job searchers, and social disruption (e.g., national strikes) induced
by the affected workers.
11
The method, scope and speed of retrenchment potentially
influence these spillover effects. In addition, the concentration of the retrenchment in a
local community may imply important local spillover effects even if there are not
economy-wide spillover effects.
91
12
The nature of the measurement problems in evaluating pre and post retrenchment
productivity will vary depending on the type of public sector employment (i.e., civil
service, military, or public sector enterprises producing goods and services).
As noted, the nature and magnitude of these adjustment costs are very difficult to
measure and as will become clear we have relatively little quantifiable information on
adjustment costs that we can use in our comparison across countries. The most direct
evidence available are from recent studies of privatization in transition economies (see,
e.g., the studies in Commander and Coricelli (1995)) for which the massive restructuring
implies that adjustment costs take center stage. Even in these latter studies, the
information on the nature of the adjustment process is often quite limited. In general, and
particularly in these transition economies, characterizing these adjustment costs is at the
heart of the debate regarding gradualism versus a big push in the efforts to reduce the role
of the public sector in the economy.
Another difficult measurement/conceptual problem is evaluating the pre and post
retrenchment productivity of workers in the public sector. While there may be
widespread agreement that there are redundant public sector workers, quantifying this is
extremely difficult.
12
Further, some of the need for retrenchment may reflect more of a
need of restructuring than simply downsizing. For example, low morale and productivity
of public sector employees may reflect poor organizational structure and/or wage
compression. Thus, a program may involve organizational structure and wage structure
changes that appear to be financially expensive but are well-motivated by such factors.
This raises concerns about the use of simple financial indicators that do not adequately
incorporate such factors to evaluate the need and success of a program. In terms of (3),
this discussion implies that the public sector productivity term needs to be broadly
92
interpreted. That is, retrenching the marginal worker will yield the direct loss of the
worker’s output (if any) but may also yield productivity gains from the reorganization
that accompanies downsizing.
Beyond the above problems of measurement and assessment, it is often politically
necessary to implement the retrenchment scheme via a voluntary program. This
necessitates the provision of some incentive payment in form of severance pay, pensions,
etc. (denoted INCENTIVE below) such that for an individual worker:
INCENTIVE > W
public
- (W
private
- C
individual
) (4)
In principle, with complete information about worker heterogeneity, it is possible
to design an optimal incentive scheme for each worker (see Diwan 1993a, 1993b and
Rama (1997) for a formal analysis). One well-recognized problem that immediately
emerges is that if a simple, common incentive package is offered to all workers,
heterogeneity across workers implies that only those with lowest rent will depart. It is
important to emphasize that some aspects of the selection process may be unfavorable,
while others may be advantageous. For example, in environments with wage
compression, a common package offered to a wide class of workers implies that the most
skilled and capable workers will leave. Alternatively, individuals with better outside
opportunities and/or low adjustment costs are more likely to leave -- other things equal,
this voluntary aspect of selection will be favorable.
The implication of this discussion is that an optimal incentive scheme must be
individually tailored to reflect worker heterogeneity in rents and adjustment costs but that
this is difficult to implement in practice in the face of imperfect information. This
tension is at the center of the debate about the pros and cons of voluntary and involuntary
93
13
For a discussion on optimal schemes using either different types of severance contracts
or alternatively auctions, see Levy and Mclean (1996).
retrenchment programs. As noted above, voluntary programs have the advantage that
they yield some favorable voluntary selection of workers with good outside opportunities
and/or low adjustment costs without requiring the policy maker to have complete
information. However, workers with good outside opportunities are likely the most
productive public sector workers so that the voluntary selection may be adverse. The
problem with adverse selection is apt to be especially severe in environments with wage
compression
13
. Involuntary programs potentially permit specific targeting of groups of
workers on observable characteristics but don’t take advantage of favorable voluntary
selection on unobservable worker characteristics. Further, involuntary schemes may
also yield high adjustment costs. For example, mass involuntary layoffs may involve
substantial private and social adjustment costs and in this case the nature of the safety net
takes on particular importance. Finally, the political will (ability) to undertake and
sustain schemes with an involuntary component may be lacking.
Putting these pieces together suggests that our comparison and analysis of
alternative programs should consider the following factors. First, we must consider the
factors leading to retrenchment (fiscal crisis, overstaffing, morale problems given wage
compression, etc.) since they may provide insights about the relationship between wages
and productivity in the private and public sectors. Second, several factors influence the
adjustment costs including the scope and speed of retrenchment, the mechanism used
(involuntary/voluntary), whether the scheme involves targeting (e.g., skill or age biased),
and the nature of labor market flexibility and the safety net. Many of these same factors
94
14
Type of the World Bank documents included: Staff Appraisal Report (SAR),
Memorandum and President's Recommendation (MOP), President's Report (PR),
Supervision memorandum, Project Completion Report (PCR), Implementation
Completion Report (ICR), Project Performance Audit Report (PPAR), Operations
Evaluation Studis, Country Economic Memorandum (CEM), Sector and other economic
reports.
are important (e.g., the method used and the nature of targeting) in terms of the need for
individual tailoring of plans given worker heterogeneity. Finally, the magnitudes of the
financial costs (e.g. W
public
and INCENTIVE) are relevant to characterize the magnitude
of the transfers.
4.3 Data Collection
The survey of public sector retrenchment programs across developing and
transition countries was carried out on two dimensions. The first involved collection of
internal World Bank documents relating to the macro-economic and public sector
adjustment in these countries. A preliminary portrait of the issues relating to each
country was drawn using myriad types of World Bank Documents
14
. These were, in
turn, supplemented by a variety of external sources. The second step was interviewing
World Bank officials associated with the retrenchment programs to obtain more direct
information and assessment. The interviews typically yielded further acquisition of
relevant documents. The countries, the retrenchment programs and the associated
officials were selected using the World Bank electronic management information system
(MIS) and the adjustment lending database called ALCID. See appendix Table 4A.1 for
a list of surveyed programs/countries. The projects and loans listed on the World Bank
95
15
The selection of programs was mostly restricted to this list. In general, World Bank
staff mobility with movements away from early programs reduced usefulness of such
interviews.
16
The detailed country summaries total more than 200 pages including extensive
references to documents, papers, and other sources beyond those cited in the paper.
MIS go back to mid-eighties.
15
The interviews were conducted for more than eight
weeks during the summer of 1995.
The interview transcripts, documents, reports and articles relating to each
program were synthesized and summarized using a uniform outline. Information
presented relates to four broad aspects - characterizing retrenchment, the nature of labor
turnover and institutions, cost-benefit analyses in financial and economic terms, and
monitoring and evaluation of the program. A complete listing of the individual country
summaries are available upon request.
16
As will become apparent, the information collected and assimilated ranges from
quantitative information about the scope, speed, and financial costs and benefits to
qualitative information about the factors precipitating the retrenchment as well as
characteristics of the program such as the methods used and the productivity gains
observed in the public sector. The quality and completeness of the information collected
varies substantially across programs. This blend of quantitative and qualitative
information yields that the analysis that follows is primarily descriptive -- an attempt at
summarizing the basic facts and drawing out observable patterns. Not surprisingly, a
host of institutional and idiosyncratic factors appear to be important as one learns the
details of each individual program. These institutional and idiosyncratic factors serve as
an additional important caution for interpreting the analysis of basic patterns that follows
96
17
As of February 1997, the World Bank allowed lending for severance payments provided
it entails no organization closure.
in sections 4.4.1-4.4.5.
4.4 Survey and Analysis of Cross country Evidence
4.4.1 Basic Facts
We begin the analysis of the information we have collected by characterizing
some basic facts about the programs for which we gathered information. Tables 4.1-4.2
present summary statistics about the programs we surveyed (summary information on
each program is provided in Table 4A.1 in the appendix). As noted in the introduction,
our analysis is based on 41 programs in 37 countries. The World Bank has not
historically provided direct assistance for public sector employment retrenchment
programs
17
. However, under the umbrella of a comprehensive assistance package, an
agreement with a country often stipulates that the country use domestic counterpart funds
for specific purposes. Under such umbrella agreements, public sector retrenchment may
be part of the overall package. In our survey, about 65 percent of the programs had this
“umbrella” connection to World Bank financing.
Table 4.1 makes clear that total separations are greater than the total employment
reduction reflecting rehiring and new hires. Most programs did not experience rehiring
of the same workers that had departed but a nontrivial fraction (20 percent) experienced
significant rehiring. New hires are also relatively rare ( about 13 percent of programs).
Some programs represented only good intentions with no workers actually separated.
One of the programs included in our survey is Hungary with a massive public sector
97
18
The total cost is driven in part by enormous increases in pension and safety net
expenditures in Poland. The increases for Poland reported in Table 4A.1 are, from all
accounts, real but they do skew totals and averages somewhat. However, much of the
subsequent analysis is not sensitive to outliers on this dimension. That is, much of the
analysis is based on the percent of programs that exhibit various characteristics.
retrenchment program. This program is a clear outlier on a number of dimensions
(discussed below) but we felt it was important to include large programs among the
transition economies as points of contrast. The method of employment reduction is
categorized into three basic categories: involuntary (hard) refers to layoffs, involuntary
(soft) refers to employment reductions generated by strict enforcement of rules such as
mandatory retirement and the removal of ghost workers, and voluntary refers to programs
in which employment reductions were achieved through workers voluntarily quitting
(e.g., early retirements). As is evident from the table, involuntary-hard reductions
dominate total employment reductions. However, this is primarily driven by the massive
involuntary reductions in Eastern Europe. In contrast, the use of voluntary and
involuntary-soft measures are the prevailing patterns in Latin America, Asia, and Africa.
Table 4.2 indicates that the total financial costs of the program are large
(exceeding 12 billion dollars) with the average program having total cost of 400 million
dollars.
18
The financial costs took a variety of forms including severance payments,
increased pension liabilities, and enhancements to the worker safety net (which we use a
shorthand term of the various worker assistance programs including unemployment
benefits, job search assistance, training, and relocation assistance designed to aid in the
process of worker reallocation). As is clear from the minimum and the maximum, the
scope and mix of the packages varied considerably across countries. As the patterns
98
across continents show, the transition economies in Eastern Europe relied heavily on the
worker safety net, while countries in the Southern Hemisphere and Asia relied much
more heavily on direct compensation (severance pay and enhanced pensions, i.e.,
payments made directly to workers) to workers. The latter pattern is particularly striking
for severance pay (Asia-92 percent, Latin America-74 percent, and Africa-51 percent of
total costs). In section 4.4.6, we return to this observation in the context of discussing the
likely differences in the nature of labor market adjustment costs across countries.
A wide variety of circumstances led to the retrenchment programs across
countries. The most prominent factors leading to retrenchment were fiscal crises and as
part of a general effort to reduce the size of government in the economy. However, in
some cases, the compelling factors appeared to be more structural problems with the type
and mix of government workers. Wage compression among public sector workers
leading to morale and staffing problems was a common complaint. Overstaffing
(including the problem of “ghost workers” -- workers on the payroll who did not exist but
someone was collecting the paycheck) was another common complaint. Finally, the
downsizing of the military played a prominent role as well. Not surprisingly, many
programs had multiple factors leading to the retrenchment.
The differences in the factors generating retrenchment are closely linked to the
mix of packages used across countries and continents. For some countries, the
retrenchment episode is viewed as a one-time event correcting a perceived relatively
narrow problem (e.g., ghost workers, productivity/morale problems due to wage
compression) in a particular public sector agency or enterprise. These one-time event
cases typically used some form of direct compensation (i.e., severance and enhanced
99
19
The inclusion of Hungary in our 41 programs has a clear influence on employment-
weighted results. For the most part, our results are depicted on a unweighted basis so that
outliers such as Hungary do not play a disproportionate role.
pension) and accordingly were more likely to use voluntary methods of retrenchment. In
contrast, in some countries, the public sector retrenchment is part of a fundamental
change in the role of the public sector in the economy (e.g., the transition economies).
In the latter cases, there is typically much greater attention to institutional changes in the
programs (e.g., unemployment benefits, relocation assistance, training assistance) that act
as the safety net for worker reallocation.
Much of the detailed information we collected involves the method used for
employment reduction, the nature of compensation provided and the degree of targeting.
Summary information on these characteristics is provided in Tables 4.3-4.5. Many
programs (but less than half) used both involuntary and voluntary reduction methods.
Relatively few programs used all three methods. While voluntary programs appear to be
the most popular in terms of the percentage of programs with a voluntary component,
most employment reductions were actually achieved via involuntary methods
19
. The
removal of ghost workers played a relatively minor supporting role in the employment
reductions but was quite prevalent amongst the African countries in our survey.
Table 4.4 indicates that the primary form of direct compensation to workers was
some form of severance payment. Relatively few programs offered no direct
compensation. Most programs involved some enhancement of the overall safety net
intended to assist unemployed workers and workers attempting to relocate. An important
component of the safety net enhancement was often the stipulation of some form of
training assistance.
100
As discussed in section 2, the underlying theory suggests that designing a
program that targets specific individuals or groups of individuals is likely to be important
to avoid adverse selection and to promote favorable selection. We attempt to summarize
the many different approaches to targeting used in practice by classifying programs into
three admittedly crude groups: skill-biased, age-biased and neutral. Skill-biased
programs are those that restricted the program along detailed occupational or skill
groupings. For example, the restructuring of the workforce in the tax administration in
Peru was based on new and continuing workers passing a written test. Age-biased
programs are primarily those that focused on voluntary retirement (and associated
pension enhancements) as the means for retrenchment. Neutral programs are those that
offered a simple, common package to a wide class of workers with little or no attempt to
target specific groups. Some programs are classified as both skill-biased and age-biased
if they used a mix of packages that induced selection on both types of criteria.
Table 4.5 indicates that targeting on the basis of skills or age (in the latter case
primarily through early retirement programs) is important but not pervasive. Almost 20
percent of the programs used no targeting of workers. As we will see in the following
sections, the nature of targeting is closely related to a variety of measures of “success” of
the program.
4.4.2. The Relationship Between Rehires, New Hires and Other Program
Characteristics
Based upon the discussion in section 2, tailoring a program to individual
characteristics using some form of targeting is likely to be important to avoid losing the
most productive and or essential workers. The absence of targeting may yield excessive
101
20
In many programs, workers are rehired by the same organization/unit. In some cases,
they are rehired by a different branch of the public sector.
losses of workers in key areas and/or with key skills and thus necessitate either rehiring
some of the same workers
20
who departed (a sure sign of poor targeting) or new hiring.
Using the information we have collected, we have examined the simple bivariate
relationships between the likelihood of rehires and new hires and other program
characteristics (including the nature of targeting).
Table 4.6 examines the link between rehiring and program characteristics. The
most striking results involve the role of targeting and the type of compensation offered.
Programs with rehiring are much less likely to use skill or age targeting. For example,
only 25 percent of programs with rehiring also used skill targeting while more than 60
percent of programs without rehiring used skill targeting. The flip side of this is that
more than 60 percent of the programs with rehiring did no targeting while less than 10
percent of programs without rehiring did no targeting. These findings are consistent with
the view that programs that fail to target yield severe adverse selection problems with the
most critical workers departing and the subsequent need to rehire these same workers.
Some other interesting patterns emerge that are large in magnitude but not
statistically significant at conventional levels given the relatively small sample size (i.e.,
41 observations on individual programs). Specifically, programs with rehiring were also
less likely to have a safety net component and in particular some type of training
component. The potential inference here is that the pressure to rehire may be greater if
efforts are not made to assist workers in the transition to the private sector.
Table 4.7 examines the analogous relationships between new hires and program
102
21
As noted in the introduction, rehiring of the same workers that have been separated
could be part of a coherent plan of restructuring. For example, it might be difficult
politically to target specific workers for retrenchment. Given constraints on the ability to
target workers, the optimal strategy might be to induce separations en masse and then
rehire specific workers. We found no evidence for such purposive rehiring.
characteristics. The largest and most significant results again involve the nature of
targeting and the type of compensation assistance. Programs with new hires are all age-
biased with associated pension enhancements and do not include safety net components.
It is not clear, of course, that evidence of new hiring should be interpreted or treated in a
like manner with rehiring. Rehires of the same workers who departed is much more
likely an indicator of problems with the program but new hires may involve some
intended restructuring of the workforce.
21
The results on age targeting and pension
enhancements suggest that for some programs the intent was to replace older, incumbent
workers with new workers.
4.4.3 Summary Financial Indicators
One summary financial indicator that we consider is a measure of the simple,
financial break-even period reflecting the number of years that a program requires before
breaking even on financial costs and benefits. The break-even measure is calculated as
follows. Financial costs are the present discounted value of the compensation package or
safety net expenses incurred. This present discounted value is in principle calculated
over an infinite horizon. In practice, for many of the cases, the financial costs are front-
loaded via severance payments so this present discounted value is easy to calculate.
However, in cases in which the costs are in the form of continuing safety net or pension
liabilities, we generated appropriate present discounted value measures of the financial
103
22
For some programs, the evaluation of the program by World Bank staff had already
yielded such calculations.
costs.
22
Financial benefits are measured as the presented discounted value of the wage
bill savings from the retrenched workers. The annual wage bill savings are assumed to
be constant over time and equal to the wage bill savings relevant at the time of the
implementation of the program. These assumptions permit direct calculation of the
break-even period as the number of years for the present discounted value of wage bill
savings to equal the overall presented discounted value of financial costs.
For purposes of comparability of measures across countries, a common (10
percent) discount rate is used. Under these assumptions, the break-even period is a
scale-free financial indicator that permits comparing and contrasting the financial costs
and benefits across programs. Information was sufficient to calculate the break-even
measure for about 40 percent of programs.
As seen from Table 4.8, across the programs we surveyed, a non-trivial
fraction (22 percent) yielded net losses essentially implying an infinite break-even
period. The presence of net losses reflects in part the impact of rehires or new hires or
rising compensation for retained workers. Accordingly, some programs (e.g., Peru
SUNAT) yielded no wage bill savings given the rehires, new hires and/or increase in
compensation. For the programs for which we could calculate a finite break-even period,
the average break-even period is 2.27 years with the median 1.82 years, the maximum 10
years and the minimum 0 years. Programs with immediate break-even periods (zero
years) are those with involuntary reductions without direct compensation or other
assistance programs
104
How should this simple financial indicator be interpreted? If nothing else, it
provides an indicator of the financial viability of the programs and Table 4.8 indicates
that many of the programs yield relatively rapid financial payoffs. However, as discussed
in section 2, we would ideally like to quantify the present discounted value of all costs
and benefits from retrenching public sector workers and evaluate programs accordingly.
Where does this simple financial indicator fit into such a calculation? To address this
question, it is useful to return to equation (3). For purposes of discussion, suppose we
treat the loss function in (3) as a simple linear function such that a dollar of expenditures
on public sector programs yields a dollar in distortionary losses. Then, equation (3) can
be written as:
W
public
- COMP + P
private
- C
social
- P
public
> 0 (5)
Thus, we see that the break-even period measure is essentially providing an evaluation of
the program based upon only the first two terms of equation (5). A comprehensive
evaluation, even in these crude terms, requires calculating all of the terms in equation (5).
Unfortunately, data limitations imply that we cannot measure all of the
components of (5) and thus cannot literally evaluate programs on this basis. However,
for some programs we can go a few steps further which is instructive. In particular, for
some programs we can estimate the wages that retrenched workers received in the private
sector. Assuming that workers are paid their marginal product in the private sector,
ignoring adjustment costs and assuming that productivity of the redundant (retrenched)
workers is zero in the public sector, permits a crude implementation of the terms in (5)
using this additional information. In particular, for programs that we had this
information, we calculated a modified break-even measure (denoted for labeling purposes
105
23
Sufficient information was available to calculate the payback period indicator for 20
percent of the programs.
the “economic” payback period) using this additional information. Literally, the
“economic” payback period is the number of years it requires for the benefits from the
reduced public sector wage bill to be such that present discounted value of costs equals
the presented discounted value of benefits. For this calculation, the costs and benefits for
all components other than the public sector wage bill are calculated over an infinite
horizon. In addition to measuring benefits based upon retrenched workers’ earnings in
the private sector, for programs with measurable increases in productivity in the public
sector (e.g., PERU SUNAT) we included such effects in the benefits. As seen from Table
4.8, the average calculated economic payback period is 2.1 years and the median is 0.8
years across the programs for which we could generate the measure.
23
Part of the motivation for calculating comparable summary financial and related
indicators across programs is to permit relating these financial indicators to other
program characteristics. Table 4.9 characterizes the relationship between programs with
net financial losses and other program characteristics. Programs with a net financial loss
are much more likely to include a voluntary component than programs without financial
losses. Interestingly, programs with net financial losses are more likely to use targeting,
particularly age targeting through pension enhancements. This result indicates that
targeting may be expensive, at least measured in simple, financial terms. Table 4.10
provides a related look at the relationship between the magnitude of the break-even
period and other program characteristics. Given that these are at best crudely calculated,
we divided programs into two groups -- those in the lower and upper tail of the
106
distribution relative to the median break-even period (which is 1.82 years). The results
from Tables 4.10 reinforce those from Table 4.9. Programs with relatively high break-
even periods are more likely to involve a voluntary component, more likely to use
targeting, and more likely to use direct compensation (either through severance payments
or pension enhancements).
Interestingly, some of these patterns change substantially when we examine the
payback period distribution. As seen in Table 4.11, once we incorporated even crude
information about productivity gains into the calculation, we no longer find that targeting
is especially expensive in terms of implying a high payback period. Not surprisingly, we
still find that programs with high calculated payback periods are more likely to involve
direct compensation and in particular severance pay.
To sum up, simple calculations of financial break-even periods indicate that many
of the programs have relatively rapid financial payoffs. Alternatively, a non-trivial
fraction are financial losers. However, caution must be used in interpreting such
financial indicators as measures of relative success of programs. There are many relevant
components of the private and social costs and benefits not captured by these measures.
Even the limited attempt we made to incorporate relevant economic costs and benefits
indicates that the inferences based upon narrowly defined financial indicators may be
quite misleading.
4.4.4 Highlights of Individual Programs
We now turn to summarizing the experiences for a selected set of individual
programs. The objective here is threefold. First, consideration of individual programs
provides a more comprehensive view of the scope and heterogeneity across programs.
107
24
In general, productivity measurement in service sector, specially public sector is
difficult. Most work uses quantitative or technical measures. See Griliches (1992).
Second, the discussion of individual programs permits illustration of some of the key
patterns detected in the discussion of basic patterns. Third, our summary tables and
related analyses admittedly leave much out. Part of the reason for this is lack of data
suitable for quantifying effects in a summary fashion. For example, quantifying the
nature of productivity gains associated with retrenchment as well as the adjustment costs
is difficult both in principle and in practice given available data
24
. However, in some
cases, qualitative information about these issues are available which sheds some
additional light.
Programs in six countries (Peru, Argentina, Uganda, Ghana, India, and Hungary)
are highlighted. Key characteristics are presented in Table 4.12. In the discussion, the
ordering of the country/program cases reflects a rough attempt to select cases that
highlight the role of the following issues: (i) targeting; (ii) productivity gains; and (iii)
worker safety net and labor market adjustment. This ordering is only approximate since
all of these issues are present for every program. In the summary statistics in Table 12,
bold entries in the table highlight the relevant key issue that motivates our inclusion of
this program for special discussion.
PERU: Two programs were initiated as part of a broader exercise of fiscal
austerity and adjustment. Among the problems were the under-funded pension system,
the posting of “ghost” employees by the regions to enhance employment based federal
transfers, and erosion in public salaries making it difficult to hire and retain qualified
workers.
108
25
The primary source of information on these two programs is a World Bank (principal
external donor for both projects) document: World Bank, 1994. Peru Public Expenditure
Review. Report No. 13190-PE
26
Further, federal government staff reductions were offset by increases in employment by
regional governments with most of it re-hiring of erstwhile federal staff.
With the support of external donors, Peru initiated two labor adjustment
programs in 1991, one for the Peru civil service and the other for the Peru tax
administration authority called SUNAT. These were respectively completed in 1993 and
1992.
25
The civil service program used all three involuntary and voluntary employment
reduction methods - separating about 250,000 workers over three years. Induced
departures employed both lump-sum severance and pension enhancements. Little
evidence on skill targeting was found. Targeting by age was implicit in the use of
pension enhancements to induce voluntary separation. Poor targeting aggravated the
shortage of human resources with many of the most qualified staff leaving. The poor
targeting and accompanying shortages probably led to the significant rehiring of the
separated workers.
26
In total, 163,000 workers of the originally retrenched workers were
rehired. Severance packages of about $1000 were provided to less than half (112000) of
the workers separated. This limited the direct financial losses associated with the
significant rehiring. Our simple measure of the break-even period for this program was
2.6 years, more than the median of 1.82 years for all the programs studied. However, this
measure does not reflect the loss of productivity associated with the shuffling of the same
workers in and out of the same positions.
Interestingly, the other program in Peru appears to be a model of good targeting.
The SUNAT (tax authority) program also used a mix of voluntary and involuntary
109
27
Deflating wages using the CPI instead of the exchange rate returns an increase of
similar magnitude. The instantaneous increase would drop from 20-fold to 18.6 times.
28
This discussion neglects the welfare costs (e.g., reduction in hours and output) due to
increased enforcement of tax collections.
reduction methods. Voluntary separations came with an enhanced pension. Involuntary
workers were selected on the basis of a written test. Thus, targeting was worker specific
and objectively determined. Two-thirds of the workforce (2034 workers) was separated.
Subsequently, SUNAT hired 1309 new workers, again based on a written test. Since
productivity and competence were objectively established and necessary skills acquired
afresh, little basis remained for rehiring (skilled but severance-induced separated)
workers. Rehiring was barred for ten years and none was found. The average salary for
affected workers increased from US$50 per month to US$1000 per month
27
. Tax
collections more than doubled and so did SUNAT’s revenues (2 percent of tax
collections.) These were insufficient to cover the salary increase and the scheme incurred
a net financial loss of US$47 million in present value terms. However, the entire
increase/improvement in tax collections can be interpreted as substantial gain for the
government (and in principle the economy
28
). Incorporating the tax collection gains into
our calculation of the payback period yields a payback period of only 0.002 years. This
highlights the importance of evaluating program performance along multiple dimensions,
not just organization-level financial costs and benefits but the broader impact of the
program. It is worth noting that neither of these two programs emphasized job search,
relocation, training or other forms of assistance to reduce the adjustment costs for the
retrenched workers.
110
29
See Robbins (1996) for a case study on public sector retrenchment in Argentina since
1990. This paper examines issues of adverse selection and post-retrenchment worker
status using micro data.
30
World Bank, 1995. Argentina Public Sector Reform Project Completion Report. Report
No. 14781.
31
In the source reports (Reports no. 14781, 14809 and the World Development Report
1994,) we found no evidence of rehiring. However, some anecdotal evidence indicates
occurrence of rehiring.
ARGENTINA: With external assistance from the World Bank, Argentina
undertook programs for the federal administration and the railroads (as part of a broader
program for public enterprises) during the years 1990-92 and 1991-94 respectively
29
.
Given chronic public sector deficits and endemic inflation, extensive public sector reform
was considered essential. Wages accounted for 70 percent of federal expenditure
excluding interest payments and transfers. Employee redundancy level was estimated at
50 percent
30
. Thus, staff restructuring was an important component of overall reform.
The federal administration program retrenched 400,000 workers. No explicit
targeting by worker was found. Involuntary separations implied some implicit targeting
in the selection of redundant workers. Some evidence indicates an age bias. Also,
targeting by function was attempted in that the federal government’s role was restricted
to the provision of security, social services and economic management. Rehiring was
barred by law and none was found
31
, despite the large scale of retrenchment and the
absence of targeting. The railroads’ program separated about 73,000 workers using
multiple methods. It used both voluntary and involuntary reduction methods. An
important feature of Argentinean programs is the heterogeneity in compensation
amounts. In federal administration, average severance payment per worker was US$
111
32
World Bank, 1995. Argentina Public Sector Reform Project Completion Report. Report
No. 14781.
33
World Bank, 1995. Public Enterprise Adjustment Loan Audit Report. Report No.
14809.
3000 amounting to a one-time cost of US$ 425 million
32
. In railways, the average
severance per worker was US$12000 or a one shot payment of US$ 360
33
million in
railways . Annual wage bill saving was expected to be US$1000 million in the federal
service and US$238 million in the railroads. Calculated break-even periods were 0.4
years and 1.6 years respectively in the federal administration and the railroads, with the
difference reflecting the higher compensation per worker in railroads.
While difficult to measure, some evidence suggests that the restructuring in both
railways and federal administration was characterized by substantial productivity gains.
Railways experienced an increase in freight miles per worker and passenger miles per
worker of about six to seven times. In federal administration, a 50 percent cut in
processing time is claimed.
UGANDA: High fiscal deficits combined with inadequate pay levels and wage
compression led to a civil service labor restructuring program in 1992 - 94.
Concomitantly, the military reduced size in line with peace-time national priorities and
conducted a program in 1992 - 95. Both were partly supported by external financing.
The civil service program employed all three reduction methods restricting the package
offer to a fraction of the workers separated. Separations were targeted by skill and tenure.
Some critical jobs and personnel were “ring-fenced” that is, protected from retrenchment.
Such workers did not have the option of obtaining enhanced benefits associated with
112
34
Republic of Uganda, 1994. Management of Change, Context, Vision, Objectives,
Strategy and Plan: Civil Service Reform.
35
Although an evaluation of the civil service staff performance was not available at the
time of our data collection, measurement indicators to evaluate the performance
improvements were planned by the EDI of the World Bank.
36
World Bank, 1995. From Swords to Ploughshares: The Uganda Demobilization and
Reintegration of Ex-combatants in Uganda. Confidential Mimeo.
voluntary separation. The gross number of separations was approximately 150,000. The
average compensation per worker was US$320
34
or about twice the annual GDP per
capita. However this total expense of US$ 16 million constituted a measured net
financial loss since there was no saving on the wage bill. The wage bill increased due to
salary revisions for the remaining civil service staff and was intended to improve service
performance. Interviews with program advisors revealed a strong motivation for the
salary hike, viz., to achieve parity between the civil service average wage and the living
wage in Uganda so as to motivate the civil service staff and generate productivity
increases.
35
The military program was a post-war retrenchment exercise. It used both
voluntary and involuntary reduction methods to reduce strength by 44000 employees.
Skill targeting was based on soldier performance and military requirements. Average
compensation was US$955, provided partly in kind and distributed over several months.
The calculated break-even period was 2.7 years. In Uganda, some measurement of post-
retrenchment earnings for veterans was available. Actual post-separation earnings found
for the military programs
36
were up to 71 percent of GDP per capita in Uganda.
Including these earnings of a total of US$ 5.4 million per year in Uganda’s case yields a
113
calculated payback period of 1.2 years. The rarely observed actual data on post-
separation earnings is interesting, particularly given that the earnings are lower than the
average. While this evidence is too idiosyncratic to draw general inferences, it does
highlight the difficulty in generating the appropriate cost-benefit calculations. Using
average earnings in the private sector as a proxy (even when this is available ) is
obviously inappropriate in this case.
An interesting feature of the military program was the safety net nature of the
compensation that enabled veterans to reintegrate into civilian society. Assistance in kind
was tailored to the intended post-separation occupation, for instance farmers were
provided land and farm implements. Further the assistance was known to be temporary
(six months), creating an incentive for the veteran to hasten adjustment and thereby
limiting program costs.
GHANA: Motivated by a very large public sector wage bill as a fraction of total
government expenditures, the Government of Ghana initiated a program for the civil
service in 1987 with partial-financing from the World Bank. The program concluded in
1992. It was driven by voluntary departures induced by additional severance and
complemented by soft involuntary measures - removing ghost workers and enforcing
retirement age. Voluntary departure, with compensation, was conditional on
employment being non-critical to the performance of the unit. Necessary skills were thus
sought to be protected. 73,000 workers were separated. More than three-fourths of the
total workers separated received the package. Average compensation was US$700. The
formula of two months’ base pay for each year of uninterrupted service exceeded the
legally required four months’ base pay in most cases. Anecdotal evidence indicated
114
37
Alderman, H., S. Canagarajah and S. Younger (1994).
some rehiring in subvented organizations (attached to ministries but not covered under
budgetary process.) The calculated break-even period was 1.8 years. Despite a
generous package offer, the BEP was relatively low.
Regarding assistance to workers to making a transition to the private sector,
extensive training through Program of Action to Mitigate the Social Costs of Adjustment
(PAMSCAD) was planned. However assistance for job search & placement and for
retraining was inactive. Courses in entrepreneurial development were offered. For
potential farmers inputs including land were provided. In rural areas retrenchees were
also eligible to participate in Food-for-Work schemes. A sample survey of the retrenched
workers
37
found that about 10 percent of the retrenchees quit the labor force. Of the rest
97 percent were re-employed by the second year, about 20 percent in formal sector wage
jobs and the rest in self-employment or informal sector jobs.
INDIA: Fiscal and external payments’ deficits led India to initiate a stabilization
and structural adjustment program in 1991. A program to support public sector labor
restructuring was required. This retrenchment support program was directed at public
enterprises declared to be sick (with several years’ of accumulated losses). A voluntary
retrenchment program in the sick public sector textile firms separated about 70,000
workers over more than two years (1993-94). The average cost per worker was about
US$17000. The formula used was 30 days’ wages for each year worked compared with
the legally required 15 days’ wages for each year of permanent service. The scheme
incurred a net loss of US$276 million in present value terms given the exorbitant
115
38
The high compensation reflects in part the effects of a rigid anti-retrenchment and anti-
closure law in India. For a discussion of this, see Basu, Fields and Debgupta, (1996).
39
PA Consulting Group, 1993. Social Safety Net Adjustment Credit Technical Assistance -
Final Report. Submitted to the World Bank.
40
See, Commander et. al. (1995).
compensation
38
. Given the incentives, the voluntarily departing workers may have had
long tenures increasing the severance amount granted. No explicit targeting mechanism
by tenure or age was found. On skills, the maximum skill level among the retrenchees
was that of a supervisor.
Results from a sample survey
39
indicated that all retrenched workers remained in
the labor force and 80 percent were re-employed. Of the total surveyed, 32 percent were
in wage jobs, one-fourth of them in the same industry, that is, textiles. The rest (48
percent) were self-employed.
HUNGARY: Reform and stabilization programs aimed at a transition to a
competitive economic system led to mass dismissals in state enterprises during 1990-
1992. About 1.7 million workers constituting 8.7 percent of the country’s labor force
were separated. This is extremely large compared with other programs’ scope - Ghana
(0.2 percent of labor force,) India (0.01 percent.) The annual wage saving amounting to
US$298 million but was exceeded by the increase in the annual safety net expenditures of
US$858 million. The exercise was clearly financially costly but reflects a fundamental
change in the structure of the economy. A comprehensive set of institutions designed to
act as a safety net for worker reallocation is very common in western economies and
Hungary is obviously attempting to follow that model.
Surveys
40
indicate that retrenched workers remained in the labor force. Most of
116
the re-employed were in the private sector, in trade and service industries. The average
private manufacturing wage was found to be about 70 percent of state firm wage
(providing some indication of the size of the rent in public sector firms.) The nature of
the labor market adjustment appeared to change as the transition process proceeded and
accelerated. Early on, many of the retrenched workers left voluntarily and left the labor
force (e.g., via voluntary retirement). Others left voluntarily and transited directly to
other jobs. However, as involuntary layoffs increased, an increased fraction of the
retrenched workers entered unemployment. Moreover, the available evidence suggests
that unemployed workers had an increasingly difficult time finding jobs. Early in the
transition in February 1991, the outflow rate from unemployment implied a steady-state
duration of unemployment of 7 months. By November 1992, the implied steady-state
duration was 50 months.
4.4.5 Aggregate Factors
Another means of evaluating the programs is to look at relevant macroeconomic
indicators prior to and during the program including indicators of fiscal posture.
Examining such macro indicators provides a rough independent check of the factors that
lead to a retrenchment episode. Further, evaluating macro indicators prior to and during
the program provides another means of evaluating the impact of the program. In many
cases, the actual retrenchment program is too small by itself to have important macro
effects. An exception are the programs in Eastern Europe (e.g., Hungary) in which there
was massive public sector downsizing. However, even in cases where the specific
program we analyzed is too small by itself to have macro effects, the program may have
been part of a larger effort to restructure and downsize the public sector. Thus, the
117
41
Several sources were used for this purpose including the Penn World Tables, the cross-
country labor market database being generated by Rama (1995), the World Bank World
Tables, and the International Finance Statistics.
42
The data are not complete for all variables and all countries over this entire horizon.
The unemployment rate series is probably the worst in this regard. Considering
deviations from country-specific means mitigates the problems with limited data
availability. However, appropriate caution should be used in considering these results
(particularly for unemployment) with the additional usual caveats that there is likely
considerable measurement error in the aggregates available by country.
results that follow can be interpreted in this light.
For each of the programs we surveyed, we acquired data on the unemployment
rate, the real GDP growth rate, the ratio of the current budget deficit to GDP, the ratio of
domestic debt to GDP, the ratio of Foreign Debt/GDP and the ratio of government
spending to GDP.
41
To remove the influence of idiosyncratic effects across countries, all
variables are characterized in terms of deviations from country-specific means. Country-
specific means for each variable are calculated from data for 1980-92.
42
Table 4.13
presents the patterns of these variables for the five years prior to the initiation of the
retrenchment program in each country and during the program (e.g., for a specific
country if the retrenchment program began in 1990 and ended in 1992 the prior year
calculations reflect data for 1980 to 1989 while the calculations during the program
reflect data averaged over 1990 to 1992).
Unemployment rises in the years preceding a program and GDP growth rates fall
in the years leading up to a program. Interestingly, the unemployment rate appears to fall
somewhat during programs while GDP growth remains below average. These patterns
are roughly consistent with the view that retrenchment programs are often precipitated by
economic crises. The evidence on fiscal indicators is somewhat mixed at first glance.
118
43
The discussion that follows on the formal and informal labor market is inadequate on a
number of dimensions and does not reflect the rich body of research that has been
conducted on the role of formal, informal and rural sectors for labor market dynamics in
developing countries (e.g., see Mazumdar (1989) for an overview of this research). Our
point here is simply to highlight the potential importance of these considerations for
cross-country differences in the nature of labor market adjustment.
The government share of GDP rises steadily prior to the onset of a program and falls
during the program. Somewhat surprisingly, there is not an accompanying pattern for the
deficit to GDP ratio and the debt to GDP ratio. For both of these domestic fiscal
indicators, there is an improvement in the fiscal posture preceding the program. However,
this appears to in part reflect the rising foreign debt to GDP ratio that precedes a program
and continues during a program. Putting these pieces together is consistent with the view
that these retrenchment programs are often part of a more general austerity program that
are supported in part by foreign assistance.
4.4.6 Missing Pieces -- Measurement and Characterization of Adjustment Costs
The discussion in the above subsections provides a range of information about the
experiences of retrenchment programs. In terms of relevant factors for evaluation, one
of the biggest missing pieces is associated with the conceptual and measurement
problems in quantifying private and social adjustment costs. While quantification is
beyond the scope of readily available information, there are a few qualitative
observations that can be made regarding the cross-country variation in labor market
adjustment costs. Specifically, qualitative evidence suggests that the considering the
impact of the formal and informal sectors is important.
43
In countries for which the
informal sector plays a large role (e.g., Ghana), many of the retrenched public sector
workers found employment relatively quickly, albeit mostly not in the formal wage
119
sector. In contrast, in the transition economies, retrenched workers from public sector
enterprises are facing increasing difficulty leaving unemployment because of difficulties
in finding employment in the formal wage sector.
Stated very roughly, these observations suggest the following potentially
important difference across countries in terms of the relevant labor market adjustment.
One set of countries seem to yield relatively quick adjustment for retrenched public
sector workers with the informal sector absorbing retrenched workers. Other countries
seem to yield long adjustment with retrenched workers from public sector enterprises
experiencing long spells of unemployment while seeking jobs in the formal private
sector. These differences are further reflected in the mix of policies we observe across
countries. For example, in Africa and to some extent Latin America, many of the
programs we surveyed involved direct compensation and relatively little changes in the
formal worker safety net. In contrast, in transition economies, the resources for
retrenched workers have primarily been devoted to the worker safety net.
This contrast is likely overstated in a number of respects. First, employment in
the informal sector may involve substantial underemployment. Second, workers reported
to be officially unemployed are arguably often engaged in some unreported informal
sector work activity. Nevertheless, even this rough characterization highlights the
potentially important role of differences across countries in the structure of economies
and in turn the differences in the nature of labor market adjustment costs.
What information would we have liked to have to quantify adjustment costs? To
be fair, this question is difficult to answer in advanced economies as well as for
developed and transition economies. Nevertheless, from advanced economies (and some
120
44
See, e.g., OECD (1993) and OECD (1996).
transition economies) we have learned a great deal about labor market flexibility and
adjustment by examining the flows of workers and the flows of jobs (see, e.g., Davis,
Haltiwanger and Schuh (1996)). For example, some advanced economies exhibit high
inflows and outflow rates to unemployment (e.g., the United States and Canada) while
others exhibit low inflows and outflow rates from unemployment (e.g., France and the
Italy).
44
The variation in the inflow and outflow rates to unemployment translate into
striking differences in the importance of long-term unemployment across countries. In
the United States the percent of long term unemployed (less than 12 months) in the early
1990s was about 6 percent while the equivalent figure in Italy was 71 percent. Even this
limited and indirect evidence yields two immediate inferences for the analysis of public
sector retrenchment. First, the low outflow rates from unemployment and accompanying
importance of long term unemployment in some countries implies that the magnitude of
the adjustment costs are potentially very large. Second, the variation across countries is
substantial implying that evaluation of a specific program in a specific country will
depend critically on the nature of the labor market adjustment in the country.
4.5 Concluding Remarks.
This survey and analysis of cross country experiences with retrenchment reflects
the gathering and processing of a wealth of information about individual programs.
Details about the nature of the compensation packages, the employment reduction
methods used, the nature of targeting, the worker safety net components of the program,
the financial costs and benefits and the nature of adjustment (including the presence of
121
significant rehiring of the same workers) have been documented and examined. One of
the primary objectives and thus contributions of this paper is to gather all of this
information into one place in an easily accessible manner.
A number of interesting findings emerge from analysis of basic facts. Theory
suggests that individually tailoring programs to account for worker heterogeneity and to
avoid adverse selection problems is very important. Consistent with this view, we found
that programs that used targeting on the basis of skills and age, used multiple methods of
employment reduction, and used a combination of compensation packages that included
enhancements of the safety net for assisting the reallocation of worker were much less
likely to exhibit problems with rehiring. We also found that programs with this multi-
dimensional approach with targeting tended to be financially expensive. However, both
quantitative and qualitative information suggest that there is a potentially large payoff in
productivity gains and in reducing adjustment costs.
Our analysis of financial indicators indicates that there is considerable
heterogeneity in the financial viability across the programs. Many of the programs have
a rapid financial payoff in the sense that the wage bill savings from retrenchment quickly
cover the financial costs of the programs from worker compensation and assistance.
Alternatively, a non-trivial fraction of the programs are clear financial losers. However,
we emphasize that while simple financial indicators are of obvious interest, they are
inappropriate for evaluating the relative success of programs because they omit many of
the potentially relevant private and social costs and benefits.
Unfortunately, many of the relevant private and social costs and benefits are
difficult to quantify given data (and conceptual) limitations. Especially difficult to
122
quantify are the individual and economy-wide (social) adjustment costs that are incurred
as part of restructuring programs. The data required to assess these adjustment costs are
generally not available and there are accompanying difficult conceptual issues. This
problem has both micro and macro dimensions. First, on the micro side, little effort has
been provided to collect data systematically on the post-separation experiences of
retrenched workers. Second, on the macro side, characterizing and understanding the
nature of these adjustment costs requires an understanding of the myriad of factors (e.g.,
institutions) that affect the labor market adjustment processes within individual countries.
The large observed differences across even advanced economies on this dimension
provides prima facie evidence in support of the need for this type of information. Third,
in countries with massive public sector retrenchment episodes, the labor market
adjustment process is endogenous to the retrenchment program itself. Given the
importance of the issues, more resources should be devoted to developing the data
necessary to evaluate these adjustment costs.
123
Chapter 5: Conclusions
The direct measurement of adjustment costs establishes the existence of
substantial costs for workers. Weaker but disturbing evidence points to generational
effects where job prospects of displaced workers’ children are adversely affected.
The economy while growing seems unable to absorb displaced workers or
otherwise provide income security and a cushion against accompanying effects
such as on children education. Transition supporting institutions exist in some
measure but their coverage is restricted and where these exist, effectiveness is low.
Employers are adopting complex means to alter workforce in the backdrop of
restrictive law, increasing competition and changing market conditions. These are
legal restructuring using Voluntary Retirement Schemes (VRS), illegal
restructuring using lockouts, and disputed restucturing with resolution sought
through labor courts and tribunals. Though labor adjustment is taking place, the
methods are costly. Given cost-cutting pressures from increasing competition,
costly adjustment reduces the gains from such adjustment.
The econometric analysis establishes that adjustment is slow. There is evidence
that after job security restrictions in 1976, (i) there is a one-time fall (about 28
percent) in labor demand and (ii) a fall in the speed of adjustment (3 percent) and
124
additionally so (38 percent) for large industries. The extended coverage of the job
security regime in 1982 led to these effects becoming larger for covered industries
and slightly smaller for uncovered industries based on average employment size.
These effects are not reversed by the economic reform package of 1991. No
significant increase in speed of adjustment is evident post-1991 though there may
have been a one-time boost to labor demand. Though more adjustment and more
VRS programs are taking place post-1991, the speed of adjustment has not really
increased.
Deregulation begun prior to 1991 and liberalization from 1991 have increased
the extent and intensity of competition. Interacting competition measures with
adjustment variables sheds better light on the effects. Competition increases
average employment and reduces the sluggishness of labor adjustment or increases
the speed of adjustment for medium sized industries (size 100 and above by average
employment size) but the opposite holds for larger industries (size 300 and above
by average employment size). For the latter reuslts, pre-existing labor hoarding
may be responsible.
The survey and analysis of cross country experiences with retrenchment reflects
the gathering and processing of a wealth of information about individual programs.
125
Details about the nature of the compensation packages, the employment reduction
methods used, the nature of targeting, the worker safety net components of the
program, the financial costs and benefits and the nature of adjustment (including
the presence of significant rehiring of the same workers) have been documented
and examined. Both quantitative and qualitative information suggest that there is a
potentially large payoff in productivity gains and in reducing adjustment costs from
programs with a multi-dimensional approach. The analysis of financial indicators
indicates that there is considerable heterogeneity in the financial viability across the
programs - many of the programs have a rapid financial payoff but a non-trivial
fraction of the programs are clear financial losers. However, while simple financial
indicators are of obvious interest, these are inappropriate for evaluating the relative
success of programs because they omit many of the potentially relevant private and
social costs and benefits. Unfortunately, many of the relevant private and social
costs and benefits are difficult to quantify given data (and conceptual) limitations.
Should the job security regime be liberalized and consequent mandated
separation costs be reduced? The debate on this issue has been pending resolution
since the initiation of reforms in 1991 in India. Both the economic and political
costs are immense. The evidence of substantial adjustment costs and sluggish
126
employment adjustment here, though modest, urges a qualified yes.
Several issues remain to be examined. The assumption that ignoring capital
doesn’t bias estimates of labor based on studies of the U.S. economy may be less
true for the Indian economy. Within the existing framework here, data on trade
unions and on external trade should be incorporated. As of now, the trade union
data based on voluntary but intermittent reporting is at best indicative. Data on
external trade can be used once the industrial and trade classifications are mapped.
Finally, heterogeneity arising from idiosyncratic shocks is lost due to the use of
industry level data. The industry data based estimates are revealing but the
magnitudes may well vary substantially at the unit level. Some of the ambiguity
emerging here may be resolved.
127
Table 2.1A: Employment Aggregates
Year
Population Labour
force
Employ-
ment
Agriculture Industry Services Unemploy-
ment rate*
million million million million million million percent
1971 548 - 180 130 20 30 -
1981# 685 - 223 148 - - -
1983 718 309 303 207 43 52 1.9
1987/88 790 333 324 208 52 64 2.8
1991 816 - 306 186 - - -
1993/94 895 382 374 242 58 77 2.0
1999/2000 1004 406 397 238 69 90 2.2
2001 1025 - 403 235 - - -
Source: Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Notes:
# Excludes the state of Assam
* Using usual principal and subsidiary status (upss)
Indian Census data for 1971, 1981, 1991, 2001; Rest- NSSO quinquennial household surveys
Table 2.1B: Growth Rates - Employment and GDP
Period
Reference GDP Employ-
ment
Five-year
plan
Annual
average
Annual
average
1951-56 I 3.6 0.39
1956-61 II 4.2 0.85
1961-66 III 2.8 2.03
1967-69 Annual plans 3.9 2.21
1969-74 IV 3.3 1.99
1974-79 V 4.8 1.84
1980-85 VI 5.7 1.73
1985-90 VII 5.8 1.89
1990-92 Annual plans 3.4 1.50
1992-97 VIII 6.5 2.44
1997-2002 IX 5.8 1.07
Source: Planning Commission, Government of India, Five-Year Plan Documents
128
Table 2.2: Employment in Organised Sector
Year ending
as of
March31
Organised
employment -
public sector
Organised
employment -
private sector
Organised
employment -
total
Growth -
public
sector
Growth -
private
sector
Growth -
total
organised
million million million percent percent percent
1975 12.87 6.80 19.67 - - -
1976 13.36 6.84 20.21 3.8 0.6 2.7
1977 13.88 6.87 20.74 3.8 0.4 2.7
1978 14.44 7.04 21.48 4.1 2.5 3.6
1979 15.05 7.21 22.25 4.2 2.3 3.6
1980 15.08 7.24 22.32 0.2 0.4 0.3
1981 15.48 7.40 22.88 2.7 2.2 2.5
1982 15.88 7.61 23.48 2.5 2.9 2.6
1983 16.46 7.55 24.01 3.7 -0.7 2.2
1984 16.86 7.43 24.29 2.4 -1.6 1.2
1985 17.27 7.31 24.58 2.4 -1.6 1.2
1986 17.68 7.38 25.06 2.4 0.9 1.9
1987 18.03 7.36 25.39 1.9 -0.2 1.3
1988 18.32 7.39 25.71 1.6 0.4 1.3
1989 18.51 7.48 25.98 1.0 1.2 1.1
1990 18.77 7.58 26.35 1.4 1.4 1.4
1991 19.06 7.68 26.73 1.5 1.2 1.4
1992 19.21 7.85 27.06 0.8 2.2 1.2
1993 19.33 7.85 27.18 0.6 0.1 0.4
1994 19.45 7.93 27.38 0.6 1.0 0.7
1995 19.47 8.06 27.53 0.1 1.6 0.5
1996
19.43 8.51 27.94
-0.2 5.6 1.5
1997 19.56 8.69 28.25 0.7 2.0 1.1
1998 19.42 8.75 28.17 -0.7 0.7 -0.3
1999 19.42 8.70 28.11 0.0 -0.6 -0.2
2000 19.31 8.65 27.96 -0.5 -0.6 -0.5
2001 19.14 8.65 27.79 -0.9 0.1 -0.6
2002 18.77 8.43 27.21 -1.9 -2.5 -2.1
2003 18.58 8.42 27.00 -1.0 -0.2 -0.8
Sources:
Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Economic Survey, Ministry of Finance, Government of India
129
Table 2.3: Disputes data
Year Industrial Disputes Strikes Lockouts Strikes-% change Lockouts-% change Lockouts-% share Loss (Rs. Mill)
No. Mandays
lost
No. Mandays
lost
No. Mandays
lost
No.-%
change
Mandays
lost - %
chg
No.-%
change
Mandays
lost - %
chg.
Of no.
disputes
of lost
mandays
Mandays lost*Avg.
Labor Cost/
manday
1970 2889 20563381 2598 14749067 291 5814314
- - - - 10.1 28.3 78.3
1971 2752 16545636 2478 11802589 274 4743047
-4.6 -20.0 -5.8 -18.4 10.0 28.7 68.2
1972 3243 20543916 2857 13748262 386 6795654
15.3 16.5 40.9 43.3 11.9 33.1 -
1973 3370 20626253 2958 13862211 412 6764042
3.5 0.8 6.7 -0.5 12.2 32.8 -
1974 2938 40262417 2510 33643524 428 6618893
-15.1 142.7 3.9 -2.1 14.6 16.4 116.6
1975 1943 21900931 1644 16706369 299 5194562
-34.5 -50.3 -30.1 -21.5 15.4 23.7 113.7
1976 1459 12745735 1241 2798975 218 9946760
-24.5 -83.2 -27.1 91.5 14.9 78.0 239.8
1977 3117 25320072 2691 13410141 426 11909931
116.8 379.1 95.4 19.7 13.7 47.0 286.6
1978 3187 28340199 2762 15423344 425 12916855
2.6 15.0 -0.2 8.5 13.3 45.6 332.0
1979 3048 43853518 2708 35803816 340 8049702
-2.0 132.1 -20.0 -37.7 11.2 18.4 225.6
1980 2856 21925026 2501 12017919 355 9907107
-7.6 -66.4 4.4 23.1 12.4 45.2 310.5
1981 2589 36583564 2245 21208130 344 15375434
-10.2 76.5 -3.1 55.2 13.3 42.0 538.6
1982 2483 74614764 2029 52112874 454 22501890
-9.6 145.7 32.0 46.3 18.3 30.2 878.7
1983 2488 46858464 1993 24921446 495 21937018
-1.8 -52.2 9.0 -2.5 19.9 46.8 -
1984 2094 56025240 1689 39956861 405 16068379
-15.3 60.3 -18.2 -26.8 19.3 28.7 -
1985 1755 29239466 1355 11486803 400 17752663
-19.8 -71.3 -1.2 10.5 22.8 60.7 954.6
1986 1892 32748228 1458 18823648 434 13924580
7.6 63.9 8.5 -21.6 22.9 42.5 837.6
1987 1799 35358372 1348 14026081 451 21332291
-7.5 -25.5 3.9 53.2 25.1 60.3 1404.1
1988 1745 33946925 1304 12529895 441 21417030
-3.3 -10.7 -2.2 0.4 25.3 63.1 1635.2
1989 1786 32663377 1397 10695112 389 21968265
7.1 -14.6 -11.8 2.6 21.8 67.3 -
1990 1825 24086170 1459 10639687 366 13446483
4.4 -0.5 -5.9 -38.8 20.1 55.8 1307.3
1991 1810 26428092 1278 12428333 532 13999759
-12.4 16.8 45.4 4.1 29.4 53.0 1527.2
1992 1714 31258744 1011 15132101 703 16126643
-20.9 21.8 32.1 15.2 41.0 51.6 2019.5
1993 1393 20300653 914 5614515 479 14686138
-9.6 -62.9 -31.9 -8.9 34.4 72.3 2027.9
1994 1201 20983082 808 6651054 393 14332028
-11.6 18.5 -18.0 -2.4 32.7 68.3 2092.6
1995 1066 16289569 732 5719961 334 10569608
-9.4 -14.0 -15.0 -26.3 31.3 64.9 1757.0
1996 1166 20284803 763 7817869 403 12466934
4.2 36.7 20.7 18.0 34.6 61.5 2382.9
1997 1305 16971389 793 6295365 512 10676024
3.9 -19.5 27.0 -14.4 39.2 62.9 2040.6
1998 1097 22061984 665 9349108 432 12712876
-16.1 48.5 -15.6 19.1 39.4 57.6 2679.4
1999 927 26786856 540 10625171 387 16161685
-18.8 13.6 -10.4 27.1 41.7 60.3 4176.7
2000 771 28763121 426 11958694 345 16804427
-21.1 12.6 -10.9 4.0 44.7 58.4 4036.6
2001 674 23766809 372 5562765 302 18204044
-12.7 -53.5 -12.5 8.3 44.8 76.6 4282.5
2002 579 26585919 295 9664537 284 16921382
-20.7 73.7 -6.0 -7.0 49.1 63.6 3907.7
Source: Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Notes: Data pertains to work stoppages where at least 10 workers are involved dirsectly or indirectly.
130
Table 2.4A: Enforcement - Industrial Tribunals and Labor Courts - Dispute Resolution
Year Central Government Industrial Tribunals and Labour Courts
Industrial
disputes
pending
beginning
of year
Industrial
disputes
received
during the
year
Industrial
disputes
disposed
during the
year
Industrial
disputes
pending
end of year
Applica-
tions
pending as
on January
1
Applica-
tions
received
during the
year
Applica-
tions
disposed
during the
year
Applica-
tions
pending
end of year
Percent
disputes
disposed
Percent
applica-
tions
disposed
1981 - - - - - - - - - -
1982 1093 558 337 1314 - - - -
20.4 -
1983 1298 643 602 1206 - - - -
31.0 -
1984 1206 834 652 1173 3359 1844 2129 2796
32.0 40.9
1985 1173 603 591 1185 2796 2842 2151 3485
33.3 38.2
1986 1185 909 456 1648 3485 2390 1720 4116
21.8 29.3
1987 1651 1191 509 2333 4131 4469 2042 6564
17.9 23.7
1988
- - - - - - - - - -
1989 2810 1450 803 3457 9142 3943 3891 9224
18.8 29.7
1990 3505 1504 747 4262 9514 1556 3339 7731
14.9 30.2
1991 4266 1440 1005 4701 7731 2535 2143 8123
17.6 20.9
1992 4702 616 403 4915 8165 1783 955 899
7.6 9.6
1993 5093 989 440 5642 9829 816 2432 8213
7.2 22.8
1994 5807 1211 649 6369 8399 1519 1814 8104
9.2 18.3
1995 6369 1310 1275 6404 8104 1842 3934 6012
16.6 39.6
1996 6409 1042 1149 5928 6033 1350 2510 4849
15.4 34.0
1997 6310 1460 977 6793 4776 1114 1990 3900
12.6 33.8
1998 7127 1280 724 7683 4337 1363 1002 4698
8.6 17.6
1999 7576 1622 770 8428 4203 234 499 3939
8.4 11.2
2000 8398 2035 722 9711 3328 1310 835 3803
6.9 18.0
2001 - - - - - - - - - -
2002 - - - - - - - - - -
Source: Indian Labour Yearbook, Labour Bureau, Ministry of Labour, Government of India
Notes: Data pertains to work stoppages where at least 10 workers are involved dirsectly or indirectly.
These statistics are collected on a voluntary basis from the primary units. Due to their voluntary nature,
the quality of data is not always upto the mark.
131
Table 2.4B: Enforcement - Other Acts
Year Factories Act 1948 Shops and Establishments Act
Factories
on register
Factories
Inspected
Percent
Inspec-
tions
Factories
Number
of Convic-
tions
Convictions
as % of
inspections
Shops
and
Establ
no.
Shops and
Establ
Inspected
Percent
Inspections
Shops &
Establ
Number of
prosecu-
tions
Cases
disposed
off by
courts
Amount of
fines
realized
(Rs.'000)
Prosecu-
tions as %
of
inspections
1970 84426 62200 73.67 8678 13.95 2320218 2050696 88.38 111703 78450 2167063 5.45
1971 88089 63018 71.54 9567 15.18 2404510 2026412 84.28 105358 89121 2735810 5.20
1972 93053 67379 72.41 9320 13.83 2529173 3135065 123.96 105697 82904 2285066 3.37
1973 - - - - - 2371370 1958748 82.60 96945 94410 2827370 4.95
1974 104680 75547 72.17 9946 13.17 2468137 1946783 78.88 117664 106774 3096046 6.04
1975 112237 79720 71.03 8696 10.91 2584000 - - - - - -
1976 110427 79813 72.28 12139 15.21 2691000 1933294 71.84 129784 111604 4062020 6.71
1977 125164 89088 71.18 14338 16.09 3268000 2116729 64.77 177654 119648 4280568 8.39
1978 114071 80054 70.18 13167 16.45 3315000 1991363 60.07 179951 111088 4225743 9.04
1979 122931 79258 64.47 14920 18.82 3614000 1978296 54.74 220631 152634 6217451 11.15
1980 141028 85336 60.51 14814 17.36 3749000 1941097 51.78 221413 151988 5489004 11.41
1981 149893 95804 63.91 15554 16.24 3839000 2101856 54.75 255963 163370 6120948 12.18
1982 128341 85215 66.40 25865 30.35 3358218 2061775 61.39 208497 142192 6082111 10.11
1983 141434 92938 65.71 28049 30.18 3358218 - - - - - -
1984 171966 113054 65.74 26911 23.80 3392275 - - - - - -
1985 178696 107255 60.02 20224 18.86 3004721 - - - - - -
1986 165637 104435 63.05 19400 18.58 3321566 2086835 62.83 167456 157185 8974302 8.02
1987 185096 111660 60.33 26214 23.48 3428856 2173180 63.38 183798 177969 10779696 8.46
1988 128660 64771 50.34 7715 11.91 3401593 2340044 68.79 188991 156014 11147169 8.08
1989 155925 80900 51.88 18566 22.95 3941589 2431273 61.68 213898 198622 13258789 8.80
1990 122070 78969 71.51 6020 7.62 3293342 2172367 65.96 132596 108545 8879175 6.10
1991 89030 67342 75.64 7846 11.65 3899937 2436067 62.46 141092 110012 11105661 5.79
1992 138907 103157 74.26 13650 13.23 3656245 2055961 56.23 129613 101788 11866481 6.30
1993 - - - - - 3813607 2020591 52.98 142098 119434 14883.73 7.03
1994 84431 63084 74.72 2889 4.58 3767703 1905291 50.57 125716 107902 15467.27 6.60
1995 111742 61216 54.78 3675 6.00 3973776 2059766 51.83 127500 98690 15182.86 6.19
1996 147310 87564 59.44 7723 8.82 5157303 2057470 39.89 154176 125306 17324.97 7.49
1997 86053 58620 68.12 6380 10.88 5472569 2238485 40.90 128179 107441 19317.54 5.73
1998 258440 141930 54.92 8672 6.11 5541409 2136833 38.56 130483 106731 27785.79 6.11
1999 246252 115006 46.70 5378 4.68 5800916 2160467 37.24 115745 106961 17701.02 5.36
2000 88702 50832 57.31 3290 6.47 5536095 2127202 38.42 115092 105387 21547.45 5.41
2001 - - - - - 6023103 2080560 34.54 105943 990939 24199.06 5.09
See table 2.4A for sources and notes. Some excessively high percentage changes may be based on provisional data and may be disregarded.
132
Table 2.4C: Enforcement - Standing Orders (Employment) Act
Year ------------------------------------------------------- Standing Orders (Employment) Act 1946 -------------------------------------------------------------
Number of
establish-
ments
covered
Number of
employees
covered
No. establ
with
certified
orders beg.
of year
No. empl
covered by
certified
orders beg.
of year
Applica-
tions for
certification:
B/F as on
January 1
Applica-
tions for
certification
received
during-year
Applica-
tions for
certification
disposed
during-year
Applica-
tions for
certification
pending end
of year
No. establ
with
certified
orders end
of year
No. empl
covered by
certified
orders end
of year
Percent
application
s disposed
1970 17289 4103107 10936 3024605 1003 261 236 1082 11011 3209336
18.7
1971 18050 4270037 9283 3017326 1028 229 200 1057 9273 2953962
15.9
1972 - - - - - - - - - - -
1973 18745 4384403 10654 3447583 1145 291 230 1206 10702 3423249
16.0
1974 18011 3652803 9529 2711364 921 880 886 922 10399 2817040
49.2
1975 18573 3919894 8869 2626824 923 252 272 909 9124 2711932
23.1
1976 19162 4213827 10507 2703510 895 459 389 965 10884 3165424
28.7
1977 16189 3499811 8453 2295211 751 296 214 833 8584 2311033
20.4
1978 16196 3586615 8495 2303386 833 876 211 1498 8704 2351696
12.3
1979 17871 3308485 8769 2478435 653 283 245 681 8952 2536809
26.2
1980 25455 4231811 10617 3376040 961 171 162 970 11053 3943004
14.3
1981 18265 4309188 11076 3515010 370 237 192 1015 11258 3563515
31.6
1982 17509 3572919 8896 2857817 1015 240 175 1080 9078 2926281
13.9
1983 25650 5423826 14838 4329623 1735 447 677 1505 15161 4431945
31.0
1984 29434 5456601 15323 4681486 1751 594 546 1799 15909 4904807
23.3
1985 - - - - - - - - - - -
1986 - - - - - - - - - - -
1987 40776 5695305 26019 4606395 1829 556 529 1856 26563 4712175
22.2
1988 43992 6107072 26426 4729950 1852 726 658 1920 27089 4789207
25.5
1989 41987 5939797 23763 4763322 1867 628 658 1837 24421 4810413
26.4
1990 43305 5967538 24038 4683810 1833 1771 1509 2095 24646 4762349
41.9
1991 47100 6342735 27082 4966685 2161 738 725 2174 27803 4847477
25.0
1992 42638 5591083 23776 3378160 1334 581 566 1349 24342 4239426
29.6
1993 28410 3976780 13091 3040555 1934 597 455 2076 13291 2854244
18.0
1994 35417 4184413 20625 3074903 1321 469 385 1405 21010 3095465
21.5
1995 30069 3354608 11285 2925828 1957 470 385 2042 11670 2891996
15.9
1996 42566 4999714 11961 2989455 2117 552 562 2107 12503 3049822
21.1
1997 46024 5352919 12823 3201230 2101 530 461 2170 13284 3162004
17.5
1998 42199 5095292 12831 3011426 2070 483 471 2082 13280 3094684
18.4
1999 46115 4808860 12106 2567342 1690 328 354 1664 12460 2613654
17.5
2000 47681 5263181 13019 2822547 1740 444 489 1695 13437 2883562
22.4
2001 68850 6765485 14848 3385791 2068 447 368 2147 15289 3435837
14.6
See table 2.4A for sources and notes
133
Table 2.5: Workers' Adjustment Assistance - National Renewal Fund (NRF)
Year
Budgeted
(Rupees
million)
Released
(Rupees
million)
Total no.
workers
covered
Total no.
workers
covered -
cumulative
No. workers
counseled
via EACs*
No. workers
retrained
via EACs*
No. workers
redeployed
via EACs*
No. workers
covered by
EACs* -
cumulative
1992/93 8297 5667 38626 38626 - - - -
1993/94 10200 4781 31613 70239 - - - -
1994/95 7000 2508 3979 74218 7500 1500 234 9234
1995/96 - 2096 8946 83164 11493 5644 704 27075
1996/97 - 1906 15264 98428 28862 18927 3898 78762
1997/98 2350 - 8733 107161 - - - -
1998/99 - - - - - - - -
1999/2000 - - - - - - - -
2000/01 - - - - - - - -
Sources:
Indian Labour Yearbooks, 1994, 1995, 1996, 1997, 1998, Labour Bureau, Ministry of Labour, Government of India
PRAGYA (1999) "Managing the Negative Impacts of Retrenchments and Layoffs", Ministry of Labour, Government of India
(PRAGYA: 1994/95 data for EACs)
Notes:
* Employee Assistance Centres (EACs): 5 sanctioned in 1994/95 - Bombay, Indore, Ahmedabad, Kanpur, Calcutta
134
Table 2.6: Adjustment Costs - Labor Restructuring
Pragya 1999 TISS Study 2001, Bombay
40 cases** MNC Mill NTC Mills
Continuing-VRS Continuing-VRS Sick -VRS
Workers
No. employed pre-VRS/lockout/dispute 2000+ on average - -
No. displaced in VRS/lockout/dispute - 2000 20000
No. employed post VRS/lockout/dispute
Sample size 190 355 251
VRS - average amount (in Rupees '000) - 500-700 60-250
Worker Dues* held back na na na
Amount - avg./max. Rupees per worker na na na
Time-Years held back na na na
Age: Percent>50 years 24.2 59.5 42.0
Education: Percent>=secondary (8+ yrs) - 72.6 24.8
Assistance
Counseling (Percent of sample) - 21.4 4.7
Re-training (Percent of sample) - 0.0 2.0
Re-deployment help (% of sample) - 0.0 2.0
Post-VRS employment (% of sample) 92.2 26.2 68.6
Of which percent self-employed 57.3 89.0 70.3
Looking for employment (% sample) - 21.7 17.3
Post-VRS avg. earnings (2002 Rs./month)# - 4423 1041
pre-VRS avg. earnings (2002 Rs./month) - - 4486
Percent loss - avg. earnings - - 76.8
Impact on health - reduced expenses (% sample) - 15.7 23.5
Impact on children's education - Neutral Adverse
Percent cases - children drop-out - 1.2 13.8
Percent loss-avg. earnings -children ## - 46.0 46.0
Sources: Reports prepared for the Ministry of Labour, Governmant of India
PRAGYA (1999) "Managing the Negative Impacts of Retrenchments & Layoffs"
Tata Institute of Social Sciences (TISS) (2001), "Life After Voluntary Retirement - Examining
the Human Face of the Structural Adjustment Programme"
Institute for Human Development (IHD) (2004), "Rehabilitation of Cotton Textile Mill
Workers after the Closure of the Mills in Indore"
Notes: na Not Applicable; MNC Multinational company; NTC (STC) National (State) Textile Corp.
* Worker dues may be wages, provident fund, gratuity, and retrenchment or closure compensation
** Covers three industries - textiles, engineering, pharmaceuticals
# Inflation adjusted: Pre- & post-VRS earnings in 1991 rupees; deflation using CPI-industrial workers
## Loss in earnings by comparing average labor cost per manday in ASI-census (proxy for organised
sector) & ASI-sample (proxy-unorganised) in year of adjustment, i.e., 1997 (2002 figure = 45.4%)
135
Table 2.6: Adjustment Costs - Labor Restructuring (contd.)
IHD Study 2004, Indore
NTC - A NTC - B NTC - C
Sick -VRS Sick -VRS Sick -VRS
Workers
No. employed pre-VRS/lockout/dispute 3814 3000 1766
No. displaced in VRS/lockout/dispute 35%, 100% 38%, 100% 100%
No. employed post VRS/lockout/dispute None None None
Sample size 130 82 60
VRS - average amount (in Rupees '000) 125, 200-300 125, 200-300 200-300
Worker Dues* held back na na na
Amount - avg./max. Rupees per worker na na na
Time-Years held back na na na
Age: Percent>50 years 19.2 22.2 60.0
Education: Percent>=secondary (8+ yrs) ------------------------- 52.8% -------------------
Assistance
Counseling (Percent of sample) - - -
Re-training (Percent of sample) Yes Yes Yes
Re-deployment help (% of sample) Yes Yes Yes
Post-VRS employment (% of sample) 51.5 42.0 36.7
Of which percent self-employed 17.9 58.8 27.3
Looking for employment (% sample) 23.9 5.6 14.3
Post-VRS avg. earnings (2002 Rs./month)# 1500 1500 1500
pre-VRS avg. earnings (2002 Rs./month) 3190 3190 4253
Percent loss - avg. earnings 53.0 53.0 64.7
Impact on health - reduced expenses (% sample) --------------------------- 5.7% --------------------
Impact on children's education ------- Adverse, no data reported -----------
Percent cases - children drop-out - - -
Percent loss-avg. earnings -children ## - - -
Sources and Notes: see first page of table 2.6
136
Table 2.6: Adjustment Costs - Labor Restructuring (concld.)
Private - D Private - E STC - F
Lockout Lockout Sick -VRS
Workers
No. employed pre-VRS/lockout/dispute 1100 6000 1600
No. displaced in VRS/lockout/dispute 100% 100% 100%
No. employed post VRS/lockout/dispute 135 reporting None None
Sample size 35 122 59
VRS - average amount (in Rupees '000) None None 6-150
Worker Dues* held back Yes Yes na
Amount - avg./max. Rupees per worker - 90000 max na
Time-Years held back 18 years 12 years na
Age: Percent>50 years 74.3 59.0 30.0
Education: Percent>=secondary (8+ yrs) --------------------- 52.8% -----------
Assistance
Counseling (Percent of sample) None None -
Re-training (Percent of sample) None None Yes
Re-deployment help (% of sample) None None Yes
Post-VRS employment (% of sample) 62.9 82.8 58.3
Of which percent self-employed 22.7 21.8 28.6
Looking for employment (% sample) 25.0 14.3 7.7
Post-VRS avg. earnings (2002 Rs./month)# 1500 1500 1500
pre-VRS avg. earnings (2002 Rs./month) 5293 1595 3190
Percent loss - avg. earnings 71.7 6.0 53.0
Impact on health - reduced expenses (% sample) ----------------------- 5.7% ---------------
Impact on children's education ----- Adverse, no data reported ----
Percent cases - children drop-out - - -
Percent loss-avg. earnings -children ## - - -
Sources and Notes: see first page of table 2.6
137
Table 2.7: Industrial Disputes Act 1947 with Amendments
Date Section Coverage Provision or Institutional Change Flexibility
1920 Trade
Disputes' Act
India Established Courts of Enquiry, Conciliation Boards;
Forbade strikes in public utility services without notice;
but, no machinery for settling of "industrial disputes"
+
1929 Trade
Disputes' Act
India State intervention in settling industrial disputes - provided
conciliation machinery for peaceful settlement
-
Policy of "laissez faire" continued, selective intervention
at best
WW-
II
Defense of
India Rules
(temp)
India Power to appoint industrial tribunals & enforce tribunal
awards
-
1946 Industrial
Employ-ment
(Standing
Order) Act
India Provision for devising, framing, and certifying of
standing orders covering service conditions, and making
these known to workers
-
1947 Chap 1-VIII:
Sections 1-40
India Embodies essential principles of Rule 81A, Defense of
India Rules and certain principles of Trade Disputes Act
1929 about investigation and settlement of industrial
disputes. Seeks to achieve social justice on the basis of
collective bargaining, conciliation, arbitration, or finally
compulsory adjudication. Dismissals, and a decision to
grant or decline permission to retrench are industrial
disputes. Retrenchment means the termination by
employer of the service of a worker for any reason other
than disciplinary action; excludes termination upon
voluntary retirement, superannuation, non-renewal of
a contract, continued ill-health, and cases where the
appointment letter stipulates discharge from service
without notice or reasons.
-
1947 3 India Works committee comprising representatives of
employers and workmen to be constituted in every
industrial establishment employing 100 or more
workers. Industry means any business, trade,
undertaking, manufacture… Member-workers to be
selected in consultation with their trade union. Committee
tasked with promoting measures for securing &
preserving good relations.
-
1947 3A,3B Gujarat Additional: Joint management council instead of works
committee, for every industrial establishment employing
500 or more workers. Duties include promoting
productivity growth and training workers to understand
management responsibilities.
+
1947 3 Maharashtra Additional: Recognized union, if any, shall appoint its
nominees to represent workers
-
138
Table 2.7: Industrial Disputes Act 1947 with Amendments (contd.)
Date Section Coverage Provision or Institutional Change Flexibility
1947 3 Rajasthan Additional: State shall appoint Registrar of Unions for the
whole state and assistant registrar for any local area.
-
1947 7, 7A, 7B India Labor courts, regional and national industrial
tribunals constituted; strikes and lockouts prohibited
pending proceedings in tribunals
-
1953 25A-25J
(Chapter VA)
added
India Influenced by decisions of judiciary, bipartite and
tripartite agreements, and actual implementation of IDA
provisions, substantial modifications made.
-
1953 25A India New addition: Exemption from liability to pay lay-off
compensation or maintaining muster rolls: industrial
establishments with less than 50 workers per working day
in the preceeding month or those of a seasonal nature
+
1953 25B India ?? New addition: Eligibility is that worker's name is
borne on the establishment muster rolls and service is
continuous of at least one year.
-
1953 25C India Newly added: Compensation specified in case of lay-
off (50% of wages and dearness allowance) - 45 days
maximum during any 12 months. Not applicable to
closures. Wages as per section S.2(iv)(d) of the Payment
of Wages Act. Eligibility is that worker's name is borne
on the establishment muster rolls and service is
continuous of at least one year.
-
Immediate provocation for tightening law was imminent
downsizing in (cotton?) textiles mills - closing one shift -
possibly rendering thousands of workers unemployed
with no income.
1953 25D India First time added: Muster rolls entry recording worker's
availability for work at the establishment mandatory for
lay-off compensation, muster rolls to be made available
for entry by the employer.
-
1953 25E India New addition: Provides for exceptions to the general
provision for lay-off compensation: (i) employer offers
alternate employment in the same or any other
estalishment, (ii) worker does not present himself for
work at the appointed time, (iii) lay-off due to strike or
slow-down of production in another part of
establishment.
+
139
Table 2.7: Industrial Disputes Act 1947 with Amendments (contd.)
Date Section Coverage Provision or Institutional Change Flexibility
1953 25F India New addition: Retrenchment compensation to follow a
simple yardstick of length of service; on grounds of
humane public policy and consideration that involuntary
employment may result in general economic insecurity.
Compensation formula: 15 days average pay for every
completed year of continuous service or any part
thereof in excess of six months.
-
1953 25J India Employee rights under a contract cannot be derogated by
reason of any provision in the Act
-
1956 25C India Amended: Layoff compensation to extend beyond 45
days to all days laid off.
-
1956 25FF India Amended and added: To check indiscriminate resort to
layoff & retrenchment via closures, amendment provided
that termination of employment due to closure or
transfer be considered as retrenchment.
-
Immediate cause was a Supreme Court of India ruling
(Hariparshad Shivshanker Shukla vs. A.D. Diwakar, per
S.K. Das J) that denied compensation to a worker
unemployed due to a closure/transfer.
1956 25H India Amended?: Employer to offer an opportunity for re-
employment of retrenched workers if further employment
proposed.
-
1956 25I India Deleted, covered under section 33C: Recovery of dues to
workers may be recovered by the government in the same
manner as arrears of land revenue or as public demand by
govt. on an application to the govt. by the worker/s.
-
1956 33C India Added: recovery of money dues to workers under chap
VA (lay-off and rerenchment), also money due under
settlements or awards
-
1957 25FFF India New addition: If closure due to specified "unavoidable
circumstances" beyond the employer's control, then
retrenchment compensation is reduced to a maximum
of 3 months' average pay of the (eligible) worker.
Unavoidable circumstances refers to financial difficulties
only, or expiry of lease or license on or after April 1,
1967. For construction work, compensation applies only
beyond two years of construction project, as per section
25F.
+
140
Table 2.7: Industrial Disputes Act 1947 with Amendments (concld.)
Date Section Coverage Provision or Institutional Change Flexibility
1964 25B India Amended: Defined "continuous service" for "a period" as
all uninterrupted service for that period, including periods
of interruptions due to sickness, authorised leave,
accident, illegal strike, lockout or cessation of work not
due to the worker; deemed to be a year (six months) if
worker actually worked for 240 (120) days and for
miners 190 (95) days.
-
1964 25FFF India Amended: "completed year of continuous service"
replaced by "completed year of service" for eligibility for
compensation [see 1956, section 25B, and 1957, 25FFF]
-
1964 25FFF India Amended: unavoidable circumstances for closure include
accumulation of undisposed stocks [see 1957, section
25FFF]
+
1964 25H India Amended: Workers to be considered for re-employment
if they are citizens of India.
-
1964 25J India Amended: Provisions of chap. VA prevail over those of
other acts including Standing Orders that are inconsistent
with it. If the inconsistent provision more beneficial to
worker, it'll override chap VA
-
1965 25C India Amended: Compensation specified in case of lay-off -
1971 25FFA West Bengal Employer to give 60 days' notice to state government
(West Bengal) for closure - to enable govt. to take
measures, if required
-
1972 25FFA India Employer to give 60 days' notice to state government for
closure
-
1976 25K-255,
chap VB
India Industrial establishments (as per section 25L) employing
300 or more workers: require "prior permission" to
lay-off or retrench workers, and for closure
-
1982 25K-255??,
(Chapter VB
added)
India Industrial establishments (as per section 25L) requiring
"prior permission" to lay-off or retrench workers, and for
closure: coverage expanded to 100 or more workers
-
Sources:
2005, The Industrial Disputes Act, 1947 with amendments, BARE ACT with short comments,
Professional Book Publishers, New Delhi
2001, G.B. Pai, Labour Law in India, Butterworths India., New Delhi
Discussions with experts: Labour administration officials, Jurists, Lawyers
141
Table 3.1
Summary Statistics
India: Industry (3-digit) Data, 1973-1997
Variable Obs Unit Mean Std. Dev. Min Max
Factories 3400 Number 645 1067 1 10702
Workers 3400 Number 36921 71490 31 783025
Mandays 3400 Thousand 10695 22214 10 250334
Value of output 3400 Rupees
thousand
6772286 14500000 1059 170000000
Wage 3400 Rupees 8706 5240 575 58180
Unit severance pay 3400 Rupees 24292 14140 1090 60867
Unit hiring cost 3400 Rupees 735 362 123 1722
Marginal expected
adjustment cost -
historical memory 3264 Rupees 9416 6667 123 34073
Marginal expected
adjustment cost - five
year memory 2720 Rupees 11548 7759 153 49038
Market share 3400 Percent 14 19 0.002 100
Price Cost Margin 3398 Percent 33 12 1.000 96
Source: Author's dataset (truncated version of Annual Survey of Industries, 1973-1997)
Note: Rupee values in constant 1982/83 terms.
142
Table 3.2
India: Job flows
(Percent of Employment*)
(3-digit Industry Level)
Year
Job creation
rate
Job
destruction
rate
Net
employment
rate
Job
reallocation
rate
Excess job
reallocation
rate
1973
- - - - -
1974 7.4 4.4 3.0 11.8 8.8
1975 7.4 3.5 4.0 10.9 6.9
1976 6.4 2.9 3.5 9.4 5.9
1977 6.4 0.9 5.5 7.3 1.8
1978 4.5 2.4 2.2 6.9 4.8
1979 7.4 1.7 5.7 9.2 3.5
1980 4.7 3.2 1.5 7.8 6.4
1981 4.6 3.6 0.9 8.2 7.3
1982 5.8 2.7 3.1 8.4 5.3
1983 3.2 6.7 -3.6 9.9 6.3
1984 4.4 6.1 -1.7 10.6 8.8
1985 3.5 7.8 -4.3 11.3 7.0
1986 2.9 3.4 -0.5 6.3 5.8
1987 7.9 3.6 4.3 11.5 7.2
1988 4.0 4.4 -0.4 8.5 8.1
1989 8.0 5.5 2.5 13.5 11.0
1990 4.1 4.4 -0.3 8.5 8.2
1991 4.2 3.9 0.3 8.1 7.8
1992 7.2 2.2 5.1 9.4 4.3
1993 3.7 4.0 -0.3 7.7 7.4
1994 5.8 1.8 4.0 7.6 3.6
1995 10.3 1.6 8.7 12.0 3.3
1996 4.0 5.2 -1.2 9.2 8.0
1997 5.1 6.0 -0.8 11.1 10.3
Average(73-97) 5.5 3.8 1.7 9.4 6.6
Variance(73-97) 3.4 2.9 9.4 3.3 4.9
Correlation with net 0.9 -0.8 1.0 0.1 -0.6
Avg(73-75) 7.4 3.9 3.5 11.3 7.9
Var(73-75) 0.0 0.4 0.5 0.4 1.8
Avg(76-82) 5.7 2.5 3.2 8.2 5.0
Var(76-82) 1.3 0.8 3.4 0.8 3.4
Avg(83-90) 4.7 5.3 -0.5 10.0 7.8
Var(83-90) 4.1 2.4 8.2 4.9 2.7
Avg(91-97) 5.8 3.5 2.3 9.3 6.4
Var(91-97) 5.5 2.9 14.0 2.9 7.1
Avg(76-97) 5.4 3.8 1.5 9.2 6.5
Var(76-97) 3.5 3.3 10.4 3.3 5.3
* Employment is averaged over current and previous year.
Source: Author's dataset (truncated version of Annual Survey of Industries, 1973-1997)
143
Table 3.3
Job Flows: Comparisons
JC JD JR NT
Industry - level
US-4I 2.48 3.63 6.11 -1.15
India - 3I 5.5 3.8 9.4 1.7
Establishment - level
Canada 10.9 11.1 22 -0.2
USA 8.8 10.2 19 -1.4
Denmark 16 13.8 29.8 2.2
France 13.9 13.2 27.1 0.7
Norway 7.1 8.4 15.5 -1.3
U.K. 10.2 11.5 21.7 -1.3
Chile 12.9 13.9 26.8 -1
Colombia 12.5 12.2 24.7 0.3
Estonia 9.7 12.9 22.6 -3.2
Morocco 18.6 12.1 30.7 6.5
Sources:
Davis and Haltiwanger (1998)
Author's dataset - India
144
Table 3.4
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1975 1973-1982 1973-1997
With 76-97, 82-97, & 91-97 period dum
With 76-
82
dummy
No ind.
year
dum
Ind.
dum
only
Industry
& year
dummies
Constant 0.777 *** 0.741 *** 0.841 *** 0.942 *** 0.896 ***
Wage -0.102 *** -0.096 *** -0.101 *** -0.106 *** -0.101 ***
Mandays 0.488 *** 0.218 *** 0.202 *** 0.211 *** 0.207 ***
Output 0.064 *** 0.057 *** 0.052 *** 0.048 *** 0.048 ***
Lagged employment 0.960 *** 0.959 *** 0.959 *** 0.962 *** 0.961 ***
Time 0.115 *** -0.004 ** -0.002 ** -0.002 0.001
Average Emplt. Size(>=100) dummy -0.661 *** 0.555 0.352 0.446 0.541
Wage*Size100 0.051 ** 0.027 0.031 0.017 0.009
Mandays*Size100 -0.381 *** 0.010 0.020 0.020 0.031
Output*Size100 -0.031 ** -0.066 ** -0.050 -0.043 -0.043
Lagged employment*Size100 0.011 0.019 0.015 0.008 0.007
Average Emplt. Size(>=300) dummy 6.846 *** 2.761 * 2.497 1.885 1.880
Wage*Size300 -0.650 *** -0.344 ** -0.321 * -0.269 -0.272
Mandays*Size300 1.069 *** 0.205 0.180 0.152 0.150
Output*Size300 0.547 *** 0.379 *** 0.377 *** 0.381 *** 0.380 ***
Lagged employment*Size300 -0.692 *** -0.447 *** -0.441 *** -0.438 *** -0.435 ***
Year76 dummy - -0.533 *** -0.634 *** -0.618 *** -0.563 ***
Wage*Year76 - 0.077 *** 0.083 *** 0.083 *** 0.074 ***
Mandays*Year76 - -0.162 *** -0.155 *** -0.150 *** -0.139 ***
Output*Year76 - -0.045 *** -0.042 *** -0.041 ** -0.039 **
Lagged employment*Year76 - 0.032 ** 0.034 * 0.030 0.031
Average Emplt. Size(>=100)*Year76 - -0.552 -0.219 -0.307 -0.349
Wage*Size100*Year76 - -0.016 -0.012 -0.006 -0.002
Mandays*Size100*Year76 - -0.005 -0.004 -0.018 -0.025
Output*Size100*Year76 - 0.056 0.028 0.025 0.026
Lagged employment*Size100*Year76 - -0.014 -0.007 -0.002 -0.003
Average Emplt. Size(>=300)*Year76 - -3.564 * -3.177 -2.706 -2.822
Wage*Size300*Year76 - 0.455 ** 0.397 * 0.356 0.372 *
Mandays*Size300*Year76 - -0.373 -0.310 -0.267 -0.281
Output*Size300*Year76 - -0.351 *** -0.327 *** -0.329 *** -0.332 ***
Lagged employment*Size300*Year76 - 0.377 *** 0.362 *** 0.360 *** 0.361 ***
Year82 dummy - - 0.000 0.043 dropped
Wage*Year82 - - -0.005 -0.010 -0.011
Mandays*Year82 - - -0.007 -0.004 -0.006
Output*Year82 - - 0.018 0.020 * 0.018
Lagged employment*Year82 - - -0.025 * -0.027 ** -0.026 *
Average Emplt. Size(>=100)*Year82 - - 0.436 0.367 0.201
Wage*Size100*Year82 - - -0.001 0.002 0.014
Mandays*Size100*Year82 - - 0.073 0.045 0.020
Output*Size100*Year82 - - -0.022 -0.016 -0.013
Lagged employment*Size100*Year82 - - 0.001 -0.008 -0.010
Average Emplt. Size(>=300)*Year82 - - -0.807 -0.928 -0.201
Wage*Size300*Year82 - - 0.108 0.128 0.058
Mandays*Size300*Year82 - - 0.441 0.462 0.556 *
Output*Size300*Year82 - - -0.094 * -0.107 * -0.095 *
Lagged employment*Size300*Year82 - - 0.149 ** 0.164 *** 0.151 **
145
Table 3.4
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1975 1973-1982 1973-1997
With 76-97, 82-97, & 91-97 period dum
With 76-
82
dummy
No ind.
year
dum
Ind.
dum
only
Industry
& year
dummies
Year91 dummy - - 0.146 0.224 dropped
Wage*Year91 - - -0.008 -0.014 -0.011
Mandays*Year91 - - 0.050 0.072 0.065
Output*Year91 - - -0.006 -0.004 0.000
Lagged employment*Year91 - - 0.011 0.008 0.004
Average Emplt. Size(>=100)*Year91 - - -0.519 -0.483 -0.304
Wage*Size100*Year91 - - -0.027 -0.038 -0.053
Mandays*Size100*Year91 - - -0.262 * -0.264 * -0.229
Output*Size100*Year91 - - 0.055 0.056 0.054
Lagged employment*Size100*Year91 - - -0.037 -0.032 -0.029
Average Emplt. Size(>=300)*Year91 - - 4.466 ** 4.548 ** 3.633
Wage*Size300*Year91 - - -0.437 -0.455 -0.339
Mandays*Size300*Year91 - - 0.117 0.001 0.012
Output*Size300*Year91 - - 0.082 0.087 0.059
Lagged employment*Size300*Year91 - - -0.132 -0.144 -0.118
Price Cost Margin - - - - -
Lagged emplt*PCM - - - - -
Price Cost Margin*Size100 - - - - -
Lagged emplt*PCM*Size100 - - - - -
Price Cost Margin*Size300 - - - - -
Lagged emplt*PCM*Size300 - - - - -
Pr > chi2 0.000 0.000 0.000 0.000 0.000
No. of observations 272 1224 3264 3264 3264
Notes:
(1) * significant at 10% level, ** significant at 5% level, *** significant at 1% level
(2) Errors robust to heteroscedasticity and autocorrelation
(3) All variables in logs, except time; wage, mandays, output, average size, PCM lagged one period
(4) Industry dummies at 2-digit level
(5) Year dummies from 1979 onward, to capture gradual liberalization process, begun in 1979
146
Table 3.4 (contd.)
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1997
m NO interaction of market variables & period dummies
Market
variables
No dummies
Market variables
(Ind dum)
Market
variables (Ind &
year dum)
Constant 0.776 *** 1.068 *** 1.069 ***
Wage -0.110 *** -0.114 *** -0.110 ***
Mandays 0.200 *** 0.213 *** 0.206 ***
Output 0.058 *** 0.054 *** 0.053 ***
Lagged employment 0.955 *** 0.938 *** 0.933 ***
Time -0.002 ** -0.002 -0.001
Average Emplt. Size(>=100) dummy 2.594 *** 2.382 ** 2.543 **
Wage*Size100 0.044 0.031 0.022
Mandays*Size100 -0.033 -0.049 -0.036
Output*Size100 -0.065 -0.056 -0.056
Lagged employment*Size100 -0.204 ** -0.187 ** -0.195 **
Average Emplt. Size(>=300) dummy -1.489 -2.438 -2.765
Wage*Size300 -0.327 * -0.266 -0.267
Mandays*Size300 0.232 0.200 0.199
Output*Size300 0.389 *** 0.397 *** 0.397 ***
Lagged employment*Size300 -0.075 -0.055 -0.023
Year76 dummy -0.643 *** -0.621 *** -0.548 ***
Wage*Year76 0.082 *** 0.080 *** 0.070 ***
Mandays*Year76 -0.149 *** -0.150 *** -0.134 ***
Output*Year76 -0.041 *** -0.038 *** -0.035 **
Lagged employment*Year76 0.034 * 0.029 0.028
Average Emplt. Size(>=100)*Year76 -0.002 -0.026 -0.093
Wage*Size100*Year76 -0.029 -0.028 -0.021
Mandays*Size100*Year76 0.002 0.010 0.000
Output*Size100*Year76 0.030 0.028 0.027
Lagged employment*Size100*Year76 -0.015 -0.010 -0.009
Average Emplt. Size(>=300)*Year76 -3.423 -2.994 -3.021
Wage*Size300*Year76 0.421 * 0.389 * 0.396 *
Mandays*Size300*Year76 -0.290 -0.271 -0.272
Output*Size300*Year76 -0.306 *** -0.312 *** -0.311 ***
Lagged employment*Size300*Year76 0.340 *** 0.337 *** 0.333 ***
Year82 dummy 0.005 0.043
Wage*Year82 -0.006 -0.012 -0.013
Mandays*Year82 -0.015 -0.015 -0.020
Output*Year82 0.019 0.022 * 0.020 *
Lagged employment*Year82 -0.027 ** -0.030 ** -0.028 **
Average Emplt. Size(>=100)*Year82 0.374 0.352 0.190
Wage*Size100*Year82 0.011 0.013 0.023
Mandays*Size100*Year82 0.101 0.079 0.051
Output*Size100*Year82 -0.023 -0.018 -0.016
Lagged employment*Size100*Year82 0.002 -0.008 -0.009
Average Emplt. Size(>=300)*Year82 -0.394 -0.349 0.286
Wage*Size300*Year82 0.071 0.077 0.015
Mandays*Size300*Year82 0.484 0.537 0.597 *
Output*Size300*Year82 -0.104 * -0.116 ** -0.109 *
Lagged employment*Size300*Year82 0.159 ** 0.175 *** 0.164 ***
147
Table 3.4 (contd.)
GLS Estimates: Quadratic Convex Adjustment Costs
Employment 1973-1997
m NO interaction of market variables & period dummies
Market
variables
No dummies
Market variables
(Ind dum)
Market
variables (Ind &
year dum)
Year91 dummy 0.152 0.223
Wage*Year91 -0.008 -0.014 -0.011
Mandays*Year91 0.048 0.064 0.056
Output*Year91 -0.007 -0.005 -0.001
Lagged employment*Year91 0.011 0.008 0.004
Average Emplt. Size(>=100)*Year91 -0.847 -0.916 -0.706
Wage*Size100*Year91 -0.021 -0.016 -0.032
Mandays*Size100*Year91 -0.358 ** -0.383 ** -0.339 *
Output*Size100*Year91 0.051 0.044 0.042
Lagged employment*Size100*Year91 -0.011 -0.003 -0.001
Average Emplt. Size(>=300)*Year91 4.613 ** 4.662 ** 3.633
Wage*Size300*Year91 -0.464 -0.469 -0.326
Mandays*Size300*Year91 -0.002 -0.066 0.032
Output*Size300*Year91 0.085 0.088 0.051
Lagged employment*Size300*Year91 -0.144 -0.156 -0.121
Price Cost Margin 0.034 -0.012 -0.024
Lagged emplt*PCM -0.001 0.005 0.006
Price Cost Margin*Size100 -0.647 *** -0.577 ** -0.593 ***
Lagged emplt*PCM*Size100 0.064 *** 0.057 ** 0.059 ***
Price Cost Margin*Size300 1.145 *** 1.213 *** 1.301 ***
Lagged emplt*PCM*Size300 -0.106 *** -0.113 *** -0.121 ***
Pr > chi2 0.000 0.000 0.000
No. of observations 3264 3264 3264
Notes:
(1) * significant at 10% level, ** significant at 5% level, *** significant at 1% level
(2) Errors robust to heteroscedasticity and autocorrelation
(3) All variables in logs, except time; wage, mandays, output, average size, PCM lagged one period
(4) Industry dummies at 2-digit level
(5) Year dummies from 1979 onward, to capture gradual liberalization process, begun in 1979
148
Table 3.5
GLS Estimates: Quadratic Convex Adjustment Costs - Parsimonious
Employment 1976-1982 1976-1997
82-97 & 91-97 period dummies
With 76-82
dummy
NO Market
Variables
Market
Variables
Constant 0.276 *** 0.274 *** 0.171
Wage -0.026 *** -0.030 *** -0.032 ***
Mandays 0.073 *** 0.074 *** 0.071 ***
Output 0.013 *** 0.016 *** 0.018 ***
Lagged employment 0.989 *** 0.989 *** 0.994 ***
Average Emplt. Size(>=100) dummy 0.077 ** 0.199 * 0.189
Lagged employment*Size100 -0.006 -0.018 -0.017
Average Emplt. Size(>=300) dummy 0.635 ** 0.507 * 0.528 *
Lagged employment*Size300 -0.051 ** -0.039 * -0.041 *
Year82 dummy - 0.045 0.036
Lagged employment*Year82 - -0.008 -0.009
Average Emplt. Size(>=100)*Year82 - 0.100 0.102
Lagged employment*Size100*Year82 - -0.009 -0.009
Average Emplt. Size(>=300)*Year82 - -0.926 ** -0.942 **
Lagged employment*Size300*Year82 - 0.074 ** 0.075 **
Year91 dummy - dropped 0.015
Lagged employment*Year91 - 0.002 0.002
Average Emplt. Size(>=100)*Year91 - -0.084 -0.091
Lagged employment*Size100*Year91 - 0.010 0.011
Average Emplt. Size(>=300)*Year91 - 0.578 0.586
Lagged employment*Size300*Year91 - -0.050 -0.051
Price Cost Margin - - 0.031
Lagged emplt*PCM - - -0.002
Pr > Chi2 0.000 0.000 0.000
Observations 952 2992 2992
Notes:
(1) Significant at 1% level***, at 5% level**, at 10%level*
(2) Errors robust to heteroscedasticity and autocorrelation, all variables in logs
(3) Year dummies from 1979 onward, to capture gradual liberalization process begun in 1979,
except with period dummy 1991-97, where year dummies until 1990 included
149
Table 3.6
Estimates: Quadratic Convex Adjustment Costs - Parsimonious (System GMM Estimators)
Employment 1976-1982 1976-1997 1976-1997: Market Var.
82-97 & 91-97 period dum 82-97 & 91-97 period dum
With
76-82
dum
Instru-
ment
set 1
Instru-
ment
set 2
Instru-
ment
set 1
Instru-
ment
set 2
Constant -1.165 1.087 0.927 2.093 *** 2.085
Wage 0.142 -0.251 -0.240 -0.448 *** -0.115
Mandays 0.166 0.400 * 0.383 * 0.288 *** 0.283 *
Output 0.040 0.199 0.138 0.333 *** 0.083
Lagged employment 0.960 *** 0.880 *** 0.974 *** 0.780 *** 0.838 *
Average Emplt. Size(>=100) dummy 0.699 5.426 * 4.476 * 5.735 *** 4.859 *
Lagged employment*Size100 -0.075 -0.558 * -0.458 * -0.593 ** -0.502 *
Average Emplt. Size(>=300) dummy -1.525 0.163 2.384 -0.196 2.812
Lagged employment*Size300 0.159 0.083 -0.130 0.124 -0.159
Year82 dummy - 0.703 0.845 * 0.742 0.862 **
Lagged employment*Year82 - 0.008 -0.253 ** -0.003 ** -0.268 ***
Average Emplt. Size(>=100)*Year82 - 6.294 -10.633 ** 4.847 -8.964 **
Lagged employment*Size100*Year82 - -0.612 1.091 ** -0.467 0.928 **
Average Emplt. Size(>=300)*Year82 - -6.840 2.169 -5.011 0.498
Lagged employment*Size300*Year82 - 0.642 -0.329 0.470 ** -0.169
Year91 dummy - dropped dropped dropped dropped
Lagged employment*Year91 - -0.093 ** 0.159 * -0.085 0.170 **
Average Emplt. Size(>=100)*Year91 - -10.441 7.268 -9.134 4.464
Lagged employment*Size100*Year91 - 1.049 -0.741 0.921 -0.461
Average Emplt. Size(>=300)*Year91 - 12.862 2.620 10.149 ** 4.317
Lagged employment*Size300*Year91 - -1.238 -0.137 -0.988 -0.319
Price Cost Margin - - - -0.021 -0.657
Lagged emplt*PCM - - - -0.010 0.061
Hansen Test: Pr>Chi2 0.216 0.285 0.450 0.214 0.586
Arellano-Bond Test: No AR(1) 0.001 0.000 0.000 0.000 0.000
Arellano-Bond Test: No AR(2) 0.807 0.170 0.271 0.098 0.279
Observations 1224 3264 3264 3264 3264
No. of industry groups 136 136 136 136 136
Notes: (1) Significant at 1% level***, at 5% level**, at 10%level*
(2) Errors robust to heteroscedasticity, all variables in logs
(3) Year dummies from 1979 onward, to capture gradual liberalization process, begun in 1979,
except with period dummy 1991-97, where year dummies until 1990 included
(4) Using lagged instruments leads to losing first three years' observations, i.e., 1973-1975; variables
instrumented are the first panel - lagged wage, mandays, ouput, employment- and market variables
where included; with industry groups no. 136 and multiple lagged instruments, the no. of
variables instrumented for is constrained
(5a) Instrument set 1 also instruments for period dummy 1982 and its interactions
(5b) Instrument set 2 also instruments for period dummy 1991 and its interactions
(6) Lags restricted to account for some serial correlation AR(1) in residual disturbance
(7) No. of observations reports all years used including those used only for lagged instruments
150
Table 4.1
Scope of Retrenchment: Number of Workers
New Hires Rehires Separations By Instrument Employment
Vol-soft Invol-soft Invol-hard Reduction Continent
41002 282307 5033875 577188 752809 1178394 4710566 TOTAL - All cases
1743 15900 909705 97878 140896 104035 892062 Africa
19000 3000 255311 172711 11000 0 233311 Asia
0 0 2853559 105000 132000 883259 2853559 Europe
20259 263407 1015300 201599 468913 191100 731634 Latin America
0 0 37500 4800 1800 0 32900 MEDIAN - All cases
0 0 10061 3696 6861 1922 10061 Africa
0 0 23425 14373 0 0 21925 Asia
0 0 172959 17500 0 42000 172959 Europe
0 0 56409 4599 0 0 26000 Latin America
1025 7058 125847 18037 27882 39280 117764 AVERAGE - All cases
116 1060 60647 7529 14090 10404 59471 Africa
2375 375 31914 28785 2200 0 29164 Asia
0 0 407651 26250 33000 176652 407651 Europe
2026 26341 101530 22400 58614 19110 73163 Latin America
0 0 247 0 0 0 0 MINIMUM - All cases
0 0 247 0 0 0 247 Africa
0 0 6500 0 0 0 6500 Asia
0 0 35000 0 0 0 35000 Europe
0 0 2034 0 0 0 0 Latin America
19000 163059 1661000 112000 424095 547300 1661000 MAXIMUM - All cases
1743 7500 547200 59810 75000 57000 541200 Africa
19000 3000 69466 69466 7000 0 69466 Asia
0 0 1661000 70000 132000 547300 1661000 Europe
18100 163059 424095 112000 424095 100000 405995 Latin America
Source: Author's calculations
151
Table 4.2
Scope of Retrenchment: Financial Costs
FINANCIAL BENEFITS (In $ million) FINANCIAL COSTS (In $ million)
Wage bill Total Cost per Safety Enhanced Severance Total Cost
savings-annual worker ($) Net Pension Payments Continent
2834 4206 na 3268 6098.29 2707.89 12074.10 TOTAL - All cases
96 673 na 224.86 20.00 255.17 500.03 Africa
-153 179 na 48.40 70.73 1280.73 1399.86 Asia
1446 1846 na 2995 5588.36 0.00 8583.02 Europe
1445 1508 na 0.00 419.20 1171.99 1591.19 Latin America
10 17 1085 0.00 0.00 12.50 42.04 MEDIAN - All cases
8 10 1476 0.00 0.00 12.50 20.00 Africa
9 9 1040 24.20 0.00 24.87 50.00 Asia
723 923 616 858.51 25.00 0.00 454.26 Europe
222 222 4735 0.00 0.00 112.00 280.00 Latin America
123 168 4142 192.23 243.93 87.35 402.47 AVERAGE - All cases
11 61 3630 32.12 2.00 19.63 38.46 Africa
-22 26 3651 24.20 14.15 182.96 199.98 Asia
723 923 3817 998.22 1397.09 0.00 2145.76 Europe
289 302 5695 0.00 69.87 167.43 265.20 Latin America
-157 -157 0 0.00 0.00 0.00 0.00 MINIMUM - All cases
0 0 320 0.00 0.00 0.00 2.12 Africa
-157 -157 0 0.00 0.00 0.00 0.00 Asia
298 298 24 6.00 0.00 0.00 6.00 Europe
-15 -15 0 0.00 0.00 0.00 2.30 Latin America
1148 1548 17108 2130 5538.36 1140.00 7668.51 MAXIMUM - All cases
30 542 13166 199.80 20.00 80.31 199.80 Africa
83 350 17108 48.40 70.73 1140.00 1188.40 Asia
1148 1548 14012 2130 5538.36 0.00 7668.51 Europe
1000 1063 16000 0.00 418.30 425.00 530.30 Latin America
Source: Author's calculations
Note: Percentages computed from these tables (e.g. severance as percent of total cost) may
not match percentages given in table 5 that are based on a count of programs.
152
Table 4.3
Distribution of Employment Reduction Method
Type: Percent of Programs: Percent of
Workers:
Involuntary (hard -- layoffs) 41.5 47.0
Involuntary (soft -- enforcement of rules, removal
of ghost workers, etc.)
65.0 30.0
Involuntary (removal of ghost workers) 22.5 3.3
Voluntary 77.5 23.0
Used Both Voluntary and Involuntary (either) 42.5 19.2
Used All Methods 17.5 13.4
Source: Author's calculations.
Note: Percent of workers may differ from those computable from Table 1 on account of
missing values that are subsumed in totals in Table 1. Here, missing values are excluded.
153
Table 4. 4
Distribution of Compensation and Transition Assistance
Type: Percent of programs:
Severance Payment 68.3
Pension Enhancement 29.3
No Direct Compensation 14.6
Safety Net 63.4
Safety Net (Training) 53.7
Unknown 12.2
Source: Author's calculations
154
Table 4.5
Distribution of Targeting
Type: Percent of programs:
Skill-biased 53.7
Age-biased 51.2
Neutral 19.5
Unknown 2.4
Source: Author's calculations
155
Table 4.6
Relationship Between Rehiring and Program Characteristics
Program
Characteristic:
Percent of Programs with Rehiring Percent of Other
Programs
Employment Reduction
Method:
Involuntary (hard) 50.0 39.4
Involuntary (soft) 62.5 65.6
Voluntary 62.5 81.3
Used All Methods 12.5 18.8
Nature of Targeting:
Skill-biased 25.0* 60.6
Age-biased 12.5** 60.6
Neutral 62.5** 9.1
Compensation --
Assistance:
Severance Payment 75.0 66.7
Pension Enhancement 37.5 27.3
No Direct Compensation 12.5 15.2
Safety Net 50.0 66.7
Safety Net (Training) 37.5 57.6
Source: Author's calculations. Table entries provide the percent of programs with indicated
column characteristic that also have indicated row characteristic. * indicates that column
difference in row is statistically significant at 10% level; ** indicates that column difference is
statistically significant at 5% level.
156
Table 4.7
Relationship Between New Hires and Program Characteristics
Program
Characteristic:
Percent of Programs with New Hires Percent of Programs
without New Hires
Employment Reduction
Method:
Involuntary (hard) 0.0** 47.2
Involuntary (soft) 80.0 62.9
Voluntary 80.0 77.1
Used All Methods 0.0 20.0
Nature of Targeting:
Skill-biased 40.0 55.6
Age-biased 100** 44.4
Neutral 0.0 22.2
Compensation --
Assistance:
Severance Payment 60.0 69.4
Pension Enhancement 60.0 25.0
No Direct Compensation 0.0 16.7
Safety Net 0.0** 72.2
Safety Net (Training) 0.0** 61.1
Source: Author's calculations. Table entries provide the percent of programs with indicated column
characteristic that also have indicated row characteristic. * indicates that column difference in row
is statistically significant at 10% level; ** indicates that column difference is statistically significant
at 5% level.
157
Table 4.8
Summary Statistics for Financial Indicators
_____________________________________________________________________
Average Median Min Max % programs
With Net Loss
Estimated Financial
Break-even Period 2.3 1.8 0 10 22
Estimated Payback
Period 2.1 0.8 0 10 -
_____________________________________________________________________
Source: Author’s Calculations
158
Table 4.9
Relationship Between Net Losses and Program Characteristics
Program
Characteristic:
Percent of Programs with Net
Losses
Percent of Other Programs
Employment Reduction Method:
Involuntary (hard) 44.4 40.6
Involuntary (soft) 50.0 68.8
Voluntary 87.5 75.0
Used All Methods 25.0 15.6
Nature of Targeting:
Skill-biased 66.7 50.0
Age-biased 88.9** 40.6
Neutral 0.00* 25.0
Compensation --
Assistance:
Severance Payment 66.7 68.8
Pension Enhancement 55.6** 21.9
No Direct Compensation 22.2 12.5
Safety Net 55.6 65.6
Safety Net (Training) 55.6 53.1
Source: Author’s Calculations. Table entries provide the percent of programs with indicated column characteristic
that also have indicated row characteristic. * indicates that column difference in row is statistically significant at
10% level; ** indicates that column difference is statistically significant at 5% level.
159
Table 4.10
Relationship Between Financial Break-even Period (BEP) Program Characteristics
Program
Characteristic:
Percent of Programs with below
Median BEP
Percent of Programs with Above
Median BEP
Employment Reduction Method:
Involuntary (hard) 50.0 33.3
Involuntary (soft) 66.7 70.6
Voluntary 33.3** 94.1
Used All Methods 0.0 23.5
Nature of Targeting:
Skill-biased 0.0** 66.7
Age-biased 16.7** 66.7
Neutral 66.7** 5.6
Compensation --
Assistance:
Severance Payment 66.7 83.3
Pension Enhancement 16.7 38.9
No Direct Compensation 33.3 11.1
Safety Net 66.7 50.0
Safety Net (Training) 50.0 50.0
Source/Notes: Author’s Calculations Table entries provide the percent of programs with indicated column
characteristic that also have indicated row characteristic. * indicates that column difference in row is statistically
significant at 10% level; ** indicates that column difference is statistically significant at 5% level.
160
Table 4.11
Relationship Between Payback Period (PBP) and Program Characteristics
Program
Characteristic:
Percent of Programs with below
Median PBP
Percent of Programs with above
Median PBP
Employment Reduction Method:
Involuntary (hard) 60.0 18.2
Involuntary (soft) 80.0 81.8
Voluntary 20.0** 100
Used All Methods 0.0 18.2
Nature of Targeting:
Skill-biased 20.0 54.5
Age-biased 40.0 36.4
Neutral 40.0 27.3
Compensation --
Assistance:
Severance Payment 60.0** 100.0
Pension Enhancement 40.0 18.2
No Direct Compensation 40.0** 0.0
Safety Net 60.0 45.5
Safety Net (Training) 60.0 36.4
Source/Notes: Author’s Calculations. Table entries provide the percent of programs with indicated column
characteristic that also have indicated row characteristic. * indicates that column difference in row is statistically
significant at 10% level; ** indicates that column difference is statistically significant at 5% level.
161
Table 4.12
Key Characteristics: Selected Retrenchment Programs
Uganda Peru Argentina
Military Civil Service SUNAT Civil Service India Hungary Ghana Railroads Federal govt.
Targeting
Yes Yes Yes No No Yes Yes No No Skill bias
No Yes Yes No Yes Yes Yes Yes Yes Age bias
- - - Yes - - - - - Neutral
Reduction Method
Yes Yes No Yes No Yes No No No Involuntary - hard
Yes Yes Yes Yes No na Yes Yes Yes Involuntary - soft
Yes Yes Yes Yes Yes na Yes Yes No Voluntary
No No No Yes No No No No No Rehires
No No Yes No No No No No Yes New Hires
Financial Indicators
2.7 Net Loss Net Loss 2.6 Net Loss Net loss 1.82 1.56 0.41 BEP
1.2 na 0.0023 na na na 1.66 na na PBP
Productivity Gains
No na Yes na na na No Yes Yes Organizational
(Monetary) (Quantitative) (Quantitative)
Yes na na na na na Yes na na Worker
(Monetary) (Monetary)
Yes Yes No No Yes Yes Yes No No Safety Net Provision
(most workers (most workers
re-employed) re-employed)
Source: Author's calculations
162
Table 4.13
Conditions before and during programs: deviations from country-specific means
Measure: Five Years
before Program
Four Years before
Program
Three Years before
Program
Two Years before
Program
One Year before
Program
During Program
Unemployment rate -0.01 -0.26 -0.05 1.09 0.87 -0.18
Real GDP Growth
Rate**
-0.05 0.82 -0.70 -3.25 -0.41 -2.08
Deficit/GDP
**
0.012 -0.005 -0.008 -0.010 -0.012 -0.034
Domestic
Debt/GDP**
-0.036 -0.049 -0.041 -0.036 -0.063 -0.129
Foreign Debt/GDP -0.050 -0.058 0.026 0.055 0.098 0.149
Government
Spending/GDP**
0.020 0.021 0.021 0.014 0.001 0.007
Source/Notes: Tabulations based upon a variety of sources including the Penn World Tables, World Bank World Tables, the International Finance
Statistics, and the Rama (1995) Cross-Country Labor Market Database. See text for methodology. The deficit numbers are equal to expenditures less
revenues and ignore any grants and/or loans. * indicates that the F-test for the null hypothesis that all of the coefficients in the row are equal is rejected
at the 10 percent level. ** indicates that the F-test is rejected at the 5 percent level.
163
Table 4A.1: Summary Statistics for all Countries
Employment (In Numbers) Continent
By Instrument Employment
Vol-soft Invol-soft Invol-hard Reduction Country/Case NO.
Africa
3300 5721 1040 10061 Benin Civil Service 1
- - - 1585 Burkina Faso Joint Railway Line 2
3696 - 2804 6500 Cameroon Civil Service 3
247 - 0 247 Cape Verde Public Enterprises 4
725 2275 0 1100 Central African Republic Civil Service 5
0 8000 - 8000 Congo Civil Service 6
0 - - 541200 Ethiopia Military 7
59810 14000 0 73810 Ghana Civil Service 8
5800 25000 - 30800 Kenya Civil Service 9
- - - - Malawi Civil Service 10
- - - 1900 Mauritania Public Enterprises 11
0 0 57000 49500 Namibia Military 12
4300 1800 0 4357 Senegal Civil Service 13
6000 1100 20852 27452 Sierra Leone Civil Service 14
5000 75000 11339 91339 Uganda Civil Service 15
9000 8000 11000 44211 Uganda Military 16
Asia
21250 4000 0 22250 Bangladesh Jute Sector Public Enterprises 17
- - - 50000 Cambodia Civil Service 18
0 7000 - 7000 China Shenyang Region Reform 19
69466 0 0 69466 India Public Enterprises 20
- - - 21600 Lao PDR Civil Service 21
7495 0 0 7495 Pakistan Public Enterprises 22
6500 - 0 6500 Pakistan Sindh Region Reform 23
68000 0 0 49000 Sri Lanka Civil Service 24
Europe
0 0 253000 253000 Albania Public Enterprises 25
- - - 1661000 Hungary Public Enterprises 26
- 132000 40959 172959 Kazakhstan Public Enterprises 27
70000 0 42000 112000 Macedonia Public Enterprises 28
0 0 547300 547300 Poland Public Enterprises 29
- - - 72300 Russian Federation Coal Sector 30
35000 - 0 35000 Turkey Public Enterprises 31
Latin America
30000 42818 0 72818 Argentina Public Enterprises 32
0 424095 0 405995 Argentina Federal Administration 33
4599 0 0 4251 Bolivia Mining - public corporation 34
0 0 100000 0 Brazil Civil Service 35
0 0 91100 91100 Chile Civil Service and parastatal organizations 36
12000 0 0 12000 Colombia Tourism and Transport Ministry 37
40000 0 0 40000 Ecuador Civil Service 38
3000 2000 0 4150 Mexico SOCEFI - Ministry of Trade & Industry 39
112000 - 0 100595 Peru Civil Service 40
- - 0 725 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
Note: In several cases, the number of workers separated by instrument do not add up to total
separations owing to partial availability of information.
164
Table 4A.1: Summary Statistics for all Countries (contd.)
Employment (In Numbers) Continent
New Hires Rehires Separations Employment
Reduction Country/Case NO.
Africa
0 0 10061 10061 Benin Civil Service 1
0 0 1585 1585 Burkina Faso Joint Railway Line 2
0 0 6500 6500 Cameroon Civil Service 3
0 0 247 247 Cape Verde Public Enterprises 4
0 1900 3000 1100 Central African Republic Civil Service 5
0 0 8000 8000 Congo Civil Service 6
0 6000 547200 541200 Ethiopia Military 7
0 0 73810 73810 Ghana Civil Service 8
0 0 30800 30800 Kenya Civil Service 9
- - - - Malawi Civil Service 10
0 0 1900 1900 Mauritania Public Enterprises 11
0 7500 57000 49500 Namibia Military 12
1743 0 6100 4357 Senegal Civil Service 13
0 500 27952 27452 Sierra Leone Civil Service 14
0 0 91339 91339 Uganda Civil Service 15
0 0 44211 44211 Uganda Military 16
Asia
0 3000 25250 22250 Bangladesh Jute Sector Public Enterprises 17
0 0 50000 50000 Cambodia Civil Service 18
0 0 7000 7000 China Shenyang Region Reform 19
0 0 69466 69466 India Public Enterprises 20
0 0 21600 21600 Lao PDR Civil Service 21
0 0 7495 7495 Pakistan Public Enterprises 22
0 0 6500 6500 Pakistan Sindh Region Reform 23
19000 0 68000 49000 Sri Lanka Civil Service 24
Europe
0 0 253000 253000 Albania Public Enterprises 25
0 0 1661000 1661000 Hungary Public Enterprises 26
0 0 172959 172959 Kazakhstan Public Enterprises 27
0 0 112000 112000 Macedonia Public Enterprises 28
0 0 547300 547300 Poland Public Enterprises 29
0 0 72300 72300 Russian Federation Coal Sector 30
0 0 35000 35000 Turkey Public Enterprises 31
Latin America
0 0 72818 72818 Argentina Public Enterprises 32
18100 0 424095 405995 Argentina Federal Administration 33
0 348 4599 4251 Bolivia Mining - public corporation 34
0 100000 100000 0 Brazil Civil Service 35
0 0 91100 91100 Chile Civil Service and parastatal organizations 36
0 0 12000 12000 Colombia Tourism and Transport Ministry 37
0 0 40000 40000 Ecuador Civil Service 38
850 0 5000 4150 Mexico SOCEFI - Ministry of Trade & Industry 39
0 163059 263654 100595 Peru Civil Service 40
1309 0 2034 725 Peru SUNAT - Tax collecting Authority 41
Source: Author's Calculations
165
Table 4A.1: Summary Statistics for all Countries (contd.)
FINANCIAL COSTS (In $ million) Continent
Cost per Safety Enhanced Severance Total Cost
(a) worker (USD) Net Pension Payments Country/Case NO.
Africa
6424 0.00 0.00 21.20 21.20 Benin Civil Service 1
- - - - - Burkina Faso Joint Railway Line 2
1997 - 0.00 7.38 7.38 Cameroon Civil Service 3
10260 0.06 0.00 2.47 2.53 Cape Verde Public Enterprises 4
- - - - - Central African Republic Civil Service 5
- - - - - Congo Civil Service 6
(d) 365 199.80 0.00 0.00 199.80 Ethiopia Military 7
700 - 0.00 41.87 41.87 Ghana Civil Service 8
3448 - 20.00 0.00 20.00 Kenya Civil Service 9
- 0.00 0.00 20.00 20.00 Malawi Civil Service 10
(b) 4910 - 0.00 9.33 9.33 Mauritania Public Enterprises 11
(d) 658 25.00 - 12.50 37.50 Namibia Military 12
(b) 13166 0.00 - 80.31 80.31 Senegal Civil Service 13
353 0.00 0.00 2.12 2.12 Sierra Leone Civil Service 14
(c) 320 - - 15.79 15.79 Uganda Civil Service 15
(b) 955 - 0.00 42.20 42.20 Uganda Military 16
Asia
2621 - 0.00 55.70 55.70 Bangladesh Jute Sector Public Enterprises 17
(b) 1000 - 0.00 50.00 50.00 Cambodia Civil Service 18
0 - 0.00 0.00 0.00 China Shenyang Region Reform 19
17108 48.40 - 1140.00 1188.40 India Public Enterprises 20
(b) 470 - - 10.16 10.16 Lao PDR Civil Service 21
3318 - 0.00 24.87 24.87 Pakistan Public Enterprises 22
- - - - - Pakistan Sindh Region Reform 23
1040 0.00 70.73 0.00 70.73 Sri Lanka Civil Service 24
Europe
(d) 24 6.00 0.00 0.00 6.00 Albania Public Enterprises 25
(d) 517 858.51 0.00 0.00 858.51 Hungary Public Enterprises 26
- - - - - Kazakhstan Public Enterprises 27
714 - 50.00 0.00 50.00 Macedonia Public Enterprises 28
(d) 14012 2130.2 5538.36 0.00 7668.51 Poland Public Enterprises 29
- - - - - Russian Federation Coal Sector 30
- - - - - Turkey Public Enterprises 31
Latin America
12000 0.00 0.00 360.00 360.00 Argentina Public Enterprises 32
(d) 1002 0.00 0.00 425.00 425.00 Argentina Federal Administration 33
16000 0.00 0.00 73.59 73.59 Bolivia Mining - public corporation 34
- - - - - Brazil Civil Service 35
0 - - 0.00 - Chile Civil Service and parastatal organizations 36
- - - - - Colombia Tourism and Transport Ministry 37
5000 - 0.00 200.00 200.00 Ecuador Civil Service 38
- - - - - Mexico SOCEFI - Ministry of Trade & Industry 39
4735 0.00 418.30 112.00 530.30 Peru Civil Service 40
(b) 1131 0.00 0.90 1.40 2.30 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
(a): Amount of severance per (voluntarily retrenched) worker.
(b): Average severance amount per (all retrenched) worker.
(c): Avg. sev. using all workers (excluding ghost workers, no.=42000)
166
Table 4A.1: Summary Statistics for all Countries (contd.)
FINANCIAL BENEFITS (In $ million) Continent
Other Wages Wages Wages Wage bill Total
New hires Rehires Separatees savings-annual Country/Case NO.
Africa
- - - - 0 0 Benin Civil Service 1
- - - - - - Burkina Faso Joint Railway Line 2
- - - - - - Cameroon Civil Service 3
- - - - - - Cape Verde Public Enterprises 4
- - - - 8 8 Central African Republic Civil Service 5
- - - - - - Congo Civil Service 6
542 - - - - 542 Ethiopia Military 7
- - - 24 24 24 Ghana Civil Service 8
- - - 6 6 6 Kenya Civil Service 9
- - - 10 10 10 Malawi Civil Service 10
- - - - - - Mauritania Public Enterprises 11
- - - - - 34 Namibia Military 12
- - - - 30 30 Senegal Civil Service 13
- - - 1 1 1 Sierra Leone Civil Service 14
- - - - 0 0 Uganda Civil Service 15
- 0 0 17 17 17 Uganda Military 16
Asia
- 0 - - 18 18 Bangladesh Jute Sector Public Enterprises 17
- 144 - 18 -126 -126 Cambodia Civil Service 18
- 0 0 3 3 3 China Shenyang Region Reform 19
- 0 0 83 83 83 India Public Enterprises 20
- - - - 9 9 Lao PDR Civil Service 21
332 - - 18 18 350 Pakistan Public Enterprises 22
- - - - - - Pakistan Sindh Region Reform 23
- - - - -157 -157 Sri Lanka Civil Service 24
Europe
- - - - - - Albania Public Enterprises 25
- - - - 298 298 Hungary Public Enterprises 26
- - - - - - Kazakhstan Public Enterprises 27
- - - - - - Macedonia Public Enterprises 28
400 - - 1148 1148 1548 Poland Public Enterprises 29
- - - - - - Russian Federation Coal Sector 30
- - - - - - Turkey Public Enterprises 31
Latin America
0 - - 237 237 237 Argentina Public Enterprises 32
63 - - - 1000 1063 Argentina Federal Administration 33
0 - - - - - Bolivia Mining - public corporation 34
0 - - - 0 0 Brazil Civil Service 35
- - - - - - Chile Civil Service and parastatal organizations 36
- - - - - - Colombia Tourism and Transport Ministry 37
- - - - - - Ecuador Civil Service 38
- - - - - - Mexico SOCEFI - Ministry of Trade & Industry 39
- - - - 222 222 Peru Civil Service 40
- 16 0 1 -15 -15 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
167
Table 4A.1: Summary Statistics for all Countries (contd.)
PERFORMANCE INDICATORS Continent
Net econ. Payback Net fin. Break-even
gain(loss) period(In yrs) gain(loss) period(In yrs) Country/Case NO.
Africa
- - -21 - Benin Civil Service 1
- - - - Burkina Faso Joint Railway Line 2
- - - - Cameroon Civil Service 3
- - - - Cape Verde Public Enterprises 4
- - - - Central African Republic Civil Service 5
- - - - Congo Civil Service 6
* 365 0.30 * 342 0.40 Ethiopia Military 7
- 1.66 - 1.82 Ghana Civil Service 8
- 3.02 - 3.60 Kenya Civil Service 9
- - - 2.00 Malawi Civil Service 10
- - - - Mauritania Public Enterprises 11
** 115 0.40 ** 48 1.10 Namibia Military 12
- - -409 - Senegal Civil Service 13
- - - 1.87 Sierra Leone Civil Service 14
- - -16 - Uganda Civil Service 15
- 1.20 - 2.70 Uganda Military 16
Asia
- - - 3.40 Bangladesh Jute Sector Public Enterprises 17
- - -1436 - Cambodia Civil Service 18
93 0.00 29 - China Shenyang Region Reform 19
- - -276 - India Public Enterprises 20
- - - 1.10 Lao PDR Civil Service 21
- - # 501 1.44 Pakistan Public Enterprises 22
- - - 0.00 Pakistan Sindh Region Reform 23
- - -1802 - Sri Lanka Civil Service 24
Europe
- - - - Albania Public Enterprises 25
- - -561 - Hungary Public Enterprises 26
- - - - Kazakhstan Public Enterprises 27
- - - - Macedonia Public Enterprises 28
- - -6121 - Poland Public Enterprises 29
- - - - Russian Federation Coal Sector 30
- - - - Turkey Public Enterprises 31
Latin America
- - - 1.56 Argentina Public Enterprises 32
- - - 0.41 Argentina Federal Administration 33
- ## 10.00 - ## 10.00 Bolivia Mining - public corporation 34
- - - - Brazil Civil Service 35
- - - - Chile Civil Service and parastatal organizations 36
- - - - Colombia Tourism and Transport Ministry 37
- - - - Ecuador Civil Service 38
- - - - Mexico SOCEFI - Ministry of Trade & Industry 39
- - - 2.60 Peru Civil Service 40
- 0.002 -47 - Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
* Including arms' imports savings.
** For three year program duration, defence expenditure savings
reversed thereafter.
# Including privatization proceeds.
168
Table 4A.1: Summary Statistics for all Countries (contd.)
ECON. BENEFITS (In $ mill) ECON..COSTS (In $ mill) Continent
Non-finl (productivity rise Total Non-financial Total
Worker Organization (production loss) Country/Case NO.
Africa
- - - - 21 Benin Civil Service 1
- - - - - Burkina Faso Joint Railway Line 2
- - - - 7 Cameroon Civil Service 3
- - - - 3 Cape Verde Public Enterprises 4
- - 8 - - Central African Republic Civil Service 5
- - - - - Congo Civil Service 6
23 - 565 0 200 Ethiopia Military 7
2 - 26 - 42 Ghana Civil Service 8
1 - 7 - 20 Kenya Civil Service 9
- - 10 - 20 Malawi Civil Service 10
- - 0 - 9 Mauritania Public Enterprises 11
67 0 101 - 38 Namibia Military 12
- - 30 - 80 Senegal Civil Service 13
- - 1 - 2 Sierra Leone Civil Service 14
- - 0 - 16 Uganda Civil Service 15
5 - 23 - 42 Uganda Military 16
Asia
- - 18 - 56 Bangladesh Jute Sector Public Enterprises 17
- - -126 - 50 Cambodia Civil Service 18
- 12 15 0 0 China Shenyang Region Reform 19
- - 83 - 1188 India Public Enterprises 20
- - 9 - 10 Lao PDR Civil Service 21
- - 18 - 25 Pakistan Public Enterprises 22
- - - - - Pakistan Sindh Region Reform 23
- - -157 - 71 Sri Lanka Civil Service 24
Europe
- - 0 - 6 Albania Public Enterprises 25
- - 298 - 859 Hungary Public Enterprises 26
- - - - - Kazakhstan Public Enterprises 27
- - - - 50 Macedonia Public Enterprises 28
- - 1148 - 7669 Poland Public Enterprises 29
- - - - - Russian Federation Coal Sector 30
- - - - - Turkey Public Enterprises 31
Latin America
- quantitative ris 237 - 360 Argentina Public Enterprises 32
- quantitative ris 1000 - 425 Argentina Federal Administration 33
- - - - 74 Bolivia Mining - public corporation 34
0 0 0 - 0 Brazil Civil Service 35
- - - - - Chile Civil Service and parastatal organizations 36
- - - - - Colombia Tourism and Transport Ministry 37
- - - - 200 Ecuador Civil Service 38
- - - - - Mexico SOCEFI - Ministry of Trade & Industry 39
- - 222 - 530 Peru Civil Service 40
- 947 933 0 2 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
169
Table 4A.1: Summary Statistics for all Countries (concld.)
Time Period Continent
To From Country/Case NO.
Africa
1994 1991 Benin Civil Service 1
1994 (Ongoing) Burkina Faso Joint Railway Line 2
1994 1989 Cameroon Civil Service 3
1997 1992 Cape Verde Public Enterprises 4
1990 1987 Central African Republic Civil Service 5
1995(Ongoing) Congo Civil Service 6
1995 1991 Ethiopia Military 7
1992 1987 Ghana Civil Service 8
1997 1994 Kenya Civil Service 9
1994 (Ongoing) Malawi Civil Service 10
1994 1990 Mauritania Public Enterprises 11
1994 1991 Namibia Military 12
1991 1989 Senegal Civil Service 13
1997 1992 Sierra Leone Civil Service 14
1994 1992 Uganda Civil Service 15
1995 1992 Uganda Military 16
Asia
1995 1994 Bangladesh Jute Sector Public Enterprises 17
1995/6 (Proposed) Cambodia Civil Service 18
2002 1994 China Shenyang Region Reform 19
1994 1993 India Public Enterprises 20
1993 1989 Lao PDR Civil Service 21
1993 1991 Pakistan Public Enterprises 22
1997 1993 Pakistan Sindh Region Reform 23
1992 1991 Sri Lanka Civil Service 24
Europe
1992 1992 Albania Public Enterprises 25
1992 1990 Hungary Public Enterprises 26
1993 1993 Kazakhstan Public Enterprises 27
1993 1988 Macedonia Public Enterprises 28
1993 1990 Poland Public Enterprises 29
1996 (proposed) Russian Federation Coal Sector 30
1994 1993 Turkey Public Enterprises 31
Latin America
1994 1991 Argentina Public Enterprises 32
1992 1990 Argentina Federal Administration 33
1994 1991 Bolivia Mining - public corporation 34
1991 1990 Brazil Civil Service 35
1977 1973 Chile Civil Service and parastatal organizations 36
1992 1990 Colombia Tourism and Transport Ministry 37
1994 1992 Ecuador Civil Service 38
1992 1989 Mexico SOCEFI - Ministry of Trade & Industry 39
1993 1991 Peru Civil Service 40
1992 1991 Peru SUNAT - Tax collecting Authority 41
Source: Author's calculations
170
References
Alderman, H., S. Canagarajah and S. Younger, (1994), ”A Comparison of
Ghanian Civil Servants’ Earnings Before and After Retrenchment.” World Bank
Mimeo.
BARE ACT (2005). The Industrial Disputes Act 1947, with amendments,
Professional Book Publishers, New Delhi
Basu, Kaushik, Gary Fields, and Shub Debgupta, (1996), ”Retrenchment, Labor
Laws, and Government Policy: An Analysis with Special Reference to India,” paper
presented at World Bank Conference on Public Sector Retrenchment and Efficient
Compensation Schemes.
Bentolila, Samuel and Bertola, Giuseppe (1990). Firing Costs and Labour
Demand: How Bad is Eurosclerosis?. Review of Economic Studies 57, 381-402.
Bertola, Giuseppe, (1990), ”Job Security, Employment and Wages,” European
Economic Review, 34, 851-886.
Bertola, Giuseppe and Richard Rogerson (1996). Institutions and Labor
Reallocation. NBER Working Paper Series. Working Paper 5828, November 1996.
Blanchard, Olivier J. (1997). Labor Market Flexibility and Aggregate
Employment Volatility: A Comment. Comment on paper by Cabrales and
Hopenhayn (1997). Carnegie-Rochester Conference Series on Public Policy 46,
189-228.
171
Blanchard, Olivier, Simon Commander and Fabrizio Coricelli, (1995),
”Unemployment and Restructuring in Eastern Europe,” in Unemployment,
Restructuring and the Labor Market in Easter Europe and Russia, (Commander and
Coricelli, eds), Washington, D.C.: EDI World Bank.
Blanchard, Olivier and Peter Diamond, (1989), ”The Beveridge Curve,”
Brookings Papers on Economic Activity, 1, 1-60.
Blanchard, Olivier and Peter Diamond, (1990), ”The Cyclical Behavior of
Gross Flows of Workers in the United States,” Brookings Papers on Economic
Activity, 2, 85-155.
Blundell, Richard and Stephen Bond (1998). Initial Conditions and Moment
Restrictions in Dynamic Panel Data Models. Journal of Econometrics 87,115-143.
Blundell, Richard, Stephen Bond, and Frank Windmeijer (2000). Estimation in
Dynamic Panel Data Models: Improving on the Performance of the Standard GMM
Estimator. Advances in Econometrics: Nonstationary Panels, Panel Cointegration
and Dynamic Panels, Volume 15, 53-91.
Caballero, Ricardo J., Eduardo M.R.A.Engel and John Haltiwanger (1997).
Aggregate Employment Dynamics: Building from Microeconomic Evidence.
American Economic Review, 87(1), March 1997.
Caballero, Ricardo J. and Eduardo M.R.A.Engel (1999). Explaining Investment
Dynamics in U.S. Manufacturing: A Generalized (S,s) Approach. Econmetrica
67(4), July 1999.
172
Cabrales, Antonio and Hugo A. Hopenhayn (1997). Labor Market Flexibility
and Aggregate Employment Volatility. Carnegie-Rochester Conference Series on
Public Policy 46, 189-228.
Commander, Simon and Farbrizo Coricelli, eds. (1995), Unemployment,
Restructuring and the Labor Market in Easter Europe and Russia, Washington,
D.C.: EDI World Bank.
Commander, Simon, Janos Kollo, Cecilia Ugaz, and Balacs Vilagi, (1995),
”Hungary” in Unemployment, Restructuring and the Labor Market in Easter Europe
and Russia, (Commander and Coricelli, eds), Washington, D.C.: EDI World Bank.
Cooper, Russell and John Haltiwanger, (1996), ”Evidence on Macroeconomic
Complementarities,” Review of Economics and Statistics, 78, 78-93.
Davis, Steven J. and John Haltiwanger (1992). Gross Job Creation, Gross Job
Destruction, and Employment Reallocation.Quarterly Journal of Economics,
107(3), 819-63.
Davis, Steven J., John Haltiwanger, and Scott Schuh, (1996), Job Creation and
Destruction, Cambridge: MIT Press.
Davis, Steven J. and John Haltiwanger (1998). Gross Job Flows. Handbook of
Labor Economics, edited by Orley Ashenfelter and David Card (1998).
Datt, Ruddar (2000). Lockouts in India - A Case Study of West Bengal, Indian
Institute for Human Development, New Delhi.
173
Datta, R.C., (TISS) (2001). Life After Voluntary Retirement - Examining the
Human Face of the Structural Adjustment Programme, Tata Institute of Social
Sciences, Mumbai.
Debroy, Bibek and P.D. Kaushik, editors (2005). Reforming the Labour Market,
Academic Foundation, New Delhi.
Deshpande, L.K., Alakh Sharma, Anup Karan, Sandip Sarkar (2004).
Liberalisation and Labour, Institute for Human Development, New Delhi
Domowitz, Ian, R. Glenn Hubbard and Bruce Petersen (1988). Market Structure
and Cyclical Fluctuations. The Review of Economics and Statistics. Volume 70
(1), February 1988, 55-66.
Diwan, Ishac, (1993a), ”Public Sector Retrenchment and Efficient Severance
Schemes.” The World Bank, mimeo.
Diwan, Ishac, (1993b), ”Public Sector Retrenchment and Severance Pay: Nine
Propositions,” The World Bank, mimeo.
Fallon, Peter R. and Lucas, Robert E.B. (1991). The Impact of Changes in Job
Security Regulations in India and Zimbabwe. The World Bank Economic Review,
5(3).
Griliches, Zvi (editor), (1992), Output Measurement in the Service Sectors.
NBER Studies in Income and Wealth, Volume 56. University of Chicago Press.
174
Hall, Robert E., (1995), ”Lost Jobs,” Brookings Papers on Economic Activity,
1, 221-256.
Hamermesh, Daniel (1992). A General Model of Dynamic Labor Demand.
Review of Economic Statistics, 74(4), November 1992.
Hamermesh, Daniel (1995). Labor Demand and the Source of Adjustment
Costs. The Economic Journal, 105 (May), 620-634.
Hamermesh, Daniel and Gerard A. Pfann (1996). Adjustment Costs in Factor
Demand. Journal of Economic Literature, Vol. XXXIV (September), 1264-1292.
Haltiwanger, John and Manisha Singh (1999). Cross-Country Evidence on
Public Sector Retrenchment. The World Bank Economic Review, Vol.13 No.1,
May 1999.
Hopenhayn, Hugo A. and Richard Rogerson (1993). Job Turnover. Journal of
Political Economy, 101(5).
Indian Labour Yearbooks, Labor Bureau, Ministry of Labor, Shimla, various
issues.
Institute for Human Development (IHD) (2004). Rehabilitation of Cotton
Textile Mill Workers after the Closure of the Mills in Indore, New Delhi.
Lindauer, David and Barbara Nunberg (editors), (1994), Rehabilitating
Government: Pay and Employment Reform in Africa, Washington, D.C.: World
Bank.
175
Lazear, Edward, (1990), ”Job Security Provisions and Employment,” Quarterly
Journal of Economics, 105, 699-726.
Levy, Anat and Richard Mclean, (1996), ”Optimal and Sub-Optimal
Retrenchment Schemes:: An Analytical Framework,” paper presented at World
Bank Conference on Public Sector Retrenchment and Efficient Compensation
Schemes.
Mazumdar, Dipak, (1989), ”Microeconomic Issues of Labor Markets in
Developing Countries: Analysis and Policy Implications,” EDI Seminar Paper No.
40, World Bank.
Mills, Bradford, David E. Sahn, Edward E. Walden, and Stephen D. Younger,
(1993), ”Public Finance and Employment: An Analysis of Public Sector
Retrenchment Programs in Ghana and Guinea,” mimeo.
National Labour Commission (2002). Report of the National Commission on
Labour: 2002, New Delhi
Nickell, Stephen (1978). Fixed Costs, Employment and Labour Demand over
the Cycle. Economica, 45(180), November 1978, 329-45.
OECD, (1993), Employment Outlook, July, Paris.
OECD, (1996), Job Creation and Loss: Analysis, Policy and Data Development,
Paris.
176
PA Consulting Group, 1993. Social Safety Net Adjustment Credit Technical
Assistance - Final Report. Submitted to the World Bank.
Papola, T.S. (1994). Structural Adjustment, Labor Market Flexibility and
Employment. Indian Journal of Labor Economics, 37(1), 1994.
Pai, G.B. (2001). Labour Law in India, Butterworths India., New Delhi.
Pfann, Gerard A. and Franz C. Palm (1993). Asymmetric Adjustment Costs in
Non-linear labor Demand Models for the Netherlands and U.K. Manufacturing
Sectors. The Review of Economic Studies, 60(2), 397-412.
Planning Commission (2001). Report of the Task Force on Employment
Opportunities, New Delhi.
Planning Commission (2002). Report of the Steering Committee on Labour and
Employment for the 10th Five Year Plan(2002-07), New Delhi.
PRAGYA (1999). Managing the Negative Impacts of Retrenchments &
Layoffs, Sushant Lok, Gurgaon.
Rama, Martin, (1994), ”Flexibility in Sri Lanka’s Labor Market,” Policy
Research Working Paper 1262, The World Bank, March.
Rama, Martin, (1995), ”Labor Market Cross-Country Database – Glossary,”
mimeo.
Rama, Martin, (1997), ”Efficient Public Sector Downsizing,” mimeo.
177
Rama, Martin and Donna Macisaac (1996), ”Activity, Earnings and Welfare
after Retrenchment: Central Bank Employees in Ecuador,” paper presented at
World Bank Conference on Public Sector Retrenchment and Efficient
Compensation Schemes.
Republic of Uganda, 1994. Management of Change, Context, Vision,
Objectives, Strategy and Plan: Civil Service Reform.
Robbins, Donald, (1996), ”Public Sector Retrenchment: A Case Study of
Argentina,” paper presented at World Bank Conference on Public Sector
Retrenchment and Efficient Compensation Schemes.
Roberts, Mark J. and Tybout, James R. (1996). Industrial Evolution in
Developing Countries. A World Bank Book. Oxford University Press.
Shea, John (1997). Instrument Relevance in Multivariate Linear Modela: A
Simple Measure. The Review of Economics and Statistics, Vol. 79, No.2, 348-352.
Svejnar, Jan and Katherine Terrell, (1991), ”Reducing Labor Redundancy in
State-Owned Enterprises,” PRE Working Paper No. 792, World Bank, October.
World Bank, 1994. Peru Public Expenditure Review. Report No. 13190-PE.
World Bank, 1994. Shenyang Industrial Reform Project Appraisal Report.
Report No. 12415-CHA.
178
World Bank, 1995. Argentina Public Sector Reform Project Completion Report.
Report No. 14781.
World Bank, 1995. Public Enterprise Adjustment Loan Audit Report. Report
No. 14809.
World Bank, 1995. From Swords to Ploughshares: The Uganda Demobilization
and Reintegration of Ex-combatants in Uganda. Confidential Mimeo.
179
doc_185905729.pdf