Description
This paper investigates the extent and the determ inants of changes in labour market regulation in a large sample of developing and developed countries since 1960. The existing evidence tends to focus on the level of regulation (not on changes) and is often confined to the post11995 period (when not, it focuses on OECD or Latin America.) Our results for the 1995 cross section replicate previous findings: legal origins matter more than per capita GDP or institutional factors.
The Dynamics of the Regulation of Labour
in Developing and Developed Countries since 1960*
Nauro F. Campos
Brunel University, CEPR and IZA
[email protected]
Jeffrey B. Nugent
University of Southern California and IZA
[email protected]
This draft: October 2010
Abstract: This paper investigates the extent and the determinants of changes in
labour market regulation in a large sample of developing and developed countries
since 1960. The existing evidence tends to focus on the level of regulation (not on
changes) and is often confined to the post-1995 period (when not, it focuses on
OECD or Latin America.) Our results for the 1995 cross-section replicate previous
findings: legal origins matter more than per capita GDP or institutional factors.
Yet, when we consider changes in the level of labor regulation from 1960 to 2004,
the effect of legal origins diminishes greatly, while per capita GDP and other
structural reforms move to center stage. Our results suggest labour market reforms
are associated with higher levels of per capita GDP and of financial liberalization,
while trade liberalization tends to slow down labour market reforms.
Keywords: employment protection legislation, labour market institutions, reforms
* The authors are grateful for comments received on earlier drafts from Olivier Bargain,
Armando Barrientos, Andrea Bassanini, Tito Boeri, Suzana Brixiova, Ralitza Dimova, Werner
Eichhorst, Ioannis Ganoulis, Robert Holzmann, Prudence Kwenda, Giovanna Labartin, Hartmut
Lehmann, Joaquim Oliveira-Martins, Carmen Pages, Giovani Peri, Stefano Scarpetta, Kunal Sen,
Zahra Siddique, Jan Svejnar, Katherine Terrell and participants at fRD “Tracking Structural
Reforms” meetings (Milan), European Commission (Brussels), Manchester University, IZA
Topic Week on “Political Economy of Labour Market Reform in Transition and Emerging
Economies” (Bonn) and World Bank/IZA “Employment and Development” conference (Cape
Town). Jacob Burnett, Kannika Damrongplasit, Rob Lee, Ladan Masoudie, and James Ng
provided excellent research assistance. All remaining errors are our own.
1
I. Introduction
In recent years, an enormous literature has arisen concerning the construction of various
alternative indicators of labor market characteristics, most often concerning rigidity or
flexibility. These indexes have been based on various different kinds of measures: (1) measures
of actual activity, such as the extent of labor turnover, the number of strikes, labor force
participation rates, unemployment rates, (2) subjective opinion surveys of employers, workers or
other parties concerning job satisfaction, the rigidity of labor markets, the competiveness of
labor markets, and discrimination in labor markets, (3) tax wedges (distortions measured in
terms of the gap between what workers receive and employers pay) and (4) codified
characterizations of what the labor laws and other labor market regulations say with respect to
different kinds of constraints on employers, workers or intermediaries in the market. Within this
latter class of measures, some apply to individual worker rights, while others to collective rights
(such as the right to form and join a union) and in both cases, distinguishing between the laws
themselves and the practice or enforcement of these laws.
1
Each approach has advantages and
disadvantages.
The present paper attempts to extend the individual rights and law-based component of the
third approach in several respects, namely, by adding countries and more importantly extending
it backwards in time from the late 1990s for some countries to the early 1950s. This extension
allows us to study the dynamics of labor market reform across a fairly large number of both
developing and developed countries. Specifically, it does so for a single relatively
comprehensive measure of labor market rigidity based on comparisons of labor laws across
countries and over time. This index is an Employment Laws Rigidity index (ELR) intentionally
designed to be consistent with the seminal study of Botero et al (2004) for 85 countries in 1997.
1
See for example Kucera (2002) and Greenhill, Mosley and Prakash (2009).
2
Quite a few studies have attempted to up-date similar indexes for large samples of countries
(e.g., World Bank's Doing Business project, the EU’s LABREF). For going backwards in time,
however, studies have been limited to two regions (OECD as in Blanchard and Wolfers 2000 and
Allard 2005) and Latin America (Heckman and Pages 2004).
2
The index we develop is for
Employment Law Rigidity (ELR), a (de jure) index. Our extension increases the number of
countries to 145 for at least one time period and to approximately 100 countries for our panel
data analysis (5-year averages from 1950-54 to 2000-04.) The resulting dataset is referred to as
the Labor Market Rigidities (LAMRIG) dataset.
The largest part of the paper is Section II which describes how these indexes are constructed,
which varies somewhat by region and country but always in a way consistent with the Botero et
al study. This data construction exercise was made possible in large part by the on-line
availability of comprehensive databases of labour laws, such as NATLEX and LEXADIN.
3
In
Section III we make use of LAMRIG to describe the changes over time in the ELR indexes for
specific regions and countries. We use seven countries to illustrate both how the changes in the
ELR indexes have been calculated and to identify possible political economy determinants of
these changes.
Section IV is devoted to several exercises aimed at assessing the applicability of the index.
In order to assess the usefulness of this new index we conduct several exercises. First, we restrict
our analysis to the cross-section for 1995-1999 (the period coinciding with that in Botero et al.
2004) and repeat their analysis concerning the determinants of labor market regulatory rigidities.
2
In the case of collective relations laws, Greenhill, Mosley and Prakash (2009) have extrapolated the series put
together by Kucera (2002) back to 1986 for a sample of 90 developing countries.
3
One of our main sources was NATLEX, available at http://www.ilo.org/dyn/natlex/natlex_browse.home. It is
maintained by the International Labour Organization (ILO)'s International Labour Standards Department and
contains labour law records for more than 150 countries since the late 1940s. LEXADIN is a law database covering
various subjects (such as labor laws but also including electoral, commercial, civil, banking, etc) and is available at
http://www.lexadin.nl/wlg/legis/nofr/legis.php
3
For this cross-section, we fully replicate their results, demonstrating the greater importance of
legal origins than those of per capita GDP and/or political factors. Second, however, when we
extend the analysis to the panel and to changes over time (defining reforms as negative changes
in the ELRs we are thus able to examine the determinants of labor market reforms), our results
diverge from those of Botero et al. (2004). For example, when we use a random-effects model
with clustered standard-errors at the country level to explain labor market reform, the influence
of legal origins is much less significant, but still present. To the aforementioned models of ELRs
and reforms in ELRs that included the legal tradition measures, income and political variables,
we add some measures of factors suggested by our country case studies. These include measures
of economic structure, such as the share of agriculture in GDP, the share of natural resource
exports in GDP and income inequality, various measures of other types of economic reform and
market development, and finally both economic and political crises. Third, we test for the
relevance of other determinants of labor market reforms, such as economic and political crises,
structural factors and other structural reforms. Along with the reduced role of legal origins, we
find evidence suggesting that inertia, per capita GDP and other types of reforms all matter. In
particular, our results suggest that financial liberalization supports labor market reform, while
trade liberalization seems to slow it down.
The paper is concluded with Section V which contains our conclusions and suggestions for
future research. A detailed appendix of data sources and further details on the construction of the
ELR indexes contained in LAMRIG is available upon request as well as on-line.
II. Constructing the ELR Indexes Across Countries and over Time
The purpose of this section is to describe the methods used in constructing the Labor Market
Rigidities (LAMRIG) dataset. To that end it identifies the data sources for constructing an index
4
of rigidities “Employment Law Restriction (ELR) Index” that is consistent across countries and
over time. We begin with a brief overview and then go on to provide more details for the OECD
and Latin American regions for which take advantage of existing studies. This is followed by a
less precise but illustrative discussion of how the ELR indexes are constructed for countries
outside these regions.
A. Overview
To study labor market reforms, either determinants or effects, one needs time series data
on some kind of a labor market index that one thinks is relevant to the issues in which one is
interested, such as unemployment (duration or level), labor turnover, dualism, growth, or
structural change. Yet, because there is no single data set of any kind that covers the more than
100 countries studied here on a consistent basis for anything more than a few years
4
, the present
study makes use of data from several different major sources and many other country-specific
special sources for countries and years not covered by the major sources.
4
One labor index that may be considered close to what one might want in this respect is the one by Forteza and
Rama (2006) and Rama and Artecona (2002) based on ILO conventions signed by each country. This has good
coverage (more than one hundred countries and over time. But, since this index bases much on country’s having
approved of various ILO conventions on such social issues as non-discrimination in employment that are often not
adhered to in practice, few seem to have been persuaded that this is a useful index for examining unemployment and
other issues. It also has the disadvantage of having almost no variation over time once these conventions had been
signed by the individual country (which in many cases was quite early). With respect to the rules governing unions
and collective bargaining, Kucera (2002) takes advantage of data from a number of other sources such as the
International Confederation of Free Trade Unions, the US State Department’s Country Reports on Human Rights
Practices, and ILO Reports of the Committee on Freedom of Association to construct a Freedom of Association
index based on 37 different indicators. The individual indicators are then weighted by their assumed relative
importance and then aggregated the weighted sum into an index representing the average of such scores for the
period 1993-7. Greenhill et al (2009) distinguish between the indicators which pertain to the laws and those which
pertain to practice (captured by the reported numbers of law violations) and then extend the two sets of indexes
backward to 1986 and forward to 2002 for developing countries. Neither of these indexes is utilized here since the
rights with respect to creation and participation in unions need not necessarily relate to flexibility/rigidity in the use
of labor. Interestingly, however, these authors use bilateral trade patterns to show that the level of these freedom to
organize indexes (especially the laws-oriented) can be linked to the extent to which the individual developing
country exports to developed countries with higher labor standards. Another source for measuring the degree of
regulation of labor markets is now Canada’s Fraser Institute. Since 1975, this institute had been scoring countries
on a number of sub-indicators of economic freedom, such as strength of property rights, freedom from price and
wage controls, restrictions on trade, financial transactions and product markets. These were then aggregated into an
overall index of Economic Freedom. In 2001 the Institute began to include scores on six additional subcomponents,
all relevant to measuring the freedom of labor markets. While at first this was limited to 58 countries, the country
coverage has grown somewhat over time.
5
To our knowledge, there are only a very few labour laws rigidity indexes relevant to
unemployment, the size of the informal sector and related phenomena that have reasonable
cross-country and over time coverage going back from the present to the late 1980s or beyond.
Aside from the Forteza and Rama index if ILO Conventions, almost all of these, e.g., Blanchard
and Wolfers (2001), OECD (2004), Allard (2005a) do so only for developed countries. These
studies built upon a whole series of earlier attempts (e.g., Lazear 1990, Grubb and Wells 1993,
Addison and Gosso 1996, Nickell 1997, Layard and Nickell 1998 and OECD 1999) to construct
such an index for developed countries. The Blanchard and Wolfers (2001) study constructs a
series “NEWEP” for 26 OECD countries going back from the 1995-99 period (more exactly
about 1996) to the 1960s in five year intervals.
5
We have used that one in the past but for greater
consistency over time, availability on an annual basis and slightly longer backward coverage, for
the 21 countries she has studied we have now switched to the estimates of Employment
Protection Legislation (EPL) of Allard (2005a). They are in principle comparable to those of
OECD (2004) but exclude two minor subcomponents ( delay in the notification and
compensation for unfair dismissal), for which information could rarely be found in the
legislation. For the remaining five OECD countries not covered by Allard but covered by
Blanchard and Wolfers, namely,Iceland, Korea, Luxemburg, Mexico and Turkey, we have based
on our indexes on Blanchard and Wolfers (2001), though in the case of Korea also on various
other sources as discussed below.
The other multi-country sources with some time coverage as well are Heckman and
5
Nickell et al (2003) have annualized the Blanchard and Wolfers series. More recently, the European Union (EU)
has constructed a somewhat similar set of indexes called the Labor Market Reform Database (LABREF) with
somewhat more detail on certain policy-related aspects of labor legislation, but only for each year between 2000 and
2006 for each EU member. These labor market reform indexes include pension, labor taxation and other aspects.
Both Arpaia et al (2007) and Bassanini and Venn (2007) describe the indexes and relate them to different effects on
labor. Arpaia (2007) focuses on the effects of the indexes on labor market participation (of all workers but especially
of older ones) while Bassanini and Venn (2007) examine the effects of the indexes on labor productivity.
6
Pages (2000 and 2004) which cover most countries of the Latin America and Caribbean (LAC)
region, going back from the late 1990s only to the late 1980s. For the most part, the LAC
indexes are available primarily only at intervals a decade apart, not annually. For this reason, we
extend these indexes backwards and forwards with use of information on the labor laws and
other studies where available.
As has been noted in many surveys, e.g., Bertola (2008), Freeman (2008), Djankov and
Ramalho (2008), the data on countries outside of these two regions is much more limited in time
coverage. Indeed, for them we had to base our data construction work on a wide variety of
sources but with methods designed to be as consistent as possible with those used for the OECD
and LAC regions. We will come to these regions after describing how the OECD and LAC
regions are dealt with in greater detail.
Even for the OECD and LAC data sets, their comparability is made more difficult by the
fact that, although similar in spirit, the Heckman and Pages (2000 and 2004) Job Security Index
and Allard (EPL) are built up from sources, methods and index aggregation procedures that are
by no means identical.
The Heckman and Pages Job Security Index (JS) is defined as the discounted value of
dismissing a worker at an expected date in the future based on the likelihood and costs of
dismissal implied by the labor laws and regulations (but excluding the costs of court actions). It
makes use of a common discount rate of 8 percent, an assumed turnover rate of 12 percent
and
the country and period-specific cost (inclusive of those related to seniority) of dismissing a
worker for either justified or unjustified reasons.
As indicated above, Allard (2005a) made use of 16 of the original 18 aspects of EPL used
in OECD (2004) but obtained the data, not from questionnaires cross-checked with the
individual countries as in the OECD study, but rather from direct examination of the laws
7
themselves based on ILO’s NATLEX supplemented with OECD sources. The 16 indicators were
aggregated first into three separate indicators (laws protecting workers with regular contracts,
those affecting workers with fixed term (temporary) and regulations applying to dismissals), and
then into a single EPL index. Both the scoring of the sub-components and their weighting into
the various components and indexes have been controversial since virtually any method is
subjective.
6
While the scales of the indexes (the EPL of Allard and NEWEP of Blanchard) are
almost identical ranging from 0-4, that of EP of Heckman and Pages) is quite different, the latter
ranging from 0-18. Unfortunately, none of these indexes reflects by any means all of the labor
market institutions (such as wage flexibility, team production, job rotation, social dialogue,
pension plans of different types, and workers use of the courts) that one might think would
exercise influence on economic outcomes of various sorts (Freeman 2008).
7
Yet, as indicated
both above and further below, each of them captures a number of important (largely common)
dimensions of labor regulations and thus may be regarded as a measure of the restrictiveness of
labor laws and regulations for firms in their use of labor. Since both also allow internally
consistent comparisons over time, we deem it valuable to make use of them together.
To mitigate the problem of the differences between the Allard-Blanchard and Wolfers
and Heckman-Pages indexes, we make use of another source, namely Djankov et al (2003)
revised as Botero et al (2004), that uses a closely related methodology for constructing an
employment laws restrictiveness index (ELR). Their ELR has much greater country coverage
(originally 85 countries) but only for a single time period 1997. This index is constructed as the
total of the scores on three different subcomponents, alternative employment contracts (part-
6
Indeed, as shown by Addison and Teixeira (2003), the various variants of the aggregate indices that have arisen are
not always highly correlated and their application to issues like unemployment rates has sometimes resulted in
opposite findings. These and other authors also point out that what is relevant in constructing these indices may also
vary from industry to industry.
7
Allard (2005b) creates for the same 21 OECD countries in her 2005a EPL indexes of unemployment benefits
based in part on tax treatment and subsidies with duration and the conditions for qualification.
8
time, fixed term etc.), conditions of employment (mandatory rest, maximum hours of work
without overtime, overtime pay premium, leaves for holidays and maternity, etc.) and job
security (restrictions on dismissal, mandatory notice periods, severance payments). In this
respect the ELRs of Botero et al are somewhat broader in scope than the other studies. Because it
has scored each of 85 countries, including almost all of the 21 countries for which we have used
Allard’s ELR,
8
3 of the 5 other countries for which we have used NEWEP from Blanchard and
Wolfers and the 21 LAC countries
9
covered by Heckman and Pages (2000, 2004) for the year
1997, we use the ELR of Botero et al (2004) in order to provide consistency between the indexes
coming from each of the other sources in the LAMRIG database.
10
A major advantage of this
index is that it comes with a complete matrix of how the authors scored each country on each of
the 31 subindicators on which their ELRs are constructed. These details facilitate matching of
their evaluations for the year 1997 with information concerning changes in the laws over time
from NATLEX and other sources. The fact that the studies with over time coverage on JS and
EP do not include our indicator for the working conditions component of the Botero et al ELR
indexes is not crucial since these typically change less frequently over time than other
components of the labor laws.
Each of the three subcomponent scores in ELR is in turn an average of scores on a 0-1
8
These countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland,
Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and
United States.
9
The LAC countries covered by Heckman and Pages include: Argentina, Barbados, Belize, Bolivia, Brazil, Chile,
Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guyana, Honduras, Jamaica, Nicaragua,
Panama, Paraguay, Peru, Trinidad and Tobago, Uruguay, and Venezuela. They also included Mexico but for this
country we have used Blanchard and Wolfers because of its more complete time coverage.
10
Another reason for choosing the broader ELR index of Botero et al (2004) as the lynch-pin for our construction
of LAMRIG is that these authors have shown it to have important consequences for labor market outcomes, perhaps
stronger ones than have been obtained by other authors. For example, in their recent recapitulation of their results
and attempt to rebut the criticisms raised of their study, La Porta et al (2008) have shown that a two standard
deviation increase in ELR implies a 1.99 percentage point reduction in the male labor force participation rate, a 2.32
percentage point increase in the unemployment rate, and a 5.67 percentage point increase in the unemployment rate
of young males.
9
scale on the list of indicators, many of which are “yes” –“no” answers as to whether a certain
regulation exists or not. Hence, the maximum score is 3 and the minimum 0. Moreover, the 85
country sample of Botero et al (2004) has subsequently been extended in the closely related
Rigidity of Employment (ROE) Indexes developed in the World Bank’s Doing Business Surveys
so that, for subsequent years 2003 and 2007 at least, an additional forty or more countries can
potentially be added to the sample. The ROE indexes are, like the Botero et al (2004) ELR
indexes, based on three subcomponents, namely, difficulty of hiring, rigidity of hours, and
difficulty of firing. Each of these depends on most of the same indicators as in Botero et al
(2004). In this case, however, each of the subcomponents scores is on a 0-100 scale and since the
aggregate ROE index is an average of these, it, too, is on a 0-100 scale. Hence, while
methodologically and substantively the two indexes are almost identical, their scales are
different: (0-100 for ROE instead of 0 – 3 for ELR). As described in greater detail below, the
difference in scale was overcome by establishing an average conversion factor between each of
these Doing Business publications and numbers from the Botero study. The Allard (1995a) EPL
scores use a 0-4 scale and hence also have to be converted to Botero’s 0-3 scale and normalized
so as to be consistent with the Botero figures for 1995-9.
Both the country-specific Heckman and Pages Job Security Indexes and the Blanchard-
Wolfers NEWEP indexes are then converted into indexes with bases 1995-9 = 1.0. Using the
NATLEX data, the over-time variations in these indexes are then applied to the country-specific
1995-9 values in the Botero et al (2004) to construct over time variations in the country-specific
ELR indexes for the countries included in the OECD and LAC samples. Since they were roughly
on the same scale, the Allard indices were left as they were but aggregated from annual scores to
their five-year averages. A similar procedure is applied to the more fragmentary evidence of
over-time changes in the relevant components of labor laws for the remaining countries in the
10
samples afforded by the Botero et al (2004) and subsequent Doing Business Surveys. Below we
describe how these indices were also updated to 2000-4.
While others may wish to keep the various sub-indexes separate for use in different kinds
of application, for the present purposes we keep the focus on a single broad indicator of the
restrictiveness of employment laws. To avoid getting bogged down in the various possible
weighting systems, moreover, we simply accept the equal weighting of the three sub-components
as suggested by Botero et al. (2004).
The end result is an incomplete panel of Employment Laws Restrictiveness (ELR)
indicators for well over 100 countries measured as 5-year averages ranging from 1950-54
through 2000-04, a dataset we call LAMRIG for Labor Market Rigidities. For some years there
are as many as 145 countries with ELR scores. As has been pointed out by quite a few analysts
(e.g., Eichhorst et al 2007, Freeman 2005, 2008), whether higher scores are looked as desirable
or undesirable is subjective. For example, employers associations and individual employers
typically view them as harmful to investment, employment, productivity and the long run
interests of workers. But, those supporting labor interests and those interested in “public
welfare” and human rights often see them as good, helping to increase the legitimacy of working
outside the home for individual workers and thereby creating larger and better organized labor
markets. Others (Boeri et al (2000), Nicoletti et al (2000) and Amable et al. (2007)) view the
“goodness” or “badness” of such indexes to be more complex, depending on the identity and
magnitude of other market imperfections, regulations and so on. We are agnostic on this, but
given considerable evidence suggesting that higher scores are associated with higher informality
or unemployment rates and lower labor force participation rates, we do use the term “reform” to
refer to a reduction in these indexes and “reform reversal” to refer to an increase in these
indexes.
11
B. More Detail on Constructing the Indexes for the OECD and LAC Regions
As indicated above, for the 21 OECD countries covered by Allard (2005a), her series
were used to construct an index from 1950-4 through 2000-4 and then converted to a 0-3 scale so
as to coincide with those of Botero et al (2004) for the period 1995-9. For three of the OECD
countries covered by Blanchard and Wolfers but not by Allard (Korea, Mexico and Turkey), the
Botero numbers were extrapolated backwards to 1960-4 based on an index with 1995-9 = 1
constructed from Blanchard and Wolfers NEWEP, although as described below additional
information from other studies has been used in the case of Korea.
Two countries covered by Blanchard and Wolfers but not Allard, Luxemburg and
Iceland, were not covered in Botero et al (2004). They were, however, covered in the World
Bank’s Doing Business 2008 (pertaining to 2007) as “Rigidity of Employment Index”
11
. This
index was scaled differently but based on similar methodology to the Employment Laws Index
of Botero et al (2004) as explained in Doing Business. The values from the Doing Business scale
for these two countries were then extrapolated backwards to 1995-9 on the basis of information
in NATLEX or LEXADIN website and other sources and converted to those of Botero et al
(2004) scale on the basis of the conversion factor found for neighboring countries. The resulting
values were then interpolated back from 1995-9 on the basis of the Blanchard and Wolfers
(2000) data base for NEWEP.
As indicated above, for Latin American and Caribbean (LAC) countries for which 1995-
1999 values of ELR were available from Botero et al (2004) (other than Mexico which was
covered in Blanchard and Wolfers (2000)), the interpolation backwards was based on the Job
Security Index of Heckman and Pages (2000), certain refinements thereof for the dates of
11
This index combines sub-indexes for (1) alternative employment contracts, (2) conditions of employment, and (3)
job security. See also World Bank (2004).
12
reforms contained in Heckman and Pages (2004) and information from the NATLEX and
LEXADIN websites, and other sources identified in the Appendix.
12
Specifically, the Heckman
and Pages numbers were calculated as the total costs of firing a worker relative to wages in 1987
and in 1999. We calculated the ratio between the 1987 number and the 1999 number for each
nation in the Heckman and Pages study, and used this as an index to extrapolate the index
number in the Botero study backwards to the e 1985-1989 period for each nation. For the 1990-
4 value we made use of NATLEX and other studies to identify changes and if these occurred to
approximate the magnitude based on the relative importance of the items changed based on
NATLEX or LEXIDIN. Similarly, NATLEX was also used to extrapolate the ELR indexes
backward in time in a way consistent with the Botero (2004) matrix of the scores assigned to
each of the 31 components of ELR.
For those LAC countries (Costa Rica, El Salvador, Guatemala, Haiti, Honduras,
Nicaragua and Paraguay) for which there was no observation for the Employment Laws Index in
Botero et al (2004), but there did exist a score on the closely related Rigidity of Employment
Index for either 2003 or 2007 from the World Bank Doing Business volumes for 2004 and 2008,
a similar procedure was used as that described above for Iceland and Luxemburg to interpolate
backwards to 1995-99. For Chile, even though over time values were included in the Heckman
and Pages (2000) study, because of its greater detail and longer time coverage, the index was
interpolated back to 1960 based on the Job Security Index data presented by Montenegro and
Pages 2004 (Figure 7.1). Wherever possible, these indexes for Latin America and the Caribbean
were interpolated backwards from there to the late 1980s based on Heckman and Pages (2004)
12
This index was computed as the expected discounted cost at the time a worker is hired of dismissing the worker at
some time in the future based on existing labor law (but excluding the costs of court actions). It makes use of a
common discount rate of 8 percent and assumed turnover rate of 12 percent
12
and the costs (inclusive of those
related to seniority) of dismissing a worker (for justified and unjustified reasons).
13
and to earlier years based on the various provisions in the earlier employment laws from
NATLEX and LEXIDIN. In cases where there was no new Employment Law between dates
covered, such as in Haiti between 1984 and 1995-99, the resulting index values were assumed to
remain constant between those dates. For LAC countries not included in Heckman and Pages for
which Djankov et al (2003) or Botero et al (2004) was available or could be constructed based
on conversions from the subsequent Doing Business surveys, namely, Guatemala and Haiti, the
values were interpolated backwards solely based on changes in the provisions of the employment
laws as reported in NATLEX and LEXIDIN. Other LAC countries, Belize and Suriname had to
be dropped from the present study for lack of data.
To repeat, the 1995-9 values for all these countries have been constructed in such a way
as to be consistent with what the Botero et al (2004).
C. Constructing the Indexes for Countries in Other Regions
As indicated above, for countries outside of the OECD and LAC regions, data on
employment rigidities are much less complete and rarely if ever already developed into an index
over time. Some early studies identifying the effects of employment laws were Fallon and Lucas
(1991, 1993). They identified law changes in both India and Zimbabwe that had the effect of
tightening labor regulations and claimed that in both cases the result was lower formal sector
employment of industrial labor.
Once again, our first step is to make use of the Botero et al (2004) and subsequent Doing
Business surveys for 2004 and 2008, for arriving at values of the index for1995-99 that would be
comparable to those for the Botero et al (2004) indicators. In these cases, we went country by
country, making use of NATLEX, LEXADIN and published papers covering the country in
order to interpolate the indexes backward in time to the extent possible.
For example, Gerardo Sicat’s article “Reforming the Philippine Labor Market” provided
14
us with ample data of Philippine labor regulations. This source discusses changes in labor
law/regulation and the actual impact they have on labor regulation (and the ease of employing
workers) – whether they make regulation stricter or looser. Starting with our 1999 Botero
number, we assessed the impact of previous changes in labor regulation and worked our way
backward.
For India we made use of Dutt (2002), Dutta Roy (2004), Besley and Burgess (2004) and
Pages and Ahsan (2008) and other materials identified in Section IIIB below. The latter two of
the studies identified above make use of state-specific changes to the federal-level Industrial
Disputes Act of 1947. This is relevant because in India’s federal system states are also granted
the power to regulate industry, labor, health and other matters. A problem with the state level
data is that some states were liberalizing while others were tightening regulations making it
difficult to aggregate them into all-India changes. As noted below, we did so very crudely based
on the number of states moving in either direction, the magnitudes of these changes and the sizes
of the respective states. Note also that Bhattachajea (2006) has criticized Besley and Burgess
(2004) and Pages and Ahsan (2008) though more for their analysis of the effects than for the
scoring of the amendments. It should also be admitted that most of these indexes for India
pertain exclusively to manufacturing (and even within manufacturing there may be differences).
For the remainder of our countries (primarily Africa, Asia and the Middle East), the International
Labor Office’s NATLEX database provided us with the majority of our data. Similar to our
process for the Philippines, we gauged the effect of previous changes in labor regulation and
adjusted Botero ELRs across different time periods accordingly. In each such case, we gain
made considerable use of the appendix materials for Botero (2004) containing the scores
assigned to the countries included in that study for all 31 subindicators. For the nations for which
a 1995-1999 Botero ELR did not exist, we used the subsequent rankings and indexes on ease of
15
hiring and firing workers in the Doing Business Surveys of 2004, 2007 and 200 as described
above for OECD and Latin American countries. There were, however, some instances where we
found inconsistencies between Botero et al (2004) and Doing Business evaluations that could not
be attributed to labor law changes and therefore did not use this method of conversion. For Iran,
for example, which was not included in Botero et al (2004) but was included in subsequent
editions of Doing Business, we have made use of the relevant components of the index of Labor
Market Flexibility by Gholam Reza (2009) which covers the entire period 1960-2006 that also
includes minimum wages and other unemployment insurance requirements. For a few countries
that have received considerable attention by individual scholars or international organizations,
we have made considerable use of these studies. Illustrative examples of this are given in the
following section.
III. Some General Patterns and Illustrative Country Examples
As has been widely observed, labor market regulations tend to be much more static over time
than other types of regulations. This may well be attributable to the numbers of parties that are
generally involved in making changes to labor regulations (firms of different ownership types,
sizes, sectors, workers of different types, managers of different backgrounds, government
bureaucrats from different ministries, labor unions of different types and sizes, employers
organizations, intermediaries between business and labor, and the judiciary).
Despite the fact that there are quite a few countries that have experienced little or no change
in their ELR scores over the entire period, in each region, there are also countries whose scores
have changed from one 5 year period to another, resulting in some interesting differences over
time as well as across countries and regional or other groupings. Section A examines some of
these patterns of change in regional or other groupings. Section B identifies some changes over
16
time in some individual countries that are in neither the OECD nor the LAC regions and thus
with over time changes in labor regulations not covered in the major existing sources identified
in the previous section.
A. Some Regional Patterns
On average, the change in the average of all country ELR scores in the LAMRIG data base
over the period 1960-4 to 1995-9 have been modest. There were some notable changes in the
1950s and especially in the last ten years. At the regional level, however, there has been more
volatility, but again primarily in the 1950s and 1960s and again after 1995. This is reflected in
Figures 1 and 2. An important caveat to interpretation of both Figures 1 and 2 is that, except for
the high income high-middle countries of Europe, the early year portions of these curves are
based on very few and not necessarily representative countries.
Figure 1 presents the average ELR scores over the same period of time for six regional
groupings, namely, Latin America including the Caribbean(LAC), Asia, “Europe” (defined to
include, Canada, the US, Australia and New Zealand), Sub Sahara Africa (SSA), Transition
Countries of Central and Eastern Europe and Central Asia and the Middle East and North Africa
(MENA). In this case, there are two regions, Asia and MENA, with very little change in the
average ELR scores over time. Asia exhibited very little change over the whole period,
remaining below 1.5 for the entire period before 2000-4 at which time the average ELR has
edged upwards. While MENA’s index revealed little change except for the noticeable increase in
the 1950s. Its ELR also remained near but slightly below 1.5 for the entire period. In contrast,
SSA’s average ELR has risen rather steadily from below 1 in 1950-4 well over 1.5 in 2000-4.
The ELR average for “Europe” rose even more steeply from less than 0.5 in 1950-4 to well over
1.5 in the early 1990s before dropping off very slightly since then. By contrast, LAC’s ELR
started well above 1.5, rose further in the 1960s but has declined to only about 1.2 by 2000-4.
17
The ELR for Transition region started at about about 2.3 in 1060-4 but has fallen to a little below
2 since 1965-9.
Figure 2 presents the average ELR scores in LAMRIG over time for four different levels of
per capita income, low, lower middle, upper middle and high. Reminiscent of the pattern for the
“Europe” region in Figure 1, the ELR average for high income countries started from an
extremely low level of just over 0.4 in 1950-4, rose rather steadily to well over 1.6 by 1985-9
before leveling off to about 1.6 in 2000-4. The low income and lower-middle income countries
started the 1950s with quite a bit higher ELR indexes but experienced quite sharp increases in
their respective ELRs in the 1950s, leveled off in the 1970s and 1980s before rising further in the
last decade. The ELR of the lower-middle income countries in 2000-4 was 1.8 the highest of off
the groups. By contrast, the ELR of the upper-middle income countries displayed a rather
inverted U pattern, starting with the highest level of any region in 1950-4, rising further to 1.6 by
the mid-1960s but the falling gradually to about 1.4 by 2000-4.
B. Some Illustrative Country Cases
Given (1) the varying sample sizes of the region averages reported in Figures 1 and 2, (2)
the earlier observation that many countries experienced little or no change in ELR scores over
time and (3) the possibility that the ELR scores of different countries within any one of these
regional or income groupings may move in different directions resulting in little change in group
average scores, in this section we turn to some individual country experiences.
In their exposition of their indicators of the restrictiveness of labor laws, Botero et al
(2004) illustrated the indexes and the relevance of differences in legal tradition by comparing
New Zealand and Portugal, two countries at fairly similar income levels(at least in the late
1990s) but different legal traditions and ELR scores in 1997. In particular, Portugal was an
example of a country with French Civil Law background and a high ELR Index of 2.36 (3.7 on
18
Allard’s EPL index) while New Zealand was an English Common Law country with a low ELR
index of 1.06 (0.7 on Allard’s EPL index). They illustrated the sizeable differences between
these countries with reference to the various subcomponents of their index, making the sizable
difference between their respective overall index scores understandable. As shown in Table
III.1, in our LAMRIG data base, New Zealand has had a low ELR score of 0.48 (based on our
conversion of Allard’s EPL to the 0-3 Botero scale) for the entire time between 1975-9 and
1995-9 before rising slightly to 0.50 in 2000-4. In the 1960’s, however, its ELR score was lower
still at 0.14. Portugal, by contrast, had its high ELR score of about 2.4 ever since 1985-89.
Notably, however, in the early 1950s and even in the 1960-4 period, Portugal’s score of 0.066
was slightly lower than that of New Zealand’s at that time. Clearly, if the 1950s or 1960-4 scores
had been used, this comparison would not have served the purpose of showing that the French
civil law tradition gives rise to greater restrictiveness in labor legislation than does the common
law tradition. Moreover, with such sizeable changes in relative rankings over time, it is unclear
why the legal tradition should matter much since the legal tradition almost never changes over
time. As indicated above, the changes over time in OECD and Latin American countries have
already been rather extensively documented by the several already cited studies done on these
regions.
Therefore, to illustrate some other interesting differences in ELR scores over time, also in
Table III.1 we show the ELR index values for seven countries from outside of these regions, two
giants from Asia: India and China, another large country from Asia; Philippines, a medium sized
country from SSA: Ethiopia, and three small countries Botswana and Zambia from SSA and
Jordan from MENA. Clearly, China’s ELR started high with a score of 2.5 (the highest of all the
countries in the table in 1960-4 ) but it declined in the late 1970s, again in the late 1980s and still
again in each of the next three periods, reaching a relatively moderate ELR of 1.42 by 2000-4.
19
By contrast, Ethiopia, India and especially Philippines have seen their ELR averages rise over
time so as to have the next highest scores behind only Portugal by 2000-4. The ELR scores for
the other three countries have displayed more interesting patterns. Jordan’s was steady at a
relatively high value of 2.7, before falling substantially in 1995-9 and then rising again slightly
in 2000-4. Botswana’s ELR index started very low at 0.9 in 1970-4, rose gradually to 1.3 in the
1990s before falling to 1.05, the lowest of all countries except New Zealand in 2000-4. Zambia’s
ELR has fluctuated a bit more but remained fairly low over the whole period.
To obtain some insight into the changes in some of these countries and again the role of
legal tradition, in the following subsections we will provide short political economy accounts of
four of these countries, two (India and Zambia) classified as English Common Law countries,
Jordan classified as French Civil Law and China classified as having a Socialist legal tradition.
In the Botero et al (2004) indexes for 1997 and hence the 1995-9 period in LAMRIG, these four
countries all had scores in between those rather extreme scores for New Zealand and Portugal.
India
The main legislation concerning labor regulations in India dates back to two important
laws at more or less the time of the country’s independence, the Industrial Employment Act of
1946 and the Industrial Disputes Act of 1947. A third important act is the Trade Union Act
which dates all the way back to 1926. The stipulations of this latter act have changed little over
time and been quite accommodating to unions and union activity. While the Industrial
Employment Act allowed for layoffs on grounds of financial necessity, as well as for medical or
disciplinary reasons, when financial reasons are given as the justification, the Industrial Disputes
Act comes into play. While the latter act was amended on various early occasions (1964, 1965
and 1971), the most important amendments were those of 1976, 1982, and 1984.
The 1976 amendment made prior approval by the government of any such layoffs
20
mandatory. While at that time this provision applied only to firms with 300 or more workers, in
1982 the provision was changed to apply to all firms with 100 or more workers. Employers who
violated these regulations could be fined heavily and forced to reinstate the dismissed workers.
These changes are reflected in the increase ELR index for India in Table III.1 from 1.1 to 1.15 in
1975-9 and 1.2 in 1980-4. Under India’s federal constitution, its states also have the power to
impose regulations of these types. Indeed, several studies, Besley and Burgess (2004) and Ahsan
and Pages (2008) have made extensive use of state level amendments to these laws.
13
While in
each period there were some states that amended their laws in ways that were more restrictive
and others that amended in a less restrictive direction, during the late 1980s the number of states
classified as increasing restrictions outnumbered and were generally more important states than
those liberalizing the laws. This accounts for the final increase in the index from 1.2 to 1.3 in
1985-9. After this, despite liberalization of various other regulations, the labor laws of India
were not appreciably modified as noted by Saha (2007). But, on the other hand, the 2004 Doing
Business reported a considerably higher score which converts to 1.53 on LAMRIG.
Saha (2007) also cites an estimate that only 10 percent of the requests to government for
layoffs on financial conditions were approved. Hence, the inability to lay off workers could be
regarded as very restrictive. The reason why, even at the end of the period, India’s scores on this
index are not higher is that in terms of the restrictiveness of work hours, overtime and overtime
pay, and to a lesser extent alternative types of contracts, India has low restrictiveness scores.
China
As indicated above, China had the highest ELR index of all the countries included in
Table 3 and indeed was tied (with Panama and Poland) for the highest in the entire LAMRIG
data set for 1960-64. At that time, China not only had highly regulated labor market markets by
13
For some criticism of the studies and the conclusions drawn from them see Bhattacharjea (2006).
21
virtue of the labor law, with industrial employers, then largely state enterprises, unable to lay off
workers under virtually any circumstances and providing life-time employment, but also workers
had virtually no physical mobility thanks to the tightly controlled household registration system
(hukou) that had been set up in the mid-1960s.(Brooks and Tao, 2002, Reutersward, 2005) This
system restricted households, by province, within province by local community and rural-urban,
and sector (agriculture and non-agriculture). Not only could workers not be dismissed but also
had to be given housing and other benefits. Workers had little to no choice as to where they
would work.
Only beginning in the late 1970s did employers have a right to offer incentives to
workers such as bonuses for productivity (Brooks and Tao, 2002). In 1986 and especially in
1988 with the People’s Industrial Enterprise Act and the Regulation of Private Enterprise Act,
both of 1988, labor regulations were further lightened. Employers wanting to terminate workers
for financial reasons were from that time only forced to consult with authorities and labor unions
over the layoffs but were not required to receive approvals. At the same time, fixed term
contracts were introduced on a limited scale, especially for foreign and non-state enterprises
which were by then being encouraged.
In 1992 enterprises were allowed to set their wage rates in an autonomous way as long
as they stayed within certain government -set required bounds. Even SOEs were allowed to be
closed down on financial grounds. The 1994 Labor Act allowed for collective contracts, floating
wage rules and layoffs for financial, disciplinary or medical reasons. This was facilitated socially
by the state itself assuming responsibility for the re-training and financial support of laid off
workers (OECD 2010). More widespread approval was also given to the use of fixed term
contracts and employers were allowed to terminate a worker prior to the completion of the fixed
term for financial reasons, subject only to 30 days notice and specified compensation. Firms
22
were increasingly relieved of their requirement to provide housing for their workers. Also in the
1990s the hukou system was relaxed in small towns at least in certain provinces so as to allow
more migration from agricultural to non-agricultural jobs and even from rural to small town
locations (Brooks and Tao 2002). Quotas on the number of hukou changes in a given jurisdiction
were also abolished in some locations (OECD 2010). In late 1999 a Contract Law was passed
which came into force the following year. Although this law did not deal directly with
employment contracts (Cooney et al 2007), it may have indirectly through allowing for labor
contracting within service contracts. Beginning in 2002 laid-off workers no longer received the
retraining in Re-employment Centers under the auspices of the state enterprises, receiving
instead only unemployment benefits. Fixed term contracts were allowed to (and in fact were
reported to) increase rapidly in the 2000-4 period suggesting, perhaps why the 2004 Doing
Business assigned a lower score on the Employment Laws Index during this period equivalent to
1.42 on the Botero et al (2004) ELR. We have adapted that figure even though as Cooney et al
indicate, the next new Labor law, the Labor Contract Law, did not come into effect until 2007.
These successive changes in the relevant labor laws and regulations of China account for
the indicated reductions in the ELR index for China in Table III.1 from 2.5 in 1960- for all the
periods through 1980-4 to 1.8 in 1985-9 and 1.62 beginning in 1990-4. These changes were no
doubt associated with the increasing market-orientation of Chinese leaders and the accumulating
evidence that these liberalizing economic reforms were paying off. There was also gradual
increase in the relative importance of town and village enterprises and subsequently international
joint ventures and private enterprises in which inter-firm and international competition must
have helped to generate pressures for liberalizing labor markets. In the process the importance of
firm-provided housing and residence based provisions of food and other benefits also
diminished, and as noted above a proliferation of new types of employment contracts.
23
Jordan
Jordan is different from all the other countries discussed in this section in that it is a
monarchy. Monarchies, especially those of relative short duration like Jordan’s and where the
monarch has considerable political power, face the problem of how to maintain legitimacy in the
eyes of its citizens without elections for the executive and democracy which could limit their
power. The provision of economic security is one important means for endowing itself with a
sense of legitimacy on the part of its citizenry. Lacking the funding for a full-fledged social
insurance system, for most of its post-WWII history Jordan’s monarch-led government has
attempted to do this through its government and public enterprises providing rather well-paying
jobs to many of its citizens with reasonably generous retirement schemes and job protection.
Not surprisingly, over much of this period, most Jordanian public enterprises were known to be
over-staffed.
The country’s Trade Union Law of 1953 confirmed the right of workers to form unions
and the right to strike. This law was amended on several occasions but with only minor import,
e.g., in 1956 to prohibit civil service workers from unionizing or striking, and in 1976 to limit
the number of labor unions. From the detailed subindexes for Jordan’s overall score on the
Botero index, Jordan is rated very restrictive in the use and cost of overtime, the hours of work
and pay scales being treated quite similarly to those in the state bureaucracy. The Labor Code of
1960 applied with little revision until its replacement by the Labor Law of 1996. It was also rated
quite restrictive in terms of the lack of provision for fixed term or other types of labor contracts.
We assigned an ELR score of 1.7 to Jordan for all years prior to the 1995-9 period.
The Labor Law of 1996 and Act No. 36 of 1997, however, relaxed restrictions quite
significantly. First, it gave more specific recognition to fixed term contracts which would not
require compensation on completion of the term of employment specified in the contract.
24
Second, it allowed for employment on a trial basis not to exceed three months. Third, it
recognized still another type of contract, namely, that of indefinite duration which would apply
to workers working on piece-rate and other contracts. Employers were permitted to suspend even
these contracts for economic or technical reasons as long as the Ministry of Labor was notified
(but its approval was not necessary). Workers could even be dismissed from fixed term contracts
before completion of the terms as long as notification was provided and specified compensation
provided. It was also made clear that the rights to join a labor union and be represented by it did
not apply to non-Jordanian workers (Bitar 2004, Shawabkey2006).
14
Based on these
considerations Jordan’s score on ELR index for 1995-9 was reduced to 1.46 from its earlier
value of 1.7.
Lying behind the change were three factors: (1) the delayed but growing interest of the
Jordanian government in privatization and viewing growth of the private sector as the way to
deal with the country’s high and rising unemployment rate
15
, (2) the prior liberalization of other
aspects of Jordan’s economy, and (3) the fact that in 1996 negotiations were under way to
stimulate Jordanian exports in low wage garment industries by signing the Qualifying Industrial
Zones (QIZ) protocol with Israel and the US.
With respect to (2), as noted by Pripstein-Posusney (2004) the delay in privatization was
especially notable since the other liberalization measures (with respect to product markets,
financial markets and international trade) had commenced in 1991 and 1992, triggered by the
severe economic crisis that Jordan suffered as a result of the Gulf War. These reforms and
proclamations that privatization would also take place were also supported financially by the
14
Non-Jordanian workers have become an increasingly important component not only of the agricultural and
construction labor force but also of manufacturing employment.
15
Advocacy for increased reliance on the private sector became stronger after King Abdullah succeeded his long-
time predecessor King Hussain in 1999.
25
World Bank and other agencies. The fear of loss of jobs in the process of privatization and the
adverse effect that this might have for legitimacy of the monarchy might well have contributed
to this. Even when privatization started in earnest after 2000, Pripstein-Posusney (2004) informs
us that clauses in the specific privatization agreements specified that the new firms would have
to retain workers for at least two years after the sale and that the proceeds would be used to
provide retraining and other help to impacted workers.
With respect to (3) the QIZ Protocol, formulated making QIZs in Jordan considered as an
extension of Israel which already had a FTA agreement with the US.
16
Chief among the
industries subject to high tariffs and quotas was the garments industry, a low-wage, labor
intensive sector. Garments were the main activity to attract investment, most of which was in the
form of FDI which came to Jordan to take advantage of these special incentives. Yet clearly to
be competitive with countries with much lower wage rates like China, India, Pakistan and
Bangladesh, This put pressure on Jordanian officials to keep labor laws such that they would not
undermine these activities that soon came to provide a large share of total industrial employment
(Nugent and AbdelLatif, 2009).
However, continuing inflation in the country and press calling attention to some abuses
of labor among QIZ employers led Jordan’s Labor Ministry to impose minimum wage
regulations on all labor contracts, thereby causing us to raise the ELR for Jordan to 1.52 for the
2000-4 period.
Zambia
Interestingly enough, the motivation for the changes over time in the ELR index for
Zambia bears some resemblance to that of Jordan. Like Jordan, Zambia has been heavily
16
As further incentive to Jordan for having signed a Peace Treaty with Israel, Jordanian QIZ industries in which at
least 8 percent of the final cost had to come from Israel, were exempted from quotas as well as tariffs.
26
dependent on the public sector for its formal economic activity. Its government and state
enterprises provided the vast majority of formal sector enterprises until the mid 1990s when
privatization got under way. In 1991 there were some 282 state enterprises, many of the largest
being in the mining, manufacturing and service sectors.
While in terms of the labor force as a whole the vast majority has long been engaged in
self employment, microenterprises, subsistence farming and other informal activities, the formal
sector contracts, largely in the public sector were of two types, fixed term contracts of 6 months
duration with no commitment for renewal at the end of the contract, and permanent contracts,
that would terminate primarily only with the death or retirement of the worker. The labor law,
the Employment Act of 1965, was very simple and not very restrictive since short term contracts
were feasible. Since state enterprises in which the bulk of the formal sector labor force was
employed virtually never went into default, the issue of dismissal of permanent or even fixed
term workers for financial reasons virtually never arose. On the basis of matching the various
provisions of the law to the Botero et al 2004 scoring of the sub-indexes, we assigned Zambia a
rather low score of 1.05 for the years up to the 1985-9 period. The Industry and Labour
Relations Act allowed for labor unions, one in each industry.
However, in the 1980s the many distortions arising from monopoly labor unions and the
country’s highly protective trade regime and various other product market imperfections began
to take their toll on the economic viability of the country’s economy and even state enterprises
started to face financial difficulties. In this environment, a new Employment Act was introduced
in 1982 with subsequent set of regulations. The revisions of the labor regulations had the effect
of slightly tightening the conditions under which workers could be dismissed. On the basis of the
tighter job security regulations, the score assigned to Zambia on the ELR index rose to 1.3
beginning in the 1985-9 period. With a new democratically elected government coming to
27
power after 1990, the government obtained financial support from the International Fund and
adopted a package of liberalizing policies, including trade and product market reforms and
committed itself to privatization of the state enterprises. The new government looked to the
private sector as the preferred route to prosperity. But this meant encouraging private firms to
enter and grow in size and private investors to buy up the ailing state enterprises. To that end, the
government passed a number of new laws, the Privatization Act of 1992 and later the
Employment Act of 1997.
This 1997 labor law explicitly identified a number of additional types of labor contracts,
temporary, part-time, fixed term and so on giving private sector employers more options. This
explains why the ELM index score assigned to Zambia was lowered to 1.15 in the period 1995-9.
Rather remarkably, by 2000 the vast majority of the 282 state enterprises that existed in 1991
were privatized and the government was able to collect sizable amounts of privatization
revenues. These revenues were to be used in part for retraining and otherwise assisting
redundant workers. Yet, in terms of formal sector employment, the privatization experience was
very disappointing. Formal sector employment fell by some 15 percent over the decade of the
1990s. Most of the dismissed workers went into the informal sector or unemployment. The
decline in formal employment was quite remarkable given that population was growing at a rate
of 3.8 percent per annum over the same period. Several observers (e.g., Petrauskis 2005 and
Nyirenda and Shikwe 2003) attribute this failure to (1) the law itself which made it easy for the
private employers to substitute informal short-term workers for formal ones, (2) the failure of the
government to use the privatization proceeds to retrain laid-off workers, and (3) the failure of
either government administrators or the courts to enforce the provisions of the employment laws
(made difficult by the vagueness of some of its provisions).
Perhaps as a result of the unpopularity of the privatization and the 1997 labor law, by
28
Statutory Instrument No. 2 of 2002, various labor regulations were tightened slightly, including
those for minimum wages, maternity leave, hours of work and other conditions of work. As a
result, we roughly calibrated this to justify an increase of 0.1 on the ELR to a level of 1.25 for
the period 2000-4.
A Brief Recap of the Cases
Of the four cases treated here, three different legal traditions were represented. Just as in
the previous Portugal New Zealand comparison, even if there may have been some notable
differences in these scores between legal traditions at one point in time as argued by Botero et al
(2004), from the above discussion it is by no means clear that different legal traditions played
much of a role in explaining changes in these scores over time.
17
Indeed, since the legal
traditions didn’t change over time, it would seem doubtful that they could contribute to the
explanation of changes in the ELR scores over time. From the above discussion of the four cases,
we believe that a case could be made for economic crises, political crises (or changes) and
perhaps certain changes in economic structure. We will therefore pay some attention to these
factors in the empirical work below attempting to explain changes in the ELR indexes across
countries and over time in the LAMRIG dataset.
IV. Trying to Assess the ELR/LAMRIG Index
In this section we discuss the methodology we choose to assess the appropriateness of our
ELR/LAMRIG index and present the auxiliary data required for this task.
In terms of the econometric methodology, we first try to replicate the results in Botero et al.
(2004). At the outset, the most clear-cut difference is that while their sample has 85 countries,
ours contains 142 countries for the same period (1995-9). Our first task is to determine whether
17
Perhaps a case could be made for Socialist law contributing to the high score for China in the early periods.
29
or not we can replicate their results in a cross-sectional setting with this extended sample.
Accordingly, from their Table IV, the first model we estimate takes the form:
i i i i i
LO GDP LAMRIG ? ? ? ? + + + =
2 1
(1)
where LAMRIG
i
is our index of Labor Market Rigidity for country i, GDP
i
is the log of per
capita GDP, and LO
i
is a set of dummy variables for legal origins (namely, French, German,
Scandinavian, Socialist and English.) Botero et al. estimate this model by OLS with robust
standard errors and data for 85 countries in year 1997. They find that legal origins are a
substantially more important determinant of labor market reform than per capita GDP. They
argue that this result favors the legal theories of institutional changes (and, by the same token,
belittles the other two theories they identify, the efficiency and political theories.)
We then subject this baseline model to various robustness checks. In particular, we ask
whether the taking into account the over time variation of our index affects the basic results. We
investigate this first by fitting a simple pooled OLS model to the data (by simple we mean we
run OLS on the pooled data without taking into account the panel structure.) We also run two
split-sample exercises. The first estimates this baseline model separately for OECD and non-
OECD countries. The second estimates the model separately for the pre- and post-1980
observations. The rationale for the first split of the samples is that richer countries may face quite
different constraints than poorer countries. If so, this could be reflected by important differences
in the determinants of LAMRIG between the samples. The justification for pre and post 1980
split is the widespread perception that this year somewhat marked the beginning of an era more
favorable to market-oriented reforms than the previous 20 years or so (which were marked by
interventionist import-substitution strategies, especially in developing countries).
Our second step in terms of estimation strategy is to exploit more decisively the panel
30
structure of our data. The fixed-effects estimator would be a natural starting point but one of the
most important variables in the Botero et al (2004) exercise, legal origins, are time-invariant.
Therefore our starting point is the random-effects estimator.
18
The model we estimate takes the
form:
it i ti it
LO GDP LAMRIG ? ? ? ? + + + =
2 1
(2)
where again LAMRIG
it
is our index of Labor Market Rigidity for country i at period t. The
subscript t refers to a 5-year period, where the measure is the average over the whole period. The
nine periods included are: 1960-1964, 1965-1969, 1970-1974, 1975-1979, 1980-1984, 1985-
1989, 1990-1994 and 1995-2000.
In order to minimize country-specific errors, we clustered the
standard errors at the country level. Using the random-effects estimator, we also carry out the
same split-samples checks as for the baseline model, namely for OECD versus non-OECD and
pre- and post-1980.
Thus far we have talked only about the levels of LAMRIG. Yet we define reform as
changes in these levels. The third step in our estimation strategy is therefore to estimate changes
in levels of LAMRIG. In order to do that, we recognize that labor market reform in one period is
related to the amount of past progress in this reform. Hence, we capture this by enlarging the
baseline Botero et al. model with a one-period (i.e. 5 year) lag of LAMRIG. The model we
estimate takes the form:
it t i i ti t i it
X LO GDP LAMRIG LAMRIG ? ? ? ? ? ? + + + + + = ?
? ? 1 , 4 3 2 1 , 1
(3)
where ?LAMRIG
it
is the change in our index of Labor Market Rigidity for country i between
period t and period t-1, with periods defined as before. This model will be estimated at first using
the random-effects with standard errors clustered at the country level. Next, we re-estimate it by
18
The Hausman test contrasts the fixed- and random-effects estimator for models containing only the time-variant
variables in these models. In this case, it is not very helpful because the test runs for a model that only contains per
capita GDP.
31
adding variables for four different groups of factors (in X
i, t-1
) namely, political crises, economic
shocks, structural factors and other reforms. As the notation indicates, we always enter these
factors lagged one-period. This is not only to minimize endogeneity concerns but also because
the time window we use is somewhat lengthy and it may be that the reform occurs early in the
window and any of these four factors later. By lagging these right-hand side variables we
eliminate this concern.
As for the data we use in the estimation of the above models, the measure of the
dependent variable in the analysis is the LAMRIG index, discussed above in considerable detail.
The other two sets of variables in the baseline model are GDP and legal origins. Per capita GDP
is from the Penn World Tables and the legal origins classification is provided by Botero et al.
(2004).
In terms of structural factors, we collected data on the following variables from the
World Development Indicators: the Gini coefficient, the Government share of GDP, the ratio of
foreign aid to GDP, the share of natural resources exports in total exports and the share of
agriculture in GDP.
For economic crises we include several different measures,
19
namely, the largest single
year GDP fall in percentage points that occurred in each five-year period (Max fall GDP), the
number of years of negative GDP growth (between zero and five for each of the 5-year average
period), the current account balance (CAB)
20
, the number of years in a debt crisis within each
five year period (Debt Crisis),
and a dummy variable for periods in which annual inflation was
above 50%.
Regarding political crises, we limit our attention to the following indicators. The first
19
For a review, see Furman and Stiglitz (1998) and Ishihara (2005).
20
CAB is an inverse measure of crisis.
32
group comprises count variables for both the assassination of important political leaders and
general strikes during each five year period.
Both of these variables originate from Banks (2005).
The second group comprises the democracy measure (from the POLITY IV data set) and also the
Political Constraints Index (POLCON) provided by Henisz (2000). The Polity IV democracy
variable is used to control for relative levels of democratic freedoms (coded in a 1 to 10 scale,
with 10 indicating the highest level of democracy). The stronger is democracy, the more the
median voter might be expected to exercise influence. Yet, because the median voter is more
likely to be a worker or even a union member, the influence of democracy on labor market
liberalization could be ambiguous or perhaps even negative. POLCON measures the number of
veto points in a political system, the expectation being that the more potential vetoes need to be
circumvented, the less likely it is that labor market reform will be adopted. The third and last
group contains a measure of the intensity of civil war and of the intensity of international armed
conflicts. Data for constructing these measures is from the Correlates of War project at the
University of Michigan.
Finally we investigate the role of other structural reforms potentially affecting the
probability that labor market reform is implemented. We focus on financial reform and on trade
liberalization.
21
We proxy financial reform by two measures: the share of credit to the private
sector in GDP, and an index of financial development that reflects not the overall size of the
financial system but its efficiency levels. In the case of trade liberalization, we use four
measures. One is the length in years of uninterrupted trade liberalization derived from the
Appendix 2-B of Warcziarg and Welch (2003). Another measure is a measure of trade openness
from PWT (openk, exports plus imports as a share of GDP). Thirdly, we use the trade
21
For a discussion of the relationship between trade liberalization and labor market reform see Fajnzylber
and Maloney (2005), and references therein. Idem for financial reform and labor market reform, see
Pagano and Volpin (2008).
33
liberalization index developed by Campos, Nugent and Hsiao (2006). The later further extends
the Sachs and Warner (1995) measure of trade openness that was already corrected and extended
from 1970-1989 to 1990-99 by Wacziarg and Welch (2003).
22
Since Rodriguez and Rodrik
(2000) and Rodriguez (2006) provide a powerful critique of the efforts of Sachs and Warner
(1995) to relate their “open” measure to cross-country growth rates, we have incorporated these
views in this measure of trade reform. A major objection of these authors was that the cross-
sectional evidence on growth rates relied on only two of the five S-W criteria, namely, export
marketing boards (XMB) and black market premium (BMP). In effect, this suggests that the S-W
index of openness, even when extended by Warcziarg and Welch, in fact has little to do with
trade restrictions in the form of tariffs and non-tariff barriers. Rodriguez (2006) also had some
specific quarrels with the way XMBs were treated in their classifications of certain countries.
Finally, he followed Warcziarg and Welch (2003) in using a lower tariff rate threshold (20%
instead of the 40% in the original S-W) to distinguish “open” from “closed”.
23
Since most
countries in the world had fallen below the 40% threshold by the mid- 1990s, this change has the
effect of giving more weight to tariff barriers in the classification. Therefore, we construct an
alternative measure that takes advantage of more recent information on XMBs (from World
Bank and other sources) so as to distinguish between those marketing boards that in practice
discriminate against producers for export markets from and those which do not, as well as these
other suggestions. Given the view expressed most strongly in Rodriguez and Rodrik (2000), that
22
More specifically, these authors defined a country as closed (i.e., open =0) if it had any one of the
following: (1) an average tariff rate of 40 per cent or more, (2) non-tariff barriers covering 40 per cent or
more of trade, (3) a black market exchange rate that is depreciated by 20 percent or more relative to the
official exchange rate, (4) a state marketing agency or board for major exports, and (5) a socialist
economic system (as defined by Kornai 1992).
23
This was used to show something that Warcziarg and Welch (2003) had already shown, namely that the
positive relation between growth and open found by Sachs and Warner (1995) and others disappears
when the lower threshold is used or when the period studied is that after 1990.
34
as far as growth is concerned, the key reform was BMP reduction, as another alternative we use
simply BMP.
V. Assessing LAMRIG
Next we turn to an assessment of LAMRIG based on an examination of the determinants of the
ELR indexes across countries and over time. Given that the lynch-pin for our construction of
LAMRIG was the Botero et al (2004) data set for 85 countries, we begin our assessment in Table
V.1 by trying to replicate some of the findings of that study. That study’s basic finding was
presented in their Table IV relating their ELR index to the log of per capita GNP, and dummy
variables for Socialist, French, German and Scandinavian legal origins. The omitted legal origin
was English Common Law. The results they reported for their sample of 85 countries is reported
in column (1) of Table V.1. As can be seen, the explanatory power of the model was quite high
and although the income per capita measure was insignificant the four legal origin dummy
variables had highly significant positive effects on ELR.
In column (2) of this table we repeat their analysis based on the LAMRIG dataset for the
same year but using our larger sample of 142 countries. Notice that the effect of income per
capita is now negative and significant but all four legal origin dummy variables still have
positive and highly significant effects on LMR (from now on designated LAMRIG). But our
more fundamental extension of their dataset is its extension over time going back to the early
1950s in quite a few cases. OLS estimates of the pooled panel data (now consisting of 792
observations) are provided in column (3). In this case, the negative effect of higher income per
capita is no longer significant (as in Botero et al (2004)) but all four legal origin dummies retain
their positive and significant effects. Note that the effects of all of them on LAMRIG are now
considerably stronger than in the original Botero et al (2004). Given our earlier observation that
35
in the 1950, 1960s and even 1970s, the ELRs were rising before stabilizing and declining in
some cases in recent years, in columns (4) and (5) we break the sample into pre and post-1980
observations. While the results are very similar for the Socialist and French legal origin
dummies, there are some notable differences in other respects. When split this way the negative
coefficient of the Log Per Capita GDP is again statistically significant but quite a bit larger in the
Pre1980 sample. On the other hand, the impacts of the German and Scandinavian Legal Origin
dummies are larger and more statistically significant in the post 1980 sample. In each of the
columns so far, the effects of the legal origin variables are quite consistent consistent with
Botero et al.(2004) in that the socialist, French Civil Law and German Civil Law traditions in
that order all have large positive effects on LAMRIG than the omitted Common Law tradition.
The only difference is that in some of these cases, columns (2), (4) and (5) the Scandinavian
Legal Origin has the largest positive effect.
Columns (6) and (7) provide the corresponding comparison between OECD countries and
non-OECD countries, reflecting in large part differences in income. Notice that in our case, the
non-OECD sample is considerably larger than the OECD sample. Again the various types of
Civil Law dummies are shown to have significant positive influences in both samples (when
there is sufficient variation of these variables in the sample to allow coefficients to be estimated).
Both the French and Scandinavian Legal Origin variables have considerably larger effect in the
OECD sample than in the non-OECD one. The most striking difference between the samples,
however, is the difference in the effect of Log Per Capita GDP, quite large and positive in the
case of the OECD sample, but negative and significant in the non-OECD sample. This would
seem to help explain the opposing trends in LAMRIG indexes between high income and
“Europe” regions and some of the other regions and income groups in Figures 1 and 2.
Given the aforementioned absence of change over time in the legal tradition upon which each
36
country’s legal system is based, as noted above, when fixed effects are used, the parameters for
legal origin dummies cannot be estimated. We proceed therefore in the rest of our empirical
analysis to estimate not the levels of LAMRIG but rather the changes in LAMRIG (i.e., reform
reversals in labor market regulations). As explained in Section IV, this makes it appropriate to
estimate the relationships in the LAMRIG panel with random effects and standard errors
clustered at the country level with equation (3) above.
Table V.2 reports the results obtained for changes in LAMRIG first for the full sample (721
observations) and then for the same subsamples as in columns (3)-(7) in Table V.1 but based on
this more appropriate (RE) estimation procedure. Once again, we find considerable variation
across samples in the effects of Log Per Capita GDP, positive and significant in the post 1980
sample and negative and significant once again in the OECD sample. In all the other cases
including the full sample, the coefficient of Log Per Capita GDP is not statistically significant.
With the minor exception of the Scandinavian Legal Origin dummy (for which there is little
variation in our samples), the coefficients of the various Civil Law Origin variables are no longer
statistically significant. This confirms what we seemed to find in Table III.1. Note, that in great
contrast to the estimates in Table V.1, the results are very weak with no more than one
explanatory variable being significant in any column except for the OECD sample where there
are two.
24
Given the rather glaring weakness of these results, in subsequent tables we add the lagged
level of LAMRIG to the right hand side and a series of other determinants suggested by our brief
review of the several country cases of Section III and by other literature.
Table V.3 reports the results when the added variable is one or another of the following
24
We have also investigate whether there are important non-linearities in the effects of per capita GDP but did not
find evidence for this to be the case.
37
structural variables: Income Gini, the government share in GDP, the share of foreign aid in GDP,
natural resource exports as a share of total exports and the share of agriculture in GDP. As
expected, the effect of Lagged LAMRIG is always negative and significant indicating that there
seems to be a convergence process going on in labor market regulations. This is quite consistent
with the quite different trends between countries with initially low LAMRIG indexes and those
with initially high ones in Figures 1 and 2 and the upward trends for Portugal and New Zealand
whose initial LAMRIG scores were very low, and the downward trends from initially high
scores for China and Jordan in Table III.1. Countries with high LAMRIG index values are likely
to reform, i.e., lower their LAMRIG scores over time, whereas those countries with low
LAMRIG scores are more likely to introduce reform reversals by raising their rigidity scores.
Consistent with the results for the full sample (column 1) of Table V.2, the coefficients of Log
Per Capita GDP are not statistically significant, except in column (1) where the Income Gini is
the added structural variable. Somewhat surprisingly, the coefficients of the French,
Scandinavian and Socialist Legal Origin dummy variables are once again positive and
significant, though of course much smaller than in the estimates presented in Table V.1 for the
level of LAMRIG. In most cases, the coefficients of the Socialist and Scandinavian Civil law
dummies are also positive and significant, though again much smaller than before. None of the
individual structural indicators has a significant effect on the change in LAMRIG. One should
notice, however, that due to missing observations on these additional variables, the sample sizes
are considerably smaller in this table, especially in columns (1) and (5).
In Table V.4 we present estimates similar to those of Table V.3 for changes in LAMRIG but
in this case with five different measures of economic crises, in each case lagged to avoid the
simultaneity and other problems identified in Section IV. Column (1) presents the results when
the crisis is a debt crisis. Columns (2) –(5) report the corresponding results when the crises
38
pertain to inflation rates above 30% per annum, a period including a year with the largest fall in
GDP during the period covered, the number of years of falling GDP within the five year period,
and the current account, respectively.
Once again, the effects of Lagged LAMRIG are consistently negative and significant, those
of French and Socialist legal origins positive and significant. The effects of Log Per Capita GDP
are negative and significant in columns (1) and (2) but negative and not significant in columns
(3)- (5). None of the economic crisis variables turns out to have a significant effect on the change
in LAMRIG. As in Table V.3, the French, Scandinavian and Socialist Legal Origin variables all
have small positive and significant effects.
25
Table V.5 substitutes political variables including political crises for the economic crisis and
structural variables included in Tables V.3 and V.4. Column (1) adds Democracy, Column (2)
uses instead the political constraints index (POLCON), column (3) assassinations, column (4)
strikes, and columns (5) and (6) international and civil wars, respectively. As with the economic
crises, these measures are all based on lagged values. Democracy has a negative but insignificant
effect on the change in LAMRIG as does POLCON which is often considered another measure
of democracy reflecting a system of checks and balances. So too neither strikes, international
war and civil wars have significant effects on the change in LAMRIG. The one political crisis
measure with an effect that is somewhat significant is that for Assassinations in column (3). This
variable has a negative effect that is significant at the 10 percent level. (We trust that labor
market reform zealots will not go so far as to recommend assassinations as a means of bringing
about such reforms!). The findings of previous tables of significant negative effects of lagged
LMRIG and the positive effects of the French, Scandinavian and Socialist Legal Origin measures
25
We have also run all these spcifications for each legal origin sub-sample and do find that this had an impact in the
main results.
39
are all retained in this table as well.
Finally, in Table V.6, to our basic specification we add alternative measures of other types or
reforms, again lagged. In columns (1) – (3) we present the results for three alternative measures
of trade reforms. Column (4) presents estimates when the added variable is the black market
premium (BMP), an inverse measure of trade reform. Columns (5) and (6) present results for two
alternative measures of financial market reform/development, namely, the share of credit to the
private sector in GDP and the Financial Reform Index, respectively. Again all the standard
results apply. In addition in columns (1), (2), (4), (5) and (6) the effect of Log Per Capita GDP is
again negative and significant (in most of these cases at the 5 percent level). The effects of the
various lagged other reform measures vary considerably from case to case. Trade openness as
measured by the first two measures in columns (1) and (2) reveal positive and significant effects
on LAMRIG changes, meaning reform reversal. In the same spirit, an increase in the BMP
premium has the effect of reducing LAMRIG. The two, financial reform measures raising the
share of private credit in GDP and the financial reform index, by contrast have negative effects
on changes in LAMRIG though only in the first case is the effect significant at the 10 percent
level.
Taken together, the results could be interpreted as providing a somewhat positive assessment
of LAMRIG. In particular, the cross-section results with the same specification as in Botero et al
(2004) but based on our considerably larger data set replicate and perhaps even strengthen the
Botero et al results for the legal origin variables. The estimated impact of per capita income,
however, is different between the smaller Botero sample and the larger LAMRIG sample for
1997. In particular, while the effect is negative and significant in some of our specifications,
especially those in Table V.6, the most important findings in this respect is the way its effect
differs in different types of countries, being most strongly negative in the OECD sample of Table
40
V.2 and most significantly positive in the post 1980 sample of the same table.
With respect to the legal origin dummies, the positive and significant influences cant in the
Botero study are confirmed when the larger LAMRIG data set is used. This is true for both the
cross section and pooled estimates for levels of LAMRIG. Although these effects largely
disappear when it is the changes in LAMRIG that are explained in Table V.2, they come back in
again though in much smaller magnitudes when the Lagged LAMRIG is included among the
explanatory variables.
Of the other variables we have introduced one at a time in the subsequent tables, only a few
have significant effects. One is the Assassinations measure of Political crises which seems to
have a weak negative effect on LAMRIG changes, implying LMR reduction or reform. The
others are three of the four trade reform measures which in each case have positive influences,
suggesting reform reversal and the private credit/GDP measure which has a significant negative
effect on LAMRIG changes, suggesting that lagged reforms of this sort facilitate labor market
reform whereas the opposite is the case for the aforementioned measures of trade reform.
VI. Conclusions and Future Research
The substantive results presented here are clearly only a beginning of a fuller analysis.
We would like to further examine the robustness of the results, e.g., when several of the
additional variables are retained in the estimating equation at the same time, or with more refined
measures of some of the variables used. Similarly, in view of the differences in some of the
effects between pre and post 1980 samples and between OECD and non-OECD samples, it
would be desirable to examine the robustness of the results of the more inclusive specifications
to different samples. Yet, even thus far, we find several interesting results:
(1) That, when the dependent variable is changes in the LAMRIG indexes, legal origin measures
41
still exert significant though smaller influence than when as in the Botero study it is the level of
the indexes that is estimated
(2) That none of the (a) structural variables (income Gini, Government share in GDP, Foreign
Aid to GDP, the share of natural resource exports in total exports and the share of agriculture) or
(b) economic crisis or (c) political crisis measures (except Assassinations) seems to have much
of an influence on LAMRIG changes.
(3) Consistent with the findings of other studies, labor market reforms may be affected
significantly by other reforms. Our evidence, though limited to trade and financial reforms, is
that lagged trade reform may set back labor market reform but that financial reforms may
encourage labor market reforms.
(5) With respect to future research, in addition to the additional robustness checks and
improvements in some of the measures of variables identified above, it is our intent to:
(a) Further improve on LAMRIG by digging deeper into the ever-improving availability
of information on labor laws over time and across countries,
(b) Possibly to follow the lead of some researchers on OECD countries to annualize the
data on LAMRIG as well as the related variables used to explain changes therein
over time,
(c) To extend the use of LAMRIG to examine its effects on labor market and other
phenomena as Botero and many others have with somewhat smaller data sets.
42
References
Addison, John T. and J. Grosso 1996. “Job Security Provisions and Employment: Revised
Estimates” Industrial Relations 35 (4), 585-603.
Addison, John T.and Paulino Teixeira 2003. “The Economics of Employment Protection”,
Journal of Labor Research 24 (1) 85-129.
Allard, Gayle 2005a. “Measuring Job Security over Time: In Search of a Historical Indicator for
EPL (Employment Protection Legislation)” Instituto de Empresa, Spain Working Paper
WP05-17.
Allard, Gayle 2005b. “Measuring the Changing Generosity of Unemployment Benefits: Beyond
Existing Indicators” Instituto de Empresa, Spain IE Working Paper WP05-18.
Alesina, A. and A. Drazen (1991). “Why Are Stabilizations Delayed?” American Economic
Review 81, 1170-88.
Arpaia, Alfonso, Peghe Braila and Fabiana Pierini 2007. “Tracking Labor Market Reforms in the
EU Using the LABREF Database”, Paper presented at the IZA - Fondazione Rodolfo
Debenedetti Workshop: Measurement of Labor Market Institutions, IZA, Bonn, July 4, 2007.
Banks, A. (2005), “Banks' Cross-National Time-Series Data Archive,” electronic database.
Bassaanini, Andrea and Danielle Venn 2007. Assessing the Impact of Labour Market Policies on
Productivity: A Difference-in-Difference Approach. Paper presented at the IZA - Fondazione
Rodolfo Debenedetti Workshop: Measurement of Labor Market Institutions, Bonn, July 2007.
Bertola, Guiseppe 1999. “Micoeconomic Perspectives on Aggregate Labor Markets”, in
Handbook of Labor Economics, v 3, O. Ashenfelter and D. Card, eds, Elsevier Science, 2985-
3027.
Bertola, Guiseppe 2008. Labor Market Regulation: Motives, Measures, Effects
Besley, Timothy and Robin Burgess 2004. “Can Labor Regulation Hinder Economic
Performance? Evidence from India” Quarterly Journal of Economics 119, (1), 91-134.
Bhattacharjea, Aditya 2006. Labour Market Regulation and industrial Performance in India: A
critical Review of Empirical Evidence” Indian Journal of Labour Economics 49 (2), 211-232.
Bitar, Fouad 2004. National Labour Profile: Jordan, International Labor Office
(http://www.ilo.org/public/english/dialogue/ifpdial/info/national/jo.htm)
Blanchard, O. and J. Wolfers (2000). “The Role of Shocks and Institutions in The Rise of
European Unemployment: The Aggregate Evidence,” Economic Journal, 110, 1-33.
Boeri, Tito , G. Nicoletti and S. Scarpetta 2000. “Regulations and Labor Market Performance”
CEPR Discussion Paper 2420.
43
Botero, J., S. Djankov, R. La Porta, F. Lopez-de-Silanes, and A. Shleifer (2004). “The
Regulation of Labor,”Quarterly Journal of Economics 119: 1339-1382.
Botero et al Labor Dataset_qje_dataforweb_2005.xls
Brooks, Ray and Ran Tao 2002. China’s Labor Market Performance and Challenges IMF
Working Paper 02/210
Bruno, M. (1996). Deep Crises and Reform: What Have We Learned? World Bank.
Bruno, M. and W. Easterly (1996). “Inflation’s Children: Tales of Crises that Beget Reforms,”
American Economic Review Papers and Proceedings 86, 213-17.
Campos, N., Hsiao, C. and J. Nugent (2009) “Crises, What Crises? New Evidence on the
Relative Roles of Political and Economic Crises in Begetting Reforms,” IZA DP.
Deakin, Simon, Pruya Lele and Mathias Siems 2007. “The Evolution of Labor Law: Calibrating
and Comparing Regulatory Regimes” International Labour Review 146 (1), 133-162.
Di Tella, Rafael and Robert MacCulloch 1999. “The Consequences of Labor Market Flexibility:
Panel Evidence Based on Survey Data” European Economic Review 49 (5), 1225-1259.
Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes and Andrei Shleifer, 2004, “The
Regulation of Labor,” National Bureau of Economic Research, Working Paper 9756.
Djankov, Simeon and Rita Ramalho 2008. “Employment Laws in Developing Countries”
Drazen, A. (2000). Political Economy in Macroeconomics, Princeton University Press.
Drazen, A. and V. Grilli (1993). “The Benefit of Crises for Economic Reforms,” American
Economic Review, 83 (3), 598-607.
Dutt, Pusha Labor Market Outcomes and Trade Reforms: The Case of India
Dutta Roy, Sudipta 2004. Employment Dynamics in Indian Industry: Adjustment Lags and the
Impact of Job Security Regulations” Journal of Development Economics 73 (1), 233-256.
Eichhorst, Werner, Michael Feil and Christoph Braun 2007. “What Have We Learned?
Assessing Labor Market Institutions and Indicators”
Fallon, Peter R. and Robert E.B. Lucas 1991. “ The Impact of Changes in Job Security
Regulations in India and Zimbabwe” World Bank Economic Review 5 (3), 395-413.
Fallon, Peter R. and Robert E.B. Lucas 1993. Job Security Regulations and the Dynamic
Demand for Industrial Labor in India and Zimbabwe” Journal of Development Economics 40,
214-235.
44
Fajnzylber, P and W Maloney (2005), “Labor demand and trade reform in Latin America,”
Journal of International Economics 66 (3): 423-446.
Forteza, Alvaro and Martin Rama (2006). “Labor Market “Rigidity” and the Success of
Economic Reforms across More than 100 Countries,” Journal of Policy Reform 9 (1) 75-106.
Freeman, Richard B. 2008. “Labor Market Institutions Around the World”, London: LSE CEP
Discussion Paper no 844.
Greenhill, Brian, Layna Mosley and Aseem Prakash 2009. “Trade-Based Diffusion of Labor
Rights: A Panel Study, 1986-2002” American Political Science Review 103 (4),
Grubb, D. and W. Wells 1993. “Employment Regulation and Patterns of Work in EC Countries”,
OECD Economic Studies 21.
Haddad, Gholam Reza, K. “The Impacts of Globalization on the Earning Inequality (the Case
Study of Iran,” Sharif University, mimeo, 2009.
Heckman, James J., and Carmen Pages 2000. “The Cost of Job Security Regulation: Evidence
from Latin American Labor Markets” NBER Working Paper 7773.
Heckman, James J., and Carmen Pages, eds. Law and Employment: Lessons from Latin America
and the Caribbean. New York: University of Chicago Press, 2004.
Henisz, W. (2000). “The Institutional Environment for Multinational Investment”, Journal of
Law Economics and Organization, 16 (2), 334-364.
Joh, Sung Wook 2001. Korea’s Corporate Sector: Crisis and Reform” in O. Yul Kwon and
William Shepherd. Eds. Korea’s Economic Prospects. Cheltenham: Elgar
Kucera, David 2002. “Core Labour Standards and Foreign Direct Investment” International
Labour Review 141 (1-2), 31-69.
Kwon, O. Yul 2001. Korea’s International Business Environment Before and After the Financial
Crisis” in O. Yul Kwon and William Shepherd. Eds. Korea’s Economic Prospects.
Cheltenham: Elgar, 245-265.
La Porta, Rafael, Florencio Lopez-de-Silanes and Andrei Shleifer 2008. “The Economic
Consequences of Legal Origins”, Journal of Economic Literature 46 (2), 285-332.
Lazear Edward 1990. “Job Security Provisions and Employment” Quarterly Journal of
Economics 105, 699-726.
Magruder, Jeremy R. “High Unemployment Yet Few Small Firms: The Role of South African
Labor Regulations”,
Montenegro, Claudio and Carmen Pages 2004. “Who Benefits from Labor Market Regulations?
Chile 1960-1998” in Heckman and Pages, ed. Law and Employment : Lessons from Latin
America and the Caribbean. New York: University of Chicago Press, 2004, 401- 434.
45
NATLEX. November 2008. International Labour Organization <natlex.ilo.org>.
Nicoletti, G. R.C.G. Haffner, S. Nickell, S. Scarpetta and G. Zoega 2000. European Integration,
Liberalization and Labor Market Reform” in G. Bertola, T. Boeri and G. Nicoletta, eds.
Welfare and Employment in a United Europe. Cambridge: MIT Press
Nickell, Stephen 1997. “Unemployment and Labor Market Rigidities: Europe versus North
America” Journal of Economic Perspectives 11, 55-74.
Nickell, Stephen.and R. Layard l 1999. Labour Market Institutions and Economic Performance,
CEP Discussion Paper 407 "Handbook of Labor Economics., V. 3 Ed by O. Ashenfelter and
David Card, 3029-3084.
Nickell, Stephen, L. Nunziata, W. Ochel and G. Quintini 2003. “The Beveridge Curve,
Unemployment and Wages in the OECD from the 1960s to the 1990s” in P Aghion, R.
Frydman, J. Stiglitz and M. Woodford, eds. Knowledge, Information and Expectations in
Modern Macroeconomics: In Honor of Edmund S. Phelps. Princeton University Press.
Nugent, Jeffrey B. and Abla AbdelLatif 2009. “A Quiz on the Net Benefits of Trade Creation
and Trade Diversion in the QIZs of Jordan and Egypt”. Los Angeles, CA. USC.
OECD 1999. OECD Employment Outlook 1999. Paris OECD
OECD 2004. OECD Employment Outlook 2004. Paris OECD
OECD 2006 Governance in China . Paris: OECD , especially Ch 1, 11.
OECD 2010. Economic Survey China Vol. 2010/6 , February, 2010.
Pagano, M. and P. Volpin (2008), “Labor and finance”, London Business School, mimeo
Persson, T. (2002). “Do political institutions shape economic policy?” Econometrica 70, 883-
905.
Persson, T. and G. Tabellini (2000). Political Economics: Explaining Economic Policy, MIT
Press.
Polity IV Project (2002). “Political Regime Characteristics and Transitions, 1800-2002”,
available at http://www.cidcm.umd.edu/inscr/polity/.
Pripstein Posusney, Martha 2004. “Globalization and Labor Protection in Oil-Poor Arab
Countries Racing to the Bottom?”in Ibrahim Saif, ed.,. The Jordanian Economy in a
Changing Environment. Amman: University of Jordan, Center for Strategic Studies, 115-151.
Rama, Martin and Raquel Artecona 2002. “ A Database of Labor Market Indicators across the
Countries”
46
Reutersward, A. 2005. “Labor Protection in China”: OECD, Social Employment and Migration
Working Paper No 30.
Rodrik, D. (1996). “Understanding Economic Policy Reform,” Journal of Economic Literature
34 (1), 9-41.
Sicat, Gerardo. “Reforming the Philippine Labor Market,” The Philippine Review of Economics.
Volume XLI, No 2, December 2004.
Sohn, Chan-Hyun and Junsok Yang 1998. Korea’s Economic Reform Measures under the IMF
Program. Seoul: Korea Institute for International Economic Policy.
World Bank (2004). Doing Business in 2004: Understanding Regulation. Washington, D.C. and
New York: World Bank and Oxford University Press.
World Bank 2006. Doing Business 2007: How to Reform. Washington, D.C.: World Bank .
World Bank 2008. Doing Business 2009. Washington, D.C. : World Bank Publications, 2008.
Yang,Junsok 2002. Update on Korean Economic Reforms and Issues in Korea’s Future
Economic Competitiveness Seoul: Korea Institute for International Economic Policy, Discussion
Paper 02-03.
ADDITIONAL INTERNET SOURCES:
http://www.allacademic.com//meta/p_mla_apa_research_citation/1/0/0/1/7/pages100174/p100174-1.php
http://www.state.gov/e/eeb/ifd/2007/80670.htm
http://www.lexadin.nl/wlg/
http://www.hrw.org/reports/pdfs/e/ethiopia/ethiopia913.pdf
47
Figure 1. Rigidity of Employment Protection Legislation across Selected Countries
0
0.5
1
1.5
2
2.5
1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
Zambia Botswana Ethiopia Philippines Jordan
0
0.5
1
1.5
2
2.5
1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
New Zealand Portugal
0
0.5
1
1.5
2
2.5
1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
China India Brazil
48
Table III.1
Country Scores on the Employment Law Rigidity Index (ELR) of Selected Countries Over Time
Country 1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
China 2 2 2 2 2 1.8 1.62 1.62 1.41
India 1.1 1.1 1.15 1.2 1.3 1.3 1.3 1.53
Jordan 1.7 1.7 1.7 1.7 1.7 1.7 1.7 1.46 1.52
Korea 1.44 1.44 1.44 1.44 1.55 1.6 1.36 1.19
New Zealand 0.13 0.13 0.27 0.48 0.48 0.48 0.48 0.48 0.51
Portugal 0.06 0.31 1.27 2.29 2.29 2.39 2.49 2.42 2.42
Brazil 2.06 2.06 2.06 2.06 2.06 2.06 2.4 2.4 2.1
Zambia 1.05 1.05 1.05 1.3 1.3 1.15 1.25
Botswana 0.9 0.9 1.0 1.0 1.3 1.3 1.05
Ethiopia 1.3 1.3 1.3 1.3 1.53 1.53 1.53
Philippines 1.4 1.4 1.6 1.6 1.6 1.69 1.69 1.61 1.80
49
Table V.1
The Extent of Labor Regulation, Legal Origins and Per Capita GDP
(Dependent variable: Level of LAMRIG, Labor Market Rigidity)
[1] [2] [3] [4] [5] [6] [7]
Botero et al. Xsection PoolOLS Pre1980 Post1980 OECD Non-OECD
Log Per Capita GDP -0.001 -0.0716** -0.00224 -0.160*** -0.0790*** 0.428*** -0.0419***
[0.0116] [0.028] [0.016] [0.033] [0.018] [0.038] [0.016]
Socialist Legal Origin 0.2943*** 0.698*** 0.956*** 0.839*** 0.839*** 0.788***
[0.0453] [0.11] [0.075] [0.081] [0.081] [0.078]
French Legal Origin 0.2474*** 0.477*** 0.621*** 0.669*** 0.569*** 1.109*** 0.453***
[0.0381] [0.066] [0.041] [0.069] [0.041] [0.091] [0.034]
German Legal Origin 0.1553** 0.474*** 0.335*** 0.192 0.535*** 0.476*** 0.333***
[0.0702] [0.12] [0.096] [0.16] [0.089] [0.15] [0.096]
Scandinavian Legal Origin 0.3865*** 0.901*** 0.562*** 0.503*** 1.128*** 1.074***
[0.0462] [0.11] [0.12] [0.16] [0.080] [0.092]
Constant 0.3072*** 1.860*** 1.067*** 1.983*** 1.826*** -3.211*** 1.566***
[0.1038] [0.24] [0.13] [0.24] [0.15] [0.34] [0.12]
Observations 85 142 792 324 468 222 570
R-squared 0.44 0.36 0.22 0.24 0.35 0.56 0.23
Notes: Robust standard errors in brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
50
Table V.2
Labor Market Reform, Legal Origins and Per Capita GDP
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5]
PoolOLS Pre1980 Post1980 OECD Non-OECD
Log Per Capita GDP -0.00214 0.00186 0.0380*** -0.0515*** -0.00813
[0.0058] [0.0069] [0.013] [0.017] [0.0050]
Socialist Legal Origin 0.03 0.0305 0.0305 0.0518**
[0.024] [0.025] [0.025] [0.024]
French Legal Origin 0.00364 -0.0169 0.0341 0.0491 -0.00478
[0.013] [0.015] [0.022] [0.030] [0.0088]
German Legal Origin -0.0153 -0.0603 0.0133 0.039 -0.0459
[0.029] [0.043] [0.032] [0.027] [0.033]
Scandinavian Legal Origin 0.0980*** -0.0933* 0.182*** 0.0691**
[0.036] [0.049] [0.050] [0.034]
Constant 0.0541 0.0167 -0.223** 0.514*** 0.0815**
[0.043] [0.053] [0.090] [0.14] [0.037]
Observations 721 397 324 220 501
Number of countries 133 133 90 23 110
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
51
Table V.3
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP and Structural Factors
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5]
Structural Factor: Income Gini Govt Share of Foreign Aid Natural Res Agric Share
GDP to GDP Exports (%) in GDP
Lagged LAMRIG -0.192*** -0.162*** -0.146*** -0.169*** -0.134***
[0.053] [0.019] [0.017] [0.019] [0.020]
French Legal Origin 0.0917** 0.103*** 0.0962*** 0.104*** 0.0945***
[0.043] [0.021] [0.020] [0.020] [0.023]
German Legal Origin 0.0746 0.0383 0.0304 0.059 0.0153
[0.066] [0.039] [0.035] [0.039] [0.032]
Scandinavian Legal Origin 0.168** 0.173*** 0.168*** 0.198*** 0.171***
[0.080] [0.023] [0.022] [0.031] [0.024]
Socialist Legal Origin 0.120** 0.169*** 0.152*** 0.178*** 0.210***
[0.049] [0.035] [0.041] [0.041] [0.027]
Log Per Capita GDP -0.0331*** -0.00199 -0.00464 -0.00731 0.00501
[0.013] [0.0060] [0.0060] [0.0056] [0.014]
Lagged Structural Factor -0.00114 0.000795 0.000442 0.000479 0.0325
[0.0017] [0.00066] [0.00088] [0.00031] [0.060]
Constant 0.567*** 0.203*** 0.220*** 0.261*** 0.122
[0.17] [0.053] [0.048] [0.048] [0.12]
Observations 150 600 539 599 415
Number of countries 101 122 120 125 96
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
52
Table V.4
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP and Economic Crises/Shocks
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5]
Economic Crises/Shocks: Debt Crises High Inflation Max Fall of
Years of
Negative
Current
Account
(>30% p.a.) GDP GDP Growth Crises
Lagged LAMRIG -0.181*** -0.234*** -0.161*** -0.156*** -0.147***
[0.031] [0.031] [0.018] [0.018] [0.016]
French Legal Origin 0.103*** 0.133*** 0.100*** 0.0978*** 0.101***
[0.027] [0.029] [0.020] [0.020] [0.021]
German Legal Origin 0.048 0.0924* 0.0405 0.039 0.0254
[0.044] [0.053] [0.037] [0.035] [0.031]
Scandinavian Legal Origin 0.172*** 0.274*** 0.171*** 0.170*** 0.166***
[0.045] [0.043] [0.022] [0.022] [0.033]
Socialist Legal Origin 0.209*** 0.283*** 0.183*** 0.178*** 0.172***
[0.042] [0.043] [0.040] [0.036] [0.056]
Log Per Capita GDP -0.0149** -0.0154** -0.00412 -0.00454 -0.00467
[0.0066] [0.0069] [0.0058] [0.0056] [0.0057]
Lagged Economic Crises -0.00119 0.02 0.000592 0.0104 0.0297
[0.012] [0.023] [0.0016] [0.011] [0.021]
Constant 0.349*** 0.408*** 0.234*** 0.228*** 0.215***
[0.075] [0.068] [0.047] [0.046] [0.047]
Observations 453 538 606 606 528
Number of countries 124 123 125 125 118
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
53
Table V.5
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP and Political Crises/Factors
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5] [6]
Political Crises/Factors: Democracy POLCON Assassinations Strikes International Civil war
conflict (war) (intensity)
Lagged LAMRIG -0.146*** -0.212*** -0.145*** -0.147*** -0.181*** -0.153***
[0.019] [0.028] [0.017] [0.017] [0.037] [0.018]
French Legal Origin 0.0889*** 0.122*** 0.0976*** 0.0976*** 0.0913*** 0.115***
[0.020] [0.026] [0.020] [0.020] [0.030] [0.025]
German Legal Origin 0.0471 0.0997** 0.0547 0.0561 0.137** 0.0930***
[0.031] [0.044] [0.035] [0.036] [0.060] [0.030]
Scandinavian Legal Origin 0.181*** 0.275*** 0.163*** 0.165*** 0.184***
[0.024] [0.044] [0.022] [0.022] [0.025]
Socialist Legal Origin 0.145*** 0.189*** 0.153*** 0.150*** 0.180*** 0.178***
[0.040] [0.048] [0.041] [0.041] [0.062] [0.058]
Log Per Capita GDP -0.000665 -0.00873 -0.00529 -0.00527 -0.00631 -0.000246
[0.0081] [0.012] [0.0053] [0.0055] [0.0085] [0.0058]
Lagged Political Crises -0.00231 -0.0534 -0.0129* -0.0133 -0.0092 0.00743
[0.0030] [0.060] [0.0071] [0.015] [0.0068] [0.0052]
Constant 0.203*** 0.350*** 0.227*** 0.229*** 0.286*** 0.175***
[0.048] [0.077] [0.045] [0.046] [0.065] [0.048]
Observations 570 538 567 567 316 485
Number of countries 119 122 120 120 76 93
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
54
Table V.6
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP, Trade and Financial Reforms
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5] [6]
Other Reforms
Wacziarg
Open C.N.Hsiao PWT openk BMP Credit Private Financial Ref
Uninterrupted Trade Lib Sector/GDP Index
Lagged LAMRIG -0.163*** -0.163*** -0.148*** -0.211*** -0.194*** -0.238***
[0.019] [0.019] [0.018] [0.025] [0.027] [0.028]
French Legal Origin 0.113*** 0.111*** 0.0979*** 0.130*** 0.130*** 0.145***
[0.022] [0.021] [0.021] [0.024] [0.031] [0.028]
German Legal Origin 0.0489 0.0556 0.0416 0.0649 0.0312 0.104**
[0.040] [0.041] [0.032] [0.052] [0.053] [0.050]
Scandinavian Legal Origin 0.167*** 0.163*** 0.165*** 0.265*** 0.232*** 0.284***
[0.022] [0.022] [0.023] [0.040] [0.048] [0.040]
Socialist Legal Origin 0.240*** 0.207*** 0.155*** 0.279*** 0.233***
[0.040] [0.046] [0.038] [0.030] [0.050]
Log Per Capita GDP -0.0213** -0.0196** -0.00396 -0.0147** -0.0137* -0.0161**
[0.0089] [0.0086] [0.0053] [0.0070] [0.0080] [0.0072]
Lagged Other Reforms 0.0827*** 0.0702** 0.00000975 -0.000003*** -0.000000120* -0.0277
[0.028] [0.027] [0.00013] [0.00000027] [0.000000072] [0.034]
Constant 0.336*** 0.326*** 0.219*** 0.371*** 0.347*** 0.437***
[0.066] [0.065] [0.044] [0.062] [0.077] [0.068]
Observations 588 580 579 514 326 538
Number of countries 114 120 117 108 84 118
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
doc_721490958.pdf
This paper investigates the extent and the determ inants of changes in labour market regulation in a large sample of developing and developed countries since 1960. The existing evidence tends to focus on the level of regulation (not on changes) and is often confined to the post11995 period (when not, it focuses on OECD or Latin America.) Our results for the 1995 cross section replicate previous findings: legal origins matter more than per capita GDP or institutional factors.
The Dynamics of the Regulation of Labour
in Developing and Developed Countries since 1960*
Nauro F. Campos
Brunel University, CEPR and IZA
[email protected]
Jeffrey B. Nugent
University of Southern California and IZA
[email protected]
This draft: October 2010
Abstract: This paper investigates the extent and the determinants of changes in
labour market regulation in a large sample of developing and developed countries
since 1960. The existing evidence tends to focus on the level of regulation (not on
changes) and is often confined to the post-1995 period (when not, it focuses on
OECD or Latin America.) Our results for the 1995 cross-section replicate previous
findings: legal origins matter more than per capita GDP or institutional factors.
Yet, when we consider changes in the level of labor regulation from 1960 to 2004,
the effect of legal origins diminishes greatly, while per capita GDP and other
structural reforms move to center stage. Our results suggest labour market reforms
are associated with higher levels of per capita GDP and of financial liberalization,
while trade liberalization tends to slow down labour market reforms.
Keywords: employment protection legislation, labour market institutions, reforms
* The authors are grateful for comments received on earlier drafts from Olivier Bargain,
Armando Barrientos, Andrea Bassanini, Tito Boeri, Suzana Brixiova, Ralitza Dimova, Werner
Eichhorst, Ioannis Ganoulis, Robert Holzmann, Prudence Kwenda, Giovanna Labartin, Hartmut
Lehmann, Joaquim Oliveira-Martins, Carmen Pages, Giovani Peri, Stefano Scarpetta, Kunal Sen,
Zahra Siddique, Jan Svejnar, Katherine Terrell and participants at fRD “Tracking Structural
Reforms” meetings (Milan), European Commission (Brussels), Manchester University, IZA
Topic Week on “Political Economy of Labour Market Reform in Transition and Emerging
Economies” (Bonn) and World Bank/IZA “Employment and Development” conference (Cape
Town). Jacob Burnett, Kannika Damrongplasit, Rob Lee, Ladan Masoudie, and James Ng
provided excellent research assistance. All remaining errors are our own.
1
I. Introduction
In recent years, an enormous literature has arisen concerning the construction of various
alternative indicators of labor market characteristics, most often concerning rigidity or
flexibility. These indexes have been based on various different kinds of measures: (1) measures
of actual activity, such as the extent of labor turnover, the number of strikes, labor force
participation rates, unemployment rates, (2) subjective opinion surveys of employers, workers or
other parties concerning job satisfaction, the rigidity of labor markets, the competiveness of
labor markets, and discrimination in labor markets, (3) tax wedges (distortions measured in
terms of the gap between what workers receive and employers pay) and (4) codified
characterizations of what the labor laws and other labor market regulations say with respect to
different kinds of constraints on employers, workers or intermediaries in the market. Within this
latter class of measures, some apply to individual worker rights, while others to collective rights
(such as the right to form and join a union) and in both cases, distinguishing between the laws
themselves and the practice or enforcement of these laws.
1
Each approach has advantages and
disadvantages.
The present paper attempts to extend the individual rights and law-based component of the
third approach in several respects, namely, by adding countries and more importantly extending
it backwards in time from the late 1990s for some countries to the early 1950s. This extension
allows us to study the dynamics of labor market reform across a fairly large number of both
developing and developed countries. Specifically, it does so for a single relatively
comprehensive measure of labor market rigidity based on comparisons of labor laws across
countries and over time. This index is an Employment Laws Rigidity index (ELR) intentionally
designed to be consistent with the seminal study of Botero et al (2004) for 85 countries in 1997.
1
See for example Kucera (2002) and Greenhill, Mosley and Prakash (2009).
2
Quite a few studies have attempted to up-date similar indexes for large samples of countries
(e.g., World Bank's Doing Business project, the EU’s LABREF). For going backwards in time,
however, studies have been limited to two regions (OECD as in Blanchard and Wolfers 2000 and
Allard 2005) and Latin America (Heckman and Pages 2004).
2
The index we develop is for
Employment Law Rigidity (ELR), a (de jure) index. Our extension increases the number of
countries to 145 for at least one time period and to approximately 100 countries for our panel
data analysis (5-year averages from 1950-54 to 2000-04.) The resulting dataset is referred to as
the Labor Market Rigidities (LAMRIG) dataset.
The largest part of the paper is Section II which describes how these indexes are constructed,
which varies somewhat by region and country but always in a way consistent with the Botero et
al study. This data construction exercise was made possible in large part by the on-line
availability of comprehensive databases of labour laws, such as NATLEX and LEXADIN.
3
In
Section III we make use of LAMRIG to describe the changes over time in the ELR indexes for
specific regions and countries. We use seven countries to illustrate both how the changes in the
ELR indexes have been calculated and to identify possible political economy determinants of
these changes.
Section IV is devoted to several exercises aimed at assessing the applicability of the index.
In order to assess the usefulness of this new index we conduct several exercises. First, we restrict
our analysis to the cross-section for 1995-1999 (the period coinciding with that in Botero et al.
2004) and repeat their analysis concerning the determinants of labor market regulatory rigidities.
2
In the case of collective relations laws, Greenhill, Mosley and Prakash (2009) have extrapolated the series put
together by Kucera (2002) back to 1986 for a sample of 90 developing countries.
3
One of our main sources was NATLEX, available at http://www.ilo.org/dyn/natlex/natlex_browse.home. It is
maintained by the International Labour Organization (ILO)'s International Labour Standards Department and
contains labour law records for more than 150 countries since the late 1940s. LEXADIN is a law database covering
various subjects (such as labor laws but also including electoral, commercial, civil, banking, etc) and is available at
http://www.lexadin.nl/wlg/legis/nofr/legis.php
3
For this cross-section, we fully replicate their results, demonstrating the greater importance of
legal origins than those of per capita GDP and/or political factors. Second, however, when we
extend the analysis to the panel and to changes over time (defining reforms as negative changes
in the ELRs we are thus able to examine the determinants of labor market reforms), our results
diverge from those of Botero et al. (2004). For example, when we use a random-effects model
with clustered standard-errors at the country level to explain labor market reform, the influence
of legal origins is much less significant, but still present. To the aforementioned models of ELRs
and reforms in ELRs that included the legal tradition measures, income and political variables,
we add some measures of factors suggested by our country case studies. These include measures
of economic structure, such as the share of agriculture in GDP, the share of natural resource
exports in GDP and income inequality, various measures of other types of economic reform and
market development, and finally both economic and political crises. Third, we test for the
relevance of other determinants of labor market reforms, such as economic and political crises,
structural factors and other structural reforms. Along with the reduced role of legal origins, we
find evidence suggesting that inertia, per capita GDP and other types of reforms all matter. In
particular, our results suggest that financial liberalization supports labor market reform, while
trade liberalization seems to slow it down.
The paper is concluded with Section V which contains our conclusions and suggestions for
future research. A detailed appendix of data sources and further details on the construction of the
ELR indexes contained in LAMRIG is available upon request as well as on-line.
II. Constructing the ELR Indexes Across Countries and over Time
The purpose of this section is to describe the methods used in constructing the Labor Market
Rigidities (LAMRIG) dataset. To that end it identifies the data sources for constructing an index
4
of rigidities “Employment Law Restriction (ELR) Index” that is consistent across countries and
over time. We begin with a brief overview and then go on to provide more details for the OECD
and Latin American regions for which take advantage of existing studies. This is followed by a
less precise but illustrative discussion of how the ELR indexes are constructed for countries
outside these regions.
A. Overview
To study labor market reforms, either determinants or effects, one needs time series data
on some kind of a labor market index that one thinks is relevant to the issues in which one is
interested, such as unemployment (duration or level), labor turnover, dualism, growth, or
structural change. Yet, because there is no single data set of any kind that covers the more than
100 countries studied here on a consistent basis for anything more than a few years
4
, the present
study makes use of data from several different major sources and many other country-specific
special sources for countries and years not covered by the major sources.
4
One labor index that may be considered close to what one might want in this respect is the one by Forteza and
Rama (2006) and Rama and Artecona (2002) based on ILO conventions signed by each country. This has good
coverage (more than one hundred countries and over time. But, since this index bases much on country’s having
approved of various ILO conventions on such social issues as non-discrimination in employment that are often not
adhered to in practice, few seem to have been persuaded that this is a useful index for examining unemployment and
other issues. It also has the disadvantage of having almost no variation over time once these conventions had been
signed by the individual country (which in many cases was quite early). With respect to the rules governing unions
and collective bargaining, Kucera (2002) takes advantage of data from a number of other sources such as the
International Confederation of Free Trade Unions, the US State Department’s Country Reports on Human Rights
Practices, and ILO Reports of the Committee on Freedom of Association to construct a Freedom of Association
index based on 37 different indicators. The individual indicators are then weighted by their assumed relative
importance and then aggregated the weighted sum into an index representing the average of such scores for the
period 1993-7. Greenhill et al (2009) distinguish between the indicators which pertain to the laws and those which
pertain to practice (captured by the reported numbers of law violations) and then extend the two sets of indexes
backward to 1986 and forward to 2002 for developing countries. Neither of these indexes is utilized here since the
rights with respect to creation and participation in unions need not necessarily relate to flexibility/rigidity in the use
of labor. Interestingly, however, these authors use bilateral trade patterns to show that the level of these freedom to
organize indexes (especially the laws-oriented) can be linked to the extent to which the individual developing
country exports to developed countries with higher labor standards. Another source for measuring the degree of
regulation of labor markets is now Canada’s Fraser Institute. Since 1975, this institute had been scoring countries
on a number of sub-indicators of economic freedom, such as strength of property rights, freedom from price and
wage controls, restrictions on trade, financial transactions and product markets. These were then aggregated into an
overall index of Economic Freedom. In 2001 the Institute began to include scores on six additional subcomponents,
all relevant to measuring the freedom of labor markets. While at first this was limited to 58 countries, the country
coverage has grown somewhat over time.
5
To our knowledge, there are only a very few labour laws rigidity indexes relevant to
unemployment, the size of the informal sector and related phenomena that have reasonable
cross-country and over time coverage going back from the present to the late 1980s or beyond.
Aside from the Forteza and Rama index if ILO Conventions, almost all of these, e.g., Blanchard
and Wolfers (2001), OECD (2004), Allard (2005a) do so only for developed countries. These
studies built upon a whole series of earlier attempts (e.g., Lazear 1990, Grubb and Wells 1993,
Addison and Gosso 1996, Nickell 1997, Layard and Nickell 1998 and OECD 1999) to construct
such an index for developed countries. The Blanchard and Wolfers (2001) study constructs a
series “NEWEP” for 26 OECD countries going back from the 1995-99 period (more exactly
about 1996) to the 1960s in five year intervals.
5
We have used that one in the past but for greater
consistency over time, availability on an annual basis and slightly longer backward coverage, for
the 21 countries she has studied we have now switched to the estimates of Employment
Protection Legislation (EPL) of Allard (2005a). They are in principle comparable to those of
OECD (2004) but exclude two minor subcomponents ( delay in the notification and
compensation for unfair dismissal), for which information could rarely be found in the
legislation. For the remaining five OECD countries not covered by Allard but covered by
Blanchard and Wolfers, namely,Iceland, Korea, Luxemburg, Mexico and Turkey, we have based
on our indexes on Blanchard and Wolfers (2001), though in the case of Korea also on various
other sources as discussed below.
The other multi-country sources with some time coverage as well are Heckman and
5
Nickell et al (2003) have annualized the Blanchard and Wolfers series. More recently, the European Union (EU)
has constructed a somewhat similar set of indexes called the Labor Market Reform Database (LABREF) with
somewhat more detail on certain policy-related aspects of labor legislation, but only for each year between 2000 and
2006 for each EU member. These labor market reform indexes include pension, labor taxation and other aspects.
Both Arpaia et al (2007) and Bassanini and Venn (2007) describe the indexes and relate them to different effects on
labor. Arpaia (2007) focuses on the effects of the indexes on labor market participation (of all workers but especially
of older ones) while Bassanini and Venn (2007) examine the effects of the indexes on labor productivity.
6
Pages (2000 and 2004) which cover most countries of the Latin America and Caribbean (LAC)
region, going back from the late 1990s only to the late 1980s. For the most part, the LAC
indexes are available primarily only at intervals a decade apart, not annually. For this reason, we
extend these indexes backwards and forwards with use of information on the labor laws and
other studies where available.
As has been noted in many surveys, e.g., Bertola (2008), Freeman (2008), Djankov and
Ramalho (2008), the data on countries outside of these two regions is much more limited in time
coverage. Indeed, for them we had to base our data construction work on a wide variety of
sources but with methods designed to be as consistent as possible with those used for the OECD
and LAC regions. We will come to these regions after describing how the OECD and LAC
regions are dealt with in greater detail.
Even for the OECD and LAC data sets, their comparability is made more difficult by the
fact that, although similar in spirit, the Heckman and Pages (2000 and 2004) Job Security Index
and Allard (EPL) are built up from sources, methods and index aggregation procedures that are
by no means identical.
The Heckman and Pages Job Security Index (JS) is defined as the discounted value of
dismissing a worker at an expected date in the future based on the likelihood and costs of
dismissal implied by the labor laws and regulations (but excluding the costs of court actions). It
makes use of a common discount rate of 8 percent, an assumed turnover rate of 12 percent
and
the country and period-specific cost (inclusive of those related to seniority) of dismissing a
worker for either justified or unjustified reasons.
As indicated above, Allard (2005a) made use of 16 of the original 18 aspects of EPL used
in OECD (2004) but obtained the data, not from questionnaires cross-checked with the
individual countries as in the OECD study, but rather from direct examination of the laws
7
themselves based on ILO’s NATLEX supplemented with OECD sources. The 16 indicators were
aggregated first into three separate indicators (laws protecting workers with regular contracts,
those affecting workers with fixed term (temporary) and regulations applying to dismissals), and
then into a single EPL index. Both the scoring of the sub-components and their weighting into
the various components and indexes have been controversial since virtually any method is
subjective.
6
While the scales of the indexes (the EPL of Allard and NEWEP of Blanchard) are
almost identical ranging from 0-4, that of EP of Heckman and Pages) is quite different, the latter
ranging from 0-18. Unfortunately, none of these indexes reflects by any means all of the labor
market institutions (such as wage flexibility, team production, job rotation, social dialogue,
pension plans of different types, and workers use of the courts) that one might think would
exercise influence on economic outcomes of various sorts (Freeman 2008).
7
Yet, as indicated
both above and further below, each of them captures a number of important (largely common)
dimensions of labor regulations and thus may be regarded as a measure of the restrictiveness of
labor laws and regulations for firms in their use of labor. Since both also allow internally
consistent comparisons over time, we deem it valuable to make use of them together.
To mitigate the problem of the differences between the Allard-Blanchard and Wolfers
and Heckman-Pages indexes, we make use of another source, namely Djankov et al (2003)
revised as Botero et al (2004), that uses a closely related methodology for constructing an
employment laws restrictiveness index (ELR). Their ELR has much greater country coverage
(originally 85 countries) but only for a single time period 1997. This index is constructed as the
total of the scores on three different subcomponents, alternative employment contracts (part-
6
Indeed, as shown by Addison and Teixeira (2003), the various variants of the aggregate indices that have arisen are
not always highly correlated and their application to issues like unemployment rates has sometimes resulted in
opposite findings. These and other authors also point out that what is relevant in constructing these indices may also
vary from industry to industry.
7
Allard (2005b) creates for the same 21 OECD countries in her 2005a EPL indexes of unemployment benefits
based in part on tax treatment and subsidies with duration and the conditions for qualification.
8
time, fixed term etc.), conditions of employment (mandatory rest, maximum hours of work
without overtime, overtime pay premium, leaves for holidays and maternity, etc.) and job
security (restrictions on dismissal, mandatory notice periods, severance payments). In this
respect the ELRs of Botero et al are somewhat broader in scope than the other studies. Because it
has scored each of 85 countries, including almost all of the 21 countries for which we have used
Allard’s ELR,
8
3 of the 5 other countries for which we have used NEWEP from Blanchard and
Wolfers and the 21 LAC countries
9
covered by Heckman and Pages (2000, 2004) for the year
1997, we use the ELR of Botero et al (2004) in order to provide consistency between the indexes
coming from each of the other sources in the LAMRIG database.
10
A major advantage of this
index is that it comes with a complete matrix of how the authors scored each country on each of
the 31 subindicators on which their ELRs are constructed. These details facilitate matching of
their evaluations for the year 1997 with information concerning changes in the laws over time
from NATLEX and other sources. The fact that the studies with over time coverage on JS and
EP do not include our indicator for the working conditions component of the Botero et al ELR
indexes is not crucial since these typically change less frequently over time than other
components of the labor laws.
Each of the three subcomponent scores in ELR is in turn an average of scores on a 0-1
8
These countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland,
Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and
United States.
9
The LAC countries covered by Heckman and Pages include: Argentina, Barbados, Belize, Bolivia, Brazil, Chile,
Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guyana, Honduras, Jamaica, Nicaragua,
Panama, Paraguay, Peru, Trinidad and Tobago, Uruguay, and Venezuela. They also included Mexico but for this
country we have used Blanchard and Wolfers because of its more complete time coverage.
10
Another reason for choosing the broader ELR index of Botero et al (2004) as the lynch-pin for our construction
of LAMRIG is that these authors have shown it to have important consequences for labor market outcomes, perhaps
stronger ones than have been obtained by other authors. For example, in their recent recapitulation of their results
and attempt to rebut the criticisms raised of their study, La Porta et al (2008) have shown that a two standard
deviation increase in ELR implies a 1.99 percentage point reduction in the male labor force participation rate, a 2.32
percentage point increase in the unemployment rate, and a 5.67 percentage point increase in the unemployment rate
of young males.
9
scale on the list of indicators, many of which are “yes” –“no” answers as to whether a certain
regulation exists or not. Hence, the maximum score is 3 and the minimum 0. Moreover, the 85
country sample of Botero et al (2004) has subsequently been extended in the closely related
Rigidity of Employment (ROE) Indexes developed in the World Bank’s Doing Business Surveys
so that, for subsequent years 2003 and 2007 at least, an additional forty or more countries can
potentially be added to the sample. The ROE indexes are, like the Botero et al (2004) ELR
indexes, based on three subcomponents, namely, difficulty of hiring, rigidity of hours, and
difficulty of firing. Each of these depends on most of the same indicators as in Botero et al
(2004). In this case, however, each of the subcomponents scores is on a 0-100 scale and since the
aggregate ROE index is an average of these, it, too, is on a 0-100 scale. Hence, while
methodologically and substantively the two indexes are almost identical, their scales are
different: (0-100 for ROE instead of 0 – 3 for ELR). As described in greater detail below, the
difference in scale was overcome by establishing an average conversion factor between each of
these Doing Business publications and numbers from the Botero study. The Allard (1995a) EPL
scores use a 0-4 scale and hence also have to be converted to Botero’s 0-3 scale and normalized
so as to be consistent with the Botero figures for 1995-9.
Both the country-specific Heckman and Pages Job Security Indexes and the Blanchard-
Wolfers NEWEP indexes are then converted into indexes with bases 1995-9 = 1.0. Using the
NATLEX data, the over-time variations in these indexes are then applied to the country-specific
1995-9 values in the Botero et al (2004) to construct over time variations in the country-specific
ELR indexes for the countries included in the OECD and LAC samples. Since they were roughly
on the same scale, the Allard indices were left as they were but aggregated from annual scores to
their five-year averages. A similar procedure is applied to the more fragmentary evidence of
over-time changes in the relevant components of labor laws for the remaining countries in the
10
samples afforded by the Botero et al (2004) and subsequent Doing Business Surveys. Below we
describe how these indices were also updated to 2000-4.
While others may wish to keep the various sub-indexes separate for use in different kinds
of application, for the present purposes we keep the focus on a single broad indicator of the
restrictiveness of employment laws. To avoid getting bogged down in the various possible
weighting systems, moreover, we simply accept the equal weighting of the three sub-components
as suggested by Botero et al. (2004).
The end result is an incomplete panel of Employment Laws Restrictiveness (ELR)
indicators for well over 100 countries measured as 5-year averages ranging from 1950-54
through 2000-04, a dataset we call LAMRIG for Labor Market Rigidities. For some years there
are as many as 145 countries with ELR scores. As has been pointed out by quite a few analysts
(e.g., Eichhorst et al 2007, Freeman 2005, 2008), whether higher scores are looked as desirable
or undesirable is subjective. For example, employers associations and individual employers
typically view them as harmful to investment, employment, productivity and the long run
interests of workers. But, those supporting labor interests and those interested in “public
welfare” and human rights often see them as good, helping to increase the legitimacy of working
outside the home for individual workers and thereby creating larger and better organized labor
markets. Others (Boeri et al (2000), Nicoletti et al (2000) and Amable et al. (2007)) view the
“goodness” or “badness” of such indexes to be more complex, depending on the identity and
magnitude of other market imperfections, regulations and so on. We are agnostic on this, but
given considerable evidence suggesting that higher scores are associated with higher informality
or unemployment rates and lower labor force participation rates, we do use the term “reform” to
refer to a reduction in these indexes and “reform reversal” to refer to an increase in these
indexes.
11
B. More Detail on Constructing the Indexes for the OECD and LAC Regions
As indicated above, for the 21 OECD countries covered by Allard (2005a), her series
were used to construct an index from 1950-4 through 2000-4 and then converted to a 0-3 scale so
as to coincide with those of Botero et al (2004) for the period 1995-9. For three of the OECD
countries covered by Blanchard and Wolfers but not by Allard (Korea, Mexico and Turkey), the
Botero numbers were extrapolated backwards to 1960-4 based on an index with 1995-9 = 1
constructed from Blanchard and Wolfers NEWEP, although as described below additional
information from other studies has been used in the case of Korea.
Two countries covered by Blanchard and Wolfers but not Allard, Luxemburg and
Iceland, were not covered in Botero et al (2004). They were, however, covered in the World
Bank’s Doing Business 2008 (pertaining to 2007) as “Rigidity of Employment Index”
11
. This
index was scaled differently but based on similar methodology to the Employment Laws Index
of Botero et al (2004) as explained in Doing Business. The values from the Doing Business scale
for these two countries were then extrapolated backwards to 1995-9 on the basis of information
in NATLEX or LEXADIN website and other sources and converted to those of Botero et al
(2004) scale on the basis of the conversion factor found for neighboring countries. The resulting
values were then interpolated back from 1995-9 on the basis of the Blanchard and Wolfers
(2000) data base for NEWEP.
As indicated above, for Latin American and Caribbean (LAC) countries for which 1995-
1999 values of ELR were available from Botero et al (2004) (other than Mexico which was
covered in Blanchard and Wolfers (2000)), the interpolation backwards was based on the Job
Security Index of Heckman and Pages (2000), certain refinements thereof for the dates of
11
This index combines sub-indexes for (1) alternative employment contracts, (2) conditions of employment, and (3)
job security. See also World Bank (2004).
12
reforms contained in Heckman and Pages (2004) and information from the NATLEX and
LEXADIN websites, and other sources identified in the Appendix.
12
Specifically, the Heckman
and Pages numbers were calculated as the total costs of firing a worker relative to wages in 1987
and in 1999. We calculated the ratio between the 1987 number and the 1999 number for each
nation in the Heckman and Pages study, and used this as an index to extrapolate the index
number in the Botero study backwards to the e 1985-1989 period for each nation. For the 1990-
4 value we made use of NATLEX and other studies to identify changes and if these occurred to
approximate the magnitude based on the relative importance of the items changed based on
NATLEX or LEXIDIN. Similarly, NATLEX was also used to extrapolate the ELR indexes
backward in time in a way consistent with the Botero (2004) matrix of the scores assigned to
each of the 31 components of ELR.
For those LAC countries (Costa Rica, El Salvador, Guatemala, Haiti, Honduras,
Nicaragua and Paraguay) for which there was no observation for the Employment Laws Index in
Botero et al (2004), but there did exist a score on the closely related Rigidity of Employment
Index for either 2003 or 2007 from the World Bank Doing Business volumes for 2004 and 2008,
a similar procedure was used as that described above for Iceland and Luxemburg to interpolate
backwards to 1995-99. For Chile, even though over time values were included in the Heckman
and Pages (2000) study, because of its greater detail and longer time coverage, the index was
interpolated back to 1960 based on the Job Security Index data presented by Montenegro and
Pages 2004 (Figure 7.1). Wherever possible, these indexes for Latin America and the Caribbean
were interpolated backwards from there to the late 1980s based on Heckman and Pages (2004)
12
This index was computed as the expected discounted cost at the time a worker is hired of dismissing the worker at
some time in the future based on existing labor law (but excluding the costs of court actions). It makes use of a
common discount rate of 8 percent and assumed turnover rate of 12 percent
12
and the costs (inclusive of those
related to seniority) of dismissing a worker (for justified and unjustified reasons).
13
and to earlier years based on the various provisions in the earlier employment laws from
NATLEX and LEXIDIN. In cases where there was no new Employment Law between dates
covered, such as in Haiti between 1984 and 1995-99, the resulting index values were assumed to
remain constant between those dates. For LAC countries not included in Heckman and Pages for
which Djankov et al (2003) or Botero et al (2004) was available or could be constructed based
on conversions from the subsequent Doing Business surveys, namely, Guatemala and Haiti, the
values were interpolated backwards solely based on changes in the provisions of the employment
laws as reported in NATLEX and LEXIDIN. Other LAC countries, Belize and Suriname had to
be dropped from the present study for lack of data.
To repeat, the 1995-9 values for all these countries have been constructed in such a way
as to be consistent with what the Botero et al (2004).
C. Constructing the Indexes for Countries in Other Regions
As indicated above, for countries outside of the OECD and LAC regions, data on
employment rigidities are much less complete and rarely if ever already developed into an index
over time. Some early studies identifying the effects of employment laws were Fallon and Lucas
(1991, 1993). They identified law changes in both India and Zimbabwe that had the effect of
tightening labor regulations and claimed that in both cases the result was lower formal sector
employment of industrial labor.
Once again, our first step is to make use of the Botero et al (2004) and subsequent Doing
Business surveys for 2004 and 2008, for arriving at values of the index for1995-99 that would be
comparable to those for the Botero et al (2004) indicators. In these cases, we went country by
country, making use of NATLEX, LEXADIN and published papers covering the country in
order to interpolate the indexes backward in time to the extent possible.
For example, Gerardo Sicat’s article “Reforming the Philippine Labor Market” provided
14
us with ample data of Philippine labor regulations. This source discusses changes in labor
law/regulation and the actual impact they have on labor regulation (and the ease of employing
workers) – whether they make regulation stricter or looser. Starting with our 1999 Botero
number, we assessed the impact of previous changes in labor regulation and worked our way
backward.
For India we made use of Dutt (2002), Dutta Roy (2004), Besley and Burgess (2004) and
Pages and Ahsan (2008) and other materials identified in Section IIIB below. The latter two of
the studies identified above make use of state-specific changes to the federal-level Industrial
Disputes Act of 1947. This is relevant because in India’s federal system states are also granted
the power to regulate industry, labor, health and other matters. A problem with the state level
data is that some states were liberalizing while others were tightening regulations making it
difficult to aggregate them into all-India changes. As noted below, we did so very crudely based
on the number of states moving in either direction, the magnitudes of these changes and the sizes
of the respective states. Note also that Bhattachajea (2006) has criticized Besley and Burgess
(2004) and Pages and Ahsan (2008) though more for their analysis of the effects than for the
scoring of the amendments. It should also be admitted that most of these indexes for India
pertain exclusively to manufacturing (and even within manufacturing there may be differences).
For the remainder of our countries (primarily Africa, Asia and the Middle East), the International
Labor Office’s NATLEX database provided us with the majority of our data. Similar to our
process for the Philippines, we gauged the effect of previous changes in labor regulation and
adjusted Botero ELRs across different time periods accordingly. In each such case, we gain
made considerable use of the appendix materials for Botero (2004) containing the scores
assigned to the countries included in that study for all 31 subindicators. For the nations for which
a 1995-1999 Botero ELR did not exist, we used the subsequent rankings and indexes on ease of
15
hiring and firing workers in the Doing Business Surveys of 2004, 2007 and 200 as described
above for OECD and Latin American countries. There were, however, some instances where we
found inconsistencies between Botero et al (2004) and Doing Business evaluations that could not
be attributed to labor law changes and therefore did not use this method of conversion. For Iran,
for example, which was not included in Botero et al (2004) but was included in subsequent
editions of Doing Business, we have made use of the relevant components of the index of Labor
Market Flexibility by Gholam Reza (2009) which covers the entire period 1960-2006 that also
includes minimum wages and other unemployment insurance requirements. For a few countries
that have received considerable attention by individual scholars or international organizations,
we have made considerable use of these studies. Illustrative examples of this are given in the
following section.
III. Some General Patterns and Illustrative Country Examples
As has been widely observed, labor market regulations tend to be much more static over time
than other types of regulations. This may well be attributable to the numbers of parties that are
generally involved in making changes to labor regulations (firms of different ownership types,
sizes, sectors, workers of different types, managers of different backgrounds, government
bureaucrats from different ministries, labor unions of different types and sizes, employers
organizations, intermediaries between business and labor, and the judiciary).
Despite the fact that there are quite a few countries that have experienced little or no change
in their ELR scores over the entire period, in each region, there are also countries whose scores
have changed from one 5 year period to another, resulting in some interesting differences over
time as well as across countries and regional or other groupings. Section A examines some of
these patterns of change in regional or other groupings. Section B identifies some changes over
16
time in some individual countries that are in neither the OECD nor the LAC regions and thus
with over time changes in labor regulations not covered in the major existing sources identified
in the previous section.
A. Some Regional Patterns
On average, the change in the average of all country ELR scores in the LAMRIG data base
over the period 1960-4 to 1995-9 have been modest. There were some notable changes in the
1950s and especially in the last ten years. At the regional level, however, there has been more
volatility, but again primarily in the 1950s and 1960s and again after 1995. This is reflected in
Figures 1 and 2. An important caveat to interpretation of both Figures 1 and 2 is that, except for
the high income high-middle countries of Europe, the early year portions of these curves are
based on very few and not necessarily representative countries.
Figure 1 presents the average ELR scores over the same period of time for six regional
groupings, namely, Latin America including the Caribbean(LAC), Asia, “Europe” (defined to
include, Canada, the US, Australia and New Zealand), Sub Sahara Africa (SSA), Transition
Countries of Central and Eastern Europe and Central Asia and the Middle East and North Africa
(MENA). In this case, there are two regions, Asia and MENA, with very little change in the
average ELR scores over time. Asia exhibited very little change over the whole period,
remaining below 1.5 for the entire period before 2000-4 at which time the average ELR has
edged upwards. While MENA’s index revealed little change except for the noticeable increase in
the 1950s. Its ELR also remained near but slightly below 1.5 for the entire period. In contrast,
SSA’s average ELR has risen rather steadily from below 1 in 1950-4 well over 1.5 in 2000-4.
The ELR average for “Europe” rose even more steeply from less than 0.5 in 1950-4 to well over
1.5 in the early 1990s before dropping off very slightly since then. By contrast, LAC’s ELR
started well above 1.5, rose further in the 1960s but has declined to only about 1.2 by 2000-4.
17
The ELR for Transition region started at about about 2.3 in 1060-4 but has fallen to a little below
2 since 1965-9.
Figure 2 presents the average ELR scores in LAMRIG over time for four different levels of
per capita income, low, lower middle, upper middle and high. Reminiscent of the pattern for the
“Europe” region in Figure 1, the ELR average for high income countries started from an
extremely low level of just over 0.4 in 1950-4, rose rather steadily to well over 1.6 by 1985-9
before leveling off to about 1.6 in 2000-4. The low income and lower-middle income countries
started the 1950s with quite a bit higher ELR indexes but experienced quite sharp increases in
their respective ELRs in the 1950s, leveled off in the 1970s and 1980s before rising further in the
last decade. The ELR of the lower-middle income countries in 2000-4 was 1.8 the highest of off
the groups. By contrast, the ELR of the upper-middle income countries displayed a rather
inverted U pattern, starting with the highest level of any region in 1950-4, rising further to 1.6 by
the mid-1960s but the falling gradually to about 1.4 by 2000-4.
B. Some Illustrative Country Cases
Given (1) the varying sample sizes of the region averages reported in Figures 1 and 2, (2)
the earlier observation that many countries experienced little or no change in ELR scores over
time and (3) the possibility that the ELR scores of different countries within any one of these
regional or income groupings may move in different directions resulting in little change in group
average scores, in this section we turn to some individual country experiences.
In their exposition of their indicators of the restrictiveness of labor laws, Botero et al
(2004) illustrated the indexes and the relevance of differences in legal tradition by comparing
New Zealand and Portugal, two countries at fairly similar income levels(at least in the late
1990s) but different legal traditions and ELR scores in 1997. In particular, Portugal was an
example of a country with French Civil Law background and a high ELR Index of 2.36 (3.7 on
18
Allard’s EPL index) while New Zealand was an English Common Law country with a low ELR
index of 1.06 (0.7 on Allard’s EPL index). They illustrated the sizeable differences between
these countries with reference to the various subcomponents of their index, making the sizable
difference between their respective overall index scores understandable. As shown in Table
III.1, in our LAMRIG data base, New Zealand has had a low ELR score of 0.48 (based on our
conversion of Allard’s EPL to the 0-3 Botero scale) for the entire time between 1975-9 and
1995-9 before rising slightly to 0.50 in 2000-4. In the 1960’s, however, its ELR score was lower
still at 0.14. Portugal, by contrast, had its high ELR score of about 2.4 ever since 1985-89.
Notably, however, in the early 1950s and even in the 1960-4 period, Portugal’s score of 0.066
was slightly lower than that of New Zealand’s at that time. Clearly, if the 1950s or 1960-4 scores
had been used, this comparison would not have served the purpose of showing that the French
civil law tradition gives rise to greater restrictiveness in labor legislation than does the common
law tradition. Moreover, with such sizeable changes in relative rankings over time, it is unclear
why the legal tradition should matter much since the legal tradition almost never changes over
time. As indicated above, the changes over time in OECD and Latin American countries have
already been rather extensively documented by the several already cited studies done on these
regions.
Therefore, to illustrate some other interesting differences in ELR scores over time, also in
Table III.1 we show the ELR index values for seven countries from outside of these regions, two
giants from Asia: India and China, another large country from Asia; Philippines, a medium sized
country from SSA: Ethiopia, and three small countries Botswana and Zambia from SSA and
Jordan from MENA. Clearly, China’s ELR started high with a score of 2.5 (the highest of all the
countries in the table in 1960-4 ) but it declined in the late 1970s, again in the late 1980s and still
again in each of the next three periods, reaching a relatively moderate ELR of 1.42 by 2000-4.
19
By contrast, Ethiopia, India and especially Philippines have seen their ELR averages rise over
time so as to have the next highest scores behind only Portugal by 2000-4. The ELR scores for
the other three countries have displayed more interesting patterns. Jordan’s was steady at a
relatively high value of 2.7, before falling substantially in 1995-9 and then rising again slightly
in 2000-4. Botswana’s ELR index started very low at 0.9 in 1970-4, rose gradually to 1.3 in the
1990s before falling to 1.05, the lowest of all countries except New Zealand in 2000-4. Zambia’s
ELR has fluctuated a bit more but remained fairly low over the whole period.
To obtain some insight into the changes in some of these countries and again the role of
legal tradition, in the following subsections we will provide short political economy accounts of
four of these countries, two (India and Zambia) classified as English Common Law countries,
Jordan classified as French Civil Law and China classified as having a Socialist legal tradition.
In the Botero et al (2004) indexes for 1997 and hence the 1995-9 period in LAMRIG, these four
countries all had scores in between those rather extreme scores for New Zealand and Portugal.
India
The main legislation concerning labor regulations in India dates back to two important
laws at more or less the time of the country’s independence, the Industrial Employment Act of
1946 and the Industrial Disputes Act of 1947. A third important act is the Trade Union Act
which dates all the way back to 1926. The stipulations of this latter act have changed little over
time and been quite accommodating to unions and union activity. While the Industrial
Employment Act allowed for layoffs on grounds of financial necessity, as well as for medical or
disciplinary reasons, when financial reasons are given as the justification, the Industrial Disputes
Act comes into play. While the latter act was amended on various early occasions (1964, 1965
and 1971), the most important amendments were those of 1976, 1982, and 1984.
The 1976 amendment made prior approval by the government of any such layoffs
20
mandatory. While at that time this provision applied only to firms with 300 or more workers, in
1982 the provision was changed to apply to all firms with 100 or more workers. Employers who
violated these regulations could be fined heavily and forced to reinstate the dismissed workers.
These changes are reflected in the increase ELR index for India in Table III.1 from 1.1 to 1.15 in
1975-9 and 1.2 in 1980-4. Under India’s federal constitution, its states also have the power to
impose regulations of these types. Indeed, several studies, Besley and Burgess (2004) and Ahsan
and Pages (2008) have made extensive use of state level amendments to these laws.
13
While in
each period there were some states that amended their laws in ways that were more restrictive
and others that amended in a less restrictive direction, during the late 1980s the number of states
classified as increasing restrictions outnumbered and were generally more important states than
those liberalizing the laws. This accounts for the final increase in the index from 1.2 to 1.3 in
1985-9. After this, despite liberalization of various other regulations, the labor laws of India
were not appreciably modified as noted by Saha (2007). But, on the other hand, the 2004 Doing
Business reported a considerably higher score which converts to 1.53 on LAMRIG.
Saha (2007) also cites an estimate that only 10 percent of the requests to government for
layoffs on financial conditions were approved. Hence, the inability to lay off workers could be
regarded as very restrictive. The reason why, even at the end of the period, India’s scores on this
index are not higher is that in terms of the restrictiveness of work hours, overtime and overtime
pay, and to a lesser extent alternative types of contracts, India has low restrictiveness scores.
China
As indicated above, China had the highest ELR index of all the countries included in
Table 3 and indeed was tied (with Panama and Poland) for the highest in the entire LAMRIG
data set for 1960-64. At that time, China not only had highly regulated labor market markets by
13
For some criticism of the studies and the conclusions drawn from them see Bhattacharjea (2006).
21
virtue of the labor law, with industrial employers, then largely state enterprises, unable to lay off
workers under virtually any circumstances and providing life-time employment, but also workers
had virtually no physical mobility thanks to the tightly controlled household registration system
(hukou) that had been set up in the mid-1960s.(Brooks and Tao, 2002, Reutersward, 2005) This
system restricted households, by province, within province by local community and rural-urban,
and sector (agriculture and non-agriculture). Not only could workers not be dismissed but also
had to be given housing and other benefits. Workers had little to no choice as to where they
would work.
Only beginning in the late 1970s did employers have a right to offer incentives to
workers such as bonuses for productivity (Brooks and Tao, 2002). In 1986 and especially in
1988 with the People’s Industrial Enterprise Act and the Regulation of Private Enterprise Act,
both of 1988, labor regulations were further lightened. Employers wanting to terminate workers
for financial reasons were from that time only forced to consult with authorities and labor unions
over the layoffs but were not required to receive approvals. At the same time, fixed term
contracts were introduced on a limited scale, especially for foreign and non-state enterprises
which were by then being encouraged.
In 1992 enterprises were allowed to set their wage rates in an autonomous way as long
as they stayed within certain government -set required bounds. Even SOEs were allowed to be
closed down on financial grounds. The 1994 Labor Act allowed for collective contracts, floating
wage rules and layoffs for financial, disciplinary or medical reasons. This was facilitated socially
by the state itself assuming responsibility for the re-training and financial support of laid off
workers (OECD 2010). More widespread approval was also given to the use of fixed term
contracts and employers were allowed to terminate a worker prior to the completion of the fixed
term for financial reasons, subject only to 30 days notice and specified compensation. Firms
22
were increasingly relieved of their requirement to provide housing for their workers. Also in the
1990s the hukou system was relaxed in small towns at least in certain provinces so as to allow
more migration from agricultural to non-agricultural jobs and even from rural to small town
locations (Brooks and Tao 2002). Quotas on the number of hukou changes in a given jurisdiction
were also abolished in some locations (OECD 2010). In late 1999 a Contract Law was passed
which came into force the following year. Although this law did not deal directly with
employment contracts (Cooney et al 2007), it may have indirectly through allowing for labor
contracting within service contracts. Beginning in 2002 laid-off workers no longer received the
retraining in Re-employment Centers under the auspices of the state enterprises, receiving
instead only unemployment benefits. Fixed term contracts were allowed to (and in fact were
reported to) increase rapidly in the 2000-4 period suggesting, perhaps why the 2004 Doing
Business assigned a lower score on the Employment Laws Index during this period equivalent to
1.42 on the Botero et al (2004) ELR. We have adapted that figure even though as Cooney et al
indicate, the next new Labor law, the Labor Contract Law, did not come into effect until 2007.
These successive changes in the relevant labor laws and regulations of China account for
the indicated reductions in the ELR index for China in Table III.1 from 2.5 in 1960- for all the
periods through 1980-4 to 1.8 in 1985-9 and 1.62 beginning in 1990-4. These changes were no
doubt associated with the increasing market-orientation of Chinese leaders and the accumulating
evidence that these liberalizing economic reforms were paying off. There was also gradual
increase in the relative importance of town and village enterprises and subsequently international
joint ventures and private enterprises in which inter-firm and international competition must
have helped to generate pressures for liberalizing labor markets. In the process the importance of
firm-provided housing and residence based provisions of food and other benefits also
diminished, and as noted above a proliferation of new types of employment contracts.
23
Jordan
Jordan is different from all the other countries discussed in this section in that it is a
monarchy. Monarchies, especially those of relative short duration like Jordan’s and where the
monarch has considerable political power, face the problem of how to maintain legitimacy in the
eyes of its citizens without elections for the executive and democracy which could limit their
power. The provision of economic security is one important means for endowing itself with a
sense of legitimacy on the part of its citizenry. Lacking the funding for a full-fledged social
insurance system, for most of its post-WWII history Jordan’s monarch-led government has
attempted to do this through its government and public enterprises providing rather well-paying
jobs to many of its citizens with reasonably generous retirement schemes and job protection.
Not surprisingly, over much of this period, most Jordanian public enterprises were known to be
over-staffed.
The country’s Trade Union Law of 1953 confirmed the right of workers to form unions
and the right to strike. This law was amended on several occasions but with only minor import,
e.g., in 1956 to prohibit civil service workers from unionizing or striking, and in 1976 to limit
the number of labor unions. From the detailed subindexes for Jordan’s overall score on the
Botero index, Jordan is rated very restrictive in the use and cost of overtime, the hours of work
and pay scales being treated quite similarly to those in the state bureaucracy. The Labor Code of
1960 applied with little revision until its replacement by the Labor Law of 1996. It was also rated
quite restrictive in terms of the lack of provision for fixed term or other types of labor contracts.
We assigned an ELR score of 1.7 to Jordan for all years prior to the 1995-9 period.
The Labor Law of 1996 and Act No. 36 of 1997, however, relaxed restrictions quite
significantly. First, it gave more specific recognition to fixed term contracts which would not
require compensation on completion of the term of employment specified in the contract.
24
Second, it allowed for employment on a trial basis not to exceed three months. Third, it
recognized still another type of contract, namely, that of indefinite duration which would apply
to workers working on piece-rate and other contracts. Employers were permitted to suspend even
these contracts for economic or technical reasons as long as the Ministry of Labor was notified
(but its approval was not necessary). Workers could even be dismissed from fixed term contracts
before completion of the terms as long as notification was provided and specified compensation
provided. It was also made clear that the rights to join a labor union and be represented by it did
not apply to non-Jordanian workers (Bitar 2004, Shawabkey2006).
14
Based on these
considerations Jordan’s score on ELR index for 1995-9 was reduced to 1.46 from its earlier
value of 1.7.
Lying behind the change were three factors: (1) the delayed but growing interest of the
Jordanian government in privatization and viewing growth of the private sector as the way to
deal with the country’s high and rising unemployment rate
15
, (2) the prior liberalization of other
aspects of Jordan’s economy, and (3) the fact that in 1996 negotiations were under way to
stimulate Jordanian exports in low wage garment industries by signing the Qualifying Industrial
Zones (QIZ) protocol with Israel and the US.
With respect to (2), as noted by Pripstein-Posusney (2004) the delay in privatization was
especially notable since the other liberalization measures (with respect to product markets,
financial markets and international trade) had commenced in 1991 and 1992, triggered by the
severe economic crisis that Jordan suffered as a result of the Gulf War. These reforms and
proclamations that privatization would also take place were also supported financially by the
14
Non-Jordanian workers have become an increasingly important component not only of the agricultural and
construction labor force but also of manufacturing employment.
15
Advocacy for increased reliance on the private sector became stronger after King Abdullah succeeded his long-
time predecessor King Hussain in 1999.
25
World Bank and other agencies. The fear of loss of jobs in the process of privatization and the
adverse effect that this might have for legitimacy of the monarchy might well have contributed
to this. Even when privatization started in earnest after 2000, Pripstein-Posusney (2004) informs
us that clauses in the specific privatization agreements specified that the new firms would have
to retain workers for at least two years after the sale and that the proceeds would be used to
provide retraining and other help to impacted workers.
With respect to (3) the QIZ Protocol, formulated making QIZs in Jordan considered as an
extension of Israel which already had a FTA agreement with the US.
16
Chief among the
industries subject to high tariffs and quotas was the garments industry, a low-wage, labor
intensive sector. Garments were the main activity to attract investment, most of which was in the
form of FDI which came to Jordan to take advantage of these special incentives. Yet clearly to
be competitive with countries with much lower wage rates like China, India, Pakistan and
Bangladesh, This put pressure on Jordanian officials to keep labor laws such that they would not
undermine these activities that soon came to provide a large share of total industrial employment
(Nugent and AbdelLatif, 2009).
However, continuing inflation in the country and press calling attention to some abuses
of labor among QIZ employers led Jordan’s Labor Ministry to impose minimum wage
regulations on all labor contracts, thereby causing us to raise the ELR for Jordan to 1.52 for the
2000-4 period.
Zambia
Interestingly enough, the motivation for the changes over time in the ELR index for
Zambia bears some resemblance to that of Jordan. Like Jordan, Zambia has been heavily
16
As further incentive to Jordan for having signed a Peace Treaty with Israel, Jordanian QIZ industries in which at
least 8 percent of the final cost had to come from Israel, were exempted from quotas as well as tariffs.
26
dependent on the public sector for its formal economic activity. Its government and state
enterprises provided the vast majority of formal sector enterprises until the mid 1990s when
privatization got under way. In 1991 there were some 282 state enterprises, many of the largest
being in the mining, manufacturing and service sectors.
While in terms of the labor force as a whole the vast majority has long been engaged in
self employment, microenterprises, subsistence farming and other informal activities, the formal
sector contracts, largely in the public sector were of two types, fixed term contracts of 6 months
duration with no commitment for renewal at the end of the contract, and permanent contracts,
that would terminate primarily only with the death or retirement of the worker. The labor law,
the Employment Act of 1965, was very simple and not very restrictive since short term contracts
were feasible. Since state enterprises in which the bulk of the formal sector labor force was
employed virtually never went into default, the issue of dismissal of permanent or even fixed
term workers for financial reasons virtually never arose. On the basis of matching the various
provisions of the law to the Botero et al 2004 scoring of the sub-indexes, we assigned Zambia a
rather low score of 1.05 for the years up to the 1985-9 period. The Industry and Labour
Relations Act allowed for labor unions, one in each industry.
However, in the 1980s the many distortions arising from monopoly labor unions and the
country’s highly protective trade regime and various other product market imperfections began
to take their toll on the economic viability of the country’s economy and even state enterprises
started to face financial difficulties. In this environment, a new Employment Act was introduced
in 1982 with subsequent set of regulations. The revisions of the labor regulations had the effect
of slightly tightening the conditions under which workers could be dismissed. On the basis of the
tighter job security regulations, the score assigned to Zambia on the ELR index rose to 1.3
beginning in the 1985-9 period. With a new democratically elected government coming to
27
power after 1990, the government obtained financial support from the International Fund and
adopted a package of liberalizing policies, including trade and product market reforms and
committed itself to privatization of the state enterprises. The new government looked to the
private sector as the preferred route to prosperity. But this meant encouraging private firms to
enter and grow in size and private investors to buy up the ailing state enterprises. To that end, the
government passed a number of new laws, the Privatization Act of 1992 and later the
Employment Act of 1997.
This 1997 labor law explicitly identified a number of additional types of labor contracts,
temporary, part-time, fixed term and so on giving private sector employers more options. This
explains why the ELM index score assigned to Zambia was lowered to 1.15 in the period 1995-9.
Rather remarkably, by 2000 the vast majority of the 282 state enterprises that existed in 1991
were privatized and the government was able to collect sizable amounts of privatization
revenues. These revenues were to be used in part for retraining and otherwise assisting
redundant workers. Yet, in terms of formal sector employment, the privatization experience was
very disappointing. Formal sector employment fell by some 15 percent over the decade of the
1990s. Most of the dismissed workers went into the informal sector or unemployment. The
decline in formal employment was quite remarkable given that population was growing at a rate
of 3.8 percent per annum over the same period. Several observers (e.g., Petrauskis 2005 and
Nyirenda and Shikwe 2003) attribute this failure to (1) the law itself which made it easy for the
private employers to substitute informal short-term workers for formal ones, (2) the failure of the
government to use the privatization proceeds to retrain laid-off workers, and (3) the failure of
either government administrators or the courts to enforce the provisions of the employment laws
(made difficult by the vagueness of some of its provisions).
Perhaps as a result of the unpopularity of the privatization and the 1997 labor law, by
28
Statutory Instrument No. 2 of 2002, various labor regulations were tightened slightly, including
those for minimum wages, maternity leave, hours of work and other conditions of work. As a
result, we roughly calibrated this to justify an increase of 0.1 on the ELR to a level of 1.25 for
the period 2000-4.
A Brief Recap of the Cases
Of the four cases treated here, three different legal traditions were represented. Just as in
the previous Portugal New Zealand comparison, even if there may have been some notable
differences in these scores between legal traditions at one point in time as argued by Botero et al
(2004), from the above discussion it is by no means clear that different legal traditions played
much of a role in explaining changes in these scores over time.
17
Indeed, since the legal
traditions didn’t change over time, it would seem doubtful that they could contribute to the
explanation of changes in the ELR scores over time. From the above discussion of the four cases,
we believe that a case could be made for economic crises, political crises (or changes) and
perhaps certain changes in economic structure. We will therefore pay some attention to these
factors in the empirical work below attempting to explain changes in the ELR indexes across
countries and over time in the LAMRIG dataset.
IV. Trying to Assess the ELR/LAMRIG Index
In this section we discuss the methodology we choose to assess the appropriateness of our
ELR/LAMRIG index and present the auxiliary data required for this task.
In terms of the econometric methodology, we first try to replicate the results in Botero et al.
(2004). At the outset, the most clear-cut difference is that while their sample has 85 countries,
ours contains 142 countries for the same period (1995-9). Our first task is to determine whether
17
Perhaps a case could be made for Socialist law contributing to the high score for China in the early periods.
29
or not we can replicate their results in a cross-sectional setting with this extended sample.
Accordingly, from their Table IV, the first model we estimate takes the form:
i i i i i
LO GDP LAMRIG ? ? ? ? + + + =
2 1
(1)
where LAMRIG
i
is our index of Labor Market Rigidity for country i, GDP
i
is the log of per
capita GDP, and LO
i
is a set of dummy variables for legal origins (namely, French, German,
Scandinavian, Socialist and English.) Botero et al. estimate this model by OLS with robust
standard errors and data for 85 countries in year 1997. They find that legal origins are a
substantially more important determinant of labor market reform than per capita GDP. They
argue that this result favors the legal theories of institutional changes (and, by the same token,
belittles the other two theories they identify, the efficiency and political theories.)
We then subject this baseline model to various robustness checks. In particular, we ask
whether the taking into account the over time variation of our index affects the basic results. We
investigate this first by fitting a simple pooled OLS model to the data (by simple we mean we
run OLS on the pooled data without taking into account the panel structure.) We also run two
split-sample exercises. The first estimates this baseline model separately for OECD and non-
OECD countries. The second estimates the model separately for the pre- and post-1980
observations. The rationale for the first split of the samples is that richer countries may face quite
different constraints than poorer countries. If so, this could be reflected by important differences
in the determinants of LAMRIG between the samples. The justification for pre and post 1980
split is the widespread perception that this year somewhat marked the beginning of an era more
favorable to market-oriented reforms than the previous 20 years or so (which were marked by
interventionist import-substitution strategies, especially in developing countries).
Our second step in terms of estimation strategy is to exploit more decisively the panel
30
structure of our data. The fixed-effects estimator would be a natural starting point but one of the
most important variables in the Botero et al (2004) exercise, legal origins, are time-invariant.
Therefore our starting point is the random-effects estimator.
18
The model we estimate takes the
form:
it i ti it
LO GDP LAMRIG ? ? ? ? + + + =
2 1
(2)
where again LAMRIG
it
is our index of Labor Market Rigidity for country i at period t. The
subscript t refers to a 5-year period, where the measure is the average over the whole period. The
nine periods included are: 1960-1964, 1965-1969, 1970-1974, 1975-1979, 1980-1984, 1985-
1989, 1990-1994 and 1995-2000.
In order to minimize country-specific errors, we clustered the
standard errors at the country level. Using the random-effects estimator, we also carry out the
same split-samples checks as for the baseline model, namely for OECD versus non-OECD and
pre- and post-1980.
Thus far we have talked only about the levels of LAMRIG. Yet we define reform as
changes in these levels. The third step in our estimation strategy is therefore to estimate changes
in levels of LAMRIG. In order to do that, we recognize that labor market reform in one period is
related to the amount of past progress in this reform. Hence, we capture this by enlarging the
baseline Botero et al. model with a one-period (i.e. 5 year) lag of LAMRIG. The model we
estimate takes the form:
it t i i ti t i it
X LO GDP LAMRIG LAMRIG ? ? ? ? ? ? + + + + + = ?
? ? 1 , 4 3 2 1 , 1
(3)
where ?LAMRIG
it
is the change in our index of Labor Market Rigidity for country i between
period t and period t-1, with periods defined as before. This model will be estimated at first using
the random-effects with standard errors clustered at the country level. Next, we re-estimate it by
18
The Hausman test contrasts the fixed- and random-effects estimator for models containing only the time-variant
variables in these models. In this case, it is not very helpful because the test runs for a model that only contains per
capita GDP.
31
adding variables for four different groups of factors (in X
i, t-1
) namely, political crises, economic
shocks, structural factors and other reforms. As the notation indicates, we always enter these
factors lagged one-period. This is not only to minimize endogeneity concerns but also because
the time window we use is somewhat lengthy and it may be that the reform occurs early in the
window and any of these four factors later. By lagging these right-hand side variables we
eliminate this concern.
As for the data we use in the estimation of the above models, the measure of the
dependent variable in the analysis is the LAMRIG index, discussed above in considerable detail.
The other two sets of variables in the baseline model are GDP and legal origins. Per capita GDP
is from the Penn World Tables and the legal origins classification is provided by Botero et al.
(2004).
In terms of structural factors, we collected data on the following variables from the
World Development Indicators: the Gini coefficient, the Government share of GDP, the ratio of
foreign aid to GDP, the share of natural resources exports in total exports and the share of
agriculture in GDP.
For economic crises we include several different measures,
19
namely, the largest single
year GDP fall in percentage points that occurred in each five-year period (Max fall GDP), the
number of years of negative GDP growth (between zero and five for each of the 5-year average
period), the current account balance (CAB)
20
, the number of years in a debt crisis within each
five year period (Debt Crisis),
and a dummy variable for periods in which annual inflation was
above 50%.
Regarding political crises, we limit our attention to the following indicators. The first
19
For a review, see Furman and Stiglitz (1998) and Ishihara (2005).
20
CAB is an inverse measure of crisis.
32
group comprises count variables for both the assassination of important political leaders and
general strikes during each five year period.
Both of these variables originate from Banks (2005).
The second group comprises the democracy measure (from the POLITY IV data set) and also the
Political Constraints Index (POLCON) provided by Henisz (2000). The Polity IV democracy
variable is used to control for relative levels of democratic freedoms (coded in a 1 to 10 scale,
with 10 indicating the highest level of democracy). The stronger is democracy, the more the
median voter might be expected to exercise influence. Yet, because the median voter is more
likely to be a worker or even a union member, the influence of democracy on labor market
liberalization could be ambiguous or perhaps even negative. POLCON measures the number of
veto points in a political system, the expectation being that the more potential vetoes need to be
circumvented, the less likely it is that labor market reform will be adopted. The third and last
group contains a measure of the intensity of civil war and of the intensity of international armed
conflicts. Data for constructing these measures is from the Correlates of War project at the
University of Michigan.
Finally we investigate the role of other structural reforms potentially affecting the
probability that labor market reform is implemented. We focus on financial reform and on trade
liberalization.
21
We proxy financial reform by two measures: the share of credit to the private
sector in GDP, and an index of financial development that reflects not the overall size of the
financial system but its efficiency levels. In the case of trade liberalization, we use four
measures. One is the length in years of uninterrupted trade liberalization derived from the
Appendix 2-B of Warcziarg and Welch (2003). Another measure is a measure of trade openness
from PWT (openk, exports plus imports as a share of GDP). Thirdly, we use the trade
21
For a discussion of the relationship between trade liberalization and labor market reform see Fajnzylber
and Maloney (2005), and references therein. Idem for financial reform and labor market reform, see
Pagano and Volpin (2008).
33
liberalization index developed by Campos, Nugent and Hsiao (2006). The later further extends
the Sachs and Warner (1995) measure of trade openness that was already corrected and extended
from 1970-1989 to 1990-99 by Wacziarg and Welch (2003).
22
Since Rodriguez and Rodrik
(2000) and Rodriguez (2006) provide a powerful critique of the efforts of Sachs and Warner
(1995) to relate their “open” measure to cross-country growth rates, we have incorporated these
views in this measure of trade reform. A major objection of these authors was that the cross-
sectional evidence on growth rates relied on only two of the five S-W criteria, namely, export
marketing boards (XMB) and black market premium (BMP). In effect, this suggests that the S-W
index of openness, even when extended by Warcziarg and Welch, in fact has little to do with
trade restrictions in the form of tariffs and non-tariff barriers. Rodriguez (2006) also had some
specific quarrels with the way XMBs were treated in their classifications of certain countries.
Finally, he followed Warcziarg and Welch (2003) in using a lower tariff rate threshold (20%
instead of the 40% in the original S-W) to distinguish “open” from “closed”.
23
Since most
countries in the world had fallen below the 40% threshold by the mid- 1990s, this change has the
effect of giving more weight to tariff barriers in the classification. Therefore, we construct an
alternative measure that takes advantage of more recent information on XMBs (from World
Bank and other sources) so as to distinguish between those marketing boards that in practice
discriminate against producers for export markets from and those which do not, as well as these
other suggestions. Given the view expressed most strongly in Rodriguez and Rodrik (2000), that
22
More specifically, these authors defined a country as closed (i.e., open =0) if it had any one of the
following: (1) an average tariff rate of 40 per cent or more, (2) non-tariff barriers covering 40 per cent or
more of trade, (3) a black market exchange rate that is depreciated by 20 percent or more relative to the
official exchange rate, (4) a state marketing agency or board for major exports, and (5) a socialist
economic system (as defined by Kornai 1992).
23
This was used to show something that Warcziarg and Welch (2003) had already shown, namely that the
positive relation between growth and open found by Sachs and Warner (1995) and others disappears
when the lower threshold is used or when the period studied is that after 1990.
34
as far as growth is concerned, the key reform was BMP reduction, as another alternative we use
simply BMP.
V. Assessing LAMRIG
Next we turn to an assessment of LAMRIG based on an examination of the determinants of the
ELR indexes across countries and over time. Given that the lynch-pin for our construction of
LAMRIG was the Botero et al (2004) data set for 85 countries, we begin our assessment in Table
V.1 by trying to replicate some of the findings of that study. That study’s basic finding was
presented in their Table IV relating their ELR index to the log of per capita GNP, and dummy
variables for Socialist, French, German and Scandinavian legal origins. The omitted legal origin
was English Common Law. The results they reported for their sample of 85 countries is reported
in column (1) of Table V.1. As can be seen, the explanatory power of the model was quite high
and although the income per capita measure was insignificant the four legal origin dummy
variables had highly significant positive effects on ELR.
In column (2) of this table we repeat their analysis based on the LAMRIG dataset for the
same year but using our larger sample of 142 countries. Notice that the effect of income per
capita is now negative and significant but all four legal origin dummy variables still have
positive and highly significant effects on LMR (from now on designated LAMRIG). But our
more fundamental extension of their dataset is its extension over time going back to the early
1950s in quite a few cases. OLS estimates of the pooled panel data (now consisting of 792
observations) are provided in column (3). In this case, the negative effect of higher income per
capita is no longer significant (as in Botero et al (2004)) but all four legal origin dummies retain
their positive and significant effects. Note that the effects of all of them on LAMRIG are now
considerably stronger than in the original Botero et al (2004). Given our earlier observation that
35
in the 1950, 1960s and even 1970s, the ELRs were rising before stabilizing and declining in
some cases in recent years, in columns (4) and (5) we break the sample into pre and post-1980
observations. While the results are very similar for the Socialist and French legal origin
dummies, there are some notable differences in other respects. When split this way the negative
coefficient of the Log Per Capita GDP is again statistically significant but quite a bit larger in the
Pre1980 sample. On the other hand, the impacts of the German and Scandinavian Legal Origin
dummies are larger and more statistically significant in the post 1980 sample. In each of the
columns so far, the effects of the legal origin variables are quite consistent consistent with
Botero et al.(2004) in that the socialist, French Civil Law and German Civil Law traditions in
that order all have large positive effects on LAMRIG than the omitted Common Law tradition.
The only difference is that in some of these cases, columns (2), (4) and (5) the Scandinavian
Legal Origin has the largest positive effect.
Columns (6) and (7) provide the corresponding comparison between OECD countries and
non-OECD countries, reflecting in large part differences in income. Notice that in our case, the
non-OECD sample is considerably larger than the OECD sample. Again the various types of
Civil Law dummies are shown to have significant positive influences in both samples (when
there is sufficient variation of these variables in the sample to allow coefficients to be estimated).
Both the French and Scandinavian Legal Origin variables have considerably larger effect in the
OECD sample than in the non-OECD one. The most striking difference between the samples,
however, is the difference in the effect of Log Per Capita GDP, quite large and positive in the
case of the OECD sample, but negative and significant in the non-OECD sample. This would
seem to help explain the opposing trends in LAMRIG indexes between high income and
“Europe” regions and some of the other regions and income groups in Figures 1 and 2.
Given the aforementioned absence of change over time in the legal tradition upon which each
36
country’s legal system is based, as noted above, when fixed effects are used, the parameters for
legal origin dummies cannot be estimated. We proceed therefore in the rest of our empirical
analysis to estimate not the levels of LAMRIG but rather the changes in LAMRIG (i.e., reform
reversals in labor market regulations). As explained in Section IV, this makes it appropriate to
estimate the relationships in the LAMRIG panel with random effects and standard errors
clustered at the country level with equation (3) above.
Table V.2 reports the results obtained for changes in LAMRIG first for the full sample (721
observations) and then for the same subsamples as in columns (3)-(7) in Table V.1 but based on
this more appropriate (RE) estimation procedure. Once again, we find considerable variation
across samples in the effects of Log Per Capita GDP, positive and significant in the post 1980
sample and negative and significant once again in the OECD sample. In all the other cases
including the full sample, the coefficient of Log Per Capita GDP is not statistically significant.
With the minor exception of the Scandinavian Legal Origin dummy (for which there is little
variation in our samples), the coefficients of the various Civil Law Origin variables are no longer
statistically significant. This confirms what we seemed to find in Table III.1. Note, that in great
contrast to the estimates in Table V.1, the results are very weak with no more than one
explanatory variable being significant in any column except for the OECD sample where there
are two.
24
Given the rather glaring weakness of these results, in subsequent tables we add the lagged
level of LAMRIG to the right hand side and a series of other determinants suggested by our brief
review of the several country cases of Section III and by other literature.
Table V.3 reports the results when the added variable is one or another of the following
24
We have also investigate whether there are important non-linearities in the effects of per capita GDP but did not
find evidence for this to be the case.
37
structural variables: Income Gini, the government share in GDP, the share of foreign aid in GDP,
natural resource exports as a share of total exports and the share of agriculture in GDP. As
expected, the effect of Lagged LAMRIG is always negative and significant indicating that there
seems to be a convergence process going on in labor market regulations. This is quite consistent
with the quite different trends between countries with initially low LAMRIG indexes and those
with initially high ones in Figures 1 and 2 and the upward trends for Portugal and New Zealand
whose initial LAMRIG scores were very low, and the downward trends from initially high
scores for China and Jordan in Table III.1. Countries with high LAMRIG index values are likely
to reform, i.e., lower their LAMRIG scores over time, whereas those countries with low
LAMRIG scores are more likely to introduce reform reversals by raising their rigidity scores.
Consistent with the results for the full sample (column 1) of Table V.2, the coefficients of Log
Per Capita GDP are not statistically significant, except in column (1) where the Income Gini is
the added structural variable. Somewhat surprisingly, the coefficients of the French,
Scandinavian and Socialist Legal Origin dummy variables are once again positive and
significant, though of course much smaller than in the estimates presented in Table V.1 for the
level of LAMRIG. In most cases, the coefficients of the Socialist and Scandinavian Civil law
dummies are also positive and significant, though again much smaller than before. None of the
individual structural indicators has a significant effect on the change in LAMRIG. One should
notice, however, that due to missing observations on these additional variables, the sample sizes
are considerably smaller in this table, especially in columns (1) and (5).
In Table V.4 we present estimates similar to those of Table V.3 for changes in LAMRIG but
in this case with five different measures of economic crises, in each case lagged to avoid the
simultaneity and other problems identified in Section IV. Column (1) presents the results when
the crisis is a debt crisis. Columns (2) –(5) report the corresponding results when the crises
38
pertain to inflation rates above 30% per annum, a period including a year with the largest fall in
GDP during the period covered, the number of years of falling GDP within the five year period,
and the current account, respectively.
Once again, the effects of Lagged LAMRIG are consistently negative and significant, those
of French and Socialist legal origins positive and significant. The effects of Log Per Capita GDP
are negative and significant in columns (1) and (2) but negative and not significant in columns
(3)- (5). None of the economic crisis variables turns out to have a significant effect on the change
in LAMRIG. As in Table V.3, the French, Scandinavian and Socialist Legal Origin variables all
have small positive and significant effects.
25
Table V.5 substitutes political variables including political crises for the economic crisis and
structural variables included in Tables V.3 and V.4. Column (1) adds Democracy, Column (2)
uses instead the political constraints index (POLCON), column (3) assassinations, column (4)
strikes, and columns (5) and (6) international and civil wars, respectively. As with the economic
crises, these measures are all based on lagged values. Democracy has a negative but insignificant
effect on the change in LAMRIG as does POLCON which is often considered another measure
of democracy reflecting a system of checks and balances. So too neither strikes, international
war and civil wars have significant effects on the change in LAMRIG. The one political crisis
measure with an effect that is somewhat significant is that for Assassinations in column (3). This
variable has a negative effect that is significant at the 10 percent level. (We trust that labor
market reform zealots will not go so far as to recommend assassinations as a means of bringing
about such reforms!). The findings of previous tables of significant negative effects of lagged
LMRIG and the positive effects of the French, Scandinavian and Socialist Legal Origin measures
25
We have also run all these spcifications for each legal origin sub-sample and do find that this had an impact in the
main results.
39
are all retained in this table as well.
Finally, in Table V.6, to our basic specification we add alternative measures of other types or
reforms, again lagged. In columns (1) – (3) we present the results for three alternative measures
of trade reforms. Column (4) presents estimates when the added variable is the black market
premium (BMP), an inverse measure of trade reform. Columns (5) and (6) present results for two
alternative measures of financial market reform/development, namely, the share of credit to the
private sector in GDP and the Financial Reform Index, respectively. Again all the standard
results apply. In addition in columns (1), (2), (4), (5) and (6) the effect of Log Per Capita GDP is
again negative and significant (in most of these cases at the 5 percent level). The effects of the
various lagged other reform measures vary considerably from case to case. Trade openness as
measured by the first two measures in columns (1) and (2) reveal positive and significant effects
on LAMRIG changes, meaning reform reversal. In the same spirit, an increase in the BMP
premium has the effect of reducing LAMRIG. The two, financial reform measures raising the
share of private credit in GDP and the financial reform index, by contrast have negative effects
on changes in LAMRIG though only in the first case is the effect significant at the 10 percent
level.
Taken together, the results could be interpreted as providing a somewhat positive assessment
of LAMRIG. In particular, the cross-section results with the same specification as in Botero et al
(2004) but based on our considerably larger data set replicate and perhaps even strengthen the
Botero et al results for the legal origin variables. The estimated impact of per capita income,
however, is different between the smaller Botero sample and the larger LAMRIG sample for
1997. In particular, while the effect is negative and significant in some of our specifications,
especially those in Table V.6, the most important findings in this respect is the way its effect
differs in different types of countries, being most strongly negative in the OECD sample of Table
40
V.2 and most significantly positive in the post 1980 sample of the same table.
With respect to the legal origin dummies, the positive and significant influences cant in the
Botero study are confirmed when the larger LAMRIG data set is used. This is true for both the
cross section and pooled estimates for levels of LAMRIG. Although these effects largely
disappear when it is the changes in LAMRIG that are explained in Table V.2, they come back in
again though in much smaller magnitudes when the Lagged LAMRIG is included among the
explanatory variables.
Of the other variables we have introduced one at a time in the subsequent tables, only a few
have significant effects. One is the Assassinations measure of Political crises which seems to
have a weak negative effect on LAMRIG changes, implying LMR reduction or reform. The
others are three of the four trade reform measures which in each case have positive influences,
suggesting reform reversal and the private credit/GDP measure which has a significant negative
effect on LAMRIG changes, suggesting that lagged reforms of this sort facilitate labor market
reform whereas the opposite is the case for the aforementioned measures of trade reform.
VI. Conclusions and Future Research
The substantive results presented here are clearly only a beginning of a fuller analysis.
We would like to further examine the robustness of the results, e.g., when several of the
additional variables are retained in the estimating equation at the same time, or with more refined
measures of some of the variables used. Similarly, in view of the differences in some of the
effects between pre and post 1980 samples and between OECD and non-OECD samples, it
would be desirable to examine the robustness of the results of the more inclusive specifications
to different samples. Yet, even thus far, we find several interesting results:
(1) That, when the dependent variable is changes in the LAMRIG indexes, legal origin measures
41
still exert significant though smaller influence than when as in the Botero study it is the level of
the indexes that is estimated
(2) That none of the (a) structural variables (income Gini, Government share in GDP, Foreign
Aid to GDP, the share of natural resource exports in total exports and the share of agriculture) or
(b) economic crisis or (c) political crisis measures (except Assassinations) seems to have much
of an influence on LAMRIG changes.
(3) Consistent with the findings of other studies, labor market reforms may be affected
significantly by other reforms. Our evidence, though limited to trade and financial reforms, is
that lagged trade reform may set back labor market reform but that financial reforms may
encourage labor market reforms.
(5) With respect to future research, in addition to the additional robustness checks and
improvements in some of the measures of variables identified above, it is our intent to:
(a) Further improve on LAMRIG by digging deeper into the ever-improving availability
of information on labor laws over time and across countries,
(b) Possibly to follow the lead of some researchers on OECD countries to annualize the
data on LAMRIG as well as the related variables used to explain changes therein
over time,
(c) To extend the use of LAMRIG to examine its effects on labor market and other
phenomena as Botero and many others have with somewhat smaller data sets.
42
References
Addison, John T. and J. Grosso 1996. “Job Security Provisions and Employment: Revised
Estimates” Industrial Relations 35 (4), 585-603.
Addison, John T.and Paulino Teixeira 2003. “The Economics of Employment Protection”,
Journal of Labor Research 24 (1) 85-129.
Allard, Gayle 2005a. “Measuring Job Security over Time: In Search of a Historical Indicator for
EPL (Employment Protection Legislation)” Instituto de Empresa, Spain Working Paper
WP05-17.
Allard, Gayle 2005b. “Measuring the Changing Generosity of Unemployment Benefits: Beyond
Existing Indicators” Instituto de Empresa, Spain IE Working Paper WP05-18.
Alesina, A. and A. Drazen (1991). “Why Are Stabilizations Delayed?” American Economic
Review 81, 1170-88.
Arpaia, Alfonso, Peghe Braila and Fabiana Pierini 2007. “Tracking Labor Market Reforms in the
EU Using the LABREF Database”, Paper presented at the IZA - Fondazione Rodolfo
Debenedetti Workshop: Measurement of Labor Market Institutions, IZA, Bonn, July 4, 2007.
Banks, A. (2005), “Banks' Cross-National Time-Series Data Archive,” electronic database.
Bassaanini, Andrea and Danielle Venn 2007. Assessing the Impact of Labour Market Policies on
Productivity: A Difference-in-Difference Approach. Paper presented at the IZA - Fondazione
Rodolfo Debenedetti Workshop: Measurement of Labor Market Institutions, Bonn, July 2007.
Bertola, Guiseppe 1999. “Micoeconomic Perspectives on Aggregate Labor Markets”, in
Handbook of Labor Economics, v 3, O. Ashenfelter and D. Card, eds, Elsevier Science, 2985-
3027.
Bertola, Guiseppe 2008. Labor Market Regulation: Motives, Measures, Effects
Besley, Timothy and Robin Burgess 2004. “Can Labor Regulation Hinder Economic
Performance? Evidence from India” Quarterly Journal of Economics 119, (1), 91-134.
Bhattacharjea, Aditya 2006. Labour Market Regulation and industrial Performance in India: A
critical Review of Empirical Evidence” Indian Journal of Labour Economics 49 (2), 211-232.
Bitar, Fouad 2004. National Labour Profile: Jordan, International Labor Office
(http://www.ilo.org/public/english/dialogue/ifpdial/info/national/jo.htm)
Blanchard, O. and J. Wolfers (2000). “The Role of Shocks and Institutions in The Rise of
European Unemployment: The Aggregate Evidence,” Economic Journal, 110, 1-33.
Boeri, Tito , G. Nicoletti and S. Scarpetta 2000. “Regulations and Labor Market Performance”
CEPR Discussion Paper 2420.
43
Botero, J., S. Djankov, R. La Porta, F. Lopez-de-Silanes, and A. Shleifer (2004). “The
Regulation of Labor,”Quarterly Journal of Economics 119: 1339-1382.
Botero et al Labor Dataset_qje_dataforweb_2005.xls
Brooks, Ray and Ran Tao 2002. China’s Labor Market Performance and Challenges IMF
Working Paper 02/210
Bruno, M. (1996). Deep Crises and Reform: What Have We Learned? World Bank.
Bruno, M. and W. Easterly (1996). “Inflation’s Children: Tales of Crises that Beget Reforms,”
American Economic Review Papers and Proceedings 86, 213-17.
Campos, N., Hsiao, C. and J. Nugent (2009) “Crises, What Crises? New Evidence on the
Relative Roles of Political and Economic Crises in Begetting Reforms,” IZA DP.
Deakin, Simon, Pruya Lele and Mathias Siems 2007. “The Evolution of Labor Law: Calibrating
and Comparing Regulatory Regimes” International Labour Review 146 (1), 133-162.
Di Tella, Rafael and Robert MacCulloch 1999. “The Consequences of Labor Market Flexibility:
Panel Evidence Based on Survey Data” European Economic Review 49 (5), 1225-1259.
Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes and Andrei Shleifer, 2004, “The
Regulation of Labor,” National Bureau of Economic Research, Working Paper 9756.
Djankov, Simeon and Rita Ramalho 2008. “Employment Laws in Developing Countries”
Drazen, A. (2000). Political Economy in Macroeconomics, Princeton University Press.
Drazen, A. and V. Grilli (1993). “The Benefit of Crises for Economic Reforms,” American
Economic Review, 83 (3), 598-607.
Dutt, Pusha Labor Market Outcomes and Trade Reforms: The Case of India
Dutta Roy, Sudipta 2004. Employment Dynamics in Indian Industry: Adjustment Lags and the
Impact of Job Security Regulations” Journal of Development Economics 73 (1), 233-256.
Eichhorst, Werner, Michael Feil and Christoph Braun 2007. “What Have We Learned?
Assessing Labor Market Institutions and Indicators”
Fallon, Peter R. and Robert E.B. Lucas 1991. “ The Impact of Changes in Job Security
Regulations in India and Zimbabwe” World Bank Economic Review 5 (3), 395-413.
Fallon, Peter R. and Robert E.B. Lucas 1993. Job Security Regulations and the Dynamic
Demand for Industrial Labor in India and Zimbabwe” Journal of Development Economics 40,
214-235.
44
Fajnzylber, P and W Maloney (2005), “Labor demand and trade reform in Latin America,”
Journal of International Economics 66 (3): 423-446.
Forteza, Alvaro and Martin Rama (2006). “Labor Market “Rigidity” and the Success of
Economic Reforms across More than 100 Countries,” Journal of Policy Reform 9 (1) 75-106.
Freeman, Richard B. 2008. “Labor Market Institutions Around the World”, London: LSE CEP
Discussion Paper no 844.
Greenhill, Brian, Layna Mosley and Aseem Prakash 2009. “Trade-Based Diffusion of Labor
Rights: A Panel Study, 1986-2002” American Political Science Review 103 (4),
Grubb, D. and W. Wells 1993. “Employment Regulation and Patterns of Work in EC Countries”,
OECD Economic Studies 21.
Haddad, Gholam Reza, K. “The Impacts of Globalization on the Earning Inequality (the Case
Study of Iran,” Sharif University, mimeo, 2009.
Heckman, James J., and Carmen Pages 2000. “The Cost of Job Security Regulation: Evidence
from Latin American Labor Markets” NBER Working Paper 7773.
Heckman, James J., and Carmen Pages, eds. Law and Employment: Lessons from Latin America
and the Caribbean. New York: University of Chicago Press, 2004.
Henisz, W. (2000). “The Institutional Environment for Multinational Investment”, Journal of
Law Economics and Organization, 16 (2), 334-364.
Joh, Sung Wook 2001. Korea’s Corporate Sector: Crisis and Reform” in O. Yul Kwon and
William Shepherd. Eds. Korea’s Economic Prospects. Cheltenham: Elgar
Kucera, David 2002. “Core Labour Standards and Foreign Direct Investment” International
Labour Review 141 (1-2), 31-69.
Kwon, O. Yul 2001. Korea’s International Business Environment Before and After the Financial
Crisis” in O. Yul Kwon and William Shepherd. Eds. Korea’s Economic Prospects.
Cheltenham: Elgar, 245-265.
La Porta, Rafael, Florencio Lopez-de-Silanes and Andrei Shleifer 2008. “The Economic
Consequences of Legal Origins”, Journal of Economic Literature 46 (2), 285-332.
Lazear Edward 1990. “Job Security Provisions and Employment” Quarterly Journal of
Economics 105, 699-726.
Magruder, Jeremy R. “High Unemployment Yet Few Small Firms: The Role of South African
Labor Regulations”,
Montenegro, Claudio and Carmen Pages 2004. “Who Benefits from Labor Market Regulations?
Chile 1960-1998” in Heckman and Pages, ed. Law and Employment : Lessons from Latin
America and the Caribbean. New York: University of Chicago Press, 2004, 401- 434.
45
NATLEX. November 2008. International Labour Organization <natlex.ilo.org>.
Nicoletti, G. R.C.G. Haffner, S. Nickell, S. Scarpetta and G. Zoega 2000. European Integration,
Liberalization and Labor Market Reform” in G. Bertola, T. Boeri and G. Nicoletta, eds.
Welfare and Employment in a United Europe. Cambridge: MIT Press
Nickell, Stephen 1997. “Unemployment and Labor Market Rigidities: Europe versus North
America” Journal of Economic Perspectives 11, 55-74.
Nickell, Stephen.and R. Layard l 1999. Labour Market Institutions and Economic Performance,
CEP Discussion Paper 407 "Handbook of Labor Economics., V. 3 Ed by O. Ashenfelter and
David Card, 3029-3084.
Nickell, Stephen, L. Nunziata, W. Ochel and G. Quintini 2003. “The Beveridge Curve,
Unemployment and Wages in the OECD from the 1960s to the 1990s” in P Aghion, R.
Frydman, J. Stiglitz and M. Woodford, eds. Knowledge, Information and Expectations in
Modern Macroeconomics: In Honor of Edmund S. Phelps. Princeton University Press.
Nugent, Jeffrey B. and Abla AbdelLatif 2009. “A Quiz on the Net Benefits of Trade Creation
and Trade Diversion in the QIZs of Jordan and Egypt”. Los Angeles, CA. USC.
OECD 1999. OECD Employment Outlook 1999. Paris OECD
OECD 2004. OECD Employment Outlook 2004. Paris OECD
OECD 2006 Governance in China . Paris: OECD , especially Ch 1, 11.
OECD 2010. Economic Survey China Vol. 2010/6 , February, 2010.
Pagano, M. and P. Volpin (2008), “Labor and finance”, London Business School, mimeo
Persson, T. (2002). “Do political institutions shape economic policy?” Econometrica 70, 883-
905.
Persson, T. and G. Tabellini (2000). Political Economics: Explaining Economic Policy, MIT
Press.
Polity IV Project (2002). “Political Regime Characteristics and Transitions, 1800-2002”,
available at http://www.cidcm.umd.edu/inscr/polity/.
Pripstein Posusney, Martha 2004. “Globalization and Labor Protection in Oil-Poor Arab
Countries Racing to the Bottom?”in Ibrahim Saif, ed.,. The Jordanian Economy in a
Changing Environment. Amman: University of Jordan, Center for Strategic Studies, 115-151.
Rama, Martin and Raquel Artecona 2002. “ A Database of Labor Market Indicators across the
Countries”
46
Reutersward, A. 2005. “Labor Protection in China”: OECD, Social Employment and Migration
Working Paper No 30.
Rodrik, D. (1996). “Understanding Economic Policy Reform,” Journal of Economic Literature
34 (1), 9-41.
Sicat, Gerardo. “Reforming the Philippine Labor Market,” The Philippine Review of Economics.
Volume XLI, No 2, December 2004.
Sohn, Chan-Hyun and Junsok Yang 1998. Korea’s Economic Reform Measures under the IMF
Program. Seoul: Korea Institute for International Economic Policy.
World Bank (2004). Doing Business in 2004: Understanding Regulation. Washington, D.C. and
New York: World Bank and Oxford University Press.
World Bank 2006. Doing Business 2007: How to Reform. Washington, D.C.: World Bank .
World Bank 2008. Doing Business 2009. Washington, D.C. : World Bank Publications, 2008.
Yang,Junsok 2002. Update on Korean Economic Reforms and Issues in Korea’s Future
Economic Competitiveness Seoul: Korea Institute for International Economic Policy, Discussion
Paper 02-03.
ADDITIONAL INTERNET SOURCES:
http://www.allacademic.com//meta/p_mla_apa_research_citation/1/0/0/1/7/pages100174/p100174-1.php
http://www.state.gov/e/eeb/ifd/2007/80670.htm
http://www.lexadin.nl/wlg/
http://www.hrw.org/reports/pdfs/e/ethiopia/ethiopia913.pdf
47
Figure 1. Rigidity of Employment Protection Legislation across Selected Countries
0
0.5
1
1.5
2
2.5
1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
Zambia Botswana Ethiopia Philippines Jordan
0
0.5
1
1.5
2
2.5
1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
New Zealand Portugal
0
0.5
1
1.5
2
2.5
1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
China India Brazil
48
Table III.1
Country Scores on the Employment Law Rigidity Index (ELR) of Selected Countries Over Time
Country 1960-4 1965-9 1970-4 1975-9 1980-4 1985-9 1990-4 1995-9 2000-4
China 2 2 2 2 2 1.8 1.62 1.62 1.41
India 1.1 1.1 1.15 1.2 1.3 1.3 1.3 1.53
Jordan 1.7 1.7 1.7 1.7 1.7 1.7 1.7 1.46 1.52
Korea 1.44 1.44 1.44 1.44 1.55 1.6 1.36 1.19
New Zealand 0.13 0.13 0.27 0.48 0.48 0.48 0.48 0.48 0.51
Portugal 0.06 0.31 1.27 2.29 2.29 2.39 2.49 2.42 2.42
Brazil 2.06 2.06 2.06 2.06 2.06 2.06 2.4 2.4 2.1
Zambia 1.05 1.05 1.05 1.3 1.3 1.15 1.25
Botswana 0.9 0.9 1.0 1.0 1.3 1.3 1.05
Ethiopia 1.3 1.3 1.3 1.3 1.53 1.53 1.53
Philippines 1.4 1.4 1.6 1.6 1.6 1.69 1.69 1.61 1.80
49
Table V.1
The Extent of Labor Regulation, Legal Origins and Per Capita GDP
(Dependent variable: Level of LAMRIG, Labor Market Rigidity)
[1] [2] [3] [4] [5] [6] [7]
Botero et al. Xsection PoolOLS Pre1980 Post1980 OECD Non-OECD
Log Per Capita GDP -0.001 -0.0716** -0.00224 -0.160*** -0.0790*** 0.428*** -0.0419***
[0.0116] [0.028] [0.016] [0.033] [0.018] [0.038] [0.016]
Socialist Legal Origin 0.2943*** 0.698*** 0.956*** 0.839*** 0.839*** 0.788***
[0.0453] [0.11] [0.075] [0.081] [0.081] [0.078]
French Legal Origin 0.2474*** 0.477*** 0.621*** 0.669*** 0.569*** 1.109*** 0.453***
[0.0381] [0.066] [0.041] [0.069] [0.041] [0.091] [0.034]
German Legal Origin 0.1553** 0.474*** 0.335*** 0.192 0.535*** 0.476*** 0.333***
[0.0702] [0.12] [0.096] [0.16] [0.089] [0.15] [0.096]
Scandinavian Legal Origin 0.3865*** 0.901*** 0.562*** 0.503*** 1.128*** 1.074***
[0.0462] [0.11] [0.12] [0.16] [0.080] [0.092]
Constant 0.3072*** 1.860*** 1.067*** 1.983*** 1.826*** -3.211*** 1.566***
[0.1038] [0.24] [0.13] [0.24] [0.15] [0.34] [0.12]
Observations 85 142 792 324 468 222 570
R-squared 0.44 0.36 0.22 0.24 0.35 0.56 0.23
Notes: Robust standard errors in brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
50
Table V.2
Labor Market Reform, Legal Origins and Per Capita GDP
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5]
PoolOLS Pre1980 Post1980 OECD Non-OECD
Log Per Capita GDP -0.00214 0.00186 0.0380*** -0.0515*** -0.00813
[0.0058] [0.0069] [0.013] [0.017] [0.0050]
Socialist Legal Origin 0.03 0.0305 0.0305 0.0518**
[0.024] [0.025] [0.025] [0.024]
French Legal Origin 0.00364 -0.0169 0.0341 0.0491 -0.00478
[0.013] [0.015] [0.022] [0.030] [0.0088]
German Legal Origin -0.0153 -0.0603 0.0133 0.039 -0.0459
[0.029] [0.043] [0.032] [0.027] [0.033]
Scandinavian Legal Origin 0.0980*** -0.0933* 0.182*** 0.0691**
[0.036] [0.049] [0.050] [0.034]
Constant 0.0541 0.0167 -0.223** 0.514*** 0.0815**
[0.043] [0.053] [0.090] [0.14] [0.037]
Observations 721 397 324 220 501
Number of countries 133 133 90 23 110
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
51
Table V.3
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP and Structural Factors
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5]
Structural Factor: Income Gini Govt Share of Foreign Aid Natural Res Agric Share
GDP to GDP Exports (%) in GDP
Lagged LAMRIG -0.192*** -0.162*** -0.146*** -0.169*** -0.134***
[0.053] [0.019] [0.017] [0.019] [0.020]
French Legal Origin 0.0917** 0.103*** 0.0962*** 0.104*** 0.0945***
[0.043] [0.021] [0.020] [0.020] [0.023]
German Legal Origin 0.0746 0.0383 0.0304 0.059 0.0153
[0.066] [0.039] [0.035] [0.039] [0.032]
Scandinavian Legal Origin 0.168** 0.173*** 0.168*** 0.198*** 0.171***
[0.080] [0.023] [0.022] [0.031] [0.024]
Socialist Legal Origin 0.120** 0.169*** 0.152*** 0.178*** 0.210***
[0.049] [0.035] [0.041] [0.041] [0.027]
Log Per Capita GDP -0.0331*** -0.00199 -0.00464 -0.00731 0.00501
[0.013] [0.0060] [0.0060] [0.0056] [0.014]
Lagged Structural Factor -0.00114 0.000795 0.000442 0.000479 0.0325
[0.0017] [0.00066] [0.00088] [0.00031] [0.060]
Constant 0.567*** 0.203*** 0.220*** 0.261*** 0.122
[0.17] [0.053] [0.048] [0.048] [0.12]
Observations 150 600 539 599 415
Number of countries 101 122 120 125 96
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
52
Table V.4
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP and Economic Crises/Shocks
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5]
Economic Crises/Shocks: Debt Crises High Inflation Max Fall of
Years of
Negative
Current
Account
(>30% p.a.) GDP GDP Growth Crises
Lagged LAMRIG -0.181*** -0.234*** -0.161*** -0.156*** -0.147***
[0.031] [0.031] [0.018] [0.018] [0.016]
French Legal Origin 0.103*** 0.133*** 0.100*** 0.0978*** 0.101***
[0.027] [0.029] [0.020] [0.020] [0.021]
German Legal Origin 0.048 0.0924* 0.0405 0.039 0.0254
[0.044] [0.053] [0.037] [0.035] [0.031]
Scandinavian Legal Origin 0.172*** 0.274*** 0.171*** 0.170*** 0.166***
[0.045] [0.043] [0.022] [0.022] [0.033]
Socialist Legal Origin 0.209*** 0.283*** 0.183*** 0.178*** 0.172***
[0.042] [0.043] [0.040] [0.036] [0.056]
Log Per Capita GDP -0.0149** -0.0154** -0.00412 -0.00454 -0.00467
[0.0066] [0.0069] [0.0058] [0.0056] [0.0057]
Lagged Economic Crises -0.00119 0.02 0.000592 0.0104 0.0297
[0.012] [0.023] [0.0016] [0.011] [0.021]
Constant 0.349*** 0.408*** 0.234*** 0.228*** 0.215***
[0.075] [0.068] [0.047] [0.046] [0.047]
Observations 453 538 606 606 528
Number of countries 124 123 125 125 118
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
53
Table V.5
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP and Political Crises/Factors
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5] [6]
Political Crises/Factors: Democracy POLCON Assassinations Strikes International Civil war
conflict (war) (intensity)
Lagged LAMRIG -0.146*** -0.212*** -0.145*** -0.147*** -0.181*** -0.153***
[0.019] [0.028] [0.017] [0.017] [0.037] [0.018]
French Legal Origin 0.0889*** 0.122*** 0.0976*** 0.0976*** 0.0913*** 0.115***
[0.020] [0.026] [0.020] [0.020] [0.030] [0.025]
German Legal Origin 0.0471 0.0997** 0.0547 0.0561 0.137** 0.0930***
[0.031] [0.044] [0.035] [0.036] [0.060] [0.030]
Scandinavian Legal Origin 0.181*** 0.275*** 0.163*** 0.165*** 0.184***
[0.024] [0.044] [0.022] [0.022] [0.025]
Socialist Legal Origin 0.145*** 0.189*** 0.153*** 0.150*** 0.180*** 0.178***
[0.040] [0.048] [0.041] [0.041] [0.062] [0.058]
Log Per Capita GDP -0.000665 -0.00873 -0.00529 -0.00527 -0.00631 -0.000246
[0.0081] [0.012] [0.0053] [0.0055] [0.0085] [0.0058]
Lagged Political Crises -0.00231 -0.0534 -0.0129* -0.0133 -0.0092 0.00743
[0.0030] [0.060] [0.0071] [0.015] [0.0068] [0.0052]
Constant 0.203*** 0.350*** 0.227*** 0.229*** 0.286*** 0.175***
[0.048] [0.077] [0.045] [0.046] [0.065] [0.048]
Observations 570 538 567 567 316 485
Number of countries 119 122 120 120 76 93
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
54
Table V.6
Labor Market Reform, Inertia, Legal Origins, Per Capita GDP, Trade and Financial Reforms
Dependent variable: Change of LAMRIG, Labor Market Rigidity
Random-Effects Panel Estimator with Standard Errors Clustered at Country Level
[1] [2] [3] [4] [5] [6]
Other Reforms
Wacziarg
Open C.N.Hsiao PWT openk BMP Credit Private Financial Ref
Uninterrupted Trade Lib Sector/GDP Index
Lagged LAMRIG -0.163*** -0.163*** -0.148*** -0.211*** -0.194*** -0.238***
[0.019] [0.019] [0.018] [0.025] [0.027] [0.028]
French Legal Origin 0.113*** 0.111*** 0.0979*** 0.130*** 0.130*** 0.145***
[0.022] [0.021] [0.021] [0.024] [0.031] [0.028]
German Legal Origin 0.0489 0.0556 0.0416 0.0649 0.0312 0.104**
[0.040] [0.041] [0.032] [0.052] [0.053] [0.050]
Scandinavian Legal Origin 0.167*** 0.163*** 0.165*** 0.265*** 0.232*** 0.284***
[0.022] [0.022] [0.023] [0.040] [0.048] [0.040]
Socialist Legal Origin 0.240*** 0.207*** 0.155*** 0.279*** 0.233***
[0.040] [0.046] [0.038] [0.030] [0.050]
Log Per Capita GDP -0.0213** -0.0196** -0.00396 -0.0147** -0.0137* -0.0161**
[0.0089] [0.0086] [0.0053] [0.0070] [0.0080] [0.0072]
Lagged Other Reforms 0.0827*** 0.0702** 0.00000975 -0.000003*** -0.000000120* -0.0277
[0.028] [0.027] [0.00013] [0.00000027] [0.000000072] [0.034]
Constant 0.336*** 0.326*** 0.219*** 0.371*** 0.347*** 0.437***
[0.066] [0.065] [0.044] [0.062] [0.077] [0.068]
Observations 588 580 579 514 326 538
Number of countries 114 120 117 108 84 118
Notes: Robust standard errors in
brackets,
*** denotes significant at 1%, ** at 5% and * at 10%.
doc_721490958.pdf