Description
Post-communism is a term describing the period of political and economic transformation or "transition" in former communist states located in parts of Europe and Asia, in which new governments aimed to create free market-oriented capitalist economies.
ABSTRACT
Title of dissertation: PRIVATIZATION AND NEW ENTRY IN
POST-COMMUNIST TRANSITION:
THE IMPACT ON RESTRUCTURING
Polina Vlasenko, Doctor of Philosophy, 2004
Dissertation directed by: Professor Allan Drazen
Department of Economics
This work suggests a connection between the lack of restructuring of the privatized
?rms in transition and the high level of regulatory barriers faced by the private sector. In
this work I suggest that the potential entry of the newly created private ?rms signi?cantly
a?ects the incentives of the privatized ?rms regarding restructuring. Since the entry of
new businesses is threatening to the ine?cient existing enterprises, managers of these
enterprises have an incentive to use their political power to restrict the new entry. Given
that the restructuring often involves politically unpopular measures, such as shedding of
excess labor, politicians may prefer to see no restructuring of the privatized ?rms, thus
creating a possibility for the privatized enterprises to lobby with politicians for the creation
of entry barriers.
The ?rst chapter presents the background on privatization and restructuring in
transition and reviews the existing literature on the subject. It also outlines the argument
for the connection between new entry and the behavior of the existing enterprises.
The second chapter presents an analytical model that investigates the conditions
under which managers of the existing enterprises are likely to be successful in lobbying
with politicians for the restriction of new entry. This chapter also discusses some examples
of policies that would make such lobbying less likely, thus bringing an economy closer to
the e?cient equilibrium with more restructuring and low entry barriers.
The third chapter uses data from the World Business Environment Survey to explore
whether the threat of new entry induces managers of the existing ?rms to lobby with
politicians, resulting in high entry barriers and low restructuring by the existing ?rms.
Speci?cally, I investigate how the competition created by the new entrants a?ects the
probability that high regulatory barriers are erected. The results indicate that the presence
of excess employment (i.e. lack of restructuring) at the existing ?rms coincides with the
new entrants facing high regulatory barriers. Furthermore, higher competition from the
new entrants may result in new ?rms facing higher entry barriers, lending support to
the argument that the level of regulatory barriers is in?uenced by the existing ?rms via
lobbying.
PRIVATIZATION AND NEW ENRTY IN POST-COMMUNIST
TRANSITION: THE IMPACT ON RESTRUCTURING
by
Polina Vlasenko
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial ful?llment
of the requirements for the degree of
Doctor of Philosophy
2004
Advisory Commmittee:
Professor Allan Drazen, Chair/Advisor
Professor Roger Betancourt
Professor Nuno Lim˜ao
Professor Peter Murrell
Professor Margaret M. Pearson
c Copyright by
Polina Vlasenko
2004
TABLE OF CONTENTS
List of Tables iv
List of Figures vi
1 Introduction 1
1.1 Background and Literature on Privatization and Restructuring in Transition 1
1.2 Incorporating New Entry – Description of the Theoretical Framework . . . 4
1.3 Outline of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Theoretical Model 13
2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Political Bene?ts from Excess Employment . . . . . . . . . . . . . . . . . . 14
2.2.1 General setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3 Political Bene?ts From Total Employment . . . . . . . . . . . . . . . . . . . 24
2.3.1 General Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.2 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4 Applications and Policy Relevance . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1 Stylized Facts Revisited . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.2 Policy Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3 Empirical Evidence 36
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Data and Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3 Estimation Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4 Interpretation of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4.1 Barriers faced by new ?rms vs. Excess employment at old ?rms . . . 49
3.4.2 Alternative Speci?cations and Robustness Checks . . . . . . . . . . . 59
3.4.3 Barriers faced by the new ?rms vs. Threat to the old ?rms from
new entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
ii
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
A Indi?erence Curves for the Case of ? < 1. 75
A.1 Objective Functions and General Setup . . . . . . . . . . . . . . . . . . . . 75
A.2 E?ect of ? ? on IC
m
and IC
p
. . . . . . . . . . . . . . . . . . . . . . . . . . 78
A.3 E?ect of ? ? on equilibrium values on R and L . . . . . . . . . . . . . . . . 79
B Characterization of the Contract Curve. 81
Bibliography 85
iii
LIST OF TABLES
3.1 Transition countries in WBES dataset: number of ?rms and fraction of old
and new ?rms by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2 Variables and Relevant Survey Questions . . . . . . . . . . . . . . . . . . . 40
3.3 Variables and Relevant Survey Questions (cont’d) . . . . . . . . . . . . . . . 41
3.4 Barriers (TxReg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 50
3.5 Barriers (AntComp) faced by the new ?rms and excess employment at the
old ?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . 51
3.6 Higher barriers (bl reg) faced by the new ?rms correspond to higher fraction
of the old ?rms with excess employment (ordered probit regressions). . . . . 52
3.7 Bariers (lab reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 53
3.8 Bariers (env reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 54
3.9 Barriers (fir reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 55
3.10 Barriers (hit reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 56
3.11 Barriers (tadm reg) faced by the new ?rms and excess employment at the
old ?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . 57
3.12 Level of barriers faced by the old ?rms is not systematically positively
related to the presence of excess employment at the old ?rms (ordered
probit regressions). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.13 Level of barriers faced by the old ?rms is not positively related to the
presence of excess employment at the old ?rms (ordered probit regressions,
sample of old ?rms restricted to ?rms established prior to 1992). . . . . . . 62
3.14 Barriers faced by the new ?rms and non-restructuring by the old ?rms, as
measured by the change in investment (ordered probit regressions). . . . . . 63
3.15 Barriers faced by the new ?rms and excess employment at the old ?rms
(ordered probit regressions with country-speci?c intercept). . . . . . . . . . 64
iv
3.16 Higher barriers (TxReg) faced by the new ?rms correspond to higher com-
petition faced by the old ?rms (ordered probit regressions). . . . . . . . . . 66
3.17 AntComp barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions). . . . . . . . . . . . . . . 67
3.18 Higher barriers (bl reg) faced by the new ?rms correspond to higher com-
petition from new entry faced by the old ?rms (ordered probit regressions). 68
3.19 env reg barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions). . . . . . . . . . . . . . . 69
3.20 fir reg barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions). . . . . . . . . . . . . . . 70
v
LIST OF FIGURES
2.1 Pro?t of an old ?rm as a function of entry barriers. . . . . . . . . . . . . . . 16
2.2 Politician’s bene?ts from excess employment. . . . . . . . . . . . . . . . . . 16
2.3 Tax collections from the new ?rms as a function of entry barriers. . . . . . . 16
2.4 Politician will choose R
?
? 0 for any given level of excess employment
¯
L;
?R
?
??
< 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Manager’s and politician’s indi?erence curves with ? = 1. . . . . . . . . . . 19
2.6 As w ?, IC
m
becomes steeper. ? = 1. . . . . . . . . . . . . . . . . . . . . . 20
2.7 As ?
?
(R) ?, IC
m
becomes ?atter. ? = 1. . . . . . . . . . . . . . . . . . . . . 20
2.8 IC
p
becomes steeper due to B
?
(L) ?. ? = 1. . . . . . . . . . . . . . . . . . . 22
2.9 IC
p
becomes ?atter due to m ?. ? = 1. . . . . . . . . . . . . . . . . . . . . 22
2.10 IC
p
becomes ?atter due to |N
?
(R)| ?. ? = 1. . . . . . . . . . . . . . . . . . 23
2.11 No bargaining is possible. ? = 1. . . . . . . . . . . . . . . . . . . . . . . . . 23
2.12 For any given level of barriers R, manager would choose employment on
ˆ
L
RF
without bargaining. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.13 Politician chooses R
?
> 0 if bargaining is not allowed. ? = 1. . . . . . . . . 27
2.14 Politician chooses R
?
= 0 if bargaining is not allowed. ? = 1. . . . . . . . . 28
2.15 Indi?erence curves for manager and politician. ? = 1. . . . . . . . . . . . . 28
2.16 IC
m
shifts due to w ? or ?
R
? . ? = 1. . . . . . . . . . . . . . . . . . . . . . 30
2.17 IC
p
shifts due to B
?
(·) ? or m ? or |N
?
(·)| ?. ? = 1. . . . . . . . . . . . . . . 30
A.1 Slope of the politician’s indi?erence curve for case ? < 1. . . . . . . . . . . 77
A.2 Politician’s indi?erence curve for case ? < 1. . . . . . . . . . . . . . . . . . 77
A.3 Indi?erence curves of the manager and the politician for the case of ? < 1. . 78
A.4 Shift in the politician’s indi?erence curve when ? ?; ?
1
> ?
0
. . . . . . . . . 80
A.5 Shift in the politician’s indi?erence curve when ? ?; ?
1
> ?
0
(di?erent case). 80
B.1 Contract curve assuming B
???
(L) < 0, ?
???
(R) > 0, ?N
???
(L) < 0. . . . . . . . 82
B.2 Contract curve in the general case. . . . . . . . . . . . . . . . . . . . . . . . 82
B.3 Shift in the contract curve when w ?. . . . . . . . . . . . . . . . . . . . . . . 83
B.4 Shift in the contract curve when ?
?
(·) ? (assuming ?
??
(·) does not change). . 84
B.5 Shift in the contract curve when B
?
(·) ? (assuming B
??
(·) does not change). 84
vi
B.6 Shift in the contract curve when |N
?
(·)| ? (assuming N
??
(·) does not change). 84
B.7 Shift in the contract curve when m ?. . . . . . . . . . . . . . . . . . . . . . 84
vii
Chapter 1
Introduction
1.1 Background and Literature on Privatization and Restructuring in Transition
Large-scale privatization of state-owned enterprises is one of the major ingridients of the
market-oriented reforms in most transition countries. State-owned enterprises are priva-
tized in order to introduce market incentives for the ?rm and its manager in hopes that
it will improve e?ciency and performance of the ?rm. However, success of privatization
programs varies widely across countries. A large literature on privatization in transition
generally ?nds that success of policies aimed at inducing enterprise restructuring varies
widely depending on reform strategy and, furthermore, similar policies have di?erent ef-
fects in di?erent regions and countries.
Across transition countries we observe that privatization has not always led to the
restructuring and improvement in performance of the privatized ?rms. A large number
of empirical studies investigated performance of privatized ?rms in transition economies
trying to identify factors that in?uence restructuring. An extensive survey of empirical
studies on the determinants of enterprise restructuring in transition economies with quan-
titative summary of the results can be found in Djankov and Murrell (2002). They identify
a number of factors a?ecting enterprise restructuring: ownership structure of the enter-
prise (state owned, privatized, concentration of ownership shares among di?erent owners),
identity of the owners (most notably, insiders vs. outsiders in privatized enterprises), ex-
tent of the competition faced by the enterprise, presence of the soft budget constraints,
etc. In addition, the impact of these factors seems to di?er by region. Usually regional
comparison is drawn between Eastern Europe and countries of the Commonwealth of In-
dependent States (denoted CIS, it includes all former republics of the Soviet Union except
for Estonia, Latvia and Lithuania).
The main empirical ?ndings of the literature on enterprise restructuring can be
summarized as follows (this summary closely follows summary ?ndings in Djankov and
Murrell (2002)). Privatization leads to more enterprise restructuring on average, but the
e?ect depends on the identity of owners. Privatization to outsiders with concentrated
ownership (as opposed to insiders, such as managers and workers) has the largest positive
e?ect on restructuring, both in Eastern Europe and in the CIS. Privatization to workers
1
did not have positive e?ect on restructuring in Eastern Europe and had a signi?cant
negative e?ect in the CIS. For example, Frydman et al. (1999) ?nds that privatization
has no bene?cial e?ect on ?rm performance for the ?rms controlled by insider owners,
but has pronounced positive e?ect on ?rms with outsider owners. They also found that
insider-owned ?rms are virtually indistinguishable in their performance from state-owned
?rms, except that they increase their employment by more (!!!) than the state-owned
?rms do. Empirical studies also ?nd that elimination (or reduction) of soft budgets has a
positive e?ect on restructuring. Increased competition enhances restructuring in Eastern
Europe, but not in the CIS. Furthermore, import competition was even damaging to the
restructuring in many CIS countries.
Several studies have attempted to consider the e?ect of privatization taking into
account other reforms necessary in transition. They usually ?nd that privatization alone
may not be enough to produce improvements in enterprise performance – other comple-
mentary reforms, such as hardening of the soft budgets, institutions for protection of
property rights, and the ability of owners to monitor and control management, have to
be in place for privatization to have a positive e?ect (see, for example, Sachs, Zinnes and
Eilat (2000)).
Finally, several ?ndings in the literature support the view that bringing in new
managers (new human capital), as opposed to simply giving ownership incentives to the
old managers, leads to restructuring and improved performance of the enterprises (Djankov
and Murrell (2002), Barberis, N. et al. (1996)).
Several theoretical models exist that predict outcomes consistent with these em-
pirical ?ndings. There is a strand of literature that argues that privatization leads to
improved enterprise performance to the extent that it provides equity incentives to the
manager. If managerial e?ort is not observable then giving the manager high ownership
of the enterprise induces him to exert higher e?ort, thus increasing enterprise e?ciency
(see Holmstrom (1979)). However, empirical literature on transition economies demon-
strates that privatization does not always lead to restructuring, which means that giving
equity incentives to the managers does not always work. Thus, in the context of transition
economies, there are theoretical models demonstrating that giving ownership incentives
to the managers may not produce restructuring. Speci?cally, Shleifer and Vishny (1994)
and Boycko, Shleifer, Vishny (1996) show that politicians may in?uence managers of the
2
privatized enterprises and thus prevent them from restructuring. Speci?cally, they argue
that politicians may have political reasons to prefer no restructuring, for example, if politi-
cian derives political bene?ts from high employment at the enterprise and restructuring
involves layo?s. In such case, the politician will try to a?ect employment decision of the
privatized enterprise to get them to employ more workers than is economically e?cient.
In exchange the politician would provide subsidies to the enterprise. Such collusion is
possible as long as marginal bene?ts from extra employment for the politician is higher
than his marginal cost of providing subsidies to the ?rm. Thus, these models provide an
explanation as to why privatization does not necessarily lead to restructuring.
A di?erent theoretical argument calls attention to what can be termed the human
capital argument for restructuring. This argument emphasizes that managers need to not
only have correct incentives, but also an appropriate human capital to be able to function
e?ectively, thus role of managerial turnover is important (see Rosen (1992)). Under this
argument privatization works because it brings new managers with the appropriate human
capital to the enterprise. Without the right human capital, giving cash ?ow incentives
to the manager would not be conductive to restructuring. This theoretical argument was
tested in Barberis, N. et al. (1996) for the case of privatization in Russia and they found
that bringing in new managers (the ones with the new human capital) is more likely to
lead to restructuring than privatizing the business to the existing manager.
One of the “stylized facts” of empirical privatization literature repeated in many
studies is that privatization to outsiders is more conductive to restructuring than privati-
zation to insiders (managers and workers). However, I am not aware of theoretical models
that speci?cally explain why this must be the case.
Another empirical ?nding mentioned above is concerned with the e?ects of competi-
tion. As was indicated above, increased competition had a positive e?ect on restructuring
in the Eastern Europe but negative in the CIS (especially for the case of he import com-
petition). Again, I am not aware of a convincing theoretical explanations of this result.
Thus, here are some questions and empirical “stylized facts” that a theoretical model
of privatization should be able to explain:
1. Why is it that privatization does not always lead to restructuring?
2. Why are the outside owners more conductive to restructuring than insiders?
3
3. Why are the new managers more likely to undertake restructuring than the old
managers, even if the old managers are given ownership incentives?
4. Why does an increased competition sometimes lead to less rather than more restruc-
turing?
5. Why do e?ects of seemingly similar reforms di?er in Eastern Europe and CIS?
Answers to some of these questions were given in the theoretical literature to various
extent. For example, Shleifer and Vishny (1994) provides a convincing explanation for
why privatization may not always work (question 1 above). Also, one possible answer to
question 3 is provided by the human capital argument. In a centrally planned economy
managers of the state enterprises were presumably selected for their ability to deal with
ministerial o?cial, negotiate favorable terms for allocation of inputs in the shortage econ-
omy, address political concerns, lobby for assistance etc. In contrast, managers of private
?rms are selected for their ability to run the ?rms e?ciently in economic terms. Thus,
privatization works to the extent that it brings such new managers to run the privatized
enterprises. This gives a reason why new managers might be more likely to restructure.
However, I feel that there is a room for additional theoretical explanations. I hope to con-
vince the reader that the theoretical framework developed here can be fruitfully applied
to answer several questions on the list above.
1.2 Incorporating New Entry – Description of the Theoretical Framework
Here I propose a theoretical framework for analyzing privatization in the transition economies
that will provide answers to some of the questions posed at the end of the previous section.
We start with two key observations:
• In many transition countries privatization of state enterprises coincided with liber-
alization reform that allowed creation and entry of new private ?rms, which was
forbidden or restricted under the centrally planned system.
• While the development of the private sector is widely believed to be essential for
successful transition reforms, often the new private businesses face many obstacles -
extensive regulations, high taxation, corruption etc – when trying to establish and
operate their business.
4
I believe that these two observations are interconnected. The entry of new private
?rms, or even simply the possibility of such entry, a?ects incentives of the existing ?rms
regarding restructuring. Since, in transition countries, the new entering ?rms are likely to
be much more e?cient than the existing ones, existing ?rms realize that they would su?er
from competition created by the new entrants. Therefore, the existing ?rms (whether
state-owned or privatized) would like to see the entry of new ?rms restricted. One way to
restrict the new entry would be for the existing ?rms to lobby with politicians demanding
creation of the entry barriers which would limit the creation of new ?rms. This would
explain why the entry barriers for new businesses are high in many transition countries.
But why would such lobbying necessarily lead to less restructuring? When the existing
?rms lobby with politicians demanding the creation of entry barriers, they have to provide
politician with incentives to ful?ll their demands. In many cases, a politician may be in-
terested in preventing ?rms from restructuring, e.g. for the same reasons as in Shleifer and
Vishny (1994) – politician derives political bene?ts from excess employment and restruc-
turing involves layo?s. Thus, managers of the existing ?rms would lobby with politicians
demanding the creation of entry barriers in exchange for keeping their employment high
(i.e. not restructuring).
Why do I think that such lobbying is likely to occur and is likely to be successful?
First, we know that in many transition countries the politicians often did not get replaced
since the pre-reform period. By ”politicians” here I mean ministry o?cials, local gov-
ernment o?cials etc. In many transition countries, certainly in the former Soviet Union,
many of these ”lower level politicians” did not change, they are the same people holding
essentially the same responsibilities even if the title of their job has changed. In addition
to large amounts of anecdotal evidence, Shleifer (1997) refers to a study that compared
political and economic elites in Poland and Russia in 1993
1
. The study found that in Rus-
sia 83% of the current political elite and 53% of the current economic elite were former
Communist Party members. For comparison, in Poland they found that only 30% of the
current political elite were former Party members. This provides evidence that in some
transition countries the politicians in power (whether in political or economic elite) have
not changed much since the pre-transition period.
1
Szelenyi et al. (1995), reviewed by Karpinski (1996)
5
Second, in many transition countries state enterprises were privatized to their man-
agers. The managers of enterprises in a centrally planned economy are selected for their
ability to negotiate with ministry o?cials, lobby for assistance, get preferred allocation
of inputs etc., and thus they have good ties with the politicians. When, during privati-
zation, these managers get ownership of the ?rm they would still retain their ties with
the politicians. Furthermore, the fact that many politicians did not get replaced since the
pre-reform period, makes it even easier for the managers of (now privatized) ?rms to lobby
with politicians for their preferred policies. Clearly, the old managers (the managers that
kept their position since pre-transition period) will ?nd it easier to lobby with the politi-
cians than the new managers. This may be an additional reason why the old managers
are less likely to restructure.
The idea that formerly state-owned ?rms have good ties with politicians is supported
by Hellman et al (2000). They look at three types of interactions between ?rms at the
state: administrative corruption (payment of bribes “to get things done”), state capture
(purchase of laws, decrees and regulations), and in?uence (ability to in?uence content and
application of laws and regulations without necessarily resorting to uno?cial payments).
Here I argue that the existing ?rms would use their in?uence to restrict new entry. This
is consistent with the ?nding in Hellman et al (2000) that “In?uential ?rms appear to
be classic incumbent ?rms inherited from the socialist system. They are large, usually
state-owned, with good access to public o?cials and a dominant position in their own
market.” (Hellman et al (2000), p.15)
Thus, under certain plausible conditions, managers of privatized ?rms would lobby
for the creation of entry barriers. The politician would agree to such demands if the ?rms
can provide him with some political bene?ts. For example, if the ?rms are able to organize
their workers to vote for a particular politician, then the politician would be interested
in keeping these ?rms happy and he also would want to have employment at these ?rms
high. Therefore, an economy can end up in a situation when existing ?rms lobby with
politicians demanding creation of entry barriers and in exchange they are o?ering political
bene?ts in the form of ine?ciently high employment, which means that these ?rms do not
restructure. As a result of such lobbying, economy ends up in an equilibrium with high
entry barriers and no restructuring.
6
In order for this story to convincingly apply to transition countries, we have to
establish the following points:
• Entry of new ?rms is threatening for the existing ?rms
• Managers of the existing ?rms prefer to lobby for the creation of entry barriers rather
than to undertake restructuring and compete with the new entrants in the market
setting
• We indeed observe high entry barriers in transition countries
I will now try to provide evidence in support of these points. The new entering ?rms
can a?ect the existing ?rms in one of three ways:
1. Product competition – new ?rms either produce or import products that are sub-
stitutes to the ones produced by the existing ?rms. It is often argued that because
of high ?xed capital requirements, new ?rms would not be able to produce products
competing with the existing producers. However, this does not mean that entry
of new ?rms would not create any competition at all – new ?rms can always im-
port products, since organizing an enterprise for importing ?nished goods has much
lower capital requirements. Given the widespread shortages and poor quality of
consumer products produced in the Soviet Union, imported products became major
competitors for the domestic producers starting in the earliest years of transition.
2. Competition for suppliers – this type of competition would be damaging to the ex-
isting ?rms if their production process requires multiple suppliers and the output
cannot be produced unless all inputs are supplied. Because of the way centrally
planned economies were structured, most large enterprises in transition countries
are likely to have this type of supply structure. As shown in Blanchard and Kremer
(1997), when existing ?rms have this type of supply structure increased opportuni-
ties in the private sector may cause breakdown in the supply chain resulting in a
dramatic decrease in output. Entry of new private ?rms presents exactly this type
of “increased private sector opportunities”.
3. Competition for skilled labor. In centrally planned economies, most certainly in the
Soviet Union, salaries paid to employees essentially did not vary with level of e?ort.
7
Consequently, the level of e?ort exerted by most employees was low. However, some
employees have better quali?cations and/or ability for their jobs, but under the
socialist system they still get the same (usually, low) wage. In contrast, at a private
?rm such employees would be more valuable and therefore would receive higher
wages. At the start of transition reforms in the Soviet Union it was widely believed
that private companies would pay much higher wages to the quali?ed employees
than the state-owned enterprises. This belief turned out to be true, and in many
state-owned enterprises (especially in the service sector, where skilled employees
are the main asset of the enterprise) employees with best skills left as soon as any
opportunities were available in the private sector.
Faced with such competition (or a possibility of such competition), manager of an
existing enterprise has two possible courses of action: make his ?rm more e?cient, so that
it will be able to successfully compete in the marketplace (that is, restructure) or try to
restrict the competition, for example, by lobbying for creation of entry barriers
2
. Which
of these two action the manager would take depends on their relative costs and bene?ts.
I argue that under certain conditions lobbying for the creation of entry barriers will be
chosen over restructuring. Some of these conditions are:
• Human capital of the manager – if the manager simply does not know how to create
an enterprise that functions e?ciently in the market economy, his chances of beating
the competition are slim, and thus he is more likely to choose to lobby for the entry
barriers.
• Lobbying skills or political ties of the manager – as argued above, managers that
remained since the planning era are likely to have very good ties with politicians
because they have been involved in negotiations with politicians throughout their
career. This makes lobbying for the creation of entry barriers relatively less costly,
possibly making it a preferred choice. This argument essentially says that managers
have “comparative advantage in politics” – it is relatively less costly for them to
lobby than to undertake restructuring.
2
I would like to note here that incentives of privatized enterprises are potentially di?erent from those
of the state-owned enterprises. To the extent that the state-owned enterprises are more likely to obtain
subsidies if needed, competition from new entrants would be more of a problem for privatized ?rms than
state-owned ?rms.
8
• Unpro?table enterprises – some of the state-owned enterprises in transition economies
are not pro?table and, furthermore, some of these enterprises cannot be made prof-
itable even if restructured (they are termed negative value added enterprises in the
transition literature). Obviously, managers of such enterprises have only one choice
for keeping their enterprise in existence – lobby for assistance and restrict competi-
tion by creating the entry barriers.
Given that conditions described above can plausibly exist, can we ?nd evidence of
high entry barriers for the new ?rms in transition economies? Djankov et al. (2002) look
at the o?cial regulation of entry of start-up ?rms in 75 countries. They measure the
number of procedures that a new ?rm has to go through in order to become operational
(business registration and post-registration procedures), the length of time it takes, and
the o?cial cost of this process. They ?nd that in most countries the cost of entry is
substantial. In transition countries cost of entry is also high. For example, they ?nd that
in Ukraine a ?rm has to go through 11 procedures, which take 21 business days and cost
20% of per capita GDP. In Georgia, a new ?rm has to go through 12 procedures, which
take 70 business days and cost 28% of per capita GDP, while in Russia a new ?rm has to
go through 16 procedures, which take 69 business days and cost 37% of per capita GDP.
For comparison, a new ?rm in United States has to complete 4 procedures, which take
seven business days and cost less that 1% of per capita GDP.
It should be emphasized that Djankov et al. (2002) consider o?cial number of
procedures, and o?cial time and cost. To the extent that actual experience di?ers from
o?cial regulations, due to delays as well as side payments and bribes that may be required
in order to complete the registration process, their estimates of time and cost of entry
may be understated. It is indeed easy to ?nd anecdotal evidence that delays, bureaucratic
barriers and side payments are widespread in some countries. Consider the following quote
from Shleifer (1997):
Many Russian entrepreneurs, particularly founders of small businesses,
complain about the di?culties of starting and operating a business in Rus-
sia. They always point to multiple permits, inspections, and registrations, all
requiring interactions with multiple o?cials many of whom need to be bribed
before the necessary documents are issued. . . . To compare with the sit-
9
uation in Poland, consider the February 1996 comment of a wealthy Polish
businessman on the di?culty of opening a shop in Poland: ”Oh, it is very,
very di?cult. There are so many shops and so much competition that it is
impossible to make money.”
Entrepreneurs in other transition economies also complain about the di?culty of starting
and operating a business because of di?cult registration procedures, excessive regulation
and taxation, and widespread corruption. Kaufmann (1997) describes a number of bar-
riers that prevent development of private businesses in Ukraine. He lists a number of
factors that are seen by private ?rms as signi?cant barriers to their development: non-
transparent and time consuming company registration, cumbersome and ambiguous tax
laws, customs regulations, and specialized licensing procedures; very high payroll taxes
(in Ukraine employer-paid payroll taxes reached up to 40% of the total wage bill at that
time); highly varying corporate pro?t taxes with tax exemptions that are signi?cant and
discriminatory. Many private ?rms in Ukraine indicate that curruption is widespread, and
practice of uno?cial payments to o?cials, such as tax or health inspectors, is commonplace
and these payments are often sizable.
Hashi and Mladek (2000) use a survey of small and medium businesses in ?ve Eastern
European countries to determine which ?scal and regulatory measures impede the entry of
new ?rms in the early stages of transition. They consider entry barriers by the following
categories: registration and licensing, real estate regulations, labor and employment laws,
?scal regulations (taxes and social security contributions), and export/import regulations.
They found that taxes and other contributions are seen by the enterprises as creating the
largest obstacles, closely followed by renting and leasing of real estate, registration and
licensing, and export-import regulations.
From the above description, we can summarize that new ?rms in transition economies
are faced with the following types of barriers:
• Complicated, time-consuming and poorly de?ned registration and licensing proce-
dures; many of these regulations are ambiguous, thus creating opportunities for
corruption.
• Complicated and ambiguous tax laws that change frequently; often high taxes and
social security contributions.
10
• Limited access to commercial real estate. Often in transition countries the real
estate is owned by local governments, which gives the politicians additional power
over local businesses and creates further opportunities for corruption. Hashi and
Mladek (2000) report that the majority of surveyed ?rms saw rent increase at short
notice and shortage of premises as two main problems with real estate.
• Export-import and foreign exchange regulations that disrupt foreign trade. For ex-
ample, Kaufmann (1997) reports that in Ukraine customs regulations are extensive
and foreign currency purchases and private foreign currency loans are under admin-
istrative control.
• Finally, corruption is reported to be widespread in many of these countries, with
extralegal payments to o?cial as an established practice.
In conclusion, descriptive evidence presented here seems to indicate that new ?rms
face substantial barriers to entry in many transition countries. I also tried to demonstrate
that the entry of new ?rms is likely to be damaging to the existing ?rms, and moreover,
managers of the existing ?rms are likely to use their political ties to restrict competition
from new ?rms by creating the high entry barriers observed in these countries.
1.3 Outline of Thesis
The rest of the thesis is organized as follows. Chapter 2 presents a theoretical model
that describes the interactions between manager of the existing ?rm and a politician.
The model is used to determine which factors a?ect whether the bargaining between the
managers and the politician takes place as well as the likely outcome of such bargaining
for the restructuring and the presence of entry barriers. The model is then used to explain
the empirical stylized facts presented in section 1.1 above and to o?er some examples of
policies that can improve the outcome in terms of restructuring and promotion of new
entry.
Chapter 3 o?ers empirical support for the model presented in chapter 2. Using data
from the World Business Environment Survey conducted by the World Bank, I show that
in transition countries high entry barriers faced by the new ?rms do coincide with the old
?rms not restructuring, and, additionally, higher competitive threat from the new entrants
11
sometimes results in higher entry barriers for the new ?rms. Such ?nding are consistent
with the bargaining equilibrium where the existing ?rms lobby with the politician for the
creation of entry barriers in the face of a threat from new entry.
Appendices A and B contain some extensions and the technical derivations related
to the model in chapter 2.
12
Chapter 2
Theoretical Model
2.1 Overview
In this section I set up a model in which the existing ?rms lobby with politicians demanding
the creation of entry barriers in exchange for keeping their employment high. Such a
model would help us understand which factors a?ect whether such lobbying takes place
and provide insights into which institutional features are important for avoiding such
ine?cient outcomes.
Here is the basic argument. The existing ?rms want to have high entry barriers
because new entry creates competition and hurts their pro?ts. The politician wants to
have employment at state and privatized ?rms as high as possible because he derives
personal political bene?ts from it. I will make the nature of these bene?ts more precise
below. I assume that existing state/privatized ?rms are politically organized and thus can
collectively bargain with the politician. This is consistent with the earlier argument I made
about managers of privatized ?rms having good ties with politicians and it is supported by
?ndings in Hellman et al. (2000) about the in?uential ?rms. The new entering ?rms, on
the other hand, are assumed to not be politically organized and thus they do not engage
in lobbying with the politician. Bargaining between managers of the existing ?rms and
the politician determines the level of employment at the existing ?rms and the level of
entry barriers that the new ?rms would face if they tried to enter.
A note on terminology. In the remainder of this document, by the existing state or
privatized ?rms I mean those enterprises that existed before the start of transition reforms
and before substantial new entry was allowed. From now on, I will call these enterprises
the old ?rms. These old ?rms may be completely or partially privatized or even completely
state-owned. The new entering ?rms will be called the new ?rms and they are private by
de?nition. These are the ?rms that were created after the start of transition reforms once
the creation of private businesses was allowed.
I will set up two versions of the model. The di?erence between them is in the
speci?cation of how the politician derives bene?ts from high employment at the old ?rms.
13
2.2 Political Bene?ts from Excess Employment
2.2.1 General setup
The model has the following players
1
: manager of an old ?rm, politician, Treasury and
the new ?rms. The manager of an old ?rm has control over employment at that ?rm,
and the politician has control over the entry barriers. The manager and the politician
are active players – they can engage in bargaining, which determines the level of entry
barriers and employment at the old ?rms. The new ?rms are passive players, they do not
engage in bargaining, they take the level of entry barriers as given and decide whether or
not to enter and operate their business. If they do enter, they pay taxes to the Treasury.
The Treasury is also a passive player, it does not attempt to in?uence the old ?rms or
the politician. It collects revenues from both the old and the new ?rms in the manner
described below.
We assume that the manager of the old ?rm receives fraction ? of its net pro?ts, and
thus he cares about maximizing net pro?ts of the ?rm. Fraction (1 ? ?) of net pro?ts of
the old ?rm goes to the Treasury. Varying ? from 0 to 1 allows the model to incorporate
partial privatization: ? close to zero means that the ?rm is almost completely state owned
and the manager’s ownership of cash ?ows is low, while ? close to unity means that the
?rm is almost completely private. (1 ??) can also be used to represent the tax on pro?ts
collected by the Treasury.
The entry of new ?rms negatively impacts the pro?ts that the old ?rm can earn.
Since higher entry barriers reduce new entry, the pro?t of the old ?rm increases with the
level of entry barriers.
Thus, the objective function of the manager of an old ?rm is assumed to be:
U
m
= ?(?(R) ?wL) (2.1)
where
?(R) – maximum pro?t that the ?rm can earn with no excess employment;
L – excess employment; L ? 0
w – wage;
R – level of entry barriers, R ? 0; lower barriers lead to more new ?rms entering and thus
1
This version of the model is based on Shleifer and Vishny (1994).
14
pro?ts of the existing ?rms su?er: ?
?
(R) > 0 and ?
??
(R) < 0. We also assume: ?(0) = 0,
lim
R?0
?
?
(R) = ?, and lim
R??
?
?
(R) = 0.
Faced with some level of entry barriers, the manager, in the absence of bargaining
with the politician, will choose some level of employment to maximize pro?ts. This em-
ployment does not show up in the objective function (2.1) because we assume that ?(R)
is already optimized over employment. L denotes only the excess employment, which
represents any workers employed is excess of the pro?t maximizing level of employment.
We assume for simplicity that the production function is Leontief and therefore the ex-
cess workers cannot be productively employed. Thus, the excess workers do not produce
anything, they are simply paid wage w. Figure 2.1 depicts a function ?(R), the shape of
which is consistent with our assumptions about its derivatives.
The politician in this model cares about his personal political bene?ts and about
the Treasury’s revenue. The Treasury’s revenue consists of (1 ? ?) fraction of the old
?rm’s pro?t plus the tax collections from the new ?rms. The politician’s preferences are
summarized by the following objective function:
U
p
= B(L) +m{(1 ??)(?(R) ?wL) +N(R)} (2.2)
where
B(L) – political bene?ts from excess employment; B
?
(L) > 0, B
??
(L) < 0; B(0) = 0,
lim
L?0
B
?
(L) = ?, and lim
L??
B
?
(L) = 0;
N(R) – represents taxes collected from the new entering ?rms; higher barriers mean that
fewer ?rms enter and thus tax collections are lower: N
?
(R) < 0, N
??
(R) < 0; N(0) =
¯
N,
where
¯
N is the maximum amount of taxes that can be collected from the new ?rms,
lim
R?0
N
?
(R) = 0, and N
?
(R = R
max
) ? ?, where R
max
is the level of barriers at which
no new ?rms are entering
2
;
m – weight that the politician puts on Treasury’s revenues.
In this section we assume that the politician derives political bene?ts B(L) only
from excess employment at the old ?rms (not from total employment). Such assumption is
consistent with a story where these excess jobs are ”patronage jobs” – the excess workers
are ”friends and family” of the politician and he derives bene?ts if they are employed.
2
Figures 2.2 and 2.3 depict functions B(L) and N(R) consistent with our assumptions on their deriva-
tives.
15
R
? (R)
Figure 2.1: Pro?t of an old ?rm as a func-
tion of entry barriers.
L
B (L)
Figure 2.2: Politician’s bene?ts from ex-
cess employment.
R
N(R)
Figure 2.3: Tax collections from the new ?rms as a function of entry barriers.
Another story consistent with this formulation of political bene?ts is the case when these
“excess workers” realize that they hold their jobs only due to the politician’s in?uence,
and thus they are more likely to support this politicians in an election.
An alternative interpretation of the objective function of the politician can be
given as follows. B(L) represents some personal bene?ts that the politician receives
from excess employment, while the term in curly braces is a form of “social welfare”.
(1 ? ?)(?
?
(R) ? wL) enters social welfare because it represents the surplus produced by
the old ?rms, and, similarly, N(R) represents the surplus produced by the new ?rms. In
this interpretation, the politician cares about his personal bene?ts and about the social
welfare, and m represents the weight he puts on social welfare.
We now determine the equilibrium values for excess employment and the entry
barriers in the absence of bargaining. This would be our baseline case and would tell us
16
the threat points of the manager and the politician.
For a given level of entry barriers,
¯
R, the manager solves:
max
L
?(?(
¯
R) ?wL) (2.3)
s.t. L ? 0
Obviously, the solution is L
?
= 0. Thus, for any level of entry barriers, in the
absence of bargaining the manager will always choose zero excess employment, which is
the economically e?cient outcome.
For a given level of excess employment,
¯
L, politician solves:
max
R
B(
¯
L) +m{(1 ??)(?(R) ?w
¯
L) +N(R)} (2.4)
s.t. R ? 0
FOC: (1 ??)?
?
(R) = ?N
?
(R).
If ? = 1, then solution is R
?
= 0.
If ? < 1, then we can always ?nd such R
?
> 0 that FOC holds
3
. Moreover,
?R
?
??
< 0 (2.5)
Thus, in the absence of bargaining the politician will choose a positive level of entry
barriers as long as ? < 1.
In order to analyze the e?ect of bargaining we will derive the indi?erence curves
of the manager and the politician in (R, L) space. Totally di?erentiating the manager’s
utility function (2.1), we get the following expressions for the slope of the manager’s
indi?erence curve:
dR
dL
=
w
?
?
(R)
> 0 (2.6)
d
2
R
dL
2
= ?
w?
??
(R)
[?
?
(R)]
3
> 0 (2.7)
Similarly, the expression for the slope of the politician’s indi?erence curve is:
dR
dL
= ?
B
?
(L) ?mw(1 ??)
m[(1 ??)?
?
(R) +N
?
(R)]
(2.8)
3
Given our assumptions on derivatives, ?N
?
(R) is everywhere upward sloping and (1 ? ?)?
?
(R) is
everywhere downward sloping. Moreover, ?N
?
(0) is close to zero, while ?
?
(0) is very large. Therefore,
there will always be a positive value R
?
> 0 that satis?es FOC above. Figure 2.4 shows how the solution
for R
?
depends on the value of ?.
17
The ?rst derivative does not have a monotonic sign and its shape changes with the value
of ?. Putting a sign on the second derivative is even harder. Therefore, here I will analyze
a simpler version of the model where we restrict ? = 1
4
.
Assuming ? = 1, the politician’s indi?erence curve has the following derivatives:
dR
dL
= ?
B
?
(L)
mN
?
(R)
> 0 (2.9)
d
2
R
dL
2
= ?
B
??
(L)mN
?
(R) + [B
?
(L)]
2
N
??
(R)
N
?
(R)
[mN
?
(R)]
2
< 0 (2.10)
Recall that with ? = 1, politician’s threat point is R
?
= 0. We can now plot
indi?erence curves of the politician and the manager, which is done in ?gure 2.5.
When bargaining is not allowed the equilibrium point is the origin, with R
?
= 0 and
L
?
= 0 – no entry barriers and no excess employment. However, when bargaining between
the manager and the politician occurs, the resulting equilibrium will be somewhere on the
contract curve between points A and A
?
in ?gure 2.5. The exact allocation will depend
on the relative bargaining strengths of the manager and the politician. The important
thing to note is that with bargaining, irrespective of where on the contract curve the
equilibrium point is, we will always have R
?
> 0 and L
?
> 0, that is, we have entry
barriers and positive excess employment. Such outcome is not economically e?cient since
it does not promote restructuring of the old ?rms and prevents entry of the new, more
e?cient, ?rms.
2.2.2 Comparative Statics
Certain parameters in the model a?ect the slope of the indi?erence curves and thus can
a?ect the equilibrium allocation. The manager’s indi?erence curve shifts if:
• w ? – excess labor becomes more costly to the ?rm ? IC
m
becomes steeper; this
case is depicted on ?gure 2.6;
• ?
?
(R) ? – entry of the new ?rms is more damaging ? IC
m
becomes ?atter; this case
is depicted on ?gure 2.7.
First we consider case of an increase in w. As w increases, excess labor becomes
more costly to the ?rms, and the manager becomes less willing to trade excess labor for the
4
Appendix A contains the derivation of politician’s indi?erence curve for the case of ? < 1.
18
R
?N’ (R)
(1??
0
)?’ (R)
(1??
1
)?’ (R)
R
*
0
R
*
1
?
1
>?
0
Figure 2.4: Politician will choose R
?
? 0
for any given level of excess employment
¯
L;
?R
?
??
< 0
L
IC
m
R
IC
p
c
c
A’
A
Figure 2.5: Manager’s and politician’s in-
di?erence curves with ? = 1.
entry barriers. This results in a shift of the manager’s indi?erence curve up from IC
m
to
IC
?
m
and a shift of the contract curve from cc to c
?
c
?5
, all of which is depicted on ?gure 2.6.
Notice that if the initial bargaining equilibrium was located between points A
?
and A
??
on
the initial contract curve cc, then after the shift in IC
m
the new bargaining equilibrium,
which will be locates on the BB
?
portion of the new contract curve will de?nitely have a
lower level of L. The bargaining equilibrium is more likely to be on A
?
A
??
section of the
initial contract curve if the politician has greater bargaining power and is able to capture
almost the entire surplus from bargaining. At the opposite extreme, if the manager has
greater bargaining power and is able to force the politician to his reservation utility, an
increase in w would result in both lower L and lower R (movement from pt. A to pt. B on
the graph). The e?ect of an increase in w on the level of barriers for other combinations
of relative bargaining powers is harder to determine, since it would depend not only on
the relative bargaining strength of the manager and the politician but also on the size of
the shift in IC
m
.
Thus, when excess labor becomes more costly for the old ?rm, the equilibrium level
of excess employment resulting from the bargaining process is likely to fall, and it is more
likely to do so if the equilibrium is located at the extremes of the bargaining range, i.e.
either the politician has all of the bargaining power or the manager does. The e?ect on
5
Shifts of the contract curve are derived in Appendix B
19
IC
p
IC
m
R
L
c
c
c’
c’
A
A’
B
A’’
B’
IC
m
’
Figure 2.6: As w ?, IC
m
becomes
steeper. ? = 1.
IC
p
IC
m
R
L
c
c
c’
c’
A
A’
B
B’’
B’
IC
m
’
Figure 2.7: As ?
?
(R) ?, IC
m
becomes
?atter. ? = 1.
the equilibrium level of barriers is de?nite only in the case when the manager has greater
bargaining power and is able to force the politician to his reservation utility – in that case
the equilibrium level of barriers will fall in response to an increase in wages. For other
combinations of relative bargaining powers the e?ect on the level of barriers is ambiguous.
Similarly, when ?
?
(R) increases the entry of new ?rms becomes more damaging to
the old ?rms and IC
m
becomes ?atter (?gure 2.7). In this case, the entry barriers become
more valuable for the manager, and he would be willing to trade more excess employment
for them. Note that there is a part of the new contract curve (section B
?
B
??
on the graph)
that lies outside the initial bargaining range, AA
?
. The level of excess employment is
higher on B
?
B
??
than on any part of AA
?
. If the manager has most of the bargaining
power and is able to force the politician to his reservation utility, then an increase in
?
?
(R) will result in higher L and higher R (movement from pt. A to pt. B on the graph).
On the other hand, if the politician has most of the bargaining power and is able to force
the manager to his reservation utility, then an increase in ?
?
(R) will result in higher L, but
the e?ect on R is ambiguous (movement from pt. A
?
to pt. B
?
). Thus, as the new entry
becomes more damaging for the existing ?rms, the level of excess employment resulting
from the bargaining process is likely to increase if the distribution of bargaining powers
between the manager and the politician is very uneven, while level of barriers is likely to
increase only if the manager has most of the bargaining power.
20
The politician’s indi?erence curve shifts for the following reasons:
• B
?
(L) ? – politician’s marginal bene?ts from excess employment increase ? IC
p
becomes steeper
• |N
?
(R)| ? – with higher barriers Treasury loses tax revenues faster ? IC
p
becomes
?atter
• m ? – politician puts smaller weight on Treasury revenue ? IC
p
becomes steeper
Figure 2.8 shows the case of an increase in B
?
(L), i.e. when the politician’s marginal
bene?ts from excess employment increase thus making excess employment more valuable
to the politician. Note that there is a portion BB
??
of the new contract curve that lies
completely outside of the initial bargaining range AA
?
and the level of entry barriers on
BB
??
is higher than anywhere on AA
?
. Therefore, if the manager has most of the bargaining
power, an increase in B
?
(L) would result in higher R but the e?ect on L is ambiguous
(movement from pt. A to pt. B). On the other hand, if the politician has most of the
bargaining power and is able to force the manager to his reservation utility, an increase
in B
?
(L) would result in both higher L and higher R (movement from pt. A
?
to pt. B
?
on the graph). Thus, as politician’s marginal bene?ts from excess employment increase,
the level of barriers resulting from the bargaining process is likely to increase, while the
level of excess employment would increase only if the politician has most of the bargaining
power.
Figure 2.9 shows the case when IC
p
becomes steeper due to a decrease in m, i.e.
when the weight on the Treasury’s revenue in the politician’s utility function drops, thus
making entry barriers less costly to the politician. This case is similar to the case of
an increase in B
?
(L), the only di?erent being the slope of the new contract curve. The
e?ect on the level of barriers and excess employment is also similar. Speci?cally, as m
decreases, the level of barriers resulting for the bargaining process is likely to increase
for any relative bargaining strengths of the manager and the politician, while the level of
excess employment is more likely to increase if the politician has most of the bargaining
power.
Figure 2.10 shows the case when IC
p
becomes ?atter due to an increase in |N
?
(R)|.
In this case an increase in the entry barriers results in larger loss of revenues for the
21
IC
p
IC
p
’
IC
m
R
L
c
c
c’
c’
A
A’
B
B’’
B’
Figure 2.8: IC
p
becomes steeper due to
B
?
(L) ?. ? = 1.
IC
p
IC
p
’
IC
m
R
L
c
c
c’
c’
A
A’
B
B’’
B’
Figure 2.9: IC
p
becomes ?atter due to
m ?. ? = 1.
Treasury, thus entry barriers become more costly to the politician. We now have a portion
AA
??
of the initial contract curve that lies completely outside of the new bargaining range
BB
?
and has a higher level of R. The bargaining equilibrium is more likely to be on the
AA
??
portion of the initial contract curve if manager has most of the bargaining power.
In this case, increase in |N
?
(R)| would result in lower R, while level of L might increase
or decrease (movement from pt. A to pt. B on the graph). At the other extreme, if the
politician is able to force the manager to his reservation utility, an increase in |N
?
(R)|
would result in both lower R and lower L (movement from pt. A
?
to pt. B
?
). Thus, as tax
collections from the new ?rms become more sensitive to the level of barriers, the level of
barriers resulting for the bargaining process would likely decrease, while the level of excess
employment would decrease only if the politician has most of the bargaining power.
Finally, it is possible to have an extreme case where slopes of IC
m
and IC
p
are
such that no bargaining between the manager and the politician is possible. Such case is
depicted on ?gure 2.11. In this case, even if lobbying is allowed it will not occur and the
economy will not deviate from the e?cient equilibrium of R = 0 and L = 0. Therefore,
any policies that change slopes of IC
m
and IC
p
such as to bring the economy closer to the
case depicted in ?gure 2.11 are desirable from the economic e?ciency standpoint. Based
on the above discussion, IC
m
and IC
p
are likely to be closer to the case depicted on ?gure
2.11 if the excess labor is costly or the new entry if not damaging for the existing ?rms,
and if politician’s marginal bene?t from excess employment is low, if the politician puts a
high weight on the tax collection and if the entry barriers hurt tax collection considerably.
22
IC
p
’
IC
p
IC
m
R
L
c
c
c’
c’
A
A’ B
A’’
B’
Figure 2.10: IC
p
becomes ?atter due to
|N
?
(R)| ?. ? = 1.
L
IC
m
R
IC
p
Figure 2.11: No bargaining is possible.
? = 1.
Some examples of policies that would produce such outcomes are described in section 2.4.
2.2.3 Conclusion
To summarize, the factors described in the comparative statics section in?uence the level
of entry barriers and the level of excess employment achieved in the equilibrium through
two channels:
1. by a?ecting whether the bargaining is possible, i.e. whether we are in the situation
depicted on ?gure 2.11 or not
2. by a?ecting the equilibrium outcome given that the bargaining takes place
Taking into account both of these channels, policies that make excess employment
more costly (increase w) are likely to lead to lower excess employment – both because the
bargaining is less likely to occur and because the bargaining outcome is likely to result in
lower L. Such policies may also lead to a lower level of entry barriers by making bargaining
less likely to occur.
Similarly, a higher competitive threat from new entry (higher ?
?
(R)) is likely to
lead to higher excess employment (both because bargaining is more likely to take place
and because the bargaining outcome is likely to result in higher level of L) and possibly
a higher level of entry barriers (because of higher probability that the bargaining will
take place). This does not mean, however, that competition should be discouraged. It
23
calls to our attention the fact that high competition from the new entrants does not
automatically lead to more restructuring by the old ?rms. If bargaining between the old
?rms and the politician is possible, a higher competitive threat from new entry would give
the old ?rms stronger incentives to lobby for the creation of entry barriers. Therefore, in
order for the competition to promote restructuring some additional steps (such as legal
and administrative reform) need to be taken to eliminate the possibility of bargaining.
Our model also indicates that higher marginal political bene?ts from excess employ-
ment, such as higher bene?ts from holding o?ce, are likely to lead to a higher level of the
entry barriers (both due to higher likelihood of bargaining and a higher level of R in the
bargaining outcome) and possibly a higher level of excess employment (since bargaining
is more likely to occur). For the same reasons, a lower weight put by the politician on
Treasury’s tax collections is also likely to lead to a higher level of barriers and possibly a
higher level of excess employment.
Finally, higher responsiveness of the tax collections from new ?rms to the level of
barriers or higher e?ectiveness of barriers in preventing entry (higher |N
?
(R)|) is likely
to lead to a lower level of barriers (both because of the lower probability of bargaining
taking place and a lower R in the bargaining outcome) and possibly a lower level of excess
employment.
Some examples of policies that would a?ect w, ?
?
(R), B
?
(L), m, and |N
?
(R)| are
discussed in the section 2.4.
2.3 Political Bene?ts From Total Employment
2.3.1 General Setup
In this section I modify the model developed above. We now assume that the politician
derives bene?ts from total employment at the old ?rms. This could be the case, for
example, if people are more likely to vote for the politician if they are employed, or if the
old ?rms are better at organizing their workers to vote for a particular politician
6
.
This new assumption requires some changes in the objective functions. We introduce
new notation.
ˆ
L denotes the pro?t maximizing level of employment, that is, the level of
6
There is anecdotal evidence that the old ?rms in the former Soviet Union do try to in?uence the voting
behavior of their employees, whereas newly created private ?rms never make such attempts.
24
employment that the old ?rm would choose in the absence of bargaining. L stand for
excess employment, as before. Again, as before, we assume that the excess workers do not
produce anything. In this model the politician derives bene?ts from the total employment
L
total
=
ˆ
L+L. With these modi?cations, the objective functions of the manager and the
politician can be written as follows:
U
m
= ?(?(R,
ˆ
L) ?wL) (2.11)
U
p
= B(
ˆ
L +L) +m{(1 ??)(?(R,
ˆ
L) ?wL) +N(R)} (2.12)
For this version of the model I will again consider the simpler case with ? = 1. With
? = 1 the objective functions become,
U
m
= (?(R,
ˆ
L) ?wL)
U
p
= B(
ˆ
L +L) +mN(R)
We assume:
• ?
LL
< 0; ?
R
> 0, ?
RR
< 0; ?
LR
> 0
• B
?
(·) > 0, B
??
(·) < 0
• N
?
(R) < 0, N
??
(R) < 0
Assumptions on the derivatives of N(·) and B(·) are the same as in the earlier
version of the model. Assuming ?
LL
< 0 is standard for pro?t functions and represents
diminishing marginal product of labor. Assumptions ?
R
> 0 and ?
RR
< 0 correspond
to similar assumptions about pro?t function ?(R) in the previous section. Assumption
?
LR
> 0 means that as the level of entry barriers increases, the marginal product of labor
increases at the old ?rms. In particular, this means that as the level of entry barriers
rises, the pro?t maximizing level of employment will also rise, and the manager of the
old ?rm will choose to employ more workers even without bargaining. This assumption is
important for the derivations that follow.
We now proceed as in the previous section: we derive the threat points of the
manager and the politician, derive the shape of their indi?erence curves and see how
shifts in the indi?erence curves might a?ect the bargaining outcome.
25
Threat Points.
Maintaining the assumption of ? = 1, for any given level of barriers,
¯
R, the manager
solves:
max
ˆ
L,L
(?(
¯
R,
ˆ
L) ?wL) (2.13)
s.t. L ? 0,
ˆ
L ? 0
FOC:
?
L
(
¯
R,
ˆ
L) = 0
FOC implicitly de?nes the level of employment that the manager would choose with no
bargaining as a function of the level of barriers. We represent this relationship by the
reaction function
ˆ
L
RF
=
ˆ
L(R). This reaction function represents the level of employment
that the manager would choose for any given level of entry barriers without any interaction
with the politician. Thus, this function represents the manager’s “threat curve”.
The solution for excess employment is, obviously, L = 0.
The slope of the reaction function
ˆ
L
RF
=
ˆ
L(R) is
7
:
dR
dL
= ?
?
LL
?
LR
> 0 (2.14)
For any given level of excess employment,
¯
L, and any given reaction function
ˆ
L
RF
,
the politician solves:
max
R
B(
ˆ
L
RF
+
¯
L) +mN(R) (2.15)
s.t. R ? 0
FOC for an interior solution is:
?
B
?
(·)
mN
?
(R)
= ?
?
LR
?
LL
FOC says that the politician’s indi?erence curve has to be tangent to the manager’s
reaction function. Thus, the resulting level of barriers will be R
?
? 0. Figure 2.13 shows
the case when the politician chooses a positive level of entry barriers in the absence of
bargaining, and ?gure 2.14 shows a case when the politician chooses zero barriers in the
absence of bargaining.
7
Our assumption of ?LR > 0 is crucial for deriving the slope of the reaction function. The reaction
function is shown in ?gure 2.12, where it is drawn as a straight line for simplicity.
26
L
total
=L+L
R
L
RF ^
^
Figure 2.12: For any given level of bar-
riers R, manager would choose employ-
ment on
ˆ
L
RF
without bargaining.
L
total
=L+L
R
IC
p
L
RF
E
R
*
^
^
Figure 2.13: Politician chooses R
?
> 0 if
bargaining is not allowed. ? = 1.
Indi?erence Curves.
We now derive the shape of the indi?erence curves for the manager and the politi-
cian. The slope of manager’s indi?erence curve is given by
dR
d(
ˆ
L +L)
=
w ??
L
(·, ·)
?
R
(·, ·)
(2.16)
and the slope of the politician’s indi?erence curve is given by
dR
d(
ˆ
L +L)
= ?
B
?
(
ˆ
L +L)
mN
?
(R)
> 0 (2.17)
The expression for the slope of the politician’s indi?erence curve is almost identical to
that in the previous section (see equation (2.9)), therefore, we can easily draw it. In
order to draw the indi?erence curve for the manager, note that at the intersection with
the reaction function
ˆ
L
RF
, ?
L
(·, ·) = 0 since the reaction function represents those values
of
ˆ
L that maximize pro?t. If ?
L
(·, ·) = 0, then the slope of the manager’s indi?erence
curve is positive, according to equation (2.16). So, at the intersection with the reaction
function
ˆ
L
RF
, IC
m
is upward sloping. Also notice that for all levels of employment higher
than
ˆ
L
RF
we have ?
L
(·, ·) < 0 and thus IC
m
is de?nitely upward sloping. IC
m
can be
downward sloping only for very low levels of L
total
, which we do not depict on the graphs.
Figure 2.15 shows managers reaction function
ˆ
L
RF
and the indi?erence curves for
the manager and the politician. If bargaining is not allowed the equilibrium point is
27
L
total
=L+L
R
IC
p
L
RF
^
^
Figure 2.14: Politician chooses R
?
= 0 if
bargaining is not allowed. ? = 1.
L
total
=L+L
IC
m
R
IC
p
L
RF
E
R
E
L
E
total
A
A’
c
c
^
^
Figure 2.15: Indi?erence curves for man-
ager and politician. ? = 1.
point E. However, if bargaining is allowed, the equilibrium will be somewhere on the
AA
?
portion of the contract curve, and the levels of entry barriers and employment will
be higher than in the no-bargaining case. On the graph, the level of excess employment
is the di?erence between the value of L
total
on the contract curve and the value of
ˆ
L
RF
corresponding to the same level of R. Of course, excess employment is positive for all
points on the contact curve AA
?
, which means that bargaining between the manager and
the politician results in some excess employment, as well as a higher level of barriers.
2.3.2 Comparative Statics
We now explore the e?ects of shifts in the indi?erence curves. The manager’s indi?erence
curve becomes steeper if either w increases (excess labor becomes costlier) or ?
R
decreases
(new entry becomes less damaging). Such shift in the indi?erence curve is shown in ?gure
2.16
8
. Similar to the earlier version of the model, after the shift in IC
m
the level of excess
employment (and the level of total employment) is likely to fall if the distribution of
bargaining strengths between the manager and the politician is very uneven. In addition,
if the manager has higher bargaining strength and is able to force the politician to his
reservation utility, the equilibrium level of entry barriers is also likely to fall.
The politician’s indi?erence curve becomes steeper if B
?
(·) increases (marginal po-
litical bene?ts from employment increase), m decreases (politician puts lower weight on
Treasury’s revenue), or |N
?
(·)| decreases (tax collections from the new ?rms are less sen-
8
The shift in the contract curve is not shown on the ?gure to keep it visually simple.
28
sitive to the level of barriers). The e?ect of such shift in the indi?erence curve is shown
on ?gure 2.17. Original IC
p
was tangent to
ˆ
L
RF
at point E. When IC
p
becomes steeper,
it will be tangent to
ˆ
L
RF
at a di?erent point, E
?
. Even with no bargaining, the politi-
cian will now choose higher level of barriers corresponding to the allocation at E
?
, which
makes sense since the barriers are now less costly and employment is more valuable for
the politician. In addition, bargaining can still occur. In the case drawn in ?gure 2.17,
once IC
p
shifts to IC
?
p
the equilibrium level of barriers will be higher. Moreover, the level
of barriers will be higher irrespective of the relative bargaining strengths of the politician
and the manager, since the new contract curve lies entirely above the original contract
curve. The e?ect on the excess employment is ambiguous and depends on the size of the
shift in IC
p
and on the slope of the reaction function
ˆ
L
RF
.
Of course, it is possible that the shift in IC
p
is very slight, in which case the new
point of tangency to the reaction function, E
?
, would be very close to the original point E
and the new “bargaining lens” will be very close to the initial one, except that it will be
shifted up somewhat. In such case, it is still possible to get a higher level of barriers in the
new equilibrium if the bargaining outcome is located at the extremes of the contract curve
– either close to IC
p
or close to IC
m
. Therefore, as long as the distribution of bargaining
powers between the manager and the politician is very uneven, we should see an increase
in the level of barriers in this case.
Finally, in this speci?cation of the model, similar to the previous section, it is
possible to have slopes of IC
m
and IC
p
such that no bargaining between the manager and
the politician is possible. Steeper IC
m
and ?atter IC
p
move us closer to this case and
thus policies that have this e?ect on the slopes of indi?erence curves are desirable.
To summarize, conclusions from the model where the politician derives political
bene?ts from total employment are generally similar to the conclusions from the model
where the politician derives bene?ts from excess employment only (see section 2.2.3).
Lower w, higher ?
R
, higher B
?
(·), lower m, and lower |N
?
(·)| make it more likely that
the bargaining would take place, thus leading to both higher barriers and higher excess
employment. In addition, given an uneven distribution of the bargaining strengths between
the manager and the politician, lower w or higher ?
R
is likely to lead to higher excess
employment and higher total employment, while higher B
?
(·), lower m, or lower |N
?
(·)| is
likely to lead to a higher level of barriers.
29
L
total
= L+L
IC
m
R
IC
p
L
RF
E
R
E
L
E
total
A
A’
c
c
IC
m
’
^
^
Figure 2.16: IC
m
shifts due to w ? or
?
R
? . ? = 1.
L
total
=L+L
IC
m
R
IC
p
L
RF
E
R
E
L
E
total
A
A’
c
c
IC
m
’
E’
IC
p
^
^
Figure 2.17: IC
p
shifts due to B
?
(·) ? or
m ? or |N
?
(·)| ?. ? = 1.
2.4 Applications and Policy Relevance
2.4.1 Stylized Facts Revisited
In this section I will demonstrate how the theoretical framework developed above can
be used to explain the “stylized facts” found in the empirical literature on enterprise
restructuring (listed on page 3).
Outsiders vs. Insiders as Owners.
Insider owners are most often the managers of the enterprise who also were its
managers before privatization. Thus, the insider owners usually have better ties with the
politicians than outsiders do, because they were originally selected for their ability to deal
with o?cials, they have dealt with these politicians previously, etc. Outsider owners, on
the other hand, are less likely to have good ties with the politicians. Thus, insider owners
are more likely to engage in lobbying with the politician making the equilibrium with
bargaining a more likely outcome, which results in higher levels of both entry barriers and
excess employment (less restructuring).
New vs. Old Managers.
Here the reasoning is similar to the above example, except that it is even more
straightforward. An old manager, by de?nition has been the manager of the enterprise
under the old planning system. Thus he has all the characteristics (the ability to negotiate
with o?cials, to lobby for allocation of inputs, etc.) that would make it easy for him to
lobby with the politician. Moreover, an old manager is unlikely to have the human capital
30
required to function e?ciently in the market economy. This would make him more likely to
choose lobbying for the entry barriers instead of restructuring and facing the competition.
Thus, with old managers we are much more likely to end up in the equilibrium with
bargaining, while new managers may choose to improve e?ciency of the enterprise instead
and face the competition in a market setting. As shown above, the equilibrium with
bargaining implies less restructuring by the enterprise.
Increased Competition Sometimes Leads to Less Restructuring.
Increased competition can be captures by an increase in ?
?
(R) in the model. The
e?ect of ?
?
(R) ? is depicted on ?gure 2.7. If the relative bargaining powers of the manager
and the politician are very uneven and, in particular, if the politician has most of the
bargaining power and is able to force the manager to the allocation close to his reservation
utility, then ?
?
(R) ? would lead to an increase in the excess employment L, which means
less restructuring. To the extent that politicians have more bargaining power in CIS than
in Eastern Europe, increased competition will tend to produce less restructuring in CIS
but not in Eastern Europe.
In addition, an increase in ?
?
(R) may move the equilibrium from the case where
no bargaining is possible (depicted on ?gure 2.11) to the case where there is scope for
bargaining (depicted on ?gure 2.5), thus leading to higher excess employment and higher
barriers.
Finally, increased competition may shift the manager’s choice toward choosing lob-
bying for entry barriers instead of becoming competitive by restructuring. With increased
competition, restructuring in order to successfully face the competition requires more ef-
fort and better skills in the e?cient operation of the ?rm. Thus, if managers compare
their skills in the e?cient operation of the ?rm to their lobbying skills, with increased
competition some managers may decide to switch to lobbying.
CIS vs. Eastern Europe.
I will o?er some observations that may explain why similar reform policies may
have di?erent e?ects in CIS and Eastern Europe. This can perhaps be explained by
systematically di?erent values of certain institutional parameters in Eastern Europe and
in CIS:
• Eastern Europe may have higher m – politicians are more accountable in Eastern
Europe and therefore they put higher weight on social welfare. Higher m makes the
31
indi?erence curve of the politician ?atter and may eventually lead to the situation
when no bargaining is possible (depicted on ?gure 2.11).
• Some Eastern European countries allowed limited creation of small private businesses
in years prior to the transition reforms and mass privatization, thus an additional
entry of new private businesses may not be as damaging to the exiting ?rms. This
means that ?
?
(R) is lower in Eastern Europe than in the CIS and, consequently, the
equilibrium level of excess employment is lower (the e?ect is opposite from the one
depicted on ?gure 2.7).
• Eastern European countries undertook more decentralization in the last decades of
the planning system than did Soviet Union. As a result, managers of enterprises in
Eastern Europe did not have to rely on politicians for the routine decisions about
operations of the enterprise. Thus, managers’ ties with politicians are arguably not
as strong in Eastern Europe as in CIS. In addition, in Eastern European countries
a larger fraction of politicians was replaced during the democratic reforms, which
would further weaken the ties between existing ?rms and politicians. Hence, the
managers in Eastern Europe are less likely to lobby with politicians, making the
equilibrium with no bargaining more likely.
• Various institutional details may make B
?
(L) lower in Eastern Europe than in the
CIS. For example, more active media may expose collusion and deals between man-
agers and politician, threatening politician’s career; elections are better supervised,
thus making it more di?cult for managers to deliver votes in support of the politi-
cian. Also, it is possible that politician’s “private value of holding an o?ce” is
much higher in the CIS than in the Eastern Europe, for example, because holding
a political o?ce in CIS opens up extensive opportunities for corruption and private
enrichment. B
?
(L) represents politician’s bene?ts from excess employment, which
may be thought of as a product of the increased probability of being (re)elected and
the value of holding an o?ce. If the value of holding o?ce is much higher in CIS
than in Eastern Europe, B
?
(L) will also be higher in CIS. All this would make B
?
(L)
lower in Eastern Europe than in the CIS, making politician’s indi?erence curve ?at-
ter. As a result, bargaining is less likely to take place, and even with bargaining
the equilibrium level of entry barriers and of excess employment may be lower in
32
Eastern Europe.
2.4.2 Policy Relevance
The theoretical framework presented here can be used to evaluate which policies would
reduce the possibility of the existing ?rms lobbying for entry barriers and thus would
promote restructuring and growth of the new private sector. We have shown that in
certain cases lobbying between the manager and the politician is impossible (such case is
depicted on ?gure 2.11). Any policies that make the manager value high entry barriers
relatively less than low excess employment (making his indi?erence curve steeper) and/or
policies that make the politician value tax collections relatively more than the personal
bene?ts he derives from excess employment (making his indi?erence curve ?atter) would
bring us closer to this case. Some examples of these policies are:
• Elimination of barter payments and wage arrears. When excess labor is more costly
to the ?rm (high w), the manager is less willing to trade excess labor for barriers and
thus less likely to lobby with the politician. Barter payments, especially for wages,
and toleration of wage arrears make excess labor less costly to the ?rm. Therefore,
elimination of the barter payments is a desirable policy.
• Uno?cial economy. When marginal bene?ts from barriers are low for the existing
?rms (low ?
?
(R)), managers are less willing to “pay for barriers” with excess employ-
ment. ?
?
(R) is low when either the new entry is not damaging for the existing ?rms
or barriers are not e?ective in preventing the new entry. Strangely enough, uno?cial
economy may help in this case. If it is very easy (not costly) for a ?rm to operate
uno?cially, then high barriers to entry into the o?cial sector are irrelevant. With
extensive uno?cial economy, the new ?rms are threatening to the existing ones but
high entry barriers will not eliminate this threat, thus the existing ?rms would not
expend resources to create them.
There is another way in which uno?cial economy can in?uence the situation. Ac-
cording to our model, if tax collections fall sharply with a rise in the entry barriers
(high |N
?
(R)|), then politician may not be willing to erect entry barriers. Once
again, if it is relatively easy for the new ?rms to operate in uno?cial economy, then
they would escape into uno?cial economy whenever “o?cial” entry barriers rise.
33
Therefore, any increase in the entry barriers would lead to a sharp drop in tax col-
lections from the new ?rms
9
. Of course, this e?ect matters only if politician puts a
positive value on tax collections, i.e. if m > 0.
• Financing local budgets from local tax collections. When the politician cares a lot
about Treasury’s tax collections (high m), he is less willing to create entry barriers
since they hurt tax collections. An example of policy that would make a local
politician put a very large weight on the tax collection is a ?scal arrangement where
the local budget is ?nanced entirely by tax collection from local small and medium
enterprises (as opposed to allocations from a central budget). Since most of the new
?rms are small and medium enterprises, such policy would give the local politician
incentives to promote development of the new private sector.
• Transparent election process and active media. When the politician’s bene?ts from
excess employment are low (low B
?
(L)), he is less willing to trade the entry barriers,
which are costly, for the excess employment, which is not very valuable to him. More
transparent election process, more active media, and low corruption may make B
?
(L)
lower by exposing illegal dealings between managers and the politician and reducing
potential private gains from holding an o?ce. All of this reduces the probability
that the bargaining will take place, leading to lower barriers for the new ?rms and
more restructuring by the old ?rms.
2.5 Conclusions
This chapter presented a model of interaction between the politician and the manager of
an existing enterprise in the environment where new entry is allowed and is potentially
harmful to the existing enterprises. In such environment, managers of the existing en-
terprises have incentives to bargain with the politician for the creation of entry barriers.
We have shown that managers of privatized enterprises are more likely to lobby for the
creation of entry barriers if their ?rms would be substantially harmed by the new entry,
or if the cost of excess employment is low for the ?rms. At the same time, the politician
is more likely to agree to the creation of entry barriers if he derives substantial bene?ts
9
This in no way implies that steps should be taken to promote uno?cial economy. I am simply pointing
out that the existence of uno?cial economy may have positive consequences.
34
from the political support provided by the privatized ?rms, if he puts low weight on the
Treasury’s tax collections, or if the level of entry barriers does not signi?cantly a?ect tax
revenues collected from the new ?rms. When managers of the existing ?rms are successful
in their bargaining with the politician, the economy ends up in an equilibrium with a
higher level of barriers, fewer new ?rms entering, and a higher level of excess employment
(lower restructuring) at the old ?rms.
These ?ndings emphasize that allowing, or even promoting, the entry of new busi-
nesses may not be enough to stimulate development of the new private sector and to
promote competition and restructuring of the old ?rms. Attention should be paid to ad-
ditional institutional and administrative reforms that would prevent the possibility of old
?rms bargaining with politicians.
35
Chapter 3
Empirical Evidence
3.1 Overview
Having presented the theoretical argument in chapter 2, in this chapter we will turn to
providing relevant empirical evidence. The theoretical model outlined in chapter 2 argues
that, under certain conditions, managers of the old ?rms would lobby with politicians
demanding creation of high entry barriers in exchange for keeping employment at their
enterprises at levels above economically e?cient (i.e. not restructuring). Such collusion
is more likely to occur when the new entry is threatening and/or excess employment is
not costly to the old ?rms, and when the politician derives high political bene?ts from
excess employment and does not have incentives to raise the Treasury’s tax collections.
Unfortunately, some of these things are very di?cult to measure. Therefore, I will take
the following approach in this empirical exercise.
On the very basic level the model argues that the level of barriers faced by the new
?rms is in?uenced by what the old ?rms do, i.e. whether they lobby with politicians.
Since lobbying is di?cult to observe, we will look at the factors that would make such
lobbying likely (such as new entry being threatening to the old ?rms, etc) and see if they
can explain the level of barriers faced by the new ?rms. In addition, if the lobbying takes
place, we would expect to see both excess employment and a high level of entry barriers.
Without structurally modeling the channel through which the political lobbying operates,
I will try to see if the presence of excess employment at the old ?rms explains the level of
barriers faced by the new ones.
3.2 Data and Variables
I use data from the World Business Environment Survey (WBES)
1
, conducted by the
World Bank in late-1999 to mid-2000. The survey covers over 10, 000 ?rms in 80 countries,
among them nine countries of Eastern Europe and twelve countries of the former Soviet
Union. The list of countries and the number of ?rms surveyed in each country can be
found in table 3.1. The summary of survey questions used to create relevant variables is
1
c “The World Business Environment Survey” (WBES) 2000, The World Bank Group
36
given in tables 3.2 and 3.3
2
.
The objective of our analysis is to see how the level of barriers faced by the new
?rms is in?uenced by the characteristics of the old ?rms, including their level of excess
employment. However, we need to de?ne more carefully which old ?rms might be inter-
ested in preventing entry of a speci?c new ?rm, i.e. we need to determine what is the
“relevant group” of the old ?rms for each new ?rm. For example, old ?rms in Poland
are not likely to lobby for the creation of entry barriers to limit new entry in Uzbekistan.
Similarly, even within the same country, old ?rms in car manufacturing are not likely to
lobby for barriers to restrict new entry into the construction industry. Conceptually, the
old ?rms in the same industry and the same country would have incentives to act together
to prevent entry of the new ?rms in this industry
3
. WBES survey has some limited infor-
mation about industry classi?cation of ?rms: it classi?es all ?rms into ?ve broad sectors –
manufacturing, services, agriculture, construction, and other. We will use this information
to create variables characterizing “relevant old ?rms”.
The data is organized as follows. Each new ?rm is a unit of observation (which
?rms are considered “new” is precisely de?ned below). For each new ?rm we observe its
characteristics, such as the level of barriers it faces, its age, size, the country and the
sector it belongs to. We also observe characteristics of the old ?rms in the same country
and sector (country-sector cell) as this new ?rm. The idea is that certain characteristics
of the old ?rms in the country-sector cell in?uence the level of barriers faced by the new
?rms in the same country-sector cell.
2
Hellman et al. (2000) describe this data set for transition countries in more detail. They also indicate
that “The sample was structured to be fairly representative of the domestic economies, with speci?c quotas
placed on size, sector, location, and export orientation.” The sample was heavily weighted toward privately
owned ?rms. (Hellman et al. (2000) p. 5)
3
This is not strictly true. Domestic manufacturers might be interested in preventing entry of the
?rms that would import competing products, and these importing ?rms would fall under retail sector,
thus making the old ?rms in manufacturing trying to prevent new entry into the retail services sector.
Therefore, the analysis here would miss this important channel through which new entry in?uences the old
?rms. However, were we able to include this channel, we should ?nd an even stronger relationship between
the characteristics of the old ?rms and the level of barriers faced by the new ?rms. Unfortunately, in this
dataset it is impossible to determine which ?rms are importing competing products, which would bias our
results against ?nding a relationship between the behavior of the old ?rms and the entry barriers faced by
the new ones.
37
I will now describe in more detail how the variables are constructed.
New vs. Old Firms. The new ?rms are those ?rms that were established as
“originally private from the time of start up” (see table 3.3). All other ?rms are treated
as the old ?rms
4
. The fraction of old and new ?rms by country is presented in table 3.1.
A consistency check was performed using the age of ?rms. Those ?rms that claim to be
“originally private from the time of start up” and were established prior to 1989 were
removed from the sample because they would not constitute “the new ?rms” in the sense
of the model described here, and, furthermore, it was impossible for any private ?rms to
exist in Soviet Union prior to 1989.
5
.
Barriers. The measures of barriers are constructed only for the new ?rms. There
are several alternative measures of barriers that can be constructed given the questions in
WBES survey (see table 3.2). These questions ask how problematic are various regulations
and anti-competitive practices for the operation of a ?rm, ranging from “no obstacle” to
“major obstacle”. Since it is di?cult to say a priori that one measure is better than
others, and it is also di?cult to construct a meaningful aggregation of these measures, I
will consider each variable as a separate measure of the level of barriers, with one important
exception.
Two types of regulatory barriers considered in the survey, namely customs regula-
tions and foreign currency/exchange regulations (variables cus reg and frk reg), are only
relevant if a ?rm is engaged in the international trade. Since not all ?rms are engaged in
international trade, these variables may have systematic measurement errors. Speci?cally,
some ?rms may indicate that these types of regulations are no obstacle to the operation
of their business not because there are no barriers in these areas, but simply because the
?rms are not engaged in the international trade and therefore these barriers are irrelevant.
4
This de?nition of the new ?rms might still include some ?rms that should be classi?ed at “old”. It has
been a common practice in transition countries for managers of large state-owned enterprises to register
new ?rms on the basis of their existing enterprises as means of diverting assets. Such falsely new ?rms
are likely to have extremely good ties with politicians and would behave like the old ?rms in our model,
making it more di?cult for us to ?nd the expected relationship between the characteristics of the old ?rm
and level of barriers faces by the new ones. Unfortunately, it is impossible to identify such ?rms in the
data, but we should keep this in mind when interpreting the results.
5
This procedure removed 60 observations from the total of 1673 observations for new ?rms.
38
Country Number of Firms Old Firms (fraction) New Firms (fraction)
CIS
Armenia 125 49.6% 50.4%
Azerbaijan 128 35.2% 64.8%
Belarus 125 65.6% 34.4%
Georgia 129 44.2% 55.8%
Kazakhstan 127 61.4% 38.6%
Kyrgyzstan 125 68.8% 31.2%
Moldova 125 70.4% 29.6%
Russia 525 51.0% 49.0%
Ukraine 225 45.3% 54.7%
Uzbekistan 125 70.4% 29.6%
Baltics
Estonia 132 47.0% 53.0%
Lithuania 112 25.0% 75.0%
Eastern Europe
Bulgaria 125 39.2% 60.8%
Croatia 127 73.2% 26.8%
Czech Rep 137 27.0% 73.0%
Hungary 129 38.0% 62.0%
Poland 225 39.1% 60.9%
Romania 125 33.6% 66.4%
Slovakia 129 41.1% 58.9%
Slovenia 125 68.0% 32.0%
Albania 163 44.2% 55.2%
Total 3288 49.1% 50.9%
Table 3.1: Transition countries in WBES dataset: number of ?rms and fraction of old and
new ?rms by country
39
Variables and Relevant Survey Questions
Survey Question Possible Answers Variable Values
Measures of Barriers
TxReg – General measure of regulatory barriers
AntComp – General measure of anti-competitive barriers
How problematic are these di?erent 1 – no obstacle Answers about
factors for the operation and growth 2 – minor obstacle taxes and regulations and
of your business: ?nancing, 3 – moderate anti-competitive practices are
infrastructure,taxes and regulations, obstacle recorded as ordered variables
policy instability, in?ation, exchange 4 – major obstacle with values {1,2,3,4}.
rate, functioning of the judiciary,
corruption, street crime, organized
crime, anti-competitive practices by
gov’t or private enterprises?
bl reg – Business Licensing Regulations
cus reg – Customs/Foreign Trade Regulations
lab reg – Labor Regulations
frk reg – Foreign Currency/Exchange Regulations
env reg – Environmental Regulations
?r reg – Fire, Safety Regulations
tadm reg – Tax Regulations/Administration
hit reg – High Taxes
How problematic are these di?erent 1 – no obstacle The answers are recorded as
regulatory areas for the operation 2 – minor obstacle ordered variables
and growth of your business: 3 – moderate with values {1,2,3,4}.
business licensing, customs/foreign obstacle
trade regulations, labor regulations, 4 – major obstacle
foreign currency/exchange
regulations, environmental
regulations, ?re, safety regulations,
tax regulations/administration,
high taxes?
Measures of Excess Employment
LgrS, LgeS, LiSni, LndSni
Have your company’s sales, Increased LgrS=1 if empl-t increased by
investment, exports, imports, Decreased more (decreased by less, in %)
employment and debt changed in Did not change than sales; 0 otherwise.
real terms over the last three LgeS=1 if change in empl-t
years? is greater than or equal to the
By what percentage have they Percentage change change in sales; 0 otherwise
increased or decreased? is given LiSni=1 if empl-t increased
while sales decreased or stayed
constant; 0 otherwise
LndSni =1 if empl-t increased
or stayed constant while sales
decreased or stayed constant;
0 otherwise
Table 3.2: Variables and Relevant Survey Questions
40
Variables and Relevant Survey Questions (cont’d)
Survey Question Possible Answers Variable Values
Measures of Competition
comp hi – Intensity of Competition
Thinking of your None comp hi =1 if the
?rm’s major product 1-3 answer is “More ;
line in the domestic More than 3 than 3”; 0 otherwise
market, how many
competitors do you face?
comp SME – Identity of Major Competitor
From which of the 1 - Domestic small and medium enterprise comp SME=1 for
following would you 2 - Domestic large private enterprise answer #1;
say your company 3 - Foreign ?rm producing in 0 otherwise.
faces the biggest domestic market
competitive threat? 4 - State-owned enterprise
5 - Micro-enterprise/informal sector
6 - Legal imports
7 - Smuggled goods
8 - My ?rm has no e?ective competitors
Controls
new – Indicator variable for new ?rms
How was your ?rm 1 - Originally private, from new=1 for
established? time of start up answer #1;
2 - Privatization of a stated-owned ?rm 0 otherwise.
3 - Private subsidiary of a formerly
state-owned ?rm
4 - Joint venture, domestic and foreign
private owners
5 - State-owned company
6 - State owned some stock
in the company
7 - Agricultural co-operative
8 - State collective farm
9 - Transportation co-operative
age – Age of the ?rm in years
In what year was your actual year is given age = 2000 ?year
?rm founded?
sector – Sector/industry to which the ?rm belongs
1 - manufacturing Used to create
2 - services dummy variables
3 - other for sectors
4 - agriculture
5 - construction
size – Size of the ?rm
Based on number 1 - small ( < 50 employees) Used to create
of employees 2 - medium (between 50 and 500) dummy variables
3 - large (over 500 employees) for size
Table 3.3: Variables and Relevant Survey Questions (cont’d)
41
For this reason, I will not use measures of barriers based on variables cus reg and frk reg.
Each measure of barriers is used as a separate dependent variable in the regressions.
It should be noted that these several measures of barriers may be measuring con-
ceptually di?erent things. Generally speaking, regulations can act as the entry bariers
per se (meaning that they make it more di?cult to start a business) or as an additional
burden on businesses that already exist. While both types of regulatory intervention can
be used to limit the development of the new private ?rms, regulations that are speci?cally
designed as the entry barriers would be a pre?erred choice if the old ?rms are trying to
prevent new entry. Regulations that put an additional burden on an already existing
business might also be used to drive out the new private ?rms, therefore, old ?rms may
also be interested in seeing such regulation created. One of the variables in the dataset –
bl reg, business registration and licensing regulations – is the closest measure available of
the purely entry barriers. Therefore, if lobbying between the old ?rms and the politician
takes place, we would expect the e?ect of it to be most visible in bl reg variable. We will
still use the other measures of barriers since the old ?rms might still want to lobby for the
creation of these regulation in order to drive out the existing new ?rms.
Excess Employment. The measures of excess employment are constructed only
for the old ?rms. The WBES survey provides information on whether company’s sales,
employment, investment, exports, imports, and debt changed over the past three years
and by what percentage they have changed (see table 3.2). In order to create the excess
employment variables I look at the relative movement of sales and employment. Obvi-
ously, excess employment can be present with any combination of movement in sales and
employment, even if both sales and employment decrease (or increase). Therefore, I try
to identify those cases where excess employment is most likely to be present. I consider
several alternative measures of excess employment:
• LgrS is a dummy variable that equals one if employment at the ?rm has increased by
more (or decreased by less, in percentage terms) that sales did, i.e. if the change in
employment is greater than the change in sales with decrease denoted by a negative
change
• LgeS is a dummy variable that equals one if the change in employment at the ?rm
is greater than or equal to the change in sales, with decrease denoted by a negative
42
change
• LiSni is a dummy variable that equals one if employment at the ?rm has increased,
while sales decreased or stayed constant
• LndSni is a dummy variable that equals one if employment at the ?rm has increased
or stayed constant, while sales decreased or stayed constant
Clearly, the ?rms that increase their employment while experiencing a decline (or
no change) in sales are most likely not restructuring – these ?rms are captured by the
variable LiSni. However, even if sales of a ?rm do increase, but at the same time the
?rms increases its employment even more, this might indicate the lack of restructuring as
well. I capture this case with the variable LgrS. Finally, excess employment might also be
present in cases when a ?rm experiences no change in either sales or employment. Since
we are looking only at the old ?rms, if their level of employment relative to sales did
not change, they are likely to have the same level of excess employment that they had
prior to privatization, and most ?rms under the centrally planned system had some excess
employment. This is the reason for using variables LgeS and LndSni.
A note of caution is necessary here. Since the data on change in sales and em-
ployment comes from an answer to a survey question given by the management of the
?rm rather than from an independent sourse of data, the usual concerns of misreporting
are present and magni?ed by the fact that in the transition coutries uno?cial economy
is often substantial and thus the ?rms might misrepresent their sales and employment
considerably. In addition, the information on the percentage change in sales comes from
a question that asks: “By what percentage have sales changed in real terms over the last
three years?” Such question may be especially di?cult to answer for ?rm managers in the
transition countries because these countries have experienced periods of high and volatile
in?ation
6
. It may be easier to determine whether sales have increased or decreased in real
terms than to give a speci?c percentage of the increase (decrease). Therefore, the variables
LiSni and LndSni, which are based on the answer to the questions of whether the sales
6
In addition, many managers of ?rms in the formerly centrally planned economies may have a limited
understanding of economic concepts and therefore are likely to misunderstand what is meant by the phrase
“in real terms”.
43
(employment) have increased (decreased), are likely to be based on more accurate informa-
tion. Thus, these varaiables might be more reliable, even though they provide less precise
measurements since they measure only the dicrete changes in sales and employment.
Finally, I should point out that since the measures of excess employment are based
only on the relative changes in sales and employment they basically measure changing
labor productivity and thus should be interpreted with caution. There may be cases
when the excess employment variables indicate the presence of excess employment for
the ?rms that actually do restructute. For example, if a ?rm engages in restructuring
in the form of introducing new product line, modernizing its equipment, etc., it would
require additional workers but the results would not show up in the increased sales right
away. Our measures of excess employment may mistakingly indicate presence of excess
employment at such ?rms. We try to address such concerns below in section 3.4.2 where
we perform some robustness checks for our results.
Excess employment enters our regressions only through the characteristics of the
old ?rms in each country-sector cell. Thus, for each new ?rm in country i and sector j,
we calculate the fraction of the old ?rms with excess employment in the same country-
sector cell (out of all old ?rms in that country-sector cell). This fraction is calculated for
each measure of excess employment, and resulting variables are: LgrS F, LgeS F, LiSni F,
LndSni F. These variables enter as explanatory variables in the regressions.
Competition and Treat from New Entry. The measures of competition are
constructed for the old ?rms only, because it is the competition from new entrants or the
possibility of such competition that can induce the old ?rms to lobby for the creation
of entry barriers. I will use three measures of competition, re?ecting both the level of
competition and the identity of the major competitor (see table 3.3).
The level of competition is de?ned as high if an old ?rm faces more than three
competitors for its major product line in the domestic market. The dummy variable
comp hi = 1 for the old ?rms that face high competition. We are also interested to
know if the old ?rms face competition primarily from the new ?rms. While precisely
this question is not asked in the survey, a close approximation can be obtained with the
questions about identity of the major competitor. If the major competitor of an old ?rm
is a “domestic small and medium private enterprise” – it is most likely a new entrant.
Thus, the dummy variable compSME = 1 for those old ?rms whose major competitor is
44
a “domestic small and medium private enterprise” (SME).
For each new ?rm, we calculate the fraction of the old ?rms that face high competi-
tion (out of all old ?rms in the same country-sector cell); the resulting variable is comp hiF.
Similarly, for each new ?rm, we also calculate the fraction of the old ?rms whose major
competitor is a private domestic SME (out of all old ?rms in the same country-sector
cell); the resulting variable is compSME F. comp hiF and compSME F enter as explana-
tory variables in the regressions.
Another variable – relative number of new vs. old ?rms in a sector – can be thought
of as an alternative measure of the competition from new ?rms. A larger fraction of new
?rms in a sector means that the new ?rms are more likely to create signi?cant competition
for the old ?rms, and thus the old ?rms would have even more incentives to create entry
barriers, leading to a higher level of barriers faced by the new ?rms. This variable, New F,
is measured as a fraction of new ?rms in each country-sector cell (out of all ?rms in that
country-sector cell), and it enters as an explanatory variable in the regressions.
It should be noted that all three variables used to measure the level of competition
(comp hiF, compSME F, and New F) are potentially endogenous, i.e. the level of entry
barriers can in?uence the values of these variables. This issue is addressed in more detail
below.
3.3 Estimation Procedure
Our model argues that under certain conditions (high threat from new entry, low cost of
excess employment, etc.) the managers of the old ?rms will lobby with politicians for the
creation of entry barriers. If such lobbying takes place, we should observe both excess
employment and a high level of entry barriers. Therefore, our empirical exercise will be
structured as follows.
First, we want to check whether we indeed observe that high level of entry bar-
riers faced by the new ?rms corresponds to presence of excess employment (i.e. non-
restructuring) at the old ?rms. If we observe this correspondence, it means that there is a
possibility that lobbying between the managers of old ?rms and the politician takes place.
Having established that a high level of entry barriers corresponds to the presence of excess
employment, we will then focus on one of the factors that induces the old ?rms to create
45
entry barriers – the threat from new entry. If new entry is very threatening, managers
of the old ?rms will lobby with the politician for the creation of entry barriers, and we
should observe a high level of barriers.
Since all dependent variables in our data are ordered qualitative variables, we use
ordered probit estimation procedure. We run regressions using observations for the new
?rms only. Tables 3.2 and 3.3 provide summary of the survey questions used to create all
variables.
The following measures of barriers were used as dependent variables (higher values
mean more severe burden of regulations):
• TxReg - taxes and regulations in general; ordered variable with values {1, 2, 3, 4}
• AntComp - anti-competitive practices by government and private enterprises; or-
dered variable with values {1, 2, 3, 4}
• bl reg - business registration and licensing regulations; ordered variable with values
{1, 2, 3, 4}
• lab reg - labor regulations; ordered variable with values {1, 2, 3, 4}
• env reg - environmental regulations; ordered variable with values {1, 2, 3, 4}
• fir reg - ?re, safety regulations; ordered variable with values {1, 2, 3, 4}
• hit reg - high taxes; ordered variable with values {1, 2, 3, 4}
• tadm reg - tax administration and regulations; ordered variable with values {1, 2,
3, 4}
The independent variables in the regressions are:
• EE - equals to 1 if the new ?rm is located in Eastern Europe, 0 otherwise;
• Balt - equals to 1 if the new ?rm is located in the Baltic states of the former Soviet
Union, 0 otherwise;
• SizeS - equals to 1 if ?rm is small size, 0 otherwise;
• SizeM - equals to 1 if ?rm is medium size, 0 otherwise (the omitted category is large
size);
46
• sector1, sector2, sector3, sector4 - dummy variable for each sector (the omitted
category is sector 5 (construction));
• Age - age of the new ?rm in years in 2000 (the time when the survey was taken);
• New F - fraction of new ?rms out of all ?rms in the country-sector cell;
• compSME F - fraction of old ?rms whose major competitor is domestic private
SME out of all old ?rms in the country-sector cell;
• comp hiF - fraction of old ?rms who face high competition out of all old ?rms in
the country-sector cell;
• LgrS F, LgeS F, LiSni F, LndSni F - fraction of old ?rms with excess employ-
ment (according to measures LgrS, LgeS, LiSni, LndSni described above) out of all
old ?rms in the country-sector cell
We run two sets of regressions. The ?rst one regresses the level of barriers faced by
the new ?rms on the excess employment measures (and the region, size, age, and sector
controls). These regressions would establish whether there is a positive correspondence
between the level of barriers faced by the new ?rms and the presence of excess employment
at the old ?rms. The results for these regressions are presented in tables 3.4 though 3.11.
The second set of regressions regresses the level of barriers on the measures of competition
from new entry (and the region, size, age, and sector controls). These regressions would
establish if higher competition from the new entry results in a higher level of entry barriers.
The results for these regressions are presented in tables 3.16 through 3.20. If we establish
both that higher competition from the new entry results in a higher level of barriers and
that the higher level of barriers corresponds to the presence of excess employment, then
we can take it as an indirect evidence that competition from new entry prompts the old
?rms to lobby with politicians for the creation of entry barriers in exchange for keeping
excess employment.
I will now discuss the expected signs for the explanatory variables in the regressions.
The regional dummies, EE and Balt, are intended to capture the di?erence between
regions in the general level of barriers and regulations that is due to factors unrelated to
economics (cultural traditions, historical developments, legal systems etc.) as well as
factors related to economics but not captured by the model. Literature on transition
47
economies generally concludes that Baltic states resemble countries of Eastern Europe
much more than they do other countries of the former Soviet Union, therefore, I do not
group Baltic states
7
with other countries of the former Soviet Union. Our model does not
predict a speci?c sign for these dummy variables, but anecdotal evidence would suggest
that the level of entry barriers is lower in the Eastern Europe than in the former Soviet
Union.
Likewise, the size dummies and sector dummies are intended to capture di?erences
in the general level of regulations faced by ?rms due to their size and the nature of their
business. For example, large ?rms might face more labor regulations due to the number of
their employees, or ?rms in agriculture might face more licensing regulations, since they
have to deal with the land ownership and handling of food products. Our model does not
predict a speci?c sign for these variables.
Age variable is introduced to capture the fact that some barriers are more of an issue
for young ?rms that are just starting out (such as the business registration and licenses
that are required to open a business). Therefore, burden of some barriers might diminish
with the age of the ?rm.
The competition measures New F, compSME F, and comp hiF, would have a
positive sign if higher competition from the new entrants induces the old ?rms to lobby
with politicians for the creation of entry barriers. On the other hand, the level of barriers
can also in?uence these competition measures (even if no lobbying by the old ?rms takes
place): if the entry barriers are high, we would expect fewer new ?rms to enter the sector,
leading to lower values of New F and compSME F, and lower values of comp hiF if
competition comes at least in part from new entrants. Note that higher level of barriers in
the absence of lobbying by the old ?rms would lead to lower competition, thus implying
negative coe?cients on New F, compSME F, and comp hiF. But, if the lobbying by
the old ?rms takes place, higher values of New F, compSME F, and comp hiF would
lead to higher level of entry barriers, implying positive coe?cients. Thus, because of
the endogeneity of our competition measures, we are biased against ?nding a positive
relationship between the level of barriers and competition from new entry. This means
that if we do ?nd a positive relationship despite the bias, we can interpret it as evidence
supporting the idea that higher competition from the new ?rms induces the old ?rms to
7
Estonia and Lithuania in our sample.
48
lobby for entry barriers.
If the measures of excess employment, LgrS F, LgeS F, LiSni F, and LndSni F,
have positive coe?cients, it would provide support to the idea that the old ?rms lobby
for the creation of entry barriers in exchange for excess employment. If the coe?cient is
negative or not signi?cant, then either the lobbying does not take place or the old ?rms
reward politicians by means other than excess employment.
It is possible, however, that even if the old ?rms do lobby with politicians for the
creation of entry barriers in exchange for excess employment, it would be di?cult to ?nd
this relationship in our data. If one large existing ?rm lobbies for the creation of entry
barriers in exchange for excess employment, the other existing ?rms in the same sector
would get the bene?t of lower new entry without providing excess employment. Since
we measure excess employment as a fraction of the old ?rms in the sector with excess
employment, such “free-riding” by some old ?rms would obscure the relationship between
the entry barriers and excess employment, once again biasing us against ?nding a positive
relationship between the level of barriers the new ?rms face and the presence of excess
employment at the old ?rms.
3.4 Interpretation of Results
3.4.1 Barriers faced by new ?rms vs. Excess employment at old ?rms
We ?rst run regressions of the level of barriers faced by the new ?rms on excess employment
at the old ?rms. For each measure of barriers, we run an ordered probit regression with
the regional dummies, size dummies, sector dummies, age, and the measures of excess
employment as independent variables. The results for these regressions for each measure
of barriers are presented in tables 3.4 though 3.11.
The obvious result is that excess employment measures do not have uniformly posi-
tive and statistically signi?cant coe?cients in all equations. Only the equations with bl reg
(business registration and licensing regulations) as the dependent variable have positive
and signi?cant coe?cients for all measures of excess employment (see table 3.6). Thus, it
seems that as more old ?rms in a sector have excess employment, the new ?rms in that
sector are more likely to face high burden of business registration and licensing regulations.
49
Dependent variable - TxReg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 1.1086*** 1.0644*** 1.1524*** 1.1982***
[.000] [.001] [.000] [.000]
EE -0.0994 -0.1134* -0.1271* -0.1301**
[.137] [.086] [.052] [.047]
Balt -0.4149*** -0.3801*** -0.4053*** -0.4166***
[.000] [.000] [.000] [.000]
SizeS 0.6497** 0.6646** 0.6861** 0.6953**
[.025] [.022] [.018] [.016]
SizeM 0.6172** 0.6249** 0.6362** 0.6409**
[.035] [.033] [.029] [.028]
sector1 -0.0526 0.0030 0.0161 0.0114
[.653] [.979] [.886] [.919]
sector2 -0.1990* -0.1578 -0.1367 -0.1390
[.060] [.124] [.178] [.170]
sector3 -0.2028 -0.2133 -0.1864 -0.2014
[.452] [.429] [.489] [.456]
sector4 -0.1256 -0.0743 -0.0553 -0.0558
[.392] [.605] [.701] [.695]
Age -0.0034 -0.0029 -0.0039 -0.0045
[.779] [.814] [.752] [.712]
LgrS F 0.4719**
[.037]
LgeS F 0.2307
[.150]
LiSni F 0.0390
[.922]
LndSni F -0.1508
[.419]
µ
3
0.6886*** 0.6892*** 0.6884*** 0.6887***
[.000] [.000] [.000] [.000]
µ
4
1.5957*** 1.5953*** 1.5937*** 1.5944***
[.000] [.000] [.000] [.000]
Nobs 1577 1577 1577 1577
LR(zero slopes) 33.84*** 31.50*** 29.44*** 30.08***
[.000] [.000] [.001] [.001]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.4: Barriers (TxReg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
50
Dependent variable - AntComp
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.9768*** 0.9736*** 1.0049*** 1.0304***
[.002] [.002] [.001] [.001]
EE -0.2445*** -0.2667*** -0.2784*** -0.2756***
[.000] [.000] [.000] [.000]
Balt -0.2625** -0.2339** -0.3047*** -0.2472**
[.015] [.032] [.006] [.023]
SizeS -0.0881 -0.0629 0.0741 -0.0497
[.756] [.824] [.794] [.861]
SizeM -0.1444 -0.1313 -0.1359 -0.1245
[.614] [.646] [.635] [.663]
sector1 -0.2354** -0.1723 -0.1679 -0.1596
[.037] [.115] [.121] [.141]
sector2 -0.2573** -0.2026** -0.1987** -0.1903*
[.012] [.040] [.042] [.052]
sector3 -0.0112 -0.0096 0.0227 -0.0061
[.968] [.972] [.935] [.983]
sector4 -0.1232 -0.0590 -0.0942 -0.0471
[.389] [.673] [.503] [.735]
Age -0.0063 -0.0062 -0.0050 -0.0069
[.604] [.611] [.680] [.574]
LgrS F 0.5312**
[.020]
LgeS F 0.1474
[.367]
LiSni F 0.9767**
[.018]
LndSni F 0.0065
[.973]
µ
3
0.6612*** 0.6601*** 0.6613*** 0.6601***
[.000] [.000] [.000] [.000]
µ
4
1.3349*** 1.3319*** 1.3352*** 1.3315***
[.000] [.000] [.000] [.000]
Nobs 1409 1409 1409 1409
LR(zero slopes) 36.99*** 32.37*** 37.16*** 31.55***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.5: Barriers (AntComp) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
51
Dependent variable - bl reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.3499 0.2834 0.4040 0.3204
[.282] [.392] [.213] [.330]
EE -0.2946*** -0.3195*** -0.3486*** -0.3349***
[.000] [.000] [.000] [.000]
Balt -0.2425** -0.1804* -0.3233*** -0.1837*
[.023] [.094] [.003] [.088]
SizeS -0.1019 -0.0711 -0.0827 -0.0531
[.734] [.813] [.782] [.859]
SizeM -0.2788 -0.2605 -0.2636 -0.2490
[.357] [.389] [.383] [.411]
sector1 -0.0678 0.0223 0.0354 0.0644
[.552] [.840] [.748] [.560]
sector2 -0.0849 -0.0090 0.0057 0.0308
[.412] [.929] [.954] [.758]
sector3 -0.0630 -0.0708 -0.0169 0.0000
[.821] [.798] [.952] [.999]
sector4 0.3904*** 0.4715*** 0.4210*** 0.5077***
[.006] [.001] [.003] [.000]
Age -0.0234** -0.0224* -0.0218* -0.0226*
[.054] [.066] [.074] [.063]
LgrS F 0.8847***
[.000]
LgeS F 0.3963**
[.013]
LiSni F 1.7134***
[.000]
LndSni F 0.4157**
[.024]
µ
3
0.5863*** 0.5848*** 0.5870*** 0.5849***
[.000] [.000] [.000] [.000]
µ
4
1.3488*** 1.3415*** 1.3485*** 1.3401***
[.000] [.000] [.000] [.000]
Nobs 1540 1540 1540 1540
LR(zero slopes) 87.30*** 77.19*** 89.78*** 76.09***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.6: Higher barriers (bl reg) faced by the new ?rms correspond to higher fraction
of the old ?rms with excess employment (ordered probit regressions).
52
Dependent variable - lab reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.0483 -0.0185 -0.0067 -0.1106
[.883] [.956] [.984] [.740]
EE 0.5312*** 0.5568*** 0.5526*** 0.5619***
[.000] [.000] [.000] [.000]
Balt 0.6085*** 0.5984*** 0.5729*** 0.6248***
[.000] [.000] [.000] [.000]
SizeS 0.1162 0.0802 0.0765 0.0688
[.703] [.793] [.802] [.821]
SizeM 0.1206 0.1017 0.1007 0.0971
[.695] [.741] [.743] [.752]
sector1 -0.1984* -0.2619** -0.2604** -0.2436**
[.079] [.016] [.016] [.025]
sector2 -0.2870*** -0.3461*** -0.3454*** -0.3358***
[.005] [.000] [.000] [.001]
sector3 -0.0718 -0.0874 -0.0773 -0.0449
[.790] [.746] [.774] [.868]
sector4 0.0189 -0.0547 -0.0693 -0.0472
[.894] [.694] [.620] [.732]
Age -0.0171 -0.0163 -0.0159 -0.0148
[.161] [.184] [.194] [.225]
LgrS F -0.4237*
[.053]
LgeS F 0.0537
[.736]
LiSni F 0.3430
[.383]
LndSni F 0.3756**
[.044]
µ
3
0.7319*** 0.7307*** 0.7307*** 0.7316***
[.000] [.000] [.000] [.000]
µ
4
1.5634*** 1.5618*** 1.5623*** 1.5654***
[.000] [.000] [.000] [.000]
Nobs 1545 1545 1545 1545
LR(zero slopes) 96.82*** 93.18*** 93.83*** 97.13***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.7: Bariers (lab reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
53
Dependent variable - env reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.1355 0.0359 0.1193 0.0141
[.688] [.917] [.723] [.967]
EE -0.0147 0.0118 0.0009 0.0067
[.832] [.864] [.988] [.922]
Balt -0.0616 -0.0483 -0.0459 -0.0400
[.584] [.671] [.689] [.724]
SizeS 0.1146 0.0725 0.1021 0.0735
[.714] [.817] [.744] [.814]
SizeM 0.2002 0.1751 0.1914 0.1764
[.525] [.578] [.543] [.575]
sector1 -0.1799 -0.2294** -0.2123* -0.2045*
[.118] [.040] [.056] [.065]
sector2 -0.2730*** -0.3204*** -0.3017*** -0.2978***
[.009] [.002] [.003] [.003]
sector3 -0.4259 -0.4572 -0.4379 -0.4046
[.153] [.126] [.142] [.175]
sector4 0.1646 0.1102 0.1464 0.1309
[.256] [.438] [.306] [.354]
Age -0.0126 -0.0113 -0.0129 -0.0107
[.325] [.380] [.314] [.403]
LgrS F -0.2630
[.246]
LgeS F 0.1818
[.275]
LiSni F -0.3750
[.366]
LndSni F 0.3186*
[.098]
µ
3
0.6569*** 0.6564*** 0.6570*** 0.6565***
[.000] [.000] [.000] [.000]
µ
4
1.3898*** 1.3904*** 1.3896*** 1.3917***
[.000] [.000] [.000] [.000]
Nobs 1463 1463 1463 1463
LR(zero slopes) 28.67*** 28.51*** 28.14*** 30.05***
[.001] [.001] [.002] [.001]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.8: Bariers (env reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
54
Dependent variable - fir reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const -0.0184 -0.1396 -0.0419 -0.1821
[.959] [.701] [.907] [.615]
EE -0.0815 -0.0583 -0.0759 -0.0668
[.225] [.382] [.251] [.313]
Balt 0.0269 0.0497 -0.0072 0.0661
[.798] [.640] [.947] [.534]
SizeS 0.3586 0.3084 0.3272 0.3105
[.289] [.360] [.332] [.356]
SizeM 0.4455 0.4140 0.4278 0.4169
[.190] [.222] [.207] [.218]
sector1 -0.1689 -0.2086* -0.1920* -0.1691
[.136] [.057] [.078] [.122]
sector2 -0.2946*** -0.3336*** -0.3157*** -0.2988***
[.004] [.001] [.002] [.003]
sector3 -0.3482 -0.3740 -0.3414 -0.2916
[.214] [.183] [.224] [.299]
sector4 -0.1126 -0.1598 -0.1633 -0.1284
[.431] [.253] [.247] [.356]
Age -0.0238* -0.0221* -0.0226* -0.0212*
[.054] [.074] [.068] [.086]
LgrS F -0.1366
[.534]
LgeS F 0.2881*
[.076]
LiSni F 0.5019
[.204]
LndSni F 0.5304***
[.005]
µ
3
0.7678*** 0.7684*** 0.7682*** 0.7697***
[.000] [.000] [.000] [.000]
µ
4
1.5081*** 1.5099*** 1.5086*** 1.5127***
[.000] [.000] [.000] [.000]
Nobs 1552 1552 1552 1552
LR(zero slopes) 25.28*** 28.03*** 26.50*** 32.82***
[.005] [.002] [.003] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.9: Barriers (fir reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
55
Dependent variable - hit reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 1.5333*** 1.6199*** 1.5407*** 1.6309***
[.000] [.000] [.000] [.000]
EE -0.2619*** -0.2756*** -0.2638*** -0.2711***
[.000] [.000] [.000] [.000]
Balt -0.4148*** -0.4348*** -0.4068*** -0.4406***
[.000] [.000] [.000] [.000]
SizeS 0.3225 0.3451 0.3300 0.3416
[.301] [.268] [.289] [.272]
SizeM 0.2965 0.3074 0.2999 0.3049
[.346] [.328] [.340] [.332]
sector1 0.0893 0.1077 0.0969 0.0802
[.484] [.385] [.432] [.517]
sector2 -0.0384 -0.0186 -0.0313 -0.0432
[.739] [.868] [.779] [.699]
sector3 -0.2739 -0.2541 -0.2746 -0.3061
[.330] [.367] [.329] [.278]
sector4 -0.2019 -0.1769 -0.1857 -0.2022
[.203] [.252] [.234] [.188]
Age -0.0081 -0.0092 -0.0084 -0.0096
[.544] [.493] [.530] [.475]
LgrS F 0.0408
[.862]
LgeS F -0.1972
[.251]
LiSni F -0.1373
[.742]
LndSni F -0.2918
[.146]
µ
3
0.4401*** 0.4399*** 0.4403*** 0.4399***
[.000] [.000] [.000] [.000]
µ
4
1.1700*** 1.1700*** 1.1701*** 1.1704***
[.000] [.000] [.000] [.000]
Nobs 1570 1570 1570 1570
LR(zero slopes) 32.99*** 34.28*** 33.07*** 35.06***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.10: Barriers (hit reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
56
Dependent variable - tadm reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.7711** 0.7452** 0.8017*** 0.8262***
[.013] [.018] [.010] [.009]
EE -0.1873*** -0.1967*** -0.2048*** -0.2063***
[.004] [.002] [.001] [.001]
Balt -0.3094*** -0.2875*** -0.2993*** -0.3059***
[.002] [.005] [.004] [.003]
SizeS 0.4177 0.4289 0.4419 0.4455
[.143] [.133] [.121] [.118]
SizeM 0.4700 0.4753* 0.4821* 0.4838*
[.102] [.098] [.094] [.092]
sector1 -0.0816 -0.0489 -0.0382 -0.0414
[.468] [.653] [.725] [.703]
sector2 -0.1766* -0.1494 -0.1372 -0.1392
[.083] [.132] [.163] [.157]
sector3 -0.4650* -0.4707* -0.4561* -0.4641*
[.075] [.072] [.081] [.076]
sector4 -0.2194 -0.1827 -0.1688 -0.1703
[.123] [.188] [.228] [.217]
Age 0.0159 0.0163 0.0156 0.0153
[.179] [.168] [.188] [.198]
LgrS F 0.3055
[.153]
LgeS F 0.1402
[.367]
LiSni F -0.0043
[.991]
LndSni F -0.0798
[.660]
µ
3
0.5568*** 0.5566*** 0.5567*** 0.5568***
[.000] [.000] [.000] [.000]
µ
4
1.3782*** 1.3777*** 1.3775*** 1.3776***
[.000] [.000] [.000] [.000]
Nobs 1574 1574 1574 1574
LR(zero slopes) 29.50*** 28.27*** 27.46*** 27.65***
[.001] [.002] [.002] [.002]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.11: Barriers (tadm reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
57
Since bl reg is the variable which is closest to measuring purely entry barriers, this result
is consistent with the old ?rms lobbying with politicians to restrict new entry through the
creation of entry barriers.
Other measures of barriers are positively related to some of the measures of excess
employment. Speci?cally, equations for TxReg have positive and signi?cant coe?cient on
LgrS F, and equations for AntComp have positive and signi?cant coe?cients on both
LgrS F and LiSni F, while coe?cients on other excess employment measures are not
statistically signi?cant. Equations for env reg have positive and signi?cant coe?cient on
LndSni F, and equations for fir reg have positive and signi?cant coe?cients on both
LgeS F and LndSni F. None of the excess employment measures has a statistically sig-
ni?cant coe?cient in the equations with hit reg and tadm reg as the dependent variables.
Therefore, there might be a relationship between excess employment at the old ?rms and
the extent to which taxes and regulations, anti-competitive practices, environmental reg-
ulations, and ?re/safety regulations create obstacles for the new ?rms. However, there
seems to be no relationship between the level of excess employment and the extent to
which tax administration or high taxes are obstacles for the new ?rms
8
.
The equations with lab reg as dependent variable di?er from other equations in
this group in several respects. First of all, there is a positive and signi?cant coe?cient
on LndSni F but negative and signi?cant coe?cient on LgrS F. Therefore, we cannot
say if excess employment corresponds to higher or lower level of labor regulation barriers.
In addition, both EE and Balt have positive and statistically signi?cant coe?cients in
regressions with lab reg as a dependent variable, while in all other regressions in this group
EE and Balt have either negative and signi?cant coe?cients, or coe?cients that are not
statistically signi?cant. It would seem that Eastern Europe and Baltic states usually have
lower level of regulations, except for labor regulations
9
.
We have established that there is a strong positive relationship between the level
8
This does not mean, however, that high taxes or the tax administration do not create obstacles for the
enterprise development in these countries. To the contrary, in the survey the majority of ?rms surveyed,
both old and new, indicated that high taxes and tax administration present either moderate or major
obstacle for the operation of their enterprise.
9
This may be due to the fact that most labor regulations are not enforced in the former Soviet Union,
and thus most ?rms simply ignore them.
58
of business licensing regulations faced by the new ?rms (measured by bl reg) and excess
employment at the old ?rms. Our results also suggest that there might be a positive
relationship between the barriers measured by TxReg, AntComp, env reg, and fir reg
and at least some measures of excess employment. There is, however, no relationship
between the barriers measured by hit reg or tadm reg and excess employment, and the
direction of relationship between the labor regulations and excess employment depends
on which measure of excess employment is used.
The coe?cients on other variables in the regressions also have intuitive signs, which
I will describe using the equation with business registration and licensing regulations as
an example (results are presented in table 3.6). Eastern Europe and Baltic states seems
to have lower level of regulatory barriers than the former Soviet Union, in line with the
anecdotal evidence. Size (number of employees) of the ?rms does not seem to in?uence
the burden of business licensing regulations, while sector does in?uence the burden of
regulations, with agricultural ?rms (sector 4) facing higher burden of business registration
and licensing regulations, which is not surprising. Also, the burden of business registration
and licensing regulations diminishes with the age of the ?rm, which should be expected.
If we look at the general burden of taxes and regulations (regression with TxReg as
the dependent variable, presented in table 3.4), we ?nd that, once again, Eastern Europe
and Baltic states have lower level of barriers, that small and medium ?rms are more
likely to ?nd taxes and regulations to be a major obstacle, and the burden of taxes and
regulations does not seems to change much with the age of the ?rm.
To summarize, according to our results, it is very likely that the old ?rms engage
in lobbying for the creation of the business registration and licensing regulations. This
is not surprising, since business registration and licensing regulations are closest to the
purely entry barriers. It is less likely but also possible that they lobby for the creation of
environmental or ?re/safety regulations, as well as the general anti-competitive practices.
It is unlikely that the old ?rms lobby for the creation of labor regulations, for high taxes,
or complicated tax administration regulations.
3.4.2 Alternative Speci?cations and Robustness Checks
We will now perform several robustness checks on our results.
59
First, our measures of excess employment may simply re?ect poor performance by
the old ?rms, even if their poor performance is not related to any lobbying activities. In
other words, it is possible that excess employment at old ?rms is in?uenced by the level
of regulatory barriers in the sector. If the level of regulations and barriers is determined
exogenously (not through lobbying by the old ?rms), then high level of regulations and
barriers would adversely a?ect all ?rms in the sector, whether new or old. As a result,
given the way our variables are constracted, we may see a positive relationship between
the level of barriers faced by the new ?rms and poor performance (excess employment) at
the old ?rms. We can check for this possibility by regressing the level of barriers faced by
the old ?rms on the excess employment at the old ?rms. If the old ?rms do not engage in
lobbying, then high level of barriers faced by the old ?rms should lead to poor performance
by these ?rms and thus should be positively related to our measures of excess employment
at the old ?rms. However, if the old ?rms do lobby for the creation of barriers designed
to restrict new entry, then such entry barriers should not be damaging to the old ?rms
themselves and we should see either negative or non-signi?cant coe?cients on the excess
employment measures.
Summary results for these regressions are presented in table 3.12, where we present
only the coe?cients on the excess employment variables. The regression speci?cation in
these equations is identical to that in tables 3.4 though 3.11, except that we use observa-
tions for the old ?rms only
10
.
Most coe?cients in table 3.12 are not statistically signi?cant and some of them are
negative and statistically signi?cant. This seems to indicate that there is no strong positive
correspondence between the level of barriers faced by the old ?rms and the presence of
excess employment at these ?rms, i.e. high level of regulatory barriers does not induce poor
performance among the old ?rms. This ?nding is consistent with the level of regulatory
barriers being determined by the old ?rms lobbying with politicians.
However, some coe?cients in table 3.12 corresponding to LiSni F measure of excess
employment are positive and signi?cant. This may be due to the fact that our de?nition
of the old ?rms might include some ?rms that are actually new, such as joint ventures. As
a consistency check, we run the same speci?cation of regressions, but restrict our sample
to the old ?rms that were established in 1991 or earlier, because these are more likely
10
Full regression results are available from the author.
60
Observations for OLD ?rms only
Variables LgrS F LgeS F LndSni F LiSni F N of Obs
TxReg 0.3742 0.3117* -0.5101** -0.7249 1546
[.183] [.097] [.033] [.207]
AntComp -0.0419 -0.2149 -0.5625** 0.8324 1378
[.885] [.266] [.026] [.177]
bl reg 0.3515 0.1167 0.0593 1.1543* 1476
[.234] [.554] [.814] [.072]
lab reg -0.0844 -0.1277 0.0191 1.3696** 1533
[.751] [.500] [.938] [.207]
env reg 0.1044 -0.0832 -0.0142 1.6975*** 1493
[.708] [.658] [.953] [.004]
fir reg -0.2357 -0.0658 0.2211 1.6456*** 1535
[.398] [.727] [.358] [.004]
hit reg 0.2041 -0.1227 -1.1024*** -0.8153 1543
[.516] [.550] [.000] [.191]
tadm reg 0.2309 0.0351 -0.6324*** -0.3481 1545
[.396] [.847] [.007] [.538]
– measures of barriers are dependent variables in the regressions
– using observations for old ?rms only
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– full regression speci?cation can be seen in table 3.4
Table 3.12: Level of barriers faced by the old ?rms is not systematically positively related
to the presence of excess employment at the old ?rms (ordered probit regressions).
to be the classic incumbent ?rms and therefore likely to have political in?uence and be
able to lobby for their preferred policies. Summary of the results from this estimation
is presented in table 3.13, where we again present only the coe?cients on the excess
employment variables. In the restricted sample, none of the coe?cients are positive and
statistically signi?cant.
Thus, our results suggest that it is unlikely that the regulatory barriers are deter-
mined exogenously. Instead, our results are consistent with the argument that old ?rms
lobby for the creation of entry barriers, which is why these regulatory barriers do not
a?ect the old ?rms adversely.
It was mentioned earlier (see page 44) that our measures of excess employment
may mistakingly indicate the presence of excess employment at the ?rms that actually do
restructure by engaging in expansion and investment activities, which require additional
workers but a?ect sales only with a long lag. It is also possible that lobbying for the
creation of entry barriers leads to the lack of restructuring by the old ?rms, as measured
61
Observations for OLD ?rms established prior to 1992 only
Variables LgrS F LgeS F LndSni F LiSni F N of Obs
TxReg 0.2317 0.4491 -0.1485 -0.1294 1546
[.630] [.138] [.708] [.895]
AntComp -0.5888 -0.7777** -0.7234* 1.4045 1378
[.243] [.014] [.085] [.175]
bl reg -0.2031 -0.1663 -0.0391 -0.1179 1476
[.722] [.613] [.926] [.921]
lab reg -0.3494 -0.0784 0.3147 1.0997 1533
[.466] [.794] [.415] [.259]
env reg -0.3661 -0.0515 -0.0029 1.0895 1493
[.456] [.866] [.994] [.269]
fir reg -0.8569* 0.0746 0.3847 0.3457 1535
[.090] [.806] [.332] [.730]
hit reg 0.9058 0.4979 -0.6403 -0.0635 1543
[.104] [.133] [.130] [.953]
tadm reg -0.2038 -0.0024 -0.5128 -1.3657 1545
[.671] [.994] [.184] [.167]
– measures of barriers are dependent variables in the regressions
– using observations for old ?rms established prior to 1992 only
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– full regression speci?cation can be seen in table 3.4
Table 3.13: Level of barriers faced by the old ?rms is not positively related to the presence
of excess employment at the old ?rms (ordered probit regressions, sample of old ?rms
restricted to ?rms established prior to 1992).
by aspects other than excess employment. Therefore, we want to check that our result
– high entry barriers for the new ?rms corresponding to the lack of restructuring by the
old ?rms – does not depend solely on our de?nition of the excess employment variables.
We do this by using an alternative measure of non-restructuring based on the change in
investment:
• InvD F is a fraction of old ?rms in the country-sector cell whose investment has
decreased in real terms over the past three years
• InvCD F is a fraction of old ?rms in the country-sector cell whose investment has
decreased or stayed constant in real terms over the past three years
We then regress the level of barriers faced by the new ?rms on the region dummies, size
dummies, sector dummies, age, and these investment measures, using observations for the
new ?rms only. Results are presented in table 3.14, where we present only the coe?cients
62
Measure lack of restructuring by change in investment
Dependent Variables InvD F InvCD F N of Obs
TxReg 0.4719** 0.2307 1577
[.037] [.150]
AntComp 0.5312** 0.1474 1409
[.020] [.367]
bl reg 0.8847*** 0.3963** 1540
[.000] [.013]
lab reg -0.4237* 0.0537 1545
[.053] [.736]
env reg -0.2630 0.1818 1463
[.246] [.275]
fir reg -0.1366 0.2881* 1552
[.534] [.076]
hit reg 0.0408 -0.1972 1570
[.862] [.251]
tadm reg 0.3055 0.1402 1574
[.153] [.367]
– measures of barriers are dependent variables in the regressions
– using observations for new ?rms established in 1989 and later
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– full regression speci?cation can be seen in table 3.4
Table 3.14: Barriers faced by the new ?rms and non-restructuring by the old ?rms, as
measured by the change in investment (ordered probit regressions).
on the investment measures.
The key result is the same as before – the level of business licensing regulations faced
by the new ?rms is positively related to the fraction of the old ?rms in the same sector
that do not restructure. There is also some evidence that taxes and regulations (TxReg)
and anti-competitive measures (AntComp) faced by the new ?rms are positively related
to non-restructuring by the old ?rms.
As a ?nal robustness check, we allow for country ?xed e?ects, instead of grouping
countries into regions. Speci?cally, we regress the level of barriers faced by the new
?rms on the country-speci?c intercept, size dummies, sector dummies, age, and the excess
employment measures. Summary of the results is presented in table 3.15, where we present
only the coe?cients on the excess employment variables.
Our key result still survives – the level of business licensing regulations faced by the
new ?rms is positively related to the fraction of the old ?rms with excess employment.
Thus, our results suggest that old ?rms do lobby for the creation of business licensing
63
Observations for NEW ?rms only, Country E?ects
Variables LgrS F LgeS F LndSni F LiSni F N of Obs
TxReg 0.1685 0.2239 0.0070 -0.0253 1577
[.576] [.382] [.981] [.963]
AntComp -0.0060 -0.0362 -0.0336 0.0285 1409
[.242] [.885] [.908] [.958]
bl reg 0.9222*** 0.6183** 0.6939** 1.0303* 1540
[.002] [.017] [.019] [.053]
lab reg -0.2243 0.1242 0.0774 -0.2055 1545
[.444] [.621] [.789] [.695]
env reg 0.0233 0.1899 0.3071 -0.2636 1463
[.939] [.475] [.312] [.643]
fir reg 0.2162 0.3268 0.4335 0.2868 1552
[.465] [.208] [.136] [.593]
hit reg 0.1701 0.2442 0.1285 0.2512 1570
[.591] [.383] [.690] [.655]
tadm reg 0.2446 0.2107 -0.0621 0.0688 1574
[.391] [.390] [.826] [.893]
– measures of barriers are dependent variables in the regressions
– using observations for new ?rms established after 1989 only
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– regressions include country-speci?c intercept
Table 3.15: Barriers faced by the new ?rms and excess employment at the old ?rms
(ordered probit regressions with country-speci?c intercept).
regulations to prevent entry of the new ?rms.
3.4.3 Barriers faced by the new ?rms vs. Threat to the old ?rms from new entry
In the next step we investigate whether the competitive threat from new entry results
in a higher level of barriers for the new ?rms, which would be consistent with old ?rms
lobbying for the creation of entry barriers. Clearly, this exercise makes sense only for those
measures of barriers that were positively related to excess employment in the ?rst group
of regressions, namely bl reg and maybe TxReg, AntComp, env reg, and fir reg. For
these measures of barriers, we run an ordered probit regression with the regional dummies,
size dummies, sector dummies, age, and the measures of competition from new entry as
independent variables. The results for these regressions are presented in tables 3.16 though
3.20. For each measure of barriers we run two speci?cations: with and without interaction
64
terms between the competition measures and the regional dummies
11
.
In all equations at least one measure of competition from new entry has a positive
and statistically signi?cant coe?cient. Higher fraction of new ?rms in a sector (higher
New F) results in higher barriers for new ?rms as measured by TxReg, bl reg, env reg,
and fir reg.
As more old ?rms in a sector face high competition (comp hiF increases), the new
?rms in those sectors are more likely to face high barriers as measured by AntComp,
bl reg, and fir reg. This e?ect is smaller in Eastern Europe, i.e. in Eastern Europe high
competition is less likely to translate into high barriers for the new ?rms.
As more old ?rms in a sector face competition primarily from private domestic SMEs
(compSME F increases), the new ?rms in those sectors are more likely to face high barri-
ers as measured by TxReg, but less likely to face high barriers as measured by AntComp
and bl reg. This relationship di?ers by region: in Eastern Europe high compSME F is
less likely to translate into high TxReg barriers but more likely to translate into high
AntComp and bl reg barriers. As mentioned earlier, negative relationship between the
level of barriers and compSME F may be due to reverse causality: lower AntComp and
bl reg barriers lead to more new ?rms entering, increasing the number of private domestic
SMEs, which leads to higher compSME F. However, even with this possibility of reverse
causality, in all equations at least one of the competition measures has a positive and
statistically signi?cant coe?cient.
The e?ect of competition from the new ?rms on each measure of barriers can be
summarized as follows. For bl reg: as the fraction of new ?rms in a sector increases and/or
as the fraction of old ?rms facing high competition increases, the new ?rms in those sectors
are more likely to face high barriers in the form of business licensing regulations, but this is
less of a problem in Eastern Europe. This ?nding is consistent with the result previously
found in the literature on transition – competition promotes restructuring in Eastern
Europe but does not do that in CIS. On the other hand, as the fraction of old ?rms
facing competition primarily from private domestic SMEs increases, the new ?rms are
less likely to face high barriers if they are in the former Soviet Union, but more likely to
face high barriers in they are in the Eastern Europe or the Baltic states, which could be
11
We run regressions with regional dummy variables instead of country ?xed e?ects in order to be able
to introduce these interaction terms.
65
Dependent variable - TxReg
Variables I II
Const 0.5763 0.4857
[.127] [.250]
EE -0.1715** 0.4381
[.017] [.108]
Balt -0.4460*** -1.0584**
[.000] [.024]
SizeS 0.6865** 0.6789**
[.018] [.019]
SizeM 0.6529** 0.6489**
[.026] [.027]
sector1 0.1626 0.0774
[.182] [.534]
sector2 -0.2007** -0.2897***
[.051] [.007]
sector3 -0.0162 -0.1945
[.953] [.486]
sector4 0.0829 -0.0520
[.580] [.739]
Age -0.0006 0.0012
[.960] [.922]
New F 0.6019*** 0.3735
[.009] [.132]
Comp hiF 0.0688 0.2494
[.702] [.467]
EE ×Comp hiF -0.4410
[.290]
Bl ×Comp hiF 0.7430
[.355]
CompSME F 0.5815*** 1.1258***
[.005] [.002]
EE ×CompSME F -0.9234**
[.038]
Bl ×CompSME F 0.3622
[.751]
µ
3
0.6939*** 0.7000***
[.000] [.000]
µ
4
1.6059*** 1.6192***
[.000] [.000]
Nobs 1577 1577
LR(zero slopes) 44.79*** 62.84***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.16: Higher barriers (TxReg) faced by the new ?rms correspond to higher compe-
tition faced by the old ?rms (ordered probit regressions).
66
Dependent variable - AntComp
Variables I II
Const 0.6427* 0.8508**
[.085] [.041]
EE -0.2618*** -0.0732
[.000] [.787]
Balt -0.2862** -1.9367***
[.012] [.000]
SizeS -0.0218 -0.0555
[.939] [.845]
SizeM -0.0927 -0.1014
[.746] [.723]
sector1 -0.0949 -0.1269
[.425] [.296]
sector2 -0.2043** -0.1329
[.040] [.205]
sector3 0.0876 0.0213
[.759] [.941]
sector4 -0.0046 -0.0719
[.975] [.638]
Age -0.0074 -0.0055
[.546] [.653]
New F 0.2010 -0.0979
[.391] [.694]
Comp hiF 0.3445* 0.5449
[.060] [.107]
EE ×Comp hiF -0.6886*
[.098]
Bl ×Comp hiF 2.3058***
[.007]
CompSME F -0.0352 -0.6673*
[.876] [.063]
EE ×CompSME F 0.9765**
[.029]
Bl ×CompSME F -0.6798
[.590]
µ
3
0.6612*** 0.6675***
[.000] [.000]
µ
4
1.3338*** 1.3459***
[.000] [.000]
Nobs 1409 1409
LR(zero slopes) 35.97*** 56.99***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.17: AntComp barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions).
67
Dependent variable - bl reg
Variables I II
Const -0.1042 -0.1402
[.787] [.742]
EE -0.3975*** -0.164
[.000] [.545]
Balt -0.3157*** -1.2835**
[.005] [.011]
SizeS -0.0074 -0.0242
[.980] [.936]
SizeM -0.2001 -0.2006
[.510] [.508]
sector1 0.1845 0.2130*
[.123] [.084]
sector2 -0.0023 0.0684
[.982] [.521]
sector3 0.1264 0.1007
[.657] [.729]
sector4 0.6381*** 0.6171***
[.000] [.000]
Age -0.0222** -0.0183
[.069] [.136]
New F 0.6624*** 0.5463**
[.004] [.026]
Comp hiF 0.1762 0.5918*
[.333] [.077]
EE ×Comp hiF -0.8721**
[.035]
Bl ×Comp hiF 0.0473
[.956]
CompSME F -0.0274 -0.8567**
[.893] [.012]
EE ×CompSME F 1.2171***
[.005]
Bl ×CompSME F 3.2439***
[.009]
µ
3
0.5861*** 0.5911***
[.000] [.000]
µ
4
1.3408*** 1.3530***
[.000] [.000]
Nobs 1540 1540
LR(zero slopes) 79.82*** 103.47***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.18: Higher barriers (bl reg) faced by the new ?rms correspond to higher compe-
tition from new entry faced by the old ?rms (ordered probit regressions).
68
Dependent variable - env reg
Variables I II
Const -0.3175 -0.2391
[.431] [.591]
EE -0.0175 -0.3391
[.814] [.226]
Balt -0.1128 0.1763
[.341] [.735]
SizeS 0.1063 0.1198
[.734] [.702]
SizeM 0.2105 0.2214
[.505] [.483]
sector1 -0.1204 -0.0892
[.322] [.473]
sector2 -0.3332*** -0.3098***
[.001] [.004]
sector3 -0.3086 -0.2317
[.311] [.452]
sector4 0.2130 0.2805*
[.153] [.071]
Age -0.0110 -0.0116
[.392] [.369]
New F 0.3822 0.4960**
[.109] [.050]
Comp hiF 0.1725 -0.0052
[.366] [.988]
EE ×Comp hiF 0.3487
[.419]
Bl ×Comp hiF -0.4946
[.572]
CompSME F 0.1746 0.0313
[.414] [.930]
EE ×CompSME F 0.2260
[.620]
Bl ×CompSME F 0.3269
[.785]
µ
3
0.6574*** 0.6581***
[.000] [.000]
µ
4
1.3932*** 1.3952***
[.000] [.000]
Nobs 1463 1463
LR(zero slopes) 31.77*** 34.55***
[.001] [.005]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.19: env reg barriers faced by the new ?rms and competition from new entry faced
by the old ?rms (ordered probit regressions).
69
Dependent variable - fir reg
Variables I II
Const -1.0319** -1.1246**
[.014] [.014]
EE -0.1415* -0.0691
[.052] [.802]
Balt -0.1399 0.0486
[.209] [.921]
SizeS 0.4033 0.3939
[.234] [.245]
SizeM 0.5176 0.5110
[.129] [.134]
sector1 0.0406 0.0606
[.734] [.619]
sector2 -0.3599*** -0.3314***
[.000] [.002]
sector3 -0.0666 -0.0506
[.816] [.862]
sector4 0.0773 0.0865
[.597] [.571]
Age -0.0209* -0.0208*
[.091] [.095]
New F 1.0533*** 1.0766***
[.000] [.000]
Comp hiF 0.4300** 0.6358*
[.020] [.064]
EE ×Comp hiF -0.2922
[.490]
Bl ×Comp hiF -0.3897
[.633]
CompSME F -0.0217 -0.2997
[.918] [.398]
EE ×CompSME F 0.4308
[.334]
Bl ×CompSME F 0.3466
[.760]
µ
3
0.7762*** 0.7765***
[.000] [.000]
µ
4
1.5262*** 1.5267***
[.000] [.000]
Nobs 1552 1552
LR(zero slopes) 48.61*** 49.69***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.20: fir reg barriers faced by the new ?rms and competition from new entry faced
by the old ?rms (ordered probit regressions).
70
due to reverse causality as explained above. Another possible explanation of this result
comes from the fact that the variable bl reg measures mainly the entry barriers, i.e. the
regulations designed to prevent the entry of new businesses. If a new ?rm has already
entered (and has already gone through the business registration and licensing) and is strong
enough to be a major competitor, creating business registration and licensing regulations
may not be helpful for the old ?rms at this point. The old ?rms may choose to lobby
for the creation of some other type of regulations that can drive out an already existing
new ?rm. Thus, an increase in compSME F may not lead to a higher level of business
registration and licensing regulations, but may lead to an increased level of other types of
regulations.
For TxReg: as the fraction of the new ?rms in a sector increases and/or as the frac-
tion of the old ?rms facing competition primarily from private domestic SMEs increases,
the new ?rms in those sectors are more likely to face high barriers in the form of taxes
and regulations, but this is less of a problem in Eastern Europe.
For AntComp: as the fraction of old ?rms facing high competition in a sector in-
creases, the new ?rms in that sector are more likely to face anti-competitive measures.
This is less of a problem in Eastern Europe, but more of a problem in Baltic states.
In addition, as the fraction of old ?rms facing competition primarily from domestic pri-
vate SMEs increases, the new ?rms in that sector are less likely to face anti-competitive
measures, except in Eastern Europe.
For env reg: as the fraction of new ?rms in a sector increases, the new ?rms in the
sector are more likely to face high barriers in the form of environmental regulations. High
competition and competition from private domestic SMEs seems to have no e?ect.
For fir reg: as the fraction of new ?rms in a sector increases and/or as the fraction
of old ?rms facing high competition increases, the new ?rms in those sectors are more
likely to face high barriers in the form of ?re/safety regulations.
3.5 Conclusions
Overall, our empirical results show that a higher threat from new entry leads to a higher
level of barriers faced by the new ?rms (as measured by TxReg, AntComp, bl reg, env reg,
and fir reg), and simultaneously, higher levels of barriers are positively related to at least
71
one measure of excess employment at the old ?rms. The e?ect is especially clear for
business licensing regulations measure, which is positively related to all four measures of
excess employment. Our ?ndings are consistent with the idea that a higher threat from
new entry creates incentives for the old ?rms to lobby with politicians for the creation
of entry barriers in exchange for excess employment, thus resulting simultaneously in the
high level of barriers faced by the new ?rms and the presence of excess employment at the
old ?rms. Our results also suggest that the e?ect of the threat from new entry on the level
of barriers may vary be region, possibly due to the institutional and cultural di?erences
that make it more di?cult for old ?rms to lobby with politicians in some countries. These
results are consistent with the empirical ?ndings previously established in the literature
on privatization and restructuring in transition.
72
Conclusions
In this paper we have argued that allowing creation and entry of the new private businesses
at the start of transition reforms had an important impact on the incentives of the existing
?rms to restructure. Privatization coinciding with opening up of new entry may create
incentives for the managers of the existing (state-owned or privatized) ?rms to lobby with
politicians for the restrictions on new entry resulting in slow restructuring of the old ?rms,
high level of administrative barriers to entry, and slow development of the new private
sector. This does not mean, however, that the new entry should not be allowed. On
the contrary, those new ?rms that enter despite the high entry barriers are more e?cient
than old ?rms and their entry increases overall e?ciency of the economy. This model
calls our attention to the fact that simply allowing the entry of new businesses will not
necessarily result in substantial new entry and higher competition for the old ?rms. In
order to ensure that new entry will create competition and promote restructuring at the
old ?rms, policymakers should pay attention to the complimentary institutional reforms
designed to prevent collusion between politicians and managers of the existing old ?rms.
Certain reforms can be undertaken to reduce the probability of managers of the old
?rms successfully lobbying with politicians. Such lobbying is less likely to succeed if the
excess labor is costly for the old ?rms (e.g. barter payments are eliminated), if the barriers
are not e?ective in preventing new entry or the tax collections are very sensitive to the
level of barriers (e.g. because of uno?cial economy), if the politician is more accountable
to the treasury for tax collections (e.g. if local budgets are ?nanced entirely through
local tax collections), or if the politician’s bene?ts from excess employment are low (e.g.
because of more transparent election process and less corruption).
The empirical evidence, based on data from the World Business Environment Survey
by World Bank, is consistent with the idea that competitive threat from the new entrants
may lead the old ?rms to lobby for the creation of entry barriers. Speci?cally, we see that
the new ?rms in sectors with a large proportion of old ?rms facing high competition or
competition primarily from the private domestic SMEs, or in sectors with a large number
of new ?rms relative to the old ?rms are more likely to encounter high administrative and
regulatory barriers. Since higher level of barriers faced by the new ?rms corresponds to the
presence of excess employment at the old ?rms, especially for the business registration and
licensing barriers, our ?ndings are consistent with the model where the old ?rms provide
73
excess employment in exchange for the desired level of barriers set by the politician.
Our results also suggest that the e?ect of the threat from new entry on the level of
barriers may vary be region, possibly due to the institutional and cultural di?erences that
make it more di?cult for old ?rms to lobby with politicians in certain countries.
74
Appendix A
Indi?erence Curves for the Case of ? < 1.
A.1 Objective Functions and General Setup
In chapter 2 we assumed that ? = 1, that is, old ?rms are entirely private. ? = 1 also
implies that the Treasury does not receive any revenue from the old ?rms (does not receive
any tax collections). This is unrealistic and thus we would like to relax this assumption
on the value of ?.
Here I will derive the indi?erence curves for the manager and the politician for the
case when ? is allowed to vary between 0 and 1 and the politician derives bene?ts from
excess employment only. Recall that in this case utility functions of the manager and the
politician are
1
:
U
m
= ?(?(R) ?wL) (A.1)
U
p
= B(L) +m{(1 ??)(?(R) ?wL) +N(R)} (A.2)
Indi?erence curves have the following slopes (see equations (2.6) and (2.8)).
The manager’s indi?erence curve:
dR
dL
=
w
?
?
(R)
> 0 (A.3)
The politician’s indi?erence curve:
dR
dL
= ?
B
?
(L) ?mw(1 ??)
m[(1 ??)?
?
(R) +N
?
(R)]
(A.4)
Note that the slope of the manager’s indi?erence curve does not depend on ?. Thus, as ?
changes between 0 and 1, manager’s preferred combination of entry barriers R and excess
employment L does not change. This also means that the manager’s threat point stays at
L
?
= 0 (see page 17 for derivation of the threat points for the manager and the politician).
Of course, even though the slope of the manager’s indi?erence curve is not a?ected by
changes in ?, the value of manager’s utility will be a?ected (e.g. as ? ?, if R and L stay
unchanged, U
m
will increase). The situation is di?erent for the politician. Changes in ?
would lead to changes in the value of U
p
(if R and L are kept constant) and to changes in
the slope of politician’s indi?erence curve.
1
these are equations (2.1) and (2.2) from chapter 2
75
Let’s see how the politician’s indi?erence curve looks like for ? ? (0, 1). Recall that
we made certain assumptions on the functional forms of functions B(·), ?(·), and N(·)
(see page 14 and ?gures 2.1, 2.2, and 2.3). Under these assumptions (B
?
(·) > 0, ?
?
(·) > 0,
and N
?
(·) < 0) we cannot put a de?nite sign on the slope of the politician’s indi?erence
curve in equation (A.4).
However, using our assumptions on functional forms we can make the following
arguments. For very low values of L, B
?
(L) is positive and very large, therefore for very
low values of L we will have B
?
(L) > mw(1??) and thus the numerator in (A.4) is positive
for low values of L. As L ?, B
?
(L) ? and eventually we will get B
?
(L) < mw(1 ? ?). We
can summarize this argument in the following proposition.
Proposition 1 For any given values of m, w, and ?, such that mw(1 ? ?) = 0, we can
?nd such value of L > 0, denote it by L
??
, that B
?
(L
??
) = mw(1 ??) and
?L < L
??
we have B
?
(L) > mw(1 ??)
?L > L
??
we have B
?
(L) < mw(1 ??)
Note that, if the values of m, w, and ? are such that mw(1 ??) = 0, then B
?
(L) >
mw(1 ??) for all values of L, that is, L
??
= ? (recall our assumption that B
?
(L) ? 0 as
L ? ?). This was the case when we assumed the value of ? = 1.
Similarly, for very low levels of R, ?
?
(R) is large and positive, while N
?
(R) is small
in absolute value and negative. Thus, for low levels of R we have |(1??)?
?
(R)| > |N
?
(R)|
and the denominator in equation (A.4) is positive (see also ?gure 2.4). As R ?, ?
?
(R) ? and
|N
?
(R)| ?, and thus eventually we will have |(1??)?
?
(R)| < |N
?
(R)| and the denominator
of (A.4) will be negative. We can summarize this argument in the following proposition.
Proposition 2 For any given value of ? ? (0, 1), we can ?nd such value of R > 0, denote
it by R
??
, that (1 ??)?
?
(R
??
) = ?N
?
(R
??
) and
?R < R
??
we have (1 ??)?
?
(R) +N
?
(R) > 0
?R > R
??
we have (1 ??)?
?
(R) +N
?
(R) < 0
76
R
L
R
**
L
**
+
?
+
?
Figure A.1: Slope of the politician’s in-
di?erence curve for case ? < 1.
R
L
R
**
L
**
IC
p
Figure A.2: Politician’s indi?erence curve
for case ? < 1.
Note that at R = R
??
the denominator of (A.4) is equal to zero and thus
dR
dL
is
unde?ned at that point. Also, notice that the condition that de?nes R
??
in proposition 2
is identical to the ?rst order condition in the politician’s optimization problem (equation
(2.4)), where politician chooses R for a given level of L. Thus, our R
??
is identical to R
?
,
which is the politician’s threat point for ? < 1.
Combining signs for the numerator and the denominator of the expression in (A.4)
(and not forgetting the overall minus sign), we can determine the slope of the politician’s
indi?erence curve depending on the values of R and L (see ?gure A.1).
Further, note that R
??
is the threat point of the politician. That is, for any level
of excess employment L chosen by the manager, the politician would choose the level of
barriers R
??
in the absence of bargaining. If bargaining does occur the politician may
choose a di?erent level of barriers. However, manager of the old ?rm will never engage
in bargaining trying to set a lower level of entry barriers (given the utility function of
the manager). Thus, the level of entry barriers will never be below R
??
. Therefore, the
relevant space for the politician’s indi?erence curve is values of R > R
??
. Hence, IC
p
should have the shape depicted on ?gure A.2 (here we show the indi?erence curve that
goes through the point corresponding to the uncooperative equilibrium of R = R
??
and
L = 0).
This shape can also be derived mathematically. Along the politician’s indi?erence
77
curve
d
2
R
dL
2
=
d
dL
_
?
B
??
(L) ?mw(1 ??)
m(1 ??)?
?
(R) +mN
?
(R)
_
= (A.5)
=
?1
{m[(1 ??)?
?
(R) +N
?
(R)]}
3
×
×
_
B
??
(L)[m(1 ??)?
?
(R) +mN
?
(R)]
2
+
+
_
B
?
(L) ?mw(1 ??)
¸
2
_
(1 ??)?
??
(R) +N
??
(R)
¸
_
The long term in curly braces is negative, given our assumptions about functions
B(·), ?(·), and N(·). Thus, sign of
d
2
R
dL
2
= sign of [(1 ? ?)?
?
(R) + N
?
(R)]. For R > R
??
we have (1 ? ?)?
?
(R) + N
?
(R) < 0 (see proposition 2) and thus, along the politician’s
indi?erence curve,
d
2
R
dL
2
< 0 for R > R
??
,
which corresponds to the shape of IC
p
depicted in the ?gure A.2. We can also show the
indi?erence curves for politician and manager together, which is done on ?gure A.3. Here,
again, I have shown the indi?erence curves going through the point corresponding to the
uncooperative equilibrium of R = R
??
and L = 0.
R
L
R
**
L
**
IC
p
IC
m
0
Figure A.3: Indi?erence curves of the manager and the politician for the case of ? < 1.
A.2 E?ect of ? ? on IC
m
and IC
p
As mentioned above, changes in the value of ? do not a?ect the manager’s indi?erence
curve, they only a?ect the manager’s level of utility, U
m
.
78
A change in ? a?ects the politician’s indi?erence curve in three ways:
• it a?ects the slope of the indi?erence curve,
dR
dL
given in equation (A.4), for any given
value of R and L
• it a?ects the value of L
??
, as de?ned in proposition 1
• it a?ects the value of R
??
, as de?ned in proposition 2
We consider these e?ects in turn.
1. Slope – as ? increases, if R and L are kept constant,
dR
dL
will also increase. Thus the
slope of IC
p
becomes steeper on the upward sloping part (L < L
??
), may change
from downward to upward sloping for some values of L around L
??
, and becomes
?atter (less negative) on the downward sloping part.
2. Value of L
??
– L
??
is de?ned by B
?
(L
??
) = mw(1 ??) (see proposition 1). Thus, as
? increases, L
??
increases as well.
3. Value of R
??
– R
??
is de?ned by (1 ? ?)?
?
(R
??
) +N
?
(R
??
) = 0 (see proposition 2).
Thus, as ? increases, R
??
will decrease.
This means that when ? increases, new IC
p
will cross the old IC
p
once and only
once. However, it is impossible to tell (without knowing the functional form for the
indi?erence curve) where they would cross: on the upward sloping part before old L
??
,
between old L
??
and new L
??
, or on the downward sloping part after new L
??
. Two of
these cases are depicted on ?gures A.4 and A.5.
A.3 E?ect of ? ? on equilibrium values on R and L
Now that we know how change in ? a?ects the indi?erence curves of the manager and the
politician, we can analyze the e?ects of an increase in ? on the equilibrium level of entry
barriers and excess employment.
Because the value of ? a?ects the politician’s threat point R
??
, an increase in ?
would lead to a lower equilibrium level of barriers even in the absence of bargaining.
That is, as the manager’s share of the ?rm’s pro?t increases (?rm moves closer to being
79
R
L
R
**
0
L
**
0
IC
0
p
0
R
**
1
?
1
>?
0
L
**
1
IC
1
p
Figure A.4: Shift in the politician’s indif-
ference curve when ? ?; ?
1
> ?
0
.
R
L
R
**
0
L
**
0
IC
0
p
0
R
**
1
?
1
>?
0
L
**
1
IC
1
p
Figure A.5: Shift in the politician’s indif-
ference curve when ? ?; ?
1
> ?
0
(di?er-
ent case).
completely private), in the absence of bargaining the politician will choose lower level of
entry barriers.
If bargaining does take place, then with an increase in ? IC
p
becomes steeper, while
slope of IC
m
does not change. However, since R
??
decreases, the point of the uncooperative
equilibrium (R = R
??
and L = 0) also shifts down, forcing IC
m
to shift down without
changing its slope. As a result, if the politician has higher bargaining power and the
equilibrium allocation is at the point on the contract curve close to IC
m
, an increase in ?
is likely to produce a (large) increase in excess employment. The e?ect on the equilibrium
level of entry barriers is ambiguous.
80
Appendix B
Characterization of the Contract Curve.
Here I will consider the contract curve for the case when the politician derives bene?ts
from excess employment only. The objective is to derive the shape of the contract curve
(its slope) and to characterize how the contract curve shifts when the indi?erence curves
of the manager and/or politician shift due to change in the model parameters.
Along the contract curve we have slope of IC
m
= slope of IC
p
. Using the expres-
sions for slopes of the indi?erence curves (see equations (2.6) and (2.8)), we get:
w
?
?
(R)
= ?
B
?
(L) ?mw(1 ??)
m[(1 ??)?
?
(R) +N
?
(R)]
(B.1)
Multiplying through, we get:
wm(1 ??)?
?
(R) +wmN
?
(R) = ??
?
(R)B
?
(L) +wm(1 ??)?
?
(R) (B.2)
Note that the terms containing ? cancel out and the resulting expression, which implicitly
de?nes R as a function of L along the contract curve, is:
wm(?N
?
(R)) = ?
?
(R)B
?
(L) (B.3)
Totally di?erentiating expression (B.3) we obtain the expression for the slope of the
contract curve:
dR
dL
¸
¸
¸
¸
cc
=
B
??
(L)?
?
(R)
wm(?N
??
(R)) ?B
?
(L)?
??
(R)
(B.4)
Given our assumptions on the shape of B(·), ?(·), and N(·) functions, we can show that
dR
dL
¸
¸
¸
¸
cc
< 0.
Thus, the contract curve is downward sloping, which should not be a surprise.
Can we ?nd
d
2
R
dL
2
¸
¸
¸
cc
? The expression for the second derivative along the contract
curve is:
dR
dL
¸
¸
¸
¸
cc
=
1
[wm(?N
??
(R)) ?B
?
(L)?
??
(R)]
2
× (B.5)
×
__
B
???
(L)?
?
(R) +B
??
(L)?
??
(R)
dR
dL
_
_
wm(?N
??
(R)) ?B
?
(L)?
??
(R)
¸
?(B.6)
? B
??
(L)?
?
(R)
_
wm(?N
???
(R))
dR
dL
?B
??
(L)?
??
(R) ?B
?
(L)?
???
(R)
dR
dL
__
(B.7)
81
R
L
CC
Figure B.1: Contract curve assuming
B
???
(L) < 0, ?
???
(R) > 0, ?N
???
(L) < 0.
R
L
CC
Figure B.2: Contract curve in the general
case.
In order to be able to evaluate the sign of
d
2
R
dL
2
¸
¸
¸
cc
we need to know the signs of
B
???
(·), ?
???
(·), and N
???
(·). I am reluctant to make any assumptions about the signs of
third derivatives, especially because I am not sure what the interpretation of the third
derivative is in this case. I can only point out that the sight of
d
2
R
dL
2
¸
¸
¸
cc
will be unambiguous
only if we assume the following:
B
???
(L) < 0
?
???
(R) > 0
?N
???
(L) < 0
Under these assumptions
d
2
R
dL
2
¸
¸
¸
cc
< 0 and the contract curve looks like the one pictured on
?gure B.1.
If we do not assume the above signs for the third derivatives, then the sign of
d
2
R
dL
2
¸
¸
¸
cc
is ambiguous and the contract curve does not have a uniform second derivative and
probably looks like the one pictured on ?gure B.2, i.e. it is downward sloping, but the
second derivative changes its value.
We now consider the shifts in the contract curve in response to changes in the model
parameters
1
.
1
When depicting shift of the contract curve on the ?gures, I show contract curve as a downward sloping
straight line for simplicity.
82
CC
CC’
R
L
Figure B.3: Shift in the contract curve when w ?.
1. w ? –
¸
¸
dR
dL
¸
¸
?, contract curve shifts down and becomes ?atter (?gure B.3)
2. ?
?
(R) ? – assuming that ?
??
(·) does not change,
¸
¸
dR
dL
¸
¸
?, contract curves shifts up and
becomes steeper (?gure B.4)
3. B
?
(L) ? – assuming that B
??
(·) does not change,
¸
¸
dR
dL
¸
¸
?, contract curves shifts up
and becomes ?atter (?gure B.5)
4. |N
?
(R)| ? – assuming that N
??
(·) does not change,
¸
¸
dR
dL
¸
¸
does not change, contract
curves shifts down and does not change its slope unless N
??
(·) changes (?gure B.6)
5. m ? –
¸
¸
dR
dL
¸
¸
?, contract curves shifts down and becomes ?atter (?gure B.7)
6. ? ? – since the expression that implicitly de?nes the contract curve (equation (B.3))
does not depend on ?, changes in ? do not a?ect either the slope or the position of
the contract curve.
83
CC
CC’
R
L
Figure B.4: Shift in the contract curve
when ?
?
(·) ? (assuming ?
??
(·) does not
change).
CC
CC’
R
L
Figure B.5: Shift in the contract curve
when B
?
(·) ? (assuming B
??
(·) does not
change).
CC
CC’
R
L
Figure B.6: Shift in the contract curve
when |N
?
(·)| ? (assuming N
??
(·) does not
change).
CC
CC’
R
L
Figure B.7: Shift in the contract curve
when m ?.
84
BIBLIOGRAPHY
[1] Barberis, Nicholas; Maxim Boycko, Andrei Shleifer, Natalia Tsukanova, (1996) ”How
Does Privatization Work? Evidence from the Russian Shops.” Journal of Political
Economy, 104(4).
[2] Blanchard, O., and Michael Kremer, (November 1997) ”Disorganization.” Quarterly
Journal of Economics, 112(4).
[3] Boycko, M., Andrei Shleifer, and Robert Vishny, (1996) ”A Theory of Privatization,”
Economic Journal, 106.
[4] Carlin, Wendy, Steven Fries, Mark Scha?er, Paul Seabright, (2001). ”Competition
and Enterprise Performance in Transition Economies: Evidence from a Cross-country
Survey”, Working paper, Department of Economics, University College, London.
[5] De Melo, Martha, Cevdet Denizer, and Stoyan Tenev (2001), “Circumstance and
Choice: the Role of Initial Conditions and Policies in Transition Economies,” World
Bank Economic Review, 15(1).
[6] Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer,
(2002). ”The Regulation of Entry,” Quarterly Journal of Economics, 117(1).
[7] Djankov, Simeon and Peter Murrell, (2002). ”Enterprise Restructuring in Transition:
A Quantitative Survey”, Journal of Economic Literature, vol. XL.
[8] Frydman, Roman, Cheryl Gray, Marek Hessel, Andrzej Rapaczynski, (1999). ”When
Does Privatization Work? The Impact of Private Ownership on Corporate Perfor-
mance in the Transition Economies.” Quarterly Journal of Economics, 114(4)
[9] Hashi, Iraj and Jan Mladek, (2000). “Fiscal and Regulatory Impediments to the Entry
of New Firms in Five Transition Economies”, Journal of East-West Business, 6(2).
[10] Hellman, Joel S., (January 1998). ”Winners Take All: The Politics of Partial Reform
in Postcommunist Transition.” World Politics, 50.
[11] Hellman, Joel S., Geraint Jones, and Daniel Kaufmann, (September 2000). “Seize the
State, Seize the Day: State Capture, Corruption and In?uence in Transition”, World
Bank Policy Research Paper, no. 2444.
85
[12] Holmstrom, Bengt, (1979). ”Moral Hazard and Observability.” Bell Journal of Eco-
nomics, 10(1).
[13] Karpinski, Jakub, (1996). “Sociologists Compare Nomenklatura Members in Contem-
porary Elites”, Transition, World Bank, May 1996.
[14] Kaufman, Daniel, (1997). ”Why is Ukraine’s Economy – and Russia’s – Not Grow-
ing?”, Transition, World Bank, April 1997.
[15] Kaufmann, Daniel, and Aleksander Kaliberda, (1997). ”Integrating the Uno?cial
Economy into the Dynamics of Post-Socialist Economies”. In B. Kaminski, ed. Eco-
nomic Transition in Russia and the New States of Eurasia. Armonk: M. E. Sharpe.
[16] Rosen, Sherwin, (1992). ”Contracts and the Market for Executives”, in Contract
Economics, edited by Lars Werin and Hans Wijkander, Oxford University Press,
Oxford, England.
[17] Sachs, Je?rey, Cli?ord Zinnes, Yair Eilat, (2000). ”The gains from Privatization in
Transition Economies: Is ”Change of Ownership” Enough?”, CAER II Discussion
Paper 63.
[18] Shleifer, Andrei, (May 1997). ”Government in Transition”, European Economic Re-
view, 41, nos. 3-5.
[19] Shleifer, Andrei and Daniel Treisman, (2000). Without a Map: Political Tactics and
Economic Reform in Russia, MIT Press, Cambridge, Massachusetts.
[20] Shleifer, Andrei, and Robert Vishny, (November 1994). ”Politicians and Firms.”
Quarterly Journal of Economics, 109(4).
[21] Szelenyi, Ivan, Daniel Treisman, and Edmund Wnuk-Lipinski, eds. (1995). Elites in
Poland, Russia, and Hungary: Change or Reproduction?, Warsaw, Polish Academy
of Sciences.
86
doc_379464092.pdf
Post-communism is a term describing the period of political and economic transformation or "transition" in former communist states located in parts of Europe and Asia, in which new governments aimed to create free market-oriented capitalist economies.
ABSTRACT
Title of dissertation: PRIVATIZATION AND NEW ENTRY IN
POST-COMMUNIST TRANSITION:
THE IMPACT ON RESTRUCTURING
Polina Vlasenko, Doctor of Philosophy, 2004
Dissertation directed by: Professor Allan Drazen
Department of Economics
This work suggests a connection between the lack of restructuring of the privatized
?rms in transition and the high level of regulatory barriers faced by the private sector. In
this work I suggest that the potential entry of the newly created private ?rms signi?cantly
a?ects the incentives of the privatized ?rms regarding restructuring. Since the entry of
new businesses is threatening to the ine?cient existing enterprises, managers of these
enterprises have an incentive to use their political power to restrict the new entry. Given
that the restructuring often involves politically unpopular measures, such as shedding of
excess labor, politicians may prefer to see no restructuring of the privatized ?rms, thus
creating a possibility for the privatized enterprises to lobby with politicians for the creation
of entry barriers.
The ?rst chapter presents the background on privatization and restructuring in
transition and reviews the existing literature on the subject. It also outlines the argument
for the connection between new entry and the behavior of the existing enterprises.
The second chapter presents an analytical model that investigates the conditions
under which managers of the existing enterprises are likely to be successful in lobbying
with politicians for the restriction of new entry. This chapter also discusses some examples
of policies that would make such lobbying less likely, thus bringing an economy closer to
the e?cient equilibrium with more restructuring and low entry barriers.
The third chapter uses data from the World Business Environment Survey to explore
whether the threat of new entry induces managers of the existing ?rms to lobby with
politicians, resulting in high entry barriers and low restructuring by the existing ?rms.
Speci?cally, I investigate how the competition created by the new entrants a?ects the
probability that high regulatory barriers are erected. The results indicate that the presence
of excess employment (i.e. lack of restructuring) at the existing ?rms coincides with the
new entrants facing high regulatory barriers. Furthermore, higher competition from the
new entrants may result in new ?rms facing higher entry barriers, lending support to
the argument that the level of regulatory barriers is in?uenced by the existing ?rms via
lobbying.
PRIVATIZATION AND NEW ENRTY IN POST-COMMUNIST
TRANSITION: THE IMPACT ON RESTRUCTURING
by
Polina Vlasenko
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial ful?llment
of the requirements for the degree of
Doctor of Philosophy
2004
Advisory Commmittee:
Professor Allan Drazen, Chair/Advisor
Professor Roger Betancourt
Professor Nuno Lim˜ao
Professor Peter Murrell
Professor Margaret M. Pearson
c Copyright by
Polina Vlasenko
2004
TABLE OF CONTENTS
List of Tables iv
List of Figures vi
1 Introduction 1
1.1 Background and Literature on Privatization and Restructuring in Transition 1
1.2 Incorporating New Entry – Description of the Theoretical Framework . . . 4
1.3 Outline of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Theoretical Model 13
2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Political Bene?ts from Excess Employment . . . . . . . . . . . . . . . . . . 14
2.2.1 General setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3 Political Bene?ts From Total Employment . . . . . . . . . . . . . . . . . . . 24
2.3.1 General Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.2 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4 Applications and Policy Relevance . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1 Stylized Facts Revisited . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.2 Policy Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3 Empirical Evidence 36
3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Data and Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3 Estimation Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4 Interpretation of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4.1 Barriers faced by new ?rms vs. Excess employment at old ?rms . . . 49
3.4.2 Alternative Speci?cations and Robustness Checks . . . . . . . . . . . 59
3.4.3 Barriers faced by the new ?rms vs. Threat to the old ?rms from
new entry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
ii
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
A Indi?erence Curves for the Case of ? < 1. 75
A.1 Objective Functions and General Setup . . . . . . . . . . . . . . . . . . . . 75
A.2 E?ect of ? ? on IC
m
and IC
p
. . . . . . . . . . . . . . . . . . . . . . . . . . 78
A.3 E?ect of ? ? on equilibrium values on R and L . . . . . . . . . . . . . . . . 79
B Characterization of the Contract Curve. 81
Bibliography 85
iii
LIST OF TABLES
3.1 Transition countries in WBES dataset: number of ?rms and fraction of old
and new ?rms by country . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2 Variables and Relevant Survey Questions . . . . . . . . . . . . . . . . . . . 40
3.3 Variables and Relevant Survey Questions (cont’d) . . . . . . . . . . . . . . . 41
3.4 Barriers (TxReg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 50
3.5 Barriers (AntComp) faced by the new ?rms and excess employment at the
old ?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . 51
3.6 Higher barriers (bl reg) faced by the new ?rms correspond to higher fraction
of the old ?rms with excess employment (ordered probit regressions). . . . . 52
3.7 Bariers (lab reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 53
3.8 Bariers (env reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 54
3.9 Barriers (fir reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 55
3.10 Barriers (hit reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . . . 56
3.11 Barriers (tadm reg) faced by the new ?rms and excess employment at the
old ?rms (ordered probit regressions). . . . . . . . . . . . . . . . . . . . . . 57
3.12 Level of barriers faced by the old ?rms is not systematically positively
related to the presence of excess employment at the old ?rms (ordered
probit regressions). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.13 Level of barriers faced by the old ?rms is not positively related to the
presence of excess employment at the old ?rms (ordered probit regressions,
sample of old ?rms restricted to ?rms established prior to 1992). . . . . . . 62
3.14 Barriers faced by the new ?rms and non-restructuring by the old ?rms, as
measured by the change in investment (ordered probit regressions). . . . . . 63
3.15 Barriers faced by the new ?rms and excess employment at the old ?rms
(ordered probit regressions with country-speci?c intercept). . . . . . . . . . 64
iv
3.16 Higher barriers (TxReg) faced by the new ?rms correspond to higher com-
petition faced by the old ?rms (ordered probit regressions). . . . . . . . . . 66
3.17 AntComp barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions). . . . . . . . . . . . . . . 67
3.18 Higher barriers (bl reg) faced by the new ?rms correspond to higher com-
petition from new entry faced by the old ?rms (ordered probit regressions). 68
3.19 env reg barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions). . . . . . . . . . . . . . . 69
3.20 fir reg barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions). . . . . . . . . . . . . . . 70
v
LIST OF FIGURES
2.1 Pro?t of an old ?rm as a function of entry barriers. . . . . . . . . . . . . . . 16
2.2 Politician’s bene?ts from excess employment. . . . . . . . . . . . . . . . . . 16
2.3 Tax collections from the new ?rms as a function of entry barriers. . . . . . . 16
2.4 Politician will choose R
?
? 0 for any given level of excess employment
¯
L;
?R
?
??
< 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Manager’s and politician’s indi?erence curves with ? = 1. . . . . . . . . . . 19
2.6 As w ?, IC
m
becomes steeper. ? = 1. . . . . . . . . . . . . . . . . . . . . . 20
2.7 As ?
?
(R) ?, IC
m
becomes ?atter. ? = 1. . . . . . . . . . . . . . . . . . . . . 20
2.8 IC
p
becomes steeper due to B
?
(L) ?. ? = 1. . . . . . . . . . . . . . . . . . . 22
2.9 IC
p
becomes ?atter due to m ?. ? = 1. . . . . . . . . . . . . . . . . . . . . 22
2.10 IC
p
becomes ?atter due to |N
?
(R)| ?. ? = 1. . . . . . . . . . . . . . . . . . 23
2.11 No bargaining is possible. ? = 1. . . . . . . . . . . . . . . . . . . . . . . . . 23
2.12 For any given level of barriers R, manager would choose employment on
ˆ
L
RF
without bargaining. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.13 Politician chooses R
?
> 0 if bargaining is not allowed. ? = 1. . . . . . . . . 27
2.14 Politician chooses R
?
= 0 if bargaining is not allowed. ? = 1. . . . . . . . . 28
2.15 Indi?erence curves for manager and politician. ? = 1. . . . . . . . . . . . . 28
2.16 IC
m
shifts due to w ? or ?
R
? . ? = 1. . . . . . . . . . . . . . . . . . . . . . 30
2.17 IC
p
shifts due to B
?
(·) ? or m ? or |N
?
(·)| ?. ? = 1. . . . . . . . . . . . . . . 30
A.1 Slope of the politician’s indi?erence curve for case ? < 1. . . . . . . . . . . 77
A.2 Politician’s indi?erence curve for case ? < 1. . . . . . . . . . . . . . . . . . 77
A.3 Indi?erence curves of the manager and the politician for the case of ? < 1. . 78
A.4 Shift in the politician’s indi?erence curve when ? ?; ?
1
> ?
0
. . . . . . . . . 80
A.5 Shift in the politician’s indi?erence curve when ? ?; ?
1
> ?
0
(di?erent case). 80
B.1 Contract curve assuming B
???
(L) < 0, ?
???
(R) > 0, ?N
???
(L) < 0. . . . . . . . 82
B.2 Contract curve in the general case. . . . . . . . . . . . . . . . . . . . . . . . 82
B.3 Shift in the contract curve when w ?. . . . . . . . . . . . . . . . . . . . . . . 83
B.4 Shift in the contract curve when ?
?
(·) ? (assuming ?
??
(·) does not change). . 84
B.5 Shift in the contract curve when B
?
(·) ? (assuming B
??
(·) does not change). 84
vi
B.6 Shift in the contract curve when |N
?
(·)| ? (assuming N
??
(·) does not change). 84
B.7 Shift in the contract curve when m ?. . . . . . . . . . . . . . . . . . . . . . 84
vii
Chapter 1
Introduction
1.1 Background and Literature on Privatization and Restructuring in Transition
Large-scale privatization of state-owned enterprises is one of the major ingridients of the
market-oriented reforms in most transition countries. State-owned enterprises are priva-
tized in order to introduce market incentives for the ?rm and its manager in hopes that
it will improve e?ciency and performance of the ?rm. However, success of privatization
programs varies widely across countries. A large literature on privatization in transition
generally ?nds that success of policies aimed at inducing enterprise restructuring varies
widely depending on reform strategy and, furthermore, similar policies have di?erent ef-
fects in di?erent regions and countries.
Across transition countries we observe that privatization has not always led to the
restructuring and improvement in performance of the privatized ?rms. A large number
of empirical studies investigated performance of privatized ?rms in transition economies
trying to identify factors that in?uence restructuring. An extensive survey of empirical
studies on the determinants of enterprise restructuring in transition economies with quan-
titative summary of the results can be found in Djankov and Murrell (2002). They identify
a number of factors a?ecting enterprise restructuring: ownership structure of the enter-
prise (state owned, privatized, concentration of ownership shares among di?erent owners),
identity of the owners (most notably, insiders vs. outsiders in privatized enterprises), ex-
tent of the competition faced by the enterprise, presence of the soft budget constraints,
etc. In addition, the impact of these factors seems to di?er by region. Usually regional
comparison is drawn between Eastern Europe and countries of the Commonwealth of In-
dependent States (denoted CIS, it includes all former republics of the Soviet Union except
for Estonia, Latvia and Lithuania).
The main empirical ?ndings of the literature on enterprise restructuring can be
summarized as follows (this summary closely follows summary ?ndings in Djankov and
Murrell (2002)). Privatization leads to more enterprise restructuring on average, but the
e?ect depends on the identity of owners. Privatization to outsiders with concentrated
ownership (as opposed to insiders, such as managers and workers) has the largest positive
e?ect on restructuring, both in Eastern Europe and in the CIS. Privatization to workers
1
did not have positive e?ect on restructuring in Eastern Europe and had a signi?cant
negative e?ect in the CIS. For example, Frydman et al. (1999) ?nds that privatization
has no bene?cial e?ect on ?rm performance for the ?rms controlled by insider owners,
but has pronounced positive e?ect on ?rms with outsider owners. They also found that
insider-owned ?rms are virtually indistinguishable in their performance from state-owned
?rms, except that they increase their employment by more (!!!) than the state-owned
?rms do. Empirical studies also ?nd that elimination (or reduction) of soft budgets has a
positive e?ect on restructuring. Increased competition enhances restructuring in Eastern
Europe, but not in the CIS. Furthermore, import competition was even damaging to the
restructuring in many CIS countries.
Several studies have attempted to consider the e?ect of privatization taking into
account other reforms necessary in transition. They usually ?nd that privatization alone
may not be enough to produce improvements in enterprise performance – other comple-
mentary reforms, such as hardening of the soft budgets, institutions for protection of
property rights, and the ability of owners to monitor and control management, have to
be in place for privatization to have a positive e?ect (see, for example, Sachs, Zinnes and
Eilat (2000)).
Finally, several ?ndings in the literature support the view that bringing in new
managers (new human capital), as opposed to simply giving ownership incentives to the
old managers, leads to restructuring and improved performance of the enterprises (Djankov
and Murrell (2002), Barberis, N. et al. (1996)).
Several theoretical models exist that predict outcomes consistent with these em-
pirical ?ndings. There is a strand of literature that argues that privatization leads to
improved enterprise performance to the extent that it provides equity incentives to the
manager. If managerial e?ort is not observable then giving the manager high ownership
of the enterprise induces him to exert higher e?ort, thus increasing enterprise e?ciency
(see Holmstrom (1979)). However, empirical literature on transition economies demon-
strates that privatization does not always lead to restructuring, which means that giving
equity incentives to the managers does not always work. Thus, in the context of transition
economies, there are theoretical models demonstrating that giving ownership incentives
to the managers may not produce restructuring. Speci?cally, Shleifer and Vishny (1994)
and Boycko, Shleifer, Vishny (1996) show that politicians may in?uence managers of the
2
privatized enterprises and thus prevent them from restructuring. Speci?cally, they argue
that politicians may have political reasons to prefer no restructuring, for example, if politi-
cian derives political bene?ts from high employment at the enterprise and restructuring
involves layo?s. In such case, the politician will try to a?ect employment decision of the
privatized enterprise to get them to employ more workers than is economically e?cient.
In exchange the politician would provide subsidies to the enterprise. Such collusion is
possible as long as marginal bene?ts from extra employment for the politician is higher
than his marginal cost of providing subsidies to the ?rm. Thus, these models provide an
explanation as to why privatization does not necessarily lead to restructuring.
A di?erent theoretical argument calls attention to what can be termed the human
capital argument for restructuring. This argument emphasizes that managers need to not
only have correct incentives, but also an appropriate human capital to be able to function
e?ectively, thus role of managerial turnover is important (see Rosen (1992)). Under this
argument privatization works because it brings new managers with the appropriate human
capital to the enterprise. Without the right human capital, giving cash ?ow incentives
to the manager would not be conductive to restructuring. This theoretical argument was
tested in Barberis, N. et al. (1996) for the case of privatization in Russia and they found
that bringing in new managers (the ones with the new human capital) is more likely to
lead to restructuring than privatizing the business to the existing manager.
One of the “stylized facts” of empirical privatization literature repeated in many
studies is that privatization to outsiders is more conductive to restructuring than privati-
zation to insiders (managers and workers). However, I am not aware of theoretical models
that speci?cally explain why this must be the case.
Another empirical ?nding mentioned above is concerned with the e?ects of competi-
tion. As was indicated above, increased competition had a positive e?ect on restructuring
in the Eastern Europe but negative in the CIS (especially for the case of he import com-
petition). Again, I am not aware of a convincing theoretical explanations of this result.
Thus, here are some questions and empirical “stylized facts” that a theoretical model
of privatization should be able to explain:
1. Why is it that privatization does not always lead to restructuring?
2. Why are the outside owners more conductive to restructuring than insiders?
3
3. Why are the new managers more likely to undertake restructuring than the old
managers, even if the old managers are given ownership incentives?
4. Why does an increased competition sometimes lead to less rather than more restruc-
turing?
5. Why do e?ects of seemingly similar reforms di?er in Eastern Europe and CIS?
Answers to some of these questions were given in the theoretical literature to various
extent. For example, Shleifer and Vishny (1994) provides a convincing explanation for
why privatization may not always work (question 1 above). Also, one possible answer to
question 3 is provided by the human capital argument. In a centrally planned economy
managers of the state enterprises were presumably selected for their ability to deal with
ministerial o?cial, negotiate favorable terms for allocation of inputs in the shortage econ-
omy, address political concerns, lobby for assistance etc. In contrast, managers of private
?rms are selected for their ability to run the ?rms e?ciently in economic terms. Thus,
privatization works to the extent that it brings such new managers to run the privatized
enterprises. This gives a reason why new managers might be more likely to restructure.
However, I feel that there is a room for additional theoretical explanations. I hope to con-
vince the reader that the theoretical framework developed here can be fruitfully applied
to answer several questions on the list above.
1.2 Incorporating New Entry – Description of the Theoretical Framework
Here I propose a theoretical framework for analyzing privatization in the transition economies
that will provide answers to some of the questions posed at the end of the previous section.
We start with two key observations:
• In many transition countries privatization of state enterprises coincided with liber-
alization reform that allowed creation and entry of new private ?rms, which was
forbidden or restricted under the centrally planned system.
• While the development of the private sector is widely believed to be essential for
successful transition reforms, often the new private businesses face many obstacles -
extensive regulations, high taxation, corruption etc – when trying to establish and
operate their business.
4
I believe that these two observations are interconnected. The entry of new private
?rms, or even simply the possibility of such entry, a?ects incentives of the existing ?rms
regarding restructuring. Since, in transition countries, the new entering ?rms are likely to
be much more e?cient than the existing ones, existing ?rms realize that they would su?er
from competition created by the new entrants. Therefore, the existing ?rms (whether
state-owned or privatized) would like to see the entry of new ?rms restricted. One way to
restrict the new entry would be for the existing ?rms to lobby with politicians demanding
creation of the entry barriers which would limit the creation of new ?rms. This would
explain why the entry barriers for new businesses are high in many transition countries.
But why would such lobbying necessarily lead to less restructuring? When the existing
?rms lobby with politicians demanding the creation of entry barriers, they have to provide
politician with incentives to ful?ll their demands. In many cases, a politician may be in-
terested in preventing ?rms from restructuring, e.g. for the same reasons as in Shleifer and
Vishny (1994) – politician derives political bene?ts from excess employment and restruc-
turing involves layo?s. Thus, managers of the existing ?rms would lobby with politicians
demanding the creation of entry barriers in exchange for keeping their employment high
(i.e. not restructuring).
Why do I think that such lobbying is likely to occur and is likely to be successful?
First, we know that in many transition countries the politicians often did not get replaced
since the pre-reform period. By ”politicians” here I mean ministry o?cials, local gov-
ernment o?cials etc. In many transition countries, certainly in the former Soviet Union,
many of these ”lower level politicians” did not change, they are the same people holding
essentially the same responsibilities even if the title of their job has changed. In addition
to large amounts of anecdotal evidence, Shleifer (1997) refers to a study that compared
political and economic elites in Poland and Russia in 1993
1
. The study found that in Rus-
sia 83% of the current political elite and 53% of the current economic elite were former
Communist Party members. For comparison, in Poland they found that only 30% of the
current political elite were former Party members. This provides evidence that in some
transition countries the politicians in power (whether in political or economic elite) have
not changed much since the pre-transition period.
1
Szelenyi et al. (1995), reviewed by Karpinski (1996)
5
Second, in many transition countries state enterprises were privatized to their man-
agers. The managers of enterprises in a centrally planned economy are selected for their
ability to negotiate with ministry o?cials, lobby for assistance, get preferred allocation
of inputs etc., and thus they have good ties with the politicians. When, during privati-
zation, these managers get ownership of the ?rm they would still retain their ties with
the politicians. Furthermore, the fact that many politicians did not get replaced since the
pre-reform period, makes it even easier for the managers of (now privatized) ?rms to lobby
with politicians for their preferred policies. Clearly, the old managers (the managers that
kept their position since pre-transition period) will ?nd it easier to lobby with the politi-
cians than the new managers. This may be an additional reason why the old managers
are less likely to restructure.
The idea that formerly state-owned ?rms have good ties with politicians is supported
by Hellman et al (2000). They look at three types of interactions between ?rms at the
state: administrative corruption (payment of bribes “to get things done”), state capture
(purchase of laws, decrees and regulations), and in?uence (ability to in?uence content and
application of laws and regulations without necessarily resorting to uno?cial payments).
Here I argue that the existing ?rms would use their in?uence to restrict new entry. This
is consistent with the ?nding in Hellman et al (2000) that “In?uential ?rms appear to
be classic incumbent ?rms inherited from the socialist system. They are large, usually
state-owned, with good access to public o?cials and a dominant position in their own
market.” (Hellman et al (2000), p.15)
Thus, under certain plausible conditions, managers of privatized ?rms would lobby
for the creation of entry barriers. The politician would agree to such demands if the ?rms
can provide him with some political bene?ts. For example, if the ?rms are able to organize
their workers to vote for a particular politician, then the politician would be interested
in keeping these ?rms happy and he also would want to have employment at these ?rms
high. Therefore, an economy can end up in a situation when existing ?rms lobby with
politicians demanding creation of entry barriers and in exchange they are o?ering political
bene?ts in the form of ine?ciently high employment, which means that these ?rms do not
restructure. As a result of such lobbying, economy ends up in an equilibrium with high
entry barriers and no restructuring.
6
In order for this story to convincingly apply to transition countries, we have to
establish the following points:
• Entry of new ?rms is threatening for the existing ?rms
• Managers of the existing ?rms prefer to lobby for the creation of entry barriers rather
than to undertake restructuring and compete with the new entrants in the market
setting
• We indeed observe high entry barriers in transition countries
I will now try to provide evidence in support of these points. The new entering ?rms
can a?ect the existing ?rms in one of three ways:
1. Product competition – new ?rms either produce or import products that are sub-
stitutes to the ones produced by the existing ?rms. It is often argued that because
of high ?xed capital requirements, new ?rms would not be able to produce products
competing with the existing producers. However, this does not mean that entry
of new ?rms would not create any competition at all – new ?rms can always im-
port products, since organizing an enterprise for importing ?nished goods has much
lower capital requirements. Given the widespread shortages and poor quality of
consumer products produced in the Soviet Union, imported products became major
competitors for the domestic producers starting in the earliest years of transition.
2. Competition for suppliers – this type of competition would be damaging to the ex-
isting ?rms if their production process requires multiple suppliers and the output
cannot be produced unless all inputs are supplied. Because of the way centrally
planned economies were structured, most large enterprises in transition countries
are likely to have this type of supply structure. As shown in Blanchard and Kremer
(1997), when existing ?rms have this type of supply structure increased opportuni-
ties in the private sector may cause breakdown in the supply chain resulting in a
dramatic decrease in output. Entry of new private ?rms presents exactly this type
of “increased private sector opportunities”.
3. Competition for skilled labor. In centrally planned economies, most certainly in the
Soviet Union, salaries paid to employees essentially did not vary with level of e?ort.
7
Consequently, the level of e?ort exerted by most employees was low. However, some
employees have better quali?cations and/or ability for their jobs, but under the
socialist system they still get the same (usually, low) wage. In contrast, at a private
?rm such employees would be more valuable and therefore would receive higher
wages. At the start of transition reforms in the Soviet Union it was widely believed
that private companies would pay much higher wages to the quali?ed employees
than the state-owned enterprises. This belief turned out to be true, and in many
state-owned enterprises (especially in the service sector, where skilled employees
are the main asset of the enterprise) employees with best skills left as soon as any
opportunities were available in the private sector.
Faced with such competition (or a possibility of such competition), manager of an
existing enterprise has two possible courses of action: make his ?rm more e?cient, so that
it will be able to successfully compete in the marketplace (that is, restructure) or try to
restrict the competition, for example, by lobbying for creation of entry barriers
2
. Which
of these two action the manager would take depends on their relative costs and bene?ts.
I argue that under certain conditions lobbying for the creation of entry barriers will be
chosen over restructuring. Some of these conditions are:
• Human capital of the manager – if the manager simply does not know how to create
an enterprise that functions e?ciently in the market economy, his chances of beating
the competition are slim, and thus he is more likely to choose to lobby for the entry
barriers.
• Lobbying skills or political ties of the manager – as argued above, managers that
remained since the planning era are likely to have very good ties with politicians
because they have been involved in negotiations with politicians throughout their
career. This makes lobbying for the creation of entry barriers relatively less costly,
possibly making it a preferred choice. This argument essentially says that managers
have “comparative advantage in politics” – it is relatively less costly for them to
lobby than to undertake restructuring.
2
I would like to note here that incentives of privatized enterprises are potentially di?erent from those
of the state-owned enterprises. To the extent that the state-owned enterprises are more likely to obtain
subsidies if needed, competition from new entrants would be more of a problem for privatized ?rms than
state-owned ?rms.
8
• Unpro?table enterprises – some of the state-owned enterprises in transition economies
are not pro?table and, furthermore, some of these enterprises cannot be made prof-
itable even if restructured (they are termed negative value added enterprises in the
transition literature). Obviously, managers of such enterprises have only one choice
for keeping their enterprise in existence – lobby for assistance and restrict competi-
tion by creating the entry barriers.
Given that conditions described above can plausibly exist, can we ?nd evidence of
high entry barriers for the new ?rms in transition economies? Djankov et al. (2002) look
at the o?cial regulation of entry of start-up ?rms in 75 countries. They measure the
number of procedures that a new ?rm has to go through in order to become operational
(business registration and post-registration procedures), the length of time it takes, and
the o?cial cost of this process. They ?nd that in most countries the cost of entry is
substantial. In transition countries cost of entry is also high. For example, they ?nd that
in Ukraine a ?rm has to go through 11 procedures, which take 21 business days and cost
20% of per capita GDP. In Georgia, a new ?rm has to go through 12 procedures, which
take 70 business days and cost 28% of per capita GDP, while in Russia a new ?rm has to
go through 16 procedures, which take 69 business days and cost 37% of per capita GDP.
For comparison, a new ?rm in United States has to complete 4 procedures, which take
seven business days and cost less that 1% of per capita GDP.
It should be emphasized that Djankov et al. (2002) consider o?cial number of
procedures, and o?cial time and cost. To the extent that actual experience di?ers from
o?cial regulations, due to delays as well as side payments and bribes that may be required
in order to complete the registration process, their estimates of time and cost of entry
may be understated. It is indeed easy to ?nd anecdotal evidence that delays, bureaucratic
barriers and side payments are widespread in some countries. Consider the following quote
from Shleifer (1997):
Many Russian entrepreneurs, particularly founders of small businesses,
complain about the di?culties of starting and operating a business in Rus-
sia. They always point to multiple permits, inspections, and registrations, all
requiring interactions with multiple o?cials many of whom need to be bribed
before the necessary documents are issued. . . . To compare with the sit-
9
uation in Poland, consider the February 1996 comment of a wealthy Polish
businessman on the di?culty of opening a shop in Poland: ”Oh, it is very,
very di?cult. There are so many shops and so much competition that it is
impossible to make money.”
Entrepreneurs in other transition economies also complain about the di?culty of starting
and operating a business because of di?cult registration procedures, excessive regulation
and taxation, and widespread corruption. Kaufmann (1997) describes a number of bar-
riers that prevent development of private businesses in Ukraine. He lists a number of
factors that are seen by private ?rms as signi?cant barriers to their development: non-
transparent and time consuming company registration, cumbersome and ambiguous tax
laws, customs regulations, and specialized licensing procedures; very high payroll taxes
(in Ukraine employer-paid payroll taxes reached up to 40% of the total wage bill at that
time); highly varying corporate pro?t taxes with tax exemptions that are signi?cant and
discriminatory. Many private ?rms in Ukraine indicate that curruption is widespread, and
practice of uno?cial payments to o?cials, such as tax or health inspectors, is commonplace
and these payments are often sizable.
Hashi and Mladek (2000) use a survey of small and medium businesses in ?ve Eastern
European countries to determine which ?scal and regulatory measures impede the entry of
new ?rms in the early stages of transition. They consider entry barriers by the following
categories: registration and licensing, real estate regulations, labor and employment laws,
?scal regulations (taxes and social security contributions), and export/import regulations.
They found that taxes and other contributions are seen by the enterprises as creating the
largest obstacles, closely followed by renting and leasing of real estate, registration and
licensing, and export-import regulations.
From the above description, we can summarize that new ?rms in transition economies
are faced with the following types of barriers:
• Complicated, time-consuming and poorly de?ned registration and licensing proce-
dures; many of these regulations are ambiguous, thus creating opportunities for
corruption.
• Complicated and ambiguous tax laws that change frequently; often high taxes and
social security contributions.
10
• Limited access to commercial real estate. Often in transition countries the real
estate is owned by local governments, which gives the politicians additional power
over local businesses and creates further opportunities for corruption. Hashi and
Mladek (2000) report that the majority of surveyed ?rms saw rent increase at short
notice and shortage of premises as two main problems with real estate.
• Export-import and foreign exchange regulations that disrupt foreign trade. For ex-
ample, Kaufmann (1997) reports that in Ukraine customs regulations are extensive
and foreign currency purchases and private foreign currency loans are under admin-
istrative control.
• Finally, corruption is reported to be widespread in many of these countries, with
extralegal payments to o?cial as an established practice.
In conclusion, descriptive evidence presented here seems to indicate that new ?rms
face substantial barriers to entry in many transition countries. I also tried to demonstrate
that the entry of new ?rms is likely to be damaging to the existing ?rms, and moreover,
managers of the existing ?rms are likely to use their political ties to restrict competition
from new ?rms by creating the high entry barriers observed in these countries.
1.3 Outline of Thesis
The rest of the thesis is organized as follows. Chapter 2 presents a theoretical model
that describes the interactions between manager of the existing ?rm and a politician.
The model is used to determine which factors a?ect whether the bargaining between the
managers and the politician takes place as well as the likely outcome of such bargaining
for the restructuring and the presence of entry barriers. The model is then used to explain
the empirical stylized facts presented in section 1.1 above and to o?er some examples of
policies that can improve the outcome in terms of restructuring and promotion of new
entry.
Chapter 3 o?ers empirical support for the model presented in chapter 2. Using data
from the World Business Environment Survey conducted by the World Bank, I show that
in transition countries high entry barriers faced by the new ?rms do coincide with the old
?rms not restructuring, and, additionally, higher competitive threat from the new entrants
11
sometimes results in higher entry barriers for the new ?rms. Such ?nding are consistent
with the bargaining equilibrium where the existing ?rms lobby with the politician for the
creation of entry barriers in the face of a threat from new entry.
Appendices A and B contain some extensions and the technical derivations related
to the model in chapter 2.
12
Chapter 2
Theoretical Model
2.1 Overview
In this section I set up a model in which the existing ?rms lobby with politicians demanding
the creation of entry barriers in exchange for keeping their employment high. Such a
model would help us understand which factors a?ect whether such lobbying takes place
and provide insights into which institutional features are important for avoiding such
ine?cient outcomes.
Here is the basic argument. The existing ?rms want to have high entry barriers
because new entry creates competition and hurts their pro?ts. The politician wants to
have employment at state and privatized ?rms as high as possible because he derives
personal political bene?ts from it. I will make the nature of these bene?ts more precise
below. I assume that existing state/privatized ?rms are politically organized and thus can
collectively bargain with the politician. This is consistent with the earlier argument I made
about managers of privatized ?rms having good ties with politicians and it is supported by
?ndings in Hellman et al. (2000) about the in?uential ?rms. The new entering ?rms, on
the other hand, are assumed to not be politically organized and thus they do not engage
in lobbying with the politician. Bargaining between managers of the existing ?rms and
the politician determines the level of employment at the existing ?rms and the level of
entry barriers that the new ?rms would face if they tried to enter.
A note on terminology. In the remainder of this document, by the existing state or
privatized ?rms I mean those enterprises that existed before the start of transition reforms
and before substantial new entry was allowed. From now on, I will call these enterprises
the old ?rms. These old ?rms may be completely or partially privatized or even completely
state-owned. The new entering ?rms will be called the new ?rms and they are private by
de?nition. These are the ?rms that were created after the start of transition reforms once
the creation of private businesses was allowed.
I will set up two versions of the model. The di?erence between them is in the
speci?cation of how the politician derives bene?ts from high employment at the old ?rms.
13
2.2 Political Bene?ts from Excess Employment
2.2.1 General setup
The model has the following players
1
: manager of an old ?rm, politician, Treasury and
the new ?rms. The manager of an old ?rm has control over employment at that ?rm,
and the politician has control over the entry barriers. The manager and the politician
are active players – they can engage in bargaining, which determines the level of entry
barriers and employment at the old ?rms. The new ?rms are passive players, they do not
engage in bargaining, they take the level of entry barriers as given and decide whether or
not to enter and operate their business. If they do enter, they pay taxes to the Treasury.
The Treasury is also a passive player, it does not attempt to in?uence the old ?rms or
the politician. It collects revenues from both the old and the new ?rms in the manner
described below.
We assume that the manager of the old ?rm receives fraction ? of its net pro?ts, and
thus he cares about maximizing net pro?ts of the ?rm. Fraction (1 ? ?) of net pro?ts of
the old ?rm goes to the Treasury. Varying ? from 0 to 1 allows the model to incorporate
partial privatization: ? close to zero means that the ?rm is almost completely state owned
and the manager’s ownership of cash ?ows is low, while ? close to unity means that the
?rm is almost completely private. (1 ??) can also be used to represent the tax on pro?ts
collected by the Treasury.
The entry of new ?rms negatively impacts the pro?ts that the old ?rm can earn.
Since higher entry barriers reduce new entry, the pro?t of the old ?rm increases with the
level of entry barriers.
Thus, the objective function of the manager of an old ?rm is assumed to be:
U
m
= ?(?(R) ?wL) (2.1)
where
?(R) – maximum pro?t that the ?rm can earn with no excess employment;
L – excess employment; L ? 0
w – wage;
R – level of entry barriers, R ? 0; lower barriers lead to more new ?rms entering and thus
1
This version of the model is based on Shleifer and Vishny (1994).
14
pro?ts of the existing ?rms su?er: ?
?
(R) > 0 and ?
??
(R) < 0. We also assume: ?(0) = 0,
lim
R?0
?
?
(R) = ?, and lim
R??
?
?
(R) = 0.
Faced with some level of entry barriers, the manager, in the absence of bargaining
with the politician, will choose some level of employment to maximize pro?ts. This em-
ployment does not show up in the objective function (2.1) because we assume that ?(R)
is already optimized over employment. L denotes only the excess employment, which
represents any workers employed is excess of the pro?t maximizing level of employment.
We assume for simplicity that the production function is Leontief and therefore the ex-
cess workers cannot be productively employed. Thus, the excess workers do not produce
anything, they are simply paid wage w. Figure 2.1 depicts a function ?(R), the shape of
which is consistent with our assumptions about its derivatives.
The politician in this model cares about his personal political bene?ts and about
the Treasury’s revenue. The Treasury’s revenue consists of (1 ? ?) fraction of the old
?rm’s pro?t plus the tax collections from the new ?rms. The politician’s preferences are
summarized by the following objective function:
U
p
= B(L) +m{(1 ??)(?(R) ?wL) +N(R)} (2.2)
where
B(L) – political bene?ts from excess employment; B
?
(L) > 0, B
??
(L) < 0; B(0) = 0,
lim
L?0
B
?
(L) = ?, and lim
L??
B
?
(L) = 0;
N(R) – represents taxes collected from the new entering ?rms; higher barriers mean that
fewer ?rms enter and thus tax collections are lower: N
?
(R) < 0, N
??
(R) < 0; N(0) =
¯
N,
where
¯
N is the maximum amount of taxes that can be collected from the new ?rms,
lim
R?0
N
?
(R) = 0, and N
?
(R = R
max
) ? ?, where R
max
is the level of barriers at which
no new ?rms are entering
2
;
m – weight that the politician puts on Treasury’s revenues.
In this section we assume that the politician derives political bene?ts B(L) only
from excess employment at the old ?rms (not from total employment). Such assumption is
consistent with a story where these excess jobs are ”patronage jobs” – the excess workers
are ”friends and family” of the politician and he derives bene?ts if they are employed.
2
Figures 2.2 and 2.3 depict functions B(L) and N(R) consistent with our assumptions on their deriva-
tives.
15
R
? (R)
Figure 2.1: Pro?t of an old ?rm as a func-
tion of entry barriers.
L
B (L)
Figure 2.2: Politician’s bene?ts from ex-
cess employment.
R
N(R)
Figure 2.3: Tax collections from the new ?rms as a function of entry barriers.
Another story consistent with this formulation of political bene?ts is the case when these
“excess workers” realize that they hold their jobs only due to the politician’s in?uence,
and thus they are more likely to support this politicians in an election.
An alternative interpretation of the objective function of the politician can be
given as follows. B(L) represents some personal bene?ts that the politician receives
from excess employment, while the term in curly braces is a form of “social welfare”.
(1 ? ?)(?
?
(R) ? wL) enters social welfare because it represents the surplus produced by
the old ?rms, and, similarly, N(R) represents the surplus produced by the new ?rms. In
this interpretation, the politician cares about his personal bene?ts and about the social
welfare, and m represents the weight he puts on social welfare.
We now determine the equilibrium values for excess employment and the entry
barriers in the absence of bargaining. This would be our baseline case and would tell us
16
the threat points of the manager and the politician.
For a given level of entry barriers,
¯
R, the manager solves:
max
L
?(?(
¯
R) ?wL) (2.3)
s.t. L ? 0
Obviously, the solution is L
?
= 0. Thus, for any level of entry barriers, in the
absence of bargaining the manager will always choose zero excess employment, which is
the economically e?cient outcome.
For a given level of excess employment,
¯
L, politician solves:
max
R
B(
¯
L) +m{(1 ??)(?(R) ?w
¯
L) +N(R)} (2.4)
s.t. R ? 0
FOC: (1 ??)?
?
(R) = ?N
?
(R).
If ? = 1, then solution is R
?
= 0.
If ? < 1, then we can always ?nd such R
?
> 0 that FOC holds
3
. Moreover,
?R
?
??
< 0 (2.5)
Thus, in the absence of bargaining the politician will choose a positive level of entry
barriers as long as ? < 1.
In order to analyze the e?ect of bargaining we will derive the indi?erence curves
of the manager and the politician in (R, L) space. Totally di?erentiating the manager’s
utility function (2.1), we get the following expressions for the slope of the manager’s
indi?erence curve:
dR
dL
=
w
?
?
(R)
> 0 (2.6)
d
2
R
dL
2
= ?
w?
??
(R)
[?
?
(R)]
3
> 0 (2.7)
Similarly, the expression for the slope of the politician’s indi?erence curve is:
dR
dL
= ?
B
?
(L) ?mw(1 ??)
m[(1 ??)?
?
(R) +N
?
(R)]
(2.8)
3
Given our assumptions on derivatives, ?N
?
(R) is everywhere upward sloping and (1 ? ?)?
?
(R) is
everywhere downward sloping. Moreover, ?N
?
(0) is close to zero, while ?
?
(0) is very large. Therefore,
there will always be a positive value R
?
> 0 that satis?es FOC above. Figure 2.4 shows how the solution
for R
?
depends on the value of ?.
17
The ?rst derivative does not have a monotonic sign and its shape changes with the value
of ?. Putting a sign on the second derivative is even harder. Therefore, here I will analyze
a simpler version of the model where we restrict ? = 1
4
.
Assuming ? = 1, the politician’s indi?erence curve has the following derivatives:
dR
dL
= ?
B
?
(L)
mN
?
(R)
> 0 (2.9)
d
2
R
dL
2
= ?
B
??
(L)mN
?
(R) + [B
?
(L)]
2
N
??
(R)
N
?
(R)
[mN
?
(R)]
2
< 0 (2.10)
Recall that with ? = 1, politician’s threat point is R
?
= 0. We can now plot
indi?erence curves of the politician and the manager, which is done in ?gure 2.5.
When bargaining is not allowed the equilibrium point is the origin, with R
?
= 0 and
L
?
= 0 – no entry barriers and no excess employment. However, when bargaining between
the manager and the politician occurs, the resulting equilibrium will be somewhere on the
contract curve between points A and A
?
in ?gure 2.5. The exact allocation will depend
on the relative bargaining strengths of the manager and the politician. The important
thing to note is that with bargaining, irrespective of where on the contract curve the
equilibrium point is, we will always have R
?
> 0 and L
?
> 0, that is, we have entry
barriers and positive excess employment. Such outcome is not economically e?cient since
it does not promote restructuring of the old ?rms and prevents entry of the new, more
e?cient, ?rms.
2.2.2 Comparative Statics
Certain parameters in the model a?ect the slope of the indi?erence curves and thus can
a?ect the equilibrium allocation. The manager’s indi?erence curve shifts if:
• w ? – excess labor becomes more costly to the ?rm ? IC
m
becomes steeper; this
case is depicted on ?gure 2.6;
• ?
?
(R) ? – entry of the new ?rms is more damaging ? IC
m
becomes ?atter; this case
is depicted on ?gure 2.7.
First we consider case of an increase in w. As w increases, excess labor becomes
more costly to the ?rms, and the manager becomes less willing to trade excess labor for the
4
Appendix A contains the derivation of politician’s indi?erence curve for the case of ? < 1.
18
R
?N’ (R)
(1??
0
)?’ (R)
(1??
1
)?’ (R)
R
*
0
R
*
1
?
1
>?
0
Figure 2.4: Politician will choose R
?
? 0
for any given level of excess employment
¯
L;
?R
?
??
< 0
L
IC
m
R
IC
p
c
c
A’
A
Figure 2.5: Manager’s and politician’s in-
di?erence curves with ? = 1.
entry barriers. This results in a shift of the manager’s indi?erence curve up from IC
m
to
IC
?
m
and a shift of the contract curve from cc to c
?
c
?5
, all of which is depicted on ?gure 2.6.
Notice that if the initial bargaining equilibrium was located between points A
?
and A
??
on
the initial contract curve cc, then after the shift in IC
m
the new bargaining equilibrium,
which will be locates on the BB
?
portion of the new contract curve will de?nitely have a
lower level of L. The bargaining equilibrium is more likely to be on A
?
A
??
section of the
initial contract curve if the politician has greater bargaining power and is able to capture
almost the entire surplus from bargaining. At the opposite extreme, if the manager has
greater bargaining power and is able to force the politician to his reservation utility, an
increase in w would result in both lower L and lower R (movement from pt. A to pt. B on
the graph). The e?ect of an increase in w on the level of barriers for other combinations
of relative bargaining powers is harder to determine, since it would depend not only on
the relative bargaining strength of the manager and the politician but also on the size of
the shift in IC
m
.
Thus, when excess labor becomes more costly for the old ?rm, the equilibrium level
of excess employment resulting from the bargaining process is likely to fall, and it is more
likely to do so if the equilibrium is located at the extremes of the bargaining range, i.e.
either the politician has all of the bargaining power or the manager does. The e?ect on
5
Shifts of the contract curve are derived in Appendix B
19
IC
p
IC
m
R
L
c
c
c’
c’
A
A’
B
A’’
B’
IC
m
’
Figure 2.6: As w ?, IC
m
becomes
steeper. ? = 1.
IC
p
IC
m
R
L
c
c
c’
c’
A
A’
B
B’’
B’
IC
m
’
Figure 2.7: As ?
?
(R) ?, IC
m
becomes
?atter. ? = 1.
the equilibrium level of barriers is de?nite only in the case when the manager has greater
bargaining power and is able to force the politician to his reservation utility – in that case
the equilibrium level of barriers will fall in response to an increase in wages. For other
combinations of relative bargaining powers the e?ect on the level of barriers is ambiguous.
Similarly, when ?
?
(R) increases the entry of new ?rms becomes more damaging to
the old ?rms and IC
m
becomes ?atter (?gure 2.7). In this case, the entry barriers become
more valuable for the manager, and he would be willing to trade more excess employment
for them. Note that there is a part of the new contract curve (section B
?
B
??
on the graph)
that lies outside the initial bargaining range, AA
?
. The level of excess employment is
higher on B
?
B
??
than on any part of AA
?
. If the manager has most of the bargaining
power and is able to force the politician to his reservation utility, then an increase in
?
?
(R) will result in higher L and higher R (movement from pt. A to pt. B on the graph).
On the other hand, if the politician has most of the bargaining power and is able to force
the manager to his reservation utility, then an increase in ?
?
(R) will result in higher L, but
the e?ect on R is ambiguous (movement from pt. A
?
to pt. B
?
). Thus, as the new entry
becomes more damaging for the existing ?rms, the level of excess employment resulting
from the bargaining process is likely to increase if the distribution of bargaining powers
between the manager and the politician is very uneven, while level of barriers is likely to
increase only if the manager has most of the bargaining power.
20
The politician’s indi?erence curve shifts for the following reasons:
• B
?
(L) ? – politician’s marginal bene?ts from excess employment increase ? IC
p
becomes steeper
• |N
?
(R)| ? – with higher barriers Treasury loses tax revenues faster ? IC
p
becomes
?atter
• m ? – politician puts smaller weight on Treasury revenue ? IC
p
becomes steeper
Figure 2.8 shows the case of an increase in B
?
(L), i.e. when the politician’s marginal
bene?ts from excess employment increase thus making excess employment more valuable
to the politician. Note that there is a portion BB
??
of the new contract curve that lies
completely outside of the initial bargaining range AA
?
and the level of entry barriers on
BB
??
is higher than anywhere on AA
?
. Therefore, if the manager has most of the bargaining
power, an increase in B
?
(L) would result in higher R but the e?ect on L is ambiguous
(movement from pt. A to pt. B). On the other hand, if the politician has most of the
bargaining power and is able to force the manager to his reservation utility, an increase
in B
?
(L) would result in both higher L and higher R (movement from pt. A
?
to pt. B
?
on the graph). Thus, as politician’s marginal bene?ts from excess employment increase,
the level of barriers resulting from the bargaining process is likely to increase, while the
level of excess employment would increase only if the politician has most of the bargaining
power.
Figure 2.9 shows the case when IC
p
becomes steeper due to a decrease in m, i.e.
when the weight on the Treasury’s revenue in the politician’s utility function drops, thus
making entry barriers less costly to the politician. This case is similar to the case of
an increase in B
?
(L), the only di?erent being the slope of the new contract curve. The
e?ect on the level of barriers and excess employment is also similar. Speci?cally, as m
decreases, the level of barriers resulting for the bargaining process is likely to increase
for any relative bargaining strengths of the manager and the politician, while the level of
excess employment is more likely to increase if the politician has most of the bargaining
power.
Figure 2.10 shows the case when IC
p
becomes ?atter due to an increase in |N
?
(R)|.
In this case an increase in the entry barriers results in larger loss of revenues for the
21
IC
p
IC
p
’
IC
m
R
L
c
c
c’
c’
A
A’
B
B’’
B’
Figure 2.8: IC
p
becomes steeper due to
B
?
(L) ?. ? = 1.
IC
p
IC
p
’
IC
m
R
L
c
c
c’
c’
A
A’
B
B’’
B’
Figure 2.9: IC
p
becomes ?atter due to
m ?. ? = 1.
Treasury, thus entry barriers become more costly to the politician. We now have a portion
AA
??
of the initial contract curve that lies completely outside of the new bargaining range
BB
?
and has a higher level of R. The bargaining equilibrium is more likely to be on the
AA
??
portion of the initial contract curve if manager has most of the bargaining power.
In this case, increase in |N
?
(R)| would result in lower R, while level of L might increase
or decrease (movement from pt. A to pt. B on the graph). At the other extreme, if the
politician is able to force the manager to his reservation utility, an increase in |N
?
(R)|
would result in both lower R and lower L (movement from pt. A
?
to pt. B
?
). Thus, as tax
collections from the new ?rms become more sensitive to the level of barriers, the level of
barriers resulting for the bargaining process would likely decrease, while the level of excess
employment would decrease only if the politician has most of the bargaining power.
Finally, it is possible to have an extreme case where slopes of IC
m
and IC
p
are
such that no bargaining between the manager and the politician is possible. Such case is
depicted on ?gure 2.11. In this case, even if lobbying is allowed it will not occur and the
economy will not deviate from the e?cient equilibrium of R = 0 and L = 0. Therefore,
any policies that change slopes of IC
m
and IC
p
such as to bring the economy closer to the
case depicted in ?gure 2.11 are desirable from the economic e?ciency standpoint. Based
on the above discussion, IC
m
and IC
p
are likely to be closer to the case depicted on ?gure
2.11 if the excess labor is costly or the new entry if not damaging for the existing ?rms,
and if politician’s marginal bene?t from excess employment is low, if the politician puts a
high weight on the tax collection and if the entry barriers hurt tax collection considerably.
22
IC
p
’
IC
p
IC
m
R
L
c
c
c’
c’
A
A’ B
A’’
B’
Figure 2.10: IC
p
becomes ?atter due to
|N
?
(R)| ?. ? = 1.
L
IC
m
R
IC
p
Figure 2.11: No bargaining is possible.
? = 1.
Some examples of policies that would produce such outcomes are described in section 2.4.
2.2.3 Conclusion
To summarize, the factors described in the comparative statics section in?uence the level
of entry barriers and the level of excess employment achieved in the equilibrium through
two channels:
1. by a?ecting whether the bargaining is possible, i.e. whether we are in the situation
depicted on ?gure 2.11 or not
2. by a?ecting the equilibrium outcome given that the bargaining takes place
Taking into account both of these channels, policies that make excess employment
more costly (increase w) are likely to lead to lower excess employment – both because the
bargaining is less likely to occur and because the bargaining outcome is likely to result in
lower L. Such policies may also lead to a lower level of entry barriers by making bargaining
less likely to occur.
Similarly, a higher competitive threat from new entry (higher ?
?
(R)) is likely to
lead to higher excess employment (both because bargaining is more likely to take place
and because the bargaining outcome is likely to result in higher level of L) and possibly
a higher level of entry barriers (because of higher probability that the bargaining will
take place). This does not mean, however, that competition should be discouraged. It
23
calls to our attention the fact that high competition from the new entrants does not
automatically lead to more restructuring by the old ?rms. If bargaining between the old
?rms and the politician is possible, a higher competitive threat from new entry would give
the old ?rms stronger incentives to lobby for the creation of entry barriers. Therefore, in
order for the competition to promote restructuring some additional steps (such as legal
and administrative reform) need to be taken to eliminate the possibility of bargaining.
Our model also indicates that higher marginal political bene?ts from excess employ-
ment, such as higher bene?ts from holding o?ce, are likely to lead to a higher level of the
entry barriers (both due to higher likelihood of bargaining and a higher level of R in the
bargaining outcome) and possibly a higher level of excess employment (since bargaining
is more likely to occur). For the same reasons, a lower weight put by the politician on
Treasury’s tax collections is also likely to lead to a higher level of barriers and possibly a
higher level of excess employment.
Finally, higher responsiveness of the tax collections from new ?rms to the level of
barriers or higher e?ectiveness of barriers in preventing entry (higher |N
?
(R)|) is likely
to lead to a lower level of barriers (both because of the lower probability of bargaining
taking place and a lower R in the bargaining outcome) and possibly a lower level of excess
employment.
Some examples of policies that would a?ect w, ?
?
(R), B
?
(L), m, and |N
?
(R)| are
discussed in the section 2.4.
2.3 Political Bene?ts From Total Employment
2.3.1 General Setup
In this section I modify the model developed above. We now assume that the politician
derives bene?ts from total employment at the old ?rms. This could be the case, for
example, if people are more likely to vote for the politician if they are employed, or if the
old ?rms are better at organizing their workers to vote for a particular politician
6
.
This new assumption requires some changes in the objective functions. We introduce
new notation.
ˆ
L denotes the pro?t maximizing level of employment, that is, the level of
6
There is anecdotal evidence that the old ?rms in the former Soviet Union do try to in?uence the voting
behavior of their employees, whereas newly created private ?rms never make such attempts.
24
employment that the old ?rm would choose in the absence of bargaining. L stand for
excess employment, as before. Again, as before, we assume that the excess workers do not
produce anything. In this model the politician derives bene?ts from the total employment
L
total
=
ˆ
L+L. With these modi?cations, the objective functions of the manager and the
politician can be written as follows:
U
m
= ?(?(R,
ˆ
L) ?wL) (2.11)
U
p
= B(
ˆ
L +L) +m{(1 ??)(?(R,
ˆ
L) ?wL) +N(R)} (2.12)
For this version of the model I will again consider the simpler case with ? = 1. With
? = 1 the objective functions become,
U
m
= (?(R,
ˆ
L) ?wL)
U
p
= B(
ˆ
L +L) +mN(R)
We assume:
• ?
LL
< 0; ?
R
> 0, ?
RR
< 0; ?
LR
> 0
• B
?
(·) > 0, B
??
(·) < 0
• N
?
(R) < 0, N
??
(R) < 0
Assumptions on the derivatives of N(·) and B(·) are the same as in the earlier
version of the model. Assuming ?
LL
< 0 is standard for pro?t functions and represents
diminishing marginal product of labor. Assumptions ?
R
> 0 and ?
RR
< 0 correspond
to similar assumptions about pro?t function ?(R) in the previous section. Assumption
?
LR
> 0 means that as the level of entry barriers increases, the marginal product of labor
increases at the old ?rms. In particular, this means that as the level of entry barriers
rises, the pro?t maximizing level of employment will also rise, and the manager of the
old ?rm will choose to employ more workers even without bargaining. This assumption is
important for the derivations that follow.
We now proceed as in the previous section: we derive the threat points of the
manager and the politician, derive the shape of their indi?erence curves and see how
shifts in the indi?erence curves might a?ect the bargaining outcome.
25
Threat Points.
Maintaining the assumption of ? = 1, for any given level of barriers,
¯
R, the manager
solves:
max
ˆ
L,L
(?(
¯
R,
ˆ
L) ?wL) (2.13)
s.t. L ? 0,
ˆ
L ? 0
FOC:
?
L
(
¯
R,
ˆ
L) = 0
FOC implicitly de?nes the level of employment that the manager would choose with no
bargaining as a function of the level of barriers. We represent this relationship by the
reaction function
ˆ
L
RF
=
ˆ
L(R). This reaction function represents the level of employment
that the manager would choose for any given level of entry barriers without any interaction
with the politician. Thus, this function represents the manager’s “threat curve”.
The solution for excess employment is, obviously, L = 0.
The slope of the reaction function
ˆ
L
RF
=
ˆ
L(R) is
7
:
dR
dL
= ?
?
LL
?
LR
> 0 (2.14)
For any given level of excess employment,
¯
L, and any given reaction function
ˆ
L
RF
,
the politician solves:
max
R
B(
ˆ
L
RF
+
¯
L) +mN(R) (2.15)
s.t. R ? 0
FOC for an interior solution is:
?
B
?
(·)
mN
?
(R)
= ?
?
LR
?
LL
FOC says that the politician’s indi?erence curve has to be tangent to the manager’s
reaction function. Thus, the resulting level of barriers will be R
?
? 0. Figure 2.13 shows
the case when the politician chooses a positive level of entry barriers in the absence of
bargaining, and ?gure 2.14 shows a case when the politician chooses zero barriers in the
absence of bargaining.
7
Our assumption of ?LR > 0 is crucial for deriving the slope of the reaction function. The reaction
function is shown in ?gure 2.12, where it is drawn as a straight line for simplicity.
26
L
total
=L+L
R
L
RF ^
^
Figure 2.12: For any given level of bar-
riers R, manager would choose employ-
ment on
ˆ
L
RF
without bargaining.
L
total
=L+L
R
IC
p
L
RF
E
R
*
^
^
Figure 2.13: Politician chooses R
?
> 0 if
bargaining is not allowed. ? = 1.
Indi?erence Curves.
We now derive the shape of the indi?erence curves for the manager and the politi-
cian. The slope of manager’s indi?erence curve is given by
dR
d(
ˆ
L +L)
=
w ??
L
(·, ·)
?
R
(·, ·)
(2.16)
and the slope of the politician’s indi?erence curve is given by
dR
d(
ˆ
L +L)
= ?
B
?
(
ˆ
L +L)
mN
?
(R)
> 0 (2.17)
The expression for the slope of the politician’s indi?erence curve is almost identical to
that in the previous section (see equation (2.9)), therefore, we can easily draw it. In
order to draw the indi?erence curve for the manager, note that at the intersection with
the reaction function
ˆ
L
RF
, ?
L
(·, ·) = 0 since the reaction function represents those values
of
ˆ
L that maximize pro?t. If ?
L
(·, ·) = 0, then the slope of the manager’s indi?erence
curve is positive, according to equation (2.16). So, at the intersection with the reaction
function
ˆ
L
RF
, IC
m
is upward sloping. Also notice that for all levels of employment higher
than
ˆ
L
RF
we have ?
L
(·, ·) < 0 and thus IC
m
is de?nitely upward sloping. IC
m
can be
downward sloping only for very low levels of L
total
, which we do not depict on the graphs.
Figure 2.15 shows managers reaction function
ˆ
L
RF
and the indi?erence curves for
the manager and the politician. If bargaining is not allowed the equilibrium point is
27
L
total
=L+L
R
IC
p
L
RF
^
^
Figure 2.14: Politician chooses R
?
= 0 if
bargaining is not allowed. ? = 1.
L
total
=L+L
IC
m
R
IC
p
L
RF
E
R
E
L
E
total
A
A’
c
c
^
^
Figure 2.15: Indi?erence curves for man-
ager and politician. ? = 1.
point E. However, if bargaining is allowed, the equilibrium will be somewhere on the
AA
?
portion of the contract curve, and the levels of entry barriers and employment will
be higher than in the no-bargaining case. On the graph, the level of excess employment
is the di?erence between the value of L
total
on the contract curve and the value of
ˆ
L
RF
corresponding to the same level of R. Of course, excess employment is positive for all
points on the contact curve AA
?
, which means that bargaining between the manager and
the politician results in some excess employment, as well as a higher level of barriers.
2.3.2 Comparative Statics
We now explore the e?ects of shifts in the indi?erence curves. The manager’s indi?erence
curve becomes steeper if either w increases (excess labor becomes costlier) or ?
R
decreases
(new entry becomes less damaging). Such shift in the indi?erence curve is shown in ?gure
2.16
8
. Similar to the earlier version of the model, after the shift in IC
m
the level of excess
employment (and the level of total employment) is likely to fall if the distribution of
bargaining strengths between the manager and the politician is very uneven. In addition,
if the manager has higher bargaining strength and is able to force the politician to his
reservation utility, the equilibrium level of entry barriers is also likely to fall.
The politician’s indi?erence curve becomes steeper if B
?
(·) increases (marginal po-
litical bene?ts from employment increase), m decreases (politician puts lower weight on
Treasury’s revenue), or |N
?
(·)| decreases (tax collections from the new ?rms are less sen-
8
The shift in the contract curve is not shown on the ?gure to keep it visually simple.
28
sitive to the level of barriers). The e?ect of such shift in the indi?erence curve is shown
on ?gure 2.17. Original IC
p
was tangent to
ˆ
L
RF
at point E. When IC
p
becomes steeper,
it will be tangent to
ˆ
L
RF
at a di?erent point, E
?
. Even with no bargaining, the politi-
cian will now choose higher level of barriers corresponding to the allocation at E
?
, which
makes sense since the barriers are now less costly and employment is more valuable for
the politician. In addition, bargaining can still occur. In the case drawn in ?gure 2.17,
once IC
p
shifts to IC
?
p
the equilibrium level of barriers will be higher. Moreover, the level
of barriers will be higher irrespective of the relative bargaining strengths of the politician
and the manager, since the new contract curve lies entirely above the original contract
curve. The e?ect on the excess employment is ambiguous and depends on the size of the
shift in IC
p
and on the slope of the reaction function
ˆ
L
RF
.
Of course, it is possible that the shift in IC
p
is very slight, in which case the new
point of tangency to the reaction function, E
?
, would be very close to the original point E
and the new “bargaining lens” will be very close to the initial one, except that it will be
shifted up somewhat. In such case, it is still possible to get a higher level of barriers in the
new equilibrium if the bargaining outcome is located at the extremes of the contract curve
– either close to IC
p
or close to IC
m
. Therefore, as long as the distribution of bargaining
powers between the manager and the politician is very uneven, we should see an increase
in the level of barriers in this case.
Finally, in this speci?cation of the model, similar to the previous section, it is
possible to have slopes of IC
m
and IC
p
such that no bargaining between the manager and
the politician is possible. Steeper IC
m
and ?atter IC
p
move us closer to this case and
thus policies that have this e?ect on the slopes of indi?erence curves are desirable.
To summarize, conclusions from the model where the politician derives political
bene?ts from total employment are generally similar to the conclusions from the model
where the politician derives bene?ts from excess employment only (see section 2.2.3).
Lower w, higher ?
R
, higher B
?
(·), lower m, and lower |N
?
(·)| make it more likely that
the bargaining would take place, thus leading to both higher barriers and higher excess
employment. In addition, given an uneven distribution of the bargaining strengths between
the manager and the politician, lower w or higher ?
R
is likely to lead to higher excess
employment and higher total employment, while higher B
?
(·), lower m, or lower |N
?
(·)| is
likely to lead to a higher level of barriers.
29
L
total
= L+L
IC
m
R
IC
p
L
RF
E
R
E
L
E
total
A
A’
c
c
IC
m
’
^
^
Figure 2.16: IC
m
shifts due to w ? or
?
R
? . ? = 1.
L
total
=L+L
IC
m
R
IC
p
L
RF
E
R
E
L
E
total
A
A’
c
c
IC
m
’
E’
IC
p
^
^
Figure 2.17: IC
p
shifts due to B
?
(·) ? or
m ? or |N
?
(·)| ?. ? = 1.
2.4 Applications and Policy Relevance
2.4.1 Stylized Facts Revisited
In this section I will demonstrate how the theoretical framework developed above can
be used to explain the “stylized facts” found in the empirical literature on enterprise
restructuring (listed on page 3).
Outsiders vs. Insiders as Owners.
Insider owners are most often the managers of the enterprise who also were its
managers before privatization. Thus, the insider owners usually have better ties with the
politicians than outsiders do, because they were originally selected for their ability to deal
with o?cials, they have dealt with these politicians previously, etc. Outsider owners, on
the other hand, are less likely to have good ties with the politicians. Thus, insider owners
are more likely to engage in lobbying with the politician making the equilibrium with
bargaining a more likely outcome, which results in higher levels of both entry barriers and
excess employment (less restructuring).
New vs. Old Managers.
Here the reasoning is similar to the above example, except that it is even more
straightforward. An old manager, by de?nition has been the manager of the enterprise
under the old planning system. Thus he has all the characteristics (the ability to negotiate
with o?cials, to lobby for allocation of inputs, etc.) that would make it easy for him to
lobby with the politician. Moreover, an old manager is unlikely to have the human capital
30
required to function e?ciently in the market economy. This would make him more likely to
choose lobbying for the entry barriers instead of restructuring and facing the competition.
Thus, with old managers we are much more likely to end up in the equilibrium with
bargaining, while new managers may choose to improve e?ciency of the enterprise instead
and face the competition in a market setting. As shown above, the equilibrium with
bargaining implies less restructuring by the enterprise.
Increased Competition Sometimes Leads to Less Restructuring.
Increased competition can be captures by an increase in ?
?
(R) in the model. The
e?ect of ?
?
(R) ? is depicted on ?gure 2.7. If the relative bargaining powers of the manager
and the politician are very uneven and, in particular, if the politician has most of the
bargaining power and is able to force the manager to the allocation close to his reservation
utility, then ?
?
(R) ? would lead to an increase in the excess employment L, which means
less restructuring. To the extent that politicians have more bargaining power in CIS than
in Eastern Europe, increased competition will tend to produce less restructuring in CIS
but not in Eastern Europe.
In addition, an increase in ?
?
(R) may move the equilibrium from the case where
no bargaining is possible (depicted on ?gure 2.11) to the case where there is scope for
bargaining (depicted on ?gure 2.5), thus leading to higher excess employment and higher
barriers.
Finally, increased competition may shift the manager’s choice toward choosing lob-
bying for entry barriers instead of becoming competitive by restructuring. With increased
competition, restructuring in order to successfully face the competition requires more ef-
fort and better skills in the e?cient operation of the ?rm. Thus, if managers compare
their skills in the e?cient operation of the ?rm to their lobbying skills, with increased
competition some managers may decide to switch to lobbying.
CIS vs. Eastern Europe.
I will o?er some observations that may explain why similar reform policies may
have di?erent e?ects in CIS and Eastern Europe. This can perhaps be explained by
systematically di?erent values of certain institutional parameters in Eastern Europe and
in CIS:
• Eastern Europe may have higher m – politicians are more accountable in Eastern
Europe and therefore they put higher weight on social welfare. Higher m makes the
31
indi?erence curve of the politician ?atter and may eventually lead to the situation
when no bargaining is possible (depicted on ?gure 2.11).
• Some Eastern European countries allowed limited creation of small private businesses
in years prior to the transition reforms and mass privatization, thus an additional
entry of new private businesses may not be as damaging to the exiting ?rms. This
means that ?
?
(R) is lower in Eastern Europe than in the CIS and, consequently, the
equilibrium level of excess employment is lower (the e?ect is opposite from the one
depicted on ?gure 2.7).
• Eastern European countries undertook more decentralization in the last decades of
the planning system than did Soviet Union. As a result, managers of enterprises in
Eastern Europe did not have to rely on politicians for the routine decisions about
operations of the enterprise. Thus, managers’ ties with politicians are arguably not
as strong in Eastern Europe as in CIS. In addition, in Eastern European countries
a larger fraction of politicians was replaced during the democratic reforms, which
would further weaken the ties between existing ?rms and politicians. Hence, the
managers in Eastern Europe are less likely to lobby with politicians, making the
equilibrium with no bargaining more likely.
• Various institutional details may make B
?
(L) lower in Eastern Europe than in the
CIS. For example, more active media may expose collusion and deals between man-
agers and politician, threatening politician’s career; elections are better supervised,
thus making it more di?cult for managers to deliver votes in support of the politi-
cian. Also, it is possible that politician’s “private value of holding an o?ce” is
much higher in the CIS than in the Eastern Europe, for example, because holding
a political o?ce in CIS opens up extensive opportunities for corruption and private
enrichment. B
?
(L) represents politician’s bene?ts from excess employment, which
may be thought of as a product of the increased probability of being (re)elected and
the value of holding an o?ce. If the value of holding o?ce is much higher in CIS
than in Eastern Europe, B
?
(L) will also be higher in CIS. All this would make B
?
(L)
lower in Eastern Europe than in the CIS, making politician’s indi?erence curve ?at-
ter. As a result, bargaining is less likely to take place, and even with bargaining
the equilibrium level of entry barriers and of excess employment may be lower in
32
Eastern Europe.
2.4.2 Policy Relevance
The theoretical framework presented here can be used to evaluate which policies would
reduce the possibility of the existing ?rms lobbying for entry barriers and thus would
promote restructuring and growth of the new private sector. We have shown that in
certain cases lobbying between the manager and the politician is impossible (such case is
depicted on ?gure 2.11). Any policies that make the manager value high entry barriers
relatively less than low excess employment (making his indi?erence curve steeper) and/or
policies that make the politician value tax collections relatively more than the personal
bene?ts he derives from excess employment (making his indi?erence curve ?atter) would
bring us closer to this case. Some examples of these policies are:
• Elimination of barter payments and wage arrears. When excess labor is more costly
to the ?rm (high w), the manager is less willing to trade excess labor for barriers and
thus less likely to lobby with the politician. Barter payments, especially for wages,
and toleration of wage arrears make excess labor less costly to the ?rm. Therefore,
elimination of the barter payments is a desirable policy.
• Uno?cial economy. When marginal bene?ts from barriers are low for the existing
?rms (low ?
?
(R)), managers are less willing to “pay for barriers” with excess employ-
ment. ?
?
(R) is low when either the new entry is not damaging for the existing ?rms
or barriers are not e?ective in preventing the new entry. Strangely enough, uno?cial
economy may help in this case. If it is very easy (not costly) for a ?rm to operate
uno?cially, then high barriers to entry into the o?cial sector are irrelevant. With
extensive uno?cial economy, the new ?rms are threatening to the existing ones but
high entry barriers will not eliminate this threat, thus the existing ?rms would not
expend resources to create them.
There is another way in which uno?cial economy can in?uence the situation. Ac-
cording to our model, if tax collections fall sharply with a rise in the entry barriers
(high |N
?
(R)|), then politician may not be willing to erect entry barriers. Once
again, if it is relatively easy for the new ?rms to operate in uno?cial economy, then
they would escape into uno?cial economy whenever “o?cial” entry barriers rise.
33
Therefore, any increase in the entry barriers would lead to a sharp drop in tax col-
lections from the new ?rms
9
. Of course, this e?ect matters only if politician puts a
positive value on tax collections, i.e. if m > 0.
• Financing local budgets from local tax collections. When the politician cares a lot
about Treasury’s tax collections (high m), he is less willing to create entry barriers
since they hurt tax collections. An example of policy that would make a local
politician put a very large weight on the tax collection is a ?scal arrangement where
the local budget is ?nanced entirely by tax collection from local small and medium
enterprises (as opposed to allocations from a central budget). Since most of the new
?rms are small and medium enterprises, such policy would give the local politician
incentives to promote development of the new private sector.
• Transparent election process and active media. When the politician’s bene?ts from
excess employment are low (low B
?
(L)), he is less willing to trade the entry barriers,
which are costly, for the excess employment, which is not very valuable to him. More
transparent election process, more active media, and low corruption may make B
?
(L)
lower by exposing illegal dealings between managers and the politician and reducing
potential private gains from holding an o?ce. All of this reduces the probability
that the bargaining will take place, leading to lower barriers for the new ?rms and
more restructuring by the old ?rms.
2.5 Conclusions
This chapter presented a model of interaction between the politician and the manager of
an existing enterprise in the environment where new entry is allowed and is potentially
harmful to the existing enterprises. In such environment, managers of the existing en-
terprises have incentives to bargain with the politician for the creation of entry barriers.
We have shown that managers of privatized enterprises are more likely to lobby for the
creation of entry barriers if their ?rms would be substantially harmed by the new entry,
or if the cost of excess employment is low for the ?rms. At the same time, the politician
is more likely to agree to the creation of entry barriers if he derives substantial bene?ts
9
This in no way implies that steps should be taken to promote uno?cial economy. I am simply pointing
out that the existence of uno?cial economy may have positive consequences.
34
from the political support provided by the privatized ?rms, if he puts low weight on the
Treasury’s tax collections, or if the level of entry barriers does not signi?cantly a?ect tax
revenues collected from the new ?rms. When managers of the existing ?rms are successful
in their bargaining with the politician, the economy ends up in an equilibrium with a
higher level of barriers, fewer new ?rms entering, and a higher level of excess employment
(lower restructuring) at the old ?rms.
These ?ndings emphasize that allowing, or even promoting, the entry of new busi-
nesses may not be enough to stimulate development of the new private sector and to
promote competition and restructuring of the old ?rms. Attention should be paid to ad-
ditional institutional and administrative reforms that would prevent the possibility of old
?rms bargaining with politicians.
35
Chapter 3
Empirical Evidence
3.1 Overview
Having presented the theoretical argument in chapter 2, in this chapter we will turn to
providing relevant empirical evidence. The theoretical model outlined in chapter 2 argues
that, under certain conditions, managers of the old ?rms would lobby with politicians
demanding creation of high entry barriers in exchange for keeping employment at their
enterprises at levels above economically e?cient (i.e. not restructuring). Such collusion
is more likely to occur when the new entry is threatening and/or excess employment is
not costly to the old ?rms, and when the politician derives high political bene?ts from
excess employment and does not have incentives to raise the Treasury’s tax collections.
Unfortunately, some of these things are very di?cult to measure. Therefore, I will take
the following approach in this empirical exercise.
On the very basic level the model argues that the level of barriers faced by the new
?rms is in?uenced by what the old ?rms do, i.e. whether they lobby with politicians.
Since lobbying is di?cult to observe, we will look at the factors that would make such
lobbying likely (such as new entry being threatening to the old ?rms, etc) and see if they
can explain the level of barriers faced by the new ?rms. In addition, if the lobbying takes
place, we would expect to see both excess employment and a high level of entry barriers.
Without structurally modeling the channel through which the political lobbying operates,
I will try to see if the presence of excess employment at the old ?rms explains the level of
barriers faced by the new ones.
3.2 Data and Variables
I use data from the World Business Environment Survey (WBES)
1
, conducted by the
World Bank in late-1999 to mid-2000. The survey covers over 10, 000 ?rms in 80 countries,
among them nine countries of Eastern Europe and twelve countries of the former Soviet
Union. The list of countries and the number of ?rms surveyed in each country can be
found in table 3.1. The summary of survey questions used to create relevant variables is
1
c “The World Business Environment Survey” (WBES) 2000, The World Bank Group
36
given in tables 3.2 and 3.3
2
.
The objective of our analysis is to see how the level of barriers faced by the new
?rms is in?uenced by the characteristics of the old ?rms, including their level of excess
employment. However, we need to de?ne more carefully which old ?rms might be inter-
ested in preventing entry of a speci?c new ?rm, i.e. we need to determine what is the
“relevant group” of the old ?rms for each new ?rm. For example, old ?rms in Poland
are not likely to lobby for the creation of entry barriers to limit new entry in Uzbekistan.
Similarly, even within the same country, old ?rms in car manufacturing are not likely to
lobby for barriers to restrict new entry into the construction industry. Conceptually, the
old ?rms in the same industry and the same country would have incentives to act together
to prevent entry of the new ?rms in this industry
3
. WBES survey has some limited infor-
mation about industry classi?cation of ?rms: it classi?es all ?rms into ?ve broad sectors –
manufacturing, services, agriculture, construction, and other. We will use this information
to create variables characterizing “relevant old ?rms”.
The data is organized as follows. Each new ?rm is a unit of observation (which
?rms are considered “new” is precisely de?ned below). For each new ?rm we observe its
characteristics, such as the level of barriers it faces, its age, size, the country and the
sector it belongs to. We also observe characteristics of the old ?rms in the same country
and sector (country-sector cell) as this new ?rm. The idea is that certain characteristics
of the old ?rms in the country-sector cell in?uence the level of barriers faced by the new
?rms in the same country-sector cell.
2
Hellman et al. (2000) describe this data set for transition countries in more detail. They also indicate
that “The sample was structured to be fairly representative of the domestic economies, with speci?c quotas
placed on size, sector, location, and export orientation.” The sample was heavily weighted toward privately
owned ?rms. (Hellman et al. (2000) p. 5)
3
This is not strictly true. Domestic manufacturers might be interested in preventing entry of the
?rms that would import competing products, and these importing ?rms would fall under retail sector,
thus making the old ?rms in manufacturing trying to prevent new entry into the retail services sector.
Therefore, the analysis here would miss this important channel through which new entry in?uences the old
?rms. However, were we able to include this channel, we should ?nd an even stronger relationship between
the characteristics of the old ?rms and the level of barriers faced by the new ?rms. Unfortunately, in this
dataset it is impossible to determine which ?rms are importing competing products, which would bias our
results against ?nding a relationship between the behavior of the old ?rms and the entry barriers faced by
the new ones.
37
I will now describe in more detail how the variables are constructed.
New vs. Old Firms. The new ?rms are those ?rms that were established as
“originally private from the time of start up” (see table 3.3). All other ?rms are treated
as the old ?rms
4
. The fraction of old and new ?rms by country is presented in table 3.1.
A consistency check was performed using the age of ?rms. Those ?rms that claim to be
“originally private from the time of start up” and were established prior to 1989 were
removed from the sample because they would not constitute “the new ?rms” in the sense
of the model described here, and, furthermore, it was impossible for any private ?rms to
exist in Soviet Union prior to 1989.
5
.
Barriers. The measures of barriers are constructed only for the new ?rms. There
are several alternative measures of barriers that can be constructed given the questions in
WBES survey (see table 3.2). These questions ask how problematic are various regulations
and anti-competitive practices for the operation of a ?rm, ranging from “no obstacle” to
“major obstacle”. Since it is di?cult to say a priori that one measure is better than
others, and it is also di?cult to construct a meaningful aggregation of these measures, I
will consider each variable as a separate measure of the level of barriers, with one important
exception.
Two types of regulatory barriers considered in the survey, namely customs regula-
tions and foreign currency/exchange regulations (variables cus reg and frk reg), are only
relevant if a ?rm is engaged in the international trade. Since not all ?rms are engaged in
international trade, these variables may have systematic measurement errors. Speci?cally,
some ?rms may indicate that these types of regulations are no obstacle to the operation
of their business not because there are no barriers in these areas, but simply because the
?rms are not engaged in the international trade and therefore these barriers are irrelevant.
4
This de?nition of the new ?rms might still include some ?rms that should be classi?ed at “old”. It has
been a common practice in transition countries for managers of large state-owned enterprises to register
new ?rms on the basis of their existing enterprises as means of diverting assets. Such falsely new ?rms
are likely to have extremely good ties with politicians and would behave like the old ?rms in our model,
making it more di?cult for us to ?nd the expected relationship between the characteristics of the old ?rm
and level of barriers faces by the new ones. Unfortunately, it is impossible to identify such ?rms in the
data, but we should keep this in mind when interpreting the results.
5
This procedure removed 60 observations from the total of 1673 observations for new ?rms.
38
Country Number of Firms Old Firms (fraction) New Firms (fraction)
CIS
Armenia 125 49.6% 50.4%
Azerbaijan 128 35.2% 64.8%
Belarus 125 65.6% 34.4%
Georgia 129 44.2% 55.8%
Kazakhstan 127 61.4% 38.6%
Kyrgyzstan 125 68.8% 31.2%
Moldova 125 70.4% 29.6%
Russia 525 51.0% 49.0%
Ukraine 225 45.3% 54.7%
Uzbekistan 125 70.4% 29.6%
Baltics
Estonia 132 47.0% 53.0%
Lithuania 112 25.0% 75.0%
Eastern Europe
Bulgaria 125 39.2% 60.8%
Croatia 127 73.2% 26.8%
Czech Rep 137 27.0% 73.0%
Hungary 129 38.0% 62.0%
Poland 225 39.1% 60.9%
Romania 125 33.6% 66.4%
Slovakia 129 41.1% 58.9%
Slovenia 125 68.0% 32.0%
Albania 163 44.2% 55.2%
Total 3288 49.1% 50.9%
Table 3.1: Transition countries in WBES dataset: number of ?rms and fraction of old and
new ?rms by country
39
Variables and Relevant Survey Questions
Survey Question Possible Answers Variable Values
Measures of Barriers
TxReg – General measure of regulatory barriers
AntComp – General measure of anti-competitive barriers
How problematic are these di?erent 1 – no obstacle Answers about
factors for the operation and growth 2 – minor obstacle taxes and regulations and
of your business: ?nancing, 3 – moderate anti-competitive practices are
infrastructure,taxes and regulations, obstacle recorded as ordered variables
policy instability, in?ation, exchange 4 – major obstacle with values {1,2,3,4}.
rate, functioning of the judiciary,
corruption, street crime, organized
crime, anti-competitive practices by
gov’t or private enterprises?
bl reg – Business Licensing Regulations
cus reg – Customs/Foreign Trade Regulations
lab reg – Labor Regulations
frk reg – Foreign Currency/Exchange Regulations
env reg – Environmental Regulations
?r reg – Fire, Safety Regulations
tadm reg – Tax Regulations/Administration
hit reg – High Taxes
How problematic are these di?erent 1 – no obstacle The answers are recorded as
regulatory areas for the operation 2 – minor obstacle ordered variables
and growth of your business: 3 – moderate with values {1,2,3,4}.
business licensing, customs/foreign obstacle
trade regulations, labor regulations, 4 – major obstacle
foreign currency/exchange
regulations, environmental
regulations, ?re, safety regulations,
tax regulations/administration,
high taxes?
Measures of Excess Employment
LgrS, LgeS, LiSni, LndSni
Have your company’s sales, Increased LgrS=1 if empl-t increased by
investment, exports, imports, Decreased more (decreased by less, in %)
employment and debt changed in Did not change than sales; 0 otherwise.
real terms over the last three LgeS=1 if change in empl-t
years? is greater than or equal to the
By what percentage have they Percentage change change in sales; 0 otherwise
increased or decreased? is given LiSni=1 if empl-t increased
while sales decreased or stayed
constant; 0 otherwise
LndSni =1 if empl-t increased
or stayed constant while sales
decreased or stayed constant;
0 otherwise
Table 3.2: Variables and Relevant Survey Questions
40
Variables and Relevant Survey Questions (cont’d)
Survey Question Possible Answers Variable Values
Measures of Competition
comp hi – Intensity of Competition
Thinking of your None comp hi =1 if the
?rm’s major product 1-3 answer is “More ;
line in the domestic More than 3 than 3”; 0 otherwise
market, how many
competitors do you face?
comp SME – Identity of Major Competitor
From which of the 1 - Domestic small and medium enterprise comp SME=1 for
following would you 2 - Domestic large private enterprise answer #1;
say your company 3 - Foreign ?rm producing in 0 otherwise.
faces the biggest domestic market
competitive threat? 4 - State-owned enterprise
5 - Micro-enterprise/informal sector
6 - Legal imports
7 - Smuggled goods
8 - My ?rm has no e?ective competitors
Controls
new – Indicator variable for new ?rms
How was your ?rm 1 - Originally private, from new=1 for
established? time of start up answer #1;
2 - Privatization of a stated-owned ?rm 0 otherwise.
3 - Private subsidiary of a formerly
state-owned ?rm
4 - Joint venture, domestic and foreign
private owners
5 - State-owned company
6 - State owned some stock
in the company
7 - Agricultural co-operative
8 - State collective farm
9 - Transportation co-operative
age – Age of the ?rm in years
In what year was your actual year is given age = 2000 ?year
?rm founded?
sector – Sector/industry to which the ?rm belongs
1 - manufacturing Used to create
2 - services dummy variables
3 - other for sectors
4 - agriculture
5 - construction
size – Size of the ?rm
Based on number 1 - small ( < 50 employees) Used to create
of employees 2 - medium (between 50 and 500) dummy variables
3 - large (over 500 employees) for size
Table 3.3: Variables and Relevant Survey Questions (cont’d)
41
For this reason, I will not use measures of barriers based on variables cus reg and frk reg.
Each measure of barriers is used as a separate dependent variable in the regressions.
It should be noted that these several measures of barriers may be measuring con-
ceptually di?erent things. Generally speaking, regulations can act as the entry bariers
per se (meaning that they make it more di?cult to start a business) or as an additional
burden on businesses that already exist. While both types of regulatory intervention can
be used to limit the development of the new private ?rms, regulations that are speci?cally
designed as the entry barriers would be a pre?erred choice if the old ?rms are trying to
prevent new entry. Regulations that put an additional burden on an already existing
business might also be used to drive out the new private ?rms, therefore, old ?rms may
also be interested in seeing such regulation created. One of the variables in the dataset –
bl reg, business registration and licensing regulations – is the closest measure available of
the purely entry barriers. Therefore, if lobbying between the old ?rms and the politician
takes place, we would expect the e?ect of it to be most visible in bl reg variable. We will
still use the other measures of barriers since the old ?rms might still want to lobby for the
creation of these regulation in order to drive out the existing new ?rms.
Excess Employment. The measures of excess employment are constructed only
for the old ?rms. The WBES survey provides information on whether company’s sales,
employment, investment, exports, imports, and debt changed over the past three years
and by what percentage they have changed (see table 3.2). In order to create the excess
employment variables I look at the relative movement of sales and employment. Obvi-
ously, excess employment can be present with any combination of movement in sales and
employment, even if both sales and employment decrease (or increase). Therefore, I try
to identify those cases where excess employment is most likely to be present. I consider
several alternative measures of excess employment:
• LgrS is a dummy variable that equals one if employment at the ?rm has increased by
more (or decreased by less, in percentage terms) that sales did, i.e. if the change in
employment is greater than the change in sales with decrease denoted by a negative
change
• LgeS is a dummy variable that equals one if the change in employment at the ?rm
is greater than or equal to the change in sales, with decrease denoted by a negative
42
change
• LiSni is a dummy variable that equals one if employment at the ?rm has increased,
while sales decreased or stayed constant
• LndSni is a dummy variable that equals one if employment at the ?rm has increased
or stayed constant, while sales decreased or stayed constant
Clearly, the ?rms that increase their employment while experiencing a decline (or
no change) in sales are most likely not restructuring – these ?rms are captured by the
variable LiSni. However, even if sales of a ?rm do increase, but at the same time the
?rms increases its employment even more, this might indicate the lack of restructuring as
well. I capture this case with the variable LgrS. Finally, excess employment might also be
present in cases when a ?rm experiences no change in either sales or employment. Since
we are looking only at the old ?rms, if their level of employment relative to sales did
not change, they are likely to have the same level of excess employment that they had
prior to privatization, and most ?rms under the centrally planned system had some excess
employment. This is the reason for using variables LgeS and LndSni.
A note of caution is necessary here. Since the data on change in sales and em-
ployment comes from an answer to a survey question given by the management of the
?rm rather than from an independent sourse of data, the usual concerns of misreporting
are present and magni?ed by the fact that in the transition coutries uno?cial economy
is often substantial and thus the ?rms might misrepresent their sales and employment
considerably. In addition, the information on the percentage change in sales comes from
a question that asks: “By what percentage have sales changed in real terms over the last
three years?” Such question may be especially di?cult to answer for ?rm managers in the
transition countries because these countries have experienced periods of high and volatile
in?ation
6
. It may be easier to determine whether sales have increased or decreased in real
terms than to give a speci?c percentage of the increase (decrease). Therefore, the variables
LiSni and LndSni, which are based on the answer to the questions of whether the sales
6
In addition, many managers of ?rms in the formerly centrally planned economies may have a limited
understanding of economic concepts and therefore are likely to misunderstand what is meant by the phrase
“in real terms”.
43
(employment) have increased (decreased), are likely to be based on more accurate informa-
tion. Thus, these varaiables might be more reliable, even though they provide less precise
measurements since they measure only the dicrete changes in sales and employment.
Finally, I should point out that since the measures of excess employment are based
only on the relative changes in sales and employment they basically measure changing
labor productivity and thus should be interpreted with caution. There may be cases
when the excess employment variables indicate the presence of excess employment for
the ?rms that actually do restructute. For example, if a ?rm engages in restructuring
in the form of introducing new product line, modernizing its equipment, etc., it would
require additional workers but the results would not show up in the increased sales right
away. Our measures of excess employment may mistakingly indicate presence of excess
employment at such ?rms. We try to address such concerns below in section 3.4.2 where
we perform some robustness checks for our results.
Excess employment enters our regressions only through the characteristics of the
old ?rms in each country-sector cell. Thus, for each new ?rm in country i and sector j,
we calculate the fraction of the old ?rms with excess employment in the same country-
sector cell (out of all old ?rms in that country-sector cell). This fraction is calculated for
each measure of excess employment, and resulting variables are: LgrS F, LgeS F, LiSni F,
LndSni F. These variables enter as explanatory variables in the regressions.
Competition and Treat from New Entry. The measures of competition are
constructed for the old ?rms only, because it is the competition from new entrants or the
possibility of such competition that can induce the old ?rms to lobby for the creation
of entry barriers. I will use three measures of competition, re?ecting both the level of
competition and the identity of the major competitor (see table 3.3).
The level of competition is de?ned as high if an old ?rm faces more than three
competitors for its major product line in the domestic market. The dummy variable
comp hi = 1 for the old ?rms that face high competition. We are also interested to
know if the old ?rms face competition primarily from the new ?rms. While precisely
this question is not asked in the survey, a close approximation can be obtained with the
questions about identity of the major competitor. If the major competitor of an old ?rm
is a “domestic small and medium private enterprise” – it is most likely a new entrant.
Thus, the dummy variable compSME = 1 for those old ?rms whose major competitor is
44
a “domestic small and medium private enterprise” (SME).
For each new ?rm, we calculate the fraction of the old ?rms that face high competi-
tion (out of all old ?rms in the same country-sector cell); the resulting variable is comp hiF.
Similarly, for each new ?rm, we also calculate the fraction of the old ?rms whose major
competitor is a private domestic SME (out of all old ?rms in the same country-sector
cell); the resulting variable is compSME F. comp hiF and compSME F enter as explana-
tory variables in the regressions.
Another variable – relative number of new vs. old ?rms in a sector – can be thought
of as an alternative measure of the competition from new ?rms. A larger fraction of new
?rms in a sector means that the new ?rms are more likely to create signi?cant competition
for the old ?rms, and thus the old ?rms would have even more incentives to create entry
barriers, leading to a higher level of barriers faced by the new ?rms. This variable, New F,
is measured as a fraction of new ?rms in each country-sector cell (out of all ?rms in that
country-sector cell), and it enters as an explanatory variable in the regressions.
It should be noted that all three variables used to measure the level of competition
(comp hiF, compSME F, and New F) are potentially endogenous, i.e. the level of entry
barriers can in?uence the values of these variables. This issue is addressed in more detail
below.
3.3 Estimation Procedure
Our model argues that under certain conditions (high threat from new entry, low cost of
excess employment, etc.) the managers of the old ?rms will lobby with politicians for the
creation of entry barriers. If such lobbying takes place, we should observe both excess
employment and a high level of entry barriers. Therefore, our empirical exercise will be
structured as follows.
First, we want to check whether we indeed observe that high level of entry bar-
riers faced by the new ?rms corresponds to presence of excess employment (i.e. non-
restructuring) at the old ?rms. If we observe this correspondence, it means that there is a
possibility that lobbying between the managers of old ?rms and the politician takes place.
Having established that a high level of entry barriers corresponds to the presence of excess
employment, we will then focus on one of the factors that induces the old ?rms to create
45
entry barriers – the threat from new entry. If new entry is very threatening, managers
of the old ?rms will lobby with the politician for the creation of entry barriers, and we
should observe a high level of barriers.
Since all dependent variables in our data are ordered qualitative variables, we use
ordered probit estimation procedure. We run regressions using observations for the new
?rms only. Tables 3.2 and 3.3 provide summary of the survey questions used to create all
variables.
The following measures of barriers were used as dependent variables (higher values
mean more severe burden of regulations):
• TxReg - taxes and regulations in general; ordered variable with values {1, 2, 3, 4}
• AntComp - anti-competitive practices by government and private enterprises; or-
dered variable with values {1, 2, 3, 4}
• bl reg - business registration and licensing regulations; ordered variable with values
{1, 2, 3, 4}
• lab reg - labor regulations; ordered variable with values {1, 2, 3, 4}
• env reg - environmental regulations; ordered variable with values {1, 2, 3, 4}
• fir reg - ?re, safety regulations; ordered variable with values {1, 2, 3, 4}
• hit reg - high taxes; ordered variable with values {1, 2, 3, 4}
• tadm reg - tax administration and regulations; ordered variable with values {1, 2,
3, 4}
The independent variables in the regressions are:
• EE - equals to 1 if the new ?rm is located in Eastern Europe, 0 otherwise;
• Balt - equals to 1 if the new ?rm is located in the Baltic states of the former Soviet
Union, 0 otherwise;
• SizeS - equals to 1 if ?rm is small size, 0 otherwise;
• SizeM - equals to 1 if ?rm is medium size, 0 otherwise (the omitted category is large
size);
46
• sector1, sector2, sector3, sector4 - dummy variable for each sector (the omitted
category is sector 5 (construction));
• Age - age of the new ?rm in years in 2000 (the time when the survey was taken);
• New F - fraction of new ?rms out of all ?rms in the country-sector cell;
• compSME F - fraction of old ?rms whose major competitor is domestic private
SME out of all old ?rms in the country-sector cell;
• comp hiF - fraction of old ?rms who face high competition out of all old ?rms in
the country-sector cell;
• LgrS F, LgeS F, LiSni F, LndSni F - fraction of old ?rms with excess employ-
ment (according to measures LgrS, LgeS, LiSni, LndSni described above) out of all
old ?rms in the country-sector cell
We run two sets of regressions. The ?rst one regresses the level of barriers faced by
the new ?rms on the excess employment measures (and the region, size, age, and sector
controls). These regressions would establish whether there is a positive correspondence
between the level of barriers faced by the new ?rms and the presence of excess employment
at the old ?rms. The results for these regressions are presented in tables 3.4 though 3.11.
The second set of regressions regresses the level of barriers on the measures of competition
from new entry (and the region, size, age, and sector controls). These regressions would
establish if higher competition from the new entry results in a higher level of entry barriers.
The results for these regressions are presented in tables 3.16 through 3.20. If we establish
both that higher competition from the new entry results in a higher level of barriers and
that the higher level of barriers corresponds to the presence of excess employment, then
we can take it as an indirect evidence that competition from new entry prompts the old
?rms to lobby with politicians for the creation of entry barriers in exchange for keeping
excess employment.
I will now discuss the expected signs for the explanatory variables in the regressions.
The regional dummies, EE and Balt, are intended to capture the di?erence between
regions in the general level of barriers and regulations that is due to factors unrelated to
economics (cultural traditions, historical developments, legal systems etc.) as well as
factors related to economics but not captured by the model. Literature on transition
47
economies generally concludes that Baltic states resemble countries of Eastern Europe
much more than they do other countries of the former Soviet Union, therefore, I do not
group Baltic states
7
with other countries of the former Soviet Union. Our model does not
predict a speci?c sign for these dummy variables, but anecdotal evidence would suggest
that the level of entry barriers is lower in the Eastern Europe than in the former Soviet
Union.
Likewise, the size dummies and sector dummies are intended to capture di?erences
in the general level of regulations faced by ?rms due to their size and the nature of their
business. For example, large ?rms might face more labor regulations due to the number of
their employees, or ?rms in agriculture might face more licensing regulations, since they
have to deal with the land ownership and handling of food products. Our model does not
predict a speci?c sign for these variables.
Age variable is introduced to capture the fact that some barriers are more of an issue
for young ?rms that are just starting out (such as the business registration and licenses
that are required to open a business). Therefore, burden of some barriers might diminish
with the age of the ?rm.
The competition measures New F, compSME F, and comp hiF, would have a
positive sign if higher competition from the new entrants induces the old ?rms to lobby
with politicians for the creation of entry barriers. On the other hand, the level of barriers
can also in?uence these competition measures (even if no lobbying by the old ?rms takes
place): if the entry barriers are high, we would expect fewer new ?rms to enter the sector,
leading to lower values of New F and compSME F, and lower values of comp hiF if
competition comes at least in part from new entrants. Note that higher level of barriers in
the absence of lobbying by the old ?rms would lead to lower competition, thus implying
negative coe?cients on New F, compSME F, and comp hiF. But, if the lobbying by
the old ?rms takes place, higher values of New F, compSME F, and comp hiF would
lead to higher level of entry barriers, implying positive coe?cients. Thus, because of
the endogeneity of our competition measures, we are biased against ?nding a positive
relationship between the level of barriers and competition from new entry. This means
that if we do ?nd a positive relationship despite the bias, we can interpret it as evidence
supporting the idea that higher competition from the new ?rms induces the old ?rms to
7
Estonia and Lithuania in our sample.
48
lobby for entry barriers.
If the measures of excess employment, LgrS F, LgeS F, LiSni F, and LndSni F,
have positive coe?cients, it would provide support to the idea that the old ?rms lobby
for the creation of entry barriers in exchange for excess employment. If the coe?cient is
negative or not signi?cant, then either the lobbying does not take place or the old ?rms
reward politicians by means other than excess employment.
It is possible, however, that even if the old ?rms do lobby with politicians for the
creation of entry barriers in exchange for excess employment, it would be di?cult to ?nd
this relationship in our data. If one large existing ?rm lobbies for the creation of entry
barriers in exchange for excess employment, the other existing ?rms in the same sector
would get the bene?t of lower new entry without providing excess employment. Since
we measure excess employment as a fraction of the old ?rms in the sector with excess
employment, such “free-riding” by some old ?rms would obscure the relationship between
the entry barriers and excess employment, once again biasing us against ?nding a positive
relationship between the level of barriers the new ?rms face and the presence of excess
employment at the old ?rms.
3.4 Interpretation of Results
3.4.1 Barriers faced by new ?rms vs. Excess employment at old ?rms
We ?rst run regressions of the level of barriers faced by the new ?rms on excess employment
at the old ?rms. For each measure of barriers, we run an ordered probit regression with
the regional dummies, size dummies, sector dummies, age, and the measures of excess
employment as independent variables. The results for these regressions for each measure
of barriers are presented in tables 3.4 though 3.11.
The obvious result is that excess employment measures do not have uniformly posi-
tive and statistically signi?cant coe?cients in all equations. Only the equations with bl reg
(business registration and licensing regulations) as the dependent variable have positive
and signi?cant coe?cients for all measures of excess employment (see table 3.6). Thus, it
seems that as more old ?rms in a sector have excess employment, the new ?rms in that
sector are more likely to face high burden of business registration and licensing regulations.
49
Dependent variable - TxReg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 1.1086*** 1.0644*** 1.1524*** 1.1982***
[.000] [.001] [.000] [.000]
EE -0.0994 -0.1134* -0.1271* -0.1301**
[.137] [.086] [.052] [.047]
Balt -0.4149*** -0.3801*** -0.4053*** -0.4166***
[.000] [.000] [.000] [.000]
SizeS 0.6497** 0.6646** 0.6861** 0.6953**
[.025] [.022] [.018] [.016]
SizeM 0.6172** 0.6249** 0.6362** 0.6409**
[.035] [.033] [.029] [.028]
sector1 -0.0526 0.0030 0.0161 0.0114
[.653] [.979] [.886] [.919]
sector2 -0.1990* -0.1578 -0.1367 -0.1390
[.060] [.124] [.178] [.170]
sector3 -0.2028 -0.2133 -0.1864 -0.2014
[.452] [.429] [.489] [.456]
sector4 -0.1256 -0.0743 -0.0553 -0.0558
[.392] [.605] [.701] [.695]
Age -0.0034 -0.0029 -0.0039 -0.0045
[.779] [.814] [.752] [.712]
LgrS F 0.4719**
[.037]
LgeS F 0.2307
[.150]
LiSni F 0.0390
[.922]
LndSni F -0.1508
[.419]
µ
3
0.6886*** 0.6892*** 0.6884*** 0.6887***
[.000] [.000] [.000] [.000]
µ
4
1.5957*** 1.5953*** 1.5937*** 1.5944***
[.000] [.000] [.000] [.000]
Nobs 1577 1577 1577 1577
LR(zero slopes) 33.84*** 31.50*** 29.44*** 30.08***
[.000] [.000] [.001] [.001]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.4: Barriers (TxReg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
50
Dependent variable - AntComp
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.9768*** 0.9736*** 1.0049*** 1.0304***
[.002] [.002] [.001] [.001]
EE -0.2445*** -0.2667*** -0.2784*** -0.2756***
[.000] [.000] [.000] [.000]
Balt -0.2625** -0.2339** -0.3047*** -0.2472**
[.015] [.032] [.006] [.023]
SizeS -0.0881 -0.0629 0.0741 -0.0497
[.756] [.824] [.794] [.861]
SizeM -0.1444 -0.1313 -0.1359 -0.1245
[.614] [.646] [.635] [.663]
sector1 -0.2354** -0.1723 -0.1679 -0.1596
[.037] [.115] [.121] [.141]
sector2 -0.2573** -0.2026** -0.1987** -0.1903*
[.012] [.040] [.042] [.052]
sector3 -0.0112 -0.0096 0.0227 -0.0061
[.968] [.972] [.935] [.983]
sector4 -0.1232 -0.0590 -0.0942 -0.0471
[.389] [.673] [.503] [.735]
Age -0.0063 -0.0062 -0.0050 -0.0069
[.604] [.611] [.680] [.574]
LgrS F 0.5312**
[.020]
LgeS F 0.1474
[.367]
LiSni F 0.9767**
[.018]
LndSni F 0.0065
[.973]
µ
3
0.6612*** 0.6601*** 0.6613*** 0.6601***
[.000] [.000] [.000] [.000]
µ
4
1.3349*** 1.3319*** 1.3352*** 1.3315***
[.000] [.000] [.000] [.000]
Nobs 1409 1409 1409 1409
LR(zero slopes) 36.99*** 32.37*** 37.16*** 31.55***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.5: Barriers (AntComp) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
51
Dependent variable - bl reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.3499 0.2834 0.4040 0.3204
[.282] [.392] [.213] [.330]
EE -0.2946*** -0.3195*** -0.3486*** -0.3349***
[.000] [.000] [.000] [.000]
Balt -0.2425** -0.1804* -0.3233*** -0.1837*
[.023] [.094] [.003] [.088]
SizeS -0.1019 -0.0711 -0.0827 -0.0531
[.734] [.813] [.782] [.859]
SizeM -0.2788 -0.2605 -0.2636 -0.2490
[.357] [.389] [.383] [.411]
sector1 -0.0678 0.0223 0.0354 0.0644
[.552] [.840] [.748] [.560]
sector2 -0.0849 -0.0090 0.0057 0.0308
[.412] [.929] [.954] [.758]
sector3 -0.0630 -0.0708 -0.0169 0.0000
[.821] [.798] [.952] [.999]
sector4 0.3904*** 0.4715*** 0.4210*** 0.5077***
[.006] [.001] [.003] [.000]
Age -0.0234** -0.0224* -0.0218* -0.0226*
[.054] [.066] [.074] [.063]
LgrS F 0.8847***
[.000]
LgeS F 0.3963**
[.013]
LiSni F 1.7134***
[.000]
LndSni F 0.4157**
[.024]
µ
3
0.5863*** 0.5848*** 0.5870*** 0.5849***
[.000] [.000] [.000] [.000]
µ
4
1.3488*** 1.3415*** 1.3485*** 1.3401***
[.000] [.000] [.000] [.000]
Nobs 1540 1540 1540 1540
LR(zero slopes) 87.30*** 77.19*** 89.78*** 76.09***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.6: Higher barriers (bl reg) faced by the new ?rms correspond to higher fraction
of the old ?rms with excess employment (ordered probit regressions).
52
Dependent variable - lab reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.0483 -0.0185 -0.0067 -0.1106
[.883] [.956] [.984] [.740]
EE 0.5312*** 0.5568*** 0.5526*** 0.5619***
[.000] [.000] [.000] [.000]
Balt 0.6085*** 0.5984*** 0.5729*** 0.6248***
[.000] [.000] [.000] [.000]
SizeS 0.1162 0.0802 0.0765 0.0688
[.703] [.793] [.802] [.821]
SizeM 0.1206 0.1017 0.1007 0.0971
[.695] [.741] [.743] [.752]
sector1 -0.1984* -0.2619** -0.2604** -0.2436**
[.079] [.016] [.016] [.025]
sector2 -0.2870*** -0.3461*** -0.3454*** -0.3358***
[.005] [.000] [.000] [.001]
sector3 -0.0718 -0.0874 -0.0773 -0.0449
[.790] [.746] [.774] [.868]
sector4 0.0189 -0.0547 -0.0693 -0.0472
[.894] [.694] [.620] [.732]
Age -0.0171 -0.0163 -0.0159 -0.0148
[.161] [.184] [.194] [.225]
LgrS F -0.4237*
[.053]
LgeS F 0.0537
[.736]
LiSni F 0.3430
[.383]
LndSni F 0.3756**
[.044]
µ
3
0.7319*** 0.7307*** 0.7307*** 0.7316***
[.000] [.000] [.000] [.000]
µ
4
1.5634*** 1.5618*** 1.5623*** 1.5654***
[.000] [.000] [.000] [.000]
Nobs 1545 1545 1545 1545
LR(zero slopes) 96.82*** 93.18*** 93.83*** 97.13***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.7: Bariers (lab reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
53
Dependent variable - env reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.1355 0.0359 0.1193 0.0141
[.688] [.917] [.723] [.967]
EE -0.0147 0.0118 0.0009 0.0067
[.832] [.864] [.988] [.922]
Balt -0.0616 -0.0483 -0.0459 -0.0400
[.584] [.671] [.689] [.724]
SizeS 0.1146 0.0725 0.1021 0.0735
[.714] [.817] [.744] [.814]
SizeM 0.2002 0.1751 0.1914 0.1764
[.525] [.578] [.543] [.575]
sector1 -0.1799 -0.2294** -0.2123* -0.2045*
[.118] [.040] [.056] [.065]
sector2 -0.2730*** -0.3204*** -0.3017*** -0.2978***
[.009] [.002] [.003] [.003]
sector3 -0.4259 -0.4572 -0.4379 -0.4046
[.153] [.126] [.142] [.175]
sector4 0.1646 0.1102 0.1464 0.1309
[.256] [.438] [.306] [.354]
Age -0.0126 -0.0113 -0.0129 -0.0107
[.325] [.380] [.314] [.403]
LgrS F -0.2630
[.246]
LgeS F 0.1818
[.275]
LiSni F -0.3750
[.366]
LndSni F 0.3186*
[.098]
µ
3
0.6569*** 0.6564*** 0.6570*** 0.6565***
[.000] [.000] [.000] [.000]
µ
4
1.3898*** 1.3904*** 1.3896*** 1.3917***
[.000] [.000] [.000] [.000]
Nobs 1463 1463 1463 1463
LR(zero slopes) 28.67*** 28.51*** 28.14*** 30.05***
[.001] [.001] [.002] [.001]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.8: Bariers (env reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
54
Dependent variable - fir reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const -0.0184 -0.1396 -0.0419 -0.1821
[.959] [.701] [.907] [.615]
EE -0.0815 -0.0583 -0.0759 -0.0668
[.225] [.382] [.251] [.313]
Balt 0.0269 0.0497 -0.0072 0.0661
[.798] [.640] [.947] [.534]
SizeS 0.3586 0.3084 0.3272 0.3105
[.289] [.360] [.332] [.356]
SizeM 0.4455 0.4140 0.4278 0.4169
[.190] [.222] [.207] [.218]
sector1 -0.1689 -0.2086* -0.1920* -0.1691
[.136] [.057] [.078] [.122]
sector2 -0.2946*** -0.3336*** -0.3157*** -0.2988***
[.004] [.001] [.002] [.003]
sector3 -0.3482 -0.3740 -0.3414 -0.2916
[.214] [.183] [.224] [.299]
sector4 -0.1126 -0.1598 -0.1633 -0.1284
[.431] [.253] [.247] [.356]
Age -0.0238* -0.0221* -0.0226* -0.0212*
[.054] [.074] [.068] [.086]
LgrS F -0.1366
[.534]
LgeS F 0.2881*
[.076]
LiSni F 0.5019
[.204]
LndSni F 0.5304***
[.005]
µ
3
0.7678*** 0.7684*** 0.7682*** 0.7697***
[.000] [.000] [.000] [.000]
µ
4
1.5081*** 1.5099*** 1.5086*** 1.5127***
[.000] [.000] [.000] [.000]
Nobs 1552 1552 1552 1552
LR(zero slopes) 25.28*** 28.03*** 26.50*** 32.82***
[.005] [.002] [.003] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.9: Barriers (fir reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
55
Dependent variable - hit reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 1.5333*** 1.6199*** 1.5407*** 1.6309***
[.000] [.000] [.000] [.000]
EE -0.2619*** -0.2756*** -0.2638*** -0.2711***
[.000] [.000] [.000] [.000]
Balt -0.4148*** -0.4348*** -0.4068*** -0.4406***
[.000] [.000] [.000] [.000]
SizeS 0.3225 0.3451 0.3300 0.3416
[.301] [.268] [.289] [.272]
SizeM 0.2965 0.3074 0.2999 0.3049
[.346] [.328] [.340] [.332]
sector1 0.0893 0.1077 0.0969 0.0802
[.484] [.385] [.432] [.517]
sector2 -0.0384 -0.0186 -0.0313 -0.0432
[.739] [.868] [.779] [.699]
sector3 -0.2739 -0.2541 -0.2746 -0.3061
[.330] [.367] [.329] [.278]
sector4 -0.2019 -0.1769 -0.1857 -0.2022
[.203] [.252] [.234] [.188]
Age -0.0081 -0.0092 -0.0084 -0.0096
[.544] [.493] [.530] [.475]
LgrS F 0.0408
[.862]
LgeS F -0.1972
[.251]
LiSni F -0.1373
[.742]
LndSni F -0.2918
[.146]
µ
3
0.4401*** 0.4399*** 0.4403*** 0.4399***
[.000] [.000] [.000] [.000]
µ
4
1.1700*** 1.1700*** 1.1701*** 1.1704***
[.000] [.000] [.000] [.000]
Nobs 1570 1570 1570 1570
LR(zero slopes) 32.99*** 34.28*** 33.07*** 35.06***
[.000] [.000] [.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.10: Barriers (hit reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
56
Dependent variable - tadm reg
ExEmpl = LgrS F LgeS F LiSni F LndSni F
Const 0.7711** 0.7452** 0.8017*** 0.8262***
[.013] [.018] [.010] [.009]
EE -0.1873*** -0.1967*** -0.2048*** -0.2063***
[.004] [.002] [.001] [.001]
Balt -0.3094*** -0.2875*** -0.2993*** -0.3059***
[.002] [.005] [.004] [.003]
SizeS 0.4177 0.4289 0.4419 0.4455
[.143] [.133] [.121] [.118]
SizeM 0.4700 0.4753* 0.4821* 0.4838*
[.102] [.098] [.094] [.092]
sector1 -0.0816 -0.0489 -0.0382 -0.0414
[.468] [.653] [.725] [.703]
sector2 -0.1766* -0.1494 -0.1372 -0.1392
[.083] [.132] [.163] [.157]
sector3 -0.4650* -0.4707* -0.4561* -0.4641*
[.075] [.072] [.081] [.076]
sector4 -0.2194 -0.1827 -0.1688 -0.1703
[.123] [.188] [.228] [.217]
Age 0.0159 0.0163 0.0156 0.0153
[.179] [.168] [.188] [.198]
LgrS F 0.3055
[.153]
LgeS F 0.1402
[.367]
LiSni F -0.0043
[.991]
LndSni F -0.0798
[.660]
µ
3
0.5568*** 0.5566*** 0.5567*** 0.5568***
[.000] [.000] [.000] [.000]
µ
4
1.3782*** 1.3777*** 1.3775*** 1.3776***
[.000] [.000] [.000] [.000]
Nobs 1574 1574 1574 1574
LR(zero slopes) 29.50*** 28.27*** 27.46*** 27.65***
[.001] [.002] [.002] [.002]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.11: Barriers (tadm reg) faced by the new ?rms and excess employment at the old
?rms (ordered probit regressions).
57
Since bl reg is the variable which is closest to measuring purely entry barriers, this result
is consistent with the old ?rms lobbying with politicians to restrict new entry through the
creation of entry barriers.
Other measures of barriers are positively related to some of the measures of excess
employment. Speci?cally, equations for TxReg have positive and signi?cant coe?cient on
LgrS F, and equations for AntComp have positive and signi?cant coe?cients on both
LgrS F and LiSni F, while coe?cients on other excess employment measures are not
statistically signi?cant. Equations for env reg have positive and signi?cant coe?cient on
LndSni F, and equations for fir reg have positive and signi?cant coe?cients on both
LgeS F and LndSni F. None of the excess employment measures has a statistically sig-
ni?cant coe?cient in the equations with hit reg and tadm reg as the dependent variables.
Therefore, there might be a relationship between excess employment at the old ?rms and
the extent to which taxes and regulations, anti-competitive practices, environmental reg-
ulations, and ?re/safety regulations create obstacles for the new ?rms. However, there
seems to be no relationship between the level of excess employment and the extent to
which tax administration or high taxes are obstacles for the new ?rms
8
.
The equations with lab reg as dependent variable di?er from other equations in
this group in several respects. First of all, there is a positive and signi?cant coe?cient
on LndSni F but negative and signi?cant coe?cient on LgrS F. Therefore, we cannot
say if excess employment corresponds to higher or lower level of labor regulation barriers.
In addition, both EE and Balt have positive and statistically signi?cant coe?cients in
regressions with lab reg as a dependent variable, while in all other regressions in this group
EE and Balt have either negative and signi?cant coe?cients, or coe?cients that are not
statistically signi?cant. It would seem that Eastern Europe and Baltic states usually have
lower level of regulations, except for labor regulations
9
.
We have established that there is a strong positive relationship between the level
8
This does not mean, however, that high taxes or the tax administration do not create obstacles for the
enterprise development in these countries. To the contrary, in the survey the majority of ?rms surveyed,
both old and new, indicated that high taxes and tax administration present either moderate or major
obstacle for the operation of their enterprise.
9
This may be due to the fact that most labor regulations are not enforced in the former Soviet Union,
and thus most ?rms simply ignore them.
58
of business licensing regulations faced by the new ?rms (measured by bl reg) and excess
employment at the old ?rms. Our results also suggest that there might be a positive
relationship between the barriers measured by TxReg, AntComp, env reg, and fir reg
and at least some measures of excess employment. There is, however, no relationship
between the barriers measured by hit reg or tadm reg and excess employment, and the
direction of relationship between the labor regulations and excess employment depends
on which measure of excess employment is used.
The coe?cients on other variables in the regressions also have intuitive signs, which
I will describe using the equation with business registration and licensing regulations as
an example (results are presented in table 3.6). Eastern Europe and Baltic states seems
to have lower level of regulatory barriers than the former Soviet Union, in line with the
anecdotal evidence. Size (number of employees) of the ?rms does not seem to in?uence
the burden of business licensing regulations, while sector does in?uence the burden of
regulations, with agricultural ?rms (sector 4) facing higher burden of business registration
and licensing regulations, which is not surprising. Also, the burden of business registration
and licensing regulations diminishes with the age of the ?rm, which should be expected.
If we look at the general burden of taxes and regulations (regression with TxReg as
the dependent variable, presented in table 3.4), we ?nd that, once again, Eastern Europe
and Baltic states have lower level of barriers, that small and medium ?rms are more
likely to ?nd taxes and regulations to be a major obstacle, and the burden of taxes and
regulations does not seems to change much with the age of the ?rm.
To summarize, according to our results, it is very likely that the old ?rms engage
in lobbying for the creation of the business registration and licensing regulations. This
is not surprising, since business registration and licensing regulations are closest to the
purely entry barriers. It is less likely but also possible that they lobby for the creation of
environmental or ?re/safety regulations, as well as the general anti-competitive practices.
It is unlikely that the old ?rms lobby for the creation of labor regulations, for high taxes,
or complicated tax administration regulations.
3.4.2 Alternative Speci?cations and Robustness Checks
We will now perform several robustness checks on our results.
59
First, our measures of excess employment may simply re?ect poor performance by
the old ?rms, even if their poor performance is not related to any lobbying activities. In
other words, it is possible that excess employment at old ?rms is in?uenced by the level
of regulatory barriers in the sector. If the level of regulations and barriers is determined
exogenously (not through lobbying by the old ?rms), then high level of regulations and
barriers would adversely a?ect all ?rms in the sector, whether new or old. As a result,
given the way our variables are constracted, we may see a positive relationship between
the level of barriers faced by the new ?rms and poor performance (excess employment) at
the old ?rms. We can check for this possibility by regressing the level of barriers faced by
the old ?rms on the excess employment at the old ?rms. If the old ?rms do not engage in
lobbying, then high level of barriers faced by the old ?rms should lead to poor performance
by these ?rms and thus should be positively related to our measures of excess employment
at the old ?rms. However, if the old ?rms do lobby for the creation of barriers designed
to restrict new entry, then such entry barriers should not be damaging to the old ?rms
themselves and we should see either negative or non-signi?cant coe?cients on the excess
employment measures.
Summary results for these regressions are presented in table 3.12, where we present
only the coe?cients on the excess employment variables. The regression speci?cation in
these equations is identical to that in tables 3.4 though 3.11, except that we use observa-
tions for the old ?rms only
10
.
Most coe?cients in table 3.12 are not statistically signi?cant and some of them are
negative and statistically signi?cant. This seems to indicate that there is no strong positive
correspondence between the level of barriers faced by the old ?rms and the presence of
excess employment at these ?rms, i.e. high level of regulatory barriers does not induce poor
performance among the old ?rms. This ?nding is consistent with the level of regulatory
barriers being determined by the old ?rms lobbying with politicians.
However, some coe?cients in table 3.12 corresponding to LiSni F measure of excess
employment are positive and signi?cant. This may be due to the fact that our de?nition
of the old ?rms might include some ?rms that are actually new, such as joint ventures. As
a consistency check, we run the same speci?cation of regressions, but restrict our sample
to the old ?rms that were established in 1991 or earlier, because these are more likely
10
Full regression results are available from the author.
60
Observations for OLD ?rms only
Variables LgrS F LgeS F LndSni F LiSni F N of Obs
TxReg 0.3742 0.3117* -0.5101** -0.7249 1546
[.183] [.097] [.033] [.207]
AntComp -0.0419 -0.2149 -0.5625** 0.8324 1378
[.885] [.266] [.026] [.177]
bl reg 0.3515 0.1167 0.0593 1.1543* 1476
[.234] [.554] [.814] [.072]
lab reg -0.0844 -0.1277 0.0191 1.3696** 1533
[.751] [.500] [.938] [.207]
env reg 0.1044 -0.0832 -0.0142 1.6975*** 1493
[.708] [.658] [.953] [.004]
fir reg -0.2357 -0.0658 0.2211 1.6456*** 1535
[.398] [.727] [.358] [.004]
hit reg 0.2041 -0.1227 -1.1024*** -0.8153 1543
[.516] [.550] [.000] [.191]
tadm reg 0.2309 0.0351 -0.6324*** -0.3481 1545
[.396] [.847] [.007] [.538]
– measures of barriers are dependent variables in the regressions
– using observations for old ?rms only
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– full regression speci?cation can be seen in table 3.4
Table 3.12: Level of barriers faced by the old ?rms is not systematically positively related
to the presence of excess employment at the old ?rms (ordered probit regressions).
to be the classic incumbent ?rms and therefore likely to have political in?uence and be
able to lobby for their preferred policies. Summary of the results from this estimation
is presented in table 3.13, where we again present only the coe?cients on the excess
employment variables. In the restricted sample, none of the coe?cients are positive and
statistically signi?cant.
Thus, our results suggest that it is unlikely that the regulatory barriers are deter-
mined exogenously. Instead, our results are consistent with the argument that old ?rms
lobby for the creation of entry barriers, which is why these regulatory barriers do not
a?ect the old ?rms adversely.
It was mentioned earlier (see page 44) that our measures of excess employment
may mistakingly indicate the presence of excess employment at the ?rms that actually do
restructure by engaging in expansion and investment activities, which require additional
workers but a?ect sales only with a long lag. It is also possible that lobbying for the
creation of entry barriers leads to the lack of restructuring by the old ?rms, as measured
61
Observations for OLD ?rms established prior to 1992 only
Variables LgrS F LgeS F LndSni F LiSni F N of Obs
TxReg 0.2317 0.4491 -0.1485 -0.1294 1546
[.630] [.138] [.708] [.895]
AntComp -0.5888 -0.7777** -0.7234* 1.4045 1378
[.243] [.014] [.085] [.175]
bl reg -0.2031 -0.1663 -0.0391 -0.1179 1476
[.722] [.613] [.926] [.921]
lab reg -0.3494 -0.0784 0.3147 1.0997 1533
[.466] [.794] [.415] [.259]
env reg -0.3661 -0.0515 -0.0029 1.0895 1493
[.456] [.866] [.994] [.269]
fir reg -0.8569* 0.0746 0.3847 0.3457 1535
[.090] [.806] [.332] [.730]
hit reg 0.9058 0.4979 -0.6403 -0.0635 1543
[.104] [.133] [.130] [.953]
tadm reg -0.2038 -0.0024 -0.5128 -1.3657 1545
[.671] [.994] [.184] [.167]
– measures of barriers are dependent variables in the regressions
– using observations for old ?rms established prior to 1992 only
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– full regression speci?cation can be seen in table 3.4
Table 3.13: Level of barriers faced by the old ?rms is not positively related to the presence
of excess employment at the old ?rms (ordered probit regressions, sample of old ?rms
restricted to ?rms established prior to 1992).
by aspects other than excess employment. Therefore, we want to check that our result
– high entry barriers for the new ?rms corresponding to the lack of restructuring by the
old ?rms – does not depend solely on our de?nition of the excess employment variables.
We do this by using an alternative measure of non-restructuring based on the change in
investment:
• InvD F is a fraction of old ?rms in the country-sector cell whose investment has
decreased in real terms over the past three years
• InvCD F is a fraction of old ?rms in the country-sector cell whose investment has
decreased or stayed constant in real terms over the past three years
We then regress the level of barriers faced by the new ?rms on the region dummies, size
dummies, sector dummies, age, and these investment measures, using observations for the
new ?rms only. Results are presented in table 3.14, where we present only the coe?cients
62
Measure lack of restructuring by change in investment
Dependent Variables InvD F InvCD F N of Obs
TxReg 0.4719** 0.2307 1577
[.037] [.150]
AntComp 0.5312** 0.1474 1409
[.020] [.367]
bl reg 0.8847*** 0.3963** 1540
[.000] [.013]
lab reg -0.4237* 0.0537 1545
[.053] [.736]
env reg -0.2630 0.1818 1463
[.246] [.275]
fir reg -0.1366 0.2881* 1552
[.534] [.076]
hit reg 0.0408 -0.1972 1570
[.862] [.251]
tadm reg 0.3055 0.1402 1574
[.153] [.367]
– measures of barriers are dependent variables in the regressions
– using observations for new ?rms established in 1989 and later
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– full regression speci?cation can be seen in table 3.4
Table 3.14: Barriers faced by the new ?rms and non-restructuring by the old ?rms, as
measured by the change in investment (ordered probit regressions).
on the investment measures.
The key result is the same as before – the level of business licensing regulations faced
by the new ?rms is positively related to the fraction of the old ?rms in the same sector
that do not restructure. There is also some evidence that taxes and regulations (TxReg)
and anti-competitive measures (AntComp) faced by the new ?rms are positively related
to non-restructuring by the old ?rms.
As a ?nal robustness check, we allow for country ?xed e?ects, instead of grouping
countries into regions. Speci?cally, we regress the level of barriers faced by the new
?rms on the country-speci?c intercept, size dummies, sector dummies, age, and the excess
employment measures. Summary of the results is presented in table 3.15, where we present
only the coe?cients on the excess employment variables.
Our key result still survives – the level of business licensing regulations faced by the
new ?rms is positively related to the fraction of the old ?rms with excess employment.
Thus, our results suggest that old ?rms do lobby for the creation of business licensing
63
Observations for NEW ?rms only, Country E?ects
Variables LgrS F LgeS F LndSni F LiSni F N of Obs
TxReg 0.1685 0.2239 0.0070 -0.0253 1577
[.576] [.382] [.981] [.963]
AntComp -0.0060 -0.0362 -0.0336 0.0285 1409
[.242] [.885] [.908] [.958]
bl reg 0.9222*** 0.6183** 0.6939** 1.0303* 1540
[.002] [.017] [.019] [.053]
lab reg -0.2243 0.1242 0.0774 -0.2055 1545
[.444] [.621] [.789] [.695]
env reg 0.0233 0.1899 0.3071 -0.2636 1463
[.939] [.475] [.312] [.643]
fir reg 0.2162 0.3268 0.4335 0.2868 1552
[.465] [.208] [.136] [.593]
hit reg 0.1701 0.2442 0.1285 0.2512 1570
[.591] [.383] [.690] [.655]
tadm reg 0.2446 0.2107 -0.0621 0.0688 1574
[.391] [.390] [.826] [.893]
– measures of barriers are dependent variables in the regressions
– using observations for new ?rms established after 1989 only
– p-values in brackets
– * signif. at 10%, ** signif. at 5%, *** signif. at 1%
– regressions include country-speci?c intercept
Table 3.15: Barriers faced by the new ?rms and excess employment at the old ?rms
(ordered probit regressions with country-speci?c intercept).
regulations to prevent entry of the new ?rms.
3.4.3 Barriers faced by the new ?rms vs. Threat to the old ?rms from new entry
In the next step we investigate whether the competitive threat from new entry results
in a higher level of barriers for the new ?rms, which would be consistent with old ?rms
lobbying for the creation of entry barriers. Clearly, this exercise makes sense only for those
measures of barriers that were positively related to excess employment in the ?rst group
of regressions, namely bl reg and maybe TxReg, AntComp, env reg, and fir reg. For
these measures of barriers, we run an ordered probit regression with the regional dummies,
size dummies, sector dummies, age, and the measures of competition from new entry as
independent variables. The results for these regressions are presented in tables 3.16 though
3.20. For each measure of barriers we run two speci?cations: with and without interaction
64
terms between the competition measures and the regional dummies
11
.
In all equations at least one measure of competition from new entry has a positive
and statistically signi?cant coe?cient. Higher fraction of new ?rms in a sector (higher
New F) results in higher barriers for new ?rms as measured by TxReg, bl reg, env reg,
and fir reg.
As more old ?rms in a sector face high competition (comp hiF increases), the new
?rms in those sectors are more likely to face high barriers as measured by AntComp,
bl reg, and fir reg. This e?ect is smaller in Eastern Europe, i.e. in Eastern Europe high
competition is less likely to translate into high barriers for the new ?rms.
As more old ?rms in a sector face competition primarily from private domestic SMEs
(compSME F increases), the new ?rms in those sectors are more likely to face high barri-
ers as measured by TxReg, but less likely to face high barriers as measured by AntComp
and bl reg. This relationship di?ers by region: in Eastern Europe high compSME F is
less likely to translate into high TxReg barriers but more likely to translate into high
AntComp and bl reg barriers. As mentioned earlier, negative relationship between the
level of barriers and compSME F may be due to reverse causality: lower AntComp and
bl reg barriers lead to more new ?rms entering, increasing the number of private domestic
SMEs, which leads to higher compSME F. However, even with this possibility of reverse
causality, in all equations at least one of the competition measures has a positive and
statistically signi?cant coe?cient.
The e?ect of competition from the new ?rms on each measure of barriers can be
summarized as follows. For bl reg: as the fraction of new ?rms in a sector increases and/or
as the fraction of old ?rms facing high competition increases, the new ?rms in those sectors
are more likely to face high barriers in the form of business licensing regulations, but this is
less of a problem in Eastern Europe. This ?nding is consistent with the result previously
found in the literature on transition – competition promotes restructuring in Eastern
Europe but does not do that in CIS. On the other hand, as the fraction of old ?rms
facing competition primarily from private domestic SMEs increases, the new ?rms are
less likely to face high barriers if they are in the former Soviet Union, but more likely to
face high barriers in they are in the Eastern Europe or the Baltic states, which could be
11
We run regressions with regional dummy variables instead of country ?xed e?ects in order to be able
to introduce these interaction terms.
65
Dependent variable - TxReg
Variables I II
Const 0.5763 0.4857
[.127] [.250]
EE -0.1715** 0.4381
[.017] [.108]
Balt -0.4460*** -1.0584**
[.000] [.024]
SizeS 0.6865** 0.6789**
[.018] [.019]
SizeM 0.6529** 0.6489**
[.026] [.027]
sector1 0.1626 0.0774
[.182] [.534]
sector2 -0.2007** -0.2897***
[.051] [.007]
sector3 -0.0162 -0.1945
[.953] [.486]
sector4 0.0829 -0.0520
[.580] [.739]
Age -0.0006 0.0012
[.960] [.922]
New F 0.6019*** 0.3735
[.009] [.132]
Comp hiF 0.0688 0.2494
[.702] [.467]
EE ×Comp hiF -0.4410
[.290]
Bl ×Comp hiF 0.7430
[.355]
CompSME F 0.5815*** 1.1258***
[.005] [.002]
EE ×CompSME F -0.9234**
[.038]
Bl ×CompSME F 0.3622
[.751]
µ
3
0.6939*** 0.7000***
[.000] [.000]
µ
4
1.6059*** 1.6192***
[.000] [.000]
Nobs 1577 1577
LR(zero slopes) 44.79*** 62.84***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.16: Higher barriers (TxReg) faced by the new ?rms correspond to higher compe-
tition faced by the old ?rms (ordered probit regressions).
66
Dependent variable - AntComp
Variables I II
Const 0.6427* 0.8508**
[.085] [.041]
EE -0.2618*** -0.0732
[.000] [.787]
Balt -0.2862** -1.9367***
[.012] [.000]
SizeS -0.0218 -0.0555
[.939] [.845]
SizeM -0.0927 -0.1014
[.746] [.723]
sector1 -0.0949 -0.1269
[.425] [.296]
sector2 -0.2043** -0.1329
[.040] [.205]
sector3 0.0876 0.0213
[.759] [.941]
sector4 -0.0046 -0.0719
[.975] [.638]
Age -0.0074 -0.0055
[.546] [.653]
New F 0.2010 -0.0979
[.391] [.694]
Comp hiF 0.3445* 0.5449
[.060] [.107]
EE ×Comp hiF -0.6886*
[.098]
Bl ×Comp hiF 2.3058***
[.007]
CompSME F -0.0352 -0.6673*
[.876] [.063]
EE ×CompSME F 0.9765**
[.029]
Bl ×CompSME F -0.6798
[.590]
µ
3
0.6612*** 0.6675***
[.000] [.000]
µ
4
1.3338*** 1.3459***
[.000] [.000]
Nobs 1409 1409
LR(zero slopes) 35.97*** 56.99***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.17: AntComp barriers faced by the new ?rms and competition from new entry
faced by the old ?rms (ordered probit regressions).
67
Dependent variable - bl reg
Variables I II
Const -0.1042 -0.1402
[.787] [.742]
EE -0.3975*** -0.164
[.000] [.545]
Balt -0.3157*** -1.2835**
[.005] [.011]
SizeS -0.0074 -0.0242
[.980] [.936]
SizeM -0.2001 -0.2006
[.510] [.508]
sector1 0.1845 0.2130*
[.123] [.084]
sector2 -0.0023 0.0684
[.982] [.521]
sector3 0.1264 0.1007
[.657] [.729]
sector4 0.6381*** 0.6171***
[.000] [.000]
Age -0.0222** -0.0183
[.069] [.136]
New F 0.6624*** 0.5463**
[.004] [.026]
Comp hiF 0.1762 0.5918*
[.333] [.077]
EE ×Comp hiF -0.8721**
[.035]
Bl ×Comp hiF 0.0473
[.956]
CompSME F -0.0274 -0.8567**
[.893] [.012]
EE ×CompSME F 1.2171***
[.005]
Bl ×CompSME F 3.2439***
[.009]
µ
3
0.5861*** 0.5911***
[.000] [.000]
µ
4
1.3408*** 1.3530***
[.000] [.000]
Nobs 1540 1540
LR(zero slopes) 79.82*** 103.47***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.18: Higher barriers (bl reg) faced by the new ?rms correspond to higher compe-
tition from new entry faced by the old ?rms (ordered probit regressions).
68
Dependent variable - env reg
Variables I II
Const -0.3175 -0.2391
[.431] [.591]
EE -0.0175 -0.3391
[.814] [.226]
Balt -0.1128 0.1763
[.341] [.735]
SizeS 0.1063 0.1198
[.734] [.702]
SizeM 0.2105 0.2214
[.505] [.483]
sector1 -0.1204 -0.0892
[.322] [.473]
sector2 -0.3332*** -0.3098***
[.001] [.004]
sector3 -0.3086 -0.2317
[.311] [.452]
sector4 0.2130 0.2805*
[.153] [.071]
Age -0.0110 -0.0116
[.392] [.369]
New F 0.3822 0.4960**
[.109] [.050]
Comp hiF 0.1725 -0.0052
[.366] [.988]
EE ×Comp hiF 0.3487
[.419]
Bl ×Comp hiF -0.4946
[.572]
CompSME F 0.1746 0.0313
[.414] [.930]
EE ×CompSME F 0.2260
[.620]
Bl ×CompSME F 0.3269
[.785]
µ
3
0.6574*** 0.6581***
[.000] [.000]
µ
4
1.3932*** 1.3952***
[.000] [.000]
Nobs 1463 1463
LR(zero slopes) 31.77*** 34.55***
[.001] [.005]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.19: env reg barriers faced by the new ?rms and competition from new entry faced
by the old ?rms (ordered probit regressions).
69
Dependent variable - fir reg
Variables I II
Const -1.0319** -1.1246**
[.014] [.014]
EE -0.1415* -0.0691
[.052] [.802]
Balt -0.1399 0.0486
[.209] [.921]
SizeS 0.4033 0.3939
[.234] [.245]
SizeM 0.5176 0.5110
[.129] [.134]
sector1 0.0406 0.0606
[.734] [.619]
sector2 -0.3599*** -0.3314***
[.000] [.002]
sector3 -0.0666 -0.0506
[.816] [.862]
sector4 0.0773 0.0865
[.597] [.571]
Age -0.0209* -0.0208*
[.091] [.095]
New F 1.0533*** 1.0766***
[.000] [.000]
Comp hiF 0.4300** 0.6358*
[.020] [.064]
EE ×Comp hiF -0.2922
[.490]
Bl ×Comp hiF -0.3897
[.633]
CompSME F -0.0217 -0.2997
[.918] [.398]
EE ×CompSME F 0.4308
[.334]
Bl ×CompSME F 0.3466
[.760]
µ
3
0.7762*** 0.7765***
[.000] [.000]
µ
4
1.5262*** 1.5267***
[.000] [.000]
Nobs 1552 1552
LR(zero slopes) 48.61*** 49.69***
[.000] [.000]
p-values in brackets
* signif. at 10%, ** signif. at 5%, *** signif. at 1%
µ
3
and µ
4
are cuto? parameters in ordered probit regression
Table 3.20: fir reg barriers faced by the new ?rms and competition from new entry faced
by the old ?rms (ordered probit regressions).
70
due to reverse causality as explained above. Another possible explanation of this result
comes from the fact that the variable bl reg measures mainly the entry barriers, i.e. the
regulations designed to prevent the entry of new businesses. If a new ?rm has already
entered (and has already gone through the business registration and licensing) and is strong
enough to be a major competitor, creating business registration and licensing regulations
may not be helpful for the old ?rms at this point. The old ?rms may choose to lobby
for the creation of some other type of regulations that can drive out an already existing
new ?rm. Thus, an increase in compSME F may not lead to a higher level of business
registration and licensing regulations, but may lead to an increased level of other types of
regulations.
For TxReg: as the fraction of the new ?rms in a sector increases and/or as the frac-
tion of the old ?rms facing competition primarily from private domestic SMEs increases,
the new ?rms in those sectors are more likely to face high barriers in the form of taxes
and regulations, but this is less of a problem in Eastern Europe.
For AntComp: as the fraction of old ?rms facing high competition in a sector in-
creases, the new ?rms in that sector are more likely to face anti-competitive measures.
This is less of a problem in Eastern Europe, but more of a problem in Baltic states.
In addition, as the fraction of old ?rms facing competition primarily from domestic pri-
vate SMEs increases, the new ?rms in that sector are less likely to face anti-competitive
measures, except in Eastern Europe.
For env reg: as the fraction of new ?rms in a sector increases, the new ?rms in the
sector are more likely to face high barriers in the form of environmental regulations. High
competition and competition from private domestic SMEs seems to have no e?ect.
For fir reg: as the fraction of new ?rms in a sector increases and/or as the fraction
of old ?rms facing high competition increases, the new ?rms in those sectors are more
likely to face high barriers in the form of ?re/safety regulations.
3.5 Conclusions
Overall, our empirical results show that a higher threat from new entry leads to a higher
level of barriers faced by the new ?rms (as measured by TxReg, AntComp, bl reg, env reg,
and fir reg), and simultaneously, higher levels of barriers are positively related to at least
71
one measure of excess employment at the old ?rms. The e?ect is especially clear for
business licensing regulations measure, which is positively related to all four measures of
excess employment. Our ?ndings are consistent with the idea that a higher threat from
new entry creates incentives for the old ?rms to lobby with politicians for the creation
of entry barriers in exchange for excess employment, thus resulting simultaneously in the
high level of barriers faced by the new ?rms and the presence of excess employment at the
old ?rms. Our results also suggest that the e?ect of the threat from new entry on the level
of barriers may vary be region, possibly due to the institutional and cultural di?erences
that make it more di?cult for old ?rms to lobby with politicians in some countries. These
results are consistent with the empirical ?ndings previously established in the literature
on privatization and restructuring in transition.
72
Conclusions
In this paper we have argued that allowing creation and entry of the new private businesses
at the start of transition reforms had an important impact on the incentives of the existing
?rms to restructure. Privatization coinciding with opening up of new entry may create
incentives for the managers of the existing (state-owned or privatized) ?rms to lobby with
politicians for the restrictions on new entry resulting in slow restructuring of the old ?rms,
high level of administrative barriers to entry, and slow development of the new private
sector. This does not mean, however, that the new entry should not be allowed. On
the contrary, those new ?rms that enter despite the high entry barriers are more e?cient
than old ?rms and their entry increases overall e?ciency of the economy. This model
calls our attention to the fact that simply allowing the entry of new businesses will not
necessarily result in substantial new entry and higher competition for the old ?rms. In
order to ensure that new entry will create competition and promote restructuring at the
old ?rms, policymakers should pay attention to the complimentary institutional reforms
designed to prevent collusion between politicians and managers of the existing old ?rms.
Certain reforms can be undertaken to reduce the probability of managers of the old
?rms successfully lobbying with politicians. Such lobbying is less likely to succeed if the
excess labor is costly for the old ?rms (e.g. barter payments are eliminated), if the barriers
are not e?ective in preventing new entry or the tax collections are very sensitive to the
level of barriers (e.g. because of uno?cial economy), if the politician is more accountable
to the treasury for tax collections (e.g. if local budgets are ?nanced entirely through
local tax collections), or if the politician’s bene?ts from excess employment are low (e.g.
because of more transparent election process and less corruption).
The empirical evidence, based on data from the World Business Environment Survey
by World Bank, is consistent with the idea that competitive threat from the new entrants
may lead the old ?rms to lobby for the creation of entry barriers. Speci?cally, we see that
the new ?rms in sectors with a large proportion of old ?rms facing high competition or
competition primarily from the private domestic SMEs, or in sectors with a large number
of new ?rms relative to the old ?rms are more likely to encounter high administrative and
regulatory barriers. Since higher level of barriers faced by the new ?rms corresponds to the
presence of excess employment at the old ?rms, especially for the business registration and
licensing barriers, our ?ndings are consistent with the model where the old ?rms provide
73
excess employment in exchange for the desired level of barriers set by the politician.
Our results also suggest that the e?ect of the threat from new entry on the level of
barriers may vary be region, possibly due to the institutional and cultural di?erences that
make it more di?cult for old ?rms to lobby with politicians in certain countries.
74
Appendix A
Indi?erence Curves for the Case of ? < 1.
A.1 Objective Functions and General Setup
In chapter 2 we assumed that ? = 1, that is, old ?rms are entirely private. ? = 1 also
implies that the Treasury does not receive any revenue from the old ?rms (does not receive
any tax collections). This is unrealistic and thus we would like to relax this assumption
on the value of ?.
Here I will derive the indi?erence curves for the manager and the politician for the
case when ? is allowed to vary between 0 and 1 and the politician derives bene?ts from
excess employment only. Recall that in this case utility functions of the manager and the
politician are
1
:
U
m
= ?(?(R) ?wL) (A.1)
U
p
= B(L) +m{(1 ??)(?(R) ?wL) +N(R)} (A.2)
Indi?erence curves have the following slopes (see equations (2.6) and (2.8)).
The manager’s indi?erence curve:
dR
dL
=
w
?
?
(R)
> 0 (A.3)
The politician’s indi?erence curve:
dR
dL
= ?
B
?
(L) ?mw(1 ??)
m[(1 ??)?
?
(R) +N
?
(R)]
(A.4)
Note that the slope of the manager’s indi?erence curve does not depend on ?. Thus, as ?
changes between 0 and 1, manager’s preferred combination of entry barriers R and excess
employment L does not change. This also means that the manager’s threat point stays at
L
?
= 0 (see page 17 for derivation of the threat points for the manager and the politician).
Of course, even though the slope of the manager’s indi?erence curve is not a?ected by
changes in ?, the value of manager’s utility will be a?ected (e.g. as ? ?, if R and L stay
unchanged, U
m
will increase). The situation is di?erent for the politician. Changes in ?
would lead to changes in the value of U
p
(if R and L are kept constant) and to changes in
the slope of politician’s indi?erence curve.
1
these are equations (2.1) and (2.2) from chapter 2
75
Let’s see how the politician’s indi?erence curve looks like for ? ? (0, 1). Recall that
we made certain assumptions on the functional forms of functions B(·), ?(·), and N(·)
(see page 14 and ?gures 2.1, 2.2, and 2.3). Under these assumptions (B
?
(·) > 0, ?
?
(·) > 0,
and N
?
(·) < 0) we cannot put a de?nite sign on the slope of the politician’s indi?erence
curve in equation (A.4).
However, using our assumptions on functional forms we can make the following
arguments. For very low values of L, B
?
(L) is positive and very large, therefore for very
low values of L we will have B
?
(L) > mw(1??) and thus the numerator in (A.4) is positive
for low values of L. As L ?, B
?
(L) ? and eventually we will get B
?
(L) < mw(1 ? ?). We
can summarize this argument in the following proposition.
Proposition 1 For any given values of m, w, and ?, such that mw(1 ? ?) = 0, we can
?nd such value of L > 0, denote it by L
??
, that B
?
(L
??
) = mw(1 ??) and
?L < L
??
we have B
?
(L) > mw(1 ??)
?L > L
??
we have B
?
(L) < mw(1 ??)
Note that, if the values of m, w, and ? are such that mw(1 ??) = 0, then B
?
(L) >
mw(1 ??) for all values of L, that is, L
??
= ? (recall our assumption that B
?
(L) ? 0 as
L ? ?). This was the case when we assumed the value of ? = 1.
Similarly, for very low levels of R, ?
?
(R) is large and positive, while N
?
(R) is small
in absolute value and negative. Thus, for low levels of R we have |(1??)?
?
(R)| > |N
?
(R)|
and the denominator in equation (A.4) is positive (see also ?gure 2.4). As R ?, ?
?
(R) ? and
|N
?
(R)| ?, and thus eventually we will have |(1??)?
?
(R)| < |N
?
(R)| and the denominator
of (A.4) will be negative. We can summarize this argument in the following proposition.
Proposition 2 For any given value of ? ? (0, 1), we can ?nd such value of R > 0, denote
it by R
??
, that (1 ??)?
?
(R
??
) = ?N
?
(R
??
) and
?R < R
??
we have (1 ??)?
?
(R) +N
?
(R) > 0
?R > R
??
we have (1 ??)?
?
(R) +N
?
(R) < 0
76
R
L
R
**
L
**
+
?
+
?
Figure A.1: Slope of the politician’s in-
di?erence curve for case ? < 1.
R
L
R
**
L
**
IC
p
Figure A.2: Politician’s indi?erence curve
for case ? < 1.
Note that at R = R
??
the denominator of (A.4) is equal to zero and thus
dR
dL
is
unde?ned at that point. Also, notice that the condition that de?nes R
??
in proposition 2
is identical to the ?rst order condition in the politician’s optimization problem (equation
(2.4)), where politician chooses R for a given level of L. Thus, our R
??
is identical to R
?
,
which is the politician’s threat point for ? < 1.
Combining signs for the numerator and the denominator of the expression in (A.4)
(and not forgetting the overall minus sign), we can determine the slope of the politician’s
indi?erence curve depending on the values of R and L (see ?gure A.1).
Further, note that R
??
is the threat point of the politician. That is, for any level
of excess employment L chosen by the manager, the politician would choose the level of
barriers R
??
in the absence of bargaining. If bargaining does occur the politician may
choose a di?erent level of barriers. However, manager of the old ?rm will never engage
in bargaining trying to set a lower level of entry barriers (given the utility function of
the manager). Thus, the level of entry barriers will never be below R
??
. Therefore, the
relevant space for the politician’s indi?erence curve is values of R > R
??
. Hence, IC
p
should have the shape depicted on ?gure A.2 (here we show the indi?erence curve that
goes through the point corresponding to the uncooperative equilibrium of R = R
??
and
L = 0).
This shape can also be derived mathematically. Along the politician’s indi?erence
77
curve
d
2
R
dL
2
=
d
dL
_
?
B
??
(L) ?mw(1 ??)
m(1 ??)?
?
(R) +mN
?
(R)
_
= (A.5)
=
?1
{m[(1 ??)?
?
(R) +N
?
(R)]}
3
×
×
_
B
??
(L)[m(1 ??)?
?
(R) +mN
?
(R)]
2
+
+
_
B
?
(L) ?mw(1 ??)
¸
2
_
(1 ??)?
??
(R) +N
??
(R)
¸
_
The long term in curly braces is negative, given our assumptions about functions
B(·), ?(·), and N(·). Thus, sign of
d
2
R
dL
2
= sign of [(1 ? ?)?
?
(R) + N
?
(R)]. For R > R
??
we have (1 ? ?)?
?
(R) + N
?
(R) < 0 (see proposition 2) and thus, along the politician’s
indi?erence curve,
d
2
R
dL
2
< 0 for R > R
??
,
which corresponds to the shape of IC
p
depicted in the ?gure A.2. We can also show the
indi?erence curves for politician and manager together, which is done on ?gure A.3. Here,
again, I have shown the indi?erence curves going through the point corresponding to the
uncooperative equilibrium of R = R
??
and L = 0.
R
L
R
**
L
**
IC
p
IC
m
0
Figure A.3: Indi?erence curves of the manager and the politician for the case of ? < 1.
A.2 E?ect of ? ? on IC
m
and IC
p
As mentioned above, changes in the value of ? do not a?ect the manager’s indi?erence
curve, they only a?ect the manager’s level of utility, U
m
.
78
A change in ? a?ects the politician’s indi?erence curve in three ways:
• it a?ects the slope of the indi?erence curve,
dR
dL
given in equation (A.4), for any given
value of R and L
• it a?ects the value of L
??
, as de?ned in proposition 1
• it a?ects the value of R
??
, as de?ned in proposition 2
We consider these e?ects in turn.
1. Slope – as ? increases, if R and L are kept constant,
dR
dL
will also increase. Thus the
slope of IC
p
becomes steeper on the upward sloping part (L < L
??
), may change
from downward to upward sloping for some values of L around L
??
, and becomes
?atter (less negative) on the downward sloping part.
2. Value of L
??
– L
??
is de?ned by B
?
(L
??
) = mw(1 ??) (see proposition 1). Thus, as
? increases, L
??
increases as well.
3. Value of R
??
– R
??
is de?ned by (1 ? ?)?
?
(R
??
) +N
?
(R
??
) = 0 (see proposition 2).
Thus, as ? increases, R
??
will decrease.
This means that when ? increases, new IC
p
will cross the old IC
p
once and only
once. However, it is impossible to tell (without knowing the functional form for the
indi?erence curve) where they would cross: on the upward sloping part before old L
??
,
between old L
??
and new L
??
, or on the downward sloping part after new L
??
. Two of
these cases are depicted on ?gures A.4 and A.5.
A.3 E?ect of ? ? on equilibrium values on R and L
Now that we know how change in ? a?ects the indi?erence curves of the manager and the
politician, we can analyze the e?ects of an increase in ? on the equilibrium level of entry
barriers and excess employment.
Because the value of ? a?ects the politician’s threat point R
??
, an increase in ?
would lead to a lower equilibrium level of barriers even in the absence of bargaining.
That is, as the manager’s share of the ?rm’s pro?t increases (?rm moves closer to being
79
R
L
R
**
0
L
**
0
IC
0
p
0
R
**
1
?
1
>?
0
L
**
1
IC
1
p
Figure A.4: Shift in the politician’s indif-
ference curve when ? ?; ?
1
> ?
0
.
R
L
R
**
0
L
**
0
IC
0
p
0
R
**
1
?
1
>?
0
L
**
1
IC
1
p
Figure A.5: Shift in the politician’s indif-
ference curve when ? ?; ?
1
> ?
0
(di?er-
ent case).
completely private), in the absence of bargaining the politician will choose lower level of
entry barriers.
If bargaining does take place, then with an increase in ? IC
p
becomes steeper, while
slope of IC
m
does not change. However, since R
??
decreases, the point of the uncooperative
equilibrium (R = R
??
and L = 0) also shifts down, forcing IC
m
to shift down without
changing its slope. As a result, if the politician has higher bargaining power and the
equilibrium allocation is at the point on the contract curve close to IC
m
, an increase in ?
is likely to produce a (large) increase in excess employment. The e?ect on the equilibrium
level of entry barriers is ambiguous.
80
Appendix B
Characterization of the Contract Curve.
Here I will consider the contract curve for the case when the politician derives bene?ts
from excess employment only. The objective is to derive the shape of the contract curve
(its slope) and to characterize how the contract curve shifts when the indi?erence curves
of the manager and/or politician shift due to change in the model parameters.
Along the contract curve we have slope of IC
m
= slope of IC
p
. Using the expres-
sions for slopes of the indi?erence curves (see equations (2.6) and (2.8)), we get:
w
?
?
(R)
= ?
B
?
(L) ?mw(1 ??)
m[(1 ??)?
?
(R) +N
?
(R)]
(B.1)
Multiplying through, we get:
wm(1 ??)?
?
(R) +wmN
?
(R) = ??
?
(R)B
?
(L) +wm(1 ??)?
?
(R) (B.2)
Note that the terms containing ? cancel out and the resulting expression, which implicitly
de?nes R as a function of L along the contract curve, is:
wm(?N
?
(R)) = ?
?
(R)B
?
(L) (B.3)
Totally di?erentiating expression (B.3) we obtain the expression for the slope of the
contract curve:
dR
dL
¸
¸
¸
¸
cc
=
B
??
(L)?
?
(R)
wm(?N
??
(R)) ?B
?
(L)?
??
(R)
(B.4)
Given our assumptions on the shape of B(·), ?(·), and N(·) functions, we can show that
dR
dL
¸
¸
¸
¸
cc
< 0.
Thus, the contract curve is downward sloping, which should not be a surprise.
Can we ?nd
d
2
R
dL
2
¸
¸
¸
cc
? The expression for the second derivative along the contract
curve is:
dR
dL
¸
¸
¸
¸
cc
=
1
[wm(?N
??
(R)) ?B
?
(L)?
??
(R)]
2
× (B.5)
×
__
B
???
(L)?
?
(R) +B
??
(L)?
??
(R)
dR
dL
_
_
wm(?N
??
(R)) ?B
?
(L)?
??
(R)
¸
?(B.6)
? B
??
(L)?
?
(R)
_
wm(?N
???
(R))
dR
dL
?B
??
(L)?
??
(R) ?B
?
(L)?
???
(R)
dR
dL
__
(B.7)
81
R
L
CC
Figure B.1: Contract curve assuming
B
???
(L) < 0, ?
???
(R) > 0, ?N
???
(L) < 0.
R
L
CC
Figure B.2: Contract curve in the general
case.
In order to be able to evaluate the sign of
d
2
R
dL
2
¸
¸
¸
cc
we need to know the signs of
B
???
(·), ?
???
(·), and N
???
(·). I am reluctant to make any assumptions about the signs of
third derivatives, especially because I am not sure what the interpretation of the third
derivative is in this case. I can only point out that the sight of
d
2
R
dL
2
¸
¸
¸
cc
will be unambiguous
only if we assume the following:
B
???
(L) < 0
?
???
(R) > 0
?N
???
(L) < 0
Under these assumptions
d
2
R
dL
2
¸
¸
¸
cc
< 0 and the contract curve looks like the one pictured on
?gure B.1.
If we do not assume the above signs for the third derivatives, then the sign of
d
2
R
dL
2
¸
¸
¸
cc
is ambiguous and the contract curve does not have a uniform second derivative and
probably looks like the one pictured on ?gure B.2, i.e. it is downward sloping, but the
second derivative changes its value.
We now consider the shifts in the contract curve in response to changes in the model
parameters
1
.
1
When depicting shift of the contract curve on the ?gures, I show contract curve as a downward sloping
straight line for simplicity.
82
CC
CC’
R
L
Figure B.3: Shift in the contract curve when w ?.
1. w ? –
¸
¸
dR
dL
¸
¸
?, contract curve shifts down and becomes ?atter (?gure B.3)
2. ?
?
(R) ? – assuming that ?
??
(·) does not change,
¸
¸
dR
dL
¸
¸
?, contract curves shifts up and
becomes steeper (?gure B.4)
3. B
?
(L) ? – assuming that B
??
(·) does not change,
¸
¸
dR
dL
¸
¸
?, contract curves shifts up
and becomes ?atter (?gure B.5)
4. |N
?
(R)| ? – assuming that N
??
(·) does not change,
¸
¸
dR
dL
¸
¸
does not change, contract
curves shifts down and does not change its slope unless N
??
(·) changes (?gure B.6)
5. m ? –
¸
¸
dR
dL
¸
¸
?, contract curves shifts down and becomes ?atter (?gure B.7)
6. ? ? – since the expression that implicitly de?nes the contract curve (equation (B.3))
does not depend on ?, changes in ? do not a?ect either the slope or the position of
the contract curve.
83
CC
CC’
R
L
Figure B.4: Shift in the contract curve
when ?
?
(·) ? (assuming ?
??
(·) does not
change).
CC
CC’
R
L
Figure B.5: Shift in the contract curve
when B
?
(·) ? (assuming B
??
(·) does not
change).
CC
CC’
R
L
Figure B.6: Shift in the contract curve
when |N
?
(·)| ? (assuming N
??
(·) does not
change).
CC
CC’
R
L
Figure B.7: Shift in the contract curve
when m ?.
84
BIBLIOGRAPHY
[1] Barberis, Nicholas; Maxim Boycko, Andrei Shleifer, Natalia Tsukanova, (1996) ”How
Does Privatization Work? Evidence from the Russian Shops.” Journal of Political
Economy, 104(4).
[2] Blanchard, O., and Michael Kremer, (November 1997) ”Disorganization.” Quarterly
Journal of Economics, 112(4).
[3] Boycko, M., Andrei Shleifer, and Robert Vishny, (1996) ”A Theory of Privatization,”
Economic Journal, 106.
[4] Carlin, Wendy, Steven Fries, Mark Scha?er, Paul Seabright, (2001). ”Competition
and Enterprise Performance in Transition Economies: Evidence from a Cross-country
Survey”, Working paper, Department of Economics, University College, London.
[5] De Melo, Martha, Cevdet Denizer, and Stoyan Tenev (2001), “Circumstance and
Choice: the Role of Initial Conditions and Policies in Transition Economies,” World
Bank Economic Review, 15(1).
[6] Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer,
(2002). ”The Regulation of Entry,” Quarterly Journal of Economics, 117(1).
[7] Djankov, Simeon and Peter Murrell, (2002). ”Enterprise Restructuring in Transition:
A Quantitative Survey”, Journal of Economic Literature, vol. XL.
[8] Frydman, Roman, Cheryl Gray, Marek Hessel, Andrzej Rapaczynski, (1999). ”When
Does Privatization Work? The Impact of Private Ownership on Corporate Perfor-
mance in the Transition Economies.” Quarterly Journal of Economics, 114(4)
[9] Hashi, Iraj and Jan Mladek, (2000). “Fiscal and Regulatory Impediments to the Entry
of New Firms in Five Transition Economies”, Journal of East-West Business, 6(2).
[10] Hellman, Joel S., (January 1998). ”Winners Take All: The Politics of Partial Reform
in Postcommunist Transition.” World Politics, 50.
[11] Hellman, Joel S., Geraint Jones, and Daniel Kaufmann, (September 2000). “Seize the
State, Seize the Day: State Capture, Corruption and In?uence in Transition”, World
Bank Policy Research Paper, no. 2444.
85
[12] Holmstrom, Bengt, (1979). ”Moral Hazard and Observability.” Bell Journal of Eco-
nomics, 10(1).
[13] Karpinski, Jakub, (1996). “Sociologists Compare Nomenklatura Members in Contem-
porary Elites”, Transition, World Bank, May 1996.
[14] Kaufman, Daniel, (1997). ”Why is Ukraine’s Economy – and Russia’s – Not Grow-
ing?”, Transition, World Bank, April 1997.
[15] Kaufmann, Daniel, and Aleksander Kaliberda, (1997). ”Integrating the Uno?cial
Economy into the Dynamics of Post-Socialist Economies”. In B. Kaminski, ed. Eco-
nomic Transition in Russia and the New States of Eurasia. Armonk: M. E. Sharpe.
[16] Rosen, Sherwin, (1992). ”Contracts and the Market for Executives”, in Contract
Economics, edited by Lars Werin and Hans Wijkander, Oxford University Press,
Oxford, England.
[17] Sachs, Je?rey, Cli?ord Zinnes, Yair Eilat, (2000). ”The gains from Privatization in
Transition Economies: Is ”Change of Ownership” Enough?”, CAER II Discussion
Paper 63.
[18] Shleifer, Andrei, (May 1997). ”Government in Transition”, European Economic Re-
view, 41, nos. 3-5.
[19] Shleifer, Andrei and Daniel Treisman, (2000). Without a Map: Political Tactics and
Economic Reform in Russia, MIT Press, Cambridge, Massachusetts.
[20] Shleifer, Andrei, and Robert Vishny, (November 1994). ”Politicians and Firms.”
Quarterly Journal of Economics, 109(4).
[21] Szelenyi, Ivan, Daniel Treisman, and Edmund Wnuk-Lipinski, eds. (1995). Elites in
Poland, Russia, and Hungary: Change or Reproduction?, Warsaw, Polish Academy
of Sciences.
86
doc_379464092.pdf