POLICY RESEARCH WORKING           PAPER         28 84
Privatization and Labor Force Restructuring
around the World
Alberto C bong
Florencio Lopez-de-Silanes
The World Bank
Development Research Group
Public Service
Septcmber 2002



| Poi icr  RESEA1RCH \VWOkKIN(; PAiPHl 2884
Abstract
SoImle critics of privatization argue that poor labor force      owners. Compulsorv skill-bascd programs are the onlI
restructuring is a key concer-ni and th;at goverinimlelits      type of program that is miarginally associated with higher
should establishi better retrenichliimenit programs. Using       prices and lower rchirinig rates after privatization, but the
ncw data from a sample of 400 companies in the world,           political and economilic costs of this policy mav' make it
Chong anid l-6pez-de-Silanes test competinig thcories            somewhat impractical. While a qualified non-
about the wisdom of retrencilhimenlt programs and their         intervention policy appears to be the safest bet in labor
cffect on prices paid by buyers, and rehilrinlg policies hb     retrcichimlenit before privatization, another one might be
private owners after- privatizatiotn. Thc restilts show that    to set up a social safety' net or labor reallocation program
adverse sccctioni plagues retrenchlimlenit programs carried    hefore privatization, and theln let the new private owners
Out by governments before privatization. Controlling for        decidle who is redulindant and who is not. Setting up the
cndogeneitv, several labor retrnchiclimienit policies yield a    program before privatization may hlelp with the political
negative imiipact oln net privatization priccs. In              viability of tllc process and letting the newv owyners
confirimiationi of the adverse selectioni argUmient, various     manage the retrenchmienit may help avoid adverse
types of voluIntary downsizing lead to a highier frequelicy     selection.
of rehil-inig of the same wvorkcrs by the new private
I his paper-a prodtict of PulliC Services, Developmcnit Research GrouIp-is part of a larger effort in the group tO
understand the labor implicationis of public sector reform. The study was funded by thie Bank's Research Support BLudget
ulider the rescarch project "Ptublic Sector Downsizing" (Rl'O 683-69). Copics of thc paper are available free from the World
13ank, 18 18 H Street NW, Washinigtoni DCl 20433. Please contact Hedy Sladovicil, room MC3-3 11, tclephonie 202-473-
7698, fax 202-522- 1154, email address lsladovichi(L/gxvorldbanik.org. Policy Research Working Papers are also posted on
th'e  Web    at http ://ccon .worldhank. org. The        autlors    m11av  be  contacted    at albertoch( )iadb. org      or
florcicio.lopczdcsilanesC(iyale.eduL. September 2002. (40 pages)
Tse PolicY Research \W\rkinig Paper SCries dissenjinates the findlings / iVork/ in prOgress t(r encouraIge the exchange ofideas abouit
d/eve4/oipmienu issues. .  h/etc of he series is to get e findindgs (nit qiickIv, cv cin if thb presentations are less tban full- -polished. The
papers Larm- the iiini5es if th7e alIt/iors and shoulidh be cit eaccordingly. The findoigs, initerpretationis, and coniclusions expressed in this
paper ire eiitirely those o,l t(li aut/ors. The)y do iio it necesssrily represeiit the vieu' of the \Worldk Bank. its Executive Directors. or the
c  iiiit?ie s th'eY ri'/pr5'IisWt.
Plroduced by the Research Advisory Staff



Privatization and Labor Force Restructuring
around the World
Alberto Chong
Inter-American Development Bank
and Georgetown University Public Policy Institute
Albertoch(Iadb.Org
Florencio Lopez -de-Silanes
Yale University School of Management
and National Bureau of Economic Research
florencio.lopezdesilanes@yale.edu
We are grateful to Alexander Dyck, Martin Rama, Jaime Saavedra, and seminar
participants at a conference at Fondazione Eni Enrico Mattei, and the Latin American and
Caribbean Economic Association Meeting for comments and suggestions. Virgilio Galdo
and Alice Hong provided research assistance and Madison Boeker provided editing. The
Research Committee of the World Bank supported this project through grant RPO 683-69.
The Yale School of Management provided support through its International Institute of
Corporate Governance.






1.    Introduction
In the last two decades many countries have embarked on major privatization programs,
but there are still several countries that have been reluctant to privatize. This is
particularly true in developing countries, as reflected by the fact that state-owned
enterprises in these countries still account for more than 10 percent of gross domestic
product, 20 percent of investment, and about 5 percent of formal employment (Kikeri
1999). While there is growing evidence of the benefits of privatization (for example
Megginson and Netter 2001, La Porta and L6pez-de-Silanes 1999), the unwillingness to
privatize appears to be associated with a rather negative perception of the privatization
process with respect to the labor force. Critics argue that poor labor force restructuring is a
key concern and that governments should establish better retrenchment programs.
On the other hand, opponents of the government's restructuring of to-be-privatized
firns argue that public unions can influence the future of politicians, effectively reducing
the government's bargaining power (Freeman 1986; L6pez-de-Silanes, Shleifer and
Vishny 1997). It has also been argued that it is not worth spending resources in
restructuring the labor force before privatization, as governments may not be able to
distinguish the particular workers that should be retained (Haltiwanger and Singh 1999,
Rama 1999). Govermnents that administer human resources risk retrenching the wrong,
more productive personnel. This may result in the loss of know-how that, at a minimum,
may help solve short-run post-privatization efficiency problems and, at worst may be
linked with permanent damage to the productive structure of the firm. Dismissing the
workers that the new owners would rather keep may not add value to the firm and
consequently may reduce privatization prices. This is particularly true in developing
countries where available information is even more lacking (Rama 1999).
Although labor restructuring is one of the most difficult and sensitive issues in
privatization, the empirical literature available is quite scarce as a result of the lack of data
to address these issues. Our paper contributes to this literature in three ways: by creating a
new cross-country database with detailed information about labor retrenchment policies
before privatization and labor rehiring efforts after the firms go into private hands; by
documenting the effects of different labor restructuring policies on the net privatization
prices paid by buyers; and by analyzing worker rehiring in privatized firms that were
subjected to various types of retrenchment programs.
The paper pays particular attention to the quality of various targeting processes. In
fact, not all targeting is created equal. From an empirical perspective, it may well be the
case that a state-owned enterprise that uses targeting when restructuring labor prior to
privatization may be able to fetch a higher privatization price by applying a targeting
mechanism that may prove very costly afterwards. An example may be the case of age-
biased retrenchment, where retrenchment is targeted on older and typically more
expensive workers. Although privatization prices may be thought to increase as a result of
this policy, as the future operation costs of the firm are expected to improve, the net
1



impact on future efficiency is unclear as some of these workers may also have been the
more productive, experienced or better trained.
Unlike the previous literature, which focused on the effects of restructuring on
prices for the case of Mexico (L6pez-de-Silanes 1997), the nature of our data allows us to
look at two complementary slices of the picture before and after privatization across
countries. We are able to analyze the impact of a large set of labor policies before
privatization, and observe the private firms' reaction in terms of rehiring of the previously
fired workers. Our analysis uses these two types of results to provide an evaluation of the
price effect of labor restructuring policies and the managerial quality of the downsizing
efforts carried out by the governrment before privatization.
The database was constructed by randomly selecting 400 firms privatized between
1982-2000 around the world. We obtained pre- and post-privatization data by sending a
detailed questionnaire to the CEO of the privatized firm and accessing privatization files.
We followed up with each of the firms and corroborated their answers with several public
sources and data for these firms coming from international financial agencies and
privatization ministries. The result is a comprehensive cross-country (see Appendix I for a
list of countries) database with firm characteristics, detailed labor restructuring policies
before privatization, and labor rehiring policies after the firm entered the private sphere.'
The basic thrust of our results is that adverse selection plagues retrenchment
programs carried out by governments before privatization. Controlling for endogeneity,
2
labor downsizing does not do much in terms of net privatization prices. This finding may
appear somewhat counterintuitive, as, according to the conventional wisdom, sellers will
always want the government to downsize prior to privatization. However, this is fairly
consistent with the political view on prior restructuring before privatization (L6pez-de-
Silanes 1997). Results also show that after controlling for endogeneity, pay cuts do not
increase prices while employment guarantee programs forced on the buyer do carry a
significant discount in prices.
To further study the results in the area of downsizing, we focus on the nature of the
retrenchment process prior to privatization and its impact on rehiring. Confirming the
adverse selection argument, several kinds of voluntary downsizing lead to a higher
frequency of rehiring of the same workers by the new private owners.
The only exception is compulsory skill-biased programs as they are marginally
associated with higher prices and lower rehiring rates after privatization in some
1. An additional benefit of this data is that it allows us to consider the issue of failed privatizations on
prices, rarely treated in the literature. Research on privatization imnplicitly sets aside failed privatizations
since data is typically available only for companies that actually attract bidders. This potentially
underestimates the imnpact of labor changes for these companies, which are likely the ones that went through
more thorough restructuring efforts prior to privatization. Methodologically, this issue is addressed by using
a simple non-linear procedure for the entire, truncated, distribution of privatization prices.
2. As defined by the amount that accrues to the government after all costs are taken into account,
adjusted by shares sold and divided by average sales during the three years prior to privatization.
2



specifications. One could argue that the nature of the program itself, typically based on
written exarns or panel reviews, may partly explain these results. Unfortunately, this
policy is one of the most politically difficult to implement by the government and requires
a tough stance from the authority. Results show that the managerial quality of the
government may have an impact in the results, but as a general policy it appears that not
much time and effort should be spent on labor restructuring before privatization.
The paper is organized as follows. The next section describes the data collection
process and discusses the empirical methodology. In section 3, we test whether labor
restructuring has an impact on privatization prices and whether such results hold when
failed privatization and or potential endogeneity are taken into account. Section 4 extends
our results on prices for specific types of downsizing measures. Section 5 provides a new
look at the effect on prior downsizing by the government by analyzing the nature of post-
privatization re-hires by private owners. Section 6 concludes.
2.     Data and Methodology
Our sample was formed based on a list compiled by the authors of about 1,500
privatizations around the world covering the period 1982-2000.3 The two main sources for
this list are the World Bank Privatization database and Privatisation International, which
together arguably provide the largest source of privatization transactions in the world.
From this original list, we selected a random sample of 400 big and small firms to whom
we sent a detailed questionnaire designed by the authors. The questionnaire was addressed
to the CEO with a recommendation to direct it to the chief financial officer and the
director of human resources of the firm.4
In order to ensure the quality of our data, we employed four additional sources.
First, we took advantage of the fact that in several developing countries many
privatizations have been performed as part of structural adjustment or other lending
programs with the support of the World Bank. We were able to access a wide range of
World Bank's internal documents to double check and in some instances, complement the
information collected in our survey. In particular, we made extensive use of the World
Bank's electronic Intranet system called ImageBank, which allows full access to such
documents.5 Second, we also made broad use of NEXIS to search for a number of national
3. We excluded voucher privatizations. As it has been discussed elsewhere, there are fundamental
differences between such privatization technique and others, which would have made comparisons
particularly difficult (Boycko, Shleifer, and Vishny, 1994).
4. While directors of human resources answered 71.4 percent of the labor part of the questionnaires,
public relations managers answered 16.6 percent of them. Personnel working at human resources
departments other than managers answered about 5.5 percent, and personnel working at public relations
answered 6.5 percent, usually on behalf of managers. Financial information was typically provided by the
office of the chief financial officer or, in the case of small fimns, by the office of the CEO.
5. Typical World Bank documents include Country Economic Reports, Staff Appraisal Reports,
President's Reports, Supervision documents, Project Completion Reports, Audit Reports, Operation
Evaluation Studies, and Sector Reports. We covered about 63.4 percent of the sample.
3



and international publications.6 Third, whenever possible we interviewed officials from
the World Bank, the International Monetary Fund, and the Inter-American Development
Bank who were directly associated with the privatization programs in different countries.7
Finally, when necessary, we directly contacted the privatization offices or corresponding
ministries (e.g., finance, industry) of each country to request specific pre-privatization
information missing.8 Whenever we found discrepancies we contacted again the national
privatization agencies and the firms themselves to clarify the issues.9
We organized the questionnaire in four areas. The first area covered, pre-
privatization firm characteristics, and asked about sales, assets, profits, liabilities,
management changes, and sector of origin. The second area covered pre-privatization
labor characteristics and policies, and asked about number of blue and white collar
workers, presence and incidence of unions, number of strikes, political affiliation of
unions, labor restructuring measures and targets. The third area focused on the
privatization process, and in particular, on privatization prices, transaction methods used,
shares sold, and foreign participation. Finally, the fourth area included post-privatization
labor re-hiring policies. We tried to get both dummy variables and exact number of
workers for every possible category. However, we were not able to achieve this, as most
respondents did not provide enough numerical information about the workers involved in
various retrenchment and rehiring programs, so we settled for dummy variables that tell us
if the policies were undertaken or not.'I Table 1 provides definitions of the specific
variables that we collected.
Table 2 shows the results of our efforts to gather data. Out of the 400 cases
targeted we ended up collecting data for 308 privatizations comprising 84 countries for the
period 1982-2000. Of those, 16 are failed privatizations or operations in which preparation
for privatization occurred but in which the sale ended up not materializing. As the table
indicates, the complete information for our 308 firms accounts for 97.21 percent of total
sales. On the other hand, 25 companies comprising about one percent of total sales
supplied quite incomplete information that could not be further completed and thus, ended
up not being used. Additionally, 26 firms did not respond to our requests, denied the
existence of information, or simply refused to provide it. Twenty-two firms, accounting
for 0.78 percent of total sales, could not be included since they were liquidated and no
longer exist. Finally, nineteen firms could not be included in our sample as they merged
6. To do this, we mainly employed the Wall Street Journal, the Financial Times, Oxford Analytica, and
the Economist Intelligence Unit. We covered around 70.1 percent of the sample.
7. We also specifically selected firms totaling about 15 percent of the total sample and double or triple-
check most of the information.
8. Using this approach we covered 73.2 percent of the sample.
9. We found most discrepancies in developing countries, in particular Africa and the Middle East.
Whenever data led to significant discrepancies that could not be reconciled the firms were eliminated from
the sample. This occurred in 6 cases in total, 4 from Africa, 1 from the Middle East, 1 from Asia. These
firms were classified under the category firms that supplied incomplete information in Table 2.
4



and no longer keep separate accounting and financial statements. The pattern in our
resulting sample in terms of both, region, year of privatization, and sector fits closely with
the compiled list of privatizations of Privatisation International and the World Bank,
particularly, when excluding voucher privatizations. This leads us to reasonably conclude
that our sample is unbiased.
Figure 1 shows the distribution of our sample by region of the world with respect
to the pattern found in the original privatization lists. Whereas 33 percent of the
privatizations in our sample are from Latin America, 8 percent from Asia, 21 percent from
Africa and the Middle East, 25 percent from developed countries, and 13 percent from
Transition Economies, the corresponding percentages for the original list are 30, 10, 20,
23, and 18 percent, respectively. Similarly, Figure 2 compares the distribution of
privatizations in our sample and the original list of privatizations. In both cases, the bulk
of privatizations are between the mid-1990s and late 1990s, where more than half of
operations in our sample were carried out. Finally, Figure 3 compares the distribution of
privatizations by broad sector category and, as before, the resulting pattern is remarkably
similar in both cases. In fact, in our sample and in the original compiled list, privatizations
occur mostly in services, industry and mining, in that order. I I
Table 3 presents summary statistics for the variables used in this paper. The set of
variables is organized according to firm attributes, privatization characteristics, labor
characteristics, labor-restructuring policies, and some basic post-privatization hiring
measures. Along the lines of L6pez-de-Silanes (1997) the net privatization price is defined
as the amount that accrues to the government after all privatization and restructuring costs
are taken into account, such as government commitments at the time of sale, and other
adjustments are made to the sale contract. This number is adjusted by the percentage of
company shares sold and divided by the average net sales during the three years prior to
privatization. The present value of the resulting number as of December 2000 is the
dependent variable employed which is labeled Net Privatization Price / Sales.'2
The labor-downsizing variable may be viewed as a basic summary measure of
labor restructuring, as it is typically the most widely employed and most relevant from a
policy perspective. This variable is defined as a dummy variable that equals one if the
firm undertook any reduction in the labor force up to three years prior to privatization, and
zero otherwise. We follow Haltiwanger and Singh (1999) and classify this variable by
type, as voluntary and compulsory, and by targeting nature, as age-biased, skill-biased,
10. In fact, only 26.3 percent of respondents provided some numerical infornation.
11. The World Bank privatization data, which are for developing and transition economies, only also
has information on number of shares sold and foreign participation. When we compare our developing
country sub-sample (231 observations) to the one from the World Bank we also find a very consistent
pattem between both databases.
12. L6pez de Silanes (1997) also uses the firms' total assets and total liabilities to develop a so-called
Privatization Q. In our case, such variable was not possible to construct. However, as a rough proxy in our
regressions we include a dummy variable that equals one when total liabilities are greater than total assets
(see Table I for variable definition).
5



and female-biased downsizing. Table 4 provides a breakdown of our sample along these
lines. Voluntary downsizing is defined as any kind of non-compulsory labor downsizing
where the worker chooses to leave, typically as a result of a monetary or non-monetary
severance compensation package by the firm. Monetary packages are usually given as a
function of two variables, wage and seniority. Typically, the older the worker and the
higher the current wage the larger the severance package.'3 Pension enhancements and
similar benefits may also be included under this category. Non-monetary packages include
any type of in-kind payment that range from training to any other similar enhancements to
the safety net intended to help workers that leave, such as food and clothes.' 4
The type of targeting employed, if any, is another useful classification of labor
downsizing. As mentioned above, three are considered in this paper: age-biased
downsizing, skill-biased downsizing, and gender-biased downsizing. Age-biased
downsizing includes any labor cut that used age as a reference. This type of downsizing
may be applied in the context of voluntary or involuntary downsizing programs. In fact,
while voluntary age-biased downsizing is found in nearly 34 percent of total downsizing
cases (and 82 percent of total voluntary cases), this kind of downsizing is not limited to
choice as it may be implemented by force, too. In fact, one-half of all the involuntary
downsizing cases are age-biased. This is shown in Table 4. The most common age-biased
downsizing programs are voluntary early retirement programs through pension
enhancements, which, as it implies, target older workers. 15 Mandatory retirement of a
specific group of older workers is relatively common in developing countries, too.
Skill-biased downsizing includes any labor cut that uses any written or oral
method, test, interview, or certification to measure skills or occupational ability as a
retrenchment reference. Thus, skill-biased programs restrict the program along detailed
occupational or skill groupings (Haltiwanger and Singh 1999). A typical example is to test
workers on general or particular labor skills in order to decide whether or not to keep
them. In recent years, the classic example of this is the Peruvian Tax Administration in the
early 1 990s, which required all workers to pass a written test for workers to keep their
jobs. Those workers that did not pass were fired and the new potential workers brought in
to replace the old ones also had to pass a written exam. As a consequence of this, turnover
in the Tax Agency reached about 30 percent. Another example is the case of the Central
Bank of Ecuador. After a disastrous attempt to downsize using voluntary programs, the
Central Bank decided to classify all its personnel in three categories: those who were
essential for its functioning, those who were clearly redundant, and those for whom it was
difficult to tell. This classification was based on the nature of the worker's unit and on the
13. Recent research proposes using compensation schemes based on additional characteristics of the
individual or her household (such as education, gender, and others) and not just wage and seniority (Rama
1999; Chong and Rama 2001).
14. This last is not uncommon in African countries.
15. This program typically improves pension benefits if the worker retires earlier than the legal or
agreed upon age. It has been applied widely in developing and developed countries.
6



worker's occupation and educational attainment. Essential workers did not have an option
to leave (they were ring fenced), redundant workers did not have the option to stay, and
the rest were offered a voluntary separation program (Rama and MacIsaac 1999).
Gender-biased retrenchment refers to labor cuts based on any implicit or explicit
gender-based indicators. Practically all gender-biased retrenchment is female biased.
Since most developed countries have laws against gender discrimination and in both,
developing and developed countries such bias is, at least explicitly, considered ethically
wrong, we expect this variable to be biased downwards. Finally, neutral downsizing refers
to any labor cuts that did not include any of the three target groups above. Though in
theory a firm may apply one, two, or more targeted downsizing mechanisms at the same
time, interestingly, the overlap of retrenchment policies is relatively small in the case of
our sample of firms, as 87 percent of firms did not apply more than one single mechanism.
On the other hand, less than 1 percent of firms in our sample applied skill, age, and
female-biased retrenchment when downsizing at the same time. This is also shown in
Table 4.16
Roughly 78 percent of our sample of firms did some labor force downsizing, most
of it compulsory, as shown in Table 5.17 Such percentage is similar when looking at
regions.'8 However, the data show great variation in the labor cuts before privatization
both in terms of their nature as well as across regions. For instance, age-biased
downsizing was the more predominant in our sample, with 49 percent of firms using it. On
the other hand, skill-biased retrenchment is used by only 13 percent of state-owned
enterprises worldwide, mostly by developed countries with 15 percent. Additionally,
Table 5 also includes two other labor restructuring policies, employment guarantees (after
privatization) and pay cuts (prior to privatization). These are dummy variables that capture
whether a specific firm proceeded with the mentioned policy in any of the three years that
preceded privatization and, as their names indicate, their interpretation is straightforward.
Table 6 provides some simple correlation of our labor downsizing measures. Two
things stand out. First, the general downsizing measure is, as expected, correlated with its
components, in particular, voluntary downsizing. Second, the voluntary downsizing
measure is significantly correlated with age-biased downsizing, as it is frequent to find
early retirement programs as a downsizing mechanism. However, other than that, there is
little significant correlation among downsizing measures.
16. Also, while voluntary and compulsory downsizing may have been used simultaneously, as the
exarmple in the Central Bank of Ecuador above shows, this was an extremely uncommon occurrence
according to our sample. In the three instances where this occurred we treated one as two separate episodes,
and in the others we chose the predominant downsizing method.
17. This percentage is strikingly simnilar to the one found by Haltiwanger and Singh (1999) for 41
retrenchment programs in 37 countries using a mainly civil sector sample. They also find that compulsory
downsizing is as predominant as voluntary downsizing (46 percent in our sample).
18. In Latin America, Africa, and industrial countries, 82, 79, and 79 percent of the firms did some
labor force downsizing, respectively.
7



Table 7 provides a first analysis of the data. We divide the sample into two groups
according to whether any labor restructuring did or did not take place in a state-owned
enterprise. The table shows the value of the mean and median of the share adjusted net
privatization prices of the firms, the difference in net price means and medians, and the t-
statistic and z-statistics associated with such difference in means and medians,
respectively. Most labor restructuring policies yield statistically significant differences in
means and medians. Interestingly, this finding does not provide support to the idea that
governments should pursue labor restructuring, quite the opposite. Governments that
restructured labor in state-owned enterprises before privatization obtained significantly
lower privatization prices in relation with those that did not restructure labor. In particular,
governments received lower revenues as a result of labor force downsizing, the key
measure. This finding, however, does not consider that other prior restructuring policies
may be playing a role, and in particular, do not take into account endogeneity problems. In
fact, it may be argued that the firms that downsize are the ones that need to do so as they
are the worst performers. According to these results, only labor restructuring through pay
cuts yield increased prices, and even in this case the difference in means is only weakly
statistically significant.
3.     Downsizing and Privatization Prices
In this section we present regression analysis on the link between labor
restructuring policies and privatization prices. Net privatization prices are regressed
against a set of variables that has been classified in four groups. The first is firm and
privatization characteristics. We use a dummy that equals one when net total liabilities are
greater than zero for the average of the three years prior to privatization. Similarly, we
include a set of dummy variables to take into account the economic sector. 19 We also
include the percentage of shares sold, a variable that takes into account whether foreign
participation was allowed, as well the type of privatization sale, in particular, public
offerings and direct sales. The second group includes firm labor characteristics, as
reflected by the presence of unions and the existence of strikes and related physical
protests on the last three years before privatization. The third group reflects labor-
restructuring policies applied prior to privatization, namely, employment guarantees, pay
cuts, and labor cuts, including whether such downsizing was compulsory or voluntary, and
whether there are any skills, age, or gender bias in the labor downsizing operation.
Finally, the last group includes country-specific macroeconomic variables, in particular,
the gross domestic product, and the rate of inflation.20
19. These dummies are not reported in the regressions. We considered the following economic sectors:
(i) mining (metallic minerals and nonmetallic minerals); (ii) manufacturing (canned fish and seafood; sugar
mills; tobacco products; beverages; textiles, clothing, and leather; wood; paper and printing; heavy
machinery; transportation equipment); (iii) services (hotels and restaurants; land and sea transportation;
communications; and recreation); (iv) others (land; unclassified firms)
20. Since the country-specific macroeconornic variables do capture any specific variation among
countries in our sample, country dunmmies are excluded when using them and vice versa. Results do not
change. Also, other macroeconomic variables were considered and results are very similar.
8



The first column in Table 8 presents our basic results. We first use a simple
ordinary least squares approach and assume that labor-restructuring policies are
exogenous. In this case, sixteen observations are excluded from our sample as they
represent failed operations, that is, privatizations of state-owned enterprises that for one
reason or another did not find a buyer and consequently have no privatization price. With
respect to the first group of variables, firm and industry characteristics, we find, as
expected, a negative and statistically significant coefficient in the case net liabilities. The
result suggests that when net total liabilities are present, the privatization price decrease
by 31 percent (L6pez-de-Silanes 1997). With respect to privatization characteristics, we
find that the coefficient of the share of the firm that was privatized yields a negative and
statistically significant link with privatization prices. This result suggests that an
additional 10 percent of privatized share decreases the privatization price by 3 percent.
Additionally, foreign participation yields a positive and statistically significant sign at 1
percent. This result suggests that allowing foreign participation is associated with a 32
percent increase in privatization prices. Public offerings yield a positive and statistically
significant sign and are associated with a 19 percent increase in the privatization price.
Direct sales yield a negative and statistically non-significant sign (Dewenter and Malatesta
1997, L6pez-de-Silanes 1997). With respect to labor characteristics we find that the
presence of unions up to three years prior to privatization is associated with a privatization
price 25 percent lower, as the sign of the coefficient is negative and statistically significant
at 1 percent. Similarly, we also find that the strikes and other forms of physical protest are
negatively linked with privatization prices though it is not statistically significant.21 These
findings are similar to the ones by L6pez-de-Silanes (1997) for the case of Mexico.22
When focusing on the set of labor policy variables, our key set of interest, we find
that the downsizing summary measure is associated with a privatization price 8 percent
lower, as the sign of the coefficient is negative and statistically significant at 5 percent.
Again this is under the assumption of exogeneity and when excluding failed
privatizations. Similarly, we find that using employment guarantees prior to privatization
is linked with a privatization price 16 percent lower, as the corresponding sign is negative
and statistically significant at 1 percent. If maximizing revenues is the sole objective of
policymakers, applying this kind of policy contradicts such an objective.23 On the other
hand, pay cuts prior to privatization yield a negative sign that is not statistically
significant.
21. Since unions and strikes are relatively highly correlated it is not a surprise that the latter yields a
statistically non-significant coefficient. When excluding the unions variable or constructing a combined
unions-strikes variable the signs are negative and statistically significant at 1 percent.
22. They are consistent with the political view of labor restructuring by which unions may try to block
privatizations which are costly to buyers (Shleifer and Vishny 1994; Lopez-de-Silanes, Shleifer and Vishny
1997; Boycko, Shleifer, and Vishny 1996).
23. However, governments frequently have more than one, and frequently contradictory objectives.
The value of this finding from a policy perspective is, perhaps, to make policymakers aware that there
appears to be a trade-off between objectives and their cost.
9



The second column in Table 8 provides results when failed privatizations, but not
endogeneity, are taken into account. As mentioned above, we use tobits (censored below
at zero) to include observations from failed privatizations from which we did obtain labor
restructuring. Our findings using this technique to account for failed privatizations are
similar to our previous results. For instance, we find unions associated with a privatization
price 25 percent lower, as the sign of the coefficient is negative and statistically significant
at 1 percent. Similarly, the downsizing variable is associated with a 12 percent decrease in
privatization prices, as the sign of coefficient is negative and statistically significant at 5
percent. Finally, employment guarantees are associated with a privatization price 18
percent lower. In general, the results provided so far support the idea that labor force
restructuring, and more importantly, labor retrenchment decrease privatization prices. A
problem with the empirical results above is, however, that they do not take into account
potential endogeneity problems. This may arise as governments try to restructure the labor
force of the state-owned enterprises before the sale in order to raise the privatization price.
The negative sign may be simply a reflection that the firms in the worst shape are
shedding labor. For instance, if the unobservable characteristics of a firm are positively
correlated with the presence of strong unions, the government may be particularly
interested in dismantle such union.
Using a method by L6pez-de-Silanes (1997) we apply a two-step instrumental
variables approach by estimating a non-linear reduced-form equation that describes the
probability that a particular labor restructuring policy may be implemented.24 The
instruments used are classified in two groups, macroeconomic-level determinants and
firm-level determinants. The macroeconomic variables considered are (1) the average
growth rate in the three years prior to privatization, (2) the average unemployment rate
three years prior to privatization, (3) the average fiscal deficits over gross domestic
product three years prior to privatization, (4) the size of the public sector, (5) openness,
(6) law origin, and (7) continental dummies. The firm-level variables included are (1) a
dummy variable to reflect whether the firm had profits greater than zero in any of the
three years prior to privatization, (2) the presence of a leading agent bank in the country,
(3) management change, (4) political affiliation of unions, and (5) sector variables. In
general, these variables correspond with the variables employed in L6pez-de-Silanes
(1997). As required in this procedure, none of these variables is statistically significant
when included in the price equation. Also the F-statistic for the excluded instruments is
statistically significant at 1 percent. Appendix 2 shows the first stage probit for the case of
the key labor downsizing summary measure.25 The set of instruments used for each labor-
restructuring variable is shown in Table 9.
24. These variables are excluded instruments, as they are not included in the privatization price
equation. This instruments have very low statistical power when included directly in the price equation, but
they are highly correlated with the labor restructuring actions of the firrn, as shown by applying F-statistics
to test for the joint hypothesis that they are all equal to zero (Lopez-de-Silanes 1997).
25. Because of space considerations, the first stage for all the other labor restructuring measures and
for the fn-rm labor conditions are not presented. We would be happy to provide them upon request.
10



The third and fourth columns in Table 8 present our findings when correcting for
endogeneity using the method above. The former excludes failed privatizations, while the
latter includes them. In general, the results do not differ much with respect to previous
results presented. In terms of firm characteristics, net total liabilities are negative and
statistically significant at 1 percent. This result suggests that when net total liabilities are
greater than zero, the privatization price decreases by 29 percent.26 With respect to
privatization characteristics the results are very similar to the regressions assuming
exogenous independent variables.27 On the other hand, the coefficients of the labor
characteristics variables are always negative and statistically significant in the case of
unions. This result suggests that unions are associated with a privatization price 25 percent
lower. Again, these results are identical to the non-instrumented findings. With respect to
our key group of variables of interest, labor-restructuring policies, we find that even
though the variables keep the same signs as the non-instrumented regressions, two out of
the three variables included (in particular, the summary downsizing measure) yield no
statistical significance. These findings indicate that when controlling for endogeneity,
labor retrenchment policies do not appear to significantly increase privatization prices.
Moreover, employment guarantees, the only labor restructuring policy that has statistical
power is, as expected, and contrary to the aim of policymakers, negatively linked with net
privatization priced as its presence is associated with a 22 percent decrease in prices.
While the conventional wisdom has it that prospective buyers will prefer governments to
get rid of labor before privatization, our results so far show that such may not be the most
adequate policy. As seen above, our findings assuming exogeneity in the explanatory
variables yield a negative link with privatization prices, whereas those taking endogeneity
into account also yield negative signs, which though non-statistically significant at
conventional levels are barely so.28
4.     Voluntary and Targeted Downsizing and Adverse Selection
Governments sometimes intervene in the labor downsizing process by using
voluntary downsizing schemes as well as indicators of skills, age, or gender of the firm's
workforce. The inclusion of these variables in our regressions are shown in columns 5-8 in
Table 8, when a dummy accounting for voluntary downsizing is included as an
explanatory variable. Voluntary downsizing schemes usually account for a very large
26. All the coefficients of economic sectors have positive signs and are not statistically significant in
the ordinary least square regression. However, they become statistically significant at 10 percent when
including failed privatizations.
27. As in the regressions that do not control for endogeneity, the direct sale variable changes sign
although, as before, it yields no statistical significance.
28. These results are actually similar to those by Lopez-de-Silanes (1997) regarding labor restructuring
in Mexico. In fact, he finds that labor cuts yield a negative sign at 10 percent statistical significance when
not controlling for endogeneity, and a positive sign also at 10 percent statistically significance. Both results
are not robust.



percentage of total labor downsizing (Haltiwanger and Singh 1999).29 The reason for their
popularity is simple. Such schemes are politically non-costly, are attractive to workers and
thus, by-pass the power of unions, and can be relatively easily designed and administered
by governments (Rama 1999, Jeon and Laffont 1999). We find that regardless of the
econometric method and inclusion of failed privatizations, this variable yields a negative
and statistically significant sign.30 Our results suggest that voluntary downsizing is
associated with about a 12 percent decrease in privatization prices. It appears that
prospective buyers do not favor labor retrenchment. This negative link may be a reflection
of adverse selection, as workers with the best outside prospects will leave and those with
the worst outside perspectives will tend to stay. Human capital of the firm has deteriorated
and the privatization price will reflect this. When the government is in charge of
downsizing, and uses voluntary schemes it will tend to separate the wrong workers from
their jobs at an excessively high cost (Haltiwanger and Singh 1999, Rama 1999). The
resulting wrong composition of separations may occur, as governments may not be able to
make the adequate decisions or apply proper mechanisms to identify whom to retain and
whom to lay-off.
Governments also try to manage the downsizing process by focusing on three
downsizing categories that are particularly predominant: age-biased, skill-biased, and
female-biased downsizing. As described in the data section, age-biased retrenchment
focuses on age as the deciding variable, skill-biased downsizing usually focus on workers
along measures of skills, for instance, written tests, and female-biased downsizing focuses
on gender as the critical retrenchment factor.31
Table 10 shows our findings using these measures. Controlling for endogeneity we
find that downsizing using age as a benchmark results in a decrease of around 10 percent
in privatization prices as the sign of the corresponding coefficients are negative and
statistically significant at 1 percent (columns 3 and 4).32 On the other hand, our findings
suggest that downsizing using skills is barely positively significant with respect to
privatization prices (columns 3 and 4 in Table 10). According to this finding, using skills
29. In our sample, for instance, voluntary downsizing accounts for about 41 percent of total
downsizing as shown in Table 5.
30. The statistical significance ranges from I percent in the simple ordinary least squares case, to 10
percent in the tobit instrumental variables case.
31. Notice, regardless of the legal issues that may preclude the use of gender or age to downsize, firms
themselves reported such actions. Also, rule enforcement tends to be much less predominant in developing
countries (Knack and Keefer 1995).
32. In the case of age-biased downsizing, ordinary least squares coefficients are statistically significant
at 1 percent (columns 1 and 2). In the case of admiinistered downsizing by skills, ordinary least squares yield
similar signs as in the instrumental variables methods, but the coefficients are not statistically significant in
the former. As we argue above, and as L6pez-de-Silanes (1997) shows, endogeneity is a problem.
12



as a benchmark indicator may increase prices by 22 percent. When controlling for
endogeneity, the female bias variable is negative but statistically non-significant.33
The results above suggest that adverse selection may be an issue since both
voluntary downsizing and age-biased downsizing appear to reduce privatization prices.
Workers that are let go using these methods are not necessarily the least productive or
least skilled as asymmetric information theory shows. Furthermore, the fact that skill-
biased downsizing yields a positive, though marginally statistical significant link with
privatization prices, provides some further evidence towards this, especially if one
believes that skill-biased retrenchment better identifies productive from unproductive
workers.
We further explore the adverse selection issue. While downsizing may be
classified as voluntary or compulsory, when introducing group targeting, the
corresponding resulting categories may be labeled as voluntary-targeted downsizing and
compulsory-targeted downsizing. Furthermore, as before, voluntary or compulsory
targeted downsizing may be age-biased, skilled-biased or female-biased driven which
results in three voluntary targeted downsizing categories and three compulsory targeted
downsizing categories (see Table 4). Voluntary targeted results are shown in Table 11 and
compulsory targeted results are shown in Table 12. The classic example of voluntary age-
biased downsizing is early retirement programs.34 As Table 11 shows, voluntary age-
biased downsizing is negative and statistically significant at one percent when using the
uncorrected method, and negative and statistically significant at 5 percent in the two-step
procedure. This result further confirms the idea that adverse selection is a problem when
applying administered downsizing in the public sector. Moreover, the voluntary skill-
biased downsizing variable is, as before, positive but it is non-significant.35 On the other
hand, Table 12 shows results when using compulsory targeted variables. As the name
implies, in this category there is no choice element by worker. Firms simply choose the
workers that will stay and those that will leave using age (for example, older workers),
skills (for example, managers in each division choose the most skilled ones), or gender, as
a retrenchment reference. Though the signs in the compulsory age-biased variable and
compulsory skill-biased variable are similar to our previous results, the age coefficient,
unlike the voluntary targeted case, is now statistically non-significant. There is no impact
on privatization prices. This result is consistent with the fact that voluntary programs are
33. In fact, this variable is never statistically significant and does change signs depending on the
methodology employed. People admitting to gender bias were few. This is clearly not surprising, as people
may not be truthful for fear of retaliation. Also, legal and particular society considerations may be at issue.
34. In the skill-biased category finding a clean example is somewhat more difficult. After all, who
would want to take a voluntary test when she will know it may be used to fire her? Somewhat like the Fifth
Amendment though, where, for all practical purposes taldng it is frequently associated with a presumption
of guilt, not taking a suggested exam may expose the worker to retaliation. While one may argue that this
mechanism is not really voluntary, from the perspective of fn-ms, ministries, and privatization agencies-
from whom we mainly got the data from-the explicit method is clearly understood as voluntary.
35. While high correlation with the voluntary age-biased variable may be a problem, as 14 out of the
20 observations that are positive are also voluntary age-biased (Table 4).
13



theoretically expected to produce somewhat larger adverse selection problems than
compulsory ones (Jeon and Laffont 1999, Levy and McLean 1997, Kahn 1985). In fact,
while in voluntary age-biased programs it may be expected that the more productive
workers may leave rather than the more unproductive ones, in compulsory programs it is
reasonable to expect that the good and the bad will leave and thus, the net effect will tend
to cancel out. On the other hand, in the case of the compulsory skill-biased downsizing
variable we find that the corresponding coefficient yields a positive and marginally
statistically significant coefficient in the two-step Tobit procedure that includes failed
privatizations, only. Compulsory exams, as a relatively good objective measure of
productivity, may help keep the more productive workers, which is reflected in an
increase in privatization prices. From a practical perspective, however, this policy
prescription is highly controversial, as its applicability will clearly depend on the political
climate of the country.36
In summary, so far we have found that when controlling for endogeneity labor
retrenchment and pay cuts do not have any bearing on net privatization prices.
Employment guarantee programs affect prices negatively as their imposition lowers the
privatization prices by 16 percent, ceteris paribus. Furthermore, our results show that, if
anything, administered labor downsizing appears to produce a selection of the wrong
group of workers, the less productive ones as voluntary and age-biased downsizing reduce
net privatization prices between 10 and 15 percent, ceteris paribus. In fact, while
governments resort to a different array of productivity-identification methods in order to
select which workers to fire and which to keep, the evidence above shows that their
application results in the opposite effect to the one originally desired, that is, lower
privatization prices instead of higher ones. In other words, it appears that the labor
downsizing process ends up marred with adverse selection problems, which appears to be
reflected in penalized prices as potential buyers suspect that govermnents do not have the
capacity to collect adequate information on the productive abilities of workers.
5.     Re-Hiring After Privatization
As much as adverse selection appears to be a reasonable factor why buyers may
not be willing to bid higher prices for state-owned enterprises on sale, it may also be the
case that some other unobserved factor may produce a negative link between labor
downsizing and privatization prices. The question is whether there is anything close to a
silver bullet that can provide further, robust, evidence on the presence of adverse selection
in the downsizing process prior to privatization. We believe re-hires, defined as workers
that were dismissed prior to privatization but were re-hired after privatization, provide a
quite reasonable proxy that allows us to understand the pervasiveness of adverse selection
36. In fact, countries where skill-biased programs have been used rather successfully did so under not-
so-democratic regimes. Two examples in our sample are Chile (Pinochet) and Peru (Fujimori).
14



in the labor downsizing process prior to privatization.37 After all, it is by no means
obvious that a firm would need to re-hire a worker that was deemed expendable a
relatively short time ago.38 According to our data nearly 70 percent of firms did hire more
personnel after privatization. However, this number is misleading for total increases in
personnel after privatization are due not only to new hires of workers not previously
associated with the firm, but, as explained above, also with re-hires. While the former may
be attributed with the natural progression of privatized firms as an on-going concern, the
latter may be reasonably linked with the quality of the downsizing prior to privatization
An indication that re-hires may be a useful measure of quality in the downsizing
managerial process prior to privatization may be illustrated by studying its relationship
with voluntary downsizing as an explanatory variable. As the theory and the empirical
evidence on net privatization prices above shows, this latter variable is a primary suspect
in producing adverse selection. Using re-hires as the dependent variable, we find that the
coefficient of the voluntary downsizing variable is positive and statistically significant at 5
percent. That is, voluntary downsizing before privatization increases the probability of re-
hiring workers after privatization.39 Results are shown in the upper panel of Table 13,
specification 1. In fact, since typical voluntary downsizing mechanisms are theoretically
flawed with adverse selection problems, this finding is not surprising (Jeon and Laffont
1999, Kahn 1985). Workers that leave voluntarily are usually those that have the highest
chances of obtaining work outside in less time. They are also the ones that are more able
to find better-matched jobs to their abilities and skills outside of the public sector and
quasi-public sector.
On the other hand, it may be argued that the incidence of re-hires after
privatization may not necessarily reflect the presence of voluntary downsizing but the
presence of high labor firing costs and related rigidities. Possible firm buyers may want to
take advantage of regulations that allow for a clean slate approach so that state-owned
enterprises are permitted to have as many workers as possible retrenched before
privatization only to be re-employed by the privatized firm. In this way, high labor costs
are avoided while the human capital of the firm is preserved. To test for this idea, two
labor cost measures are employed. The first is an indicator of labor rigidities as measured
by the extent to which the country has signed agreements with the International Labor
Organization (ILO) and comes from Rama and Artecona (2001). This variable is defined
as the cumulative number of ILO conventions ratified by the country at the time of
privatization and is based on legal documents as compiled by Rama and Artecona.
37. While the concept was first introduced by Haltiwanger and Singh (1999), we are the first to apply
rigorous econometric methods using re-hires. We work with re-hires up to 18 months after privatization. We
tested shorter periods (12 months) and longer ones (24 months) and the empirical results do not vary.
38. Nearly 35 percent of firms did some re-hiring after privatization, of which Latin America was the
most active with upwards of 40 percent, and Asia the least active with a little more than 10 percent.
39. This, when controlling for share sold, sectoral dummies (see footnote 18), macroeconomnic controls
(rate of growth, rate of inflation, initial gross domestic product), and continental dunmnies.
15



The second variable is an index of labor firing costs constructed from legislation
from Heckman and Pages (2001). This measure summarizes the tenure-severance pay
profile using a common set of dismissal probabilities across countries and computes the
expected future costs, at the time a worker is hired, of dismissing her in the future (also
see Table 1).40 We find that the coefficients of both labor rigidity measures are negative
but statistically non-significant using re-hires measures. This is also shown in the upper
panel in Table 13. It appears that labor rigidities do not change the probabilities of re-hires
when controlling for voluntary downsizing.
Re-hiring after privatization occurred not only at the firm level, but also in some
instances, from the very same departments or areas from which the workers had been
previously fired. When exploring this more restrictive measure of re-hires as the
dependent variable we find that the coefficient of the voluntary downsizing variable is not
statistically significant. This is shown in the lower panel in Table 13, specification 2. Data
are the likely culprit of this result as only 4.7 percent of the sample re-hired in the same
department or area, compared to more than 34 percent that simply re-hired. Thus, not too
much weight should be attributed to this result.41 Additionally, and similar to the re-hires
variable, labor rigidity measures yield negative and statistically non-significant
coefficients with respect to the re-hires-same variable. This is also shown in the lower
panel in Table 13.
Voluntary downsizing is only part of the story. In fact re-hires after privatizations
are closely linked with the labor-administered process applied before privatization
according to skills, age, and gender. This is shown in fist column in the upper panel of
Table 14. Using re-hires as the dependent variable, we find that the age-targeted
retrenchment variable yields a positive and statistically significant sign at 1 percent. Age-
targeted retrenchment prior to privatization increases the probability of re-hiring by a huge
31 percent after privatization. On the other hand, the skill-targeted coefficient is negative
and statistically significant at 5 percent. Skill targeting decreases the probability that firms
will employ re-hires after privatization by 16 percent. Finally, the female-biased
retrenchment is positive but it is not statistically significant. These results are quite
consistent with our findings regarding net privatization prices. In fact, this finding
suggests that adverse selection may be a problem as voluntary downsizing increases prices
but age-biased downsizing reduces prices and increases the probability of re-hiring after
privatization, while skill-biased downsizing marginally increases net privatization prices
40. Since the Heckman and Pages (2001) samnple is relatively limited we also use an alternative
measure suggested by them, law origin. They show that French law origin is very highly correlated with
labor separation costs. In fact, we find very similar results. We would be happy to provide these additional
estimations upon request.
41. Another explanation may be related with the fact that enticing the best workers back after having
them fired entails an additional cost. It is not easy to lure good workers back, especially given their potential
alternatives outside. Additional pay, perks, or higher position may be necessary. The fact that the more
restrictive measure of re-hires after privatization is not significantly linked with voluntary downsizing
before privatization is not surprising.
16



and reduces the probability of re-hiring after privatization. Even more revealing, the
results above hold when using the more restrictive re-hiring measure. These findings are
shown in the lower panel of Table 14. Again, the corresponding coefficient for the age-
biased variable is positive and statistically significant at 1 percent while the coefficient of
42
the skill-biased variable is negative and statistically significant at 5 percent.
Similar to the case with voluntary downsizing, the incidence of re-hires after
privatization may not necessarily reflect poor management, but high labor firing costs and
related rigidities. To explore this issue, we run probit regressions using the same two
measures of labor costs used above. This is also shown Table 14. When using ILO
conventions as an additional explanatory variable we find that such control is negative but
statistically non-significant. Excessive labor costs and regulations do not seem to be a
determinant on the probability of re-hires after privatization. Moreover, the signs and
statistical significance of the age and skills variables do not change. The age-bias variable
is always positive and is statistically significant at 1 percent in both re-hires and re-hires-
same. Age-biased downsizing prior to privatization increases the probability of re-hires
increase by 32 percent and increases the probability of re-hires in the same department by
16 percent. On the other hand, the skill-bias variable is always negative, implying a lower
probability of re-employment as it is statistically significant at 5 percent for both re-hires
in general and re-hires from the same department or area in the firm. Skill-biased
downsizing prior to privatization is linked with a lower probability of re-employment that
ranges between 1 percent (re-hires-same) and 17 percent (general re-hires). Very similar
results are obtained when using the Heckman-Pages firing costs variable instead. In fact,
this variable is negative but statistically non-significant suggesting that high firing costs
do not seem to have a bearing in the probability of re-hires. The signs of the skill-bias and
age-bias variables are maintained, as well as their corresponding statistical significance.
However, the statistical coefficient of the skill-bias variable decreases to 10 percent in the
case of general re-hires.43
Finally, analogous to the analysis perforned with privatization prices, Tables 15
and 16 provide evidence related with voluntary targeting and compulsory targeting,
respectively." According to our results in Table 15, voluntary age-biased downsizing
increases the probability of re-hiring between 18.2 percent and 20.1 percent, as the
corresponding coefficients in the three specifications presented are positive and
42. As before the coefficient of the female-biased variable is positive but it is not statistically
significant.
43. We also use data for temporary workers, defined as those workers that were downsized prior to
privatization but were re-hired after privatization on a temporary basis, presumably in order to take
advantage of lower labor costs. As expected, we find that both the ILO and firing cost variables increase the
probability of ternporary hiring. Also, we find that voluntary downsizing increases the probability of
temporary hiring which is also somewhat consistent with the idea that adverse selection may be present.
44. Lack of enough observations did no allow us to provide further evidence using re-hires same as the
dependent variable for these two tables.
17



statistically significant.45 This finding further provides evidence of adverse selection. On
the other hand, voluntary skill-biased downsizing yields the expected, negative sign, as
before but in this case the corresponding coefficients are statistically non-significant.
Similarly, Table 16 shows that compulsory age-biased downsizing does imply a higher
probability of re-hiring as the corresponding signs are positive and statistically significant
at 5 percent or better regardless of the specification. When labor downsizing was done
according to age and in a compulsory manner, the probability of being re-hired will
increase between 19 and 22 percent after privatization. Similarly, compulsory skill-biased
downsizing prior to privatization appears to lower the probability of re-hiring after
privatization. The results in this case are, however, not very clean as the skill-biased
measure becomes barely significant when including labor costs (Heckmnan-Pages measure)
or not significant at all (ILO conventions case).
In summary, our findings with respect to rehiring policies are consistent with the
adverse selection hypothesis in labor restructuring by the government before privatization.
What governments do before privatization does have a direct bearing on how firms behave
after privatization. Voluntary downsizing is associated with a 13-15 percent higher
probability of re-hiring workers that were previously fired prior to privatization. Age-
biased downsizing yields a higher probability of re-hiring workers, which in the case of
voluntary age-biased downsizing reaches around 20 percent, and in the case of
compulsory age-biased downsizing reaches around 22 percent, sometimes re-hiring
workers even in the same exact departments. The one exception to these results is skill-
based downsizing which in some specifications leads to significantly lower rehiring rates
by private firms.
6.    Conclusions
Despite its importance, labor has probably been the single least addressed issue in
privatization (Megginson and Netter, 2001). There is ambivalence with respect to the
optimal policy approach to labor restructuring in privatization processes as reflected by
the recommendations of development agencies around the world. In fact, such institutions
have had a difficult time taking a position on whether or not it is a good idea to restructure
a firm and, in particular, how to deal with labor force changes prior to privatization. Early
advice called for labor restructuring prior to privatization under the premise that
governments are better able to cushion any financial blow to displaced workers mainly
through safety nets (Nellis and Kikeri 1989). Subsequent recommendations called for a
less meddlesome approach by discriminating between large firms and smaller ones. It was
suggested that smaller firms with relatively little overstaffing were sold with essentially
no labor restructuring, under the logic that such a decision should be left to the new
owners who would be in better position to choose which workers they would like to retain
or dismiss (Kikeri, Nellis, and Shirley 1992). A final view came later when prior
45. However, statistical significance using the Heckman-Pages labor firing costs reaches only 10
percent.
18



restructuring in privatization, including the labor area, was found to be associated with
lower net privatization prices paid by winners (L6pez-de-Silanes 1997).
This paper is the first to provide a formal cross-country analysis of the effects of a
detailed list of labor restructuring measures before privatization to provide guidance for
those countries still embarking in the privatization process. The lack of information on
what happens to workers during the privatization process has exacerbated the fears and
concerns of workers and governments, and delayed privatization in several countries
(Kikeri, 1999).
In this paper we address some policy concerns above by testing several competing
theories that aim to answer the following key question: should governments restructure
labor before privatization, as measured by privatization prices? While as a general
principle, getting rid of redundant workers should increase the privatization price, in
practice, governments have a very difficult time identifying the genuinely redundant
workers as asymmetric information problems remain. Firing the wrong workers may even
reduce the privatization price.
Our data allows us to analyze the impact of labor restructuring measures not only
in prices, but also in the rehiring policies followed by firms after they are privatized. The
benefit of such data is that we are able to say something about the management quality of
retrenchment policies followed by governments around the world.
We find that labor retrenchmnent does not significantly impact privatization prices,
if anything, voluntary retrenchment has a negative impact on net prices, suggesting a
potential problem of adverse selection. Through a detailed analysis of various targeting
policies, the paper also shows that government administration of the downsizing process,
can be subject to political considerations and may also result in adverse selection leading
to rehiring of the same workers after privatization. We did find that it may be possible for
governments to achieve some positive results through the managing of the process by
using a skill-focused retrenchment. This type of policy is associated with lower
probability rates of rehiring of the same workers after privatization. This type of policy is
one associated with negative political effects as workers may find it too aggressive and
may react negatively towards the whole process of privatization. The fact that the political
costs of this type of program are higher, suggests that there may be some not easily
observable or measurable firm characteristics that could explain the marginally positive
results of this policies. Another reason for this result is the possibility that the level of
documentation and design of this type of program simply makes it more palatable to
buyers. The political difficulties of using such policy and the problems it might cause in
terms of the overall objective of achieving privatization should be considered.
The summary of this paper is that governments should think long and hard before
they restructure labor force in preparation for privatization. The political consequences
may be large, the impact on privatization prices is not there, and the data on rehiring
policies shows that firms where retrenchment takes place may end up losing some of its
19



most valuable workers. A qualified do not intervene appears to be the safest bet in labor
retrenchment before privatization.
Bibliography
The word "processed" describes informally reproduced works that may not be commonly
available through library systems.
Barberis, Nicholas, Maxim Boycko, Andrei Shleifer, and Natalia Tsukanova (1996). "How Does
Privatization Work? Evidence from the Russian Shops." Journal of Political Economy 104:
764-90.
Bhaskar, V., and Mushtaq Khan (1995). "Privatization and Employment: A Study of the Jute
Industry in Bangladesh." American Economic Review.85(l): 267-73.
Boycko, Maxim, Andrei Shleifer, and Robert Vishny (1994). "Voucher Privatization." Journal of
Financial Economics 35: 249-266.
Boycko, Maxim, Andrei Shleifer, and Robert Vishny (1996). "A Theory of Privatization."
Economic Journal 106:309-19.
Chong, Alberto, and Martin Rama (2001). "Do Governments Really Have to Be That Generous?"
In Devarajan, Shantayanan, Lyn Squire, and Halsey Rogers, eds., Economists 'Forum.
Washington, D.C.: World Bank.
Dewenter, Kathryn, and Paul Malatesta (1997). "Public Offerings of State-Owned and Privately-
Owned Enterprises: An International Comparison." Journal of Finance 52(4): 1659-79.
Diwan, Ishac (1994). "Public Sector Retrenchment and Severance Pay: Nine Propositions." In
Shahid Chaudhry, Gary Reid, and Waleed Malik, eds., Civil Service Reform in Latin America
and the Caribbean: Proceedings of a Conference. World Bank Technical Paper 259,
Washington, D.C.
Fallick, Bruce (1995). "A Review of the Recent Empirical Literature on Displaced Workers."
Finance and Economics Discussion Series 95-14. Federal Reserve Board, Washington, DC.
Freeman, Richard (1986). "Unionism Comes to the Public Sector." Journal of Economic
Literature 24: 41-86.
Frydman, Roman, Cheryl Gray, Marek Hessel, and Andrzej Rapaczynski (1999). "When Does
Privatization Work? The Impact of Private Ownership on Corporate Performance in
Transition Economies." Quarterly Journal of Economics 114(4): 1153-91.
Galal, Ahmed, Leroy Jones, Pankaj Tandon, and Ingo Vogelsang (1994). Welfare Consequences
of Selling Public Enterprises. Oxford: Oxford University Press.
Hall, Robert (1995). "Lost Jobs." Brookings Papers on Economic Activity 1:221-56.
Haltiwanger, John, and Manisha Singh (1999). "Cross-Country Evidence on Public Sector
Retrenchment." World Bank Economic Review 13 (1): 23-66.
Hamermesh, Daniel (1989). "What Do We Know About Worker Displacement in the U.S.?"
Industrial Relations 28(1): 51-59.
Heckman, James, and Carmen Pages (2001). "Regulation and Deregulation: Lessons from Latin
American Labor Markets." Economia 1(1): 123-45.
20



Jacobson, Louis, Robert Lalonde, and Daniel Sullivan (1993). "Earnings Losses of Displaced
Workers." American Economic Review 83: 685-709.
Jeon, Doh-Shin, and Jean-Jacques Laffont (1999). << The Efficient Mechanism for Downsizing the
Public Sector." World Bank Economic Review 13(1): 67-88.
Kahn, Charles (1985). "Optimal Severance Pay with Incomplete Information." Journal of Political
Economy 93: 435-51.
Kikeri, Sunita (1999). "Privatization and Labor: What Happens to Workers When Governments
Divest?" World Bank Technical Paper 396, Washington, D.C.
Kikeri, Sunita, John Ellis, and Mary Shirley (1992). Privatization: The Lessons of Experience.
Washington, D.C.: World Bank.
Knack, Stephen, and Philip Keefer (1995). "Institutions and Economic Performance: Cross-
Country Tests Using Alternative Institutional Measures." Economics and Politics 7(3): 207-
27.
La Porta, Rafael, Florencio L6pez-de-Silanes, Andrei Shleifer, and Robert Vishny (1998). "Law
and Finance." Journal of Political Economy 106:1113-55.
La Porta, Rafael, and Florencio Lopez-de-Silanes (1999). "The Benefits of Privatization: Evidence
From Mexico." Quarterly Journal of Economics 4: 1193-1242.
Levy, Anat and Richard McLean (1997). "Optimal and Sub-Optimal Retrenchment Schemes: An
Analytical Framework." Department of Economics, Rutgers University, NJ. Processed
Lazear, Edward (1990). "Job Security Provisions and Employment." Quarterly Journal of
Economics 105: 699-726.
Layard, Richard (1999). "Is Unemployment Lower if Unions Bargain over Employment?" In
Tackling Unemployment. New York: St Martin's Press.
L6pez-de-Silanes, Florencio (1997). "Determinants of Privatization Prices." Quarterly Journal of
Economics CXII(4): 965-1025.
L6pez-de-Silanes, Florencio, Andrei Shleifer, and Robert Vishny (1997). "Privatization in the
United States." Rand Journal of Economics 28(Autumn).
MacIsaac, Doina, and Martin Rama (1997). "Do Labor Market Regulations Affect Labor
Earnings in Ecuador?" Journal of Labor Economics 15(3): 136-65.
Megginson, William, and Jeffry Netter (2001). "From State to Market: A Survey of Empirical
Studies on Privatization." Journal of Economic Literature 39:321-89.
Nellis, John, and Sunita Kikeri (1989). "Public Enterprise Reform: Privatization and the World
Bank." World Development 17: 659-72.
Rama, Martin (1999). "Efficient Public Sector Downsizing." World Bank Economic Review 13(1):
1-22.
Rama, Martin, and Raquel Artecona (2001). A Database of Labor Market Indicators Across
Countries. Development Research Group, World Bank, Washington, D.C. Processed.
Rama, Martin, and Donna MacIsaac (1999). "Earnings and Welfare After Downsizing: Central
Bank Employees in Ecuador." World Bank Economic Review 13(1): 89-116.
21



Shleifer, Andrei, and Robert Vishny (1994). "Politicians and Firms." Quarterly Journal of
Economics 109: 995-1025.
Svejnar, Jan, and Katherine Terrell (1991). "Reducing Labor Redundancy in State-Owned
Enterprises." Policy Research Working Paper 792. Development Research Group, World
Bank, Washington, D.C.
Topel, Robert (1990). "Specific Capital and Unemployment: Measuring the Costs and
Consequences of Job Loss." Carnegie-Rochester Conference Series on Public Policy 33:
181-224.
Vickers, John, and Yarrow, George (1991). "Economic Perspectives on Privatization." Journal of
Economic Perspectives 5: 111-132.
World Bank (1995). Bureaucrats in Business: The Economics and Politics of Government
Ownership. World Bank, New York: Oxford University Press.
World Bank (2001). World Bank Privatization Database. Washington, D.C.
World Bank (2001 a). World Development Indicators. Washington, D.C.
22



Figure 1. Distribution of Privatizations by Region
(percent)
Transition Economies                                I
Developed Countries
Africa and Middle East
Asia
Central and South
America
0               10              20              30              40
* Sample                  * Original List
Source: Data collected by authors. Original list based on 1,500 firms. Sample reflects 308 firms.
Figure 2. Accumulated Distribution of Privatizations by Year
(percent)
100 -
90 
80-
70-
60-
50 -
40-
30-
20-
1 0
0*
-Sample                          -Original List
Source: Data collected by authors. Original list based on 1,500 firms. Sample reflects 308 firms.
23



Figure 3. Distribution of Privatizations by Sector
(percent)
60- _-____,                                      ,      
50-
40-
30                                                U~~~~~~~~~~~~ Sample
U Original List
20-
10
Mining        Industry       Services       Othersl/
1/ Includes land and unclassified firns.
Source: Data collected by authors. Original list based on 1,500 firms. Sample reflects 308 firms.
24



Table 1. Description of Variables
Vanable                   3   .Description                                                                                     ____
vmo.prce/sale    Thf  etpeen.se o9f teqn     psce of sal to SU S. dollar afteralpsvh o annrsiuoo csts wre take
itacotladjust.ed1 by th Percentag o  _otpsi'aae  od n  ree  yttlalsbfr  tvna
Sales               Th~~~~~~~~'e net present vaLue 2f!the thre -year avraje.frm sa!les befor pevnaoidn ntd toin   olr
Net total liabilsoes  ~~Deensy instabl equal to 1 if act total liabliies ar gete ta   ueo p to in  years pro to privaucation.
ano0there                                                               I   iccsto
Preprivancanon profits  Dsioany vanable e~~~~qual. to I if the compnymadeanyuroitp ttheyasprotoprivxatiain an  terwise
Sector              Dsinan~~~~~~~~~~~~~L.y vsial  eqalt I tO  inketot accon for economic sector. The  usmssi  osdrd ia fr
Jin0,ssdsy.ad s _eevsc. respSctively, end 0 otherw-ise (reflecting lan. atw   reoreongtutr,orucasid
*aagme1      hage          Duianry vatableequal tolIif the compay changed CEOupto thee yearsprior toprivatir-atiend 0 otiheris
Foreign pararspanon  Dunony ~~~vanable equal to I  - p earspton  wano  j5 qprocess. end 0 otherwsse.
Shape ofsold               DeeUMSagefonsnshaeisdreedattokssa1ccou -                                         -
Type of sale       Diamny ~~qvuariale equal to I to take into accoutot for mthod~ of p.tv. a on sale. Two dwstrnes arecotdrdoacun
forw Pinita pulcofs" and dret (no-compeiie  se opcvl.ad0ohrie(elcigothe tnethiods such an
Agent bank       Diamny variabl~~~~~Eequal toIlaig gn  akorga re  si  ato Procs Leading sgets ban a dfined as bankthat
orane ms ptsvazaionsthe cosmaytthotener ofoarreea  Agent ialsnea chre of obtamnon informaion
on thetawe-own~ed eteeprie ugs etulan     esre,adognz h        aets
UssionsDnm                        ntbeeult       fta  a  nosot     ie es    pio to p....rio. and0 etherwise.
Poitca.a..iaio oftn.   ito   ntbeeult     sfpo..cal.afititonfutnonisthe smenste dLpolti,cal parthlnked web he dnsg govretoen
at th ~tim ofplnatza 'n EAn 0 otherie   .        
Soilces             Dra~~~~~~~~~Nnny vasribl-e eqalt !I if there wer a id of pretests. picketing, or stikes upto three years ro to p bioriaso.a
Dowassnstg         ~~~~~~~~~~~~Durmoy vaiabe eulto I iffirem s,todeetono mSi teLa,f- 
otherwvise. Dp!owig.may becasfe  s  *   o    oploty.an mayb    targte a cTcring& to age (agebiased
(sInE-biased~~ dowain.gender penl-iae         aose)o may be nentl(no p iicular group agtd
VobailnY Dowasn `g:        Dumanyinst:abluedeua to I if there wa ay kind ofvolintny do-umina toIk labor force three er  eot psov natn
Oothmwiae VoliatNaY dLownasng ia defind as sa  o-compaly worke-bated dcisson dwns nog Typically severac
patckege!. peosio  ~enhncm nts andoter benth are !re to incentive worskers to laethe fian
Age-biased dewensang  ~Dusm.isibeeul oIite         a  n ag-biaed labor downag    t he    er  ro opeaisin        tewo
Skill-biased dewasiong  Dumrmy variable equal to I if ther wa jn kll-biased*labr dwn-,AsUn  p t he  easpirt p mstrl on0 ohrse.
SItl-biased prpram resol the program, along de'Yd cuasoa oranlgoos(aawanger and sw8b. 999 for ensamiple.
Female-blase       ~~~~~~~thro4. the Pasasi ofFwesCien or oral ema .panel interviews. and so on.Skill-biased pormssay be voharrar or snoints
d ....  doalig     Dnnyvnbeeul oIi hr            was any    .        dIowantong op to tree years peior to p fvsaion. 0 Othcewac.
Female-biased progrm ma  evltr     rivhts     ic ytsl hr      r oepii polcie onvtIn. thr ted to be r-
underrpoenyjs this woulSantcentto tlesi onhgMdo  0ia-                               ds tob
Employment Guarate        D-amyv   al   qalto? I if there was ayp omieo mlyet urne      pt treyaspir to pewtiaton 0 therwise
pay, Cut                   D!aemy enmabLe equa to If tlhere was any pa  L o [Le total salarv zt ..sg  thde welies Icree ye no pn,, so ptIaw_.aLorar.on
Re-hoes            ~~~~~~~~~~~D-armY vnable equaldto I if the piseased imrkeo-lerd preomounly tE.d we oo (ptothree yersprior topevatin
Re-lares same        mtoon~~~~~~~~YvarihleeqLo iif th  vaid- then r-edpreojoul fired woles(op to three years prio topesan)
andP plcdthe wositer inthe same departentorare from which he was orinal firedO."0ohews
Law  p ostg              Lga orgne o-f! tE- cny From wIc  opiyi ej9Ielybsduo           iepsil lega ogins cosiere
otFc alabcmon Law. French.cad Code  Getman1 coe1m1ercial code. Scandiwavian code.,n  oshtlw  L  ot.Lpr
... .... ....      Groan.~    ~slic pRodct US! PPP) inivg.: Averg ofte-ee year psortostaiztin(WoidBn.20
Gross domestic prodact   domef c -                1- Y
Infatin       Aerae      f~aon n te eunty tre er f!peter to psvszto (Wosd Bank 200p    F 
Con.inental .tsae    Dam.yqualdoIi ..o ..count for. ollw..   ruonLaaAmerica. Asia. ltddle East, Mfica. Developed Courntes, 0
Sire of the public ector  Employmenthi th erafra-gove m toh.- "  of P5!rsom asAercWNtag of th laborforce. Inldes wOrkers of both aemes
1.0 Co..n........ ..   C...slT?!"  manbe of ntntina Labor 9rgMtiOn (I0     Mcnentions ratifed by the coutry at the time of piairts
Lbr&tig ot               Meanr tha esuarnanies the tonureseverance--L......................_.  _ypoeug!~q qse f pAOit!ciscunis Temaur
comuputS thie eapectod Oiflure cost, a the time a woelcer is hired. of diasrf  g e oteftr  u  eooal  cnmccniin
Th idn is consteucted to include only firigcssta fette' dcso' ttemrl.I include the cost of proiigstoe
adac  oteat everance Pgy ?onMtina oneach ~ psileeveoftotnsa  worker cmattsn int the hfutur Mea,ore thus
unemploytont at            Averageuemlymn rate of thounuy' thee year -I.prior t1o Pr..v.....tio (WodBak20T,R aanAten.201
25



TabRe 2. Observations in the Sanmpile
Number    Percentagefsales
Firms in our final sample                                  308        97.21
Firms that supplied incomplete information                  25         1.04
Firms merged and keep no independent records                 19       0.12
Firms that were liquidated an no longer exist               22        0.78
Firms that denied or refused to give information            26        0.85
All Privatizod Firms (1982-200                             400        100.00
Note: This table breaks our world samnple between 1982 and 2000 into two groups. For each group we
provide the number of firms and the percentage of pre-privatization sales in the total. Source: Data collected
by authors.
26



Table 3. Summary Statistics
Variable                  Obs Mean Median Std. Dev.   Min     Max
Firm Characteristics:
Net Privatization Prices/sales  308  0.587  0.609  3.228  0.000  1.367
Sales                     308  1.415  0.140   3.167   0.001   21.991
Net total liabilities     308  0.432  0.000   0.496    0.000   1.000
Preprivatization profits  308  0.455  0.000   0.499    0.000   1.000
Mining                    308  0.143  0.000   0.350    0.000   1.000
Industry                  308  0.231  0.000   0.422    0.000   1.000
Services                  308  0.558  1.000   0.497    0.000   1.000
Management change         308  0.449  0.000   0.498    0.000   1.000
Privatization Characteristics:
Foreign participation     308  0.682  1.000   0.467    0.000   1.000
Share sold                308  0.509  0.506   0.282    0.010   1.000
Public offering           308  0.653  1.000   0.477    0.000   1.000
Direct Sale               308  0.198  0.000   0.399    0.000   1.000
Labor Characteristics:
Unions                    308  0.844  1.000   0.363    0.000   1.000
Strikes                   308  0.474  0.000   0.500    0.000   1.000
Labor Policies:
Downsizing                308  0.782  1.000   0.413    0.000   1.000
Voluntary downsizing      308  0.325  0.000   0.469    0.000   1.000
Age-biased downsizing     308  0.497  0.000   0.501    0.000   1.000
Skill-biased downsizing   308  0.130  0.000   0.337    0.000   1.000
Female-biased downsizing  308  0.058  0.000   0.235    0.000   1.000
Employment guarantee      308  0.282  0.000   0.451    0.000   1.000
Pay cut                   308  0.075  0.000   0.263    0.000   1.000
Re-hiring                 292  0.345  0.000   0.475    0.000   1.000
Re-hiring Same            292  0.047  0.000   0.321    0.000    1.000
Country-Specific Variables:
English common law        308  0.253  0.000   0.436    0.000    1.000
French commercial code    308  0.500  0.500   0.501    0.000    1.000
German commercial code    308  0.117  0.000   0.322    0.000    1.000
Scandinavian commercial code 308  0.019  0.000  0.138  0.000    1.000
Socialist/communist laws  308  0.110  0.000   0.314    0.000    1.000
Gross domestic product    308 25.398 25.452   1.851   19.448   28.856
Inflation                 308 109.876 11.485 292.683   0.618 1667.207
Openess                   308 31.137 28.158   31.953   0.000  314.588
Economic growth           308  3.028  2.726   3.811  -11.144   21.320
Fiscal deficits           308 -2.580  -2.279  3.475  -14.003   13.629
ILO conventions           221 54.164 52.000   28.883   1.000  123.000
Labor firing cost         151  2.526  2.718   1.216    0.443   4.756
27



Table 4. Decomposition of Labor Downsizing Measures
Downsizing
yes                                                      no
241                                                                                       67
(78. %)                                                                                  (21.6%)
Voluntary                                                                 Compulsory
Downsizing                                                                Downsizing
100                                                                       141
(41.5%)                                                                   (58.5%)
Neutral         Age-biased(^)        Sklill-biased(-)      Female-biased(-)               Age-biased(**)       Skill-biased(-")      Femalk-biased(-)                  Neutral
5                  82                   20                    7                              71                   20                        9                         62
(5.0%/)           (82.1%)                (20%)                 (7%)                         (50.4%)               (14.2%)                   (6.3%)                    (44.0%)
Age-biased and skill-biased            14       (14)                       Age-biased and skill-biased           12            (8.5)
Age-biased and female-biased            3        (3)                       Age-biased and female-biased          5             (3.5)
Skill-biased and female-biased          3        (3)                       Skill-biased and female-biased        3             (2.1)
Age-biased, skill-biased and                                               Age-biased, skill-biased and
female-biased                           0        (0)                      female-biased                          3            (2.1)
Note: This table shows the decomposition of labor downsizing cases in our sample in terms of cases and percentages (in parenthesis). Downsizing may
be voluntary or compulsory (non-voluntary). Additionally, it mnay be classified according to its targeting nature as age-biased, skill-biased, and female-
biased. Thus, voluntary and compulsory downsizing may be targeted. Since one firm may opt to pursue more than one targeted downsizing method,
biases do not add up to 100 percent. For instance, one firm may concurrently pursue age-biased downsizing and skill-biased downsizing in its downsizing
program. The corresponding numbers for all the possible combinations are shown in (*) and (**).
28



Table 5. Labor Restructuring Measures around the World
(percent)
Latin            Africa and  Developed    Transition
America    Asia   Middle East  Countries   Economies     All
Downsizing                  82.2     58.3      79.7        79.2        76.2       78.2
Voluntary downsizing        32.5     12.5      45.3        28.6         14.3      32.5
Age-biased downsizing       57.4     29.2      54.7        54.5         26.6      49.7
Skill-biased downsizing     12.5     13.9       9.4        15.6         11.9       13.0
Female-biased downsizing    5.0       8.3      14.1         0.0         4.8        5.8
Employment guarantee        8.4      20.1      51.6        13.0         52.4      28.2
Pay cut                     8.9       0.0       1.6        13.0         7.1        7.5
Sample                      32.8      7.8      20.8        25.0         13.6      100.0
Source: Data collected by authors.
Table 6. Simple Correlation of Labor Downsizing Measures
..-z  Volunamy  AWaid    ill-biased Fm1e-biased  rrploypmit
downsang domn7g     doiWmng    domsizing   gatee
Dmwsizng
VoluitarydofiziZng      0.36568     1
A4biaged dwnsizitg      0.5239a  0.4483e      1
Skill-biaseddoWMiZiDg   0.20373   0.1447   0.1184       1
Fenmle-biadowning       0.0643    0.0342   -0.0261    0.1508       1
Enployrnt garantee      0.1036    0.027     0.0113    0.0365     0.2126e      1
PNav c                  -0.0456   -00099   -0.009     -0.0584    -0.0803   -0.0788     1
a Significant at 1 percent, b Significant at 5 percent., C Significant at 10 percent.
29



Table 7. Labor Restructuring and Privatization Prices: Tests of Means and Medians
SOEs where  SOEs where              T-statistic for change
measure was measure was                  in mean
taken      not taken  Difference
(a)         (b)        (a)-(b)  Z-staftstic for change
in median2V
Downsizing
mean       0.5532      0.7085      -0.1552       3.547 a
median     0.5711      0.7070      -0.1360       3.576 a
Voluntary downsizing
mean       0.4818      0.6376      -0.1557       4.064 a
median     0.4716      0.6259      -0.1543       3.909 a
Age-biased downsizing
mean       0.5265      0.6467      -0.1202       3.320 a
median     0.5136      0.6320      -0.1184       3.184a
Skill-biased downsizing
mean       0.5616      0.5908      -0.0292       0.534
median     0.6074      0.6157      -0.0083       0.371
Female-biased downsizing
mean       0.3533      0.6015      -0.2482       3.213 a
median     0.3765      0.6150      -0.2385       2.977 a
Employment guarantee
mean       0.4200      0.6496      -0.2296       5.853 a
median     0.3664      0.6508      -0.2844       6.936 a
Pay cut
mean       0.6893      0.5787      0.1106       -1.585 c
median     0.7424      0.6006      0.1417       -1.725 c
Note: Table 7 reports mean and median values of the privatization price/sales in the group of firms both
where the labor restructuring measure was taken compared to those firms where the labor restructuring
measure was not taken. The particular measure taken is indicated in the lines (downsizing, voluntary
downsizing, age-biased downsizing...) The third column shows the difference in mean and medians
between the net privatization price of the group of firms that took the measure compared to the group that
did not. The fourth column reports the resulting t-statistics and z-statistics of the difference in means and
medians of the two groups respectively. I/ T-test for Ho about difference between means. Unequal N's 2/ Z-
test for Ho about difference between medians. Unequal N's. (Wilcoxon rank sum). a significant at I percent;
b significant at 5 percent; ' significant at 10 percent
30



Table 8. Labor Restructuring and Privatization Prices
Dependent variable is net privatization price/sales
Variables                                                       Two-Step procedure                         Two-Step procedure
OLS      TOBIT         OLS       TOBIT      OLS       TOJIT        OLS      TO81T
(1)       (2)          (3)        (4)       (5)        (6)         (7)       (S)
1.- Firm and privalization chatacteristics:
Net total liabilities                    -0.0903    -0.0918 b   -0.2113    -0.2142    -0.0887   -0.0897     -0.1455     -0.145
(0.043)   (0.044)      (0.072)    (0.062)   (0.041)    (0.043)     (0.057)   (0.053)
Share Sold                               -0.0040    -0.0037     -0.0021    -0.0028    -0.0039   -0.0036     -0.0036    -0.0033
(0,002)   (0.002)      (0.001)    (0.001)   (0.002)    (0.002)     (0.002)   (0.002)
Foreign participation                     0.1502    0.1657       0.1229     0.1376    0.1439     0.1606 '    0.1413 '  0.1558
(0.032)   (0.028)      (0.030)   (0.028)    (0.032)    (0.028)     (0.031)   (0.029)
Public offering                           0.0911    0.1339       0.0855     0.13 15   0.0927     0.1364      0.0895    0.1339
(0.042)   (0.041)      (0.043)    (0.040)   (0.041)    (0.040)     (0.043)   (0.041)
Direct Sale                              -0.0007    0.0482      -0.0079     0.0441    -0.0018    0.0481     -0.0044    0.0463
(0.046)   (0.045)      (0.048)    (0.044)   (0.046)    (0.045)     (0.048)   (0.045)
2.- Labor Characteristics:
Unions                                   -0.1484   -0.1250      -0.1536    -0.1314   -0.1487    -0.1242     -0.1641    -0.1413
(0.035)   (0.038)      (0.034)    (0.036)   (0.035)    (0.038)     (0.035)   (0.037)
Strikes                                  -0.0075   -0.0172      -0.0191    -0.0197   -0.0069    -0.0179     -0.0323   -0.0386
(0.028)   (0.026)      (0.043)    (0.044)   (0.027)    (0.026)     (0.038)   (0.040)
3.- Labor Policies:
Downsizing                               -0.0619   -0.0600       0.0284     0.0215
(0.030)  '(0.030)      (0.033)    (0.038)
Voluntary downsizing                                                                 -0.0773    -0.0692     -0.0572   -0.0561
(0.028)   (0.027)      (0.027)   (0.029)
Employment guarantee                     -0.0927 '  -0.0956 '   -0.0862     -0.091   -0.1005    -0.1024 *   -0.0996 '  -0.103
(0.032)   (0.030)      (0.032)   (0.030)    (0.031)    (0.030)     (0.033)   (0.031)
Pay cut                                  -0.0650   -0.0723       0.0803     0.0692   -0.0688    -0.0762      0.0306    0.0155
(0.045)   (0.044)      (0.091)   (0.104)    (0.044)    (0.044)     (0.093)   (0.110)
4.- Macroeconomic Variables:
Gross Domestic Product                    0.0514    0.0533       0.0525 '   0.0553 '  0.0512 '   0.0533      0.0447    0.0467
(0.009)   (0.008)      (0.009)    (0.009)   (0.009)    (0.008)     (0.010)   (0.009)
Inflation                                 0.0001    0.0001       0.0001     0.0001    0.0001     0.0001      0.0001    0.0001
(0.000)   (0.000)      (1.128)   (0.728)    (0.000)    (0.000)     (1.841)   (1.353)
Constant                                 -0.7242   -1.1126      -0.8489    -1.2934 '  -0.7211   -1.1236     -0.6119   -1.0373
(0.310)   (0.238)      (0.332)    (0.283)   (0.299)    (0.237)     (0.330)   (0.270)
Observations                                 292       308         292        308        292       308         292        308
R-squared                                   0.53                  0.546                0.536                  0.528
F                                          21.99                  24.59                23.39                  22.97
Prob > F                                   0.000                  0.000                0.000                  0.000
Pseudo R2                                             1.327                 1.3471                1.338                1.3184
LR chi2                                             314.26                  317.23               316.86                312.24
Prob > chi2                                          0.000                   0.000                0.000                 0.000
Note: The dependent variable is net privatization price/sales, defined as the amount that accrues to the
government after all privatization and restructuring costs are taken into account, such as government
commitments at the time of sale, and other adjustments are made to the sale contract. This number is
adjusted by the percentage of company shares sold and is dividesd by the average net sales during the three
years prior to privatization. The present value of the resulting number as of December 2000 is used.
Columns (1), (2) (5) and (6) consider prior restructuring measures and the rest of variable as exogenous and
provide estimates from OLS and TOBIT regressions. Column (3), (4), (7) and (8) show the second stage of
the two-step procedure in order to account for endogeneity. All regressions include sectoral controls and
firm size controls. Robust standard errors are given in parentheses. a significant at I percent; b significant at
5 percent; c significant at 10 percent.
31



Table 9. Instruments for Potentially Endogenous Variables
Agent    Managemrent      Political      Pre-        Law       Continental     Macro       F-statistic
affiliation  privatization       I/            2/            3   on ex luded
Bank        Change       of unions      profits    ongin       dummies        controls    isrnnts
Lownsizing                       yes                        yes          yes          yes          yes           yes         4.32
Voluntary downsizing             yes          yes                        yes          yes          yes           yes         3.88
Age-biased downsiang                          yes           yes          yes          yes          yes           yes         3.72
Skill-biased downsizing                       yes           yes          yes         yes           yes           yes         3.23
Fermle-biased downsizing                      yes           yes          yes          yes          yes           yes         3.09
Enploynrnt guarantee             yes                        yes          yes          yes          yes           yes         3.18
Pay cut                          yes                        yes          yes          yes          yes           yes         3.36
1/ This set includes English ComrDn Law, Genmn Conunercial Law, Scandinavian Comrercial Law.
2/ This set includes Latin Armrica, Asia, Africa and Middle East, Developed Countries.
3/ This set includes unemployment rate, fiscal deficit, openness, Rrowth, and sie of public sector.
Note: Table 9 reports the group of instruments used in the first-step regression. The rows give the names of the dependent variables in the first-step
regression. The columns describe the different groups of instruments used in each regression writing a yes if that group is used in the estimation of each
dependent variable. The last column gives the F-statistic on the excluded instruments.
32



Table 10. Labor Targeting and Privatization Prices
Dependent variable is net privatization price/sales
Variables                                                                                                         Two-Step procedure
OLS                 TOBIT                     OLS                TOBIT
(fl                  (2)                      (3)                  (4)
1.- Firm and privtiazion characterisdes:
Net total liabilities                                                -0.0891              -0.0926    i             -0.0932              -0.0952
(0.040)               (0.043)                 (0.080)              (0.069)
Share Sold                                                             -0.004             -0.0038     b            -0.0038    b         -0.0035
(0.002)               (0.002)                 (0.002)              (0.002)
Foreign participation                                                 0.1506                0.1652    '               0.142   '          0.1564
(0.032)               (0.028)                 (0.031)              (0.028)
Public offering                                                       0.0863    b           0.1289                  0.0971    b          0.1407
(0.044)               (0.040)                 (0.041)              (0.041)
Direct Sale                                                          -0.0029                0.0462                  0.0086               0.0585
(0.047)               (0.045)                 (0.046)              (0.045)
2.- Labor Characteristics:
Unions                                                               -0.1321                -0.109                 -0.1784              -0.1537
(0.036)               (0.038)                 (0.036)              (0.037)
Strikes                                                              -0.0041               -0.0128                    0.009              0.0001
(0.027)               (0.026)                 (0.054)              (0.050)
3.- LaborPolicies:
Age-biased downsizing                                                -0.0832               -0.0843                 -0.0833    '         -0.0823
(0.026)               (0.026)                 (0.029)              (0.030)
Skill-biased downsizing                                               0.0161                0.0039                  0.1616               0.1532
(0.036)               (0.037)                 (0.097)              (0.086)
Fetrale-biased downsicing                                             0.0082                0.0274                  0.0170               0.0115
(0.059)               (0.056)                 (0.095)              (0.097)
Employment guarantee                                                 -0.0992               -0.1025                 -0.1029    '         -0.1064
(0.031)               (0.030)                 (0.034)              (0.031)
Pay cut                                                              -0.0726               -0.0798                  0.0749               0.0564
(0.044)               (0.044)                 (0.119)              (0.122)
4.- Macroeconomic Varabla:
Gross Domestic Product                                                0.0539                0.0565                  0.0461               0.0483
(0.009)               (0.008)                 (0.010)              (0.009)
Inflation                                                             0.0001                0.0001                  0.0001               0.0001
(0.000)               (0.000)                 (0.000)              (0.000)
Constant                                                             -0.8018               -1.1946                 -0.3127              -0.7673
(0.291)               (0.237)                 (0.425)              (0.352)
Observations                                                              292                   308                      292                 308
R-squared                                                                 0.54                                          0.54
F                                                                       20.05                                          19.73
Prob > F                                                                0.000                                         0.000
Pseudo R2                                                                                     1.357                                        1.364
LR chi2                                                                                     321.34                                       323.04
Prob > chi2                                                                                   0.000                                        0.000
Note: The dependent variable is net privatization price/sales, defined as the amount that accrues to the
government after all privatization and restructuring costs are taken into account, such as government
commitments at the time of sale, and other adjustments are made to the sale contract. This number is
adjusted by the percentage of company shares sold and divided by the average net sales during the three
years prior to privatization. The present value of the resulting number as of December 2000 is used.
Columns (1) and (2) consider prior restructuring measures and the rest of variable as exogenous and provide
estimates from an OLS and TOBIT regressions. Columns (3) and (4) show the second stage of the two-step
procedure to take account of endogeneity. All regressions include sectoral controls and firm size controls.
Robust standard errors are given in parentheses. a significant at I percent; b significant at 5 percent; c
significant at 10 percent.
33



Table 11. Voluntary Targeting and Privatization Prices
Dependent variable is net privatization price/sales
Variables                                                                           Two-Step procedure
Ol S          TORIT               O1LS          TORIT
(l)        ~~~(2)             (1)
1.- Firm and privatization characteristics:
Net total liabilities                              -0.0863          -0.087            -0.1331        -0.1348
(0.040)         (0.043)           (0.073)        (0.065)
Share Sold                                         .0.0039         -0.0036            -0.0039        -0.0036
(0.002)         (0.002)           (0.002)         (0.002)
Foreign participation                               0.1464          0.1623  '         0.1433         0.1572
(0.032)         (0.029)           (0.031)         (0.029)
Public offering                                     0.0863          0.1308            0.0932          0.1366
(0.041)         (0.040)           (0.042)         (0.041)
Direct Sale                                        -0.0099          0.0408            0.0045          0.0545
(0.046)         (0.045)           (0.047)         (0.046)
2.- Labor Characteristics:
Unions                                             -0.1412         -0.1164            -0.1682        -0.1428
(0.035)         (0.038)           (0.036)         (0.037)
Strikes                                            -0.0077         -0.0189            -0.0262        -0.0336
(0.027)         (0.026)           (0.040)         (0.041)
3.- Labor Policies:
Voluntary age-biased downsizing                    -0.0933         -0.0823           -0.0517         -0.0494
(0.031)         (0.029)            (0.020)        (0.022)
Voluntary skill-biased downsizing                   0.0319          0.0252             0.0295         0.0136
(0.050)         (0.054)            (0.057)        (0.056)
Voluntary female-biased downsizing                 -0.1348           -0.132            0.0077         0.0134
(0.069)         (0.103)            (0.052)        (0.046)
Employment guarantee                                -0.099         -0.1005            -0.1037        -0.1071
(0.032)         (0.030)            (0.033)        (0.031)
Pay cut                                            -0.0644         -0.0723             0.0029         0.0234
(0.044)         (0.044)            (0.125)        (0.129)
4.- Macroeconomic Variables:
Gross Domestic Product                              0.0513          0.0533             0.0465          0.049
(0.009)         (0.008)            (0.010)        (0.009)
Inflation                                           0.0001          0.0001             0.0001         0.0001
(0.000)         (0.000)            (0.000)        (0.000)
Constant                                            -0.7168         -1.1205           -0.5996        -1.0611
(0.296)         (0.237)            (0.361)        (0.294)
Observations                                           292             308                292            308
R-squared                                              0.54                              0.53
F                                                     22.98                             2 1.26
Prob > F                                              0.000                             0.000
Pseudo R2                                                             1.352                           1.3558
LR chi2                                                             320.13                            313.97
Prnh > chi2                                                           non                       _          non
The dependent variable is net privatization price/sales, defined as the amount that accrues to the government
after all privatization and restructuring costs are taken into account, such as government COmmitments at the
time of sale, and other adjustments are made to the sale contract. This number is adjusted by the percentage
of company shares sold and divided by the average net sales during the three years prior to privatization.
The present value of the resulting number as of December 2000 is used. Columns (1), (2) consider prior
restructuring measures and the rest of variable as exogenous and provide estimates from an OLS and TOBIT
regressions. Columns (3), (4) show the second stage of the two-step procedure to take account of
endogeneity. All regressions include sectoral controls and firm size controls. Robust standard errors are
given in parentheses. a significant at 1 percent; b significant at 5 percent; c significant at 10 percent
34



Table 12. Compulsory Targeting and Privatization Prices
Dependent variable is net privatization price/sales
Variables                                                                 Two-Step procedure
OLS          TOBIT            OLS         TOBIT
(1)           (2)             (3)          (4)
1.- Firm and privatization characteristics:
Net total liabilities                       -0.0888        -0.0925        -0.0641       -0.0697
(0.042)       (0.044)         (0.074)      (0.069)
Share Sold                                  -0.0043        -0.0042        -0.0037       -0.0034
(0.002)       (0.002)         (0.002)      (0.002)
Foreign participation                        0.1517        0.1659 '         0.145 *     0.1584
(0.032)       (0.029)         (0.031)      (0.028)
Public offering                               0.095        0.1 356 '       0.0962         0.139
(0.044)       (0.041)         (0.043)      (0.041)
Direct Sale                                 -0.0007        0.0461          0.0004       0.0496
(0.048)       (0.045)         (0.048)      (0.046)
2.- Labor Characteristics:
Unions                                        -0.147      -0.1231          -0.1709 9    0.1477
(0.036)       (0.038)         (0.036)      (0.037)
Strikes                                     -0.0171        -0.0257         0.0278       0.0197
(0.027)       (0.026)         (0.050)      (0.050)
3.- Labor Policies:
Compulsory age-biased downsizing             -0.0211       -0.0322        -0.0137       -0.0425
(0.033)       (0.031)         (0.100)      (0.104)
Compulsory skill-biased downsizing            0.0266         0,011         0.0624        0.0722
(0.046)       (0.051)         (0.038)      (0.042)
Compulsory female-biased downsizing           0.0709        0.1081         -0.0504      -0.0559
(0.071)       (0.068)         (0.065)      (0.069)
Employment guarantee                        -0.1003        -0.1021         -0.0989 *    -0.1024
(0.032)       (0.030)         (0.035)      (0.032)
Pay cut                                      -0.0704       -0.0786         0.2176        0.2135
(0.046)       (0.045)         (0.132)      (0.142)
4.- Macroeconomic Variables:
Gross Domestic Product                        0.0517        0.0545         0.0499 *      0.0524
(0.009)       (0.008)         (0.010)      (0.010)
Inflation                                    0.0001         0.0001          0.0o00 1     0.0001
(0.000)       (0.000)         (0.000)      (0.000)
Constant                                     -0.7866       -1 .1795        -0.4776      -0.9101
(0.300)       (0.242)         (0.475)      (0.460)
Observations                                    292           308             292           308
R-squared                                       0.53                          0.53
F                                              19.49                         20.14
Prob > F                                       0.000                         0.000
Pseudo R2                                                   1.3248                       1.3182
LR chi2                                                     313.74                       312.20
Prob > chi2                                                  0.000                        0.000
The dependent variable is net privatization price/sales, defined as the amount that accrues to the government
after all privatization and restructuring costs are taken into account, such as government commitments at the
time of sale, and other adjustments are made to the sale contract. This number is adjusted by the percentage
of company shares sold and divided by the average net sales during the three years prior to privatization.
The present value of the resulting number as of December 2000 is used. Columns (1), (2) consider prior
restructuring measures and the rest of variable as exogenous and provide estimates from an OLS and TOBIT
regressions. Columns (3), (4) show the second stage of the two-step procedure to take account of
endogeneity. All regressions include sectoral controls and firm size controls. Robust standard errors are
given in parentheses. a significant at I percent; b significant at 5 percent; c significant at 10 percent.
35



Table 13. Voluntary Downsizing and Re-Hiring
Deoendent Vanable: re-hire
Probit        dF/dX        Probit        dF/dX       Probit         dF/dX
(1)                        (2)                       (3)
Voluntary downsizing  0.3476 (0.170)  [0.1292]  0.3741 (0.198) [0.1374]  0.4027 (0.240) c [0.1581]
Union.             0.7863 (0.270) a [0.2373]  0.5006 (0.313) [0.1598]    1.1636 (0.888) [0.3492]
ILO Conventons                               -0.0002 (0.004)  [-0.0001]
Labor Finng Cost                                                        -0.1199 (0.096)  [-0.0468]
Constant            -1.947 (1.334)            -3.9402 (2.052) b         -1.1906 (0.617) b
Observations          292                        214                       146
Log likelihood     -179.87                    -130.22                    -94.92
Waldchi2             16.8                       9.19                       7.92
Prob > chi2         0.010                      0.031                      0.054
Dependent Variable: re-hire sane
Probit         dF/dX       Probit         dF/dX      Probit         dF/dX
(lb)                       (2b)                       (3b)
Voluntary downsizinw  0.0758 (0.253)  [0.0068]  0.0897 (0.289)  [0.0087]  0.1010 (0.111) [0.0187]
Union              0.2795 (0.458)  [0.0204]    0.1313 (0.486) [0.0113]   0.1728 (0.346)  [0.0123]
ILO Conventions                               -0.0030 (0.004) [-0.0003]
Labor Finng Cost                                                        -0.0730 (0.152) [-0.0072]
Constant           -4.9733 (1.680) a          -3.3286 (2.328)           -3.4810 (2.012)'
Observations          292                        214                        146
Log likelihood      -51.89                     -39.81                    -37.12
Wald chi2            7.03                       6.23                       6.03
Prob > chi2         0.060                       0.07                       0.08
All Regression include: partial privatzation dunmry, sectoral dummies,cotry macro controls, and continental dumrnes.
Standard emors and nral effects are given in parentheses and brackets respectively.
a significant at I pere b significant at 5 percent; c sigrificant at 10 paent
36



Table 14. Labor Targeting and Re-Hires
Denendent Variable: re-hire
Probit         dF/dX       Probit        dF/dX        Probit        dF/dX
(1)                        (2)                        (3)
Age-biased downsizing   0.885 (0.170)  [0.3100]   0.9453 (0.198) [0.3200]    0.8354 (0.242)a [0.3114]
Skill-biased downsizing  -0.512 (0.242)b [-0.1624]  -0.5894 (0.288)b [-0.1778]  -0.5827 (0.315) c [-0.2089]
Female-biased downsizins  0.557 (0.360)  [0.2141]  0.7464 (0.450)  [0.2861]  0.0577 (0.825) [0.0223]
Union                   0.633 (0.283) b [0.1936]  0.2859 (0.325)  [0.0931]   0.9826 (0.610)  [0.3090]
ILO Conventions                                  -0.0002 (0.000)  [-0.0001]
Labor Firing Cost                                                           -0.0450 (0.118)  [-0.0175]
Constant               -1.974 (1.390)            -4.1495 (1.976)            -3.8461 (2.747)
Observations             292                        214                        146
Log likelihood        -164.49                    -118.09                     -88.48
Wald chi2               47.13                      34.17                      20.38
Prob > chi2             0.000                      0.000                      0.000
Dependent Variable: re-hire same
Probit          dF/dX      Probit         dF/dX       Probit         dF/dX
(lb)                       (2b)                       (3ba
Age-biased downsizing  0.9907 (0.248)  [0.1664]   1.0145 (0.245) [0.1612]    0.8945 (0.334) [0.1521]
Skill-biaseddownsizing  -0.1389 (0.070)b [-0.0060]  -0.2741 (0.139) [-0.0140]  -0.2615 (0.115) [-0.0131]
Female-biased downsizint 0.1108 (0.583)  [0.0164]  0.1211 (0.620)  [0.0201]  0.1614 (0.609)  [0.0193]
Union                  0.2801  (0.522)  [0.0380]  0.4801 (0.642)  [0.0488]   0.3814 (0.542) [0.0512]
ILO Conventions                                  -0.0017 (0.004)  [-0.0029]
Labor Firing Cost                                                           -0.0585 (0.149) [-0.0138]
Constant              -3.0785 (1.596)            -3.7566 (1.901)            4.1594 (2.979)
Observations             292                        214                         146
Log likelihood         -87.91                     -67.11                     -62.51
Wald chi2                20.1                      22.15                      12.21
Prob > chi2             0.001                      0.003                      0.070
All Regression include: partial privatization dummy, sectoral dummies,country macro controls, and continental dummies.
Standard errors and marginal effects are given in parentheses and brackets respectively.
a                   b
Significant at I percent: significant at 5 percent: C significant at 10 percent
37



Table 15. Voluntary Targeting and Re-Hiring
Dependent Variable: re-hire
Probit         dF/dX        Probit        dF/dX        Probit         dF/dX
(1)                         (2)                         (3)
Voluntary age-biased downsizing      0.486 (0.189)  [0.1825]   0.5438 (0.223)b [0.20121    0.4696 (0.261) [0.1850]
Voluntary skill-biased downsizing   -0.230 (0.357)  [- 0.0793]  -0.6009 (0.433)  [-0.1778]  -0.4238 (0.506)  [-0.1571]
Voluntary female-biased downsizing  -0.276 (0.727)  [-0.0932]  -0.1348 (0.779)  [ -0.0459]  -0.1241  (0.788)  [0.0323]
Union                                0.499 (0.259)  [0.1638]   0.1576 (0.292)  [ 0.0540]   0.8966 (0.495)  [0.2993
ILO Conventions                                               -0.0035  (0.004)  [ -0.0012]
Labor Firing Cost                                                                         -0.0772  (0.120)  [-0.0303]
Constant                            -0.798 (0.648)            -0.8361  (1.149)            -1.3176 (1.177)
Observations                          292                         214                         146
Log likelihood                    -177.33                      -125.82                     -88.48
Wald chi2                             22.8                      20.75                        7.87
Prob > chi2                          0.000                      0.000                       0.040
All Regression include: parbal privatization dummy, sectoral dummies,country macro controls, and continental dummies.
Standard errors and marginal effects are given in parentheses and brackets respectively.
Significant at I percent; b significant at 5 Dercent; ' significant at 10 percent
Table 16. Compulsory Downsizing and Re-Hiring
Dependent Variable: re-hire
Probit         dF/dX        Probit         dF/dX        Probit         dF/dX
(1)                         (2)                         (3)
Compulsory age-biased downsizing     0.713 (0.190) ' [0.2699]    0.726 (0.22 1) [0.2695]   0.5602 (0.257)b [0.2201]
Compulsory skill-biased downsizing  -0.862  (0.353) b [-0.242]  -0.583 (0.366)  [-0.1738]  -0.7775 (0.435)' [-0.2660]
Compulsory female-biased downsizing  0.738 (0.404)' [0.2861]     1.008 (0.538)' [0.3858]   0.3483 (0.869)  [0.1382]
Union                                0.449 (0.252)  [0.1480]     0.094 (0.301)  [ 0.0323]  0.8522 (0.490)' [0.2858]
ILO Conventions                                                -0.0056 (0.004)  [ -0.0019]
Labor Fifing Cost                                                                         -0.0496 (0.120)  [-0.0194]
Constant                            -0.572 (0.649)              0.0014 (1.120)            -0.8201  (1.164)
Observations                          292                         214                         146
Log likelihood                     -171.05                      -121.4                       -92.4
Wald chi2                            32.59                       24.44                       11.66
Prob > chi2                         0°00°0                       0000                       0.040
All Regression include: partial privatization dummy, sectoral dummies,country macro controls, and continental dummies.
Standard errors and marginal effects are given in parentheses and brackets respectively.
Significant at I percent; b significant at 5 oercent; ' significant at 10 Percent
38



Appendix 1. Country Sample
Africa:
1.- Benin                   (2)      30.- Saint Vincent  (1)      57.- Austria              (5)
2.- Cape Verde              (1)      31.- Peru         (13)       58.- Belgium              (I)
3.- Cote d'lvoire          (10)      32.- Puerto Rico   (1)       59.- Bulgaria             (6)
4.- Egypt                   (6)      33.- St. Kitts and  (2)      60.- Croatia              (3)
Nevis
5.- Gabon                   (1)      34.- Trinidad and  (2)       61.- Czech Republic       (6)
Tobago
6.- Ghana                    (8)     35.- Venezuela     (6)       62.- Denmark              (1)
7.- Kenya                    (7)     36.- Barbados      (2)       63.- Estonia              (4)
8.- Lesotho                  (1)                                  64.- Finland              (2)
9.- Senegal                 (2)    Asia:                          65.- France               (7)
10.- South Africa            (4)      37.- Bahrain       (1)       66.- Germany              (5)
11.- Tanzania                (2)      38.- China         (1)       67.- Hungary              (I)
12.- Uganda                  (6)      39.- India         (1)       68.- Ireland              (3)
13.- Zambia                  (3)      40.- Indonesia     (3)       69.- Italy                (7)
14.- Madagascar              (3)      41.- Israel        (1)       70.- Latvia               (2)
42.- Japan         (8)       71.- Lithuania            (2)
Americas:                               43.- Jordan        (1)       72.- Netherlands          (3)
15.- Argentina               (8)      44.- Korea, Rep.   (5)       73.- Poland              (10)
16.- Belize                  (2)      45.- Kuwait        (2)       74.- Portugal             (3)
17.- Bolivia                 (8)      46.- Lao PDR       (1)       75.- Russia               (3)
18.- Brazil                 (16)      47.- Malaysia      (3)       76.- Serbia               (I)
19.- Canada                  (4)      48.- Pakistan      (1)       77.- Slovak Republic      (2)
20.- Chile                   (4)      49.- Philippines    (3)      78.- Spain                (5)
21.- Colombia                 (9)     50.- Qatar         (1)       79.- Sweden               (3)
22.- Dominican Republic      (2)      51.- Singapore     (2)       80.- Switzerland           (2)
23.- El Salvador             (2)      52.- Sri Lanka      (1)      81.- Turkey                (2)
24.- Grenada                 (1)      53.- Taiwan         (1)      82.- United Kingdom      (13)
25.- Guatemala               (1)      54.- Thailand       (1)
26.- Guyana                   (3)     55.- Yemen, Rep.    (1)     Oceania:
27.- Jamaica                 (4)                                   83.- Australia            (5)
28 - Panama                   (2)   Europe:                        84.- New Zealand           (4)
29.- Mexico                   (8)     56.- Albania       (2)
39



Appendix 2. First Stage Probit
Variables                                                              Probit Model
Pre-privatization profits                                         -0.7573 a
(0.168)
Political affiliation of unions                                   -0.1823 b
(0.091)
Latin America                                                     -0.1941
(0.243)
Asia                                                               0.1790
(0.298)
Africa and Middle East                                             0.3652
(0.304)
Developed Countries                                                0.3681
(0.263)
English Common Law                                                -0.4983 c
(0.277)
German Commercial Code                                            -0.3351
(0.213)
Scandinavian Code                                                 -1.0645 b
(0.463)
Openness                                                           0.0011
(0.003)
Number of observations                                                308
Pseudo R Squared                                                    0.205
F-statistics on excluded instruments                                 4.32
Prob>F                                                              0.000
Note: This appendix p;esents the first-step regression of the two-step procedure for one of the potentially
endogenous variables. Robust standard errors are given in parentheses. Regressions include agent bank
dummy (not reported) a significant at I percent; bsignificant at 5 percent; c significant at 10 percent.
40









Policy Research Working Paper Series
Contact
Title                            Author                  Date              for paper
WPS2868 Universal(ly Bad) Service:        George R. G. Clarke    July 2002          P. Sintim-Aboagye
Providing Infrastructure Services  Scott J. Wallsten                       38526
to Rural and Poor Urban Consumers
WPS2869 Stabilizing Intergovernmental     Christian Y. Gonzalez  July 2002          B. Mekuria
Transfers in Latin America:      David Rosenblatt                          82756
A Complement to National/        Steven B. Webb
Subnational Fiscal Rules?
WPS2870 Electronic Security: Risk Mitigation  Thomas Glaessner   July 2002          E. Mekhova
In Financial Transactions-Public  Tom Kellermann                           85984
Policy Issues                    Valerie McNevin
WPS2871 Pricing of Deposit Insurance      Luc Laeven             July 2002          R. Vo
33722
WPS2872 Regional Cooperation, and the Role  Maurice Schiff       July 2002          P. Flewitt
of International Organizations and  L. Alan Winters                        32724
Regional Integration
WPS2873 A Little Engine that Could ...    Liesbet Steer          August 2002        H. Sutrisna
Domestic Private Companies and   Markus Taussig                            88032
Vietnam's Pressing Need for Wage
Employment
WPS2874 The Risks and Macroeconomic       David A. Robalino      August 2002        C. Fall
Impact of HIV/AIDS in the Middle  Carol Jenkins                            30632
East and North Africa: Why       Karim El Maroufi
Waiting to Intervene Can Be Costly
WPS2875 Does Liberte=Egalite? A Survey    Mark Gradstein         August 2002        P. Sader
of the Empirical Links between   Branko Milanovic                          33902
Democracy and Inequality with
Some Evidence on the Transition
Economies
WPS2876 Can We Discern the Effect of      Branko Milanovic        August 2002       P. Sader
Globalization on Income Distribution?                                      33902
Evidence from Household Budget
Surveys
WPS2877 Patterns of Industrial Development  Raymond Fisman       August 2002        K. Labrie
Revisited: The Role of Finance   Inessa Love                               31001
WPS2878 On the Governance of Public       Gregorio Impavido      August 2002        P. Braxton
Pension Fund Management                                                    32720
WPS2879 Externalities in Rural Development:  Martin Ravallion     August 2002       C. Cunanan
Evidence for China                                                         32301
WPS2880 The Hidden Costs of Ethnic        Soumya Alva             August 2002       T. Bebli
Conflict: Decomposing Trends in  Edmundo Murrugarra                        39690
Educational Outcomes of Young    Pierella Paci
Kosovars
WPS2881 Returns to Investment in Education:  George Psacharopoulos  September 2002  N. Vergara
A Further Update                 Harry Anthony Patrinos                    30432



Policy Research Working Paper Series
Contact
Title                            Author                  Date               for paper
WPS2882 Politically Optimal Tariffs:       Dorsati Madani         September 2002     P. Flewitt
An Application to Egypt          Marcelo Olarreaga                          32724
WPS2883 Assessing the Distributional Impact  B. Essama-Nssah      September 2002     0. Kootzemew
of Public Policy                                                            35075