BENEFITS AND COSTS OF PRIVATIZATION: EVIDENCE FROM BRAZIL
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Third version. For discussion
only. Not to be quoted.
Comments are welcomed.
BENEFITS AND COSTS OF PRIVATIZATION:
EVIDENCE FROM BRAZIL
FRANCISCO ANUATTI-NETO, MILTON BAROSSI-FILHO, A. GLEDSON DE CARVALHO
AND ROBERTO MACEDO1
I. INTRODUCTION
The Brazilian privatization program has been a major one by international standards. From
1991 to July 2001, the state transferred the control of 119 firms and minority stakes in a number of
companies. In the case of companies where the government had a majority participation in control
(hereafter, state owned enterprises, or SOEs), and in those where it had a minority participation in
control (hereafter, state owned minority participations, or SOMPs), the auctions produced US$67.9
billion in revenues, plus the transfer of US$18.1 billion in debt. The government also sold US$6
billion of minority participations in firms that remained as SOEs, obtained US$10 billion from new
concessions of public services to the private sector, and sold US$1.1 billion in scattered non-control
participations of BNDES, the National Social and Economic Development Bank, in various private
companies. This dimension, one of the largest in the world, makes the Brazilian program worthy of
special attention.
Nevertheless, the Brazilian experience has been largely ignored by the international literature.
For instance, a recent survey by Megginson and Netter (2001) recognizes the Brazilian program as
“likely to remain very influential”, because of its size and the largeness of the country.2 However,
their survey does not cover any specific study of the Brazilian program. This is due to the paucity of
studies, and also to the fact that most of existing literature was published in Brazil only, and in
Portuguese. Even in this case, however, the existing studies have their shortcomings, as will be clear
from a review that will be made in this paper. Therefore, there is room for adding to the literature,
both Brazilian and international.
It is also important to bring conclusions to the Brazilian public at large. The performance of
the economy was very disappointing in the nineties. Some groups, among them politicians and
journalists, have often expressed their frustration with privatization and other policies of the so-called
Washington consensus, which are thus blamed for the sluggish growth of the economy. In part
because of this, the program stalled since 1999. Thus, it is crucial to show the results of the
1
Anuatti-Neto, Barossi-Filho and Gledson de Carvalho: University of São Paulo and FIPE-Foundation Institute of
Economic Research. Macedo: also Mackenzie University and FAAP-Foundation Armando Álvares Penteado, São Paulo.
This paper was developed with the financial support from FIPE and from the LACRNP-Latin American and Caribbean
Research Network Program of the IDB-Interamerican Development Bank. The assistance of Economática, in providing the
main data set used in the analysis , and of Renata Domingos and Alan de Genaro Dario, in processing the data, is
gratefully acknowledged. All remaining errors are the authors´ alone.
2
Megginson and Netter (2001), p.326.privatization program as such, as this will shed light on a discussion largely based on unwarranted
conclusions.
In addition to its conclusions, this paper adds to the literature on the Brazilian program in six
major aspects. In terms of the privatized companies covered, it is the most comprehensive thus far. In
the sampling of companies, we avoided a selection bias, by including both large and small firms,
SOEs and SOMPs, as well as companies listed and unlisted in the stock exchange. In addition to tests
of means and medians, the empirical analysis also resorts to panel data analysis. The paper is also
updated, as it covers performance indicators as recent as 2000. Moreover, it has been carried by an
independent team, while most of the previous major studies had been produced by members of the
staff of the federal Brazilian agency in charge of privatization. In addition, the analysis of
performance before and after privatization is also made in comparison to the private sector.
In the conclusions, the paper shows that privatization has improved the performance of the
firms. It also reveals other benefits as well as some costs of the privatization process. In this respect,
it concludes that the privatization program could have fared better, as it also shows some
shortcomings, both of a micro and macroeconomic nature.
The paper is organized as follows. Section II describes the Brazilian privatization program and
surveys the literature on it, in particular the major studies published in Portuguese. Section III
describes the variables and the data set used in the empirical analysis. Section IV summarizes the
methodology and presents the empirical results. Section V contains a discussion of other benefits the
program, in addition to those found in Section IV, as well as some of the costs. Section VI
summarizes the major conclusions. References come next, followed by two appendices: A.1 presents
a list of the privatized companies, and A.2 covers technical procedures adopted in the tests of means
and medians.
II. THE BRAZILIAN PRIVATIZATION PROGRAM AND THE LITERATURE3
The Brazilian privatization program has three parts: (a) the federal National Program of
“Desestatization” (NPD), which started in 1991; (b) similar programs at the state level, which began
in 1996; (c) the privatization program of the telecommunications industry. The latter, also at the
federal level, started in 1997, as a program separated from the NPD, but running parallel to it. We
shall refer to it as the Telecom program. The auctions of Telecom, heavily concentrated in 1997 and
1998, produced a total of US$28.8 billion in revenues plus US$2.1 billion in transfer of debt. The
NPD came to a total of US$28.2 billion in revenues plus US$9.2 billion in transfer of debt, while the
program of the states produced a total of US$27.9 billion in revenues plus US$6.8 billion in transfer
of debt.4
The composition of the total program by industry shows that electricity responded for 31% of
the total value of the auctions, telecommunications 31%, steel 8%, mining 8%, oil and gas 7%,
petrochemicals 7%, financial 6% and others 2%. Pushed by the Telecom program, privatization
reached a peak in 1997-98, period which accounted for 69% of the total value until now. This will
3
This section and other parts of this paper draw from Macedo (2000), and updates his analysis.
4
These values exclude concessions of public services.
2have important implications for the analysis, in Section V, of the macroeconomic impact of the
program in conditions of fiscal crisis and the external imbalance.5
Before moving to the literature on the program, we address the following questions: 1) what
enterprises the government had before the program started; 2) what enterprises have been privatized;
and 3) what enterprises still remain under control of the government. We have little information on the
initial picture at the 28 Brazilian states and on what they still have to privatize. Therefore, with respect
to questions (1) and (3) we will focus on the federal level only, the most important part of the
program. With respect to what has been privatized, our information covers the whole program.
The initial picture at the federal government
In 1980 the federal government undertook a survey of all its state "entities", including
companies, foundations, port authorities, research institutes, autonomous organizations, councils in
charge of professional registry, and so forth. These institutions added to 560, among which 250
organized as firms (mainly in the form of corporations). In the eighties, some minor privatizations
occurred, and a few firms were closed. Moreover, at the start of the federal privatization program, in
1991, other firms also ceased to exist. As a result, the program started with 186 firms still under
government control. At the end of 2000, mainly because of the privatization program, this number
was reduced to 102.
What has been privatized
Table A1 Panel A in Appendix A contains a list of 37 firms privatized by the federal
government since 1990. Table A1 Panel B lists 75 firms privatized on behalf of some states by the
BNDES (National Bank for Economic and Social Development), a federal agency in charge of the
privatization program; some minority participations formerly held by the federal government; and
firms privatized by the State of São Paulo. In both tables we single out the firms included in our initial
sample and the revenue obtained from their sales.
What remains as SOEs
Table A3 in Appendix A includes the SOEs that still remain under control of the federal
government. It is a mixed bunch, since besides strictly defined SOEs, it includes hospitals, port
authorities, the Post Office, a firm in charge of agricultural research, the BNDES and others. Among
the SEOs, the major ones are in 1) the electricity industry (item 1.1 of the list), whose privatization
has been postponed; 2) the oil industry (item 1.2), of which the government has sold a minority
participation in 2000, but still keeps under control; and 3) the financial sector (item 2), in which a few
federal banks and most state banks have already been privatized. Table A3 also includes some state
banks that have been federalized for privatization. The largest financial institutions in the list are
Banco do Brasil, CEF (the National Savings Bank), and BNDES. For them privatization has been out
of consideration. Notice also that BNDES is essentially a government agency in charge of long run
financing and specific tasks, such as the privatization program. Finally, Table A3 (item 3) contains a
group of entities organized as corporations, where the government has a 100% control. Some of them
5
The BNDES (National Bank for Economic and Social Development) is the major source of data on the Brazilian
privatization program as a whole. It was given the task of managing it, including a part developed at the state level. The
reports and other documents used as sources are BNDES (1999a, 1999b and 2001).
3are government agencies disguised as companies. These firms are directly linked to the federal
budget, from which they receive practically all the resources they invest.
The government continues to add updated numbers to the results of the program, even though
the privatization program stalled after 1998. Since then, only three major SOEs have been privatized.
The most relevant was Banespa (federalized bank formerly owned by the state of the State of São
Paulo), sold in 2000 for US$3.6 billion. Banespa had been in the pipeline for many years, hampered
by court battles.
As explained before, privatization and other liberalization measures coincided with sluggish
growth and were blamed for it. Moreover, some accusations that the government had pushed too far
its efforts to bring interested groups to the Telecom auctions caused furor in the press, and led the
Minister of Telecommunications to resign in 1998. Furthermore, if continued, the program would
reach further into politically sensitive areas such as electricity, where the states are very strong; oil,
where the gigantic Petrobras still arouses strong nationalistic feelings; and the almost bicentennial
Banco do Brasil, which plays an important role in financing farmers in remote areas of the country
and, therefore, has strong political support. The government’s own failures in the economic area have
also handicapped it to further push the privatization program.
The Brazilian literature on privatization
In reviewing this literature, we will concentrate on the studies which have addressed the status
of the SOEs before and after privatization, as this is the major focus of this paper. Section VI will
refer to the literature on other issues as well.
Three studies are worth mentioning in this section. Pinheiro and Gambiagi (1997), of the
BNDES staff, presented an overall evaluation of the pre-privatization performance of federal SOEs
for the 1981-94 period. They showed disappointing figures for SOEs, both in terms of profitability
and dividends received by the Treasury. Over that whole period, the ratio of profits to net assets was a
negative 2.5% on average. Moreover, from 1988 to 1994, years for which data on dividends were
available, they accounted for only 0.4% of the equity capital owned by the federal government in the
SOEs.
One of the causes for this disappointing performance were the wage policies of the SOEs.
Macedo (1985) undertook a comprehensive analysis of wage differentials between private and SOEs.
His data consisted of wages and other characteristics of individual workers, obtained from forms filled
by the firms every year, as required by the Ministry of Labor.6 For ten industries, he compared the
wages of the workers in private firms and SOEs of approximately the same size. After controlling for
differences in education, age, gender and experience, he found sizable differentials in favor of the
workers of the SOEs. This differential, net of the workers’ characteristics, reached a peak of 80%.
This occurred when the characteristics of the workers were valued according to the criteria of the
private sector, as measured by the regression coefficients of the workers´ characteristics in the wage
equation of that sector.
6
The same kind of data will be used in the analysis of employment effects in Section VI. This data basis is known as
RAIS-Annual Survey of Social Data.
4The third study is Pinheiro (1996) and it is the most important thus far. He analyzed the
performance of 50 former SOEs before and after privatization, using data until 1994 and a
methodology adopted in other studies directed at evaluating the change in performance of firms
following privatization.7 His data covered 1 to 4 years before and after privatization for each company
and come from data sets similar to those used in this study, but complemented by questionnaires filled
by the firms and delivered to BNDES for this purpose. Unfortunately, the bank’s policy prevents the
use of the data by outsiders. The study covered eight variables: net sales, net profit, net assets,
investment, fixed investment, number of employees, debt and an index of liquidity. From these
variables, other six were derived to measure efficiency: sales and profit by employee, the rate of
return in the form of profit to sales and to net assets, and the propensity to invest, both with respect to
sales and to assets. Pinheiro separated the companies privatized in the eighties, when some minor and
scattered privatizations were undertaken, from those sold thereafter, and the number of observations
ranged from 29 to 46 (14 to 19 in the first period, and 11 to 27 in the second), depending on the
variable. No comparison was made with the performance of the private sector, as a control group.
The conclusion was that “in general, the obtained results confirm that privatization brings a
significant improvement... of the performance of the firms. Thus, for most of the variables, the null
hypothesis of no change in behavior is rejected in favor of the alternative hypotheses that privatization
increases the production, the efficiency, the profitability and the propensity to invest, reduces
employment and improve the financial indicators of the firms.”8
This paper adds to this literature in various aspects, as will become clear from the analysis that
follows. It was carried out by an independent team and covers a larger number of firms and data until
the year 2000, obtained from data that can be disclosed. We took explicit care to avoid a selection
bias, by including both large and small privatized firms, SOEs and SOMPs, as well as those listed in
the stock exchange and unlisted ones. In addition to tests of means, the empirical analysis also resorts
to panel data analysis. Moreover, the analysis of performance before and after privatization is also
made in comparison to the indicators observed in the private sector during the same periods. The
importance of this last feature must be underscored, as the Brazilian economy suffered various cycles
in the pre and post privatization periods. In summary, strong growth in 1994 and 1995, when a
minority of companies had already been privatized, and a sluggish performance thereafter, followed
by a strong recovery in 2000, when all former SOEs in our sample had been privatized. Thus,
economic cycles might have affected the performance of former SOEs. The absence of control for this
effect could have blurred the results of the impact of privatization as such. In our analysis, we
overcome this problem by adjusting the performance of the former SOEs (both before and after
privatization) with respect to the performance of private enterprises.
7
Among them, Meggison et al. (1994), as cited by the author.
8
More recently, in a seminar sponsored by BNDES to celebrate the 10th anniversary of the privatization program, Pinheiro
(2000) presented some additional and updated results, again based on data that cannot be disclosed, this time covering 55
firms. Without the form of a scientific paper, that is, with methodology, description of the data set and statistical tests, he
simply compared the periods before and after privatization, for the privatized firms in isolation, thus not comparing their
performance with those of the private firms. He found sizable increases in net operational revenues, investment, net profit,
productivity, tax collections and a reduction in employment, in some cases compensated by an expansion in contracted out
services. We will return to the question of employment in Section VI.
5III. THE DATA SET AND THE VARIABLES
The Sample
The source of our data set are the annual financial statements (balance sheets, income
statements and cash flows) of the privatized companies, as well as number of private enterprises to be
used as a control group. Brazilian accounting standards and procedures, as established by law and
regulatory agencies, have remained the same for the whole period, facilitating our analysis.9 The data
range from 1987 to 2000. The financial statements were obtained from two consulting firms,
Economática and Austin Assis, and from Getulio Vargas Foundation, a NGO. All three collect
financial statements from several sources, including those published in newspapers. We excluded
from our analysis the privatizations undertaken in the financial sector, as it has a different structure,
involves specific issues, and would have required an analysis of its own. We also excluded the cases
where the government sold only a minority participation in remaining SOEs, as well as the cases
where BNDES sold minor non-control participations in scattered companies, as part of its portfolio as
a development bank. Thus, we focused only on the sales of control packages, both of a majority and
minority nature. These procedures are among those shown in Table 1 to explain the focus of the
analysis and the coverage of the samples.
To proceed, it is necessary to differentiate between privatization contracts (or auctions) and
privatized enterprises. A number of the former SOEs were sold as a block, and the winning bidder for
an operational holding company was also given access to the control of its subsidiaries. In the case of
the Telecom sector, for instance, five amalgamated blocks of privatization auctions covered the entire
local, cellular long distance and international restructured segments.
The data set covers 66 privatization contracts, corresponding to 102 enterprises. The number
of privatization contracts is smaller than the number of companies because many were sold as
members of a multifirm conglomerate. The sample covers companies listed in the Sao Paulo Stock
Exchange, as well as unlisted ones, as shown in the Table A.1, in the Appendix. From the information
in Table 1, it can be seen that our sample of control packages covers 66% of the contracts, 63% of the
firms which they include, and 92% of the total value of the results of the auctions. The tests of means
and medians covered 73% of the firms in the sample, and 89% of the total value of the results of their
auctions.
The Variables
Given the nature of our data set, it involves essentially the same financial variables used by La
Porta and Lopez-de-Silanes (1999) in their study of the Mexican case, which has served as a reference
to a group of studies of other Latin American countries, including this one. Fifteen financial
indicators, according to seven criteria, represent this set of variables, listed as follows and described in
Table 2
9
High rates of inflation plagued the economy from 1986 to 1994, a period in which indexation following legal rules was
widespread. As the analysis will be developed in terms of ratios based on flow variables, such as operating income-to-
sales, the problems of inflation and indexation are circumvented in this fashion. In a few cases where the absolute value of
the indicator is used, the values in the Brazilian currency have been converted into dollars.
6Table 1
DESCRIPTION AND COVERAGE OF THE SAMPLE
Number of Number of Auctions Results
Contracts Companies (US$million)(*)
Financial sector 9 9 5,112.30
Minority sales in SOEs 6 6 6,164.10
PRIVATIZATION
PROGRAM BNDES participations 1,146,00
(1991 – 2000)
Control packages sales 103 147 76,878.20
Total 118 162 89,439.20
State minority control 16 16 1,299.20
SAMPLE (control
State majority control 50 86 70,709.80
package sales only)
Total 66 102 72,009.00
Mean/median tests 73 68,062.50
STATISTICAL Control packages 102 72,009.00
METHODS OF
ANALYSIS Panel SOEs 20
Private sector 158
(*) Includes transferred debt (US$17.8 billion), offers to employees in the telecommunications
industry (US$0.3 billion), and excludes concessions of new services (US$7.7 billion).
a) Profitability: OI/S (Operating Income-to-Sales), OI/PPE (Operating Income-to-
PPE10), NI/S (Net Income-to-Sales), ROA (Return on Assets) and ROE (Return on
Equity);
b) Operating Efficiency: S/PPE (Sales-to-PPE) and OC/S (Operating Costs-to-Sales);
c) Assets: Log(PPE)(Property Plants and Equipments), I/S (Investment-to-Sales),
(I/PPE Investment-to-PPE);
d) Output: Log (Sales);
e) Shareholders: Payout Ratio;
f) Finance: LTD/E (Long-Term Debt-to-Equity) and CUR (Current Ratio);
g) Taxes: NT/S (Net Taxes-to-Sales).
10
Property, plant and equipment.
7IV. THE EMPIRICAL ANALYSIS
Two different approaches were adopted to examine changes in performance after privatization:
mean and median tests and a panel data analysis.
IV.1 MEAN AND MEDIAN TESTS
For the mean and median tests, two different methods were used. In the first (hereafter,
Method I), for each indicator a comparison is made between the mean and median values of the two
years following privatization with their values in the two years before privatization.11 The second
procedure (hereafter, Method II), fully uses the information in the data set, by comparing the mean
and medians of all years after privatization with their values in all years before privatization.
Along the period over which privatization took place, the Brazilian economy experienced
several cycles: initially recession (91-92), then a recovery (93-96), then a recession (97-99) and
another recovery (2000). Thus, change in performance could reflect cyclical movements of the
economy, rather than changes due to privatization. Suppose, for instance, that after privatization the
performance of the economy had improved. As a result, the privatized firms could show or not a
better performance, but not necessarily as result of privatization as such. To circumvent this problem,
we used for control a group of private companies. In this fashion, the performance of privatized
companies was adjusted by taking the difference between the indicator for the privatized enterprise
and the average of the indicator for the control group. Thus, we followed a procedure close to the one
used by La Porta & López-de-Silanes (1999) who adopted, in their words, industry-adjusted changes
in performance for the sample of privatized firms. Our adjustment, however, could not be done by
industry, as some of privatized enterprises do not have a corresponding match in the private sector.
This is the case, for instance, of CVRD (Companhia Vale do Rio Doce), a major mining company, the
Telecoms and many companies of the energy sector. Appendix A.2 details these procedures. Tables 3
to 6 present the results of the mean and median tests for changes in performance. Table 7 summarizes
these results in their signs and significance.
Profitability
In general, the results in Tables 3 to 6 indicate an improvement of profitability for privatized
companies. Considering operating income-to-PPE, return on assets (ROA), and return on equity
(ROE), performance after privatization improves regardless of the method adopted. The improvement
of operating income-to-PPE is evident, once the coefficient is always positive and significant, at least
at the ten percent level. The statistics for ROE and ROA also are always positive. In the case of ROE,
three of the four coefficients are significant, while for ROA only two reveal significance.
11
This procedure differs from that of La Porta & López-de-Silanes (1999) that used one fixed year for the period after
privatization. In the Mexican case, privatization was heavily in a few years. In the Brazilian case, privatization has
extended over more a decade. Therefore, a fixed year for comparison would be inadequate.
8Table 2
DESCRIPTION OF THE VARIABLES
CRITERIUM VARIABLE DESCRIPTION
The ratio of operating income to sales. Operating
income is equal to sales minus operating expenses,
Operating Income/Sales minus cost of sales, and minus depreciation. Sales are
equal to total value of products and services sold
minus sales returns and discounts.
The ratio of operating income to property, plant, and
Operating Income/PPE equipment, which comprise the value of a company’s
PROFITABILITY
fixed asset adjusted for inflation.
The ratio of net income to sales. Net income is equal
Net Income/Sales to operating income minus interest expenses and net
taxes paid.
ROA Ratio of net income to total assets.
ROE Ratio of net income to equity.
Log (Sales/PPE) Sales and PPE as defined above.
OPERATING
EFFICIENCY
Operating Costs/Sales Ratio of operating expenses to sales.
Log (PPE) PPE as defined above.
The ratio of investment to sales. Investment is the
ASSETS Investment/Sales value of expenditures to acquire property, equipment,
and other capital assets that produces revenue.
The ratio of investment to property, plant, and
Investment/PPE
equipment.
OUTPUT Log (Sales) Sales as defined above.
SHAREHOLDERS Payout Ratio Ratio of total dividends to net income.
Current The ratio of current assets to current liabilities.
FINANCE
LTD/Equity Ratio of long term debt to equity
The ratio of net taxes to sales. Net taxes are equal to
NET TAXES Net Taxes/Sales corporate income taxes paid net of direct subsidies or
tax credits received during the fiscal year.
9A little different picture appears when we consider operating income-to-sales. In absolute
terms, it improves after privatization. However, when we take it in comparison with the private sector,
the change becomes negative and significant at the one percent level (Table 4). Little can be said in
terms of net income-to-sale. The sign of the coefficients varies across methods and fails to present
statistical significance.
At the firm level, various reasons could account for results of this type. At this point, the
weakness of the method to investigate in detail the sources of variance becomes apparent, and this
underscore the importance of using a different approach to test explanatory variables other than
privatization, as will be done later in this section, when we will resort to a panel data analysis.
Operating Efficiency
The results presented in Tables 3 to 6 strongly supports the presumption of an improvement in
efficiency. In all tables we observe an increase in sales to PPE and a reduction in operating costs-to-
sales. In the case of sales-to-PPE, all the statistics are positive and significant, strongly suggesting that
privatized firms became more efficient in the use of their assets.
Regarding operating-costs-to-sales, all the statistics present a negative sign, while only one of
them lacks significance at the ten percent level. As illustrated in Table 3, the mean of the two years
after privatization is 0.251, while the mean for the two years before privatization was 0.375,
representing a reduction of 33%, thus providing evidence of a reduction in costs at the operational
level.
Assets and output
An observable effect of privatization is a reduction in sales. In all tables the statistics that test
for difference in average are negative (significant at the one percent level in three of them). There is a
decrease in sales even when compared to the performance of the private sector (Tables 3 and 5). We
also observe a decrease in PPE in absolute terms. In all the tables the statistics presents a negative
sign, even though it presents significance in only one table.
Apparently, privatization had a negative impact on investment-to-sales. In all the tables the
statistics present a negative sign, even though the coefficient is significant in only two tables (Tables 3
and 4). These results seem consistent with the increase in efficiency reported above. However, when
considering investment-to-PPE that reflects the rate of investment, there is no clear picture: none of
the statistics is significant and the sign changes across tables.
Finance and shareholders
With respect to the payout ratio, we did not obtained conclusive evidence. The sign of the
coefficient is consistently negative, although never significant. This could be due to the lack of
information once this variable could be calculated only for a reduced number of firms (45).12
12
This information was available only for listed companies.
10A clearer picture appears with the indicators of financial management. We observe an increase
in the current ratio, both in absolute terms and in comparison with the private firms in our control
group. The statistics for difference in average are consistently positive and significant. However, one
observes that the adjsted mean/median is negative, meaning that former SOEs, when compared with
the control group, still present lower short-term solvency. The improvement indicates that SOEs,
having the backing of the government, are less concerned with sound financial performance.
With respect to long term debt-to-equity (LTD-to-equity), we observe that in absolute terms
privatization has a positive impact: the coefficients in Tables 3 and 5 are positive and significant.
However, when compared with the performance of the private sector firms, a different picture
emerges, as the coefficients become negative (Tables 3 and 5). In any case, the mean values after
privatization in Tables 3 and 5 (-0,108 and –0,002, respectively) indicate that the leverage of former
SOE quickly converged to values observed in the private sector.
These results on financial structure are similar to those reported by La Porta and López-de-
Silanes(1999). This can be explained by the almost null probability of insolvency of state-owned
enterprises, once their credit status is guaranteed by the government. By loosing the government
backing, these firms were forced to adjust by decreasing their LTD-to-equity and increasing the
current ratios.
Net Taxes
Our results indicate a clear decrease in net taxes-to-sale. All the coefficients are negative and
significant at the one percent level. There are two reasons to find a clear and significant decrease in
net taxes after privatization in Brazil. This variable is defined as the difference between calculated
taxes and allowed deductions. With respect to the latter, as they do not come in the form of explicit
subsidies, it is worthwhile to describe the procedures in detail, in order to interpret the results more
accurately.
Three general categories of deductions apply: fiscal incentives, compensation for previous
losses, and tax credits. Losses incurred in one particular year may be deducted from income tax over
several years. This, in particular, affected companies highly dollar indebted when the devaluation of
the real occurred in February 1999. In fact, losses of this sort were also responsible for a decrease in
net taxes even for the control group in 2000.
With respect to tax credits, an important dimension is the legal treatment of the premium paid
on asset value in mergers and acquisitions. Brazilian corporate law recognizes the premium, and it
was regulated in the mid-nineties. The taking over company is allowed to constitute a reserve account
for the premium and amortize it over period of five to ten years. When the reason for the premium
paid over assets is based on expected future profits, the rebate is allowed in a period up to five years.
This benefit reaches mergers and acquisitions in general. Thus, both the overall private sector under
restructuring and the privatized companies have been beneficiaries of these rebates. The existence of
an explicit provision in declaring premiums in concessions as expected future profits facilitates the
use of this sort of tax credits in privatization. Therefore, there is a reasonable explanation for our
result that net taxes payments have decreased after privatization.
11Table 3
CHANGE IN PERFORMANCE: TESTS OF MEANS AND MEDIANS
Method I.a
Two years before privatization versus two years after, without adjustment
MEAN AND MEAN AND
CRITERIUM VARIABLE N MEDIAN MEDIAN Z TEST
BEFORE AFTER
0.037 0.042 0.536
Operating Income/Sales 66
0.072 0.108 0.523
67 0.092 0.141 3.556*
Operating Income/PPE 67
0.035 0.107 3.566*
65 0.000 -0.008 -0.595
PROFITABILITY Net Income/Sales 65
0.034 0.039 0.677
-0.860 0.008 0.291
ROA 70
0.014 0.011 -1.287
-1.152 0.046 0.662
ROE 70
0.019 0.039 0.862
-0.207 -0.002 5.371*
Log (Sales/PPE) 64
OPERATING -0.254 0.010 5.371*
EFFICIENCY 0.375 0.251 -2.631*
Operating Costs/Sales 58
0.200 0.196 -2.917*
6.497 6.105 -1.671***
Log (PPE) 68
5.967 5.900 -1.693***
0.295 -0.032 -2.550**
ASSETS Investment/Sales 54
0.158 0.093 -2.476**
0.115 0.094 -1.202
Investment/PPE 57
0.101 0.104 0.202
6.387 6.087 -3.206*
OUTPUT Log (Sales) 65
5.746 5.460 -3.206*
71.40 55.99 -0.089
SHAREHOLDERS Payout ratio 45
30.78 48.66 0.166
0.847 1.009 2.755*
Current 70
0.745 0.866 3.089*
FINANCE
0.636 0.701 2.506**
LTD/Equity 63
0.181 0.269 2.506**
0.024 -0.010 -3.834*
NET TAXES Net Taxes/Sales 65
0.017 0.007 -3.343*
*** Significant at the 1 percent level.
** Significant at the 5 percent level.
* Significant at the 10 percent level.
12Table 4
CHANGE IN PERFORMANCE: TESTS OF MEANS AND MEDIANS
Method I.b
Two years before privatization versus two years after, with adjustment
MEAN AND MEAN AND
CRITERIUM VARIABLE N MEDIAN MEDIAN Z TEST
BEFORE AFTER
0.097 -0.430 -2.944*
Operating Income/Sales 66
0.084 0.019 -2.944*
67 -0.092 0.141 3.556*
Operating Income/PPE 67
0.005 0.222 5.713*
65 -0.004 -0.105 -1.476
PROFITABILITY Net Income/Sales 65
0.020 0.012 -1.534
-0.870 -0.014 0.824
ROA 70
0.003 -0.012 -1.369
-1.194 0.025 1.768 ***
ROE 70
-0.030 0.021 1.698 ***
-0.464 -0.379 3.027 *
Log (Sales/PPE) 64
OPERATING -0.521 -0.362 3.217 *
EFFICIENCY 0.174 0.065 -1.837
Operating Costs/Sales 58
0.014 0.021 -0.809
1.350 0.909 -1.126
Log (PPE) 68
0.812 0.700 -1.156
0.223 -0.058 -1.887 **
ASSETS Investment/Sales 54
0.117 0.066 -1.795 ***
0.038 0.024 -0.774
Investment/PPE 57
0.026 0.039 0.264
0.916 0.468 -2.537**
OUTPUT Log (Sales) 65
0.281 0.337 -2.335**
0.309 -0.263 -0.229
SHAREHOLDERS Payout ratio 45
-28.62 -5.805 0.299
-0.510 -0.250 3.238*
Current 70
-0.605 -0.250 3.768*
FINANCE
0.254 0.108 -0.210
LTD/Equity 63
-0.142 -0.325 -0.021
0.018 -0.014 -3.578*
NET TAXES Net Taxes/Sales 65
0.005 0.003 -3.575*
*** Significant at the 1 percent level.
** Significant at the 5 percent level.
* Significant at the 10 percent level.
13Table 5
CHANGE IN PERFORMANCE: TESTS OF MEANS AND MEDIANS
Method II.a
All years before and after privatization, without adjustment
MEAN AND MEAN AND
CRITERIUM VARIABLE N MEDIAN MEDIAN Z TEST
BEFORE AFTER
-0.052 0.050 1.511
Operating Income/Sales 71
0.080 0.096 1.037
0.057 0.291 3.042*
Operating Income/PPE 70
0.045 0.097 3.408*
-0.067 -0.042 0.815
PROFITABILITY Net Income/Sales 68
0.010 0.039 0.889
-0.812 0.017 2.967*
ROA 73
0.003 0.026 2.311**
-1.109 0.021 2.258**
ROE 73
0.008 0.038 2.150**
-0.207 -0.036 5.441*
Log (Sales/PPE) 68
OPERATING -0.301 0.003 5.398*
EFFICIENCY 0.428 0.245 -3.138*
Operating Costs/Sales 64
0.255 0.207 -2.756*
6.926 5.924 -0.892
Log (PPE) 71
5.928 5.835 -0.774
0.191 0.038 -1.406
ASSETS Investment/Sales 61
0.202 0.1131 -1.157
-1.735 0.1181 0.288
Investment/PPE 62
0.085 0.098 0.168
6.820 5.894 -1.819***
OUTPUT Log (Sales) 69
5.728 5.880 1.609***
34.406 30.860 -0.138
SHAREHOLDERS Payout ratio 59
38.848 42.268 1.232
0.849 1.106 2.662*
Current 73
0.843 0.905 2.642*
FINANCE
0.529 0.576 3.192*
LTD/Equity 66
0.167 0.298 3.302*
0.015 0.009 -3.821*
NET TAXES Net Taxes/Sales 68
0.018 0.006 -4.296*
*** Significant at the 1 percent level.
** Significant at the 5 percent level.
* Significant at the 10 percent level.
14Table 6
CHANGE IN PERFORMANCE: TESTS OF MEANS AND MEDIANS
Method II.b
All years before and after privatization, with adjustment
MEAN AND MEAN AND
CRITERIUM VARIABLE N MEDIAN MEDIAN Z TEST
BEFORE AFTER
-0.050 -0.005 0.107
Operating Income/Sales 71
0.072 0.036 -0.241
-0.003 0.385 6.112*
Operating Income/PPE 70
-0.010 0.207 6.387*
-0.084 -0.064 0.693
PROFITABILITY Net Income/Sales 68
0.005 0.014 0.262
-0.831 -0.003 3.130*
ROA 73
-0.017 0.003 2.736*
-1.159 -0.012 3.236*
ROE 73
-0.44 0.014 3.223*
-0.501 -0.384 4.816*
Log (Sales/PPE) 68
OPERATING -0.578 -0.355 4.871*
EFFICIENCY 0.236 0.066 -3.199*
Operating Costs/Sales 64
0.090 0.025 -2.819*
1.895 0.716 -0.925
Log (PPE) 71
0.908 0.631 -1.310
0.098 0.011 -0.394
ASSETS Investment/Sales 61
0.123 0.086 -0.730
-1.840 0.055 1.385
Investment/PPE 62
0.022 0.029 1.072
1.437 0.294 -1.006
OUTPUT Log (Sales) 69
0.243 0.285 -0.852
0.082 -5.963 -0.731
SHAREHOLDERS Payout ratio 59
-29.35 -9.292 1.169
-0.526 -0.232 3.653*
Current 73
-0.503 -0.313 3.937*
FINANCE
0.233 -0.002 -2.086**
LTD/Equity 66
-0.107 -0.238 -2.286**
0.007 0.005 -3.173*
NET TAXES Net Taxes/Sales 68
0.007 0.002 -3.534 *
*** Significant at the 1 percent level.
** Significant at the 5 percent level.
* Significant at the 10 percent level.
15Table 7
SUMMARY OF TABLES 2 TO 5
TABLES
CRITERIUM INDICATOR 3 4 5 6
Operating Income/Sales + - + +
Operating Income/PPE + + + +
PROFITABILITY Net Income/Sales - - + +
ROA + + + +
ROE + + + +
Log (Sales/PPE) + + + +
OPERATING
EFFICIENCY
Operating Cost/Sales - - - -
Log (PPE) - - - -
ASSETS Investment/Sales - - - -
Investment/PPE - - + +
OUTPUT Log (Sales) - - - -
SHAREHOLDERS Payout - - - -
Current + + + +
FINANCE
LTD/Equity + - + -
NET TAXES Net Taxes/Sales - - - -
The shade means that the coefficient is significant at least at the ten percent level
16The results of this subsection support the view that privatization brought improvements in the
performance of the firms. This updated analysis, also extended to a larger number of firms, thus
confirms previous findings of the literature on the Brazilian privatization process. It also adds to
previous findings as it included a much needed comparison with the private sector over time.
However, as already pointed out the means e medians tests leave room for a more
comprehensive analysis that could fully use the variance of the data set, and allow for examining other
aspects of the of the privatization process. This will the focus of the next subsection.
IV.2. PANEL DATA ANALYSIS
Methodological aspects
We start with a brief description of the technique adopted in this subsection. It is a dynamic
version of a panel data analysis, which fits our scope to focus on individual heterogeneities over time,
in particular the discontinuous effect of privatization. This approach is an alternative to
generalizations of constant-intercept-and-slope models for panel data, which introduce dummy
variables to account for effects of variables that are specific to individual cross-sectional units, but
stay constant over time, together with the effects that are specific to each time period, but the same for
all cross-sectional units. The analysis is dynamic because the lagged value of the independent variable
is included in the model, and the panel is unbalanced as there are missing observations for some firms
in the data set.
Two approaches for representing the individual heterogeneities were initially considered.
Firstly, under the assumption that the error terms are a random variable independent and identically
distributed with mean zero and variance σu2 , and that the individual heterogeneities are treated as
constant over time, a fixed-effect panel data could be used. Secondly, admitting the same hypothesis
on the error terms, but allowing for the individual differences to be also random, a random-effect
panel data would be the case.
According to Baltagi (2000), the fixed effects model is an appropriate specification if we are
focusing on a specific set of firms, and our inference is restricted to their behavior. Alternatively, the
random effects model is an appropriate specification if we are drawing N individuals randomly from a
large population. For example, this is usually the case for household’s panel studies.
Given the nature of our sample, and after performing a Hausman specification test, the option
was for a fixed-effects panel data model. Our independent variables are the financial indicators for
each privatized firm spanning for a time period of 14 years as previously described, plus a large group
of private firms used as a control group. In this fashion, our sample includes all the Brazilian
companies for which we could collect information, though all the appropriate control variables were
introduced in order to deal with the issues of interest in this paper.
The independent variables are a set of variables defined as follows: PRIVATIZATION, a
dummy variable that takes the value one for the privatized firms right after the year of privatization,
and zero otherwise; TRADABLE, a dummy variable that takes the value one if the company is in a
tradable goods industry, and zero otherwise; REGULATION, a dummy variable that takes the value
one if price of the firm’s product is regulated by the government, and zero otherwise;
17SPLIT/MERGERS, a dummy variable that takes the value one if company has been privatized in
these forms, and zero otherwise; MINORITY CONTROL, a dummy variable that takes the value one
if the government owned only a minority participation in the pre-privatization phase, and zero
otherwise; and LISTED, a dummy variable that takes the value one if the privatized enterprise was
listed at the São Paulo Stock Exchange, and zero otherwise.
Two other variables were added to this set: PRIVATE MEAN, defined as the private sector
annual mean value for each financial indicator, and INDICATOR (-1), the one year lagged value
financial indicator used as dependent variable in each panel regression. The role of the private mean is
again to account for macroeconomic effects which affect the performance of the privatized firms, and
which must be controlled to isolate the effects of the dummy variables. The lagged values of the
indicators are included to account for continuous changes in the firms´ efficiency over time, not
captured by the discontinuous nature of the privatization dummy.
Panel estimation is carried out by a FGLS (Feasible Generalized Least Squares) estimator,
which better fits the case because there is a lagged independent variable in the regressions and fixed
effects are indeed likely to be correlated to the independent variable. The software also allows for a
correction involving heteroscedasticity and auto-correlation in the model.
The empirical results
To organize the presentation of the panel results, they are shown in Tables 8 to 9. In the
discussion of the tests of means and medians, we focused on the impact of privatization on the various
indicators of performance. We now resume this discussion, but adding to it the insights allowed by the
additional variables now included in the regression. Focusing on them, the discussion will follow the
summary of results presented in Table 1013.
Confirming the results of the previous subsection, privatization improves the performance of
the firms covered by the extended analysis, as revealed by the sign and significance of most
coefficients. For instance, the impact is positive on returns and current liquidity, and negative on costs
and long term debt.. The already explained negative impact on taxes is again revealed.
Other variable that reveals a clear impact, although with a few more exceptions, is the dummy
for listed companies, in general the larger ones. The results caution against the bias in selecting only
firms of this type in the studies of privatization.
The dummy for private mean is positive for all indicators, and only exceptionally not
significant, reflecting the impact of overall business and macroeconomics conditions. It also cautions
against another distortion of some studies on privatization, which do not isolate the impact of
privatization from the changes in these conditions over time.
The lagged variable replicates the same results of the private mean, confirming that the best
predictor of a firms performance is its past behavior, over which other variables have an influence, but
without which their effects would be difficult to distinguish.
13
Regarding the payout ratio, the data revealed insufficient to run a panel data regression.
18The different form of privatization assumed in the case of splits and mergers has no
distinguishable effect on performance. The minority control dummy is the one that show the lowest
number of significant coefficients. A reasonable explanation is that as the government had only a
minority participation in control, the performance of these firms were already closer to the standards
of the private sector.
Regarding the other dummy variables, no definite pattern arises from the results presented in
the tables, either in terms of the signs of the coefficients or their significance. Nevertheless, in the
case of the dummy for the tradables industries, the shakier results are indicative that the exposure to
competition, aggravated from 1994 to 1999 by an overvalued exchange rate, has had a negative effect
on performance. Mixed results are also observed in the case of regulation. As its effects might be
different in the various regulated industries, a detailed analysis would be required to investigate them.
In any case, as the effects come particularly in the form of regulated prices, an analysis of the
behavior of prices by industries will be presented in the next section.
19Table 8
CHANGE IN PERFORMANCE: PANEL DATA ANALYSIS – PART I
Independent
OI/S OI/PPE NI/S ROA ROE
Variables
0.164*** 0.048 0.043*** 0.027*** 0.065***
PRIVATIZATION 0.010 0.052 0.011 0.003 0.005
0.018*** 0.003 -0.012*** -0.008*** -0.011**
TRADABLE
0.004 0.016 0.004 0.003 0.005
0.015** -0.104*** 0.032*** 0.0015 0.007
REGULATION
0.007 0.033 0.005 0.0026 0.005
0.010 -0.068 -0.118*** -0.018*** 0.002
SPLIT/MERGERS
0.037 0.093 0.011 0.003 0.003
MINORITY -0.066*** -0.041 0.019 -0.011 0.008
CONTROL 0.017 0.110 0.016 0.007 0.012
0.143*** 0.044 0.023** 0.012*** -0.008***
LISTED
0.010 0.044 0.010 0.003 0.003
1.215*** 0.844 *** 0.829*** 0.764*** 0.693***
PRIVATE MEAN
0.055 0.039 0.018 0.038 0.038
0.197***
OI/S (-1)
0.020
0.106 ***
OI/PPE (-1)
0.023
0.182***
NI/S (-1)
0.057
0.035***
ROA (-1)
0.006
0.146***
ROE (-1)
0.012
-0.210*** 0.018 -0.037*** -0.014*** -0.040***
Constant
0.010 0.055 0.011 0.004 0.006
Observations 2083 2456 2003 2546 2193
Wald χ2 1305.48 508.12 473.66 595.81 977.84
P – Value 0.000 0.000 0.000 0.000 0.000
***Significant at 1% level.
**Significant at 5% level.
*Significant at 10% level.
20Table 9
CHANGE IN PERFORMANCE: PANEL DATA ANALYSIS – PART II
Independent
Log (S/PPE) OC/S Log (PPE) I/S I/PPE
Variables
0.172 *** -0.029*** -0.436*** 112285.2 *** 14077.64
PRIVATIZATION 0.016 0.004 0.067 28895.16 10442.17
0.016 -0.002 0.029 -16.327.3- -2818.90
TRADABLE
0.013 0.002 0.022 21694.31 7359.34
-0.051*** -0.016*** 0.043** 17205.49 -95.2571
REGULATION
0.011 0.002 0.021 19716.31 6865.42
-0.062*** 0.014 *** -0.174*** 21148.36 -1780.20
SPLIT/MERGERS
0.014 0.003 0.035 37107.87 11876.80
MINORITY -0.036 0.013 -0.029 -33113.13 -79780.81***
CONTROL 0.029 0.005 0.122 44645.07 16932.30
-0.012 -0.027*** 0.275*** 62861.83 *** 21476.41
LISTED
0.014 0.003 0.065 26406.96 9684.41
0.950 *** 0.140 *** 0.697*** 64790.05 3583.57
PRIVATE MEAN 0.038 0.039 0.031 86670.65 16381.40
0.679 ***
Log (S/PPE) (-1)
0.010
0.833 ***
OC/S (-1)
0.010
0.657***
Log (PPE (-1))
0.013
-0.00914
I/S (-1)
0.02313
-0.00464
I/PPE (-1)
0.03114
-0.412*** 0.035 *** -1.334*** -111775.5 -15519.36
Constant
0.023 0.008 0.183 34414.57 12143.50
Observations 2067 2173 2591 1838 2301
Wald χ2 10732.16 57557.35 6707 16.21 23.47
P – Value 0.000 0.000 0.000 0.0627 0.0028
***Significant at 1% level.
**Significant at 5% level.
*Significant at 10% level.
21Table 10
CHANGE IN PERFORMANCE: PANEL DATA ANALYSIS – PART III
Independent NT/S
Log (Sales) Current LTD/E
Variables Net Taxes
-0.362*** 0.137 *** -0.036* -0.008***
PRIVATIZATION 0.065 0.020 0.019 0.002
0.164 *** -0.031 ** -0.052*** 0.003***
TRADABLE
0.021 0.015 0.014 0.001
-0.054** -0.022 0.056*** -0.003***
REGULATION
0.022 0.014 0.015 0.001
0.073 * -0.089 *** -0.256*** -0.004**
SPLIT/MERGERS
0.040 0.019 0.023 0.002
MINORITY -0.084 0.165 *** -0.109*** 0.030***
CONTROL 0.091 0.021 0.040 0.004
0.250 *** 0.017 0.073*** -0.004*
LISTED
0.063 0.019 0.017 0.002
0.713 *** 0.248 *** 0.846*** 1.009***
PRIVATE MEAN
0.036 0.043 0.030 0.072
0.594 ***
Log (S (-1))
0.016
0.720 ***
CUR (-1)
0.009
0.876***
LTD/E (-1)
0.030
0.032***
NT/S (-1)
0.010
-1.405*** -0.105 0.011 0.003
Constant
0.223 0.065 0.024 0.002
Observations 2129 2724 2560 1896
Wald χ2 3194.62 8008.28 1235.73 303.78
P – Value 0.000 0.000 0.000 0.000
***Significant at 1% level.
**Significant at 5% level.
*Significant at 10% level.
22Table 11
SUMMARY OF TABLES 7 TO 9
Independent Variable OI/ S OI/ NI/ S ROA ROE LOG OC/ S LOG I /S I /PPE LOG
PPE (S/PPE) (PPE) (S)
PRIVATIZATION + + + + + + - - + + -
TRADABLE + + - - - + - + - - +
REGULATION + - + + + - - + + - -
SPLIT/ + - - - + - + - - - +
MERGERS
MINORITY - - + - + - + - - - -
CONTROL
LISTED + + + + - - - + + + +
COMPANIES
PRIVATE + + + + + + + + + + +
MEAN
LAGGED + + + + + + + + + + +
VALUE
The shade means that the coefficients are significant at least at 10% level.V. COSTS AND OTHER BENEFITS OF THE PROGRAM
The improvement in the performance of the privatized firms shown in the previous section can
be viewed as a benefit, as it contributes to the efficiency of the economy as a whole. This section
addresses other benefits, as well as some costs of the program. It also helps to identify some sources
of the gains made by the privatized firms, in the form of reductions in employment and increases in
prices.
Employment
One of the weaknesses of Brazilian data is that there is no comprehensive, reliable and unified
record of the number of employees of the privatized companies before and after their sale. Financial
statements and annual reports, including those of listed firms, are not required to include information
on employment, and companies provide it at their own discretion.There are also no uniform
requirements for including payroll information in these reports and statements, which bundle wage
and salaries costs together with other operational costs.
Even when employment and payroll data are available, their analysis is handicapped for other
reasons. In Brazil, there are strong incentives for the adoption of outsourced services in several
occupations, ranging from security, cleaning, and maintenance, to accountancy, and even to blue-
collar and white collar workers in general. Outsourcing became a widespread practice to reduce labor
costs, as service providers are smaller firms and pay lower wages. In addition, one often finds workers
disguised as business owners to avoid the heavy taxation on wages and salaries.14 As most workers
prefer formal contracts with contracting firms, and unions also press for this and are more successful
with SOEs, very likely privatization would lead to an extension of the outsourcing. Thus, a reduction
in employment in a company would not necessarily mean a reduction in the jobs generated by its
activities along its chain of suppliers.15
Given this picture, first we will examine the employment effects at the industry level, for
which data from a different source are available. Then, for a limited number of former SOEs, we will
resort on employment data available from the files of Exame, a business magazine that also collects
data on the final statements and reports of the Brazilian firms, as well as employment data from the
same and other sources.
14
The incentives received a new push after new “social rights” were established by the Constitution of 1988, as detailed by
Fernandes (1998).
15
Pinheiro (2000), tackled both the direct and contracting out impact on employment, on the basis of questionnaires send
to the privatized firms by BNDES. In the first case, he found a 33% reduction in the total number of formal workers. In
the case of production workers, the reduction was 29.5%, an evidence that overstaffing was concentrated in the case of
white collar workers. In absolute numbers, he found that, excluding telecommunications, the total reduction was of 10,000
workers in relation to the year of privatization and 35,000 with respect to the year before, again showing the adjustment by
the SOEs before privatization. In the case of telecommunications, he found that 145,000 new jobs were created in the
firms contracted out by the industry to expand its services. This number might sound very high, but notice that in this
country of 170 million inhabitants, the number of fixed telephone lines increased from 9.6 per one hundred persons in
1996 to 21.4 in 2000, while the number of cellular phones raised from 1.6 to 12.9 per one hundred persons, an expansion
and maintenance that has required a lot of labor, particular in the case of fixed lines.In Brazil, the most important source of data on formal employment is RAIS (Annual Survey
of Social Data) from the Ministry of Labor and Employment. All organized firms and the government
are required to list their workers who have a formal contract together with various characteristics,
such as age, gender, years of education, length of service, wage or salary, and so forth. These dataset
allows identification only by groups of firms, as individual firms cannot be identified. This source has
consistent data for the period 1995 to 1999. Thus, it is not possible to observe the full effects of
privatization on employment in the industries where privatization occurred before 1995.
Table 12 shows data on employment for the most important privatized industries. In the public
utilities, privatization came later and in a less complete fashion in the electricity industry. One can see
that until 1997 the private sector was responsible for less than 5% of employment in this industry, less
than a third in water and sewage, a quarter of telecommunications and a fifth of gas distribution. By
1999, both in the telecommunications and gas distribution sectors the larger part of employment
moved to private companies. In electricity, water and sewage sectors, employment is still largely in
public enterprises, but with a significant mix.
Still in the case of electricity, Table 12 shows a clear reduction in employment following
privatization as revealed by the public/private mix of employment. The same holds for the minor gas
distribution sector, which covers only the limited network of gas pipelines, as most gas are distributed
in bottles by private companies. In the case of telecommunications, the impact in reducing
employment is less clear, one of the reasons being the fact that following privatization the services
provided by this sector expanded very rapidly. Worth mentioning is the case of the water and sewage
industry. Still largely in the hands of the government, and not expanding as fast as
telecommunications, its employment ranks high in stability among the industries shown in the table.
Note also the recovery of employment in petrochemicals and in iron and steel, showing that after
employment adjusts following privatization, the growth of investment and production generates new
jobs.
25Table 12
EMPLOYMENT IN SELECTED INDUSTRIES, BY PUBLIC/PRIVATE OWNERSHIP - 1995-19
Number of Employees as of December 31st
1995 1996 1997 1998
SECTOR Total Total Total Total
Public % Private % Public % Private % Public % Private % Public % Private %
39131 38060 31447 39955
Mining
18 82 18 82 1 99 1 99
14442 21546 16963 13923
Petroleum
76 24 82 18 72 28 62 38
6460 7145 8395 12563
Fertilizers
18 82 9 91 11 89 1 99
15739 14947 19018 26263
Petrochemicals
5 95 2 98 0 100 1 99
376220 369234 385064 429965
Iron & Steel
5 95 5 95 2 98 2 98
149100 128545 99871 111225
Electricity
97 3 97 3 95 5 64 36
3257 2640 1551 1763
Gas Distribution
92 8 89 11 83 17 60 40
135313 146791 159588 145375
Water & Sewage
68 32 72 28 66 34 66 34
107689 113126 117740 105284
Telecommunications
80 20 77 23 75 25 19 81
Source: Ministry of Labor and Employment (RAIS 1995)You can also read
