Professional Documents
Culture Documents
Bank Privatization in Developing
Bank Privatization in Developing
ABSTRACT
We examine how political, institutional, and economic factors are related to a country’s decision
to privatize state-owned banks. Using a comprehensive panel of 101 countries from 1982-2000,
we find that the determinants of this decision differ markedly between OECD and non-OECD
countries. In non-OECD countries, a bank privatization is more likely the lower the quality of the
nation’s banking sector, the more right-wing the government is, and the more accountable the
government is to its people. These findings are largely consistent with an expanding literature
documenting the importance of legal and political factors in the development of capital markets. In
contrast, poor fiscal conditions are the most important determinant of bank privatizations in OECD
countries.
Robert Nash
Babcock Graduate School of Management
Wake Forest University
Winston-Salem, NC 27109
Email: rob.nash@mba.wfu.edu
* Boehmer is at Texas A&M University, Nash is at Wake Forest University, and Netter is at the
University of Georgia
Bank Privatization in Developing and Developed Countries:
Cross-Sectional Evidence on the Impact of Economic and Political
Factors
1. Overview
Because state-owned banks (SOBs) often provide governments with important policy tools,
privatization of the banking sector is often regarded as a sensitive issue. Previous evidence
suggests that privatization generally improves firm performance, raises revenue for privatizing
governments, and helps develop capital markets. At a macro level, other related evidence
suggests country-specific governance systems are important in determining how well capital
Our analysis uses a comprehensive dataset of all countries that have had any
privatization activity since 1982. In about half of these countries, the state has privatized at
least one bank. We seek to isolate why some countries have privatized banks and others
have not. We examine institutional, political, and banking-sector specific variables and ask
which factors are related to the decision to sell a SOB. We also examine factors related to
when a country privatizes banks. Most importantly, we split the sample into developing and
developed countries to see if there are systematic differences based on level of economic
development.
decision to privatize (and when to privatize) a state-owned bank. The economic factors
include the fiscal condition of the country, the efficiency and size of the nation’s banking
sector, the occurrence of a banking crisis, and the level of capital market development. The
1
See, e.g., Megginson and Netter (2001), Djankov and Murrell (2002), and Denis and McConnell (2003) for
comprehensive discussions of the privatization phenomena.
political variables include measures of government stability and political risk, the economic
orientation of the government, and the accountability of the government to the public.
We find that in non-OECD countries bank privatization is more likely the lower the
quality of the nation’s banking sector, more likely with a right-wing government, and more
likely the greater the government’s accountability to the people. These political factors are not
as important in OECD countries, where poor fiscal conditions are the primary determinant of
bank privatizations. These results are generally robust to different model specifications –
univariate tests, logistic regressions explaining whether a country had privatized a bank, and a
Overall, our findings suggest that the institutional environment is a very important
with an expanding literature that stresses the importance of the legal and political environment
in the development of capital markets (see Denis and McConnell (2003)). Our work is most
directly related to Bortolotti and Pinotti (2003) who, using panel data for 21 industrialized
countries, show that the probability of privatization is significantly associated with the
characteristics of a country’s political system.2 In the case of banks, we show that this holds in
One way to study the decision process that leads to bank privatizations is to focus in
detail on specific countries. This approach makes it easier to control for many institutional,
legal, social, and economic factors that are difficult to deal with in cross-sectional studies. In
the area of bank privatization, Clarke and Cull (1997, 2000, 2002) provide excellent examples
of detailed country analyses. For example, Clarke and Cull (2002) examine how the
implementation of the Argentinean Convertibility Plan in the early 1990s affected the political
and economic incentives for provincial governments to own banks. They find that poorly
2
Bortolotti and Pinotti (2003) find that the presence of a majoritarian political system increases the likelihood of
privatization. They argue that in consensual democracies, privatization proceeds more slowly.
2
performing banks were most likely to be privatized and overstaffed and larger banks were less
likely to be sold. In addition, higher levels of provincial unemployment and higher shares of
A second approach, the one we choose in this paper, is to study a broader cross-
section of countries. While it does not allow us to investigate detailed country-specific issues,
across countries limit our ability to draw general inferences (see Megginson and Netter (2001,
p. 346)). On the other hand, we can focus on systematic variations across countries and
identify the conditions that appear to be important for bank privatizations. We believe this
broader analysis of bank privatization can add important insights for policy makers and
regulators.
The remainder of our paper is organized as follows. Section 2 describes our sample.
Section 3 outlines our empirical methodology. In Section 4, we identify economic and political
characteristics that may affect the likelihood of SOB privatization and present our results.
possible drawing their sample from two principal sources: Privatization International, a
proprietary database that attempts to include privatizations from all nations (developing and
developed); and the World Bank Privatization database. In addition, they contacted various
governments to make sure their data were complete (see Megginson, Nash, Netter, and
3
Poulsen (2004) for further description of the data). These data provide details regarding each
privatizing transaction (e.g., offer size, offer date, method of sale, percent of capital sold).
Table 1 (Panel A) provides basic descriptive statistics of the privatization activity in our
sample countries. Including both developed and developing nations, 101 countries privatized
a state-owned firm from 1982-2000. Almost exactly half of the countries that conducted a
account for approximately 11% of the number of transactions and 10% of the value privatized.
SIPs) and by direct sale to a private investor (asset sales). Our SOB transactions are
approximately evenly divided between SIPs (47%) and asset sales (53%). The average
(median) SOB privatization raises $442 million ($156 million). These values are similar to the
findings of Verbrugge et al. (1999).3 Also consistent with Verbrugge et al. (1999), we find that
governments are hesitant to relinquish control of SOBs (i.e., privatize more than 50% of the
capital). The average (median) percentage of SOB capital privatized is 47.7% (41%).
our sample, 33 (41.3%) implemented a SOB privatization. A much higher proportion (almost
86%) of OECD countries privatized a SOB during our sample period. The SOB transaction
size is smaller for non-OECD countries, where the average (median) SOB transaction is $247
million ($85 million) vs. $710 ($376) in the OECD. The government does not typically
Table 2 presents the SOB privatizing transactions by country for each year of our
sample period. Panel A (non-OECD countries) reveals that SOB privatization was relatively
3
The average (median) from Verbrugge et al. (1999) is slightly higher because they focus exclusively on SOB
privatization through public share offerings. Megginson et al. (2004) show that share-issue privatizations are much
larger than asset sales.
4
infrequent until the early 1990s. SOB transactions were heaviest in the later 1990s with 68%
occurring between 1994 and 1999. Of the non-OECD countries, Mexico conducted the most
SOB privatization transactions (20 sales during our sample period).4 Panel B provides similar
details regarding SOB privatization in OECD countries. As in the non-OECD countries, SOB
privatization activity was greatest in the 1990s with 60% of the OECD transactions from 1995-
1999.
In Table 3, we present information about the timing, quantity, and size of privatization
activity in each of our sample countries. The table shows the period of time between a
country’s first privatization (regardless of industry) and its first SOB privatization. For the initial
transaction, most countries chose a SOE from an industry other than banking. The first SOB
privatization typically occurs 1-3 years later (median of 1 year for non-OECD; 2.5 years for
OECD). Next, the table presents the number of SOB transactions compared to all
privatizations in each country. Finally, Table 3 shows the dollar-value of SOB privatizations
relative to all privatizations. We see that the average non-OECD country conducts a smaller
number of SOB transactions (9% of total transactions for non-OECD vs. 19% for OECD).
Also, SOB privatizations in non-OECD countries represent a smaller relative amount of the
3. Methodology
privatize a SOB, examine factors impacting the timing of bank privatization, and determine
whether these factors differ in developing (non-OECD) and developed (OECD) countries. To
identify differences between OECD and non-OECD countries, we provide separate analyses
4
While now an OECD country, Mexico did not become a member until 1994. We specify OECD membership as of
1982 (the start date for our sample). Later in the paper, we perform a duration analysis that explicitly considers the
year in which a country became an OECD member.
5
for these two groups. In a first step, we use the entire multi-year panel of 101 countries
(discussed above) to contrast governments that privatize a SOB to those that do not. To
address when a bank is privatized, given the country’s general intent to open its economy, we
investigate countries that have completed a privatization (whether a bank or a firm from
another industry). Eighteen of the 101 countries in our sample sold a bank in their first
privatization deal. Thirty-three others initially privatized a firm from another industry and sold a
bank subsequently. Approximately half (50 countries) of our sample privatized non-bank
the bank privatization. Second, we use a logistic regression model to estimate each variable’s
marginal effects on the likelihood of a bank privatization. Third, we estimate a duration model
with time-varying covariates that allows us to investigate how long it takes a country to
privatize a SOB. The unconditional analysis is valuable, because it provides a general picture
of potential determinants and allows us to identify the equilibrium changes associated with the
the relative influence of each variable. Specifically, we model the yearly probability that a
country privatizes a bank, and use a logit link function to associate this probability with the
potential explanatory variables discussed above. The logistic analysis is, by construction, not
able to explicitly incorporate the time dimension. A country’s decision to privatize a bank is
likely to change over time, as economic, political, and financial conditions change. In a third
approach, we explicitly model this dependency. We measure the time between a country’s
first general privatization and its first SOB privatization, and examine which factors determine
The panel nature of our data makes it necessary to control for unobservable effects
that may be associated with countries or specific years. It makes little sense to control for
6
country effects, because our primary interest lies in the cross-sectional differences between
countries. We do control for unobservable time effects by including fixed effects for each year
except the most recent. This approach yields consistent estimates when time effects are
present, but it reduces the number of observations we can use. While the univariate tests are
based on all country-years, we have to eliminate years where no country privatized a bank
from the panel estimation. For example, there is no SOB privatization in 1987; this implies that
all 1987 observations for the dependent variable are equal to zero. As a result, the fixed time
estimates would not be unique. We believe this approach is more powerful and more
condense the panel to a cross-section (which ignores information and thus reduces power).
For each group (OECD and non-OECD), we retain all years where at least one country
privatized at least one bank. The estimation results are not sensitive to this particular
results when we omit time fixed effects and use the entire panel for estimation.
value of the associated Wald Χ2 statistics. In addition, to aid interpretation of the estimates, we
compute the odds ratio (the change in the probability of a bank privatization) for a one-
likelihood-ratio test that all coefficients are jointly equal to zero to gauge the explanatory
The last analysis seeks to explain the factors that impact when a country first
privatized a bank. An advantage of our data panel is that we can pinpoint each country’s first
privatization. Because we also know when it privatized its first bank, a natural question is how
long this process takes. This is important, because political considerations make the general
decision to begin privatizing state-owned firms distinctly different from the decision to privatize
7
a bank. Using a hazard-rate model, we can estimate the determinants of the length of this
privatizing its first bank) in each period, measured by the following hazard rate:
X that have a multiplicative effect on the hazard. The baseline hazard h0(t) is left unspecified
as in the Cox (1972) model. The coefficients β are estimated via partial maximum likelihood.
As in the logistic regression, we want to capture the fact that countries have a choice
between privatizing a SOB and doing nothing. However, we now model this decision
dynamically by allowing a country to revisit the decision in every year until a bank is sold.
Another advantage of the duration model is that we can explicitly address the right-censoring
in our sample. Censoring arises, because any country that has not yet privatized a bank (by
the end of our sample period) may still do so in the future. The matrix of covariates X consists
of the same explanatory variables used in the logistic regression. Our methodology allows all
variables to vary over time. This is important because changes in these variables may affect
a country’s choice.
Our independent variables measure the economic and political factors predicted to
influence a government’s decision of whether to privatize a SOB. Sources for the explanatory
variables include the Emerging Stock Markets Factbook, International Financial Statistics,
International Country Risk Guide (PRS Group), databases from the World Bank and others,
and various academic papers. Table 4 provides a detailed list of all independent variables and
8
Table 5 presents our univariate analysis comparing country-years with no SOB
transactions to those with SOB privatizations. We report both means and medians and test
whether they are equal between the two groups of countries (using a two-sample t-test and a
non-parametric Wilcoxon rank-sum test, respectively). The univariate results provide evidence
consistent with many of our hypotheses. We focus on the non-OECD countries but also
identify instances when the influence of a factor differs between the OECD and non-OECD
subsamples. We then discuss the logistic regressions and the duration model results.
specific to the banking sector as well as factors relating to the broader economy. In the
also hypothesize how each economic factor may affect the probability that a government will
study specific to bank privatizations, Verbrugge et al. (1999) find that governments appear to
structure SOB privatizations in order to maximize the proceeds from the sale. However,
Clarke and Cull (2002) find little evidence that fiscal needs impacted the likelihood of
pronounced during periods of fiscal crisis, we expect the probability of SOB privatization to
increase as a country’s deficit widens.5 We identify the annual budget deficit for each country
from the International Financial Statistics of the IMF. We find no evidence of differences in the
deficit between non-OECD countries that privatized a bank and those that did not. For OECD
5
Megginson et al. (1994) state that governments frequently use the proceeds from privatizations to offset budget
deficits.
9
countries, the median deficit is larger in absolute terms for countries that privatized a SOB.
Governments may also consider the efficiency of the banking sector when considering
the privatization of SOBs. Privatization frees the government from providing subsidies to loss-
making SOEs and encourages the restructuring of unprofitable firms. Caprio and Klingebiel
(2000) and Verbrugge et al. (1999) show that many state-owned banks exhibit poor financial
performance, possibly because the banks were used to make politically-motivated loans. If
likely in countries with less efficient banking sectors. Consistent with this hypothesis, Clarke
and Cull (2002) find strong evidence that poorly performing SOBs are more frequently
privatized.6
We employ two metrics to gauge the efficiency of a country’s banks. Our first measure
of bank quality is the ratio of bank loans to the public sector to total bank assets. In an earlier
paper, Clarke and Cull (1997) show that less efficient banks typically make more loans to
public entities. Further, they find that a larger relative amount of lending to the public sector
increases the probability that a SOB would be privatized. Therefore, if a higher proportion of
loans to public entities are a characteristic of lower efficiency, we expect a positive relation
Our second proxy for bank quality focuses on bank capital. Lower amounts of capital
should indicate a weaker and less efficient banking sector. Our bank capital ratio is the
difference between total bank assets and total bank deposits divided by total bank assets. We
predict a negative relation between our bank capital ratio and the likelihood of SOB
privatization.
6
Clarke and Cull (2002) report that the worse performing banks (measured by ratio of net worth to liabilities and
percentage of loans that are overdue) are more likely to be privatized.
10
The univariate results support our hypotheses regarding the relation between banking
efficiency and the likelihood of SOB privatization. In the comparison of means for non-OECD
countries, we find the expected negative association between the bank capital ratio and the
likelihood a government privatizes a SOB. This does not hold in the OECD countries.
However, inspection of the bank capital ratios reveals that the OECD banks have much more
capital (approximately 46% of total assets vs. approximately 30% for the non-OECD).
Therefore, lack of adequate capital is not as critical a concern in OECD countries. The data in
Table 5 also support our expectation that privatization is more likely if banks extend more
loans to government entities. The mean and median values of the ratio of loans to government
are higher for countries that privatized a bank than those that did not. This holds in non-
OECD countries and OECD countries. It is one of the few factors that has a similar effect on
Clarke and Cull (1997, 2000, 2002) and World Bank (1995) note that governments
facing an economic crisis, such as systemic bank failures, are more likely to privatize. A
systemic banking crisis occurs when much or all of a nation’s bank capital is exhausted.
Clarke and Cull (2002), Barth, Caprio, and Levine (2000), and Caprio and Klingebiel (1996)
document banking crises around the world during our sample period. Bank insolvencies have
government deficits. Therefore, to lessen the fiscal burden and reduce the probability of future
bank failures, governments may be more likely to privatize SOBs following a systemic banking
crisis. We expect that this effect should be stronger in developing countries. We use the data
from Barth et. al. (2000) and Caprio and Klingebiel (1996) to construct a variable that takes
the value 1 if a country had a bank crisis in that year and 0 otherwise.
11
Consistent with our prediction, the univariate results indicate that a banking crisis
increases the probability of a SOB privatization in non-OECD countries – the mean and
median values of the crisis variable are higher in the countries that privatized a bank. There is
Megginson et al. (2004) present evidence that governments use privatizations to spur
the growth of fledgling financial markets. Perotti and Oijen (2001), Subrahmanyam and
Titman (1999), and McLindon (1996) note that privatization through public share offerings can
jumpstart stock-market development and trigger gains in economic growth and efficiency.7 For
example, the privatization of large banks through share offerings should enhance the liquidity
of the nation’s equity market. With more shareholders, the market becomes more efficient.
This encourages more firms to go public, and the capital market experiences rapid growth.
Verbrugge et al. (1999) document that SOB privatizations have created hundreds of
thousands of new shareholders in countries around the world. The benefits from SOB
privatization should be most significant in the equity markets of developing nations. These
countries typically have less sophisticated capital markets (Demirguc-Kunt and Huizinga,
2000) and more state-owned banks (Barth et al., 2000). Therefore, if governments use
more likely in nations with less developed equity markets. We follow Megginson et al. (2004),
Booth et al. (2001), Demirguc-Kunt and Maksimovic (1999) and others and measure equity
market development with the ratio of annual equity value-traded to market capitalization.
7
Beck, Levine, and Loayza (2000), Rousseau and Wachtel (2000), Subrahmanyam and Titman (1999), and Rajan
and Zingales (1998) also analyze the relationship between development of a nation’s financial market and greater
economic efficiency and growth.
12
The univariate tests reveal a positive relation between equity market development and
the probability of SOB privatization in non OECD countries (the mean and median values are
higher in the countries that privatized a bank). This is inconsistent with the hypothesis that
governments use SOB privatizations to stimulate the growth of domestic stock markets.
However, this result is consistent with Verbrugge et al. (1999) who find that governments often
require a well-developed equity market in order to execute a larger bank SIP. Furthermore,
while SOB privatizations can create many new shareholders, other types of privatizations
typically have an even larger impact on stock market development. Megginson et al. (2004)
show that telecom privatizations are the largest offerings in a majority of countries and create
the most new shareholders. Therefore, governments seeking to develop equity markets may
larger or more well-known than the nation’s SOBs. In addition, the univariate results do not
capture the more subtle difference between using SIP and asset sale privatizations to develop
capital markets (see Megginson et al. (2004)). This may be important, because only SIP
Similar to using privatization to bolster its capital market, a government may also use
SOB privatization to enhance the country’s private banking sector. Beck et al. (1999) report
that state-owned banks are more dominant in developing countries. We also have anecdotal
evidence that state-owned banks are notorious for making politically-motivated loans, which
are often economically unsound. As a result, governments may choose to privatize SOBs to
improve access to private funding and reduce the state’s involvement in capital allocation. We
expect governments of countries with smaller private banking sectors to be more likely to
privatize SOBs. Demirguc-Kunt and Huizinga (2000) present two proxies to measure the size
of a nation’s private banking sector. First, the aggregate assets of private banks (as a
percentage of GDP) indicate the overall size of a nation’s private banking sector. Second, the
13
aggregate credit from private banks to the private sector (as a percentage of GDP) identifies
the quantity of credit provided by private banks. We expect that the probability of SOB
development.
The univariate results provide no evidence of a relation between the likelihood of SOB
privatization and the size of a nation’s banking sector. However, we will further test for this
In addition to economic characteristics, political and legal factors also influence the
privatization decision (see Denis and McConnell (2003) and Megginson and Netter (2001) for
a more complete discussion of the influence of political and legal factors on the workings of
capital markets). Political influences are especially important in SOB privatizations because
state-owned banks provide a significant source of political rents. Verbrugge et al. (1999) and
Shleifer and Vishny (1994) describe SOBs as a powerful political tool, frequently used to
reward supporters with high-wage jobs or favorable loans. Furthermore, Claessens and
Djankov (1998) and Bortolotti et al. (2002) note that governments use SOBs to channel funds
that cover the losses of other state-owned enterprises. Such subsidies are commonly
necessary because the other SOEs are also being used to garner political favor. We identify
the following political factors which may influence the likelihood of SOB privatization.8
Clarke and Cull (2002) argue that politicians choose to privatize when the political
benefits of privatization exceed the political costs. Privatization may subject a government to
substantial political cost because it is frequently associated with layoffs or other politically
8
The measures of political risk and legal factors are necessarily imperfect. In addition, they are often correlated
with each other. Thus, the results for these variables, especially in a univariate analysis, should be viewed with
some skepticism.
14
to accept the political risk involved in a large privatization. Bortolotti and Pinotti (2003) support
this conjecture with the finding that privatization is more likely in more stable regimes.
regimes with consensual rule, at least one group’s constituents will probably be averse to
privatization. This makes obtaining approval for privatization very difficult. Clarke and Cull
(2002) find weak evidence that privatization is less likely if another party holds veto power to
block the privatization at either the legislative or executive levels. Similarly, Bortolotti and
Pinotti (2003) confirm that privatization occurs more frequently in majoritarian governments
(where it is easier to gain consent for policy decisions and the executive enjoys greater
stability). Furthermore, Svensson (1998), Clague et al. (1996), and Knack and Keefer (1995)
contend that less stable governments may lack the ability to effectively enforce property and
expect a positive relation between government stability and the likelihood of SOB privatization.
Our government stability variable is from the International Country Risk Guide (PRS Group)
and measures each country’s overall political risk. This variable includes the government’s
The results reveal that in OECD countries greater political stability is associated with
SOB privatization (means and medians of the political risk variables are significantly lower in
countries that privatized a SOB). We find the opposite result for non-OECD countries.
The executive’s economic ideology may also impact the government’s likelihood of
privatizing a SOB. Beck, Clarke, Groff, Keefer, and Walsh (2001) identify the economic
(conservative, Christian democratic, or rightist parties) as those that favor less state control
over the economy and left-wing governments (communist, socialist, or leftist parties) as those
that exert more state control. Megginson et al. (2004) and Clarke and Cull (1997) use similar
15
measures of ideology and find that a government’s economic orientation figures significantly in
its privatization decisions. We follow Clarke and Cull (1997) and expect that a state-owned
We measure political orientation with an indicator variable (Right) which takes a value
of 1 if the executive is from a right-wing party (as specified by Beck et al. (2001)). In the
univariate results for this variable, we do not find any significant differences between
We note above that SOBs may be valuable to politicians as tools to build support
through the channeling of funds to favored constituents. However, a public official’s ability to
capture rents from state-bank ownership may be limited by an institutional structure that
provides accountability to voters. Shapiro and Willig (1990) note that a well-functioning
political system restricts the ability of politicians to pursue personal interests. Additionally,
Bortolotti and Pinotti (2003) contend that the threat from facing more competitive elections
keeps public officials “on their toes” and mitigates a politician’s willingness to exploit SOEs for
political or personal gain. Since politicians who are more accountable to voters may be less
willing to expropriate value from SOEs, these politicians should view privatization as a more
viable option.9 Therefore, greater accountability to voters, by limiting the ability to extract
political benefits from SOBs, should increase the likelihood of privatization. We measure the
politician’s degree of accountability to voters with the democratic accountability index from the
International Country Risk Guide (PRS Group). We predict a positive relation between
greater government accountability is associated with more SOB privatizations (for both mean
and median). However, we again find the opposite result for OECD countries.
9
Bortolotti and Pinotti (2003) find that the likelihood of privatization increases as politicians face more competitive
elections.
16
A complication is that the accountability of government to voters may be related to
other factors. For example, in the case of banks, public pressure on politicians to privatize
state-owned banks may be exacerbated by events such as a banking crisis. Clarke and Cull
(1997, 2002) identify a significant increase in the likelihood of SOB privatization in Argentina
following the Tequila Crisis. They further note that the crisis intensified public support for
The success of earlier privatizations may also increase public pressure calling for the
government’s sale of SOBs. As Megginson and Netter (2001) summarize, privatized firms
taxes. As a result, popular support for SOB privatization may strengthen as privatization
becomes more widespread. In their study of the privatization experience in Argentina, Clarke
and Cull (1997) find that government sales of SOBs become more likely over time.
Accordingly, we expect that the probability of SOB privatization increases as time passes
since the nation’s first privatization. To capture this effect, we compute the time since first
privatization for all country-years. For both non-OECD and OECD countries, the number is
significantly higher for countries with a bank privatization. This suggests that privatization in
In this section, we first report the results of logistic regressions that model a country’s
decision to privatize a SOB. Next, we estimate a duration model that explicitly considers that
10
Government involvement in banking (e.g., politically-motivated lending) contributed to the banking crisis in many
nations. See Caprio and Klingebiel (1996).
13
While counter to our prediction, this is understandable since, as noted by Megginson et al. (2004), privatizations
of telecoms (not banks) are more frequently used when governments seek to build capital markets.
17
this decision may vary with changes in the economic, financial, and political environment over
time.
Table 6 presents our logistic regression results for the non-OECD (Panel A) and
OECD (Panel B) subsamples. We provide two models that include variables we hypothesize
to affect the likelihood of SOB privatization. As with the univariate analysis, we focus on the
Our data indicate that the bank quality variables are critical factors in a government’s
decision to privatize SOBs in developing countries. For the non-OECD countries, we find that
both of our measures of banking sector efficiency significantly affect the probability of SOB
privatization. Lower bank capital ratios suggest less efficient banks. Consistent with our
predictions, there is a significant negative relation between the bank capital ratio and the
likelihood of SOB privatization. A larger proportion of lending to the public sector is another
indicator of poorly performing banks. Table 6 (Panel A) shows that SOB privatization is more
likely if banks make more loans to the public sector. Overall, these findings suggest that
governments in non-OECD countries are more likely to privatize a SOB when the quality of
the nation’s banking sector is poor. We find no evidence of a relation between bank quality
In the non-OECD countries, the government’s fiscal condition does not appear to affect
the likelihood of SOB privatization. Panel A indicates no significant relation between the
government’s deficit and its decision whether to privatize a SOB. However, in the OECD
countries (Panel B), the government’s deficit is a significant determinant of SOB privatization.
Additionally, the regressions reveal no significant relation between equity market development
(as measured by MktDevlp) and the probability of SOB privatization. Therefore, governments
18
do not appear to use SOB privatizations as instruments to expand the size and liquidity of
equity markets.13
The regressions also provide strong support for our hypotheses regarding the effect of
political variables. First, in the non-OECD countries, right-wing governments are significantly
more likely to privatize a SOB. Verbrugge et al. (1999) contend that state-owned banks play a
governments are expected to favor less state involvement in the financial sector. We do not
find a similar relation in the OECD countries—the right-wing proxy is insignificant in Panel B.
Our second political variable, democratic accountability (Dem. Account) is positive and
significant in the non-OECD regressions. This is consistent with the hypothesis that greater
accountability to the people limits a public official’s ability to use a SOB for political advantage.
As a result, a politician finds the privatization of SOBs more acceptable because the
associated opportunity cost is lower. The data support this hypothesis in the non-OECD
countries (but not in the OECD). The final political variable measures the impact of
government stability on the decision whether to privatize a SOB. Our measure of government
stability, the Political Risk Index, is not significant in any of the models.
In sum, we find significant differences in the factors that affect bank privatization in
non-OECD and OECD countries. Political and legal factors as well as the efficiency of the
unreported tests, we estimated separate models based on different sets of economic and
political variables (i.e., additional measures of the banking sector, government power, political
risk, etc.). Generally, we find similar magnitudes and significance levels for all coefficients.
19
4.3.2. Determinants of the time until the first bank privatization
Table 7 reports results of duration models measuring determinants of the time until a
country’s first bank privatization. We provide three different specifications of the Cox (1972)
duration model that differ in how we treat the difference between OECD and non-OECD
countries. First, when we estimate separate models, there are only 21 countries in the OECD
group and consequently this regression has little power. Second, we estimate a pooled model
that explicitly allows different baseline hazards for OECD and non-OECD countries. Third, and
several countries join the OECD after 1982, this variable should capture any difference in the
baseline hazard. Moreover, it goes beyond the separate models used in the logistic analysis,
because we can directly estimate how becoming an OECD member affects the likelihood of
bank privatization. As in the logistic model, we report hazard ratios instead of coefficients
because they are easier to interpret. Each ratio measures how much the hazard (i.e., the
instantaneous risk of exiting) increases for a unit change in the covariate. Hazard ratios
greater (less) than 1 imply that the covariate increases (decreases) the probability of exit.
We find that countries whose banks are less capitalized and extend more loans to the
government, and whose public officials are more accountable to the people privatize state-
owned banks faster. For example, model III shows that the odds ratio associated with the
bank capital ratio is 0.06. This suggests that a unit (100%) increase in bank capital ratios
reduces the probability of bank privatization (in any given year) by 94%. Similarly, a 10%
In one instance, the duration results appear to point in a different direction than
expected: Better capital market development is associated with a greater likelihood of a bank
privatization. However, there are conflicting forces in the relationship between privatization
and capital market development. On one hand (more important in developing countries),
20
privatization can be used to improve the capital markets. On the other hand (more important in
developed countries), it is easier to privatize a SOB by selling shares to the public in a country
Overall, the results are similar across the three models, except for the OECD-only
model that is based on just 21 cross-sectional observations (Model I, Table 7). The pooled
models show coefficients that are similar to those in the non-OECD model, which may
suggest that the OECD-only model fails to show significant results simply because of a lack of
power. On the other hand, the OECD-only logistic results are consistent with the OECD-only
duration results. Because the OECD-only logistic results are based on a panel of 276
observations, lack of power does not appear to be the main reason for the absence of political
determinants in developed countries. In sum, the duration results are largely consistent with
the logistic regressions; this attests to the robustness of both the logistic and the duration
models.14
5. Conclusion
This study examines the economic and political factors that affect whether and when
governments decide to privatize a state-owned bank. We are most concerned with bank
(OECD) countries to assess whether the importance of these factors differs across levels of
countries from 1982-2000), we find that both economic and political factors significantly
14
Both models derive their power from the cross-section of countries, and differ only in how the time dimension is
incorporated. The logistic model treats each period as independent (and we mitigate some of the resulting
econometric problems by including fixed time effects). In contrast, the duration model explicitly includes the time-
series variation in the independent variables and models the decision to privatize conditional on time.
21
In developing countries, the most important influences on a government’s decision to
privatize a SOB relate to political conditions and the quality of the nation’s banking sector.
Consistent with Clarke and Cull (1997), we find that SOB privatization is more likely when
banks are less efficient (have lower quality). This negative relation between bank efficiency
and the probability of privatization is much stronger in the developing countries than in the
OECD countries. We also find that the government’s ideological orientation significantly
affects the likelihood of bank privatization in non-OECD countries. Specifically, the data
indicate a significant positive association between right-wing governments and the probability
of SOB privatization. Additionally, governments that have greater accountability to voters are
factors for bank privatizations in developed (OECD) countries. Instead, it seems that other
types of political factors (such as public pressure due to widening budget deficits) influence an
Our results are robust to various estimation methods and are largely consistent with
those of Clarke and Cull (2002), who study bank privatizations in Argentina. We feel that our
study adds value by expanding the analysis across several countries and throughout many
years. We also contribute to the growing literature that underscores the importance of political
22
References
Barth, J., G. Caprio, and R. Levine, 2000, Banking systems around the globe: Do regulation
and ownership affect performance and stability? Working Paper, World Bank.
Beck, T., A. Demirguc-Kunt, and R. Levine, 2000, A new database on financial development
and structure, World Bank Economic Review 14, 597-605.
Beck, T., R. Levine, and N. Loayza, 2000, Finance and sources of growth, Journal of Financial
Economics 58, 261-300.
Beck, T., G. Clarke, A. Groff, P. Keefer, and P. Walsh, 2001, New tools in comparative
political economy: The database of political institutions, World Bank Economic Review
15, 165-176.
Booth, L., V. Aivazian, A. Demiguc-Kunt, and V. Maksimovic, 2001, Capital structure in
developing countries, Journal of Finance 56, 87-130.
Bortolotti, B., M. Fantini, and D. Siniscalco, 2002, Privatization around the world: Evidence
from panel data, Journal of Public Economics, forthcoming.
Bortolotti, B., and P. Pinotti, 2003, The political economy of privatization, Working Paper,
FEEM.
Caprio, G., and D. Klingebiel, 2000, Bank insolvencies: Cross-country experience, Working
Paper, World Bank.
Claessens, S., and S. Djankov, 1998, Politicians and firms in seven Central and Eastern
European countries, Working Paper, World Bank.
Clague, C., P. Keefer, S. Knack, and M. Olson, 1996, Property and contract rights in
autocracies and democracies, Journal of Economic Growth 1, 243-276.
Clarke, G. and R. Cull, 2002, Political and economic determinants of the likelihood of
privatizing Argentine public banks, Journal of Law and Economics 45, 165-197.
Clarke, G., and R. Cull, 2000, Bank privatization in Argentina: A model of political constraints
and different outcomes, Working Paper, World Bank.
Clarke, G., and R. Cull, 1997, The political economy of privatization: An empirical analysis of
bank privatization in Argentina, Working Paper, World Bank.
Cox, D. R., 1972, Regression models and life-tables, Journal of the Royal Statistical Society
34, 187–220.
Demirguc-Kunt, A., and H. Huizinga, 2000, Financial structure and bank profitability, Working
Paper, World Bank.
Demirguc-Kunt, A., and R. Levine, 1996, Stock market development and financial
intermediaries: Stylized facts, World Bank Economic Review 10, 291-321.
Demirguc-Kunt, A., and V. Maksimovic, 1999, Institutions, financial markets, and firm debt
maturity, Journal of Financial Economics 54, 295-336.
Denis, D. and J. McConnell, 2003, International corporate governance, Journal of Financial
and Quantitative Analysis 38, 1-36.
Djankov, S., and P. Murrell, 2002, Enterprise restructuring in transition: A quantitative survey,
Journal of Economic Literature 40, 739-792.
23
International Monetary Fund, 1995, International Financial Statistics Yearbook.
Knack, S., and P. Keefer, 1995, Institutions and economic performance: Cross-country tests
using alternative institutional measures, Economics and Politics 7, 207-227.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R.W. Vishny, 1997, Legal determinants of
external finance, Journal of Finance 52, 1131-1150.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R.W. Vishny, 1998, Law and finance,
Journal of Political Economy 106, 1113-1150.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R.W. Vishny, 1999, The quality of
government, Journal of Law, Economics, and Organization 15, 222-279.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R.W. Vishny, 2002, Investor protection and
corporate valuation, Journal of Finance 57, 1147-1170.
McLindon, M., 1996, Privatization and Capital Market Development: Strategies to Promote
Economic Growth (Praeger Publishers, Westport, CT.).
Megginson, W., R. Nash, J. Netter, and A. Poulsen, 2004, The choice of private versus public
markets: Evidence from privatizations, Journal of Finance, forthcoming.
Megginson, W., and J. Netter, 2001, From state to market: A survey of empirical studies on
privatization, Journal of Economic Literature 39, 321-389.
Megginson, W., R. Nash, and M. Van Randenborgh, 1994, The financial and operating
performance of newly-privatized firms: An international empirical analysis, Journal of
Finance 49, 403-452.
Perotti, E., and P. Oijen, 2001, Privatization, stock market development, and political risk,
Journal of International Money and Finance 20, 43-69.
PRS Group, Inc., 2002, http://www.countrydata.com (PRS Group, Inc, East Syracuse, NY).
Rajan, R., and L. Zingales, 1998, Financial dependence and growth, American Economic
Review 88, 559-586.
Rousseau, P., and P. Wachtel, 2000, Equity markets and growth: Cross-country evidence on
timing and outcomes, 1980-1995, Journal of Banking and Finance 24, 1933-1957.
Shapiro, C., and R. Willig, 1990, Economic rationales for the scope of privatization, in The
Political Economy of Public Sector Reform and Privatization.
Shleifer, A., and R. Vishny, 1994, Politicians and firms, Quarterly Journal of Economics 109,
995-1025.
Subrahmanyam, A., and S. Titman, 1999, The going public decision and the development of
financial markets, Journal of Finance 54, 1045-1082.
Svensson, J., 1998, Investment, property rights, and political instability: Theory and evidence,
European Economic Review 42, 1317-1341.
Verbrugge, J., W. Megginson, and W. Owens, 1999, State ownership and the financial
performance of privatized banks: An empirical analysis, Working Paper, University of
Georgia.
World Bank, 1995, Bureaucrats in Business: The Economics and Politics of Government
Ownership, Oxford University Press, Oxford.
24
Table 1
This table presents summary statistics for our sample of 101 countries which privatized firms from
1982-2000. SOB is state-owned bank. Other refers to a privatization involving an industry other
than banking. Data are from Privatization International and the World Bank. Transaction
amounts are in millions of US$ as of the date of the privatization.
Panel A presents summary statistics for the entire sample. Panel B separates the sample into
OECD and non-OECD countries. We identify OECD membership as of 1982.
Panel A
Panel B
Number of Countries 80 33 79
Number of Countries 21 18 21
This table presents the number of privatizing transactions involving state-owned banks (SOBs) by year for each country that privatized a SOB during 1982-2000.
Panel A presents data for non-OECD countries. We measure OECD membership as of 1982.
Country 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total
Argentina 1 1 1 2 5
Brazil 3 1 1 2 7
Cape Verde 1 1
Colombia 2 3 5
Egypt 1 4 1 1 7
Ghana 2 1 1 4
Guyana 1 1
India 1 1 1 1 4 8
Indonesia 1 1 2
Israel 1 4 1 1 4 3 1 1 16
Ivory Coast 1 1 2
Jamaica 1 1 2
Kenya 1 1 1 1 1 2 1 8
Korea 1 1 2
Kuwait 1 3 5 1 10
Lebanon 1 1
Malawi 1 1
Malta 1 1
Mexico 8 9 1 1 1 20
Morocco 1 5 2 8
Mozambique 1 1 2
Pakistan 1 2 3 6
Peru 1 2 1 1 1 1 7
Philippines 1 1 2 1 1 1 7
Singapore 1 1
Sri Lanka 1 1 2
Taiwan 1 1 1 3
Thailand 1 2 3
Trinidad & Tobago 1 1
Uganda 2 2 4
Uruguay 1 1
Venezuela 1 1 1 2 2 7
Zimbabwe 1 1
Total 0 0 0 0 2 0 1 2 3 11 15 9 13 11 23 34 13 12 7 156
Table 2 (continued)
This table presents the number of privatizing transactions involving state-owned banks (SOBs) by year for each country that privatized a SOB during 1982-2000.
Panel B presents data for OECD countries. We measure OECD membership as of 1982.
Country 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total
Australia 2 1 1 2 2 1 2 11
Austria 1 1 1 1 4
Belgium 1 2 1 1 1 1 7
Denmark 1 1
Finland 1 1
France 1 2 1 1 2 2 9
Germany 1 1 1 1 1 5
Greece 1 1 2 3 3 10
Iceland 1 3 1 5
Italy 1 1 1 1 1 1 2 1 1 5 2 2 19
Netherlands 1 1 2
New Zealand 1 1 1 1 4
Norway 1 1 2 2 6
Portugal 1 3 2 2 1 2 1 12
Spain 2 1 1 4
Sweden 1 1 1 1 4
Turkey 1 1 3 2 7
UK 1 1 1 3
Total 0 0 0 1 0 0 4 8 3 7 4 14 4 14 12 14 15 13 1 114
Table 3
This table presents data about the timing, quantity, and size of privatization activity in our 80 sample countries
from non-OECD countries. We identify OECD membership as of 1982. The year columns display the year of
each country’s first privatization (regardless of industry) and its first bank privatization, respectively. Transaction
amounts are in millions of US$ as of the date of the privatization.
This table presents data about the timing, quantity, and size of privatization activity in our 21
sample countries from OECD countries. We identify OECD membership as of 1982. The year
columns display the year of each country’s first privatization (regardless of industry) and its first
bank privatization, respectively. Transaction amounts are in millions of US$ as of the date of the
privatization.
This table presents the definitions and data sources for the independent variables used in our
empirical analysis.
Variable: Turnover
Definition: Total Equity Value Traded – in millions of current U.S. $
Source: Emerging Stock Markets Factbook
Variable: MktCap
Definition: Total Equity Market Capitalization – in millions of current U.S. $
Source: Emerging Stock Markets Factbook
Variable: MktDevlp
Definition: Turnover/MktCap
Source: Emerging Stock Markets Factbook
Variable: Right
Definition: Economic Orientation of Executive Branch of Government – 1 = right-wing executive
Source: Database of Political Institutions; Beck et al. (2001)
Variable: GNI
Definition: Gross National Income – per capita, PPP-adjusted (current international $)
Source: World Bank SIMA Database
Variable: GDP
Definition: Gross Domestic Product – in millions of current, local currency
Source: World Bank SIMA Database
Variable: Deficit
Definition: Government Deficit; millions of units of local currency
Source: International Financial Statistics
Variable: Crisis
Definition: 1 if country experienced banking crisis that year
Source: Barth, Caprio, and Levine (2000) and Caprio and Klingebiel (1996)
This table presents summary statistics for our independent variables. The sample includes 101 countries
that have privatized a company since 1982. We split the sample into two groups, according to whether a
country was an OECD member as of 1982. See Table 4 for the definitions of the independent variables.
Independent Number of Median for Median for p-value of Mean for Mean for p-value of
Variable country- years years with Wilcoxon years years with t-test for
year obs without SOB priv. test for without SOB priv. difference
SOB priv. difference SOB priv.
Independent Number of Median for Median for p-value of Mean for Mean for p-value of
Variable country- years years with Wilcoxon years years with t-test for
year obs without SOB priv. test for without SOB priv. difference
SOB priv. difference SOB priv.
This table contains the results of logistic panel regressions examining a country's decision to privatize a state-owned bank. The sample includes
101 countries that have privatized a company since 1982 (bank or non-bank). We split the sample into two groups, according to whether a country
was not an OECD member in 1982 (Panel A) or an OECD member (Panel B). For each group, we exclude all years where no country privatized a
bank.
The regressions include time fixed effects (coefficients not reported). The adjusted odds ratio represents the relative change in the probability that
a bank is privatized for a one-standard deviation change in the independent variable.
Variable Hypothesized Coefficient Odds Ratio p-value Coefficient Odds Ratio p-value
Relation Estimate Estimate
Country-years with 85 73
bank privatizations
Country-years without 502 365
bank privatizations
p-value for likelihood 0.00 0.00
ratio test
Table 6 (continued)
Variable Hypothesized Coefficient Odds Ratio p-value Coefficient Odds Ratio p-value
Relation Estimate Estimate
Country-years with 84 75
bank privatizations
Country-years without 192 171
bank privatizations
p-value for likelihood 0.00 0.00
ratio test
Table 7
The table contains the results of a duration model with time-varying covariates, estimated using a partial maximum likelihood Cox (1972) model. We model
the duration between a country's first privatization, whether bank or other industry, and its first bank privatization. The sample includes 101 countries that
have privatized a company since 1982 (bank or non-bank). The odds ratio represents the relative change in the instantaneous probability that a bank is
privatized in year t, conditional on not having sold one before, for a one-unit change in the independent variable.
Number of
countries with 29 17 46 46
bank privatizations
Number of
censored
observations (no 51 4 55 55
bank privatization
by 2000)
p-value for
likelihood ratio test (0.00) (0.38) (0.00) (0.00)