Session 1: The Nature and Scope of State Banking
James A. Hanson
6th Annual Financial Markets and Development Conference
The Role of State-Owned Financial Institutions:
Policy and Practice
April 26-27, 2004
For more information, see:
Public Sector Banks and their Transformation
James A. Hanson a
Senior Financial Policy Advisor
Financial Operations and Policy Department, the World Bank
The opinions in this paper are solely those of the author and may not reflect the
opinions of the World Bank, its Executive Directors, or the countries which they
represent. The author is grateful to Jerry Caprio, Ruth Neyens, and Michael Pomerleano
for comments. This paper reflects the presentations at a World Bank conference
“Transforming Public Sector Banks”, April 9-10, 2003, Washington D.C. The
presentations and some related papers are available
Public Sector Banks and their Transformation
Public sector banks’ performance is important. Public banks still dominate the
banking systems serving the majority of people in developing countries, despite the rash
of privatizations of the last 10 years. In 2002, public sector banks represented 60 percent
or more of the banking system’s assets in Algeria, Bangladesh, China, Egypt, Ethiopia,
India, Iran, and Vietnam.1 In Indonesia, public banks, including those under control of
the Indonesian Bank Restructuring Agency, held over 60 percent of the banking system’s
assets, up from about 45 percent before the East Asian crisis. In Brazil, despite closure,
conversion into agencies or privatization of most provincial banks, including the massive
State Bank of Sao Paulo in 2001, federal banks, including the federal development bank
(BNDES), held about 1/3 of bank assets and dominate lending for agriculture, housing
and longer term projects. In Argentina, most of the provincial banks were privatized after
1995, but the public sector Bank of the Nation, the Bank of the Province of Buenos Aires,
and the other remaining provincial banks still controlled over 30 percent of bank assets.
In the transition countries, the entry of private banks and the privatization or closure of
former state banks has reduced the role of public banks sharply since 1992, particularly in
the Baltics and Central Europe, but public sector banks remains important in Russia and
the Commonwealth Independent States (Sherif, Borish, and Gross (2003). Figure 1
indicates the importance of public sector banks in recent years.2 Since 2002, major
privatizations include the remaining re-nationalized banks in Mexico, 5 banks taken over
after the crisis in Indonesia, and the two of the three remaining state banks in Pakistan.
Improving public sector banks’ performance involves first understanding the
rationale for public sector banks (discussed in Section B), and how the hopes for public
sector banks’ performance compare with the actual outcomes (Section C). To improve
performance requires a further analysis of the interaction between the owners
(government), the public sector bankers, and the clients (Section D). Numerous attempts
have been made to improve public sector banks without much success but there are some
lessons to be learned that would tend to improve their performance by making their
incentives and governance more like private banks (Section E). Privatizing a public
sector bank is an obvious transformation but a number of issues need attention to make
privatization successful, particularly in unfavorable institutional environments (Section
F). Some other options also exist for transforming public sector banks when neither
reform nor privatization seems feasible—closure, or turning the bank into an agency or a
narrow bank (Section G). A brief summary is found in Section H.
Two data bases exist on the importance of public sector banks. La Porta, Lopez de Silanes and Schliefer,
2000 and 2002, refer to the share of government ownership (direct and indirect) in the 10 largest
commercial and development banks in 92 countries in 1970, 1985, and 1995. Data in Barth, Caprio, and
Levine, 2001a, 2001b,, updated in 2004, refer to the share of government ownership in all commercial
banks in 150 countries in 2002, as reported by the country’s central bank. The La Porta, Lopez de Silanes
and Schliefer data set tends to show a higher percentage of public sector ownership, as might be expected.
Figure 1 is based on Barth, et al, 2004, Sherif, Borish, and Gros, 2003 and World Bank estimates.
B. The Rationales for Public Sector Banks
Various overlapping political and economic rationales have been advanced for
public sector banks. Politically, public sector banks may be a response to the substantial
economic and political power of large private banks. Although private banks are
government chartered and regulated, officials may nonetheless consider private banks to
be abusing their power. As the U.S. president Andrew Jackson said in the speech
accompanying his 1832 veto of the rechartering of the (private) Second Bank of the U. S.,
“It is to be regretted that the rich and powerful too often bend the acts of government to
their selfish purposes.”3 More recently, as part of de-colonization, many newly-
independent African nations nationalized banks domiciled in the former colonial power.
One of the many reasons for India’s 1969 bank nationalization to reduce the power of
economic conglomerates and concentrated lending. In justifying Mexico’s 1982 bank
nationalization, president Lopez-Portillo blamed the banks (and the “dollar exporters”)
for the peso’s devaluation. Thus, to counter the economic and political power of private
banks, the government may create public sector banks or nationalize private banks, rather
than attempt to create competition among private banks. Of course, in small countries it
is difficult to create such competition, even by the threat of entry.
Second, the “state-led” approach to economic development that was prevalent for
many years naturally involved government control of such important, potential policy
instruments as banks. Lenin, just before the October 17 Revolution said, “Without big
banks, socialism would be impossible. The big banks are the ‘state apparatus’ which we
need to bring about socialism and which we take ready made from capitalism.” (quoted in
La Porta, Lopez de Silanes, and Shleifer, 2002, p. 266). Communist regimes in Eastern
Europe, China, and Cuba, nationalized banks after World War II.4
A more moderate, post-World War II version of this rationale was that
government ownership of firms in “strategic sectors” or “commanding heights” was
critical to development and these firms needed assured, low-cost funding by government
banks. Thus, countries that adopted socialist or planned economic regimes nationalized
private banks and/or set-up public sector banks. 5 For example, India’s 1969
nationalization of 14 major banks was mainly justified by the “need to control the
commanding heights of the economy and to meet progressively the needs of development
of the economy in conformity with national policy objectives” (Preamble of the Banking
Companies [Acquisition and Transfer of Undertakings] Act of 1969). The continued
large role of public sector banks in many countries probably represents an overhang of
these models of development.6
Remini, 1967, p. 83. The Second Bank of the United States bank was a private bank that had a monopoly
on carrying out government financial activities. Remini, 1967, discusses the political issues.
La Porta, Lopez de Silanes, and Shleifer (2002) find a statistical tendency for more public sector control
of banking in Communist and ex-Communist countries.
Gerschenkron (1962) was among the first to provide academic support for the view that public sector
banks were needed to provide funds for large scale industrialization and long-term credit.
See Sherif, Borish, and Gros (2003) for a discussion of how the Transition countries in Eastern Europe
and the Commonwealth of Independent States dealt with banks after 1991.
A third, oft-cited and related economic rationale for public sector banks has been
the allocation of credits to underserved groups, not only long-term industrial credit but
credit for agriculture, small businesses, housing and export finance. This rationale was
not necessarily part of the state-directed approach to development, but a response to
perceptions of failures in financial markets and political demands. The policy involved
both redirection of nationalized banks and creation of new banks. In many countries,
separate public sector development banks were set up to intermediate between foreign
lenders and users of long-term credit (often public sector enterprises) or SME borrowers.7
Housing banks often were also set up. For example, in India, yet another rationale for
nationalization of commercial banks was the lack of bank credit for agriculture and the
importance of rural money lenders, which had been highlighted in Reserve Bank of India
studies. In addition, financing agriculture was an important need signaled in India’s
Fourth Plan (1969), particularly with the newly begun Green Revolution increasing credit
demand to finance commercial inputs. Small businessmen were also considered
neglected by India’s formal credit system. Farmers and small businessmen were thought
to be able to use credit more productively than the traditional clients. Finally, deposit
mobilization became a major goal to finance development, and public sector banks were
urged to expand their branches.8 These issues became even more important with the
1971 “Abolish Poverty” campaign of then-Prime Minister Indira Gandhi.9 Of course,
India had long had development banks for long term credits and created other second tier
public sector institutions to provide banks with credit for agriculture and small scale
industry. In Africa, concerns about foreign banks that focused on traditional export
lending and neglected new industries and non-export agriculture were one of the
rationales for bank nationalizations. In Latin America, concerns about long-term credits
and agricultural credits led to the set-up of development banks and agricultural banks. In
the Transitioncountries, sectoral banks were often set up after 1992 out of the previous
mono-bank structure (Sherif, Borish, and Gross, 2003).
The World Bank made 350 loans to more than 140 development banks from the 1950s to 1984. The
development banks intermediated these and other funds by acting second-tier lenders through banks, first-
tier, direct lenders, or both (e.g., BNDES in Brazil). The World Bank’s lending initially focused on private
development banks, but lending through public sector development banks came to dominate the program.
In India between 1969 and 1979 bank offices of public sector commercial nearly tripled, rising from about
7,200 to 22,400; in addition, regional rural banks were created in the 1970s and they had 2400 offices by
the end of the decade. The growth of public sector bank offices continued in the 1980s, albeit more slowly.
By 1992, there were about 44,000 commercial offices and 14,700 regional rural bank offices. Burgess and
Pande, 2003, find an association between the expansion of bank offices in the social banking era, the
reduction of poverty, and the expansion of non-agricultural output.
Interestingly, in India, social control of the private commercial banks was tried before nationalization
through a National Credit Council modeled on France’s Bank Nationalization Act of 1945. The Indian
Council first met in 1968 and set credit targets that the banks met. However, the targets were only a
modest reallocation of credit and the government was concerned about the banks’ condition and the
permanence of their former management (Tandon, 1989, and Sen and Vaidya, 1997). In 1969, the 14
largest private commercial banks were nationalized; this, together with the existing public sector State
Bank of India, increased the public sector banks’ share of deposits to 86 percent. Although pressures
continued for increased credits to the underserved sectors, only in March 1979 were specific priority sector
requirements set (40 percent of loans of which 18 percent was for agriculture and 10 percent for the
“weaker sectors”). In 1980, 6 more commercial banks were nationalized, raising the public sector
commercial banks’ share to 92 percent of bank assets. In addition, India’s public sector and quasi-public
sector development banks were and are relatively large.
Thus, the third rationale argued that public sector banks were needed to meet
some perceived market failures, to be remedied by public bank loans (rather than
Treasury grants). The public sector banks were a substitute for priority sector lending by
private banks, or to ensure the implementation priority sector lending and extension of
branch networks. They were intended to change the allocation of credit within a market
system and their funding benefited from the political and economic power of the state.10
Yet, at the same time, the lack of credit also reflected the government and public
enterprises’ preemption of an increasing fraction of bank funds and the difficulties of
mobilizing deposits and allocating them an environment of repressed interest rates and
uncertain legal, political and economic conditions.11
Fourth, public sector control of banks has often arisen from government takeovers
of weak banks in the aftermath of crises. For example, the nationalization of the Indian
banks in 1969 may also have reflected concerns regarding the frequent bankruptcies of
the private banks. Numerous banks were taken over in the 1980s (Sheng, 1996, p.
21-23). During the 1990s, crises and failed privatizations led to nationalizations, or re-
nationalizations, at least temporarily, of banks in countries as diverse as the Czech
Republic, Hungary, Mexico, Uganda, and Indonesia.12
C. The Performance of Public Sector Banks
In general public sector banks have performed poorly, although a few public
sector banks have reasonably well. Public sector banks played a large role in Singapore
and the State Bank of Mauritius is recognized as one of the best commercial banks in that
country.13 The approach of Bank Rakyat Indonesia (BRI) to small credits is recognized
as one of the best in the world.14 Over 90 percent of BRI’s small loans continued to be
performing during the Asian crisis, although BRI’s corporate lending was always weak
and subject to the same massive corporate defaults as the other Indonesian banks.
Public sector banks typically are able to raise money at lower rates than private banks because of the
implicit state guarantee. Public sector banks also have benefited from directed credit to them.
La Porta, Lopez-Silanes, and Schleifer (2002) find a correlation between public sector banks and
countries that they characterize as having less investor protection, less developed financial markets and
more interventionist than common law countries. As is well known, high inflation countries tend to have
lower ratios of bank deposits to GDP and high inflation inherently makes long-term lending difficult.
Insolvent private Indonesian banks were managed by the Indonesian Bank Restructuring Agency
(founded in 1998). Almost all of the banks that IBRA managed had been re-privatized by end-2003.
The State Bank of Mauritius is owned by a number of public sector entities, not the government directly.
See, for example, J. Yaron, Benjamin, and Piprek, 1997 and M. Robinson, 2001 and 2002. BRI currently
has about 2.9 million small-scale loans that carry market-based interest rates covering the cost of funds and
expenses. About 30 percent of loans are under Indonesian Rupiah 2 million (about US$200). BRI has an
enviable record of loan recovery on small credits that it maintained during the 1997-1998 crisis. This
recovery probably reflects its close supervision of loans and its incentives to staff and operational units for
good loan performance, within the context of decentralized authority. BRI also has a good training
program. It has some 27 million deposit accounts. However, BRI’s performance on small scale lending
was not always good. In the 1970s under the BIMAS program, it followed the usual approach to providing
subsidized credits to support the Green Revolution. As in India, this activity helped spread the Green
Revolution but the credits often went to wealthier farmers; were linked to input packages that often were
unsuited to the borrowers’ needs; and often were not repaid. BRI’s current success reflects its 1984 shift to
the approach described above, made possible by the general financial liberalization in 1983. See, for
example, Robinson, 2002, and Cole and Slade, 1996.
Unfortunately, the performance of most public sector banks has been weak,
particularly in terms of large non-performing loans (NPLs).15 For example, in China,
recent official estimates suggest NPLs of around 24 percent in the four public sector
commercial banks, equivalent to over 30 percent of GDP; estimates of private analysts
are much higher. In Brazil, the estimated cost of recapitalizing the state (provincial
banks) included some $20 billion for the bank of Sao Paulo (Banespa) before it was
privatized; in 2001. In addition, the Brazilian finance ministry has estimated that the
recapitalization of the federal banks, when completed, may cost over $50 billion, mostly
for the recapitalization of the Federal Mortgage Bank (Caixa Economica Federal). In
Africa and Eastern Europe, a number of costly restructurings of public sector banks have
occurred, often more than once for the same bank.16 The 1980s also witnessed massive
NPLs in public sector banks (See Sheng, 1996, pp. 21-24). These problems are of course
not limited to developing countries; the cost of cleaning up France’s Credit Lyonnaise
has been estimated at more than $20 billion by the European Commission.
In countries where public and private banks co-exist, public banks typically have
much higher ratios of NPLs to total loans. For example, in Bangladesh recently, non-
performing loans in the public sector banks were over 40 percent of loans, about 3 times
the rate in private banks.17 In India, the ratio of NPLs to loans in the public sector banks
in 2000 (14 percent) was about 25 percent higher than in the “old” private sector banks
which had a similar clientele (Hanson, 2003). In Latin America, Argentina’s provincial
banks reported NPL ratios in 1991 of over 50 percent, more than five times those in the
largest private banks (World Bank, 2001, p. 132). In Indonesia prior to the East Asian
crisis, the ratio of NPLs to credit in the public sector banks was three to four times larger
than in the large local private sector banks; and about 20 percent larger than the small
private banks (World Bank, 1997). Of the gross cost of the crisis in the Indonesian
banking system (liquidity credits to banks and replacement of bad loans by government
bonds), roughly 54 percent was due to the public banks, though they accounted for less
than 45 percent of banks’ assets before the crisis.18
Public sector banks also seem to have a correlation with higher spreads and
overhead costs, in other words banking systems with a large share of public banks tend to
be less efficient. La Porta, Lopez de Silanes, and Shleifer (2002) show a statistically
significant, positive relation between spreads and overhead costs and the proportion of
government ownership in a banking system. Barth, Caprio, and Levine et. al., (2001a)
also show a positive correlation but the degree of statistical significance is lower.
Reasons for the public sector banks’ higher spreads may be their higher costs and higher
National figures on banks’ NPLs should be taken as indicators, rather than definitive figures given the
significant differences in national systems of bank regulation and supervision. In crises, NPLs typically
turn out to be much larger than previously reported figures.
For some African examples, see Box 1; for Eastern Europe, see Sherif , Borish, and Gross, 2003,. Zoli,
2001, and Tang, Zoli, and Klytchnikova, 2000.
These figures include special (development) banks which have by far the highest NPL ratios.
See Enoch, et al, 2001 for data on liquidity credits by banks and Indonesian Finance Ministry data for
figures on recapitalization bonds by bank. For purposes of comparison, the liquidity credit data are
converted into dollars by monthly exchange rates and the bonds are converted into dollars at the exchange
rate at the date of their issue.
costs of provisions of their higher NPLs, as is the case in India (Hanson, 2003).
Meanwhile, the non-public banks in the system may not compete business away, either
because they choose to lead a profitable, non-competitive life or because their ability to
compete is limited by government regulations protecting the public sector banks, for
example on branching or placement of treasury deposits.
Of course public sector banks performance may also be rated in terms of service
to underserved clients but the rhetoric and the reality of their clientele should be
distinguished. First, many public sector banks have tended to lend mainly to the public
sector—that was often a rationale for their founding—but the converse is that they often
lend less to the private sector, including SMEs. For example, in Argentina, the provincial
banks largely lent to the state governments. In Ghana, the state banks main lending was
to the government (Box 1). In India, the banking sector, which is dominated by public
sector banks, has typically lent nearly 40 percent of deposits to the government and
public sector enterprises (Hanson, 2003). Barth, Caprio, and Levine (2001b) show a
strong negative statistical relationship between state-owned bank assets and private sector
Second, the public sector banks are associated with a concentration of credit to
large borrowers. La Porta, Lopez de Silanes and Schliefer (2002) find that the larger the
share of public banks initially, the greater the share of bank lending to the largest 20
firms in the future. The links between state agricultural banks’ and large agricultural
borrowers has been documented in various studies, for example, Adams and Vogel
(1986), Adams et al (1984), Gonzalez-Vega (1984), Yaron (1994, 1997). A well known
example is Costa Rica in the mid 1970s. The public sector Banco Nacional’s interest rate
subsidy on agricultural credits was equal to about 4 percent of GDP and 20 percent of
agricultural value added and about 80 percent of the credit went to 10 percent of the
borrowers; the average subsidy on these loans alone would have put each recipient into
the upper 10 percent of the income distribution (World Bank, 1989).
The negative cross-country correlations between public sector banks and later
growth are not surprising, given the generally poor micro-performance of public sector
banks. As noted, Barth, Caprio, and Levine (2001b) show a strong negative statistical
relationship between state-owned bank assets and private sector credit, which is a key
factor in growth. La Porta, Lopez de Silanes, and Shleifer (2002) find the share of
government ownership of banks in 1970 is negatively correlated with per capita GDP
growth from 1965 to 1995 in a large cross-country sample, controlling for other standard
determinants of growth. The negative correlation is particularly strong for low-income
countries. Interestingly, this correlation between state banks and growth remains even
taking into account distortions such as inflation, importance of state owned enterprises
and an index of government intervention. La Porta, Lopez de Silanes, and Shleifer
(2002) also find a correlation between the size of public sector banking in 1970 and
slower productivity growth over the 1965-1990 period.
D. Explanations for Poor Performance
Why has public sector bank performance been poor and associated with slow
development? One obvious explanation is the difference in credit allocation between
public sector banks and private sector bank, which is the rationale for the existence of
public sector banks. Thus, unless private banks really didn’t recognize the loans with
best return/risk characteristics, public sector banks would do worse than private banks.
In fact, the results of credit allocation by public sector banks seems to have been worse,
not better—their non-performing loans are generally higher and their profits lower as
shown in Section C. A non-performing loan means that the use of the credit was not
sufficiently productive to service the loan and presumably alternative uses of the credit
would have been productive. Alternatively, the recipient of the loan might use the loan
proceeds for private profit that did not raise output and choose not to repay the loan.19
This means that the other clients of the lending institution or the taxpayers provided the
borrower, in effect, with a subsidy. In this case output of the borrower might grow as a
result of the loan, but other investors might be negatively affected because of higher taxes
and higher credit costs. Thus either the productivity of the credits of public sector banks
are worse than those of private banks, or that the recipients of their credits receive a
subsidy that must be covered by taxes elsewhere in the system.
The lower efficiency in public sector bank loans and higher NPLs on the micro-
level translate into bank crises and less growth at the macroeconomic level. Moreover, in
most countries there appears to be no offsetting distributional benefit from any
improvement in credit access or from credit subsidies from public sector bank lending.
In the view of La Porta, Lopez de Silanes, and Schliefer (2002) “[the evidence is]
consistent with the political view of government ownership of firms, including banks,
according to which ownership politicizes the resource allocation process and reduces
efficiency.” (pp. 290). For example, in India and Indonesia, even small credit allocations
were directed by government at some points in time.20 In this view, public sector banks,
like other public sector enterprises are instruments to transfer wealth to supporters and,
through bribes, to politicians and bureaucrats (Schleifer, 1998). No doubt part of the
public sector banks’ poor performance across countries reflects these issues.
Yet, even the best intentioned governments, aiming, for example, to improve
provision of credit to underserved sectors with public sector banks, face at least five
severe problems inherent in public sector banks . These problems hinder even the best
motivated efforts to improve public sector bank performance.
Measured output would not grow if credits were used for capital flight, which could yield good private
returns but measured negative national returns.
In India during the late 1980s, parliamentarian organized loan melas (fairs) at which banks simply gave
loans to farmers on their signature. In Indonesia during the 1970s and early 1980s, BRI was often directed
to provide credits to farmers by the Minister of Agriculture.
First, public sector banks suffer from multiple objectives, which makes it difficult
to set-up appropriate incentives. Private banks only face the difficult task of balancing
return and risk. But public sector banks face additional objectives—serve the
underserved, develop parts of the economy, provide employment, and meet the sporadic
demands from the state to correct financial problems. How can the government set-up
incentives for the manager to meet these diverse demands? How can the owner, the state,
judge how well the manager meets these objectives? Obviously trade-offs will be
necessary between the objective of prudent, profitable banking and, say, the provision of
access to farmers or small and medium industry. But these tradeoffs are usually are hard
to specify, especially in advance, and, as discussed below, information is weak. The
typical result is that managers are not really held accountable as long as they meet the
sporadic requests of government and major problems do not emerge. In this context, the
public bank managers, who are typically low paid, tend to avoid risks, ignore problems,
and focus on a long career in public service.
Moreover, one of the objectives often becomes providing employment in the
banks themselves and increasing staff welfare. Public sector banks are overstaffed
compared to private banks, typically. Overstaffing is often greater at lower levels—in
India the ratio of management employees (officers) to assets in public sector banks were
5 times foreign banks figures, while total employees are 7 times higher (Sarkar, 1999).
While salaries are usually low for the management, at the lower levels they are often high
relative to similarly qualified workers. The result is that wage costs are often fairly high
as a percentage of margins. In addition, public sector banks often offer their employees
low interest loans (See Hanson, 2001, for the case of India).
Second, information is lacking with which the owner, the state, could judge
performance and hold the management accountable. Information on all banks is obscure
to outsiders—the risk and return on assets may change very quickly as may funding costs
on market borrowings. All banks face difficulties in providing transparent, up-to date
information, particularly in developing economies. However, the public banks’ problems
are usually much worse. Public sector banks tend to lack good information systems,
because of widespread operations, weak national communications, and the lack of
information technology personnel as well as funds to hire them, let alone buy equipment.
The information problem is political, not just technical. The owner, the state,
often requests a public sector bank to make interventions outside the bank’s stated
objectives, for example to benefit well-connected supporters or correct problems that
have arisen. The owner often wants as little public information as possible about such
interventions. Hence it has only limited incentives to demand regular, up-to-date
transparent information on what a public sector bank is doing. The bank’s management
also has little incentive to ensure good information on which it can be judged, in order to
ensure it maintains a “comfortable” life. Thus, it is to both the government’s and the
banks’ management’s advantage to limit information on what the bank is doing and how
well it does it. The result, unfortunately, is a lack of good information on performance
and undiscovered problems that become obvious in a general crisis.
The information problem and conflicts of interest limit the supervision of public
sector banks. Information for supervisors, a political as well as a technical problem, is
lacking. If the banks’ managers themselves lack information, then supervisors have even
less information. Lacking information, supervisors are unable to provide a strong,
separate channel for analysis of a public sector bank’s condition.
Moreover, the government, faces a conflict of interest between its role as owner
of the bank—which may seek regulatory forbearance—and as the defender of taxpayers
interest. The government, lacking funds to recapitalize the banks, may apply pressure for
easy supervision or encourage regulatory forbearance, such as easing of income
recognition, provisioning, and restructuring norms. Supervisors may not even supervise
the public sector development banks or use a different procedure for them. Of course,
supervisors in developing countries tend to have little independence from the finance
ministry, which is the owner of the banks. But, even if supervisors were independent,
they would lack the political power to withdraw a weak public sector bank’s license.
The market also provides no discipline to public sector banks or demand
information about them. Public sector bank deposits carry an implicit government
guarantee. Thus depositors in and lenders to these banks can expect to be fully repaid,
irrespective of their activities, except where faith in the government’s credit has declined.
Third, even if public sector banks lent to the same clients as private banks, they
face a political difficulty in lending at market rates and in collecting loans and executing
collateral. This is true not only when they deal with the political powerful, but when they
deal with poor, underserved populations. Lending rates would have to be well above
prime rates to cover risks of default on long-term loans in an unstable macroeconomic
environment, or on loans to underserved clients who face high risks and about whom
little is known. But, such rates are typically politically impossible, so the public sector
banks set lending rates well below what the market would charge. Rates also tend to be
adjusted slowly to inflation. As a result, in high inflation situations, public sector banks’
capital is quickly eroded. This may explain many of the problems of public sector
development banks in the inflationary Latin American countries.21 In less inflationary
environments, an important issue in public sector banks poor performance is non-
performing loans. Here too, public sector banks face political problems. Collection is
limited either because the borrowers have strong political connections or because the
government would be seen as anti-poor by enforcing loan contracts and evicting
mortgagees or farmers.
Fourth, public banks often face a related problem—a culture of non-payment.
Borrowers may consider the loan a transfer by the government, not requiring payment. If
a noticeable number of borrowers either decide not to pay, or cannot, then borrowers in
general may decide that loan repayment is foolish, would put them at a competitive
disadvantage, etc., and then a general default may arise. Execution of collateral would
Indexing was used in some Latin American countries to limit the problem of keeping interest rates in line
with inflation. However, when relative prices changed sharply, indexed loan contracts, like standard loan
contracts, encountered problems.
tend to limit the culture of non-payment, but even if public banks could collect politically
they would probably find it difficult because of the weak legal framework in most
Finally, fifth, corruption, and capture by unintended beneficiaries, noted initially,
are problems that come on top of the foregoing problems that face well-meaning
governments. Bank loans at less-than-market rates are desirable, even more desirable if
pressures for repayment are limited. Such loans must be allocated by something other
than market forces and interest rates set by banks. Borrowers will use political pressures
and bribery. At the same time, the bankers, who receive civil service salaries, may seek
bribes as a way of dividing the “rents” on the loans; the bribe is often added into the
loan.22 As a result, lines of credit for small borrowers may often go to larger firms or
larger farmers, as discussed above.
E. Transforming Public Sector Banks
Public sector banks’ poor performance typically has led to bankruptcies followed
by attempts at “reform.” The reform usually begins with announcements that the banks
will be made as good as the best commercial banks. The typical reform package involves
a) New management;
b) Writing off or removal of bad loans and their replacement with government debt
or equity or by government takeover of an equal amount of the bank’s debt;23
c) Efforts to reduce costs including reducing staff and branch offices, particularly
those overseas, and, in some cases, mergers; and
d) Improving information technology and risk-management methodology.
These reform packages have typically been unsuccessful. First, the reforms are
often too small and often leave a substantial amount of bad debts on the books, but do not
provide a new approach to collect these debts. One study finds a need for multiple
recapitalizations in almost 25 percent of the cases of restructuring (Klingebiel and
Honohan, 2002). Box 1 discusses the repeated attempts at reform in three African
countries. Moreover, the write-offs and the easy treatment that the bad debtors get may
encourage further defaults.
Second and more importantly, the reforms do not address the fundamental
problems of state banks described above: the multiple objectives of the bank, the
government’s own ad hoc interventions in the bank, the lack of information on the bank’s
performance, the lack of accountability because of the multiple objectives, the
See World Bank, 1997, Box 5.13, for an example from Indonesia. Recapitalized banks are sometimes
given non-tradeable, low return assets as capital. The growth of recapitalization bonds in some countries
has been massive, see Hanson, 2003.
The bad loans that are removed may be placed in an asset management company or a treasury agency
charged with debt collection. In some cases the government may write off bank liabilities to the
government, guarantee loans, or assume the non performing debt of public enterprises to the bank. Tang,
Zoli, and Klychnikova, 2000, discuss the various approaches that have been used in Eastern Europe.
Enoch, Garcia, and Sundarajan, 1999, provide a general discussion with country examples.
interventions, and the desire to obscure what the banks actually do, as well as difficulties
in charging enough and collecting on loans, let alone any problems of corruption
To put it another way, substantial bad loans will almost certainly recur unless
changes occur in the basic character of state bank along the following lines:
a) Eliminate government interference in their lending,
b) Improved incentives for their management and staff to lend and collect well, and
an information system that permits accountability,
c) Interest rates that cover costs of lending to high-risk clients, and
d) Eliminate the culture of non-payment by the borrowers through enforcement of
Box 1. The Rise, Reprieve, and Fall of State Banks in Africa
“…Where the same strong interests that derailed earlier reforms still dominate a country's politics,
outcomes from bank privatization will tend to be disappointing … Most African countries opted to create at
least one large state bank after independence to support indigenous industries and state ventures and to
make banking services available for the broad population, including those in rural areas. In many countries,
these big state banks still dominate the banking sector and, after decades of politicized management and
soft budget constraints, have been difficult to restructure or privatize. The disappointing results from
restructuring and the problems in privatization can be seen from three [African] countries that attempted
banking reform programs during the 1990s: Ghana, Tanzania, and Uganda.
Ghana started economic reforms in the early 1980s after a politically unstable period of heavy state
involvement in the economy. The state owned three commercial banks, three development banks, and the
Cooperative Bank. There were also two foreign banks and a merchant bank.
All the state-owned banks were restructured and recapitalized under the financial reforms that started in
1987, with bad loans removed to an AMC. Management was improved through extensive technical
assistance. Both before and after restructuring, the primary function of the Ghanaian banks has been
funding the deficit of central government and public enterprises (this averaged 73 percent of domestic
credit in the 1990s). The very high T-Bill yields received by the banks helped offset the continued loan
losses from other lending.
Bank privatization has been a stop-go process, being held up, for example, by disagreement between the
privatization agency and external estimates of values on the price. With the program years behind
schedule, the government decided to sell some shares in two state commercial banks domestically even
before finding a strategic investor. This made it difficult subsequently to reduce the price to attract a
strategic investor. Eventually, in late 1996, the government dropped its requirement that the strategic buyer
should be a bank, and managed to sell the Social Security Bank to a consortium of foreign investment
funds. By 1998, this newly privatized bank had about 13 percent of total banking system assets.
The largest bank, Ghana Commercial Bank (GCB), continued to have problems even after the restructuring
of the late 1980s. With the failure of a planned sale in 1996 to a Malaysian manufacturing firm, it remains
government-controlled, with just 41 percent held by Ghanaians after the initial public offering (IPO). In
preparation for privatization in the mid-1990s, it was found that there were serious reconciliation problems
in the accounts and shortcomings in management, and some of the loss making branches had never been
closed. In 1997, the senior management of GCB had to be removed in the wake of a check fraud scandal.
Twelve banks had been nationalized in 1967 and merged into a dominant commercial bank, National Bank
of Commerce (NBC), which had a virtual monopoly for 25 years. The only other financial institutions were
a small [state] cooperative bank…and a few specialized state banks for housing.
By the mid-1980s, the NBC was insolvent, illiquid, and losing money at an alarming rate. Restructuring
moved a significant portion of the NPIs out of the bank, closed some loss-making branches, and retrenched
staff, but operating costs as a percentage of assets doubled and spreads became negative in 1992. The bank
was recapitalized in 1992, but as the losses continued to mount, restructuring intensified with an "action
plan" in 1994 that changed the board of directors, curtained lending and laid off further staff. However, the
salaries of the remaining staff were doubled by the new board of directors, thus offsetting the reductions in
costs. The benefits from removing bad loans to the AMC were short-lived. By 1994, 77 percent of the
remaining loans were nonperforming.
In 1995, another attempt to restructure failed. Finally, National Commercial Bank (NBC) was split in
November 1997 into two banks and a holding company. The NBC holding company took the non-banking
assets, for example, staff housing and the training center. The business bank, NBC-1997, took all lending
and 45 percent of the deposits, and a service bank took the remainder of the deposits. The National
Microfinance Bank was to provide basic depository services to the general population, and took the small
deposits but no lending. The decision to set up a microfinance bank that would keep the rural branch
network may have softened some of the political opposition to the privatization of the business bank. The
separation proved difficult. Poor financial and operational controls led to the need for significant provisions
on unreconciled balances, and there was a significant delay in producing financial statements after the split.
NBC-1997 was sold to the South African bank, Amalgamated Banks of South Africa Group (ABSA) in late
1999 with IFC participation. The microfinance bank …[could not be sold. With the support of donors, an
international development agency was contracted to manage and restructure the banks (Dressen, Dyer, and
Northrip, 2002). It] is now focusing on the provision of payments and savings services [and some
microcredits] in its 95 branches [and doing relatively well].
By the early 1990s, … the government had stakes in all nine commercial banks, and owned the largest two:
Uganda Commercial Bank (UCB), with about 50 percent of the market, and the Cooperative Bank. As of
late 1991, about one-third of the loans of UCB were non performing, and the negative net worth of the
bank was estimated at $24 million.
Timid restructuring efforts started. Loss-making branches were converted into agencies rather than being
closed. The AMC that was to take bad loans was not created until 1996 and, even then, there was a
significant lag in transferring bad loans. There was a performance agreement in 1994 between the Ministry
of Finance and the bank's board of directors, but the strategy pursued was to try to reduce the proportion of
NPLs by growing the loan portfolio. Bank supervisors did not monitor compliance. Every improvement in
profitability was temporary and losses continued to mount. By mid-1996, the financial position had
deteriorated so that its negative net worth tripled from earlier estimates.
While the government's intention was that the restructuring would culminate in privatization, management
of the UCB was actively opposed to sale. Eventually, after three years of unsuccessful attempts to
restructure the bank, it was agreed that a reputable merchant bank be selected to implement the sale, giving
the buyer greater freedom to define which assets and branches were to be purchased. Again, there was a
lag, and the merchant bank was finally hired in February 1996 and, at its request, top management was
finally changed in July 1996. Losses were mounting throughout the delay, and UCB was losing market
share. Audited financial statements for 1997 showed another fall in interest income, wiping out the core
profits advertised to investors six months earlier. With few expressions of interest, a sale agreement was
signed in late 1997 with a Malaysian industrial and real estate company, by December, 1998, however, the
deal had unraveled amid allegations of corruption.” Source: World Bank, 2001, Box. 3.3.
Without these changes in restructured banks, business continues undisturbed in its
basics and new bad loans may increase rapidly. In addition to the unchanged operations,
bad loans tend to increased because a) there is often political pressure to begin lending
again (because the bank crisis may be associated with the need for a stabilization program
that constrains other expansionary measures) and b) a public bank can raise deposits for
lending because the implicit government guarantee limits market discipline. As a result,
reformed public sector banks often return to bankruptcy in a few years, as seen in the
experience in Africa (Box 1) and the Transitioncountries (Sherif, Borish and Gross, 2003,
Zoli, 2001., and Tang, Zoli, and Klytchinikova, 2000).
The few cases where restructurings have been more successful support the
importance of the changes described above. 24 Specifically:
a) The government should define a simple objective—usually privatization over a
short time horizon—and gives the management “the power to say no” to requests
outside that objective. In other words, the government needs to avoid
intervention, except for failure to meet the well defined objective. If privatization
is the objective, but timing is left vague, then restructured banks tend to revert to
their earlier status.
b) The new management should be chosen for their capacity to manage. All of the
successful cases drew managers from reputable international banks, either a bank
itself, a twining arrangement with an international bank, or citizens who had
worked in such banks, though this alone is not sufficient for good results. In some
cases, the management (or twinning bank) was given an incentive for good
performance, based on the results of the privatizations, for example in Poland.
c) Information systems should be put in place to allow the new management to
monitor developments frequently with a short time lag, and to allow the finance
ministry and the central bank to evaluate the progress toward the objective. If
lending is aimed at targeted groups, then the information systems should also
make clear the success in reaching these beneficiaries and the full costs of doing
so, including interest rate subsidies and non-performance rates.
d) New lending should be constrained, especially large loans, to limit new non-
The bad loans typically were removed from the banks, although in the case of Poland
they were left in the banks to be resolved by the new management before privatization
(Kwalec and Kluza, 2003), and replaced by government recapitalization bonds.25
In India, where state banks continue to dominate, bank managers were given the
objective to reduce their non-performing loans after the small recapitalizations done in
1994 and 1995 (about 2 percent of GDP). Generally the banks succeeded in that goal,
through a combination of more careful lending, higher provisions than other banks (out of
higher margins), increasing their purchases of new government debt more than other
Some relatively successful reforms that include the points described below include Korea’ Seoul Bank
(Kang, 2003), Poland (Kwalec and Kluza, 2003), Pakistan (Sumro, 2003, Hussein, 2003), and AgBank in
Mongolia and NBC in Tanzania (Dressen, Dryer, and Northrip, 2002). In all these cases, the banks were
either privatized or transformed into a more narrow savings bank that focused on deposits and payments.
Placing bad loans in an asset management corporation (AMC) is likely to be expensive unless the AMC
has a strong incentive to collect and the legal framework is strong (See Klingebiel, 2000).
banks, and debt restructurings that were encouraged by regulatory changes and helped by
falling interest rates (Hanson 2003). Nonetheless, the non-performing loans before
provisioning remain about 10 percent of loans, remained undercapitalized for many years
and the state banks continue to have relatively high margins; high costs (despite savings
on wage costs brought about by the a self-financed voluntary retirement scheme in
2000/01); and profits on lending that are too low to maintain the ratio of capital to un-risk
weighted assets (Hanson, 2003, Reserve Bank of India, Trend and Progress in Banking,
various years). There has been no consolidation within the state banks, even among those
that experienced problems and required multiple capital injections, and information
technology has been a problem (Verma, 2000). Some of the private banks, including
some of the 9 licensed in 1994, also have experienced problems, as India’s information
and legal frameworks for lending remained weak. A new, large private bank, ICICI--
created recently through the merger of ICICI bank and an old private bank and then the
reverse merger of ICICI development bank into that bank—may put the public sector
banks under more pressure.
If in restructuring government seeks to make state banks as good as private banks,
the question is why not privatize them, rather than just restructure them. Presumably one
answer is that the government wishes to retain control over the credit allocation
instrument, to reward favored clients, and to keep the implicit taxes and transfers
provided by the public sector banks off the government budget. This suggests that
politicians considering privatization of public sector enterprise may apply a sort of cost-
benefit calculus, and privatization occurs when the cost of continued state ownership
exceeds the benefit of the ability to reward favored clients, provide employment, etc.
(World Bank, 1995). Of course, this calculus must be adjusted to take into account the
political support for general privatization of state-owned enterprises in the
Assessing the political calculus for privatization of banks is more complex than
for state enterprises, but the analysis seems hold. The most detailed studies relate to
Argentina; they find that provincially owned banks tended to become privatized when the
banks’ easy access to the central bank was ended, gains from seigniorage declined with
falling inflation, and runs developed on the banks in 1995 when the Tequila Crisis hit
Argentina; across provinces, badly performing banks were more likely to be privatized
while large overstaffed banks in provinces with high unemployment and large public
employment were less likely to be privatized (Clarke and Cull, 1999, 2002).27 There is
also some tentative support for this model in Brazil (Beck and Summerhill, 2003) and
across countries (Boehmer, Nash, and Netter, 2003). In Eastern Europe, the cost of serial
recapitalizations played a role (Bokros, 2003). In Mexico, the government sought a large
benefit from the sale of banks, which was needed to help finance the government (Haber,
and Kantor, 2003)
This section focuses on state bank privatization, it does not deal extensively with privatization of private
banks that have been taken over by the government during a crisis, except for the cases of Mexico and
On average, employment fell by over one-third in the privatized banks (Clarke and Cull, 2002).
The shift in national ideology appears to be another factor in privatization across
countries. Certainly ideology played an important role in the privatization of banks in the
Transitioneconomies (Table 1). The shift to a more market-based model of development
also probably contributed to the bank privatizations in Latin America. On the other hand,
less commitment to a market-based system and the continued strength of the post-
colonial elites appears to have limited privatizations in some African countries (Box 1).
Table 1. State Ownership of Banks in Transition Economies, 1993–2000
(percent of assets)
1993 1996 or 1997 2000
State Banks State Banks State Banks
Country Share of Banks Share of Banks Share of Banks
Albania 93.7 64.8
Armenia 3.2 2.6
Azerbaijan 77.6 60.4
Belarus 54.1 66.0
Bosnia and Herzegovina 55.4
Bulgaria 82.2 19.8
Croatia 36.2 5.7
Czech Republic 20.6 16.6 28.2
Estonia 25.7 6.6 0
Georgia 72.7 0 0
Hungary 74.4 16.3 8.6
Kazakhstan 4.6 28.4 1.9
Kyrgyz Republic 77.3 5.0 7.1
Latvia 24 6.9 2.9
Lithuania 53.6 54.0 38.9
Macedonia, FYR 0.0 1.1
Moldova 0.3 9.8
Poland 86.2 69.8 24.0
Romania 80.9 50.0
Russian Federation 37.0 41.9
Slovak Republic 54.2 49.1
Slovenia 40.7 42.2
Tajikistan 5.3 6.8
Uzbekistan 75.5 77.5
Yugoslavia 92.0 90.9
Unweighted Regional averages
Central and Eastern Europe 53.0 36.7
Baltics 22.5 13.9
CIS 31.9 26.0
Note : Countries without data are excluded from regional averages.
Sources : Sherif, et. al, 2003; Tang, et al, 2000; EBRD, Transition Report , 1999.
Despite the shift in ideology, governments in the Transitioncountries generally
tended to be privatized their banks after 1995 with a few exceptions (Table 1), a timing
that is later than many of the other privatizations. The relative slowness of privatization
may have reflected not just the desires of the governments to retain the instrument of
state banks—indeed, they typically retained a large interest—but macroeconomic issues
and the costliness of the initial recapitalization that would be needed in the banking
Among the larger countries in Eastern Europe, substantial bank privatization
before 1995 took place only in the Czech Republic (3 of the 4 state banks were included
in the voucher program; it was much less in Hungary (1 bank was partially privatized)
and in Poland (2 banks were partially privatized); and in all three cases the government
retained a significant ownership share in the partially privatized banks (25-30 percent).
In the Czech case, the banks were privatized through vouchers that were traded in the
stock market, with the government retaining a 30 percent share. The Czech banks,
dominated by government and owners of other voucher-privatized companies, made
loans to increasingly over-indebted companies that eventually went into default, requiring
another bank recapitalization.
Finally, other factors in bank privatizations seem to have been national
governance and opportunities for new markets, which increased the demand for bank
ownership from well-known foreign banks. In the Transitioncountries, once the
possibility of access to the European Union opened up there was a large inflow of foreign
banks. For example, in Bulgaria, foreign banks rose from 1 of 40 banks in 1994 to 25 of
35 in 2000, in the Czech Republic foreign banks rose from 13 of 55 banks in 1994 to 16
of 40 banks in 2000, in Hungary, they rose from 17 of 43 banks in 1994 to 30 of 38 banks
in 2000 and in Poland they rose from 11 of 82 banks in 1994 to 47 of 74 in 2000 (Sherif,
Borish, and Gross, 2003). Similar proportionate increases in the number of foreign banks
occurred in the Baltics. In Latin America, the rapid purchase of Mexico’s banks by well-
known international bankers in the second round of privatization probably reflected the
likelihood of better governance and less government intervention, as signaled by
NAFTA. On the other hand, concerns about relatively weak regulatory and legal
environments have limited the interest of well-known international banks in Africa and
much of Southeast Asia. However, South African banks have played an increasing role
in the southern part of Africa at the end of the 1990s.
Privatization, if done well, appears to yield significant gains. Again, the most
detailed study is of provincial banks in Argentina. Estimates suggest that the present
value of the cost of continuing to recapitalize the provincial banks at the usual rate was
2-4 times greater than the net cost of privatizing—taking the bad debts out of the bank,
recapitalizing and privatizing the bank for what was typically obtained for bank assets;
the estimated median saving was equal to about one-third of provincial government
expenditures (Clarke and Cull, 1999).28 Once privatized, the banks’ non-performing
loans dropped substantially and their credit allocation resembled the other private banks
more than the provincial banks that were not privatized (Clarke and Cull, 1999).
The range reflects differences in assumptions regarding the discount rate and rates of recapitalization and
Privatized banks also appear to be more efficient than state banks when the bank
is privatized to a strategic investor, particularly a reputable foreign investor, but
privatizations through the stock market, particularly when the government retains a
significant share, do not seem to generate much improvement in performance and may
even require further capital injections.29 These conclusions appear to be supported by a
comparison between the first rounds of privatizations in the Czech Republic (3 banks)
and Poland (2 banks) and the second rounds in these countries,30 in Argentina (Clarke and
Cull, 1999), in Brazil in a comparison between privatized and restructured banks (Beck
and Summerhill, 2003), in Nigeria (Beck, Cull, and Jerome, 2003), and in a 9 country
sample (Ochere, 2003). The partially privatized banks tended to have higher loan loss
provisions and higher labor costs than comparators (Ochere, 2003). Reflecting these
problems, the equity shares of the partially privatized banks have tended to do worse than
the market index in a 9 country sample (Ochere, 2003) and in India.
Partial privatizations may also cause problems if the intent is a later sale to a
strategic investor. The initial sale may later make it difficult to adjust the sale price to a
strategic investor, for example as occurred in Ghana (Box 1). The minority shareholders
may even be able to block the deal legally.
Privatization to a well-known, international bank acting as a strategic investor
seems to yield better results than where foreign participation is limited. This difference
may be a major factor accounting for the improved results in the second round of
privatizations in the Czech Republic and Poland and the good results so far in some other
privatizations in Eastern Europe (Bonin, Hasan, and Wachtel, 2003). The re-
privatizations in Mexico also produced much better results than the initial sales that were
limited to nationals (Haber, 2003, Haber and Kantor, 2003). Such banks bring modern
banking and risk management techniques. They also can provide some counterbalance to
the dominant political elite in small countries. Of course, even such banks may run into
problems in the environments of weak governance, weak information and weak legal
systems that prevail in many developing countries. However, such banks will usually
resolve the problems without demands for government support, since the failure to do so
would be costlier, in terms of their international reputation, than the costs of resolution.31
On the other hand, privatization to foreign investors without a reputation to protect can
often lead to problems—they are probably harder to supervise than national investors,
may simply pull out if conditions become bad, and may create problems if their
businesses run into trouble in their home country.
See the studies presented at the World Bank Conference on Bank Privatization, Nov. 20-21, 2003. Many
of the studies measure performance in terms of the degree to which banks deviate from an econometrically
estimated efficiency frontier, in addition, standard indicators of performance are explored.
As noted above the Czech government retained 30 percent of the shares in the first round privatizations,
through vouchers, and recapitalization was needed before the banks could be resold (Bonin, Hasan, and
Wachtel, 2003, and Kwalec and Kluza, 2003). In Poland’s first round of privatization, where the state
retained 30 percent of the shares, employees were sold 20 percent on preferential terms, and the rest was
divided into small and large investor lots, performance improved somewhat, but much less than in the
second round (Bonin, Hasan and Wachtel, 2003)
For example, in the recent Uruguay banking crisis, the full owned foreign banks were excluded from
central bank liquidity support.
Although bank privatization leads to benefits, doing it well has not been easy.
Delays and resistance occurs because of pressures from beneficiaries of the current
arrangements—clients, management, and staff—as well as nationalistic politics.
Governments and ministers may face legal or political difficulties in selling a bank for
less than the substantial amount that has put into it for restructuring, for example in
Ghana (Box 1) and Ecuador. The valuation problem is significant, since after
restructuring much of the bank’s assets may be government debt. Government debt in
external markets may well be selling at a large discount, and if a buyer wanted to take
country risk, it could simply buy that debt, rather than invest capital in a bank with all its
attendant problems of dealing with bad loans that have not been removed from the books
and a politically active staff accustomed to working in a public enterprise.
Above all, privatization has proved costly when the buyers proved to be unsound.
The worst case is the well-known first round of privatization in Mexico in 1991 to 1992.32
Foreign buyers were excluded and the buyers paid excessively for what appeared to be a
protected market and access to non-arms-length lending. The banks expanded credit
rapidly, not only to the industrial financial conglomerates to which they belonged (20
percent of all large loans went to insiders, see La Porta and Lopez de Silanes, 2003), but
also in mortgages. The quality of the capital in the privatized banks has also been called
into question, as it has been alleged that the buyers borrowed much of the funds with
which they bought the banks. The rising loans increasingly proved uncollectible; the
reported ratio of non-performing loans to total loans more than doubled between 1991
and 1993, reaching nearly 20 percent of loans.33 As the crisis deepened, related lending
nearly doubled; its terms were much easier and the defaults much higher than those of
unrelated borrowers (La Porta and Lopez de Silanes, 2003). The Mexican government
was forced to renationalized the bankrupt banks; it cleaned-up their balance sheets at an
estimated cost of over $70 billion (nearly 20 percent of GDP), and re-privatized them,
beginning in 1998, to international banks.
Such problems have developed in other privatizations. As discussed above, the
first rounds of partial privatizations in the Czech Republic, Poland and Hungary led to
problems, though not as big as the first round of recapitalizations; avoiding such costs
was probably a factor that contributed to the inclusion of foreign investors in the latter
rounds. In Africa, privatizations have often gone to nationals or foreign investors without
a reputation to protect, which have later created problems in countries such as
Mozambique (Mozes, 2003).
In sum, since bank privatization has large benefits, moving toward it fairly rapidly
is desirable, but a poor privatization can be costly. A desirable approach would probably
entail improving the informational and regulatory and supervisory environment while
setting a well-defined timetable for privatization. Improvements in the informational
environment would include frequent publication of data on banks, internationally audited
accounts, and setting up a system to publicize the recipients of the state banks’ loans and
See Haber and Kantor, 2003; Haber, 2003; and La Porta and Lopez-de-Silanes, 2003.
These figures follow international practice and include both the past-due interest (which was what was all
that was officially reported as non-performing before 1997) and the rolled-over principal of loans with past-
due interest. See Haber and Kantor, 2003, for the data.
subsidies on them. Subjecting the state banks to regulation and supervision similar to
private banks would provide the government with an independent source of information
regarding their condition. In addition, regulatory and supervisory standards should be
improved to ensure reasonable capital, income recognition and provisioning standards are
maintained, exposure to individual borrowers are limited, and strong prompt corrective
action is in place—issues that are very important after privatization to reduce the chance
of looting. The Argentine results suggest this approach is desirable—its regulatory and
supervisory system on the eve of privatization was one of the best in the developing
world.34 It may be desirable to hire international managers but this should not occur
much before the privatization occurs, as it is difficult for the state officials to supervise
the managers or to maintain incentives for a long time.
Recapitalization is probably necessary before privatization. The new owners
typically want a relatively clean balance sheet and do not want to sort out problems of the
previous managers.35 This is particularly true if the non-performing borrowers are state
enterprises, where collection may be politically difficult. Recapitalization should
probably not be done substantially before the privatization, lest the government decide
that privatization is not necessary (See for example, Box 1). The traditional approach to
recapitalization has been to recapitalize a bank by splitting it into a “good” bank with
deposit liabilities, performing loans, and government recapitalization bonds, and a “bad”
bank that receives that bad loans and focuses on asset recovery—the so-called Spanish
model that was used in the crisis in Spain in the early 1980s. However, in Poland some
success was achieved by keeping the non-performing loans in the banks and having new
bank managers, who knew their clients best, who could invoke relationships to improve
debt service, and who received good incentives for their performance, reduce the NPLs
before privatization (Kwalec and Kluza, 2003).
The actual privatization should be done to bidders who have passed a “fit and
proper” test, and foreign buyers should be allowed to participate. In Eastern Europe, it
was relatively easy to find good buyers, once it appeared that the European Union would
be expanded. However, recent attempts to sell banks in other parts of the world have
found it difficult to attract well-known international banks.36 When buyers are nationals
or less-well-known international banks, they may be under-capitalized and subject to
various pressures. Hence, good information on and good regulation and supervision of
the privatized banks is extremely important to limit new losses.
Minimum capital was 11.5 percent with risk weights based on market and credit risk, liquidity
requirements were 20 percent, nine of the top ten banks were well-known foreign banks, and information
on the banks was readily available (World Bank, 1998). The later problems in the banking system reflected
the macroeconomic problem of an over-indebted government, the freezing of deposits, and then the
judicially enforced unfreezing of deposits and the asymmetric conversion of deposits and loans when the
convertibility board was abolished, problems that even the best regulated and supervised banks could not
It is probably desirable to limit any “insurance” contracts against bad loans that are discovered. While
failure to provide insurance will limit the sale price, the experience with such contracts has proved costly in
some cases, notably the sale of Long Term Credit Bank (now Shinsei Bank) in Japan.
For example, in Indonesia, the sales of five banks that were taken over during the crisis received only one
bid from a well-known international bank.
G. Alternatives to Restructuring and Privatization
Three broad approaches have been used to deal with state banks when neither
restructuring nor privatization seems appropriate: closure and liquidation, conversion to
a government agency, or conversion of a state commercial bank to a savings bank
holding only government debt as assets (a narrow bank).
Closure of public banks has been used in the Transitioncountries (e.g., in
Yugoslavia, Romania (Bancorex), and Ukraine (Ukraina)), some of the provincial banks
in Brazil,37 in Africa (e.g., Benin) and in the case of numerous development banks around
the world.38 Closure has been done when the lending culture of the bank seemed well
beyond repair, and it was recognized that restructuring would simply lead to more bad
loans, a degree of realism that unfortunately was missing in many cases of restructuring.
The key point to remember with regard to closure is that banking is essential, a
particular bank is not (Andrews and Josefsson, 2003). The bank to be closed should be
thought of as a service provider and consideration given to how these services—deposit
taking, payments, loans, can be provided—an approach that unfortunately is complicated
by the demands of its public sector employees. Regarding deposits and payments,
closure is easiest in development banks, which often have not provided these and where
the source of finance was borrowing offshore or from domestic banks through directed
credit. The government could simply assume these liabilities, write-off uncollectible
loans, and place the remaining debts into asset recovery institutions and the courts. For
commercial or savings banks, the issue is more complex. Closure of commercial banks,
savings banks, and agricultural banks means depositors must be paid off or transferred.
Other banks may provide deposit and payments services to the former deposits through
their existing and new branches. Concerns are often raised that rural depositors may have
no good alternatives for protecting their savings and making and receiving payments and
remittances from their migrant family members. However, branches of the closed bank
can be sold or transferred to other intermediaries, including non-bank financial
institutions or micro-finance institutions, or narrow banks as discussed below.
Improvement in the payments systems can allow these institutions or post office banks to
handle remittances and other payments.
Another concern in closure is lending to the former borrowers, in particular, the
concern that small borrowers, may find it difficult to get credit. However, these fears are
often overstated. In particular, bad credit often drives out good credit. Elimination of
state banks that lend but do not collect can open the way for other institutions to engage
in small lending, particularly if information on borrowers and the legal framework are
improved (Box 2).
None of the provincial banks in Argentina were closed, because the central government provided
incentives to privatize them with its “Fondo Fiduciario”. Access to the Fundo meant that the provinces got
loans to cover their residual obligations after they split their banks into “good” and “bad” banks.
The membership of the Latin American Association of Financial Institutions for Economic Development
(ALIDE) declined from 171 in 1988 to 73 in 2003. The decline reflected the closure of many development
banks during the 1990s, for example in Peru and Argentina.
Box 2. Without State Banks Good Credit Can Drive Out Bad Credit for Small Borrowers
A standard criticism of closing or privatizing state banks is that it reduces small borrowers’ access to
credit. It is often argued that closure will wipe out lenders to small borrowers’ and large foreign owners
of banks will focus on only the best clients.
In fact, closure or privatization may not affect small-scale lending much, and may even increase it. The
rhetoric of small borrowers being served by state banks is often untrue, unfortunately. Various studies
suggest that large borrowers often get the majority of state banks’ loans, especially in the rural areas,
even from banks supposedly dedicated to small borrowers, as noted. Attempts to subsidize small
borrowing encourage diversion of credits from small borrowers to politically connected borrowers.
Bad credit also drives out good credit. State banks’ channeling of loans at below-market rates to small
borrowers and neglecting collection destroy the demand for sustainable rural and SME finance. Some
evidence suggests that micro credit lenders and private and foreign banks are reasonable suppliers of
credit to small borrowers when the role of state banks and subsidized credits is reduced. In Bangladesh,
small scale credit is done not through the state banks, but largely through Grameen type operations. In
India, foreign banks’ credit to SMEs has grown much faster than state banks. In Argentina, Chile, and
Peru, foreign banks do at least as well as local banks in providing small credits (Clarke, et al, 2004). The
role of foreign banks probably reflects their technology for managing small credits; in the U.S., the cost
of doing a small-scale loan has fallen below $15. Finally, private banks specializing in small credits have
expanded rapidly after the decline of state banks, for example, in Ecuador (CrediFe) and Peru
(MiBanco). Thus, removal of weak state bank credit from the market will create demand for sustainable
small scale lending and allow providers of it to expand.
The sustainable expansion of credit access can be supported by improved information on borrowers and
better creditors’ rights. Better information on borrowers not only cuts the costs of lending, but provides
an incentive for borrowers to repay in order to maintain an intangible asset—their credit rating. Attempts
to improve information on borrowers are growing in many developing countries. For example, 23
developing countries have established public credit bureaus since 1994, mostly Transitioncountries. In
many countries public and private credit bureaus are being improving. The legal and judicial framework,
particularly with regard to the definition and execution of collateral and bankruptcy, also is important to
credit access. Again, a viable threat to execute collateral, which state banks often cannot make even in
the presence of a good legal and judicial framework, provides an incentive for prompt debt service.
Finally, sustainable institutions for small-scale lending need to follow a few key principles:
a) Collect on loans; b) Charge enough to cover costs; c) Encourage staff to reduce costs, select borrowers
well, and collect loans; d) Provide deposit services which also reduces dependence on donors, and e)
Provide transparent information on accounts, clientele, and any subsidies (Yaron et al, 1997).
Institutions that follow these principles will contribute to sustainable access to credit; institutions that do
not will tend to attract non-target borrowers and reduce the sustained development of access to credit.
Although some state institutions have managed to follow these principles, notably Bank Rakyat
Indonesia’s Unit Desa program and Thailand’s Bank for Agriculture and Agricultural Cooperatives
(Yaron, et al, 1997), private institutions are likely to be more successful in their application than public
A second approach is converting a state bank or state bank offices into an agency
of the federal or state government. Such an agency can disburse funds but does not take
deposits. The advantage of this system is transparency. Private deposits are not used to
make what, in effect, are fiscal allocations, with “the bill”—the contingent liability of the
government—only coming later when “loans” are not paid. In the agency approach, the
payments to recipients are treated as expenditures and reviewed annually, as part of the
budget process. Of course the review depends upon the quality of those processes.
Finally, since no deposits are taken, there is less disruption of deposit taking by private
banks, and more possibility of market discipline for the other banks. The agency could,
of course, make loans, not expenditures. However, an agency may have even less ability
to collect it loans than a public sector bank, as Cull and Xu (2000) suggest was the case
A third approach is conversion of a state bank into a narrow savings bank, holding
only government debt as assets, an approach that limits the main problem of public sector
banks—future non-performing loans. This approach starts with the idea that the “good”
bank, that is left after the usual restructuring approach has a large share of public sector
debt and only a limited amount of loans. For example, in Indonesia in 2000, an extreme
case, the state banks had less than 20 percent of their assets in loans (net of provisions)
and nearly 70 percent in recapitalization bonds, cash and reserves, and central bank
securities; the largest (accounting for half of the state banks) having only 10 percent loans
(net of provisions) and about 75 percent in recapitalization bonds, cash and reserves, and
central bank securities. A number of banks in Transitioncountries also have relatively
low loan to asset ratios (Sherif, Borish, and Gross, 2003). In effect, such banks are
already nearly narrow banks; they could be converted into narrow banks by the sale or
transfer of the small volume of “good” loans to another bank, along with an equal amount
of deposits, which would leave the remaining deposits and the recapitalization bonds in a
Narrow banks have a long intellectual history and are favored by many
distinguished economists because they would provide safe payment services and deposits
to the public.39 Thus the conversion of a state bank into a narrow savings bank would
reduce concerns that small depositors, particularly in rural areas, would lose banking and
payments services—two of the three functions of banks. The narrow banks would,
however, not lend to private or public enterprises, the banking function in which state
banks have done poorly. Narrow banks have a number of advantages in addition to
satisfying the need for deposit and payments services—they would be less subject to
bank runs, help reduce monetary fluctuations (in currency plus demand deposits), and
protect the payments system, compared to standard, fractional-reserve, banks.40 The
Bossone (2001) provides a good review of the history of the idea. Narrow banking received an impetus
during the great depression of the 1930s, when the well-known economists Henry Simons, Lloyd Mints,
and Frank Knight (Chicago) and Irving Fisher (Yale) all argued for 100 percent reserves on demand
deposits to avoid bank panics and monetary fluctuations. Milton Friedman resurrected the “Chicago Plan”
in his Program for Monetary Stability (1959). More recently, concerns about the impact of rapid financial
innovation and the costs of banking crises have led to reconsideration of narrow banking, in, for example,
Bossone (2001), Tobin (1985), Mishkin (1999), Phillips (1995) and World Bank (2001).
narrow banks have minimal credit risk, except in the case of sovereign default,41 and
other risks can be limited by limiting their assets to government debt that is short-term or
has floating rates.42 Consequently, narrow banks’ deposits do not need guarantees or
deposit insurance as they are guaranteed by the assets. Supervision of the narrow banks
would be easy — simply checking that the investment policy is followed and deposits
equal government bond holdings. Supervision would be especially easy if the
government debt market were computerized and dematerialized (with ownership by book
entry), as is the international trend. Finally, it is likely that narrow banks that arise from
these circumstances will have plenty of assets to buy in the future, since countries where
state banks have had crises have large volumes of recapitalization bonds that are likely to
A deposit outflow from a narrow bank, or a run in the unlikely case that it occurred, is easily managed
and less complex and costly for the government than a run in a fractional-reserve, commercial bank. A
narrow bank could meet the outflow by selling government debt, assuming a liquid government debt
market exists. Even if such a market didn’t exist, the Central Bank/Government could buy the debt or
make a loan to the bank using the collateral of the government debt, and thereby avoid the risk of loss that
occurs with a lender of last resort or liquidity recycling facility. To the extent that the fall in deposits in
one narrow bank is deposited in another narrow bank, the receiving bank would demand an equal amount
of government debt, permitting the Central Bank/Government to unwind any intervention easily. Of
course, a run on a narrow bank in favor of foreign exchange would create a problem, but less than in a
fractional-reserve banking system that would generate a multiple reduction in banks’ assets and deposits to
adjust to the lower amount of bank reserves in the system. Of course, deposit shifts between narrow banks
and non-narrow banks do generate broad monetary fluctuations, because of their differential in reserve
requirements. Regarding the payments system, proponents of narrow banks typically have assumed,
implicitly, that payments services would be done by narrow banks and non-narrow banks would have to
carry out their payments through them, in order to protect the payments system. However, the increasing
use of collateralized Real Time Gross Settlements payments systems reduces the risks associated with
allowing non-narrow banks to participate in the payments system.
Recent developments in Argentina raise the risk of government default. In such circumstances, credit
risk is also likely to rise on loans to private parties, and a narrow bank would likely suffer less losses than a
commercial bank. Exactly how such a government default would affect depositors remains an open
question. The effective default on government bonds by inflation is a more common risk than an
announced default. Issuing bonds with floating interest rates to narrow banks would reduce this risk.
One possible risk is interest rate/term transformation risk, reflecting the possibility that interest rates
may rise. If a loan or government bond can be held to maturity and is paid at face value, then the nominal
risk is eliminated. However, even in that case, risks arise while the loan or government bond is maturing,
because the bank’s marked-to-market portfolio would be less than its deposits and higher deposit interest
rates would have to be paid to limit deposit outflow. These risks can be reduced by limiting narrow banks’
portfolio to short-term government bonds, or floating rate government debt. Of course, commercial banks
are not immune to such problems – typically commercial banks do some term transformation and higher
interest rates would require it to increase capital, plus create risks that debt servicing by the borrowers
might worsen and that the banks owners might opt for more speculative investments in order to recoup their
losses. Even if a narrow bank held long-term, fixed-interest government bonds, the term transformation
risk would be less for the country than with a commercial bank. The purchase or takeover of these
instruments at face value by the government/central bank would liquidate an equivalent amount of
government debt (at face value). The replacement of these instruments or the switch of deposits to another
narrow bank, either by depositors or as a result of government intervention in a narrow bank, would entail a
higher borrowing cost for the government, but the 100 percent reserve requirement would ensure that the
bank receiving the deposits demanded an equivalent amount of new, higher interest rate debt. Thus, the
potential cost to government of interest risks is less for narrow banks than for commercial banks, where
there could be losses on assets that are taken over. A second possible risk if fraud, a risk in any bank but
one that is lessened in narrow banks by the transparency of the balance sheet and the ease of supervision.
A third risk is broadening the activities of a narrow bank would create risks. Trading losses are of course
one possibility but trading could be limited. Other risks are the broadening of activities to other types of
grow because of interest costs.
In practice, few narrow banks exist except postal savings banks. The most well-
known example is the Japanese Postal Savings Bank. These and other postal saving
banks suggest that there is demand for a narrow bank, and that non-narrow banks, which
could offer better returns on deposits because of the higher rates of return on assets, will
not compete away all the demand for narrow bank deposits, particularly in rural areas.
Of course, postal banks and narrow banks could also benefit from information technology
improvements, not only to improve handling of accounts, but also remittances. Finally,
public sector savings banks have often been victims of their own success—their success
in deposit mobilization has often generated funds not only for financing central
government deficits but also state and local government infrastructure projects have low
rates of return and difficulties in repayment. To reduce the inflows, deposit rates could
be lowered, but this has also proved difficult to do politically. Of course, these problems
are a long way off in countries where there has been a major bank recapitalization and
much of the banking system’s assets consists of public sector debt.
Failed state banks have not consciously been turned into narrow banks, because
the government usually wants them to start lending again. In some recent cases they have
been turned into micro-credit institutions, which seem to be doing well, so far. Two
examples of this approach are Mongolia’s AgBank and Tanzania’s National
Microfinance Bank, which was the rural portion of the National Bank of Commerce
(Dressen, Dryer, and Northrip, 2002) . In both cases, the governments were concerned
about the political ramifications of closing a failed state bank that had attracted no
buyers, yet one that provided substantial banking services to much of the population. In
both cases, with the support of donors, an external agency was hired to manage the bank
with a contract that protected the bank from political interference. In both cases, the rural
branch network proved an asset, and improved technology allowed easier access to
deposits. Loans were kept small—in the case of the National Microfinance Bank the
average size was less than $400, or about 1.5 times national income. Loans grew sharply,
but arrears have been minimal. In both cases, the government is searching for a private
buyer. This approach represents another possible solution to the problem of maintaining
access to banking services when a rural state bank fails. The issue in turning failed state
banks into micro credit institutions is, of course, whether these institutions can continue
to collect well, or whether they will revert to the past high levels of non-performing loans
if a new wave of populism develops.
Public sector banks continue to dominate banking for the majority of people in
developing countries, although substantial privatization has occurred in Latin America
and the transition countries and to a lesser degree in Africa. Public sector banks’ micro
economic performance has been poor. Cross country evidence suggests that countries
lending which is where the problems started with state banks. An interesting issue here is the offering of
foreign currency deposits—presumably narrow banks could do this by also holding government short-term
foreign currency assets, but the question is whether these assets should be domestic government or foreign
with a large public sector bank presence have slower financial and economic
development, other things being equal.
Public sector banks’ poor performance reflects the multiple objectives facing the
banks, government intervention in credit decisions and collection, lack of incentives for
sound information and sound lending decisions, and the public’s non-payment culture
with respect to public sector banks, not to speak of corruption. Serial restructurings
typically have not worked because of the failure to address these problems. In the few
restructurings that have succeeded, even partially, governments have set well-defined
objectives that could be monitored, for example, privatization within a short time
horizon, hired managers with excellent reputations and paid them reasonably; improved
the banks information framework for monitoring the objectives; and avoided any
interventions or requests unrelated to the pre-set objectives.
Bank privatizations have reflected the high costs of the state bank approach and
the shift in ideology to a more market-based approach, again particularly in Latin
America and the countries. Bank privatization proved difficult in many cases,
particularly where foreign banks were excluded from the bidding for the banks and strong
private domestic groups took advantage of the privatizations to gain access to banks for
non-arms length loans to related companies that were not repaid. On the other hand,
privatization yielded substantial benefits when done well, especially in sales to well-
known, international banks that provided better risk management and were willing to
cover any problems developed. Concerns that such privatizations will reduce credit to
small borrowers may be overstated, in Latin America such banks have done reasonably
well in lending to small borrowers. Moreover, with the bad credit of state banks no
longer driving out good credit, countries have seen access to credit increase sustainably
through a variety of institutions.
National governance seems important in attracting well-known international
banks to bid in the sales of public sector bank. The transition countries were able to sell
many of their state banks to such banks, who sought a foothold in countries that would
join the European Union; the attraction of Mexico’s banks were increased by NAFTA.
Some Latin American and African countries have also benefited from the desire of these
banks to expand. However, in many areas, concerns about state intervention and legal
systems have limited purchasers’ interest in buying state banks, or even private banks.
Preparations for privatization, or even restructurings, would benefit from
strengthening of the regulatory and supervisory framework, the information framework
that can support market discipline, and the general legal framework. These elements are
important in limiting risks if privatizations cannot be made to well-known international
banks and such banks. They will also help in increasing access to credit. Where state
banks have proved costly but privatizations have proved difficult, countries have tried
alternatives such as a) closing state banks that had long histories of failures, b) converting
state banks into agencies of the government that disbursed funds but did not take deposits
and were subject to budget discipline, and conversions into variations on narrow banks
that provided deposit and payments services to under-served areas but limited assets to
government recapitalization bonds and micro credits. Technology improvements can
help to reduce costs in the latter institutions.
Adams, D., D. Graham, J. D. Von Pischke, eds, 1984. Undermining Rural Development
with Cheap Credit, Boulder, Colorado, Westview Press.
Adams, D. and R. Vogel, 1986. " Rural Financial Markets in Developing Countries :
Recent Controversies and Lessons" World Development #14, pp. 477-487.
Andrews, M. and M. Josefsson, 2003, What Happens after Supervisory Intervention?
Considering Bank Closure Options, IMF Working Paper #WP/03/17
Barth, J., G. Caprio, and R. Levine, 2001a, “Banking Systems Around the Globe: Do
Regulation and Ownership Affect Performance and Stability”, in F. Mishkin, ed.
Prudential Supervision, What Works and What Doesn’t, University of Chicago, Chicago,
Illinois, pp. 31-96.
__________, 2001b, The Regulation and Supervision of Banks Around the World: A New
Data Base, World Bank Policy Research Paper # 2588.
Beck, T. and W. Summerhill, 2003.“State Bank Transformation in Brazil—Choices and
Beck, T., R. Cull, and A. Jerome, 2003, “Bank Privatization and Performance: Empirical
Evidence from Nigeria”, processed, World Bank Conference on Bank Privatization, Nov.
Boehmer, E., R. Nash, and J. Netter, 2003, “Bank Privatization in Developed and
Developing Countries: Cross Sectional Evidence, on the Impact of Economic and
Bokros, L. “Transforming Public Sector Banks in Central and Eastern Europe”, ”,
presentation at the World Bank Conference on Transforming Public Sector Banks,
Privatization, April 9-10,
Bonin, J., I. Hasan, and P. Wachtel, 2003, “Privatization Matters, Bank Performance in
Transition Countries”, processed, World Bank Conference on Bank Privatization, Nov.
Bossone, B., 2001, Should Banks Be Narrowed?, IMF Working Paper #WP/01/159.
Burgess, R. and R. Pandi, 2003, “Do Rural Banks Matter? Evidence from the Indian
Social Banking Experiment”, London School of Economics, processed.