Lending of first versus lending of last resort: the Bulgarian Financial Crisis of 1996/1997 (1).

Berlemann, Michael ; Nenovsky, Nikolay

INTRODUCTION

During the last decade, several countries experienced some sort of financial crises. That is why the crisis problem is often seen as one of the dominant problems of the 1990s (Bordo et al., 2001, p. 53). It is thus not surprising that interest in financial crises recently increased considerably. In this paper, we analyse a crisis that has received relatively little attention in the (international) financial crisis literature: (2) the Bulgarian Financial Crisis of 1996/ 1997. (3) In spire of this, we suggest that the Bulgarian crisis is a good example of a twin crisis where a banking and a currency crisis occur almost simultaneously. Twin crises have been the focus of many third-generation crisis models. While most of these models, among them the moral hazard models of Dooley (2000) and Krugman (1998), were inspired by the Asian Crisis of 1997, recently some doubts have been expressed about how well these models explain important features of the crisis (sec eg Krugman, 1999). We argue that these models are nevertheless quite useful since they explain important features of crises like the one in Bulgaria in 1996/1997.

We argue that the Bulgarian crisis has some features of a fundamental crisis since Bulgaria's authorities followed an inconsistent policy mix by monetising fiscal losses while trying to stabilise the exchange rate. However, the severity of the Bulgarian crisis was primarily, but not only, due to substantial moral hazard behaviour in the banking sector. This behaviour was partially the result of the former political system and partially induced by the Bulgarian National Bank (BNB), which acted more like a tender of first resort than a lender of last resort (LOLR).

The Bulgarian example is also useful with respect to learning about the possibilities to overcome a financial crisis and the design of a financial system in a transition country that is less prone to financial crises. Bulgaria decided to introduce a second-generation currency board (CB) arrangement. Different from orthodox CB arrangements, the Bulgarian CB allows for a strictly limited LOLR function of the central bank. Thus, Bulgaria's experiences with the 1996/1997 crisis finally led to a switch from a system with a lender of first resort to a system with a strictly limited LOLR.

This paper is organised as follows. The second section briefly reviews the theoretical literature on financial crises. In the third section, we describe the development of the Bulgarian economy since 1990, which finally culminated in the crisis of 1996/1997. We also make an attempt at classifying the Bulgarian crisis with respect to the theoretical literature. We argue that the crisis can be explained by combining elements of first-generation models and third-generation moral hazard models in the tradition of Dooley (2000) and especially Krugman (1998). The fourth section describes the second-generation CB Bulgaria introduced in the aftermath of the crisis. Special attention is attached to the peculiarities of the Bulgarian CB arrangement that retained some flexibility in order to fulfil a strictly limited role as a LOLR. The paper closes with a summary of the main arguments and some conclusions.

THEORETICAL MODELS OF CURRENCY CRISES

There are at least two different types of financial crises: currency and banking crises. (4) It is standard now to distinguish between the so-called first-, second-and third-generation models of currency crises. To be able to classify the Bulgarian Crisis of 1996/1997, it is necessary to give a brief overview on the existing theoretical literature. We then focus on first- and especially third-generation models with moral hazard.

In first-generation models of Krugman (1979) and Flood and Garber (1984), countries face currency crises when governments finance their fiscal deficits by monetary expansion and at the same time try to peg their currency to enhance trade with major trade partners or to import low inflation from abroad. Since demand for domestic money does not change, individuals will exchange domestic currency against foreign currency assets. To hold the exchange rate constant, the central bank is forced to sell its foreign currency reserves until they are finally exhausted and the currency peg has to be abandoned. This kind of crisis is often referred to as a 'fundamental crisis' since the existence of a large fiscal deficit is a fundamental reason as to why it is impossible for a government to fix the exchange rate. In second-generation models with multiple equilibria in the tradition of Obstfeld (1994), currency devaluations arise for political reasons rather than being the unavoidable result of policy inconsistencies. A country will adopt a fixed exchange-rate regime when the arguments in favour of a currency peg (eg import of credibility in fighting inflation) outweigh those against the peg (eg the ability to pursue an independent monetary policy to react on asymmetric shocks). However, the cost-benefit relation might change in the course of time and might lead to the decision to abandon the currency peg. This might, for example, happen if the public lacks confidence in the government's (or central bank's) interest (or ability) to fix the exchange rate. While market expectations on the future exchange rate are obviously influenced by news on fundamentals, second-generation models also allow for currency crises caused by contagion or some sort of herding behaviour. (5)

Third-generation models typically consider the fact that currency crises and banking crises often (but not always) occur together. (6) While some third-generation models predict a currency crisis to occur as the result of a banking crisis, others imply a reverse causation. Another possibility is that a currency crisis will occur when actions are taken to prevent a banking crisis or vice versa. (7) Two basic types of third-generation models have evolved: (1) random withdrawal models (Chang and Velasco, 1998), based on an open-economy version of the Diamond and Dybvig (1983) model, and (2) moral hazard models (McKinnon and Pill, 1996; Dooley, 2000; Corsetti et al., 1999; Krugman, 1998), which focus on policy inconsistencies developed in first-generation models. We will focus solely on moral hazard models.

When the government guarantees the liabilities of financial intermediaries, a potential moral hazard problem arises since this creates a strong incentive for the financial intermediary to pursue a highly risky investment strategy. Within a class of investments with the same net present value and one good and one bad outcome, a protected intermediary will choose the one with the lowest probability of success (Freixas and Rochet, 1999, pp. 267-268). More generally, protected intermediaries will prefer investments with 'fat right tails' (Krugman, 1998, p. 4), thereby neglecting the fact that expected returns on these investments might be low or even negative. (8) This behaviour arises whenever losses from a bad outcome of the investment are 'nationalised', while profits under the good outcome are going to the owners of the intermediary. The reason why intermediaries can pursue these risky strategies is that depositors of financial intermediaries have no interest in monitoring their deposit institutions when they are protected from losses by governmental guarantees. Thus, whenever a government guarantees financial intermediaries' liabilities, it needs to institute an appropriate system of banking regulation and supervision.

Dooley's (2000) crisis model is one of the first models where moral hazard plays an important role. In his model, governments are interested in (a) holding reserve assets to self-insure against shocks to national consumption and (2) protecting the domestic financial sector by acting as a LOLR even though this also places demands on reserve assets. Dooley assumes that governments are credit-constrained, that is, they cannot borrow money on international capital markets without providing collateral in the form of reserve assets like foreign exchange or lines of credit from other governments or international organisations. In such a situation, a country can not credibly promise liquidity insurance to the domestic banking sector if the country's net assets (gross assets minus noncontingent liabilities) are not positive.

Dooley provides the following example to illustrate how his model might work when a country initially has negative or zero net reserves and experiences a macroeconomic shock like a decrease of international interest rates. Lower interest rates might substantially increase the country's net assets by reducing the value of noncontingent government liabilities. This enables the country to credibly insure the banking sector's liabilities. This will induce moral hazard behaviour on the part of domestic banks. Without appropriate banking supervision, banks have a strong incentive to seek new deposits by promising above-market yields to investors. Even if investors know that the domestic banks will not be able to pay them back in full, they are willing to invest in the country since they expect to be compensated via governmental reserve assets. The situation is stable as long as the government's net reserves are positive. The problem is that net reserves will decline as banks ask the government for assistance in meeting their deposit liabilities. As soon as net reserves are zero, investors will start to withdraw their money from the country. Once they recognise that domestic banks can no longer meet their obligations, the promises to pay above market yields are no longer credible since the insurance option does not exist anymore. This will generate an attack on governmental reserves as investors compete to avoid losses.

What makes this model different is that it shows that an attack on a country's reserves can occur even when a country does not have a fixed exchange rate regime. Nevertheless, it is obvious that a country which is running out of reserves will not be able to commit credibly to a fixed exchange rate regime.

Krugman (1998) presents a model where a fixed exchange-rate regime might collapse when governmental guarantees to a financial sector without an adequate system of banking regulation and supervision create a moral hazard problem. Krugman assumes a stochastic production function with decreasing returns to scale with respect to invested capital, When the government provides liquidity insurance, intermediaries invest in risky investments as long as the most favourable return on capital equals the world safe rate of interest. Thus, when deciding on investments, intermediaries take into account the so-called pangloss values (p. 6), thereby increasing the capital stock to an inefficiently high level. Government reserves fall when guarantees must be honoured. Unlike Dooley and first-generation models of currency crises, Krugman argues that crisis are not necessarily triggered by exhausting the stock of reserves but rather the expectation that the government will not stick to its promise to bail out private banks in the case of bankruptcy. Thus, Krugman ends up with a story that is somewhat similar to second-generation models of multiple equilibria.

THE BULGARIAN CRISIS OF 1996/1997

We now turn to a description of the Bulgarian Crisis and an attempt at classifying it with respect to the theoretical literature summarised in the previous section. We basically argue that the Bulgarian Crisis can best be described by a mixture of first- and third-generation models of currency crises. The crisis had some first-generation features since the Bulgarian Government's policy to (indirectly) monetise the subsidies to the ailing state-owned enterprises was inconsistent with the attempts of BNB to stabilise the exchange rate. However, the Bulgarian Crisis was a twin crisis and the banking sector played an important role herein. As in third-generation moral hazard models, implicit (and later on explicit) prudential guarantees caused massive moral hazard behaviour of the banking sector. In mid-1996 as doubts arose as to whether the authorities would be able to continue bailing out the banking sector's losses, a first wave of bank runs occurred. The attempt to stop the banking crisis by introducing a deposit insurance scheme was unsuccessful, because it lacked credibility due to the low foreign currency reserves. A sudden change to a highly restrictive monetary and banking policy, that is, a switch from a lender of first resort to an LOLR strategy, reinforced the banking crisis and helped to create a twin crisis.

The pre-crisis period 1990-1996

There is little doubt that the Bulgarian Crisis of 1996/1997 was a logical consequence of the development of the Bulgarian economy in the 1990-1996 period. The 'bad' dynamics of the process of transition in Bulgaria has been reported in many studies (OECD, 1999; Dobrinsky, 2000; Koford, 2000; Vutcheva, 2001; Mihov, 2002; among others), One of the most important problems was the very slow process of privatisation in Bulgaria. By 1997, only 20 per cent of the state's assets were privatised and state-owned enterprises accumulated enormous losses (see Table 1).

The government was not willing to close down these enterprises since this would have caused excessive unemployment. Therefore, the government forced state-owned commercial banks to subsidise ailing enterprises by granting excessive credit lines. Since most of these credits ex post turned out to be noncollectible, the banking system accumulated losses, too (Koford and Tschoegl, 1999; Caporale et al., 2002). State-owned enterprises and banks were rescued in several waves by issuing government securities, which led to increases in the government's internal debt (Vutcheva, 2001). Despite its de jure independence, de facto BNB was totally dependent on the government, BNB subsidised the government's strategy, (9) Even if there was no formal promise by the central bank or the government to bail out illiquid or insolvent banks until 1996, BNB always provided the necessary refinancing. Moreover, BNB (and State Saving Bank) in some cases granted direct credits to ailing state-owned enterprises, gave credits to the Ministry of Finance or bought government securities. Thus, the enormous losses of the state-owned enterprises were quickly nationalised via monetisation.

At the same time, BNB engaged in attempts at stabilising the exchange rate. As first-generation models of currency crises have shown, such an inconsistent strategy cannot succeed in the long run. In the short run, BNB was quite successful in stabilising the exchange rate (see Figure 1). However, when foreign currency reserves decreased to $500 million at the end of 1994, BNB let the lev depreciate, thereby increasing international competitiveness. Consequently, in the second half of 1994 foreign exchange reserves started growing until they reached approximately 1500 mil USD in late 1995. Early in 1996 foreign exchange reserves again started decreasing, an indication that the devaluations in 1994 had only a temporary effect.

[FIGURE 1 OMITTED]

However, this is only one part of the story explaining the Bulgarian Crisis. In line with the basic line of argument in the third-generation models of Dooley (2000) and Krugman (1998), we argue that the severity of Bulgaria's twin crises was primarily, but not only, due to systematic moral hazard problems in the banking sector. In Dooley's (2000) and Krugman's (1998) models, moral hazard behaviour is caused by explicit government bailout guarantees. Such formal guarantees did not exist in Bulgaria until the introduction of a deposit insurance scheme in raid-1996. The main reason for the moral hazard behaviour in Bulgaria was its history as a former communist country with a centrally planned economy. While market institutions changed quickly, the inherited behaviour of market participants remained. (10) In Bulgaria's previous economic system, losses were nationalised and socialised either within the country or within CEMA. Thus, in the early years of transition there was still a belief among economic agents that they were fully insured against losses or bankruptcy. We might, therefore, talk of some kind of moral hazard path dependence of economic agents' behaviour. While this moral hazard behaviour penetrated the whole Bulgarian economy, it was very pronounced in the banking sector.

The large-scale restructuring of Bulgaria's financial and banking system began at the end of 1989, reflecting the need to shift to a modern two-tier banking system. The sector-specific banks were transformed into classical commercial banks, accepting deposits from individuals. The existing 59 branches of BNB were transformed into autonomous commercial banks. By early 1991, the banking system comprised BNB, State Saving Bank and 69 commercial banks organised as joint stock companies. In March 1992, the Law on Banks and Credit Activity was adopted, establishing a regulatory framework for activities of banking institutions. In the 1990-1996 period BNB adopted a liberal licensing policy, which led to the appearance of a great variety of financial agents, most of which turned out to be ponzi pyramids. In addition, bankers often lacked sufficient training and internal controls on bank loan decisions were weak. While banks were required to collateralise their loans, the system did not work well. Poor communication among bankers and inadequate data made it difficult to identify poor credit risks. It is thus not surprising that the banking sector accumulated large amounts of bad credits.

Although there was no formal law guaranteeing bank deposits before 1996, the population expected to be compensated in cases of losses. For instance, before the crisis many financial 'ponzi pyramids, offering unusually high interest rates, collapsed. Although the public had been repeatedly warned not to place its money in these 'ponzi' pyramids, after their defaults people still claimed the government to pay back the lost money as it had been the time before. As argued earlier, this belief was primarily driven by the experiences with the previous economic system in Bulgaria. Consequently, depositors had little interest in monitoring commercial banks. Thus, the banking sector could take over excessive risk without having to pay higher risk premiums to depositors. Moral hazard incentives were even stronger for very illiquid or insolvent banks since they tried to 'gamble for resurrection'. However, not only depositors but also bank managers of both state-owned and private banks heavily relied on help from the authorities in the event of illiquidity or insolvency problems.

The public belief that their deposits were implicitly fully insured was reinforced by actions taken by the government and Bulgarian National Bank. Commercial banks were refinanced by BNB (see Table 2) on a completely subjective and discretionary basis. (11) The various forms of refinancing in lev and foreign currency included discount refinancing using private securities as collateral, Lombard refinancing using government securities as collateral and uncollateralised refinancing (deposit facilities and later arrears). (12) Large-scale uncollateralised refinancing was used during some periods; at the end of June 1996 uncollateralised refinancing was 80 per cent of total lev refinancing and 80 per cent of total foreign currency refinancing.

The State Savings Bank (DSK) was also active in refinancing other commercial banks on the interbank market and also one of the main buyers of government securities on the primary market. DSK often engaged in lending to previously important economic sectors like agriculture. Although these credits were decreasing in relative terms, DSK provided even more refinancing than BNB (OECD, 1997). (13) Thus, DSK carried out functions close to those of BNB. Both BNB and the Ministry of Finance directed DSK's activities. The explicit deposit insurance granted to DSK allowed cheap refinancing operations.

The losses of state-owned banks were always and quickly nationalised. This Bulgarian pre-crisis setup in which commercial banks were directly and almost automatically supported by the central bank without having to seek other types of funding (including the interbank market) is what we call lending of first resort.

The interest rates on loans, although they were very high at times, did not reflect true credit risk. In addition, most loans were granted under conditions where the expectation of repayment was low (Koford and Tschoegl, 1999). Private banks especially engaged in expanding credits to the many newly created private companies. Quite often these loans violated regulations that were designed to maintain bank solvency by restricting the size of loans to bank officers. An OECD (1999) analysis points out that '[u]ntil 1996, commercial credit was expanded to the non-financial sector in Bulgaria to a degree that was unprecedented relative to any other European transition economy'. As shown in Table 3, the structure of these credits was not 'healthy' and led to the accumulation of a large amount of bad loans. As a consequence, at the end of June 1994, 35 of 44 Bulgarian banks were losing money (Vutcheva, 2001). (14)

These problems might have been smaller if there had been better bank regulation and supervision. However, Bulgaria did not have good bank supervision, enforcement mechanisms or bankruptcy proceedings. (15)

The crisis period

The basic story of most second- and third-generation models is that there are two types of equilibria and that a crisis is the result from switching from the 'good' equilibrium to the 'bad' one. In the Krugman (1998) model, the crisis is finally caused by the market participants' expectation that the government will not (be able to) stick to its prior made promises to bail out private banks in the case of bankruptcy. We argue that it was exactly this change in public expectations that triggered the Bulgarian twin crises.

The first crisis wave came from the banking system when in May 1996 (16) BNB took five commercial banks, three of which were private, under conservatorship. (17) The runs on these banks were triggered by depositors' expectations that their foreign deposits would be confiscated or frozen by the government so the government could meet its interest payments on external debt due in July. At that time, Bulgaria was unable to get loans in international financial markets because of insufficient foreign currency reserves that could be used as collateral. In addition, information about the unhealthy state of several banks spread out and the population was worried that these banks would be closed. The fact that the Socialist government had no agreement with the International Monetary Fund (IMF) in 1996 reinforced this fear. The deposits from bankrupt banks were transferred to sound ones (Enoch et al., 2002).

In order to stop the panic, two strategies were implemented in parallel. On the one hand, a law on Bank Deposit Guarantees was introduced, according to which the government had to repay the full amount of individuals' deposits with bankrupt banks and 50 per cent of enterprises' deposits (BNB, 1996). (18) However, the deposit insurance scheme did not provide a credible guarantee at that time since reserves were already so low, internal and external debt was so high and there was still no IMF agreement. In addition after a period of large-scale refinancing, the BNB started to pursue a restrictive policy towards banks by increasing minimum reserve requirements, (19) raising interest rates (20) and at once selling US dollars to protect the lev exchange rate. In May 1996, the base interest rate was 108 per cent (simple annual); in September 1996 it rose to 300% and then it was reduced twice in October 1996 to 240 per cent and then 180 per cent (see Figure 2). (21)

[FIGURE 2 OMITTED]

The sharp increase in interest rates in the second half of 1996 further intensified the crisis. This policy was suggested by the IMF and was typical for emerging market countries trying to preserve their exchange rates (Chang and Velasco, 2001). However, the sudden change in BNB's banking policy was unexpected to both commercial banks and the government. High nominal interest rates caused an avalanche-like increase in internal debt and suspicions about the government's ability to service it arose. Internal debt became a classic example of 'ponzi' financing where new government securities had to be issued in order to make interest payments on previous issues of government securities. As investors' interest in these new issues was low, BNB was compelled to buy them. In addition, commercial banks suffered from increased interest rates and a new round of nine banks became bankrupt, thereby further increasing the panic. Facing the threat of a moratorium on internal debt, BNB was forced by the government and the parliament to provide extensive monetary financing of the budget deficit. (22) BNB also continued to grant direct loans to the Ministry of Finance. One of these credits, granted at the end of 1996 was 115 billion lev or 7 per cent of GDP. (23) This asymmetry in monetary policy, restrictive for banks and expansionary financing for the budget, was ineffective and even dangerous.

The funds withdrawn from commercial banks and obtained from sales of government bonds were quickly converted into dollars, because a sharp depreciation of lev was becoming more likely, the smaller the level of foreign currency reserves available to defend the exchange rate. Foreign currency was increasingly used as a store of value. While Bulgaria was not officially maintaining a fixed exchange rate, the BNB was trying to maintain the value of lev as it came under pressure in 1996. As in the Dooley (2000) model, BNB's foreign currency reserves had two functions: protect the exchange rate against devaluation and maintain liquidity in the banking system. In addition, as in first-generation models of financial crises, BNB still had to finance a large part of the budget deficit. Thus, the central bank had to juggle three nominal commitments. As a consequence, net foreign reserves dropped to 483 mil USD (Figure 1) while domestic credit increased considerably. Late in 1996, internal and external debt reached alarming levels. Internal debt was 60 per cent of GDP and the external debt was 243 per cent of GDP (Table 4).

After declining by 590 per cent in 1996, the exchange rate totally collapsed in February 1997 when the lev depreciated by almost 250 per cent. The devaluation was accompanied by a short period of hyperinflation (Figure 2). The monthly chain CPI was 44 per cent in January and 243 per cent in February and the annual inflation for 1997 was 578 per cent (BNB, 1997). This hyperinflation greatly reduced the government's internal debt and liabilities to banks. During this period, the US dollar effectively became the unit of account while the lev was the medium of exchange. The US dollar and deutsche mark became major stores of value.

Bulgaria had a high level of bank intermediation. (24) For this reason, the costs of the crisis were especially high. Zoli (2001) calculated the cost to be 26 per cent of GDP for the period of 1991-1998. Altogether, 14 commercial banks (out of 46) were closed in 1996. This represented 24 per cent of the banking system's assets. Ninety per cent of uncollateralised refinancing before the 1996/1997 crisis was concentrated in the bankrupt banks. During 1996, the population withdrew 42 per cent of its foreign currency deposits and 21 per cent of its lev deposits. This was almost 70 per cent of Bulgaria's foreign currency reserves. During 1996, depositors lost more than 50 per cent of their savings.

The financial crisis was accompanied by a deep political crisis and mass demonstrations. The social turmoil culminated on 10 January 1997 when the Parliament was attacked. At that time, information on the possibility of blocking deposits and internal debt default leaked out (Roussenova, 2002). On 4 February 1997, major political parties took a principal decision to introduce a CB arrangement. (25) The new President took office on 20 January 1997, the socialist party abdicated from power and a caretaker government was appointed. The Bulgarian Crisis came to an end when the exchange rate stabilised at the end of February 1997 and de facto the Bulgarian CB started to operate (formally the CB started in July 1997). In March 1997, the inflation rate dropped drastically. Parliamentary elections were held on 19 April 1997. In April 1997, a new agreement with the IMF was reached. In the course of time, economic agents started to adjust their behaviour to the forthcoming formal establishment of the CB in July 1997.

The Bulgarian Crisis was foremost a closed economy crisis. In contrast to the Asian Crisis where foreign capital outflows were significant, capital outflows were very small, about $240 million (BNB, 1996). This was due to the fact that during 1990-1996 foreign capital inflows, particularly portfolio investments, were very small. (26)

SECOND-GENERATION CBS AND THE LOLR

As argued in the last section, the Bulgarian Crisis was caused by a combination of excessive monetisation of fiscal debt and moral hazard behaviour in the banking sector resulting from the belief that BNB would always provide refinancing in the case of losses. When the crisis culminated in early 1997, a new monetary system had to be implemented. The new system had to fulfil two requirements: (1) In the short run, the monetary system had to stop the crisis. Thus, monetary strategies like inflation targeting, requiring considerable periods of reputation building, were not suitable. (2) The new monetary system had to provide long-term stability. EU accession was not possible unless the exchange rate was stabilised, and interest and inflation rates came down.

Bulgaria chose a CB arrangement, thereby strictly restricting the possibility of monetising the fiscal debt. However, the second-generation CB introduced in Bulgaria differs significantly from orthodox CBs. (27) The Bulgarian CB arrangement enables BNB to conduct limited monetary policy operations and provides for a strictly limited LOLR to help stabilise the banking sector. In this section, we focus on the limited LOLR function and briefly summarise the experiences during the first years under the CB arrangement. (28) This experience is quite encouraging, as expectations quickly adjusted to the new monetary policy strategy and the economy stabilised.

A strictly limited LOLR

As pointed out earlier in this paper, excessive financing by BNB through quasi LOLR transactions greatly contributed to the Bulgarian Crisis. Since in contemporary financial systems, the LOLR function is considered to be a fundamental part of the financial safety net system (Freixas and Rochet, 1999), Bulgarian authorities felt that a total elimination of LOLR functions was not appropriate. When introducing a second-generation CB, an attempt was made to distinguish between conducting monetary policy on the one hand and LOLR functions on the other. (29)

CBs provide an opportunity to return to some traditional forms of the LOLR. A CB arrangement is similar in some respects to the gold standard. There were several different forms of LOLRs under gold standard monetary systems, depending on whether they were centralised or decentralised: (30) (i) In the case of a decentralised gold standard and free banking regime, the LOLR function was carried out by private clearing house associations or by branch banking (in the case of the US and Canada), by option clause or a foreign central bank (in the case of Scotland). (31) (ii) Under the centralised (and international functioning) gold standard, LOLR function was performed by a temporary suspension of the rule of convertibility or a central bank could rely on support from foreign private banks (like the Barings crisis in 1889/1890). (32) The existence of branches and affiliates of major foreign banks in Bulgaria is thus an opportunity for importing LOLR from abroad. Introducing a CB system does therefore not automatically imply the abolishment of any form of LOLR functions. Similarly, the strategy of most CBs is to open the domestic market to foreign banks (as eg in Argentina and Estonia). In Argentina, the central bank agreed to bilateral credit lines from American and other foreign banks.

However, second-generation CBs can perform LOLR functions in at least two additional and more direct ways: (i) the inclusion of positions into the balance sheet (either on the asset or liabilities side) that are not typical for CBs (eg, deposit certificates are used in Estonia and repo operations are carried out in Lithuania) and (ii) the creation of a separate liquidity fund (which is independent of the CB), or a fund directly linked to the budget (Caprio et al., 1996; Rostowski, 2002).

The basic idea behind the Bulgarian CB is to restrict excessive monetisation of public debt by fully backing the currency while allowing for LOLR functions using excess reserves. Thus, BNB may extend loans in lev to banks through the Banking Department up to the level of central bank excess reserves (Nenovsky and Hristov, 2002). By generally restricting LOLR transactions to the amount of excess reserves, there is a natural limit to the funds that can be used for liquidity assistance. In the light of the earlier discussed path dependence of moral hazard in transition countries, Bulgaria decided to spell out the strictly restricted willingness and ability to support banks with liquidity shortages clearly and to make this restriction part of BNB law.

In contrast to Bagehot's (1992) definition, LOLR under the Bulgarian CB does not imply free refinancing and imposing a penalty rate, but rather limited refinancing at a penalty rate. According to Bagehot, an LOLR should announce in advance a policy of free lending in the case of a crisis. In Bulgaria, as well as in most contemporary CBs, the conditions under which refinancing is possible are legally determined and strictly limited. In line with Bagehot's proposal, the Bulgarian LOLR function sticks to the principle to grant credits only against good collateral. BNB can grant loans only to solvent banks (33) experiencing an acute need for liquidity that cannot be provided from other sources. These loans can only be extended against collateral of liquid assets and the loan repayment term shall not exceed 3 months. (34)

However, it is extremely important that CBs should not be allowed to drift temporarily from the predefined rules since any suspension of CB principles might negatively affect the credibility of the CB (Ho, 2002). (35) As the basic objective of second-generation CBs is monetary stabilisation, and in most of the cases these arrangements are put into practice after a period of hyperinflation or a financial crisis, any break of the CB rules (reserve money backing and fixed exchange rate by law) will likely be considered as a return to financial instability and inflation (in the theoretical models we might think of this as the cause for a change in market expectations causing the economy to jump from a 'good' equilibrium to a 'bad' one). However, even if a CB sticks to its rules there is obviously no guarantee for the market participants that the CB will not be abandoned at some time in the future. Thus, the introduction of a CB is no insurance against currency crises. However, we argue that the Bulgarian CB provides a considerably higher degree of stability as the highly discretionary monetary regime before the crisis. This is primarily due to the fact that the rules of the game are now clarified and became part of the legislation. Thus, the government not only tied its own hands but also eliminated a major source of moral hazard that arose from the belief that banks were fully insured against bankruptcy by state institutions. However, additional measures were necessary to secure a high degree of financial stability.

Under second-generation CBs, banking supervision, bank regulations and banking court proceedings are vital parts of the safety net. A system of efficient bank regulation is both a prerequisite for the central bank to be able to serve as an LOLR without creating moral hazard behaviour and to compensate for the somewhat limited function of the central bank as an LOLR. It is not surprising that in most of the countries under a CB arrangement, capital adequacy, liquidity and even equity requirements are significantly higher than those in countries with other monetary regimes. With the introduction of the Bulgarian CB, banking supervision became an integral and legally defined part of the new monetary strategy. The important role of bank supervision was underlined by the establishment of a separate department responsible for bank supervision within BNB. This department is fully responsible for licencing, analysing and controlling banks' activities. A number of laws and regulations were passed (like the law for bank insolvency, regulation for internal control, regulation for the management and supervision of the banks liquidity, etc), all of which enforce the implementation of the supervision principles. The minimal level for the total capital adequacy ratio was set to 12 per cent; primary capital adequacy ratios shall not be lower than 6 per cent (in fact, the banks maintain considerably higher total capital adequacy ratios ranging from 30 to 40 per cent, see Table 5).

Another crucial part of a safety net is a well-designed system of deposit guarantees. This system is designed to minimise moral hazard problems, and is more restrictive than in countries with discretionary monetary policy. (36) Deposit guarantees are necessary to avoid panics, even if panics do not always lead to a total contraction of the monetary base but only its reallocation between banks (as was experienced under the gold standard). Panics usually move deposits from 'suspicious' and weak banks to ones of sound reputation (information-based bank runs rather than random withdrawals). Such runs were observed during the Bulgarian Crisis when depositors transferred their funds to healthy banks like Bulbank. There were similar runs throughout the banking crisis in Argentina in 1994/1995 with a CB in operation but no safety net (Schumacher, 2000; Caprio et al., 1996). Depositors reallocated their money to big Argentine and foreign banks. In 1998, a Deposit Insurance Fund financed by commercial banks was established in Bulgaria. The entry premium is 1 per cent of the minimum capital required and the annual premium contribution is 0.5 per cent of the total amount of the deposit base for the preceding year, determined on an average daily basis. For a long time, the maximum amount reimbursed to a single person was 2,500 euros. In 2002, it was raised to 7669 euros as part of the requirements of EU convergence (by the end of 2006, the coverage limit should become 20,000 euros as in EU countries). However, since the majority of deposits are less than 5,000 euros, the current deposit guarantee might be too high, inducing moral hazard on the behalf of banks due to low monitoring incentives of depositors. (37)

Under a CB, the interbank market might also play an important role for liquidity control. However, the Bulgarian interbank market does not function very well. Banks set up internal regulations for transferring liquidity to other banks case by case. In fact, there is no dialogue among the banks, which is an important precondition of providing mutual aid in the case of a crisis (Freixas et al., 1999). This phenomenon can perhaps be explained partly by the lack of trust among commercial banks induced by the 1996/1997 crisis.

Development of the Bulgarian economy since July 1997

As argued earlier in this paper, the newly introduced monetary strategy had to fulfil two tasks: to stop the crisis in the short run and to stabilise the Bulgarian economy in the medium and long run. The first task was solved by the establishment of the Bulgarian CB. The simple fact that the CB arrangement survived shows that the credibility of the new regime was sufficiently high. External stabilisation also allowed restoring internal stability. Soon after the introduction of the CB, inflation fell sharply. While inflation expectations also decreased after the CB introduction by about 50 per cent, average inflation expectations were initially quite high (24.9 per cent). This finding might be explained by agents taking a wait-and-see strategy. (38) How far the new monetary regime will be able to solve the second and more important task of long-term stability cannot be judged definitely after only 5 years. (39) Recent developments in the Bulgarian economy suggest, however, that the CB will continue to provide a stabilising force in the longer run. Table 5 summarises the movements of the main economic indicators from 1997 to 2001.

Essential changes have occurred in the banking system. First, almost all the banks have been privatised. The last state-owned bank is expected to be privatised in 2003. Second, bank behaviour has changed substantially. This is partly due to the experience before and during the 1996/1997 crisis and partly due to the restrictions imposed by the CB arrangement and bank supervision. Banks became extremely cautious in managing their portfolios, especially their lending operations. They have maintained capital adequacy and liquidity ratios substantially above international standards. As a result, domestic credit was restructured. The share of standard exposures in total exposures increased from 58.2 per cent in 1997 to 92.3 per cent in 2001. Third, lending to the public sector significantly decreased. While the share of public sector credits in total credits was 64.4 per cent in 1997, it decreased to some 33.5 per cent in 2001. These measures indicate that phenomena like direct lending and moral hazard behaviour are considerably restricted by the new regime. In the second half of 2002, a more active lending by banks was observed, which was associated with a drop of interest rates in international markets and a redirection of banks' investments back to Bulgaria.

LOLR practices have also changed. During the 5-year existence of Bulgaria's CB, BNB never made a loan under the LOLR procedures, although two banks suffered some liquidity difficulties during this period. Both these small banks failed. (40) This experience shows that a restrictive definition of the LOLR function might be sufficient to provide the necessary stability to the financial sector, thereby preventing systematic bank runs from occurring. However, there is no certainty about what would happen if a large bank would fail and ask for liquidity assistance. While current excess reserves of the Bulgarian CB covering more than hall of the deposits of the banking system would allow for supporting even larger banks without violating the CB rules, the described institutional framework of banking supervision makes such a scenario less likely. Moreover, the presence of various foreign banks in Bulgaria contributes to a higher degree of stability of the financial system.

The restrictions imposed by the CB arrangement also induced the Bulgarian government to adopt a more conservative fiscal policy. Since introduction of the CB in 1997, the budget deficits have been extremely small. In 1998, there was even a small surplus. As a result, the internal debt in per cent of GDP had decreased from 60 per cent at the end of 1996 to 7 per cent by the end of 2001.

The switch from a discretionary monetary policy with an excessive use of the LOLR function and excessive monetisation of government debt to a second-generation CB has created stability, which is an important precondition for Bulgaria's process of EU accession. BNB balance-sheet positions currently give no reason to worry about the CB's future. However, the economy has grown slowly. Official measures of unemployment increased during the last years (although this development might be induced by a growing shadow economy). The CB has been a necessary but not sufficient precondition for sustainable economic growth. The CB arrangement is only a part of extensive institutional and legal reforms that must take place.

SUMMARY AND CONCLUSIONS

In this paper, we argue that the Bulgarian twin crises of 1996/1997 can best be explained by a combination of first- and third-generation moral hazard models of currency crises. Unlike the typical assumption of third-generation currency crisis models, moral hazard behaviour was, at least initially, not induced by explicit governmental guarantees but by the unchanged public belief that the government and the central bank would protect industrial companies and banks facing bankruptcy problems. These beliefs, which are at least partially due to Bulgaria's history as a former communist country, were reinforced by the government's actions. When the public started to worry about the governments' ability and willingness to continue this policy, a banking crisis was triggered and reinforced by a sudden change to a restrictive banking policy by BNB. Thus, an important cause of the Bulgarian Crisis of 1996/1997 was BNB's role as a lender of first resort rather than an LOLR in the pre-crisis and the crisis period.

The crisis was stopped by implementing a second-generation CB. While this arrangement allows BNB to perform some LOLR functions, these are strictly limited. This institutional setting is the response to the experiences with a highly discretionary monetary strategy in the period of 1991-1996. The CB arrangement allowed Bulgaria to import the hard-nosed reputation of the German Bundesbank and the ECB in fighting inflation. The strictly limited LOLR function provides the necessary degree of stability for the banking system. Both theorists and practitioners seem to have underestimated the special systemic moral hazard risk in transition countries primarily resulting from the inherited belief that state institutions will resolve any form of financial distress in an economy. The twin crises in Bulgaria of 1996/1997 underlines both the importance and the vulnerability of banking systems in transition countries. Since capital market-based financing is often impossible, a strong banking system is needed to provide financing of investments, which are urgently needed to catch up with Western European countries. The transformation process is a time of soft budget constraints that can end up in a deep financial crisis, as the Bulgarian example has shown. The Bulgarian experience also demonstrates that with a new institutional structure for the banking system, proper supervision and a more restrictive role for the central bank, it is possible to stabilise a crisis economy.

Table 1: Financial results (net profit) of state-owned firms
(1992-1997) in per cent of balance sheet's assets

Year Industry Construction Transport Trade Others

1992 -7.87 0.26 0.08 -1.62 0.00
1993 -12.74 -2.17 -6.99 -1.14 -4.96
1994 -4.89 -2.14 -3.28 0.43 5.37
1995 -4.24 -1.14 -5.23 -1.31 1.10
1996 -5.54 -1.18 -5.85 -1.78 -6.71
1997 2.89 0.99 3.43 2.30 2.50

Source: OECD (1999, p. 79)

Table 2: Domestic credit and refinancing to commercial banks
(1995-1997)

(Thousand BGN) 12/1995 03/1996 06/1996

Domestic assets (net) 72,030 88,870 115,107
Claims on central government (net) 25,976 35,936 28,066
 Of which Government securities 50,558 48,707 57,089
Claims on commercial banks 43,383 58,294 105,171
 In foreign currency 19,137 21,282 38,589
 Deposits 14,911 16,404 30,785
 Credits 4,226 4,878 7,804
 Shares and other equity 0 0 0
 Other 0 0 0
 Deposits+other 14,911 16,404 30,785
 In lev 24,246 37,012 66,582
 Deposits 11,419 29,176 52,937
 Credits 9,548 5,801 11,676
 Shares and other equity 1,469 1,838 1,838
 Other 1,810 197 131
 Deposits+other 13,229 29,373 53,068
Other items (net) 2,671 -5,360 -18,130

(Thousand BGN) 09/1996 12/1996 03/1997

Domestic assets (net) 134,814 157,660 200,241
Claims on central government (net) 82,142 155,007 404,375
 Of which Government securities 123,497 201,535 509,296
Claims on commercial banks 132,600 238,758 487,317
 In foreign currency 53,349 113,388 348,232
 Deposits 41,061 72,896 226,561
 Credits 12,288 40,492 120,718
 Shares and other equity 0 0 0
 Other 0 0 953
 Deposits+other 41,061 72,896 227,514
 In lev 79,251 125,370 139,085
 Deposits 54,379 12 12
 Credits 22,862 123,387 56,172
 Shares and other equity 1,838 1,838 1,838
 Other 172 133 81,063
 Deposits+other 54,551 145 81,075
Other items (net) -79,928 -236,106 -691,452

(Thousand BGN) 06/1997 09/1997 12/1997

Domestic assets (net) -1095,231 -1251,067 -1042,559
Claims on central government (net) 88,991 -13,468 178,180
 Of which Government securities 0 0 0
Claims on commercial banks 311,604 315,059 334,617
 In foreign currency 159,057 159,622 181,888
 Deposits 24,562 24,856 49,954
 Credits 114,652 111,150 110,483
 Shares and other equity 0 0 0
 Other 19,843 23,616 21,451
 Deposits+other 44,405 48,472 71,405
 In lev 152,547 155,437 152,729
 Deposits 12 18 20
 Credits 53,827 53,757 53,404
 Shares and other equity 1,838 1,838 1,838
 Other 96,870 99,824 97,467
 Deposits+other 96,882 99,842 97,487
Other items (net) -1495,826 -1552,658 -1555,356

Source: BNB analytical reporting

Table 3: Dynamics of uncollectible credits (in per cent of total
credit) (1993-1996)

 1993 1994 1995 1996 (a)

Standard exposures 7.61 17.69 25.91 43.67
Doubtful exposures (group A) 82.75 66.88 54.55 33.89
Doubtful exposures (group B) 2.19 3.46 4.18 10.67
Uncollectible exposures 7.45 11.97 15.35 11.77
Reported/required statutory provisions 7.18 23.58 23.84 105.42

Source: BNB, Banking Supervision Department

(a) Banks in liquidations are excluded. Group A--arrears up to 30
days, group B--arrears up to 90 days and uncollectible
exposures--arrears over 90 days, or case of bankruptcy or liquidation
of the credit holder.

Table 4: Dynamics of government and government guaranteed debt
(1991-2000)

 1991 1992 1993 1994 1995 1996

Dom. Debt/GDP (%) 13 19 37 52 39 60
For. debt/GDP (%) 168 127 109 129 73 243 (a)

 1997 1998 1999 2000 2001

Dom. Debt/GDP (%) 16 14 14 7 7
For. debt/GDP (%) 91 72 78 72 67

Source: BNB, fiscal services

(a) Lev devaluation should be taken into account.

Table 5: Performance of the Bulgarian economy (1997-2001)

 1997 1998 1999

GDP real growth (%) -7 3.5 2.4
Unemployment rate (%) 13.7 12.2 16
Inflation (%, eop) 578.5 1.0 6.2
Budget deficit (% of GDP) -3 1 -1
Current account (% of GDP) 10.1 -0.5 -5
Foreign direct investment (% of GDP) 4.9 4.2 6.3
Foreign reserves (billions USD) 2474.1 3051.1 3221.6
Number of banks (foreign banks) 34(14) 34(17) 34(22)
Total capital adequacy (%) 28.9 37 41.3
RDA (ROF) total for banking system (%) 5(116) 2(22) 2(21)
Domestic credit (% of GDP) 29.5 18.9 17.8
Credit on private sector (% of total credit) 35.6 38.9 55.8
Credit to public sector (% of total credit) 64.4 61.1 44.2
Standard exposures (% of total exposure) 58.2 69 73.3
Broad money (% of GDP) 25.5 28.2 28.3
Foreign currency deposits (% of total deposits) 63.5 54.1 52.9

 2000 2001

GDP real growth (%) 5.8 5
Unemployment rate (%) 17.9 17.3
Inflation (%, eop) 11.3 4.8
Budget deficit (% of GDP) -1.1 -1.5
Current account (% of GDP) -5.6 -6.2
Foreign direct investment (% of GDP) 7.9 5.1
Foreign reserves (billions USD) 3460.3 3580.3
Number of banks (foreign banks) 34(24) 35(25)
Total capital adequacy (%) 35.5 31.32
RDA (ROF) total for banking system (%) 3(23) 3(19)
Domestic credit (% of GDP) 17.4 18.5
Credit on private sector (% of total credit 67.2 66.5
Credit to public sector (% of total credit) 32.8 33.5
Standard exposures (% of total exposure) 82.7 92.3
Broad money (% of GDP) 32.2 36.9
Foreign currency deposits (% of total deposits) 52.7 51.7

Source: BNB, NSI, author calculations

(1) We are very grateful for a large number of helpful comments by Jeffrey Miller. We are also indebted to the participants of the conference 'Institutional and Organizational Dynamics in the Post-Socialist Transformation' in January 2002 in Amiens as well as Christian Hott, Alexander Karmann, Lena Roussenova and two anonymous referees. "We also thank Kalina Dimitrova for technical assistance. Financial support of Altana Stiftung is gratefully acknowledged.

(2) A possible reason for the relatively low level of international interest in the Bulgarian crisis is that Bulgaria is a small country that was not a part of the European Community. In addition, Bulgaria is not an important trading partner of major European countries and did not receive significant foreign direct investments. International investors did not worry very much about a possible Bulgarian crisis nor did they fear that such a crisis could infect other economies of interest to international investors. Furthermore, the Asian Crisis started soon after the Bulgarian Crisis.

(3) The complexity of the Bulgarian Crisis has rarely been subject to special analyses. A detailed crisis chronology can be found only in a limited number of publications. It is broadly covered by BNB annual reports (1996, 1997, 1998, 1999, 2000), OECD (1997, 1999), Balyozov (1999), Enoch et al. (2002), and partially by Filipov (1998), Sgard (1999), Mihov (2002), Nenovsky (1999), Dobrinsky (2000), Vutcheva (2001) and Roussenova (2002), In their review of banking crises in transition economies, Tang et al. (2000) point out that Bulgaria is the only transition country where a banking crisis was combined with a currency crisis.

(4) in this paper, we use 'currency crisis' to refer to a sharp sudden devaluation or depreciation. A 'banking crisis' occurs when a substantial number of banking institutions rail and/or a substantial amount of bank deposits are lost through failing banks.

(5) See, for example, the contagion models by Eichengreen et al. (1996) and Drazen (1999) or the herding model by Calvo (1999).

(6) See, for example, Dooley (2000), Krugman (1998), Chang and Velasco (1998), Buch and Heinrich (1999) or Flood and Marion (2000).

(7) For example, a central bank might decide to give liquidity assistance to banks suffering from temporary illiquidity problems, thereby increasing domestic money supply in spite of the fact that the exchange rate peg has to be abandoned because foreign currency reserves are exhausted. See for example, Chang and Velasco (1998).

(8) It should be noted that this strong form of moral hazard only applies to the case where intermediaries do not invest on their own and thus have nothing to lose when going bankrupt.

(9) Hochreiter and Kowalski (2000) argue that the legal independence of central banks in Eastern and Central Europe did not automatically lead to de facto independence. In most cases, the governments round ways to impose its fiscal interests by surmounting legal restrictions.

(10) For a general discussion of the process of transition, see Kornai (2000). He distinguishes between soft budget constraints (i) inherited from socialism and (ii) specific to the transition process.

(11) According to Balyozov (1999), during the whole period of 1995-1996 two big state owned banks regularly obtained refinancing in order to prevent shocks to the payments system.

(12) Uncollateralised refinancing is captured under 'deposits and other' in Table 2. Revised monetary data are available since 1995. The data before 1995 are not compatible with the newly revised data.

(13) For instance, refinancing by DSK and BNB as a percentage of GDP was 11.5 respectively 8.8 per cent in 1993, 8.5 respectively 6.6 per cent in 1994 and 6.0 respectively 5.2 per cent in 1995. In July 1996, DSK was directed 10 stop uncollateralised refinancing as well as participating in the deposit market (OECD, 1997, p. 109). At the end of 1996, refinancing operations fell to 2 per cent of GDP and only 0.1 per cent at the end off 1997 (data for 1993, 1994 and 1995 are taken from OECD reports and for 1996 and 1997 from BNB staff reports).

(14) The Bulgarian accounting practice in that period was guided by the intention to show better bank results. However, even with these standards in the beginning of 1996 only one or two banks had positive net worth (Enoch et al., 2002). See also Roussenova (2002) for a discussion of accounting standards in Bulgaria.

(15) BNB's limited possibilities to initiate bankruptcy, coupled with the cumbersome bankruptcy proceedings in that period, are major reasons for the delayed banking reform (Enoch et al., 2002).

(16) There were already some signs of the upcoming crisis in late 1995 when liquidity shortages arose and several rumours of bank failures were reported (Enoch et al., 2002).

(17) Conservatorship is a legal procedure (introduced in May 1996) allowing BNB to suspend the operation of a bank close to insolvency. In that case, BNB appoints a conservator who (temporarily) manages the bank.

(18) At first, individuals were allowed to withdraw their deposits in lev (without any restrictions) before the court declared its decision on closed banks (withdrawals of foreign currency deposits were allowed only by portions). Lev deposits withdrawn were quickly directed to the foreign currency market where the lev came under pressure. Later, the permission to withdraw the full amount of lev deposits was abolished and lev deposits were also blocked like deposits in foreign currency.

(19) BNB changed the minimum reserve requirements in opposite directions. First, it lowered them from 9.5 per cent to 8.5 per cent; in December 1996 BNB began raising them up to a level of 11 per cent.

(20) It should be noted that the decision to raise the base interest rate was suggested by IMF although the negative side effects have already been recognised during the former crises in Latin America in 1997-1998.

(21) The increase of the base interest rate was partially caused by BNB's active open market operations (primarily reverse repurchase agreements) in order to withdraw liquidity (Balyozov, 1999).

(22) According to the then BNB Governor Lubomir Filipov, BNB was subject to an assault by both the Bulgarian government and Parliament in raid-1996. At that time, the Parliament passed a law allowing to remove Managing Board members with qualified majority (Filipov, 1998).

(23) According to the then BNB Chief Economist, the BNB Managing Board expressed disagreement with this credit in a special letter to the Government and the Parliament (Roussenova, 2002).

(24) For a detailed survey on the financial sector in transition countries, see Bonin and Wachtel (2002).

(25) At some point, all political forces competed to introduce the CB arrangement and curry favour with the IME For more details, see Nenovsky and Rizopoulos (2003).

(26) Resident capital flight' could be considered as still another form of capital flight; the term was introduced by BNB to denote the outflow of capital from the banking system (see also Dobrinsky, 2000).

(27) The features of orthodox (or 'first-generation') CBs, typical of the colonial system, are well known in broad outlines (compare Schuler, 1992; Schwartz, 1993). An orthodox CB completely rejects monetary policy. A CB is backed by a simple and clear rule which determines the relationship between balance of payments, reserve money (or money supply) and interest rate dynamics (compare Hanke and Schuler, 1991; Williamson, 1995).

(28) For a detailed description of the Bulgarian CB, see Miller (1999), Nenovsky and Hrislov (2002) and Nenovsky et al. (2002).

(29) Some economists think that such a separation is impossible (Goodfriend and King, 1988), although in the 18th century Thornton and Bagehot tried to distinguish between them.

(30) For surveys, see Bordo (1989) or Denise (2001).

(31) See Bordo and Kydland (1996). For example, the Bank of England was supported by the Banque de France in 1890, and several more times in the beginning of the 19th century. It is possible that the European Central Bank will carry out the LOLR function for Bulgaria since Bulgaria is currently in the process of EU accession. See White (1999) and Goodhart (1987) for details.

(32) In order to guarantee the restoration of the gold parity before 1866, the Bank of England issued letters of indemnity.

(33) While Bagehot's advice to support only solvent banks with liquidity problems is implemented in Bulgaria, it is often hard to judge whether a certain bank asking for liquidity assistance is solvent or not. This information problem is even harder to solve in the short period of time the central bank has to decide on LOLR assistance (Goodhart and Huang, 1999, p. 6).

(34) BNB's Regulation N6 defines liquidity risk as a situation where the amount of the ordered but unpaid payment documents in the Banking Integrated System for Electronic Transfer (BISERA) exceeds 15 per cent of its total amount for both the last 2 days. In addition, the liquidity risk for the banking system is a condition caused by a bank delay or an announcement that the bank is going to postpone the settlement of the submitted payment documents for more than 3 days, and if the bank has at least eight per cent share of all interbank payments for each of the last five business days prior to filing a request for a loan with BNB (BNB, 1999).

(35) The importance of respecting the rules of the game during the gold standard is underlined in Bordo and Kydland (1996).

(36) This conclusion is drawn in a review by Garcia (1999). He shows that the optimal guarantee is between one and two times the annual GDP per capita. For details about the specific features of deposit guarantees in transition countries, see Hermes and Lensink (2000).

(37) Details for Bulgaria are given in Nenovsky and Petrov (2002).

(38) See the survey conducted by Carlson and Valev (2001).

(39) The recent events in Argentina teach us that a strategy which worked quite well for some years must not be adequate in the long run.

(40) In the beginning of 1999 Credit Bank was declared to be insolvent (BNB, 1998), and in the beginning of the following year Bulgarian Universal Bank went bankrupt (BNB, 2000).

REFERENCES

Bagehot, W. 1992: What a panic is and how it might be mitigated. The Economist, XXIV: 554-555, 12 May 1866, and reprinted in Collins, M (ed), Central Banking in History, Vol. I. Edward Elgar: Cambridge. pp. 236-237.

Balyozov, Z. 1999: The Bulgarian financial crisis of 1996-1997. Discussion paper no. 7, Bulgarian National Bank: Sofia.

Bonin, J and Wachtel, P. 2002: Financial sector development in transition economies: Lessons from the first decade. Bank of Finland, BOFIT Discussion papers no. 9.

Bordo, M. 1989: The lender of last resort: Some historical insights. Working paper no. 3011, NBER: Cambridge, MA.

Bordo, M and Kydland, F. 1996: The gold standard as a commitment mechanism. In: Tamim, B, Barry, E and Taylor, MP (eds). Modern Perspectives on the Gold Standard. Cambridge University Press: Cambridge. pp. 55-100.

Bordo, M, Eichengreen, B, Klingebiel, D and Martinez-Peria, MS. 2001: Is the crisis problem growing more severe? Economic Policy 32: 53-82.

Buch, CM and Heinrich, RP. 1999: Twin crises and the intermediary role of banks. International Journal of Finance and Economics 4: 313-323.

Bulgarian National Bank (BNB). 1996, 1997, 1998, 1999, 2000: Annual reports. Bulgarian National Bank: Sofia.

Calvo, GA. 1999: Contagion in emerging markets: When Wall Street is a carrier. Mimeo, University of Maryland.

Caporale, G, Miller, J, Hristov, K, Nenovsky, N and Petrov, B. 2002: The banking system in Bulgaria. In: Zeljko S. (ed). Banking Reforms in South-East Europe. Edward Elgar: London.

Caprio, G, Dooley, M, Leipziger, D and Walsh, C. 1996: The lender of last resort function under a currency board: The case of Argentina. Open Economies Review, 7(Suppl. 1): 625-650.

Carlson, J and Valev, N. 2001: Credibility of the new monetary regime: The currency board in Bulgaria. Journal of Monetary Economics 47: 581-594.

Chang, R and Velasco, A. 1998: Financial crises in emerging markets: A canonical model. Working paper no. 6606, NBER: Cambridge, MA.

Chang, R and Velasco, A. 2001: Monetary policy in a dollarized economy where balance sheets matter. Journal of Development Economics 66: 445-464.

Corsetti, G, Pesenti, P and Roubini, N. 1999: Paper tigers? A model of the Asian crisis. European Economic Review 43: 1211-1236.

Denise, C. 2001: Les determinants et les acteurs du prets international en dernier ressort pendent la periode de l'etalon-or classique (1890-1914). Quelques enseignements pour l'avenir. Revue economique 52: 415-430.

Diamond, D and Dybvig, PH. 1983: Bank runs, deposit insurance, and liquidity. Journal of Political Economy 91:401-419.

Dobrinsky, R. 2000: Transition crisis in Bulgaria. Cambridge Journal of Economics 24: 581-602.

Dooley, MP. 2000: A model of crises in emerging markets. The Economic Journal 110: 256-272.

Drazen, A. 1999: Political contagion in currency crises. NBER working paper no. 7211, NBER: Cambridge/MA.

Eichengreen, B, Rose, A and Wyplosz, C. 1996: Contagious Currency Crises. EUI working paper EUF no. 96/2, European University Institute: Florence, Italy.

Enoch, C, Gulde, A-M and Hardy, D. 2002: Banking crises and bank resolution: Experiences in some transition countries. IMF working paper no. 02/56, March.

Filipov, L. 1998: BNB's sad progress from an independent central bank to a currency board (the causes & the moral). Paper presented at Dubrovnik Conference on Central Banking in Transition, June 24-26, Croatian Central Bank, Dubrovnik, Croatia.

Flood, RP and Marion, NP. 2000: A model of the joint distribution of banking and exchange-rate crises. Paper presented at the ASSA-conference in New Orleans, January.

Freixas, X and Rochet, JC. 1999: Microeconomics of banking (4th printing). MIT Press: Cambridge/ MA.

Freixas, X, Giannini, C, Hoggarth, C and Soussa, F. 1999: Leader of last resort: A review of literature. Bank of England Financial Stability Review no. 7. London: Bank of England. pp. 151-167.

Garcia, G. 1999: Deposits insurance: A survey and best practices. Working paper no. 54, IMF: Washington, DC.

Goodfriend, M and King, M. 1988: Financial deregulation, monetary policy, and central banking. Economic Review 74:3-22.

Goodhart, CAE. 1987: Why do banks need a central bank? Oxford Economic Papers 39:75-89.

Goodhart, CAE and Huang, H. 1999: A model of the lender of last resort. Working paper no. 39, IMF: Washington, DC.

Hanke, S and Schuler, K. 1991: Teeth for the Bulgarian lev: A currency board solution. International Freedom Foundation: Washington, DC.

Hermes, N and Lensink, R. 2000: Financial system development in transition economies. Journal of Banking & Finance 24: 507-524.

Ho, C. 2002: A survey of the institutional and operational aspects of modern-day currency boards. Working paper no. 110, BIS: Basel.

Hochreiter, E and Kowalski, T. 2000: Central banks in European emerging market economies in the 1990s. Working paper no. 40, Bank of Austria: Vienna, Austria.

Koford, K and Tschoegl, A. 1999: Problems of bank lending in Bulgaria: Information asymmetry and institutional learning. MOCT-MOST: Economic Policy in Transitional Economies 9: 123-151.

Koford, K. 2000: Citizen restraints on "Leviathan" government: Transition politics in Bulgaria. European Journal of Political Economy 16: 307-338.

Kornai, J. 2000: What the change of system from socialism to capitalism does and does not mean. Journal of Economic Perspectives 14: 27-42.

Krugman, P. 1979: A model of balance of payment crises. Journal of Money, Credit and Banking 11: 311-325.

Krugman, P. 1998: What happened to Asia? mimeo.

Krugman, P. 1999: The return of depression economics. Lane: London.

McKinnon, R and Pill, H. 1996: Credible liberalizations and international capital flows: The 'overborrowing syndrome'. In: Ito, T and Krueger, A (eds). Financial Deregulation and Integration in East Asia. University of Chicago Press: Chicago. pp. 7-42.

Mihov, I. 2002: The economic transition in Bulgaria, 1989-1999. In: Blejer, M and Screb, M (eds). Transition: The First Decade. MIT Press: Cambridge/MA.

Miller, J. 1999: The currency board in Bulgaria: The first two years. Discussion paper no. 11, BNB: Sofia.

National Statistical Institute 1999: Main macroeconomic indicators. NSI: Sofia,

Nenovsky, N. 1999: Une economie en transition a-t-elle vraiment besoin d'une Banque centrale? La Caisse d'emission en Bulgarie. Revue d'etudes comparatives Est-Ouest 30:65 96.

Nenovsky, N and Hristov, K. 2002: The new currency boards and discretion: Empirical evidence from Bulgaria. Economie Systems 26: 55-72.

Nenovsky, N, Hristov, K and Mihaylov, M. 2002: Comparing institutional and organizational design of Currency Boards in transition countries. Eastern European Economics 40: 6-36.

Nenovsky, N and Petrov, B. 2002: Deposit insurance in Bulgaria problems and possible solutions during the accession process. Banks, Investments and Money 1: 2-9.

Nenovsky, N and Rizopoulos, Y. 2003: Extreme monetary regime change. Evidence from currency board introduction in Bulgaria. Journal of Economic Issues 37: 909-941.

Obstfeld, M. 1994: The logic of currency crises. Cahiers Economiques et Monetaires 43: 189-213.

Organization for Economic Cooperation and Development (OECD). 1997, 1999: OECD economic survey of Bulgaria. OECD: Paris.

Rostowski, J. 2002: The eastern enlargement of the EU and the case for unilateral euroization. Paper presented at the EU High Level Conference Exchange Rate Strategies During EU Enlargement, organized by the ICEG Europe, Budapest, November 27-30.

Roussenova, L. 2002: The 1996-1997 financial crisis in Bulgaria. European Institute: Sofia.

Schuler, KA. 1992: Currency boards. Ph.D. Thesis, Fairfax: George Mason University.

Schumacher, L. 2000: Bank runs and currency run in a system without a safety net: Argentina and the 'tequila' shock. Journal of Monetary Economics 46: 257-277.

Schwartz, A. 1993: Currency boards: Their past, present and possible future role. Carnegie-Rochester Conference Series on Public Policies, Vol. 39, Elsevier: Amsterdam, pp. 147 193.

Sgard, J. 1999: Crise financiere, inflation et Currency Board en Bulgarie (1991-1998): Les lecons d'une transition indisciplinee. Revue d'etudes comparatives Est Ouest 30:215-235.

Tang, M, Zoli, E and Klytchnikova, I. 2000: Banking crises in transition countries: Fiscal costs and related issues. Policy research working paper no. 2484, World Bank: Washington, DC.

Vutcheva, H. 2001: Ikonomicheskata politika v Balgaria prez perioda 1991-2000 (Economic Policy in Bulgaria during 1991-2000). UNSS Edition: Sofia.

White, L. 1999: The theory of monetary institutions. Malden: MA.

Williamson, J. 1995: What role for currency boards? Policy analyses in international economics no. 40, Institute for International Economics: Washington, DC.

Zoli, E. 2001: Cost and effectiveness of banking sector restructuring in transitional economies. Working paper no. 157, IMF: Washington, DC.

MICHAEL BERLEMANN (1,2) & NIKOLAY NENOVSKY (3,4,5)

(1) ifo Institute for Economic Research, Branch Dresden, Einsteinstrasse 3, D-01069 Dresden, Germany

(2) Dresden University of Technology, Dresden, Germany E-mail: berlemann@ifo.de

(3) Bulgarian National Bank, Sofia, Bulgaria,

(4) University of National and World Economy, Sofia, Bulgaria,

(5) Universite d'Orleans, Orleans, France