IMF-supported programmes in transition economies: are they effective?

Eke, Burcu ; Kutan, Ali M.

INTRODUCTION

Since the establishment of the International Monetary Fund (IMF), its organizational structure and programmes have been criticized, in terms of transparency, accountability, effectiveness, and conditionality. The IMF has been blamed for responding to a crisis either too slowly or too quickly, and for prescribing the wrong medicine. Some critics have argued that IMF programmes are time-consuming to negotiate and often ineffective. Moreover, Sachs (1997) supports the idea that IMF packages are more likely to harm economies rather than improve them by 'transforming a currency crisis into a rip-roaring economic downturn'.

Previous studies on the effectiveness of IMF programmes focus on developing economies (see, among others, Przeworski and Vreeland, 2000; Evrensel, 2002). There is limited empirical evidence on the role of IMF-supported policies in transition economies. Due to data limitations, available studies have taken a descriptive approach. Rodlauer (1995) studies the role of the IMF in the Eastern European transition process in Albania, Bulgaria, the former Czechoslovakia (Czech Republic and Slovak Republic), Hungary, Poland, and Romania. He concludes that a fundamental commitment to IMF programmes yields better macroeconomic performance and a less painful transition process, which was the case in both Poland and the Czech Republic.

Zecchini (1995) examines the role of several international financial institutions in the transition processes of both Eastern European and Central Asian countries. (1) His results show that the international financial institutions may play an important role in transition countries, especially, in helping these countries recognise their key economic problems. He also adds that the transition experience has been a learning process, not only for the countries involved, but also for international financial institutions.

Csaba (1995) analyses the relationship between Hungary and the IMF during the pre- and post-transition years. He attempts to answer the question of why IMF programmes could not lead to greater market reforms in this country, despite the fact that the government was willing to cooperate with the Fund. Csaba (1995) emphasises that the lack of a full commitment between the two parties was an important factor.

Minassian (2001) studies the relationship between Bulgaria and the IMF, using a case study approach. He argues that the IMF's forecasting ability was relatively poor in Bulgaria, resulting in misjudgments in 'policy formulation and implementation' (p 33). He also adds that, although the IMF could not be blamed for undesired macroeconomic outcomes in Bulgaria, it could have provided more flexibility in terms of target variables.

In this paper, we focus on IMF programmes in Bulgaria and Poland. Although similar programmes are implemented in each country, Poland experienced a more successful transition than Bulgaria. Indeed, in its annual reports, the IMF classified Poland's transition as a 'success story' (IMF Annual Report, 1992). An important question for us is whether IMF programmes have contributed to the success of transition economies.

To shed some lights on this issue, this study has dual objectives. First, we review IMF-supported policies in the two countries. Second, we provide empirical evidence about whether IMF programmes are able to reduce the crisis probabilities associated with key macroeconomic indicators. For this, we provide tests that offer inferences about the necessary conditions for the effectiveness of IMF programmes in reducing the likelihood of inflation, output and foreign exchange reserve crises.

The rest of the paper is organised as follows. The following section reviews the IMF's involvement in Bulgaria and Poland and compares the implementation of Fund programmes in both countries. The subsequent section introduces the methodology used to measure the effectiveness of IMF programmes. The last section summarises the findings and discusses their policy implications.

IMF PROGRAMMES IN BULGARIA AND POLAND

Table 1 presents information about the types, dates, and approved amounts of IMF financial assistance for Bulgaria and Poland. The stand-by arrangements (SB) are signed when the country is experiencing a short-term balance of payments problem, while the extended fund facility (EFF) agreements are designed for a long-term balance of payments problem. As a result, the programme period under the EFF programme is longer than that under the SB agreement. To illustrate, Table 1 shows that Poland signed a stand-by agreement with the IMF in the first quarter (Q1) of 1990 and the agreed amount was SDR545 millions. As evident in Table 1, Bulgaria has been a frequent borrower from the Fund.

IMF programmes in Bulgaria

Bulgaria, which became a member of the IMF in September 1990, is one of the largest users of IMF credit on a per capita basis among all transition economies. Since 1991, the country has signed six stand-by (SB) arrangements and adopted one extended fund facility (EFF) programme. Figure 1 shows per capita amount of credit approved and drawn in Bulgaria.

The first agreement of Bulgaria with the IMF, which was a stand-by one, was approved in March 1991. At that time, Bulgaria was suffering from macroeconomic disequilibria due to the collapse of Council for Mutual Economic Assistance (CMEA) system in 1991, on which Bulgarian economy was highly dependent. The purpose of the stand-by agreement was to liberalise prices, to promote external trade and payments systems, to eliminate monetary expansion, and to begin privatisation (IMF Annual Report, 1991). In order to achieve these targets, the IMF advised Bulgaria to modify its legislative and institutional structure, and to begin privatisation and restructuring of its state enterprises. Out of the total credit awarded, which was SDR279 million, Bulgaria was able to receive only the first three tranches; the fourth one was suspended due to the failure to meet agreed targets.

Following the suspension of the first agreement, Bulgaria and the IMF began renegotiations and signed the second stand-by agreement in April 1992. The aim of this agreement was to reduce the monthly inflation rate to 2% by the end of 1992, to hold the decline in the output over the year to about 4%, and to restrict bank financing to 3% of GDP. At the same time, the central bank was charged with conducting a credit policy consistent with inflation targeting and advocating minimum price controls (IMF Annual Report, 1992). Although Bulgaria accepted these policy changes, it could not successfully implement them. In fact, one of the key target variables, the net bank loans, exceeded the upper limit. Therefore, after the fourth tranche, the second stand-by agreement was suspended.

The next stand-by agreement, signed in April 1994, was designed to generate to a turning point in the economy by 'combining structural measures to revitalise and accelerate the transformation process with financial stabilization measures' (IMF Annual Report, 1994, p 104). In order to achieve this objective, policy targets were designed to generate conditions for sustainable growth, lower inflation, higher international reserves, and a decline in the overall balance of payments deficit. It is argued that the government's main motivation for this agreement was to reschedule its growing debt, not necessarily to achieve the targets (Stone, 2002). Because this agreement could not be completed either, the IMF decided to suspend its financing after the release of the second tranche.

The fourth stand-by agreement between the IMF and Bulgaria, represented by the new Socialist Party government, was signed in July 1996. This programme had ambitious policy targets for a country suffering from an unstable political environment, like Bulgaria. The programme's targets and actual outcomes for this agreement are shown in Table 2. The IMF targeted monthly CPI inflation in December 1996 to be reduced to 2.5%. However, the outcome was about 10 times larger than the targeted value. Although the actual monthly inflation in December 1997 in Table 2 appears to be very small, this is misleading, because Bulgaria experienced hyperinflation in 1997, with an annual CPI inflation rate of over 300%, and a monthly inflation of 242.35% in February 1997. Since the targets were missed, the IMF suspended its financing after the first tranche.

After the 1997 financial crisis, Bulgaria signed its fifth stand-by agreement. Under this agreement, an IMF programme was successfully completed first time. As part of the agreement, a currency board arrangement was implemented. The target and actual values for macroeconomic targets are presented in Table 3. This fifth stand-by agreement was closer to its target outcomes than the former ones for two reasons. First, given the economic and political situation of the country, the IMF set more reasonable targets. For example, the target value of the CPI inflation under the fourth stand-by agreement in 1996 was 105%, whereas the actual outcome was 310.8%. Under the fifth stand-by agreement, on the other hand, the target values were determined based on the political environment of the country (Minassian, 2001). Second, because of the frequent government changes, the governments were not able to fully commit themselves to the reform efforts until 1997. Hence, all prior agreements were incomplete and hence they did not improve macroeconomic performance significantly. In fact, till the fifth stand-by agreement, Bulgaria entered each new agreement in a worse condition than before. Therefore, it is not surprising to observe that the previous stand-by agreements were not successful.

Bulgaria's sixth agreement, signed in 1998, was that of an extended fund facility. This 3-year arrangement aimed to stabilise economic growth at about 4-5% per year, to raise living standards, to reduce inflation and, to improve the current account deficit. In order to achieve these objectives, the parties agreed on the continuation of the currency board regime, as well as employing cautious fiscal policies. Although during the first year most of the targets, that is, output growth rate, were missed, this agreement was also completed successfully.

Bulgaria signed its last stand-by agreement in February 2002 with an amount of SDR240 millions awarded. The country is still under this stand-by agreement. After the last review of this programme, which was completed at 4 February 2004, the IMF approved a fourth credit tranche of SDR26 millions. With this, the total amount drawn totalled SDR214 million. The fourth review of the IMF signalled a strong macroeconomic environment in Bulgaria. With a better macroeconomic and political environment, we expect that IMF programmes should be more successful in Bulgaria.

IMF programmes in Poland

Poland is viewed as one of the most successful transition economies. The authorities were able to deal with hyperinflation problems early in the transition. This success may be the result of the combination of IMF assistance and effective government policies committed to Fund programmes. Figure 2 shows the per capita amount of credit approved and drawn in Poland during its transition process.

Poland, suffering from hyperinflation and declining output level during the early 1990s, signed a stand-by agreement with the IMF in February 1990 with the aim of stabilising the economy quickly through reducing inflation and enhancing the role of market forces. The IMF advice was to employ tight fiscal and monetary policies, using the exchange rate and wage controls as the nominal anchors. Poland was expected to slow down the rate of inflation, to reduce demand pressures, and to strengthen its financial structure. However, high unemployment levels and a steep decline in output caused the suspension of the first stand-by agreement after the third tranche of credit (IMF Annual Report, 1991). In addition, some observers argued that the suspension was also driven by too optimistic programme targets (Stone, 2002).

The IMF's review of Polish economic performance in April 1991 reflected strong approval of the stabilisation and structural reform measures (IMF Annual Report, 1991). Following the review, the IMF and Polish government agreed on a 3-year extended fund facility agreement. This agreement aimed to reduce the budget deficit. Shortly after signing the agreement, both the IMF and the government realised that the agreed targets were too ambitious and the agreement therefore was suspended (Gomulka, 1995). In 1991, the Paris Club offered Poland an immediate 30% debt reduction, provided that the country was under an IMF programme. Although the extended fund facility was not successfully implemented, it nonetheless helped Poland receive a 30% debt reduction.

The Paris Club offered another 20% debt reduction to Poland in 1993, under the condition that Poland would run a successful IMF programme. Hence, Poland signed its third agreement with the IMF. This time, the IMF tolerated higher levels of budget deficit and inflation. Despite this, the Polish government proposed tight fiscal measures, which caused some debate regarding its burden on the Poles. However, the overall outcome of the implementation of Fund programmes was successful, allowing Poland to comfortably meet all the performance criteria.

The last agreement between Poland and the IMF, signed in 1994, was also linked to debt reduction, this time with the London Club. The new programme had a strong emphasis on structural and systematic reforms, such as mass privatisation of 444 large state enterprises and changes in the pension indexation rule. The Polish government had some difficulties in implementing these changes, however, due to their political sensitivity. Since 1994, Poland has not requested any IMF assistance and has managed to pay off all of its debts to the IMF.

A comparison of the implementation of fund programmes in Bulgaria and Poland

The IMF's initial involvement in both countries was very similar. Both countries were experiencing problems due to transition from a centrally planned economy to a market economy. However, the economic performance in Bulgaria and Poland turned out to be quite different. Poland did not request any IMF assistance after 1994 and paid off all its debt to the IMF in 1990s, whereas IMF involvement in Bulgaria still exists today with an increasing debt level owed to the IMF. Before the transition started, Bulgaria was even in a better economic condition than Poland, however. As Table 4 points out, Poland, had a lower GNP per capita in US dollars and lower real GDP growth rates than Bulgaria. Both countries experienced a huge decline in real GNP growth in the 1980s, compared to the 1970s; but Poland managed to perform much better than Bulgaria in the 1990s. An important question is whether Poland's success and Bulgaria's initial painful transition and its success later are affected by IMF policies and programmes. This is a complicated question and the answer depends on two key related issues: implementation and effectiveness of Fund programmes.

With respect to the implementation issue, Poland had an advantage over Bulgaria, partly due to the negative political environment in the latter. During the implementation of the first four programmes in Bulgaria, frequent government changes did not help fulfill the IMF's performance criteria. On the other hand, Polish officials were better motivated to enact target policy changes. Poland also had incomplete programmes, but that was mainly due to overly optimistic expectations of the IMF. However, the implementation of IMF programmes in Bulgaria improved after the 1997 financial crisis when the country signed the currency board arrangement. At the same, the political environment improved significantly in recent years.

Thus, the effectiveness of IMF programmes in both countries has been affected by the ability of governments to commit themselves to Fund programmes, the degree of social-political instability, and the achievability of the agreed targets between the two parties. In the next section, we provide empirical evidence on the ability of Fund programmes on reducing the crisis probabilities with respect to key macroeconomic indicators.

METHODOLOGY AND EMPIRICAL EVIDENCE

Before defining a measure of effectiveness, it is necessary to define what is meant by crisis first. In the literature, crises are generally based on shocks to output growth, inflation, foreign exchange reserves, and exchange rates (Conway, 2000). In this paper, all but exchange rates are used as measures of key macroeconomic indicators, because both Poland and Bulgaria had pegged exchange rates during most of their transition processes, including the currency board arrangement in Bulgaria since 1997.

It is assumed that if an economic variable exceeds a pre-determined threshold level for year t, then a crisis occurs. There are different ways to determine the threshold level. For instance, Conway (2000), in investigating the relationship between IMF programmes and crises for a sample of 90 developing countries, employs a threshold level based on the percentage of participation in certain IMF programmes, which is measured by dividing the number of years under an IMF programme by the total number of years of IMF membership. This measure may not be useful for transition economies, because IME programmes are relatively new in these countries. An alternative approach, which we use in this study, is the method used by Pozo and Amuedo-Dorantes (2003). Their definition of a crisis includes the deviation of an exchange market pressure index from its mean value. A crisis occurs when the value of this index is more than 1.5 standard deviations away from its mean value.

Similarly, it is also assumed here that if a variable is more than 1.5 standard deviations away from its sample mean for the year t, then a crisis has occurred. For example, official foreign exchange reserves and output crises occur when they fall 1.5 standard deviations below their average values. An inflation crisis takes place when inflation in a particular year is 1.5 standard deviations above the sample average. That is, the crisis occurs when [x.sub.it] > [[mu].sub.x] + 1.5[[sigma].sub.x], where [x.sub.it] represents the value of macroeconomic indicator in country i, at year t, [[mu].sub.x] and [[sigma].sub.x] represents the sample mean and standard deviation of the indicator variable, respectively.

In order to minimise the effects of business cycles on our definition of crisis, the data are filtered, using the Hodrick-Prescott filter. This filtering was also useful to smooth the effects of hyper-inflation which occurred in both countries. Poland experienced more than 500% yearly inflation in 1990, and Bulgaria faced more than 1000% annual inflation in 1997.

Next, we need a measure of the effectiveness of IMF programmes. According to the IMF's Articles of Agreement (Article 1), if a country experiences a crisis at year t-1, and asks for the IMF help, then, the IMF programme is expected to decrease the probability of a continuation of the crisis at year t. Therefore, the effectiveness of the IMF programmes can be measured in terms of changes in the crisis probabilities, when a country is assumed to be under an IMF programme. These probabilities can be estimated using the two-way contingency tables. General forms of contingency tables are shown below.

Crisis status at year t-1 Crisis status at year t

 Non-crisis
 Crisis (C) (NC)

Crisis (C) [n.sub.11] [n.sub.12] [n.sub.1*]
Non-crisis (NC) [n.sub.21] [n.sub.22] [n.sub.2*]

Programme status at year t-1 Crisis status at year t

 Non-crisis
 Crisis (C) (NC)

Programme (P) [n.sub.11] [n.sub.12] [n.sub.1*]
Non-programme (NP) [n.sub.21] [n.sub.22] [n.sub.2*]

The cell counts are represented by [n.sub.ij]'s and [n.sub.i*]'s, where [n.sub.ij] represents the number of occurrences in each cell and [n.sub.i*] represents number of occurrences in each row. Based on the contingency tables, conditional probabilities of a crisis status can be estimated using the formulas below.

Prob([C.sub.t]|[C.sub.t-1]) = [[pi].sub.1|1] = [n.sub.11]/[n.sub.1*], Prob(N[C.sub.t]|[C.sub.t-1]) = [[pi].sub.1|2] = [n.sub.12]/[n.sub.1*]

Prob([C.sub.t]|N[C.sub.t-1]) = [[pi].sub.2|1] = [n.sub.12]/[n.sub.2*], Prob(N[C.sub.t]|N[C.sub.t-1]) = [[pi].sub.2|2] = [n.sub.22]/[n.sub.2*]

Prob([C.sub.t]|[P.sub.t-1]) = [[pi].sub.1|1] = [n.sub.11]/[n.sub.1*], Prob(N[C.sub.t]|[P.sub.t-1]) = [[pi].sub.1|2] = [n.sub.12]/[n.sub.1*]

Prob([C.sub.t]|N[P.sub.t-1]) = [[pi].sub.2|1] = [n.sub.12]/[n.sub.2*], Prob(N[C.sub.t]|N[P.sub.t-1]) = [[pi].sub.2|2] = [n.sub.22]/[n.sub.2*]

Those in the first two lines are used to calculate the conditional probabilities for the first contingency table, and the remaining formulas are used to calculate those of the second contingency table. To explain, Prob([C.sub.t]|[P.sub.t-1]) gives the probability of an economic crisis at year t, given that there was an IMF programme in effect at year t-1. An effective IMF programme is expected to decrease the probability of crisis at year t, relative to the case where there is no IMF programme. That is,

Prob([C.sub.t]|[P.sub.t-1])<Prob([C.sub.t]|N[P.sub.t-1])

After calculating the conditional probabilities, we conduct a test of independence to conclude that whether IMF's effects on the crisis probabilities are significant or not. The likelihood ratio test statistic is calculated using the formula below:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

Test statistic: [G.sup.2] = -2 ln([LAMBDA]), which is compared to [[chi square].sub.df].

The degrees of freedom for the log-likelihood test are calculated based on the size of the contingency table. Degrees of freedom for a I x J table is calculated as follows: (I-1)(J-1). Since we have a 2 x 2 contingency table, the degrees of freedom in this case is one. A computed test statistic, which is less than the [chi square] critical value, provides sufficient evidence that there is a significant relationship between IMF programmes and crisis.

Because the effective programmes are expected to decrease the crisis probability, significance and effectiveness are not equal. Thus, an IMF programme is effective only if both necessary and sufficient conditions are met. The necessary condition requires a significant relationship between IMF programmes and the crisis, while the sufficient condition is given by

Prob([C.sub.t]|[P.sub.t-1])<Prob([C.sub.t]|[P.sub.t-1])

That is, the probability of experiencing a crisis at time t, given that there was an IMF programme at time t-1, should be less than the probability of experiencing a crisis at time t, given that there was a crisis at time t-1.

Empirical evidence

Annual data for the CPI, real GDP, and foreign exchange reserves for the period from 1985 to 2002 are obtained through the International Financial Statistics, IFS, CDROM (2004). Foreign exchange reserves are expressed in US dollars, while CPI inflation and real GDP (output) growth rate are expressed in percentages. The descriptive statistics for the variables are presented in Table 5.

Our sample period starts in 1985, because the analysis requires positive cell counts in the contingency tables. In addition, because the IMF programmes in Poland started in 1990, measuring the effectiveness of IMF programmes requires data from the previous years. The inclusion of the pre-transition data should not affect the results, because this period does not include significant crises.

Table 6 presents the estimated probabilities for the crisis at year t and year t-1 for both countries, along with the corresponding estimated probabilities, which are reported in parentheses. To illustrate, for Poland, there were eight cases foreign exchange reserves crises both at year t and year t-1. Accordingly, the conditional probability of experiencing a foreign exchange reserves crisis at year t, given that there was a similar crisis in year t-1, is 88.88%. Similarly, in Poland, there were one inflation and 12 output growth crises both at year t and year t-1 and the conditional probability of experiencing an inflation and output growth crisis at year t, given that there were these types of crisis in year t-1, was 50% for inflation and 92.30% for output growth.

When we look at the results for Bulgaria in Table 6, we observe 16 foreign exchange reserves crises, seven inflation crises, and six output crises both at year t and year t-1. The corresponding conditional probabilities of observing a crisis at year t, provided that there was a crisis last period, are 94.12, 87.50, and 66.67%, respectively.

Table 7 presents the estimated probabilities for a crisis at year t, given that there was an IMF programme at year t-1. Again, both the event frequencies and probabilities (in parentheses) are reported. For example, in Poland, there were three foreign exchange reserves crises, four inflation crises, and four output crises at year t, when there was an IMF programme in effect at year t-1. Accordingly, the corresponding conditional probabilities of observing these crises at year t are 75, 100, and 100%, respectively. Similarly, for Bulgaria, the conditional probabilities of five foreign exchange, six inflation, and three real output crises at year t, given that there was an IMF programme in year t-1, are 70.59, 25, and 66.67%, respectively.

Comparing the results of Table 6 to those of Table 7, we conclude that an IMF programme in the previous year t-1 decreases the crisis probability for Poland for foreign reserves crisis from 88.88 to 75%. For inflation and output in Poland, the crisis probabilities increase, however, especially for inflation. The results indicate that the sufficient condition for the effectiveness of IMF programmes in Poland is met only for foreign exchange reserves. On the other hand, for Bulgaria, IMF programmes reduce the crisis probabilities in all areas, but more significantly so for foreign exchange and output crises. The probability of foreign exchange reserves crises declines from 94.12 to 29.41%, while the probability of real output crisis is reduced from 66.67 to 33.33 %. The probability of inflation crisis goes down to 75% from 87.50%. Hence the results indicate that the sufficient condition for the effectiveness of IMF programmes in Bulgaria is met for every type of crisis. However, we also need to check the necessary conditions to make better inferences about the success of IMF programmes.

In Table 8, we report two alternative test statistics to check the necessary conditions, namely table probability and likelihood ratio tests. The null hypothesis for both tests is independence between IMF programmes in year t-1 and crisis in year t. A reported low P-value rejects the null, which suggests that it is quite likely to observe the occurrence of events, as reported in Tables 6 and 7.

According to the table probability results for Poland and Bulgaria in Table 8, the null hypothesis is rejected only for Bulgaria and only for inflation. The results for the likelihood ratio tests suggest a similar finding: the null can be rejected again only for inflation crises and only in Bulgaria, suggesting that IMF programmes last year are closely related to the inflation outcome in Bulgaria this year. Combining this evidence about the necessary conditions in Table 8 with the evidence on the sufficient conditions given in Tables 6 and 7, we conclude that both the necessary and sufficient condition for the effectiveness of IMF programmes is met only for Bulgaria and with respect to inflation crisis only. Thus, we conclude that IMF programmes are effective in Bulgaria but only for inflation crises. IMF programmes were not effective in Poland.

To check the sensitivity of our results to the definition of crisis, we also allowed for larger standard deviations. We used 1.75 and 2 standard deviations from the mean to redefine crises. This did not change the number of crises significantly. The only change was that Bulgaria had one additional foreign reserves crisis. The rest of the inferences in Tables 6-8 are not affected qualitatively. These results are not reported due to space considerations, but they are available upon request from the authors.

CONCLUSIONS AND POLICY IMPLICATIONS

We investigate whether IMF programmes are effective in transition countries, using the experience of Bulgaria and Poland. We first present historical evidence about the implementation of fund programmes in both countries. We note that Poland had a better implementation track than Bulgaria, because Bulgaria, during the implementation of the early fund programmes, had a higher level of political instability, which negatively affected its commitment to the programmes. Meanwhile, a better political and economic environment in Bulgaria since the signing of the currency board agreement after the 1997 financial crisis has given the government the opportunity to implement more successful policy changes under the IMF agreements than before. As a result, IMF programmes improved over time in Bulgaria. On the other hand, Poland did not ask for any additional IMF advice after 1994.

We also conduct empirical tests that may provide evidence for the effectiveness of IMF programmes. We find that IMF programmes were more effective in Bulgaria than Poland, especially for reducing the probability of inflation crises.

In interpreting these results, it needs to be emphasised that the effectiveness of IMF programmes depends on a host of factors, including the nature of IMF programmes and policies, the commitment of country's officials to perform these programmes, the degree of social and political stability, and external shocks. In addition, as transition economies, both countries experienced substantial shocks during the initial transition period, which coincided with the implementations of IMF programmes. Future research should control for the impact of such initial factors and also provide evidence from the experience of other transition economies. Therefore, our results are preliminary and should be interpreted cautiously.

Table 1: Fund programmes in Bulgaria and Poland

 Bulgaria

 Date of agreement Amount (in SDR millions)

1990
1991 SB (Q1) 279.00
1992 SB (Q2) 155.00
1993
1994 SB (Q2) 139.48
1996 SB (Q3) 400.00
1997 SB (Q2) 371.90
1998 EFF (Q3) 627.62
2002 SB (Q1) 240.00

 Poland

 Date of agreement Amount (in SDR millions)

1990 SB (Q1) 545.00
1991 EFF (Q2) 1,224.00
1992
1993 SB (Q1) 476.00
1994 SB (Q3) 333.00
1996
1997
1998
2002

Source: IMF Annual Report, various years

Table 2: Target and actual figures for the fourth stand-by agreement:
Bulgaria

Measures 1996

 Target Actual

Real GDP growth (%) NA -10.1
Consumer price index (%, December) 105 310.8
CPI inflation (%, monthly, December) 2.5 28.15
Current account balance (% of GDP) 3.1 0.2
Gross official reserves 2.8 1.4
Overall fiscal balance (% of GDP) -4.7 -10.5

Measures 1997

 Target Actual

Real GDP growth (%) 2.5 -6.9
Consumer price index (%, December) 20 578.5
CPI inflation (%, monthly, December) NA 0.23
Current account balance (% of GDP) 2 4.2
Gross official reserves 3.5 5.2
Overall fiscal balance (% of GDP) 2.6 -3.1

Source: IMF, 1997; IFS CD-ROM, 2004; and Bulgarian National Bank

Table 3: Target and actual figures for the fifth stand-by agreement:
Bulgaria

Measures 1997

 Target Actual

Real GDP growth (%) -4.8 -6.9
Consumer price index (%, December) 769 578.5
Inflation (%, monthly, December) NA 0.23
Current account balance (% of GDP) 0.1 4.2
Overall fiscal balance (% of GDP) -3.8 -3.1

Measures 1998

 Target Actual

Real GDP growth (%) 5.7 3.5
Consumer price index (%, December) 14 1
Inflation (%, monthly, December) NA 0.41
Current account balance (% of GDP) 0.4 -1.3
Overall fiscal balance (% of GDP) 0 1.1

Source: IMF, 1997; IFS CD-ROM, 2004, and Bulgarian National Bank

Table 4: Starting conditions (1990 values, unless stated)

 Bulgaria Poland

GNP per capita (USD) 2320.00 1790.00

Real GNP growth (%)
1970-79 average 7.0 5.5
1980-89 average 2.0 -0.7
1990-99 average -0.9 4.4

Money supply/GDP 1.3 0.6
External debt/GDP 50.00 77.00
External debt-service ratio 116.00 56.00

Source: IFS CD-ROM (2004) and UN Statistical Database
(http://unstats.un.org)

Table 5: Descriptive statistics for foreign exchange reserves,
inflation, and real GDP growth (before HP filter)

 Poland Bulgaria

Foreign exchange reserves
Mean 1166.6067 209.9333
Std. dev. 4976.5213 1183.8139
Min -9835.4000 -1145.3500
Max 11035.0000 4454.0000

CPI inflation
Mean 66.9788 106.9730
Std. dev. 133.8360 251.0641
Min 1.8846 2.1571
Max 556.2992 1058.3590

Real GDP
Mean -4.33302 -23.2158
Std. dev. 47.34582 51.46121
Min -182.773 -168.534
Max 56.0918 9.944185

Source: IFS CD-ROM (2004) and authors' calculation

Table 6: Contingency tables: estimated probabilities for crisis in time
t versus crisis in time t-1

 Time t

 Crisis Non-crisis

Poland

Foreign exchange reserves crises
Time t-1
 Crisis 8 (88.88%) 1 (11.12%)
 Non-crisis 1 (11.11%) 8 (88.89%)

Inflation crises
Time t-1
 Crisis 1 (50%) 1 (50%)
 Non-crisis 1 (6.25%) 15 (93.75%)

Real GDP growth rate crises
Time t-1
 Crisis 12 (92.30%) 1 (7.70%)
 Non-crisis 1 (20%) 4 (80%)

Bulgaria

Foreign exchange reserves crises
Time t-1
 Crisis 16 (94.12%) 1 (5.88%)
 Non-crisis 1 (100%) 0 (0%)

Inflation crises
Time t-1
 Crisis 7 (87.50%) 1 (12.50%)
 Non-crisis 1 (10%) 0 (90%)

Real GDP growth rate crises
Time t-1
 Crisis 6 (66.67%) 3 (33.33%)
 Non-crisis 3 (37.50%) 5 (62.50%)

Table 7: Contingency tables: estimated probabilities for crisis in
time t versus IMF programme status in time t-1

 Crisis Non-crisis

Poland

Foreign exchange reserves crises
IMF agreement
 Yes 3 (75%) 1 (25%)
 No 6 (42.86%) 8 (57.14%)

Inflation crises
IMF agreement
 Yes 4 (100%) 0 (0%)
 No 2 (14.29%) 12% (85.71%)

Real GDP growth rate crises
IMF agreement
 Yes 4 (100%) 0 (0%)
 No 10 (71.42%) 4% (28.58%)

Bulgaria

Foreign exchange reserves crises
IMF agreement
 Yes 5 (39.41%) 12 (70.59%)
 No 1 (100%) 0 (0%)

Inflation crises
IMF agreement
 Yes 6 (75%) 2 (25%)
 No 1 (10%) 9 (90%)

Real GDP growth rate crises
IMF agreement
 Yes 3 (33.33%) 6 (66.67%)
 No 3 (37.50%) 5 (62.50%)

Table 8: Results for test of independence: independence of crisis in
time-t versus programme status in time t-1

 Likelihood test:
 Table probability P-value

Poland
Foreign ex. reserves crises 0.2471 0.5765
Inflation 0.5948 1.0000
Growth rate 0.3271 0.5242

Bulgaria
Foreign ex. reserves crises 0.3333 0.3333
Inflation 0.0088 0.0128
Growth rate 0.3801 1.0000

Acknowledgements

We thank two anonymous referees, Ayse Evrensel, Jacky So, and Radcliffe Edmonds, Jr. for their comments on an earlier draft. The usual disclaimer applies.

(1) Albauia, Bulgaria, the former Czechoslovakia (Czech Republic and Slovakia), Estonia, Hungary, Latvia, Lithuania, Poland. Romania, Yugoslavia (Croatia and Slovenia), Armenia, Azerbaijan, Belarus, Georgia, Kaznkhstan, Kyrgyz Republic, Moldova, Russia, Tajikistan, Ukraine, and Uzbekistan.

REFERENCES

Conway, P. 2000: IMF programs and economic crisis: An empirical study on transition. Working paper, University of North Carolina, Chapel Hill.

Csaba, L. 1995: Hungary and the IMF: The experience of a cordial discord. Journal of Comparative Economics 20(2): 211-234.

Evrensel, AY. 2002: Effectiveness of IMF-supported stabilization programs in developing countries. Journal of International Money and Finance 21(5): 565-587.

Gomulka, S. 1995: The IMF-supported programs of Poland and Russia, 1990-1994: Principles, errors, and results. Journal of Comparative Economics 20(3): 316-346.

IMF. 1986-2003: Various Annual Reports. IMF Press, Washington, DC.

IMF. 1997: IMF Approves Stand-By Credit and a Drawing Under the CCFF for Bulgaria Press Release, 97/15.

Pozo, S. and Amuedo-Dorantes, C. 2003: Statistical distributions and the identification of currency crisis. Journal of International Money and Finance 22:591-609.

Przeworski, A. and Vreeland, J.R. 2000: The effect of IMF programs on economic growth. Journal of Development Economics 62(2): 385-421.

Rodlauer, M. 1995: The experience with IMF-supported reform programs in central and eastern Europe. Journal of Comparative Economics 20(1): 95-115.

Sachs, J. 1997: The Wrong Medicine or Asia. New York Times, November 3.

Stone, R. 2002: Lending credibility: The International Monetary Fund and the post-communist transition. Princeton University Press: Princeton, NJ.

Zecchini, S. 1995: The role of international financial restitutions in the transition processes. Journal of Comparative Economics 20(1): 116-138.

BURCU EKE (1) & ALI M KUTAN (2,3,4,5)

(1) Arizona State University, AZ, USA;

(2) Economics and Finance Department, Southern Illinois University Edwardsville, IL 62016-1102, USA. E-mail: akutan@siue.edu;

(3) The Center for European Integration Studies (ZEI), Bonn, Germany;

(4) The Emerging Markets Group (EMG), London, UK;

(5) The William Davidson Institute (WDI), MI, USA