THE LEVEL OF FOREIGN RESERVES IN EU ACCESSION COUNTRIES Jerome - - PDF document

THE LEVEL OF FOREIGN RESERVES IN EU ACCESSION COUNTRIES Jerome Vacher IMF Institute Paper presented at the conference “Exchange rate strategies during EU enlargement”, ICEG European Center, Budapest, November 27-30 2002 This version: November 2002

• Draft. Please do not quote

Abstract

The views expressed in this paper are those of the author(s) and do not necessarily represent those of the IMF or IMF policy. This paper describes research in progress by the author(s) and are published to elicit comments and to further debate.

This paper focuses on the level of foreign reserves in the 10 transition economies candidate to EU accession. The level of reserves is shown to be adequate according to a benchmark based on short term external debt and money supply, as well as according to a more traditional measure. This high level reached during transition can be explained by the effects of trade openness, financial development, lower interest rates and real convergence. In the medium to long run, some determinants of the demand for reserves and the comparison with the European Union point to a lower level of reserves. However, some strong counter forces will push reserves in the other direction on the way to EMU, depending to a large extent on the exchange rate policies followed, but above all on the magnitude of capital flows and the possibility of speculative attacks, notably during the ERM-II period. JEL Classification Numbers: F30, O57, P27 Keywords: Foreign reserves, vulnerability, currency crisis, EU candidate countries Author’s E-Mail Address: jvacher@imf.org

2

• I. Introduction ............................................................................................................................3
• II. The level of foreign reserves: recent issues ..........................................................................5
• A. Currency crises and reserve adequacy ......................................................................5
• B. The demand for foreign reserves and optimal reserves: a recent renewal ................7
• III. Reserves and reserve adequacy in the 10 EU candidate countries ....................................10
• A. The evolution of reserves: some facts.....................................................................10
• B. Reserve adequacy in the EU candidate countries....................................................14
• C. The demand for reserves: some evidence ...............................................................20
• IV. The level of reserves: some implications for the road to EU and EMU............................25
• A. Reserves and convergence ......................................................................................25
• B. The specifics of the road to EMU: implications for the level of reserves...............28
• V. Conclusion ..........................................................................................................................34
• VI. References..........................................................................................................................36
• VII. APPENDIX ......................................................................................................................43

......................................................................................................................................52

3

• I. INTRODUCTION

Foreign reserves play an essential role in all economies and more so in emerging economies dependent on capital flows and illiquid domestic markets. Often, when compared to GDP, they represent a volume of significant economic importance. The level of reserves plays a substantial and growing role for the private sector to assess sovereign creditworthiness and country risk. Recently, there has been a renewed attention to foreign reserves and liquidity that can be linked with several developments in emerging economies and economic analysis. First, the recent emerging market crises showed how reserves, which in many cases were considered to stand at a satisfactory level, could be wiped out in a matter of weeks or even days and force governments to costly devaluations. Sovereign liquidity was not properly evaluated by many market participants (e.g. Mexico, Thailand, notably because of a lack of transparency). Second, in the same vein, it was also noted that some countries which resisted well during the Asian crisis were holding large amounts of reserves1; this has led to some calls for “self protection” or “self insurance” (Feldstein, 1999), in the form, for example, of a higher level of reserves. After the different crisis episodes, some countries also committed to rebuild reserves, even in the case

• f newly floating regimes (e.g. Korea). Third, the economic analysis of currency crises,

especially in an attempt to identify predictive indicators (early warning systems) also led to greater emphasis on liquidity and reserves. Consequently, the effort is now on trying to better

1 Fischer (1999) pinpointed how countries with high reserves survived the Asian crisis (e.g.

Taiwan).

4 assess an adequate, if not optimal, level of reserves given the increasing openness of capital accounts, financial liberalization and development. Transition countries candidate to the EU2 are now, for the majority of them, among the most advanced emerging economies. They had to move, in a short period of time, from no openness and inconvertibility of their currencies, to the status of emerging economy, “converging” economy for some, and finally, member of the EU in a few years’ time (probably 2004 at the earliest). Not only are they moving very quickly from one stage to an other but they have experienced a great variety of economic policies, and more specifically of exchange rate policies: in particular, all the exchange rate arrangements were experimented at some point or another (pegs, currency boards, crawling pegs and bands, floats). A comparative analysis of the level of reserves per se in transition countries candidate to the EU is now needed, with the hindsight of a few years’ of transition . After a short review of recent issues on the level of foreign reserves (Part II), it is shown that the level of reserves is high in EU candidate countries and can be characterized as adequate, even if there are substantial differences in levels between countries; the overall high demand for reserves can be explained by several determinants (Part III). In the long run, it can be inferred that the level of reserves will go down and converge to the EU level. However, how fast this level will go down, will depend on the transition path to EMU and the different exchange rate policies adopted, the size of capital flows to be expected in the region, the risk of speculative attacks, factors mostly pushing for a high level of reserves (Part IV).

2 Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia

and Slovenia.

5

• II. THE LEVEL OF FOREIGN RESERVES: RECENT ISSUES
• A. Currency crises and reserve adequacy

The role of reserves and liquidity was emphasized in studies which tried to explain external vulnerability and to point out the most efficient economic signals to predict currency crises (early warning systems - EWS). This led to a depart from the traditional reserve adequacy measure in terms of imports which can still be applied as the main benchmark for the least advanced economies less integrated in the world financial markets, but along with other measures for emerging economies. As a consequence, reserves in emerging economies are now routinely compared to two main measures of reserve adequacy:

• reserves/ short term external debt: the contribution of higher external liquidity to a decrease in

vulnerability would suggest full coverage of total short term external debt as a practical rule for reserve adequacy for countries which could be faced with redemption or absence of rollover at short notice. A real exchange rate overvaluation and high current account deficits would call for even higher levels of reserves (Bussière and Mulder, 1999). As a policy answer to this specific factor of external vulnerability, the “Greenspan-Guidotti rule” proposes that a country should be capable of living without foreign borrowing up to one year, with an average maturity of external liabilities exceeding a certain threshold e.g. 3 years, and, if possible, a liquidity at risk standard3.

• reserves/ M2: additionally to an “external drain”, led by a high level of short term external

debt and roll over problems, monetary authorities may have to face an “internal drain”, as well.

3 Greenspan (1999). Liquidity at risk rests on the idea of applying value at risk models used in

commercial banks to sovereign analysis, focusing on liquidity and maturity of assets and liabilities; just as a bank would be vulnerable to a run on its deposits (mostly short term), a country could be vulnerable to a run on its external debt especially if it is mostly market determined (e.g. bonds).

6 Following the Mexican crisis especially, it has been suggested to turn the ratio of reserves to money supply as an indicator of reserve adequacy (Calvo, 1996). The ratio of reserves to the broad money supply is the appropriate standard for reserve adequacy for countries with a pegged exchange rate, and can be a good measure for evaluating the potential demand for foreign assets from domestic agents. In effect, most EWS studies have shown a predictive role for such an indicator; moreover, the evidence points to higher reserves/M2 in countries with fixed exchange rates, a probable sign of a greater attention paid by the authorities to this risk4. However, money based measures suffer from several drawbacks (IMF, 2000): for example, in countries where the confidence in money is high and the demand for domestic money larger the ratio of reserves to money tends to be, other things equal, smaller. Therefore, this ratio, when low does not necessarily mean that the country will experience a crisis; it only worked well as an explanation in some countries who had experienced a crisis. In effect, there is no “rule of thumb” devised yet for this measure contrary to reserves on imports or reserves on short term external debt. It should be noted, that the role of liquidity and hence reserves has also been reconsidered in the theoretical framework of currency crises, by so called “third generation” models, but also in first and second generation models. Reserves play an important role in most models: crises are caused by reserve depletion and macroeconomic imbalances (“1st generation”), by a political decision to devalue pushed by a growing cost due to rising interest

4 It may also reflect different policies towards accumulation of reserves, sterilization, and

development of the financial sector through policies modifying the money multiplier; a high ratio may be due to a low M2 which would indicate a late development of the financial system as in the case of most transition economies, to the relative exception of the Czech Republic, Hungary and Slovakia.

7 rates of avoiding reserve depletion (“2nd generation”), by features of the financial system (“3rd generation”), but most of the time the level of reserves is exogenous. Flood and Marion (1999), for example, propose a cross generation framework where the commitment to the fixed exchange rate is made state dependent in the first generation model (the post attack reserve level is not a fixed number anymore, but is chosen optimally period by period to minimize a government loss function) providing a potential reconciliation of first and second generation

• models. Disyatat (2001) suggests an optimising model of currency crisis where an endogenously

determined threshold level of reserves is a function of fundamental economic variables. This model helps to explain why in most countries affected by a currency crisis the minimum level of foreign reserves is not zero contrary to the underlying assumption of traditional models of currency crises.

• B. The demand for foreign reserves and optimal reserves: a recent renewal

The second question which was traditionally addressed in the literature was: what are the determinants of reserve holdings and the optimal level of reserves ? One of the current puzzles is why some countries maintain a high level of reserves. The global level of reserves has significantly augmented, contrary to what could have been expected; at the end of 1999, reserves represented 9% of global GDP, 3,5 times what they were at the end of 1960 and 50% higher than in 1990. The average level for reserves in emerging economies in the 1990s was 14,9%, much higher than in the preceding decades (around 6%); it is also higher than the

8 average for developing countries (12,4% of GDP) and developed countries (7,7% of GDP)5. This is in particular striking in some emerging economies with floating exchange rate regimes, which in theory should be holding a lesser amount of reserves, a phenomenon which was explained by the “fear of floating” (Calvo and Reinhart, 2000). The literature on demand for reserves and reserve adequacy were initially closely linked due to the presence and important role of trade and trade variability in the demand equations. This link between reserve adequacy and demand for reserves has been somewhat severed with the recent crises and the increasing role of the capital account. However, research is under way to better assess the importance of the capital account in the demand for reserves. Flood and Marion (2002) recall this evolution: “In the mid-1960s, the debate about needed reforms of the Bretton Woods system led researchers to ask whether reserves levels were “adequate” and were distributed optimally across countries. In the late 1970s and early 1980s, researchers were interested in whether the “demand for reserves” had substantially changed after the demise of Bretton Woods. They were also curious about whether developed and developing countries differed in their “demand for reserves”. Eventually attention was directed away from reserves holdings by the widespread assumption that international reserves would be stable-and probably low- in an era of increased exchange rate flexibility and very high capital mobility”. Two main theories address the issue of the level of reserves held: the demand of reserve literature and the standard monetary approach to the balance of payments. Models of reserve accumulation focus solely on the portfolio decision of a central bank along the lines of optimal

5 Flood and Marion (2002); the measures are for reserves including gold. Comparing reserves to

GDP provides an indication of the economic weight of reserves in a given country, it is not intended to be an indicator of reserve adequacy.

9 inventory control models or precautionary saving. They apply models of money demand, such as the buffer stock approach. These models allow to assess the determinants of reserves and, for some of them, determine an optimal level of reserves. Heller (1966) was the first to analyse the needed level of reserves in terms of a rational optimising decision, focusing on the precautionary motive for holding reserves (reserves allow to smooth consumption and production in case of balance of payments deficit) and introducing an opportunity cost for holding reserves. Central banks choose an optimal level of reserves to balance the macroeconomic adjustment cost incurred in the absence of reserves with the opportunity cost of holding reserves. Reserve holdings turn out to be a stable function of just a few variables: the adjustment cost, the opportunity cost and reserve volatility (Frenkel and Jovanovic, 1981). Most models and empirical studies6 have then focused on four main variables to explain the holding

• f foreign reserves: external payments variability, marginal propensity to import, scale variables

(output or imports), opportunity cost. One strong criticism addressed to this approach is the fact that the level of reserves might be determined by policies not in the hands of central banks, i.e. that central banks appear to many as not necessarily able to chose their desired level of reserves. In effect, the assumption of perfect elasticity supply of reserves must be fulfilled to justify ex post an isolated estimation of reserve estimation. Partly to take into account this concern, empirical studies have integrated both the demand for reserve literature and elements of the monetary approach of the balance of payments, starting with Edwards (1984), where the evolution of domestic money demand has an influence on the holding of reserves. More

6 See Lehto (1994) for an exhaustive listing of these studies and a discussion of the variables;

for a more detailed review of the early literature see Grubel (1971).

10 recently, Aizenman and Marion (2002) proposed a model where holding foreign reserves can be viewed as a form of precautionary saving for economies with conditional access to global capital markets and costly domestic tax collection; in addition, in the presence of loss aversion (non linearity of preferences), reserve holdings would tend to be larger. Several recent papers have studied with a new eye the demand for foreign reserves, with the introduction of capital account considerations. The buffer stock model of international reserve holdings works about as well in an era of high capital mobility (Flood and Marion, 2002): the prediction that increased volatility significantly increases reserve holdings seems robust, but country specific effects, such as effective exchange rate stability and a country’s financial and real side openness, together with volatility and opportunity cost elements, can explain only about 40% of the variation in countries’ reserve holdings. Trade openness continues to be an important determinant and financial deepening is associated with an increase in the reserves ratio (Burke and Lane, 2001). However, the influence of recent currency crisis episodes on reserve holdings might not be well apprehended by traditional models of demand for reserve; in the case of East Asia, traditional models largely under predict the post crisis and recent accumulation of reserves (Aizenman and Marion, 2002).

• III. RESERVES AND RESERVE ADEQUACY IN THE 10 EU CANDIDATE COUNTRIES
• A. The evolution of reserves: some facts

Figure 1 shows the monthly evolution of non gold reserves in the 10 candidate countries, plus Russia and Ukraine as a matter of comparison, from 1994 to 2002. Poland , the

11 most important accession country, holds the largest amount of reserves. Since the spring of 2001, reserves in Russia have outpaced Poland’s7. Most of the big countries have registered a first build up of their reserves until the first half of 1996, corresponding with a first wave of capital inflows. Russia clearly shows a much more volatile profile of its reserves, linked to the price of oil in some respect, and the start of the depletion of reserves can be seen in mid 1997 well before the “Russian crisis”8. Poland also clearly stands out by the evolution of its reserves. It has experienced well known complications in its monetary management starting in 1994, due to high capital inflows, for some part on an unrecorded basis. This has led to some periods of intense sterilization efforts9. The Polish exchange rate and monetary policy10 had to find a balance between the level of foreign reserves, appreciation of the currency and control of inflation. After the first build up in Polish reserves, a second period of accumulation began in mid 1997, before levelling off at the time of the Russian crisis. It also corresponds with a period of greater exchange rate flexibility (larger bands and then float). In the Czech Republic and Hungary, the trend of reserve accumulation has levelled off much earlier and the tendency looks similar, although there are some differences between the two countries. The currency crisis in the Czech Republic shows a somewhat limited drop of reserves in May 1997, whereas reserves started to decline nearly one

7 Russia holds a significant amount of gold reserves; however, taking them into account does

not change by much the profile of its level of reserves.

8 For an account of the Russian crisis, reserves and the decision to default, see Kharas, Pinto

and Ulatov (2001).

9 Sterilization costs in 1995-1997 were of the order of 1-1,5% of GDP. 10 For an historical overview see Kokoszczynski (2001), and Nuti (2000).

12 year before. After the Russian crisis, reserves in Hungary recorded a decline as the country was said to be more sensitive to the crisis (especially banks) and as the central bank was managing the forint in a narrow band. An index of speculative pressure can indicate it rather clearly (Krkoska, 2000). In comparison, the Czech Republic was not so much affected by the Russian

• crisis. In addition, since the second half of 2001 reserves in the Czech republic have augmented

significantly, with a notable rise in Spring 2002. Overall, it must be recalled that there were only three main episodes of currency crises which ended in devaluations : March 1995 in Hungary, May 1997 in the Czech Republic and September 1998 in Slovakia. Bulgaria also registered a severe currency and monetary crisis prior the implementation of a currency board in July 1997, followed by a successful macroeconomic stabilization. The short number of currency crises and the shortness of time series make it difficult to build early warning indicators for transition economies. Examples of such attempts include Krkoska (2000) using a VAR model for the four biggest countries (Poland, Czech Republic, Hungary and Slovakia), with an index of speculative pressure as a dependent variable, Brugemann and Linne (2002) and Schardax (2002) applying the signal approach. When compared to GDP (Figure 2), countries can be roughly divided in 4 different groups: - more than 20%: Bulgaria, Czech Republic, Hungary, Slovakia;

• between 15 and 20%: Estonia, Poland and Slovenia;
• between 10 and 15%: Latvia and Lithuania;
• less than 10%: Romania, Russia and Ukraine.

The low levels for Latvia and Lithuania may be surprising, in light of their small size and fixed exchange rate regimes. In the case of Latvia, there is a clear declining trend. The build up of

13 reserves in the case of Poland and the decline of Czech reserves in 1997 is even more

• pronounced. The massive surge of reserves in Bulgaria is mainly linked to the adoption of a

currency board in summer 1997 after a period of hyperinflation and rapid currency depreciation; Bulgaria has now the highest level of reserves to GDP in the area, and is clearly atypical. Such diversity of outcomes, however, can not be explained at first glance by differences in exchange rate policies.

14

• B. Reserve adequacy in the EU candidate countries

The traditional measure of reserve adequacy (in weeks of imports as shown in table 1) shows the very comfortable levels reached by Poland (between 6 and 7 months) and Russia. No EU candidate country has a lower ratio than the average of industrial countries. Turkey for example, shows a high level of reserves by this measure, which given the subsequent currency crisis highlights the importance of liquidity considerations to be taken into account and which in fact, in the case of Turkey, had been recognized for several years, including by market participants. The evolution of reserves/M2 (Figure 3) displays much more stable and predictable trends than for reserves/GDP. Bulgaria is again characterized by a high level of reserves but most of the countries have reserves/M2 ratios situated between 30 and 50%. Even countries like Russia and Ukraine, do not stand out contrary to the trend for reserves/GDP. In these countries, as reserves are low, the outcome for the ratios might be related to the low level of M2, consequent to a low monetization of the economy. Reserves cover short term external debt, in all the countries, except Estonia (Figure 4). Following De Beaufort Wijnholds, Kapteyn (2001), a simple benchmark of reserve adequacy can be built including the possibility of internal and external drains, country risk and exchange rate regime. In De Beaufort Wijnholds, Kapteyn (2001), this benchmark appears quite reasonable for the year 1999 and a sample of emerging countries; it is intended not to work as a perfect indicator of vulnerability, as a variety of variables would have to be taken into account (with no guarantee of success in terms of timing). Rather, it is a rough measure of reserve

15

• adequacy. It does not intend to be an indicator for an optimal level of reserves either, as such an

indicator would have to take into account a wider range of variables, notably an opportunity cost. To get an idea of the use of this benchmark for a longer period, the benchmark is built for 5 years (1996-2000), for all the CEEC-10, plus Russia and Turkey11. The two most important variables are M2 (“internal drain”) and short term external debt (“external drain”). The exchange rate regime factor (ER) rests on the assumption that “corner solutions” are preferable: currency boards and independent floats (with the possibility of a run of 5 to 10% of M2, as opposed to 10-20% for managed floats and pegs). The country risk (CR) is built on a scale similar to the Economist Intelligence Unit sovereign index (from 20 to 80), based on Standard and Poor’s ratings in this case to build the time series. The short term external debt by residual maturity (STED) comes from the joint BIS-IMF-OECD-World Bank external debt

• database12. The calculation formula is rather simple:

CR (M2 . ER) + STED A band of adequate reserves can then be drawn, with a higher and lower bound13.

11 It should be noted, that the relevance of the indicator is of course very dependent on a series

• f data, above all the exact amount of short term external debt and less so on the classification
• f the exchange rate regime.

12 series G, H and I: liabilities to banks due within a year, debt securities issued abroad due

within a year, non bank trade credits due within a year.

13 Details for the different variables are in tables 2a and 2b.

16 Reserve adequacy in millions USD (Source: own calculations based on IMF, BIS-World Bank-OECD data)

1996 1997 1998 1999 2000 Reserves < Benchmark(Years) BULGARIA Reserves low bound 1493 929 685 594 634 Reserves high bound 1943 1101 799 716 767 Actual reserves 484 2249 2831 3083 3342 1 CZECH REPUBLIC Reserves low bound 5957 6543 8966 6575 7056 Reserves high bound 6809 7284 9718 7476 7992 Actual reserves 12352 9734 12542 12806 13019 ESTONIA Reserves low bound 148 497 583 1339 1129 Reserves high bound 175 530 605 1366 1159 Actual reserves 637 758 811 853 921 2 HUNGARY Reserves low bound 6729 6791 9213,4 7872 7555 Reserves high bound 7570 7641 10069 8654 8185 Actual reserves 9720 8408 9319 10954 11190 LATVIA Reserves low bound 97 135 295 402 512 Reserves high bound 138 189 352 464 588 Actual reserves 654 704 728 840 851 LITHUANIA Reserves low bound 328 234 511 772 860 Reserves high bound 362 279 552 817 913 Actual reserves 772 1010 1409 1195 1311 POLAND Reserves low bound 5856 6428 9845 10335 10592 Reserves high bound 7884 8580 12386 12826 12761 Actual reserves 17844 20407 27325 26354 25657 ROMANIA Reserves low bound 1855 1900 3058 2071 2347 Reserves high bound 2395 2377 3697 2683 2944 Actual reserves 2103 3803 2867 2687 3922 1 RUSSIA Reserves low bound 30893 37601 22849 13658 14885 Reserves high bound 34696 41908 27059 15198 17104 Actual reserves 11276 12895 7801 8457 24264 4 SLOVAKIA Reserves low bound 1980 2876 3285 2682 2425 Reserves high bound 2531 3409 3833 3250 3008 Actual reserves 3419 3230 2869 3371 4022 1 SLOVENIA Reserves low bound 668 899 967 936 1577

17

Reserves high bound 816 1053 1144 1123 1757 Actual reserves 2297 3315 3638 3168 3196 TURKEY Reserves low bound 18294 25565 29774 31598 38662 Reserves high bound 22320 30183 34878 33999 44229 Actual reserves 16436 18658 19489 23346 22488 5

These results can lead to two main conclusions. First, the level of reserves, has been adequate according to this benchmark. In some cases, the actual level of reserves has been way above the adequacy ratio: such is the case for Poland, Czech Republic, Hungary, Latvia,

• Slovenia. For currency boards, the level is generally more than adequate, especially in the case
• f Bulgaria (the jump to a currency board is visible) and Lithuania. As for Estonia, 1999 and

2000 show a lower than adequate level of reserves; it could be essentially due to a rapid build up in short term external debt, in the wake of high current account deficits and declining FDI

• flows. For Latvia, the importance of short term external debt is linked with its role as a regional

banking center (IMF, 2002)14. In the case of currency boards, the debate remains open on what measures of reserve adequacy should be applied; in the case of transition economies with currency boards, M2 was comparatively small as well as the money multiplier. As currency boards have a level of reserves representing the equivalent of the monetary base, they consequently have a high level of reserves compared to M2. This could change with the development of the financial sector. Second, the systematic misalignment of the level of reserves in Russia and Turkey is striking under this simple benchmark, with an under than adequate level before and after the crisis. In Russia, a higher price of oil, a subsequent boom in

14 Two thirds of the short term external debt consists of non resident deposits matched with

deposits in foreign banks and secure liquid foreign assets in OECD countries.

18 reserves (which continued in 2001) and the conversion of part of the short term external debt in 1998 helped reach a better adequacy recently. It should be added that the measure suffers from several drawbacks, which come after testing it on the transition countries for several years. First and foremost, it is just a benchmark taking into consideration financial and money based measures which does not rest solidly on a set of empirical studies in particular for the money based measures; intuitively, it can be shown that some of the countries which have a high level of reserves are also very open commercially (such as the Czech Republic, Hungary, Bulgaria). Second, because of the method of calculation, for countries with a high short term debt and low M2, and because of the major role of short term external debt, a narrow band of adequate reserves is obtained. Third, it does not take into consideration current account deficits, which can be large and persistent in the region and play a role in currency crises when they reach a high level (in the case of the Czech Republic, Slovakia, coupled with high fiscal deficits and large contingent debt) when it has been suggested to add to a rule based on short term external debt the amount of current account deficit (although this might be too strong a prescription). Still, real appreciation and persistent account deficits in the accession countries would probably push in favour of a higher level of

• reserves. Fourth, it is not intended to function as a vulnerability index; this can be shown in the

case of the Czech Republic, where by this measure, the level of reserves was adequate. However, in case there are other vulnerabilities at play, and there is a systematic misalignment, it can give an indication of the potential for a run and the subsequent deepness of the crisis for countries systematically misaligned (e.g. Russia and Turkey). A one time drop of reserves under the level indicated by the benchmark (Czech Republic, Slovakia) reflects often the

• utcome of a currency crisis or an ongoing crisis; it becomes “worrying” though as this

19 situation is lasting. Lastly, the transition economies show some specific economic developments, notably a quick trade integration before financial integration and financial development (a lower M2/GDP ratio on average than other major emerging economies). As a complement to this measure, a different measure of the possibility of an internal drain is used, with the share of foreign currency deposits in M2, as a substitute for the previous money based measure (Figure 6). In some rare instances among emerging economies, deposit dollarization ratios have steadied in some transition economies or even come down with the stabilization of the economy and development of the financial sector, especially in Poland (Figure 7). Deposit dollarization can be an indicator of the vulnerability of the financial sector and the potential need of central bank intervention in case of distress, as well as an indicator of the perceived stability of the exchange rate by agents. As such, foreign currency deposits can be considered as an additional claim, akin to a part of the foreign debt redeemable at short notice15. Interestingly, the profile and evolution are roughly the same. In a world of higher capital mobility and greater sophistication of capital markets, it is not clearly established how such type

• f measures can capture all the implications for reserve adequacy.

Still, it remains that, overall, reserves in the 10 EU candidates are high and adequate when measured by both traditional ratios or newer methods.

15 The measure is dependent however on the regulations adopted in each country; Slovenia is

the only country which does not allow foreign currency deposits and hence is excluded from the sample.

20

• C. The demand for reserves: some evidence

Turning to the determinants of reserves, tests for the demand of reserves usually tried to find out the variables having an influence on the holdings of reserves which in the present case appear large, and try to explain the differences between countries. However, several characteristics make it even more difficult to test the demand for reserves in transition economies: time series are short; there are frequent changes in policy, and notably exchange rate policy: not only the candidate countries all have different exchange rate policies but they have experienced different regimes during a relatively short span. Perhaps the thorniest issue empirically, if one focuses on a traditional and theoretically fundamental variable in the demand for reserves, is the potential effect of the opportunity cost namely the interest rate. The issue of the opportunity cost is probably the most debated in the literature on the demand for reserves and optimal reserves. Theoretically, it is defined as the difference between the highest possible marginal productivity forgone from an alternative investment in domestic fixed assets and the yield on international reserves. Ben Bassat and Gottlieb (1992) argue that the difficulty of finding a significant opportunity cost effect in many empirical studies is due to measurements which do not correspond to the theory (at least the coefficients can have the right sign, i.e. negative, but in several cases are not significant16). It is

16 Also, frequent multicollinearity makes it difficult to use the interest rate with other variables

such as M2, financial openness or short term external debt.

21 notably difficult to get an adequate measure of the interest rate, be short term, long term, on assets denominated in foreign currency, nominal or real17 18. Both intuitively and based on the numerous past studies on demand for reserves19, the factors that might explain this high level of reserves in EU accession countries and the differences between them are many: trade and terms of trade shocks, growing financial integration and development, real convergence, decreasing interest rates, past volatility of reserves and exchange rates… However, the shortness of time series, uneven availability of data and frequent policy changes make it difficult to assess with all the precision the potential various determinants of reserves. Here, the model for the demand for reserves is assumed to be: Ln R= β0 + β1 Ln ROP + β2 Ln MON + β3 Ln GDP + β4 Ln I + u Where R is reserves scaled by GDP, ROP is trade openness or real openness, MON is M2/GDP as an indicator of financial development, GDP is GDP per capita as a proxy for the

17 One good measure would be a spread on foreign currency assets: eurobond spreads, for

example, but they are not available for all countries at all times. Some studies use local rates in domestic currency but this is imperfect given the exchange rate risk. The use of nominal rates is also problematic; however, for transition countries, the use of real interest rates would make the verification of the influence of interest rates even more specific due to the fact that real interest rates were negative for some years.

18 Some studies use short term interest rates in emerging countries; sometimes the lending rate

is used (e.g. in Flood and Marion, 2002) mostly because of higher availability. In this specific case, it does not take care over of the high intermediation margins found in most emerging countries and specifically in transition economies. Also, short term interest rates can be determined by monetary policy, for example to fight against inflation pressures. Such is the case in some of the most advanced accession countries, above all Poland and Hungary, where the yield curve is inverted because of the so called “convergence play” (expectations of lower inflation in the medium to long term and of EU accession).

19 See Lehto (1994) for a review.

22 level of development and real convergence and I is the spread of lending rates for the

• pportunity cost (see data appendix).

A simple test is run using annual data for 1993 to 2000; due to the small sample of countries and number of observations, pooled time series with cross section data were applied. In order to deal with heteroscedascity and obtain best linear unbiased estimators, Generalized Least Squares (GLS) with cross section weights were utilized. Weighted statistics take into account country specific effects and provide a higher explanatory power. In a first version, short term external debt (as in Burke and Lane, 2001) and a variable for financial openness (as in Flood and Marion, 2002) were also used as explanatory variables but turned out to be

• insignificant. A dummy (D) for currency board countries was also introduced. All the

coefficients are significant and have the expected sign, including the dummy for currency board countries which shows, as expected, a higher demand for reserves20:

Ln R = 2,200 + 0, 168 ln ROP + 0,769 ln MON – 0,232 ln I – 0,376 ln GDP + 0,159 D + resid. (13,420) (2,213) (14,821) (-2,358) (-13,335) (3,607)

20 T-values in parenthesis. See the appendix for more detailed results.

23

Dependent Variable: Reserves Method: GLS (Cross Section Weights) Sample: 1993 2000 Included observations: 8 Number of cross-sections used: 10 Total panel (balanced) observations: 80 Convergence achieved after 37 iteration(s) White Heteroskedasticity-Consistent Standard Errors & Covariance Variable Coefficient

• Std. Error

t-Statistic Prob. Constant 2.199689 0.163907 13.42033 0.0000 Real openness 0.167952 0.075906 2.212627 0.0300 Interest rate

• 0.232325

0.098511

• 2.358368

0.0210 Dummy 0.158664 0.043989 3.606902 0.0006 Financial devt. 0.769220 0.051899 14.82158 0.0000 GDP per capita

• 0.375662

0.028171

• 13.33520

0.0000 Weighted Statistics R-squared 0.997685 Mean dependent var 7.392569 Adjusted R-squared 0.997528 S.D. dependent var 9.525427 S.E. of regression 0.473571 Sum squared resid 16.59597 F-statistic 6377.468 Durbin-Watson stat 0.917341 Prob(F-statistic) 0.000000 Unweighted Statistics R-squared 0.393783 Mean dependent var 2.713461 Adjusted R-squared 0.352823 S.D. dependent var 0.588675 S.E. of regression 0.473573 Sum squared resid 16.59611 Durbin-Watson stat 0.484139

Although more empirical work would be needed, the implications of these results are fourfold :

• trade openness had a positive influence: it is consistent with the increasing trade

integration experienced during the transition years, the presence of several small open economies as well as a transaction and precautionary motive for the financing of trade and potential terms of trade shocks;

• financial development had a high and positive influence; two main and non mutually

restrictive interpretations can be brought forward: first, the monetary authorities follow a precautionary motive and try to compensate the development of the financial sector by holding

24 a larger amount of reserves, notably due to the risk of “internal drain” and potential intervention as a lender of last resort; second, following a standard monetary approach to the balance of payments, higher money demand during the transition period due to macroeconomic stabilization and lower inflation is partly satisfied by means of an increase in reserves (a Krugman type balance of payments crisis in reverse);

• the opportunity cost is linked with the progressive decrease in interest rates; high interest

rates in an individual country have a negative influence on the accumulation of reserves, and, consequently, reserves increase with lower inflation and interest rates21; the opportunity cost is, however, slightly lower than in Edwards (1985) or Landel-Mills (1989) who worked on a sample of emerging economies;

• there is a scale effect: development and real convergence lessens the need to hold
• reserves. A higher level of development can bring a comparatively lesser need of reserves, a fact

consistent with most of the findings in the literature22.

21 As in Flood and Marion (2002) and other studies, all the variables studied are in nominal

terms.

22 See Lehto (1994).

25

• IV. THE LEVEL OF RESERVES: SOME IMPLICATIONS FOR THE ROAD TO EU AND

EMU

• A. Reserves and convergence

If reserves have been broadly adequate until now, it is obviously difficult to assume that they will stay so over the medium term. Reserve adequacy measures will, of course, depend on a large extent on the evolution of the ratio denominators (M2 and short term external debt). It can be expected, especially for countries which have currently low ratios, that with monetization and the development of financial systems, M2/GDP will continue to grow, with, other things equal, a probable corresponding deterioration of the reserves/M2 ratio; however, as there seems to be a partial offset by reserves, the deterioration could be lower as in some countries at least more financial development would also lead to an accumulation of reserves. In fact, and importantly, a deterioration of the reserves/M2 ratio would not matter if the financial system is perceived as more stable and sophisticated as well as the currency. Evidence on the 4 low income countries shows a convergence of reserves/M2 over time (Figure 8) to a level lower than the actual average level in candidate countries. As for short term external debt, it is difficult to assess its evolution; on the one hand, decreasing country risk can lead investors to favour longer maturities; on the other hand, when markets become more efficient and liquid, short term financing is easier. The recent evidence just points to a deterioration of the ratio in some countries (the three Baltic countries, Poland and Slovenia). The evidence from the 4 low income countries (Ireland, Spain, Portugal, Greece) who accessed the EU and then the Euro area, is also that their level of reserves compared to GDP

26 converges to a lower value, but is still higher than for the average of the EU (Figure 9). Also, the evidence points, in fact, more to a divergence of the level of reserves - which was close to the bigger countries average in the 1970s - with a slight return to these countries’ levels . For the “Euro-5”23, the level of foreign reserves is very stable at a level of 4-5% of GDP, which corresponds to a little less than the average for developed countries. Currently, reserves in the Eurosystem stand at 5% of GDP, whereas in the 10 candidate countries they represent 18,5% of GDP, a proportion that has more than doubled since 1993. These evolutions took place comparatively on a longer time span than for the current applicant countries. If there are diseconomies of scale in holding reserves, then the level of reserves in accession countries after having gone up, would go down as a share of GDP, with the real convergence taking place. Real convergence might take some time though and there are still uncertainties on its

• speed. Trade openness seems to have a strong influence on the holdings of reserves, both as a

transaction motive and a precautionary motive. Very open candidate countries tend to hold a higher amount of reserves (such as Czech Republic, Hungary or Bulgaria; Figure 10), and there has been a strong trend towards openness which will probably not break up in the near future even if it might stabilize at some point. In the case of transition economies, the impact of financial openness (Figure 11) appears more difficult to gauge yet, as it is ambiguous (financial

• penness can increase precaution with the possibility of capital flow reversal, and can mean

better access to world financial markets) but it can be expected to play a larger role in the future. Accounting for reserve holdings, might be linked to a complex set of determinants. In the

23 The core countries at the start of the European Community and currently members of the euro

area: Belgium, France, Germany, Italy, and the Netherlands.

27 Aizenman-Marion (2002) framework, the following evolution could be considered as likely for EU accession countries and the possible convergence of reserves: in a first stage of transition, these countries lower their discount rates, have less political instability and less political corruption (an interesting comparison is with the FSU countries), giving a first boost to the level

• f reserves. As overall sovereign risk goes down, tax collection becomes less costly, hence

demand for reserves would go down (adjustments are easier to finance). The issue of a potential effect of interest rates as an opportunity cost is especially important for the accession countries as they register a decrease in their interest rates, from a high level, in the wake of their nominal convergence to the EU and forthcoming accession. More fundamentally, besides the above mentioned estimation problems, the interest rate can be considered as having an influence on the holding of reserves in, actually, two ways: as a traditional opportunity cost, and as a borrowing cost, something which is not well apprehended in the literature on the demand for reserves. On the borrowing side, when borrowing costs decrease, it should be easier for a country to borrow internationally and build up its reserves; however, this borrowing possibility can also lead to an easier “adjustment“ to balance of payments difficulties, and lower or moderate the need to hold reserves in a permanent fashion (especially if the possibility to borrow is perceived as durable). This possibility to borrow at all times might be an explanation of the lower level of reserves held by developed economies24. If this possibility to borrow is not certain, and if the authorities are prone to a higher loss aversion, then the cost of acquiring reserves matter less, and the temptation of borrowing reserves for

24 A similar issue than the “original sin” approach (difficulty to borrow internationally in its

• wn currency).

28 uncertain times may be stronger. Assessing the cost of borrowing is similar to estimating sterilization costs in the case of quasi fiscal costs; borrowing short term to boost reserves can be however costly and ineffective under several circumstances (Kletzer and Mody, 2000). For the most advanced countries candidate to the EU, one could argue that borrowing to build up reserves would make sense, as they pay lower spreads25; on the other hand, if these countries are not vulnerable to a currency crisis, in the sense that there is no perceived shortage of reserves, such a deliberate strategy makes less sense. Higher liquidity can also affect positively spreads (the interest rate is somewhat endogenous) and therefore lower the cost of reserves. In the case of accession countries however, one can infer that the dynamics are different and that the main determinants in lower spreads are the convergence dynamics, and that there is less chance of reverse causality than in

• ther emerging economies; still, high capital inflows can lead to lower interest rates and

growing reserves at the same time. Also, similarly, sterilization policies linked to an inflow of foreign reserves might account for higher interest rates, an additional explanation for the possible endogeneity of interest rates.

• B. The specifics of the road to EMU: implications for the level of reserves

The rhythm at which the level of reserves will evolve will depend to a great extent on the path to EU accession and EMU: that is mostly on the exchange rate policies adopted, the extent of capital flows, and on the vulnerability to currency crises on the way to EMU. It may

25 Although this is more true in the longer end of the curve.

29 be argued that candidate countries might not feel the need to hold so much reserves on the approach to EMU and that they could adjust quickly to close to the EU level. However, several factors may lead to a different evolution with less convergence than

• expected. First, the determinants of reserve holdings might have changed, within a different

environment of high capital mobility and greater openness to capital flows. This changing environment might have two consequences: a greater loss aversion with the use of reserves as a “war chest”, and the vanishing role of the opportunity cost (also as a consequence of the first trend). Precautionary motives might dominate by far: there is a feeling that the level of reserves should not be lowered as a matter of precaution, even now that some floating regimes are in

• place26. For example, when incorporating loss aversion, the probability of crisis would

continue to increase the demand for reserves. In addition, EU accession countries, which will have fully liberalized capital accounts, and as a part of a general trend in emerging economies, might be less sensitive to the opportunity cost of holding reserves, in a world where the effects

• f high capital mobility and higher loss aversion tend to predominate; moreover, “fixers” might

be less sensitive to an opportunity cost. In a situation of convergence, declining interest rates could have a positive influence on the willingness to hold reserves; if insensitive to interest rates, the remaining convergence trend of interest rates will not have an influence on a level of reserves which is high27. More generally, from a practical view, the importance of an

26 Wyznikiewicz (2001) for Poland. 27 For some countries (Poland, Czech Republic, Hungary) which have an inverted yield curve,

the opportunity cost if having an influence is rather conducive, recently, to a relatively higher demand for reserves - if it is the longer end of the yield curve that matters. (continued…)

30

• pportunity cost can be questioned nowadays: what are the true alternatives to holding a large

amount of reserves ? The only true alternative would be to modify the exchange rate policy if this policy of high reserves is judged as too costly. Second, supply factors might be dominant and determine the level of reserves. In front

• f these pressures and a different environment, the reaction function of the central bank might

be different. Indeed, there might be strong counter trends, going against a convergence of the level of reserves to the EU’s. Analyses focusing on the demand for reserves do not include supply side factors to determine the equilibrium reserve holdings. In most analyses, supply side factors (such as the provision of international liquidity by the United States or the willingness of foreign creditors to finance investments) are taken as exogenous. The question can be asked if supply side considerations are not dominant in the case of EU accession countries; faced with high capital inflows, their only tool is the exchange rate; however, fearful of real exchange rate appreciation, the monetary authorities tend to accumulate reserves; limiting real appreciation, they also limit current account deficits and facilitate the accumulation of reserves… High capital inflows can lead to reserve accumulation; they can also lead to potential reversals, in which case those reserves are highly needed, leading to consider the current level of reserves as adequate but not extremely high28. Overall balances have been positive for the most part (Figure 12) in recent years: at least from a strict accounting point of view, they explain naturally part of the build up in reserves. The low income EU countries, especially Portugal, Spain to a lesser

28 For an overview of exchange rate and monetary policies, capital inflows and EU accession,

see Corker et alii (2000).

31 extent have experienced a very important rise in their level of reserves, in the wake of important capital flows at the period of EU accession. In the late 1980s, greater exchange rate stability (narrow ERM), recent accession of Spain and Portugal, successful macroeconomic stabilizations, coupled with the certain prospect of the Single market led to important inflows29. There is evidence that the candidate countries will experience high capital inflows (Lipschitz et alii., 2002), even if it can be assumed that high flows were already experienced by the most advanced countries. Third, a different calendar of enlargement would also mean that the convergence effects would not take place as expected. An important difference between the current accession countries and Greece, Portugal and Spain, is that the time lag between EU accession and EMU, is probably going to be much shorter for the current accession countries; some countries see EU accession in 2004-2005 and EMU in 2007, i.e. two to three years, versus 11 years for Spain and Portugal, more so for Greece and Ireland. If history repeats itself, and if high capital inflows take place around the time of accession (i.e. after 2004 for most countries), the level of reserves might register an additional boost, especially if countries chose to maintain a given level of exchange rate – with the attached difficulties in monetary control. The transition to EMU will be shorter and the convergence trend seen for previous accession countries might not take place. Hence, the question of “sustainable capital account regimes” (Begg and alii, 2001) will gain in importance.

29 1n 1987-1991, capital inflows averaged 5,4% of GDP in Portugal and 4,5% in Spain.;

Ireland’s case was a bit different, however, notably due to its current account profile, already high level of reserves and early EU accession.

32 As a matter of fact, it can be argued forcefully that accession countries might be vulnerable to self fulfilling speculative attacks (2nd generation models) as they are in the ERM II, just as Spain and Italy have been vulnerable as well to tests by the markets30. These tensions could arise because of a combination of sizeable current account deficits, real exchange rate appreciation, and willingness to respect the Maastricht criteria for public finance and inflation. The accession countries might well face a trade-off between exchange rate stability (no nominal appreciation) and inflation targets31. A strategy of nominal exchange rate stability would imply unsterilized interventions (sterilized interventions would be a costlier strategy), accumulation of reserves, a faster growing money supply and prices; on the other hand, a strategy of flexible exchange rates would entail a risk of distress in the financial and corporate sectors, but would be safer and potentially less problematic in case of capital flow reversals (Begg et alii, 2001); in this latter case, the accumulation of reserves would be theoretically less important, if authorities stick of course to an independent float, especially during the ERM II. Nonetheless, it is not sure that the ERM II bands will be sufficiently wide to allow for the effects of prices and cases of sudden reversals. In this regard, the demands of some accession countries for earlier monetary integration might be well founded (Schweickert, 2001) or at least for the preservation of their current arrangements (Gulde et alii, 2000). The fact that most of these countries have benefited from capital inflows considered as more stable does not mean that they are not vulnerable, in

30 For example, Gros (2001). 31 There are strong pressures notably on prices, in particular due to the Balassa-Samuelson

effect; large capital inflows could reinforce real appreciation by the pressure on the nominal exchange rate and the Balassa Samuelson effect, as FDI also leads to higher productivity growth.

33 theory, to a portfolio capital outflow (Buiter and Grafe, 200232), especially at a time when capital account transactions will be fully liberalized and financial markets more developed. It is also probable that the risk of contagion cannot be ruled out, especially if one country is tempted to follow a different path; encouragement would then be given to test other candidate countries which would share some of the same characteristics. The risk of a speculative attacks and the exchange rate policies followed will influence the level of reserves, with the former reinforcing the loss aversion of authorities and precautionary motive for holding reserves. In turn, the level

• f reserves held by each accession country not only means that they are in unequal positions

when facing the risk of speculative attacks but also that it will determine in some part the exchange rate policies followed, e.g. the possibility and the decision to “euroize” early. Accession countries if staying on a trend of high reserves might be well positioned to counter a speculative attack. Both possible speculative attacks and the exchange rate policies followed will play a role

• n the level of reserves with which the accession countries will become part of the EMU.

Considering this, the question of reserve adequacy in the accession countries might be asked again at the outset or during the time of the ERM II. On the one hand, if reserves stay at a high level and a level which is considered more than adequate, the question of the relevance of holding so much reserves might be asked, especially when entering EMU. On the other hand, if reserves are not adequate, there are some mechanisms in the ERM II that will make intervention

32 “What matters here is the capacity or ability (of resident and/or non resident economic agents)

to go short in the domestic currency and to go long in foreign exchange in any of a wide range

• f spot, forward or contingent claims markets. The manner in which capital has, in the past,

entered a country need bear no relationship to the manner in which capital can, at some later date, leave the country”, p 8.

34 compulsory for European central banks; the EU members might be pushed to use their excess reserves to give additional support the monetary authorities for example. There was some debate before EMU on the excess level of reserves in the EU (Gros, Schobert, 199933). This debate was shadowed later on by the relative weakness of the euro, but might well come to the light again if it turns out that despite a high level of reserves EU accession countries require some additional support34, or if accession countries enter the Euro area with an “excessive” level of reserves.

• V. CONCLUSION

The level of foreign reserves is high and adequate in countries candidate to EU accession, contributing to expectations of lesser external vulnerability than other emerging economies and hence an easier transition to EU membership. However, uncertainties remain on the road to EMU and on the policies to be adopted by both current European Union members and accession countries. Uncertainty, in turn, remains an important component of the demand for reserves. If the role of reserves is now better understood, the analysis of the determinants of its level stays a promising field of investigation, as well as the behaviour of central banks.

33 Some calculations were made at the time, reaching for example around 100 billion USD

(Masson, Turtelboom, 1997).

34 For the supporters of unilateral euroisation (Buiter, Grafe, 2002), support should be given to

the countries chosing to acquire the necessary monetary base by a loan of the ECB, which would be become a grant at the time of membership.

35 Notably, further research is needed on the links between the level of reserves, reserve adequacy and financial development. Reserve adequacy, an “old” notion, is itself changing and might continue to evolve, drawing on the analysis of crisis episodes. Better appreciation of reserve behaviour in accession countries provides indeed useful guidelines for other transition and “converging” economies, in the context of higher trade and financial integration.

36

• VI. REFERENCES

Aizenman, Joshua and Marion, Nancy (2002): “The high demand for international reserves in the Far East: What’s going on ?”, NBER Working Paper No.9266 Begg, David, Barry Eichengreen, Laszlo Halpern, Jurgen Von Hagen, Charles Wyplosz (2001): „ Sustainable regimes of capital movements in accession countries“, CEPR report, February Ben Bassat, Avraham and Gottlieb, Daniel (1992): “The effect of opportunity cost on international reserve holdings”, Review of Economics and Statistics 74, 329-332 Bruggemann, Axel and Linne, Thomas (2002): “Are the Central and Eastern European transition countries still vulnerable to a financial crisis ? Results from the signal approach”, BOFIT Discussion paper 5, Bank of Finland Buiter, Willem and Grafe, Clemens (2002): “Anchor, float or abandon ship: exchange rate regimes for accession countries”, CEPR Discussion Paper No.3184

37 Burke, Dominic and Lane, Philip (2001): “The empirics of foreign reserves”, Open Economies Review, 12(4), October, 423-434. Bussière, Matthieu and Mulder, Christian (1999): “ External vulnerability in emerging market economies: how high liquidity can offset weak fundamentals and the effects of contagion ”, IMF working paper n°99/88, July Calvo, Guillermo (1996): “Capital flows and macroeconomic management: Tequila lessons”, International Journal of Finance and Economics, July Calvo, Guillermo and Reinhart, Carmen (2000): “Fear of floating”, NBER working paper 7993, November Corker, Roger, Beaumont, Craig and Iakova, Dora (2000): “Exchange rate regimes in selected advanced transition economies – Coping with transition, capital inflows and EU accession”, IMF Policy Discussion Paper n°3 De Beaufort Wijnholds, J. Otto, and Kapteyn, Arend (2001): “Reserve adequacy in emerging market economies”, IMF working paper n°143, September Disyatat, Piti (2001): « Currency crises and foreign reserves: a simple model », IMF working paper n°01/18

38 Edwards, Sebastian (1985): “On the interest rate elasticity of the demand for international reserves: some evidence from developing countries”, Journal of International Money and Finance, 4, 287-295 Edwards, Sebastian (1984): “The demand for international reserve and monetary equilibrium: some evidence from developing countries”, Review of Economics and Statistics, 60, 495-500 Feldstein, Martin (1999): « A self-help guide for emerging markets », mars-avril, Foreign Affairs n°78 Fischer, Stanley (1999): “On the need for an international lender of last resort”, Journal of Economic Perspectives, vol. 13, n°4, fall 1999, 85-104 Flood, Robert and Marion, Nancy (1999): “Perspectives on the recent currency crisis literature”, International Journal of Finance and Economics 4, n°1, 1-26, reprinted in Guillermo Calvo, Rudi Dornbusch and Maurice Obtsfeld, Money, capital mobility and trade: Essays in honor of Robert A. Mundell, Cambridge; MIT Press, 2001, 207-250

39 Flood, Robert, and Marion, Nancy (2002): “Holding international reserves in an era of high capital mobility”, IMF Working Paper 02/62, April Frenkel, Jacob and Jovanovic, Boyan (1981): « Optimal international reserves: a stochastic framework », Economic Journal 91, p507-14 Greenspan, Alan (1999): « Currency reserves and debt » paper presented at the World Bank conference on reserves management, Washington, April Gros, Daniel (2001): “Five years to the euro for the CEE-3 ?”, CEPS Policy Brief n°3, April Gros, Daniel and Schobert, Franziska (1999): “Excess foreign exchange reserves and

• vercapitalisation in the euro system”, CEPS working document n°128

Gulde, Anne-Marie, Kähkönen, Juha and Keller, Peter (2000): “Pros and cons of currency board arrangements in the lead-up to EU accession and participation in the Euro zone”, IMF Policy Discussion Paper n°1 Heller, Robert (1966): “Optimal international reserves”, Economic Journal, vol.76 (June), pp.296-311

40 International Monetary Fund (2002): “The Republic of Latvia: 2001 Article IV consultation”, IMF Country Report 02/10, January International Monetary Fund (2000): “Debt – and reserve – related indicators of external vulnerability”, paper prepared by the Policy Development and Review Department, March 23, http://www.imf.org/external/np/pdr/debtres/index.htm Kharas, Homi and Pinto, Brian and Ulatov, Sergei (2001): “An analysis of Russia’s 1998 meltdown: fundamentals and market signals”, Brookings Papers on Economic Activity, 1:1 Kletzer, Kenneth and Mody, Ashoka (2000): „Will self protection policies safeguard emerging markets from crises ?“, unpublished. Kokoszczynski, Ryszard (2001): “From fixed to floating: other country experiences: the case of Poland”, paper presented at the IMF High Level Seminar “Exchange rate regimes : hard peg or free floating ?”, Washington DC, March Krkoska, Libor (2000): “Assessing macroeconomic vulnerability in Central Europe”, EBRD working paper n°52, June

41 Landel-Mills, J.M. (1989): “The demand for international reserves and their opportunity cost”, IMF Staff Papers 36, September, 708-731 Lehto, Taru (1994): “The level of a central bank’s international reserves: theory and cross country analysis”, Bank of Finland discussion paper 15/94 Lipschitz, Leslie, Lane, Timothy and Mourmouras, Alex (2002): “Capital flows to transition economies: master or servant ?”, IMF working paper 02/11 Masson, Paul and Turtelboom, Bart (1997): “Characteristics of the euro, the demand for reserves and policy coordination under EMU”, IMF working paper n°58 Nuti, Domenico Mario (2000): “The Polish zloty, 1990-1999: success and underperformance”, American Economic Review, vol.90 n°2, May Schardax, Franz (2002): “An early warning model for currency crises in Central and Eastern Europe”, Focus on Transition, 1/2002, pp. 108-124, Austrian National Bank

42 Schweickert, Rainer (2001): “Assessing the advantages of EMU enlargement for the EU and the accession countries: a comparative indicator approach”, Kiel working paper n°1080, Kiel Institute of World Economics Wyznikiewicz, Dorota (2001): “Problems of optimal reserve level determination”, Bank I Kredyt, January-February, National Bank of Poland

43

• VII. APPENDIX
• 1. Reserves minus gold for 10 Central and Eastern European countries, plus

Russia and Ukraine

(source: IMF)

0.00 5000.00 10000.00 15000.00 20000.00 25000.00 30000.00 35000.00 40000.00 45000.00 1 9 9 6 M 2 1 9 9 6 M 6 1 9 9 6 M 1 1 9 9 7 M 2 1 9 9 7 M 6 1 9 9 7 M 1 1 9 9 8 M 2 1 9 9 8 M 6 1 9 9 8 M 1 1 9 9 9 M 2 1 9 9 9 M 6 1 9 9 9 M 1 2 M 2 2 M 6 2 M 1 2 1 M 2 2 1 M 6 2 1 M 1 2 2 M 2 2 2 M 6 Months USD Millions

BULGARIA CZECH REPUBLIC ESTONIA HUNGARY LITHUANIA POLAND ROMANIA RUSSIA SLOVAK REPUBLIC SLOVENIA UKRAINE

Poland Russia Czech Republic Hungary Ukraine

44 Table 1. Reserves/imports (in weeks of imports)

CZECH REPUBLIC ESTONIA HUNGARY POLAND ROMANIA RUSSIA SLOVAKIA UKRAINE NON OIL DEVELOPING COUNTRIES INDUSTRIAL COUNTRIES 1993 12,7 22,5 27,8 11,3 7,9 9,2 3,3 0,9 21,6 11,5 1994 17,5 13,8 24,3 14,2 15,3 3,7 12,9 3,1 22,1 11,3 1995 27,3 11,8 41,4 26,4 8,0 10,9 19,0 3,4 22,0 11,0 1996 21,9 10,3 31,9 25,0 9,6 7,8 15,6 5,5 23,0 11,5 1997 17,6 8,9 21,2 25,1 17,5 8,5 15,6 7,1 23,6 11,2 1998 21,5 9,1 18,9 30,6 12,6 6,4 10,9 2,7 27,7 10,7 1999 22,6 10,8 20,4 29,9 13,4 10,1 14,7 4,6 29,5 10,7 2000 20,0 11,3 18,2 27,3 15,6 25,7 15,7 n.a. 26,6 10,2

Source: International Financial Statistics Yearbook, IMF

• 2. Reserves minus gold as a % of GDP

(source: IM F)

5 10 15 20 25 30 B U L G A R I A C Z E C H R E P U B L I C E S T O N I A H U N G A R Y L A T V I A L I T H U A N I A P O L A N D R O M A N I A R U S S I A S L O V A K I A S L O V E N I A U K R A I N E 1993 1994 1995 1996 1997 1998 1999 2000

45

• 3. RESERVES/M2

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% B U L G A R I A C Z E C H R E P U B L I C E S T O N I A H U N G A R Y L A T V I A L I T H U A N I A P O L A N D R O M A N I A S L O V A K I A S L O V E N I A 1993 1994 1995 1996 1997 1998 1999 2000

• 4. COVERAGE RATIO RESERVES / SHORT TERM EXTERNAL DEBT

1 2 3 4 5 6 7 8 9 10 B U L G A R I A C Z E C H R E P U B L I C E S T O N I A H U N G A R Y L A T V I A L I T H U A N I A P O L A N D R O M A N I A S L O V A K I A S L O V E N I A

1996 1997 1998 1999 2000

46 Table 2a. Indicators for the calculation of reserve adequacy / EU candidates countries

1996 1997 1998 1999 2000 BULGARIA M2 6914 3297 3511 3758 4201 ER 0,1-0,2 0,07-0,15 0,05-0,1 0,05-0,1 0,05-0,1 CR 0,65 0,65 0,65 0,65 0,63 STED 1044 780 571 472 502 CZECH REPUBLIC M2 42579 37061 37619 36048 37451 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,2 0,2 0,2 0,25 0,25 STED 5106 5802 8214 5674 6120 ESTONIA M2 1235 1475 1516 1798 1955 ER 0,05-0,1 0,05-0,1 0,05-0,1 0,05-0,1 0,05-0,1 CR 0,45 0,45 0,3 0,3 0,3 STED 120 464 580 1312 1100 HUNGARY M2 18687 18875 21409 22364 21010 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,45 0,45 0,4 0,35 0,3 STED 5888 5942 8357 7089 6925 LATVIA M2 1191 1547 1626 1774 2174 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,35 0,35 0,35 0,35 0,35 STED 55 81 238 340 436 LITHUANIA M2 1356 1818 2082 2243 2614 ER 0,05-0,1 0,05-0,1 0,05-0,1 0,05-0,1 0,05-0,1 CR 0,5 0,5 0,4 0,4 0,4 STED 294 189 469 727 808 POLAND M2 50689 53789 63526 66423 67763 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,4 0,4 0,4 0,375 0,32 STED 3829 4277 7304 7844 8424 ROMANIA M2 9829 8670 10425 8747 8525 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,55 0,55 0,6125 0,7 0,7 STED 1314 1423 2419 1458 1751 SLOVAKIA M2 13379 13260 13277 12625 12964 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,4125 0,4 0,4125 0,45 0,45

47

STED 1428 2346 2738 2114 1842 SLOVENIA M2 7397 7732 8883 9341 9002 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,2 0,2 0,2 0,2 0,2 STED 520 744 789 749 1397

Table 2b. Indicators for the calculation of reserve adequacy / Russia and Turkey

(M2 and STED in M USD)

1996 1997 1998 1999 2000 RUSSIA M2 69149 78329 64775 40003 55457 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,55 0,55 0,65 0,75 0,8 STED 27090 33293 18639 12158 17104 TURKEY M2 67100 71051 78523 96957 90897 ER 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 0,1-0,2 CR 0,6 0,65 0,65 0,65 0,6125 STED 14268 20946 24670 25296 33095

48

• 5. Reserve adequacy, 1996-2000

49

50

• 8. RESERVES / M2 (%)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999

Greece Ireland Portugal Spain

Ireland Greece Spain Portugal

• 9. FOREIGN RESERVES

(% GDP)

0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999

SPAIN PORTUGAL GREECE IRELAND EURO 5

Spain Euro 5 Portugal Ireland Greece

51

• 10. TRADE OPENNESS

(IMPORTS AND EXPORTS, % GDP)

0% 20% 40% 60% 80% 100% 120% 140% 160% 180% B U L G A R I A C Z E C H R E P U B L I C E S T O N I A H U N G A R Y L A T V I A L I T H U A N I A P O L A N D R O M A N I A S L O V A K I A S L O V E N I A

1993 1994 1995 1996 1997 1998 1999 2000

• 11. FINANCIAL OPENNESS

(CAPITAL INFLOWS AND OUTFLOWS ON GDP)

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% B U L G A R I A C Z E C H R E P U B L I C E S T O N I A H U N G A R Y L A T V I A L I T H U A N I A P O L A N D R O M A N I A S L O V A K I A S L O V E N I A 1993 1994 1995 1996 1997 1998 1999 2000

52

• 12. Overall balances

53

54

55 Figure 6. Deposit dollarization in EU accession countries

DEPOSIT DOLLARIZATION IN ACCESSION COUNTRIES

10 20 30 40 50 60 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

years % of M2

BULGARIA CZECH REPUBLIC ESTONIA LITHUANIA POLAND ROMANIA SLOVAKIA

56 Figure 7. Reserve adequacy (alternative benchmark)

57

58 Data appendix All series are obtained from the IMF’s International Financial Statistics database, except for short term external debt. Reserves excluding gold. Series “.1LDZF”, quoted in $. Interest rates. Lending rates in series “60C..ZF” were used; the interest rate differential was constructed as ln((1+i)/(1+is)), where is is the US interest rate corresponding to the definition used for the national interest rate.

• GDP. GDP series in domestic currency (series “99B..ZF”), converted in $ using the period

average exchange rate (series “..RF.ZF”). GDP/capita is GDP divided by population (IFS). Real openness . Sum of Imports (series “71.DZF”) and Exports (series “70.DZF”) quoted in $, scaled by GDP.

• M2. The sum of M1 (series 34…ZF) and quasi money (series 35…ZF). The M2 series are

quoted in national currency and converted into $ using the period average exchange rate (series “..RF.ZF”). For Greece, Ireland and Portugal, the sum of currency in circulation, demand deposits and other deposits was used; for Spain, M2 was used (serie 18459MB.ZF). Financial openness (FOP). Sum of capital inflows (sum of absolute value of series “78BEDZF”, “78BGDZF”, “78BIDZF”) and outflows (sum of absolute value of series “78BDDZF”, “78BFDZF”, “78BHDZF”, “78CADZF”). Short term external debt by residual maturity. Series G, H and I: liabilities to banks due within a year, debt securities issued abroad due within a year, non bank trade credits due within a year in the joint BIS-IMF-OECD-World Bank external debt database.

59