土耳其1994年的货币危机

___s I113 POLICY RESEARCH WORKING PAPER The 1994 Currency Crisi s Huge req.iir-erits ,-,,blic sector torroving !r- 1993 and in Turkey early i99-4, combined with major polic\ errors in financitia the deficit. led to Oya Celasuin Turkey's cor-ency crashl in 1994 The World Bank Development Research Group U April 1998 l  POLICY RESEARCH WORKING PAI'ER 1913 Summary findings As a result of Turkey's currency crisis in 1994, output Celasun argues that huge requirements for public fell 6 percent, inflation rose to three-digit levels, the sector borrowing in 1993 and early 1994, combined with Central Bank lost half of its reserves, and the exchange major policy errors in financing the deficit, led to the rate (against the U.S. dollar) depreciated by more than currency crash. As a result of interventions to control half in the first three months of the year. inter-est rates and Treasury borrowing at the same time, Celasun presents stylized facts associated with the the market for domestic borrowing almost disappeared, government's debt-financing mechanisms and other the government turned to monetization for financing, relevant macroeconomic variables to show the system's and the value of the overappreciated Turkish lira inherent fragility at the time of the crisis and to clarify plummeted. the extent to which different factors contributed to the crisis. This paper-a product of the Development Research Group -is part of a larger effort in the group to study currency crises. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please conltact Kari Labrie, room MC3-347, telephone 202-473--1001, fax 202-522-3518, Internet address klabrie @worldbank.org. April 1998. (44 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to enzcordage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polisbed. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center  The 1994 Currency Crisis in Turkey Oya Celasun Macroeconomics and Growth Group Development Research Department, The World Bank Views expressed here are those of the author and does not necessarily represent official opinions of the World Bank. I would like to thank, without implication, Merih Celasun, Cevdet Denizer and Carmen Reinhart for very helpful suggestions.  I. INTRODUCTION The objective of this paper is to present an analysis of the stylized facts associated with the currency crisis in 1994., and briefly interpret the experience in light of the recent crisis literature. Although interest in currency and banking crises have intensified in the aftermath of the ERM, Mexican and more recently Thai crises, the dynamics of the Turkish crisis, -which is different than the aforementioned crises owing to the fact that the exchange rate was governed by a managed float, rather than being fixed - has not been discussed widely in the literature. In the aftermath of the crisis in 1994, the Turkish economy contracted by 6%, the highest level of annual output loss in the history of the Turkish Republic. In the first quarter of 1994, the Turkish Lira (TL) was devalued more than 50% against the US$, the Central Bank lost half of its reserves, interest rates skyrocketed, and the inflation rate reached three digit levels. A stabilization program, later supported by an IMF Stand-By was launched on April 5th, 1994, but no success has yet been achieved in implementing any of the structural adjustment measures. Turkey experienced large and growing fiscal and external imbalances following the capital account liberalization in 1989, until the first quarter of 1994, and during that period the real exchange rate appreciation was no less than 20 %. Against this background of rising and very high PSBR, (about 10 and 12% in 1992 and 1993 respectively) there were remarkable policy mistakes committed on the monetary front. Towards lowering the very high levels of domestic public debt stock through cutting interest rates on Treasury bills, there was a shift towards deficit financing through monetization beginning in the last months of 1993. Several auctions of short term maturity Treasury bills were canceled one after another and the Treasury started to rely on I  cash advances from the Central Bank instead. Still, the announced government budget for 1994 did not contain any measures towards tightening. While these caused increasing levels of anxiety in the financial sector, Turkey's credit rating was downgraded by some major international agencies. The commercial banks that had engaged in heavy offshore borrowing in 1992-93 and held mainly TL denominated assets, hastened the process of acquiring foreign currency to close their open foreign currency positions and there was some capital flight. The Central Bank, aiming to defend the currency and to contain the loss of foreign currency reserves, started to heavily intervene in the interbank market and raised the overnight rate to record levels. Yet, the Central Bank still went on losing reserves - selling foreign currency to the commercial banks, and the commercial banks which were able to buy foreign currency from the Central Bank at relatively inexpensive rates started to lose their own reserves as residents started to withdraw their foreign exchange deposits. The liquidity build-up through excessive creation of domestic credit to the public sector in the form of cash advances to the Treasury by the Central Bank, and the decline in total foreign exchange reserves in the first quarter of 1994 finally had its impact on the parity: from about 15,000 TL/$ in January 1994, the parity more than doubled to 35,000 TL/$ by the first days of April 1994. After the crisis, some academic and policy circles in Turkey strongly defended the idea that the crisis was the natural outcome of the interventions in the domestic borrowing market: had there not been any attempted intervention on the rates at which the Treasury was borrowing (by canceling auctions, fixing the upper limit or by offering small amounts), the crisis could have been 2  avoided'. This view implied the determinants of the crisis to be within the domestic capital markets, and thus neglected the importance of the fundamentals. Others suggested that such a sharp real exchange rate correction was inevitable, given the external imbalances. Another line of argument was that the highly deteriorated fundamentals, namely soaring public deficits provided a backdrop where avoiding such an event was almost unavoidable. The extent to which underlying disequilibria in macroeconomic variables and/or policy errors contributed to the crisis is an issue considered in the analysis of stylized facts. Section II describes the evolution of the external balances following the trade liberalization of the early 80s, and discusses the broad economic conditions during 1990-93. Section III describes the fiscal imbalances and the public sector financing mix during the same time. Section IV describes the events in the financial markets that led to the eventual crash in the first quarter of 1994. Section V summarizes the effects of this substantial devaluation and fiscal tightening on the output of the economy, that is, how the demand and supply side factors interacted and led to the contraction after the first quarter of 1994. Section VI briefly discusses whether the elements of the Turkish Crisis are explained in the recent crisis literature, and section VII concludes. I Ozatay (1996) presents a detailed account of the developments in the Treasury borrowing market in late 1993-1994. 3  II. GROWING EXTERNAL IMBALANCES AND THE BROAD ECONOMIC CONDITIONS IN 1990-93 The 1978-80 debt crisis marked the end of the inward orientation of the Turkish economy and hence the import substitution motive in the trade regime. The 1980-83 period under military rule was characterized by economic stabilization and trade liberalization at the same time. Real exchange rate depreciation and export promoting strategies led to strong export growth. Restrictive wage policies enhanced saving mainly in the public sector, curbed domestic absorption and hence promoted export expansion. The real depreciation of the TL and the repression of real wages continued during the 1984-87 period of civilian administration and supported the trade reforms of the period. Imports were liberalized gradually after 1983, and in 1989 tariffs were lowered to a large extent as part of a program aiming to fight inflation and in 1990 nearly all quantity and price restrictions were removed. From 1984 on, Turkish citizens were allowed to hold deposits denominated in foreign currency. Starting in 1988 and by the end of 1989, the process of capital account liberalization was completed; capital flows were fully liberalized in the external accounts. This reversed the major exchange rate trends that prevailed till then, the cumulative appreciation of the real exchange rate amounted to no less than 20% during 1989-90. The liberalization of capital flows increased the interest rates as well, as in many financial liberalization episodes2. The 1989 tariff 2 See Saracoglu (1996) and World Bank (1997). 4  reductions combined with currency appreciation led to an import boom3 and deteriorated the trade balance in 1990: the deficit doubled in 1990. Table 1 documents the balance of payments and some other key econornic indicators of Turkey, after 1990, when trade and financial liberalization were complete. Figure 1 shows the evolution of the real exchange rate (TL/$) after 1986 together with a fitted simple linear time trend. There has been a sustained tendency for the real exchange rate to decrease over time (i.e. to appreciate), and the real exchange rate was over appreciated, that is, "stayed below its trend", after capital flows were liberalized, from the second half of 1988 to January 1994, when a sharp correction in the nominal rate came to make up for the inflation differential. Figure 2 shows a TL/$-DM real exchange rate index together with a real labor cost index. The sharp increase in the real labor cost that coincided with the real exchange rate appreciation shows how the economy became less competitive after 1988. Domestic inputs (labor) became more costly, and the real cost of tradables declined. The export-led growth of early to mid 80s was replaced with domestic demand led growth and external imbalances widened, the trade deficit went up from 3% of GNP in 1992 to 8.5% in 1993. With regards to the saving - investment gaps implied by the current account deficits of 1990-93; private savings in Turkey during that period have been almost constant, if not slightly increasing, but the absence of inflation accounting implies that private disposable income is not corrected for inflation tax, whereas consumption is deflated fully, which produces an upward bias 3Real wage repression, a politically unsustainable aspect of macroeconomic adjustment during 1981-87 could not be sustained after the 1987 elections. The real wage recovery was rapid; the 1988-89 period saw a sharp increase in wages: 129% in private and 188% in the public sector. (A further strain on the public sector which was burdened by the debt repayments after 1985) Celasun (1995) decomposes the sources of import growth between 85-90 using the input- output framework, and finds that 60% of his growth was attributable to domestic final demand expansion. This trend must have continued well till the import explosion of 1993. 5  in the measurement of private savings. The public sector however, has been in a continuous state of dissaving during the same period. (See Figures 3a and 3b) Given that the currency crisis was triggered by the eventual mismanagement of this growing amount of public sector debt, we now turn to analyzing the public sector. III. FISCAL IMBALANCES AND PATTERNS OF FINANCING IN 1990-93 The liberalization of the capital account in Turkey took place against a background of macro-populism and mounting fiscal imbalances. The public sector borrowing requirement (PSBR) of Turkey rose steadily between 1988 and 1993, as shown in Figures 4-5, where the gap between public sector revenue and expenditure is widening between 1989 and 1993, and the PSBR is increasing along with the primary deficit, which excludes the interest payments of the non-financial public sector. The gap between the PSBR and the primary deficit started to widen after 1992, as interest payments on existing debt became an increasing burden. The reasons for this growth in public expenditures were increases in the total wage bill of the government, generous agricultural support policies, worsening performance of the state owned economic enterprises (SEE), the increased cost of military operations in the southeastern region of the country, and increased interest payments after 1992. The privatization of the SEEs has been a source of "expected public revenue" especially after 1993, but no success in this area was achieved as the Constitutional Court deemed the Privatization Law unconstitutional in 1994. (This ruling of the court was removed in early 1997) After 1989, the borrowings of the public sector became increasingly dependent on foreign 6  savings. It was agreed in early 1989 that the Central Bank's financing of the Treasury would not exceed 15% of the total budgetary appropriations. The Central Bank started to implement a monetary program in 1989, with the aim of restructuring its balance sheet4. The Central Bank was restricting credit to commercial banks too, and liquidity would be created basically against foreign assets. The financing of public sector deficits were shifted to domestic borrowing, and the share of external borrowing was to be reduced. External borrowing was delegated to private financial institutions, mainly to commercial banks, which were the main source of demand for domestic debt instruments. As the foreign exchange purchases of the Central Bank became the main source of money creation, the ultimate source of public debt financing were short term capital inflows. Ekinci (1996) notes that of the 7.2 billion dollars of external debt accumulation in 1990, 3.8 billion was short term, and 60% of that amount was in the form of short-term foreign liabilities of the commercial banking sector. The medium term success of this program in terms of lowering inflation in the absence of any fiscal adjustment however was low: as will be described below, the increased burden of domestic debt and the eventual shift towards Central Bank financing of the Treasury would lead to the crisis of 1994, eventually increasing the inflation plateau. The share of domestic debt (as opposed to external financing) increased until 1992, when it constituted almost all of the financing, but then suppressed in 1993, with an important share in it being Central Bank advances. In the crisis year of 1994 however, domestic borrowing rebounded not only to finance the government deficit but also the repayments of accumulated external debt. Next I turn to briefly describing the patterns of the financing mix during the last years leading up to the crisis. 4 Ekinci (1996) notes that the program mainly involved targeted growth rates for different entries in the balance sheet, and one of the objectives was to reduce foreign currency liabilities to residents, implying reverse currency substitution. 7  The capital flows to Turkey are shown in Figure 6a, and the interest rate differential in Figure 6b. In 1990 the flows were strong, and conditions for domestic bond financing were favorable. In 1991 however the Gulf War took place and led to uncertainties and minor panics in the financial markets. This resulted in increases in interest rates, shortening of debt maturity, and also put constraints on foreign financing. The widening of the fiscal deficit had impacts on the governments financing policy mix and patterns of financing started to change after 1991. In November 1991 there were general elections and the government changed. The new government announced a program aiming at lowering inflation, through reducing the public deficit, but it soon became clear that the high level of interest payments were seen as a potential area for savings. In 1992, facing high levels of domestic debt service payments, the government increased the share of money financing. It used almost all of its short term advances from the Central Bank up to its legal limit during the first half of the year, shifted towards longer maturities in its domestic financing and abandoned its policy of keeping external borrowing at about the level of principal repayments, and borrowed about $1 million in international bond markets. In the second half of 1992 however, it became evident that reliance on short term cash advances from the Central Bank to keep interest rates from rising resulted in pressure on the TL/$ exchange rate, and hence on the Central Bank's foreign exchange position, thus the Treasury accepted a 10% increase on the 3 monthly T-bills and also obtained another $1.5 billion of external funds. With regards to the currency crisis of early 1994, analyzing the debt financing mix of 1993, especially that of the second half, is important. Not only were public sector expenditures 8  booming at that time5, but there was also a shift towards money financing of expenditures, and cancellations of Treasury auctions at the end of 1993. Ozatay (1996) argues that the Turkish government had become "insolvent" already by the end of 1992, and that the timing of the crisis specifically at the beginning of 1994 was due to the interventions in the domestic borrowing market. He tests the stationarity (as a condition for sustainability) of the discounted real domestic debt stock from 1985.07 on, and finds that there is a unit root in the process, no matter whether the sample end point is 1992.12, or 1993.08 or 1993.12. He concludes that the economy was vulnerable to a funding crisis as of the end of 1992, but the loss of confidence caused by developments in the borrowing process of the Treasury at the end of 1993 determined the timing of the crisis. Comparing the case of Turkey with the funding crises in several European countries in the 1920's ( France, Belgium, Italy ,Portugal and Greece - most of which had actually corrected their fiscal fundamentals by that time), he concludes that such crises were triggered by problems in debt management policy: namely by offering interest rates at less than market clearing levels and attributes the timing of the Turkish crisis to the debt mismanagement of late 1993 to early 1994. The first half of 1993 saw an increased burden of interest payments on domestic and foreign financing of 1992. Yet the Treasury's budgeting program restated the goals of lowering the interest payments and lengthening the maturity of domestic borrowing. Still, the majority of financing came from short term borrowing, the interest rates on which were gradually reduced. 5 It should be noted that a significant reason for the large public expenditures of late 1993 and early 1994 was the local elections to be held on March 27,1994. Because of these elections, the prices on SEE produced goods were not increased "on time", and changes in these were lagging well behind the inflation rate. The public, having learned from past experiences, knew with certainty that these government administered prices would go up in a few days after the local elections. 9  (By 13% on 3-monthly, and 2% points on 6-monthly bills) The rates on longer term maturities were increased. $1.7 billion was borrowed in international markets, and half of the legal limit on Central Bank advances was used. It was the second half of 1993 that saw a great deal of policy changes. In the beginning of 1993, there was a change in party leadership of the leading coalition partner DYP, and Ms. Ciller was elected as PM in mid June. Towards the end of July, the Central Bank governor resigned as a result of disagreements between him and the PM on the conduct of monetary policy. In the meanwhile, it was often stated by the government that the most important short term policy goal was to lower the burden of the share of interest payments on short term debt, by lowering the nominal interest rates. Thereafter until the beginning of 1994, instead of trying to correct the fundamentals that led to this problem, the government tried to control the interest rates, that is attacked the symptoms of the problem rather than the cause. This attempt of trying to lower interest rates on debt at such high levels of PSBR (12.4 % in 1993) proved to be a very dangerous one. VI. MONETIZATION, DOLLARIZATION, AND THE TURBULENCE IN THE FINANCIAL MARKETS IN LATE 1993-EARLY 1994 Based on two laws passed in August and October 1993, the Short Term Cash Advance (STA) facility6 of the Turkish Central Bank to the Treasury was extended. In August 1993, the 6 STA is the facility through which the Central Bank extends domestic credit to the Treasury, that is, the public sector. The Treasury in turn, gives interest free paper to the Central Bank. 10  accumulated debt of the Central Bank to the Treasury due to former STA's. were canceled, and through an annexed budget, an additional 26.8 trillion TL was made available to the Treasury. (The original limit was 28.8 trillion) The developments at the domestic borrowing market as described below explains the reasons for the Treasury to use up not only all of these resources by the end of 1993, but also 53% of the legal limit for 1994 (TL 52.4 trillion) in only the first 3 weeks of 1994. Figure 7 shows the path of domestic credit extended by the Central Bank to the Treasury. The sharp and sustained increase after October 1993 until April 1994 shows the extent of liquidity that was pumped in to the system. Figures 8a and 8b show how during the same months the Central Bank was losing its foreign exchange reserves. To assess the extent of excess liquidity in the system, a money demand equation was estimated in the fashion of Kaminsky and Reinhart(1996). In the model, the measure of excess real balances are associated with sustained positive residuals in a money demand equation (the measure used is narrow money, i.e. Ml), implying money creation over and above money demand7. The residuals of the regression are indeed positive during the whole of 1993, and the last two months of 1992 implying excess liquidity in the system prior to the crisis. The possible sources of the sustained positive residuals are discussed and results of the regression and a graph of the residuals are presented in the appendix. 7 Kaminsky and Reinhart (1996) use this measure as one of the many indicators that may signal a forthcoming balance of payments or banking crisis. The motivation is due to possible excess supply prior to the crisis due to deficit financing a la Krugman (1979), or a decrease in money demand a la Calvo and Mendoza (1996). The pattern emerging from this indicator in their cross-country framework is not clear, and they note the shortcomings of the money demand estimation, including measurement and stability issues. As for the 1994 Turkish crisis however, the regression results, -subject to the problems mentioned above-, seem to be clearly indicative of excess liquidity in the economy before the crash. 11  During the second half of 1993, the Treasury continued to finance the growing deficit, but at the same time limited the share of 3-month paper to an insignificant amount. The average maturity of domestic debt stock was on a steadily declining path till then, as shown in Figure 9. Against that background, the rates on 3-month bills were suppressed down 4-5% points, while 6- 12 monthly borrowing increased with increased rates of 2-3% points. External funding amounting to US$2 billion was obtained and reliance on Central Bank financing increased. In the last quarter of 1993, the Treasury started to cancel auctions altogether. In November 1993, 4 out of 5 were canceled, namely those with 3, 6 and 9 monthly maturities. Similarly, in December 1993, no bills with maturities shorter than a year were auctioned. The auctions of September, October, November and December of 1993 are summarized in Table 2. The cancellations of auctions, and the acceptance rates on 3, 6, 9 monthly maturity auctions points towards the often announced aim of the Treasury: to save on interest rates, and to increase the maturity. ' The low amount of offers by the market participants for all maturities, in the last two months of 1993 (shown under the entry "Amount Offered") compared to the previous two months is striking. The "% Accepted" entry refers to the percentage of these offers that were accepted by the Treasury. The amount of borrowing from 3 and 6 month paper was zero, and for the 9 and 12 month bills, both the offered amounts and rates accepted were low. As domestic debt rollover was substituted with monetization, credit to the govenment from the Central Bank amounted to half of the annual legal limit of 1993, only in the last quarter of 1993. This corresponded to 30% of the international reserves of the Central Bank at that time. (The 8The timing of this attempt and perhaps the attempt itself- instead led to much higher interest rates to prevail after the crisis of March 1994, since it induced a heavy reliance on monetization and triggered several runs on the TL. 12  evolution of Central Bank Net Foreign Assets is shown in Figures 8a, 8b.) The Treasury, after having canceled the auctions in November 1993 on the basis of high rates (for example 73% in November) would return to the domestic borrowing market in January 1994, having used up half of the 1994 legal limit on STA's in only the first 3 weeks of the year. this time by fixing the maximum interest rate at 94% in simple annual terms. Yet the demand for these bills were extremely low. The announced budget for 1994 contained no measures for fiscal correction, price increases on SEE goods were delayed but expected to take place immediately. after the local elections in the first couple of days of April 1994, thus inflationary expectations were high (Figure 10 shows the evolution of WPI and CPI ) and residents were abandoning TL denominated assets in favor of foreign ones. Currency substitution, as proxied by the ratio of M2Y to M2 where M2Y is a broad money measure including foreign currency deposits is shown in Figure 11. The share of domestic currency deposits in M2Y fell from 53 percent to 42 percent between December 1993 to April 1994 9. The evolution of foreign currency deposits at commercial banks is shown in Figure 12. The decline the stock amount of deposits during the third and fourth months of 1994 corresponds to the period of large withdrawals, and then the return to TL denominated assets, mainly government paper, in May 1994. The domestic borrowing market would in fact disappear until May 1994, when the Treasury managed to borrow substantial amounts again, but at compounded annual rates around 400%. Rates on 3, 6 and 9 montly bills is shown in Figure 13. Returning to the situation in early 1994, as the domestic borrowing market of the Treasury collapsed, the mechanism through which commercial banks borrowed at foreign markets, and at 90ECD Economic Survey on Turkey, 1995, p.23. 13  the domestic market bought mainly T-bills to reap huge profits was temporarily broken. The domestic credit extended by the Central Bank to the Treasury reached record amounts. 53% of the limit of STA's were extended in only the first 3 weeks, which approximately corresponded to 30 % of the net foreign assets of the Central Bank, This liquidity pumped in to the system at a time when the demand for TL was low, led to several runs against the TL in January, February and March 1994. The Central Bank, prior to- and in between the official devaluations, was heavily intervening in the overnight market to defend the parityv