资本流动，宏观经济管理和金融体制：土耳其，1989-97

WPs JI'v, POLICY RESEARCH WORKING PAPER 2141 Capital Flows, Between 1989-97, large private capital flows to Turkey Macroeconomic contributed to economic growth Yet chronic and high fiscal deficits - coupled with the Financial System an inconsistent financial sector regulatory framework - left the banking system Turkey, 1989-97 and the economy vulnerable to capital flow reversals and Oya Celasun external shocks Cevdet Denizer Dong He The World Bank Europe and Central Asia Region Poverty Reduction and Economic Management Sector Unit July 1999  POLICY RESEARCH WORKING PAPER 2141 Summary findings Recent developments in a number of emerging one of the key issues of Turkey's macroeconomic economies have heightened interest in the relationship management. Only by reducing its interest expenses can between macroeconomic management and financial fiscal deficits be reduced and greater stability be regulation, in an environment of open capital accounts achieved. and large-scale movements of private capital. The Turkish banking system, in becoming increasingly Celasun, Denizer, and He analyze the Turkish integrated with international financial markets, has experience with capital flows in a macroeconomy become vulnerable to shifts in market confidence. Banks characterized by chronically high inflation and fiscal borrowed abroad in response to macroeconomic deficits. They study the relationship between capital imbalances to benefit from high interest rates on flows, macroeconomic management, and vulnerability in domestic loans and government paper. In the process, the financial system. the banks have exposed themselves to interest rate risk, Their analysis highlights the importance of fiscal policy to foreign-exchange risk, and to large credit risks. in an era of large capital flows. Fiscal imbalances To reduce the Turkish economy's vulnerability to contributed both to real exchange rate appreciation and external shocks, financial regulation must be high real interest rates in Turkey. The high interest rates strengthened simultaneously with the achievement of the government must pay on domestic debt have become macroeconomic stability. This paper-a product of the Poverty Reduction and Economic Management Sector Unit, Europe and Central Asia Region - is part of a larger effort in the region to examine the relationship between capital flows and economic management. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact LianaNathaniel, room F3P-198, telephone 202-458-9569, fax 202-974-4396, Internet address lnathanielCifc.org. Policy Research Working Papers are allso posted on the Web at http://www.worldbank.org/html/dec/Publications/Workpapers/ home.html. The authors may be contacted at celasun@econ.umd.edu, cdenizer@ifc.org, or dhe@imf.org. July 1999. (58 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage 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 polished. 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  Capital Flows, Macroeconomic Management, and the Financial System: The Turkish Case, 1989-97 by Oya Celasun University of Maryland Cevdet Denizer International Finance Corporation/World Bank and Dong He International Monetary Fund When this paper was written, Dong He was Country Economist in the Poverty Reduction and Economic Management Sector Unit in the Europe and Central Asia Regional Office (ECSPE), World Bank; Oya Celasun was Summer Intern in ECSPE, and Cevdet Denizer was Economist at the World Bank Development Institute. Comments from Merih Celasun, Ajay Chhibber, Wafik Grais, Roumeen Islam, and Samuel Otoo are gratefillly acknowledged. Cevdet Denizer thanks to Amar Bhattacharya for providing financial support and encouragement during the early stages of this paper. JEL Classification Numbers: E2, E4, E5, E6, F3, F4 Keywords: Turkey, capital flows, external vulnerability, financial regulation, macroeconomic management Author's E-Mail Address: dhe(imf. org, cdenizer@ifc.org, celasun@econ.umd.edu  -2 - I. Introduction Turkey liberalized its capital account in 1989, taking an important step in its efforts to integrate its economy with the rest of the world. Since then capital inflows increased significantly and the financial system became increasingly linked with external markets. While in most other emerging market economies financial opening was accompanied or preceded by fiscal and structural reforms, this was not the case in Turkey. The new regime began under conditions of chronic and high inflation that averaged about 70 percent in the late 1980s. Due to persistently poor public finances, this situation continued throughout the 1990s and achieving stabilization still remains a challenge. The financial sector has been operating in this unstable environment and reforms to ensure its soundness have been uneven. Liberalization of interest rates, easing entry of new foreign and local banks, and permits for new products and services in the early 1980s led to a more dynamic financial sector. However, improvements to the legal and supervisory frameworks to anticipate and deal with weaknesses in the banking system have not been fully sufficient. Transparency, consolidated supervision, orderly exit mechanisms, and accounting standards are still policy issues to be addressed. Turkey's experience with capital flows raises interesting questions. To begin with, what factors account for capital inflows with poor fiscal fundamentals? What are the characteristics of such flows and how did they impact consumption, investment and growth? An analysis of economic management during the 1989-1997 period in general and the post 1994 crisis period in particular is also called for, in order to draw policy lessons and to offer some thoughts for a possible stabilization program. There are also important issues related to the financial sector. How did capital flows, the volatile economic environment, and regulatory policies affect the system's operations? Has the financial sector become weaker and exposed to higher currency and credit risk as a result of real exchange rate targeting policy? These are the main questions we attempt to answer in this paper. In light of the current global financial turmoil, our goal is to assess emerging vulnerabilities that relate to capital flows in the Turkish financial system. Our overall message is that capital inflows contributed to economic growth through their positive impact on private consumption and investment, but also rendered monetary policy ineffective and inflation path unchecked, given the particular policy mix of real exchange rate targeting and high fiscal deficits. Although the authority's conscious policy in recent years of maintaining external competitiveness helped reduce the currency's vulnerability to speculative attacks, it led to an economy that lacks a nominal anchor. Capital flows also enabled the country to delay its structural reforms. The failure of successive governments to deal with structural problems resulted in the economy operating under a cloud of vulnerability, particularly in terms of the worsening debt dynamics. In other words, the underlying causes of stop-go stabilization policies, which generate considerable uncertainty about the overall policy framework, and volatile economic growth, are still present.  -3- The overall macroeconomic situation, which sets the broad incentive structure for the operations of banks, has affected the financial sector. As a result of predictable depreciation of the currency, banks borrow at cheaper rates abroad and lend it at high rates domestically, expecting to earn more in net interest income than they lose from currency depreciation. As a result of the moral hazard created by extensive government guarantees and the lack of effective prudential regulation and supervision, the banks take on substantial risks in their interest rate exposures due to maturity mismatches, in unhedged foreign currency positions, and in potential loan losses. Such risks and the existence of large confidence-sensitive claims by nonresidents make the banking sector vulnerable to business cycles and shifts in market sentiment. Although the state banks have much less exposure in market risks, their operations are distorted by their heavy involvement in quasi-fiscal activities and the failure of the government to reimburse the banks for the subsequent "duty losses". Their losses are a major source of liquidity risk to the banking system. The rest of the paper is organized in three main parts. First, we analyze capital flows data and the determinants of capital flows and their impact on the real sectors of the economy. Seconcl, we consider macroeconomic management issues relating to capital flows. Thirdly, we review the interactions of capital flows with the financial sector. Section II presents the basic capital inflows and outflows data and, discusses the main characteristics of these flows and provides a decomposition of the capital account. In section III the setting for capital account liberalization and the initial conditions with regard to the key macroeconomic aggregates are presented, followed by some econometric evidence on the determinants of capital flows. Section IV discusses the effects of capital flows on consumption and investment. Section V turns to macroeconomic management issues, including the exchange rate, monetary and fiscal policies, that relate to capital flows and point out macroeconomic vulnerabilities. Section VI turns to the financial sector and analyzes banking intermediation of capital flows and issues surrounding the process. Section VII concludes. II. The Characteristics of Capital Flows 11.1 The Record Despite the unstable macroeconomic environment, capital inflows to Turkey increased steadily after 1990, 'with net capital inflows reaching more than four percent of GNP in 1996 and 1997. At the aggregate level, the volatility of capital inflows reflected the volatility in economic activity. Turkey's GNP grew by more than six percent each year in the 1990s, except in 1991 (the year of the Gulf Crisis which saw GNP growing by only 0.3%) and in 1994 (the year oif a severe currency crisis when GNP contracted by six percent). In parallel, total net capital inflows were also above two percent of GNP each year, except in 1991 and in 1994., when there was a net capital outflow. In comparative terms, the relative level (in terms of GNP) of net capital inflows to Turkey has been higher than that to Brazil before the implementation of the Real Plan, close to the level of capital inflows to Indonesia, but lower than those to Mexico and Thailand. In  -4 - terms of debt stock, at the end of 1996, Turkey's total extemal debt reached 43% of GNP, as compared to 24% for Brazil, 49% for Mexico, 50% for Thailand, and 60% for Indonesia. Table 1 Turkey: Net Capital Inflows (US$ million) 1990 1991 1992 1993 1994 1995 1996 1997 Financial Account 4,037 (2,397) 3,648 8,963 (4,194) 4,643 8,763 8,616 Direct Investment 700 783 779 622 559 772 612 554 Portfolio Investment 547 623 2,411 3,917 1,158 237 570 1,634 Equity (45) 56 300 431 994 120 198 (42) Debt 592 567 2,111 3,486 164 117 372 1,676 Other Investment 2,790 (3,803) 458 4,424 (5,911) 3,634 7,581 6,428 of which: short-term 3,000 (3,020) 1,396 3,054 (5,127) 2,305 5,945 1,761 Monetary Authority (130) (1,060) 336 1,024 1,397 1,632 1,339 1,097 General Govrnment 503 330 (1,310) (1,953) (2,516) (1,991) (2,232) (1,406) Banks 1,510 (2,199) (374) 1,265 (4,612) 1,692 4,494 1,256 Other Sectors 835 (880) 1,806 4,088 (180) 2,301 3,980 5,481 Net Erros & Omissions (469) 948 (1,190) (2,222) 1,766 2,355 (1,782) (2,523) Sources: International Financial Statistics and Central Bank of Turkey. T able 2 _Turkey: Net Capital Inflows (% of GNP) 1990 1991 1992 1993 1994 1995 1996 1997 FinancialAccount 2.7% -1.6% 2.3% 4.9% -3.2% 2.7% 4.7% 4.4% Direct Investment 0.5% 0.5% 0.5% 0.3% 0.4% 0.5% 0.3% 0.3% Portfolio Investment 0.4% 0.4% 1.5% 2.2% 0.9% 0.1% 0.3% 0.8% Euity 0.0% 0.0% 0.2% 0.2% 0.8% 0.1% 0.1% 0.0% Debt 0.4% 0.4% 1.3% 1.9% 0.1% 0.1% 0.2% 0.9% Other Investment 1.8% -2.5% 0.3% 2.4% -4.5% 2.1% 4.1% 3.3% of which: short-term 2.0% -2.0% 0.9% 1.7% -3.9% 1.3% 3.2% 0.9% Monetary Authority -0.1% -0.7% 0.2% 0.6% 1.1% 1.0% 0.7% 0.6% General Govrnment 0.3% 0.2% -0.8% -1.1% -1.9% -1.2% -1.2% -0.7% Banks 1.0% -1.4% -0.2% 0.7% -3.5% 1.0% 2.4% 0.6% Other Sectors 0.5% -0.6% 1.1% 2.2% -0.1% 1.3% 2.2% 2.8% Net Erros & Omissions -0.3% 0.6% -0.7% -1.2% 1.3% 1.4% -1.0% -1.3% Sources: International Financial Statistics and Central Bank of Turkey. Table 3 Net Capital Inflows of Selected Countries (% of GNP) 1990 1991 1992 1993 1994 1995 1996 1997 Mexico 3.3% 8.2% 7.6% 8.6% 3.9% -4.4% 1.3% - Brazil -1.3% -1.3% 1.6% 1.8% 1.5% 4.3% - - Thailand 10.8% 12.2% 8.7% 8.6% 8.7% 13.3% 10.8% -10.0% Indonesia 4.1% 4.6% 4.6% 3.7% 2.3% 5.3% 5.0% 0.7% Source: International Financial Statistics.  - 5 - There were a number of salient features of capital flows to Turkey. First of all, capital flows have been very much a two-way phenomenon. While inflows have been much larger than outflovvs, Turkish investments abroad were also substantial in a number of years. Turkish banks made large investments in foreign assets before 1994, and Turkish portfolio investments in foreign securities (particularly debt securities) reached more than one billion US dollars in 1996. Given that the current account balance in Turkey has been on average less than two percent of GNP in the 1990s, Turkey cannot be said to be a large capital net importer. Secondly, net foreign direct investments in Turkey have been very small - it has never been more than 0.5% of GNP in the 1990s. This is a somewhat puzzling observation given that Turkey has a v,ery dynamic private manufacturing sector, and Turkey is a major manufacturing base for a number of important multinational corporations (e.g. in the automobile industry). Thus, although there may well be a reasonably good presence of international firms in Turkey, such presence has not brought about a large financial inflow with it. Thirdly, portfolio investments have not been as important as deposits, loans and trade credits. In most years of the 1990s (except in 1997), deposits and credits were dominated by short-term flows. Within the portfolio investment category, investments in equity securities have been less important than investments in debt securities. It is likely that such debt portfolio inflows were mostly bonds issued abroad by the Turkish government and private sector residents rather than foreign investment in domestic Treasury bills and government bonds. There is not much information available as to the magnitude of foreign investment in domestic government securities. Fourthly, the public sector has been a net repayer to rather than a net borrower from the foreign sector, though it still holds the bulk of the external debt stock, particularly that of the medium-long term external debt. Fifthly, foreign loans and credits made directly to non-financial private sector borrowers seem to have been more important that the loans and credits made through the banking sector. In terms of the distribution of the external debt stock, the banking sector's foreign debt had been much larger than the non-financial private sector before 1994, but the situation has changed considerably since. By the end of 1996 and 1997, the external debt stock of the non-financial private sector was twice as large as that of the commercial banking sector. It should be noted, however, that the loans and credits borrowed by the non-financial private sector have, often carried guarantees by the domestic banking sector, as we shall see later. As a result the banking sector is not less exposed. Finally, there is considerable uncertainty regarding the size of short-term flows in and out of Turkey. Apiart from the fact that there have been sizable "net errors and omissions", the current account records very large unclassified invisible earnings. Whereas such items could perhaps be largely accounted for by so called "shuttle trade" before 1996, the balances  -6 - for 1996 and 1997 have been estimated with "shuttle trade" already being taken into account in merchandise exports. Therefore there could be substantial amount of short-term flows which were captured by the official statistics on "short-term" capital inflows. In terms of the volatility of the flows, it is noteworthy that, during 1989-1997, foreign direct investments had the lowest volatility in all category of flows, whereas net errors and omissions had the highest volatility. Loans, deposits and trade credits were more volatile than portfolio investments, but most interestingly, there was no substantial difference in volatility between short-term loans and deposits and long-term loans and deposits. Table 4 Turkey: Volatility of New Capital Flows (Coefficient of Variation) 1989 1990 1991 1992 1993 1994 1995 1996 1997 89-97 Capital Account Balance 4.2 0.5 -1.6 1.0 1.0 -1.8 1.9 0.9 0.9 2.1 Foreign Direct Investment 0.6 0.5 0.5 0.5 0.6 0.6 1.2 0.9 0.9 0.8 Portfolio Investment 0.9 1.6 2.4 1.0 1.4 3.2 19.5 7.5 3.3 2.8 Other Long-term Flows -2.6 -10.8 -4.3 -1.8 1.3 -1.9 -30.9 1.8 0.4 6.2 Other Short-term Flows -4.1 0.7 -1.2 2.2 2.0 -0.8 2.9 0.8 3.2 5.6 Net Error& Omissions 2.0 -6.8 3.2 -2.4 -2.3 5.5 1.8 -2.0 -2.1 -15.5 Source: Own calculation based on monthly central bank balance of payments statistics.  -7 - Table 5 Turkey: External Debt (in billions of US Dollars; end of period) Old Series New Series 1992 1993 1994 1995 1996 1996 1997 (By borrower) Medium-Long Term 42.9 48.8 54.3 57.6 59.2 64.1 69.6 _ Public Sector 39.8 42.8 48.2 50.0 48.8 51.1 49.2 of which: Consolidated EBudget 25.8 28.3 30.4 31.1 30.2 31.4 30.8 State Owned Enterprises 5.1 5.4 5.5 4.8 4.4 4.9 4.7 Central Bank 6.2 6.6 8.6 10.5 10.7 10.7 10.3 Private Sector 3.2 6.0 6.1 7.6 10.4 13.0 20.4 of which: Banks - - - - - 2.7 5.1 Non-financial Companies - - - - - 10.2 14.1 Short Term 12.7 18.5 11.3 15.7 20.5 20.5 22.6 Central Bank 0.6 0.7 0.8 1.0 1.0 1.0 0.9 Deposit Money Banks 7.2 11.1 4.7 6.6 8.5 8.4 8.5 Other Sectors 4.9 6.7 5.8 8.0 11.0 11.1 13.2 (By type of credit) Medium-Long Term 42.9 48.8 54.3 57.6 59.2 64.1 69.6 of which: Project and Program Credits 21.8 21.8 25.2 23.6 22.1 - Bond Issues 9.3 12.6 13.8 14.2 14.8 - Short Term 12.7 18.5 11.3 15.7 20.5 20.5 22.6 Credits 10.1 15.4 8.0 11.2 15.0 - Credits for Imports 2.6 4.8 3.8 5.4 8.3 - Pre-Export Credits 0.9 1.1 1.4 1.6 1.6 - FX Credits to BEnks 5.1 8.7 2.2 3.2 3.9 - Other 1.5 0.8 0.6 1.0 1.2 - Deposits 2.6 3.1 3.3 4.5 5.5 - Source: Turkishn Treasury. 11.2 Decomposition of the Capital Account The balance of payments identity allows decomposing the capital account surplus (i.e. net capital inflows) plus statistical discrepancies , into the current account deficit, and the accumulation of official foreign exchange reserves. The current account can further be  -8 - decomposed in to the net resource balance deficit, that is net trade in goods and services, plus net factor payments and transfers. Table 6 shows the decomposition of the flows until 1997 in millions of US$, whereas Table 7 shows the decomposition as percentages of the Capital Account, inclusive of net errors and omissions. The negative sign on most of the elements of the Net Factor Payments and Transfers column in Table 6 indicates a surplus (net inflow) on the account. Although a sharp increase in the level of the trade deficit took place after 1989, no clear pattern emerges on the allocation of the capital account among its three components in the pre 1994 period. The last three years in the table show a larger allocation to reserve growth compared to the current account. This is consistent with the post crisis policy of keeping the real exchange rate constant, which resulted in heavy foreign exchange intervention. Since inflows were particularly strong in 1995-1997, the extent of reserve accumulation has been substantial, as seen in Table 6. In the pre-crisis period, in the years which saw heavy inflows such as 1993 and 1990, larger shares were for current account deficit allocation as opposed to reserve accumulation. Chart 1 shows the evolution of the capital account and its allocation between current account and reserve accumulation. Chart 1 Allocation Of Capital Account, Million $ 8 600 400 200 -200 -400 -600 -8000 uD ~co O a) 0- CC '3) CV SI'C R) I CC ZC 0 00 a C ))O CO CD C) 03 c) aJ) CD CC) cV 13 Reserve Accumulation n Current Account m Capital Account incl Net Errors and Omissions The large and increasing size of recorded trade deficits of the recent years (which now include an estimate of shuttle trade) with respect to capital inflows are noteworthy. Financing of the trade deficits was made possible by the strong positive balance on the invisibles account, especially non-interest and non-tourism revenues. Part of these were likely to be short-term capital flows, which could be subject to sudden reversals. The  - 9 - relatively healthy current account position of Turkey therefore relies on unrecorded flows, which may pose problems in the future. Table 6: Allocation of Capital Account, US $ Billions Capital Capital Account incl Reserve Current Net Resource Net Factor Account Net Account Payments Balance Errors and Accumulation Balance Balance Deficit and Transfers Omissions 1986 2.1 2.3 0.8 -1.5 3.1 -1.6 1987 1.9 1.8 1.0 -0.8 3.2 -2.4 1988 -1.0 -0.7 0.9 1.6 1.8 -3.4 1989 0.8 1.8 2.8 1.0 4.2 -5.2 1990 4.0 3.9 1.3 -2.6 9.6 -6.9 1991 -2.4 -1.3 -1.0 0.2 7.3 -7.6 1992 3.6 2.5 1.5 -1.0 8.2 -7.2 1993 9.0 6.7 0.3 -6.4 14.2 -7.7 1994 -4.2 -2.4 0.2 2.6 4.2 -6.8 1995 4.6 7.0 4.7 -2.3 13.2 -10.9 1996 8.8 7.0 4.6 -2.4 10.6 -8.2 1997 8.6 6.1 3.3 -2.8 15.5 -12.7 Table 7 Allocation of Capital Account, % of Total Reserve Accumulation Current Account Net Factor Payments Net Resource Balance and Transfers Deficit 1986 35 65 -72 137 1987 55 45 -137 182 1988 -126 226 478 -252 1989 153 -53 -288 234 1990 33 67 -176 243 1991 80 20 593 -574 1992 60 40 -294 333 1993 5 95 -115 210 1994 -8 108 282 -174 1995 67 33 -155 189 1996 65 35 -117 152 1997 55 45 -209 254 III. The Liberalization of the Capital Account and Determinants of Capital Flows  -10- The surge in capital flows to developing countries in the early 90s are believed to be partly associated with common external factors, such as the recession in industrialized countries, and low interest rates in the United States. Yet, the countries receiving the largest share of capital flows were also those that had undertaken fundamental fiscal and structural reforms. In contrast, liberalization of the capital account and the increase in capital inflows took place against a background of considerable macroeconomic imbalances in Turkey, such as deteriorating fiscal fundamentals, and high inflation. The absence of a stronger and persistent increase in the flows to Turkey in the early 1 990s as experienced in other developing countries points towards the relative importance of domestic and regional factors in determining the flows. To better understand the determinants and effects of foreign capital inflows to Turkey, we first describe the macroeconomic setting under which capital account liberalization took place. The path of fiscal deficits was of great importance with respect to the timing of capital account liberalization, and financing patterns were in turn affected significantly by the easing of the external borrowing constraint. We will then present econometric evidence on the determinants of various components of the capital account. 11I.1 Capital Account Liberalization: The Process and the Motive After experiencing a severe debt crisis in 1978-80, Turkey abandoned its inward oriented policy stance and embarked on an export oriented growth strategy. The key elements of this change have been trade, capital account and financial sector liberalization. By the mid-1980s all quantitative restrictions on trade were lifted and only minimal controls on the current account remained.' The impressive export performance in the early 1980s benefited to a great extent by significant alterations in relative prices (Celasun and Rodrik,1 989), which served to enhance Turkey's creditworthiness in international capital markets. Turkey maintained a competitive real exchange rate throughout the 1981-88 period, which was supported by a repressed real wage regime (Celasun, 1990, Boratav, 1990). With increased political contestability from 1987 onwards, however, repression of real wages became politically unsustainable. The real wage boom of 1989-90 and further populist wage policies from then on had adverse impacts on Turkey's public finances, which were already burdened by a sizable external debt servicing burden in the latter half of the 1980s. The policy of maintaining a competitive real exchange rate, which helped the private sectors export performance in the mid 1980s, implied capital losses on foreign debt and a deterioration of the terms of trade of the public sector vis a vis the private sector. Failing to achieve a counterbalancing improvement in the primary stance, the government abandoned the real exchange rate rule in 1989, and after that, the exchange rate appreciated in real I The drive to liberalize foreign trade culminated in a customs union with the European Union in early 1996.  -11- terms.2 (See Chart 3) The appreciation of the exchange rate not only eased the servicing of foreign official debt, but the slower crawl of the exchange rate also implicitly served as a nominal anchor and helped to control inflation in an environment of deteriorating fiscal deficits. Against this background, the move to fully liberalize the capital account started in 1989 and through a series of decrees Turkey accepted IMF's Article VIII in 1990. This marked the completion of the external financial liberalization process, which was initiated in 1984 when Turkish residents were allowed to hold foreign exchange denominated deposit accounts. While the goal of capital account liberalization was put forward as further integration with international capital markets, and in particular the European Union, Celasun and Arslan (1996) suggest that easing of the financial constraint on surging public expenditures was en important consideration underlying this decision. Events in 1989 seem to confirm this view. In that year, chronic inflation became the major issue on the policy agenda. Aiming to limit the monetization of fiscal deficits, the Central Bank and Treasury came to an agreement to constrain the Central Bank financing to 15% of total budgetary appropriations. With the share of net external financing by the public sector also being rather limited, domestic borrowing became the main source of financing the deficits. Yet, with the exchange rate following a path of real appreciation, lending by the domestic banks to the public sector was based on a rapid build up of short term foreign debt. Ekinci (1996) notes that, with the Central Bank creating reserve money mainly against foreign reserve accumulation, amd external borrowing being delegated to domestic financial institutions, short term capital inflows became the ultimate financing source of fiscal deficits. 11I.2 The Determinants of Capital Flows: Some Econometric Evidence To capture the main determinants of net capital inflows, we follow the literature and regress capital flows on a constant, the uncovered interest differential between Turkish three month T-bills and the TL equivalent of three-month LIBOR rate, and the growth rate of real GDP or the industiial production index. Estimation was done for total, portfolio and short term capital flows, using monthly data for 1990-1997 as well as foreign direct investment using quarterly data for the period 1990-19973. The results are presented in Table 8. 2 No objective behind the move to a (managed) float from a real exchange rate targeting rule was ever officially announced. With increasing capital inflows, the feasibility of real exchange raLte depreciation had decreased substantially and the Central Bank seemed not to have another option but allow for some real appreciation. 3 Since portfolio flows are mainly composed of debt instruments placed by Turkish residents in foreign markets, foreign interest rates are likely to be most relevant measure of opportunity costs of these flows. Therefore we broke down the uncovered differential into the continued  - 12 - As expected, the uncovered T-bill interest differential is significant in explaining short term capital flows. Total capital flows turned out to be significantly explained by the lagged first difference of the interest differential, not its lagged level. Growth rate of real GDP, however, does not significantly (at 10 percent level) affect short term or total capital flows. The insignificance of the growth of real GDP variable stands in contrast to some other country studies (see for example Corbo and Desormeaux 1996) and seems to be suggestive of the fact that, the most important pull factor of capital flows is the short run interest rate differential rather than growth opportunities in the economy. While capital flows do significantly affect the real variables in the economy as we discuss below, we were unable to find significant effects of the dynamics of GDP growth on short term or total capital flows. For portfolio flows, the foreign interest rate is a significant regressor, with a negative sign. As the opportunity cost of placing debt securities in foreign markets increase, this component of capital flows are negatively affected. This is consistent with the observation that most flows in this category are debt related rather than equity investment. For foreign direct investment, it turns out that, only the lagged real GDP growth rate is significant, other than the constant. This is an expected result, because these types of flows are necessarily longer term, and not related to short run arbitrage opportunities in the financial markets. LIBOR rate and the US equivalent of the T-Bill rate, and included these two in our regression.  - 13 - TABLE 8 Dependent Variaible: Total Capital Flows (TCF) Variable Coefficient T-Statistic c 234 3.18** TCF(-1) 0.40 4.06** AUIP(-1) 4794 2.62** G(-1) 803 1.20 RL= 0.23 Durbin-Watson Statistic= 2.18 F-Statistic=8.00 (P-value=0.00) Dependent Variable: Short Term Capital Flows (STC) Variable Coefficient T-Statistic c 67 1.20 STC(-1,) 0.42 2.35** UIP(-1) 2199 1.76* G(-1) 131 0.21 AR(1) -0.16 -0.81 RL= 0.13 Durbin-Watson Statistic= 2.09 F-Statistic=2.94 (P-value=0.03) Dependent Variable: Portfolio Flows (PI) Variable Coefficient T-Statistic c 57798 2.09** PI(-I) 0.02 0.17 LIBOR(- l) -56234 -2.04** TB3(-l)i -1181 -1.57 G(-1) -66 -0.19 RI= 0.09 Durbin-Watson Statistic= 1.94 F-Statistic=l1.99 (P-value=0. 10) Dependent Variable: Foreign Direct Investment (FDI) Variable Coefficient T-Statistic c 162 5.82** FDI(-1) 0.02 0.13 UIP(-1) 9.96 0.13 G(-1) 142 2.56** R2= 0.23 Durbin-Watson Statistic= 1.78 F-Statistic=2.60 (P-value=0.07) Note: **: Significant at 5 %, * :Significant at 10 % . IV. Impact Of Capital Flows On the Real Economy Capital flows and macroeconomic developments clearly have impacts on one another through various channels. For example, the high growth rates of GDP in the most recent  - 14- years are by and large associated with growth in private consumption expenditure biased towards durables, and also private investment, which increased demand for imports of intermediary goods and external financing.4 Increased capital inflows due to financial developments may stimulate aggregate demand by increasing the stock of loanable funds in the financial system, and domestic credit. In Turkey, the path of economic growth has been closely associated with the amount of capital inflows. (See Chart 2). We attempt here to analyze whether capital flows have an independent impact on different components of aggregate demand, once other standard determinants are controlled for. In doing so, we follow the approach of Kamin and Wood (1997) applied to the case of Mexico, and a cross section of other Pacific Basin countries. We estimate separate econometric models relating consumption and investment to a standard set of determinants, and capital inflows. We also attempt to distinguish the effects of capital flows on various breakdowns of aggregate demand, such as private and public consumption, and private and public investment. Chart 2 GDP Growth and Capital Flows -6 -6000 f-- GOP Growth Rate (e6) lett scale -Capital Account (Mill. USS), right scale IV. 1 Consumption We use a standard set of easily quantifiable determinants of consumption in our consumnption equations. While consumption should be positively related to income, and negatively to the real interest rate, the availability of credit may also be positively related to See SPO (1998).  - 15 - consumption. Here, we used real M2 (inclusive of foreign currency deposits) as a proxy for the stock of bank loans. We also add the capital account to our equation. Since capital flows are assumed to affect real variables indirectly through their effect on interest rates and the availability of credit, the inclusion of these variables in the regression would be expected to reduce the coefficient on capital flows. The basic equation we estimate is as follows: (1) C = Po + PGDP + P2rir + P3M2 + 4KA + E We estimate this equation for private and public consumption separately, using quarterly data for the period 1987-96. The first column in Table 3 shows the estimated coefficients when the equation excludes real M2.5 While real income and capital flows are very significant and positive, the real interest rate is found to negatively effect private consumption. The broad results are not significantly altered when real M2 is added to the equation, as seen in the second column. Real M2 enters the equation with a significant positive coefficient. When the availability of credit as proxied by real M2 is controlled for, the significance of the real interest rate declines, but not by a large extent. Interestingly, the results are altered considerably when public consumption is considered. While one would expect that easier financing due to capital inflows would increase public consumption, the effect of capital flows on public consumption are insignificant. The results are shown in the last two columns of Table 10. The inclusion of real M2 does not alter this result. Although the real interest rate becomes significant along with real M2, capital flows are sltill insignificant. We estimated error correction versions of the equations as well, and finally estimated a parsimonious version of the equations, removing insignificant explanatory variables. The results are broadly the same: private consumption is positively related to capital flows, but public consumption is not. The results are in Table 11. 5 The estimation method used was maximum likelihood allowing for an AR(1) error structure.  - 16- Table 10 Results for Real Consumption Dependent Variable Private Consumption Public Consumption Without M2 With M2 Without M2 With M2 Constant 4211 2244 334 -23.10 (7.81**) (3.47) (2.11**) (-0.19) Real GDP 0.49 0.49 0.06 0.04 (24.96**) (22.42**) (7.68**) (3.64**) Real Interest Rate -1870 -1321 -703 -2188 (-2.23**) (-1.72*) (-1.58) (-5.10**) Real M2 2.26E-09 1.08E-09 (2.67**) (3.82**) Capital Account 0.13 0.13 0.004 -0.02 (2.96**) (3.10**) (0.20) AR(I) 0.69 0.51 0.94) (4.76**) (3.12**) -0.62 -0.72 (-5.03 **) ( 5.92**) 0.97 0.97 0.51 0.65 Durbin-Watson Statistic 2.26 2.05 2.08 1.80 F-Statistic 266 237 8.69 12.08 p-value of F-Statistic 0.00 0.00 0.00 0.00 Note: T- Statistics in parentheses. * :significant at 10% level, **:significant at 5% level  - 17- Table 11 Results for Real Consumption, Error Correction Version Dependent Variable Change in Private Change in Public Consumption Consumption Constant 936 665 (1.23) (0.94) Real GDP(-1) -0.05 -0.03 (-1.33) (-0.82) Real Interest Rate(- 1) 2288 -1118 (1.26) (-0.67) Capital Account(-1) 0.08 0.038 (1.28) (0.61) AReal GDP 0.47 0.019 (15.8**) (0.69) AReal GDP(-1) 0.09 0.09 (4.37**) (4.58**) ACapital Account 0.17 0.03 (3.06**) (0.59) RL 0.98 0.62 Durbin-Watson Statistic 2.39 3.15 F-Statistic 292 8.44 p-value of F-Statistic 0.00 0.00 Note: T-Statistics in parentheses. * :significant at 10% level, **:significant at 5% level IV.2 Investment Our results for investment seem less robust to specification. We still use the three monthly deposit rate as a proxy of the opportunity cost of funds, due to the lack of reliable lending rate data. Although the levels of lending and deposit rates are different, the variation in the deposit rate would be expected to track the variation in the lending rate reasonably well. The results of the basic equation estimation for private fixed capital formation are in the first two columns of Table 12. Though the capital account does not significantly affect private fixed capital formation whether real M2 is included or not, the real interest rate becomes significantly negative only when it is included. But the sign of the coefficient of real M2 is negative, which is not plausible. The results for fixed capital formation by the public sector, as shown in the last two columns of Table 11 are not different. Capital flows do not significantly enter any of the two equations, while the implausible sign on real M2 persists. Error correction versions of the estimation are summarized in Table 13. It can seen that public investment still is not related to capital flows, but capital flows, in past level, or past changes, do effect private fixed capital formation. Based on the error correction  - 18 - specification results, we can not reject the significant effect of capital flows on private investment, although the effect is not as clear as it is for consumption. Table 12 Results for Real Fixed Capital Formation (FCF) Dependent Variable Private FCF Public FCF Without M2 With M2 Without M2 With M2 Constant 3180 80134 969 -1551 (2.72**) (0.03) (2.43**) (3.88**) Real GDP 0.08 0.08 0.04 0.09 (4.60**) (5.13**) (2.5**) (4.12) Real Interest Rate 810 -152 -6089 -4616 (0.78) (-3.40**) (-2.74**) 0.14) Real M2 -3.08E-09 -1.9E-09 (-1.65) (-2.85**) Capital Account -0.01 -0.04 -0.03 0.007 (-0.19) (-0.82) (-0.58) AR(1) 0.91 0.99 (0.14) (12.4**) -0.40 -0.52 (I16.81** (-2.25**) ( 3.08**) Ri 0.88 0.89 0.22 0.37 Durbin-Watson Statistic 1.90 2.15 1.59 1.51 F-Statistic 61.8 51.3 2.44 3.89 p-value of F-Statistic 0.00 0.00 0.07 0.01 Note: T-Statistics in parentheses. * :significant at 10% level, **:significant at 5% level  - 19- Table 13 Results for Real Fixed Capital Formation Error Correction Version Dependent Variable Change in Private Fixed Change in Public Fixed Capital Formation Capital Formation Constant 1517 2473 (4.25**) (3.95**) Real GDP(-1) -0.07 -0.12 (-4.76**) (-4.36) Real Interest Rate(-1) 3024 1763 (3.06**) (1.02) Capital Account(- l) 0.13 0.11 (2.56**) (1.29) AReal GDP(-1) -0.007 0.15 (-0.57) (6.65**) ACapital Account(-1) 0.10 -0.05 (2.16**) (-0.61) 0.72 0.58 Durbin-Watson Statistic 2.18 2.83 F-Statistic 16.6 9.30 p-value of F-Statistic 0.00 0.00 Note: T-Statistics in parentheses. * :significant at 10% level, **:significant at 5% level V. Macroeconomic Management During Capital Flows As already indicated earlier, Turkey opened up its capital account under conditions of large fiscal imbalances and high and chronic inflation in sharp contrast to the experience of East Asian and most Latin American countries. What were the effects of these inflows on important variables such as the exchange rate and interest rates, and how did Turkey manage its economy under these circumstances? In this section we first review the evolution of exchange rates and consider how they interacted with the overall economy. Next, we review fiscal and monetary policies with a view to assessing whether there was an internal consistency among these policy tools given the capital inflows and the openness of the economy. V. 1 Capital Flovvs and the Exchange Rate The recent literature on capital flows to developing countries has shown that capital inflows are associated with the appreciation of the real exchange6. To consider the evolution 6 See for example Calvo, Leiderman and Reinhart (1993), (1996),and Fernandez-Arias and Montiel (1995).  -20 - of the real exchange rate Chart 3 plots three series. Like many Latin American countries that experienced heavy capital flows, it is clear that Turkey also experienced real exchange rate appreciation7. The abandonment of the real exchange rate rule in 1989 coupled with liberalization of the capital account indeed led to sharp real appreciation during 1989-90 which continued at a more moderate scale until 1994. With no fiscal adjustment and the persistence of high inflation, choosing the exchange rate (govemed with a managed float) as an implicit nominal anchor inevitably led to such real appreciation when combined with capital inflows. This policy process backfired when further deterioration of the fiscal stance combined with government's attempt to control interest rates on its domestic bonds-a fundamental policy error when capital account is open-led to a correction of the real exchange rate in the 1994 financial crisis, which is clearly visible in Chart 3. Chart 3 Turkey: Real Exchange Rate (1987=100) 140 130 . . . . \ 120 110 - 90 80 l l 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 - - RER: (1US$+1.5DM)ITL - RER: Trade Weighted (IMF) ..--- RER: US$/TL The response to the crisis was a stabilization program announced on April 5, 1994, which was supported by a three year stand-by arrangement with the IMF. Demand for 7 Capital flows are believed to cause exchange rate appreciation insofar as the increased domestic absorption associated with the capital flow from abroad puts pressure on the non traded goods sector, and increase its relative price.  - 21 - Turkish Lira denominated assets recovered due to very high real interest rates in the second quarter of 1994. The program, however, had only short term success in meeting the fiscal adjustment targets. Exchange rate targets were announced from mid 1994 to mid 1995, and foreign borrowing; resumed in early 1995. With political tensions and early elections taking over the agenda in late 1995, the program was not implemented and the last tranche of the Stand-by was never disbursed. That marked the return of the real exchange rule back on to the policy agenda. Central BaLnk credit to the Treasury was gradually phased out in the post crisis period. The Central Bank,, having achieved greater independence, has chosen the relative stability of the real exchange rate as its objective. In the absence of any lasting structural and fiscal adjustment in the aftermath of the crisis, this policy has served to reduce uncertainty in the foreign exchange market and maintain competitiveness of the export sector. In our view, another important objective was to mitigate the build up of excessive short term foreign debt associated with arbitrage opportunities that arise due to real appreciation in an uncertain environment. It is interesting to note the differences in the three real exchange rate series in the chart above. The trade weighted series show a much more appreciated real exchange rate in 1993, before the crisis, consistent with the observation that currency crises are often preceded by steep exchange rate appreciation8. The same series also indicates a slightly larger upward change (appreciation) in the real exchange rate index since 1994. The post 1995 stability of the US$-TL real exchange rate index compared to the other two indices is also noteworthy. The Central Bank., mainly intervening in the market for US dollars, seems to have been more successful in keeping this index constant. With a strengthening US dollar, the TL has appreciated against the currencies of Turkey's main trade partners. The appreciation in the latter half of 1997 was partly dLue to a conscious strategy by the Central Bank to minimize potential speculation against the TL during the East Asian crisis. Preliminary data suggest that the appreciation of the currency in real terms in the first half of the 1998 was around 8-10%. However, it appears that the rate of nominal depreciation in the last quarter of 1998 picked up and it is possible that real appreciation for the whole year could be below 8 percent. It must also be noted that the Central Bank's intervention in the foreign exchange market has been instrumental in avoiding a larger real appreciation of the currency. As our analysis of the decomposition of capital flows in section II.2 showed, the C'entral Bank's accumulation of reserves over the 1995-1997 period was quite large, which was consistent with its constant real exchange rate policy. While not much is known about the long run equilibrium exchange rate for Turkey at present, the overall post 1995 appreciation is indicative of the difficulties in attaining a constant real exchange rate by the Central Bank when capital flows in and there are serious fiscal imbalances at the same time. 8 See Kaminsky and Reinhart (1996)  - 22 - Apart from the direct effects of capital inflows on exchange rates, there are other possible indirect effects that need to be pointed out. As a result of exchange rate and fiscal policies becoming increasingly inconsistent with each other, the currency steadily appreciated prior to the 1994 crisis (Chart 3) and this real appreciation was associated with a stronger import boom and a relatively weak effect on exports. That an expected future devaluation (due for example to inconsistent government spending versus the managed exchange rate regime) can lead to a consumption boom is a well established result in the theoretical literature, and implies a temporary surge in imported goods demand, which results in increased home good demand and a real appreciation.9 Elements of this "temporariness" hypothesis seem to be present in the pre-1994 crisis experience. Another mechanism seems to have the wealth and income affects that has been due to the domestic and asset yield differentials, which we found in section III.2 to be the key factor pulling capital inflows to Turkey. To the extent that such effects led to a higher demand for non-traded goods, the real exchange rate would tend to appreciate. V. 2 Financial Integration and Interest Rates The liberalization of the capital account established a strong link between domestic and external markets and this was expected to lead to a convergence of local and foreign interest rates. However, over the last decade and particularly after the 1994 crisis, domestic ex-post real rates have diverged significantly from foreign rates which were the main cause of capital inflows as shown in section 111.2. Chart 4 compares three month T-bills real rates in TL (i.e., after adjusting for inflation), three month T-bill return in US$ (i.e., after adjusting for lira depreciation), and the LIBOR. It is clear that Turkish rates, ex-post, have been high by international standards since 1989. It is also visible that there has been a large volatility and in a few occasions returns have been negative in real terms. While high real rates have been a source of concern, the discussion in Turkey has rarely been in an open economy context and policymakers often ignored or did not fully grasp the implications of open capital account for interest rate determination - lack of such an understanding was the reason behind the policymakers attempt to manipulate the auction rates of T-bills in 1993/1994 which triggered the currency crisis. 9 See Calvo and Vegh (1993).  - 23 - Chart 4 Turkey: Real Interest rates Ex-post annual return on 3-month T-BilIs 150 1994 MS : 294 1994 M6 :202 -50 -100 3-month T-Bill Return in US$ --- 3-month T-Bill Return in TL---LBO Given the oDpenness of financial markets in Turkey, the starting point is the uncovered interest parity condition. It states that with no barriers to capital mobility, risk neutrality, and no country risk, returns for simtilar types and maturity of assets would be equalized through arbitrage, and deviations from parity will be unpredictable and white noise (Frankel & Okongwu 1995, Edwards, 1998). The fact that there maybe restrictions or taxes on capital flows, and there ar-e foreign exchange and country risks can put a wedge between domestic and foreign rates, which will leave investors indifferent between holding foreign and domestic instruments. This portfolio equilibrium differential can be expressed in real terms as: R(TR) = R(US) + R + TX