The internal transfer problem : Turkey

THE WORLD BANK Internal Discussion Paper EUROPE, MIDDLE EAST AND NORTH AFRICA REGION Report No. IDP-0083 Turkey: The Internal Transfer Problem by Alfredo Thorne February 1991 (Revised Version) Office of the Vice President Furope, Middle East and North Africa Region Discussion Papers are not formal publications of the World Bank. They present preliminary and unpolished results of country analysis or research that is circulated to encourage discussion and comment; citation and the use of such a paper should take account of its provisional character. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. TURKEY: THE INTERNAL TRANSFER PROBLEM By: Alfredo Thorne January, 1991 The author is a Financial Economist in the Trade and Finance Division in the Technical Department of Europe, Middle East and North Africa region. ABSTRACT This paper describes how Turkey is coping with the so-called "internal transfer problem". A successful positive net external debt transfer in countries, such as Turkey, where most of the external debt is public, presupposes a successful internal transfer. It is not sufficient for a country to generate the required foreign exchange, but it requires that both the public and private sectors generate the required net savings (savings less investment) and that the financial sector is capable of performing an efficient mobilization of savings. The main conclusions of this paper are that the fiscal adjustment in recent years (the data and analysis in this paper extend only until end-1988) has not been generating the sufficient domestic resources to service the foreign debt without drawing on resources from the private sector: the size of the JSal public sector deficit (measured through changes in liabilities) has shown very little change since 1986. The public sector, in order to meet its external commitments, while at the same time postponing its fiscal adjustment, has increased its reliance on private sector savings. Also, it is concluded that the public sector's increased demand for private savings coupled with an under-developed financial sector, has contrained the efficient mobilization of domestic resources. This, in turn, has placed an excessive stress on the financial sector. Evidence of this was the high real domestic lending rates resulting from the public sector's reliance on domestic debt; and the increased inflation levels which were partly explained by the public sector's use of money finance. This also resulted from the presence of non-performing loans in banks' portfolios. The paper also offers two annexes where a more technical discussion on the topics of measurement of public sector deficits and on the demand for assets in Turkey is undertaken. TURKEY: THE INTERNAL TRANSFER PROBLEM TABLE OF CONTENTS Page No. Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v CHAPTER I: PUBLIC DEBT AND FISCAL DEFICITS A. Background . . . . . . . . . . . . . . . . . . . . . . . . . 1 B. Structure of Domestic Debt . . . . . . e b . . . . . . . 2 Structure of the Net Domestic Public Debt . . . . . . . . . 2 C. Public Sector Deficits and Financing . . . . . . . . . . . . 8 Below-the-Line Public Sector Deficits . . . . . . . . . . 8 Public Sector Real Deficits and their Sources of Finance . . . . . . . . . . . . . . . . . . . . . . . . 12 D. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 20 CHAPTER II; THE INTERNAL TRANSFER PROBLEM A. Background . . . . . . . . . . . . . . . . . . . . . . . . . 22 B. Resource Transfer between the Public and Private Sectors . . 22 The Transfer of Real Resources . . . . . . . . . . . . . . 23 Mechanism Used to Transfer the Resources . . . . . . . . . 27 C. Implications for the Domestic Financial Sector . . . . . . . 30 Pressure of Public Sector Deficits on the Domestic Financial Market . . . . . . . . . . . . . . . . . . . . 30 The Effect of Public Sector Deficits on Real Interest Rates . . . . . . . . . . . . . . . . . . . . . . . . . . 38 The Effect of Public Sector Deficits on Bank Behavior . . . 45 D. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 50 ANNEX 1: A TECHNICAL NOTE ON THE ESTIMATION OF PUBLIC SECTOR DEFICITS 1. The Algebra of Fiscal Deficits . . . . . . . . . . . . . . . 52 Nominal Fiscal Deficts . . . . . . . . . . . . . . . . . . 53 Real Fiscal Deficits . . . . . . . . . . . . . . . . . . . 54 Quasi-Fiscal Deficit . . . . . . . . . . . . . f i c i . . 55 Pressure of the Fiscal Deficit on the Domestic Financial Market . . . . . . . . . . . . . . . . . . . . . . . . . 59 2. Methodology and Sources of Data . . . . . . . . . . . . . . . 60 Foreign Debt Finance . . . . . . . . . . . . . . . . . . . 60 Domestic Debt Finance . . . . . . . . . . . . . . . . . . . 65 Money Finance . . . . . . . . . . . . . . . . . . . . . . . 67 Quasi-Fiscal Deficit . . . . . . . . . . . . . . . . . . . 70 - ii - ANNEX 2: DEMAND FOR ASSETS IN TURKEY: A PRELIMINARY WORK 1. The Model . . . . . . . . . . . . . . . . . . . . . . . . . . 79 2. Emperical Results . . . . . . . . . . . . . . . . . . . . . . 82 3. Inflation Tax . . . . . . . . . . . . . . . . . . . . . . . . 86 4. References . . . . . . . . . . . . . . . . . . . . . . . . . 88 LIST OF TABLES Table 1.1 Domestic Public Debt Outstanding as Reported by Treasury 3 Table 1.2 Structure of Public Sector Debt . . . . . . . . . . . . . 5 Table 1.3 Below-the-Line PSBR Estimates . . . . . . . . . ... . . . 10 Table 1.4 Above-the-Line Consolidated PSBR Estimates . . . . . . . . 11 Table 1.5 Real Consolidated Public Sector Deficit . . . . . . . . . 14 Table 1.6 Breakdown of Adjusted Money Finance . . . . . . . . . . . 17 Table 2.1 Real Net Savings . . . . . . . . . . . . . . . . . . . . . 24 Table 2.2 Pressure of the Consolidated Public Sector Deficit on the Domestic Financial Market . . . . . . . . . . . . . 31 Table 2.3 Consolidated Banking Sector Balance Sheet . . . . . . . . 35 Table 2.4 Alternative Short-Term Investment Opportunities for Large Depositors . . . . . . . . . . . . . . . . . . . . . 50 Table A.1.1 Balance Sheet and Profit and Loss Account of a Central Bank . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Table A.1.2 Foreign Exchange Liabilities of the Public Sector . . . . 61 Table A.1.3 Nominal Net Domestic Debt of the Public Sector and Adjusted Base Money . . . . . . . . . . . . . . . . . . . 66 Table A.1.4 Profit and Loss Account of the Central bank . . . . . . . 72 Table A.1.5 Structure of Public Sector Debt . . . . . . . . . . . . . 73 Table A.1.6 Real Consolidated Public Sector Deficit . . . . . . . . . 74 Table A.1.7 Breakdown of Nominal Fiscal Deficit Above-the-Line Government's Estimates . . . . . . . . . . . . . . . . . . 77 Table A.1.8 Breakdown of Nominal Fiscal Deficit Estimates through Changes in Net Total Public Sector Liabilities . . . . . . 78 - iii - Table A.2.1 Long-Run Coefficients of Regression Results . . . . . . . 84 Table A.2.2 Interest Rates Offered by Banks on 1 Year Foreign Exchange Time Deposits . . . . . . . . . . . . . . . . . . 89 LIST OF FIGURES Figure 1.1 Composition of Stock of External Debt . . . . . . . . . . 6 Figure 1.2 Share of Foreign and Domestic Finance . . . . . . . . . . 15 Figure 2.1 Net Real Savings Composition . . . . . . . . . . . . . . . 26 Figure 2.2 Ratio of Principal and Interest to New Borrowings . . . . 34 Figure 2.3 6 Month Deposit and Bill Real Interest Rates in TL and in US$ . . . . . . . . . . . . . . . . . . . . . . 39 Figure 2.4 Lending Real Interest Rates in TL and in US$ . . . . . . . 39 Figure 2.5 Interbank Nominal Foreign Exchange Rates . . . . . . . . . 43 Figure 2.6 Overnight Nominal Interest Rates . . . . . . . . . . . . . 43 Figure 2.7 Deposit, Bonds and Bill Real Interest Rates in TL . . . . 44 Figure 2.8 Lending and Deposit Real Interest Rates in TL . . . . . . 44 Figure 2.9 Bank's Average Real Spread . . . . . . . . . . . . . . . . 46 Figure 2.10 Proportion of Banks' Gross TL Spread Taxed . . . . . . . . 46 Figure 2.11 Proportion of Bank's Gross TL Spread Taxed (Assuming Different Non-Performing Assets/Performing Assets) . . . . 47 Figure A.1.1 RER and International Price Indexes . . . . . . . . . . . 65 Figure A.2.1 Laffer Curve on Inflation Tax . . . . . . . . . . . . . . 87 -iv - LIST OF EQUATIONS Equation (1.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Equation (2.1) . . . . . . . . * * . . . . * . . . . . . . . . . . . 23 Equation (A.1.1) . . . . . * . * * * * * . * . . . . . . . . . . . . . 53 Equation (A.1.2 . . . . * . . * * * . * * * * * . . . . . . . . . . . 53 Equation (A.1.3) . . . . . * * * * * * * * * * * . . . . . . . . . . . 54 Equation (A.1.4) . . . . * . * . * * . * * * . * . . . . . . . . . . 54 Equation (A.1.5) . . . . * * . . . * * * * * . * . . 5 . . . . . . . . 56 Equation (A.1.6) . . . . . . . . * * * . * . . . . 7 . . . . . . . . . 57 Equation (A.1.7) . . . . . . . * * * * * * * . * . . . . . . . . . . . 57 Equation (A.1.8) . . * * . * . * * * * * . * . . . . . . . . . . . . . 57 Equation (A.1.9) . . . . * . * * . * * . . . . . * 8 . . . . . . . . . 58 Equation (A.1.10) . . . . * * * * . . * . . * . . . . . . . . . . . . . 59 Equation (A.1.11) . . . . * * * * . * . . * 6 . . . . . . . . . . . . . 62 Equation (A.1.11a) . . * * * * * * * * . * . . * . . . . . . . . . . . 62 Equation (A.1.12) , . . * . * * * . * * * . . * . 8 . . . . . . . . . 68 Equat4on (A.1.13) . . . * * * * * * . * * * . . . 6 . . . . . . . . . 68 Equation (A.1.14) . * . * * * * * . * . * . . 9 . . . . . . . . . . . 69 Equation (A.1.15) . . . * * * . * . * . * * . . . . . . . . . . . . . 69 Equation (A.1.16) . . * . . . . . . . . 7 * . . . . . . . . . . . . . 70 Equation (A.1.16a) . . . * * * * * . * * * * . . . . . . . . . . . . . 70 Equation (A.2.1) . . . * * * * * * * . * . . * . . . . . . . . . . . 79 Equation (A.2.2) . . . * * * * * * * * * . . * . . . . . . . . . . . 81 Equation (A.2.3) . . . * * * * * * . * * 8 . . . . . . . . . . . . . 82 Equation (A.2.4) . . . . . . . . . . . . . . . .4 Equation (A.2.5) . . . * . * . * * * . . * . . . . . . . . . . . . . 84 WP50\OC\ATTOCTUR TURKEY: THE INTERNAL TRANSFER PROBLEM PREFACE Turkey, during the early 1980s, undertook an adjustment program which enabled it to regain full creditworthiness in international capital markets and thereby to access the capital markets for voluntary lending. This foreign debt strategy contrasts with that adopted by other developing countries, notably the heavyly-indebted Latin American countries which have resorted to debt resche- dulings and recently have also benefitted frcm some debt forgiveness under the Brady scheme. However, the cost for Turkey of adopting this debt strategy has been having to make a positive net foreign transfer. First, since the mid-1980s, Turkey started transferring net resources abroad by attaining a surplus in the Non-Interest Current Account (NICA); and second, since the late 1980s, Turkey had to repay increasing amounts of principal resulting from the 1978-82 rescheduled debt. In this way Turkey enhanced further its international creditworthiness. During the early 1980s Turkey regained creditworthiness in international markets (i.e., it reduced its key debt ratios) by achieving high growth rates of exports and of GDP. This paper focuses on the implications of the net transfer abroad on the domestic economy. The first section reviews the problem of the public sector debt, the fiscal deficits ar.d their link to the external debt transfer strategy; and the second section focuses on the effects that the external debt transfer and the lack of sufficient fiscal adjustment is having on private sector investment. - vi Two annexes provide a more technical discussion of these issues, as well as a discussion of the methodology used for calculating the estimates of real fiscal deficits and of domestic debt. This document asserts that the strategy of producing a net transfer abroad without first making a strong fiscal adjustment increased the burden on the private sector. Also, this undue pressure on private sector savings was transmitted to the financial sector. In brief, this strategy was based on transferring domestic resources -- i.e., savings -- from the private to the public sector so that the latter could perform the required net foreign transfer. The document recommends as necessary preconditions for easing the costs of this strategy a strong fiscal adjustment and the adoption of measures leading to more efficient mobilization of private sector savings. I would like to acknowledge especially the comments and useful suggestions of my colleagues: Jeffrey Balkind, Roberto Rocha and Manuel Hinds. This document served as an input for the World Bank report entitled: Turkey: A Strategy for Managing Debt .Borrowings and Transfers under Macroeconomic Adjustment. Washington D.C.: World Bank, published country study, May, 1990. CHAPTER 1: PUBLIC DEBT AND FISCAL DEFICITS A. BACKGROUND 1.01 Turkey, during the 1980s, regained its creditworthiness in the inter- national markets by pursuing an economic reform program which led to a reduction of the ratios of total foreign debt to exports, total service of foreign debt to exports, and foreign debt to GNP. This was achieved by a rapid increase in exports and high GNP growth.1 Yet, Turkey could not reduce the total stoL' of foreign debt, because of substantial exchange rate losses incurred (see Figure 1.1); in fact, the foreign debt stock increased. Also, to this added the deficits in the current account balance during 1980-87. However, since 1988 the authorities have moved to confront the other aspect of the debt problem: how to produce a net positive transfer abroad.2 In brief, this implies that authorities need to free domestic resources, convert them into foreign currency, and transfer them abroad. This is the so-called transfer problem. 1.02 Successful production of such a transfer will depend chiefly on the capacity of the public sector to raise these resources (to the extent that most of the foreign debt is public). This chapter opens the discussion of the internal transfer by examining public sector readiness to meet the transfer problem. We thus discuss the structure of public debt, the size of fiscal deficits, and their source of finance. This discussion will be continued in the following chapter, where we analyse the implications of the size of the deficit and its financing on the domestic financial sector. 1/ For a detailed discussion of the foreign debt strategy see: The World Bank. Turkey: A Strategy for Managing Debt. Borrowings and Transfers under Macroeconomic Adjustment. Washington, D.C.: The World Bank, published country study, May 1990; and The World Bank. Turkey: External Dbt,. Fiscal Poliey and Sustainable Growth Report No. 7162-TU. Washington, D.C.: The World Bank, September, 1988. 2/ This is define as the surplus in the 1gal current account balance; and the latter is the current account balance corrected for international inflation (see Table 2.1). -2- 1.03 This chapter is organized as follows. In section B the structure of domestic debt is analyzed, and stock of domestic debt estimated. Section C discusses the size of the public sector deficit and its sources of finance. Annex 1 explains formally some of the definitions, methodology and sources of data used in the calculation of the estimates used in the discussion in the text. B. STRUCTURE OF DOMESTIC DEBT 1.04 The structure of total public debt and the relation between domestic and foreign public debt can be observed via the trend in net (i.e., liabilities less assets) domestic public and foreign debt, Structure of the Net Domestic Public Debt 1.05 The first step is to determine the actual stock of net domestic public debt. Table 1.1 shows the Treasury's estimates of its domestic public debt; two observations are worth noting. First, the table is not a comprehensive measure of consolidated public sector debt, since it does not include a portion of the Treasury's stock of debt which is held by the Central Bank, and a portion of the debt of the State Economic Enterprises (SEEs), Extra-Budgetary Funds (EBFs), and the amount of net debt of the Central Government that is outstanding with the domestic financial sector. Nor does it include public sector assets. Second, Treasury classifies the ite.n 'consolidation' as domestic debt which in reality is part of foreign debt. About 90% of this item is made up by loans granted to the Treasury by the Central Bank financed with Central Bank foreign liabilities, and exchange rate losses accumulated on these liabilities. By Central Bank law, the exchange rate losses of Central Bank's foreign liabilities are a claim on the Treasury; yet, the Treasury has not always paid them to the Central Bank. Moreover, the Central Bank has not charged interest on this debt. However, consolidations of public sector debt (which most governments do and which Turkey 3- also does, recognizing that this process is difficult due to data constraints) have been attempted and therefore various items cancel each other out in the process of measuring domestic public debt. TABLE 1.1 DOMESTIC PUBLIC DEBT OUTSTANDING AS REPORTED BY TREASURY (as % of GNP)(1) 1980-83 1984 1985 1986 1987 1988 Average NET DEBT STOCK 16.1 21.1 21.3 23.4 23.1 21.1 (at end-period) Bonds 2.4 2.4 3.2 3.4 3.2 3.8 Bills 1.0 1.3 1.5 1.8 2.6 2.0 Consolidation 9.7 14.9 14.2 15.9 15.4 13.6 Central Bank Adv. 3.1 2.4 2.4 2.3 1.9 1.6 SOURCE: Treasury (1) Since stocks are measured at end-of-year prices while GNP is measured at mid-year prices, nominal stocks of debt were converted to mid-year 1980 real prices before computing its ratio to real GNP at 1980 prices. For this reason these ratios might differ from the official ones. See Annex 1 for the estima- tion method. 1.06 Table 1.2 reports the estimates of the stock of net domestic public debt corrected for these two aspects,3 along with two estimates of the stock of net foreign public debt are also reported. Three key observations emerge: First, the net domestic public debt (7.6% of GNP in 1988) is small relative to the net foreign public debt both including capital losses (47% of GNP in 1988) and excluding the capital losses/gains (28% of GNP in 1988). However, since 1984 net domestic debt has tended to increase. Yet, before analyzing the effects of this size of domestic debt in the economy (Chapter 2) it is necessary to compare the increase in the stock of net domestic public debt with the supply of domestic funds (i.e., the flows of private savings and the stock of net domestic public 3/ These estimates were obtained using the same methodology as in World Bank. Op. Cit., September, 1988. debt with the stock of financial wealth, such as M2X). Furthermore, there are high costs to this debt in both nominal and real terms. For instance, in 1986 the nominal interest expense of the net domestic public debt was 2.7% of GNP, while that of the net foreign public debt was 3.5% of GNP; in real t-erms these ratios were 1.4% and 1.8% respectively,4 despite the fact that the size of net domestic public debt stock was less than one quarter that of the net foreign public debt stock (including capital losses). 1.07 Second, the most important source of domestic public finance has been Central Bank -- monetary -- finance followed by bonds and bills. Bank credit fluctuated considerably and increased sharply only in 1988, chiefly due to the reliance of SEEs on this source. Notably, monetary finance has been consistently important while bonds and bills, as a source of finance for the public sector, have become increasingly important over time. 1.08 Third, the most important borrowers of funds were the Central Bank, the Central Government and the SEEs. However, this statement needs to be qualified because it excludes transfers within the public sector; data availability does not permit the necessary corrections. For example, a large proportion of the monetary component of the Central Bank's finance was transferred to the Central Government as Central Bank advances, while similar transactions occurred between the Central Government and the SEEs. Also, as will be discussed later, the Central Bank has to be finance the servicing of the Treasury's foreign debt on the Central Bank balance-sheet (for example, by expanding the money supply to purchase the foreign exchange resources). That servicing usually includes payment of the exchange rate losses accumulated on this foreign debt (i.e., the consolidation item in Table 1.1), which, by Central Bank's Charter, are a Treasury liability, yet is rarely repaid by the Treasury, 4/ See: The World Bank. QP Cit., September, 1988, Table 3.16. 1986 is the only year for which there is an estimate of total interest expenses on net domestic public debt (i.e., including SEEs). -5- TABLE 1.2 STRUCTURE OF PUBLIC SECTOR DEBT (as % of GNP)(1) 1980-83 1984 1985 1986 1987 1988 Average I. TOTAL DOMESTIC DEBT PLUS 9.98 10.81 12.06 11.35 12.07 13.53 CENTRAL BANK FINANCE(2) (A + B) A. DOMESTIC DEBT 4.85 2.60 4.42 4.26 6.38 7.55 (=1+2+3-4+5+6) SOURCES: 1. Bonds and Bills 3.33 3.78 4.65 5.19 5.81 5.80 2. Net Bank Credit 2.84 0.50 1.07 1.41 1.95 2.33 3. EBFs -1.32 -1.68 -1.30 -2.34 -1.38 -0.58 BORROWERS: 4. Central Government 2.63 2.97 4.23 4.18 5.45 5.57 5. SEEs 3.54 1.32 1.48 2.43 2.31 2.55 6. EBFs -1.32 -1.68 -1.30 -2.34 -1.38 -0.58 B. CENTRAL BANK 5.13 8.21 7.63 7.09 5.69 5.98 FINANCE(2) II.TOTAL NET FOREIGN DEBT 19.67 23.56 24.51 26.91 27.46 27.71 (Excl. K losses/gains)(3) IIa.TOTAL NET FOREIGN DEBT 24,06 33.32 36.55 45.67 46.40 46.76 (Incl. K losses/gains) MEMO: ACCUMLATED FOREIGN 4.39 9.76 12.04 18.76 18.94 19.05 EXCHANGE LOSSES(4) (IIa-II) SOURCES: Central Bank, Quarterly Bulletin; and Treasury. See Annex 1 for fur- ther discussion on the derivation of these estimates. (1) Since stocks are measured at end-of-year prices while GNP is measured at mid-year prices, nominal stocks of both domestic and foreign debt were con- verted to mid-year 1980 real prices before computing its ratio to real GNP at 1980 prices. (2) This is the finance appropiated through monetary expansion adjusted for the subsidized credit extended to the private sector, i.e., rediscounts. (3) This estimate excludes the capital loses (gains) caused by a real effective devaluation (appreciation) in the domestic currency and by changes in international exchange rates. For the estimates for the period 1980-86 the actual currency composition of the foreign debt was used and, for the remaining period the currency composition of end-1986 was used. (4) These are the foreign exchange losses accumulated as a result of fluctuations of international and domestic exchange rates. . 6 - 1.09 Finally, another very important feature of Turkey's debt structure to be seen in Table 1.2 is the large size of the net stock of foreign public debt, including the capital losses relative to the stock of foreign public debt excluding them capital losses (rows IIa and II in Table 1.2). Although capital losses (see "Memo" item in Table 1.2) are not resources borrowed, they are a liability which will have to be serviced. In general, the greater the stock of net foreign public debt that includes the capital losses, the greater would have to be the transfer of resources abroad, and thus the greater the pressure of the foreign debt service on domestic resources. Although the extent of this pressure will depend on the extent to which the country has access to more foreign resources, this indicator also has a domestic dimension: it represents future claim on the country's wealth, a claim which needs to be paid -- and generated -- in foreign currency. 1.10 The actual size of the total stock of foreign public debt including capital losses (row IIa in Table 1.2) can be decomposed into three factors: the stock of foreign public net debt excluding capital losses which is the debt actually contracted; the effect of changes in changes in international exchange rates and in currency composition of the stock of debt and the effect of changes in the real effective domestic exchange rate, i.e., Turkish Lira to the U.S. dollar. The net sum of the last two effects made up the valuation effect or capital losses. In Figure 1.1 the foreign debt (expressed as percentage of the total debt including capital losses) have been decomposed into these three effects; this figure, therefore, enables us to appreciate the factors that had had a major importance in the growth of the foreign debt. For instance, for the period 1979 to 1984 this figure indicates that the public foreign net debt increased, chiefly, as a result of the real effective devaluation of the Tl;' A real effective devaluation/appreciation of the Tl to the U.S. dollar can be defined as a real devaluation/appreciation of the T1 relative to the basket of currencies with which Turkey trades. A closer observation to the real effective exchange rate index (RER) -- see Figure A.1.1 in Annex 1.1 -- would suggest that most of the real devaluation took place between 1980 and 1986, and after that period the RER index shows an appreciation. -7. FIGURE 1.1 COMPOSITION OF STOCK OF EXT DEBT As % of Stock in TI 150 140 1 130 120-i 110 100 90 80 70 -1 60 - so - 40 - -30 J 240- 10 -10 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 M R Stock + Inter Ps 0 RDevTI and the change in international exchange rates and in currency composition of the debt -- less dollar-denominated -- had the favorable effect of reducing its current value. From 1984 to 1987 the total foreign debt increased as result of the fluctuation in international currencies (depreciation of the U.S. dollar relative to the other international currencies) and in currency-composition. i.e., a reduction in dollar-denominated debt. Figure 1.1 is also clear in showing that the capital losses generated between 1984 and 1988 as a result of the depreciation of the dollar relative to the rest of international currencies is comparable to the capital gains generated between 1980 and 1984 as a result of 8 - the appreciation of the dollar.6 In fact, this figure suggests that most of the capital losses recorded in 1988 are the result of a real effective devaluation of the Tl. C. PU1BLIC SFCTOR DEFICITS AND FINANCING 1.11 This section is devoted to analyzing the sources of financing of the public sector deficit. For this purpose the concept of below-the-line fiscal deficit or fiscal deficit estimated through changes in net public sector lia- bilities is introduced. These estimates will then be used in the following chapter to analyse the impact of the fiscal deficit on the domestic financial sector. First, the below-the-line nominal fiscal deficit is reported. Second, after introducing the concept of rgal public deficit (the nominal deficit corrected for inflation), we discuss the sources of finance. Below-the-Line Public Sector Deficits 1.12 An alternative method of estimating the nominal Public Sector deficit is by estimating it through changes in public sector net nominal liabilities (liabilities less assets). This is the so-called below-the-line method. Table 1.37 reports these estimates for the period 1984-88. To arrive at the consolidated PSBR we have made two adjustments to the below-the-line non-financial PSBR. First, we have added the so-called "quasi-fiscal" deficit, since in the case of Turkey the Central Bank is involved in activities that in most other countries would correspond with public sector activity (e.g., rediscount lines). Second, as defined in the IMF's definition of the PSBR, we have included the net lending to the private sector, which is defined as the loans granted to the private 6/ See also The World Bank. Op. Cit., September, 1988 for further elaboration on this print. 7/ All estimates in this table were calculated in nominal terms. See Annex 1 for a discussion of the methodology. -9- sector.8 These calculations (Non-Financial PSBR and Central Bank's quasi-fiscal deficit), in line with Government's methodology, exclude the valuation effects (actual and/or accrued) arising from capital losses on the net foreign public sector liabilities. These "capital losses" originate from the fluctuations in the real exchange rate of the Tl and international currencies. Currency fluc- tuations only affect the deficit estimate in Table 1.3 (and the Government's, see Table 1.4) to the extent that it affects the T1 equivalent of interest payments on the net foreign public debt. However, as we will see in the following section, when the foreign debt is amortized, as has been happening in Turkey since the early 1980s, the realized portion of these capital losses should be considered as part of the financeable deficit, namely the PSBR. However, due to lack of sufficient data, we have not calculated the realized capital losses. 1.13 A similar set of estimates but calculated using the above-the-line method are reported in Table 1.4. In this case also the total above-the-line LM is defined as the sum of the officially calculated non-financial PSBR, plus the above-the-line Central Bank quasifiscal deficit -- estimated using Central Bank's Profit and Loss statements -- and plus the same net lending used in Table 1.3. 8/ There is a problem with the estimate of net lending in Table 1.3 and in Table 1.4 which in the future can be corrected; it includes all rediscounts granted by the Central Banks to the private sector. A more accurate measure should include only those rediscounts arising from the quasi-fiscal activities, that is, it should exclude the rediscounts granted by the Central Bank to banks to meet short-term liquidity needs. However, to the extent that net lending is added to both the Governement's and the Bank's estimates to obtain the respective PSBR estimates, our conclusions would not be affected. - 10 - TABLE 1.3 BELOW-THE-LINE CONSOLIDATED PSBR ESTIMATES (As percentage of GNP) 1984 1985 1986 1987 1988 TOTAL BELOW-THE-LINE CONSOLI- 10.6 9.4 8.9 11.2 8.7 DATED PSBR (-1+2+3)(1,2) 1 Below-the-Line Non- 2.7 5.0 6.2 7.6 5.8 Financial PSBR (2,3) 2 Below-the-Line Central 9.9 3.5 1.6 2.0 2.1 Bank's Quasi-fiscal Deficit(2,4) 3 Total Net Lending(5) -2.0 0.9 1.1 1.6 0.8 Source: See Annex 1.1. (1) PSBR was estimated as suggested in the IMF's methodology, by adding the nominal deficit plus net lending to the private sector. (2) These estimates exclude the capital losses (gains) caused by a real effective devaluation (appreciation) in the domestic currency and by changes in international exchane rates. For the estimates for the period 1980-86 the actual currency composition of the foreign debt was used, and, for the remaining period, the currency composition of end-1986 was used. (3) This is the non-financial nominal public sector PSBR estimated through changes in nominal liabilities of the consolidated public sector. See Annex 1.1 for the methodology used in arriving at these estimates. (4) This is a below-the-line deficit estimated using the Central Bank's balance sheet. See Table A.1.8 and Annex 1.1 for more discussion. (5) This is defined as the sum of MHF's loans to the private sector plus Central Bank's rediscounts granted to the private sector. For the period 1983-84, it was assumed that MHF's loans to the private sector were zero. 1.14 Although '.n theory both methods for estimating the nominal PSBR (i.e., below- and above-the-line) should yield similar estimates, this is not completely apparent from Tables 1.3 and 1.4.9 This is in spite of the fact that we had applied a compatible methodology in estimating both the below- and the above- the-line PSBR estimates. The differences are partly explained by data consistency problems, and partly as a consequence of the different treatment of capital losses in the Central Bank's profit and loss and balance-sheet accounts. 9/ In future, Bank work would, very likely, further refine these estimates and, thus provide a clearer explanation for the differences that remain between the above- and below-the-line estimates. - 11 - TABLE 1.4 ABOVE-THE-LINE CONSOLIDATED PSBR ESTIMATES (As percentage of GNP) 1984 1985 1986 1987 1988 TOTAL ABOVE-THE-LINE CONSOLI- 5.5 6.7 6.4 10.0 7.2 DATED PSBR (=1+2+3)(1) 1 Official Non-Financial 6.5 4.6 4.7 7.8 6.4 Above-the-Line PSBR (2) 2 Above-the-Line Central 0.9 1.1 0.8 0.6 0.0 Bank's Quasifiscal Defi- cit(3) 3 Total Net Lending(4) -2.0 0.9 1.1 1.6 0.8 Source: See Annex 1.1. (1) PSBR was estimated as suggested in the IMF's methodology, by adding the nominal deficit plus net lending to the private sector. (2) Is the government estimate of the PSBR and excludes the quasi-fiscal deficil. of the Central Bank. (3) This is a cash-flow deficit estimated using the Central Bank's Profit and Loss statements, i.e., it is an above-the-line estimate. See Table 1.5.4 for a detailed breakdown in Annex 1.1 for the methodology. (4) This is defined as the sum of MHF's loans to the private sector plus Central Bank's rediscounts granted to the private sector. For the period 1983-84, it was assumed that MHF's loans to the private sector were zero. 1.15 Two aspects are worth underlining from Tables 1.3 and 1.4. First, except for 1984, both methods yield very close estimates for the non-financial PSBR. In 1984 there was a large consolidation of public sector debt. The differences, as explained before and more formally in Annex 1.1, have to do with the different treatment of the foreign exchange capital losses in the Central Bank's profit and loss account and in the balance sheet. Second, taking 1985 as the benchmark, both consolidated PSBRs estimates show a very a similar trend. The nominal PSBR increases up to 1987 and falls in 1988, this is in spite of the fact that the - 12 - inflation level increased sharply between 987 and 1988. Thus, both the above- and below-the-line estimates show an improvement in fiscal improvement in 1988 of about 2.0-2.5 percentage points of GNP.10 Public Sector Real Deficits and their Sources of Finance 1.16 During inflationary periods, as explained before, nominal deficits can become an unreliable indicator for assessing public sector performance. In general, a country experiencing high inflation and with a large public debt stock, as was the case of Turkey in the mid and late 1980s, will experience an increase in its nominal public deficit regardless of whether expenditures were reduced or revenues increased. That is because nominal interest rates usually adjust fully for inflation, thus increasing the size of the nominal deficit. Moreover, as we will see in the following chapter, savings estimates not corrected for inflation over-estimate private sector savings. It is therefore convenient to undertake the analysis of public sector performance, and that of savings in the following chapter, using estimates corrected for inflation. As defined in Annex 1.1, real public deficits are the sum of the so-called primary deficit (total revenues less non-interest expenditures) plus the real interest expense on net public debt, i.e., net domestic public debt and net foreign public debt. 1.17 The estimate of the real deficit calculated through changes in the ZAl consolidated public sector's net liabilities and its sources of finance are reported in Table 1.5.11 The real deficit is a useful measure of effective fiscal performance, i.e., fiscal performance corrected for inflation. It indicates the 10/ The above-the-line estimates though show a greater improvement than the below-the-line. 11/ For this calculation, we have treated inflation-tax as a financing item, as we implicitely assume that the inflation rate is determined, "ceteris paribus", by the proportion of the deficit financed by money creation. In other words, it is through the inflation mechanism that resources are transferred from the private to the public sector. To treat the inflation-tax as a tax revenue item, as it is sometimes found in the literature (i.e., an above-the-line item), would imply assuming that the inflation rate is not affected by the proportion of the deficit financed with monetary creation, i.e., that is given ex-ante. . 13 - size of the resources claims of the public sector on other sectors' real resources (i.e., private and foreign), and how these resources were transferred to the public sector. Table 1.5 suggests that after the fiscal contraction in 1986, real deficits increased in 1987 and then decline in 1988. Real deficits as in the official nominal PSBR shows a strong expansion in 1987, but the 1988 contraction is less sizeable -- but nonetheless sizeable -- to that shown in the officially nominal PSBR. Table 1.5 is clear in indicating that this was the result of the government's policy of reducing its reliance on foreign finance. 1.18 Table 1.5 also reports the foreign exchange capital losses incurred every year. As discussed earlier, these are incurred as a result of fluctuations in the international currencies in which Turkey's foreign debt is denominated and as a result of real devaluations in domestic currency. Table 1.5 indicates that Turkey incurred substantial capital losses up to 1987 and, in the period 1987-88 these were very small.12 Although, all of these capital losses are not an immediate claim and, thus excluded from the definition of public sector deficit in Table 1.5, they are a future claim. In practice, however, the realized capital losses -- the proportion of the capital losses paid when the foreign debt is serviced -- should be included in the definition of the deficit,1 but it was not possible to make this estimate with the data available. It is important, however, to underline that capital losses become a realized claim when the foreign debt comes due, which in the case of Turkey, is occurring all the time (see Chapter 2 of World Bank, Q2 Cit.). To this extent, the estimate of the real deficit in Table 1.5 is underestimated; this is particularly the case of 1987 and 1988 when Turkey made significant payments on her foreign debt. 12/ See Chapter IV of World Bank, Turkey: Debt Management and Borrowing Strategy Under Macroeconomic Adjustment. Report No. 7732-TU. Washington D.C.: World Bank, February 21, 1990, for a discussion of currency composition. 13/ The Government, for instance, in its estimate of the PSBR only considers the capital losses in the interest component, i.e., the interest expense is estimated as the product of the interest times the stock evaluated at current exchange rates. -14 - TABLE: 1.5 REAL CONSOLIDATED PUBLIC SECTOR DEFICIT (As percentage of GNP)(1) 1980-83 1984 1985 1986 1987 1988 Average TOTAL REAL PS DEFICIT 1.68 9.69 6.41 5.39 6.52 4.13 (Excl. K losses on For. Debt) A.ADJUSTED MONEY FINAN- 1.70 5.74 2.57 1.74 1.82 3.40 CE(2) 1. Inflation Tax 1.84 2.56 2.80 1.92 2.75 3.23 2. Seignorage -0.15 3.18 -0.23 -0.18 -0.93 0.17 B.NET DOMESTIC DEBT -1.42 -0.22 1.92 0.06 2.39 1.03 FINANCE 3. Bonds and Bills -0.07 0.94 1.03 0.77 0.95 -0.13 4. Net Bank Credit -0.83 -1.32 0.58 0.40 0.63 0.33 5. EBFs -0.52 0.16 0.31 -1.10 0.81 0.83 C.NET FOREIGN DEBT 1.41 4.17 1.93 3.58 2.31 -0.30 FINANCE (Excl. K losses of For. Debt)(3) D.NET FOREIGN DEBT 4.97 6.06 4.61 10.88 3.71 -0.56 FINANCE (Incl. K losses of For. Debt) MEMO: FOREIGN EXCHANGE 3.56 1.89 2.68 7.30 1.40 -0.26 LOSSES (D-C)(5) SOURCE: Central Bank, Quartefly Bulletin; and Treasury. See Annex 1 for fur- ther discussion on the derivation of these estimates. (1) Since stocks are measured at end-of-year prices while GNP is measured at mid-year prices, nominal stocks of both domestic and foreign debt were con- verted to mid-year 1980 real prices before computing its ratio to real GNP at 1980 prices. (2) This is the portion of money finance which is used to finance the deficit, i.e., after deducting the portion returned to the private sector (see Table 1.5 for a breakdown). (3) This estimate excludes the capital losses (gains) caused by a real effective devaluation (appreciation) in the domestic currency and by changes in international exchange rates. For the estimates for the period 1980-86 the actual currency composition of the foreign debt was used, and, for the remainin period, the currency composition of end-1986 was used. (5) These are the capital losses (gains) incurred each year as a result of fluctuations in international exchange rates and of real devaluation of the domestic exchange rate. A negative sign indicates a gain. - 15 - 1.19 A close observation of the sources of finance indicates that since 1984, with the exception of 1986, a large proportion of the real deficit had been financed with domestic resources. This can be more clearly appreciated in Figure 1.2, where the share ol domestic and foreign financing in relation to the total deficit has been plotted. Figure 1.2 shows that since 1986, domestic finance as a share of the total real deficit has increased sharply while the share of foreign finance has fallen. FIGURE 1.2 SHARE OF FOREIGN AND DOMESTIC FINANCE (As percentage of Total Real Deficit) 130- 120- 110- 100- 90- 80 - 70- 60 SO -50 x 40 30 - 20 - 10 - 0 - -10- -20- -30- 1981 1982 1983 1984 1985 1986 1987 1988 Years 0 Foreign + Domestic - 16 - 1.20 The most important sources of domestic finance are money finance and net bank credit. Bonds and bills were an important source of finance during the period 1985-87. In 1988, however, their use declined sharply. This relative decline will be discussed in Chapter 2. It was, however, a direct consequence of the Treasury's decision to cap real interest rates by restricting the volume of public debt instruments sold. It is important, however, to underline the different nature of net domestic credit (i.e., the sum of bank credit, bonds and bills and EBFs finance) and money finance. While availability of net domestic credit is market determined, depending on the willingness of bankers to lend to the public sector and o- the availability of private sector financial assets, money finance is, to a certain extent, enforced: the private sector is forced to transfer resources to the public sector. These resources are extracted through money finance which is the sum of seignorage (the increased real demand for base money as the economy grows) and the inflation tax (the revenue obtained from non-interest bearing asset holders during inflationary periods). The inflation tax, in particular, is an instrument used to force the private sector to transfer savings to the public sector. And, as shown in Table 1.5 (row 1), the inflation tax has been an important source of financing of the real public sector deficit: it has contributed, on average, 2% of GNP. 1.21 Table 1.6, which provides a breakdown of the sources of adjusted money finance (I.e., the portion of money finance used to finance the real fiscal deficit) suggests that the two main sources were an increase in banks' reserves and an increase in currency in circulation. There was also a one-time gain obtained when rediscounts to the private sector were reduced in 1984. This explains the sharp increase in money finance in that year. 1.22 On average, banks' reserves were a very important source of finance during the period 1980-85 and in 1988. However, these events are explained by two different reasons. On average, the increase in reserve finance during the - 17 - period 1980-85 was explained by an increase in T1 denominated deposits (in fact, the reserve requirement ratio on sight and time deposits was reduced from 35% in December 1982 to 19% in December 1985), in 1988 the increase in reserve finance was explained by a sharp increase in the reserve requirement ratio on sight and time deposits from 10% in August 1987 to 25% for sight deposits and 14% for time deposits in October 1988.14 However, since deposits denominated in T1 (and more precisely M2) had been falling since 1986, the increase in reserve finance in 1988 was a once-and-for-all effect. In the future, if deposits remain constant, an increase in reserve requirement would require an increase in the reserve requirement ratio, i.e., an increased tax on banks. TABLE: 1.6 BREAKDOWN OF ADJUSTED MONEY FINANCE (As percentage of GNP) 1980-83 1984 1985 1986 1987 1988 Average ADJUSTED MONEY FINANCE 1.70 5.74 2.57 1.74 1.82 3.40 (1+2+3+4)(1) 1. Currency 1.33 1.02 0.99 1.03 1.47 1.15 2. Reserves 2.16 2.57 1.93 0.94 1.03 2.82 3. Rediscounts -1.64 1.99 -0.26 -0.43 -0.79 0.70 4. Other -0.15 0.16 -0.09 0.20 0.11 0.13 SOURCE: Central Bank, Quarterly Bulletin; and Treasury. (1) This is the portion of the money finance which is used to finance the public deficit, i.e., after deducting the portion returned to the private sector through rediscounts. 14/ In May 12, 1989 reserve requirement on banks deposits were, once again, changed; and these changes were made effective April 14. Reserve requirements on sight deposits were changed from 25% to 20%, on 3-month deposits from 14 to 20% and for deposits up-to-1-year from 14 to 10%. Also, the Central Bank introduce deposits with maturities varying from 2 to 5 years with variable interest rates and impose reserve requirements which fluctuate between 8% and 2%. Taking into account the quasi-money composition, it is very likely that the ratio of reserve requirements to quasi-money would increase, thus increasing public finance from this source and the taxation on banks. - 18 - 1.23 The amount of finance provided by increased currency was, on average, between 1% and 1.5% of GNP. And, although, for 1988 Table 1.6 shows a drop to 1.1% of GNP from 1.5% of GNP in 1987, thus suggesting a more restrictive monetary policy, this was not the case for the first three quarters of 1988. Up to the third quarter, as shown in Table A.1.6, money finance was greater than 1.6% of GNP. In general, the greater the reliance on this source, the greater the inflation pressures. Further, since inflation discourages individuals' demand for currency, it also introduces inflationary pressures by reducing the demand for currency relative to its supply.15 In Turkey, as we conclude in Annex 2.1, demand for currency is very sensitive to inflation level. This implies that stabilizing the portion of the deficit financed with respect to currency issued may involve a higher inflation. This relationship is illustrated by the inflation-tax figure in Annex 2.1. 1.24 Before moving on, it is important to enquire into the fiscal implications of a net positive public sector transfer abroad.16 The fact that only in 1988 the public sector showed a net public sector transfer abroad (i.e., the net foreign public finance excluding capital losses was still positive, see row C in Table 1.5)17 should not inhibit us from drawing the full implications of a public sector net transfer abroad. This would be a counterfactual exercise. The data shows (see Table 1.5 and Figure 1.2) a sharp fall in dependence on foreign financing both as percentage of GNP and as a share of the total real fiscal deficit. 15i To reduce the currency supply, the authorities relied on banks reserves (i.e., required reserves and liquid reserves), thus benefiting from both sources of money finance: currency and banks' reserves. 16/ The surplus on the current account, which measures the total net positive transfer abroad, can be broken into the portions corresponding to the public and private sectors. 17/ The surplus in the current account suggests that it was the private sector that had a net transfer abroad and not the public sector. 19 - 1.25 It is useful to look at the implications of a net positive public sector transfer abroad (i.e., a reduction in net foreign public debt excluding capital losses) through the use of the public sector debt-dynamics identity shown as equation (1.1),18 (1.1) 6*= d+(r-g)b+(r*-g)b*- ri-6, where 6, b* and m are the ratios of the partial derivatives of net domestic public debt, net foreign public debt and adjusted stock of money base to GNP, b, b* and d are the ratios to GNP of net domestic public debt, net foreign public debt and primary deficit denominated in T1, g is the rate of growth of the economy and r and r* are the real interest rates on net domestic public debt and net foreign public debt. 1.26 This identity asserts that for the ratio of net foreign debt to GNP to fall, the sum of the increa'se in net domestic debt and in the increase in the adjusted money base needs to be greater than the primary deficit, assuming that average real interest rates on net domestic public debt and on net foreign public debt are equal to the rate of growth of the economy. However, if these average real interest rates are greater than the rate of growth of the economy, as seems to be the case in Turkey, then the sum of the increase in real net domestic public debt and the increase in adjusted money base will need to be financed. 1.27 Although, on average, the average real interest rates on net domestic public debt and net foreign public debt are higher than the growth rate, one might think that in 1988 the average real interest rate on net domestic public debt was smaller than the rate of growth. The reason for this was that it was artificially held down. The public sector reduced its cost of domestic debt by 18/ This identity is derived from the real fiscal deficit equation in Annex 1.1. See also Stanley Fischer, 'The Ecoitorics of the Government Budget Constraint' . Lecture delivered at the Central Bank of Pakistan. March 1989. - 20 - forcing the private sector to hold low-yielding debt instruments, such as base money in inflationary periods, and by capping the rate on bonds and bills in 1988. For instance, the real weighted interest rate of bonds and bills declined from 15% in 1986 and 7% in 1987 to -3% in 1988. However, since the real effective interest rate paid can be artificially held down only in the short term, it is more appropriate to measure it at opportunity cost (the market rate), which was unambiguously greater than the growth rate (see discussion and evidence in Chapter 2). 1.28 Equation (1.1) thus underlines Turkey's future fiscal dilemma in trying to make a net positive public sector transfer abroad: either it reduces its primary deficit and the difference between the average real interest rates and the rate of growth of the economy; or it must increase its net domestic debt and the adjusted money base. Obviously, the latter alternative would be more prob- lematic, for as we will argue in the following chapter, the large public sector deficit is already showing signs of stress in the domestic financial sector. D. CONCLUSIONS 1.29 Our conclusion is that, although Turkey has reduced the size of the fiscal deficit in 1988, the decline in foreign financing since 1986 has resulted in increased reliance on domestic debt, which illustrates the difficulties that the public sector is facing in making an external transfer. Moreover, increases in the domestic debt of the public sector have also increased public deficits, as the cost of this debt is very high. Also, excessive reliance on the domestic financial sector and on the instruments used to channel resources from the private to the public sectors -- in particular the greater reliance on money finance -- has contributed to the high inflation :.nd the high real interest rates. 1.30 The public sector need to make an external transfer would demand a greater reduction of its primary deficit (i.e., its non-interest current account) - 21 - while at the same time attempting to reduce inflation and the high real interest rates through a comprehensive economic program. Moreover, reform of the current public debt auction system is also required - - to make it more transparent and competitive -- and to rely less on money finance and more on efficient domestic debt financing. Such policies would help to ease the external transfer diffi- culties. - 22 CHAPTER II: THE INTERNAL TRANSFER PROBLEM A. BACKGROUND 2.01 As discussed in earlier parts of this report, Turkey faces the challenge of producing a net positive transfer abroad. Its success will depend chiefly on the public sector's capacity to raise these resources (as most of the foreign debt is public) but also on the capacity of the financial system to mobilize them, by encouraging private sector savings in public debt instruments. This chapter, by discussing the implications of public financial policy for the domestic financial sector, will attempt to assess the ability of the public and private domestic financial sectors to meet the transfer challenge. 2.02 This chapter is organized as follows. Section B looks at the resource transfer in a savings and investment framework (i.e., how much of the net rial private savings needs to be transferred to the public sector to finance its deficit). Section C looks at the implications of the public sector's increased reliance on the domestic financial sector for the mobilization of domestic resources, taking into account the undeveloped and fragile nature of this market. Annex 2.1 offers some preliminary estimates for a demand for assets model and the interest and income elasticity of substitutions are reported. Also, using these results, the inflation-tax schedule on currency holdings is reported. B. RESOURCE TRANSFER BETWEEN THE PUBLIC AND PRIVATE SECTORS 2.03 This section discusses first the Xgal (inflation-corrected) resource transfer between sectors, and second the mechanism by which these resources were transferred. For this purpose we use the savings-investment framework that links the financial requirements of the real fiscal deficits with the supply of resources - 23 - from the foreign and private sectors. This framework indicates the size of the transfer of resources needed for a given public sector deficit, and the sectors which finance it. The Transfer of Real Resources 2.04 From the real net savings accounting identity the real budget deficit (minus the net real public savings) is equal to the sum of real private and foreign net savings. This can be written as equation (2.1), (2.1) Real Budget Deficit -RNS" = RNSP+RNSI, where RNS denotes real net savings (i.e., savings less investment) and superscripts pu, p and f stand for public, private, and foreign sectors. 2.05 Table 2.1 reports on real net savings estimates by sector and composition; and in Figure 2.1 the real net savings of the public, private and foreign sectors are plotted together. Real net savings estimates are different from those reported by official accounts, as they have been corrected for inflation. The real deficit is as defined in the previous section (see Table 1.4), real foreign savings is the deficit on the current account less the foreign inflation component of nominal interest expenses on the net foreign debt,' and real net private savings are a residual -- as in the n&tional accounts. . We used net foreign public debt as a proxy for total net foreign debt. The inflation component of nominal interest expenses on the net foreign public debt is the product of the foreign inflation confronted by Turkey (i.e., that of her main trade partners and thus incorporates devaluations/appreciations of Turkey's main trade partners' currencies) and the previous period stock of net foreign public debt denominated in Tl. 24 - TABLE: 2.1 REAL NET SAVINGS(l) (As percentage of GNP) 1980-83 1984 1985 1986 1987 1988 Average Total Net Savings: 0.0 0.0 0.0 0.0 0.0 0.0 (-A+B+C) A. Foreign Sector 1.9 2.0 1.4 3.3 0.7 -3.0 B. Public Sector(2) -1.7 -9.7 -6.4 -5.4 -6.5 -4.1 Gross Savings 10.0 0.0 5.0 8.2 6.8 6.2 Investment 11.7 9.7 11.4 13.6 13.3 10.3 C. Private Sector -0.2 7.7 5.1 2.1 5.9 7.2 Gross Savings 8.9 17.3 14.5 13.1 17.9 20.5 Market Determined(3) 7.1 14.7 11.7 11.2 15.1 17.2 Inflation tax (4) 1.8 2.6 2.8 1.9 2.7 3.2 Investment 9.1 9.6 9.4 11.0 12.0 13.3 SOURCE: SPO, National Accounts and Tables 1.5 and 2.2. (1) Figures in this table are different from official data because of the two corrections made. First, net and gross savings figures have been adjusted for inflation. This correction affects the official foreign savings (net and gross) as it excludes the inflation component on interest expenses of the foreign debt; and the public and private savings (net and gross), as it also deducts the the inflation component on the interest expenses of the domestic debt. Second, public sector savings (net and gross) have been adjusted to include the so-called quasi-fiscal deficit of the Central Bank. Investment figures are the official ones. (2) This is the same as the total real deficit in Table 1.5. (3) These are defined as the difference between gross savings and inflation tax or forced savings. (4) This is the adjusted inflation tax (same as row A in Table 1.5), i.e., after deducting the portion of the gross inflation tax which in effect remains with the private sector. 2.06 The interpretation of these estimates is also different from that of the official net nominal savings estimates. Real net savings estimates in Table 2.1 indicates the proportion of real net private savings and real foreign savings (i.e., real income from abroad) claimed by the public sector to finance its real fiscal deficit. The reason for uadertaking this exercise in XgAl terms is that (as we said) nominal net savings over-estimate the net private sector's savings - 25 - during inflationary periods. In particular, these estimates assume that indi- viduals consider as income all of the inflationary component of interest payments, which are really compensation for the depreciation of their financial assets during inflationary periods. 2.07 The increasing importance of real private savings as a source of finance since 1985 and therefore, the declining importance of foreign savings since the same year is apparent in Table 2.1. In fact, since 1988 real net foreign savings show a negative sign, indicating that real net private savings in addition to providing resources to finance the public deficit, have also provided resources to sustain a net real transfer abroad (i.e., a private sector net transfer). Both observations underline the links between the strategy to encourage a total net positive transfer abroad (sum of public and private transfers) and the domestic financial markets. In particular, for a given xal fiscal deficit, a net positive transfer abroad (a surplus in the real current account, see Table 2.1) has to be financed by increasing private savings. Obviously, a reduction in the real fiscal deficit will ease the burden on the private sector. 2.08 Two other aspects emerge from Table 2.1. First, this Table suggests that a large portion of the real net private savings transfer to the public sector occurred through the inflation tax. For example, in the period 1985-88 (when the inflation tax was rising), on average, 40% of the real net savings transferred from the private to the public sector was accounted for by the inflation tax. Although this might indicate the public sector's preference for low-cost funds, to the extent that the real interest rate on bank credit and bonds and bills increased (see Section C) and that the inflation tax can be considered as forced savings, it can also be seen as an indication of the private sector's reluctance to finance the deficit. In other words, had the public sector decided not to use the money finance, the public deficit would have had to be cut or would have had to bear the higher cost of domestic finance. - 26 - FIGURE 2.1 NET REAL SAVINGS COMPOSITION (as percentage of GNP) 8 6 4 3 2 - / 1/ 0 -6 - -7- -8- -9 - 10 IllIs I 1980 1981 1982 1983 1984 1985 1986 1987 1988 Years 0 Net public + Net private Net foreign 2.09 Second, in comparing the foreign savings in Table 2.1 and the increase in foreign public debt in Table 1.5 (row C), it is possible to observe years, such as in 1985, 1987 and 1988, when the increase in net foreign public debt was greater than the foreign savings (i.e., increase in total debt). This suggests that the private sector was making a net positive transfer abroad while the public sector was benefiting from a net negative real transfer (i.e., the net foreign public debt was increasing). - 27 - Mechanism used to Transfer the Resources 2.10 The mechanism whereby the market determined net savings (excluding the inflation tax) transfer was achieved has important implications for the current strategy of reducing the foreign savings, cutting the public sector deficit, and increasing private sector investment. This is the link between high fiscal deficits and high private net savings. 2.11 As discussed at length in a previous report,2 a large portion of the resources transferred from the private to the public sector occurred through high real interest rates and through a reduction in the proportion of credit allocated to the private sector,3 Both had the effect of generating the net private savings necessary to finance the public sector deficit, while at the same time reducing the deficit in the current account of the balance of payments (which is the same as a reduction in foreign savings). This explains how high public sector deficits were consistent with low foreign savings. 2.12 However, the regression results on the savings and investment functions reported in that report4 also predicted that most of the increase in net savings would have come from a reduction in private investment, rather than from an increase in private savings. How then can this reduction in private investment be reconciled with the observed high export growth and moderate increase in private investment? (Private investment increased from 9.9% of GNP in 1980 to 13.3% of GNP in 1988.) This implies that the effects of high real interest rate and the declining share of private credit on private investment and savings were not the only mechanisms by which a high public deficit was made consistent with 2/ See, The World Bank. Op. Cit-, September, 1988. 3/ In the report it is argued that the share of credit to the private sector increased, thus counterbalancing the negative effect of the interest rates. However, as will be shown in the following section, a detailed analysis of the consolidated balance sheet of the banking system indicates that the share of credit to the private sector experienced a small reduction. See also Tables 1.4 and 2.1. 4 See The World Bank. Op. Cit., September 1988. . 28 - low foreign savings. In fact, to counterbalance these negative effects on private investment, Turkey relied on a complex system of subsidies to the private sector that sustained the growth of exports and of private investment.5 These subsidies aimed at increasing the profitability of these export industries and of a subset of investors (the priority sectors) by reducing the cost of capital through subsidizing the interest rate or by making credit more available. 2.13 However, at the same time, and as a result, the public sector's deficit was increasing. To encourage or force savers to hold public debt, the public sector had to resort to both non-market and market mechanisms. The most notable non-market mechanism has been the inflation tax or forced savings in low-yielding savings instruments; and the most important market mechanism has been the tax exempt interest income on Treasury bonds and bills.6 2.14 Up to now this seems to have been a very important ingredient of the overall strategy followed by Turkey, with very successful results. In terms of the net savings balances in Figure 2.1 and Table 2.1, this strategy has reduced the public sector's present and future revenues and penalized marginal investors or exporters, those who were not eligible for these subsidies. The public sector teduces its present revenues through the granting of subsidies to exporters and investors; and it reduces its future revenues by exempting bond holders of tax on interest income. The penalization occurs because those investors and exporters who were not favored have to pay the market cost, which is higher than it would be had the public sector not provided these subsidies. Moreover, newcomers receive 5/ There are also other important subsidies such as those granted on mortgage loans aimed at reducing the cost of housing for low income earners. 6/ It is not clear, though, how a tax imposition on interest income from bond. and bills would affect the demand for these instruments. It might happen -- as happened before -- that the market determined interest rate will increase, thus leaving the after-tax rate of return unchanged. - 29 - no benefits because most benefits are granted as a rebate on a tax liability. Both of these groups, the excluded and the newcomers, have to pay for the cost of this strategy. 2.15 However, this strategy of subsidizing exports and investment is running into difficulties for two main reasons: (i) because of the size of the deficit (more discussion will follow in Section C) and the adverse effect it causes, notably crowding-out of private sector investment which outweighs the benefits from the subsidy scheme; and (ii) because if the public sector continues to grant subsidies through the exemption of taxes, thus reducing its future revenue, it will also be increasing its future deficit. This can be avoided if the new income generated by exporters or investors outweighs the loss in revenue from tax exemptions. However, the higher these tax exemptions and the longer they are maintained, the less likely that will be. 2.16 Sharply cutting the deficit by discouraging private sector exports and investment (through a cut in the subsidies) can produce a sharp fall in export and total growth. This can mean loss of international creditworthiness if Turkey experiences a dramatic increase in its ratios of total foreign debt to exports and of total foreign debt to GNP. This difficulty calls for a strategy aimed at cutting the fiscal deficit while at the same time making private investment and exports grow -- keeping the ratios of the stock of foreign debt to GNP and to exports constant. For a given current account balance, the only way to move to a lower fiscal deficit and higher growth rate seems to be by increasing private savings faster than the increase in private investment, and by phasing out the subsidies in a carefully predetermined sequence. Failure to do this could result in a recession or an increase in the current account deficit, both of which would place great strains on the economy -- particularly when Turkey still faces a debt problem. 30 - C. IMPLICATIONS FOR THE DOMESTIC FINANCIAL SECTOR 2.17 The public sector's reliance on the domestic financial system has had important effects on the functioning of the domestic financial market. First, the pressure of the domestic public debt on the supply of financial savings is analyzed; second, we discuss the effect of reliance on the domestic financial sector and of high nominal deficits on the real interest rates; and third, the effect on banks' behavior and profitability is discussed. Pressure of Public Sector .Deficits on the Domestic Financial Market 2.18 There are two complementary ways of looking at the problem of fiscal pressure on the domestic financial market. First, one could construct an indicator for the pressure that public sector deficits exert on the domestic financial system; and second, one could look at the share of domestic credit allocated to the public sector. Table 2.2 reports on an estimate for the first indicator, and Table 2.3 reports on the consolidated banking sector balance sheet. 2.19 The indicator for the pressure exerted by the public sector deficit on the domestic financial sector is not the real deficit, because the government usually needs to raise money to refinance the payment of the nominal interest payments and not just to make real interest payments.' What is needed is a cash-flow estimate of the fiscal deficits: the amount of nominal resources that the public sector needs to raise from the domestic financial system, for a given supply of foreign resources, in order to meet its payments commitments. Obviously, if foreign resources increase, the pressure on the domestic financial sector " In addition, it is also possible to suggest that individuals cannot always anticipate with errorless precision the inflation component of the nominal interest rate payments received on their financial assets (e.g., public sector liabilities), thus to a certain degree, they suffer from money illusion. See, for discussion: V. Tanzi; M. I. Blejer and M. 0. Teijeiro. "The Effects of Inflation on the Measurement of Fiscal Deficits", in Blejer, M. and Ke-Young Chu (Eds.). Measurement of Fiscal Impact: Methodological Issues. Washington D.C.: IMF, June 1988. 31 - TABLE 2.2 PRESSURE OF THE CONSOLIDATED PUBLIC SECTOR DEFICIT ON THE DOMESTIC FINANCIAL MARKET (As percentage of GNP)(1) 1980-83 1984 1985 1986 1987 1988 Average TOTAL PRESSURE (A+B)(2) 2.42 6.59 5.73 2.94 6.69 8.56 A. MONEY FINANCE (from 1.70 5.74 2.57 1.74 1.82 3.40 Table 1.6) B. NET NOMINAL DOMESTIC 0.72 0.85 3.16 1.20 4.87 4.88 DEBT FINANCE(3) (1+2+3) 1. Bonds and Bills 1.10 2.23 2.52 2.06 3.42 3.08 2. Net Bank Credit 0.43 -0.84 0.86 0.73 1.41 1.52 3. EBFs -0.81 -0.53 -0.22 -1.60 0.04 0.28 MEMO ITEMS: (in percentage) Ratio of Pressure of Fis- 23.25 33.97 35.80 18.16 35.18 33.43 cal Deficits to Nominal Private Savings Ratio of Pressure of Fis- 15.83 68.29 53.78 28.91 65.46 86.42 cal Deficits to CBs Pri- vate Financial Savings Ratio of Stock of Net 23.27 10.36 16.23 14.40 23.29 29.73 Domestic Public Debt to M2X SOURCE: Central Bank, Quarterly Bulletin; and Treasury. (1) Since stocks are measured at end-of-year prices while CNP is measured at mid-year prices, nominal stocks of both domestic and foreign debt were con- verted to mid-year 1980 real prices before computing its ratio to real GNP at 1980 prices. (2) Is an estimate of the pressure on the domestic financial market of the public sector deficits (see discussion in the text). diminishes; but if the public sector needs to transfer resources abroad (i.e., negative foreign financing), the pressure on the domestic financial sector will increase. - 32 - 2.20 As defined more formally in Annex 1.1, the pressure of the fiscal deficit on the domestic financial sector can be defined as the sum of money finance and net nominal domestic public finance, i.e., includes the increase in the real net domestic finance plus the inflation conponent. Table 2.2 reports these estimates. To the extent that the indicator of fiscal pressure on domestic financial markets uses the concept of net nominal domestic finance, it is closer to the definition of the nominal deficit; yet, since it excludes foreign finance, it is smaller than the nominal deficit. The indicator for fiscal pressure on domestic financial markets measures oml the pressure exerted in the domestic financial market. 2.21 This pressure was greatest in 1984-85 and 1987-88, and generaly increased after 1986, from 2.9% of GNP in 1986 to 6.7% and 8.6% of GNP in 1987 and 1988. 1988 is the year of maximum pressure. These increases are a clear evidence of the reduction of net foreign finance; as net foreign finance started to decline and deficits remained unchanged, reliance on the domestic financial sector increased. In terms of sources, the greatest pressure was exerted by money financing. The sharp increase in 1987 and 1988 was explained by both components money finance and nominal net domestic finance. The former increased from close to 1.7% of GNP in 1986 to 3.4% in 1988; the latter from 1.2% of GNP in 1986 to 4.9% of GNP in 1988. 2.22 In the last three rows of Table 2.2, the indicator of the fiscal pressure on the domestic financial market is compared with nominal private savings and the Central Bank's definition of private financial savings,8 and the stock of net domestic public debt (row A in Table 1.2) is compared with private financial assets in the banking sector (M2X). These ratios measure the claim of public sector resources relative to their supply; one measures in terms of flows and 8/This flow of private financial savings was obtained from Table 6 pp 11 of Sak, Gdven and Sevtap Sdngu "Reflections on Securitization in the Context of Turkey". Paper presented to the CMB/OECD Conference on Current Issues in Turkish Capital Markets. Ankara: Central Bank, September, 1989. A problem with this definition of private financial savings, however, is that it excludes currency. - 33 - the other of stocks. The first ratio of pressure on the financial markets to nominal private savings indicates that this pressure was close to its peak in 1988 -- near its level of 1984. In 1988, claims of the fiscal sector amounted to 33% of nominal private savings. The ratio of pressure on the financial markets to private financial savings (i.e., the supply of financial resources in this market), indicates that in 1988 it reached its peak point - - more than 80% of the private financial savings were allocated to finance the PS deficit.9 The ratio of the stock of net domestic debt to M2X reached its peak in 1988. The demand for private financial savings has increased very fast; while in 1984 the ratio of net domestic public debt to M2X was 10%, in 1988 this increased to 30%. As we will see later, this increasing pressure is already generating signs of stress in the domestic financial sector, as is indicated by the high real interest rates and increasing inflation rates. 2.23 Another way to illustrate the pressure that the fiscal deficit exerts on the domestic financial system is to look at the cash-flow of the public sector debt; that is, the ratio of new debt issued to the payment of principal and interest. The higher the ratio, the greater the cost to the public sector and the greater the pressure on domestic financial resources. In Figure 2.2, this ratio and its interest and principal components for bonds and bills are plotted together.10 Once again, the evidence confirms the high cost of the debt in cash-flow terms and the heavy pressure on the domestic financial market; on average, for every one Ti issued of bonds and bills only 0.2 TI goes to finance the primary deficit (i.e., Dt in Annex 1.1). 9/This ratio of pressure of the pressure of the fiscal deficits on to private financial savings overestimates the true pressure on the financial market as it excludes currency holdings and as shown in Table 6.2 the public sector in 1988 financed a significant proportion of its deficit with money finance. The trend -- as suggested by the other two ratios -- is more likely to be correct. 10/The formulas for variables in Figure 2.2 are: P/NI, (P+I)/NI and I/NI respectively; where, NI is new issue of bonds and bills, P is repayment of principal and I is nominal interest repaid. Although it would have been desirable to make this calculation for the total domestic debt, the available data was insufficient. * 34 - FIGURE 2.2 Ratio of Prin and Int to New Borrowings For Bonds and Bills 0.9 - 0.8 - 0.7 - 0.6 / 0.4 0.3 0.2 0.1 0 T1111 1984 1985 1986 1987.1 1987.2 1987.3 1987.4 1988.1 1988.2 1988.3 1988.4 Years/Quarters 0 Principal + Prin + Int interest 2.24 A summary of the consolidated banking sector balance sheet is reported in Table 2.3. By accounting identity, the sum of the net foreign assets (foreign assets less foreign liabilities) and domestic liabilities of the banking system (sum of rows A and D in Table 2.3) is equal to the sum of domestic credit (sum of rows B and C in Table 2.3). For example, a decrease in the net foreign assets (i.e., an increase in foreign liabilities greater than the increase in foreign assets) of the banking system -- such as an increase in the Central Bank's foreign debt -- has to be matched by an increase in domestic credit (sum of rows B and C) or a decrease in M2 (row D). However, Table 2.3 also indicates that a strategy - 35 - aimed at making a net positive transfer abroad, that is, an increase in net foreign assets, would require a decrease in domestic credit or an increase in M2. This table thus enables us to see more clearly the effects of public sector reliance on the domestic supply of financial resources. TABLE 2.3 CONSOLIDATED BANKING SECTOR BALANCE SHEET (As % of GNP) 1983 1984 1985 1986 1987 1988 A. Net Foreign Assets -3.36 -6.85 -10.88 -15.11 -18.93 -14.80 (A-D-B-C) B. Domestic Credit 22.46 25.95 27.47 31.86 29.85 N.A. 1. Public Sector Net 5.20 11.60 11.69 12.09 11.21 N.A. Public Sector,gross 16.93 23.06 25.80 29.17 29.40 27.34 Bonds(Pub. Adm.) 1.54 3.13 4.47 5.08 5.46 5.47 Other(Credits) 15.39 19.92 21.33 24.09 23.94 21.87 2. Private Sector, 17.27 14.35 15.77 19.77 18.64 15.67 gross C. Other Items, net 5.15 4.70 8.29 9.15 11.19 N.A. Gross(unclassified) 9.00 7.76 9.42 11.24 16.67 15.67 D. Broad Money (M2) 24.25 23.80 24.88 25.91 22.11 19.56 1. Narrow Money (Ml) 14.31 10.35 9.80 11.16 11.08 8.28 2. Quasi-Money 9.94 13.45 15.08 14.75 11.03 11.28 MEMO ITEMS: Broad Money (M2X)(1) 24.42 25.09 27.23 29.58 27.39 25.36 FEX(2) 0.17 1.29 2.34 3.68 5.28 5.80 SOURCE: For the period 1983-87, Central Bank worksheets; and for 1988, Central Bank, Quarterly Bulletin. (1) This is row D (M2) plus FEX. (2) These are the residents Foreign exchange deposits. 2.25 Overall, this table suggests that for the period 1983-86, the expansion in domestic credit (chiefly, the increase in net credit to the public sector) and in the item "other items " (row C) was chiefly financed by a decrease in net foreign assets -- chiefly, an inflow of foreign liabilities to the Central Bank - 36 -- and by an increase in M2. The table shows that the faster increase was experienced in net domestic credit to the public sector from 5.2% of GNP in 1983 to 12% in 1986 and in "other items"11 from .2% of GNP in 1983 to 9% in 1986. A close inspection of "other items" suggests that 90% of it is accounted for by the exchange rate losses on foreign liabilities, which are, according to the Central Bank's Charter, a Treasury liability. During the period 1983-86, the sum of net credit to the public sector and in "other items" increased by about 11% of GNP. This was financed by a decrease in net foreign assets on the order of 12% of GNP, and by an increase of more than 1% of GNP in M2; domestic credit to the private sector (row 2) showed a very moderate increase during the same period. 2.26 The period 1987-88 suggests a different conclusion. The increase in net foreign assets and the fall in M2 in 1988 relative to 1987 was chiefly financed by a fall in net domestic credit to the private sector. The sum of credit to the public sector and the item "other items" as suggested by the gross figures, showed no significant change. 2.27 The observations from Table 2.3 for the periods 1983-86 and 1987-88 is consistent with earlier observations that while the availability of foreign sources of finance had led to an expansion of the public sector, once these started to dry up, private sector credit began to fall, yet the domestic cr-dit to the public sector showed a slight fall. 2.28 Using evidence discussed in Chapter 1, two mechanisms were most used by the authorities to encourage the private sector to finance the increase in net foreign assets. /11It is important to note that bank credit to the public sector in Table 2.3 also includes banks' holdings of bonds and bills. Also, official deposits must be collateralized 65 percent by Treasury securities. Banks in Turkey are the most important holders; individuals' holdings have been negligible and are limited to those trading in the REPO market. 37 - 2.29 First, there was monetary expansion (monetary financing) and its resulting inflation. In general, when the Central Bank expands the monetary base by generating more inflation, it encourages the private sector to increase its nominal money holdings, and therefore, to increase private sector holdings of nominal net assets. These resources could then be used to pay for the foreign liabilities by purchasing foreign exchange in the domestic market, and also by increasing the net domestic credit to the public sector, if the Treasury's foreign liabilities were the ones that were reduced. 2.30 Second, there was an increase in real interest rates. Real lending interest rates remained high until the end of 1987 and after a short-term decline in early-1988, they peaked up again by mid-1988 (see Figure 2.4). In general, an increase in real interest rates encourages private sector asset holders to hold more financial assets (i.e., M2) and borrowers to demand less credit, thereby releasing financial resources to finance the increase in the net foreign financial assets or the net credit of the public sector. The exchange rate works in a similar way: expectation of a real devaluation encourages the private sector to increase its holdings of foreign-denominated assets and discourages new borrowing in foreign currencies.12 However, when there is a large proportion of foreign liabilities, as in Turkey, the necessary resource transfer generated by a real devaluation needs to be much larger. The private sector needs to accumulate sufficient net assets to outweigh the increase in the domestic currency value of the foreign liabilities (i.e., to compensate for the exchange rate losses). In Turkey, to the extent that the private sector does not increase its asset 12/The only situation under which total net assets of the private sector will not increase is if the real interest rate on loans does not fully compensate for the real devaluations. - 38 - holding in the same proportion (i.e., M2) by which Central Bank's foreign liabilities increase, real devaluations need to be complemented with an increase in real interest rates or by monetary financing.13 2.31 The upshot in Turkey, which appears very clear from Table 2.3, was that despite the rapid increase in real lending (uncontrolled) rates since 1984 (fluctuating between 10% and 40%), credit to the private sector decreased only in 1987 and 1988. This could, of course, reflect the importance of subsidized credit, but perhaps more important, it might also reflect the high proportion of non-performing loans14 and thus of distress borrowing in the financial system, and the fact that nominal interest expenses are tax-deductible. However, this presents a serious problem for a strategy aimed at increasing net foreign assets by reducing domestic credit. The demand for credit does not seem very sensitive to changes in the real interest rate. The Fffect of Public Sector Deficits on Real Tnterest Rates 2.32 In Turkey, real interest rates are affected by both the increasing public sector reliance on the domestic financial sector and by the decreasing foreign savings. The high level of real interest rate is a widespread concern in Turkey. And interest rates are also affected by the policy of subsidizing exports and investments, which have caused the segmentation of the financial market. 2.33 During 1980-88, Turkey pursued an exchange rate policy consistent with a reduction in the current account deficit. One instrument used to achieve this was by trying to devalue its real exchange rate with Turkey's main trade partners. Moreover, in general, the lower the current account deficit desired (i.e., the 13/For a similar argument see: OED, The Overview of Structural Adjustment Loans I-V. Washington D.C.: The World Bank, April 1988. 14/Although it is difficult to provide supportive evidence on the importance of non-performing loans as these are classified as performing loans in Turkish banks' balance-sheets, this is widely accepted among bankers and authorities. 39 - lower the borrowing from abroad), the higher the real exchange rate needed. To further encourage the increase in exports, Turkey also used an export subsidy scheme. FIGURE 2.3 FIGURE 2.4 6 M Deposit. and Bills Rates in TI & US Lending interest Rates in TL and US RAw .ate.mt setw* .C.*5steet W.te" I ,0 30 - 60 - 10 '40 0 1984 1985 186 198 19l 1989 1984 19 196 1987 1988 *Mo U.-thTt us.thYeaw 0 LISOh .o US * Tsi t 1 *T# 6 . De TI 0 Lead t. * Lot Us 2.34 This policy had an effect on the domestic market via the effect of the exchange rate devaluation on the real interest rate.15 It not only encouraged currency substitution, from Ti-denominated assets into foreign currencies (see Annex 6.1 for an estimation of the elasticity of substitution),16 but, to the extent that the domestic interest rate is market determined, domestic real interest rates adjusted for the expected real devaluation. This not only meant that real domestic interest rates became high (see Figure 2.4), but also that 15/1n equilibrium and with full capital mobility, the real interest rate in the domestic market should be equal to the international real interest rate plus the real devaluation plus a risk premium. 16/In recent years, currency substitution has been further encouraged by making the cost for banks' of reserve requirements on FEX lower than those deposits denominated in Tl. See below. - 40 they were subject to a high variance as a consequence of the instability in the financial market, as asset holders do not know ex-ante the future levels of real domestic interest rates and of the real exchange rate. 2.35 The real effective foreign and domestic deposit and lending interest rates are compared in Figures 2.3 and 2.4.17 The two most noticeable aspects of these figures are: the high level of both foreign and domestic interest rates, the higher level of the domestic real lending interest rate relative to the international real lending interest rate and their high variance.18 Starting in 1984, real interest rates, on average, varied between -20% and above 60% (see Figure 2.4). But perhaps more important, the banks' and individuals' management of their portfolios became unstable. Although speculation is necessary for arbitrage, one cannot disregard its adverse effects in terms of both banks' primary objective of lending to increase production capacity, and of the authorities' monetary management -- by encouraging currency substitution (see Annex 2.1 for more discussion).19 Both of these are now typical features of Turkey's financial markets. 2.36 The October 1988 run on the Tl is an example both of banks' difficulties in placing their funds with low-risk clients, and of the monetary authorities' inability to conduct a stable monetary policy (c.f., the rapid increase in the dollar real deposit and lending rates). The increase in foreign exchange deposits 17/Foreigr. rates were converted into real T1 rates by applying the formula: r - (((l+i*), '-+p*))(1+e)-1)100, where r is the international real interest rate expressed in Tl, i* is the nominal interest rate in the international market, p* is the international price index measured as a weighted average of Turkey's main trade partners and e is the effective real rate of exchange with Turkey's main trade partners. Notice that since (1+p*) is in both the denominator and the numerator (used to calculate the real exchange rate) it cancels out. For the calculation of all real interest rates, the last quarter annualized inflation rate was used. 18/The high level of the real deposit and lending rates of the US market is explained chiefly, by the strong real devaluation of the T1 undertaken by Turkey. See formula in note 17. 19/A high domestic inflation also encourages currency substitution and, thus contributes to the loss of control by the authorities of the monetary management. 41 - (although restricted) and in banks' foreign exchange assets and liabilities are examples of the banks' difficulties in finding low-risk borrowers. For example, total foreign liabilities of deposit money banks increased from 10% of M2X in 1983 to 53% at the end of 1987; and foreign exchange deposits increased from being 6% of deposit money banks' foreign liabilities in 1983 to 51% in 1987.20 This does not imply that the authorities should abandon their real devaluation policy which has been crucial in reducing their current account deficit, but rather that this policy needs to be consistent with the pressure that the financial sector can absorb without causing uneasiness in the financial market. This point becomes more relevant when it is realized that many banks are carrying a sizeable proportion of non-performing assets, and are thus vulnerable to speculation. 2.37 The October 1988 crisis is a good illustration of the effects of a speculative surge on banks' stability. In Figures 2.5 and 2.6, the interbank nominal exchange rate (TL/US$) and the interbank overnight annual nominal interest rates are plotted. The crisis started when banks and foreign exchange traders in general, anticipated a nominal devaluation and started to speculate against the Turkish liraby buying foreign currency. The banks' behavior could be explained by the acceleration of inflation and by the fact that in the past the Central Bank always tried not to let nominal devaluation lag too much behind inflation. However, the banks forced the nominal devaluation much faster than the Central Bank would have wanted in this instance. 2.38 Moreover, banks, on average, favoured time deposits denominated ill foreign currencies by offering nominal interest rates above those offered in the international markets such as the LIBOR interest rates. For instance, in 1988 the bank offering the highest interest on foreign deposits was paying 10.9% on dollar-denominated deposits and 8.1% on DM denominated deposits, while the LIBOR interest rates on both currencies for similar maturity deposits were 8.2% and 20/Data obtained from: OECD, Op. Cit., Paris: OECD, 1988. Table 16, page 40. -42 4.2% (see Annex 2, Table A.2.2). Banks offering the lower interest rates were offering 7.8% and 4.8% respectively, lower than the international average for dollar-denominated deposits and higher for deposits in DM. A similar conclusion could be drawn from 1986-87 figures. In general, in countries that are subject to foreign exchange controls -- even though these are fairly relaxed in Turkey -- commercipl banks do not need to offer interest rates on foreign deposits above internatio.ial levels, as they are subject to little competition from abroad. Furthermore, the attractiveness of these deposits, as discussed in Annex 2.1, arises from the devaluation and from inflation. This evidence then underlines Turkish banks' eagerness to capture these type of deposits and to increase their overall volume of deposits. In fact, these deposits grew very fast in the recent past. 2.39 Figure 2.5 shows the rapid nominal devaluation of the T1 in the interbank foreign exchange market as banks started to speculate against it. The nominal exchange rate went from close to 1700 Tl/US$ in early September 1988 to close to 2000 in early October. Figure 2.5 also shows the sharp fall in the interbank exchange rate over a very short period (three days) as a result of Central Bank intervention. The Central Bank intervened in the market first by increasing the overnight nominal interest rate from 40% annualised nominal interest rate on October 12 to 329% by October 14, to close a source of speculative funds. Second, the Central Bank started selling in the foreign exchange market at a lower price, and in less than three days the interbank nominal exchange rate fell from close to 2000 Tl/US$ to close to 1600 Tl/US$. This discussion thus underlines the risk involved in foreign exchange trading, and its effects on the domestic financial system. It also illustrates the authorities' difficulties in conducting a stable monetary policy. 2.40 Domestic real interest rates are also influenced by the proportion of the deficit financed with domestic resources and the instruments used. As discussed - 43 - FIGURE 2.5 FIGURE 2.6 INTERBANK FOREIGN EXCHANGE RATES OVERNIGHT NOMINAl. INTEREST RATES IMIUM AND UAlIUUM U,XiuUU AND UAlUUM 44.340- 172 100 100 20. 1 .. . . . . . . . . . . . .. . . . .. .. . . . .. . .. . .I a . . . . .. , , , , .. . . ..0,. , , , n . , n n . StPT 2C80 o ,P -c s O 00 'II 220. II 0001. 4 e10v0 40 700 S'T4T 14 Xv" T ' o earlier, the deficit is financed using two main instruments. First, by raising finance through the market mechanism, i.e., auctioning of bonds and bills, and borrowing from banks at the market interest rate; and second, by resorting to non-market instruments such as money finance or the imposition of reserve requirements and liquidity ratios. Both produce a similar effect: financial crowding out. Yet, one operates through a general increase in the interest rate, while the other works by quantity rationing, and in the end, also affects the real interest rates. The more money finance is used the higher and more uncertain the inflation rate. This greater uncertainty associated with inflation (as shown in Annex 2.1) discourages demand for financial assets, thus affecting the real interest rate. The greater the reserve requirements on banks, the lower the supply of loan4ole resources, which increases the real lending interest rate; to the extent that this becomes a tax on the financial institutions, it also increases the spread between lending and deposit rates. 2.41 The most relevant domestic interest rates are plotted in Figures 2.7 and 2.8. The two different pressures on domestic real interest rates are clear .44 - from these figures. First there is the close relation between the real deposit and the real Treasury bill interest rates (in Figure 2.7), which suggests that deposit rates also incorporate the pressure that the deficit exerts on the financial market. In other words, there is a close substitution between both savings instruments. Moreover, this close relation is also warranted from the functioning of the REPO market (see below for more discussion), which, to a large extent is a sub-market which enables banks to increase yields to their preferred customers and to give banks more flexibility in their allocation of resources. The point is that the higher the portion of the deficit financed domestically, the higher will be the level of the Treasury bill rate, and this will be transmitted to the deposit rates. FIGURE 2.7 FIGURE 2.8 Deposit. Bonds and Bills Rates in TI Lending and Deposit Rates in TI 30.0 40 ' (Ral0eees u*I ha 0e1o atnet to.o 0 --10 .40 -\0*,11.11 1984 1086 1986 87 188 989 1984 1986 986 . 988 .989 UWt.011*(h61 O Tr a 6 Me9 48il 6 M* 0ep..t L*.e4*4 6a14 I T beges0e 6 Me 8.peset 2.42 The sharp fall in the real interest rates of Treasury bills' in 1988 (see Figure 2.7) was the consequence of the Treasury's decision to lower these interest rates by changing the mechanics of the auctioning. The auctioning changed from a system where the pre-anounced volume determined the interest rate, to one in which the interest level determined the volume -- allowing a cap on interest rates. As a result, the volume of real bonds and bills issued showed almost no 45 increase in 1988 (see Table 1.4), yet to compensate for the lack of finance, the public sector had to increase its use of Bank credit. In the end, to the extent that the pressure of the fiscal deficit was not reduced, it still fed into the real interest rate and the capping of the interest rates on bonds and bills could only be seen as a temporary measure. The Effects of Public Sector Deficits on Banks' Behavior 2.43 The use of reserve requirements affects both interest rates and also banks' behavior. Figure 2.8 shows these other implications, indicating both the high real interest rates, and also the wide spread between lending and borrowing rates. On average, Turkish banks' need a high spread to finance the imposition of reserve requirements and of other taxes, and to compensate them for the income not received on their non-performing assets. Moreover, as will be shown, the fact that a significant share of assets are non-performing makes banks more sensitive to reserve requirements and other taxes. 2.44 Two different indicators for the spread between lnnding and deposit real interest rates are plotted together in Figure 2.9. The first (the effective) indicator measures the effective spread for the banks once the cost of reserve requirements and other taxes had been taken into account; and the second (the gross spread) indicator measures only the difference between the weighted average interest received on loans and that paid on deposits, i.e., it excludes the cost for banks of reserve requirements and of other taxes. Hence, the difference between these two measures is an estimate of the share of the banks' gross spread (or gross profitability) transferred to the public sector through the imposition of reserve requirements or other taxes. Both spreads are measured in real terms and exclude operational costs, which are above 4% in Turkey,21 and banks' spread on foreign exchange trading. The ratio of the spread taxed to the total gross spread in plotted in Figure 2.9 and, suggests that between 1986 and 1988, more 21/See Op. Cit. OECD, pp. 69, Table 27. -46- than half of the spread was transferred to the public sector, and also that in 1988 this proportion was reduced and banks' profitability increased. This change coincided with the period when the Treasury was pushing interest rates down. Yet this reversal did not last for long, and in October 1988, when banks' reserve requirements were increased and deposit interest rates were liberalized,22 the proportion of the gross spread taxed increased from around 20% to above 70% (see Figure 2.10) and, as a result, banks profitability fell sharply. FIGURE 2.9 FIGURE 2.10 Bank's Average Real Spread Proportion of Bks Gross TI Spread Taxed 20 16018 9818 1986 1918818 so ?0 30 70 0 f0f 02 .10 . . . . . . . . .o0. . . . . . 19S6 19*e Ill$"1oS "s18 UMlTtAlk MONTWTIAR 0 Iftn.it, 2.45 However, Figures 2.9 and 2.10 do not take into account the fact that banks are not receiving all the interest due on their loans, as they hold non-performing loans. In Figure 2.11 the same two estimates for the proportion of gross spread taxed are plotted but it is now assumed three different hypothesis in relation to the average proportion of non-performing assets in the banking system (0%, 10% and 25%). The conclusion, although subject to further analysis, is very clear: if banks do have on average between 10% and 25% of non-performing assets, then the public sector would be taxing almost aU of the available 22/Banks' profitability also fell in October 1988 as a consequence of the lib- eralization of domestic deposit interest rates and the fact that some banks set their domestic deposit interest rates very high (see Figure 2.7). -47- profitability of banks. Moreover, Figure 2.11 also shows the dramatic increase in the cost of reserve requirements in October 1988. Assuming banks do carry a proportion of non-performing loans, it is possible to suggest that, on average, and before accounting for operational costs, banks in Turkey are unprofitable in their Tl operations after payments for the cost of reserve requirements and other taxes. Therefore, the conclusion reached after taking into account that banks hold a proportion of non-performing loans is very different to that reached assuming that they do not have non-performing loans. FIGURE 2.11 Proportion of Bks Gross TI Spread Taxed Assuming Different Non-Per/Per Assets 500- 450- 400 - 350 - 300- 0 250- CL 200- ISO 1 00- .50 1 50 1986 1987 1988 1989 MONTH/YEAR 0 0% + 10% o 25% 48 2.46 Moreover, as is well known, when banks become unprofitable in QAJh terms because of their non-performing assets -- that is, if balance-sheets do not record these assets as non-performing .23 banks will search for activities that will yield them a quick and substantial return. However, in doing so they will overexpose themselves. Examples of this behavior is the size of banks, unregulated off-balance sheet activities relative to total assets, which, on average are around 40% of total assets.24 In Turkey banks have been resorting to unregulated off-balance sheet activities such as the REPO and foreign exchange markets as a means to increase their profits in a very short time. Although banks trading in the REPO and foreign exchange market do not necessarily increase their risk exposure (as happens in many other countries), the fact that Turkish banks hold a high proportion of non-performing assets -- thus reducing their cash-flow -- makes them more prone to take more risk, as they need the high profits. Problems arise because of the banks' inability to match inflows with outflows. Moreover, the rirk assumed in these operations is not fully reflected in banks balance-sheets. Banks will therefore, become vulnerable to unexpected short-term shocks, such as an unexpected increase in interest rates or devaluation. 2.47 Figures 2.5 and 2.6 illustrate very clearly the risk of trading in the foreign exchange market. These swift changes in the exchange rate suggest that banks can, in a few days, lose a very large amount of money. 2.48 The so-called Repurchasement Agreement (REPO) market is another source of instability for banks . In brief, this market consists of a sub-market for deposits and loans where banks offer the instruments to their preferred or wholesale customers. When banks want to increase the interest paid on the short-term deposits (of less than a month) they do so by selling Treasury bills 23/When non-performing loans are classified as performing, profits recorded in their financial statements will not be adjusted for the fact that the accrued income on these loans was not received. 24/The ratio of off-balance sheet activities for the consolidated banking sector balance sheet increased from 34% in December 1988 to 40% in August 1988. - 49 - to their preferred customers, for a period of less than the maturity (which is three months or more). The Treasury bills, however, very rarely change hands, remaining in bank security boxes, and thus banks use these several times. Banks also offer the opposite service to customers who have invested in Treasury bills and want to borrow against them. Table 2.4 shows estimates for the real interest rates on REPOs, Treasury bills, and sight deposits: REPO deposits' real interest rates are greater than those on sight deposits; and Treasury bills' interest rates are greater than those on REPO deposits. This then suggests that both banks and depositors benefit from the REPO market; the banks profit from the difference between the Treasury bills and the REPO real deposit interest rates, and the depositors profit from the difference between the sight deposit real interest rates (which is the opportunity cost) and the REPO deposit interest rate. Similarly, on the loan side, borrowers benefit by not having to sell their Treasury bills to get the cash needed, and banks increase their yield on the loan (the interest charged on the loan plus the yield received on the Treasury bill during the period of the loan). Moreover, and very important, REPO deposits are not subject to reserve requirements, thus enabling banks to increase the amount of resources lent. 2.49 The risk for banks arises, as in the foreign exchange market, from a maturity mis-match (i.e., exposure). The operation of the REPO market makes it difficult for banks to guarantee a perfect match, especially in an environment, characterised by swift movements in interest rates and expectations. The October crisis made it apparent that these shocks do happen. Moreover, the fact that Central Bank does not control this market, and that the banks do not report on their volume of activity, make the REPO market a source for banks to increase their profitability, but also for increasing banks exposure. 50 - TABLE: 2.4 ALTERNATIVE SHORT-TERM INVESTMENT OPPORTUNITIES FOR LARGE DEPOS- ITORS Repurchaz,ment Agreements Real Interest Real Interest (REPO)(1) Rates Rates Average Average on 3 Months on Sight Deposits Real Interest Maturity Treasury (less than 1 Rates(2) (No of Days) Bills(2) month)(3) (%) (%) (%) 1988 Jan -41.0 17.5 -35.9 -49.1 Feb -39.5 21.5 -31.5 -42.8 Mar -24.6 18.8 -17.0 -34.7 Apr -21.4 16.2 -13.3 -31.9 May -11.7 22.2 0.4 -23.5 Jun 9.5 15.0 22.8 -6.0 Jul 14.0 13.8 28.5 2.0 Ago 6.9 13.0 7.3 -10.0 Sep -9.8 23.0 -3.5 -28.6 Oct -7.7 21.5 -12.3 -26.7 Nov -24.0 19.3 -18.8 N.A. SOURCE: Central Bank and Treasury (1) These are the interest rates and average maturities on the REPO made between the Central Bank and the Banks. (2) These are the interest rates on the last transaction of the month. Real rates were obtained by using the last quarter inflation rate. i.e., backward rate. (3) This is the weighted average of the interest rate offered by deposit money banks on sight deposits. D. CONCLUSIONS 2.50 The conclusion here is that by financing its deficits with instruments such as reserve requirements the public sector is forcing banks to over-stretch themselves, chiefly because by taxing a large portion of their cash-flow profits (i.e., after excluding the income not received on non-performing loans), the government's policy is encouraging banks to overexpose themselves. In the end, the use of high reserve requirements could prove self-defeating, because a major bank failure will have consequences both on the fiscal deficit and on monetary policy, thus affecting overall macroeconomic policy. - 51 - 2.51 Our overall conclusion is that Turkey's financial sector is showing signs of being overstretched. This is partly due to its underdeveloped nature, particularly the low supply of financial savings, and partly to the increased demand for domestic resources to cover public sector deficits; fiscal deficits have not declined since 1986, and dependence on domestic resources to finance them has increased. The latter process is producing adverse effects, such as high inflation and high real interest rates. 2.52 In consequence we recommend that the authorities focus on reducing the budget deficit and on undertaking institutional reforms of the financial market which will permit this market to absorb the increase in domestic public debt needed to produce a transfer of resources abroad. Both reforms will take time and demand careful planning and sequencing. These reforms will determine the pace at which resources can be transferred abroad. -52- ANNEX 1 A TECHNICAL NOTE ON THE ESTIMATION OF PUBLIC SECTOR DEFICITS' A.1.01 In this annex the definitions, methodology and data sources used in estimating the nominal and real fiscal deficits and the pressure of the fiscal deficits on the domestic financial market are discussed. Also, it incorporates the so-called quasi-fiscal deficit of the Central Bank into the discussion of the fiscal deficits. The algebraic definitions of the fiscal deficits are discussed in the first section; and the methodology and sources of information used in estimating the real and nominal fiscal deficits through changes in net public sector liabilities are explained in the second section. A. THE ALGEBRA OF FISCAL DEFICITS A.1.02 In this section the algebraic definition of the real and nominal deficits, the pressure of the deficits on the financial sector and of the quasi-fiscal deficits are derived and discussed. Equations will be defined using the continous time approach, thus enabling the reader a clearer understanding of the underling concepts. However, calculations in Chapter 1 were made using the discrete time approach due to data limitations.2 1/ This methodology is the same as developed by EMTTF for the purpose of The World Bank. Op- Cit-, September, 1988. 2/ The difference between the continous and discrete time approaches is that in the discrete time approach the time period is specifically shown, this though makes this approach cumbersome. The continous time approach has the advantage of being more faithful to the data-generating process (i.e., how data is generated in real life). However, available information enabled us only to use the discrete time approach for our calculations. -53- Nominal Fiscal Deficits A.1.03 The nominal deficit including foreign exchange losses is defined as in equation (A.1.1), (A.1.1) D+i*B+(i*+/^)*B**E=HI+B+1B**E+B**E, where, D is the primary deficit, B and B* are the stocks of nominal net domestic and net foreign public debt denominated in T1 and in dollars respectively, i is the weighted average nominal average interest rates on net domestic debt denominated in Tl, i* is the weighted average international nominal interest rate denominated in dollars, H is the stock of money used to finance the public sector, E is the nominal exchange rate of Ti/US $, a dot above the variable denotes the partial derivative with respect to time (i.e., dx/dt), a hat denotes the rate of change (i.e., (dx/dt)*(l/x)) and t is a time index. Stocks, as is common, are evaluated at end-of-year prices. A.1.04 By decomposing t into (E)/(E) in the left hand side of equation (A.1.1) and cancelling similar terms in the left and right hand side of the equation, it is possible to arrive to the definition of the nominal deficit excluding capital losses expressed as equation (A.1.2), (A.1.2) D+i*B+i*B3**E=I+B+B**E, where variables are as defined before. A.1.5 The left hand side of equation (A.1.2) is the definition of the fiscal deficit from above-the-line and the right hand side is the definition from -54- below-the-line or, through changes in net liabilities of the public sector. As it is apparent from this equation, errors in applying the correct exchange rate in either side of the equation will yield errors in the estimation. Real Fiscal Deficits A.1.06 The real fiscal deficit including capital losses can be derived in a similar way as we derived the nominal fiscal deficit excluding capital losses. Let the real deficit including foreign exchange losses be defined as in equation (A.1.3), (A.1.3) D+r*b+(r*+O)*b**e= [+6+6**e+b**6, where a lower case variable indicates the variable in real terms (i.e., xt - X./Pt, where Pt is the end-of-period price index) r is the real weighted average interest rates on net domestic public debt denominated in T1, r* is the real weighted international interest rate, e is the real effective exchange rate and the rest of variables are as before. A.1.07 By rearranging and cancelling similar terms at the left and right hand side of the equation, it is possible to arrive at the definition of real fiscal deficits excluding capital losses expressed as equation (A.1.4), (A.1.4) D+r*b+r**b**e= H+6+6**e, where variables are as defined above. As before, the left hand side of equation (A.1.4) is the-above-the-line definition of real fiscal deficit and the right hand side is the below-the-line definition. -55- Ouasi-Fiscal Deficit A.1.08 A useful extension of the definition of consolidated fiscal deficits is the definition of quasi-fiscal deficit of the Central Bank. As is well known, the Central Bank in Turkey performs certain activities that would have corresponded to the fiscal sector; the quasi-fiscal deficit definition is an attempt at measuring the net position of the Central Bank resulting from these activities. The quasi-fiscal deficit, therefore, should be interpreted as a transfer from the public to the private sector. Later on, in Section B, we will describe the methodology and discuss an estimate for the quasi-fiscal deficit based on the Profit and Loss Account of the Central Bank. A.1.09 An example -- suited for Turkey -- of Balance Sheet and Profit and Loss Account of a Central Bank is reported in Table A. 1. 1. Making net foreign liabilities as NFL- FL-FA and adjusted money base as H-H'-CP, the increase in net worth in period t (i.e., Nt) is defined from the balance sheet as expressed by equation (A.1.5) and from the Profit and Loss Account as expressed by equation (A.1.6), -56- Table A.1.1: BALANCE SHEET AND PROFIT AND LOSS ACCOUNT OF A CENTRAL BANK BALANCE SHEEI Assets Liabilities Foreign Assets (FAxE) Foreign Liabilities (FLxE) Credit to the Government(Cg) Domestic Liabilities (DL) Capital Losses on Credit to the Gov- Unadjusted Base Money (II') ernment (C9') Net Worth (Nw) Credit to the Banks (Cb) Credit to the Private Sector(C") PROFIT AND LOSS ACCOUNT Debit Credit Interest Paid on Foreign Liabili- Interest Received on Foreign Assets ties((i*+R)xEXFL) ((i** t)xExFA) Interest on Domestic Liabilities (ix DL) Interest Received on Credit to the Government (ix C") Net Income (NV) Interest Received on Credits to Banks (i x<C') Interest Received on Credits to the Private Sector (ixCl) -57. (A..5) ;v = cj+ cjI+ b . VFL - F -ATL* -li - 5L, (A. 1. 6) .i,' i-Cg+ Ct i -Cp-(i*+ )*V LE iDL, where variables are as defined before. A.1.10 Using equations (A.1.5) and (A.1.6), let us define the quasi-fiscal deficit including foreign and domestic exchange rate losses as expressed by equation (A.1.7), (A.1.7) QFD = (i*+P)*NFL*E+i*DL-i'Cg-i* .Cb-i*jC p = NPL*- E +NFL* E_ +H1 +LL- ('g-C0 -C', where QFD denotes quasi-fiscal deficit and the rest of variables are as defined before. The first line of this equation defines the QFD from the income side, i.e., above-the-line, and the second line defines it from the liability side, i.e., from below-the-line. As before, after cancelling similar terms at both sides of equation (A.1.7), let define theQFD free from international and domestic currency fluctuations (QFDf) as in equation (A.1.8), (A.1.8) QFD' = i**NFL*E+i*DL-i*C9-i* .Cb-_i*.C"p = NPL*E+H1+ DL-C _6C-Cb, and let us define the real quasi-fiscal deficit free from internationa'. and domestic currency fluctuations (qfdf) as in equation (A.1.9), -58- (A. 1. 9) qfd = r*f*+ * -r*.Cg-rcb-*c =nfl*o+// +dI -cr-cl- b where a lower case variable indicates the variable in real terms (i.e. , xt - Xt/Pt, where Pt is the end-of-period price index). A.1.11 In all three equations (A.1.7, A.1.8 and A.1.9), the first line defines the quasi-fiscal deficit from the income side, i.e., above-the-line, and the second line from the liability side i.e., below-the-line. These definitions assume that Central Bank's operation costs are zero and that all income/expen- ditures in the Profit and Loss Account generates cash-flow. A more refined definition would have excluded from the above definitions those activities typical of a Central Bank, such as the credit to banks (Cb)3 and the issue of own liabilities for open market operations (DL) purpose. However, in the case of Turkey, - - as shown the Central Bank's Balance Sheet -- none of them are important. Moreover, open market operations are performed with government securities making it difficult to distinguish an open market operation from credit to the government. A.1.12 A characteristic of the Turkish Central Bank -- depicted in our defi- nitions of QFD -- is the inclusion of the item "C91", which is defined as the capital losses on Central Bank's net foreign liabilities resulting from fluctuations in exchange rates. By Central Bank law, the exchange rate losses of Central Bank's foreign liabilities are a claim on the Treasury; yet the Treasury has not always paid them to the Central Bank. For instance, in the 1987 Balance Sheet this item represent more than 40% of the total Central Bank's 3/ This is the case, for instance, with the very short-term line of credits that central banks open for commercial banks, and banks use them only for very short-term liquidity needs. However, when this credit turns into more long-term loans because of some banks' solvency problems, then it is possible to argue that these loans should be included in the definition of quasi-fiscal deficit. The central bank is in effect making a transfer to the private sector. .59- assets. The Central Bank, by including the C91 on the asset side, understates the true size of the quasi-fiscal deficit. Moreover, for this reason and because the item C81 exceeds the total capital losses from cross and domestic currencies fluctuations -- calculated as the difference between the CBs NFL at current international and domestic exchange rates and the total CBs NFL at fix 1980 international and domestic exchange rates -- the above-the-line QFD is larger than the below-the line one. As we argued in Chapter 1 and in Section B in this Annex, it is this inconsistency between the Balance Sheet and the Profit and Loss Account of the Central Bank which results in the discrepancy found when the consolidated fiscal deficit from above-the-line and below-the-line. Pressure of the Fiscal Deficit on the Domestic Financial Market A.1.13 Using the definition for nominal fiscal deficit in equation (A.1.2), it is possible to define fiscal deficits' pressure on domestic financial markets as in equation (A.1.10): (A.1.10)PFI = D+i*B+(i**B**E-8 **E} where PFM is the indicator for pressure on the domestic financial market of the public deficit and the rest of variables are as defined before. A.1.14 The term in brackets () in equation (A.1.10) is equal to the non-interest current account of the public sector with opposite sign, or the net inflow of foreign finance for the public sector with opposite sign; the more negative this term is relative to the total fiscal deficit -- i.e., the greater the inflow of public foreign finance -- the lower the pressure of the deficit on the domestic financial market. This term will become positive when the country is performing -60- a transfer abroad, i.e., when the country is re-paying her foreign debt and, thus, resulting in a greater pressure on the financial market than that exerted by the fiscal deficit. B. METHODOLOGY AND SOURCES OF DATA A.1.15 This section focuses on the methodology and sources of data used to estimate the deficit through changes in net liabilities of the public sector. In what follows, the discussion of the methodology for calculating foreign debt finance, domestic debt finance, money finance and quasi-fiscal deficits is undertaken. Foreign Debt Finance A.1.16 Table A.1.2 reports in the first column the main items of the net foreign liabilities of the public sector (foreign assets and liabilities of the Central Bank and foreign liabilities of the government and SEE's), in the second and third columns the data for 1987 and 1988 for this same breakdown; and in the fourth column the sources from where the data were obtained. A.1.17 First, the nominal net foreign liabilities free from domestic and international currency fluctuations was estimated. For this purpose the non-dollar portion of the foreign public sector net foreign liabilities was converted to 1980 exchange rates by using the 1980 non-dollar to dollar exchange rates. A.1.18 The ratio of nominal net foreign finance (i.e., the increase in net nominal foreign liabilities of the public sector) free from currency fluctuations to GNP was calculated: (i) by estimating the increase in net foreign liabilities of the public sector free from currencies fluctuations denominated in dollar (i.e., by calculating B*,= *,- B*,-.); (ii) by multiplying it by the current annual average T1/US $ exchange rate; and (iii) by dividing it by the nominal GNP. -61- TABLE A.1.2: FOREIGN EXCHANGE LIABILITIES OF THE PUBLIC SECTOR (in millions of US Dollars) 1987.IV 1988.IV DATA SOURCES(*) j I..I..I... . .. . . . . . . . . . . . . . . . . . I- - -- - . . . . . . . . . . . . . . . . . . . I. FOREIGN EXCHANGE ASSETS OF THE | | Central Bank's CENTRAL BANK I | Quarterly Bulletin I I I 11.1 RESERVES j 1,765.60 I 2,330.42 I C88 la.4 11.2 GOLD 1,627.03 1,424.93 j CBS la.1+2 11.3 INCONVERTIBLES j 156.36 ) 153.81 CBS 1a.5 11.4 OTHER FX ASSETS j 1,947.36 2,306.69 CBS la.39 11.5 TOTAL ASSETS OF THE CENTRAL BANK I 5,496.35 I 6,215.85 I I I I I 1 II. FOREIGN EXCHANGE LIABILITIES OF THE I I CENTRAL BANK I I I I 1 I I I 111.1 OVERDRAFTS I 289.78 I 169.30 | CBB 1p.2 I 111.2 INCONVERTIBLES j 386.53 I 361.22 I CBS 1p.32 | 111.3 DRESDNER I 5,913.00 I 6,208.00 CB Worksheets I 111.4 DIRECT BORROWINGS AND CTLDs 4,179.07 j 2,876.09 | CBS 2p.27 I 111.5 LETTERS OF CREDIT j 106.23 I 106.46 I CBS 1p.30 111.6 BANK DEPOSITS AND OTHER j 4,995.20 j 5,457.20 | CBB 2p.25-(II.1+1l.2+II.3) 111.7 TOTAL FOREIGN LIABILITIES OF THE 115,869.81 115,178.27 j CBS 2p.28 I CENTRAL BANK I ( I I I I II 1111. FOREIGN EXCHANGE LIABILITIES OF I | I Central Bank's OF GOVERNMENT AND SEEs | | | Outstanding External SI I ( Debt worksheets 1111.1 MEDIUM-LONG TERM 116,974.00 117,727.00 1 1111.2 IMF | 770.00 1 299.00 j 1111.3 SHORT-TERM I 491.00 | 242.00 1111.4 18RD j 6,289.50 J 6,129.70 I 1111.5 TOTAL FOREIGN LIABILITIES 124,524.50 124,397.70 j OF GOVERNMENT AND SEEs j j j IIV. NET FOREIGN EXCHANGE LIABILITIES ) I I OF THE PUBLIC SECTOR 134,897.96 133,360.12 1 IV. PRICES AND EXCHANGE RATES j j (State Planning Organization, II I Main Economic Indicators IV.1 EXCHANGE RATE TL/USS j 991.18 I 1,795.04 f IV.2 EXT. CURR. PRICE INDEX(--) j 1.17 I 1.14 I IV.3 EXT. ADJ. PRICE INDEX(--) 1.27 ( 1.29 f V.4 DOMESTIC PRICES 11.35 I 19.89 I IV.5 RER INDEX I 1.32 j 1.34 I IV.6 DOMESTIC AVERAGE PRICE INDEX(**) | 8.89 I 15.60 I IV.7 AVERAGE EXC RATE TL/USS j 855.68 I 1,416.49 I (*) For instance, CBS 1a.5, indicates Central Bank's Quarterly Bulletin, Table la, row 5. (*) Base 1980=1 -62- A.1.19 Second, the ratio of the real net foreign finance free from currencies fluctuations to real GNP was estimated: (i) by estimating the stock of net foreign liabilities at 1980 dollar prices by using an international weighted average price index. This price index was estimated using a geometric average formula as expressed by equation (A.1.11), jP*/ 0 ,Ihre:P*t = (P* /EU o )a 1- after taking natural logs at both sides of the equation this can be written as: (A.1.11a) t-n Z aLn(P/E'/$ I where: a' is a weight that takes the value of the proportion of total foreign trade performed with currency i, P is the end-year Consumer Price Index of country with currency i, E's is the end-year exchange rate of currency i with respect to the US dollar in 1980, Ln is the natural log, FT denotes a product and denotes a sum. A.1.20 It is important to notice from this methodology that both the stock of net foreign liabilities and the international weighted average price index used (i.e., numerator and denominator) exclude fluctuations in the international exchange rates. A.1.21 (ii) By calculating the stock of net foreign debt of the public sector at mid-1980 Tl prices (i.e., in real Tl terms). This was done by multiplying the stock of foreign liabilities free from currencies fluctuations by the end-1980 exchange rate of T1 to US $ and by dividing this product by an end-of-period -63- domestic price index with base 100 in mid-1980. And (iii), the ratio of the annual change in stock of real net foreign liabilites free from currency fluc- tuations at mid-1980 Ti prices to current GNP at 1980 prices was estimated. The use of this ad-hoc domestic price index assures us that both the numerator (the change in stock in real net foreign liabilities) and the denominator (the current GNP in real terms) were evaluated at the same prices: mid-1980. As it is well known, when this correction is not performed in inflationary economies this price effect can introduce important distortions in the final result. This price effect results from the fact that the numerator is evaluated at end-of-year prices (the case of stocks in real terms) while the denominator is evaluated at mid-year prices (the case of flows in real terms). A.1.22 The estimate of net foreign finance including international and domestic currencies fluctuations as a percentage of GNP in Table 1.4 was estimated (i), by estimating the TI stock of net foreign liabilities including currency fluc- tuations at mid-1980 prices. This was estimated by multiplying the dollar- denominated stock of net foreign liabilities without correction for currency fluctuations by the current end-1980 Ti/US $ exchange rate and dividing it by the ad-hoc end-of-period domestic price index with base at mid-1980. And (ii), by calculating the annual change in the mid-1980 stock of net foreign liabilites and dividing it by the current GNP at 1980 prices. Capital losses were obtained as the difference between the estimates which include capital losses and those which exclude them. A.1.23 The two international price indexes -- including (Int Cut PS) and excluding (Int Adj PS) international currencies fluctuations -- and the real effective exchange rate index (RER) are plotted in Figure A.1.1. As in the IMF methodology, in this figure an appreciation in the RER is shown as an increase in the price index and a devaluation as a decrease. The difference between the -64- international price index which includes the fluctuation of international cur- rencies and that which excludes them, is the effect of fluctuation in international currencies; and the difference between RER and the international price index which includes the fluctuation in the international currencies is the effect of the real effective devaluation (appreciation) of the Turkish Lira against the basket of international currencies. Both the two international price indexes and the RER have the same weights for international currencies and correspond to Turkey's share of the total trade performed with each currency. By comparing Figure A.1.1 with Figure 1.1 one can see the effect that changes in international prices, international currencies fluctuation and of real effective devaluation (appreciation) of the TI have had on the stock of net external public sector debt. -65- FIGURE A.1.1 RER AND INTERNATIONAL PS INDEXES INCL AND EXCL CURRENCIES FLUCTUATIONS 1 .4 I .3 - 1.2- 1.1 0.9 0.8- 0.7- 0.6111111 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 o INT ADJ PS + INT CUR PS + RER INDEX Domestic Debt Finance A.1.24 Table A.1.3 reports in the first column the decomposition of the domestic debt, in the next two columns the estimates for 1987 and 1988 in nominal terms and in the last column the sources where the data was obtained. A.1.25 First, the ratio of nominal net domestic debt finance to GNP was calculated: (i) by estimating the annual increase in net domestic debt at current prices; and (ii), by dividing it by GNP at current prices. -66- TABLE A.1.3: NOMINAL NET DOMESTIC DEBT OF THE PUBLIC SECTOR AND ADJUSTED BASE MONEY (in million of current TI) 1987.IV I 1988.IV DATA SOURCES(*) -- - -- - - -- - - -- - - -- - - --- - -- - - --.-- - - -- - - .. . . .. . . .. . . .. . . I. NOMINAL NET DOMESTIC PUBLIC SECTOR DEBT j 4,757,356.20 j 9,655,380.00 I I I I I 1. CENTRAL GOVERNMENT | 4,064,835.20 I 7,134,000.00 I I A. BONDS AND BILLS ( 4,330,564.20 ( 7,421,000.00 Treasury's I 1.1. Bonds I 2,407,250.50 I 4,880,000.00 j worksheets | 1.2. Bills I 1,923,313.70 2,541,000.00 I I B. NET CREDITS I (265,729.00)1 (287,000.00)1 I 1.3. Bank Credits 1 1,118,959.00 | 1,795,400.00 I I a. Deposit Money Banks 1,094,543.00 1,748,800.00 I CBB 3a.10+3a.12 I I b. Development Banks 24,416.00 j 46,600.00 I CBS 5a.7 I 1.4. Bank Deposits J 1,384,688.00 ) 2,082,400.00 I | a. Deposit Money Banks j 1,172,696.00 j 1,777,000.00 | CBS 3p.31 I I b. Development Banks I 211,992.00 ( 305,400.00 I CBB 5p.29 I I I I I I I 2. SEEs 1 1,721,730.00 j 3,265,500.00 j I A. NET CREDITS I | | 2.3. Bank Credits | 2,862,058.00 I 4,630,100.00 I ( a. Deposit Money Banks I 2,248,900.00 I 3,425,100.00 j CBS 3a.18 b. Development Banks ( 613,158.00 1,205,000.00 ( CBS 5a.13 ( 2.4. Bank Deposits 1,140,328.00 1,364,600.00 | | I a. Deposit Money Banks | 1,138,293.00 1,355,200.00 I CBS 3p.36 I b. Development Bankt. ( 2,035.00 | 9,400.00 I CBS 5p.30 I I I I I I 3. EBFs j(1,029,209.00)1 (744,120.00)1 EBF's 3.1. R.S.Certificates j 766,480.00 1 721,480.00 Information I 3.2. EBFs Deposits I 1,795,689.00 j 1,465,600.00 ] I I I I I II. NOMINAL NON-ADJUSTED AND ADJUSTED BASE MONEYJ I | 1. NON-ADJUSTED BASE MONEY I 5,080,532.00 ( 9,060,300.00 ( a. Currency in Circulation j 2,274,737.00 | 3,425,600.00 I CBS 4ap.23 b. Vault Cash I 769,368.00 j 1,092,800.00 CBS 3a.1 j c. Free Reserves ( 438,491.00 j 794,000.00 j CB 8mk.12 d. Required Reserves j 1,597,936.00 I 3,747,900.00 I CS 1p.13 I 2. ADJUSTED BASE MONEY I 4,247,311.00 | 7,657,100.00 j A. REDISCOUNTS I 1,036,379.00 I 1,736,100.00 j | a. Deposit Money Banks I 893,119.00 1,289,300.00 | CBS 2a.17-2p.38 I I b. Other Fin. Inst. j 143,260.00 j 446,800.00 j CBS Sp.28-5a.4+3a.4 I B. IMPORT DEPOSITS I 203,158.00 I 332,900.00 | CBS 1p.31 I I I I I I Jill. PRICES AND OUTPUT (1980 = Base) IState Planning Organization I Main Economic Indicators DOMESTIC PRICES(END) I 11.35 I 19.89 j DOMESTIC PRICES(AVE) I 8.89 j 15.60 j NOMINAL GNP 1 58,387.20 I 100,386.00 | | REAL GNP(CPI) I 6,567.74 | 6,436.61 j INFLATION(END) I 55.05 I 75.21 I (*) For instance, CBS 4ap.23, indicates Central Bank's Quarterly Bulletin, Table 4ap, column 23. -67- A.1.26 Second, the real net domestic debt finance to real current GNP was calculated (i) by calculating the stock of net domestic debt at 1980 prices by dividing the stock of nominal net domestic debt by the ad-hoc end-of-year domestic price index with base in mid-1980; (ii), by calculating the annual differences in the stock of net domestic debt at 1980 prices; and (iii) by dividing the change in the stock of net domestic debt at 1980 prices by the current GNP at 1980 prices. Money Finance A.1.27 Table A.1.3 also reports the money finance estimates. As before, the first column shows the breakdown, the next two columns the estimates for 1987 and 1988 at current prices and the last column the sources where the information was obtained. The adjusted money base estimate, which is defined as the portion of base money used to finance the public sector, was obtained by adding currency in circulation plus total banks' reserves (vault cash, required and free banks' reserves) plus import deposits less rediscounts granted to the private sector. A.1.28 The ratio of money finance to GNP was estimated first, by calculating the annual change in the adjusted money stock at current prices; and second, by dividing it by GNP at current prices. More formally this can be written as: H1, ,- where H is the stock of adjusted money stock at current prices and GNP is the GNP at current prices. A.1.29 It is common to find in the literature different methods and definitions for money finance, seignorage and inflation tax; it is therefore helpful to discuss the definitions and methods used in our analysis and their relation to others. While the most accurate definition would have been to use continous form equation (i.e., an integral), the data available only rendered it possible to use a discrete form equation (i.e., changes in end-year stocks). A.1.30 The inflation tax and seignorage estimates reported in Table 1,4 are defined as the portion of seignorage and inflation tax used to finance the public sector deficit. The difference between the total money finance (i.e., that estimated from the non-adjusted money stock) and the portion of money finance used to finance the public sector deficit (i.e., the adjusted money finance), are the resources returned to the private sector, e.g., through rediscounts. A.1.31 Formally, the true seignorage and inflation tax can be defined as equations (A.1.12) and (A.1.13),4 (A.1.12) ( t" - t- I) St gnp, and (A.1.13) tITx h P gnpt Hi CNP, where: St is the seignorage, ITt is the inflation tax, hN--, gnp,*-)- (i.e., lower case variables denote the variable in real terms), Ht is the adjusted money stock, T, is the average domestic consumer price index with base 1980, P, is the end-of-period domestic price index with base mid-1980, P, is the inflation rate and GNPt is the GNP at current prices. A.1.32 Seignorage as expressed by equation (A.1.12) is defined in the same way as we defined the real net foreign and real net domestic finance. A problem of estimating seignorage and inflation tax as expressed by equations (A.1.12) and (A.1.13), is that they do not add to the adjusted money finance. And, because we wanted a definition of seignorage and of inflation tax that would add up to 4 /A more precise estimate would had been obtained using integrals. -69- the adjusted money finance, we decided to define seignorage as equation A.1.12 and inflation tax as the difference between money finance and seignorage. Formally, the definition of inflation tax used can be written as equation (A.1.14), IT' = - GNPt gnpt by expanding and rearranging equation (A.1.14), it is possible to arrive to an expression that relates our definition of inflation tax (IT') to the true inflation tax (IT) as defined by equation (A.1.13). This is expressed as equation (A.1.15), (A..15 ITt + (HI-H-1) Pt-T,P I'=+ *, (1+?,) GNPt Pj where variables are defined as before. A.1.33 Differences between IT and IT' are very small (less than 0.5% of GNP). except for 1981 when the rate of inflation increased sharply. As it is apparent from equation (A. 1.15), these differences depend on the rate of inflation measured by the end-of-period domestic prices index, on the difference between the average and end-of-period domestic price indexes and on the size of the nominal money finance (second term in equation A.1.15). In general, during periods of high inflation IT' underestimates IT and during periods of decelaration of inflation IT' overestimates IT. Examples of the latter are 1982 and 1984. A.1.34 A less troublesome method to decompose the adjusted money finance term between seignorage and inflation tax -- which was used in The Fiscal Policy Report -- would have been by using equation (A.1.16), -70- = + GNP, GNP CA1P. this equation can also be expressed in terms of the true seignorage and inflation tax as defined by equations (A.1.12 and A.1.13), (A.1.16a) Ht - -1 Pt Pl-1 = St* -+ ITt*--- GNPt P, P, where variables are as defined before. A.1.35 From equation (A.1.16a) it is apparent that seignorage and inflation tax as defined by equation (A.1.16) is slightly different from the true seignorage and inflation tax (S and IT). In general, if inflLtion is increasing then the estimated seignorage and inflation tax using equation (A.1.16) will underestimate the true seignorage (S) and overestimate the true inflation tax (IT). However, as mentioned before, differences are small and will only become worrisome when the inflation rate experiences large fluctuations. At the end, the decision of how to define seignorage and inflation tax will depend on the particular interest of the author and on the purpose of the study. Quasi-Fiscal Deficit A.1.36 The quasi-fiscal deficit, was estimated both from the Profit and Loss Account and from the Balance Sheets of the Central Bank. The above-the-line quasi-fiscal deficit estimate which, added to the government PSBR estimate -- also estimated through above the line -- will yield the above-the-line consolidated -71- fiscal deficit estimate. Both estimates of the consolidated fiscal deficit from below-the-line and above-the-line are reported in Tables 1.3 and 1.4 and its detail is reported in Tables A.1.4, A.1.7 and A.1.8. A.1.37 The above-the-line quasi-fiscal estimates are reported in Table A.1.4. In this table the Profit and Loss Account of the Central Bank is reported; the items which generate more cash-flow have been put at the top of the table and those which generate less cash-flow are at the bottom of the table. At the extreme bottom of this table the Net Profits as reported in the Profit and Loss Account of the Central Bank and an estimate for the cash-flow Net Profits (i.e., the cash-flow quasi-fiscal deficit) are reported. This latter estimate intends to correct for those revenues/expenditures which, although recorded in the Profit and Loss Account, are not effectively received. An example is the revaluation of gold. A cash-flow Net Profit or Loss indicates the true net profits effectively received or, if it is a loss, the demand for additional liabilities to finance it. In the case of a Central Bank this is particularly important because it has the advantage of being able to issue reserve money as its liability: the Central Bank finances its losses by forcing the private sector to increase its nominal money holdings through the creation of a higher level of inflation. A.1.38 The cash-flow estimate of net profits of the Central Bank was calculated by deducting from the reported "Net Profits" those items which we believe do not generate a cash-flow net income. As mentioned before, this is the case of the revaluation from gold, but also of other items such as the interest on Public Sector (Treasury and SEE's) assets. Although this method for estimating the cash-flow Net Profits can be debatable and, therefore needs further refinement as more data becomes available, the fact that for the period 1983-88 this estimate is very close to "Net Interest Received" suggests that it is a relatively effective -72- TABLE A.1.4: PROFIT AND LOSS ACCOUNT OF THE CENTRAL BANK OF THE REPUBLIC OF TURKEY (As percentage of GNP) ..... .. . . . . . . . . ... . . . . . . . 1983 19864 1985 1986 198? 19881 I....... ................ ....................................... I. NET INTEREST RECEIVED (*A-) I -0.11 *0.88 *1.08 *0.48 0.0? 0.07 | A. INTEREST RECEIVED (1) I 1.17 0.70 0.62 0.92 1.45 1.94 I | 1. Interest received on various advances I 1.06 0.59 0.37 0.39 0.66 a.- Advances against treasury guaranteed bills | 0.08 0.02 0.06 0.08 0.18 | b.- Advances commercial bills I 0.68 0.30 0.07 0.07 0.09 c.- Advances medium-term credits j 0.21 0.19 0.19 0.18 0.20 I d.- Advances agricultural bills ( 0.04 0.04 0.02 0.03 0.10 | e.- Others I 0.05 0.03 0.03 0.03 0.09 I | 2. Interest received from foreign correspondents | 0.09 0.10 0.17 0.38 0.30 I a.* Current and time deposits accounts | 0.09 0.10 0.17 0.22 0.13 | b.- Portfolio accounts ( 0.00 0.00 0.00 0.17 0.18 I 3. Other Interest I 0.02 0.01 0.07 0.14 0.13 I a.- Interest charged on due debts of Public Sector Institutions | 0.00 0.00 0.05 0.13 0.11 I b.- Penalty Interest on Reserve Requirements I 0.01 0.00 0.00 0.00 0.00 I c.- Others ( 0.01 0.01 0.02 0.01 0.02 | 8. INTEREST PAID (1) I 1.28 1.58 1.70 1.40 1.38 1.87 I I 1. Interest paid to foreign correspondents I 0.27 0.33 0.46 0.53 0.39 I I 2. Others | 1.01 1.25 1.24 0.88 0.96 I I a.- Interest paid on foreign exchange accounts I 0.22 0.14 0.25 0.72 0.77 | | i.e., on Dresdner Bank scheme. I b.- interest paid on foreign exchange deposit accounts 0.00 0.01 0.12 0.12 0.11 I i.e., on voluntary foreign exchange deposits. c.- Interest paid on reserve requirements 0.79 1.10 0.87 0.03 0.06 I (including TI reserve requirements up to 1985) I I I d.- Others j 0.00 0.00 0.00 0.00 0.02 | II. NET COMMISSIONS RECEIVED j -0.02 0.00 *0.02 0.00 0.02 0.06 | A. COMMISSIONS RECEIVED 0.05 0.04 0.03 0.03 0.04 0.15 S. COMMISSIONS PAID I 0.07 0.04 0.05 0.03 0.03 0.08 |llt. NET PROFITS FROM GOLD AND FOREIGN EXCHANGE OPERATIONS I 0.24 0.99 1.18 0.58 0.05 0.07 A. PROFITS FROM GOLD AND FOREIGN EXCHANGE OPERATIONS 0.24 0.99 1.18 0.58 0.06 0.10 | | a.- Gold trading profits and evaluation difference 0.00 0.84 0.97 0.39 b.- Foreign exchange trading profits 0.23 0.11 0.10 0.11 c.- Interest earned on swas 0.00 0.00 0.00 0.07 d.- Arbitrage profits j 0.01 0.01 0.01 0.01 j e.- Earnings on portfolio accounts held with I 0.00 0.00 0.10 0.00 I foreign correspondents I f.- Others 0.00 0.03 0.00 0.00 0.00 0.00 I 8. LOSSES FROM GOLD AND FOREIGN EXCHANGE OPERATIONS I 0.00 0.00 0.00 0.00 0.02 0.03 ) (IV. OTHER NET RECEIPTS I 0.00 -0.02 *0.01 0.01 -0.01 -0.05 I I A. PROFITS FROM SECURITIES 0.00 0.00 0.00 0.00 0.00 0.00 I 8. OTHER PROFITS | 0.04 0.02 0.02 0.04 0.03 0.00 | C. OTHER EXPENDITURES AND LOSSES j 0.04 0.04 0.03 0.04 0.05 0.05 j IV. PROVISIONS AND DEPRECIATIONS ALLOWANCES I 0.01 0.02 0.02 0.03 0.02 0.03 j IVI. PERSONNEL EXPENDITURES 0.05 0.05 0.05 0.05 0.06 0.06 IVII NET PROFITS j 0.05 0.03 0.01 0.03 0.04 0.06 I IMENO: CASH-FLOU NET PROFITS (a OUASI-FISCAL SURPLUS) (2) *0.08 -0.86 -1.10 -0.82 -0.56 0.02 I j(uVII-(I.A.1.a+I.A.1.e+I.A.2.b+I.A.3.a+1II.A.a+II.A.c+tI.A.f+IV.A)+V) I .... ........................................................** ....** ....I. *. *. ************. ****.*.***. ****. *.*.*.. I SOURCE: CENTRAL BANKS OF THE REPUBLIC OF TURKEY, ANNUAL REPORT (Various years); and CENTRAL BANK worksheets. (1) For 1987 the breakdown of items "Interest Paid" and "Interest Received" do not match with the total because the breakdown used was a provisional estimate, while the total was a definitive estimated reported in Central Bank's Annual Report. (2) For 1987 estimated the reported breakdown for "Interest Received* and "Interest Paid" was used and since the breakdown for item "Net Profits from Gold and Foreign Exchange Operations" was unavailable, all of III was deducted; and for 1988, when a similar breakdown was unavailable, "Cash-Flow Net Profits" was defined as "Net Profits" less (III1IVa-V). TABLE A.1.5: STRUCTURE OF PUBLIC SECTOR DEBT (as Percentage of GNP) (1) | 1979 1980 1981 1982 1983 1984 1985 1986 1987.1 1987.11 1987.III 1987.IV 1988.1 1988.!1 1988.111 1988.IV it. TOTAL DOMESTIC DEBT PLUS CENTRAL | BANK FINANCE (A + 8) (1) | II I A. Domestic Debt (1+2+3=4+5+6) j 9.21 7.49 4.54 4.36 5.02 2.60 4.42 4.26 4.47 4.31 6.79 6.38 6.41 7.49 8.06 7.55 I1 I Instruments I 1. Bonds and Bills I 3.70 3.65 3.29 3.32 3.05 3.78 4.65 5.19 5.30 5.28 6.56 5.81 5.43 5.71 5.89 5.80 2. Net Bank Credit I 5.51 3.98 2.18 3.26 1.95 0.50 1.07 1.41 1.03 0.81 1.91 1.95 1.78 2.52 2.85 2.33 | I 3. EBFs (1 0.00 -0.14 -0.93 -2.22 -1.98 -1.68 -1.30 -2.34 -1.86 -1.78 -1.68 -1.38 -0.80 -0.74 -0.68 -0.58 | I I I Borrowers J I I 4. Central Government j 2.97 2.78 2.07 3.45 2.22 2.97 4.23 4.18 4.90 4.44 6.15 5.45 5.16 6.31 5.17 5.57 I 5. SEEs j 6.23 4.84 3.40 3.13 2.78 1.32 1.48 2.43 1.42 1.65 2.31 2.31 2.05 1.92 3.56 2.55 6. EBFs (3) 0.00 -0.14 -0.93 -2.22 -1.98 -1.68 -1.30 -2.34 -1.86 -1.78 -1.68 -1.38 -0.80 -0.74 -0.68 -0.58 | AlI I I I . Central Bank Finance (2) I 6.38 4.16 4.59 6.44 5.39 8.21 7.63 7.09 6.03 6.32 5.92 5.69 5.32 5.59 6.45 5.98 II I 1I. TOTAL NET FOREIGN DEBT (4) I 17.43 20.53 18.55 18.81 20.78 23.56 24.51 26.91 N.A. N.A. N.A. 27.46 N.A. 27.18 27.53 27.71 (Exct. K Losses/Gains) I I ~II 1ita. TOTAL NET FOREIGN DEBT 12.13 20.53 21.86 24.61 29.22 33.32 36.55 45.67 N.A. N.A. N.A. 46.40 N.A. 45.12 47.26 46.76 (Incd. K Losses/Gains) I IMEMO: ACCUMULATED FOREIGN EXCH LOSSES (5)1 -5.30 0.00 3.31 5.80 8.44 9.76 12.04 18.76 N.A. N.A. N.A. 18.94 N.A. 17.94 19.73 19.05 SOURCE: Central Bank, Quarterly Bulletin; and Treasury. (1) Since stocks are measured at end-of-year prices white GNP is measured at mid-year prices, stocks of both domestic and foreign debt were converted to 1980 mid-year prices before computing its ratio to real GNP at 1980 prices. (2) This is the finance appropriated through monetary expansion adjusted for the subsidized credit extended to the private sector, i.e., rediscounts. (3) For the first three quarters of 1987 and 1988, it was assumed the net debt of end-1987 and end-1988 respectively. (4) This estimate for the stock of foreign debt excludes the capital losses (gains) caused by a real devaluation (appreciation) in the domestic currency and by changes in international exchange rates. For the estimates for the period 1980-86, the actual currency composition of the foreign debt was used and, for the remaining period the currency composition of end-1986 was used. (5) These are the foreign exchange losses accumulated as a result of fluctuations of international and domestic exchange rates. TABLE A.1.6: REAL CONSOLIDATED PUBLIC SECTOR DEFICIT (as Percentage of GNP) (1) | 1980 1981 1982 1983 1984 1985 1986 1987.1 1987.11 1987.111 1987.IV 1988.1 1988.11 1988.111 1988.IV ITOTAL REAL PS DEFICIT | 0.83 -1.06 5.32 1.65 9.69 6.41 5.39 N.A. N.A. N.A. 6.52 N.A. 2.76 4.43 4.13 | (Exct. K tosses of For. Debt) I I I IA. ADJUSTED MONEY FINANCE(2) I 0.67 1.79 3.47 0.85 5.74 2.57 1.74 0.26 0.80 0.75 1.82 0.73 1.34 2.78 3.40 I 1 I I 1. Inflation Tax [ 3.25 0.99 1.27 1.86 2.56 2.80 1.92 0.54 0.95 1.35 2.75 0.86 1.35 2.04 3.23 I 2. seignorage I -2.58 0.79 2.21 -1.00 3.18 -0.23 -0.18 -0.28 -0.14 -0.60 -0.93 -0.13 -0.01 0.74 0.17 I I I 1B. NET DOMESTIC DEBT FINANCE I -2.15 -2.39 0.17 -1.30 -0.22 1.92 0.06 0.67 0.42 2.86 2.39 -0.49 1.21 1.65 1.03 I I I 3. Bonds and Bills I -0.22 -0.09 0.29 -0.25 0.94 1.03 0.77 0.67 0.54 1.78 0.95 -0.12 0.00 0.06 -0.13 | I 4. Net Bank Credit I -1.78 -1.50 1.24 -1.28 -1.32 0.58 0.40 -0.23 -0.48 0.61 0.63 -0.90 0.60 0.89 0.33 | 5. E8Fs(3) I -0.14 -0.80 -1.36 0.23 0.16 0.31 -1.10 0.23 0.36 0.47 0.81 0.53 0.61 0.70 0.83 I I II IC. NET FOREIGN DEBT FINANCE(4) 2.30 -0.46 1.68 2.10 4.17 1.93 3.58 N.A. N.A. N.A. 2.31 N.A. 0.21 0.00 *0.30 I (Exct. K tosses of For. Debt) I I ID. NET FOREIGN DEBT FINANCE I 7.84 2.85 4.42 4.77 6.06 4.61 10.88 N.A. N.A. N.A. 3.71 N.A. -0.46 0.75 -0.56 I (Inct. K tosses of For. Debt) I I II IMEMO: FOREIGN EXCHANGE LOSSES (5) (0-C) 5.54 3.31 2.74 2.67 89 2.68 7.30 N.A. N.A. N.A. 1.40 N.A. -0.67 0.75 -0.26 1 SOURCE: Central Bank, Quarterly Bulletin; and Treasury. (1) Stocks were converted to average 1980 prices before calculating its ratio to GNP. (2) This is the portion of money finance which is used to finance the deficit, i.e., after deducting the portion returned to the private sector (see Table 5. for a breakdown). (3) For the first three quarters of 1987 and 1988 it was assumed that the change in net debt was equal to that of end-1987 and of end-1988 respectively. (4) This estimate excludes the capital loses (gains) caused by a real effective devaluation (apreciation) in the domestic currency and by changes in international exchange rates. For the estimates for the period 1980-86 the actual currency composition of the foreign debt was used and, for the remaining period the currency composition of end-of-1986 was used. (5) These are the losses incurred each year due to fluctuations in international exchange rates and to real devaluations in the domestic exchange rate. -75- estimate of the cash-flow generated. Most of the surplus recorded in the reported "Net Profits" is explained by the net income accrued but not received, such as the revaluation of gold. A.1.39 However, as mentioned in Chapter 1, this estimate of cash-flow Net Profits is under-estimated relatively to the below-the-line quasi-fiscal deficit because the expenditures on Central Bank's net foreign liabilities in the Profit and Loss Account are not fully captured. A way to illustrate this is by using our below-the-line quasi-fiscal definition, expressed as equation (A.1.17), (A.1.17) Q D = i + NiPL *E + YL- C g, by defining equation (A.1.5) (which is the below-the-line net worth identity) in terms of ', expressed as equation (A.1.5a), (A.1.5a) _=e + egI b - N L*E- NFL - Nis'""- 9L, and substituting equation (A.1.5a) into equation (A.1.17). After cancelling-out and ze-arrangin. similar terms our below-the-line can be defined as equation (A.1.17a), (A.1.17a) (FL)bI=(6g- NFL*L )+Cb ral where variables are as defined in Section A and the superscripts al and bl denote above- and below-the-line respectively. A.1.40 Equation (A.1.17a) defines our below-the-line QFD as the difference between the so-called "devaluation account" and the estimated capital losses due to fluctuations in domestic and international exchange rates -- the term in -76- parenthesis --, the loans granted to private banks -- which we assumed to be are negligible -- and the QFD estimated from the Profit and Loss Account -- i.e., QFDa'-NNWL'. It is apparent from our earlier discussion and from the above- the-line estimates reported in Table A.1.7 that the largest item accounting for rhe difference between the below- and above-the-line QFD is the term in parenthesis in equation (A.1.17a). However, a definitive conclusion would require more information on what is contained in the item C91 and on the Profit and Loss Account. -77- TABLE: A.1.7: BREAKDOWN OF NOMINAL FISCAL DEFICIT ABOVE-THE-LINE GOVERNMENT'S ESTIMATES (1) (As percentage of GNP) 1984 1985 1986 1987 1988 IPSBR (= A+B) I 5.4 6.6 6.6 10.1 6.8 I IA. NET LENDING (1) | -2.0 0.9 1.1 1.6 0.8 I I I 1 1B. TOTAL NOMINAL DEFICIT (=I+II+11+IV+V) j 7.4 5.7 5.5 8.5 6.0 I (1. CENTRAL BANK (2) I 0.9 1.1 0.8 0.6 0.0 I I I I |II. CENTRAL AND LOCAL GOVERNMENT (3) I 4.8 2.3 3.6 4.4 4.2 j I 2.1 FOREIGN FINANCE | 1.8 -0.8 0.3 -0.3 0.6 I 2.2 CENTRAL BANK NET ADVANCES I 1.0 1.0 0.7 0.6 0.7 | 2.3 DOMESTIC FINANCE I 2.0 2.1 2.6 4.1 2.9 2.3.1 BONDS AND BILLS (4) I 2.5 2.6 2.2 3.4 3.1 I 2.3.2 NET BANK'S CREDIT (5) I -0.5 -0.5 0.4 0.7 -0.2 I I I 1111. NON-FINANCIAL SEEs j 2.7 3.1 3.4 4.4 2.5 I I 3.1 FOREIGN FINANCE I 1.6 1.5 2.2 2.2 1.1 I 3.2 CENTRAL BANK NET ADVANCES I -0.3 0.3 0.0 1.0 0.1 I 3.3 DOMESTIC FINANCE (5) | 1.4 1.3 1.3 1.2 1.4 IIV. FINANCIAL SEE's *.4 0.0 -0.1 -0.2 0.0 V. EBFs -0.6 -0.8 -2.2 -0.7 -0.7 j 4.1 FOREIGN FINANCE 0.0 0.0 0.0 1.1 0.1 | 4.2 DOMESTIC FINANCE (5) -0.6 -0.8 -2.2 -1.8 -0.8 -........ ..................................... .......................................... SOURCE: Central Bank, Quarterly Bulletin; and Treasury. (1) These estimates were calculated by the Government from revenues and expenditures. (2) This is defined as the sum of NHFOs loans to the private sector plus Central Bank*s rediscounts granted to the private sector. For 1984, it was assumed that MHF's loans to the private sector were zero. This definition is different to that used by the IMF. (3) This is the above-the-line quasi-fiscal deficit estimated in Table A.5.4. (4) In addition to the Central and Local Governments, it also includes the Local Administration, Revolving Funds and Social Security Institutions. (5) This is an estimate of net new issues of Treasury bills and bonds (i.e., bills and bonds sales less repayment of principal), using the Treasury data. (6) It also includes changes in the cash balance position. (7) IMPs definition excludes this item, as it is intended to be a non-financial PS8R estimate. -78- TABLE A.1.8: BREAKDOWN OF NOMINAL FISCAL DEFICIT ESTIMATED THROUGH CHANGES IN NET TOTAL PUBLIC SECTOR LIABILITIES (As percentage of GNP) 1984 1985 1986 1987 1988 IPSBR (a A*B) 10.6 9.4 8.9 11.2 8.8 III IA. NET LENDING (1) *2.0 0.9 1.1 1.6 0.8 IB. TOTAL NOMINAL DEFICIT (=I+I1+11l+IV) 12.6 8.5 7.8 9.6 7.9 | (Exct K losses) FOREIGN FINANCE (2) j 6.1 2.7 4.9 2.9 -0.4 I DOMESTIC FINANCE j 0.9 3.2 1.2 4.9 4.9 BONDS AND BILLS I 2.2 2.5 2.1 3.4 3.1 NET BANKS CREDIT *0.8 0.9 0.7 1.4 1.5 E8F's -0.5 -0.2 -1.6 0.0 0.3 j MONEY FINANCE 5.7 2.6 1.7 1.8 3.4 I I I It. CENTRAL BANK (3) j 9.9 3.5 1.6 2.0 2.1 | 1.1 FOREIGN FINANCE (2) I 4.6 2.0 0.8 1.7 -0.6 1.2 MONEY FINANCE | 5.7 2.6 1.7 1.8 3.4 I 1.3 NET ADVANCES TO PUBLIC SECTOR -0.4 -1.0 -0.9 -1.5 -0.8 I 1.3.1 TO CENTRAL GOVERNMENT -1.9 -0.7 -0.7 -0.6 -0.4 I 1.3.2 TO PUBLIC ENTERPRISES | 1.5 *0.4 -0.2 -1.0 -0.3 I I I I 111. CENTRAL AND LOCAL GOVERNMENT I 5.1 3.9 4.7 5.2 3.6 | 2.1 FOREIGN FINANCE (2),(4) 1.4 0.6 2.8 0.9 0.1 | | 2.2 CENTRAL BANK NET ADVANCES | 1.9 0.7 0.7 0.6 0.4 | 2.3 DOMESTIC FINANCE | 1.9 2.7 1.2 3.7 3.1 | 2.3.1 BONDS AND BILLS | 2.2 2.5 2.1 3.4 3.1 j 2.3.2 NET BANK'S CREDIT j -0.4 0.1 -0.8 0.3 0.0 I I I 1111. SEEs | -1.9 1.3 2.9 2.1 1.9 I j 3.1 FOREIGN FINANCE (2),(4) | 0.1 0.2 1.2 0.1 0.1 I I 3.2 CENTRAL BANK NET ADVANCES j -1.5 0.4 0.2 1.0 0.3 I 3.3 DOMESTIC FINANCE I -0.5 0.7 1.5 1.1 1.5 I I 3.3.1 NET BANK'S CREDIT -0.5 0.7 1.5 1.1 1.5 I I I JIV. EBFs I -0.5 -0.2 -1.4 0.2 0.3 I 4.1 FOREIGN FINANCE (2),(4) j 0.0 0.0 0.2 0.2 0.1 4.2 DOMESTIC FINANCE -0.5 -0.2 -1.6 0.0 0.3 ------- -- - - - - - - - - - - - - - - - - - - - . . . . . . . . . . . . . . . . . . . . . . . . . SOURCE: Central Bank, Quarterly Bulletin; and Treasury. (1) This is defined as the sum of MHF0s toans to the private sector plus Central Bank's rediscounts granted to the private sector. For 1984, it was assumed that MHNF's loans to the private sector were zero. (2) This estimate excludes the capital tosses (gains) caused by a reat effective devaluation (appreciation) in the domestic currency and by changes in international exchange rates. (3) This is a below-the-Line estimate for the Central Bank's quasi-fiscal deficit. See Annex 5.1 for its definition and further discussion. (4) To estimate the foreign finance breakdown between General Government, SEEs and EBFs we have applied the share of the increase in foreign finance without correction for capital losses for each of these sectors, i.e., we have assumed the the currency composition of the General Government's. SEE's and EBF's is the same as that of the rest of the public sector i.e., total public sector's foreign debt less that of the Central Bank. -79- ANNEX 2 Demand for Assets in Turkey' A Preliminary Work A.2.01 The purpose of this annex is to define a simple framework to estimate the real interest elasticity of substitution of the demand for assets. In this sense, it should be regarded as an extension of the estimates done in a previous report.1 This annex will enable us: (i) to measure the sensitivity of real interest rates to public sector's financial policy; and (if.) to estimate the inflation rate at which the inflation-tax revenue is maximised. First, the model is dercribed; and second, the results are reported and discussed. Available data on assets and interest rates permitted us only to estimate a partial model, yet results on this exercise are useful. Further refinement of the model will be helpful to understand how exogeneous shocks (e.g., changes in monetary policy) are transmitted through the financial system and into private sector's savings and investment decisions. The Model A.2.02 In short, this model asserts that the desired demand for each asset relative to total wealth is a function of the vector of real rates of return and total wealth.2 This is equation (A.2.1). (A.2.1) .I * -=ci+ b IIR I+ di Ih'/ I/ See: The World Bank. Q, Cit., September, 1989. Volume II. 2/ See Brainard and Tobin(1968), and although they use income instead of wealth, we decided to specify it with wealth as a proxy for permanent income. Furthermore, empirical results favour the use of the latter. -80- where A,* is the desired holding of asset i, W is total wealth, R, is the real rate of return of the asset j and ci, b,, and d, are coefficients. Since by definition Z,"., -= 1. then the constraints ". b,, =O. Z.d,. and 7J.1c,= I must hold in the stationary state. A.2.03 There are two problems with this model. First, as stressed by Brainard and Tobin(1968) this is only valid for the stationary state; in real life savers do not always hold their desired portfolio, they take some time to adjust to the different variables that determine their demand for assets and it will be more reasonable to think that in the short run they are in disequilibrium. This calls for dynamic modelling. This problem is very important, for many empirical studies for developed and developing countries have used a model that assumes that investors are always in their long-run equilibrium, which is certainly not true.3 Furthermore, dynamic modelling implies that investors will not only respond to current and lagged values of the interest variables, but also to the extent that they are in disequillbrium in their desired portfolio. That is, demand for asset i at period t will be a function of the past values of their holdings of asset i; of their holdings of other assets; of the real rates of return on asset i (i.e., own real interest rate) and of the real rate of return of other asset holdings (i.e., cross real interest rates). Moreover, since adjustment to desired holdings 3/ For developing countries studies and a survey of them see, e.g. Gupta(1984). The only study that to our knowledge attempts a similar approach for a developing country is that made by Ortmeyer(1985) for Korea. For a more detail discussion of the model and probable shortcomings see Brainard and Tobin(1968) and Fried- man(1977). For an example of an empirical study using equation (A.2.1) see Taylor and Clements(1983). -81- is not immediate, current and lagged values of all variables should be included. More formally, this can be expressed as equation (A.2.2). This is a simultaneous stock adjustment system. (A.2.2) - z A :1 =t- I Zl(A *-A ) where Ai and Ak are as before and Zik are the coefficients of adjustment in relation to past holdings of the same and other assets, i.e. own and cross. If k-i, as in a non-simultaneous model then O<Zik<l. A.2.04 The problem with this model is that it is very difficult to estimate the simultaneous dynamic adjustment of the whole system because of the available computer programms and because it implcitly assumes the need for a large number of degrees of freedom, which are usually restricted as is our case. For these reasons we decided to estimate the demand function for each asset independently, i.e. to assume that k-i in equation (A.6.2). A.2.05 Second, information on assets, interest rates and wealth is usually limited, especially in developing countries. Moreover, in these countries unorganised markets are known to be important. Information on these is practically non-existent. There are two assets (or groups of assets) which are especially important: those traded in the unorganised market, and tangible assets, especially real estate acquisitions. The latter is a very attractive outlet for investors in countries with underdeveloped capital markets. A.2.06 For the econometric modelling the dynamic model of Davidsonet al(1978) was adopted.4 The dependent variable is the log of the ratio of the demand for asset i to total wealth, and the independent variables are: the real interest rates of all assets considered and the log of wealth. To the extent 4/ See Thorne (1986), Chapter 4 for a more detail discussion of this model. -82- that real interest rates can be negative - and they were - they cannot be expressed in logs. In the stationary state (i.e., when g-0) the solution of the dynamic model will be given by equation (A.2.3) below. (A.2.3) At1* c , cb +dto -= K, w\,here: K = e ' It/ where the variables are defined as above and lower case variables denotes the log of the variable. In the stationary state the ratio of asset i to total wealth is a constant and this will be given by the level of real interest rates and wealth. By definition W, -- I.. This seems to be a more adequate way to estimate the Brainard-Tobin model for a developing country, for it explicitly asserts that investors are in disequilibrium in the short-run and that only in the long-run do they achieve their desired portfolio. It makes use of both short and long-run information contained in time-series data. Empirical Results A.2.07 Turkish data on assets, interest rates and wealth is scarce. For this reasons we decided to estimate the model only for two assets, using three real intetest rates and using current income rather than permanent income. Further studies will be needed to refine these results. The assets used were currency holdings (i.e., non-interest bearing assets) and Turki--h lira (Tl) denominated interest bearing assets. The latter was defined to include demand and time deposits (i.e., M2 less currency holdings). The interest rates used were: the annual rate of inflation with a negative sign as the real rate of return on non-inte.act bearing T1 assets; a weighted average interest raLe on Tl denominated interest bearing assets (weights vary according to private sector holdings of each asset); and a weighted average interest rate on foreign currency denominated assets (weights vary according to foreign currency .83- composition of residential portion of foreign exchange deposits). It would ' had been desireable to include the rest of assets that comprise total private sector's wealth, which would had enable us to close the general equilibrium model, but data was not available. In particular, it would have been important to include the demand for bonds and bills and the interest rate on this debt instruments, as this would have enable us to measure more accurately the effect of the deficit in the private sector's demand for assets. Future extensions of the model can yield very useful results. A.2.08 Equations (A.2.4) and (A.2.5) report the best regression results using OLS (ordinary least squares) for the equations of demand for currency and of interest bearing assets in T1. The data used was quartely for the period 1981.3 to 1988.2. A.2.09 The long run or stationary-state coefficients are reported in Table A.2.1. The most important observations from the long-run coefficients results are the following. First, the strongest effects on the demand for assets are the effects of the own real rate of interest and of current income. In the case of the demand for currency, for instance, a decrease in the inflation rate by 10 percentage points will increase the ratio of currency to GNP by 2%. Similarly, an increase by 10 percentage points in the real interest rate on these assets will increase the ratio of interest bearing assets to GNP in 4%. -84- Equation (A.2.4) Equation (A.2.5) Dependent Variable: LN(C/GNP) Dependent Variable: LN(FA/GNP) Constant -1.37 LN(FA/GNP),-4 0.40 (-2.78) (9.01) (-pm) 0.22 (p,) 0.80 (1.46) (5.48) R -0.11 (-pe-) -0.57 (0.78) (-3.59) RfZ -0.33 R{" 0.24 (-5.16) (2.10) Ln(RGNP,) 1.31 R1x1 0.025 (2.68) (0.35) Ln(RGNP.,) -1.76 Ln(RGNP,) -0.24 (-3.57) (-9.36) R 0.80 T 0.79 DW 1.82 DW 1.47 SSR 0.084 SSR 0.047 where C is currency holdings in TL, FA is interest bearing assets denominated in Tl, GNP and RGNP are nominal and real GNP, R is the real annual interest rate using the last quarter annualized infaltion rate, p is the annual inflationr rate, superscripts fa and fx denote real interest of interest bearing asset%, in TI and in foreign exchange respectively, LN is the natural log, T is the statistic for goodness of fit for the equation adjusted for degrees of freedom. DW is the Durbin Watson statistic and SSR is the sum of squared residual. Table A.2.1 : Long-run Coefficients of Regression Results Variable -p Ria Ri^ Income Elas- ticitv 0.219 0.0 -0.33 -0.45 GNP 0.38 0.41 0.0 -0.4 GNP where variables are as before. The income elasticity is computed relatint i. to the dependent variable (i.e., - ; and 5"). The income elasticity : respect to C and FA can be calculated as 0.55 and 0.6 respectively. -85- A.2.10 Second, the closer substitute to hold currency are foreign exchange deposits (FEX). The coefficient of the real interest rate of this alternative asset is even higher than that of the negative of the rate of inflation. This suggests that a real devaluation of the Tl will encourage asset holders to switch to FEX even if the inflation rate do not increase. An increase in inflation will further reinforce this switch. To the extent that both effects have been in operation in Turkey in the last two years, they explain the fast decline of currency holdings and the rapid increase in FEX's. Notice also that neither the coefficient of the real interest rate on interest bearing assets inTl in the demand for currency holding's equation nor the coefficient of the real interest rate on FEX in the demand for interest bearing assets equation is not statistically significant different from cero. This reinforces the above conclusion that in Turkey FEX are a close substitute for currency. (It still remains to be seen the regression results for the other demand for assets held by the private sector.) The implication of this (which needs further analysis) is that strong real devaluations coupled with an increasing rate of inflation will make authorities lose control of monetary management (e.g., policies aimed at curbing inflation using monetary policy). Moreover, this also means that the base of the inflation-tax will shrink very fast, thus authorities to collect the same inflation tax will need to produce a higher inflation rate to compensate for the decline in currency holdings. This higher inflation rate, however, will further discourage private sector's holdings of currency. A.2.11 Third, income elasticity of demand for currency and of interest bearing assets in T1 is small. This is even if the income elasticity is computed on C and FA -- 0.55 and 0.6 respectively -- and not on their ratios to #NP. However, the interpretation of these coefficients is not clear because they were estimated using current income and not permanent income or wealth. It might well be the case that the relatively low coefficients are simply -86- indicating the high variance of current income relatively to permanent income. In any case the coefficients indicate that asset holders have a relatively low propensity to increase their financial assets holdings as their current income increases. The difference is allocated to the increase of other asset holdings (assets from outside the organised financial market and/or assets not modelled) and/or to the increase in consumption. A.2.12 The interpretation of a positive effect of the negative of the rate of inflation is somewhat striking as one would have expected a negative coefficient: the lower the inflation the greater the demand for cash-balances and, thus, the lower the demand for interest bearing assets. The fact that this coefficient is showing a positive effect might be indicating that at higher inflation rates the greater is the risk of holding long maturity assets. The reason for this is that asset holders are uncertain about what the realized inflation rate would be; and although past expereince provide a way to form expectations about its future level, it is always inaccurate and asset holders' losses can be large. Inflation Tax A.2.13 These results can be used to estimate what would be the maximum inflation tax for the given demand function for currency. To estimate the total inflation tax one would need to add the inflation tax collected on the other non-remunerated deposits at the Central Bank such as banks' reserve requirements. Moreover, to see the full implications of an increasing inflation rate (i.e., a decrease in the real rate of return on currency holdings) one would need of the other demand for assets functions (not modelled) to work out how the new equilibrium is attained once an exogeneous shock such as an increase in the inflation rate is experienced. In general, for a given wealth, a sudden increase in the inflation rate by encouraging asset holders to reduce their currency holdings will exert the strongest downward pressure on the -87- real interest rates of assets with which currency holding is a closer substitute. The strength will depend on the elasticity of substitution of currency holdings relative to inflation and the volume of resources freed. A.2.14 In Figure A.2.1 the inflation tax on currency holdings and the demand for currency holdings are plotted together. In this simulation it was assumed that only the inflation rate varies and that the other variables remain constant. While this is a simplifying assumption, it is clear that this is not a realistic assumption as real interest rates do vary with inflation -- i.e., the higher the inflation rate the greater asset holders uncertainty about future inflation rate levels - - and similarly with current and permanent income: production and investment is also affected by future uncertainty about relative prices. Furthermore, the currency demand function can also experience a structural shift, and this needs to be further analysed. The curve in Figure A.2.1 is useful only as an illustration of what would be the curve if the assumptions hold. A.2.15 This Figure shows that the inflation tax on currency holdings increases up to a maximum which is reached at an annual inflation rate of around 100-140 percent. Thereafter the inflation tax declines steadily. This same figure also shows the steady decline in the demand for currency as inflation rises. -88- FIGURE A.2.1 LAFFER CURVE OF INVLATION TAX On Currency Holdings 2.6 2.4 4. 2.2 2 1.8 1.6 1.4 0.6- 0.4 - 0.2 1 t I 20 40 60 80 100 140 180 200 250 300 350 400 4'30 b00 600 700 800 1000 Annual Inflation Rates in % 0 lit Tax + Dem for Curr References Brainard, W. and Tobin, J. "Pitfalls in Financial Model Building". American Economic Review (Papers and Proceedings) 58 (1968), 99-122. Davidson, J. E. H., Hendry, D. F., Srba, F. and Yeo, S. "Econometric Modelling of the Aggregate Time-Series Relationship Between Consumers' Expenditure and Income in the UK". Economic Journal 88 (1978), 661-692. -89- Friedman, B. "Financial Flow Variables and the Short-run Determination of Long-term Interest Rates". Journal of Political Economy 85, (1977), 661-689. Gupta, K. L. Finance and Economic Growth in Developing Countries. London: Croom-Helm, 1984. Ortmeyer, D. L. "A Portfolio Model of Korean Household Saving Behaviour, 1962-1976". Economic Development and Cultural Change 33 (1985), 575-599. Taylor, J. C. and Clements, K. W. "A Simple Portfolio Allocation Model of Financial Wealth". European Economic Review 23 (1983), 241-251. Thorne, A. E. The Determinants of Savings in a Developing Economy: The Case of Peru 1960-1984. Ph.D. thesis. Oxford: University of Oxford, 1986. -90- TABLE: A.2.2 INTEREST RATES OFFERED BY BANKS ON 1 YEAR FOREIGN EXCHANGE TIME DEPOSITS (FEX)(1) (In percentage) Name of Share in Share of Bank Overall FEXs in Total Liabi- lities 1986 1987 1988 Total FEX(2) of each Bank US $ DM US $ DM US $ DM (%) (%) Ziraat Bankasi 20.3 22 7.8 4.8 7.8 4.8 7.8 4.8 Akbank 10.1 24 7.8 4.8 7.8 4.8 7.8 4.8 Garanti Bankasi 4.5 32 8.8 5.2 8.8 5.8 9.0 6.3 Disbank 0.9 33 9.8 6.8 8.5 6.0 8.5 5.5 Sumerbank 1.1 22 11.0 9.0 9.0 7.0 8.0 7.0 Imar Bankasi 2.4 95 12.5 9.1 12.0 9.1 10.9 8.1 Memo: LIBOR(3) 7.0 4.6 7.6 4.2 8.2 4.2 SOURCES: Central Bank; IMF, International Financial Statisitics and Banks Association of Turkey. (1) These are the annual average interest rates, except for 1988 which is the average of the last 10 months. (2) Is the share of each bank FEX relatively to the total of the financial system. (3) US dollar deposits are annual average interest rates on 1 year maturity deposits and DM deposits are annual average on 6 month maturity deposits.