External debt, inflation and the public sector : towards fiscal policy for sustainable growth

THE WORLD BANK .2EP- co 2 2 Working Paper EUROPE, MIDDLE EAST AND NORTH AFRICA REGION ,P =@ Report No. EMNI Efternal Debt, Inflatin and the Public Sector: Towards Fiscal Policy for Sustainable Growth Sweder van Wijnbergen June 1988 Office of the Vice President Europe, Middle East and North Africa Region Working 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. EXTERNAL DEBT, INFLATION AND THE PULIC SECTOR: TOWARD FISCAL POLICY FOR SUSTAINABLE GROW BY SWEDER VAN WIJNBERGEN June 1988 The author is Principal Economist at the World Bank and member of the National Bureau of Economic Research. This paper was prepared for the senior policy seminar on Adjustment Policies in Izair, March 28-29, 1988. The paper draws heavily on joint work with several colleagues in the World Bank. In particular the author would like to thank Ajay Chhibber, Roberto Rocha, and, above all, Ritu Anand for many stimulating discussions, and Kemal Dervis for comments on an earlier draft. 1 Introduction The 1982-83 recession in the USA and the ensuing rise in interest rates and collapse in commodity prices triggered the debt crisis that has dominated macroeconomics in the developing countries ever since. While the almost exclusive focus on Latin America would lead one to believe otherwise, other regions have not escaped the problems such adverse shifts ia the world environment cause. A strategy to deal with external debt and the f,rml!lation of internal policies that allow sustainable growth within the limits of creditworthiness and macroeconomic stability are at the forefront in most developing countries. These issues are also the subject of this paper. The state is set by a brief overview of external debt developments since 1980. We use this overview to highlight what choices need to be made to arrive at an external debt strategy, and the trade-offs involved. At issue is, whether restrictive expenditure policies should be pursued to improve current account performance. To what extent will such policies come at the cost of future output growth, thus undermining the benefits of any debt reduction that does take place? Are there alternatives, that allow satisfactory output growth within the limits set by creditworthiness constraints? What is the role of exchange rate policy in all this? A real depreciation stimulates exports, an essential element of maintaining creditworthiness; on the other hand, real depreciation causes capital losses on external debt, thus raising the burden of any given level of debt in terms of the goods that need to be exported to service the debt. Sustainability of an external debt strategy and its social costs and economic benefits depend, to a large extent, on the internal policies that form the counterpart of any external adjustment undertaken. External -2- adjustment requires a ransfer to be made to foreigners (or adjustment to a lover transfer to be received from them); internal adjustment deals with the way the matching internal surplus of savings over investment is brought about. To this process, and the role the public sector can play in it, the paper turns next. The central question here is how to bring about the necessary surplus of savings over investment at levels of investment high enough to sustain output growth. An important part of any program of internal adjustment is the extent to which the public sector contributes directly towards the necessary improvement in the savings surplus. To this end, the fiscal deficit will typically need to be brought down. Any remaining deficit has to be financed by issue of domestic or foreign debt or by revenue from monetization. But macroeconomic target for inflation and output growth, in addition to the constraints on debt issue implied by continued creditworthiness and solvency, impose restrictions on each financing method; hence the issue of fiscal consistency. Do these targets and constraints allow the government to raise sufficient revenue to cover the deficit on which it has decided as part of its internal adjustment program? The absence of such consistency forebodes future policy change and so undermines the credibility of the fiscal program envisaged. The final part of the paper discusses the interactions between fiscal deficits and macroeconomic variables upon which fiscal consistency hinges. The paper draws on empirical work on Turkey to show the trade-off between fiscal policy adjustment and sustainable inflation. We discuss the impact of financial sector reform, economic growth and real exchange rate policies on this trade-off; how it is affected by the interest rates on foreign and domestic debt; and when and how postponing adjustment adversely affects the terms at which this trade-off takes place. -3- 2 A Brief Historical Overview: en Debt Output Growth and the Real Exchange Rate Debt-output ratios in the Mediterranean -region range from a low 24 percent in Algeria -tto a high of more than 100 percent in Morocco, both in 1986. The average for the region increased from 35% in 80/81 to almost 50% in 85/86. The median value of almost 60 percent in 1986 puts the region well into Latin American territory. This value is higher than the average for the group of 15 "high-debt" countries listed in the IMF's WEO. Thus, by current international standards, exte-nal debt is high in the EMENA region. However, it is important to see such measures in perspective. Countries like the US and the UK also relied extensively on external borrowing during corresponding periods in their economic history. Britain financed much of its industrial revolution in the early nineteenth century by borrowing from cash-rich Holland. With that process completed as the century progressed, Britain itself turned into a lender and financed much of the economic expansion in the USA and Argentina, at that time a dynamic economic power. The American move towards the West and the extension of Argentina's railroad system were financed by borrowing from abroad. It took to the middle of the current century for the US to reverse the tables and turn itself into the net lender it was until the deficit period of the last few years. JL/ Counties included are Morocco, Tunisia, Egypt, Algeria, Turkey, Pakistan, Portugal, Poland, Hungary and Yugoslavia. I will occasionally use the World Bank acronym "EMMA" to refer to this group. Calculated using the official exchange rate, which is severely overvalued. A more realistic exchange rate would lead to a much higher ratio. -4- The historical examples show that extensive debt accumulation occurred before; they also demonstrate that the borrower-lender cycles that are a part of this process often stretch themselves out over many decades. It has often taken that long for major borrowers to turn around and become lenders. From this vantage point, the current emphasis on short term solutions to what has become known as the "debt-crisis" may very well be unwarranted. An essential features of the two or three successful examples of external debt accumulation mentioned is that the high rate of foreign borrowing fuelled substantial investment and thus output growth. The high output growth and accompanying increases in productivity made it possible for the countries involved to eventually engage in extended lending rather than borrowing as time progressed and investment needs declined. This element is perhaps the most worrisome aspect of the current debt situation: in almost all debtor countries output growth has fallen to a post-war low. The fifteen high-debt countries listed in the IMF's WEO saw their output growth fall from over 5 percent per annum in the seventies to only 1 percent in the 1980s. EMENA is no exception to this rule: in oil importers and oil exporters alike, the growth rate fell by close to four percentage points, a fall that is even more pronounced if the one exception, Turkey, is excluded. The importance of high output growth is brought out clearly by a close look at Turkey's performance since its series of debt reschedulings in the late seventies (see Table 1). In Turkey, the ratio of gross debt to output has increased from 28 percent to 56 percent between the end of 1980 and 1986. On this account, Turkey has moved in line with the average for the 15 "high-debt" countries (see Table 1). In fact it is surprising that the debt-output ratio did not rise more rapidly in Turkey than it did in the "high-debt" countries: as a -5- percentage of GNP, Turkey ran a much lower non-interest current account surplus than the group of "high-debt" cour.tries did on average after their respective debt crises (-0.25 percent of GNP for Turkey over the period 1980-1986 versus 2.6 percent over 1982-1986 for the "high-debt" countries). This apparent inconsistency is explained by the much higher growth rate that Turkey managed to sustain since 1982. Turkey's debt-output ratio followed a path similar to that of the "high-debt" countries, not so much because of large surpluses, but because of its high output growth. This is where Turkey is most strikingly different from the "high-debt" countries. Figure 1 shows Turkey's growth rate since 1980 compared with the growth rate in the countries that the IMF classifies as having debt-servicing difficulties. Turkey's growth rate exceeds the real growth rate in the "high-debt" countries by 4 to 5 percentage points in almost every year since 1980. On average over these six years, the real growth rate in the Turkish economy has exceeded the average growth rate in the "high-debt" countries by no less than 4 percentage points. Higher growth reduces debt-output ratios as time goes by, or at least slows down their rate of increase. Against this process works the impact of interest payments on debt incurred in the past. Higher real interest ratzs increase debt-output ratios through accelerating debt-service costs. For any given net resource transfer, the debt-output ratio will increase (fall) further if real interest rates exceed (fall short) of the real growth rate of the economy. From this perspective, the world environment has turned distinctly unfavorable. Real interest rates were negative by any measure in the seventies, but have increased rapidly until 1985-86. Figure 2 shows the real interest rate on foreign debt for Turkey. The comparison has turned sharply negative. Even for Turkey, which grew at a much faster pace than the -6- Table 1: MEASURES OF THE OVERALL DEBT BURDEN 1980 1981 1982 1983 1984 1985 1986 Turkey: Debt (US$ billion) 16.3 16.9 17.6 18.2 20.8 25.5 32.5 Medium/long term 13.8 14.7 15.9 16.0 17.6 20.8 25.6 Short term 2.5 2.2 1.8 2.3 3.2 4.8 6.9 Debt/GNP 28.0 28.6 32.8 35.6 41.5 47.9 55.9 Debt/exports 284.1 198.3 175.0 192.9 130.5 194.5 260.5 Current Account Surplus/GNP -5.04 -2.83 -1.55 -3.57 -2.81 -1.90 -2.63 Non-Interest Current Account Surplus/GNP -3.89 -0.81 1.17 -0.36 0.36 1.39 1.04 Countries with Recent Debt-Servicing Problems: Debt/GDP 33.6 38.5 45.5 50.0 51.1 52.2 54.8 Debt/exports 151.2 185.8 241.5 254.3 247.2 263.9 302.4 Current Account Surplus/GDP -3.6 -5.9 -5.5 -2.0 -0.9 -0.5 -1.8 Non-Interest Current Account Surplus/GDP -0.5 -1.7 -0.5 2.8 4.1 4.2 2.5 Notes: For comparability the debt figures reported here for Turkey refer to gross debt. The debt-export ratio refers to year-end debt to exports of goods and services during the year. Countries with recent debt-servicing problems are defined as those which incurred external payment arrears in 1985 or rescheduled their debt during the period from end-1983 to end-1986. Source: Undersecretariat of Treasury and Foreign Trade, Central Bank and World Economic Outlook (IMF). -7- Fimue 1: REAL INTEREST RATE ON FOREIGN DE8T 9* 0 7* 8 S 22 £ 3 0 1 1 I -2 1ag tSAS sAt ISS3 194 Isgs s : Real Interest Rate Corrected fot Cross*Currency Effects Figure 2: REAL OUTPUT GROWTH. TUREY~w AND "HIGH 0Egl" COUNTRIES 6- 19058 S2383ae SStB a u-xY+ HIN0 7CUN4t -8- rest of the region, real interest rates no longer fall short of the real growth rate in the economy. Table 1 shows another striking feature where Turkey differs from most "high-debt" countries, including the EMENA region. In debtor countries across the world, the ratio of debt to exports rose in line with the debt-output ratio. On this measure, Turkey has been much more successful than the "high-debt" countries. Alone among the debtor countries, Turkey saw its debt-export ratio fall by a third after 1980, with not much deterioration afterwards. The ratio of exports (of goods and ion-factor-services) to GNP hovered between 5 and 7 percent of GNP between 1967 and 1980. The reform measures implemented since have caused a dramatic turn around. Exports jumped to 11 percent of GNP in 1890, up from 7.15 percent in 1979, and have been increasing as a share of GNP ever since. Exports reached 20.7 percent of GNP in 1985, then fell back to 18 percent in 1986 because of developments in the Middle East but have more than recovered in 1987. Exports are estimated to have grown by 30 percent in real terms in 1987. The net effect of this is that while the Turkish debt-output ratio has steadily deteriorated, the ratio of debt to exports, after a substantial improvement between 1980 and 1981, has by and large stayed constant since. Empirical analysis shows that the real depreciation effected since 1980 was a major contributing factor to the successful export drive. Without any real depreciation, exports would, the analysis suggests, have increased by a few percentage points of GNP only. This fact needs to be taken into account when assessing the impact of the real exchange rate on the debt. The counterpart of this real depreciation, however, has been a substantial capital loss on Turkey's external debt. This was a major contributing factor to the increase in the debt-output ratio; it accounts for - 9 - more than half of the increases in the debt-output ratio between 1980 and 1986. Empirical results show, however, that the debt-export ratio will in fact improve after a real devaluation: exports will increase enough in volume terms to offset the negative prite effect. Clearly, the debt-e,xport ratio would have been much more unfavorable without the depreciation that actually took place. A real devaluation causes a capital loss on foreign debt and thus a reduction in national wealth. Higher exports cannot undo this, but increased export orientation eases access to foreign capital markets. It is doubtful that Turkey would have had the access to external markets it did enjoy without the successful export performance generated by the reform program. The real depreciation of the exchange rate was an essential component of that program. One other country in EMENA, Morocco, has followed a somewhat similar strategy, although with a more moderate real exchange rate change than Turkey. Its non-traditional exports have increased accordingly, both in absolute terms and, in fact, with respect to Morocco's external debt. The effect on the aggregate debt-export indicator has been masked, however, by the substantial decline in phosphate earnings over the same period. Morocco's exchange rate policy has of course little bearing on the world price of phosphates and is unlikely to have had much of an impact on traditional exports for that reason. But in the absence of the policy of real devaluation, manufacturing expe-ts would not have grown as they have, with obvious negative impact on the debt-export ratio. 3 Towards the Formulation of an External Debt Strategy This brief survey suggests that for an analysis of external adjustment, three factors are of major importance. First, the non-interest - 10 - current account, as the most fundamental measure of the net resource transfers between a borrowing country and the rest of the world. Second, exchange rate developments, both between the borrowers and its trading partners (captured by the real exchange rate), and between the country's trading partners and creditors themselves (oross-currency exchange rates). Third, the way real interest rates paid on external debt interact with the growth rate of the economy to set the pace at which the dynamics of debt and output growth unfold over time. These are in fact the three factors to which an) increase in the debt-output ratio can be traced. I" The first term equals the non-interest current account deficit of the balance of payments. This is the most fundamental measures of a country's external (im)balance: it equals the difference between total expenditures (net of interest payments on foreign debt) and nationally generated income. Its counterpart is the net resource transfer the country receives from foreigners: the increase in debt minus interests payments made. If the non-interest current account is zero, the increase in debt exactly equals -' The decomposition is based on accounting identities. Define the debt-output ratio b* as: b* = (B*/P*).e/y; e= E.P*/P e is the real exchange rate, B* is the dollar value of foreign debt, P* the dollar-based, export-weighted price index of foreign goods, and E the nominal exchange rate of the local currency against the dollar. P is the local price index. Increases in the debt-output ratio can be traced to the following components: b* = -nicasy + (r* - n)b* + eb* a " " indicates changes and a " " percentage changes. nicasy is the ratio of the non-interest current account surplus to GNP. r* is the average real interest rate on foreign debt and n the real growth rate of GNP. - 11 - interest payments; the debt grows at the rate of interest in this case. Aslong as there is a surplus on the non-interest current account, foreign borrowing is less than interest payments to foreigners; or, to put it another way, the growth in foreign borrowing is less than the rate of interest. In that case, a net resource transfer to he rest of the world takes place. The opposite will happen when there is a deficit on the non-interest current account: in that case the debt will grow faster than the rate of interest. A debt growing faster than the rate of interest will eventually lead to insolvency. The second component captures what might be called an autonomous effect inherent in the mechanics of debt, real interest rates and output growth. If the non-interest current account is zero, the numerator of the debt-output ratio grows at the rate of interest; the denominator obviously grows at the (real) growth rate of the economy. Therefore, if the real interest rate exceeds (falls short of) the real growth rate of the economy, the debt-output ratio rises (falls) if the non-interest current account is zero. This term therefore measures the dynamics inherent in the interplay between real interest rates and real output growth. This is referred to as the debt dynamics component in this chapter. If real interest rates exceed the real growth rate by a substantial margin, the dynamics term will contribute significantly to increases in the debt-output ratio; the room for non-interest current account deficits will be limited accordingly. The final term measures the capital loss a country incurs on its external debt when the exchange rate depreciates in real terms. The debt-output ratio measures the debt in terms of home goods; if their relative value falls, as it does after a real depreciation, the debt-output ratio necesaarily rises. Against that must be set the favorable impact of the real - 12 - devaluation on exports, an important determinant of creditworthiness. We now turn to a discussion of the role these three factors play in different external debt strategies. In the current world environment, an external debt strategy consists of two choices. The first choice is between two alternative ways of restraining the ratio of external debt to GNP: (a) effect a net resource transfer to creditors through sufficiently large surpluses on the non-interest current account; (b) pursue a high output growth policy; high growth slows down the extent to which external debt feeds on itself through escalating debt service costs in an unstable manner. Option (a) is the one pursued by most Latin-American and Eastern European debtor countries since 1981/1982. The problem with this approach is vividly demonstrated by their experience. The only reliable and practically implementable way of bringing about a surplus on the non-interest current account is through substantial cuts in expenditure. This may, however, cause substantial loss of output. One reason is the potential short-run recessionary impact of expenditure cutbacks. A more fundamental cause of output losses arises because the expenditure cutbscks are likely to come out of investment, thus slowing down output growth. But this opens up the possibility that gains made through improvements in the non-interest current account are offset by a widening excess of real interest rates over real output growth rates. Effectively, what is gained on the numerator is at least partially lost again because of a slowdown in the rate of increase in the denominator of the debt-output ratio. This is what happened in most of the "high-debt" countries. While Turkey did make substantial external adjustment, it did not take this route and thus avoided the destabilizing spiral in which ost other "high-debt" countries seem to be trapped. - 13 - Option (b) relies on a policy geared towards high output growth, to slow down the dynamic process of debt feeding on itself through escalating debt service costs as a share of GNP. The main problem with a low-trade-surplus/high growth strategy is that the Government needs to make sure that the extra expenditure the lower trade surplus allows is indeed channeled into productive, trade-oriented capital accumulation. Even if this is done, either through increased public sector investment or thorough incentives for private investment or both, the strategy could fail because of a potential clash with the export drive that, we will argue, should also be part of a successful external debt strategy. Higher investment expenditure will invariably increase aggregate demand for home goods and put upward pressure on the real exchange rate. The growth strategy would then crowd-out exports and jeopardize creditworthiness by diverting production away from traded goods. The only way out is an active attempt to create room for exports by restraining public and private consumption. This would also alleviate any pressure on imports and the trade balance that could result if no such consumption restraint would accompany the increased investment expenditure. All these issues really concern internal adjustment problems, to which we turn in Section 5. The discussion of the pros and cons of each option also demonstrates that they are, in practice, mutually exclusive. Running high surpluses on non-interest current account will typically lead to slower growth, as investment falls. As a consequence, the debt-dynamics term increases as t:. growth rate falls below the real interest rate on external debt. Conversely, higher growth and the investment expenditure it requires is almost certainly going to require continued net resource transfers from abroad. - 14 - The second choice concerns the role of the real exchange rate. A real depreciation raises the debt-output ratio but lowers the ratio of debt to exports. Should a country opt for real depreciation and export orientation and simply accept the associated losses on external debt? Is there an alternative, involving less exchange rate depreciation? A real appreciation lowers the ratio of external debt to output by lowering the relative price of foreign goods (in which the foreign debt is expressed), in terms of home goods (of which GNP is made up, by definition). However, a steady real appreciation implies a steady increase in the relative price of home goods, which would, in the absence of policy changes, induce an increasing excess supply of them. The only way this can be avoided is to raise the one component of demand for home goods that is likely to be both least price-sensitive and under control of policy makers, Government expenditure. Such a strategy would see Gov ernment expenditure rise and exports fall as time goes by. In addition, domestic consumers would increasingly shift from more expensive some goods to foreign goods. As a consequence such a strategy is likely to lead to a deteriorating trade balance, which would eventually force its abandonment. The anticipation of such events is what is behind the exchange rate crises that have characterized many Latin American countries over the past few years. Turkey has in fact followed the opposite strategy: its concerted export drive has been based on a commitment to an exchange rate strategy designed to maintain or steadily improve Turkey's external competitiveness. This requires real depreciation of the exchange rate. Empirical evidence widely supports the view that such a strategy is essential to maintain creditworthiness. Commercial credit ratings invariably put a great deal of emphasis on the degree of expo:t orientation in the economy. The conclusion seems clear: there is little alternative to an - 15 - exchange rate based export orientation as an esential component of an external debt strategy, the associated capital losses on external debt notwithstanding. 4 A Digession the Implications of Cross-Currency Exchange Rate Fluctuations Cross-currency exchange rate fluctuations cause a major problem for the interpretation of year-to-year changes in the dollar value of external debt, For example, a substantial and rising part of Turkish externil debt is denominated in hard currencies such as the DM, the Yen and the Swiss Franc. The share of these three currencies in the total external debt (measured net of foreign ssets of the banking system) rose from 22 percent in 1980 to 40 percent in 1986. The ratios are rather similar when evaluated at fixed cross-currency rates: 22 percent in 1980 versus 38 percent in 1986. This reflects the fact that by 1986 the dollar had by and large lost all the ground it gained between 1980 and 1984 against the DN and the Yen. In 1984, however, the share of the DM, Yen and Swiss Franc was 31 percent when evaluated at actual rates, with the share at constant rates at a high 36 percent. With such a large component of the debt denominated in non-dollar currencies, cross-currency rate fluctuations can obviously cause substantial fluctuations in the total (dollar) value of the debt. The capital loss in 1986 was a hefty 6.4 percent of GNP. However, to a large extent these losses simply offset capital gains made earlier, during the years the dollar appreciated against the DM, Yen and Swiss Franc. Interpretation of such fluctuations should be set against the likelihood of their recurrence in the future. A once-off loss is deplorable since it reduces the borrower's wealth, but does not call for as much of a policy change as a loss that is likely to happen again in the future. Such a - 16 - policy response could include expenditure reduction, in line with downward revisions of the country's net wealth, and attempts to restructure the currency composition of the-debt. Both are more important for capital losses that are likely to occur again, in whatever direction, than in the case of a once-off event. Figure 3 shows the extent to which this has been taking place. Figure 3A shows the exchange rate of the dollar against a basket made up of the Dm, the Japanese Yen and the Swiss Franc, with weights proportional to their respective weights in Turkish external debt at the end of 1980. !' The figure shows how almost the entire rise in the dollar between 1980 and 1984 was offset by the fall in the dollar in 1985 alone. Figure 3B plots two measures of Turkey's external debt (public and private sector combined, both net of foreign assets), expressed as a share of real GNP. The lower one, labeled b*, is the actual debt, converted into "Turkish goods" (constant TL) by multiplication with the end of period real exchange rate and subsequently expressed as a share of real GNP. 1' The second measure, b*, is similar, but converts all non-dollar debt into dollars at 1980 exchange rates. This procedure eliminates the impact of cross-exchange rates. This procedure eliminates the impact of cross-currency exchange rate changes on the dollar value of the debt. ' The weights are rescaled so that they add up to one. i.e. b*=(B*/P*).e/y = B*E/Py. e is the real, and E the nominal exchange rate (E in terms of the dollar and e in terms of an export weighted basket of foreign competitors' prices). P is the Turkish price index, for which we used the CPI. y is real GNP. P* is a dollar-based, trade-weighted index of foreign prices. It is a geometric average of the dollar-based WPI indices in the six major export markets outside the Middle East, with their respective (1980) share in Turkish exports as weights (rescaled so that they add up to one). - 17 - The graph demonstrates two things. First, the two measures may give very different answers to the question of how much Turkish debt increased in any given year. For example, during 1986 the actual ratio of debt against GNP increased by almost 10 percentage points. However, once one eliminates the impact of cross-currency fluctuations, the increase was only 3.5 percentage points of GNP. The second point is that over the entire period under consideration, 1980-1986, the difference between the two measures is actually quite small. The two measures of debt are equal in 1980 by definition; after that, until late 1984 the actual value increased more slowly under the impact of the rapid appreciation of the dollar. Thereafter the difference goes in the other direction, as the dollar began its slide. By the end of 1986, the two measures are once again close, at 51.1 percent for the corrected measure and 51.4 percent of GNP for the actual value inclusive of exchange rate losses due to cross-currency fluctuations. This implies that it does not matter for the outcome of the assessment of the adjustment effort over the whole period which debt indicator is used; but by eliminating once-off capital losses due to cross-currency exchange rate fluctuations, one probably gets a better view on the year-to-year adjustment effort. The discussion so far has focused on the effects of cross-currency rates on the dollar value of the debt. There are, however, also other channels through which countries are exposed to cross currency exchange rate risks. Composition of production and expenditure is rarely matched in any country, even if the aggregate levels are close. The differences are made up through international trade. As a consequence, the net trade position with any individual country is often unbalanced. This in turn implies that relative price changes between trading partners may have substantial income ** 18 - liure 3: u A: WEIGHTED CROSS-CURRENCY EXCHANGE RATE 30 (US$ v. M. V md VM) 125* 120* Its* 5- 95 - *0 *n0 1 1 1I2 1I3 1934 135 Ing Note: 1 currency coaposition of the debt at the erd of 1980 was used to derive the weights. Fivre IB: CROSS-CURRENCY EFFECT ON FOREIGN DEBT (Nt de ae o f O01P) 32 - SO 48 - 44 42 40 38 -. 36- it 34- 32- 30- 28- 26- 24 - 22 J 20 I28 1nS 1982 19W3 1984 Ins InIl 0 ACTU&L 0t8 ) + O1v CORR. FOR cC (*) - 19 - effects on the debtor country under consideration. Moreover, empirical work on exchange rates and prices widely supports the view that relative prices are influenced by nominal exchange rate fluctuations. Exposure to exchange risk raises the question whether active management of the currency composition of the external debt could be beneficial in reducing exchange risk exposure. What is the optimal currency composition of a country's external debt (net of reserves) taking into account exchange risk exposure through commodity trade and the exposure arising directly out of the currency composition of the debt itself? To what extent can the composition of debt be altered to achieve insurance against terms of trade shocks? It is easy to tell, with the benefit of hindsight, what a better policy would have been. Exposures to exchange rates could have been hedged by matching in the right fashion the currency composition of external liabilities with the effective currency composition of cash flows. However, such hindsight does not provide practical policy rules on how a country, given its external debt situation, can hedge the economy against future exchange movements. This is an area of active research, where practically useful results can be anticipated in the months to come. Intuition already suggests the directions in which they are likely to go. A country with a large bilateral trade deficit against say Germany should probably avoid having a major share of its external debt denominated in DM. If such a country would maintain a large DM ahare in its debt, an appreciation of the DM would hurt the country twice: its commodity terms of trade would deteriorate at the same time that it incurs capital losses on its external debt. More sophisticated results are being devised, but a basic rule seems clear: there should not be a close match between the structure of a - 20 - borrower's bilateral trade deficits and between the currency composition of its debt. In fact in some cases the same line of reasoning can be applied to individual commodities instead of trading partners. Take the example of Algeria. One major commodity dominates export trade: oil and oil based products. Clearly, Algeria's terms of trade are completely dominated by the price of oil. A country in Algeria's position should give serious consideration to issue debt indexed to the price of oil. Such oil-indexed bonds have in fact been floated by countries like Norway, and are being studied for Mexico. The advantage is clear: whenever the terms of trade would turn against Algeria, its debt burden would fall. In effect Algeria would be buying partial terms of trade insurance. Of course insurance comes at a cost; the average cost of such debt is likely to be somewhat higher than non-indexed debt. The difference constitutes what amounts to an insurance premium. 5 Solvency, Creditworthiness and Foreign Debt: What is a Sustainable Current Account Deficit? - Assume that the borrowing country has considered the evidence and decided that the Latin American and Eastern Europe example is not particularly attractive. It thus decides to take option (b): pursue a high growth strategy within the limits set by solvency and creditworthiness constraints. The formulation of such a strategy would start with a decision on what exactly constitutes a sustainable current account deficit; this strategy calls for borrowing up to that limit and internal adjustment policies that make sure that the additional borrowing is translated into investment and output growth. In this Section we discuss one approach to sustainable borrowing. The next Section covers the matching internal adjustment programs. - 21 - Assessing a country's room for.external borrowing involves two considerations: solvency and creditworthiness. Solvency concerns ability to pay and is intricately linked to the non-interest current account, real interest and output growth rates, and, finally, the initial level of debt. To remain solvent, a country should not plan expenditures higher than its current and future income (discounted) minus its initial debt. This implies that the non-interest current account surplus should at least equal the initial debt times the difference between the real interest cost of foreign debt and the real output growth rate. A A number of important consequences follow from this. First, a country with a higher income or a country with a lower debt, can borrow more than a country with a higher debt-output ratio. Second, the more expensive a country's external debt, the higher its surplus o the non-interest current account should be, if it is to maintain solvency. Third, the higher a ' The current discounted value of income less expenditure equals (Y-C-I)/(r*-n), if real interest rates and growth rates are constant. Y is national income before foreign interest payments; C and I aggregate consumption and investment expenditure; r* the average real interest rate on foreign debt; and n the real growth rate of the economy. Y-C-I equals the non-interest current account surplus. If this expression should not fall short of the initial debt, the following must hold: or (2) NICA > (r*-n)B* Expressing NICA and B* as shares of GNP and indicating them by lower case letters gives the expression discussed in the text: (3) nica > (r*-n)b* Strictly speaking, this formula is only valid if output growth rates and the real interest rate are likely to remain roughly constant. -22 - country's growth rate, the more leeway it has in borrowing without jeopardizing solvency. The latter conclusion reflects the vicious circle so many debtor countries face: slow growth implies less room for external borrowing, which in turn causes slower growth and so on. For most countries, solvency constraints are not very restrictive. Turkey's ratio of net foreign debt to GNP equals 51 percent. Even if the average real interest rate on its external debt remains as high as 8 percent, solvency would only require a surplus on the non-interest current account of one percent of GNP for a real output growth rate of 6 percent. On the plausible assumption of an average real interest on foreign debt of 6 percent, this would imply a lower limit of zero on the non-interest current account. A continued deficit on the non-interest current account. A continued deficit on the non-interest current account would eventually jeopardize solvency at current levels of interest rates and projected output growth rates. However, solvency is not the only consideration. Ability to pay does not necessarily imply willingness to repay. Creditworthiness (which depends on lenders' assessment of a country's ability and willingness to repay) therefore often imposes tighter constraints than solvency alone. Repayment requires not only a sufficiently high value of wealth to be able to repay, but also the generation of a surplus of traded goods production over traded goods consumption (net exports). This is likely to be much more burdensome in a country with most of its resources in non-traded goods sectors than in an outward-oriented country. But if it is more burdensome, a country might be more tempted not to repay, even if solvency requirements are met. Hence the importance of debt-export ratios in the assessment of creditworthiness. Assessing the precise limits imposed by creditworthiness constraints is difficult for several reasons. First of all, while debt-export ratios are - 23 - important, they are a biased estimate of the ratio of a country's debt to its output of tradable goods. Some domestically produced tradables are likely to be sold at home rather than exported. So the true measure lies somewhere between the debt-output ratio (which also counts non-tradables) and the debt-export ratio, which excludes tradable goods produced and sold at home. In a recent study completed for Turkey we followed an approach pioneered by the French economist Daniel Cohen (1985, 1987). This approach chooses the ratio in between the debt-output (D/Y) and the debt-export (D/X) ratios in such a way that there are no incentives to overvalue or undervalue the exchange rate simply to mechanically improve creditworthiness indicators. The precije way in which this ratio is derived is presented in that report; it is influenced by the price elasticity of export demand and output supply. The outcome for Turkey places a 60 percent weight on debt-export ratio and a 40 percent weight on the ratio of debt to GNP. This construct is referred to as the debt-resource ratio, D/R. A second, more fundamental problem, involves not so much the choice of any particular creditworthiness indicator, but how to assess whether the value of the indicator chosen is too high or not (high values indicate low creditworthiness). An indicator is too high (creditworthiness too low) if at that value the burden of servicing the debt exceeds the likely penalty on non-compliance to repayment terms. The problem with this definition is that nobody really knows how high that penalty is. We have followed Cohen in a very simple but forceful approach to this issue. The cost of default is not known, but if a country has not defaulted at the current value of its debt-resource ratio, that value is, by implication, not yet too high. Otherwise the country would have defaulted already. A cautious borrowing policy then is a policy that will prevent a rising debt-resource ratio. - 24 - One important caveat: it does not follow from this analysis that a borrowing policy designed to rapidly lower debt-resource ratios is necessarily a good idea. While it is true that lower debt-resource ratios indicate higher creditworthiness, the transitional costs of reaching that lower ratio clearly raise the cost of servicing the existing debt. Since creditworthiness involves comparing the cost of default with the cost of servicing the current debt, such a strategy, which has been imposed on many high-debt countries, would lower rather than increase current creditworthiness. How much foreign borrowing is compatible wita maintaining the debt-resource ratio at its current value, and hence maintaining the level of creditworthiness? Since the debt-resource ratio is a weighted average of the debt-output and the debt-export ratio, it will depend on the growth rate of the borrowing country and of its trading partners. The growth rate of its trading partners is one of the determinants of a country's likely export growth. The other determinant is the elasticity of demand for the borrowing country's exports with respect to income in the countries to which it exports. Consider the following example for Turkey. Empirical analysis suggests that the income elasticity of demand for Turkey's exports is high: 1.6 with respect to the OECD and 4 with respect to the oil-exporting countries in the Gulf region. This results in a weighted value of 2. Thus, if the weighted output in Turkey's trading partners grows by 4 percent, Turkey's exports are likely to grow by 8 percent. The results are presented in Table 2. The table gives the maximum increase in foreign debt that will avoid a rising debt-resource ratio, for Z Weighted by their share in Turkey's exports. - 25 - Table 2: ALLOWABLE FOREIGN BORROWING: SUSTAINABLE CURRENT ACCOUNT DEFICITS (percent of GNP) .... .... ... .. ............ *4* * ** *~* * *# ~ ** 9* *O ***@ ** Output Output Growth Rates of Turkey's Trading Partners Growth ................... .............................................. of Turkey 0 1 2 3 3.5 4 3 1.12 1.21 1.29 1.38 1.42 1.46 4 1.49 1.58 1.66 1.75 1.79 1.83 5 1.87 1.95 2.04 2.12 2.17 2.21 6 2.24 2.33 2.41 2.50 2.54 2.58 7 2.61 2.70 2.78 2.87 2.91 2.95 different growth rates at home and abroad. The table lists increases in debt and hence gives the feasible current account deficit. The table lists on ita vertical axis various alternative growth rates for Turkey, ranging from 3 to 7 percent. On the horizontal top axis, it lists potential growth rates for Turkey's trading partners, aggregated using their respective shares in Turkey's exports. The numbers indicate, as expected, that lower growth rates, whether at home or abroad, allow for less debt accumulation. In fact for zero growth rate at home and abroad, the formula indicates that no further borrowing is possible (this possibility is outside the range of the table). Raising the domestic output growth rate by 4 percentage points allows an extra current account deficit of 1.5 percent of GNP for given foreign output growth rate. A slump abroad lowers borrowing potential: if growth in trading partner countries falls from, say, 4 percent to zero, the amount of feasible debt accumulation goes down by 0.3 percentage points of GNP. 6 Internal Adjustment: Towards a Consistent Fiscal Poley Once the feasible current account deficit has been decided upon, a matching internal adjustment program needs to be set up. An internal - 26 - adjustment program consists of a set of policies that will bring about a fiscal deficit and a private savings surplus over investment just enough to match the external current account target. The challenge is to design this package in such a way that total investment, private and public, will be high enough to allow output to grow at this target rate. This involves once again a two stage design. First how much should the public sector contribute to the required improvement in the surplus of aggregate savings over investment? The issue here simply is, how much should the fiscal deficit be cut back. Onca this has become clear, the difference between the targets for fiscal deficits and external balance need to be made up by the private net savings surplus. The policy instruments that have the most influence on this are interest rates and tax policy. Now they should be used in this context is the subject of the next Section. In this Section we discuss the fist question, how large should the deficit be. In the long run, the size of government needs to be determined on the basis of views on the role of the public sector in the economy and the distortionary costs of raising the revenue necessary to finance the associated expenditure. Such considerations are however of little help for the medium run focus that is appropriate for the issues discussed here. Instead we suggest a more modest approach. This approach starts from the assumption that the government has certain target values for such variables as inflation, output growth and so on. In addition there are the constraints imposed by sustainability of the current account deficit, as we saw in the previous Section. Similar considerations play in the analysis of the domestic debt issue. Such considerations imply restrictions on the feasible public sector deficit, as we will argue below. Consistency with other macroeconomic targets provides - 27 - policy makers with an answer to the question: how large should the deficit be? Optimality of fiscal deficits is a more complicated target, satisfaction of which needs entirely unavailable data. Consistency with other stated macroeconomic targets, however, is much easier to assess, and is anyhow a sensible requirement. Absence of consistency is a clear signal that one policy or another will need to be changed in the future; the government surely does not want its hand forced by private speculators acting on such signals. 6.1 Consistency of Fiseal Policy Consistency analysis starts from the mundane observation that there are three sources of financing public sector expenditure beyond what can be obtained from the regular tax system: external borrowing, monetization and issue of domestic interest-bearing debt. The amount that can be expected from each source will depend on other macroeconomic targets, such as inflation, output growth, interest rates and so on. The revenue from these three sources of financing can be combined into the calculation of a "financeable deficit". This is defined as the deficit that does not require more financing than is compatible with sustainable external borrowing, existing targets for inflation and output growth, and a sustainable internal debt policy. A/ Underlying the framework suggested here to calculate the financeable deficit is a model describing private portfolio choice as a function of / A simple version of this framework was first used in the report "Fiscal Policy and Tax Reform in Turkey" and is described in Anand and van Wijnbergen (1987). The current version incorporates external debt considerations and implications of the financial structure for inflation tax revenues. It is presented in detail in the report on Turkey mentioned before and more fully in van Wijnbergen, Anand and Rocha (1988). - 28 - inflation, output and interest rates. This gives the amount of currency, demand deposits and time deposits the private sector is willing to hold given output, inflation and the level and structure of interest rates. This is coupled with a simple financial sector model incorporating reserve requirements and other bank regulatory policies to derive the demand for reserves by commercial banks. The demand for reserves is then added to the demand for currency already derived to get an estimate of the total demand for base money given inflation, interest rates, and so on. All this is used to derive total revenue from monetization for different output growth rates, interest and inflation rates and regulatory policies. The amount of revenue that can be collected through monetization depends critically on inflation and financial structure. Inflation influences both components of base money. Righer inflation will reduce demand for cash balances. It also changes the amounts the private sector is willing to hold as demand and time deposits. It will thus influence both the private sector's demand for currency and the commercial bants' demand for reserves. Both influence aggregate demand for base money directly. The structure of the banking system, the particular regulatory framework in which it operates and the interest rates on their deposits determine the level of reserves banks need to keep for any given inflation rate. Bank reserves, in turn, are one of the two components of reserve money, the basis over which the inflation tax!is levied. Thus the regulatory framework within which the banking system needs to operate has an important impact on the amount of revenue the government can expect from monetization. To revenue from monetization must be added the revenue the government can expect from external and internal debt issue given its external borrowing policies and debt management approach. This was discussed in Section 5. The - 29 - results of such an exercise for Turkey are summarized in Tables 3 and 4. Underlying these Tables are various targets and assumptions: a real growth rate of 6% a year is the most important one. We furthermore assumed the current values for reserve requirements and nominal interest rates on demand and time deposits. As to liquidity requirements, only the part over which no interest. is paid is incorporated; the remainder is included in the definition of interest-bearing public sector debt held by the banking system. The public sector can expect slightly in excess of 3% of GNP from issue of internal aiQ external debt, if sustainability and creditworthiness constraints are to be met. Table 3 first assesses potential revenues from seignorage and the inflation tax for various inflation rates. Listed are demand for currency, demand deposits and time deposits as a function of interest rates and so on for various inflation rates. It then calculates revenue from inflation tax and seignorage and adds the two to arrive at total revenue from monetization. The table shows a number of things. 1" First, both components of base money are very sensitive to inflation. As inflation rises from 15% to say 60%, demand for currency falls from 3 percent of GNP down to 2.5%. Demand for deposits goes down more (note that nominal interest rates are kept fixed by assumption in this Tablo). The combined total of demand and time deposits falls from a predicted 25% of GNP at 15 percent inflation down to 18% at an inflation rate of sixty percent. Not surprisingly, total demand for base money, listed in the column MB, also falls: from 7.8% of GNP at 15 percent 1 Measuring asset-stock-to-GNP ratios and revenue from the inflation tax involves a number of intricate corrections for differences between beginning-of-period and average price levels and so on. - 30 - Table 3: INFLATION TAX AND SEIGNORAGE AT VARIOUS INFLATION RATES (percent of GNP) Revenue Inflation Demand Time Base Inflation Tax from Rate Currency Deposits Deposits Money Revenue Monetization 15 3.0 7.5 17.5 6.8 1.0 1.4 20 2.9 7.3 16.7 6.6 1.2 1.6 25 2.9 7.1 16.0 6.4 1.4 1.8 30 2.8 6.9 15.3 6.2 1.6 2.0 35 2.7 6.7 14.7 6.0 1.8 2.2 40 2.7 6.5 14.1 5.8 2.0 2.3 45 2.6 6.3 13.5 5.7 2.1 2.5 50 2.6 6.1 13.0 5.5 2.2 2.6 55 2.5 6.0 12.5 5.4 2.4 2.7 60 2.5 5.8 12.1 5.2 2.5 2.8 Table 4: FINANCEABLE DEFICIT AT VARIOUS INFLATION TARGETS (percent of GNP) Inflation Financeable Actual Required Deficit Rate Deficit Deficit Reduction in 1986 15 4.4 5.7 1.3 20 4.6 5.7 1.1 25 4.8 5.7 0.9 30 5.0 5.7 0.7 35 5.2 5.7 0.5 40 5.3 5.7 0.4 45 5.5 5.7 0.2 50 5.6 5.7 0.1 55 5.7 5.7 0.0 60 5.8 5.7 -0.1 - 31 - inflation down to 5.92 at 60 percent inflation. It is clear from the next column that higher inflation leads to higher revenue from inflation tax: it goes up from 1% of GNP at 15 percent inflation to 2.8% at an inflation rate of 60 percent. Total revenue from monetization also rises but at a slightly lower rate, because the other component, seignorage, actually declines as inflation rises. This is a negligible effect, however. There are two important conclusions from these results. First, inflation tax revenue goes up with higher inflation, but less than proportionally. For any given ratio of base money, higher inflation increases revenue from inflation tax one for one; however, as inflation rises, the ratio of base money to GNP falls, thus eroding the base over which the inflation tax is levied. Over the range of inflation rates that are relevant for Turkey, the latter effect will not dominate: the Table shows that revenue from inflation tax goes up with inflation for the inflation rates shown. Inflation tax revenues only start falling with rising inflation for inflation rates well above 200 percent a year. The second point is the reason we include this analysis here to begin with: inflation clearly has an important impact on the revenue the authorities can expect from monetization, and hence on the financeable deficit. Table 4 adds up the revenue the government can expect from external borrowing subject to the constraints outlined in Section 5, from monetization for different inflation targets, and from the issue of interest-bearing domestic debt. The total is called the financeable deficit: a deficit of that size is sustainable and will not compromise any of the macroeconomic targets mentioned. One additional assumption needs to be mentioned. In the calculations underlying the Table, it is assumed that issue of interest-bearing domestic debt is kept down to a rate that will maintain the - 32 - ratio of domestic debt to total GNP. The reason for not allowing a faster rate of domestic debt issue is the high interest ate it currently carries; at 121 a year, it is well above the real growth rate of the economy. At this rate, debt-service will escalate as a percentage of GNP if more extensive use is made of debt-issue to finance the deficit; we will explore this at greater length in the next Section. The Table shows, first of all, the financeable deficit as a function of the inflation rate. A target of 50%, close to the 1987 inflation rate on a year-end-to-year-end basis, allows a deficit of 6% of GNP; an inflation target of 20 percent would allow only 4.9% of GNP. If the financeable deficit is subtracted from the actual deficit, one obtains the cut in the deficit necessary to achieve macroeconomic consistency (the column labeled RDR, for Required Deficit Reduction). The actual deficit is the deficit actually registered over 1986, net of capital losses on external debt. 10' The actual deficit in 1986 is compatible with a sustained inflation rate of almost 50%: the RDR turns negative when inflation goes from 45% to 50%. The Table also shows that a target rate for inflation of 20% implies a required deficit reduction (RDR) of one percent of GNP. However, the Tables are drawn up under the assumption of constant nominal interest rates. In particular, the time deposit rate is kept fixed at 55%. This would imply a JIG' Capital loses on external public sector debt are excluded, not because they would not constitute a real increase in public sector liabilities, but because they are unlikely to recur in the future. This is certainly the case with the cross-currency-fluctuations component; while nobody can accurately predict major exchange rate movements, there is a general consensus that the dollar has "bottomed out". The assumption of no real depreciation for given cross-currency rates may be more contentious; we explore the consequences of alternative scenarios below.over the five years under consideration. This is taken up further below. - 33 - real rate of interest of 29 percent, clearly an unsustainable situation. Real rates on bonds would have to rise to similar levels for the Treasury to be able to issue them, with predictable consequences for debt-service cost. An alternative scenario would lower the nominal rate of interest in line with inflation in order to maintain real rates of interest. This will lead to lover demand for time deposits by comparison. But empirical analysis suggests that some of this shift (almost a third) will go into demand deposits. This moderates the impact of lower time deposits on base money demand and hence on the basis for the inflation tax. The net effect is a decrease in the financeable deficit at 20 percent inflation, and hence an increase in the required deficit reduction, from 1% to 1.2% of GNP. Several comments are in order. First, 1.2 percentage point of GNP is in fact a large adjustment. It would, for example, require a 9.4% cut in public sector investment, or a 13.7 percent cut in public sector consumption. Second, a larger cut will be needed if instead of a zero real depreciation, the real exchange rate should be expected to depreciate at a positive rate. Third, although accurate numbers are not yet available, indications are that the fiscal deficit has widened substantially in 1987. The required deficit reduction would be commensurately larger. 6. 2 Fiscal Implications of Financial Sector Policies The previous section demonstrated the importance of revenue from monetization in the financing of government expenditure. In such circumstances, changes in financial regulation may have important fiscal consequences. Changes in reserve requirements, shifts out of domestic assets, changes in the interest rate structure on deposits etc. all influence the level of reserve money the private sector and the commercial banks will hold - 34 - for any given inflation rate. Fiscal consequences should therefore be taken into account when recommending reforms affecting any of these variables. Consider for example changes in reserve requirements. These were recently increased from 10 percent to 14 percent on all domestic currency deposits. This clearly raised the level of required reserves for any given deposit interest rate structure and inflation rate. Hence base money demands went up and revenue from monetization increased. Empirical analysis suggests that the level increase in base money is likely to have yielded a once-off gain of one percentage point of GNP extra revenue; in addition, since the level of base money demand will stay higher as long as these reserve requirements are kept at 14% instead of 10%, there are recurrent gains in both inflation tax and seignorage. This is because the tax is now levied over a higher base. As a consequence, the increased reserve requirements eased to fiscal adjustment burden by 0.25% of GNP in each year the reserve requirements are kept at 14%. This lowered the sustainable inflation rate by more than 10 percentage points to the 50% level it is at now. In many countries, reserve requirements are different against deposits of different maturity. Turkey's value of 14% is not unusual as a reserve requirement ratio against demand deposits. But, at least in many other OECD countries, reserve requirements against time deposits are much lower. We can use the same framework to assess the fiscal implications of lowering the reserve ratio applicable to time deposits to, say, 5%. This would have a substantial impact on demand for base money, since reserves held against time deposits are a major component of it. The equilibrium level of base money demand would drop by no less than 1.5 percent of GNP in response to such a regulatory change. This would present a once-off revenue loss of that magnitude for the public sector. In addition, future revenue from inflation - 35 - tax and seignorage would be reduced, since the level of base money demand would be lower for any given inflation rate. The loss would be substantial: at an inflation rate of 40%, the combined loss in inflation tax and seignorage because of this cut in reserve requirements would be 0.4 percent of GNP each year. Any such reform measure should therefore be accompanied by fiscal measures to offset what is quite a substantial negative budgetary impact. * Increases in demand deposit rates have similar consequences. An increase in the demand deposit rate to 50% (which would make it positive even at 1987's high December-to-December inflation rate I-i) trigger a substantial shift out of cash balances: almost 1.3 percentage points of GNP. Since reserve requirements on demand deposits are only 14%, this lowers demand for base money by 86% of the shift. The econometric analysis also suggests that there would be an additional shift out of non-financial assets into demand deposits, of about equal size, but this would raise demand for base money by not more than 14% of the shift. The net impact on the level of base money demand and hence on revenue from monetization would thus be negative. By coincidence the magnitude of the required fiscal adjustment is almost identical to what is required after a cut in reserve requirements back down to 102. An additional consideration should be the impact of differential reserve requirements on monetary control. With uniform reserve requirements, shifts between different deposits do not influence the demand for base money. However, any portfolio shift between demand and time deposits will influence base money demand if there is a significant difference between reserve requirement ratios applicable to the two types of deposits. This would complicate monetary policy considerably. .V ' The December-to-December inflation rate is probably a misleading indicator of the underlying "core inflation rate" for 1987. Extensive pubic sector price level adjustments caused a shift in the price level of 12% in December 1987 alone. - 36 - One should exercise care in interpreting such results. Pointing out the negative fiscal consequences of say cuts in reserve requirements does not imply that no such cuts should be undertaken. High reserve requirements carry efficiency costs that have not been incorporated in this analysis. It does mean, however, that reform packages incorporating measures like this should also specify to which extent and in which manner the fiscal consequences should be dealt with. 6. 3 Fiscal Implications of Debt Management What would have happened if Turkey had not followed its policy of a relaxed external deficit and only moderate internal debt issue? In particular, what are the fiscal consequences of a debt substitution policy followed in many debtor countries? Many of them in effect paid off relatively cheap external debt from revenue raised by issuing much more expensive domestic debt. Assume that Turkey had not increased its external debt at all between 1980 and 1986, other than what was caused by capital loses due to exchange rate depreciation, but instead had issued internal debt. The study on Turkey referred to before showed that after correction for cross-currency exchange rate fluctuations and real depreciation of the TL, Turkey's debt-output ratio went up by only 13.8 percentage points of GNP. The rest was due to capital losses. What would have happened if Turkey, instead of increasing its external debt-output ratio by 13.8 percent of GNP, had issued an equivalevt amount of internal debt instead? First the results of a mechanical debt swap: a once-off sale of domestic debt to retire an equivalent amount of external debt. This effectively amounts to a debt-buy-back scheme. This experiment considers only - 37 - the budgetary consequences of changing one type of debt instrument for another. It does not consider the transfer problem associated with effecting any transfer of resources to foreigners. Such a scheme becomes problematic when domesti. real interest rates are substantially higher than the average real interest cost of foreign debt. In that case the budgetary situation deteriorates. This would also be an issue in Turkey: over the 1988-1992 period, real rates at home are projected to be 6 percentage points above the average real cost of foreign debt. As a consequence, the increased interest burden caused by such a debt swap would raise the actual fiscal deficit by 0.8 percent of GNP in each subsequent years and the required deficit reduction for consistency with 20 percent inflation rises to 2.1 percent of GNP, up from 1.2 percent of GNP in the base case. Alternatively, the equilibrium inflation rate would jump to 85 percent per year, up from 50 percent, if no fiscal adjustment would be undertaken. A straight asset swap was, however, not the form in which this debt substitution was implemented in most high-debt countries. In order to effect the implied transfer to foreigners, the government needs to find a way to increase either its own surplus or the net private savings surplus by a matching amount. Typically, the dorestic counterpart of the increased external transfer was a gradual increase in domestic debt issue, absorbed through an increase in the private net savings surplus. This in turn required higher real interest rates. Such a strategy would be much worse from a budgetary point of view. The reason is that this scheme would in fact raise the cost of the internal debt beyond its already high level and thus worsen the impact on the budget further. Assume that such a debt substitution strategy would be implemented over the next five years, the time horizon taken in this chapter. Since over that period real interest cost of foreign debt is - 38 - assumed to equal the real output growth rate, the enLire adjustment would need to come out of the non-interest current account. To achieve the target reduction of 13.8 percentage points of GNP over a five-year period thus requires a substantial positive shift (2.7 percent of GNP, 13.8 divided by 5) in the non-interest current account in each year. Inducing an increase in net private savings requires a rie in the real interest rate. The empirical analysis discussed elsewhere (see Annex IV of Van Wijnbergens et. al. 1988) suggests that such a large increase requires an increase in domestic real interest rates of almost 7 percentage points. This would not only raise the servicing costs of the additional domestic debt created during such a policy, but also he cost of debt incurred earlier as it gets refina-nced. This is important because by now most of Turkey's internal debt has a short mattirity (by December 1986, 16 percent of the internal debt had a maturity of one year or less). The impact on the budget would be large. To sustain consistency with a 20 percent inflation target after such a debt substitution policy would now require a reduction in the fiscal deficit of 3.6 percent of GNP. This is almost double the adjustment necessary after a straight asset swap. The budget deterioration would in fact be so large, that covering it through monetization would no longer be feasible. Increased debt issue would be even worse because of the high real interest rates. Finally, external debt would not be available by the very design of the scheme, which was to reduce external debt. A fiscal cutback would thus be unavoidable and would have to be substantial. This raises the issue of whether output growth could in fact be sustained. This is explored further in Section D, but the numbers presented here should already indicate that it is highly unlikely. - 39 - 6.4 Fiseal Consequences of Exchange Rate Poliey Another issue concerns exchange rate policy. Turkey has followed an aggressive export promotion policy, in which the exchange rate has been one of the major instruments. Turkey's export-weighted real exchange rate has depreciated by an average 6 percent in real terms since 1980. This has been the most important factor behind Turkey's extremely successful export drive. Its counterpart, however, has been increasing capital losses on its foreign debt. Nevertheless, in spite of increases in its debt-output ratio, export growth was so high that the debt-export ratio has remained fairly stable since 1981. The trade-off then is clear. Continued depreciation of the real exchange rate will help to maintain export growth in excess of the growth rate of real GNP, but at the cost of an escalating debt burden as measured by the debt-output ratio. Empirical analysis reported in the Turkey report (Van Wijnbergen et al, 1988), shows that the net impact on the debt-export ratio, is, however, positive. The trade-off then will depend on which target is adopted for external borrowing. If the solvency oriented debt-output ratio is the constraint on external borrowing, higher real depreciation will severely restrict the room for fiscal deficits. With the debt-to-export ratio as the constraint on external borrowing, however, this result will be reversed as a consequence of the high price elasticity of Turkish exports. Table 5 demonstrates these effects numerically using the consistency model just described. The table presents the fiscal cutbacks required for consistency with a 20 percent inflation rate for different rates of real depreciation. It does so for two difference scenarios. In the first scenario (columns 2 and 2 in Table 5), external borrowing is restricted to just the amunt that would leave the debt-output ratio unaffected (this is a positive - 40 - amount, since output is growing). The table clearly shows how, under such an external debt policy, capital losses on foreign debt due to real depreciation of the exchange rate severely restrict fiscal policy. A required fiscal deficit cut of 1.2 percent of GNP at zeto real exchange rate depreciation jumps to 3.0 percent at a five percent real depreciation. If the real exchange the depreciates by 10 percent on average over the five year period, the required deficit reduction (RDR) increases to no less than 4.9 percent of GNP. The main reason for this is reduced room for external financing: this falls from 2.5 percent at zero depreciation to 0.8 percent at a real depreciation of 5 percent (the columns under FCA, feasible current account deficit, in the table). At a real depreciation of 10 percent there is no room for external borrowing at all. Table 5: REAL EXCHANGE RATE DEPRECIATION, FISCAL ADJUSTMENT, AND FEASIBLE EXTERNAL BORROWING (Percent of GNP) REAL EXCHANGE DEBT-OUTPUT DEBT EXPORT RATE DEPRECIATION TARGET TARGET FCA RDR FCA RDR (1) (2) (3) (4) (5) 0.0 2.6 1.2 2.5 1.3 5.0 0.8 3.0 3.6 0.2 10.0 0.0 4.9 4.7 -0.9 FCA = Feasible Current Account deficit RDR = Required Deficit Reduction L_ This is the deficit reduction required (with respect to the 1986 fiscal deficit) for consistency with a 20% inflation target. - 41 - The results are very different if the target is to maintain a constant debt-to-exports ratio. In that case, real depreciation eases room for external borrowing since the volume effect on exports offsets the valuation effect on debt. Feasible external financing under this scenario goes up from 2.5 percent at zero real depreciation to a very high 3.6 percent at 5 and 4.7 percent of GNP at 10 percent real depreciation. The extra fiscal room this gives is reflected in the corresponding RDR row. Both options are obviously too extreme. A strict debt-output target would be too restrictive a guidance for external borrowing. Pursuing an exchange-rate based export promotion policy while ignoring the favorable impact this has on creditworthiness would unduly restrict external borrowing. The real depreciation necessary for the export promotion strategy would cause capital losses on external debt. Maintaining the debt-output ratio would then require a reduction in foreign borrowing. It might, in fact, by the fiscal restraint it would necessitate, threaten the export boom that the real depreciation was intended to produce. This could happen if the fiscal restraint would directly or indirectly, lead to reduced investment in export sectors. The other policy, targeting the debt-export ratio, would clearly also carry high risks. If the export boom falters, the economy would be left with a high debt-output ratio and the possibility of a sudden cut-off from external funds. 6.5 FiNal Implications of Variations in Output Growth Higher growth allows more internal debt issue, since the target is a constant debt-output ratio; it will also increase demand for real money balances by both banks and the private sector, thus increasing the scope for revenue from monetization for any given inflation rate. Bence more growth allow a larger deficit and less need for fiscal adjustment. This is at the - 42 - core of the conflict between stabilization policy and growth: if stabilization policies cut output growth, further fiscal adjustment is needed for macroeconomic consistency. This adjustment may, in turn, slow growth further. Table 6 indicates the extent of the trade-off. A four percent growth target instead of 6 percent reduces financing room by about one percentage point of GNP: for a 20 percent inflation target, the required deficit reduction consistent with a 20 percent inflation target (RDR) becomes 2.3 percent of GNP at 4 percent real growth instead of 1.2 percent at 6 percent real output growth. A major recession brings it out more starkly: a sustained period of only 2 percent growth in real income would raise the required adjustment necessary for coneistency with a 20 percent inflation target to no less than 3.3 percent of GNP. Numbers this large raise the spectre of self-fulfilling prophecies: a deficit reduction this severe could easily validate the low growth rate on which it was premised. 6.6 Summing Up The analysis has until now focused on the revenue the government can expect from various sources of financing given its macroiconomic targets. Reducing the fiscal deficit to what ig financeable given those macroeconomic Table 6: FISCAL IMPLICATIONS OF OUTPUT-GROWTH Required Deficit Reduction for a Output Growth 20% Infl. Target (percent) (percent of GNP) 2 3.3 4 2.2 6 1.2 - 43 targets makes sure that the fiscal policy is at least sustainable. If this adjustment is made, achieving the stated macroeconomic targets will not be jeopardised by fiscal crises, high inflation or escalating interest payments. Nowever, it does not guarantee that those macroeconomic targets can or will be achieved; only that the fiscal deficit is not inconsistent with them. Whether the targets can be achieved depends on two major factors. First, will the private sector in fact generate a sufficiently large surplus of private savings over private investment for the economy to achieve its external targets, given the fiscal deficit? Second, this surplus should be achieved at sufficiently high levels of investment to meet output growth targets given the public investment program. The extent to which public sector policy can play a role in this process is the subject of the next Section. 7 Internal Adjustment: Puble Sector Polcy and Private Savgs and Investment behavior The analysis assumes that the borrower has opted for a growth-oriented strategy within the constraints sustainability of foreign borrowing imposes, rather than to rely on high surpluses on the non-interest current account to keep the debt-output ratio in check. The key factor determining success or failure of such a strategy is an internal adjustment program that relies sufficiently on reduced consumption rather than reduced investment to generate the internal surplus that is required. If consumption does not fall, either external targets or output growth will need to be sacrificed; the former, if investment is not reduced and the latter, if it is. The central question thus is whether external restraint and consistency requirements for fiscal deficits leave enough room for public and private investment and satisfactory output growth. Can external balance and - 44 - output growth be reconciled, or is there an inherent conflict between these two objectives? It is here that the interaction between private sector savings and investment decisions and fiscal policy becomes important. The way consistency between internal policies and external target is brought about determines whether fiscal plans and external targets can both be met without jeopardizing output growth: does the private sector run a surplus at high levels of savings and investment or at low levels? If the surplus is achieved by increasing savings for sustained investment levels, output growth can be maintained. If however the adjustment comes mostly out of investment cutbacks for given private savings rates, external adjustmen- is bought at the cost of lower output growth. This Section focuses on the role that fiscal policy and real interest rates can play in bringing about these developments. There are several channels through which fiscal policy influences the size of the private sector's net savings surplus and the level of investment at which any given surplus is achieved. First, fiscal policy may exert a direct influence on the net private savings surplus through real interest rate-based crowding-out. The overall fiscal deficit is important for this channel. But if high real interest rates are maintained to create the room for higher fiscal deficits without a matching current account deterioration, how can output growth be maintained? High real interest rates presumably slow down at least private investment, thus slowing down output growth. Fiscal policy can play a role in avoiding such a slowdown in two different ways. The first one focuses on policy instruments that to some extent will focus the effect of high real interest rates toward consumption restraint while shielding private investment. Investment incentives, tax measures and credit -45 - policy all play a role here. Second, output growth depends on aggregate investment, not just on private investment. There is therefore a role for public investment in reconciling external balance an output growth. Government investment itself results in capital accumulation. So negative output effects of fiscal deficits through real interest-based crowding-out of private investment can be offset to some extent by shifting the composition of government expenditure away from consumption to investment. The composition of government expenditure and not just the overall deficit is an important part of a successful internal adjustment program. In addition to this direct substitution effect, there is a more indirect channel through which the composition of government expenditure influences private investment. Public sector investment, especially in infrastructure, often stimulates rather than replaces private investment expenditure. Public sector investment in, e.g.: roads, will make investment more attractive for the private sector in places that were inaccessible before. This channel is one reason why private sector investment in Turkey has in fact not suffered that much from the continued high real interest rates over the past five years. 8 The role of Public Sector Investment: the case of Turkey Large fiscal deficits have until now not prevented a satisfactory current account performance. The price for this has been the need to maintain increasingly high real rates of interest. Empirical analysis shows that in Turkey such a policy is effective by restraining private consumption, and, to a lesser extent, private investment expenditure. Deleterious effects on output growth have until now been avoided. High public sector investment has been an important explanation of why output growth did not slow down. - 46 - Figure 4 shows the results of simulation runs made with an econometric model used in the report on Turkey mentioned before. The runs are designed to bring out the role of public sector investment in the growth process. Interest rates were varied, but fiscal deficits were adjusted so as to maintain external balance targets. First, the fiscal cutbacks necessary to sustain external balance, as interest rates are lowered, were assumed to come entirely from government consumption. Public sector investment remains constant by assumption. The figure shows that a five percentage cut in interest rates will cause a drop in the private sector's surplus of savings over investment of 2.1 percentage points of GNP (see Figure 4A, upper right). 1' A substantial part of the decline in net private savings comes from increased investment by the private sector in response to the lower real interest rates. Since public sector investment was fixed by assumption, output growth goes up, by 0.5 percentage point of GNP on average over the five year period the model was run (see Figure 4A, upper left; the base run simulates the period between 1981 and 1986). The results are very different when the fiscal cutbacks are assumed, perhaps more realistically, to come also from public sector investment rather than from consumption. Assuming that all government expenditure would be cut back proportionally implies that 60 percent of the cut comes from reductions in the public sector's investment program. The results are summarized in Figure 4B. Now while the lower interest rates stimulate private investment, the cut in public sector investment more than offsets this: as a result, output growth actually declines by an average 0.5 percentage point of GNP over In the run, the spread between lending rates and deposit rates was kept constant. A five percentage points cut in borrowing rates thus implies a five percentage points cut in lending rates too. - 47 - Figure 4: TE EFFECT OF CHANGES LN FISCAL DEFICIT ON I.YTEREST RATE A.D QUUT GH Fig. 4A: Entire Fiscal Cut from Govermnt Consumption 4 .. OUTPUT GROWTHf RATZ n PRtVATE SAVINGS 4. Fig 4: 60 Percent of Fiscal Cut from Pubitc Sector Investment Is 8 p p 1- p , , .* * 4.5. 4 *1 *a -' a* OUTPUT G0tWith tATE NET PatVATE SAVENGS - 48 - the five-year simulation period. Shifting from no cut in pubic sector investment to letting 60 percent of the fiscal adjustment come out of cutbacks in public investment therefore causes a full percentage point drop in GNP growth for the five years over which the model was run. There is, moreover, a vicious circle aspect to this policy experiment. Cutting public sector investment reduces output growth, which in turn will lead to less of a private sector's savings surplus. As a consequence, fiscal deficits and hence public sector investment need to be cut further to maintain external balance, growth slows down more and so on. As a result, a five percentage point cutback in real interest rates requires a cut in the fiscal deficit of 2.1 percentage points of GNP if external balance is to be maintained trough reduced government consumption. However, with 60 percent of the costs coming from public sector investment, deficits need to be reduced by 2.8 percentage points of GNP, a full 0.7 percentage point of GNP more. The arguments presented here do not imply a blanket endorsement of ever increasing public sector investment; public sector investment of course does come at a cost. -' They do highlight, however, that public sector investment has played an important role in Turkey's strong growth performance over the last few years. Moreover, they show that stabilization programs relying on reductions in public sector investment could have high and permanent negative output effects through the mechanisms demonstrated. These are in addition to any output effects that may arise because of short-run macroeconomic problems, which are not covered here. "1' Other expenditure components need to be cut or alternative means of financing need to be found; each carries its own cost. - 49 - An issue for concern is the resulting composition of investment. To avoid competition between public and private investment, public investment should focus on areas where it is complementary to private investment rather than a substitute for it. Such areas are infrastructure, health and education. However, it then becomes important not to slant private investment in the same direction through improperly structured investment incentives. This is currently an issue in Turkey. The adverse effect of high interest rates has predominantly been on investment in manufacturing. Its share in private investment 'ell, in response to strong public sector support for investment in housing through the Mass Housing Fund. As a consequence, investment by the private sector seems to have shifted away from the main export sectors, agriculture and manufacturing. The share of each sector in total private fixed investment has fallei by 5 percentage points between 1981 and 1987. Such a change in the composition of investment is a potential cause of future problems. The combined effect of these private and public sector developments has been a substantial decline of the share of total investment going towards the main export sectors. This has an impact on the link between export growth-real exchange rate-cost of external debt identified in previous Sections. With less capital in the tradable sectors, a larger shift in the real exchange rate is necessary to achieve a given expert target. This in turn increases the capital losses Turkey will sustain on its external debt. A reorientation of investment incentives towards the main export sectors should receive serious consideration, if this unfavorable shift in the export-growth/cost-of-external-debt trade-off is to be avoided. Lower real interest rates and a reallocation of incentives towards the traded goods - 50 - sectors would allow the export drive to continue with a slower increase in the debt-cutput ratio. 9 Summing Up This paper has provided a guided tour around the building blotck of an external debt strategy and the policy implications the necessity of a matching internal adjustment program lead to. By way of summing up, we would like to comment on two issues: the role of real interest rates and the actual implementation of the measures discussed. High real interest rates in a growth oriented adjustment progrum sound like a prima facie contradiction in terms. High growth requires high investment, and high real rates clearly slow investment down. However, high real rates may be necessary to make sure that a sufficiently large private savings surplus is generated to make fiscal deficits and external balance targets consistent. As was discussed before, investment incentives and tax measures could be used to make sure that most of the effect of the high real rates is shifted towards consumption rather than investment. It is clearly true that a larger public sector deficit cutback generates less need for high real rates. This would obviously be desirable; artificially high interest rates, at least to the extent that they exceed world interest rates, are a price distortion just like any other wedge between domestic and world prices. However, cutting fiscal deficits implies welfare costs, too. While there is often room for reducing government consumption expenditure, there is a stage where further cuts cause excessive damage to the quality of government services. Similarly, public investment programs often include inefficient projects. But we have demonstrated that public investment also has an important role to play in achieving sustainable growth. Excessive - 51 - cuts will jeopardize that policy goal. Finally, higher taxation, too, carries its costs. Higher taxes are almost always possible, and often desirable. However, higher tax rates invariably imply higher price distortions and increased tax evasion. This way of reducing deficits has its limits, too. The conclusion is that lower fiscal deficits for given interest rates and higher interest rats for given deficits are both adjustment mechanisms that cause welfare costs one way or another. A properly designed adjustment program should therefore include some of each, so as to minimize the overall welfare costs of the adjustment program as a whole. A final comment, to put the approach taken in this paper in perspective. The crisp sequential way in which the whole process was discussed is to some extent a simplification, useful for presentational purposes, but a underestimate of the difficulties likely to be encountered in practice. One example should suffice to demonstrate this. We argued that, once consistency calculations have indicated the sustainable size of the fiscal deficit, high real rates may be needed to guarantee a matching surplus of private savings over investment. Furthermore, additional investment incentives may be needed to guarantee that this surplus is brought about at a sufficiently high level of private investment. But both these policies will in turn have a negative impact on the budget and hence change the need for deficit reduction. High real interest rates will raise the cost of internal public sector debt, if at least the private sector is to be convinced to hold the extra debt voluntarily. Subsequent investment incentives to shield private investment will also imply budgetary costs. The conclusion is twofold. First, this whole process is likely to need several iterations, until all components are internally consistent. It is almost certainly not possible to seriously design such a program without - 52 - the aid of some form of quantitative analysis, if only to clearly bring out all the interdependencies and their quantitative significance. But models are by necessity imprecise, and often difficult to parametrise. Their results are important, but should for that reason be used in addition to Intuitive and informed judgment; quantitative policy analysis is only one element, although an important one, in the design of good economic policy. That qualification also leads to the second and final conclusion. The very lack of precision of qvantitative analysis, and the unpredictability of external events, calls for continuous reassessment and substantial flexibility in policy making. Consistent and predictable policy measures are an importarc precondition for credibility of any program; however, policy credibility is not enhanced by clinging to views and policy measures that have been overtaken by events. There is little room for dogma in the design and implementation of economic policy. - 53 - References Anand, R. and S. van Wijnbergen (1987), "Inflation and the Financing of Government Expenditure: an Introductory Analysis", mimeo, World Bank. Buiter, W. (1985), "A Guide to Public Sector Deficits", Economic Policr. vol. I. Cohen, D. (1985), "How to evaluate the Solvency of an Indebted Nation", Economic Policy, vol. I. Cohen, D. (1987), "External and Domestic Debt Constraints of LDCs: a Theory with Numerical Applications to Brazil and Mexico", Wotld Bank Quarterly Economic Review Sargent, T. and N. Wallace (1982), "Some Unpleasant Monetary Arithmetics", Ouarterly Review, Federal Reserve Bank of Minnesota. van Wijnbergen, S., R. Anand and R. Rocha (1988), "Inflation, External Debt and Financial Sector Reform: a Quantitative Approach to Consistent Fiscal Policy", mimeo, World Bank.