How much fiscal adjustment is enough? The case of Colombia

| Policy, Planning, and Research WORKING PAPERS Macroeconomic Adjustment -and Growth j Country Economics Department The World Bank June 1989 WPS 201 How Much Fiscal Ad justment Is Enough? The Case of Colombia William R. Easterly Colombia's impressive fiscal adjustment during 1985-87 was due to structural changes in fiscal policy-not to the coffce boom. Furtherreduction ofthefiscaldeficit isrequired toreduce interest rates and inflation to more manageable levels. The Policy, Planning, and Rcsearch Compicx distnbutLs PPR Working Papers to dissermnate the findings of work in progress and to encourage the exchangc of ideas among Bank staff and all o-hers intcrcsted in development issues. Thesc papers carry the names of the authcm, reflect only their views, and should be used and cited accordingly The findings, interpretations, and conclusions arc the authors' own They should not bc aitrbuted to thc 'orld Bank. s sBoard of Dircctors. iLs management, or any of iLs mcmbercounties  Plc,Planning, and Research . Maroeconomic Adjustment and Growth Colombia's. impressive fiscal adjustment during development Iending as a percentage of GDP 1985-87 was due to structural changes in fiscal fell slightly during the same period. policy, concludes Easterly-not simply to such Perhaps partly becat e of public debt fortuitous events as the coffee boom. behavior, real interest rates remained very high Losses of public financial institutions were and inflation accelerated slightly. Improving on important in some other Latin American coun- adjustment would probably require reducing tries but there is no evidence that they were a interest rates and inflation. major factor in Colombia. The data suggest that Easterly's model simulations suggest that to the Banco de la Republica and other public reduce interest rates to more manageable levels financial institutions suffered a small quasi- would requirc continued reduction of 1' - fiscal fiscal loss, but that that loss was not the domi- deficit, below levels currently envisioned. To nant factor in fiscal behavior. reduce inflation would require even tighter fiscal Although impressive, the fiscal adjustment policy. fell short of actually improving the govern- The magnitudes of required deficit reduction ment's net financial position. Total public debt do not seem out of reach however, even allow- Ps a percentage of GDP was roughly unchanged ing for uncertainty about the figures. Continued from its 1984 value at the end of 1987, even policy initiatives would help Colombia confront after correcting for the effect of currency devalu- the fiscal challenges of the 1 990s. ation on dollar-denominated instruments. Public This paper is a product of the Macroeconomic Adjustment and Growth Division, Country Economics Department. Copies are avaiiable free from the World Bank, 1818 H Street NW,Washington DC 20433. Please contact Raquel Luz, room NI 1- 057, extension 61760 (42 pages with tables). The PPR Working Paper Series disseminates the findings of work undcr way in the Bank's Policy, Planning, and Research Complex. An objcetive of the series is to get these findings out quickly, even if presentations are less *han fully polished. Thc findings, interp7etations, and conclusions in these papers do not necessarily represent official policy of the Bank. Produced at the PPR Dissemination Center  How Much Fiscal Adjustment is Enough? The Case of Colombia by William R. Easterly Table of Contents I. Introduction 1 Structural fiscal deficit adjustment, 1985-87 2 Profits/losses of public financial intermediaries 4 Financing of public sector deficits 8 Evaluation of sustainable deficits 13 Interpretation of results 18 Conclusions 19 Table 1: Structural Trends in Fiscal Policy 21 Table 2: Cuenta Especial de Cambios-Banco de la Republica 22 Table 3: Interest Rates for Fondos Financieros and Hypothetical Operating Losses 23 Table 4: Net Financing of Total Public Sector- Inflation Adjusted 24 Table 5: Public Debt Ratios 25 Table 6: Regressions for Simulation Model 26 Table 7: Fiscal Policy Simulation: Interest Rates and Inflation Unchanged 28 Table 8: Fiscal Policy Simulation: Fall in Interest Rates 29 Table 9: Fiscal Policy Simulation: Fall in Interest Rates and Inflation 30 Bibliography 31 Appendix: Model of government deficit finance 33 I am grateful to members of the Departamento de Investigaciones Econ6micas of Banco de la Repuiblica for valuable suggestions and assistance, to E.C. Hwa, Peter Miovic, Bela Balassa and members of CECMG for extremely useful conments, and to Susan Hume for excellent research assistance.  Introduction A key component of the largely successful adjustment program Colombia has followed since 1985 has been the 4mprovement in the fiscal deficit. Starting from a position of severe imbalances in both the external and fiscal accounts prior to 1985. Colombia had achieved surpluses in both accounts by 1986. Although both have since reverted to deficit, the levels are much lower than before the onset of adjustment. At the same time, other indicators of economic performance have been favorable. Substantial growth retutned in 1986-87 (over 5 percent), and export performance was buoyant, at least partially in response to the major exchange rate adjustment that accompanied the adjustment program. Inflation performance has been good for most of the adjustment period, remaining relatively constant at around 22 percent despite the rapid rate of currency devaluation. Only recently (in 1988) has inflation accelerated, which may in part reflect temporary supply shocks.1 Despite the apparent success of the fiscal adjustment, there are several nagging questions that remain. One question is how much the fiscal adjustment that was achieved was the result of temporary and/or fortuitous events, as opposed to fundamental policy changes which have lasting effects. A related question is how the deficit reduction was divided between reductions in investment and increases in noninterest current saving. There are also more fundamental questions about how fiscal adjustment should be defined and whether the traditional analysis of the nonfinancial public sector deficit is adequate. Public financial l.The nature of the imbalances is summarized nicely in Thomas (1985). A sumnuy of the adjustment program can be found in Ocampo (1987), and Garcia (198'j).  -2- intermediaries, especially the central bank, sometime contribute to public deficits through losso: suffered on their financial operations. Development credit extanded either by public finan2ial intermediaries or the nonfinancial public sector has to be financed just as do traditional public expenditures. To capture &ll of these factors, the most comprehensive definition of the public sector would be through its total financing requirements. This is also useful from the standpoint of macroeconomic analysis, since it is through its financing needs that the public sector comes to affect interest rates, private investment, inflation, and other macroeconomic variables. The consideration of public flnancing requirements leads naturally to the analysis of public debt behavior. This is the 'bottom linew of fiscal analysis, since the public debt captures the long-term cost of fiscal policies. The composition of public debt also affects domestic interest rates. It may also have fiscal implications if differentials exist between domestic and external interest rates. The fiscal adjustment of 1985-87 will therefore also be evaluated from the standpoint of the impact on public debt. Finally, and perhaps most importantly, the future sustainability of current public deficits must be analyzed. This can be done by evaluating the implications of future deficits for inflation and interest rates, and checking the consistency with policy targets for these variables. Structural fiscal deficit adjustment, 1985-87 This section will evaluate the question of how much thi adjustment achieved during 1985-87 reflected temporary factors. There are several temporary shocks to consider in the fiscal accounts. The public sector benefited from the advent of a fortuitous rise in coffee prices during 1986, which yielded enormous profits to the quasi-public National Coffee Federation (?NC in Spanish). This also contributed extra revenue to the National  -3- Government through the coffee export tax and through special transfers from the FNC. The rise in coffee prices was temporary, as they fell again in 1987.2 However, by that time, substantial new finds in oil reserves had begun yielding additional profits through the state-owned oil company ECOPETROL. Finally, a major tax reform in 1986 had a large but temporary effect on the budget through a one-time tax amnesty that took effect in 1987. The effect of these factors is shown in Table 1, which shows the overall public deficit adjusted for the surpluses or deficits of the FNC, ECOPETROL, and the temporary items in the National Government budget. The unadjusted consolidated public deficit improves almost 7 percentage points of GDP between 1984 and 1986, when a small surplus is achieved. There is some slippage in 1987, when it reverts to deficit again, so that the total adjustment over 1984-87 is a little more than 5 percentage points. The corrected 'structural deficit' shows much smoother behavior. There is steady improvement from 1984 to 1987, with the total adjustment amounting to slightly less than 5 percentage points. Thus, the adjustment from 1984 to 1987 is almost entirely due to structural factors, with the temporary factors mainly affecting the path of the adjustment and the level of the remaining deficit. Table 1 also shows the composition of the adjustment. We first remove the external and internal interest payments to get the 'structural primary deficit". This reflects the impact of the current fiscal policies, as opposed to the legacy of past deficits as reflected in inteorest on debt. A deficit on this account exceeding the revenue from money creation is not sustainable in the long-run. We see that a structural primary surplus had in fact been achieved by 1987. The improvement in the structural primary deficit 2.See Pizano (1988) for a discussion of the effect of the 'coffee boom' on public finances. Cuddington (1986) contains a more general discussion of the role of coffee in the economy.  -4- amounted to about 6 percentage points of GDP over 1984-87. This exceeds the overall fiscal adjustment since the increase in public interest payments offset part of the improvement in the primary account. Finally, we break out investment expenditure to see the composition of the primary deficit reversal. The "structural primary current deficit"--i.e. the primary deficit less investment--was negative throughout the period. This simply means that non-interest public saving was positive. This component of the deficit improves about 3 percent of GDP over 1984-87. Thus, half of the structural fiscal adjustment was due to improved current saving and the other half due to cutbacks in investment. It is unclear to what extenc cutbacks in investment reflect actual improvement in the long-run fiscal picture. If productive investment was cut, this will lower future public income and so does not represent fiscal improvement. If projects were cut that were additions to sectors with excess capacity or that were otherwise- unproductive, however, then this would be genuine fiscal adjustment. Past analysis of public investment in Colombia suggests that some of the investments cut were indeed in excess capacity sectors, such as the electric sector. The conclusion is that at a minimum 3 percentage points of the fiscal adjustment was a long-run improvement, while at least part of the 3 percentage points that correspond to the cuts in public investment was an improvement. Profits/losses of public financial intermediaries The central bank, Banco de la Replblica, is traditionally excluded from most measures of fiscal behavior in Colombia. However, it has some revenues and expenditures which are analogous to those in the nonfinancial public sector. Table 2 shows an account of the central bank which cover most, though not all, of its operations. This is the account known as Cuenta  -5- Especial do Cambios (CEC).3 It includes net income from foreign exchange reserves, which represent net interest received on the country's international reserves, of which Banco de la Republica is the sole custodian. Other reserve income comes from sales and/or revaluation of gold. The capital gains on holdings of foreign currency show up on a realization basis as currency is bought and sold. On the experditure sie, the main item is the interest expenditure on central bank bonds called titulos canjeables (TC's) (this category also includes another type of central bank bond known as titulo de participacion, but for accounting purposes all of the bonds are classified as TC's). Since some of these bonds are dollar-denominated, this expenditure includes also capital losses suffered on the bonds from currency devaluation. These bonds are used in the open-market operations of Banco de la ReR,iblica to control the money supply. Besides administrative costs, the other main expenditure item is the exchange rate differential, which reflects differences. between the exchange rate at which the central bank values its reserves and the price at which they are sold to the government for foreign debt repayment. The balance on the CEC fluctuates bet een plus and minus one percent of GDP. Intenest on TC's grows steadily as open-market operations became more important in 1986-87. However, interest on foreigr. exchange reserves is also growing, as is the profit on the purchase and sale of foreign exchange, both of which reflect the rising level of net international reserves. A deficit of -0.3 percent of GDP in 1985 thus becomes a surplus of 0.6 percent of GDP by 1987. However, the profit on the sale and purchase of foreign exchange essentially reflects nominal capital gains on international reserves, which 3.A superb explanation and analysis of the CEC is contained in Jaramillo and Montenegro (1982).  -6- should be excluded to be consistent with standard deficit definitions.4 The balance excluding this item shows a deficit of between 0.2 and 0.5 percent of GDP over 1984-88. The CEC is not a comprehensive measure of the profit or loss of the central bank on financial intermediation. The most notable omission is the interest revenue and expenditure of the Fondos Financieros, which are the entities within the central bank that carry out development lending. The Fondos raise financing through mandatory purchases of their bonds at below- market rates by the financial system (known as *forced investments'). They then lend at a subsidized rate to the target sectors. The sketchy data available seems to indicate that the main loss-maker among the Fondos is the Fondo Financiero Agropecuario, which carries out agricultural lending. Recent adjustments in controlled interest rates applicable to the FFAP have raised its borrowing rate faster than the rate it charges on loans, as shown in table. 3. Other fondos have a positive margin on their lending. However, to estimate the profit or loss of these entities, we need data on their operating costs and their level of defaults on loans. Table 3 shows some hypothetical calculations which assume a 5 percent operating cost ratio and a 10 percent rate of default. This yields a deficit for the Fondos shown of between 0.1 and 0.2 percent of GDP over 1985-88. It should be stressed that this is only a hypothetical calculation, however. Other public financial intermediaries exist outside the central bank. These also engage in development lending and any patential losses or gains they realize should also be considered in an evaluation of fiscal behavior. One potential loss-maker is Cala Agraria, the public bank responsible for lending to the agricultural sector. It is estimated to have 4.Teijeiro (1989) has a good discussion of principles of measuring central bank deficits.  -7- had cumulative losses equal to 0.7 percent of GDP in 1985, part of which were concealed by the transfer of a note payable by the government. The stock of losses had been reduced to 0.4 percent of GDP by the end of 1987. Although Caja Agraria continued to run small losses in 1986-87, the rapid growth of GDP allowed the cumulative losses to decline in relative terms. Another important public financial entity is the Pondo de Garantia de Instituciones Financieras, which was created to deal with the crisis in the financial system which began in 1982. Substantial resources were contributed to the Fondo de Garantia by Banco de la Repoblica and by the Coffee Fund, with a flow of gross credit of 1.6 percent of GDP during 1986 from these two entities, as we will see below. However, the evaluation of profits or losses in an economic or accounting sense raises many difficult problems. The only source of income for the Fondo de Garantia is interest on its loans to troubled banks. It is difficult to value these loans at present, since their value depends on the return of the troubled banks to profitability, which is highly uncertain. As in the case of the U.S. savings and loan crisis, the fiscal cost of the support to the financial system may not be known for several years. A system of deposit insurance has been proposed to give the Fondo de Garantia another source of income through premiums paid on deposits. However, the system appears to face severe political and legal obstacles to its implementation. It is clear from this discussion that data on profits or losses of public financial intermediaries, including the central bank, are fragmentary and incomplete. The next section will present another method of estimating the balances of these entities from the net financing they utilize, but this approach also has pitfalls. It is clear that more research and data-gathering efforts are necessary to evaluate the fiscal burden of financial intermediation performed by public entities. Although these preliminary  calculations shov the deficits of pubLc financial entitieos to be small in absolute terms, they can be important at the margin when the authorities try to alter fiscal policy. Financing of public sector deficits This section examines the magnitude and significance of the fiscal deficits during 1985-87 from the financing side. The results are-based on a flow-of-funds exercise utilizing data on Banco de la RepAblica, the financial system, direct public borrowing from the private sector, and external debt flows.5 We consider six classes of economic agents: (1) Banco de la Rep6blica, (2) the nonfinancial public sector, (3) public financial intermediaries (Banco Central Hipotecario, Financiera ElActrica Nacional, Ca1a Aeraria, and Cala Social), (4) private financial intermediariec. (5) the nonfinancial private sector, and (6) external agents. The changes in these stocks represent borrowing or asset accumulation by each sector (as well as revaluation of liabilities or assets). Table 4 shows the results of the flow-of-funds exercise for 1985-87. The table presents the financing of the consolidated nonfinancial public sector and Banco de la Rep6blica in inflation-adjusted terms as a percent of GDP. The inflation adjustment subtracts the part of the flow of domestic financing that merely compensates for the erosion of the real value of debt outstending. The financing flo,ws also exclude the revaluation of external assets and liabilities caused by depreciation of the peso against the dollar, as well as that caused by depreciation of the dollar against non-dollar currencies .6 5.An earlier analysis of deficit financing is contained in Restrepo (1987), which uses somewhat different data and methodology. 6.Herrera (1988) has an excellent analysis on how to correct for the effect of inflation and devaluation on the change in financial wealth of the public sector.  -9 - The estimates of net financing shown in Table 4 are somewhat higher than the conventional measures of nonfinancial public sector deficits. The memo part of the table compares the conventional measure with the net financing unadjusted for inflation. There are two sources of discrepancy. One is that Banco de la Republica is found to have net financing requirements (after correcting for capital gains on foreign exchange reserves) of around 1 percent of GDP. This provides another estimate of the possible loss suffered on quasi-f3.scal operations by Banco de la Repa6blica. However, such an estimate should be viewed witia caution since there are many complications in making valuation adjustments and in treating central bank and government accounts consistently. This estimate is larger than that indicated by the calculations on the operations through the CEC and Fondos Financieros discussed above. Tracing the source of this possible loss would require further research. Even after removing Banco de la Rep6blica, we still find tiat the consolidated nonfinancial public sector has a somewhat higher net financing requirement in 1985-86 than that indicated by the conventional measure. The financing estimate may well be more comprehensive, since the conventional estimate is based on a less than complete sample of public enterprises, local governments, and national decentralized public entities. The financing estimate also includes some entities that are financial in nature, such as the Instituto de Credito Territorial (although not the public financial intermediaries such ae PEN and Caja Agraria mentioned earlier). However, in 1987 we find that the financing estimate is nearly identical to the conventional nonfinancial public sector deficit. Since the table consolidates the nonfinancial public sector and Banco de la Rep6iblica, money creation is shown as simply another way to finance the consolidated net deficit. The inflation correction is not  -10- appropriate for monetary financing, however, uince the inflationary erosion of the monetary liability is a tax on the holder which helps finance the public sector. The real change in the monetary stock represents the change in demand for money as a means of payment, the revenues of which accrue to the government. This is presented as seignorage" in the table. In addition, the rate of inflation times the pre-existing stock gives an "inflation tax, to the government. The sum of these two items is simply the nominal change in the money stock. This breakdown is shown in the table for currency and financial system reserves on deposits. The table shows the Inflation-adjusted financing requirement of the public sector to decrease from 6.2 percent of GDP in 1985 to 2.1 percent in 1987. Except for financial support and other rediscounts to the banking system, the nonmonetary public assets decline in real terms over the period. The support of the banking system is concentrated in 1986. Thus, although it appears from the conventional fiscal accounts that most of the fiscal adjustment took place in 1986 and that there was slippage in 1987, the data on gross financing needs tell a different story. The large commitments for financial support of the banking system made the overall financing needs fall much less than the net deficit in 1986. In 1987, by contrast, the reduction in the flow of lerling to the banking system more than offset an increase in the net deficit of the total public sector, so that the financing requirement fell. A major change in the type of public deficit financing is also evident over the period. While in 1985 net external lending was still accounting for about half of total financing, it was drastically reduced in 1986-87. Gross external financing was significant in 1986, but practically all of it went into reserve accumulation. Thus, the need for internal financing actually increased in 1986, even though the total financing  -11- requirement fell significantly. In 1987, the reduction in financing needs and a small increase in net external lending allowed the domestic financing to fall again. Examination of the composition of domestic financing shows that monetary financing was the most consistently important. The sum of currency creation and reserves held by the banking system amounted to about 1.7 percent of GDP in all three years. The forced investments (including both inversiones del encaie and inversiones obligatorias) are surprisingly unimportant as a source of finance.7 Bond sales were important in 1985-86, but turn negative in 1987. Lending by the financial system is also volatile--very significant in 1986, much less so in 1985 and 1987. Lending from the public financial intermediaries is larger than that from the private financial system. Another perspective on fiscal policy in these years comes from examingng the ratios of total public debt to GDP, shown in table 5. These ratios capture the long run impact of fiscal policy, since they measure the extent to which fiscal policy increases or lowers the requirement for future government saving. An increase in the ratio of government debt to GDP would require some future increase in government saving to pay the debt service. Table 5 shows that the initial year of the adjustment program was not successful in reversing the fiscal deterioration, as the debt ratio increased sharply in 1985. External debt increases particularly strongly.8 7.Forced investments are the mandatory holdings of liaoilities of the Fondos Financieros described earlier. 8.This is not due to the major currency devaluation of 1985, as the external debt figures are evaluated at the 1987 real exchange rate. We also correct for revaluation of the external debt due to deprec4ation of the dollar against other industrial currencies. The dollar-denominated domestic liabilities of Banco de la RepUblica are also corrected for valuation changes.  -12- In 1986-87, the strong fiscal adjustment reduces the overall debt ratio by an amount that roughly offsets the increase of 1985. Surprisingly, the strong fiscal adjustment during 1985-87 did not actually lower the public debt ratio.9 Table 5 is also insightful in shoxring the changing composition of public debt. The composition of debt at the end of 1984 was heavily weighted towards external sources, which accounted for 80 percent of total public debt. These proportions were roughly maintained during 1985. During 1986-87, however, there was a shift towards internal debt as the external debt ratio declined. The composition of internal debt was changing at the same time. Forced investments and bonds increased in 1985, but then declined in 1986-87. The expansion in internal debt in 1986-87 was mainly fueled by lending by public financial institutions, and to a lesser extent, by the private financial system. Meanwhile, public financial assets were roughly constant over 1984-86. Decreased development lending by the Fondos Financieros was offset by the increase in public financial support of the financial system. In 1987, public financial assets decline as a percent of GDP, as both development lending and lending to the financial system decline. Table 5 also shows that the real base money stock declines over 1984-87. This reflects a fall in the real demand for the money base, meaning the potential for financing through money creation was also being eroded. Thus, at the end of 1987, the public sector's financial position had not 9.As noted earlier, however, there are many complications involved in making this calculation, so some margin of error should be allowed for in interpreting this result. Possible complications include correcting for valuation changes, classification of assets and liabilities and differing accounting methods between the government and central bank. Robinson and Stella (1988) recommend excluding central bank debt associated with normal monetary operations, but it is difficult to see the economic justification for doing so. Further research is needed in this area.  -13- improved in absolute terms compared to the end of 1984. However, the public sector's financial position was much better at the end of 1987 than it would have been in the absence of fiscal correction. Evaluation of sustainable deficits To evaluate the sustainable fiscal deficit and the financing tradeoff in financing it, a simple model is used that relates the portfolio behavior of the private sector to the financing needs of the public sector.10 The Appendix contains an algebraic presentation of the model. This section summarizes briefly the model and then will summarize the results. The financial behavior of the private sector is modeled with a standard Tobin-style portfolio model with some simplifications. The demand for currency depends only on inflation and real GDP, as shown in the regression in Table 6.11 This implies that the demand for currency is strictly a transactions demand, with the real demand declining as the *tax' on currency holdings increases. The demand for non-currency domestic financial assets depands on real interest rates and real GDP, as shown in the second regression in Table 6. These results must be translated into portfolio demands for each of the assets in the model. We make the simplifying assumption that all domestic lO.This model is in the spirit of the approach of Anand and van Wijnbergen (1989) and van Wijnbergen, et.al. (1988) to modelling inflation and sustainable deficits in Turkey, with the addition of an endogenous interest rate and an analysis of transitional portfolio shifts. Easterly (1989) derives static and dynamic results for a similar type of model. ll.Steiner (1988) has an alternative estimate of the demand for currency as a function of nominal interest rates. His equations show much lower elasticities.  -14- noncurrency assets are perfect substitutes and carry the same interest rate. The shares of each of these assets in the domestic non-currency portfolio are assumed to remain the same as in 1987. The real demand for currency is independent of the size of the portfolio, with the desired ratio of currency to GDP depending only on the inflation rate. The share of the noncurrency assets which go to domestic assets depends on the domestic real interest rate, with the elasticity given by the regression in Table 6. The remainder goes into foreign currency assets. The asset side of private financial behavior is assumed to be strictly separated from the liability side.12 The demands for credit by the private sector are modeled as part of the investment decision. Regression 3 in Table 6 shows the ratio of private investment to GDP as a function of the real interest rate. We assume that private foreign borrowing is rationed by external capital markets and/or the government. Lending by public financial institutions and the central bank is determined exogenously as a matter of government credit policy. Domestic borrowing then becomes the residual source of finance for investment, and thus is a function of real interest rates as in regression 3. The financial institutions have their behavior determined largely by reguilation. They are required to set aside fixed percentages of their portfolio in reserves, reserve investments, and forced investments.13 The 12.This could reflect, for example, an institutional distinction between consumers and private firms. 13.Montes and Carrasquilla (1986) have a model of regulatory determinants of the interest rate structure. Correa (1986) also has an insightful discussion of the effects of the financial regulations.  private financial system supplies the credit demanded by the nonfinancial private sector at a given interest rate, then supplies the remainder to the government. Public financial intermediaries deliver credit to the public and private sectors in fixed proportions, which are based on the 1987 portfolio shares. The external sector is determined on the assumption that there is external credit rationing and a government target for its own external borrowing. The external credit rationing takes the form of fixed external debt ratios for the private sector and financial system. The government sets targets for its external borrowing and reserve accumulation at the central bank, and external flows are determined accordingly. The model is closed by endogenously determining the government deficit on the basis of available financing. The inflation rate and interest rate are set exogenously, which then implies a given financeable government deficit. This can be thought of as determining the consistency of the fiscal deficit with macroeconomic targets for interest rates and inflation. Table 7 shows a simulation of the model which calculates the deficit consistent with roughly unchanged inflation and real interest rates over 1988-92. We assume the 28.5 percent inflation that took place during 1988 is reduced to 24 percent in 1989, then continues at this rate for the rest of the period. Real interest rates remain constant at their 1988 levels, which implies that the internal debt of the government will stay roughly constant relative to GDP. The external debt ratio also is assumed to stay constant. The resulting financing supplied to the government amounts to 4.8 percent of GDP in 1988, then declines to 4.4 percent of GDP in 1989-92.14 After projecting the development lending flows, this implies net financing for the 14.The decline in financing is because of the reduction in inflation.  -16- consolidated public sector plus central bank of 3.7 percent in 1988 and 3.5 percent for the rest of the period. After allowing for the net financing of the central bank and the residual between the financing definition of the public deficit and the conventional definition, a conventional nonfinancial public deficit of 2.7 percent of GDP is estimated for 1988. This declines slightly to 2.4 percent of GDP over 1989-1992. As shown in the table, this implies a primary surplus of 1.3 percent of GDP, as compared to a primary surplus of 2.0 percent in 1987. However, this outcome is not the most desirable because of the high real interest rates that are required to maintain this financing level. Table 7 shows the estimate of the model that the real lending interest rate necessa.., to finance the deficit in 1988 was 15.7 percent, an increase of 1.9 percentage points over 1987. Since controls were in place during part of 1988, this can be interpreted as the *shadow" or market-clearing interest rate. The model implies that it is necessary for interest rates to remain at this high level to finance the projected nonfinancial public deficit of 2.4 percent over 1989-92. This is likely to be inconsistent with the ccatinuing revival of private investment necessary to support growth. This projection also supposes that the current structure of interest rates remains in place. Thus, real interest rates on forced investments continue to be negative, while the interest rate on development credits remains below the rate for commercial loans. This implies that the benefits of the high real interest rate in attracting financial savings is not fully realized, since real deposit rates are six percentage points below loan rates. We thus consider an alternative simulation -- shown in table 8 -- in which real loan interest rates will be steadily reduced, with a total reduction of 9 percentage points over the period. This is accomplished in part by changing the interest rate structure through increasing the real  -17- interest rate paid on forced investments and by moving the interest rate on development credits towards the rate on commercial loans. Thus, the spread between d&posit and loan --ter is reduced and real deposit rates fall only 6 percentage points. The fall in real interest rates reduces the domestic debt financing to the government. This reduces the total financeable deficit to 1.9 percent in 1989 and 1.6 percent by 1991. However, the financeable deficit increases again to 2.1 percent of GDP in 1992 (and following years) after interest rates stabilize at the lower level. It is necessary for the deficit to decrease more in the short run than in the long run because the one-time portfolio shift reduces the domestic financing available during the transition. The change in interest rate structure also affects central bank financing requirements. There are two offsetting effects. On one hand, the increase in rates paid on forced investments without a compensating increase in development credit interest rates increase central bank losses. On the other hand, the reduction in overall interest rates lower interest costs of central bank bonds and reduce central bank financing requirements. The net effect is to leave the financing requirement roughly unchanged. The scenario in Table 8 still has the shortcoming of continued high inflation of 24 percent. Therefore, in Table 9 we present a simulation in which inflation is reduced by nearly 14 percentage points over 1989-91, stabilizing at a rate of 10.6 percent in 1992. The same real interest rate decline and change in structure that held in the previous simulation is assumed here. The reduction in inflation requires a reduction in money creation that reduces the net financing of the total public sector from 4.8 percent of GDP in 1988 to 2.5 per cent of GDP in 1991, increasing again to 2.9 percent of GDP in 1992. The decline in nominal interest rates reduce central bank financing requirements from 0.8 percent to 0.4 percent, while the nominal  -18- flow of development lending also falls. The conventional financeable deficit must be temporarily reduced to 1.3 percent of GDP in 1991, after which it stabilizes at a long run level of 1.7 percent of GDP in 1992 and after. Interpretation of results The limitations of this kind of model should be well understood. Although the model can capture the transitions from one financial equilibrium to another, it does not include other short-term shocks that perturb financial markets and the general price level (bad harvests, financial panics, etc.). Thus, interest rates or inflation may move in the short run for many other reasons than those in the model. However, the model is useful to illustrate the fundamentals that determine interest rates and inflation in the absence of short-term disturbances. The exact magnitudes calculated for required deficit reduction should also be interpreted cautiously, since they depend on many parameters whose values can only be approximated. The deficit reductions appear to be quite modest in view of the significant decreases in real interest rates and inflation in the simulation. This reflects the low elasticities with respect to interest rates of investment and financial asset demands implied by the results of table 6. A reduction in interest rates thus does not lead private credit demand to increase much, nor private financial savings to decrease greatly. This result is crucial to the results and thus would bear further study. The results are also sensitive to the projected growth rate (4.5? over 1989-92 in the current simulation). A lower growth rate would decrease the financing available for a given debt ratio, and thus require a greater deficit reduction. For example, if growth were to only be 2? over the period, then the deficit in the simulation of reduced interest rates and inflation  -19- would have to be reduced to 0.62 of GDP by 1991, as compared to 1.32 with the higher growth rate. Conclusions The fiscal adjustment during 1985-87 was impressive compared to the previous large fiscal deficits. It was due to structural changes in acal policy and not simply to fortuitous events such as the coffee boom. Although losses of public financial institutions were important in some other Latin American countries, there is no direct evidence that they were a major factor in Colombia. A small quasi-fiscal loss of Banco de la Reptiblica and other other public financial institutions is suggested by the data, but it was not the dominant factor in fiscal behavior. Although the fiscal adjustment was impressive, it fell short of actually improving the net financial position of the government. Total public debt as a percent of GDP was roughly unchanged from its 1984 value at the end of 1987, even when we correct for the effect of currency devaluation on dollar-denominated instruments. Public development lending as a ratio to GDP fell slightly from 1984 to 1987. Perhaps in part because of this public debt behavior, real interest rates remained very high, while inflation accelerated slightly. To build upon the adjustment achieved thus far would likely require reductions in interest rates and inflation. The model presented attempts to calculate the fiscal deficit reductions that would be necessary to achieve this objective. The results of the simulations suggest the difficult challenges faced by fiscal policy in the years ahead. To reduce interest rates to more manageable levels would require continued reduction in the fiscal deficit, below levels currently envisioned. To also attain the laudable goal of inflation reduction would require even tighter fiscal policy. Hlowever, the  -20- deficit magnitudes suggested do not seem out of reach, even if we allow for uncertainty as to the exact figures. This suggests that in additlon to the commendable efforts shown thus far, continued policy initiative would be very helpful in confronting the fiscal challenges of the 1990's in Colombia.  - 21 - TABLE 1: STRUCTURAL TRENDS IN FISCAL POLICY Proj Proj Proj Percent of CDP (- d.ficit/-surplus) 1964 1965 190t 1907 19U 1969 1990 Total consolidated public sector deficit 6.76 .15U -0.10 1.60 2.90 2.91 2.30 corrected for: FNC -0.41 1.80 -.161 0." 0.85 0.10 0.82 ECOPETROL -0.15 1.10 0.24 -0.98 0.t4 -0.09 -0.40 National govprimsnt--t4mporary iteos\1 0.09 0.11 0.44 0.70 0.10 0.18 0.18 Structural deficit 7.42 8.95 8.22 2.6e 2.86 2.72 2.51 corrected for: External Interest 1.68 1.94 2.27 2.09 8.01 8.42 8.48 Domeotic interest 0.77 O.9 0.70 0.9S 0.99 0.90 0.61 Structural prima-y deficit 5.02 1.02 0.20 -1.10 -1.66 -1.60 -1.78 corrected for: Fixed capital formtion t.73 *.27 6.40 5.62 5.99 7.91 *.17. Structural primary currwet detIcit -8.76 -7.26 43.20 -S."2 -7.61 -9.51 -9.9 \1 Includes coffee tax (2.51), ECOPETROL trannfere and bockpeymnt of duties. FWC transtfr., Decreto 899-1966, and special revenue from the tax amnesty (In 1907).  - 22 - TABLE 2: Cuenta Especial de Cambios--Banco de la Republica till Juno PERCENT OF GDP 1984 1985 1986 1987 1988 Net income from: 1.38 0.15 0.79 1.53 1.29 foreign exchange reserves (net) 0.13 0.09 0.22 0.42 0.34 interest earnings 0.26 0.25 0.41 0.54 0.57 interest payments 0.13 0.16 0.18 0.12 0.23 other net reserve income 0.44 0.01 0.21 0.29 0.04 gold 0.43 0.02 0.11 0.29 0.03 other 0.01 -0.00 0.09 0.00 0.01 purchase and sale of foreign exchange 0.77 -0.01 0.29 0.79 0.93 exchange rate differentials-external credit 0.04 0.04 0.07 0.03 -0.02 Expenditure on: 0.78 0.46 0.82 0.92 0.81 Titulos canjeables 0.30 0.30 0.72 0.83 0.73 Administration costs of CEC 0.07 0.05 0.04 0.04 0.04 Reserves-exchange rate differentials 0.40 0.11 0.06 0.06 0.04 Net balance 0.60 -0.32 -0.03 0.61 0.48 Net balance excl foreign exchange transaction -0.17 -0.31 -0.32 -0.18 -0.46 Source: Contraloria de la Republica, Infonme Financiero  - 23 - Tablo 8: Interest ratee for Fondoo Financieroo and hypothetical operating losses 1985 1966 1987 1968 (Noveber) Interest rates paid on crodit rooources (percent) FFAP 16.7 16.2 19.6 22.a FIP 22.7 21.1 24.a 28.8 FFI 29.7 21.7 22.0 24.4 FCE 24.0 21.8 24.0 26.4 Level of crodit resources by Fond. (percent of CDP) FFAP 1.70 1.65 1.48 1.21 FIP 0.15 0.18 0.12 0.10 FFI 0.10 0.07 0.04 0.03 FCE NA 0.04 0.18 0.17 Average Interost rate for four Fondoo 16.6 20.6 23.0 Interest rates recolved on loans by Fondo. FFAP 16.6 19.8 19.1 19.0 FIP 25.1 24.5 26.9 26.9 FFI 22.0 22.1 23.4 24.7 FCE 10.8 19.0 28.8 25.O PROEXPO 22.0 22.0 22.0 NA Loans by Fondo (percent of COP) FFAP 1.69 1.58 1.41 1.16 FIP 0.15 0.12 0.11 0. ;^ FFI 0.17 0.12 0.10 0.0 FCE NA 0.18 0.15 0.1 Average interest rate on ionding by Fondon 19.0 20.2 20.4 (except PROEXPO) Hypothotical operating profit (+)/ lose (-) of Fondoo Financieroo (percent of GDP)\ FFAP -0.15 -0.18 -0.18 -0.18 FIP -0.01 -0.01 -0.01 -0.01 FFI -0.02 0.00 0.01 0.00 FCE NA 0.01 -0.02 -0.02 Total four Fondoc -0.12 -0.14 -0.17  - 24 - TABLE 4: Not finnelng flow of consolidatod central bank and nonfinancial pubilc s*ctor--Inf ltion adjusted (perceont of GO) 19N 19#O 197 Currency hold by public 1.1 0.7 1.2 S Ignorag. 0.2 -0.1 0.3 Inflation tox 0.9 0.9 0.0 Reserves 0.6 1.0 0.7 $*1gnorage 0.1 0.6 0.2 Inflation tax 0.5 0.5 0.6 Forced Investments 0.4 0.0 -0.1 by private financial system 0. 0.0 -0.1 by public financial institutions 0.1 0.0 0.0 Bonds of nonfinancial private s*eor 1.0 0.4 -0.1 Lending by private financial system -0.4 0.0 0.4 Lending by public financial system 0.7 1.8 0.3 Total domsetic finance 3.4 4.3 2.8 Exter eal lending 3.6 8.0 -0.9 Capital losses -) -10.4 -7.6 -7.7 Nominal change 14.0 10.5 6.8 Forel n exchange resrves -0.