Abstract
MOST ANALYSES OF sovereign external debt assume that the capacity to raise foreign exchange revenue is the binding constraint in the repayment decision.1 Alternatively, since the repayment of foreign debt—most of which is owed by the government—is financed by the transfer of resources from the private to the public sector, several studies have focused on the fiscal constraint.2 This paper extends the fiscal approach. By viewing changes in payments to the different classes of creditors as signals of changes in seniority, we argue that not only the level but also the distribution of the primary fiscal balance between domestic and external creditors determines repayment and debt values.
MOST ANALYSES OF sovereign external debt assume that the capacity to raise foreign exchange revenue is the binding constraint in the repayment decision.1 Alternatively, since the repayment of foreign debt—most of which is owed by the government—is financed by the transfer of resources from the private to the public sector, several studies have focused on the fiscal constraint.2 This paper extends the fiscal approach. By viewing changes in payments to the different classes of creditors as signals of changes in seniority, we argue that not only the level but also the distribution of the primary fiscal balance between domestic and external creditors determines repayment and debt values.
A new aggregation scheme is used to measure the distribution of government financing between foreign and domestic creditors, and the influence of the level and financing of the primary fiscal surplus on the market value of foreign loans is estimated. According to the empirical results, the profile of the sources of fiscal financing influenced external debt values from 1985 to 1989, suggesting that the decline in prices during this period (Table 1) reflected a fundamental improvement in the seniority of domestic debt at the expense of foreign bank debt. We argue that this change in relative creditor standing was the cost-minimizing response of governments to internal and external capital flight that drained the domestic financial “tax base” subject to indirect taxation. A secondary objective of the paper is to show that previous studies have neglected an important reason for the decline in loan values from 1985 to 1989: the increase in market interest rates.
I. A Simple Model
Our analysis starts from the conventional assumption that the aggregate market value of claims on a government depends on the expected present value of payments to all creditors. Because debt is financed by the transfer of resources from the private to the public sector, the average value of all classes of debt depends on the expected present value of the fiscal surplus. But prices and values of different classes of credits also depend on their expected seniority of payment, and one of the interesting problems associated with sovereign debt is that the contractual structure of the debt does not provide a clear basis for seniority. Given the lack of clarity, the debtor government can treat a class of creditors as senior by making payments to that creditor at the expense of other creditors. The government can even borrow from a junior creditor to pay a senior creditor if it has the ability to coerce the junior creditor. This, of course, generates strong incentives for the junior creditor to escape the coercive relationship with the debtor government. Since the debtor government will discriminate among creditors to minimize the present value of the costs of debt service, it makes sense to aggregate debt according to broad creditor groups.
Secondary-Market Prices of External Debt
Secondary-Market Prices of External Debt
| 1985 | 1986 | 1987 | 1988 | 1989 | |
|---|---|---|---|---|---|
| Brazil | 77.2 | 75.6 | 56.6 | 47.8 | 30.3 |
| Mexico | 72.9 | 59.1 | 54.5 | 48.6 | 39.8 |
| Argentina | … | 65.7 | 50.8 | 25.6 | 16.2 |
| Venezuela | 81.2 | 76.9 | 66.3 | 51.4 | 37.2 |
| Philippines | … | 64.5 | 64.8 | 52.1 | 47.4 |
| Algeria | … | 95.2 | 95.2 | 90.5 | 76.1 |
| Chile | 67.4 | 67.7 | 64.6 | 59.5 | 60.7 |
| Nigeria | … | 50.5 | 31.5 | 27.5 | 24.7 |
| Peru | 36.9 | 20.5 | 13.5 | 6.4 | 4.8 |
| Colombia | 83.0 | 84.9 | 81.2 | 64.7 | 59.4 |
| Ecuador | 68.1 | 66.2 | 48.2 | 25.5 | 14.0 |
| Uruguay | … | 64.6 | 69.5 | 60.3 | 55.8 |
| Zaire | … | 25.1 | 25.1 | 21.4 | 19.7 |
| Costa Rica | … | 47.1 | 31.0 | 13.9 | 15.1 |
| Sudan | … | 11.6 | 6.2 | 5.4 | 2.3 |
| Zambia | … | 17.3 | 19.4 | 19.6 | 20.5 |
| Bolivia | … | 7.0 | 10.3 | 11.3 | 11.1 |
| Average | 69.5 | 52.9 | 46.4 | 37.2 | 31.5 |
Secondary-Market Prices of External Debt
| 1985 | 1986 | 1987 | 1988 | 1989 | |
|---|---|---|---|---|---|
| Brazil | 77.2 | 75.6 | 56.6 | 47.8 | 30.3 |
| Mexico | 72.9 | 59.1 | 54.5 | 48.6 | 39.8 |
| Argentina | … | 65.7 | 50.8 | 25.6 | 16.2 |
| Venezuela | 81.2 | 76.9 | 66.3 | 51.4 | 37.2 |
| Philippines | … | 64.5 | 64.8 | 52.1 | 47.4 |
| Algeria | … | 95.2 | 95.2 | 90.5 | 76.1 |
| Chile | 67.4 | 67.7 | 64.6 | 59.5 | 60.7 |
| Nigeria | … | 50.5 | 31.5 | 27.5 | 24.7 |
| Peru | 36.9 | 20.5 | 13.5 | 6.4 | 4.8 |
| Colombia | 83.0 | 84.9 | 81.2 | 64.7 | 59.4 |
| Ecuador | 68.1 | 66.2 | 48.2 | 25.5 | 14.0 |
| Uruguay | … | 64.6 | 69.5 | 60.3 | 55.8 |
| Zaire | … | 25.1 | 25.1 | 21.4 | 19.7 |
| Costa Rica | … | 47.1 | 31.0 | 13.9 | 15.1 |
| Sudan | … | 11.6 | 6.2 | 5.4 | 2.3 |
| Zambia | … | 17.3 | 19.4 | 19.6 | 20.5 |
| Bolivia | … | 7.0 | 10.3 | 11.3 | 11.1 |
| Average | 69.5 | 52.9 | 46.4 | 37.2 | 31.5 |
In general, then, the market values of classes of credits reflect both an optimal intertemporal strategy by debtor governments toward different creditors and the endogenous reactions of creditors to the government’s choices. Although an enormous theoretical literature on enforcement technology is available to foreign creditors, scant attention has been paid to the response of domestic creditors or to the relative standing of the two major classes of creditors.3
A simple static framework illustrates the financing problem faced by governments.4 The government budget constraint is
where G is fixed government expenditure, τ is the tax rate, Y is income, D is domestic debt, and F is foreign debt. For the reasons discussed above, we assume that total financing costs (interest and non interest) of domestic and external financing (Cd and Cf)are increasing and decreasing, respectively, as functions of the proportion of domestic financing(d):
Determining the exact shape of the Cd and Cf functions would be possible only with accurate data concerning the costs and amounts of financing. However, as we feel confident that the assumed signs of the first derivatives are reasonable, total financing costs become
The government chooses the mix of financing that minimizes total financing costs.
The Cd curve shown in Figure 1 is zero at the origin, where the primary balance is financed only by external debt, and increases to d = 1 (its maximum), where the deficit is financed entirely domestically. An analogous relationship holds for Cf reading leftward from the d =1 vertical line. Total financing costs are minimized at the trough of the C curve, where the marginal cost of an extra unit of each financing source is equal.
Now consider a reduction in income, which, given fixed government expenditure (G), necessitates a higher level of financing. Figure 2 shows two sets of financing cost curves for income levels Y0 and Y1 where
In a dynamic setting, the government attempts to minimize the expected present value of the costs, as represented in Figure 2. The interesting difference between the two markets for financing is that access to the international market was lost almost entirely in 1982 for most debtor countries considered here (see Table 1). In terms of Figure 2, the Cf(d) shifts to the right immediately because new external debt carries a market-clearing interest rate approximated by the risk-free rate divided by the market price of existing debt. For our sample of countries, this market rate was two to ten times the risk-free rate. Because the cost of foreign borrowing had, in practical terms, become infinite, governments turned to domestic credit. Access to domestic credit markets eroded slowly during the 1980s, not because residents were slow to learn or were misinformed but because their ability to respond was constrained by capital controls and other administrative restrictions. According to the measures of capital flight reported below, residents succeeded over time in replacing their government’s liabilities with foreign assets, as an outlet for their savings. This process takes the form of a gradual upward shift in Cd(d) over time. Our conjecture is that debtor governments had nearly exhausted their domestic market power by 1987 so that the yield on new internal debt approached the yield on existing external debt. The associated rise in debt service costs was very rapid given the very short maturity of most internal debt. The costs of continuing to tap the internal market inevitably exceeded the costs of improving the primary balance. Moreover, the expected need for access to some credit market in order to smooth unexpected changes in the primary balance may have favored reestablishing a payments record in the domestic market.
Total Cost-Minimizing Shares of Domestic and External Financing
Citation: IMF Staff Papers 1993, 001; 10.5089/9781451947137.024.A006
Total Cost-Minimizing Financing Under Different Income Levels
Citation: IMF Staff Papers 1993, 001; 10.5089/9781451947137.024.A006
The importance of a shifting profile of creditor seniority could be tested by comparing the relative rates of return on domestic versus external debt for the 17 sample countries.5 Such an analysis is ruled out, however, by the difficulty of gauging the real after-tax returns on domestic assets under a high-inflation and high-indirect-taxation regime of financial repression. Instead, we estimate the influence of the primary fiscal deficit and its allocation between domestic and external creditors on secondary-market prices of external loans.
Even after including indicators of internal financing, country-specific indicators do not explain the fall in loan prices: that is, prices early in the sample period are underfitted, while 1988 and 1989 prices are overfitted. Several studies have exploited the correlation between the portion of loan prices unexplained by country-specific indicators and measures of creditor behavior—which do not vary over countries and increase over the sample period—to support the view that creditor behavior is an important determinant of loan prices.6 If so, if prices in secondary markets reflect the regulatory and reserve positions of banks, then it seems clear that debtor countries could benefit from restructuring agreements that exploit these market imperfections.
We offer an alternative explanation for the inability of country-specific fundamentals to explain fully the fall in loan prices. From 1986 to 1989, the LIBOR (the London interbank offered rate) annual average rose from 6.86 percent to 9.28 percent, implying that the value of any payment stream would have been decreasing. Moreover, the empirical evidence supports the view that the expected partial payments on floating-rate debt are unaffected by changes in market interest rates, so that the market value of both fixed-rate and floating-rate sovereign debt is affected by changes in the rate at which expected payments are discounted. Including LIBOR in the regression improves the fit of the model and smooths out the intertemporal pattern of residuals. It follows that verification of hypotheses concerning creditor behavior should be based on indicators that have explanatory power over and above that of marketwide interest rates.
II. Measuring Domestic Financial Repression
A debtor government can make net payments to nonresident creditors if it can capture domestic resources through taxation or net borrowing from residents. If taxation is used, the resident gets a receipt. If borrowing is used, the resident gets some type of financial government liability. If the government subsequently acts to reduce the value of its liability, the borrowing becomes an alternative form of taxation. At the limit, a tax receipt is the same as a defaulted government bond. The most obvious form of this sort of taxation is the inflation tax on the government’s monetary liabilities, although in many cases this is not the most important form of taxation. Financial repression can be defined as any regulatory mechanism that limits residents’ access to investments other than their government’s liabilities. In the case of developing countries, an important alternative for residents has been foreign-currency-denominated assets in offshore markets. For a variety of reasons, developing countries have limited residents’ access to such markets through legal restrictions. In addition, registered financial intermediation in domestic markets is often a government monopoly. In return for the right to engage in financial intermediation, banks and other financial intermediaries have been forced to hold government liabilities at below market yields.
Because such a tax base depends on a system of legal restrictions, it is very difficult to quantify. Moreover, as the government imposes higher tax rates on this base, residents can be expected to intensify efforts to avoid the tax by reducing holdings of money and other government liabilities and by evading taxation by acquiring foreign financial assets and unregulated domestic assets. The government’s response usually involves more controls as well as redesigned government securities. The shifting legal structure and the response of governments makes time series of financial aggregates difficult to interpret.
The tax rate relevant for the domestic liabilities of a debtor government is also difficult to quantify. Clearly, any government liability that pays a rate of interest not indexed to the rate of inflation (or the exchange rate) is potentially taxable through inflation. Measures of the maximum steady-state revenue from inflation are useful but do not capture the essence of the problem for most debtor countries. The government can impose a 100 percent tax on any of its nonindexed liabilities, including wages and pensions, through a sufficiently sudden and unexpected hyperinflation. To the extent that the government controls the lending decisions of commercial banks and imposes restrictions on deposit yields, one can think of the liabilities of the banking system as being taxed in order to support government expenditures in the form of subsidies to favored borrowers. Finally, government reserve requirements for banks are another means of implicit financial taxation.
Although direct measures of government financing from financial repression are not available, inflation (measured using the consumer price index) can serve as a gauge of the tax rate on money holdings. After the advent of external debt servicing difficulties in 1982, the inflation tax rose for most of the sample countries. Even after excluding the debtors that experienced hyperinflation—Brazil, Argentina, Peru, and Uruguay—the average rate of inflation rose from 25 percent in 1982, to 38 percent in 1987, to 45 percent during the last two years of the sample interval (Figure 3).
By the mid-1980s, the low rates of return on controlled domestic assets induced residents of developing countries to incur the costs of transferring capital to offshore financial intermediaries. The methodology of Dooley (1988) provides estimates of the stock of resident capital invested abroad to avoid the control of domestic authorities. The doubling of the stock of flight capital for the sample countries over the five years after 1982 (Figure 4) suggests that financial repression intensified as indebted country governments paid an increasing share of their available resources to foreign creditors. For most countries, capital flight increased from the early 1980s to 1987 or 1988 and then leveled off through 1989. Mexico experienced a reversal beginning in 1987, suggesting that the process of domestic reintermediation began before the end of the sample interval. These data are utilized in more formal econometric tests below.
Average Inflation a
Citation: IMF Staff Papers 1993, 001; 10.5089/9781451947137.024.A006
Source: IMF, International Financial Statistics.a Inflation is measured using the consumer price index. The 17 debtor countries, except Argentina, Brazil, Peru, and Bolivia, are shown.Stock of Flight Capitala
Citation: IMF Staff Papers 1993, 001; 10.5089/9781451947137.024.A006
Source: IMF World Economic Outlook data base.a Total for 17 debtor countries.III. A Flow Measure of External and Domestic Creditor Seniority
In the middle-income debtor countries, the domestic debt has become the dominant competitor with the external debt for financing. However, it is often difficult to construct data for domestic contractual interest obligations and for new credits, particularly in high-inflation countries. Moreover, the distinction between money and other highly liquid government obligations is often difficult to maintain in practice. This inability to directly measure the taxation of resident creditors limits empirical study of government treatment of broad classes of creditors. To overcome this problem, we develop a “flow measure” of payments to different classes of creditors. This measure, in combination with observable external debt prices, allows empirical inference concerning the relative standing of domestic and external creditors. We develop an accounting system in which net government payments to domestic asset holders is measured as a residual.
The starting point for the analysis is a measure of the debtor government’s primary fiscal surplus, which is defined as the central government’s noninterest receipts less its noninterest expenditures. It is therefore the net amount available for distribution to all domestic and foreign creditors.7 Because we have reasonably reliable data on contractual interest obligations to nonresidents and on new credits from nonresidents, net payments to these groups can be identified. By subtracting these amounts from the primary fiscal surplus, we find payments net of new borrowing from residents of the debtor country.8 Thus, we treat net payments to holders of domestic debt, including money, as the residual in the accounting system. For example, if the primary fiscal surplus in a given year was less than the net payments to external creditors, residents must have acquired government debt, including money, in order to finance debt service payments to nonresidents.
The net payment to a creditor in a given time period does not, in itself, say anything about how different creditors expect to be treated in the future. If we observe, for example, that domestic creditors make new loans to their government which more than cover the interest due on existing debt—while nonresidents make no new loans and receive full interest on existing debt—we cannot conclude that foreign creditors will always receive payments at the expense of residents. To the contrary, if neither creditor can be coerced by the debtor, only the senior creditor will make a new loan if there is any doubt about the government’s capacity to pay. In this case, domestic creditors might grant new loans because they believe they will have first claim on future fiscal surpluses.
Nevertheless, it is reasonable to suppose that over time residents would revise their expectations about their status relative to nonresident creditors as governments continued to make large net payments to nonresidents that are financed by domestic borrowing. Moreover, capital flight, accelerating domestic inflation, and occasional outright default on domestic debt suggest that the pattern of payments during the time period studied reflects coercion of residents rather than expected seniority.
The level and distribution of the primary balance aggregated over 17 indebted countries (in U.S. dollars) is shown in Figure 5. The decline in the accumulation of net government claims on domestic residents from 1985 to 1987, combined with the sharp fall in the primary surplus, reduced the amount paid to foreigners from $50 billion to $24 billion over the two-year period. We interpret these payment streams as signalling a change in relative seniority between domestic and foreign creditors, and note that the decline in payments to foreigners may provide an explanation for the pervasive fall in external debt prices in 1987.
IV. Creditor Seniority as a Fundamental Determinant of Debt Values
The empirical analysis in this section has two objectives. First, an estimation of the impact of domestic and external creditor financing of the fiscal gap on external debt prices, controlling for other key macro-economic determinants of debt values, serves as a rough and ready test of the importance of creditor seniority. However, even after accounting for the fiscal constraint, there are certain regularities in the model’s residuals, suggesting an additional omitted variable. We offer an alternative explanation to that of other studies, which have used creditor behavior to explain loan price movements after controlling for country-specific factors.
Primary Balance and Sources of Financinga
Citation: IMF Staff Papers 1993, 001; 10.5089/9781451947137.024.A006
Sources: Data are from country sources and IMF country desk officers.a Total for 17 debtor countries.Regression Results, 17 Countries from 1985 or 1986 to 1989 Dependent variable: Log of annual market price for loans
Note: T-statistics are reported in parentheses. Standard errors are adjusted for heteroscedasticity.
Regression Results, 17 Countries from 1985 or 1986 to 1989 Dependent variable: Log of annual market price for loans
| A | B | C | D | E | F | G | |
|---|---|---|---|---|---|---|---|
| Constant | 4.24 | 4.30 | 4.20 | 4.03 | 4.22 | 4.30 | 5.56 |
| (41.8) | (40.6) | (32.7) | (32.4) | (14.5) | (14.0) | (12.0) | |
| Debt exports | -0.00122 | -0.00092 | -0.00072 | -0.00076 | -0.00109 | -0.00133 | -0.0095 |
| (3.36) | (2.23) | (1.87) | (2.32) | (1.59) | (1.56) | (2.75) | |
| Imports/reserves | -0.00013 | -0.00024 | -0.00026 | -0.00013 | -0.00037 | -0.00035 | -0.00002 |
| (0.85) | (1.56) | (1.48) | (0.96) | (1.83) | (1.68) | (0.14) | |
| Arrears/debt | -0.06088 | -0.04374 | -0.04485 | -0.04293 | -0.05457 | -0.05447 | -0.04133 |
| (4.15) | (2.66) | (3.12) | (3.34) | (5.22) | (4.84) | (2.92) | |
| CPI inflation | … | -0.00019 | -0.00018 | -0.00014 | -0.00009 | -0.00007 | -0.00005 |
| (1.97) | (2.10) | (1.65) | (1.33) | (0.86) | (0.51) | ||
| Capital flight/GDP | … | -0.50112 | -0.40801 | -0.57770 | -0.12970 | -0.28682 | -0.56624 |
| (2.04) | (1.71) | (2.55) | (0.44) | (0.81) | (2.74) | ||
| Primary balance/GDP | … | … | 0.02917 | 0.05858 | 0.06802 | … | 0.06335 |
| (1.35) | (2.80) | (3.06) | (3.34) | ||||
| Payment to domestic | … | … | … | -0.03879 | -0.02654 | -0.04070 | -0.03833 |
| creditors/GDP | (4.21) | (1.26) | (2.27) | (4.18) | |||
| Primary and central | … | … | … | … | … | 0.05398 | … |
| bank balance /GDP | (2.27) | ||||||
| LIBOR | … | … | … | … | … | … | -0.02035 |
| (3.35) | |||||||
| Asia | 0.22787 | 0.33267 | 0.36369 | 0.16378 | … | … | 0.10437 |
| Africa | 0.06714 | 0.24342 | 0.27000 | 0.41587 | … | … | 0.36558 |
| R-squared | 0.61 | 0.64 | 0.65 | 0.69 | 0.59 | 0.57 | 0.73 |
| N | 75 | 75 | 75 | 75 | 55 | 55 | 75 |
Note: T-statistics are reported in parentheses. Standard errors are adjusted for heteroscedasticity.
Regression Results, 17 Countries from 1985 or 1986 to 1989 Dependent variable: Log of annual market price for loans
| A | B | C | D | E | F | G | |
|---|---|---|---|---|---|---|---|
| Constant | 4.24 | 4.30 | 4.20 | 4.03 | 4.22 | 4.30 | 5.56 |
| (41.8) | (40.6) | (32.7) | (32.4) | (14.5) | (14.0) | (12.0) | |
| Debt exports | -0.00122 | -0.00092 | -0.00072 | -0.00076 | -0.00109 | -0.00133 | -0.0095 |
| (3.36) | (2.23) | (1.87) | (2.32) | (1.59) | (1.56) | (2.75) | |
| Imports/reserves | -0.00013 | -0.00024 | -0.00026 | -0.00013 | -0.00037 | -0.00035 | -0.00002 |
| (0.85) | (1.56) | (1.48) | (0.96) | (1.83) | (1.68) | (0.14) | |
| Arrears/debt | -0.06088 | -0.04374 | -0.04485 | -0.04293 | -0.05457 | -0.05447 | -0.04133 |
| (4.15) | (2.66) | (3.12) | (3.34) | (5.22) | (4.84) | (2.92) | |
| CPI inflation | … | -0.00019 | -0.00018 | -0.00014 | -0.00009 | -0.00007 | -0.00005 |
| (1.97) | (2.10) | (1.65) | (1.33) | (0.86) | (0.51) | ||
| Capital flight/GDP | … | -0.50112 | -0.40801 | -0.57770 | -0.12970 | -0.28682 | -0.56624 |
| (2.04) | (1.71) | (2.55) | (0.44) | (0.81) | (2.74) | ||
| Primary balance/GDP | … | … | 0.02917 | 0.05858 | 0.06802 | … | 0.06335 |
| (1.35) | (2.80) | (3.06) | (3.34) | ||||
| Payment to domestic | … | … | … | -0.03879 | -0.02654 | -0.04070 | -0.03833 |
| creditors/GDP | (4.21) | (1.26) | (2.27) | (4.18) | |||
| Primary and central | … | … | … | … | … | 0.05398 | … |
| bank balance /GDP | (2.27) | ||||||
| LIBOR | … | … | … | … | … | … | -0.02035 |
| (3.35) | |||||||
| Asia | 0.22787 | 0.33267 | 0.36369 | 0.16378 | … | … | 0.10437 |
| Africa | 0.06714 | 0.24342 | 0.27000 | 0.41587 | … | … | 0.36558 |
| R-squared | 0.61 | 0.64 | 0.65 | 0.69 | 0.59 | 0.57 | 0.73 |
| N | 75 | 75 | 75 | 75 | 55 | 55 | 75 |
Note: T-statistics are reported in parentheses. Standard errors are adjusted for heteroscedasticity.
Virtually all studies of sovereign creditworthiness, particularly those of secondary loan prices, choose from a standard list of debt payment capacity indicators related to the external balance constraint, such as the debt service-to-export ratio, the reserve-to-import ratio, and the debt-to-GDP ratio.9 However, Stone (1991b) shows that published credit ratings have predictive power over and above that of the standard balance of payments fundamentals, which suggests that loan market participants have systematically used country-specific fundamentals that have not been captured in empirical studies.
The hypothesis that the expected status of bank loans relative to other credits is an important omitted fundamental is examined in the regression results presented in Table 2. Three of the standard indicators of debt payment capacity are included in the specification reported in column A. The importance of indirect taxation of domestic assets is examined in column B. Both indicators of financial repression—CPI inflation and the stock of flight capital—appear to have a significant impact on debt prices, even after controlling for external indicators of creditworthiness.
The importance of payments to all creditors is examined in the third specification reported in column C. Interestingly, the coefficient for our measure of the total amount paid to government creditors, the primary balance, enters the model with a positive sign but is not significant at conventional levels. The key regression results are presented in column D, where both the level of the primary surplus and the distribution of funds to creditors—measured here by the amount paid to domestic creditors10—have a significant impact on the market values of external debt. Furthermore, the t-statistics for the inflation and capital flight estimates indicate that market participants jointly consider financial repression and the size and distribution of payments to creditors.
The measure of government financing needs is extended to include central bank losses but not profits, since the latter are given to the central government (see Robinson and Stella (1987)). The average across the 55 available observations (all but the African and Asian countries) of central bank losses was —3.0 percent of GDP. Column E of Table 2 indicates that the key parameter estimates (those for the primary balance and payment to domestic creditors) for the subsample for which central bank losses are available are broadly similar to those for the full sample. Column F reports the impact of the broader measures of government financing needs on debt prices. The sum of central bank losses and the primary balance does not improve on the explanatory power of the latter alone, suggesting that either central bank losses are difficult to estimate or the financing of these losses is so uncertain as to obfuscate their impact on debt prices.
The inability of country-specific measures to explain the drop in loan prices has provided scope for empirical verification of an alternative view of debt price determination. Correlations between indicators of creditor behavior and debt prices have been used to support the importance of creditor behavior. Examples of such measures include proxies for loan- loss reserves (either a dummy variable for the periods of reserve changes or an aggregate measure of U.S. bank reserves for all countries),11 as well as the amount of debt concentrated in large banks12 and bank capitalization.13 These results call into question the efficacy of using secondary- market loan prices in debt restructuring since participants could benefit from agreements that exploit these market imperfections. Except for the dummy variables, these measures of creditor behavior increase steadily over time and do not vary across countries; hence, given the pattern of residuals shown in Table 3, they will have explanatory power. Of course, it would be impossible empirically to distinguish the influence of these measures of creditor behavior from any other explanatory variable that increased over time and did not vary over countries.
We propose an alternative explanation for the tendency of the residuals to move from positive to negative values over the sample period for virtually all sample countries. However, in addition to sovereign risk, these prices will reflect interest rate risk, and, given the steady increase in the LIBOR from 1986 to 1989, the prices of loans across all countries would be expected to fall over the sample period, after controlling for country-specific fundamentals. Including LIBOR, as reported in column G of Table 2, not only improves the model’s fit, but, as shown in the bottom panel of Table 3 smooths out the intertemporal pattern of the regression residuals.
Regression Residuals
Regression Residuals
| Country | 1985 | 1986 | 1987 | 1988 | 1989 | Average |
|---|---|---|---|---|---|---|
| Specification D (without LIBOR) from Table 2 | ||||||
| Brazil | 0.52 | 0.66 | 0.57 | 0.08 | -0.05 | 0.35 |
| Mexico | 0.34 | 0.31 | 0.24 | -0.13 | -0.37 | 0.08 |
| Argentina | … | 0.58 | 0.67 | -0.09 | 0.17 | 0.33 |
| Venezuela | 0.14 | 0.34 | 0.47 | 0.20 | -0.55 | 0.12 |
| Philippines | … | 0.13 | -0.03 | -0.02 | -0.08 | 0.00 |
| Algeria | … | 0.31 | 0.32 | 0.40 | 0.14 | 0.29 |
| Chile | 0.09 | 0.02 | -0.24 | -0.23 | -0.26 | -0.12 |
| Nigeria | … | -0.05 | -0.55 | -0.50 | -0.84 | -0.49 |
| Peru | 0.24 | 0.05 | -0.05 | -0.22 | -0.15 | -0.03 |
| Colombia | 0.71 | 0.60 | 0.52 | 0.33 | 0.20 | 0.47 |
| Ecuador | 0.28 | 0.41 | 0.38 | -0.35 | -0.98 | -0.05 |
| Uruguay | … | 0.23 | 0.39 | 0.16 | -0.04 | 0.19 |
| Zaire | … | 0.03 | 0.28 | 0.09 | 0.19 | 0.15 |
| Costa Rica | … | 0.02 | -0.47 | -1.24 | -1.17 | -0.71 |
| Sudan | … | 0.07 | 0.11 | 0.35 | -0.28 | 0.06 |
| Zambia | … | 0.50 | -0.34 | -0.23 | 0.00 | -0.02 |
| Bolivia | … | -1.21 | -0.46 | -0.63 | -1.01 | -0.83 |
| Average | 0.33 | 0.18 | 0.11 | -0.12 | -0.30 | |
| Specification G (with LIBOR) from Table 2 | ||||||
| Brazil | 0.63 | 0.47 | 0.44 | 0.04 | 0.11 | 0.34 |
| Mexico | 0.42 | 0.12 | 0.11 | -0.18 | -0.16 | 0.06 |
| Argentina | … | 0.42 | 0.56 | -0.06 | 0.16 | 0.27 |
| Venezuela | 0.23 | 0.17 | 0.34 | 0.24 | -0.28 | 0.14 |
| Philippines | … | 0.00 | -0.16 | -0.02 | 0.18 | 0.00 |
| Algeria | … | 0.14 | 0.23 | 0.41 | 0.35 | 0.28 |
| Chile | 0.24 | -0.15 | -0.36 | -0.22 | 0.01 | -0.10 |
| Nigeria | … | -0.30 | -0.73 | -0.55 | -0.54 | -0.53 |
| Peru | 0.35 | -0.13 | -0.22 | -0.30 | -0.17 | -0.09 |
| Colombia | 0.84 | 0.43 | 0.40 | 0.36 | 0.49 | 0.50 |
| Ecuador | 0.36 | 0.21 | 0.25 | -0.36 | -0.72 | -0.05 |
| Uruguay | … | 0.05 | 0.27 | 0.20 | 0.26 | 0.19 |
| Zaire | … | -0.17 | 0.16 | 0.16 | 0.25 | 0.10 |
| Costa Rica | … | -0.16 | -0.61 | -1.20 | -0.87 | -0.71 |
| Sudan | … | -0.12 | 0.04 | 0.43 | 0.07 | 0.11 |
| Zambia | … | 0.38 | -0.42 | -0.17 | 0.38 | 0.04 |
| Bolivia | … | -1.35 | -0.52 | -0.53 | -0.64 | -0.76 |
| Average | 0.44 | 0.00 | -0.01 | -0.10 | -0.07 | |
Regression Residuals
| Country | 1985 | 1986 | 1987 | 1988 | 1989 | Average |
|---|---|---|---|---|---|---|
| Specification D (without LIBOR) from Table 2 | ||||||
| Brazil | 0.52 | 0.66 | 0.57 | 0.08 | -0.05 | 0.35 |
| Mexico | 0.34 | 0.31 | 0.24 | -0.13 | -0.37 | 0.08 |
| Argentina | … | 0.58 | 0.67 | -0.09 | 0.17 | 0.33 |
| Venezuela | 0.14 | 0.34 | 0.47 | 0.20 | -0.55 | 0.12 |
| Philippines | … | 0.13 | -0.03 | -0.02 | -0.08 | 0.00 |
| Algeria | … | 0.31 | 0.32 | 0.40 | 0.14 | 0.29 |
| Chile | 0.09 | 0.02 | -0.24 | -0.23 | -0.26 | -0.12 |
| Nigeria | … | -0.05 | -0.55 | -0.50 | -0.84 | -0.49 |
| Peru | 0.24 | 0.05 | -0.05 | -0.22 | -0.15 | -0.03 |
| Colombia | 0.71 | 0.60 | 0.52 | 0.33 | 0.20 | 0.47 |
| Ecuador | 0.28 | 0.41 | 0.38 | -0.35 | -0.98 | -0.05 |
| Uruguay | … | 0.23 | 0.39 | 0.16 | -0.04 | 0.19 |
| Zaire | … | 0.03 | 0.28 | 0.09 | 0.19 | 0.15 |
| Costa Rica | … | 0.02 | -0.47 | -1.24 | -1.17 | -0.71 |
| Sudan | … | 0.07 | 0.11 | 0.35 | -0.28 | 0.06 |
| Zambia | … | 0.50 | -0.34 | -0.23 | 0.00 | -0.02 |
| Bolivia | … | -1.21 | -0.46 | -0.63 | -1.01 | -0.83 |
| Average | 0.33 | 0.18 | 0.11 | -0.12 | -0.30 | |
| Specification G (with LIBOR) from Table 2 | ||||||
| Brazil | 0.63 | 0.47 | 0.44 | 0.04 | 0.11 | 0.34 |
| Mexico | 0.42 | 0.12 | 0.11 | -0.18 | -0.16 | 0.06 |
| Argentina | … | 0.42 | 0.56 | -0.06 | 0.16 | 0.27 |
| Venezuela | 0.23 | 0.17 | 0.34 | 0.24 | -0.28 | 0.14 |
| Philippines | … | 0.00 | -0.16 | -0.02 | 0.18 | 0.00 |
| Algeria | … | 0.14 | 0.23 | 0.41 | 0.35 | 0.28 |
| Chile | 0.24 | -0.15 | -0.36 | -0.22 | 0.01 | -0.10 |
| Nigeria | … | -0.30 | -0.73 | -0.55 | -0.54 | -0.53 |
| Peru | 0.35 | -0.13 | -0.22 | -0.30 | -0.17 | -0.09 |
| Colombia | 0.84 | 0.43 | 0.40 | 0.36 | 0.49 | 0.50 |
| Ecuador | 0.36 | 0.21 | 0.25 | -0.36 | -0.72 | -0.05 |
| Uruguay | … | 0.05 | 0.27 | 0.20 | 0.26 | 0.19 |
| Zaire | … | -0.17 | 0.16 | 0.16 | 0.25 | 0.10 |
| Costa Rica | … | -0.16 | -0.61 | -1.20 | -0.87 | -0.71 |
| Sudan | … | -0.12 | 0.04 | 0.43 | 0.07 | 0.11 |
| Zambia | … | 0.38 | -0.42 | -0.17 | 0.38 | 0.04 |
| Bolivia | … | -1.35 | -0.52 | -0.53 | -0.64 | -0.76 |
| Average | 0.44 | 0.00 | -0.01 | -0.10 | -0.07 | |
Ideally, the spread of sovereign securities yields over risk-free securities with the same contract terms would be used as the dependent variable. However, uncertainty regarding contractual terms as well as measurement problems rule out using secondary-market sovereign-risk yields.14 Nevertheless, we argue that the impact of creditor behavior on loan prices is best gauged after controlling for interest rate changes, in addition to accounting for external and fiscal constraints.15
V. Conclusion
The empirical results reported here are the first to show that, in addition to current account developments, sovereign debt values reflect the level and financing of the fiscal gap. This is further evidence that the debt prices used in restructuring should be perceived as capturing important macroeconomic information. The explanatory power of market interest rates may merit a reexamination of the emphasis on creditors in empirical analyses of external debt prices. The empirical results verify the importance of country-specific fundamentals in the determination of sovereign risk.16
An enormous theoretical literature has addressed the potential means of contract enforcement available to external creditors.17 The empirical importance of the shifting sources of creditor financing reported here suggests that domestic creditors should be included as players in models of government repayment and creditor punishment. Endogenizing creditor seniority could enhance understanding of the consequences of financial repression as well as provide insights into the conditions underlying the resumption of voluntary lending to debtor governments by domestic investors at market rates.18
APPENDIX
The CPI inflation series are from International Financial Statistics; the primary-balance series are from country sources and IMF country desk officers: and all other series are from the confidential IMF World Economic Outlook data base. Derivation of payments to creditors is shown in the following table. All entries are in U.S. dollars.
| Contractual interest to external creditors |
| - Change in external debt stock |
| - Arrears |
| + External debt reduction |
| + Valuation change |
| = Net payments to external creditors |
| Primary balance/GDP |
| - Net payments to external creditors/GDP |
| = Net payments to domestic creditors/GDP |
| Contractual interest to external creditors |
| - Change in external debt stock |
| - Arrears |
| + External debt reduction |
| + Valuation change |
| = Net payments to external creditors |
| Primary balance/GDP |
| - Net payments to external creditors/GDP |
| = Net payments to domestic creditors/GDP |
REFERENCES
Alexander, Lewis S., and Samira Kawash, “Measuring Market Perceptions of Credit Risk on Bank Loans to Developing Countries,” (unpublished; Washington: Board of Governors of the Federal Reserve System, 1988).
Anayiotos, George, and de Pinies, “The Secondary Market and the International Debt Problem,” World Development, Vol. 18 (1990), pp. 1655–69.
Barra, Robert I., “On the Determination of the Public Debt,” Journal of Political Economy, Vol. 87(5) (1979), pp. 940–71.
Bartolini, Leonardo, and Avinash Dixit, “Market Valuation of Illiquid Debt and Implications for Conflicts Among Creditors,” IMF Working Paper 90/88 (Washington: International Monetary Fund, 1990).
Boehmer, Ekkehart, and William L. Megginson, “Determinants of Secondary Market Prices for Developing Country Syndicated Loans.” Journal of Finance, Vol. XIV (December 1990), pp. 1517–40.
Bulow, Jeremy, and Kenneth Rogoff, “A Constant Recontracting Model of Sovereign Debt,” Journal of Political Economy, Vol. 96 (February 1989), pp. 155–78.
Calvo, Guillermo A., “Are High Interest Rates Effective for Stopping High Inflation? Some Skeptical Notes,” World Bank Economic Review,, Vol. 6 (January 1992), pp. 55–69.
Dooley, Michael P., “Capital Flight: A Response to Differences in Financial Risks,” Staff Papers,International Monetary Fund, Vol. 35 (1988), pp. 422–36.
Dooley, Michael P., Richard D. Haas, and Steven Symansky, “A Note on Burden Sharing Among Creditors,” Staff Papers, International Monetary Fund, Vol. 40 (1993), pp. 226–32.
Dooley, Michael P., and others, “An Analysis of External Debt Positions of Eight Developing Countries Through 1990,” Journal of Development Economics, Vol. 21 (1986), pp. 283–318.
Easterly, William R., “Fiscal Adjustment and Deficit Financing During the Debt Crisis,” in Dealing with the Debt Crisis, ed. by Ishrat Husain and lshac Diwan (Washington: World Bank, 1989), pp. 91–113.
Eaton, Jonathan, Mark Gersovitz, and Joseph Stiglitz, “The Pure Theory of Country Risk,” European Economic Review, Vol. 30 (1986), pp. 481–514.
El–Erian, Mohamed A., “The Restoration of Latin America’s Access to Voluntary Capital Market Financing—Developments and Prospects,” Staff Papers, International Monetary Fund, Vol. 39 (1992), pp. 175–94.
Fernandez, Raquel, and Sule Ozler, “Debt Concentration and Secondary Market Prices: A Theoretical and Empirical Analysis,” NBER Working Paper No. 3961 (Cambridge, Mass,: National Bureau of Economic Research, 1992).
Guidotti, Pablo E., and Manmohan S. Kumar, Domestic Public Debt of Externally Indebted Countries, Occasional Paper No. 80 (Washington: International Monetary Fund. 1991).
Heffernan, Shelagh A., Sovereign Risk Analysis (London: Unwin Hyman Ltd, 1986).
Khor Hoe E., and Liliana Rojas-Saurez, “Interest Rates in Mexico,” Staff Papers, International Monetary Fund, Vol. 38 (1990), pp. 850–71.
Kletzer, Kenneth M., and Brian D. Wright, “Sovereign Debt Renegotiation in a Consumption-Smoothing Model” (unpublished; International Monetary Fund, 1990).
Ozler, Sule, and Harry Huizinga, “How Factors on Creditor Countries Affect Secondary Market Prices for Developing Country Debt,” WPS 622 (Washington: World Bank, 1991).
Reisen, Helmut, and Axel van Trotsenburg, Developing Country Debt: The Budgetary and Transfer Problem (Paris: Organization for Economic Cooperation and Development, Development Center Studies, 1988).
Robinson, David, and Peter Stella, “Amalgamating Central Bank and Fiscal Deficits,” IMF Working Paper 87/73 (Washington: International Monetary Fund. 1987).
Stone, Mark R. (1991a), “Are Sovereign Debt Secondary Market Returns Sensitive to Macroeconomic Fundamentals? Evidence from the Contemporary and Interwar Markets,” Journal of International Money and Finance, Vol. 10 (1991), pp. S100–S122.
Stone, Mark R. (1991b),“On the Information Content of LDC Secondary Loan Market Prices,” IMF Working Paper 91/20 (Washington: International Monetary Fund, 1991).
Vatnick, Silvina, “The Secondary Market for Debt: A Possible Explanation of How LDC Debt Prices are Determined” (Washington: The World Bank, 1988).
Michael Dooley is on leave from the Research Department. Mark R. Stone is an Economist in the European I Department. He holds a Ph.D. from the University of Wisconsin. The authors are grateful to the country desk officers at the IMF who provided data for the empirical analysis. and they thank Dragica Pitipovic–Chaffey and Youkyong K. Kwon for excellent research assistance.
The implication of recent debt restructurings for foreign official and private debt values is analyzed in Dooley, Haas, and Symansky (1993).
This paper abstracts from the spending–taxation–total borrowing trade–offs and is instead concerned with the second–order problem of minimizing financing costs along the lines of Barra (1979).
Khor and Rojas–Saurez (1990) compare yields on domestic and external debt of Mexico.
Losses incurred by non—central government public sector entities may accumulate as an unreported conditional liability, and thus should be included in the financing formula. The two most important cases are state enterprises whose income flows are not part of the central government budget and central banks with losses that are not immediately financed by newly issued government credit. Because the profit–loss results of state enterprises are available for a limited number of sample countries, this potential financing need is not analyzed in this paper. A broader definition of financing needs that encompasses central bank losses is tested in the empirical section.
See Appendix.
For example, see Heffernan (1986).
Ratios of payments to foreign versus domestic creditors could not be used as a measure of relative creditor standing because of both negative and positive flows.
Alexander and Kawash (1988) calculate sovereign–risk spreads for a limited number of countries.
In addition to the problem of controlling for interest rate risk, the use of linear regression techniques may limit the empirical modeling of sovereign debt prices, since shifting creditor seniority implies a more general functional relationship between debt prices and fundamentals. Bartolini and Dixit (1990) base a theoretical model of debt values and creditor seniority on an option pricing framework. Debt prices have a nonlinear relationship with macroeconomic fundamentals: over intermediate price ranges, a relatively small change in seniority (for a given debt servicing capacity) may result in a large change in the value of debt to the creditor. Unfortunately, because the parameter values characterizing such a functional form are country specific, and only annual data are available for most of the fundamentals, testing the hypothesis of nonlinearity must await the arrival of more price data.
The importance of fundamentals in the determination of debt values is consistent with the circumstances underlying the reentry of indebted countries into capital markets (for another view, see El–Erian (1992)).
See Bulow and Rogoff (1989), Eaton, Gersovitz, and Stiglitz (1986), and Kletzer and Wright (1990).
See Guidotti and Kumar (1991) and Calvo (1992) for analyses of government credibility and financing costs.
