I. Stocks of External Claims and Rates of Return

In this section estimates of stocks of gross claims on nonresidents and yields on these claims are presented and analyzed. The major problem in estimating stocks of external claims held by residents of the countries studied is that balance-sheet data for such claims may be much less reliable compared with similar data for external debts,³ Indeed, to the extent that claims result from capital flight, these positions might be difficult to identify. For this reason, aggregate data for the countries studied are presented.

Column 1 of Table 1 shows the aggregate stock of private and official claims on nonresidents calculated from cumulated balance of payments data. The initial value was estimated by capitalizing investment income receipts in the initial year. Column 2 shows the aggregate stock of errors and omissions in the cumulated balance of payments accounts. This item is often associated with accumulations of financial claims on nonresidents in that it could reflect, among many other things, unrecorded capital outflows.’⁴ The sum of columns 1 and 2, shown in column 3, can be taken as an aggregate cumulative capital outflow from the seven countries considered, or an increase in gross claims, as estimated from balance of payments data.

These data, however, could seriously underestimate the stock of external claims. Balance of payments data seem to underestimate the aggregate accumulation of external debt. For the countries studied, external debt in 1984 as estimated from balance of payments data was only 60 percent as large as external debt as estimated by World Bank data.⁵ If it is assumed that the World Bank data are accurate, there must have been a corresponding underestimate of “balancing transactions.” These could include any type of transaction with a nonresident. For example, imports of goods and services or purchases of financial claims on nonresidents could have been financed by the accumulation of external debt not captured in the balance of payments reporting system.⁶ In order to proceed, it is assumed that all of the balancing transactions shown in column 4 represent increases in private financial claims on nonresidents. The sum of columns 3 and 4, shown in column 5, is an estimate of the total cumulated stock of financial claims on nonresidents for the countries considered.

^{Table 1.}

Alternative Measures of Aggregate External Claims, 1977-84

(In billions of U.S. dollars)

Note: See Appendix I for definitions and sources.

^{Table 1.}

Alternative Measures of Aggregate External Claims, 1977-84

(In billions of U.S. dollars)

	Recorded		Total
	Claims on		External
	Nonresidents	Cumulated	Claims,		Total
	Other Than	Balance of	Cumulated	Unrecorded	Stock of
	Direct Investment,	Payments	Balance of	Stock of	External
	Cumulated Balance	Errors and	Payments	External	Claims
	of Payments	Omissions	(1)+(2)	Claims	(3)+(4)
Year	(1)	(2)	(3)	(4)	(5)
1971	37.2	2.4	39.6	24.2	63.8
1978	46.8	0.7	47.5	32.7	80.1
1979	58.0	-2.3	55.8	43.4	99.1
1980	63.4	3.3	66.7	56.0	122.8
1981	71.5	14.1	85.7	56.4	142.1
1982	62.7	23.1	85.8	60.2	146.0
1983	68.1	25.1	93.2	66.9	160.1
1984	88.0	23.7	111.7	71.1	182.8

Note: See Appendix I for definitions and sources.

View Table

^{Table 1.}

Alternative Measures of Aggregate External Claims, 1977-84

(In billions of U.S. dollars)

	Recorded		Total
	Claims on		External
	Nonresidents	Cumulated	Claims,		Total
	Other Than	Balance of	Cumulated	Unrecorded	Stock of
	Direct Investment,	Payments	Balance of	Stock of	External
	Cumulated Balance	Errors and	Payments	External	Claims
	of Payments	Omissions	(1)+(2)	Claims	(3)+(4)
Year	(1)	(2)	(3)	(4)	(5)
1971	37.2	2.4	39.6	24.2	63.8
1978	46.8	0.7	47.5	32.7	80.1
1979	58.0	-2.3	55.8	43.4	99.1
1980	63.4	3.3	66.7	56.0	122.8
1981	71.5	14.1	85.7	56.4	142.1
1982	62.7	23.1	85.8	60.2	146.0
1983	68.1	25.1	93.2	66.9	160.1
1984	88.0	23.7	111.7	71.1	182.8

Note: See Appendix I for definitions and sources.

With this in mind, it is natural to compare investment income receipts as recorded in the balance of payments with alternative measures of the stock of claims on nonresidents. Columns 1, 2, and 3 of Table 2 show the weighted-average ratio of investment income receipts to the three measures of gross external claims presented in columns 1, 3, and 5 of Table 1. These implicit yields can be compared with the weighted-average “market yield” shown in column 4 of Table 2⁷ Investment income receipts seem to imply plausible yields relative to market yields if it is assumed that only cumulated recorded external claims yield income that is captured in the balance of payments accounts. In Table 2, the addition of cumulated errors and omissions to the level of external claims (column 2), or this plus unrecorded claims (column 3), reduces the implied yield to implausibly low levels.

^{Table 2.}

Weighted-Average Yields on Alternative Measures of External Claims and a Measure of Capital Flight, 1977-84

(In percent)

Note: See Appendix II for definitions and sources.

^{Table 2.}

Weighted-Average Yields on Alternative Measures of External Claims and a Measure of Capital Flight, 1977-84

(In percent)

	Weighted-	Weighted-
	Average Yield	Average Yield		Weighted-
	on Recorded	on Total	Weighted-	Average
	External Claims,	External Claims,	Average Yield	Market
	Cumulated	Cumulated	on Total Stock	Yield on
	Balance of	Balance of	of External	External
	Payments	Payments	Claims	Claims
Year	(1)	(2)	(3)	(4)
1977	4.3	4.0	2.5	6.6
1978	5.7	5.6	3.3	8.2
1979	7.3	7.6	4.3	10.6
1980	10.3	9.7	5.3	12.7
1981	11.6	9.7	5.9	14.4
1982	10.5	7.7	4.5	12.0
1983	6.8	4.9	2.9	9.7
1984	7.4	5.9	3.6	10.7

Note: See Appendix II for definitions and sources.

View Table

^{Table 2.}

Weighted-Average Yields on Alternative Measures of External Claims and a Measure of Capital Flight, 1977-84

(In percent)

	Weighted-	Weighted-
	Average Yield	Average Yield		Weighted-
	on Recorded	on Total	Weighted-	Average
	External Claims,	External Claims,	Average Yield	Market
	Cumulated	Cumulated	on Total Stock	Yield on
	Balance of	Balance of	of External	External
	Payments	Payments	Claims	Claims
Year	(1)	(2)	(3)	(4)
1977	4.3	4.0	2.5	6.6
1978	5.7	5.6	3.3	8.2
1979	7.3	7.6	4.3	10.6
1980	10.3	9.7	5.3	12.7
1981	11.6	9.7	5.9	14.4
1982	10.5	7.7	4.5	12.0
1983	6.8	4.9	2.9	9.7
1984	7.4	5.9	3.6	10.7

Note: See Appendix II for definitions and sources.

These very low calculated yields on the more inclusive measures of external claims raise two interesting possibilities. One is that the data on investment income receipts are systematically understated. In some cases (the United States is an example), balance of payments data on such receipts are not measured directly but instead are estimated by applying a market yield to an estimated stock of gross claims. When this procedure is followed, the use of cumulated balance of payments statistics to estimate the stock of claims could understate the value of receipts. It would seem reasonable in this event to revise the balance of payments statistics, particularly in cases where liabilities reported by nonresidents to the subject country are substantial. Indeed, a satisfactory resolution of important problems in evaluating the role of external claims would seem to require an effort to gather data on the stock of external claims held by residents of developing countries comparable to the data available on the external debt of these countries. In cases where the understating of investment income receipts is due simply to an inappropriate estimate of the stock of external claims, a correction of this error would result in a substantial “improvement” in the current accounts of some of the countries studied.⁸

The second possibility is that unreported earnings on external claims reflect residents’ preferences for assets that are beyond the control of domestic authorities. In this case repatriation of external claims, or of the earnings thereon, might depend on a reduction of expected taxation or on other penalties on residents’ financial wealth. It is worth noting, however, that additional steps may be necessary in cases where domestic financial markets are relatively underdeveloped and not fully recovered from the difficulties experienced in recent years.

II. Capital Flight

There is no widely accepted definition of capital flight. Capital outflows per se, types of outflows such as increases in private short-term claims, or errors and omissions are not a reasonable measure of this phenomenon. The United States, for example, by the end of 1983 had accumulated roughly US$700 billion in gross claims on nonresidents. These claims were offset by a roughly equivalent stock of external debt, and, because the average yield on claims is estimated to be higher than that on liabilities, the investment income accounts showed a substantial net credit in the U.S. balance of payments accounts.⁹ The situation would of course be immediately altered if, for some reason, these foreign exchange earnings were not available to U.S. residents who had to make investment income payments.¹⁰ This possibility suggests a useful definition of capital flight. In cases where all or some portion of the income receipts are lost to the country, it is assumed that the capital outflow was motivated by the desire of residents to obtain financial assets, and earnings on those assets, that remain outside the control of the domestic authorities. Capital flight, by this definition, does not require a change in the stock of total claims on nonresidents but only that earnings on existing claims be placed outside the control of the domestic authorities. Thus, capital flight can occur, and be reversed, quite rapidly.¹¹

The appropriate way to measure capital flight by this definition is to calculate the stock of claims implied by investment income receipts and market interest rates. This stock of “interest-earning” claims is interpreted as the stock of external claims not attributable to capital flight but attributable to normal foreign portfolio investment. The difference between this estimate and the estimate of the total stock of external claims (for the latter, see Table 1, column 5) is presented in column 1 of Table 3. It should be emphasized that whether such holdings are “beyond the reach” of the authorities is a matter of fact that cannot be established here. Thus, the estimates of aggregate capital flight shown in Table 3 are, at best, approximations.¹² One can gain some perspective on the possible importance of capital flight by comparing capital flight with other financial aggregates. By these estimates, the stock of flight capital accounted for about two thirds of the total stock of external claims of the seven countries studied and was equal to about one third of their external debt.

III. Empirical Tests

This section presents empirical tests of the view that capital flight occurs in circumstances in which residents and nonresidents have perceived different risks associated with domestic investments. The existing empirical literature on risk premiums and on capital flight emphasizes the idea that residents and nonresidents do face different risks. For example, of the nine independent variables in Edwards’s (1984) paper on risk premiums and of the three independent variables in Cuddington’s (1986) paper on capital flight, only one, the rate of domestic inflation, appears in both studies. The different regression hypotheses in these studies reflect the view that capital flight is motivated primarily by inflation or exchange rate risk, whereas risk premiums charged by nonresidents reflect mainly default risk.

^{Table 3.}

Aggregate Capital Flight in Ratio to External Claims and External Liabilities, 1977-84

Note: See Appendix III for definitions and sources.

^{Table 3.}

Aggregate Capital Flight in Ratio to External Claims and External Liabilities, 1977-84

		Ratio of	Ratio of
	Aggregate	Aggregate Capital	Aggregate Capital
	Calculated	Flight to	Flight to
	Capital Flight	External Claims	External Debt
Year	(1)	(2)	(3)
1977	38.90	0.61	0.33
1978	48.18	0.60	0.33
1979	59.03	0.60	0.33
1980	71.63	0.58	0.32
1981	84.53	0.59	0.31
1982	91.62	0.63	0.30
1983	112.19	0.70	0.35
1984	121.19	0.66	0.36

Note: See Appendix III for definitions and sources.

View Table

^{Table 3.}

Aggregate Capital Flight in Ratio to External Claims and External Liabilities, 1977-84

		Ratio of	Ratio of
	Aggregate	Aggregate Capital	Aggregate Capital
	Calculated	Flight to	Flight to
	Capital Flight	External Claims	External Debt
Year	(1)	(2)	(3)
1977	38.90	0.61	0.33
1978	48.18	0.60	0.33
1979	59.03	0.60	0.33
1980	71.63	0.58	0.32
1981	84.53	0.59	0.31
1982	91.62	0.63	0.30
1983	112.19	0.70	0.35
1984	121.19	0.66	0.36

Note: See Appendix III for definitions and sources.

The different regression hypotheses can be explained by the assumption that residents of a debtor country do not fear default in the sense that financial contracts between residents (including the government) will be enforced in cases where nonresident creditors are penalized. Residents are exposed to taxation of income derived from domestic investments, and in many cases domestic instruments may offer incomplete protection from domestic inflation and exchange rate depreciation.

In contrast, nonresident investors have typically acquired claims on debtor countries that are denominated in foreign currencies and therefore are not subject to domestic inflation and exchange rate depreciation. For this reason evaluations of risks faced by nonresident investors have focused on comparisons between the country’s ability to generate the foreign exchange necessary to service its external debt and the stock of that debt.

In this section the conjecture is tested that both inflation and other taxes on residents and the default “tax” must be considered together to obtain a satisfactory explanation of capital flight. Governments faced with a fiscal deficit may be forced to impose taxes on domestic and nonresident investors in a variety of forms that tend to reduce the value of such investments. Thus, any shock to the system that increases government expenditure or reduces revenues raises the possibility that new sources of revenue will be exploited. This possibility does not imply that all financial positions will be affected equally or at the same time. The revenue generated by a change in the tax on a financial position depends on the size of the position to be taxed and on the opportunities that asset holders have to avoid the tax. An unexpected increase in the inflation rate, for example, provides revenue to the government that is dependent on the stock of fixed-interest domestic currency liabilities of the government and on the ability of holders of such liabilities to shift into alternative assets to avoid the tax. In cases where the ability to shift into a less exposed position is limited—for example, by a limitation on the right to purchase foreign exchange at the official rate—the government might be expected first to “tax” domestic residents through money creation and inflation. In response, residents will seek to acquire financial assets denominated in foreign currency. If these are not available domestically, residents will attempt to acquire claims on nonresidents denominated in foreign currency.

The incipient capital outflow must be matched either by some offsetting transaction or by a change in other variables that discourages the outflow. One possible outcome is that nonresident investors will be attracted by the fall in prices (increases in yields) that will be generated as residents attempt to liquidate domestic securities. Nonresident investors, however, will not be interested in liabilities of the debtor country that are denominated in domestic currency because such investment also exposes them to the inflation tax. Unlike residents, nonresident investors may be able to purchase claims on residents denominated in foreign currency. Moreover, external creditors may have access to explicit or implicit government guarantees not available to residents.¹³ In this case the capital outflow to avoid the inflation tax would be matched by a capital inflow that is protected from the inflation tax because of its currency denomination and that also enjoys a government guarantee.

The capital flows generated in the above example can best be understood as attempts by asset holders to arbitrage a yield differential that is generated by the inflation tax on residents. Because nonresidents can avoid this tax in ways not available to residents, capital flight is the expected outcome of this incentive structure. Similar types of capital flows among industrial countries have been called “round-trip” capital flows (see Dooley (1980-81)). In some cases fairly small tax incentives in terms of differential reserve requirements on bank deposits have led to very large gross capital inflows and outflows in the U.S. balance of payments. Such flows are not referred to as capital flight in this context, although they are analytically equivalent to what is called capital flight in developing countries.

There are limits to round-trip capital flows. At some point the stock of liabilities to nonresident investors becomes sufficiently large to make it worthwhile for the government to consider a tax on these financial positions. The inflation tax will not be effective, but other more direct measures that reduce payments on such liabilities will yield greater revenues as the stock of liabilities to nonresidents grows. As the perceived risk faced by nonresident creditors grows, the opportunities for arbi-traging the differential risks faced initially by residents compared with nonresidents will be reduced and may eventually be eliminated. At this point the round-trip capital flows will end.

In this new situation resident investors could avoid domestic taxes on financial assets only by net transfers of goods and services to nonresidents. Thus, although capital flight as measured in this paper will be much reduced or eliminated, the consequences of continued actual or expected taxation of residents’ financial assets remain serious. Indeed, the consequences will be more serious because the flow of real saving available for increases in the domestic stock of real capital is threatened, since taxes on domestic financial saving can no longer be avoided by round-trip capital flows.

The discussion above suggests that capital flight should be evident in circumstances in which residents perceive risks to income derived from domestic claims, but in which nonresidents have perceived relatively smaller risks on credits to the residents of the country studied. To evaluate this idea, the relationships between capital flight defined above, and several possible determinants of capital flight, are tested empirically. The variables considered that might be determinants of capital flight include the following.

• Domestic inflation (INF) is interpreted as measuring the extent to which the government has resorted to taxing domestic financial assets through money creation. This easily observed measure of “tax” policy can be considered a proxy for the difficulty the government is experiencing in generating revenue. Thus, even domestic investments that are indexed to the inflation rate may be considered at risk when the government is using the inflation tax.

• Financial repression (FR) measures the difference between interest rates paid on short-term assets denominated in U.S. dollars and time deposits denominated in the subject country’s domestic currency and adjusted for actual exchange rate changes. A positive value for this differential indicates that such deposits are vulnerable to the inflation tax. A differential in favor of dollar-denominated assets would suggest more capital flight.

• The risk premium (RP) is the risk premium on external debt reported in Dooley (1986) and interpreted as measuring nonresidents’ perception of the risk of being “taxed” by the subject country’s government. As this perceived risk increases, it is expected that capital flight will be reduced because the differential risk faced by resident and nonresident investors is reduced.

As discussed above, these are clearly endogenous variables that reflect basic changes in the environment faced by governments of debtor countries. For this reason, instrumental variables were used in the regressions reported below. The pressure for domestic inflation is assumed to be related to the fiscal deficit of the debtor country, expressed as a share of gross national product. The default risk faced by nonresidents is assumed to be related to the ratio of debt to exports and to the ratio of interest payments to exports.

The results of a pooled regression for annual observations for six countries for the 1976-83 time period are shown in Table 4, In general, these results support the view that capital flight is related to the relative perceptions of risks that residents and nonresidents associate with claims on residents of the countries studied.

The positive coefficient for domestic inflation (INF) in the estimation results in Table 4 can be interpreted as indicating that, where the authorities are “taxing” domestic money balances through inflation, residents prefer to hold a larger share of their financial assets in a form that is outside the control of the domestic authorities. The coefficient on the political risk premium (RP) indicates the expected negative impact on capital flight. Thus it appears that the increases in this premium in recent years, and the associated reluctance of nonresident investors to “recycle” flight capital, have tended to limit the scale of such transactions. Finally, the positive relationship between financial repression (FR) and capital flight suggests that residents acquire foreign assets in cases where the yield on domestic currency time deposits is less than short-term rates in international markets (adjusted for exchange rate changes). In some cases this may reflect controls on domestic interest rates, or it may simply reflect relatively slow adjustment of domestic interest rates to rapidly changing economic conditions.

^{Table 4.}

Determinants of Capital Flight: Pooled Data, 1976-83

^a“Estimation results” were obtained by instrumental variables and ordinary least squares with country-specific dummy variables. Annual observations for six countries (Argentina, Brazil, Chile, Mexico, the Philippines, and Venezuela) were pooled for the regression; Peru was excluded because_data on deposit rates were unavailable. Numbers in parentheses are (-statistics;${\overline{R}}^2$is the coefficient of determination, corrected for degrees of freedom.

^{Table 4.}

Determinants of Capital Flight: Pooled Data, 1976-83

Symbol		Definition and Source
	Variables
CF	Capital flight, from data reported in Table 3
INF	Domestic inflation, calculated as differences in logarithms of consumer price indices, from International Monetary Fund, International Financial Statistics (IFS) (Washington, various issues), line 64
FR	Financial repression.
		FR = [(1 + rUS/(1 + r)]/[1 + In X - In X(-1)],
		where r is the domestic time deposit interest rate (national sources), is the U.S. Treasury bill interest rate (IFS line 60c), and X is domestic currency per dollar (IFS line ac)
RP	Risk premium r, from data reported in Dooley (1986)
	Instruments
Fiscal deficit/gross national product (GNP)	Fiscal deficit from IFS line 80zf and Fund staff estimates; GNP from IFS line 99a
Debt/GNP	Debt from World Bank, World Debt Tables (Washington, various issues)
Interest payments/ GNP	London interbank offered rate (LIBOR) and IFS line 60eb, times Debt/GNP
	Estimation resultsa
CF	$\underset{\left(4.34\right)}{23.10INF}+\underset{\left(4.09\right)}{18.91}FR-\underset{(-1.66)}{1.12RP}$
${\overline{R}}^2$	= 0.83

^a“Estimation results” were obtained by instrumental variables and ordinary least squares with country-specific dummy variables. Annual observations for six countries (Argentina, Brazil, Chile, Mexico, the Philippines, and Venezuela) were pooled for the regression; Peru was excluded because_data on deposit rates were unavailable. Numbers in parentheses are (-statistics;${\overline{R}}^2$is the coefficient of determination, corrected for degrees of freedom.

View Table

^{Table 4.}

Determinants of Capital Flight: Pooled Data, 1976-83

Symbol		Definition and Source
	Variables
CF	Capital flight, from data reported in Table 3
INF	Domestic inflation, calculated as differences in logarithms of consumer price indices, from International Monetary Fund, International Financial Statistics (IFS) (Washington, various issues), line 64
FR	Financial repression.
		FR = [(1 + rUS/(1 + r)]/[1 + In X - In X(-1)],
		where r is the domestic time deposit interest rate (national sources), is the U.S. Treasury bill interest rate (IFS line 60c), and X is domestic currency per dollar (IFS line ac)
RP	Risk premium r, from data reported in Dooley (1986)
	Instruments
Fiscal deficit/gross national product (GNP)	Fiscal deficit from IFS line 80zf and Fund staff estimates; GNP from IFS line 99a
Debt/GNP	Debt from World Bank, World Debt Tables (Washington, various issues)
Interest payments/ GNP	London interbank offered rate (LIBOR) and IFS line 60eb, times Debt/GNP
	Estimation resultsa
CF	$\underset{\left(4.34\right)}{23.10INF}+\underset{\left(4.09\right)}{18.91}FR-\underset{(-1.66)}{1.12RP}$
${\overline{R}}^2$	= 0.83

^a“Estimation results” were obtained by instrumental variables and ordinary least squares with country-specific dummy variables. Annual observations for six countries (Argentina, Brazil, Chile, Mexico, the Philippines, and Venezuela) were pooled for the regression; Peru was excluded because_data on deposit rates were unavailable. Numbers in parentheses are (-statistics;${\overline{R}}^2$is the coefficient of determination, corrected for degrees of freedom.

IV. Conclusions

An important challenge for analyses of external debt is to identify factors that are quantitatively important in shaping individual countries’ current and prospective external positions. Existing studies have in general compared a very simple representation of a country’s external financial position—the ratio of interest payments or debt service payments to some measure of the country’s ability to make such payments—under alternative assumptions about economic growth in Organization for Economic Cooperation and Development countries, interest rates in creditor countries, and several other domestic and external variables. The analysis developed in this paper suggests that in many cases a more complete representation of individual debtor countries’ financial positions would be useful.

Review of historical experience for selected debtor countries suggests that different incentives faced by nonresident and resident investors seem to be an important determinant of gross external claims and liabilities of the countries studied. Moreover, the size of such gross positions may have little to do with the determinants of net external debt of the country studied. Finally, because investment income payments associated with external debt and investment income receipts associated with external claims appear to have been quite sensitive to residents’ and nonresidents’ attitudes toward claims on debtor countries, a better understanding of the behavior that determines such positions might be an important element in evaluating and improving a country’s external position.