Chapter

The Role of International Reserves and Foreign Debt in the External Adjustment Process

Editor(s):
Joaquín Muns
Published Date:
September 1984
Share
  • ShareShare
Show Summary Details
Author(s)
SEBASTIAN EDWARDS

1. INTRODUCTION

During the 1970s the developing countries underwent serious current account deficits in their balance of payments, which led, inter alia, to major increases in their foreign debt. Whereas in 1970 the public and publicly guaranteed debt of the low-income countries (excluding the People’s Republic of China and India) amounted to 16.5 percent of their gross national product (GNP), in 1980 it had risen to 31.5 percent of GNP. On the other hand, the ratio of foreign public and publicly guaranteed debt of the middle-income countries increased from 11.8 percent of GNP in 1970 to 17.4 percent of GNP in 1980.1 More recently, some developing countries—Mexico, Brazil, Costa Rica, and Argentina, for instance—have faced serious problems in servicing their foreign debt.

The analysis of the determinants of the current account of the balance of payments, and its relation to macroeconomic adjustment, has lately made significant progress. Particular emphasis has been given to the intertemporal nature of the current account, and the distinction has been made between temporary and permanent shocks.2 It has been postulated, for instance, that whereas the most appropriate policy to face temporary shock is to increase the level of foreign debt (or reduce the level of international reserves), reduction of the level of internal absorption is the optimum policy in response to a permanent shock.3

Despite renewed interest in analyzing the macroeconomic adjustment process in an open economy, most of the papers, theoretical as well as empirical, emphasize the use of a particular tool to achieve such adjustment. Thus, for instance, while certain authors have made a detailed analysis of the use of foreign debt as an instrument for adjustment, others have focused on the use of reserves, and a third group has examined in detail the role of the exchange rate and its relation to the current account.4 Exceptions to this rule are Eaton and Gersovitz (1980), who have analyzed the joint determination of international reserves and foreign debt; Edwards (1983 a), who has explicitly considered the use of the exchange rate as an adjustment tool in the analysis of demand for international reserves; and Frenkel and Aizenman (1982), who have theoretically examined the joint determination of the optimum degree of float for a currency and the level of reserves the country wishes to maintain.

The purpose of this paper is to analyze in some detail the role of international reserves and external debt in the process of macroeconomic adjustment in developing countries, with particular emphasis on Latin American countries. The discussion explicitly recognizes that both reserves and debt are alternative instruments that countries may use to achieve external balance.5 In addition, it is recognized that foreign debt plays a twofold role in the economic process. First, in the short run, changes in the external debt make it possible to cope with temporary external disequilibria. Second, foreign debt plays a long-run role connected with financing gross domestic investment. In fact, insofar as the current account deficit—which is equal to external savings—is partly or totally financed by new (net) foreign borrowing, the foreign debt will be playing a major long-term role in the development and economic growth process. It is precisely this twofold role played by foreign debt that requires developing countries to find ways to manage it carefully. In this context it is important, for example, to make the needs for external resources to finance gross investment compatible with the uses of debt for short-term purposes. Appropriate planning of the evolution of external liabilities and assets of the monetary system will prevent imbalances that may turn out very costly in the long run.6

In this paper various problems connected with the role of international reserves and foreign debt in the external adjustment process are empirically analyzed. Section 2 contains a general discussion of the subject, emphasizing the fact that most of the economic literature dealing with demand for international reserves has failed to consider explicitly the role played by debt (or exchange rate variations) in the external adjustment process. This section also presents a discussion—and a word of caution—on the use of dynamic optimization models to determine the optimum degree of debt for a country. Section 3, in turn, presents new empirical evidence on the determinants of demand for international reserves in the developing countries, using data for 1975–80. This section also presents results obtained when reserves and debt are considered to be alternative instruments for achieving external equilibrium. The main conclusions reached in this section are that: (a) the demand for reserves in the developing countries has been a stable function in the past few years, although from a legal standpoint the international monetary system has been characterized by a dirty float;7 (b) demand for reserves in the developing countries displays diseconomies of scale showing that, as a result of economic growth, such countries’ need for international liquidity will grow more than proportionately in the next few years; (c) reserves and foreign debt are substitutes, in the sense that both have been used by these countries as alternative ways to finance their external adjustment processes.

Section 4 of the paper examines a critical issue for macroeconomic management of reserves and foreign debt. The section discusses in what way the levels of debt and reserves, among other variables, affect the level of country risk perceived by the international financial community. The importance of this discussion lies in the fact that should a relation exist between these variables and the probability of default perceived by the international financial community, countries may manage these variables in such a way that the degree of risk of default perceived is kept at relatively low levels. This in turn will lead to more favorable terms for obtaining foreign funds and continuing maintenance of such loans. The results presented in this section, based on data from over 700 public and publicly guaranteed loans granted in the period 1976–80 in the Eurocurrency market, show that the higher the debt/output ratio the higher the perceived probability of default. It was further found that a higher international liquidity ratio will result in a lower perceived country risk.

In Section 5 an empirical analysis on the relation between the dynamic adjustment of international reserves and monetary market equilibrium is presented. This analysis shows that international reserve movements over time arise essentially from two factors: (a) discrepancies between desired reserves and reserves actually maintained; and (b) imbalances in the monetary market. These results—which basically correspond to the analysis presented in the monetary approach to the balance of payments—point out that situations of excess supply of money will tend to result in international reserves dropping below desired levels. The above findings, in the light of results obtained from analysis of country risk determinants, have obvious implications in terms of economic policy. Insofar as a country wishes to maintain an acceptable (conservative) degree of perceived country risk, it must be particularly careful in the way it handles its monetary policy.

Section 6 of the paper contains some considerations on the case of the Latin American countries, with emphasis on the importance of designing a coherent exchange rate policy to deal with macroeconomic adjustment problems. Finally, some concluding remarks are offered in Section 7.

2. INTERNATIONAL RESERVES AND FOREIGN DEBT IN THE MACROECONOMIC ADJUSTMENT PROCESS

There is an extensive literature, empirical as well as theoretical, on the role of international reserves in the external adjustment process, which has been reviewed, among others, by Clower and Lipsey (1968), Gruebel (1971), Williamson (1973), and Bird (1978). Generally speaking, most of the studies on the subject assume that countries maintain reserves both for financing international transactions and for facing unforeseen international payment difficulties. Hipple (1974), for instance, groups studies on international reserves under two headings: one focusing on the transaction motive for maintaining reserves; the other emphasizing the role of international payment variability as a motive for demanding reserves. Most recent studies, however, have recognized the importance of both these motives in the demand for reserves.8

Empirical studies recently conducted by Frenkel (1978, 1980) and Heller and Khan (1978) have shown that, even after the change in the monetary system’s rules of the game in 1973, the demand for international reserves in the various groups of countries remains stable. Frenkel (1980), for instance, found that in 1972 a structural change took place in the demand function for reserves in the case of developing countries. This finding—also confirmed by Heller and Khan (1978)—shows that the developing countries altered their behavior with respect to the desired amount of international liquidity even before the Bretton Woods system was officially given up in 1973.9 For practical and economic policy purposes, however, the relevant point about these results is that they show that even during the present international monetary arrangements, various countries’ demand for international liquidity has remained stable. This fact may be explained in several ways. For example, at present the international monetary system is based on a dirty float (managed parities) by the industrial countries and diverse systems—pegging to a currency, pegging to a basket, crawling peg, and so on—used by the developing countries.10 In the event that the float is dirty, countries will wish to maintain a positive level of international reserves in order to intervene in the exchange market.

Most of the studies on international reserves are based on some concept of optimality for determining the desired level of liquidity for a particular country. Heller (1966), in his pioneer work, equated the marginal benefit to the marginal cost of maintaining reserves to derive his equation on optimum quantity of reserves. Subsequent papers, explicitly based on optimality concepts, include those of Olivera (1969), Hipple (1974), Claasen (1976), Hamada and Veda (1977), and Frenkel and Jovanovic (1981). Such studies assume that maintaining reserves allows countries to survive external crises without having to incur the costs of adjusting internal absorption. This positive effect of maintaining reserves should therefore be compared with the corresponding costs, in order to determine the optimum amount of reserves. Thus, for example, in Clark’s (1970 a) well-known dynamic analysis, the costs of maintaining reserves are given by the marginal productivity of capital (representing the alternative cost of reserves) and by the cost of adjusting the quantity actually maintained to the quantity desired. Conversely, benefits are related to the ability to maintain a more stable real-income flow over time. In this way Clark (1970 a) derives a model that, for a given probability that a country runs out of reserves, simultaneously determines the optimum level of reserves and the optimum speed of adjustment between desired reserves and reserves actually maintained.

In a recent paper Frenkel and Jovanovic (1980) develop a stochastic model to determine optimum reserve levels, combining issues on the demand for money for precautionary and for transaction motives. In this study Frenkel and Jovanovic (1980) find that the optimum level of reserves will depend on the variability of international payments, the alternative cost of maintaining reserves, and a variable measuring the country’s scale.

One problem affecting most of the empirical studies on demand for reserves is that they fail to consider explicitly that changes in international reserves are only one of the possible policy tools that can be used to face external imbalances. In fact, countries undergoing payments difficulties have at least four alternatives for solving such problems: (1) they may alter the level of reserves maintained; (2) they may resort to policies aimed at expenditure switching; that is, they may devalue (Makin (1976)) or they may implement commercial policies; (3) they may borrow abroad; or (4) they may adjust their level of internal absorption. Theoretically, by means of some optimizing criterion, countries may decide simultaneously what proportion of the adjustment is to be obtained through each of the above instruments. This decision will be influenced by such considerations as the costs associated with the use of each of these instruments and the nature of the external disequilibria (i.e., temporary or permanent shocks). While some authors have recognized, at a theoretical level, the simultaneous nature of decisions to maintain international reserves and adjust the exchange rate (Edwards (1983 a)), very few have explicitly considered the simultaneous decision to use reserves and foreign debt as alternative instruments for external adjustment. (One major exception is the work done recently by Eaton and Gersovitz (1980), (1981 a), and (1981 b).) Moreover, as far as I know, no author has considered the simultaneous decision to use reserves, debt, and exchange rate adjustment to correct external imbalances.

In a recent paper Eaton and Gersovitz (1980) consider the case of an economy where the decision regarding the level of reserves and debt to be maintained forms part of an overall portfolio problem. In their empirical study these authors find that foreign debt is a substitute for reserves, in the sense that changes in debt level are used (partly) to facilitate external adjustment, “smoothing” the level of internal absorption through periods of significant variations in exports revenue.

From the standpoint of economic policy, however, one major issue that Eaton and Gersovitz (1980) fail to consider is that while international reserves essentially fill a short-run role only—to facilitate smoother external adjustments—foreign debt plays both long-run and short-run roles. In the first place, as mentioned earlier, debt accumulation and disaccumulation help to reduce the costs associated with the external adjustment process. This is a short-run role. Second, foreign debt accumulation in each period will be linked to external savings (if international reserves remain unchanged, the new net debt will be equal to external savings), and hence will play a fundamental part in financing gross domestic investment. This is therefore an essentially long-run role. It is precisely this twofold role of foreign debt that makes it necessary for countries to design financial management programs that will make the short-run variations in reserves compatible with the needs for medium- and long-term external financing. Loser (1977) has stressed the need for financial management suited to foreign debt.

A Note on Optimum Foreign Debt, External Savings, and Economic Growth

The fact that debt accumulation is generally associated with higher foreign savings, and hence presumably with higher economic growth, has led some policymakers to feel that the optimum level of debt is relatively high. The argument usually runs along the following lines: “As more debt involves more capital accumulation, it is advisable to borrow while debt cost is less than the marginal productivity of capital.” Moreover, in the economic literature some authors (e.g., Blanchard (1981)), have used dynamic optimization models to postulate that for a country such as Brazil the optimum debt level is approximately equal to 1.6 times GNP. Nevertheless, insofar as in practice debt/output ratios are substantially below 1, it is worth wondering what is wrong with the above argument.11

It appears obvious that there are at least three reasons for qualifying the notion that the optimum foreign debt level should be relatively high. First, it is possible that higher external savings are replacing domestic savings (Mikesell and Zinser (1973)). Depending on whether such substitution is total or partial, total saving will increase or remain constant.12 In other words, in this case it is possible that the higher foreign debt is being used (partly) to finance higher consumption. This seems to have been the case in Chile in 1980–81, where, in spite of significant increases in the level of foreign debt, total savings remained constant, and even declined (see Edwards (1982 a)).

Second, a higher foreign debt will result in higher cost of foreign credit, even to the point of the country concerned being excluded from the international credit market, with all the consequential costs associated with such a measure. (See Section 4, below.) This prospect will of course lead to optimum debt levels significantly below those calculated using methods such as Blanchard’s (1981).

Lastly, to the extent that the economy produces tradable and nontradable goods, the allocation of the new investment financed with higher foreign debt is a major consideration, because in the future the foreign debt will have to be repaid through current account surplus.

3. DEMAND FOR INTERNATIONAL RESERVES IN DEVELOPING COUNTRIES: EMPIRICAL RESULTS 1975–80

This section presents the results obtained from the empirical analysis of the determinants of the demand for international reserves for a selected group of developing countries during 1975–80. As discussed in the preceding section, the empirical analyses of Frenkel (1978, 1980), Heller and Khan (1978), and others, have shown that even after the international monetary system legally became a floating rate system in 1973, the various groups of countries have had stable demand for international reserves. The above authors further pointed out that a statistical difference is observed in behavior in regard to reserves before and after the collapse of the Bretton Woods system.

The results presented in this section will therefore be useful for projecting future international liquidity needs of developing countries, assuming that the international financial system will retain its present form.

Generally speaking, the relevant literature has assumed that countries demand international liquidity both to finance international transactions and to face unforeseen international payments. Such studies have assumed that the demand for reserves is a stable function of a small number of variables that usually include the size of the country, the degree of variability of its international transactions, the degree of openness to the rest of the world, and the alternative cost of maintaining reserves.

Regarding the size of the country, most authors have assumed that the larger the country (in an economic sense), the larger its international transactions and the higher its needs for international reserves. In empirical studies, the size of a country is usually measured by its domestic product or its level of imports.

With respect to the variability of international transactions, it is assumed that the more variable such transactions are, the higher will be the desired amount of reserves. For empirical purposes various measures of international payment variability have been used; while some authors have used the export variability coefficient (Kenen and Yudin (1965), Iyoha (1976)), others have used some measurement of variation in the level of reserves as such (Heller and Khan (1978), Frenkel (1974), Saidi (1981), Edwards (1983 b)).

Most of the empirical studies have assumed that the more open the economy is to the rest of the world, the more vulnerable it is to exogenous shocks affecting its international liquidity position. Therefore, according to such authors, a greater degree of openness will entail greater demand for international reserves. The degree of openness is usually measured by the average propensity to import in each country. It should be borne in mind, however, that some authors (Heller (1966), Kelly (1970), Clark (1970 b)) have assumed that the relation between the degree of openness and the desired amount of reserves is negative.

Lastly, to the extent that by maintaining resources in the form of reserves countries are incurring a cost, the theoretical analyses have assumed that the alternative cost of maintaining reserves, measured by the domestic interest rate, negatively affects the desired level of reserves. Empirical studies, however, have repeatedly failed to find significant coefficients for this variable.13 Among the reasons given to explain these results, the following stand out: (a) the variables used as proxies of the opportunity cost of maintaining reserves have been inadequate (Williamson (1973) and Bird (1978)); (b) since a high percentage of reserves are maintained in the form of interest-earning assets, the alternative cost is (approximately) zero, and the nonsignificant coefficient obtained in the empirical studies is to be expected.

In this paper two approaches have been followed to estimate the function of demand for international reserves in developing countries. First, following the traditional literature on the subject, demand equations have been estimated using a single-equation method. It has been assumed that the desired amount of international reserves depends on a variable measuring the scale of the country (GNP, y), a variable measuring the degree of openness (the average propensity to import, m), and a variable measuring the degree of export income variability (standard deviation of export series corrected by the trend, s). It was further assumed that the adequate specification of this demand for reserves is double logarithmic:

where, based on the above discussion, it is expected that a1 > 0; a2>0; and a3 > 0;. On the other hand, u is a random error with the usual characteristics.

The second approach takes into account the fact that international reserves are only one of the possible mechanisms used to face external imbalances. Specifically, it was assumed that in addition to changes in reserves, variations in the level of indebtedness could be used to relieve these disequilibria. Thus, the debt was also included as an additional explanatory variable when the demand for reserves was being estimated. The estimate was accordingly made using simultaneous estimation methods. Following the discussion in Section 2, it was assumed that the quantities demanded of both assets are determined simultaneously and may be expressed as follows:14

where R = international reserves and D = level of debt. On the other hand, I is the average propensity to invest and is included in equation (3) to account for the fact that in the long run foreign debt is used (partially) to finance the gross domestic investment of the country concerned. In this paper, the external debt demand function is not estimated, since to the extent that some countries have limited access to external credit, this function cannot be estimated using semi-conventional econometric methods. In this case we are dealing with a situation of markets in disequilibrium, which means it will be necessary to use econometric methods for disequilibrium situations, such as those proposed by Goldfeld and Quandt (1976).15 Consequently, in this paper, analysis will focus on estimating equation (2) using an instrumental variables method.

One difficulty in estimating equation (2) is that, as discussed in Section 2, in theory at least, countries may also use exchange adjustments to balance their external position. This means that in a general equilibrium setting it should be possible to estimate equation (2) simultaneously with an equation describing the determinants of such exchange rate adjustments,16 This procedure was tried in the present context with negative success: hence the results obtained are not included in this paper.

Demand for Reserves in Developing Countries: 1975–80

Equation (1) was estimated using data for 19 developing countries for the period 1975–80. Cross-sectional data for each year were used for the estimation. (See Table 4, for a list of the countries included.) Table 1 contains the results obtained, where figures in parentheses are t statistics and MSE is the mean quadratic error of the regression.

Table 1.Estimation of Demand for International Reserves, 1975–80

logRn =a0 +a1 logyn +a2 logmn +a3 logsn +mn(OLS1 =197580)

Const.log Ynlog mnlog snR2MSE
1975–2.6781.3421.4820.8730.7510.688
(–1.766)(6.452)(2.842)(1.923)
1976–1.6791.1420.7520.7820.7790.486
(–1.353)(6.745)(1.892)(2.075)
1977–2.1061.1280.7110.4550.8030.404
(–1.692)(7.082)(1.941)(1.058)
1978–2.2491.1170.7670.2300.8550.286
(–2.093)(8.502)(2.437)(0.672)
1979–2.6031.1750.8120.4020.8190.409
(–1.932)(7.210)(2.011)(1.047)
1980–2.6681.1400.8900.2360.7930.481
(–1.887)(7.372)(2.314)(0.589)
Note: t-statistics are shown in parentheses.

Ordinary least squares.

Note: t-statistics are shown in parentheses.

Ordinary least squares.

Table 2.Demand for International Reserves(Regressions with combined cross-sectional and time series data; Fuller-Batesse method)
19761977197819791980
Constant0.273–1.025–1.116–1.027–0.882
(0.141)(–0.753)(–0.920)(–0.636)(–0.482)
Income1.1081.3761.4211.5011.652
(4.227)(5.489)(6.299)(5.062)(4.745)
Average propensity–0.0480.3620.5260.4700.672
to import(–0.775)(0.967)(1.659)(1.159)(1.972)
Export variability0.8400.5320.4230.5890.473
(1.578)(1.204)(1.139)(1.316)(1.468)
Debt–0.581–0.442–0.497–0.586–0.796
(–2.351)(–1.572)(–1.946)(–1.658)(–1.794)
R20.6630.8050.8600.8110.790
MSE0.6100.4040.2810.4370.504
Note: t-statistics are shown in parentheses.
Note: t-statistics are shown in parentheses.
Table 3.Country Risk Determinants, 1976–80(Regressions with combined cross-sectional and time series data; Fuller-Batesse method)
VariableCoefficientAsymptotic value

of t statistic
Constant0.3291.422
Debt/output0.6222.512
Debt service/output0.4261.688
Reserves/output–1.155–2.164
Debt duration–0.012–0.648
Debt volume–0.001–1.340
Investment/output–0.681–0.681
Current account/output0.4351.966
σ20.022
τ20.054
Ω20.019
Table 4.Estimated Probabilities of Nonperformance, 1976–80(In percent)
Country19761977197819791980
Greece8.37.97.97.47.6
Portugal10.79.89.19.39.6
Spain8.79.08.58.28.5
Yugoslavia8.18.17.97.07.3
Argentina7.88.18.16.85.4
Brazil8.88.98.67.58.1
Colombia10.19.68.77.99.5
Ecuador7.37.57.37.57.5
Mexico8.98.89.18.17.6
Panama10.412.312.912.113.1
Uruguay9.610.58.88.19.5
Venezuela5.66.15.76.46.7
Indonesia9.79.59.08.27.8
Korea9.99.38.37.77.6
Malaysia8.28.07.57.27.0
Philippines7.57.87.47.47.8
Thailand7.97.97.78.08.1
Ivory Coast10.210.19.89.09.0
Morocco8.99.09.19.210.0

These results are interesting in several ways. To begin with, the adjustments—as measured by both R2 and MSE—are highly satisfactory, especially since we are dealing with cross-sectional data. Second, all coefficients have the expected sign and show remarkable stability over time. Furthermore, the result of an F test for stability was 1.584. Third, the fact that the coefficient of log y exceeds the unit value in all cases shows that there are diseconomies of scale in the holding of international reserves by developing countries. That is, economic growth in these countries (increases in y) will result in more than proportional increases in the desired amount of reserves. It is most interesting to find that this result, which had been obtained earlier by the author using data for the Bretton Woods period (Edwards (1983 a)), is still valid for the recent period.

Another interesting point about the results presented in Table 1 is that for every year the coefficient of log m—the average propensity to import—is positive and significant, showing that the more open the economy, the higher is the desired level of reserves. This result has some major implications in economic policy. To the extent that the trend toward greater openness (higher value of m) observed in developing countries in the recent past continues in the future, such countries will face growing needs for international liquidity.

As to the coefficient of log s, although it is positive in all cases, as expected, it is only significant at conventional levels in 1975 and 1976. This contrasts with the results obtained earlier by the author (Edwards (1983 a) and (1984)) using data for 1964–72, where the coefficient of international payment variability proved significant in most of the cases.

Simultaneous Estimation of Demand for Reserves

As mentioned earlier, in addition to estimating equation (1), a log-linear version of equation (2) was simultaneously estimated taking into account the fact that the desired quantities of debt and reserves are simultaneously determined for the various countries. The results obtained are given in Table 2.

These results are quite interesting. In the first place, the negative coefficient of log D in all years supports the hypothetical substitutability of reserves and debt. The foregoing means, as in the case of Eaton and Gersovitz (1980), that reserves and debt are used as alternative mechanisms to deal with external adjustments. However, it should be noted that the degree of significance of these coefficients is not very high: 5 percent in 1975 and 1978 and 10 percent in the remaining years. Second, the results obtained for the coefficients of log y, log m, and log s in estimating the demand for reserves confirmed the conclusions arising from the single-equation analysis presented in Table 1, although log m is now negative in 1976. In particular, for every year the coefficient of log y is greater than 1, again pointing to the presence of diseconomies of scale.

4. INTERNATIONAL RESERVES, FOREIGN DEBT, AND COUNTRY RISK

This section contains an empirical analysis of the relation between foreign debt, international reserves, and country risk. In the event that sovereign debtors may repudiate their debt without giving up (total) control over the assets financed with it, the international financial community will tend to charge interest rates proportionate to the likelihood of a given country failing to pay its debts. This probability of default is known as country risk.17

The assessment that banks and financial institutions make of the likelihood of default will be reflected in the terms under which loans are granted by them to the various countries. In an extreme case, should the probability be deemed very high, the country in question will be totally excluded from the credit market.18 On the other hand, countries participating in the credit market will tend to pay higher interest rates (and/or to be granted shorter terms), as the likelihood that they may fail to pay is perceived to be higher. In the Eurocurrency market the various assessments of default probability will be reflected in varying degrees of spread above the London interbank offered rate (LIBOR) charged to the different countries.

In this section data on 727 public and publicly guaranteed loans granted to 19 developing countries between 1976 and 1980 are used to make an empirical analysis of the determinants of country risk. From an economic policy standpoint, the most interesting issue in this analysis is determining how appropriate management of foreign debt, international reserves, and other economic variables will allow a country’s probability of default, as perceived by its creditors, to remain at comparatively low levels, thus aiding a continuous flow of foreign funds toward the country in question.

The Model

The results presented in this section are based on Feder and Just’s (1977) model on country risk and probability of default. This model assumes monopolistic competition in the international financial market and that banks maximize expected profits. In this context, the interest rate applied to country n will be given by:19

where rn is the spread over the LIBOR applied to country n; ηn is the elasticity of demand for foreign credit with respect to the interest rate; θ is equal to (1 –(1 + r*)N)/r*, where r* is the average cost of capital of banks and N is the average loan duration; Pn is the probability of default of n, which is assumed to depend upon a vector x of economic variables relevant to that country; and where h is the proportion of the loan that the bank expects to lose in case of default.

Assuming that the default probability Pn (xn) has a logistic distribution function, we may write:

where xn is the vector of probability determinants, and the β, the associated coefficient.

By combining equations (4) and (5), using the properties of the logistic distribution function, and making some simplifications. Feder and Just (1977) write the following estimable equation with combined cross-sectional and time-series data:

where un, vt and wnt are stochastic errors; β0* = μ + log h, and μ is the expected value of log [ηn/1 – ηn].

The Data

In order to estimate equation (6) the determinants of the probability of default Pn must be established. Furthermore, in this paper the constant is estimated as γo = log βo* – log θ.

In the present study a series of variables were included in vector xn:

  • The ratio of public and publicly guaranteed debt to domestic product. It is assumed that this variable, which has been used previously in country risk studies by Frank and Cline (1971), Feder and Just (1977), and Sachs (1981), among others, will have a positive coefficient showing that the higher the debt ratio, the greater the perceived default probability. Data on this variable were taken from the World Debt Tables of the World Bank.

  • Ratio of public and publicly guaranteed debt service to exports. This variable measures the cash flow problems that a country may face in paying its foreign obligations. The coefficient of this variable is also expected to be positive, showing that the greater the proportion of export earnings that must be devoted to serve the debt, the higher the probability of default. Data for this variable were also taken from the World Debt Tables.

  • Ratio of international reserves to output. This variable measures the level of international liquidity maintained by a country at a given time. It is assumed that the greater the value of this variable, the lower the probability of default assigned to this particular country by the international community. The data for computing these ratios were taken from the IMF publication International Financial Statistics.

  • Loan duration. This variable measures the average term of the loans granted in year t to the country in question. Generally speaking, it may be assumed that international banks will use loan duration as an alternative (or additional) way to discriminate among countries with varying degrees of default probability. In this connection, it may be supposed that banks use spread and loan duration simultaneously as cover against that probability. Hence, a negative coefficient would be expected for this variable in the regression analysis. The variable was constructed as a weighted average of the duration of individual loans obtained by each country. The basic data were taken from the World Bank publication Borrowing in International Capital Markets.

  • Average loan amount. This variable, based on information contained in Borrowing in International Capital Markets, measures the role of country size (from a financial standpoint) in determining the perceived probability of default.

  • Ratio of gross domestic investment to output. This variable measures the future ability of the country to pay.20 A greater propensity to invest points to higher future growth of the country’s output, hence lower probability of default. The coefficient of this variable is therefore expected to be negative in the regression analysis. Data on this variable were taken from World Development Report.

  • Ratio of current account to output. This variable, also obtained from the World Development Report, measures the requirement of foreign resources to finance gross investment. Its sign in the regression should be negative, showing that greater deficit (lower surplus) increases the probability of default. In regressions where the propensity to invest and the current account ratio are simultaneously included, the coefficient of the latter is interpreted as follows: the greater the current account deficit, with a given propensity to invest, the larger the fraction of capital accumulation financed with external savings. It is then intuitively clear that the larger the fraction of gross investment financed with external savings, the higher the probability of default.

Other variables were also considered in the regression analysis as possible default probability determinants, such as output growth rate, export variability, and size of the public sector in the economy. Their inclusion, however, did not affect the results to any significant degree.

The dependent variable r was constructed for each country and each year as the (weighted) average spread over the LIBOR. The basic data on spread applied to individual loans were taken from Borrowing in International Capital Markets, (See Table 4, for a list of the countries included.)

The Results

If it is assumed that errors un, vt and wnt in equation (6) have the following properties:

Equation (6) may be estimated following the procedure suggested by Fuller and Batesse (1974) for estimating pooled cross-sectional and time series data.

The results obtained from estimating equation (6) with the Fuller and Batesse (1974) procedure are presented in Table 3.

In the first place, the results obtained show that the international financial community takes into account the behavior of a number of economic variables in order to determine the rate of interest to be charged to any one country. As expected, the coefficient of the external debt/output ratio is positive, which indicates that a higher volume of debt is reflected in a higher interest rate. While debt, reserves, and current account ratios are significant at 5 percent, the debt service ratio is significant at 10 percent.

What is surprising to a certain extent is the high value (in absolute terms) and high level of significance of the international reserves-to-output ratio. This result suggests that with prudent management of international reserves it is possible to keep the country risk premium relatively low. What is surprising, nonetheless, is that the coefficient of the international reserves ratio should differ significantly from the debt/output coefficient. This would mean that if foreign debt is used exclusively to finance accumulation of reserves, a country may in fact bring about a reduction in the default probability assigned to it by the international financial community. This is not easy to understand, since it is well known that reserves are a highly volatile international asset liable to decrease rapidly. A recent instance of a country where a policy of debt with equivalent accumulation of international reserves was followed on purpose was Chile during the period of so-called “neutral monetary policy” (1979–82). The recent results of the Chilean case (1981–83), however, have shown that in spite of that policy the international financial community began in early 1982 to perceive a significant increase in Chile’s probability of default.

Nevertheless, the results presented in this section should produce some doubts on the international banks’ ability to discern countries of different risk levels. The fact that some of the most important variables are not significant, or have an incorrect sign, lead one to believe that the origin of the current external indebtedness crisis lies to no small extent in the inability of banks to actually identify the pertinent economic variables. However, the results presented here will be useful for developing countries, as they clearly show what variables banks have taken into account to determine individual country risks. Of particular importance is the finding that if international reserves are suitably managed, the level of perceived probability can be kept at a reasonable level.

Probability of Default

The results presented in Table 3 together with assumptions on possible values of βo may be used to obtain information on the evolution of the probability of default perceived in each year by the international financial community. By way of illustration, Table 4 presents the probabilities of default for each country between 1976 and 1980, obtained by assuming that βo = γo – 1.75.

As may be observed in Table 4, the perceived probabilities of default, computed in this fashion, differ considerably every year from one country to another and fluctuate somewhat within each country over time. Nevertheless, it may be noted that even in 1980, the international financial community had not grasped the magnitude of the problems countries such as Mexico, Argentina, and Brazil were approaching. In fact, on the basis of Table 4, the perceived probability of default by these countries tended to decrease between 1977 and 1980.

Analysis of Residuals

The preceding sections discussed the economic determinants of default risk perceived by international banks. It is clear, however, that in addition to economic variables there are political considerations that will affect the probability of default perceived for any country. Generally speaking, it will be expected that countries with more political stability will be assigned lower probabilities of default.21

While our discussion in preceding sections has not explicitly considered political stability factors, it is possible that they are reflected by the residuals of the regression analysis. In particular, a hypothesis worth pursuing is: The most politically unstable countries will show positive (average) residuals in the regression analysis, whereas more stable countries will show negative residuals. If we define average residuals as

where rnt is the observed spread and βhGLSχnt is the estimated spread using the GLS (general-least-squares) method of Fuller and Batesse (1974). A positive value of RES says that, on average, the spread corresponding to that country exceeds the model prediction. This would be an indication that such a country would be politically unstable in relation to the sample.

Table 5 contains the results obtained from calculation of such average residuals. As may be observed, the results offer some comparatively surprising points. For instance, it is interesting that Spain should have the lowest residual, which, according to our hypothesis, would class it as one of the most politically stable countries in the sample. Conversely, countries with “strong” governments, such as Korea, turn up quite high on our “instability” scale. Undoubtedly, then, although the results presented in Table 5 are not lacking in interest, the analysis of the relation between perceived probability of default and political instability requires much more work. The use of residuals obtained from regressions like equation (6) appears to be an interesting avenue along which to conduct this line of research.

Table 5.Average Residuals by Country
CountryAverage

Residual
CountryAverage

Residual
Greece–0.113Uruguay0.076
Portugal0.039Venezuela0.059
Spain–0.202Indonesia0.078
Yugoslavia0.253Korea–0.127
Argentina0.185Malaysia–0.248
Brazil0.111Philippines0.120
Colombia–0.078Thailand–0.061
Ecuador0.117Ivory Coast0.225
Mexico–0.379Morocco0.087
Panama–0.142

5. NOTE ON THE DYNAMIC ADJUSTMENT OF INTERNATIONAL RESERVES AND MONETARY POLICY22

A number of studies have analyzed the dynamic adjustment process of demand for international reserves.23 In general, these papers show that the movement of the logarithm of international reserves over time responds to the discrepancy between desired reserves (log R*) and reserves actually maintained. In this context, the following partial adjustment equation has been considered:

Notwithstanding, in an economy with a fixed exchange rate the dynamic behavior of international reserves will not respond only to the discrepancy between desired and actual reserves, but also to the situation in the monetary sector of the economy. In particular, the movement of international reserves will be affected by excess demand for (or supply of) money: with a given domestic credit, excess demand for money will result in an increase in actual international reserves. The reason for this is, as indicated by the essential message of the monetary approach to the balance of payments (Frenkel and Johnson (1976)), that in an economy with fixed exchange rates the quantity of money—both nominal and real—is determined by demand conditions.

In the event that the movement of international reserves is actually affected by imbalance in the monetary market, the partial adjustment equation (8) should be modified to take this into account. Furthermore, exclusion of monetary influence in the empirical formulation might result in biased coefficients in the estimation of demand for reserves. One way to include monetary considerations in the dynamic analysis of demand for reserves, and to integrate the analysis with the monetary approach to the balance of payments, is to use the following equation:

Equation (9) indicates that the movements of (real) international reserves over time respond to two factors: discrepancies between desired and actual reserves and excess demand for money. While parameter λ captures the speed of adjustment between desired and actual reserves, parameter a gives the proportion of excess demands for money resolved, on average, by accumulation of reserves.

Equation (9) may be estimated in different ways. One, for example, is to postulate the structural forms of log R* and log M*, insert the resulting expressions in equation (9), and estimate a reduced form, using ordinary least squares (OLS), for reserves actually maintained. One difficulty about this procedure, however, is that in the event that both demand for reserves and demand for money depend upon common variables (income level, for instance), it will not be possible to distinguish the individual coefficients of each function. An alternative procedure used in this section is to estimate equation (9) in two stages. In the first stage a partial adjustment equation is estimated for the demand for money:

In the estimation of equation (10) both M and M* are expressed in real terms, and it is assumed that long-run demand for money is characterized by a Cagan function.

Therefore the equation to be estimated is:

From the estimation of equation (12), the coefficients of long-run demand for money are obtained. Then it is possible to estimate the desired quantity of money at any time, log M*. Having obtained these values and assuming that long-term demand for international reserves may be expressed as follows:

the following equation may be obtained for dynamic analysis of adjustment of international reserves:

where log Mt* is the estimated value of the desired quantity of money, using the coefficients obtained in the first stage of our process. All the equations contained in this section assume that both the quantity of reserves and the quantity of money are expressed in real terms. This assumption was retained for the empirical estimation of the model.

Empirical Results

Equation (14) was estimated using combined cross-sectional and time-series data for 23 developing countries during 1965 and 1972. The countries considered, as well as the period of time used, respond to the need to base the analysis on cases of fixed parity. For that reason the period following the collapse of the Bretton Woods system was excluded from the analysis. In addition, the sample countries maintained a fixed exchange rate throughout the period, which simplified the use of variables expressed in a common currency (U.S. dollars).

In the first stage—estimation of demand for money—all variables were expressed in U.S. dollars and expected inflation was replaced by actual inflation. A well-known difficulty about dynamic analyses using pooled time-series and cross-sectional data is that in the event that the error contains a country-specific element the use of OLS will result in biased coefficients. For this reason, equation (14) was estimated using dummy country variables. In the second stage of the estimation process the coefficients obtained in the first stage were used to construct the excess demand for money (logM*ntlogMntl) variable to be used for estimating equation (14). Estimation of this equation using the OLS method with dummy country variables yielded the following result:

As can be seen, all the coefficients have the expected signs. Furthermore, the coefficients of the variables of scale and lagged reserves are 5 percent significant, whereas the coefficient of excess demand for money is 10 percent significant. These results show that for these countries during this period, the movements of international reserves responded with nearly the same speed to monetary imbalances and to discrepancies between actual and desired reserves. On average, a unit imbalance between desired and actual reserves is corrected in 26.4 percent in the first year through changes in the amount of reserves. On the other hand, a unit imbalance between desired and actual quantity of money will be corrected in 30 percent in that year by accumulation or disaccumulation of international reserves.

The interesting point about these results is that they explicitly integrate the monetary sector imbalance with the analysis of demand for international reserves. They show that even if a country is obviously deficient in reserves (i.e., when actual reserves are very much lower than desired reserves), it can disaccumulate reserves if the monetary policy leads to an excess supply of money. From the standpoint of external management policy, these results emphasize the need to maintain compatible monetary and external policies. Although the results were obtained for countries with a fixed rate of exchange, the main lesson—the need to maintain coherent policies—extends to any kind of exchange rate arrangement.

6. SOME CONSIDERATIONS ON THE LATIN AMERICAN CASE

In the past few years the Latin American countries have significantly increased their foreign debt. In some cases public and publicly guaranteed debt has exceeded 40 percent of gross domestic product. Table 6 presents information connected with the level of public debt in Latin American countries in 1980. Column 1 contains the level of public and publicly guaranteed debt (D) on December 31, 1980, in millions of U.S. dollars. Column 2 in turn presents the level of public and publicly guaranteed debt obtained from private creditors, also in millions of U.S. dollars (DAP).

Table 6.External Debt of Latin American Countries, 1980
(1)(2)(3)(4)(5)(6)(7)
CountryDDAPDPSGD/ySt/ySt/xT/V
(In millions of U.S. dollars)(In percent)
Argentina10,2858,3736,59315.43.017.757.7
Bolivia2,124994n.a.36.44.725.932.1
Brazil38,26031,84416,60516.63.534.567.6
Colombia4,2941,84852513.21.710.033.9
Costa Rica1,585911n.a.34.34.316.448.1
Chile4,8853,3174,69318.05.222.951.8
Dominican
Republic1,18647625318.32.412.243.4
Ecuador2,6712,017n.a.24.53.713.962.7
El Salvador50911n.a.15.31.23.73.6
Guatemala5411n.a.6.90.83.50.2
Haiti2499n.a.17.80.94.13.0
Honduras89218619136.93.99.919.4
Mexico33,49028,998n.a.20.64.932.170.6
Nicaragua1,496586n.a.71.93.814.93.9
Panama2,2761,721n.a.71.314.514.054.0
Paraguay66727115115.52.112.99.3
Peru6,2043,046n.a.33.78.125.225.2
Uruguay1,04171921110.71.923.223.2
Venezuela10,86710,444n.a.17.94.913.277.9
Q115.41.912.2
M18.03.714.0
Q334.34.922.9
Source: World Bank, World Debt Tables.n.a. indicates data not available.
Source: World Bank, World Debt Tables.n.a. indicates data not available.

Columns 4 to 6 contain some interesting debt ratios. While column 4 gives the ratio of public and publicly guaranteed debt to gross domestic product (D/y), column 5 contains the ratio of total service for this debt to product (ST/y), and column 6 contains the ratio of total debt service to exports (ST/x). The first quartile (Q1), the median (M), and the third quartile (Q3) are shown at the bottom of the table. Column 7 presents the percentage of public and publicly guaranteed debt subject to variable interest rate (LIBOR plus a specified spread, for example). Lastly, column 7 contains information for selected countries on private non–publicly guaranteed debt.

The information given in Table 6 is useful to analyze the degree of (relative) difficulty in which some countries of the region already found themselves in 1980. It is interesting, in the first place, to consider the countries whose debt-to-GDP ratio (D/y) and debt service-to-export ratio are in the top part of the distribution of these two variables. In particular, the following countries have their debt ratio in the top 25 percent: Costa Rica (34.3 percent), Bolivia (36.4 percent), Honduras (36.9 percent), Panama (71.3 percent), and Nicaragua (71.9 percent). In addition, in the following countries the debt service-to-export ratio is in the top 25 percent of the distribution: Chile (22.9 percent), Bolivia (25.9 percent), Peru (31.9 percent), Mexico (32.1 percent), and Brazil (34.5 percent). As may be observed, most of the countries that are facing serious difficulties today in the matter of external payments are found in one of these two groups. Table 7 provides information on the indebtedness of these countries in 1981.

Table 7.External Debt of Latin American Countries, 1981
(1)(2)(3)(4)(5)(6)(7)
CountryDDAPDPSGD/ySt/ySt/xT/V
(In millions of U.S. dollars)(In percent)
Argentina10,50610,93012,166n.a.n.a.18.259.4
Bolivia2,4221,174n.a.31.93.726.935.2
Brazil43,99944,79419,79216.03.131.968.2
Colombia5,0763,16890214.02.08.440.1
Costa Rica1,8541,103n.a.14.17.28.545.2
Chile4,4233,2298,13814.15.327.248.7
Dominican
Republic1,261471232n.a.n.a.n.a.40.5
Ecuador3,3922,621n.a.26.94.317.962.7
El Salvador664355n.a.19.31.3n.a.2.5
Guatemala684302n.a.8.00.63.31.2
Haiti29571n.a.18.71.35.41.9
Honduras1,22341417247.14.412.722.8
Mexico42,64238,711n.a.18.53.728.275.0
Nicaragua1,975772n.a.77.47.2n.a.2.7
Panama2,3771,817n.a.64.513.411.552.3
Paraguay70745513313.81.4n.a.7.5
Peru5,9743,956n.a.28.69.144.925.8
Uruguay1,3121,08432612.21.6n.a.33.5
Venezuela11,35211,168n.a.163.94.512.581.3
Source: World Bank, World Debt Tables.n.a. indicates data not available.
Source: World Bank, World Debt Tables.n.a. indicates data not available.

The increased foreign debt of the countries of the region has been due to a number of reasons. On the one hand, starting in the mid-1970s the Eurocurrency markets made significant amounts of resources available to developing nations. That is, with the boom in the international private capital markets (and with the recycling of petrodollars), the supply of external funds available to developing countries grew significantly. On the other hand, the financial reforms implemented in some Latin American countries (particularly in the Southern Cone) toward the late 1970s resulted in major increases in demand for external funds in those countries.

One extremely important fact that is not properly reflected in Tables 6 and 7 is that a significant volume of the new debt was channeled to the private sector of the debtor countries. Thus, for instance, the private foreign debt of Costa Rica totaled about $1,200 million in 1980, whereas public foreign debt amounted to $1,585 million. In Chile, the private sector foreign debt was approximately $10,000 million in 1981, while public and publicly guaranteed debt amounted to only $5,400 million.24 In Mexico, the private sector foreign debt amounts to about $35,000 million at the present time (1983).

The growing significance of private external debt in the recent past requires further analysis. Among major issues connected with this topic, the following at least should be mentioned.

  • Effective regulation and supervision of domestic intermediation activity with respect to the new external resources. In general, as mentioned before, the new foreign debt has gone side by side with a process of liberalization of the domestic-financing sector. Once the financial “repression” ends, new banks and financial institutions emerge in the domestic market, intermediating internal and external resources. However, as shown very clearly by recent experience in Chile and Argentina, the lack of experience of such institutions may result in disastrous management of such resources, which in both the cases mentioned ended in large losses that were eventually absorbed—totally or partially—by the respective governments (i.e., the taxpayers). This points to a need for designing extremely efficient policies for regulating the financial and banking sector. Only in this way will it be possible to prevent new flows of external resources from being misused.

  • Determination of an “optimum” foreign debt policy. To the extent that increases in foreign debt, as shown in Section 4 above, result in growing cost of debt, it is advisable to restrict the volume of a country’s debt. This argument in favor of intervening in the external loan market emerges from direct application of the idea of an “optimum tariff.” In fact, while debt cost grows with a larger volume of debt, the country concerned will benefit from imposing a tax on debt equal to 1/e, where e is the elasticity of the external funds supply curve in respect of the interest rate. It should be noted, however, that the relevant supply curve should take account of perceptions of the risk of default as felt by both creditor and debtor.

A salient feature of the behavior of the Latin American countries is that in spite of considerable increase in foreign debt no equivalent increase has been observed in total savings. While in some cases total savings have remained practically constant (Argentina and Brazil), in others, total savings have dropped significantly as percentage of the product (Chile, Peru, Uruguay). The exceptions to this rule are mainly Colombia and Venezuela.25 This suggests, as we find in the relevant literature, that a not negligible portion of the new foreign debt has been used in the region to finance consumption.

Under present circumstances, when a significant number of Latin American countries have difficulty in paying their foreign debt, it seems obvious that new efforts must be made in the management of international reserves. In particular, in the event that access to foreign credit is (temporarily) reduced in the next few years, the need for reserves to face external payment difficulties will increase. This fact arises from the results of the empirical analysis contained in Section 4 above and shows that if the use of one of the adjustment instruments utilized to date (debt) is limited, a need to make more intensive use of other available instruments will arise.

In this connection, for the next few years economic discussion in the region should focus on the design of an appropriate exchange rate policy. Recent exchange rate crises in a number of countries of the region (Argentina, Chile, Mexico) show that this precisely has been an area where economic management has been deficient. First of all, it is essential to include in the analysis the fact that these countries are situated in a world of (semi-) floating exchange rates. This means that pegging the exchange rate to a given currency (the dollar, for instance) automatically means floating with respect to the other currencies of industrial countries. As the recent (1981–82) experience in Chile has shown, this may turn out to be disastrous when the value of the currency to which the domestic currency has been pegged varies widely in the short term. The foregoing discussion, then, indicates that under present institutional arrangements prevailing in the international monetary system, the exchange policy becomes even more complex. An area of applied research that the economic policymakers of the region would do well to examine in detail is the determination of “currency baskets” suitable for guiding the country’s exchange rate policy.

The second major issue is the determination of exchange rate rules. To date many countries of the area, at varying periods, have followed exchange rules tending to keep a fixed real exchange rate. Such rules known as purchasing power parity (PPP) may, however, prove detrimental to the operation of the economy, in that they set up rigid links between the exchange rate, domestic prices, and international prices. As the real exchange rate is a key relative price, it is important that the exchange rate policy abstain from inhibiting its equilibrium changes.26

Lastly, in relation to exchange rate policy, it is also important to define the optimum degree of float or pegging. Generally speaking, from a theoretical standpoint, the optimum degree of exchange rate flexibility will depend on variables such as the degree of openness of the economy and the nature of the shocks affecting it. Frenkel and Aizenman (1982) have stated that the optimum exchange rate will tend to be more rigid as the variance of real shocks affecting the supply of goods and services is greater.

7. FINAL COMMENTS

This paper has analyzed various aspects of the external adjustment process in developing countries. Emphasis has been laid in particular on the role that international reserves and foreign debt play in such adjustment process. The empirical analysis based on data for 19 selected developing countries during the period 1975–80 shows that these countries have made use of reserves and debt as substitutes in the adjustment process.

The empirical analysis also reveals that management of both these variables—level of reserves and level of debt—affect the perception of the international financial community regarding the risk involved in lending to a given country. Higher reserves reduce the perceived country risk, while higher foreign debt increases it. This greater perceived risk will in general bring about different effects among the smaller countries. First, it will take the form of higher cost of any foreign loan they may obtain, and/or a shorter loan term. Second, if the increased risk perceived is high enough it may result in exclusion of the country in question from the foreign loan market. This suggests that in the event that developing countries wish to keep up a stable flow of external funds, they should show care in managing their reserve and debt levels in order not to raise the level of perceived risk above a conservative figure.

In this paper, also, the dynamic analysis of demand for international reserves was integrated with the analysis of equilibrium conditions in the monetary sector of the economy. The analysis performed for 23 selected developing countries with a fixed exchange rate during the period immediately prior to the collapse of the Bretton Woods system points out that the amount of reserves maintained by a country will move over time for two reasons: first, the movement will tend to be related to discrepancies between the quantity of reserves actually kept by a country and the quantity of reserves desired in the long term. Second, the movement of reserves over time will also respond to excess demand in the monetary market. This result, which is consonant with the fundamental propositions of the modern approach to the balance of payments in a monetary economy, shows that international reserve management by the economic authorities should take particular account of monetary issues.

Finally, Section 6 of this paper contains some reflections about the Latin American case. In this connection, it is stated here that in the event that a number of countries of the region face difficulties to service their foreign debt, they will face restrictions for using (new) debt as an active instrument for facilitating the adjustment process in the next few years. This suggests that these countries will face even greater needs for international reserves in the next few years. This section also points out the need to outline effective exchange rate policies that will allow these countries to face situations of external imbalance with the lowest possible cost.

BIBLIOGRAPHY

    AliberRobert Z.“The Interest Parity Theorem: A Reinterpretation,”Journal of Political EconomyVol. 88 (April1980) pp. 37084.

    BachaEdmar Lisboa and Carlos F.Díaz Alejandro“International Financial Intermediation: A Long and Tropical View” (Princeton, New Jersey: Princeton University PressMay1982).

    BilsonJohn F.O. and Jacob A.Frenkel“Dynamic Adjustment and the Demand for International Reserves,”NBER Working Paper, No. 407 (Cambridge, Massachusetts: National Bureau of Economic ResearchNovember1979).

    BirdGrahamThe International Monetary System and the Less Developed Countries (London: Macmillan1978).

    BlanchardOliver Jean“Debt and the Current Account Deficit in Brazil,”Harvard Institute of Economic Research Discussion Paper Series, No. 882 (Cambridge, Massachusetts: Harvard University PressNovember1981).

    BrownWier M.The External Liquidity of an Advanced CountryPrinceton Studies in International Finance, No. 14 (Princeton, New Jersey: Princeton University Press1964).

    BuiterWillem“Implications for the Adjustment Process of International Asset Risks: Exchange Controls, Intervention and Policy Risk, and Sovereign Risk,”NBER Working Paper, No. 516 (Cambridge, Massachusetts: National Bureau of Economic ResearchJuly1980).

    ClaasenEmil-Maria“The Optimizing Approach to the Demand for International Reserves: A Survey” in Recent Issues in International Monetary Economicsed. by Emil-MariaClaasen and PascalSalin (Amsterdam: North-Holland1976) pp. 73116.

    ClarkPeter B. (1970 a) “Optimum International Reserves and the Speed of Adjustment,”Journal of Political EconomyVol. 78 (March–April1970) pp. 35676.

    ClarkPeter B. (1970 b) “Demand for International Reserves: A Cross-Country Analysis,”Canadian Journal of EconomicsVol. 3 (November1970) pp. 57794.

    ClowerRobert and RichardLipsey“The Present State of International Liquidity Theory,”American Economic ReviewVol. 58 (May1968) pp. 58695.

    CrockettAndrew D.“Control Over International Reserves,”Staff PapersInternational Monetary Fund (Washington) Vol. 25 (March1978) pp. 124.

    DooleyMichael P. and PeterIsard“Capital Controls, Political Risk and Deviations from Interest Rate Parity,”Journal of Political EconomyVol. 88 (April1980) pp. 37084.

    DornbuschRudiger“PPP Exchange-Rate Rules and Macroeconomic Stability,”Journal of Political EconomyVol. 90 (February1982) pp. 15865.

    DornbuschRudiger and StanleyFischer“Exchange Rates and the Current Account,”American Economic ReviewVol. 70 (December1980) pp. 96071.

    EatonJonathan and MarkGersovitz“LDC Participation in International Financial Markets: Debt and Reserves,”Journal of Development EconomicsVol. 7 (March1980) pp. 321.

    EatonJonathan and MarkGersovitz (1981 a) “Debt with Potential Repudiation: Theoretical and Empirical Analysis,”Review of Economic StudiesVol. 48 (April1981) pp. 289309.

    EatonJonathan and MarkGersovitz (1981 b) Poor-Country Borrowing in Private Financial Markets and the Repudiation Issue Princeton Studies in International Finance No. 47 (Princeton, New Jersey: Princeton University PressJune1981).

    EdwardsSebastian (1982 a) “Deuda externa, ahorro domestico y crecimiento económico en Chile: Una perspectiva de largo plazo, 1982–1990,”Estudios InternacionalesVol. 15 (July–September1982) pp. 260275.

    EdwardsSebastian (1982 b) “Ajuste cambiario y reservas internacionales: Un analasis empírico,”Cuadernos de EconomiaVol. 19 (August1982) pp. 193202.

    EdwardsSebastian (1983 a) “The Demand for International Reserves and Monetary Equilibrium: Some Evidence from Developing Countries,”UCLA Working Paper No. 293 (March1983).

    EdwardsSebastian (1983 b) “The Demand for International Reserves and Exchange Rate Adjustments. The Case of LDCs, 1964–1972,”EconomicaVol. 50 (August1983) pp. 26989.

    EdwardsSebastian (1983 c) “LDC’s Foreign Borrowing and Default Risk: An Empirical Investigation,”NBER Working Paper, No. 1172 (Cambridge, Massachusetts: National Bureau of Economic ResearchJuly1983).

    EdwardsSebastian“La demanda por liquidez internacional de los países en desarrollo: Un análisis dinámico,”El Trimestre EconómicoVol. 51(1) No. 21 (January–March1984) pp. 13146.

    FederGershon and Richard E.Just“An Analysis of Credit Terms in the Eurodollar Market,”European Economic ReviewVol. 9 (May1977) pp. 22143.

    FisherFranklin M.“Tests of Equality Between Sets of Coefficients in Two Linear Regressions: An Expository Note,”EconometricaVol. 38 (March1970) pp. 36166.

    FlandersM. JuneThe Demand for International ReservesPrinceton Studies in International Finance, No. 27 (Princeton, New Jersey: Princeton University Press1971).

    FrankCharles R. and William R.Cline“Measurement of Debt Servicing Capacity: An Application of Discriminant Analysis,”Journal of International EconomicsVol. 1 (August1971) pp. 32744.

    FrenkelJacob A.“The Demand for International Reserves by Developed and Less Developed Countries,”EconomicaVol. 41 (February1974) pp. 1424.

    FrenkelJacob A.“International Reserves: Pegged Exchange Rates and Managed Float” in Public Policies in Open Economiesed. by KarlBrunner and Allan H.Meltzer (Amsterdam; New York: North-Holland1978) pp. 111140.

    FrenkelJacob A.“The Demand for International Reserves Under Pegged and Flexible Exchange Rate Regimes and Aspects of the Economics of Managed Float” in The Function of Floating Exchange Rates: Theory Guidance and Policy Implicationsed. by DavidBigman and TeizoTaya (Cambridge, Massachusetts: Ballinger1980) pp. 16995.

    FrenkelJacob A.“International Liquidity and Monetary Control,”NBER Working Paper, No. 1118 (Cambridge, Massachusetts: National Bureau of Economic ResearchMay1983).

    FrenkelJacob A. and J.Aizenman“Aspects of the Optimal Management of Exchange Rates,”Journal of International EconomicsVol. 12 (November1982) pp. 23156.

    FrenkelJacob A. and J.Aizenman and Harry G.Johnsoneds.The Monetary Approach to the Balance of Payments (Toronto: University of Toronto Press1976).

    FrenkelJacob A. and J.Aizenman and Harry G.Johnson and BoyanJovanovic“On Transactions and Precautionary Demand for Money,”Quarterly Journal of EconomicsVol. 95 (August1980) pp. 25—43.

    FrenkelJacob A. and J.Aizenman and Harry G.Johnson and BoyanJovanovic“Optimal International Reserves: A Stochastic Framework,”Economic JournalVol. 91 (June1981) pp. 50714.

    FullerWayne A. and George E.Batesse“Estimation of Linear Models with Crossed-Error Structure,”Journal of EconometricsVol. 2 (May1974) pp. 6778.

    GoldfeldStephen M. and Richard E.Quandteds.Studies in New Linear Estimation (Cambridge, Massachusetts: Ballinger1976).

    GruebelHerbert G.“The Demand for International Reserves: A Critical Review of the Literature,”Journal of Economic LiteratureVol. 9 (December1971) pp. 114866.

    HamadaKoichi and KazuoVeda“Random Walks and the Theory of Optimal International Reserves,Economics JournalVol. 87 (December1977) pp. 72242.

    HarbergerArnold C. and SebastianEdwards“Lessons of Experience Under Fixed Exchange Rates” in The Theory and Experience of Economic Developmented. by MarkGersovitz and others (London; Boston: Allen and Unwin1982) pp. 18393.

    HellerH. Robert“Optimal International Reserves,”Economic JournalVol. 76 (June1966) pp. 296311.

    HellerH. Robert and Mohsin S.Khan“The Demand for International Reserves Under Fixed and Floating Exchange Rates,”Staff PapersInternational Monetary Fund (Washington) Vol. 25 (December1978) pp. 62349.

    Hipple F.StebThe Disturbance Approach to the Demand for International ReservesPrinceton Studies in International Finance, No. 35 (Princeton, New Jersey: Princeton University PressMay1974).

    IyohaMilton A.“Demand for International Reserves in Less Developed Countries: A Distributed Lag Specification,”Review of Economics and StatisticsVol. 58 (August1976) pp. 35155.

    KellyMichael G.“The Demand for International Reserves,”American Economic ReviewVol. 60 (September1970) pp. 65567.

    KenenPeter B. and Elinor B.Yudin“The Demand for International Reserves,”Review of Economics and StatisticsVol. 47 (August1965) pp. 24250.

    LoserClaudio M.“External Debt Management and Balance of Payments Policies,”Staff PapersInternational Monetary Fund (Washington) Vol. 24 (March1977) pp. 16892.

    McDonaldDonogh C.“Debt Capacity and Developing Country Borrowing: A Survey of the Literature,”Staff PapersInternational Monetary Fund (Washington),Vol. 29 (December1982) pp. 60346.

    MachlupFritzThe Need for Monetary Reserves reprinted in International Finance (Princeton, New Jersey: 1966).

    MakinJohn H.“Exchange Rate Flexibility and the Demand for International Reserves,”Weltwirtschaftliches Archiv (Kiel) Vol. 110 (1976) pp. 22943.

    MikesellRaymond F. and James E.Zinser“The Nature of the Savings Function in Developing Countries: A Survey of the Theoretical and Empirical Literature,”Journal of Economic LiteratureVol. 11 (March1973) pp. 126.

    NerloveMarc“Further Evidence on the Estimation of Dynamic Economic Relations from a Time Series of Cross Sections,”EconometricaVol. 39 (March1971) pp. 35982.

    ObstfeldMaurice“Aggregate Spending and the Terms of Trade: Is There a Laursen-Metzler Effect?”Quarterly Journal of EconomicsVol. 97 (May1982) pp. 25170.

    OliveraJulio H.G.“A Note on the Optimal Rate of Growth of International Reserves,”Journal of Political EconomyVol. 77 (March–April1969) pp. 24548.

    RipleyDuncan and Esther C.Suss“An Approach to the Estimation of Inequality in Reserve Distribution,”Staff PapersInternational Monetary Fund (Washington) Vol. 21 (November1974) pp. 78999.

    RodríguezCarlos Alfredo“The Role of Trade Flows in Exchange Rate Determination: A Rational Expectations Approach,”Journal of Political EconomyVol. 88 (December1980) pp. 114858.

    SachsJeffrey“The Current Account and Macroeconomic Adjustment in the 1970s” Brookings Papers on Economic Activity: 1 (1981) The Brookings Institution (Washington) pp. 20182.

    SachsJeffrey“LDC Debt in the 1980s: Risk and Reforms” in Crises in the Economic and Financial Structureed. by PaulWachtel (Lexington, Massachusetts: Lexington Books1982) pp. 197243.

    SachsJeffrey and DanielCohen“LDC Borrowing with Default Risk,”NBER Working Paper, No. 925 (Cambridge, Massachusetts: National Bureau of Economic ResearchJuly1982).

    SaidiNasser“The Square-Root Law, Uncertainty and International Reserves Under Alternative Regimes: Canadian Experience, 1950–1976,”Journal of Monetary EconomicsVol. 7 (May1981) pp. 27190.

    SvenssonLars E.O. and AssafRazin“The Terms of Trade and the Current Account: The Harberger-Laursen-Metzler Effect,”Journal of Political EconomyVol. 91 (February1983) pp. 97125.

    von FurstenbergGeorge M.“New Estimates of the Demand for Non-Gold Reserves Under Floating Exchange Rates,”Journal of International Money and FinanceVol. 1 (April1982) pp. 8195.

    WhalenEdward L.“A Rationalization of the Precautionary Demand for Cash,”Quarterly Journal of EconomicsVol. 80 (May1966) pp. 31424.

    WilliamsonJohn“International Liquidity: A Survey,”Economic JournalVol. 83 (September1973) pp. 685746.

    ZellnerArnold“An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests and Aggregation Bias,”Journal of the American Statistical AssociationVol. 57 (June1962) pp. 34868.

Commentary*

JUAN GUILLERMO ESPINOSA

I would first like to extend my thanks, not formally, but very sincerely to the organizers, and particularly the International Monetary Fund, for having afforded us the opportunity to have such frank and systematic discussions of our common concerns about the regional and international economic and financial situation.

The emphatic statements already made on the subject by most participants preclude the need to refer here to the importance and timeliness of including Mr. Edwards’s topic as a separate item on the agenda.

We are only too well acquainted with the crisis of payments and new financial resources that is facing the major economies of the region. It has resulted in considerable changes in economic policy, with significant effects on our countries’ national products, revenues, employment, and external economic relations.

In trying to comment on Sebastián Edwards’s paper, I am faced with the quandary of deciding between taking an “academic” or a “pragmatic” approach. An academic approach is, to be sure, more detached from the urgencies of the moment; it is almost “atemporal.” However, the depths of the current crises in our economies and the magnitude of the consequences referred to in other papers have prompted me to take the most realistic and current approach possible.

Furthermore, various participants have referred to the urgency of finding concerted, planned ways of resolving the difficulties before us, and I do not believe that an abstract approach would contribute to this cause.

The paper presented by Edwards is valuable, I think, primarily in that it is an interesting and concise synthesis of recent writings and studies on the topic by Anglo-American scholars. The paper would merit close examination merely for its copious bibliography and numerous references concerning both methodology and the ideas put forward by a wide range of scholars in the field.

Furthermore, it has the great merit of refreshing our memories concerning a number of the major assumptions or hypotheses of prevailing theory. What is more, it reminds us of some of the principal operational concepts or precepts implicit in the normal functioning of international finance and the management of the external debt.

The facts, however, appear to run ahead of our theories. The crises would appear to be telling us that most of the explanatory factors or variables considered valid for periods of “relative normalcy” are overtaken by circumstances to the point of being rendered virtually irrelevant during periods of economic and financial crisis. That is, it seems we need one theory for periods of “relative normalcy” and another for periods of “crisis,” when the financial system is no longer reliable, when we have run out of alternative savings instruments, when expectations begin to change suddenly, and so on.

Speaking constructively, I believe that the major critique that could be made of the paper relates more to what is not said or considered in it than to any errors which may have slipped into the text before us.

In his paper, Edwards is faced with the task of revising many of the concepts used in the first part of his study, which are based on the use of traditional background information and notions of indebtedness, when toward the end of his paper, he measures them against the actual situation in Latin America, which presents a host of new scenarios, with new problems that do not necessarily respond to the “conventional wisdom” we have accumulated over the last two decades. This is perfectly understandable as regards the constraints frequently imposed on us by figures; many of us have tried to verify or contrast various hypotheses in light of the real world—a process which proves to be virtually impossible when based on questionable data, especially in this field.

In other words, the current situation, or the reality we see, prompts us to reconsider various methods and concepts which now seem to be out of touch. I do not say this in order to enter into some abstract discussion of reality, but instead to assess certain key concepts and methods in a practical way in light of the situation we face, which has been very well described by the Fund itself.

One such concept is our notion of public debt and private debt. Recent experience and the experience of many countries in the 1930s suggests that in reality there tends to be no difference between public and private indebtedness. When problems are encountered, it is the economy as a whole which has foreign exchange or does not, which has international reserves or does not, which has a positive trade balance or a current account deficit or does not. The key problem would thus appear to be that almost always, almost inexorably, pressures are exerted to have the private debt paid off, directly or indirectly, by the public sector or the country. In practice, in almost all cases, that which had been the exclusive responsibility of the private sector is slowly but surely becoming a public responsibility which must be assumed by the national economy. Although this is mentioned in the paper, it is not incorporated into the estimates. I believe that due consideration of this matter could lead to changes in both the analysis and the results.

In this regard, though, the author reminds us early in his paper that everything is still set up for our traditional approaches to the public debt even though, in the last five years, the growth of debt originating in and for the use of the private sector has far exceeded the amount of public borrowing in several countries of the region. Consequently, our results continue to confirm the findings of economic writers of the 1960s and 1970s but are not very helpful in the present situation.

What is more, the problem goes much beyond an institutional or legal definition and points to a question which is basic and central to the entire model, namely, that in our economies everything has been set up so as not to impede the growth of the private sector, which we now find overindebted, falling short of expectations, without working capital, and with prospects for an indefinite period of adjustment and contraction.

The old approach assumed, and perhaps still largely assumes, that private sector borrowing without (or even with) a government guarantee was the form of external savings that would be most likely to be invested productively, and that, in the absence of government interference in the making of payments, private sector loans would pose the smallest debt service problems.

In other words, it was expected, or rather assumed, that excess indebtedness was unlikely, which brings us to my next point.

Our traditional belief that external indebtedness plays a long-term role in the financing of gross investment in the region would appear to be one of the assumptions which is the least borne out and the most refuted by recent evidence from various countries of the region, at least from those in southern Latin America. The author himself reminds us that, in recent experience in Latin America, the pronounced increase in external indebtedness has clearly not produced an equivalent increase in total savings. Perhaps what is most impressive is that not only has there not been increased investment, but investment has actually decreased, and that a sizable proportion of new external borrowing has been used to finance consumption.

As a number of authors have indicated, the large-scale recycling of funds over the past five years was a considerable financial success, but it was deceptive. It was not a recycling of savings into investment, but in part a shift from one form of savings into another or from savings into consumption. The old, most orthodox Keynesian theory holds that when the propensity to save is greater than the tendency to invest, a depression is created; this could explain much of the phenomenon we are now witnessing.

Financial recycling from one bank to another for any purpose other than investment is not sufficient—or its effects are not yet clear. From a banker’s perspective it may represent a success, but from the standpoint of the economic well-being of the rest of the world, it may constitute a step backward. What I am trying to say, among other things, is that in this financial recycling we have lost sight of the basic criterion for granting resources that was learned during the peak time of international cooperation, the objective of which was to maximize complementary national efforts and not just the income created by each dollar of aid.

Another way of illustrating the same problem is the pace at which credit expanded, from a low proportion of gross domestic product (GDP) in 1974–75 in various economies to levels of 80 to 90 percent of GDP in 1980–81.

Similarly, the “old wisdom” that private banks properly supervise and evaluate the use to which their loans are put appears in some cases not to coincide with the current situation in the major Latin American economies. This observation must of course be attenuated or qualified for at least two reasons: first, a multitude of banks (some estimate about 1,600) are involved in this process, and second, it is a problem of shared responsibility which must urgently be remedied. In other words, if this function has not been carried out by the banks, large or small, with or without their own analysts, or by the Fund (which arrives on the scene only when so requested by the deficit countries and when problems already exist), or by the central banks, which in the majority of the cases have instead tended to see increasing revenues from external credit as a sign of approval and confidence, we are faced with a rather paradoxical, virtually self-destructive situation.

To this must be added the concrete and widely recognized fact that each of these agents inclined (for various reasons, some of them rather creditable) not to provide adequate background information on the figures involved, which makes for an unreliable and lagging data base. This problem of the quality, timeliness, and adequate dissemination of information has not yet been mentioned, and I believe it is at the heart of the problems we are discussing today.

At present it is possible to consider only semitheoretically the alternative roles played by adjustments in domestic activity, the exchange rate, reserves, and external indebtedness in the macroeconomic adjustment process of the countries of the region, after examining the most recent figures on indebtedness and interest rates, among others. At times of economic crisis and declining international liquidity, these roles are no longer the usual ones or cease to have the same significance, as in the case of indebtedness, the mentioned cases of exchange rates, and other variables mentioned. This appears to be equally true of the banking practices we know, which are relevant to periods of normalcy; although it has been recommended to banks on innumerable occasions that they not follow the practice of first lending intensely and then abruptly curtailing their lending, in real life this practice is rather frequent, even when it can be disastrous for all.

Turning to another subject—estimates of country risk—I must stress the author’s forthrightness in portraying the results obtained from estimates, despite how disheartening this must have been. These results, like those of other studies, verge on suggesting that the type of analysis or the factors considered are irrelevant and that one or more dummy variables could just as well have explained much of the dependent variable, that is, the probability that a given country will not meet its commitments (probability of default). In other words, the analytical process at present, when it considers only or preponderantly credits to the public sector or government-guaranteed credits, overlooks all of the important and potentially difficult present situation. Therefore, the type of analysis presented still requires further study and more direct observation of reality with a view to learning from experience or from actual practice for forward projections. The quality and timeliness of data are again important here, as this is an effort to estimate risk in advance rather than after the fact. The results obtained from analysis as it now stands are in some cases impressive. To mention but a few of them, the coefficients of the debt service ratio and of the average volume of loans, and the coefficient of the ratio of current account to product have been found not to be significant in estimating country risk. On the other hand, the author stresses his surprise at learning the importance, in absolute terms and in terms of significance, of the ratio between international reserves and national product. This suggests that, by keeping the reserves-to-product ratio high, somehow, a country might sustain the impression among international creditors that it is very unlikely to default.

Finally, I was quite interested in the author’s analysis of the residuals factors involved in estimates of the probability of default or the probability that a given country will not meet its commitments. As the author points out, it is clear that in addition to economic variables, considerations of a political nature will affect the community’s perception of such a probability. In other words, politically more stable countries would be expected to have a lower default probability. Recent results appear to suggest that further study is still required of both the analytical framework and factors involved. Otherwise, current analytical methods lead us to conclude that there are several hard-to-interpret situations with respect to political stability and support from the international financial system.

Other matters not considered in the paper, but which cannot be examined here in greater detail because I have already spoken so long, include the effects of capital inflows and outflows on key prices in the economy and on the relative prices of tradables and nontradables. Also not considered are the effects of this large-scale financial recycling on aggregate supply and demand, and on the changes in their makeup; that is, the structural shifts which to varying degrees and with different nuances have caused decreases in the importance of the industrial sector and a more than proportional expansion of the financial, commercial, and services sectors of various economies of the region.

Yet another question, which is actually a more general problem that must be reconsidered in our analytical methodologies and is therefore not specific to the work being commented on here, is the concept of short-term indebtedness. Until 1981, or thereabouts, short-term financing was used primarily for purposes of trade. In the past two years, however, short-term financing, which in Latin America has reached an approximate total of about $90 billion, has been increasingly used for interbank support. This changes the basic character of the lending from commercial to financial.

Finally, in these last, more general remarks, no mention has been made of some important limitations derived from a model oriented toward export promotion. On the one hand, the most recent studies of Latin American economies indicate that the export sector of each economy makes a rather small contribution to overall employment. Today, with the stagnation and sizable adjustments experienced in the region in the last two years, unemployment has grown extraordinarily. We must therefore recall that development is fundamentally a question of human progress, and a man with no job can make no progress. On the other hand, while our exports may significantly increase in the near future owing to the new trade agreements between countries of the south, the related export proceeds do not ease the external debt service burden in that the debt is to a great degree, if not completely, made up of credits from the industrial economies whose currency of exchange is the U.S. dollar or another hard currency.

Commentary*

SERGIO MÁLAGA

The study just presented to us by Sebastian Edwards is, first of all, a fine survey of the recent literature on the subject of adjustment of disequilibria in the external sector of the economies of developing countries. This topic is eminently timely in light of recent events familiar to all of us, including the severe contraction of loanable funds available to Latin America, a disproportionate external debt in that part of the world, and the inability of most of its economies to service their external debts.

Second, Edwards takes an especially close look at the role of international reserves and external debt in this adjustment process and in the determination of the risk of default. He uses his own estimates, first individually and then simultaneously, of the demand functions for both reserves and debt, with data from 19 developing countries, to get an empirical handle on a number of hypotheses, some deriving strictly from economic logic and others being less simple to deduce. Thus, he concludes from his research that: (a) the income elasticity of the demand for reserves is greater than unity while that of the demand for debt is not significantly different from unity—hence, the need of the developing countries for international liquidity and debt in the next few years will grow at rates which are, respectively, greater than and equal to those by which their incomes grow; (b) reserves, external debt, and the exchange rate are substitutes whose combined role in the process of financing external adjustment has yet to be adequately examined; (c) the greater the ratio of debt to product and the lower the ratio of reserves to product, the greater the perceived likelihood of default—hence, rational management of their levels can keep the cost of external borrowing lower than it would otherwise be; (d) the greater the openness of the economies, as measured by the average propensity to import, the greater the required level of reserves; (e) both the demand for reserves and the demand for debt are apparently insensitive to the cost of obtaining either; (f) the greater the variability of export income, the greater the demand for reserves and debt, although for most years the coefficients found are significant only in the case of the demand for debt; (g) the greater the ratio of external debt to product and the lower the ratio of reserves to product, the higher the interest rate charged by international banks as a premium for the higher level of country risk; and, finally, (h) the greater the risk perceived by banks, the higher the interest rates and the shorter the maturities on their loans.

At this stage of the analysis, one wonders whether the demand functions for reserves and debt, estimated on the basis of the observed levels of these variables, can explain the levels desired by the authorities concerned. The discrepancy between observed and desired levels may be due to the fact that, on the one hand, the monetary authorities are following a fixed exchange rate policy or keeping exchange rates within politically established or publicly announced bands and, on the other hand, the fiscal authorities are facing a massive withdrawal of international bank lending.

The first and most common of these cases is resolved by Edwards in Section 5 of his study, where he introduces coefficients of partial adjustment between the desired and observed levels of reserves and of the money supply. In this way it is established that, given fixed exchange rates, international reserves fluctuate in response to partial adjustments in the discrepancies between the actual and desired levels of reserves and money. Introduction of the demand for money into the analysis is, of course, strictly in accord with the assumptions of the monetary approach to the balance of payments; excessive domestic credit, and the resulting excess money supply, will lead to faster inflation if the exchange rate is flexible and to a loss of reserves if the exchange rate is fixed. The money demand function used has as its explanatory variables the expected inflation rate—which is not necessarily equal to the expected cost of holding money balances—and the level of income.

Edwards uses this dynamic analysis to estimate new coefficients for the explanatory variables of the demand for reserves in his first version—the level of income, the mean propensity to import, and the standard deviation of the export series—as well as coefficients for the new explanatory variables—reserves held in the preceding period and the difference between money balances desired in the period and those held in the previous period. While all the coefficients turn out to have the expected sign, only those associated with the income and reserves variables for the previous year are significant at the customary level of 5 percent; the coefficient of the excess demand for money variable is significant only at 10 percent. Curiously, however, the income elasticity of the demand for reserves now achieves a value of about 3.

I believe this dynamic analysis can be exceptionally useful in explaining and ascertaining the demand for reserves, but it is clear that much more work remains to be done. Perhaps the results achieved can be improved (a) if the expected cost, or simply the present cost, of holding money balances, instead of the period’s inflation rate, is used as an explanatory variable in estimating the demand for money. Most developing countries, especially those in Latin America, repeatedly experience prolonged periods of high inflation. In these circumstances, banks normally pay high interest rates on current accounts as a partial offset to the loss of purchasing power. My suggestion is even more valid if we use the broader definition of money. In my opinion, the inflation rate may well significantly overstate the cost of holding money balances; (b) if the opportunity cost of reserves and cost of external credit are used in estimating the demand for reserves and the demand for debt, respectively. As Edwards properly points out, earlier empirical studies have failed to find significant coefficients for proxy variables of the cost of holding reserves, such as the domestic interest rate or the difference between that rate and the rate paid on reserves deposited abroad. It is hard for me to believe, however, that the marginal social cost or the marginal social benefit of holding reserves is approximately zero. This assessment may be biased by the international liquidity crisis now being experienced by our economies, but that is not the only such crisis we have experienced and it will not be the last, at least for individual countries. Evidence of a high cost or benefit is provided by the drastic devaluations of domestic currencies which the monetary authorities, very much against their will, often have to accept—as in the recent case of Chile—as well as by the nationalization of banks in Mexico and the deferrals of external debt payment in Venezuela and Peru. This is undoubtedly a case of nonfinancial benefits and costs relating to the domestic and foreign prestige of governments and reflected in the confidence or lack of confidence in their economic management. It would therefore be difficult to incorporate this cost variable in empirical analyses. I would venture to suggest as proxy variables an index of general economic performance, including rates of product growth, employment, inflation, etc., or simply an index of the government’s popularity, which will be useful only to the extent that political costs and benefits are more influential than economic ones; (c) if we explore, for both demand functions, the degree of explanatory power of the variable “fluctuation of import value,” which in the Latin American case is undoubtedly far less important than the variable “fluctuation of export value” used by Edwards. They could be used jointly if, as appears to be the case, they are not closely correlated.

In Section 6, Edwards gives us interesting data on the external official and officially guaranteed debt of the Latin American countries. Although his data are for 1980, they clearly show the extremely high debt levels already then registered by most of the countries which now have serious external imbalances; their ratios of debt to product and debt service to exports exceeded the third quartiles of the respective distributions, namely 34.3 percent and 22.9 percent. The picture would certainly be even more troublesome if complete data on the private debt were included. Therefore, independent of the efforts the economic authorities must continue to make to hold down the social cost of the adjustment process, it would be well worth their while to follow the interesting recommendations made by Edwards in his study.

  • Highly effective policies must be developed for regulating the financial and banking system in order to prevent the effects of misallocation of credit from being transferred to the rest of the population.

  • To the extent that the cost of the debt rises as indebtedness increases, a policy must be developed to optimize private sector borrowing abroad. The need is for an optimum charge, to be applied to public borrowing transactions as well, in the form of an annual percentage rate which ideally makes the marginal social cost of new borrowing equal to the marginal cost to the borrower. Such charges already exist in many countries, although not for the above reasons, in the form of a withholding tax. Governments unfortunately tend to eliminate them or substantially reduce their rate—with the difference sometimes being absorbed by international banks—when they face balance of payments difficulties and are trying to stimulate borrowing as a means of restoring the balance. Moreover, the charge generally does not apply to official or officially guaranteed borrowing, thus adding to the disequilibria between public enterprises, which becomes more profitable, and private enterprises. The proceeds from this charge could partially finance the insurance system for deposits in banks and financial institutions, which, in keeping with the above recommendation, should be revised to introduce variable premiums directly linked to the percentages of bad portfolio investment and credit concentration. The problem with such a charge is, of course, that of estimating the elasticity of the supply of lendable funds, which, moreover, is highly volatile. Nor am I sure that it does not involve externalities imposed by the indebtedness of other economies in the same region. The massive retreat of international banks from Latin America is perhaps evidence of this. The subject is certainly worthy of attention, with a view to the prompt regulation of external credit at optimum levels.

  • It is necessary to develop an appropriate exchange policy, one which recognizes that pegging the currency to the dollar does not necessarily lead to maintenance of parity with other currencies and that the real exchange rate is also a variable. This tool, Edwards maintains, must be used in conjunction with the others—debt and reserves—in the process of restoring equilibrium to the balance of payments. This is necessary if appropriate levels of international liquidity and external debt are to be maintained.

In conclusion, I believe that Edwards’s study is excellent, because I am sure that it will motivate further research on this important subject, which is certainly necessary considering its significance at a time like the present, when most of our economies must go through a severe and painful process of adjusting their external sectors.

I am indebted to Sergio Málaga and Juan Guillermo Espinosa for their valuable comments. (The original version of this paper was written in Spanish.)

See World Bank, World Development Report, 1982.

See, for example, Sachs (1981), (1982).

It is further recognized, theoretically, that adjustments in the exchange rate play a similar role. In the empirical analysis the inclusion of such adjustments did not affect the results, hence they are not presented in this paper.

Of course while the currencies of industrial countries have tended to float (dirtily), developing countries have adhered to the most diverse exchange arrangements. See the introduction in any recent issue of the International Monetary Fund publication international Financial Statistics (IFS) for a list of such agreements.

In the event that these countries anticipated the breakdown of the Bretton Woods agreement, this result should not prove surprising.

See, for instance, various issues of the World Economic Outlook: A Survey by the Staff of the International Monetary Fund.

For debt/output ratios, see World Debt Tables of the World Bank.

There is empirical evidence that external savings partly replace domestic savings.

For a discussion of the subject, see Williamson (1973, p. 696) and Bird (1978, pp. 88–89).

There are some nontrivial problems about estimating equations (2) and (3). In the first place, in the event that some countries face quantitative restrictions regarding the amount of foreign debt they may incur, we will face the problem of estimating markets in a state of disequilibrium. See Eaton and Gersovitz (1980). This problem is not explicitly approached here. Preliminary results, however, including this problem in the estimation process, show that there are no major changes in respect of the conclusions presented here.

See also Eaton and Gersovitz (1980) Preliminary results using maximum likelihood methods for markets in disequilibirum show that the conclusions presented here are not substantially affected.

In particular, logit analysis may be used to determine the likelihood that a given country may devalue its currency to facilitate external adjustment.

See, for instance, Buiter (1980), Eaton and Gersovitz (1980), (1981 a), (1981 b), Sachs and Cohen (1982), McDonald (1982), and Edwards (1983 b).

Eaton and Gersovitz (1981 a), (1981 b), and Sachs and Cohen (1982).

For a discussion of the political risk connected with the spread of interest rates, see Aliber (1980), Dooley and Isard (1980).

This section is partly based on Edwards (1983 a).

See in this respect Dornbusch (1982) and Frenkel (1983).

The original version of this paper was written in Spanish.

    Other Resources Citing This Publication