Describing External Debt Situations: A Roll-Over Approach

International Monetary Fund. Research Dept.
Published Date:
January 1975
  • ShareShare
Show Summary Details

there has been considerable concern in recent years that a number of developing countries might, under current trends in the volume and terms of new lending, be faced with situations in which the service of external debt would absorb a rising proportion of new borrowings to a point where doubt might arise as to the overall feasibility of the borrowing process. This is only one of the types of debt problem that the comprehensive analytical efforts of the last few years have identified;1 it is probably one that is not infrequently met.

External debt operations generate inflows from loan disbursements and outflows from contractual debt service payments with a resultant inward or outward net transfer. The typical developing country needs to secure loans in amounts and on terms that allow not only for disbursements to “finance” aggregate debt repayments but also for an inward net transfer. The process of securing a net transfer by obtaining new loans in amounts in excess of debt servicing obligations is referred to in this paper as the roll-over process. The mechanics of this process have been detailed by Domar (1950), Ohlin (1966), and King (1968).

In practice, constraints on the process arise from the availability of loanable funds and from the composition of the debtor’s imports, which limits its ability to utilize new loans when they are tied to specific procurement conditions. While the net transfer is usually the major benefit to the debtor government of the borrowing policy, the costs of this policy increase in relation to the amounts borrowed and the terms under which loans are contracted, and the debtor government must seek to maintain a balance between costs and benefits. A possible approach to the analysis of external debt difficulties is, therefore, to focus on relationships between debt flows that constitute the roll-over process.

As a contribution to this approach, this paper first reviews individually a number of ratios or indicators, which are tabulated over a bench mark of 69 countries as of 1969. The paper then investigates, by means of principal-components analysis, the relationships among these ratios in order to provide a comprehensive description of debt situations in the framework of this approach. It discovers that borrowing countries are differentiated primarily by the extent to which they are involved in debt—a notation that reflects the finding that the countries with the largest debts outstanding are also those that secure the largest net transfers—and by the terms of their debt. On that basis, it formulates two major hypotheses. First, a balance needs to be maintained between the debtor’s “involvement” in debt—a concept that covers both the measure of its debt outstanding and that of the related flows—and the terms on which debt is accumulated; the more “involved” debtors need better terms to keep rolling-over their debt. Second, the growth of external debt has to be kept in line with the anticipated growth of export earnings, the critical relation being dependent on the level of debt outstanding.

One of the applications of these exercises is to evaluate the experience of renegotiation cases, that is, to investigate what was distinctive about these cases compared with the bench-mark sample. For this purpose, a sample of 13 renegotiation cases was used; the renegotiations, however, were staggered over a period extending from 1959 to 1971 and their comparison with a bench mark for 1969 requires caution. The comparison is first carried out one ratio at a time; it suggests that the renegotiation cases feature (1) a high level of debt outstanding, (2) a high level of debt service payments, (3) a pronounced concentration of debt service payments in the upcoming few years (bunching), and (4) a strong dependence on capital inflows, on both a gross and a net basis. Turning then to a comparison of debt situations as described by comprehensive sets of ratios, it appears that renegotiations occurred when a balance failed to be maintained between the extent of involvement in debt and the terms; the hypothesis of an equilibrium between the growth rate of debt and that of exports could not be tested.

Section I reviews the various ratios individually, and Section II develops their principal-components analysis.

I. Review of Selected Ratios

This section reviews a number of measures of the level and growth of debt outstanding and of the size of debt-related flows, as well as a basic indicator—debt service relative to disbursements in a given period—of the roll-over process.

level of indebtedness: stocks

A basic measure of the level of indebtedness is the ratio of debt outstanding to gross national product (GNP) or exports. As such, this ratio indicates only to what extent the borrower has been able in the past to command an inward net flow; it does not indicate whether or not there is a condition of dependency on the continuation of such an inflow.2

In 1969 the ratio of debt outstanding3 (DOUT) to GNP for 69 countries was 16 per cent on average and exceeded 18 per cent in one fourth of the countries. The ratio of DOUT to current export earnings averaged 100 per cent; however, this resulted from the presence of a few countries with very high ratios, since three fourths of the countries had a ratio lower than 86 per cent (Table 1). Renegotiation cases in the year prior to renegotiation (t-1) did not, in relation to the general standard (the 69-country sample), display high ratios of DOUT to GNP; in only 4 of 13 did the ratio exceed 18 per cent. However, in 9 of 13 the ratio of DOUT to exports (DOUT/X) exceeded 86 per cent (Table 2). The latter ratio seems more operative in distinguishing renegotiating cases from the larger sample. In the 69-country sample, only a loose correlation exists between the two ratios, so they cannot be taken as good substitutes. The better performance of DOUT/X in distinguishing the renegotiation cases suggests that DOUT/X is perhaps an operationally more useful indicator of the level of indebtedness.

Table 1.Sixty-Nine Developing Countries: Values of Indicators of The Level of Indebtedness(In per cent)
Debt OutstandingDebt Outstanding
Gross National Product

Exports of Goods and Services

Growth Rate of Debt Outstanding

Growth Rate of Exports of Goods and Services

Standard deviation12.10182.4237.346.95
First quartile8.5434.076.183.56
Second quartile12.4453.2812.976.42
Third quartile18.0985.8918.3210.67
Source: Appendix I.
Source: Appendix I.
Table 2.Thirteen Renegotiation Cases: Values of Indicators of the Level of Indebtedness(In per cent)
Debt Out-standingDebt OutstandingGrowth Rate of Debt Out-standingGrowth Rates of Exports of Goods and Services
Year of RenegotiationGross National ProductTrend Exports of Goods and Services
Country(t)(t−1)(t−1)(t−4)(t−7 to t−1)(t−7 to t−1)
Source: Appendix I.

t-6 to t-1.


Source: Appendix I.

t-6 to t-1.


The average of the growth rates of DOUT between 1965 and 1969 for the 69 countries was 17.5 per cent a year, compared with 7.3 per cent for the average of the growth rates of exports—increasing the average ratio DOUT/X from 73 per cent at the end of 1965 to 100 per cent at the end of 1970 (Table 1). The renegotiating countries display varied growth rates of DOUT in the years prior to the renegotiation, but in only 3 of 13 is the growth rate higher than the 69-country average. The export growth rates also are varied (Table 2). In all but 3 cases the ratio DOUT/X for these countries increased between t-5 (five years before the renegotiations) and t-1; however, this increase was not stronger on balance than the increase for the 69-country sample. It may therefore be concluded that as a whole the renegotiating cases involved a high level of indebtedness but were not examples of particularly high growth of indebtedness relative to export performance in the period leading up to renegotiation.

level of indebtedness: flows

Debt service payments

Debt service indicators measure the borrower’s debt service obligations as a proportion of export earnings, budgetary receipts, or savings. There are a number of reasons to focus on debt service obligations, which makes the concept of debt service indicator somewhat obscure. Basically, however, debt service payments matter because they give rise to various types of adjustment problems: debt service pre-empts untied resources, whereas new loans are likely to be “tied” in various ways; service payments are contractual obligations, and the higher their level the greater the potential impact on import capacity of a downturn in foreign exchange earnings; debt obligations may “bunch” over a short period causing a similar problem.

The simplest debt service indicator is the debt service ratio (DSR), which relates current service payments to current export earnings. The average of this ratio for 69 countries was 8.1 per cent in 1970 (Table 3), and only 9 countries had ratios in excess of 15 per cent. Over time, the ratio has been increasing rather rapidly, since the average of the ratio was 5.7 per cent in 1965 and only 5 of the 69 countries had a ratio of more than 15 per cent.

Table 3.Sixty-Nine Developing Countries: Average of Debt Service Ratios, 1965 and 1970(In per cent)
Debt ServiceDebt Service
Export of Goods and Services

Export of Goods and Services

Standard deviation5.416.42
First quartile1.903.60
Second quartile4.205.60
Third quartile6.309.90
Source: Appendix I.
Source: Appendix I.

Of the 13 renegotiation cases, 8 had ratios above 15 per cent in t-1, that is, one year before the year of renegotiation, but this was true for only 3 of them in t-4 (Table 4). The renegotiation cases as a whole thus display high debt service ratios, which were rising in the period prior to renegotiation, although this pattern was not true for a small number in the sample.

Table 4.Thirteen Renegotiation Cases: Values of Debt Service Indicators(In per cent)
Debt ServiceDebt Service1,2Debt Service1,3
Year of RenegotiationTrend exports of goods and servicesExports of goods and services,Debt,1
Source: See Appendix I.

On DRS-reported debt, including undisbursed.

1-15 year average.

Discounted 1-15 years.

Source: See Appendix I.

On DRS-reported debt, including undisbursed.

1-15 year average.

Discounted 1-15 years.

DSR is not a substitute for indicators of the level of debt outstanding. The correlation between DOUT/X and the DSR is weak and barely significant; the influence of terms is such that to be a “large” debtor does not necessarily imply having “large” debt servicing obligations during a given period of time. Given the rapid increases in DOUT during the period 1965–70 (Table 1), one might expect that the weak correlations mentioned earlier reflect to a large extent the existence of a lag between the increase in DOUT and that in the DSR related to grace periods on loans. The renegotiation cases, compared with the 69 countries, were found to be instances of both high DOUT and high DSR, which would imply that, for them, the hypothesized lag had expired.

A drawback of the DSR is that it accounts for current debt service payments only, which may be a poor indication of what payments will fall due in the near future; indeed, over a period of 5 years, for example, the DSR may change strongly (Table 3) and not uniformly for all countries. To allow for this, attempts have been made to substitute for current payments a measure of the scheduled service payments for a future period.4 These obligations can be expressed as an average over a selected period or as a sum of discounted debt service payments (SDDS) over the period, which are then divided by current (or trend) export values. For 27 nonrescheduling countries,5 the average annual payments falling due during the next 15 years on debt (including undisbursed) at the end of 1970 were divided by 1969 exports. The average ratio was 19.3 per cent, well above the 9 per cent DSR. The higher value for the average ratio reflects, inter alia, scheduled payments on loans not yet disbursed and the expiry of grace periods on older loans. By contrast, the renegotiation cases display average debt service payments over 1 to 15 years that are lower than the current payments (Table 4) for reasons that could not be ascertained specifically. Furthermore, the ratio of those debt service payments to exports for the renegotiation cases in t-1 was in all instances (except India) lower than the 27-country average, so that the renegotiation cases, which appeared as examples of high debt-service obligations at the time of the renegotiation, did not appear as such in longer-range perspective.

Another measure of debt service obligations divides the SDDS by exports (current or trend values) to derive an indicator adjusted for the terms of loans outstanding.6 However, it appears that a more fruitful application of the SDDS for that purpose is to compare it with DOUT, for which it is a theoretical proxy, namely, the ratio of SDDS to DOUT gives a summary measure of the time distribution of debt service obligations. If all service payments were equal over the next 15 years, the ratio of SDDS to DOUT7 would be 50.7 per cent; the more that payments are concentrated in the earlier years, the higher is the ratio, and conversely. One can thus say that payments in earlier years are typically higher than those in later years when the ratio SDDS/DOUT is higher than 50.7 per cent. The concentration of payments in a few years, which defines bunching situations, may be the cause of debt service difficulties. On the other hand, should a difficulty arise because debt service payments seem exceedingly large, such a difficulty may be easier to deal with when a bunching situation prevails because then there will be scope for refinancing these maturities with new, longer-maturity loans or for rescheduling existing obligations.

On the basis of the foregoing definition, the presence of some bunching of maturities is fairly typical. For the 27-country sample, the ratio of SDDS to debt outstanding (including both disbursed and undisbursed amounts) was 55.5 per cent on average and the distribution of individual cases was fairly normal. The renegotiation cases, with ratios close to or in excess of 60 per cent (Table 4), represent rather severe bunching problems,8 by reference to both the definition and the 27-country sample.9 This explains the paradox noted previously—that these cases display high DSRs but not high ratios of 15-year average service payments to exports.


Disbursements on tied loans that exceed some proportion of imports may cause distortions to the extent that those imports that can be financed by foreign loans would receive an otherwise unwarranted priority. Additionally, it may be assumed that foreign loans involve costs related to tying, which grow in direct proportion to the amount of disbursements. The ratio of disbursements to imports (DISB/M), which is primarily an indicator of the extent to which the economy relies on foreign loans, can also be viewed as an indicator of the indirect costs associated with the borrowing process. However, in individual cases this statement requires various qualifications, not only because the indirect costs associated with a given value of the ratio will vary from one country to another—depending on the country’s import structure and on the extent and nature of the tying of its borrowings by specific procurement constraints—but also because the value of the ratio itself may reflect short-term fluctuations in disbursements.

For the 69-country sample, the ratio DISB/M averaged 13.1 per cent in 1969, and in one fourth of the countries exceeded 19 per cent (Table 5). By this standard, the renegotiating countries were receiving high disbursements in t-1, since 8 out of 12 countries displayed a ratio close to or in excess of 20 per cent10 (Table 6). Moreover, the ratio was increasing in most countries in the last few years preceding renegotiation. This result would lead to rejection of the possible a priori hypothesis that a curtailment of new loans was basic to the initiation of the debt crisis.

Table 5.Sixty-Nine Developing Countries: Values of Indicators of Disbursement and Net Transfer, 1969(In per cent)
DisbursementsNet TransferDebt ServiceDebt Service
Imports1Imports1DisbursementsDebt Outstanding
Standard deviation11.688.5648.775.97
First quartile4.600.3436.007.49
Second quartile8.794.0050.109.54
Third quartile18.959.5686.7913.58
Source: See Appendix I.

Imports of goods and nonfactor services.

Source: See Appendix I.

Imports of goods and nonfactor services.

Table 6.Thirteen Renegotiation Cases: Values of Indicators of Disbursement and Net Transfer(In per cent)
Year of RenegotiationDisbursementsNet TransferDebt ServiceDebt Service
Imports1Imports1DisbursementsDebt Outstanding
Source: See Appendix I.

Imports of goods and nonfactor services.

Source: See Appendix I.

Imports of goods and nonfactor services.

Net transfer

The net transfer is the balance of debt-related flows during a period, which measures the additional (or reduced) claim on real external resources by the debtor country. The increase in the debtor’s import capacity, for an inward net transfer, represents the immediate benefit of the borrowing process. The magnitude of the net transfer in relation to total imports in a period provides a useful scale measure of the benefit and at the same time an indication of dependence on the continuation of the resource transfer.

For the 69-country sample, the ratio of net transfers to imports in 1969 average 5.9 per cent and exceeded 9.5 per cent in one fourth of the countries (Table 5). By this standard, the renegotiating countries were receiving fairly substantial net transfers in (t-1). (See Table 6.) Since these countries were heavily indebted, this result is consistent with the existence, in the 69-country sample, of a significant (although loose) positive correlation between the ratio of net transfers to imports (NT/M) and DOUT/X; the more indebted countries, which in the past have relied more heavily on foreign financing, would tend to remain more dependent on domestic resources.11

the roll-over process

Another measure of the flows associated with foreign borrowing is the ratio of debt service payments to disbursements (DS/DISB). This is a ratio of two debt-related flows and could be called an indicator of the roll-over process, expressing what proportion of loan disbursements are, in a manner of speaking, used up by scheduled debt service payments during any period. The higher the ratio, the greater is the level of disbursements required to secure a given net transfer. Under steady growth conditions, that is, disregarding short-term movements in the annual flows, the long-term value of the ratio depends on the growth rate of disbursements and the terms of the loans.12 However, for the 69-country sample, it was not possible to predict the ratio DS/DISB in relation to the growth rate and terms of disbursements; the correlation of these variables gave a poor fit and low level of significance, although the signs corresponded to expectations. This result is not too surprising, since countries in the sample have not yet been involved in the borrowing process long enough for long-run “theoretical” relations to apply, and steady growth conditions are not realized in practice. Because of this, the theoretical framework, while conceptually useful, is of little direct help in actual forecasting exercises.

For the 69-country sample, the average DS/DISB ratio is 69 per cent (Table 5) and one fifth of the countries show ratios above 100 per cent. The renegotiation cases did not, on balance, display high ratios of DS/DISB when compared with the 69-country sample (Table 6). Moreover, in only half of the cases was the ratio growing in the years preceding the renegotiation. Since the renegotiation cases did not, as previously mentioned, display high growth rates of indebtedness, their moderate DS/DISB ratios would be expected to result from borrowing on favorable terms. In fact, taking the ratio of debt service to debt outstanding (DS/DOUT)13 as a measure of loan terms, it appears that at the time of renegotiation they had not accumulated debt on very favorable terms; except for India, the ratio DS/DOUT was well above the 11.3 per cent average for the 69-country sample in the year prior to renegotiation (Tables 5 and 6). The average interest rate (ratio of interest payments to DOUT) paid by the renegotiating countries in t-1, however, ranged between 3 and 6 per cent, that is, it was largely concessional. The high DS/DOUT ratios of these cases reflect large amortization payments as a proportion of DOUT, in line with the previously mentioned bunching problems of these countries. Here, the reasonable DS/DISB ratios are explained by the maintenance of substantial disbursements as evidenced by the DISB/M ratios; this is consistent with heavy recourse to medium-term credit.

conclusion for section i

The examination of selected debt indicators for renegotiation cases and for a large sample of developing countries reveals certain characteristics of the former.

The renegotiation cases involve heavily indebted countries that are very active borrowers, not only to “finance” maturing obligations but also to secure a large and sustained net transfer (NT/M), on which they may have become dependent. Typically, their debt service obligations are large and rising (DSR). They seek large capital inflows (DISB/M) even though in the process their debt accumulates on terms (DS/DOUT) that are not favorable and, in spite of its high level, debt continues to grow faster than exports. While a satisfactory roll-over ratio (DS/DISB) is maintained, this is a consequence of increasingly larger external borrowing rather than of a cautious policy on the terms of new borrowing. Assuming that they wish to maintain a high level of net transfers in coming years, these countries have to borrow increasingly larger amounts because the conditions of past borrowing have led to a bunching of maturities; a large portion of scheduled service payments on existing obligations falls due within the new few years. On the one hand, to pursue a policy of securing high net transfers under these conditions would soon involve a borrowing level that could not be maintained because of supply or “absorption” constraints. On the other hand, the total of scheduled debt service obligations that is due over a longer period (15 years) is not exceedingly large, so that there is scope for successfully pushing the earlier maturities to a later date via a rescheduling operation. Not all of the renegotiation cases, however, fit neatly within this broad framework.

II. Relationships Among Selected Ratios

The approach of this section differs from that of Section I. Instead of reviewing the single indicators one at a time, this section focuses on the relationships among those indicators that are most relevant in differentiating individual country situations. The criterion of high or low values of an indicator is no longer the bench mark of the 69-country sample for that indicator but rather the relationship of that indicator to the others, compared with a “normal” relationship that was found to exist in the 69-country sample. The first task of this section is to provide an adequate description of the selected indicators simultaneously for that sample. From this, relationships among these indicators are identified and the assumption is made that departures from normal (or equilibrium) relationships will result in external debt problems; the relevance of this assumption is then tested.

identification of major relationships

Description of the sample

A technique called principal-components analysis is used to describe how debt indicators are interrelated in the 69-country sample. The technique as applied here consists of substituting for the set of original indicators a set of composite indicators, or components. Each component is a linear combination of all the initial indicators and expresses a relation among the indicators. In the computation process, the percentage of total sample information incorporated in the component is measured; this provides a measure of the relative importance of the component. It is possible to focus on the components that have the highest information content—by definition these are few in number—and to draw conclusions on the relative importance of the relations between single indicators that are featured by a given component. When two components are used, they can be treated as axes of a plane on which either individual indicators or countries can be plotted14 so as to represent simultaneously all pairwise relations in that plane.

A major problem in applying principal-components analysis here is adequate selection of indicators. No technique is available presently for an optimal selection by a stepwise process of a limited number of indicators from an extended list. The only possibility, therefore, is to proceed by trial and error. Among other trials, experiments were made with a list of ten indicators for which information on 12 of the renegotiation cases was available.15

Chart 1 shows the results of a first experiment, conducted with the 69-country sample and ten indicators. In this experiment, the first four components (F1 to F4)summarize a total of 79 per cent of total information, of which 56 percentage points are accounted for by the first two components—the axes in Chart 1. There is a contraposition along F1 between a group16 of indicators of relative size of debt and debt flows (DOUT/X, DOUT/GNP, DISB/M) and a group of indicators of terms and roll-over (DS/DISB, DS/DOUT); to the right are countries more deeply involved in debt but under more satisfactory conditions with respect to terms. There is a negative correlation of DSR, DS/DOUT, and DS/DISB with F2; moving up that axis leads to countries with lower DSRs and better borrowing conditions. Altogether, F1 can be interpreted essentially as a measure of the involvement in debt, while F2 reflects the terms of the loans and the size of debt service.

Chart 1.Sixty-Nine Developing Countries: Correlation Coefficients of Ten Indicators With First Two Components, 19691

The third component accounts for nearly as much information (14 per cent) as the second (18 per cent). To examine this component in more detail, the indicators have been plotted on the plane of the first and third components (Chart 2); R and E are positively (and strongly) correlated with the third component (F3), while there is a (weaker) negative correlation of DOUT/GNP and DOUT/X with it. Other things being equal, countries in the sample that had larger growth rates of debt also had larger growth rates of exports. Also, the larger the present debt, the smaller the growth rate of both debt and exports in the previous period. This latter relation does not, however, appear clearly, since the correlation coefficients of the indicators of debt outstanding with F3 are low; it is therefore best to focus on the positive correlation of E and R with F3 and on the close relationship between these two indicators. This suggests that debtors usually limit the growth rate of their debt within some proportion of the growth rate of their exports. The exact proportion in each case depends on a variety of factors, one of which is the level of debt outstanding.

Chart 2.Sixty-Nine Developing Countries: Correlation Coefficients of Ten Indicators With First and Third Components, 19691

It is questionable whether this statistical relation could be further elaborated to yield an equilibrium condition or a hypothesis on the emergence of debt crises. The relation links the growth rate of a stock (debt) to that of a flow (exports); however, the growth of debt outstanding brings about a growth of debt flows. The relation can therefore be interpreted as one between the growth of debt-related flows and that of exports. When debt-related flows grow faster than exports and also faster than imports, the debtor becomes increasingly involved or dependent on the borrowing process. On the one hand, if a country is initially little involved, this raises few problems; on the other hand, problems arise if the country is already a large debtor. It is suggested later that the more involved debtors require favorable borrowing conditions to proceed with their borrowing policy. If a large debtor cannot contain the growth of its debt in line with its foreign trade, its involvement will increase continually; it is then unlikely that the debtor will be able to balance this increasing involvement by improving its borrowing conditions. Since there are several conceivable processes by which involvement can become excessive relative to borrowing conditions, the requirement of an equilibrium between E and R should be considered as an original requirement, leaving open the question of its relation with other aspects of the borrowing process. The structure of Chart 2 does not lend itself conveniently to tests of this hypothesis, since no area of the chart is designated as critical on its basis; consequently, testing of this hypothesis was not undertaken in this paper.

Summary of major relationships

Up to this point, results have been reported with as little analytical appraisal as possible in order to respect the descriptive nature of the exercise. Reduced to essentials, the experiments with the 69–country sample have shown that:

(1) A set of indicators comprising DOUT/X, DOUT/GNP, NT/M is fairly homogeneous; these indicators tend to be closely associated with one another and to exhibit similar relationships with other indicators. As a group, these indicators can be thought of as a measure of involvement in debt, a notion that covers not only the size of debt outstanding but also the concept of current reliance on gross and net inflows. Thus defined, the degree of involvement in debt is of primary importance in differentiating individual country situations.

(2) A second set of indicators comprising DS/DOUT and DS/DISB also contributes a rather stable group, which may be interpreted as a measure of borrowing conditions, including terms and the roll-over ratio. This group is also instrumental in differentiating individual country situations.

(3) The differentiation of country cases according to the extent of their involvement in debt is not independent of their differentiation according to borrowing conditions, as the more involved debtors tend to benefit from better conditions. This constitutes the major relationship shown in the foregoing analysis.

(4) A third stable group comprises the growth rates of debt and exports. The role of this group in differentiating individual country cases is of some importance, but less so than that of the two previous groupings. It seems best at this stage to focus on the relation between the two growth rates within the group rather than on their relation as a group with other groups; the latter, however, might be a subject for useful further research. Accordingly, the second important relation derived from the foregoing analysis is a positive relation between the growth rate of debt and that of exports.

equilibrium conditions in debt developments

Having identified important characteristics of debt situations, the problem is to formulate them in terms of equilibrium conditions. For this purpose it is assumed that the sample of 69 countries represents “normal” situations, so that relations between indicators observed in that sample also may be assumed to be normal or sound; a contrario, debt problems would be expected to occur for countries for which these relationships are not maintained. Corresponding to the previous identification of two basic relationships, we can therefore postulate the existence of two equilibrium conditions. One is that there is a trade-off between involvement in debt and borrowing conditions, that is, that difficulties can be expected when a heavy involvement in debt is accompanied by unfavorable borrowing conditions. According to this hypothesis, heavy debtors (large DOUT/X and DOUT/GNP) that still rely heavily on foreign capital (large DISB/M and NT/M) can do so only if their debt is accumulated on terms (DS/DOUT) that allow for an easy roll-over (DS/DISB). The second equilibrium condition is that the growth of debt should keep in line with the growth of exports, especially for large debtors. This hypothesis, however, cannot be tested here.

Rationale of the hypothesis

The rationale underlying a trade-off between the degree of dependence on debt and the terms and roll-over ratio is as follows: when a country’s dependence on inflows of a given magnitude on given terms (reflected with a lag by DS/DOUT) results in a high roll-over ratio (DS/DISB), the borrowing policy can be maintained only by raising disbursements continuously. However, if disbursements have reached a level (indicated by DISB/M) that is striking against an import composition or supply constraint, alternative means (such as ways of reducing debt service) must be found to improve the roll-over ratio. While it clearly calls for complementary hypotheses on the factors that influence the maximum DISB/M level, this formulation embodies the core of a representation of “critical” debt situations. It is worthwhile to compare it briefly with approaches that emphasize the debt service ratio as a unique constraint. Taking the roll-over ratio (DS/DISB) and the disbursement ratio (DISB/M) as arguments for the trade-off, and specifying the function as a hyperbola, the critical function would be defined as Y = (DS/DISB) × (DISB/M) = DS/M which is an indicator very similar to the usual DSR. However, the underlying rationale of such an indicator is different from that of the usual DSR, in that DS/M as just constructed is a composite indicator of a process developing over time rather than of a single, permanent constraint. In other words, the present interpretation substitutes a roll-over approach with emphasis on the internal consistency of debt flows for the more usual servicing capacity approach.

Relevance of the hypothesis

According to the hypothesis, we would expect that the renegotiating cases did not maintain that balance between involvement in debt and borrowing conditions which is featured by the 69-country sample, and accordingly would fall in a specific quadrant of the plane defined by these composite indicators. Furthermore, among themselves, renegotiating countries should be differentiated primarily by the extent of their involvement in debt and their borrowing conditions—the more involved countries having relatively better terms and better roll-over ratio. If this were not so, it would be difficult to explain why, at the time of renegotiation, some countries were more heavily involved in debt than others. This provides the basis for two tests of the relevance of the hypothesis—one involving a principal-components analysis of the 81-country pooled sample (69 plus 12) and a plot of renegotiation cases on the plane of the first two components, and the other involving a principal-components analysis of the 12 renegotiation cases and their plot on the plane of the first two components.

Analysis of pooled sample. In the analysis of the pooled sample, the first four components summarize a total of 77 per cent of total sample variance, of which 53 percentage points are accounted for by the first two components (F1 and F2), which are taken as the axes of Charts 3 and 4. Chart 3 represents the correlation coefficients of the indicators with the first two components. The structure of this chart is substantially similar to that of Chart 1 except that the orientation of F2 is reversed, with DS/DISB and DS/DOUT having a positive correlation with that axis. Accordingly, the critical area where heavy debt involvement is compounded by unfavorable borrowing conditions would be the northeast quadrant, since this is the quadrant where both components have positive values. In Chart 4, the countries are plotted according to the values of the components for each case.

Chart 3.Pooled Sample OF 81 Countries: Correlation Coefficients of Ten Indicators With First Two Components1

It appears that 8 of the 12 renegotiation cases are located as the hypothesis suggested, with an additional case (Peru) very close to the northeast quadrant. Contrary to the hypothesis, the Philippines, Uruguay, and Yugoslavia (1971) are located below the diagonal in noncritical debt situations. Given the small amount of debts reported by the DRS for these countries and the existence of sizable debts not reported by the DRS,17 it could be expected that developments other than those in DRS-reported debts would be instrumental in these renegotiation cases; therefore, the position of these cases in Chart 5 is not a strong argument against the hypothesis. On the other hand, 9 non-renegotiating cases are plotted in the “critical quadrant,” and there is a large dispersion of the renegotiation cases within that quadrant. The hypothesis, therefore, can be considered as providing a good indication of emerging debt difficulties, but it does not fully and unequivocally explain such cases; other factors have to be allowed for.

Chart 4.Eighty-One Countries: Plotted in Plane of First two Components

Chart 5.Twelve Renegotiation CASES (t-1): Correlation Coefficients of Ten Indicators With Two Main Components1

Analysis of renegotiation sample. In the analysis of the 12 renegotiation cases, the first four components explain 90 per cent of total variance, of which 66 percentage points are accounted for by the first two components; the correlation coefficients of the indicators with the first two components are shown in Chart 5, while the plot of the cases is reported in Chart 6. On balance, Chart 5 supports the hypothesis. In this chart, the group of indicators of the extent of involvement in debt is contraposed both (along F1 to DS/DISB and (along F2) to DS/DOUT as was required by the test. However, DS/DOUT and DS/DISB no longer belong to the same cluster, and the contraposition between DS/DISB and the indicators of involvement in debt accounts for much more of the total sample variance than that between DS/DOUT and those indicators. This might imply that, in specifying the hypothesis, DS/DISB should be preferred to DS/DOUT.

Chart 6.Twelve Renegotiation Cases (t-1): Plot of Renegotiating Countries in Plane of First Two Components

The plot of the renegotiation cases on the plane of the first two components in Chart 6 conforms exactly to their plot in the pooled sample (Chart 5) and confirms that, among themselves, the renegotiation cases are differentiated primarily by a trade-off between the extent of their involvement in debt and the DS/DISB ratio.

A quantitative criterion. Since the previous tests have shown the relevance of the hypothesis, it is tempting to express it in quantitative terms. This, however, raises a number of difficult questions, that is, to select pairs of single indicators to define a functional form for the relation and to choose a critical level of deviation from the normal relation. Since this paper is not directed primarily to this stage of the analysis, only a few crude tests were performed. The best results were obtained by stating that debt renegotiations would occur if (DS/DOUT) × (DISB/M) > 266;18 only two renegotiation cases (the Philippines, 1970, and Yugoslavia, 1971) fail to be predicted correctly by this test, while eight nonrenegotiation cases are singled out as being in the danger zone.

conclusion for section II

In general, debtor countries that manage to maintain a balance between the extent of their involvement in debt and their borrowing conditions and to keep the growth of their debt in some relation to the growth of their exports are not subject to debt renegotiation. Where these equilibria fail to be maintained, there is a good probability that debt problems emerge and provide an incentive to seek a negotiated revision of debt service payments.

It appears that those debtors with relatively large outstanding debts continue to secure a significant inward transfer requiring relatively large flows of disbursements. Such debtors can be said to be heavily involved in or dependent on debt in the sense of being importantly and continuously reliant on foreign loans. However, while in general more dependent debtors secure relatively better terms and roll-over ratios, this is not true for countries in the renegotiation sample. A successful debt management policy appears to be one that involves inflows in such volume and on terms that will enable the borrower to maintain a satisfactory roll-over ratio, that is, ratio of debt service to disbursements, over time, without having to increase disbursements to a point where they become excessive in relation to imports.

The growth rate of debt for most countries in the large sample is greater than the rate of growth of exports; however, in relation to other countries, higher growth rates of debt are maintained by those countries that have higher growth rates of exports and that generally are not among the larger debtors in terms of debt outstanding. Another condition of successful debt management is to keep the growth of debt, that is, the volume of net flows, in line with the growth of exports, particularly when the debt outstanding is already large. However, this condition could not be specified sufficiently to allow for conclusive testing. Further elaboration of the hypothesis, to take into account simultaneously the complex interrelationship between export growth, borrowing conditions, and involvement in debt, would be useful.

I. The Construction of Samples

For empirical evaluation purposes, two samples were used. First is a sample of 11 countries with an experience of debt renegotiation, representing 13 instances of renegotiations. In 8 countries only the first instance of renegotiation has been retained because “short-leash” policies in rescheduling exercises would have probably resulted in an interdependence of successive renegotiations. For Argentina (1962), Turkey (1965), and Yugoslavia (1971), however, renegotiations were not felt to have resulted from a short-leash approach. In each case, a number of indicators have been estimated for the year prior to the date of renegotiation (t-1) and for the fourth year prior to renegotiation (t-4).

The rationale in using a sample of renegotiating countries is that renegotiations are the most obvious available indication that debt difficulties occurred. However, it is not the indicators at the end of the year of the renegotiation that point to these difficulties but the indicators at some earlier time, because there is a time lag between the date a renegotiation begins and the date an agreement is concluded, and because we should expect the indicators to have some predictive value. This is why testing was conducted on the indicators at t-1.

A second sample comprising 69 countries was also used, and indicators for that sample were measured as of 1969. This sample was the broadest that could be obtained for that year.

The concept of debt” used in both samples is that of the DRS, which covers foreign exchange debts that are a liability of, or guaranteed for payment by, public bodies in the debtor country. This definition would imply that some measure of control and policy coordination is exercised over the elements composing this aggregate, which makes it meaningful for a study of the momentum of debt developments.

The concept of net transfer (disbursements minus debt service), which corresponds directly to the increment in resources accruing to the country, was felt to have greater practical relevance than the concept of net flows (disbursements net of amortization payments only). Receipts on goods and services were used as the measure of foreign exchange earnings.

The sample of renegotiation cases comprises the following 13 instances of renegotiation: Argentine (1962), Brazil (1961), Chile (1965), Egypt* (1966), Ghana (1965), India (1968), Peru (1968), the Philippines (1970), Turkey (1959), Turkey (1965), Uruguay* (1965), Yugoslavia* (1965), and Yugoslavia* (1971). It covers all the countries that have experienced bilateral (indicated by an asterisk) or multilateral renegotiations during the period 1959–71, except Indonesia (1966), Liberia* (1963), and Mali* (1966), for which insufficient data were available. For each of these countries, data on GNP, population, exchange rate, and balance of payments were retrieved from the International Monetary Fund’s International Financial Statistics (IFS) and the Balance of Payments Yearbook, eventually supplemented by other sources. The debt data were retrieved mostly from the International Bank for Reconstruction and Development (IBRD), supplemented by local sources. Data on disbursements were not retrieved but were computed as the sum of change in debt over two years and amortization payments. Data on debt outstanding comprised only the disbursed portion of the loans. For the Philippines, Uruguay, and Yugoslavia, reported debts seriously underestimate total debt.

The sample of 69 developing countries was taken from the IBRD Annual Report, 1972, Statistical Annex, Table 10, “Average Amount of Loans and Grants Received (and Grant Element) by Developing Countries, 1965–1967 and 1968–70,” pages 88–91. The 75 countries in that table (omitting the East African Community) exclude Brazil, Egypt, Ghana, and the Syrian Arab Republic, which are included in the standard list of developing countries but for which detailed information was not available. Six countries have been further excluded because of lack of information; these had small outstanding debt. For each country, data on GNP, population, exchange rate, and balance of payments were retrieved from IFS and the Balance of Payments Yearbook. Debt data were retrieved from the IBRD Annual Report, 1972, supplemented by data from the 1971 Report of the Development Assistance Committee.

II. Analytical Relations

As mentioned in Section I, the share of debt service in disbursements may be regarded as a policy variable, as it depends on the growth rate of disbursements and the terms of the loans, which are policy variables. This Appendix explains the relations involved.

In what follows we consider a borrowing program defined by:

(1) Disbursements in period t, GFt (GF for gross flows);

(2) Growing annually at a constant rate (R) from an initial disbursement (GF0=X; GFt= (1 + R)tX);

(3) With no grace period nor any lag between commitment and disbursement;

(4) Repaid in T equal annuities (T is the amortization period);

(5) With an interest rate (i) charged on the debt outstanding at the beginning of the period, that is, at the end of the previous period: Dt−1. At is the flow of amortization payments during period t; It is the flow of interest payments during that period; DSt is the total debt service payments during that period: DSt = At+ It; NTt is the net transfer: NT t = GFt − At − It; debt outstanding at the beginning of period 0 is 0: D−1= 0.

Dt−1 is the sum of unamortized fractions of disbursements between 0 and t − 1. If we focus on the case where a full amortization period has already expired, that is, t > T, we have:

For t>T, nt is a constant (nT) that is positive if the growth rate of disbursements is higher than the interest rate19 and that, for i given, grows when R or T grows. Thus, a longer amortization period and/or a higher growth rate of disbursements lead to a higher nt ratio. The relation among nT, the growth rate of disbursements, and the terms can be made more explicit by a specific interpretation of the grant element. By definition, the discounted debt service on GFt, DDSt, is equal to the discounted sum of debt service payments with a discount rate d.

knowing that Ax=1TGFt and that

and that grant element is defined as GE =(GF t—DDSt)/GFt

we have GE=(1id)[11Td+1Td(1+d)T]

so that with d = R, GE = nT

The concept of grant element therefore provides a direct insight into the roll-over process.

III. Statistics
Table 7.Sixty-Nine Countries: Correlation Coefficients of Ten Indicators With Four Main Components, 19691
Percentage of total variance explained by component3818149
Source: Sample—69 countries.

See Charts 1 and 2.

Source: Sample—69 countries.

See Charts 1 and 2.

Table 8.Pooled Sample of 81 Countries: Correlation Coefficients of Ten Indicators With Four Main Components1
DSR 700.3010.5250.212−0.451
Percentage of total variance explained by component35181410
Sources: Samples—69 countries and 12 renegotiation cases.

See Chart 3.

Sources: Samples—69 countries and 12 renegotiation cases.

See Chart 3.

Table 9.Twelve Renegotiation Cases (t-1): Correlation Coefficients of Ten Indicators With Four Main Components1
Percentage of total variance explained by component47191410
Source: Sample—12 renegotiation cases.

See Chart 5.

Source: Sample—12 renegotiation cases.

See Chart 5.


Mr. Dhonte, economist in the External Finance Division of the Exchange and Trade Relations Department, received his doctorate in economics from the University of Paris. He was formerly an instructor and engaged in research at the University of Paris.

Sometimes, debt outstanding is recalculated as the sum of discounted debt service payments; this correction makes it possible to analyze debt as the source of an income stream to foreigners, with potential alienation problems involved. Here the correction has been used only as a technical device to measure the time profile of service payments. (See section on Debt service payments).

Unless defined otherwise, “debt outstanding” (DOUT) indicates disbursed amounts of debts reported under the Debtor Reporting System (DRS). A number of other terms used in this section are defined in Appendix I.

The period selected generally has been 5 or 15 years, the former when the focus is on impending developments and the latter when a more comprehensive assessment is desired. The cutoff point of 15 years appears to have been a matter of convenience; the present value of obligations falling due beyond this period (with a 10 per cent discount rate) is usually small in relation to the present value of those falling due during the period.

These 27 nonrescheduling cases were selected from the 69-country sample because of availability of data on discounted debt service payments. They had average values for 10 major debt indicators similar to those of the 69-country sample.

Using SDDS as a measure of debt service obligations changes somewhat the nature of the debt service indicator, since the sum of discounted flows is a stock figure; the SDDS is actually a measure of debt outstanding, and the ratio of SDDS to exports is analogous to that of DOUT to exports.

With a 10 per cent discount rate.

Except India, which displayed no characteristics of a bunching problem.

Given that the range of variation of this ratio is small (the interquartile range for the 27-country sample is 6.8 points, 52.0 per cent to 58.8 per cent), a 55.5 per cent average should be thought of as considerably higher than the 50.7 per cent “norm.”

Furthermore, the renegotiating countries for which this ratio is small are those that have small debts outstanding and small DSRs.

For 69 countries: NT/M=0.0286DOUT/X+3.05R2=0.36t=6.28

As a measure of terms, DS/DOUT has the drawback of being sensitive to the time pattern of disbursements. In periods of rapid growth of DOUT, the growth of DS lags behind that of DOUT because of grace periods, and the ratio is consequently reduced. The opposite is true in periods of slower growth of DOUT, which would contribute to somewhat higher DS/DOUT ratios for the renegotiating countries.

The analysis can be conducted, alternatively, using the indicators or the countries as variables. The resulting system of components is the same in both approaches, except for a scale factor. This is why the interpretation of the system based on the examination of one plot can be applied to the analysis of the other.

List of ten indicators:

  1. Reserves over debt outstanding (RES/DOUT)

  2. Debt service over disbursements (DS/DISB)

  3. Debt service over debt outstanding (DS/DOUT)

  4. Debt service ratio for 1970 (DSR 70)

  5. Debt outstanding over exports for 1970 (DOUT/X 70)

  6. Debt outstanding over GNP (DOUT/GNP)

  7. Growth rate of debt (R)

  8. Growth rate of exports (E)

  9. Net transfers over imports (NT/M)

  10. Disbursements over imports(DISB/M)

In the renegotiation cases, the values for the indicators were those of the year prior to renegotiation (t−1).

The indicators have been clustered into groups, according to a routine devised in such a way that any plot in a cluster is closer to the center of that cluster than to the center of any other cluster. The clustering routine thus defines groups that are as homogeneous as possible in respect of the two components (axes) on the chart. The use of this procedure simplifies the reading of the charts. Plots to the right of the origin along F1 (or above the origin along F2) denote a positive correlation of the indicator with F1 (or F2), and conversely; the distance between a plot and a component (axis) is proportional to the absolute value of the correlation coefficient between the component and the indicator.

These three cases were instances of bilateral renegotiations.

The ratios are expressed in percentage points; the mean relation was estimated on the 69-country sample, and the critical level of 266 represents one standard deviation departure from the mean.

See Evsey D. Domar, “The Effect of Foreign Investment on the Balance of Payments,” American Economic Review, Vol. 40 (December 1950), pp. 805-26.

Other Resources Citing This Publication