The Stabilizing Role of the Compensatory Financing Facility: Empirical Evidence and Welfare Implications

Purchases under the compensatory financing facility (CFF) accounted for nearly one fourth of total credit extended by the International Monetary Fund from 1976 to 1985. To determine the extent to which the facility served its intended purpose of stabilizing foreign exchange earnings of member countries experiencing temporary export shortfalls, a methodology is developed for evaluating the CFFs stabilizing role. The evidence is used to evaluate the facility’s role in stabilizing demand for international reserves and its contribution to net welfare gain. The results suggest the CFF has helped stabilize earnings, and net benefits from its use have been substantial.


Purchases under the compensatory financing facility (CFF) accounted for nearly one fourth of total credit extended by the International Monetary Fund from 1976 to 1985. To determine the extent to which the facility served its intended purpose of stabilizing foreign exchange earnings of member countries experiencing temporary export shortfalls, a methodology is developed for evaluating the CFFs stabilizing role. The evidence is used to evaluate the facility’s role in stabilizing demand for international reserves and its contribution to net welfare gain. The results suggest the CFF has helped stabilize earnings, and net benefits from its use have been substantial.

THE COMPENSATORY financing facility (CFF) of the International Monetary Fund, which was set up for the purpose of stabilizing foreign exchange earnings of member countries experiencing temporary export shortfalls, was in operation between February 1963 and August 1988. On August 23, 1988 the Fund established a new facility called the compensatory and contingency financing facility, which retained the basic features of the CFF and also included a contingency element for use in Fund-supported adjustment programs (see Pownall and Stuart (1988)). Although use of the CFF was modest in the earlier years, over the period 1976 to 1985 drawings averaged nearly one quarter of total credit extended by the Fund. Given the size of these operations, it is worthwhile to determine whether, and to what extent, the facility has served its intended purpose. To this end, some quantitative estimates of the stabilizing role of the facility are provided here. These estimates are then used to obtain an indication of the CFF’s role in stabilizing the demand for international reserves and of the wider welfare implications of the facility.

The paper is organized as follows. Section I provides a brief discussion of the evolution and operation of the facility and suggests why a stabilizing role is to be expected. Section II proposes a methodology that can be used to examine this stabilizing role empirically. In Section III a series of statistical exercises are performed to ascertain the significance of this role. Section IV examines the implications of the empirical results for reserve requirements of countries using the facility and the wider welfare implications. The final section summarizes the main empirical findings and conclusions.

I. Stabilizing Role of the CFF

The CFF was set up to provide timely financial assistance to members experiencing temporary shortfalls in their export earnings due to factors outside their control. The CFF is open to all Fund members, but in practice it has been used mainly by developing countries.

Main Features of the Facility

In order to qualify for a purchase, a member must meet certain conditions. First, as with the use of all Fund resources, is the general requirement of need, which is assessed by reference to the country’s overall balance of payments position, its reserves position, and developments in its reserves. A member must also demonstrate that it has experienced a temporary shortfall in its aggregate export earnings in the sense that a deviation in earnings from their medium-term trend will be of short duration.1 In addition, the shortfall has to be attributable to circumstances largely beyond the member’s control, which means that the shortfall is not due to inappropriate economic and financial policies. The member must also be willing to cooperate with the Fund.

Since 1966 access under the CFF has been in two tranches. The first tranche is available to members demonstrating a willingness to cooperate with the Fund in an effort to find, where required, appropriate solutions for their balance of payments difficulties. The second tranche is available to members that have been cooperating with the Fund. The cooperation requirement has been a feature of the facility since its establishment, and in 1983 formal guidelines were set out on use of the lower tranche—up to 50 percent of quota—and the upper tranche—up to 83 percent of quota (see International Monetary Fund (1987, pp. 88–89)). The amount of drawing is constrained by the size of the calculated shortfall, subject to a limit on outstanding drawings in relation to the member’s quota. Following approval of a request by the Fund’s Executive Board, a drawing is made in one installment and normally repaid in eight equal quarterly installments spread over the fourth and fifth years after the drawing. The rate of charge on outstanding CFF drawings is the same as that applied to other drawings from the general resources of the Fund.

The major use of the CFF has been to compensate for shortfalls in merchandise exports.2 The shortfall is computed in relation to a medium-term trend, which is defined as an average of value of earnings (in SDRs) for five years including two years of projected exports centered on the shortfall year. Since 1979 the export trend has been based on a geometric average rather than an arithmetic average as in the earlier years. The calculations have always been done on the basis of nominal rather than real values.3 The maximum CFF drawings that a member may have outstanding at any time in relation to its quota has been changed on several occasions, usually as a result of reviews of the facility by the Executive Board. Since 1984 the maximum limits have been 83 percent of quota for an export shortfall or an excess in cereal import costs, with a joint limit of 105 percent of quota for the two elements.

Total drawings under the CFF have been substantial. Although use of the CFF in the early years of the facility was modest, during 1976–85 annual drawings averaged SDR 1.3 billion—that is, nearly one quarter of total credit extended by the Fund. For individual years over this period, drawings fluctuated between SDR 241 million in 1977 to SDR 2,839 million in 1983. The share of CFF drawings in total Fund credit, however, declined from about 30 percent during 1976–79 to about 23 percent in 1980–85, as Fund assistance under the credit tranches in support of adjustment programs grew considerably faster than that under the CFF during the latter period.

In analyzing the stabilizing influence of these drawings, it should be noted that in recent years the balance of payments difficulties of countries using the CFF tended to reflect imbalances that went beyond the effects of export shortfalls. The result was that an increasing number of requests for use of the CFF had to be considered in conjunction with Fund-supported adjustment programs; at the same time, the proportion of drawings in the upper CFF tranche also increased. Of the 73 CFF drawings made from 1982 to 1985, 44 were in the upper tranche, and all but one of these were associated with Fund programs either in place at the time of the CFF request or approved concurrently with the CFF request. Of the 29 lower CFF tranche drawings, 9 were also associated with Fund programs, but in these cases the size of the shortfall, and not the test of cooperation, was the factor limiting the drawings to the lower tranche (see Kaibni (1986)).

Expectation of a Stabilizing Role for the CFF

In the context of the CFF, a stabilizing effect means a reduction in the extent of fluctuations in foreign exchange receipts with transactions under the CFF from what would have been the case without the transactions. Purchases under the CFF can be regarded as supplementing earnings from merchandise exports, whereas repurchases diminish such earnings. Although the magnitude of purchases has always been determined in relation to a member’s quota—with the consequence that purchases have not necessarily been sufficient in all cases to compensate for export shortfalls—the timing of these purchases and associated re-purchases in relation to the profile of exports over time is the crucial factor in an assessment of the stabilizing role of the facility.

Certain aspects of the CFF suggest that use of the facility would have a stabilizing influence. First, once a shortfall is identified, assistance to the member can be provided in a timely manner. Second, since 1979 the shortfall year can include projected exports for up to six months if it is expected that a shortfall will emerge, thus increasing the timeliness of assistance. Third, as noted above, the computation of the shortfall is not based on past data alone but is calculated as a deviation from trend using two years of projected exports. An accurate identification of the shortfall profile would thus enhance the stabilizing effect of the CFF.

However, certain other facets of the facility could act to destabilize earnings. First, CFF purchases may take place up to six months after the end of the shortfall year, with the effect that the time lag between the middle of the shortfall year and the purchase could be up to one year. Second, repurchases under the facility are usually made in equal quarterly installments during the period, beginning three years and ending five years after the date of purchase—a time frame that may or may not correspond to an upturn in export receipts. Third, if there are large errors in the export forecasts for the two post-shortfall years, purchases may have taken place in periods with export excesses rather than shortfalls, thereby destabilizing receipts.

II. Methodology

To assess empirically the stabilizing influence of the facility, the basic methodology is to compute an index of export instability without CFF transactions for the countries that have made CFF purchases. This index is then compared first with an index that takes into account CFF purchases and then with another one that includes both purchases and repurchases.

Basic Framework

More formally, let the export earnings of country i in years 1 to t be given by the vector Xi, where


An index of instability computed from these earnings is denoted by Ii, where


and Ii > 0. Export earnings with CFF purchases are then


where S is the amount of CFF purchases made in period k. An index based on Xip is denoted by


Since the number of observations for each individual country is limited (11 observations with an average of 2 drawings), the sample was pooled for statistical analysis. The null hypothesis H0 that purchases exercise no stabilizing influence can then be stated as


with the alternative hypothesis H1 being




and n denotes the number of countries that have made the purchases.

Similarly, with purchases and repurchases, an index of instability can be computed and compared with I.

This methodology is relatively straightforward and allows one to focus on how instability of earnings is affected by transactions under the CFF.4 The size of the effect would, as noted, depend on the amount of purchase, the speed with which the application was made, the time taken to analyze the request, the accuracy of forecasts, and the timing of the repurchase profile. If the stabilizing effects dominate, instability without the CFF would be significantly higher than with the CFF. In the discussion below, a number of different indexes that may be used to measure instability, and the appropriateness of each for this empirical exercise, are examined. The variables that can be used to measure instability are also noted and alternative schemes for allocating drawings to exports are considered.

Reference Earnings

Before the above issues are examined, an important element of methodology needs to be given some attention—that is, the use of actual export earnings streams, Xi, as the reference by which to evaluate the facility’s stabilizing role. It could be argued that the profile of these earnings (without CFF transactions) already reflects the possibility that compensatory finance will be obtained and that the “true” reference earnings, which are unobservable, would have been different. Given this argument, an accurate indication of the CFF’s stabilizing role would require an estimate of export earnings that would have occurred in the absence of the CFF. This problem is similar to the one encountered in attempts to evaluate the impact of Fund programs, and as recent studies suggest, the process of obtaining the counterfactual outcome against which the magnitude of this impact can be assessed is far from straightforward. (For a comprehensive discussion, see Khan and Knight (1985), Goldstein (1986), and Goldstein and Montiel (1986).)5

Two considerations suggest, however, that, in the case of the CFF, the use of Xi as a reference series may not necessarily impart a bias. The first is based on the premise that exporters always want to maximize the present value of their export earnings, whether or not these are going to be augmented by outside funds. So the availability of the CFF will not change their trading decisions. There may be more scope for discretion in the timing of the stream of export earnings than for the size of the stream, but this scope too is probably minor.

The second consideration is that even though a country may ultimately have access to the facility, it must first meet the test of cooperation and demonstrate that the shortfall is beyond its control. The ability to meet both these criteria cannot be known with any degree of certainty. These two considerations suggest that apart from some minor “window dressing,” it is unlikely that the “true” earnings streams would have been different from “observed” earnings. Thus, as far as the use of reference earnings is concerned, the use of the above approach appears to be appropriate.

A related point concerns cases, which have been frequent over the last few years, where the Fund’s stand-by or other programs have accompanied CFF drawings. It is possible that the policies pursued under a program itself would have an impact on the stability, and not just on the level, of export earnings. The issue again concerns which earnings stream should be taken as reference. In view of the above discussion, it would also seem appropriate to continue to take the actual earnings as the reference, even though they may differ significantly from what the earnings might have been in the absence of the program.

Overcompensated Cases

A second aspect of the methodology relates to the treatment of over-compensated cases, in which shortfalls turn out to have been overestimated and overcompensated6 in the light of ex post export data. It could be argued that the inclusion of these cases may bias the results toward finding, on average, unduly small stabilizing effects. To the extent that this paper is concerned with actual drawings, it would not be appropriate to separate out the overcompensated cases. Nevertheless, in order to examine the role that forecasting errors might have played, an attempt is made below to examine these cases separately, and use is made of “ex post” drawings—that is, drawings that would have resulted had there been no error in forecasting exports.

Methodology for Adding Drawings to Export Earnings

A third aspect of the methodology concerns the time period over which purchases are regarded as supplementing export earnings. One procedure would be to regard purchases as supplementing exports only over the drawing year. For instance, if the shortfall year was from January 1 to December 31, 1987, and the purchase was made on April 1, 1988, the entire purchase would be added to export earnings in 1988. To the extent that a member cannot be certain of drawing under the CFF, this procedure may appear appropriate. However, to the extent that the purchases may be anticipated, the benefit from purchases may be regarded as being available earlier—say, over the shortfall year itself. Thus, a second procedure would be to add purchases to export earnings in 1987. (Since the shortfall year may differ from the calendar year, the purchase can be added to the calendar year on a prorated basis.)7 A third procedure, and the preferred one for the exercise below, is to combine the above two procedures and to regard the purchases as supplementing export earnings over the shortfall and the drawing years. The distribution of the purchases over the two years would be dependent on the interval between the middle of the shortfall year and the drawing month. In the example above, this interval is nine months (the middle of the shortfall year is July 1, 1987 and the drawing month is April 1, 1988); two thirds of the interval (that is, six months) lies in 1987 and one third in 1988. Hence, under this procedure two thirds of the purchases would be considered to have been made in 1987 and one third in 1988. This is the main procedure used in the exercises below, but in order to test the sensitivity of the results, a comparison has also been undertaken with the first two procedures.

Instability Index

Many indexes are available that could be used to measure instability in export earnings.8 The choice of an index is dependent on a number of specific factors that need to be taken into account given the nature of this analysis. The first is that a measure of export instability should be corrected for the trend in exports, so as to ensure that trend changes in export earnings are separated from a measure of the variability around that trend. Unless this correction is made, any such measure will overstate the degree of instability.9 The problem arises in the determination of the length of the trend. Should it be as long as some business cycles obtaining during the time interval under examination, or should it include the entire time interval? If, say, the export cycle is five years on average, a trend based on some sort of five-year moving average might be appropriate.10 If the export cycle is actually more or less than five years, an index of this form is likely to under- or overestimate the instability present. The alternative would be to use a longer time trend focused on the entire length of the sample interval, which in the present case would be the 11-year period from 1975–85. Ideally, one should obtain an extraneous estimate of the business cycle for each country before deciding on the index. In the absence of such an estimate, both a five-year moving average and a longer time trend were used.

A second important factor in the choice of index is that although the CFF is concerned only with deviations below the trend line (that is, shortfalls), for the purpose of measuring instability, deviations both above and below the trend line should be taken into account. This can be done either by squaring the deviations or taking their absolute value. A third factor is that the measure of instability for any one year should be taken as a percentage of the trend value in that year, so that differences across countries in the magnitudes of their exports do not influence the relative effect on export deviations.

With these three factors taken into account, the following index, I1, was used in the analysis. It is computed as follows:



X¯ = centered five-year geometric moving average of export earnings

Xi = export earnings in the shortfall or middle year

t = number of years over which deviations from X¯ are computed.

As equation (5) indicates, this index is based on the average of the annual absolute percentage deviations of exports from a centered five-year moving average. In addition to satisfying the three requirements noted above, this index has the advantage of allowing for a nonlinear trend. The lower the instability, the lower will be the value of the index, with the value of zero indicating no instability.

In addition to this index, two other indexes were also computed. The first of these, I2, is identical to I1, except that the proportionate deviations from the trend are squared, thereby giving larger values greater weight than smaller ones (see Appendix for details). The second, I3, is based on deviations from an exponential time trend fitted over the sample period. The main reason for using this index was to see if the use of a time trend over a long-run period yielded results markedly different from those obtained by using a time trend based on a five-year moving average. These additional indexes thus provide a check on the results obtained using I1.

The index I1 was first computed for past CFF cases without CFF transactions.11 The index was then recomputed with these transactions, and the differences between the two were examined. The variable used in the analysis is merchandise export earnings expressed in SDRs. This is the same variable that is used in computing the shortfall and is the variable the CFF was designed to stabilize. It might be argued that an analyses using alternative variables, such as foreign exchange reserves, would be more appropriate since stability in reserves might be more important than stability in export earnings, per se. In practice, however, analyses using this variable are difficult due to its extreme variability.12 (In Section IV an attempt is made to relate any gains in earnings stability to the demand for international reserves.)

Finally, it should be noted that in the empirical analysis calendar years are taken as the period of reference. This procedure was adopted because annual data on export earnings were available only on a calendar-year basis. It is possible to interpolate data to obtain values on a monthly basis and to conduct the analysis in terms of the actual shortfall year rather than on the basis of the calendar year. But in practice, the procedures available for interpolations are highly mechanistic and are likely to introduce bias due to spurious smoothing out of seasonal fluctuations.13

III. Empirical Results

The empirical analysis was undertaken for the countries which had made CFF purchases between January 1, 1975 and December 31, 1985.14 This included 79 countries that had made 192 purchases over the 11-year period.15

Interval Between Shortfall and Compensation

Before the results of the exercise are discussed, it is worthwhile to examine the actual time interval between the shortfall period and the compensation payment. Other things being given, the shorter this interval is, the greater is the stabilizing role of the facility. Table 1 gives the number of cases with purchases with a given time lag, as well as the mean time lag. As the table indicates, the (arithmetic) mean shows an average time lag of 3.85 months between the end of the shortfall year and the date of compensation for the period as a whole. The earliest payment came 3 months before the end of the shortfall year, because the purchase was made under the early drawing provision, and the latest payment came 9 months after the end of the shortfall year. The majority of purchases took place within 6 months of the end of the shortfall year. If the whole period is divided into two—from 1975 to 1979, and 1980 to 1985— there is a marginal decline in the compensation intervals from an average of 3.9 months to 3.8 months.16 The drawings before and at the end of the shortfall year as a percentage of all purchases during each of these two periods also increased from 12.4 percent to 16.1 percent from the first to the second period, respectively. These results do not provide any prima facie expectation that the CFF would have had any destabilizing effect.

Table 1.

Interval Between Shortfall and Compensation, 1975–79 to 1980–85

article image

Figures in parentheses denote the number of cases as a percentage of total cases in the given period.

Relative to the end of the shortfall year.

Estimates of Stabilizing Influence

The main empirical results using the index I1 are given in Table 2. The first column in this table shows the value of the index without any CFF transactions for each of the 79 countries. The index is computed for the 11-year period 1975–85. Not surprisingly, it varies considerably across countries, with its value ranging from 2.36 percent to 24.95 percent. The second and third columns report, respectively, the values of this index based on export earnings plus CFF purchases, and the values of the index based on export earnings plus purchases minus repurchases. The last two columns indicate the difference between columns 1 and 2 and between columns 1 and 3, both as a percentage of column 1.

Table 2.

Estimates of Stabilizing Influence of the Compensatory Financing Facility, 1975–85

article image
article image
Note: Column 1 reports the index without CFF purchases; column 2 reports the index with CFF purchases, and column 3 reports the index with CFF purchases and repurchases. Percentage changes in columns 4 and 5 are based on unrounded values in the first three columns; increases in instability are noted by a minus sign (–).

Consider, for example, the figures for Argentina. Column 1 shows that over the period 1975 to 1985, the instability index for Argentina had a value of 9.54. Taking into account the CFF purchases made over this entire period, the index had a value of 8.77; with purchases and repurchases, the value was 9.00. This indicates that CFF purchases can be considered to have reduced the instability of earnings over this period. Even though the repurchases by themselves increased instability marginally, purchases and repurchases taken together still had a stabilizing effect. As columns 4 and 5 for Argentina show, purchases decreased instability by 8.01 percent, and purchases and repurchases together decreased it by 5.57 percent.

As the penultimate row of the table indicates, the average value of this index for the 79 countries was 9.80; with purchases it declined to 9.14, and with purchases and repurchases, it was 9.17. The average decrease in instability as a proportion of the original value of the index was 5.44 percent for purchases and 5.39 percent for purchases and repurchases together. Both these results were statistically significant at the 5 percent level.17 From the table it can also be seen that of the 79 countries, 64 had a decline in instability, and that in 34 of these 64 countries the decline was 5 percent or more. Similar results were obtained from the two additional indexes described above (see Appendix for details).

On the basis of these results, the CFF can be considered to have led to a clear decline in instability in the availability of foreign exchange earnings. It might be asked, however, whether a decline of 5 percent is significant in an economic sense. A number of factors suggest that it is indeed significant and that a larger decline could hardly have been expected. First, the measured decline in instability was the result of CFF drawings that, for most countries, were a small proportion of total export earnings or even of export shortfalls. Second, the drawings were made, on average, only two or three times during the 11-year period. For most of the years, therefore, the fluctuations in earnings from the trend, resulting in surpluses or shortfalls, were not affected in any way by borrowing under the CFF. Third, the CFF is concerned only with shortfalls; excesses (or positive deviations), which the facility was not designed to counter, would influence the value of the index and the proportionate change in it. A fourth factor concerns the way the present CFF is implemented; drawings, subject to quota, are always based on the measure of the shortfall that takes into account the value of exports in the middle year. This means that even if none of the above factors was operative, on the basis of drawings under the CFF, “perfect” stability would still not be possible, since the “shortfall” in the middle year could never be completely eliminated.18 Given all these factors, the decline in instability could not have been expected to be very large. In fact, the magnitude of the improvement in stability appears far from negligible. Over an 11-year period, the decline was more than 5 percent on average, which, given the size of CFF drawings in relation to exports and shortfall noted above, may even be regarded as significant.

A simulation exercise was undertaken to examine the extent to which these results are robust to the use of different schemes for apportioning purchases. In this exercise, the purchases were added using the alternative methods mentioned in Section II above. In the first variation the entire purchase was added to the year in which it was made. In the second variation the purchases were regarded as being available only over the shortfall year, regardless of when they were made. Here, the purchases were added to export earnings by prorating the purchases over the 12 months of the shortfall year.

The summary results of the simulation exercise are given in Table 3. The table reports the change in the values of the instability index I1, as well as of I2 and I3, following CFF purchases, and purchases and repurchases averaged over the 79 countries. As indicated in the table, with drawings added to export earnings in the purchase year, all three indexes show a decline with purchases only (1.8 percent, 1.5 percent, and 4.4 percent, respectively) and with purchases and repurchases (1.8 percent, 1.4 percent, and 4.6 percent, respectively). Although compared to earlier results (purchases distributed over shortfall and purchase years) the decline in instability is now considerably lower, it is still unambiguous and statistically significant.

Similarly, when purchases are added to the shortfall year rather than to the purchase year, there is again a clear decline in each of the three indexes with purchases alone and with purchases and repurchases. The results of this simulation exercise indicate that the conclusions obtained earlier are not seriously affected by the use of different schemes for allocating CFF purchases to countries’ export earnings.

Impact of Overcompensated Cases

The above exercises examined all the CFF cases for which the relevant data were available, including a considerable number of cases that were overcompensated in the light of ex post data.19 The cases were overcompensated due to errors in projecting exports; when actual data became available, the shortfalls were seen to have been smaller, or there may even have been an excess. All cases were examined to obtain as large a sample as possible, but it could be argued that the inclusion of overcompensated cases might bias the results toward finding, on average, small stabilizing effects. In these cases CFF drawings may actually have had some destabilizing effect, since when the drawings were made, in reality there would have been a smaller shortfall or even an excess. This would have amplified the fluctuations in export earnings rather than dampening them.

Table 3.

Earnings Instability and Timing of Purchases

article image

The entire purchase is added to a country’s export earnings in the year it is made.

The entire purchase is allocated to the shortfall year, regardless of when it was made. The purchase is added to export earnings by prorating it over the 12 months of the shortfall year.

The distribution of purchases is as in Table 2; that is, the interval between the middle of the shortfall year and the drawing month is taken into account.

The exclusion of overcompensated cases or the use of ex post data (that is, determining the shortfall and the drawing from actual rather than projected exports) may lead to a finding, on average, of a stronger stabilizing influence. Alternatively, by using ex post data one can examine the extent to which the stabilizing effect of the CFF would have been stronger had there been perfect foresight—that is, if exports could have been projected for the two postshortfall years without any error. To examine this issue, an additional simulation was carried out using index I1 and hypothetical drawings—that is, drawings that would have taken place had there been perfect foresight.20 The results showed that, as expected, the average decline in instability was greater than that obtained previously. The average decline in I1 with ex post purchases and repurchases was 6.4 percent, compared with 5.4 percent. In other words, the stabilizing influence of the facility would have improved by about one fifth if exports had been forecast without error. Although this is a substantial improvement, it is based on an extreme premise.

Table 4.

Country Characteristics and Instability

article image
Note: Countries are divided according to the World Economic Outlook (WEO) classification. Numbers in parentheses are the number of countries in each category. Purchases are distributed as in Table 2.

The four industrial countries include Australia, Iceland, New Zealand, and Spain, with the last purchases in 1976, 1982, 1976, and 1978, respectively.

Instability and Country Characteristics

It is generally acknowledged that developing countries relying predominantly on exports of primary products are likely to have less stable export earnings than countries with a more diversified export structure. The most generally accepted explanation for this difference is the greater supply instability of primary commodities.21 An important question in the context of this study is whether the effect of the CFF in stabilizing earnings differs in any systematic manner across groups of countries relying on different types of exports. The role of the CFF in different country groupings based on their predominant exports was examined as a way to address this question. The groupings are based on the Fund’s World Economic Outlook (WEO) classification and include among developing countries, fuel exporters, primary goods exporters (subdivided into agricultural goods exporters and mineral exporters), exporters of manufactures, and services and remittances receivers. The values of the instability index I1, with and without CFF, were computed for each category. The results are given in Table 4. As can be seen from the table, the export instability of exporters of manufactures was markedly lower than that of all other countries. It was even lower than that of the four industrial countries that have used the CFF in the past. Although the average instability index for the service-exporting countries is the highest, it is disproportionately affected by instability in one country. The instability for the fuel exporters is the next highest, resulting from the sharp fluctuations that have taken place in the international oil market over the last decade.

There were also some notable differences in the effect of the CFF across country groups. The smallest decline in instability was for industrial countries, exporters of manufactures, and fuel exporters, and the largest decline was for exporters of services and mineral exporters.

It may be thought that this difference could be explained by the difference across country groups in the amount of drawings relative to shortfalls. It would appear, other things being equal, that the larger the drawings relative to the shortfall, the greater the reduction in instability. To see if this was in fact the case, total shortfalls and drawings by these country groups over the period 1975 to 1985 were computed. The results indicated that the relationship between drawings relative to shortfalls and the decline in instability was certainly positive but not very strong. For instance, over this period purchases by industrial countries amounted to SDR 684 million, which, as a proportion of shortfalls, was nearly 88 percent. This was the highest proportion of drawings to shortfalls, but, as noted above, the decline in instability for this group of countries was relatively small. However, the proportion for the exporters of manufactures, at 47.1 percent, was the lowest, and the decline in instability for these countries was also the smallest. Exporters of services had the second highest proportion of drawings to shortfalls and also the largest decline in instability.22 These results suggest that although the actual amount of drawing relative to shortfall explains some of the differences in the relative efficacy of the CFF across different groups of countries, other factors, notably the timing of purchases, must also have had some influence. The results also indicate that, contrary to the conclusions reached by some recent studies, low-income countries, including agricultural goods exporters and exporters of services, were not treated in any discriminatory way by the operations of the CFF.23 If anything, the facility had a more pronounced effect in reducing instability in export earnings in this set of countries.

IV. Demand for Reserves and Welfare Implications

The above analysis has shown that use of the CFF led to a decline in the instability of countries’ export earnings. Although the magnitude of the average decline may appear small, given that it is an average for 11 years, it can be regarded as quite substantial. The CFF was, of course, designed with this objective in mind. But this is not the sole benefit of the CFF—it also provides members with funds at rates of interest that are lower than market rates. This was of particular importance, for example, over the period 1978 to 1983, when the average interest rate charged by the Fund on general resources was around 6 percent, whereas the cost of borrowing from commercial markets averaged nearly 14 percent.24 Any evaluation of the CFF would also need to take account of this aspect.

Against this benefit must be weighed the fact that the welfare of creditor countries declines when the borrowing countries receive loans at below-market interest rates. To that extent, the analysis below, which focuses only on the borrowing countries, might overstate the welfare benefits of the facility. It should be noted, however, that the creditor countries may benefit from the CFF if purchases under the facility allow the borrowing countries to sustain their imports from creditor countries.

The analysis below adopts the following approach in examining the benefits to member countries of a decline in their earnings instability. First, the impact that the decline in instability may be expected to have had on a member’s demand for international reserves is examined. Next, the net gain to the member is examined when the decline in instability is considered in conjunction with the availability of low-cost credit. Some illustrative empirical evidence is presented in both cases.

Earnings Instability and International Reserves

In a system of managed floating, which has characterized the world economy since the early 1970s, the demand for international reserves has not changed radically from the preceding period of fixed rates (see Frenkel (1978 and 1983)).25 Issues relating to the use of reserves to finance payments imbalances directly, or to manage these imbalances by intervening to influence the exchange rate, as well as discussions about the adequacy of reserves, are still of considerable importance. Since the demand for reserves is generally acknowledged to be positively related to the fluctuations in a country’s foreign exchange earnings, it is at least conceivable that the availability of funds under the CFF may exercise some influence on this demand. In other words, the fact of the availability of the CFF, partly through any stabilizing influence, may mean that the need, and hence the demand, for reserves will be lower, compared with a situation where there was no such facility. Of course, to the extent that at the time of a shortfall a country cannot be certain that it will be able to obtain drawings under the CFF, the decline in reserve holdings and an increase in imports might not be that substantial. That the amount available is also further constrained by quota limits indicates the limited effect the facility might have on a member’s reserve holdings.26 Nevertheless, one would expect the decline in instability associated with CFF drawings to lead to at least some savings of reserves. The methodology adopted below tries to obtain an estimate of the maximum savings that could have been realized.

In order to obtain such an estimate, the relationship between reserve demand and export instability has first to be estimated. For this exercise the framework developed in Frenkel (1978) to analyze the determinants of the demand for international reserves can be used.27 In this framework the demand function for reserves is assumed to depend on three variables as follows.

The first variable is the average propensity to import (m); this can be interpreted as a proxy for “openness” of an economy, indicating the extent to which it is vulnerable to external disruptions. In this interpretation, one would expect demand for reserves to be a positive function of external vulnerability.28

The second variable is a measure of the variability of international receipts and payments (σ). The choice of this variable as an argument in the demand function is based directly on the use of reserves for dealing with fluctuations in external transactions, and it is expected to have a positive relationship with reserves demand. The empirical analysis undertaken below, instead of focusing on the variability of total international receipts and payments, uses a variable that measures the variability in export earnings directly. (This variable is identical to one of the measures of instability in export earnings used earlier (Index I1).) The third variable is a scale variable, measuring the size of international transactions by the level of imports (M).

An equation for individual countries was estimated with the above three variables as the explanatory variables and with reserves as the dependent variable for a sample of countries included in Table 2 for the period 1975 to 1985.29 The estimates obtained for the coefficient on σt were then combined with the results obtained earlier on the impact on instability of drawings under the CFF to obtain an estimate of the possible savings in reserve requirements.30

The results of this exercise are given in Table 5. The first column of the table reproduces the result obtained fora2—that is, the reserve elasticity with respect to export instability. Column 2 gives the percentage decline in instability (these are the values given in the fifth column of Table 2). Column 3 gives the value of reserves in 1980. Column 4 shows the reduction in reserve demand which was obtained by multiplying columns 1, 2, and 3. The last column gives the reduction as a percentage of reserves in 1980. As the last column indicates, for most countries the decline in instability had a substantial benefit in terms of the savings in reserves. For this group as a whole, the reduction in reserves as a percentage of total reserves was over 7 percent—a far from negligible decline.

Table 5.

Compensatory Financing Facility and the Demand for Reserves: Illustrarive Results

article image
article image

For those countries for which this equation could not be estimated, an average value of the reserve elasticity (1.31) was used. For this set of countries, the decline in reserve demand was even more substantial, at 10.32 percent. Taking all the countries in the table together, the percentage reduction in reserve demand was 8.51 percent, clearly a substantial saving. Of course, it has to be acknowledged that given the nature of the CFF, countries cannot be certain of purchases under this facility, and this figure must therefore be regarded as indicating an upper bound to the savings.

Welfare Gain: Some Illustrative Results

An alternative, but complementary, approach to measuring the CFF’s benefit would be to examine the gain to a country using a standard framework for analyzing changes in economic welfare due to commodity price stabilization schemes. In this framework, the question to be addressed is what value does the country place on stabilization of earnings using the CFF. This value consists of two elements: first, the gain due to decline in instability; and second, the availability of funds at rates of interest significantly below the rates prevailing in capital markets. At the same time, any costs attached to participating in the Fund system, as well as costs relating specifically to applying for the CFF, have to be taken into account.31

In this framework, the welfare function, W, of a potential recipient of the CFF may be considered to include the following elements:



X =level of export earnings

Ix =instability of export earnings

G =the cost of CFF credit relative to credit available on commercial terms

CM =cost attached to membership in the Fund and applying for CFF.

There are a number of ways to make equation (6) empirically operational. Here the framework developed by Newbery and Stiglitz (1981) in the context of obtaining quantitative estimates of the benefits of commodity price stabilization schemes is applied. This framework is concerned with analyzing what the stabilization of the price of a particular commodity is worth to, say, a risk-averse farmer; in other words, what sum of money he would be willing to pay for the stabilization scheme to be introduced.32 If it is assumed that this concept can also be applied to policymakers in a member country, then the benefits due to lower instability can be computed in the following way: suppose that the country’s export earnings areX,0 with mean X¯0 and an index of instability σ; with transactions under the CFF, these earnings change toX1with mean X¯1 and an index of instability σx1. Then, equating expected utility


where B is the benefit to the country of lower instability. B may be regarded as the amount a country would be willing to “pay” to obtain the benefit provided by the CFF. Expanding equation (7) in a Taylor series, the following expression is obtained for the change in welfare:



ΔWW =proportionate change in the country’s welfare


R= coefficient of relative risk-aversion33

Δσx2 = change in the coefficient of variation.

The first term in equation (8) is the transfer benefit—that is, the extent to which the country gains from the CFF credit. The second term is the efficiency or risk benefit—the benefit from reducing instability. With risk-aversion (R > 0), a decrease in instability of export earnings increases the national welfare. Taking into account the costs of making the application, CA, and regarding (8) in terms of an expected change in welfare gives


In order to obtain a quantitative estimate of the change in welfare, the three terms in equation (9) must be computed. The gain in terms of foreign exchange ΔX¯/X¯ will be due to the “grant” element in CFF transactions, which is defined as the difference between the nominal value of purchases minus the present value of future repurchases and interest payments:



P = nominal value of purchases

Cj,Ij = capital repayments and interest payments, respectively, which become due at the end of year j

r=market interest rate

b = service charge as a percentage of gross credit.

Computing the second term on the right in equation (9), 1/2RE, (Δσx2), requires an estimate of the degree of risk-aversion that characterizes policymakers in each of the countries that has obtained funding under the CFF. In the absence of any available information on the magnitude of this parameter and the critical influence it can exercise on any measure of change in national welfare, three different sets of values have been experimented with for R. The first, R = 1, is invariant across countries. This is a special case of a utility function with constant relative risk-aversion. It is given by the logarithmic or Bernoullian utility function


which has unit relative risk-aversion. For this function the proportional risk premium is independent of the level of income, Y. In the present context, “risk premium” denotes the amount of foreign exchange earnings a country would be prepared to give up in order to obtain a stream of earnings that was as stable as the one provided by the availability of the CFF. With R = 1, it is assumed that this risk premium is the same across all countries.

Clearly, this is an extreme assumption—it would be more reasonable to expect the risk premium to differ across countries, depending on the degree of risk-aversion of country authorities, which in turn would depend on the specific circumstances facing the country. Purely for illustrative purposes, it is assumed here that the risk-aversion depends on the degree of export instability and on the availability of reserves relative to imports. It is assumed that the higher the export instability is, the greater is the value placed on the stability of export earnings obtained via the CFF. Similarly, the lower the availability of reserves is relative to imports, the greater is the worth of the CFF. In order to make this assumption empirically operational, one would need to obtain specific values for R for each country, according to both these criteria. In the absence of any detailed information on countries’ utility function, a cardinal measure is not possible, and it is thus not possible to obtain an absolute measure of the changes in welfare.

One can obtain an ordinal measure and compare it to the case of R = 1. The procedure adopted was as follows: in the case of instability in exports, the value of the index for individual countries was deflated by the mean value of the index. This gives a distribution of R around R = 1, which reflects the proposition that the greater the relative instability is, the greater is the worth of the CFF to the country concerned. In the case of the imports-to-reserves ratio, an average of this ratio was obtained for each country for the period 1975 to 1985, which was then expressed as a percentage of the mean of all countries in the sample. Again, this gives a distribution around R = 1 and reflects the proposition that the higher the import/reserves ratio is, the more the country would value the drawings under the CFF. In the analysis below, the three different values of R are


R = Ii/I¯

R = mi/m¯


Ii = instability index for country i

I¯ = average of the index across all countries

mi = imports-to-reserves ratio for country i

m¯=average across all countries.34

With the second and third values of R, the welfare gain from the facility would be greater, the greater is the degree of risk-aversion compared to R=1.

The estimates of the change in instability are the estimates obtained earlier in Table 2. Finally, it is assumed that there are no net costs of making a CFF application, so that the change in welfare can be computed on the basis of the information discussed above.

The results of this simulation are given in Table 6. Consider, for example, the case of Argentina. Over the period 1975 to 1985, Argentina made purchases of over SDR 831 million. Since the rate of interest charged on these purchases was considerably below the market rate, there was an implicit “savings” element in this purchase, which in present value terms amounted to over SDR 174 million. As a percentage of the average value of export earnings, this amounted to 2.5 percent (column 3). Since the average decline in instability was 5.6 percent, using equation (8) it can be computed that with R = 1, this would have led to a welfare gain of 5.27 percent of exports. If, however, it is assumed that the degree of risk-aversion varies according to the degree of earnings instability or the imports-to-reserves ratio, the increase in welfare would have been different. If this risk-aversion is assumed to be distributed around R = 1, then R=Ii/I¯ yields a welfare gain of 5.2 percent and R=mi/m¯i yields a welfare gain of 3.7 percent.

Table 6.

Welfare Gain: Some Illustrative Estimates

article image
article image
article image
article image

This pattern is repeated for other countries as well, with a significant majority showing a clear increase in welfare. The average welfare gains with the three different values of R are 5.6 percent, 6.1 percent, and

7.3 percent, respectively. Excluding outliers for the second and third values of R leads to a welfare gain of 4.0 percent and 4.4 percent, respectively.35 Given the paucity of information on R, these estimates are, of course, purely illustrative, but they indicate that under plausible assumptions, it can be shown that on average the CFF has been of considerable net benefit to Fund members, but also that this benefit has varied considerably across countries.

V. Conclusions

This paper has examined a number of issues related to the role the CFF has played in stabilizing the foreign exchange earnings of Fund member countries. It was argued that on an a priori basis it might be expected that for any given country, the stream of export earnings with the CFF—that is, exports plus purchases less repurchases—would be more stable than earnings without the CFF transactions. However, it was also noted that there are some aspects of the facility that may play a destabilizing role. It is therefore an empirical issue whether the facility has exercised a stabilizing influence, and, if so, what has been its magnitude. The basic methodology was to compute a number of different indexes to measure instability in merchandise export earnings without CFF, and then to determine whether the transactions under the CFF had made any significant difference to the values of these indexes. These results were then used to examine the impact on reserve requirements and to compute a measure of welfare gain from the facility.

Using data for 79 countries (192 CFF cases), it was first shown that the interval between the end of the shortfall year and purchases under the CFF was on average not more than three months. This, it was argued, suggested that the CFF may well have had an important stabilizing influence. Three sets of indexes were then computed, each with and without CFF transactions. The results showed that, regardless of the index, the CFF led to an improvement in the stability of earnings of about 5 percent. Although the magnitude of the improvement could be considered small in absolute terms, it was statistically significant; also, given that it is an 11-year average, it is also significant from an economic point of view. The paper then examined the robustness of these results to alternative rules for adding purchases to export earnings. Although the magnitude of their effect on stabilization differed, in general the above conclusions continued to remain valid. A further analysis excluding those cases where there had been overcompensation showed that the stabilizing effect was somewhat stronger than for the full sample. An analysis of the instability in earnings by predominant export confirmed that exporters of manufactures have a much lower instability than exporters of primary goods. There was, however, no systematic difference in the stabilizing role of CFF across these different groups of countries.

Two further sets of simulations were undertaken to examine the economic significance of the decline in instability. The first was a detailed analysis of the difference this decline is likely to have made to countries’ reserve requirements. A demand function for international reserves, with instability as an explanatory variable, was estimated for a large sample of countries; from this it was deduced that the CFF could have led to, on average, a considerable saving in reserve holdings. The second exercise used the Newbery-Stiglitz welfare framework to analyze the welfare gain to countries from reduced earnings instability. This gain consists of the benefit from reduced instability as well as the concessional rate of charge on CFF purchases. Although the results are sensitive to assumptions regarding the countries’ degree of risk-averseness, gains were not negligible.

The main conclusion to be drawn from these findings is that the CFF has performed a clear role in stabilizing fluctuations in foreign exchange earnings of developing countries. The reason has been the timely availability of substantial compensation to counter the shortfall in export earnings.


Potential for Stabilizing Role and Computation of Indexes

Suppose “perfect” stability is defined in terms of zero shortfall. Then under the present rule of the CFF, one would not obtain “perfect” stability because the drawings (subject to quota) are always based on a measure of shortfall that is computed using the shortfall itself (the middle year).

The unattainability of perfect stability can be seen readily from the following. Suppose that the actual exports are

Xi i = l,...,5,

with X3 the exports in the shortfall year.

The trend in exports is


where Π is the multiplication operator and the shortfall is given by


Adding the full amount S toX3 will not, however, eliminate the shortfall if it is recalculated. To eliminate the shortfall, the compensation should be greater than S. Accordingly, assume that the compensation S is unknown and also that the original exports are X3. There will be perfect stability if and only if S =0. If S = 0, the earnings in shortfall year plus drawings, say X3, will equal the trend value X¯1. Substituting into equation (11)


that is,


Therefore, only if the compensation is X¯1 will stability be perfect.

It can be easily seen that X¯1 will be always greater than X¯0 for anyX3 > 0. This is because X¯1 will give a trend value that will make the shortfall zero. In other words, given the existing CFF formula, even if a country were compensated for the full amount of shortfall, calculating stability ex post will yield a shortfall.

The two indexes, as noted in the text, that supplementedI1were computed as follows.


whereX¯X, and t are as defined in the text. This index is identical to I1, except that here the proportionate deviations from the trend are squared, and a square root is taken of the mean (the “root-mean-square error” formula). It is worth noting that squaring the deviation weights gives larger values greater weights than smaller ones. For example, if this index is used, deviations of five in one year and zero in the next four will be considered more than twice as unstable as deviations of one in five successive years. WithI1, there will be no difference in the two cases. To the extent one may want to give more weight to larger deviations equation (15) would be appropriate. This index could thus be regarded as complementary toI1.

Table 7.

Instability Indexes I2 and I3: Averages

(In percent)

article image

A third index, I3, is computed as follows:



et=residual from an exponential time trend

X¯ =mean of export to the sample period.

To compute this index, an exponential time trend of the form Xit=a0+a1t is fitted to export earningsXit for the period 1975 to 1985. The squares of the residuals from this are then summed and taken as a proportion of mean of exports in the sample period.36 The main reason for using this index was to see if the use of a time trend over a long-run period yields results markedly different from those obtained by using a trend based on a five-year moving average.

The results for the average of the indexesI2 and I3 are tabulated in Table 7. In the first half of the table, column 1 gives the values of the instability indexI2for the average of 79 countries. Columns 2 and 3 show, as before, the average value of the indexI2 based on export earnings plus purchases, and the value of the index based on export earnings plus purchases minus repurchases. As the first row of this table indicates, the average value of the index was 12.2; with both purchases and repurchases, it declined to about 11.5. There is thus not only a decrease of 5.1 percent in instability with purchases, but also a decrease of 5 percent with repurchases. These differences are statistically significant. It is worth noting that the value of the index is larger for each country than that obtained from index I1. This result is to be expected, since for anyX> 0, the root-mean-square error would give a larger value than the sum of the absolute value of deviations. The second half of Table 7 provides the results for index I3, which is based on an estimate of the logarithmic trend in export earnings over the period 1975 to 1985 for the average of 79 countries. Here the absolute value of the index is somewhat larger than is the case for the other two indexes. However, the change in instability due to the CFF transactions is somewhat smaller.

On average, instability declined by 3.4 percent with purchases and 3.8 percent with purchases and repurchases. These results are also statistically significant. As the standard errors indicate, the variations around the mean are quite similar for both indexes.


  • Finger, J.M., and DeRosa, D.A., “The Compensatory Finance Facility and Export Instability,” Journal of World Trade Law, Vol. 14 (January/February 1980).

    • Search Google Scholar
    • Export Citation
  • Fleming, Marcus, Rudolf Rhomberg, and Lorette Boissonneault, “Export Norms and Their Role in Compensatory Financing,” Staff Papers, International Monetary Fund, Vol. 10 (March 1963).

    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A., “International Reserves: Pegged Exchange Rates and Managed Float,” in Public Policies in Open Economies, ed. by Karl Brunner and Allan H. Meltzer (Amsterdam: North Holland, 1978).

    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A., “International Liquidity and Monetary Control,” in International Money and Credit: The Policy Roles, ed. by George M. von Furstenberg (Washington: International Monetary Fund, 1983).

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, The Global Effects of Fund-Supported Adjustment Programs, Occasional Paper 42 (Washington: International Monetary Fund, 1986).

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, and Peter Montiel, “Evaluating Fund Stabilization Programs with Multi-country Data: Some Methodological Pitfalls,” Staff Papers, International Monetary Fund, Vol. 33 (June 1986).

    • Search Google Scholar
    • Export Citation
  • Goreux, L.M., “Compensatory Financing: The Cyclical Pattern of Export Shortfalls,” Staff Papers, International Monetary Fund, Vol. 24 (November 1977).

    • Search Google Scholar
    • Export Citation
  • Goreux, L.M., Compensatory Financing Facility, IMF Pamphlet Series No. 34 (Washington: International Monetary Fund, 1980).

  • Heller, Robert H., and Mohsin S. Khan, “The Demand for International Reserves Under Fixed and Floating Exchange Rates,” Staff Papers, International Monetary Fund, Vol. 25 (December 1978).

    • Search Google Scholar
    • Export Citation
  • Herrmann, Roland, The Compensatory Financing System of the International Monetary Fund, Department of Agricultural Policy and Marketing Report No. 4 (Kiel, Federal Republic of Germany: University of Kiel, 1983).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, Selected Decisions of the International Monetary Fund (Washington: International Monetary Fund, 1987).

  • Kaibni, Nihad, “Evolution of the Compensatory Financing Facility,” Finance and Development, International Monetary Fund, Vol. 23 (June 1986).

    • Search Google Scholar
    • Export Citation
  • Khan, Mohsin S., and Malcolm Knight, Fund-Supported Adjustment Programs and Economic Growth, Occasional Paper 41 (Washington: International Monetary Fund, 1985).

    • Search Google Scholar
    • Export Citation
  • Kumar, Manmohan S., Growth, Acquisition, and Investment (Cambridge, England: Cambridge University Press, 1984).

  • Kumar, Manmohan S., “The Stabilizing Role of the Compensatory Financing Facility,” IMF Working Paper 88/108 (Washington: International Monetary Fund, 1988).

    • Search Google Scholar
    • Export Citation
  • Lizondo, J. Saul, and Donald J. Mathieson, “The Stability of the Demand for International Reserves,” Journal of International Money and Finance, Vol. 6 (September 1987).

    • Search Google Scholar
    • Export Citation
  • Manger, Jon, “A Review of the Literature on Causes, Effects and Other Aspects of Export Instability,” Wharton EFA, Inc. Report (Philadelphia, Pennsylvania: Wharton EFA, Inc., 1979).

    • Search Google Scholar
    • Export Citation
  • Massell, Benton, “Export Instability and Economic Structure,” American Economic Review, Vol. 60 (September 1970).

  • McBean, Alisdair, Export Instability and Economic Development (Cambridge, Massachusetts: Harvard University Press, 1966).

  • Newbery, David, and Joseph Stiglitz, The Theory of Commodity Price Stabilization: A Study in the Economics of Risk (Oxford, England: Clarendon Press, 1981).

    • Search Google Scholar
    • Export Citation
  • Pownall, Roger, and Brian Stuart, “The IMF’s Contingency and Compensatory Financing Facility,” Finance and Development, International Monetary Fund, Vol. 25 (December 1988).

    • Search Google Scholar
    • Export Citation
  • Soutar, Geoffrey, “Export Instability and Concentration in the Less Developed Countries,” Journal of Development Economics, Vol. 5 (September 1977).

    • Search Google Scholar
    • Export Citation
  • United Nations Conference on Trade and Development, “Commodity Export Earning Shortfalls, Existing Financial Mechanisms and the Effects of Shortfalls on the Economic Development of Developing Countries,” study prepared by the UNCTAD Secretariat (New York: United Nations, July 1987).

    • Search Google Scholar
    • Export Citation

Mr. Kumar, an economist in the Research Department, is a graduate of the London School of Economics and Political Science. He received his Ph.D. from Cambridge University where he also taught before joining the Fund. He thanks his colleagues in the Fund for helpful discussions and comments.


See, for example, Goreux (1980) and Kaibni (1986).


In 1979 coverage of the compensatory financing facility (CFF) was expanded to give countries the option of including earnings from workers’ remittances and tourism in the calculation of the earnings shortfall. In 1981 it was further extended by a provision permitting the optional inclusion of a temporary excess (an increase) in the cost of commercial cereal imports. This provision was introduced for a period of four years and renewed in 1985 for four additional years. Under the cereal decision the amount of a drawing is calculated as the sum of the export shortfall and cereal import increase, subject to limits on outstanding drawings in relation to quota.


For a rationale, and its operational significance, see Goreux (1980, pp. 5–7).


A somewhat similar methodology was used by Herrmann (1983).


The problem of estimating the counterfactual in these types of exercises is of considerable general significance. For its application in the area of capital investment, for example, see Kumar (1984, chap. 5).


Not all overestimated shortfalls translate into overcompensated purchases, however, due to the effect of quota limits.


For instance, suppose the shortfall year was from July 1, 1987 to June 30, 1988. Under this procedure, one half of the purchase would be added to export earnings in 1987 and the other half to 1988. It is not suggested here that a member would be able to anticipate fully the benefits of the CFF, but rather to provide a benchmark for evaluating alternative procedures.


See, for example, Manger (1979).


Since trend varies across countries, unless a correction is made, a cross-country comparison of instability and changes in it would also not be feasible. These considerations mean that a simple measure of instability, such as the variance of export earnings, would not be adequate.


See, for example, Fleming, Rhomberg, and Boissonneault (1963) and, more recently, Goreux (1977).


The methodology was identical for indexes I2 andI3(see Appendix).


A preliminary analysis was undertaken using this variable, but because of its extreme variability, even as a proportion of imports, the results were not very meaningful.


Using procedures such as the quadratic interpolation yields hypothetical series that may bear no relationship to the actual series.


As noted earlier, the facility was used extensively from 1975 onwards following major changes in its operations; 1985 was chosen as the cutoff date, since at the time the empirical analysis was undertaken, comprehensive data were available only up to that year.


The total number of drawings over this period was 202, undertaken by 85 countries. However lack of sufficient data on 6 countries (10 drawings) precluded analysis for them.


In 1979 a major review of the CFF resulted in changes in the formula for calculating shortfalls, coverage, and access limits (see Kaibni (1986)).


The computation of the statistical significance assumes that the observations in columns 1 and 2, and columns 1 and 3 are not independent.


For a formal proof, see the Appendix.


This overcompensated category does not include early drawing cases where prompt repurchases are mandatory if estimates of export earnings in the shortfall year turn out to be too low.


The purchases were distributed over the shortfall and purchase years as in Table 2.


There is considerable literature on this topic. For some early studies, see McBean (1966), Massell (1970), and Manger (1979).


The Spearman’s rank correlation coefficient was 0.42.


London interbank offered rate (LIBOR) on U.S. dollar deposits for six months, plus 1 percent spread. During this period, LIBOR varied from 9.2 percent in 1978 to 16.7 percent in 1981, whereas the rate of charge on purchases under the CFF has never exceeded 7 percent. (See Kumar (1988, Annex Table 5) for details.)


There is considerable literature in this area. In addition to the references cited above, see, for instance, Heller and Khan (1978) and Lizondo and Mathieson (1987).


It may be objected that this argument applies to the potential availability of any type of external funding including that from the international capital markets. This may be so in theory, but in practice for a large number of developing countries the availability of funds from other sources is even less certain.


See Frenkel (1978), especially pp. 113–21. See also Finger and DeRosa (1980). This framework adopts the “continuous equilibrium” assumption—that is, that actual reserves adjust to desired holdings during the observation period. The alternative is the disequilibrium approach, which assumes that the behavior of reserves holdings depends on the gap between desired and actual reserves (see Frenkel (1983)).


It might be noted, however, as Frenkel (1978) demonstrates, that the relationship between reserves and the propensity to import is not clear cut. In the case of an unfavorable external equilibrium in a Keynesian model of the foreign trade multiplier, the cost of output adjustment could be reduced if the authorities are able to run down their stock of international reserves to finance the deficit. Since as demonstrated, the foreign trade multiplier (and the contraction in output required in the absence of reserves) is inversely related to the marginal propensity to import, it could be argued that the cost of not having reserves, and hence the demand for reserves, is inversely related to the marginal propensity to import.


The equation estimated is identical to that in Frenkel (1978, p. 115). It is of the form lnRt=a0+a1lnmt+a2lnσt+a3lnMt+utt, is the deviation of export earnings from a trend computed as a five-year moving average). To compute ov for year T, the deviations fromt -3 to T, up to t -11 to T were averaged. The length of the averaging period was decided by a grid search when the variable became the most significant in the regression. (For a majority of countries averaging was for seven years or less.) It should be acknowledged that although the averaging procedure is appropriate for the simulation being undertaken and has been undertaken in other studies, it can produce excessive smoothness in the series (to the extent that the variable is significant, ex post, this problem may not be acute). Further, if there is any measurement error, then a moving average process would have been introduced into the error term ut.


The equation was estimated for each of the countries in the sample. In general, the overall fit of the regression for most countries was quite satisfactory; in a majority of the cases the coefficient on , was positive and statistically significant; the other two variables were also significant in a large number of cases. For details see Kumar (1988, Annex Table 6).


In theory, there should be no or negligible costs attached to applying for the CFF. For processing the application and arranging the purchase, the Fund charges an amount equivalent to 0.5 percent of the loan.


See Newbery and Stiglitz (1981, pp. 92–95). A similar framework has been used by Herrmann (1983).


This is defined as R(X)=XU(X)/U(X), where U is the social utility function; U’ and U” denote, respectively, the first and second derivatives of U with respect to the level of export earnings. See Newbery and Stiglitz (1981, p. 72).


For a number of countries, data on reserves were not available, and for these R = 1 was used instead.


Outliers were defined as cases where welfare change exceeded 15 percent.


This index is similar to that used by Soutar (1977).