In this and past Fund staff papers on officially supported export credits, quantitative analysis of developments has been based on two statistical sources. One source is the Berne Union Country Reporting Quarterly Return, commonly referred to as the “40 Countries Report,” in which the member agencies of the Berne Union report their outstanding commitments, unrecovered claims, and outstanding offers for each of 40 countries. These data are not available to the public and have been provided to the staff on a confidential basis for its use in analyzing various aggregates for individual debtor countries.

The most attractive feature of the Berne Union series is that data are collected in the way most agencies actually keep their books; that is, the concept “commitments” encompasses insured principal and interest on undisbursed as well as disbursed credits. This facilitates consistency in reporting and avoids the errors that can occur when agencies are asked to make estimates of statistical concepts for which they have no hard numbers. The Berne Union data also have the advantage of being a leading indicator, in the sense that they include agencies’ undertakings to insure future exports. The main problem with using the Berne Union numbers is that they are not comparable with other types of debt statistics and do not reflect current financing flows. They also include the insurance of certain transactions that are not exports—for example, performance bonds or insurance against expropriation of machinery used on construction projects; these amounts are, however, relatively inconsequential.

The second source is “Statistics on External Indebtedness: Bank and Trade-Related Nonbank External Claims on Individual Borrowing Countries and Territories,” published as a joint effort by the BIS and the OECD. This series was published for the first time in April 1984, and data are available on a fairly consistent basis only from December 1983. This series reports stocks of export credits on a basis readily comparable with other external debt data, that is, outstanding amounts of disbursed principal only, and attempts to bring together the information available to the BIS on banking credits and to the OECD on export credits in order to provide a breakdown of outstanding amounts among nonguaranteed bank claims, guaranteed bank claims, and nonbank officially supported export credits. However, since the concepts used do not reflect the way most export credit agencies keep their accounts, for certain creditor countries estimation by either the reporting country or the staff of the OECD is required. Another difference between the Berne Union and the BIS/OECD series is that, while the Berne Union data cover only export credit insurance agencies, the BIS/OECD series also covers the U.S. Commodity Credit Corporation, direct lending by institutions such as the Export Import Bank of Japan, and certain military credits that are not granted or covered by export credit agencies.

Both of these series suffer from the shortcomings discussed in Section III regarding the omission of certain rescheduled export credits, the carrying of transactions at historical contract exchange rates, and the nonavailability of exchange rate adjusted data. A major problem is that neither France nor the United Kingdom reports rescheduled export credits to either the OECD or the Berne Union, and that Japan and Spain report such credits only to the OECD. It also would appear—at least for the United Kingdom and Japan, but possibly also for France and Spain—that amounts in arrears are not included in the reports to the OECD and/or the Berne Union. These omissions create a serious downward bias in the reported numbers for countries that have experienced debt-servicing difficulties.

As elaborated in Section III, a second problem in using these numbers arises from the fact that, for transactions where the currency of the insurance commitment is different from the currency of the underlying export credit, agencies generally carry these transactions on their books in domestic currency at historical exchange rates specified in each insurance contract. These domestic currency amounts are then added up and the total converted to U.S. dollars at current exchange rates in the reports of both the BIS/ OECD and the Berne Union. While this is, of course, the appropriate way for agencies to keep their own accounts, as it represents their actual exposure, it means that outstanding amounts reported at “current exchange rates” do not measure the actual stock of export credits at current exchange rates, but rather the agencies’ total contingent liabilities, that is, their “exposure,” at current exchange rates.

Given that both the BIS/OECD series and the Berne Union series use data reported by creditor countries on the basis of “exposure,” in the sense used here, rather than on the basis of the currency of the underlying credit, the staff in adjusting those series for the effect of movements in exchange rates has had to use this same exposure concept.²³ The exchange rate adjusted data represent, therefore, not exchange rate adjusted export credit flows, but rather exchange rate adjusted exposure. The implications of this difference might best be seen by further elaboration of the example given in Section III. An export credit for US$100 had been insured in deutsche mark and the insurance contract had specified an exchange rate of DM 3 = US$1 to be used in calculating any claims payment. In this case Hermes would carry on its books a commitment of DM 300 in insurance cover. If the current exchange rate were DM 2 = US$1, that commitment would be reported in the unadjusted BIS/ OECD and Berne Union series as being the equivalent of US$150. If that credit were repaid, reported out-standings would drop by US$150. If, at the same time, a new US$100 credit were insured, it would be carried on Hermes’s books as DM 200 at today’s contract exchange rate of DM 2 = US$1 and reported as being the equivalent of US$100. Therefore, while the actual stock of outstanding export credits in U.S. dollars would be unaffected by this repayment and offsetting new credit, the reported totals would drop by US$50.

It must be stressed that the foregoing problem is not pervasive, as in most cases the currency of the insurance commitment is the same as the currency of the underlying export credit. Nevertheless, it may affect about 10 to 20 percent of the outstanding stocks, and the distortions introduced can be important if current exchange rates differ substantially from average historical contract rates. It should also be noted that the size and direction of the bias depends on differences in levels between current and average contract exchange rates. There is the additional problem that the rates at which, say, the deutsche mark totals are converted to U.S. dollars differ at each reporting date; this, and this alone, is the change in exchange rates for which the staff has been able to attempt a rough adjustment. The exchange rate adjusted numbers used in this report, therefore, treat a DM 200 insurance commitment as such, regardless of the currency of the underlying credit, and adjust only for current changes in the exchange rate between the deutsche mark and the U.S. dollar. In this paper, this concept is called “exchange rate adjusted exposure” whether it refers to the BIS/OECD series or the Berne Union series.

A more accurate calculation of the exchange rate adjusted exposure would have required, for each reporting date and for each data series, information on the currency composition of each agency’s exposure. Such comprehensive information was not available, and shortcuts had to be taken to work with the data at hand.

Each of the 11 agencies provided information on the currency composition of exposure, BIS/OECD basis, as of December 31, 1985, both for all developing countries and for each of the 14 sample countries. As only end-1985 data were available for most agencies, it was assumed that the end-1985 currency composition held for all years and for both the BIS/OECD and the Berne Union series. Also, the OECD totals were distributed among the 11 creditor countries studied, assuming they were the only reporting countries.

Apart from these assumptions, the exchange rate adjustment calculation used here is a standard one. First, the currency composition estimates and the end-period exchange rates were used to break down the end-period U.S. dollar totals into end-period stocks for each currency. The change in that stock for any given currency between two reporting dates was assumed to represent the true increase in exposure in that currency. These increases in exposure were then converted into U.S. dollars at the average exchange rate for the period. Exchange rate adjusted stocks were then calculated by adding these exchange rate adjusted flows, cumulatively, to the earliest actual stock data, that is, end-1982 for the Berne Union data and end-1983 for the BIS/OECD data.