The IMF's Statistical Systems in Context of Revision of the United Nations' A System of National Accounts

4 Conversion of Balance of Payments Transactions as a Source of Valuation Changes: Problems, Principles, and Practical Solutions

Vicente Galbis
Published Date:
September 1991
  • ShareShare
Show Summary Details
Pierre Luigi Parcu

Compiling a country's economic accounts requires the recording of a large number of transactions, many of which will be expressed in a currency other than the unit of those accounts. For all such transactions, it will be necessary to convert the original values into the unit of account. When improperly conducted, however, conversion will introduce misleading information into the accounts. This problem can surface in compiling domestic transactions, but it will be most prevalent in compiling international transactions.

The first section of this paper discusses the concept of valuation change in order to clarify the nature of the problems encountered by compilers; it also examines the specific role of conversion in introducing valuation changes in the accounts. Section II discusses the theoretical solution to the conversion problem. Section III identifies some practical difficulties that any attempt to apply rigorous principles will unavoidably face and also develops some practical suggestions. Section IV discusses the conversion of resident-resident transactions. The final section briefly summarizes findings and offers some concluding remarks.

I. Valuation Changes and Conversion as a Source of Such Changes

Balance of payments and other external sector accounts are statistical statements that primarily record the value of transactions effectively concluded between the residents of an economy and the rest of the world. A transaction should be recorded using the market price at which a good is acquired and disposed of by the parties involved in the transaction. This essential rule of recording implies that changes in the price of a good that occur before or after the moment at which the transaction price is defined in a contract are irrelevant for the compilation of the accounts. This point deserves clarification. The price of a good fluctuates over time as a result of changes in demand and supply for that good. From the point of view of the transactors and the compilers of the accounts, however, only the price specified in a given contract is relevant. As a consequence, any change in the price of the good being transacted—between the time the contract is concluded and the time of the acquisition and disposal of the good—is not part of the transaction but represents a valuation change. By convention, the accounts exclude all valuation changes unless and until they are realized under a new contract.

Some corollaries to the concept of valuation change deserve illustration. Valuation changes derive from the fact that, at the moment a contract is closed, the party acquiring a good assumes a valuation risk. The assumption of this risk is independent from, and antecedent to, the assumption of legal ownership of the asset and of the risks connected with it. For example, if cars being transported to a U.S. importer were lost, perhaps through a shipping accident, the importer would not be exposed to any loss if ownership of the cars still resided with the exporter. Figure 1 helps to distinguish between these two kinds of risks. The lower part of the figure shows how the ownership risk extends from the moment an economic entity receives delivery of an asset to the moment in which that entity delivers that asset to another entity. The upper part of the figure shows that the valuation risk for that asset starts at the moment an entity contracts to acquire that asset and ends when it enters into another contract to dispose of it.

Figure 1.Valuation Risk Versus Ownership Risk

As the price of an asset is fixed in a contract, the party disposing of the asset has no economic interest in price fluctuations for that particular asset once the contract is closed. From that point on, the party acquiring the asset bears the full economic burden of any change in its price. The shift of the valuation risk is full and immediate, a circumstance that gives added importance to the choice of the contract price as the price at which to record the transaction in the balance of payments and external sector accounts. A contract may not, however, specify a price for the transaction in question but may, instead, specify the basis on which to determine the contract price—for example, by reference to prices quoted on commodity markets at the time of delivery.

An interesting by-product of the way in which valuation changes are treated in the balance of payments and external accounts involves cases in which realized valuation changes may be the only entries to be shown in the accounts. This occurs when an asset is acquired and disposed of at different prices in the same accounting period.

In general, economic transactions relevant to the balance of payments and external sector accounts of an economy are conducted in many different currencies. Thus, the compiler will have to choose a currency as the unit of account in which to denominate the accounts. As the price of the transaction currency may vary in relation to the unit of account, the process of converting different economic values into the chosen unit of account may introduce valuation changes akin to the ones deriving from price changes.

For the compiler of the accounts, the valuation change is highlighted by the conversion of transaction values into a different unit of account. Of course, if the exchange rate did not change, the process of conversion would not cause any valuation change, but once the exchange rate has changed, it is through conversion that valuation changes enter into the accounts. In the given example, it is the fact that the accounts are expressed in U.S. dollars, and not the change in the exchange rate per se that produces the valuation change in the statistics.

When transactions are expressed in different currencies, and exchange rates between currencies fluctuate widely, a second source of valuation changes acquires significance. When compiling a country's economic accounts in its preferred unit of account, the conversion of transactions originally expressed in a currency other than the unit of account may require the explicit inclusion of an account for valuation changes in the statistical statement.

II. The Solution in Principle

The analogy between valuation changes resulting from price changes and those stemming from exchange rate changes clarifies the nature of the problem for balance of payments and external sector accounting and also points to a theoretical solution. For changes in asset prices expressed in the transaction currency, the solution for ensuring that unrealized valuation changes are not included in the statistics has been to use the transaction price as defined in the contract. The advantage of this practice, besides its theoretical merits, is that it allows statisticians to adopt a consistent procedure for identifying a unique price at which to record a transaction. Price changes that occur after the contract is entered into should not affect the price at which the transaction is recorded, unless the contract allows for this possibility. Thus, irrespective of the market price when the change of ownership occurs, the accounts record the transaction at the contract price.

Similarly, the exchange rate that will establish a uniquely defined value for recording a foreign transaction denominated in a currency other than the unit of account is the exchange rate existing at the contract date. If this method is not applied, the market value in terms of a unit other than the transaction currency would not have to be a single, determinate value but could vary. This principle is set out in paragraph 123 of the IMF's Balance of Payments Manual (BPM).

In principle, subsequent conversions from the domestic unit of account to other currencies or units used as a standard for international comparison would not present further difficulty. As long as the conversion—directly from the original transaction currency or indirectly from the domestic unit of accounts—is made by using the exchange rate in existence on the day of the contract, no unrealized valuation change would be included in the statistics.

Restated, then, the principle is that only one exchange rate and one price exclude all kinds of unrealized valuation changes from the accounts: these are the exchange rate prevailing on the date of the contract and the price specified in the contract.

Although the principle of choosing the exchange rate prevailing at the date of the contract follows the logical analogy between the two kinds of valuation changes, it is useful to examine the limits of this analogy to appreciate better why the conversion problem is more difficult than that of pricing foreign transactions when conversion is not necessary.

Section I reviewed the question of who bears the risk of price-induced valuation changes in the market value of assets that are exchanged and during the period that risk is borne. It should be reemphasized here that the market price that prevails when the valuation risk shifts is the one relevant for the definition of the contract price. For valuation changes stemming from a change in the price of the transaction currency, the situation is identical, irrespective of whether the transaction currency and the unit of account are the same. Such valuation changes do not show up in the accounts, however, unless the unit of account differs from the transaction currency, and conversion from the latter into the unit of account is necessary.

III. The Solution in Practice

Although the theoretical solution for avoiding the introduction of conversion-related valuation changes is clearly stated, its practical applicability, unfortunately, is severely limited by the lack of statistical information. The principle for conversion discussed in the previous section requires knowledge of the date of the contract of each transaction. But contract dates usually are not reported in documents available to the compilers. Furthermore, even in the rare instances in which the date of the contract is available, it would be extremely difficult to identify all the phases of a transaction that need to be converted using the exchange rate of the date of the contract.

Not surprisingly, a much simpler method is generally applied in practice. Transaction values are usually converted into the unit of account at an average rate for the period in which the transaction occurs. This commonly used method of conversion introduces unrealized valuation changes and also incorrectly eliminates some realized ones, as shown in the following example. Assume that, say, on a Monday a U.S. investor contracts to acquire a German bond for the price of DM 1,000, and that the deutsche mark-dollar exchange rate is DM 3 = US$1. The asset is delivered and paid for on, say, the Friday of the same week when the exchange rate is DM 2 = US$1, and on that same day is sold for DM 1,000 to a German bank with the DM 2 = US$1 exchange rate still in force.1 There are two transactions. Following the principle of conversion at the rate prevailing on the contract date, the U.S. accounts should show a debit entry of $333.3 for the acquisition of the bond and a credit entry of $500 for its sale. (Actually, because of net recording, only a net credit of $166.7 stemming from the realized valuation change should appear in the statistics.)

If in practice both transactions are converted at the period average rate for the week of, say, DM 2.5 = US$1, the debit entry for the purchase of the bond will be $400. This entry, when compared with the theoretically correct value of $333.3, introduces into the accounts an unrealized valuation change of $66.7. The sale of the bond, also converted at the weekly average rate, will result in a credit entry of $400, which, compared with the “correct” credit entry of $500, introduces another valuation error by excluding the realized valuation change.

Thus, the practical solution of using average exchange rates distorts the statistics. The degree of distortion will vary with several factors: the volatility of the exchange rates, the extension and variability of the time lag between the contracts and their realization, and the degree of dispersion of a country's transactions in many different currencies. At one extreme, exchange rate stability, the rapid execution of contracts, and a high concentration of transactions in the currency used as the unit of account are the conditions that minimize statistical distortions. At the other extreme, widespread exchange rate instability, marked variability in the pattern of trade, and dispersion in currency invoicing all add to the distortions.

There seems to be no doubt that the increased volatility of exchange rates in recent years has rendered the problem of conversion more severe and the practical compromise of using period average rates less reliable.

There may be additional practical compilation difficulties resulting from the necessity of converting from the domestic unit into another currency or unit of account used for international comparison. Because this further conversion should be implemented by applying the principle expounded in this paper, it would be necessary to maintain and use information on the date of contract to convert either directly, from the transaction currency to the international standard, or indirectly, from the transaction currency through the domestic unit into the unit of account used for the international standard.

Moreover, when the compiler has only aggregated data, it would not be feasible to apply anything but an average rate for the second stage of the conversion, even if the first stage of the conversion had been done at the correct exchange rate. This kind of practical difficulty, however, applies no matter which uniquely specified conversion method is used.

Any solution using a period average for conversion is a compromise dictated by considerations of statistical feasibility. Therefore, conceptually more satisfactory criteria for conversion should be used whenever possible. One good example is the recording of large transactions, such as the sale of a ship or an aircraft. In this case the source of information may be different from the trade returns and very likely will be sophisticated enough to allow the identification of the date of the contract.

Another case in which conversion at an average rate should be avoided is when stocks of assets at two balance-sheet dates are being compared. In the balance of payments this is particularly important when a compiler wants to calculate total changes in reserves. Conversion is necessary because the composition of reserves will, at least to some extent, reflect the composition of a country's trade, so that it is unlikely that any country will keep all of its reserves in the currency chosen as the unit of account. If a compiler wants to identify total changes in reserves between two balance-sheet dates, he cannot use period average exchange rates for conversion. Instead the compiler has to use the exchange rates at the beginning and end of each period and compute the difference between the two levels.

An example will help to clarify further the drawbacks of using period average rates. Assume that a country holds a part of its reserves in dollars and compiles the accounts in its own currency (OC). Assume, furthermore, that the stock of dollars the country holds is $1 million and, for the sake of simplicity, that these holdings do not change from the beginning to the end of the period. Consider now a case where the exchange rate between the country's domestic currency and the dollar changes during the period from US$1 per OC 1 to US$1 per OC 3. Conversion of the stock of dollars at the exchange rates in force at the beginning and end of the period will indicate a change (increase) of OC 2 million in the value of reserves. Applying the period average rate for converting the opening and closing balances would lead to the wrong conclusion: that is, that there had not been any change in the value of reserves when expressed in the unit of account. The problem here stems from the fact that stocks are measurable only at precise points in time. Therefore, the use of a period average exchange rate (a “flow” rate) is clearly misleading.

IV. Conversion of Resident-Resident Transactions

The choice of the correct exchange rate for converting international transactions expressed in a currency other than the unit of account used to compile the balance of payments and external sector accounts is based on a clear principle. This section discusses how that principle extends to transactions among residents of the same economy that have to be included in the external accounts.

The accounts are not restricted to exchanges between residents of the compiling economy and nonresidents. Indeed, the accounts for the compiling economy should also include exchanges of financial claims on another economy between residents that do not belong to one and the same sector of the compiling economy.

The number of such transactions is quite large, since most international transactions trigger a related resident-resident transaction. For example, when an exporter sells foreign exchange acquired through his transactions with the rest of the world to a commercial bank or to the monetary authority in his economy, or an importer does the opposite, a transaction between residents of that economy is to be recorded in the accounts. Furthermore, there may be resident-resident transactions that are not directly related to an international transaction. For instance, the private sector's acquisition (or disposal) of foreign exchange for speculative purposes, when contracted with another sector of the same economy, must be entered into the accounts. Often, however, such resident-resident transactions are expressed in the national currency and, therefore, will not present a conversion problem.

The conversion of resident-resident transactions can be best explained by reference to the transactions of an importer who is acquiring a foreign good and is paying for it in foreign exchange. For this purpose, it is assumed that the conclusion of the contract, the delivery, and the payment are all simultaneous and that at the time the contract is concluded the importer owns the foreign exchange for concluding the transaction (for example, in the form of a deposit in a foreign bank) and, hence, does not have to procure it. In such a case, the international transaction calls for a debit in the merchandise account, to reflect the import, and a credit in the account for short-term assets of the private sector, to reflect the reduction in foreign assets. In principle, the compiler should convert the transaction at the market spot rate of the day of the transaction when expressing the accounts in national currency.

If, instead, the importer does not own the foreign exchange but contracts with a domestic commercial bank to make the payment in foreign currency on the day he contracts to acquire the import, the accounts should again show a debit entry for the merchandise import, but the credit entry should reflect the reduction in the short-term assets of the commercial bank. The resident-resident transaction, which transfers the ownership of an amount of foreign exchange from the commercial bank to the private sector, will not show up in the accounts because the holdings of the private sector, which increase as a result of the purchase of foreign exchange from the commercial bank, are immediately brought back to their initial position by the payment abroad of the foreign exchange. The effect of the resident-resident transaction, albeit real, is offset immediately, and the accounts show only the entries for the merchandise import and the reduction in the commercial bank holdings. Even in this case, the resident-resident transaction does not create conversion problems.

When the time of the contract to buy the merchandise differs, however, from the time of the contract to buy the foreign exchange, conversion is more complex. If, for example, it is assumed that cars are paid for at delivery and the importer contracts to buy foreign exchange to pay for the cars on the day of delivery, the presence of a transaction between the importer and his commercial bank brings out any capital gain (or loss) the importer experiences, in terms of national currency, as a result of the lapse of time between the contract to buy the cars and the contract to buy foreign exchange. The exchange rate applied to the resident-resident transaction—the spot market rate at which the importer buys the foreign exchange from his commercial bank—is the correct one for that transaction, but it is not necessarily the rate to be used in converting into national currency and recording in the accounts the price contracted for the cars. The correct exchange rate for converting that price is the one prevailing when the purchase contract was concluded. In this example, the accounts should record the importer's realized capital gain (or loss) by an entry in the account of the private sector. The role of the resident-resident transaction is that of making explicit the loss or gain of the investor.

The point to be emphasized is that a conversion problem emerges only if there is a time lapse between the conclusion of a contract and its realization. Furthermore, the presence of a resident-resident transaction involving a foreign asset will make explicit, in terms of the national currency, the values to be attached to the entries in the accounts.

In summary, it can be said that in cases in which time plays a significant role—because of exchange rate changes between the contract of the international transaction and the contract of the related resident-resident transaction—a resident-resident transaction will bring out any capital gain or loss induced by the international transaction. Incidentally, the necessity of putting different values on the international transaction and the resident-resident transaction (that is, the need to convert them at different exchange rates) makes clear, in contrast with the case when there is no lapse of time between contracts or a lapse of time does not matter (because the exchange rate does not change), that the two transactions should be considered separately. Although an international transaction often triggers a resident-resident transaction, there are in fact two distinct transactions. The resident-resident transaction should not be seen as part of the contract concluded between the resident and the nonresident.

V. Conclusions

Converting different transaction currencies into a single unit of account tends to introduce valuation changes into the balance of payments and external sector accounts. This paper has argued that these valuation changes should be excluded from the statistics and that only one principle—conversion at the exchange rate prevailing at the date of the contract—completely guarantees this result.

An analysis of the widespread practical difficulties in correctly implementing this principle has led to a search for possible compromises between theory and statistical feasibility. In practice, the most commonly applied technique is to convert at the average market rate prevailing during the period in which the transaction occurred. The type of average (arithmetic or geometric) that should be used for this purpose is not a matter of great significance, since inexact results will be achieved by any practical method of conversion.

1The price of the bond in deutsche mark does not change during the week under examination.

    Other Resources Citing This Publication