Physical Currency Movements and Capital Flows

International Monetary Fund
Published Date:
June 1992
  • ShareShare
Show Summary Details
John F. Wilson

In the literature on international capital flows there is almost no mention of currency movements. “Capital” is usually understood to move through the channels of institutional finance or company accounts, where book entries record the changes in international assets and liabilities. Yet some capital moves in simpler and more tangible ways. Whenever domestic currency notes are acquired by nonresidents, foreign claims increase, constituting a capital inflow. The flow of currency back into residents’ hands represents a decrease in external liabilities—an outflow.1 The balance of payments counterpart entry to the currency flow may belong in a country’s current account, as with tourist expenditures, or the entry may belong in the capital account. An example of the latter case is a domestic bank’s shipment of currency to a foreign correspondent, whose demand account is debited in payment. Only a few major currencies—those widely accepted in global transactions or regarded as stores of value—are likely candidates for substantial international movement. The U.S. dollar is the most obvious candidate, but there are others, too.

Measuring Currency Flows

A seamless statistical system would account for changes in cross-border currency ownership as part of international capital flows. However, currency movements pose two serious problems for compilers: measurement of the flows and appropriate use of available data. For perhaps obvious reasons, measuring currency flows is difficult. Although movements through financial institutions are accessible to statistical systems, large amounts of currency surely flow through individuals’ hands and other channels. Only a few major countries (see annex) make any attempt to record or estimate international currency movements. Compilers who do gather some statistics rely almost entirely on reports from domestic banks about cross-border shipments of bank notes, and they concede that coverage is fragmentary. Proper use of the resulting data is also problematic, as it is hard to determine what portion of net currency movements should be associated with particular current or capital account transactions.2

Whether or not currency flows can be properly measured, cumulative flows of any national currency in one direction can constitute a large capital flow that is presently unmeasured in either debtor or creditor countries. In that sense, currency flows may contribute to the problems of measuring capital flows. Owing to the lack of systematic data on this subject, it was not possible for the Working Party to address currency movements in its Report on the Measurement of International Capital Flows.

If net international currency flows were small, omitting them from balance of payments calculations would make little difference. Evidence suggests, however, that net movements of several key currencies are relatively large, and the omission may play a role in both national and global statistical discrepancies.3 In this context, key-currency countries should be understood to be those whose physical currencies are candidates for transactions, for hoarding (as a store of value), or for investment holding (such as exchange rate speculation) by nonresidents.4

John F. Wilson was the Director of the technical staff. He is now a Senior Economist with the IMF’s Middle Eastern Department.

Discussions of currency stocks are seldom found in a balance of payments context. The most frequent mention of them is usually made in research attempts to estimate the size of a country’s underground economy. A number of such studies have been carried out in the past 15 years, mostly for the United States. Until recently, such work has not focused on—and sometimes has not mentioned at all—the role of cross-border movements in the growth of the currency stock.5

It is widely accepted that the U.S. dollar circulates in many parts of the world, especially those afflicted with political and economic instability. Translating this knowledge into specific estimates of either the stock of dollars held abroad or flow estimates of dollars moving across borders is a much more difficult task. Some estimates suggest that more than 50 percent of the U.S. currency stock is in foreign hands, but they are largely unsupported by systematic statistics (see annex). Little else has been written about the nonresident holdings and cross-border flows of currencies beside the dollar.

In the past few years, questions relating to currency growth have sometimes been connected to drug trafficking and other illegal activities. The framework resembles attempts to estimate domestic underground activities, except that it is more international in nature. Yet, as noted in Chapter 11 of the Report on Capital Flows, the largest portion of cross-border currency movements probably is not connected to drug trafficking at all. There are numerous channels—institutional and individual—through which domestic currency can move abroad, and the reasons for such movements are usually more prosaic than the drug or arms trade. For instance, in a general sense, the acquisition and retention of a foreign currency may be a form of capital flight resulting from political or economic uncertainty, inflation or exchange rate considerations, a desire to hide assets or avoid taxes, or many other possible factors.

Quantifying International Currency Flows

Whatever the motivation, physical currency movements are a type of poorly measured capital flow, and the size of actual movements is an important gap in capital account statistics. Although the scope for improving the formal statistics may be limited, it may be possible to identify which currencies are most affected and roughly to quantify the amount of cross- border movements over time. The following experiment in quantification is similar to previous studies of domestic underground economies in that it uses an inferential approach to the subject of unmeasured capital flows.6 Crude estimates of such flows can be obtained by comparing data on changes in currency stocks to some assumptions about how these stocks should behave over time.

Specifically, the following paragraphs examine the real per capita currency stock in eight countries over the 1970-90 period.7 All of the currencies are candidates for significant cross-border movements because of inflation, exchange rate influences, and geography. In none of these cases can the quantity of currency circulating domestically be isolated from the amount held by nonresidents, so conclusions reached in this paper are suggestive at best.

Chart 1 displays the 1970-90 movements of the nominal currency stocks and real per capita currency stocks for four of the eight countries: France, Germany, Japan, and the United States. "Real currency per capita" is the total currency stock divided by population and deflated by the consumer price index. These adjustments are intended to eliminate the upward push on currency stocks associated with growing populations and generally rising prices. The chart shows that real per capita currency stock declined sharply in France over the past two decades and rose noticeably in Germany, Japan, and the United States.

Chart 1.Currency Stock in Selected Countries, 1970-90

How should real currency balances behave over time? Expected behavior has cyclical as well as systemic features. Physical currency is a component of narrow money in all countries, and narrow money is substituted for broadly defined money when interest rates change. When interest rates rise, the opportunity cost to a domestic resident of holding domestic bank notes increases, causing per capita holdings of currency and narrow money to fall. The converse holds true for declines in interest rates.8 Real balances will therefore fluctuate with interest rates.

For the most part, however, interest rate changes are a cyclical force, not a secular one. A more important long-run influence on currency balances is technology—that is, changes in domestic payments mechanisms. Technological changes have tended to work against the demand to hold currency. Widespread use of credit and debit cards, electronic payments, and similar mechanisms available to the individual consumer should have reduced the currency intensiveness of transactions over the past two decades. Thus, while one might expect to see cyclical ups and downs in currency balances, a distinct downward drift should occur reflecting the technological advances.

The exact rate at which real currency balances should decrease is a matter of extreme conjecture and will likely vary across countries. However, each country might be thought to follow an expected “baseline” evolution of real per capita currency stocks. For present purposes, it was assumed that between 1970 and 1990 advances in payments technology in each country lowered the intensiveness of domestic currency usage by 10 percent. The difference between actual and baseline balances is an unexplained residual, representing a cumulative change in the currency stock induced by factors other than prices, population, and technology. For key currencies, this residual could largely be cross-border movements of bank notes.

Among the countries shown in Chart 1, a persistent downward drift in real per capita currency stocks has taken place only in France. There, currency balances not only decline in a way that seems consistent with changes in payments technology and consumer behavior but they drop even faster than hypothesized over an extended period. There have been striking upward movements in per capita currency balances for Germany and Japan, and to a lesser extent the United States.9

Table 1 gives further details of the calculation for all eight countries. It shows “unexplained” currency growth—the residual—in real per capita terms. This residual is then translated back into nominal currency amounts: the total quantity of currency at current prices that is “unexplained” by population, prices, and technology. Finally, the result is shown in dollar terms.

Table 1.“Unexplained” Currency Growth in Eight Countries, 1970-90(Currency units as indicated)
Local currency
Real perNominal totalBillions of
United Kingdom−62−4.7−8.4
United States21465.065.0

Real per capita stock in 1990 less 10 percent of 1970 level.

Scaled by population and consumer price index.

Translated by 1990 average exchange rate.

Real per capita stock in 1990 less 10 percent of 1970 level.

Scaled by population and consumer price index.

Translated by 1990 average exchange rate.

Apart from the possible contribution of domestic underground economies to these residuals, the signs and magnitudes of most of the results are intuitively plausible. The local currency figures in the second column might be interpreted as upper-bound estimates of cumulative capital flows associated with physical currency movements over this interval. The dollar figures, however, are equivalent to the cumulation at 1990 average exchange rates. They should not be interpreted as the dollar value of the capital flows themselves. This distinction is especially important for the strikingly large Japanese figure, where yen appreciation during the 1980s inflated the dollar equivalent of the putative currency outflow.10

Another aspect of national currency balances is relevant to speculations about their role in capital flows: the size of the bills in circulation. It has been noted that an unusually high (and rising) proportion of U.S. currency outstanding is in denominations of $100 and greater. Although this is consistent with the presumed nature of some illegal domestic activities, it is also consistent with increased hoarding and transactions (including illegal ones) outside the United States.

As noted above, there is no simple “benchmark” for currency outstanding for domestic use. Some national economies tend to be more currency intensive than others, in the sense that more transactions are traditionally done in cash, implying higher average amounts of real per capita balances. As Table 2 shows, however, the amounts of currency outstanding vary greatly across the eight countries, even apart from the movements shown in the Chart 1.

Table 2.Per Capita Real Currency Balances, 1985(U.S. dollar equivalents)
United Kingdom325
United States710
Source: International Financial Statistics.
Source: International Financial Statistics.

Switzerland is the most prominent “outlier” in the table, with impressive real per capita currency stocks, even after allowance for exchange rate movements. There has not been much change in the real demand for Swiss currency in recent years, and the figures are consistent with the notion that a certain quantity of Swiss francs are in nonresident hands.


These final comments stress that multiple influences work on currency demand. The simple fact of “rising” or “falling” real per capita currency balances can be the result of both domestic and foreign developments, and the comparisons made with baseline assumptions in this paper are merely indicative. What several findings do suggest, however, is that foreign demand for several physical currencies has been rising over the period covered by this paper, implying unmeasured capital flows into the countries of the currency issuers. As this appears to have taken place over a fairly long interval, the cumulative capital inflow has perhaps been substantial. If this were the case, the outstanding stock of currency owned by nonresidents may not be a negligible factor in the international investment positions of such countries or in those where the currencies are held.


National Statistics on Currency Movements

Only a few countries gather statistics on international movements of physical currency or use such figures in their balance of payments calculations. Compilers in three of these countries provided information to the Working Party about flows of the domestic currency between domestic and foreign residents. Principal features of the systems for Germany, Japan, and the Netherlands are described in the following paragraphs. It appears that none of the existing systems collects information on foreign currency acquired by domestic residents. A conclusion summarizes some partial information for the United States.


German authorities are well aware that the deutsche mark is used for transactions and for store-of-value purposes outside the country. In addition to currency exports by travelers, deutsche mark notes are used for transactions in certain neighboring countries, and many foreign workers in Germany send marks to their home countries, either for their families or for saving purposes. In some cases, German citizens are thought to transfer mark notes out of the country, especially during times of public debate about releasing data on bank accounts to the tax authorities.

The return of currency notes to Germany is likewise associated with tourist and travel expenditures, but it mostly results from foreign banks shipping their excess mark balances to German banks in exchange for deposit credits.

The Deutsche Bundesbank has, on occasion, conducted surveys about currency movements of non-banks, but the main source for estimating net movements is domestic banks. Nonetheless, the Bundesbank makes regular efforts to track the more obvious channels of currency movement. It receives, for instance, reports from several European central banks on the purchases and sales of deutsche mark notes by banks in those countries.11 The Bundesbank regards these figures as helpful in estimating travel expenditures in the German current account. It also estimates the exports of deutsche marks associated with travel in certain other countries. Special arrangements apply to the estimation of German expenditures in border areas with the Netherlands.

An annual survey is conducted of “cash carried by foreign workers,” but the Bundesbank does not believe that the results shed much light on transfers of currency for family-maintenance payments and saving purposes.

With regard to capital transactions, the Bundesbank assumes that cash is mainly exported to banks in Switzerland, Luxembourg, and Austria. Figures on both inward and outward shipments of notes through German banks are used to track returns of currency from these countries. Roughly speaking, the net movement of mark notes—exceeding the balance on cash transactions for travel purposes—is thought to stem from capital transactions. Adjustments are therefore made for the fact that part of the mark shipments received from Switzerland and Austria originate from other countries.


Like other major countries, Japan does not have a full set of statistics on currency movements. Under the Foreign Exchange and Foreign Trade Control Law, exports of yen currency up to £5 million are unconstrained; those exceeding £5 million, whether by banks or persons, are subject to authorization by the Ministry of Finance. Imports of yen do not require authorization when below £5 million. Foreign exchange banks are permitted to import above £5 million without prior authorization, but most other entities (except persons “who carry the Japanese currency with themselves”) are subject to authorization.

Although currency imports and exports exceeding the £5 million threshold are generally subject to authorization, few such requests are received by the Ministry of Finance, and these reports are not used for balance of payments purposes.

There are no statistical reporting requirements—for banks or others—on exports of yen currency. With regard to currency imports, banks are asked to report all imports (both above and below the £5 million threshold), but other entities are not subject to statistical reporting.

All returns of yen currency that are reported by the foreign exchange banks are attributed to expenses incurred by Japanese travelers and are factored into current account estimates of travel and tourism. Japanese authorities have no official view on amounts of yen that might be circulating outside the country.


The direct reporting system used in the Netherlands gives information on the inward and outward shipments of guilder bank notes between domestic banks and foreign commercial banks. Only these flows are used for balance of payments compilations. Outward shipments are classified as travel receipts for the Netherlands, because it is assumed that these guilders are used by foreigners to visit the Netherlands. Return shipments are in principle classified as travel expenditure by Dutch residents, assuming they result from the exchange of guilders for foreign currency by Dutch travelers abroad. There are two exceptions to the latter treatment. First, that part of the returned bank notes made up of the largest denomination, the f. 1,000 note, is regarded as capital exports by Netherlands residents, since this denomination is not generally used by travelers. Second, a part of the return shipments of smaller bank notes that comes from the home countries of migrant workers is recorded as private income transfers.

The Netherlands takes part in the exchange of information among nine European central banks on the sales and purchases of foreign bank notes. The figures on sales and purchases of guilder bank notes by foreign banks serve as a check on the recorded shipments and thereby as a check on travel expenditures and receipts. On the other hand, those figures also facilitate ad hoc studies on total cross-border movements of guilder bank notes. For that purpose, estimates are made for the sales and purchases of guilders by banks in countries that do not participate in the exchange of this information.

The Netherlands strives not only to keep up the exchange of data on the purchases and sales of bank notes but also to urge more countries to join in the effort so that better insight can be gained into cash payments. On the basis of the described model, plans are being developed by the European Community’s statistical office to establish a European system of exchange of information on sales and purchases of foreign currency.

United States

U.S. authorities were unable to respond formally to the Working Party’s request for details about statistical sources on cross-border movements of dollars, but it is generally acknowledged that large quantities of physical currency are used for transactions and for hoarding in foreign countries.12 Various studies indicate that the amount of U.S. currency in circulation—especially bills in denominations of $100 and more—is implausibly high for domestic usage alone, even if underground economic activity is taken into account. At the same time, large quantities of dollars openly circulate in many parts of the world, especially where there are unsettled economic or political conditions.

Recent information suggests that the United States has no reliable mechanism to track currency movements, although travelers and financial institutions confront reporting requirements on some imports and exports. At various times in the past, several Federal Reserve banks have compiled information on shipments of currency—involving the banks themselves or commercial banks in their regions—to and from countries such as Mexico, Panama, and several Caribbean areas, but this has not been a systematic or ongoing process. Certain U.S. commercial banks routinely send and receive large amounts of currency to and from banks in other countries, and a few of these domestic banks have sometimes provided information on currency shipments to the Federal Reserve. All such information has been irregular. On the whole, physical currency movements involving the U.S. dollar are conjectural. Although there have been discussions on the subject, no attempt has yet been made to put together an integrated reporting system, even for institutional transactors.

U.S. balance of payments estimates for both the current and capital accounts do not use any information about physical currency movements. Such fragmentary information as exists is gathered by the Federal Reserve or the U.S. Treasury and is not made available to balance of payments compilers elsewhere in the government. Currency movements are also not reported by banks on the standard reporting forms used for capital account compilations that are submitted to the Treasury.

    Other Resources Citing This Publication