Article

The Euro-Dollar Deposit Multiplier: A Portfolio Approach

Author(s):
International Monetary Fund. Research Dept.
Published Date:
January 1974
Share
  • ShareShare
Show Summary Details

I. Background of the Study

One of the most widely debated questions in the literature on the Euro-dollar market has been whether Euro-banks, 1 as a group, create deposits (or credit) to some multiple of their initial or primary deposit inflows. Almost without exception the approach used to examine this question has been the simple “fixed-coefficient” version of the money multiplier model that is so familiar to textbook discussions of the money supply process in a domestic context. According to this version of the multiplier, explicit portfolio choice by banks and the nonbank public is neglected, for it is assumed that the nonbank public holds a constant fraction of its deposit liabilities in the form of currency, and that banks hold a constant fraction of their deposit liabilities in the form of reserves. As a result, the extent to which a given primary or initial deposit inflow is multiplied in the system depends on the magnitude of the “leakage” ratios—that is, the currency and reserve ratios.

In view of the scarcity of data on Euro-banking operations (specifically, data on primary inflows and the reserve-holding behavior of Euro-banks) and information on the currency preferences of the non- Euro-bank public, it has been difficult to quantify the Euro-dollar multiplier empirically. As a result, debate on the deposit-creating potential of the Euro-banking system has proceeded largely on an intuitive level, with discussants offering subjective opinions as to the likely magnitude of the leakage ratios involved.

On this intuitive level, two schools of thought seem to have developed (Clendenning, 1971). On the one hand, Bell (1965) and, apparently, Friedman (1969) suggest that, in the absence of reserve requirements, the multiplier is potentially quite large. On the other hand, Klopstock (1968) and others suggest that, since the currency leakages from the Euro-banking system are likely to be very high (because the system can expect to recapture a relatively small fraction of its loans in the form of deposits), the multiplier is “probably in the approximate range of 0.50 to 0.90” (p. 8). Subsequently, Clendenning (1971) claims to have reconciled these opposing views. He argues that when central banks do not deposit part of their reserves in the Euro-dollar market the multiplier is likely to be quite small, but when there is redepositing by the central banks the effect is to reduce substantially the currency leakage, thereby raising the value of the multiplier.

Unfortunately, Clendenning’s reconciliation was quite shortlived. Two recent attempts by Makin (1972) and Lee (1973) to arrive at a figure for the multiplier diverge significantly in their estimates, even though both studies cover similar time periods and both claim to have taken account of central bank redepositing in the market. Makin estimated that “the long-run deposit expansion multiplier in the Eurodollar system” averaged 18.45 from the third quarter of 1964 to the fourth quarter of 1970, but Lee estimated that the multiplier averaged only 1.51 from the first quarter of 1963 to the fourth quarter of 1969.

It is somewhat surprising that few of the participants in this debate have objected to the fixed-coefficient version of the money-multiplier model as a conceptual framework for discussion of the Euro-dollar deposit multiplier. Moreover, those who have criticized this approach have not developed an alternative. Machlup (1972, pp. 9-11) for example, says that he prefers “to do without the concept” when analyzing “the growth of Euro-deposits, because the conditions under which this tool can be helpful are absent”; specifically, he states, “There are no well-specified reserve assets held by Euro-banks, nor do these banks obey any uniform and stable reserve requirements or reserve practices. Thus, we know neither the multiplicand nor the multiplier.” However, Machlup still seems to have used this model; on another occasion (1971, p. 10) he suggests that the share of “created deposits” may be over 50 per cent. Niehans (1971)2 points out the inappropriateness of the fixed leakage assumptions of the model, noting that, as the “new view” of banking theory has emphasized, these are strongly influenced by the structure of interest rates. In addition, Niehans raises a more fundamental question as to whether the multiplier can be interpreted as a measure of the “liquidity impact” of the Euro-banking system, as it is in the domestic context, but this case remains to be substantiated. 3 Finally, Masera (1972) also argues against the model’s assumptions of “fixed” or “stable” leakage ratios and an identifiable “exogenous” Euro-dollar base. Although Masera does not develop an alternative approach, he concludes (p. 182) that “preferences of both borrowers and lenders must be simultaneously taken into account to explain the existing volume of Euro-dollar assets and liabilities.”

To summarize, these criticisms seem to have centered on the fixed leakage assumptions of the model, which preclude explicit portfolio choice on the part of Euro-banks and the non-Euro-bank public. This situation suggests that there may be considerable advantage in formulating a portfolio model which explicitly recognizes that these leakage ratios are not necessarily fixed but rather are the outcome of portfolio choices by the market participants, and that, therefore, the value of these ratios is likely to vary with changes in interest rates.

This paper presents a portfolio approach to the question of the Euro-dollar deposit multiplier and estimates the value of the multiplier on this basis. Section II argues that while the multiplier model may have plausibility as a simple approximation to the money supply process in a regulated domestic system, it is inappropriate for an unregulated competitive situation such as that of the Euro-dollar market. Section III presents a portfolio model of the Euro-dollar market and uses it to derive an expression for the impact on the size of the market of an exogenous inflow of funds to it (that is, an expression for what is called the multiplier). It is shown that, in the absence of central bank redepositing in the market, this “multiplier” will be between zero and unity, because of what are called interest rate or portfolio leakages that occur as market participants adjust their portfolios in response to the interest rate changes that result from primary deposit inflows. Since the value of this multiplier, as derived, depends on the interest elasticities of the relevant functions of the model, Section IV gives estimates of these elasticities, which, together with an estimate of the volume of central banks redepositing in the market, permit estimation of the multiplier. The final section summarizes the main arguments and conclusions of the paper.

II. The Fixed Coefficient Multiplier Analysis and the Euro-Dollar Market

There is almost a complete parallel between the debate on the Eurodollar multiplier, as described above, and the debate of the late 1950s and early 1960s on the credit-creating ability of a group of nonbank financial intermediaries. In terms of this parallel, the approach of the present study to the question of the Euro-dollar multiplier follows that of the “new view” of banking theory offered by Gurley and Shaw (1960) and Tobin (1967) in the debate of the effects on monetary control of the existence of nonbank financial intermediaries.

The main conclusion of Tobin’s analysis (1967, pp. 10-11) was stated as follows:

Commercial banks do not possess, either individually or collectively, a widow’s cruse which guarantees that any expansion of assets will generate a corresponding expansion of deposit liabilities. Certainly this happy state of affairs would not exist in an unregulated competitive financial world. Marshall’s scissors of supply and demand apply to the “output” of the banking industry, no less than to other financial and nonfinancial industries. Reserve requirements and interest ceilings give the widow’s cruse myth somewhat greater plausibility. But even in these circumstances, the scale of bank deposits and assets is affected by depositor preferences and by the lending and investing opportunities available to banks.

With the exception of the references made by Niehans (1971) and Masera (1972), Tobin’s now classic argument seems to have been neglected in the discussions of the Euro-dollar market. It would seem useful, therefore, to present the essence of Tobin’s argument to demonstrate how inappropriate the fixed-coefficient multiplier model is for the Euro-dollar market.

Consider the market for loans as depicted in Chart 1. The demand for loans is represented as a negatively sloped curve labeled LL, while the supply of loan funds (deposits)4 is represented as a positively sloped curve labeled DD. If there are no regulations applicable to banking practices and if it is assumed, for simplicity, that banks choose to hold zero reserves, the equilibrium interest rate would be at the level where the supply of funds to banks (deposits) is equated to the demand for funds from banks (loans).5 Also, for simplicity, it is assumed that the margin between the deposit and loan rates is zero, although it is recognized that under competitive conditions this margin would be squeezed only to the normal level. It should be noted that the market situation just described closely approximates the situation in the Euro-dollar market.

Chart 1.Demand and Supply of Loans

Now suppose that there is some exogenous inflow of deposit funds to the banking system in the amount of E2E1. DD would shift to DD’, and the equilibrium level of deposits would shift from D1 to D2; the equilibrium interest rate would fall to r2. In other words, in order to lend out these new deposits, banks would have to lower the loan rate and, consequently, the deposit rate. This decline in the deposit rate would then discourage some depositors (E2 - D2). In these circumstances, the deposit multiplier may be represented as:

It can be seen that as long as the DD curve slopes upward and the LL curve slopes downward, the multiplier is less than unity. Only in the limiting cases (depicted in Chart 2) where either deposits are interest inelastic or where loan demand is infinitely elastic (or both) does the multiplier attain a value of unity. In this unregulated competitive situation, therefore, the simple multiplier analysis is completely inappropriate.

Chart 2.Demand and Supply of Loans Where the Multiplier Is Unity

However, the multiplier model gains more plausibility if domestic banking practices are subject to regulations such as deposit rate ceilings and reserve requirements. In these circumstances, Tobin (1967, p. 8) argues that the “marginal yield of bank loans and investments exceeds the marginal cost of deposits to the banking system.” Furthermore, he suggests that “It is the existence of this margin—not the monetary nature of bank liabilities—which makes it possible for the economics teacher to say that additional loans permitted by new reserves will generate their own deposits.”

Although not explicitly discussed by Tobin, two additional conditions are necessary for multiplication to occur, even when domestic banking practices are regulated. To demonstrate this point, consider the impostion of a deposit rate ceiling at a level, say rD which would be less than r1 in Chart 1. 6 In these circumstances, in order for multiplication to occur, not only must banks find it profitable to lend (the Tobin condition), but borrowers must find it profitable to borrow at the prevailing loan rate and depositors must be willing to acquire additional deposits at the ceiling interest rate (rD). Since competitive forces (in the form of new entry) are likely to operate to push the loan rate down toward the deposit ceiling, borrowers will find it profitable to take up loans. In addition, so long as this loan rate offers banks at least a “normal” margin above rD, the Tobin condition will be met. However, even if an exogenous increase in deposits is lent out, for multiplication to occur there must be, somewhere in the chain of transactions initiated by the borrowers’ outlays, depositors who wish to hold new deposits with the banking system. If deposit decisions are based solely on relative market interest rates, it is doubtful whether this latter condition would be met; depositors would already hold their desired volume of deposits and banks would be prevented from raising the deposit rate. However, in a domestic context where the banking system also offers a wide variety of services to its customers, there are certain nonpecuniary returns to be derived from holding bank deposits—in general terms, the benefits of direct access to the payments mechanism. 7 Under these special circumstances, a certain proportion of the loan proceeds would generally be returned to the banking system.

To expand on Tobin’s statement, it is the existence of (1) a difference between the marginal yield on banks’ loans and investments and the marginal cost of deposits, (2) an excess demand for loans at the prevailing loan rate, and (3) a demand for deposits at the deposit rate ceiling (because of the monetary nature of bank liabilities) that give the money- multiplier model some plausibility in a domestic banking system. However, in an unregulated competitive situation such as in the Euro-dollar market, where Euro-banks are merely intermediaries in the process of international payments, none of the above conditions is necessarily met.

III. The Model

The theoretical basis of the analysis which follows is a two-region 8 seven-sector general equilibrium model of the Euro-dollar market that explicitly recognizes portfolio choice by the banks and the nonbank public. This section summarizes briefly the main features of the model and derives an expression for the Euro-dollar deposit multiplier. A more complete statement of the model has been presented in a previous paper (Hewson and Sakakibara, 1973).

The basic assumption of this model with respect to Euro-banks is that they act as “pure” financial intermediaries in the sense that they have no independent influence on the equilibrium Euro-dollar rate. It is further assumed that, because of competitive forces, Euro-banks earn a constant margin between the Euro-dollar deposit and loan rates. Thus, the principal operating rule of Euro-banks may be viewed as that of adjusting the deposit rate so as to equate deposits simultaneously to loan demands.

These assumptions about the behavior of Euro-banks imply that the reserve holdings of Euro-banks are negligible. In view of the importance attached to the concept of Euro-bank reserves in the Euro-dollar literature, it seems worthwhile to present reasons for believing that such reserves are insignificant. Advocates of the fixed-coefficient multiplier approach to the Euro-dollar market find it necessary to argue that Euro-banks hold “precautionary reserves,” and to develop an empirical approximation for these precautionary reserves. Makin (1973), for example, suggests that Euro-banks hold “precautionary reserves” in the form of demand deposits with U.S. commercial banks, and as a proxy for these reserves he uses “demand deposits of private foreign banks at U.S. commercial banks, exclusive of claims on home offices of branch banks.” On this basis, Makin (p. 616) concludes that “the steady drop in the ratio of such reserves to net assets of Euro-banks from about 14 per cent in 1964 to about 5 per cent in 1970 is consistent with the effective economies of scale which may be realized in the management of precautionary balances as the volume of receipts and disbursements of a financial intermediary rises over time.” To the extent that financial intermediaries (banks and nonbanks) borrow short to lend long, and thereby run what the literature calls a “withdrawal risk” (Orr and Mellon, 1961), they may wish to hold precautionary reserves. However, as has been argued by Niehans (1971), Hayes (1971), 9 and others, Euro-banks tend to balance, on average, the maturity structures of their assets and liabilities, and therefore would seem to have no need to hold reserves as a precaution against a net withdrawal. Furthermore, to buffer the effects of possible defaults of outstanding loans, Euro-banks generally prefer to arrange stand-by lines of credit rather than to hold cash balances with U.S. commercial banks. The low reserve ratio for Eurobanks is also consistent with the hypothesis that their balances with U.S. banks are merely for transaction purposes, since such holdings are used for clearing Euro-banking operations. The declining trend observed in the reserve ratio would be consistent with economies of scale in the management of transaction balances.

Finally, it may be noted that the proxy for reserves suggested by Makin (1971, p. 390) should be interpreted as “maximum reserves”; as he states, “there is no guarantee that all foreign commercial banks holding demand deposits at United States commercial banks are in fact participants in the Euro-dollar market.” Also, the denominator of Makin’s “reserve ratio” tends to understate the total volume of Eurobank assets; 10 consequently, the true magnitude of the reserve ratio is probably lower than what he suggests. Thus, there is justification, from both a conceptual and an empirical point of view, for treating the reserves held by Euro-banks as negligible.

In this model, as originally specified, there were five endogenous interest rates—two U.S. rates, two “European” rates, and the Euro-dollar rate. 11 However, for present purposes, it is assumed that the U.S. and “European” rates are exogenous. If the policy-making authorities of these two regions regard interest rates as intermediate targets, and if they are successful in attaining these targets, it may not be very unrealistic to assume that these interest rates are exogenous.

Finally, to take account of central bank redepositing, it is assumed that “European” central banks hold a fixed ratio (cd) of their total foreign reserves (FOR) on deposit in the Euro-dollar market.12

Given the above assumptions as to the operations of Euro-banks, the equilibrium condition for the Euro-banking sector may be expressed as follows:

where Di = the demand for Euro-dollar deposits by country i

Li = the demand for Euro-dollar loans by country i.

These demand functions for both deposits and loans are assumed, according to standard portfolio theory, to be a function of net wealth, income, and rates of return. Since FOR (the level of foreign reserves of all “European” countries) is equal to the sum of deposits by the central banks of these countries in the Euro-dollar market (DEC.ED) and the accumulated overall U.S. balance of payments deficit (BUS), that is,

substituting equation (3) for FOR in equation (2) yields:

To derive an expression for the Euro-dollar deposit multiplier, it is necessary to examine the impact on the size of the Euro-dollar market of an exogenous shift (λ) of deposits from, for example, U.S. commercial banks to the Euro-dollar market. By totally differentiating equation (4) with respect to λ and rearranging the terms, the following expression for the impact of such a shift on the Euro-dollar interest rate is produced:

where rED = the Euro-dollar interest rate.

Defining the size (S) of the Euro-dollar market as the total volume of Euro-bank deposit liabilities, the Euro-dollar deposit multiplier (M=dsdλ) may be expressed as:

which, when substituting equation (5), produces:

It can be seen from equation (7) that, in the absence of central bank depositing in the Euro-dollar market (that is, cd = 0), the multiplier would be between zero and unity. The initial shift of funds from the United States to the Euro-dollar market would be offset by secondary effects occurring as a result of movements in relative interest rates—that is, what are called interest rate leakages. The shift of funds from the U.S. to the Euro-dollar market would lower the Euro-dollar rate relative to the U.S. and European rates, decreasing the relative attractiveness of Euro-dollar deposits. However, these changes in relative rates would also encourage additional loans, so that the total effect (initial and secondary) would always be positive and thus the multiplier would be between zero and unity. However, when central banks deposit part of their reserve holdings in the Euro-dollar market, the Euro-dollar deposit multiplier would be between zero and 11cd.

To arrive at a value for this multiplier, it is necessary to estimate the relevant partials of the deposit and loan functions with respect to the Euro-dollar interest rate. Instead of estimating the impact on the entire U.S. balance of payments of a change in the Euro-dollar rate, it was decided to neglect the terms involving ∂BUS/∂rED, recognizing that in doing so the value of the multiplier is in effect overestimated.

The basic forms of the demand functions for deposits and loans to be estimated in the next section are:

where r = vector of interest rates

Wi = wealth variable of country i

Yi = income variable of country i

X = vector of control variables

S = vector of speculative variables.

Since the authorities of many countries have introduced controls over flows of short-term capital—for example, the U.S. reserve requirements on U.S. commercial bank borrowing in the Euro-dollar market and the German reserve requirement on the foreign liabilities of German banks—and since such controls seem to have had considerable impact on the flow of funds in and out of Euro-banks, it was decided to include a vector of these control variables in the specification of these functions. Also, in view of the frequency and magnitude of currency speculation during the period under review, it was decided to include relevant speculative variables, such as the price of gold and the forward premiums on various currencies.

IV. Empirical Estimation of the Multiplier

The approach in this paper to the empirical estimation of the multiplier was to divide the world into three regions—the United States and Canada, Western Europe, 13 and the Rest of the World. Although such a degree of aggregation is not fully satisfactory, it was considered appropriate for present purposes. Euro-dollar deposit and loan functions were estimated for the three areas, using monthly data for the period January 1968 to December 1972.

Unfortunately, there was considerable difficulty in producing a satisfactory specification of the deposit and loan functions for the region called Rest of the World. As a result, instead of undertaking a more complete investigation of these functions at a more geographically disaggregated level (which is planned at a future date), it was decided to assume that the partials of these functions with respect to the Euro-dollar rate were such as to preserve the leakage ratio estimated for the United States and Europe. In other words, it was assumed that:

where the subscripts USC, WE and ROW refer to the United States and Canada, Western Europe, and the Rest of the World, respectively.

The actual variables used in estimating the equations for the United States and Canada and for Western Europe, with the relevant sources, were as follows:

deposit-loan variables14

LUSC=dollar claims of U. K. banks on U.S. and Canadian residents and banks (in millions of U.S. dollars)
LWE=dollar claims of the U. K. banks on Western European residents and banks (in millions of U.S. dollars)
DUSC=dollar liabilities of U. K. banks to U.S. and Canadian residents and banks (in millions of U.S. dollars)
DWE=dollar liabilities of U. K. banks to Western European residents and banks (in millions of U.S. dollars).

interest rate variables15

rED=3-month Euro-dollar deposit rate
rED1=Euro-dollar call rate 16
rUS=90-day treasury bill rate
rUS1=Federal funds rate
rUS2=Certificate of deposit (CD) issue rate
rWE=German call rate (covered)
rWE1=U. K. 90-day treasury bill rate (covered)

constraint variables17

YWE=income variable for Western Europe, defined as production index multiplied by consumer price index in all European countries belonging to the Organization for Economic Cooperation and Development except the United Kingdom
WWE=European wealth variable, defined as money supply (money and quasi-money) of Western Europe
YUSC=U.S. and Canadian income variable, defined as commercial and industrial loans of U.S. commercial banks and Canadian business loans
WUSC=U.S. and Canadian (bank) wealth variable, approximated by the sum of U.S. and Canadian domestic credit
WUSC1=U.S. and Canadian (private) wealth variable, approximated by the sum of U.S. and Canadian money and quasi-money.

control variables18

QNL=Regulation Q variable, defined as U.S. Federal funds rate/3-month CD issue rate
RQ1=legal (nominal) U.S. reserve requirement on Euro-dollar borrowing by U.S. commercial banks, defined as 0 = 68:1 to 69:8; 10 = 69:9 to 70:12; 20 = 71:1 to 72:12
RQ2=a dummy variable to capture additional restraint at times when U.S. commercial banks were increasing their indebtedness to Euro-banks, defined as 0 = 68:1 to 71:8; 20 = 71:9 to 72:12
DFR=a dummy variable to represent the period for which borrowing by U.S. commercial banks was in excess of the reserve-free base, defined for the May 1969 commitment period (roughly $10 billion) as 1 = 69:10 ~ 70:8 (0, otherwise)
DCR=a dummy variable to capture the effect of the decision by European central banks to reduce their depositing in the Euro-dollar market, defined as 1 = 71:10 to 72:12 (0, otherwise).

speculative variables19

SPF=forward premium on French franc
SPG=forward premium on deutsche mark
SPUK=forward premium on sterling
S71=a dummy variable to capture speculative activity from May 1971 to the U.S. announcement of suspension of convertibility of the dollar into gold in August 1971 and similar activity in December 1971 at the time of the Smithsonian Agreement, defined as 1 = 71:5 to 71:8 and 71:12 (0, otherwise).
A variant of this speculative variable was designated S71II, defined as 1 = 71:5 and 71:8 (0, otherwise)
S695=a dummy variable to represent speculative activity with respect to the deutsche mark in May 1969, defined as 1 = 69:5 (0, otherwise).

It may be noted that all variables involving values are in billions of U.S. dollars and that all interest rates are in per cent a year unless otherwise specified.

The following results were obtained for the deposit and loan equations relating to Western Europe and to the United States and Canada when estimates were made by two-stage least squares.

equations for Western Europe

Loans

Deposits

equations for the united states and canada

Loans

Deposits

The general performance of these equations was quite satisfactory, in view of the degree of aggregation and the difficulties encountered in developing satisfactory wealth and income proxies. Perhaps the most surprising feature was the relative stability of the coefficient on the Eurodollar rate in alternative specifications of these equations. In general, the performance of the loan equations was more satisfactory than that of the deposit equations. The deposit equation for the United States and Canada was the least satisfactory because of the high degree of multi-collinearity between the wealth and income proxies used, and this factor led to the exclusion of the wealth proxy in the final version of this equation. 21 Experimentation also revealed that the specification of this equation could be improved significantly by the incorporation of lags. Still, although there may be room for improvement in these results, as deposit and loan equations, per se, they were considered adequate for the present purposes of calculating a multiplier.

Given the estimated partials of each of these deposit and loan variables with respect to the Euro-dollar interest rate, in order to estimate the multiplier it is also necessary to determine the value of cd—the ratio of central bank deposits in the Euro-dollar market to their total reserves. There are, of course, numerous ways of estimating this ratio. It was decided that perhaps the simplest approximation was the ratio of the change in the sum of “Identified official holdings of Euro-dollars” plus “Unidentified Euro-currencies and residual” to the change in total foreign reserves of all member countries of the International Monetary Fund (Table 1). This marginal ratio was considered more appropriate for the purpose of arriving at a figure for the multiplier, 22 although it should be noted that this estimate may very well exceed the true value of cd, because the item “Unidentified Euro-currencies and residual” includes Euro-currency deposits other than Euro-dollars, plus a statistical residual. 23 Thus, in this respect also, the estimate of the multiplier presented here should be considered as a maximum.

Table 1Estimate of the Ratio of the Change in Central Bank Deposits in the Euro-Dollar Market to the Change in Total Foreign Reserves, 1968-72(In billions of SDRs)
YearA

Change in Euro-

Dollar Deposits
B

Change in

Foreign Reserve
C

Ratio (cd)

A/B
19681.92.00.950
19692.21.71.294
19707.416.50.448
19715.429.40.184
19729.823.10.424
Mean 0.660
Source: International Monetary Fund, Annual Report, 1973, Table 10, p. 36. Column A represents Δ Identified official holdings of Euro-dollars (line 4a (ii)) plus Unidentified Euro-currencies and residual (line 4g); column B represents the Total reserve change shown in Table 10.
Source: International Monetary Fund, Annual Report, 1973, Table 10, p. 36. Column A represents Δ Identified official holdings of Euro-dollars (line 4a (ii)) plus Unidentified Euro-currencies and residual (line 4g); column B represents the Total reserve change shown in Table 10.

With the estimated partials and the figure for cd, the Euro-dollar deposit multiplier may be expressed as:

Hence, in the period 1968–72, the Euro-dollar multiplier was slightly greater than unity—that is, total deposits would have increased by slightly more than the magnitude of the primary deposit inflow during this period.24 However, as noted above, this estimate of the multiplier should be interpreted as a maximum, particularly in view of the likely overestimate of cd.

V. Conclusions

This paper has presented a portfolio approach to the question of the Euro-dollar deposit multiplier. The argument parallels that of Tobin in the debate on the credit-creating ability of a group of nonbank financial intermediaries, a topic that received a great deal of attention in the monetary literature of the late 1950s and early 1960s. In line with Tobin’s main thesis, it is argued that while the fixed-coefficient multiplier model may under certain conditions serve as a useful approximation to the money supply process in a domestic banking system, where there are often important regulations on banking practices, it is not appropriate for unregulated competitive markets such as the Euro-dollar market. More specifically, it has been shown that in the absence of regulations on banking practices the multiplier would in fact be a divisor, for its value would fall between zero and unity.

A general equilibrium portfolio model of the Euro-dollar market was used to derive an expression for the Euro-dollar multiplier. It is shown, that, unlike the fixed-coefficient multiplier, which has a minimum value of unity, the Euro-dollar multiplier for a primary inflow of deposits—from, for example, the United States—would be between zero and unity in the absence of central bank redepositing in the market. The initial shift of deposits would be offset by what is called net interest rate leakages, which would occur as wealth owners adjust their portfolios in response to the changes in relative interest rates that follow the primary deposit inflow. The precise value of the multiplier would depend on the interest elasticities of the relevant functions in the model. Finally, it is argued that when central banks hold part of their reserves on deposit in the Euro-dollar market the value of the multiplier would fall between zero and 1/(1 - cd), where cd is the ratio of the central banks’ holdings of Euro-dollar deposits to their total foreign reserves.

To arrive at a figure for the Euro-dollar multiplier, deposit and loan functions are estimated for the United States and Canada and for Western Europe, using monthly data for the period 1968-72. From the estimation of the responsiveness of loans and deposits to changes in the Euro-dollar interest rate and from an estimate of the central bank Eurodollar deposit ratio (cd), it is concluded that the Euro-dollar deposit multiplier was slightly greater than unity (1.4) in the period 1968-72. However, this estimate for the multiplier should be considered as a maximum, because of the assumptions made and because of the possibility that the figure for central bank redepositing in the Euro-dollar market is on the high side.

REFERENCES

Mr. Hewson, an economist in the North American Division of the Western Hemisphere Department, is a graduate of the University of Sydney, Australia, the University of Saskatchewan, Canada, and the Johns Hopkins University.

Mr. Sakakibara, an economist in the Exchange Rate Practices Division of the Exchange and Trade Relations Department, is a graduate of the University of Tokyo, Japan, and the University of Michigan. He is at present on a leave of absence from the Ministry of Finance, Japan.

For the purposes of this paper, the term “Euro-banks” refers to “Euro-dollar banks.”

Niehans offered a partial portfolio approach by demonstrating how the fixed coefficient multiplier is affected when the “redeposit” ratio, as well as loans and deposits, are a function of “the interest rate.”

In a domestic banking system, banks tend to borrow short (demand deposits) and to lend long (for example, personal loans and mortgages) and in doing so have a direct impact on the liquidity of the nonbank public sector, for as a whole the latter lends short and borrows long. However, if, as has been suggested (see footnote 9), Euro-banks tend to “balance” or “match” the maturity composition of their loans and deposits, the Euro-banking system will have no net impact on the liquidity position of the non-Euro-bank sector. In these circumstances the Euro-banking system may be more appropriately thought of as “distributing” existing liquidity rather than “creating” new liquidity.

As developed below, it is implicitly assumed here that the intermediaries in question operate so as to equate the volume of loans and deposits; as a result, it is possible to identify the supply of loan funds with the demand for deposits.

In Tobin’s words, “In a world without reserve requirements the preferences of depositors, as well as those of borrowers, would be very relevant in determining the volume of bank deposits. The volume of assets and liabilities of every intermediary would be determined in a competitive equilibrium, where the rate of interest charged borrowers by each kind of institution just balances at the margin the rate of interest paid its creditors” (Tobin, 1967, p. 7).

Without detracting from the analysis, the previous assumption that banks hold zero reserves is maintained.

A similar point is made by Masera (1972, pp. 151-52).

The two regions are designated the United States and “Europe.” Although it is recognized that “Europe,” for the purposes of the model, includes all non-U.S. countries, this designation is consistent with the general Euro-dollar literature.

Hayes (p. 257), reporting the results of his survey of the Euro-dollar lending policies of branch offices of some U.S. banks, concludes that: “Most banks attempted to match maturities of the purchased funds with the rate commitment period on their loans, but a few were willing to assume a more venturesome position.” Also, irregular surveys of the Bank of England on the maturity structure of U.K.-based Euro-banks tend to confirm this conclusion, although these data may be deficient for this purpose to the extent that banks report claims and liabilities classified by “loan commitment” rather than “rate commitment” periods.

For the total volume of Euro-bank assets (liabilities), Makin uses estimates by the Bank for International Settlements, which relate to the operations of banks in eight European countries plus Canada and Japan and which are generally acknowledged to cover only part, but probably a large part, of Euro-banking operations. In comparison, Makin’s estimate of reserve balances relates to all private foreign banks that hold deposits in U.S. banks.

In the theoretical model, the endogenous interest rates for the United States and “Europe” were bank loan and bond rates.

It should be recognized, of course, that the holdings of Euro-dollar deposits by the “European” central banks could have been expressed as a function of the relevant interest rates in a manner similar to the other deposit functions of the model. However, as it is not clear that central banks act as portfolio choice makers in the same way as other market participants, it was decided to assume that they hold a fixed ratio of Euro-dollar deposits to total foreign reserves, although it is recognized that this assumption tends to understate the “interest leakage” and therefore to overstate the multiplier.

Western Europe is defined here as in the basic source of the data (Bank of England, Quarterly Bulletin; see, for example, March 1973, Table 23, pp. 104-105), but excludes Switzerland because the data for Switzerland represent transactions by the Bank for International Settlements.

Euro-dollar deposit and loan data were taken from the Bank of England, Quarterly Bulletin, rather than from the Bank for International Settlements, because the Bank of England reported on a monthly basis and had the desired geographical disaggregation.

The source for rED, rED1, and rUS2 is the Morgan Guaranty Trust Company of New York, World Financial Markets; the source for rUS and rUS1 is the Federal Reserve System, Board of Governors, Federal Reserve Bulletin; and the source for rWE and rWE1 is the International Monetary Fund, International Financial Statistics.

The Euro-dollar call rate was used in the loan equation for the United States and Canada, since these loans are primarily loans from the foreign branches of banks to their head offices, the bulk of which were transacted at call (Bank of England, Quarterly Bulletin (March 1970), p. 37). For a detailed discussion of U.S. commercial bank borrowing in the Euro-dollar market in the period under consideration, see Hewson and Sakakibara (1974).

Needless to say, there was considerable difficulty in developing appropriate wealth and income variables on a monthly basis. In general the wealth variable was money supply alone, since short-term security holdings were not readily available. For the United States and Canada it was decided to use the total of commercial and industrial loans in order to achieve consistency with similar work (Hewson and Sakakibara, 1974). In addition, it was felt that this variable represented an appropriate proxy for changes in the aggregate level of transactions. Unfortunately, it was not possible to define YWE in a manner similar to YUSC because of data deficiencies. However, it may be noted that the estimated partials of the deposit and loan equations were generally stable, irrespective of the specification of the constraint variables.

Sources for the constraint variables are YWE—production index and weights, from the Organization for Economic Cooperation and Development, Main Economic Indicators; consumer price index, from the International Monetary Fund, International Financial Statistics; WWE, WUSC, and WUSC1—International Monetary Fund, International Financial Statistics; and YUSC—loans of U.S. commercial banks, from Federal Reserve System, Board of Governors, Federal Reserve Bulletin, and Canadian business loans from Bank of Canada, Bank of Canada Review.

For a detailed discussion of the U.S. control variables, see Hewson and Sakakibara (1974). The sources for the control variable QNL are: Federal Reserve System, Board of Governors, Federal Reserve Bulletin, and Morgan Guaranty Trust Company of New York, World Financial Markets.

Source for speculative variables SPF, SPG, and SPUK is the International Monetary Fund, International Financial Statistics.

20

This equation largely represents U.S. bank borrowing from foreign branches and therefore closely resembles an equation that was developed on another occasion to explain Euro-dollar borrowing by U.S. banks. Consequently, the wealth variable here is essentially a loanable funds constraint and therefore could be expected to have a negative sign. See Hewson and Sakakibara (1974).

The equation for DUSC including the wealth variable was:

As an alternative estimate of cd, an average ratio could be used—that is, a ratio of the central banks’ holdings of Euro-dollar deposits to total foreign reserves. On this basis, the estimate of cd would be 0.11. However, apart from being conceptually more appropriate, the marginal ratio appeared to perform in a manner consistent with the little that is known about central bank depositing in the Euro-dollar market during this period; specifically, the sharp drop in the ratio in 1971 would seem to reflect the impact of the agreement among European central banks to limit redepositing (and to reduce it if possible), while the jump in the ratio in 1972 would be consistent with the noted increase in redepositing by some less developed countries and Japan.

More specifically, this item includes “asymmetries arising from the fact that data on U.S. and U.K. currency liabilities are more comprehensive than data on official foreign exchange holdings shown in International Financial Statistics” (International Monetary Fund, Annual Report, 1973, Table 10, footnote 8, p. 36).

It may be noted that using the alternative average estimate of cd would produce a value of 0.54 for the multiplier.

Other Resources Citing This Publication