The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
The R e v i e w is published monthly by the Research Department of the Federal Reserve Rank of St. Louis. Single-copy subscriptions are available to the public free of charge. Mail requests for subscriptions, back issues, or address changes to: Research Department, Federal Reserve Rank of St. Louis, P.O. Rox 442, St. Louis, Missouri 63166. Articles herein may be reprinted provided the source is credited. Please provide the Rank’s Re search Department with a copy of reprinted material. Inflation and Personal Saving: An Update CLAUDIA R. CAMPBELL and JEAN M. LOVATI F R O M 1965 to 1974, a decade of rapid inflation, households saved relatively more of their current in come than they had in the previous decade of gen erally stable prices. Following the 1974 recession, however, the saving response of U.S. households to inflation appeared to undergo a major change. De spite the higher average rates of inflation from 197578, the proportion of current income saved fell below that of the previous decade (Table 1). Earlier studies of inflation and household saving generally concluded that U.S. households respond to price level increases by cutting back on borrowing and spending, thereby increasing their saving.1 Most of these studies encompass the period prior to 1975, before saving rates plunged to post-World War II lows. If the positive relationship between saving and inflation no longer holds, a rising rate of inflation in the future is no guarantee of higher average rates of household saving. This development could have an adverse effect on future economic growth since lower average rates of household saving tend to restrict the future supply of funds used for investment in plant and equipment. This paper updates earlier investigations of the re lationship between inflation and saving to include the years, 1975 through 1978. In particular, it examines the long-run saving response to inflation in order to determine whether the observed impact of inflation on saving is merely a temporary phenomenon. 1Recent work on this subject indicates that the positive re sponse of saving to inflation is partially the result of uncer tainty created by high and variable rates of inflation. See Paul Wachtel, “Inflation, Uncertainty and Saving Behavior Since the M id-1950s,” Explorations in Econom ic Research (Fall 1977), pp. 558-78. Coupled with uncertainty, house hold saving has been affected by the failure of corporate stocks to provide an adequate hedge against inflation. This is discussed in Philip Cagan and Robert Lipsey, The Finan cial E ffects o f Inflation, (Cambridge, Mass.: Ballinger Pub lishing Company, 1978). Another study suggests that house holds downgrade the quality of their purchases in response to a rise in the rate of price increases, producing the observed positive saving response to inflation. See Susan Burch and Diane Wemeke, “The Stock of Consumer Durables, Inflation and Personal Saving Decisions,” The R eview o f Econom ics and Statistics (May 1975), pp. 141-54. The long-run effect of inflation on household saving was estimated previously in a 1977 study by Paul Wachtel. Wachtel found that the uncertainty gener ated by inflation helped to explain the persistent rise in saving with price level increases in the 1/1955 to III/1974 period. Using Wachtel’s model with a differ ent measure of inflation uncertainty yields a long-run response of saving to inflation uncertainty that is posi tive but statistically insignificant over the 1/1955IV/1978 sample period. However, the composition of household assets — the forms of saving — is altered Table 1 Personal Saving and Inflation Year Personal Saving Rate1 Inflation Rate2 1955-64 5 .8 % 1.4% 1965-74 6.9 4.7 1975-78 5.9 7.2 1Personal saving/disposable personal income, average o f annual rates. 2Annual rate of change in Consumer Price Index. SOURCE: Survey of Current Business. by changes in the rate of inflation. These results are consistent with traditional economic theory which in dicates that inflation has no significant impact on sav ing in the long run except, under certain circum stances, to produce readjustment in the components of household wealth. WACHTEL’S SAVING EQUATION Wachtel assumed that the long-run effect of infla tion on saving resulted from uncertainty created by higher and more variable inflation rates.2 Because households are unable to forecast prices accurately, -Some recent studies suggest that countries with higher aver age inflation rates experience more variation in the rate of inflation and that the variability of inflation contributes to the welfare costs of inflation. See A. Okun, “The Mirage of Steady Inflation,” Brookings Papers on Econom ic Activity (2 :1 9 7 1 ), pp. 485-98, D. E. Logue, and T. D. Willet, “A Note on the Relation between the Rate and Variability of Inflation,” E co nomica (May 1976); pp. 151-58, and E. Foster, “The Vari ability of Inflation,” R eview o f Econom ics and Statistics, (August 1978), pp. 346-50. Page 3 FEDERAL RESERVE BANK OF ST. LOUIS AUGUST Table 2 Derivation o f N atio n al Income and Product Accounts Saving 1978 (Billions of Dollars) (1 ) PERSONAL INCOME less Personal Tax and Nontax Payments (includes net payments to social security) $1,717.4 $259.0 (2 ) DISPOSABLE PERSONAL INCOME (includes imputed rental income from owner-occupied housing) 1,458.4 less PERSONAL OUTLAYS (3 ) Personal Consumption Expenditures (includes consumer durables and mobile homes and imputed rental payments on owner-occupied housing) (4 ) Interest Paid by Consumers to Business (5 ) Personal Net Transfer Payments to Foreigners 1,350.8 34.8 0.8 PERSONAL SAVING ( 2 - ( 3 + 4 + 5) ) (includes net investment in housing) SOURCE: Survey o f Current Business. they become uncertain about future prices and real income and, as a result, save more. Wachtel asserted that other effects of inflation on saving, such as money illusion, intertemporal substitutions, and indirect wealth and interest rate effects, have no lasting in fluence on saving behavior. In order to test this hypothesis, Wachtel used the stock adjustment demand function developed earlier by Houthakker and Taylor.3 According to the specifi cations of this model, real saving per household (q) is a linear function of the stock of accumulated real saving (s), real income per household (y), and infla tion uncertainty (X ): (1 ) q = oc + |3s + yy + ih X In addition, the stock of past real saving (s) is as sumed to depreciate at a constant rate, 5, per year. Thus, the change in the stock of real saving (s) over a given time (t) can be represented by: (2 ) s ( t ) = q ( t ) — 6 s (t) and used to transform the structural equation into one containing only flow variables. In its reduced form, Wachtel’s estimated equation was: (3 ) q t = A0 + Aiqt-i + AiAy + A^yt-i + A.AX + AsXti Because the structural parameters (3 and 5 are overidentified, nx (the inflation uncertainty coeffi3H. S. Houthakker and Lester D. Taylor, Consumer D emand in the United States, 1929-1970: Analysis and Projections, (Cambridge, Mass., Harvard University Press, 1966). Page 4 72.0 1979 cient) is not readily obtainable from the reduced-form equation.4 Nevertheless, the long-run effect of inflation uncer tainty can be calculated from the coeffi cient on the lagged variable Xt-i, where the long-run effect of uncertainty (0 X) equals A5/( 1-A ,). Wachtel estimated equation (3) using both National Income and Product Ac counts (NIA) and flow of funds (FO F) accounts saving per household, deflated by the personal consumption expendi tures deflator. Disposable personal in come, similarly deflated, was used as the income variable. Inflation uncertainty was measured by the average variance in households’ assessment of future price increases as obtained from Survey Re search Center surveys. The Saving Data NIA saving is basically the residual after de ducting current outlays for goods, services, and interest payments from current disposable per sonal income (Table 2). Disposable personal income consists of the after-tax receipts of households from wages and salaries, interest income, rent, dividends, and net transfer payments. Capital gains are not in cluded. The rental value of owner-occupied housing is imputed and added to both personal disposable income and personal consumption expenditures. Since purchases of new housing are excluded from personal consumption expenditures, net investment in housing by households is included as a component of personal saving. Nonconsumed income, held in the form of currency, demand deposits, bonds, stocks, or pension funds, is incorporated into net financial investment. Thus, the major assets into which NIA saving flows are net housing investment and net financial investment. The measure of household saving in the FOF ac counts is also a residual, in this case, from the meas ured transactions among all other sectors of the economy (Table 3). In addition to net financial in vestment and net housing investment, FOF house hold saving includes capital gains dividends, addi tions to government pension funds, and net durable goods investment. 4Two distinct values for (5 and 8 are generated from the re duced-form coefficients. For a technique to deal with this problem, see Ibid., pp. 21-25. AUGUST FEDERAL RESERVE BANK OF ST. LOUIS FOF saving during the post-war pe riod has been consistently higher than NIA saving even after adjustments for these compositional differences. Meas urement errors in both series account for some portion of the discrepancy. In ad dition, capital gains from the sale of real estate and other durable assets (art and antiques, for example) to the business sector may add to the observed differ ence in the two measures. These trans actions would amplify the discrepancy during periods of rising inflation. Table 3 Derivation of Flow o f Funds Accounts Saving (Year end, total flows) 1978 (Billions of Dollars) (1 ) WachteVs Results Wachtel obtained a significant positive response to inflation uncertainty for real NIA saving per household over the sample period, I/1955-III/1974. He found that a 1 percent rise (fall) in inflation uncertainty resulted in a $69 increase (decrease) per household in real NIA saving. The inflation uncer tainty coefficients in the reduced-form equation, however, were statistically insignificant using the similarly-deflated FOF saving data. 1979 $245.9 INCREASE IN FINANCIAL ASSETS $ 18.2 Demand Deposits & Currency Time and Savings Accounts 105.3 64.9 Credit Market Instruments (Government Securities, Corporate & Foreign Bonds, Mortgages, Commercial Paper, Money Market Funds) Investment Company Shares -1 .0 Other Corporate Equities -5 .2 Life Insurance and Pension Fund Reserves 77.8 Net Investment in Noncorporate Business -23.1 Security Credit 8. Miscellaneous Assets 9.0 GROSS INVESTMENT IN TANGIBLE ASSETS Nonfarm Homes (includes mobile homes) Consumer Durables Nonprofit Plant & Equipment $298.2 92.0 200.3 5.9 181.0 less CAPITAL CONSUMPTION ALLOWANCES Nonfarm Homes (includes mobile homes) Consumer Durables Nonprofit Plant & Equipment 32.8 142.8 5.4 117.2 (2 ) NET INVESTMENT IN TANGIBLE ASSETS (3 ) NET INCREASE IN LIABILITIES Mortgages Consumer Credit Bank Loans & Other Loans Security Credit, Trade Debt, & Miscellaneous 166.4 104.7 50.6 7.2 3.9 When FOF saving was disaggregated NET SAVING ( 1 + 2 - 3 ) 196.7 into its components — net increases in financial assets, net increases in liabili ties, and net increases in tangible assets SOURCE: Board of Governors of the Federal Reserve System. (mainly housing and durable goods) — Wachtel discovered that increased un certainty about inflation reduced net increases in lia INFLATION AND SAVING: 1955-1978 bilities and tangible assets. The reduced-form coeffi In an inflationary economy, uncertainty about fu cients on inflation uncertainty for net increases in ture prices and real income results from the unex financial assets were negative, but statistically insig pected variation of prices around the generally antici nificant. Inflation uncertainty had a positive and sig pated rate of inflation. Thus, in this analysis, the nificant effect, however, on net financial investment uncertainty variable is approximated by using an esti (net increases in financial assets less net increases in mate of unanticipated changes in the rate of inflation. liabilities). Wachtel’s equation was respecified to include meas ures of unanticipated (X ) and anticipated (Z) infla When Wachtel used actual price changes as the tion. The reduced-form equation in this analysis is: inflation variable, he found that inflation exerted a positive and statistically significant influence on both NIA and FOF saving. He concluded that inflation and the uncertainty it creates made households re luctant to acquire additional debt in order to pur chase tangible assets. As a result, real saving per household rose. (4 ) q, = Ao + Aiqt-i + A2Ay + Aay t-i + AiAX -f- A^Xt-i + A«AZ + AjZt-j The coefficients on anticipated inflation (Z) and on the estimated long-run effect (0 Z) are not expected to be statistically different from zero. This result is Page 5 FEDERAL RESERVE BANK OF ST. LOUIS AUGUST 1979 Table 4 Regression Results for N IA Savin g * 1/1955 q, = -1 0 7 .1 8 (.8 8 ) RJ = + .91 .85q-i + (1 2 .6 0 ) D.W. = .5 5 A y + (7 .0 2 ) 2.28 = - 111/1974 ,02y-i + (1 .1 9 ) 75 .1 7 (1-45) 2 3 .8 8 A X + (1 .7 6 ) 0. = 11.23X-1 — 5 4 .8 0 A Z + (2 .3 0 ) (1 .1 2 ) .4 1 Z-, (.0 4 ) 2.76 (.0 4 ) 1/1955 - IV/1978 qt = 76.45 (.6 3 ) R2 = .87 + -93q-i + (1 5 .5 0 ) D.W. = 2.24 .6 5 A y — .01y-i + (9 .8 4 ) (.4 4 ) 0, = 190.58 (1 .1 0 ) 4 0 .0 9 A X + (3 .3 6 ) 0* = 13.76X-i + (3 .4 8 ) .8 2 A Z + (.0 2 ) 4.39Z-, (.4 3 ) 60.80 (.3 9 ) *t-statistics are reported in parentheses. consistent with the explanation that only unantici pated inflation produces the uncertainty effect ob tained by Wachtel. Furthermore, economic theory suggests that fully anticipated inflation has no lasting effect on saving behavior.5 Thus, unanticipated infla tion (X ) is expected to be the only source of a positive long-run relationship between household sav ing and inflation.6 tel) and I/1955-IV/1978. A significantly different ef fect of unanticipated inflation on saving behavior before and after III/1974 would suggest that the household saving response to inflation had, in fact, changed. Since empirical evidence has shown that there is a direct relationship between lagged money growth and the fundamental rate of inflation, a 20-quarter rate of change in the narrowlv-defined money supply, Ml, was initially used as a proxy for anticipated infla tion.7 The difference between a four-quarter rate of change in the Consumer Price Index (C PI), a wellpublicized indicator of price change, and the money supply variable above was used to measure unantici pated inflation. All other data used to estimate equa tion (4) are the same as those previously used in Wachtel’s study. Consistent with the analysis above, anticipated in flation (Z) and its long-run effect (0 Z) were found to have no significant impact on saving as measured in the NIA in either sample period (Table 4). Fur thermore, the lagged variable of unanticipated infla tion had significant positive effects on NIA saving in both periods. The equation was estimated over two sample periods: I/1955-III/1974 (the period used by Wach5A wealth loss could result even with fully anticipated infla tion because of the increased costs of using money as a me dium of exchange and a store of value. This effect on real wealth could have an impact on saving and consumption. For a discussion of the costs of anticipated inflation, see John A. Tatom, “The Welfare Costs of Inflation,” this Review (November 1976), pp. 9-22. °The positive relationship between unanticipated inflation and saving is supported by the results of several studies. See, for example, F. Thomas Juster and Paul Wachtel, “Inflation and the Consumer,” Brookings Papers on Econom ic Activity (1 :1 9 7 2 ), pp. 71-121 and Joseph Bisignano, “The Effect of Inflation on Savings Behavior,” Federal Reserve Bank of San Francisco Economic Review (December 1975), pp. 22-26. 7See Denis S. Kamosky, “The Link Between Money and Prices — 1971-76,” this Review (June 1978), pp. 17-23. Inflation and NIA Saving When the long-run effect of unanticipated inflation ( 0X) is examined for the 1955-74 period, a 1 per cent rise (fall) in the rate of unanticipated inflation produced a $75 rise (fall) in real saving per house hold. Over the longer sample period, this effect becomes more than twice as great: A 1 percent rise (fall) in the rate of unanticipated inflation resulted in about a $192 rise (fall) in real saving per house hold. In neither sample period, however, was the long-run effect significanthj different from zero at the 95 percent level of confidence. To determine whether these findings depend upon the disaggregation of inflation into anticipated and unanticipated price changes, the saving relationship was reestimated using lagged and first differences of the actual rate of inflation, measured by a four-quar ter rate of change in the CPI. The initial results pre vailed: The reduced-form coefficients showed a sig- FEDERAL RESERVE BANK OF ST. LOUIS AUGUST 1979 Table 5 Regression Results for FO F Savin g * 1/1955 - 111/1974 q, = - 5 4 8 .2 4 (1 .7 7 ) Rs = + .85 .4 1 q , + (3 .9 9 ) D.W. = 2.41 .6 8 A y + (3 .9 0 ) = .1 2y-, + (2 .8 2 ) -2 7 .0 6 (1 .2 6 ) 2 2 .7 6 A X — 15.95X-, — (.7 1 ) (1 .3 8 ) 0, = 16 .0 8A Z + (.1 4 ) 5.46Z-, (.2 3 ) 9.26 (.2 3 ) 1/1955 - IV / 1978 q, = -2 8 6 .6 2 (.9 8 ) R* = .81 + .38q-! + (4 .0 0 ) D.W. = 2.36 .6 2 A y + (4 .4 4 ) 0. = .09y~, + (2 .3 8 ) -1 3 .5 3 ( 94 ) 2 1 .7 0 A X — 8.42X-, + (.8 2 ) (.9 1 ) 0. = 74.51 A Z + (.7 0 ) 23.86Z-, (1 .0 6 ) 38.44 (1 .0 5 ) *t-statistics are reported in parentheses. nificant positive relationship between saving and in flation, but this relationship was statistically insignifi cant in the long run. In summary, the existence of a significant long-run positive effect of inflation on NIA saving is not supported by the results whether a measure of inflation uncertainty or the actual infla tion rate is used. Inflation and F O F Saving When using FOF saving, Wachtel obtained a posi tive effect of inflation on saving only when the actual inflation rate was substituted for his measure of in flation uncertainty. The analysis of FOF data in this study, however, reveals no such relationship. Further more, neither unanticipated inflation nor anticipated inflation have a significant impact on FOF saving in either sample period (Table 5). Wachtel’s results showing a positive effect of actual inflation on FOF saving may be due to the estimates of depreciation of tangible assets used in his study. When Wachtel published his results, the revised estimates that are incorporated in the FOF data used in this update were not available. Although FOF saving is not significantly affected by either inflation or inflation uncertainty, its com ponents could be altered by adjustments across var ious household asset categories. Adjustments that re duce purchases of durable goods relative to other assets would appear as increased NIA saving with rising inflation. This occurs because durable goods purchases are classified as consumption expenditures in the NIA. To investigate this aspect of the impact of inflation, the saving model was estimated using, as dependent variables, the three components of FOF saving: net acquisitions of financial assets, net increases in finan cial liabilities, and net investment in tangible assets. Tangible asset acquisitions were disaggregated into net housing and net durable goods investment. As indicated in Table 6, the reduced-form coeffi cient on the lagged variable for unanticipated inflation is statistically significant and negative in the net dura ble goods investment equation over both sample pe riods. The long-run effect, which is not statistically significant from 1955-74, is significant in the longer sample period. The estimate of the long-run effect suggests that an increase (decrease) of 1 percent in the rate of unanticipated inflation induced a reduc tion (expansion) in real net durable goods investment of $45 per household in the 1955-78 sample period. This result is consistent with a rise in NIA saving in response to a rise in u nanticipated inflation. Net housing investment appears to be strongly af fected by both anticipated and unanticipated infla tion in the reduced-form equation. The long-run ef fects of anticipated and unanticipated inflation on housing investment, however, are not statistically significant in either sample period. When net durable goods and housing investment are aggregated into net increases in tangible assets, a significant long-run relationship with both antici pated and unanticipated inflation is obtained for the 1955-74 sample period. A 1 percent increase (de crease) in the rate of unanticipated inflation produces a $63 per household decrease (increase) in real tangi ble asset acquisitions. At the same time, the effect of anticipated inflation on saving is almost twice as strong, but positive: A 1 percent rise (decline) in the Page 7 FEDERAL RESERVE BANK OF ST. LOUIS rate of anticipated inflation produces a $107 per household rise (decline) in real net tangible asset investments. Over the longer sample period, the positive effect of anticipated inflation on household investment in tangible assets dissipates.8 Only unanticipated inflation continues to exert an influence on net tangible asset investment that is statis tically significant in the long run. In the period, 1/1955 to IV/1978, a decline (increase) in the rate of unanticipated inflation by 1 percent induced a rise (decrease) in net purchases of tangible assets of $102 per household, nearly double the impact of the 1/1955 to III/1974 period. AUGUST Table 6 T a n g ib le A sset Com ponent of FO F Savin g * Net Increase in Housing Investment Net Increase in Durable Goods Net Increase in Tangible Assets 1955-74 1955-78 1955-74 1955-78 1955-74 1955-78 224.51 (3 .0 4 ) -8 .6 7 (.1 6 ) -5 2 .0 7 (.5 2 ) -205.71 (2 .2 7 ) 349.58 (2 .9 7 ) 57.29 (.4 9 ) .87 (1 7 .5 4 ) 1.01 (3 1 .1 7 ) .69 (8 .8 0 ) .61 (7 .8 6 ) .73 (1 0 .5 7 ) .81 (1 2 .7 4 ) Ay -.0 0 2 (.0 7 ) .01 (.4 8 ) .19 (3 .3 0 ) .12 (2 .8 4 ) .17 (2 .4 3 ) .12 (2 .0 6 ) yt-j -.0 2 (3 .1 3 ) -.0 0 2 (.3 1 ) .01 (.8 1 ) .03 (2 .6 0 ) -.0 3 (1 .9 2 ) (.1 6 ) Ax -7 .0 5 (1 2 5 ) -15.41 (3 .3 0 ) 0.56 (.0 5 ) -1 .6 8 (.1 8 ) -4 .9 9 (3 5 ) -1 8 .4 3 (1 3 7 ) X ,-a -4 .5 6 (2 .3 2 ) -4 .3 6 (2 .9 3 ) -1 0 .9 0 (2 .5 6 ) -1 7 .7 2 (5 .2 0 ) -1 7 .1 0 (3 .1 7 ) -1 9 .1 2 (4 .2 7 ) 40.01 (2 .1 0 ) 21.53 (1 .2 2 ) 55.41 (1 .5 3 ) 20.17 (.6 4 ) 83.50 (1 .8 5 ) 25.95 (.5 8 ) 15.17 (3 .3 7 ) 5.86 (1 .5 0 ) 10.65 (1 .4 1 ) 3.18 (.4 7 ) 29.15 (3 .0 3 ) 13.11 (1-36) Constant Lagged Dependent Variable As separate components, net increases A z in financial assets and liabilities gener ally are not affected by anticipated or Z t-i unanticipated inflation (Table 7). Nei ther the reduced-form results nor the R2 long-run relationship between the infla D.W. tion variables and the financial asset and liability components of FOF saving is statistically significant in either sam 0 . ple period. When net increases in financial assets and liabilities are com bined (called net financial investment), *t-statistics are however, a statistically significant posi tive long-run response to unanticipated inflation results, but only over the 1955-78 sample period. Assuming an anticipated rate of inflation of about 6 percent from 1974 to 1976, these findings suggest that the decline in the rate of inflation from 11 per cent in 1974 to 5.6 percent in 1976 resulted in a re duction in real net financial investment of approxi mately $280 per household, or $20 billion, and a net increase in real durable goods investment of about $243 per household, or $17 billion from 1975 to 1977. Therefore, the effects of this reduction in unantici pated inflation would have contributed to the ob served decline in NIA saving in that period. sThe trend growth of money — anticipated inflation — tended to stabilize around a 6 percent annual rate after 1972, pro viding no further positive impetus to tangible asset acquisi tion. This is consistent with a one-time shift from money to goods resulting from the transition to a higher expected rate of inflation. Page 8 1979 .002 .95 .95 .91 .90 .89 .88 1.33 1.36 1.98 2.02 1.53 1.63 35.08 (1 .5 0 ) 322.96 (1-03) -3 4 .6 0 (1-77) -4 4 .8 9 (2 .8 6 ) -6 3 .3 3 (1 .9 8 ) -102.36 (1 .9 6 ) 1 18.52 (1 7 2 ) 434.07 (.1 5 ) 33.81 (1 3 4 ) 8.06 (.4 8 ) 107.96 (2 .1 7 ) 70.18 (1 .1 6 ) reported in parentheses. ALTERNATIVE SPECIFICATIONS Several alternative measures of anticipated and un anticipated inflation were used in reestimating equa tion (4). First, an anticipated inflation series was generated using forecasts of future price changes from the Livingston survey.9 These semiannual fore casts were interpolated to create a quarterly data series and the difference between the expected rate of price change generated by these forecasts and the actual inflation rate was used as the measure of un anticipated inflation. Under this specification, both unanticipated and anticipated inflation showed a pos itive long-run effect on NIA saving in the 1955-74 sample period, which contradicts the hypothesis that anticipated inflation has no long-run effect on saving. UJ. A. Livingston, “Business Outlook,” The Philadelphia Record, June 1954-December 1978. FEDERAL. RESERVE BANK OF ST. LOUIS AUGUST unanticipated inflation on NIA saving was statistically significant at the 90 per cent level over both sample periods. Table 7 Finan cial Asset and Liability Com ponents of FO F Savin g * Net Increase in Financial Assets 1955-74 1955-78 Net Increase in Liabilities 1955-74 1955-78 Net Financial Investment 1955-74 1955-78 - 9 6 4 .4 9 --929.06 (2 .2 7 ) (2 .3 0 ) 137.57 (6 3 ) -55.51 (2 6 ) .47 (4 .0 1 ) .46 (4 .9 3 ) .77 (9 .4 7 ) .93 (1 7 .3 5 ) .29 (2 .6 3 ) .50 (5 .2 1 ) Ay .60 (2 .7 3 ) .68 (3 .9 5 ) .11 (7 7 ) .07 (.6 5 ) .51 (2 .8 7 ) .49 (3 .1 4 ) y t» .16 (2 .8 1 ) .15 (2 .7 6 ) -.01 (-27) .01 (.2 4 ) .18 (3 .9 5 ) .06 (1 .4 1 ) Ax - 9 .4 9 (.2 3 ) -2 0 .4 0 (-63) -4 1 .8 8 (1 .6 2 ) -7 3 .9 8 (3 .5 0 ) -2 .7 7 (.0 9 ) .98 (.0 3 ) Constant Lagged Dependent Variable 1979 -1296.01 -3 3 8 .3 6 (1 .0 2 ) (3 .6 7 ) Although these alternative measures of inflation anticipations yield positive longrun relationships between inflation and NIA saving, they show no effect of infla tion on FOF saving. Wachtel encoun tered this same dichotomy in his analysis — the results are sensitive to the saving measure used. Conclusions As an update to previous work on the relationship between inflation and sav 16.34 -1 6 .1 4 -1 3 .8 0 26.06 -8 .0 2 X.-x -2 0 .4 3 ing, this study finds no conclusive evi (1 .9 3 ) (1 .4 6 ) (2 .6 7 ) (1 .6 4 ) (1 .3 8 ) (.7 1 ) dence that inflation has a positive long66.24 128.95 39.89 -7 6 .0 0 Az 64.41 61.43 run effect on saving. FOF saving, which (.5 7 ) (.5 0 ) (.6 7 ) (1 .4 7 ) (.4 6 ) (.4 7 ) represents net additions to household -42.51 3.91 21.32 10.49 10.92 z ,- ! - 7 .4 8 wealth, is not affected by any measure (-16) (1 .1 7 ) (1 .7 ) (.2 5 ) (.3 9 ) (.6 1 ) of inflation or inflationary anticipations .78 .71 .80 .86 .84 .86 used in the analysis. NIA saving, a nar 2.26 2.17 2.60 2.49 D.W. 2.28 2.30 rower measure, is not affected by actual inflation nor by unanticipated inflation 51.81 -6 9 .3 3 -2 0 6 .5 9 23.15 -1 4 .8 9 -3 6 .2 9 <t>* (2 .2 7 ) (1 .8 4 ) (1 .0 3 ) (1 .4 1 ) (1 .4 8 ) (.6 9 ) derived from the difference between ac tual prices and lagged money growth. -6 0 .2 2 7.76 91.58 157.03 -1 3 .2 8 20.28 (1 .5 4 ) (.5 5 ) (-16) (1 .0 5 ) (.2 5 ) (3 9 ) The use of Livingston survey and Scad ding data, however, produce a positive *t-statistics are reported in parentheses. relationship between unanticipated infla tion and NIA saving. Both Livingston and Scadding data are sensitive to the saving meas A second measure of anticipated inflation was ob ure used. tained using a series developed in a recent study by Scadding.10 His series takes into account the way Unanticipated inflation had a significant long-run in which people revise their estimates of the undereffect on the components of saving over the 1955lying inflation rate when actual prices turn out dif 78 sample period. Rising rates of unanticipated in ferently from expected. The Scadding data produced flation reduced durable goods investment and in a positive and significant relationship between NIA creased net financial investment. The observed posi saving and unanticipated inflation in the reduced-form tive relationship between inflation and NIA saving is equation. In addition, the positive long-run effect of due, in large part, to the negative effect of unantici 10John L. Scadding, “Estimating the Underlying Inflation pated inflation on durable goods purchases, which Rate,” Federal Reserve Bank of San Francisco Econom ic Review (Spring 1979), pp. 7-18. are classified as consumption expenditures in the NIA. Page 9 Does Eurodollar Borrowing Improve the Dollar’s Exchange Value? DAVID H. RESLER “In a further move to improve the international position of the dollar, the Board of Governors on August 28, 1978, announced a change in reserve require ments to make it more attractive for member banks to borrow funds in the Eurodollar market. . . . The new action involves a reduction from 4 percent to zero in the reserve requirement on foreign borrowings of m em ber banks, pri marily Eurodollars, from their foreign branches and other foreign banks.”1 Federal Reserve f c j ARLY in 1978, the dollar began to decline sharply in value in the foreign exchange markets. This dra matic decline, shown in Chart 1, precipitated several Federal Reserve policy actions, culminating in last November’s comprehensive dollar rescue effort under taken in cooperation with the Treasury. This action consisted of a combination of dollar-supporting efforts including an expansion of both direct foreign exchange intervention and swap arrangements, and an an nounced increase in the discount rate. While these ac tions seem to have successfully abated the dollar s decline, the desired improvement in the dollar’s inter national position has been modest. The action taken last November was the most dramatic of several actions taken to support the dollar.2 The quotation above identifies another such dollar-supporting move. By removing the reserve re quirements against Eurodollar borrowing, the Fed in tended to encourage the use of this source of funds in order to generate a net increase in the demand for 1Federal Reserve Bulletin (September 1978), p. 777. The reg ulations affected by this policy action are Regulations D and M. Regulation D specifies the reserve requirements member banks must meet for various liability classifications. Regulation M governs the Federal Reserve’s treatment of foreign branch banks. It is important to note that the computation of the reserve requirement against “Eurodollar borrowings” was ac tually on net balances due to foreign branches. -In addition to the action indicated in the quotation, the Fed eral Reserve has increased the discount rate several times during the past year. For an assessment of the effect of these discount rate changes on the exchange rate, see Douglas R. Mudd, “Did Discount Rate Changes Affect the Foreign Ex change Value of the Dollar During 1978?” this Review (April 1979), pp. 20-26. 10 Digitized for Page FRASER B u l l e t in , September 1978. the dollar and thereby increase its foreign exchange value. This paper examines analytically the conditions under which removal of these reserve requirements would improve the dollar’s foreign exchange value. Available data relating to Eurodollar borrowing offer little evidence that this policy initiative has fulfilled its intentions. THE EURODOLLAR MARKET: AN OVERVIEW Eurodollars are simply dollar-denominated deposits placed in a bank outside the United States. Anyone may own Eurodollars and these owners may reside in a foreign country or in the United States. They may As this article was published, the Federal Re serve announced a comprehensive change in pol icy that includes Eurodollar borrowing. Eurodol lar borrowing will be included in the calculation of “managed liabilities.” Increases in the total of these managed liabilities above a base level will be subject to an 8 percent marginal reserve re quirement. This action, however, does not re move the differential reserve requirement be tween large CDs and Eurodollar borrowing. In fact, the new policy action may further stimulate the substitution of Eurodollars for large CDs that this paper examines. AUGUST FEDERAL RESERVE BANK OF ST. LOUIS for a substantial volume of Eurodollar activity. C h a rt 1 Weighted A v e ra g e Foreign Currency V a lu e of the D o lla r 11 Index 1979 Index M arch 1973=100 Regardless of their loca tion, Eurodollar banks ( Eurobanks) perform an intermediary function simi lar to that of other banks. They issue liabilities (that is, they accept deposits) which they use to acquire earning assets, primarily loans to customers and fi nancial investments such as bonds, commercial paper, and so on. As with other intermediaries, Eurobanks’ profits are the differential between earnings received on their assets and the costs of their liabilities. Eurodollar deposits dif fer from domestic U.S. bank deposits in one often over looked but very important respect: Generally, liabili ties of Eurobanks are not “checkable deposits.” Euro 197 8 197 9 dollar depositors cannot S o u rc e : F e d e ra l R e se rve B u lletin write drafts on their depos L i The c o u n trie s in c lu d e d in the w e ig h te d - a v e ra g e fo re ig n in te re s t ra te a n d e x c h a n g e ra te se rie s a re B e lg iu m , C a n a d a , F ra n c e , G e rm a n y , Ita ly , J a p a n , the N e th e rla n d s , S w e d e n , S w itz e rla n d , and the U n ited K in g d o m . The its. In other words, Euro w e ig h ts a n d fo rm u la used in c o n stru ctin g th e se se rie s a re from " In d e x of the W e ig h te d - A v e ra g e E x c h a n g e dollars are not “money” in V a lu e of the U.S. D o lla r: R e v is io n ,” F e d e ra l R e s e rv e B ulletin (A u g u st 1978). the same sense that demand deposits and U.S. currency be private citizens, nonfinancial corporations, other are money. Eurodollars are, instead, most comparable banks or financial intermediaries, or official institutions to various “near-monies” like large denomination cer of foreign governments. tificates of deposit ( CDs) .3 Motives for holding Eurodollars are equally diverse. The primary motive, however, is that Eurodollars are short-term dollar-denominated assets which pay an at tractive yield. Those extensively engaged in interna tional trade view the market as especially convenient. With a large volume of trade ultimately conducted in dollars, the Eurodollar market provides a relatively high yielding outlet for dollar balances that obviates much of the risk and transactions costs associated with converting them into a foreign asset or with in vesting them directly in U.S. capital markets. Despite the “Eurodollar” designation, the market is not exclusively located in Europe. Though the largest part of the market’s activity takes place in London, the rest of Europe and such diverse locations as Sing apore, the Bahamas, and the Cayman Islands account THE RELATIONSHIP BETWEEN THE U.S. BANKING SYSTEM AND EURODOLLARS There are two important links between the Eurodol lar market and the U.S. banking system. First, and most important to this discussion, many Eurodollar banks are branches or subsidiaries of U.S. commercial banks. This means that U.S. parent banks have an aux•!The degree of liquidity of Eurodollars varies with the term to maturity of the deposit. The maturity of Eurodollar de posits ranges from overnight to, more typically, 30 days or more. The extent to which Eurodollars add to the world’s liquid balances and thereby represent a source of world infla tion is perhaps the most controversial aspect of the market. For a recent discussion of this problem, see Adrian W. Throop, “Eurobanking and World Inflation,” Voice of the Federal Reserve Bank o f Dallas (August 1979), pp. 8-23. Page 11 FEDERAL RESERVE BANK OF ST. LOUIS AUGUST Table 1 * Effective Cost of Bank Liabilities ID Eurodollar Borrowing (2 ) Certificates of Deposit 1977 August (3 ) Differenc 6 .5 6 % 6 .2 9 % .2 7 % September 6.83 6.57 .26 October 7.44 6.64 .80 November 7.39 7.11 .28 December 7.42 7.15 .27 1978 January 7.63 7.37 .26 February 7.58 7.33 .25 March 7.57 7.29 .28 April 7.69 7.48 .21 May 8.15 7.89 .26 June 8.68 8.32 .36 July 8.88 8.63 .25 August 8.83 8.56 .27 - .0 4 September 9.12 9.16 October 10.12 10.02 .10 November 11.51 11.65 - .1 4 December 11.62 11.63 -.01 -.2 5 1979 January 11.16 11.41 February 10.79 11.07 -.2 8 March 10.64 11.01 - .3 7 April 10.60 10.92 -.3 2 May 10.75 11.03 -.2 8 June 10.52 10.82 - .3 0 July 10.87 10.99 -.12 August 11.53 11.62 -.0 9 ♦Calculations are based on the reported daily average yield for each type o f liability and the applicable reserve requirements. SOURCE: Federal Reserve Bulletin and the Federal Reserve Bank o f St. Louis. iliary source of funds for their domestic operations. Specifically, a U.S. parent bank may use the special relationship with its branch to obtain liabilities (that is, to borrow from its branch) when domestic sources of funds become constrained. This occurred, for exam ple, in 1968-69 when restrictive monetary policy, cou pled with Regulation Q deposit ceilings, dried up domestic sources of funds, thereby encouraging U.S. banks to utilize credit lines with their foreign branches. At other times, this relationship between parent and branch has resulted in a net flow of funds from the parent to the branch. This was, in fact, typical of the market from 1975 until early this year. The second important link between the U.S. and Eurodollar banking systems centers on the Eurobanks’ demand for reserve funds. As with any financial inter Page 12 1979 mediary, a Eurobank maintains a stock of readily accessible funds (reserves) to meet day-to-day trans actions and clearing requirements. One of the most striking and controversial features of the Eurodollar system is that, unlike domestic banks, the level of re serves held by Eurobanks is not regulated. This does not mean, however, that Eurobanks hold no reserves. Profit-maximizing considerations determine the opti mal level of precautionary reserves for Eurobanks. The special characteristics of this market result in very low levels of reserves relative to total deposit volume.4 Generally, Eurobanks’ deposits with U.S. banks serve as precautionary reserves for the Eurodollar market. THE EFFECT OF EURODOLLARS ON THE DOLLAR S FOREIGN EXCHANGE VALUE As previously noted, U.S. banks often obtain liabil ities from the Eurodollar market by borrowing from their own branches or from other Eurobanks. Like other forms of foreign borrowing, this practice in creases U.S. liabilities to foreigners and lowers (raises) the short-term international capital account deficit (surplus). Falling deficits or rising surpluses generally indicate an increasing demand for dollars which in turn implies a rising value of the dollar in foreign exchange mar kets.5 This is the connection between Eurodollar bor rowing and the foreign exchange rate that the August 28, 1978 policy action attempted to exploit. The connection between the net liquidity deficit and the foreign exchange rate, however, is more compli cated when Eurodollars are borrowed because such borrowing need not result in a currency conversion. To see this point more clearly, consider the following example: When a U.S. resident borrows from a for eigner, he usually issues a dollar-denominated IOU. 4For both a theoretical and empirical discussion of optimal Eurodollar reserves, see John H. Makin, “Identifying a Reserve Base for the Euro-Dollar System,” Journal o f Finance (June 1973), pp. 609-17 and David H. Resler, A Study o f the Euro-Dollar Market: Its Origin and Interaction with U.S. Monetary Policy, unpublished Ph.D. dissertation (The Ohio State University, 1977). 5It is important to note that increased borrowing by U.S. banks tends to improve (lower) the U.S. balance-of-payments defi cit as measured on a net liquidity basis. It need not and probably does not, however, exert any impact on the “official settlements” balance. This balance is based only on official governmental settlements. In the case above, no intergovern mental transactions are involved. For a detailed discussion of this distinction, see Donald S. Kemp, “Balance of Pay ments Concepts — What Do They Really Mean?” this Review (July 1975), pp. 14-23. FEDERAL RESERVE BANK OF ST. LOUIS To purchase this debt instrument, the foreigner first acquires dollars through the foreign exchange market, thereby increasing the demand for dollars. If, how ever, the foreigner already possesses dollar-denomi nated assets such as Eurodollars, the transaction does not involve the foreign exchange market even though the U.S. net liquidity deficit falls. Thus, Eurodollar borrowing need not increase the demand for dollars in the foreign exchange markets. But, can Eurodollar borrowing produce a net in crease in the demand for dollars? The answer is a qualified yes. Elimination of the reserve requirements against Eurodollar borrowing effectively reduces the cost of this source of funds. This tends to increase the total demand for Eurodollar borrowings, thereby bid ding up the Eurodollar loan (and deposit) rate. If the higher relative yield on Eurodollars produces an increase in the general level of U.S. interest rates, it may induce a substitution of dollars for other curren cies. When this occurs, the demand for dollars and the dollar exchange rate will increase. On the other hand, the higher yield on Eurodollars may induce only a substitution among dollar assets. Owners of domestic dollar CDs or U.S. Treasury bills, for in stance, may switch to Eurodollars. The extent to which Eurodollars are substituted for other dollardenominated assets, then, is the key factor in evalu ating the effect this policy action has on the foreign exchange value of the dollar. EVALUATING THE CHANGE IN RESERVE REQUIREMENTS When a bank meets a reserve requirement, the cost of its funds includes both the interest expense and the earnings foregone on the idle balances (reserves) it must hold. The elimination of reserve requirements against Eurodollar borrowing lowers the effective cost of these funds to U.S. banks.6 When making portfolio decisions about their Hability structure, banks com pare the effective cost of funds for alternative liabil ities. Thus, in assessing the relative attractiveness of Eurodollar borrowings, the effective cost of these funds must be compared with alternative liabilities. Eurodollar borrowings can be considered a substi tute for large denomination ($100,000 or more) CDs issued by U.S. banks. The effective cost of funds for Specifically, the effective cost ( C j ) of any liability ( j ) can be written as: Ci = ij/( 1 - r j ) , where ij and rj are the interest rate and required reserve ratio for the liability. AUGUST 1979 these two liabilities and the differences between them over the last two years are reported in Table l.7 While a modest cost advantage in favor of Eurodollar bor rowing emerged temporarily in September 1978, a per sistent cost advantage in favor of Eurodollar borrow ing has prevailed only since November 1978 when the Federal Reserve increased the reserve require ment against large CDs from 6 percent to 8 percent. The cost differential fell dramatically following this action.8 Data presented in Table 1 show that the elimination of reserve requirements against Eurodollar borrowing did little by itself to encourage a preferential shift by U.S. banks toward borrowing Eurodollars. The Fed’s action of November 1, raising reserve requirements on CDs, however, appears to have eventually encour aged Eurodollar borrowing. A persistent effective cost differential in favor of Eurodollar borrowing began to emerge in November 1978. Since U.S. banks’ cost of funds had become higher in the domestic CD market than in the Euro dollar market, it is reasonable to expect that U.S. banks would have attempted to reduce their CD holdings relative to borrowing in the Eurodollar market. One way for banks to replace CDs with Eurodollars without endangering well-established customer rela tionships is to encourage their depositors to place CDs directly with the banks’ foreign branches. U.S. banks could then borrow from these branches at a lower ef fective cost. This transaction produces offsetting short term dollar flows with no net change in the demand for dollars. The Federal Reserve recognized this po tential in its August 28 announcement when it . . reemphasized the importance of compliance by U.S. banks with its previous requests not to solicit or to 7Data in column 1 of Table 1 tend to overstate the effective cost of Eurodollar borrowings. The reason is that, as noted in footnote 1, the relevant reserve requirement applies to net balances due to foreign branches. Since the aggregate net position of the banking system was negative preceding the policy revision, only a small number of banks could have been net borrowers from the market. It is only for these banks that the calculated effective cost of Eurodollar funds is appropriate. 8A brief digression on the characteristics of this cost differen tial should prove illuminating. In constructing Table 1, the Eurodollar borrowing rate is the three-month interbank loan rate as published by the Federal Reserve. This reported rate represents the Eurobank’s opportunity cost of lending to a U.S. (i.e., its parent) bank. A U.S. bank may be willing to borrow from its Eurobank branch even when the cost differ ential favors the CD market. This may occur if earnings and costs of the parent and branch are differentially treated under the relevant tax laws for the two banks. Thus, even a small positive cost differential may be consistent with a domestic bank’s preference for Eurodollar borrowing. Page 13 AUGUST FEDERAL RESERVE BANK OF ST. LOUIS 1979 duce any changes in the foreign branches’ liabilities to U.S. nonbanks. Table 2 Large CDs and Eurodollar Deposits of U .S. Residents (Billions of Dollars, Not Seasonally Adjusted) Lorge Denomination CDs U.S. Nonbanks’ Eurodollar Deposits at Foreign Branches of U.S. Banks January $76.4 N.A. February 76.9 N.A. March 80.2 N.A. April 81.4 N.A. May 84.6 $20.1 June 86.3 21.6 July 87.3 23.0 August 88.0 24.3 September 90.3 21.8 October 90.8 24.7 November 96.4 25.9 December 99.5 25.0 101.1 30.5 February 99.6 31.5 March 97.5 33.0 April 92.6 33.5 May 88.9 34.8 June 84.4 35.3 July 84.0 N.A. August 86.4 N.A. September 89.8 N.A. January tCE : Federal Reserve Bu lletin and Board o f Governors o f the Federal Reserve System. encourage deposits by U.S. residents at their foreign branches .. ,”9 Data suggest that very little of this direct transfer has occurred (Table 2, column 2). Eurodollar deposits of U.S. nonbank residents have increased steadily since May 1978 but have shown no dramatically sharper rise when large CDs have fallen. These data, however, probably understate the value of CDs that U.S. resi dents have replaced with Eurodollar deposits. Instead of transferring deposits to branches of U.S. banks, U.S. residents mav have established Eurodollar ac counts with foreign banks. These banks could then sell Eurodollar CDs in a secondary market to U.S. foreign branches. The net effect of these transactions is the same as when U.S. residents deposit funds directly with the branches. The important difference, however, is that the transactions outlined here would not pro9Federal Reserve Bulletin (September 1978), p. 778. Page 14 Any empirical assessment of Eurodollar borrowing by U.S. banks must begin with a word of caution: Since Eurodollar borrowings are not directly reported by U.S. banks, available data provide only approxima tions of the actual borrowing volume. In October of this year,7 the Federal Reserve Board initiated reporting of new data that provide useful approximations for Eurodollar borrowing.10 These data record net balances due to directly related for eign institutions. The data measure the net direction of the flow of funds between the U.S. banking system and the Eurodollar market. Eurodollar borrowing by U.S. banks represents only part of the net flow of funds and may be offset by loans from U.S. banks to Eurobanks. Nevertheless, changes in net balances due to directly related foreign institutions represent a rea sonable proxy for changes in Eurodollar borrowing. For instance, an increase of $1 billion in the “net balances” is interpreted as an increase in Eurodollar borrowing of $1 billion. Data for this measure of Euro dollar borrowing are given in Table 3. J Data reported in Table 3 reveal that Eurodollar bor rowing by U.S. banks changed very little in the four months immediately following the change in reserve requirements. At the same time, the data indicate that Eurodollar borrowing has increased sharply since Jan uary 1979. Column 1 shows that, in January 1979, the net flow of dollars from U.S. banks to their own branches began to reverse itself. The net outflow fell substantiallv J each month and finally J became a net inflow from Eurobanks in May 1979. This flow reversal is attributable to the extensive Eurodollar borrowing by U.S. banks. The data reveal that U.S. banks have increased their Eurodollar borrowing from their own branches by $19 billion since the beginning of the year. Over the same period, total net balances due to related foreign institutions increased by more than $26 billion. Both data are essentially consistent with the incentive pattern reported in Table 1. The data sug gest that the increase in Eurodollar borrowing this year can be attributed less to the Fed’s elimination of reserve requirements against Eurodollar borrowing than to the Fed’s increase in reserve requirements against large CDs. 10In the past, most researchers measured borrowing with gross claims (in dollars) of foreign branch banks on their parent U.S. bank. This measured only Eurodollar borrowings from their own branches but did not record borrowing from other Eurobanks nor did it account for borrowing by nonmember U.S. banks. Nevertheless, these data were the only useful proxies for Eurodollar borrowing. AUGUST FEDERAL RESERVE BANK OF ST. LOUIS U.S. BANKS HAVE REDUCED CDs IN FAVOR OF EURODOLLARS The overall success of the August policy action in terms of its effect on the dollar’s exchange value de pends on whether this Eurodollar borrowing is substi tuted for more conventional liabilities, such as large denomination CDs. If this has occurred, there is little reason to believe that the increased borrowing by U.S. banks has produced a net increase in the demand for dollars in foreign exchange markets. To evaluate the extent of this liability substitution ( “round-tripping”), the behavior of large CDs over this period must be examined. Data on this liability (Table 2) reveal a substantial reduction in the total amount of CDs out standing since the beginning of the year. From the January peak of $101.1 billion, CDs fell to $84.0 bil lion in July, a drop which accompanies the emergence of a relative cost disadvantage for CDs (reported in Table 1). It is interesting to note that, as CDs fell by about $17 billion from January to July, liabilities of U.S. banks to their foreign branches rose by $17.4 bil lion. The general pattern in this data suggests an ap parent switching of Eurodollars and large CDs.11 In August and September, data on the volume of CDs and preliminary data on Eurodollar borrowing both show an increase in response to strong U.S. credit demands. This suggests that, since the cost advantage in favor of Eurodollar borrowing has now virtually disappeared, both liabilities will grow in response to overall credit demand. IMPLICATIONS OF “ROUND-TRIPPING” FOR MONETARY CONTROL So far, the discussion has ignored any effect this substitution of Eurodollar borrowing for domestic CDs may have on the U.S. money supply. Since the primary advantage to U.S. banks from borrowing Eurodollars is that these liabilities are not subject to reserve requirements, the substitution of Eurodollar borrowing for CDs “liberates” reserves. For example, suppose a U.S. bank allows its CDs to decline by $1 million and offsets this outflow by borrowing $1 mil lion from its foreign branch. The bank’s total liability position is unchanged by the transaction. The bank’s asset side, however, shows that the transaction has 11The data on Eurodollar borrowing is not sufficiently accurate to warrant the conclusion that this switchover has been complete, since it seems inappropriate to argue that only Eurodollars have replaced CDs. More extensive use by do mestic money managers of other short-term financial instru ments including repurchase agreements and commercial paper has probably also diminished their use of CDs. 1979 Table 3 Net Balances Due to Directly Related Foreign Institutions (Billions of Dollars) (1 ) Domestically Chartered Banks (2 ) ForeignRelated Institutions (3 ) Total $-2.1 Changes in Euro dollar Borrow ing* $ -1 3 .6 $1 1.5 February -1 3 .2 1 1.0 -2 .2 $-0.1 March - 1 4 .7 1 1.2 -3 .5 -1 .3 January April -11.1 10.8 -0 .3 3.2 May -1 1 .7 12.0 0.3 0.6 June -1 1 .8 13.3 1.5 1.2 July - 9.5 12.6 3.1 1.6 August -1 0 .5 12.9 2.4 -0 .7 September -1 0 .3 14.8 4.5 2.1 October - 9.9 17.0 7.1 2.6 November - 9.8 18.0 8.2 1.1 December - 1 0 .7 17.0 6.3 -1 .9 January -10.1 16.4 6.3 0.0 February - 6.3 18.3 12.0 5.7 March - 4.5 20.8 16.3 4.3 April - 1.9 20.8 18.9 2.6 May 2.5 20.6 23.2 4.3 June 5.8 21.7 27.5 4.3 July 6.3 22.8 29.1 1.6 August 8.9 23.8 32.7 3.6 * Equals Changes in Column 3. SOURCE: Federal Reserve Bulletin and the Board o f Governors of the Federal Reserve System. generated an additional $.08 million in excess reserves which it can then lend. Lending these newly gener ated excess reserves increases the U.S. money supply unless the increase in excess reserves is offset by Fed eral Reserve open market operations. Of course, such an increase in the money supply could prove counterproductive to the Fed’s ob jective of improving the dollar’s foreign exchange value. If the faster growth of money leads to a higher expected rate of inflation in the United States and, hence, lowers the value of the dollar in the future, the dollar’s current foreign exchange value will also fall as speculators attempt to minimize the anticipated exchange rate loss. Unless Federal Reserve open market operations off set this increase in reserves, there will be a multiple expansion of the money supply equal to the money multiplier times the newly liberated reserves. Under this assumption, the reduction in CDs of $17 billion Page 15 from January to July (if offset by an equal increase in Eurodollar borrowing) would have resulted in about a $3.4 billion increase in M l.12 This amounts to roughly 40 percent of the increase in M l (not seasonally ad justed) that occurred from January to July 1979, and suggests that increases in Eurodollar borrowing have contributed to a more rapid expansion of the money supply. Since foreign exchange rates are sensitive to differential rates of anticipated inflation (and, hence, money growth), Eurodollar borrowing of this magni tude would indeed have affected the dollar’s exchange value, but in a direction opposite to that intended by the Federal Reserve Board. SUMMARY By raising the reserve requirement on large CDs after eliminating the reserve requirement for Eurodol lar borrowings, the Federal Reserve induced U.S. banks to borrow from their foreign branches. The combination of these two policy changes contributed to a rapid expansion in Eurodollar borrowing. These policies would have to be judged a success were their 12This calculation assumes a constant money multiplier of 2.5. sole intent to increase Eurodollar borrowing. While the elimination of reserve requirements against Euro dollars should increase demand for Eurodollars, it need not increase the demand for dollars in the for eign exchange market. However, the stated objective was to encourage Eurodollar borrowing which, in turn, would increase the foreign exchange value of the dollar. The link between Eurodollar borrowing and the foreign exchange value of the dollar, however, is more tenuous than that implicit in the Fed’s actions. Though the data do not permit a definitive analysis, available evidence suggests that a by-product of these policy actions has been the substitution of Eurodollar borrowing for CDs. This kind of substitution does not involve foreign exchange transactions and therefore has little direct effect on the dollar’s exchange value. There may, however, be an indirect effect on the foreign exchange value of the dollar. Substitution of reserve-free Eurodollar borrowing for reservable CDs has the potential to increase the U.S. money supply. Unless Federal Reserve open-market operations offset the increase in reserves that this substitution produces, the more rapid growth of money that results may ac tually depress the dollar’s foreign exchange value.