The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Review ________ ___ Vol. 67, No. 6 June/July 1985 5 Controlling Federal Outlays: Trends and Proposals 12 The Monetary Control Art, Reserve IiL\es and the Stock Prices of Commercial Banks 21 Recent Changes in Handling Bank Failures and Their Effects on the Banking Industry 29 Are Weighted Monetary Aggregates Better Than Simple-Sum M l? The Review is published 10 times per year by the Research and Public Information Department o f the Federal Reserve Bank o f St. Louis. Single-copy subscriptions are available to the public free o f charge. Mail requests for subscriptions, back issues, or address changes to: Research and Public Information Department, Federal Reserve Bank o f St. Louis, P.O. Bo* 442, St. Louis, Missouri 63166. The views expressed are those o f the individual authors and do not necessarily reflect official positions o f the Federal Reserve Bank o f St. Louis or the Federal Reserve System. Articles herein may be reprinted provided the source is credited. Please provide the Bank's Research and Public Information Department with a copy o f reprinted material. Federal Reserve Bank o f St. Louis Review June/Julv 1985 In This Issue . . . The federal government in its February 1985 budget announced cuts o f about $507 billion relative to current services estimates for the 1986-90 period. In the first article o f this Review, “Controlling Federal Outlays: Trends and Proposals," Keith M. Carlson assesses the significance o f these proposed reductions by comparing them with some longer-term trends in federal outlays. He concludes that, while attempts to cut the proportion o f total federal outlays to GNP have been unsuccessful thus far, the administration’s proposals, if achieved, w ould reduce outlays relative to GNP. Historical comparisons show, however, that the mix o f outlays between defense and nondefense has been altered quite dramati cally, and a continuation o f this reversal is proposed for the future. * * ♦ In the second article in this issue, “The Monetary Control Act, Reserve Taxes and the Stock Prices o f Commercial Banks,” G. J. Santoni shows that the reserve requirements im posed on financial institutions have the properties o f a tax. With the aid o f some simple examples, the author demonstrates that this reserve tax varies with the interest rate and, prior to 1980, had differential effects on the earnings streams and capital values o f banks. These differential tax effects are important. Equity considerations aside, they artificially raise the operating cost o f some firms relative to others engaged in essentially the same business. This distorts the rates o f production among the differentially taxed firms and lowers the value o f output for given costs. Santoni discusses an important change in the tax em bodied in the Monetary Control Act o f 1980. Specifically, the act raised reserve requirements for firms that are not members o f the Federal Reserve System w hile lowering them for member banks. The paper shows that the more uniform reserve requirements have significantly reduced the differential effect o f interest rate changes on the stock prices o f these two types o f banks. The legislation has raised the after-tax earnings streams and stock prices o f member banks, other things the same, w hile lowering both for nonmem ber banks. * * * In the early 1980s, the FDIC became concerned about a lack o f market discipline im posed by large depositor's on the risks assumed by banks. The FDIC has attempted to prom ote greater market discipline by allowing the uninsured depositors o f some failed banks to suffer losses. There is some evidence, however, of a double standard in these actions: the cases in which uninsured depositors have been exposed to losses involve relatively small banks, whereas the unin sured depositors o f the Continental Illinois Bank w ere protected from losses by the FDIC in May of last year. In the third article in this Review, “ Recent Changes in Handling Bank Failures and their Effects on the Banking Industry," R. Alton Gilbert investigates whether the FDIC s actions have induced uninsured depositors to shift their accounts to the relatively large banks that appear to be too large to be allowed to fail. There is no evidence o f a shift o f uninsured deposits from small to large banks. Instead, 3 In This Issue 4 there has been a small rise in the share o f large-denomination time deposits at relatively small banks in the past year. * * # The introduction o f new financial instruments and recent financial deregula tion have obfuscated the distinction between m oney and near-money assets. One response to this situation has been the introduction o f w eighted monetary aggregates, the construction o f which attempts to extract the m onetary services provided by assets. In the final article o f this issue, "Are W eighted Monetary Aggregates Better Than Simple-Sum M l? ” Dallas S. Batten and Daniel L. Thornton investigate the properties o f two w eighted monetary aggregates and compare these with the measure M l. Influencing econom ic activity is a central objective of monetary policy, so an important attribute for a monetary aggregate to possess is a close, predictable relationship with econom ic activity. Th e evidence, however, indicates that the relationship between the tw o weighted measures and eco nom ic activity, as measured by nominal GNP, has been no m ore predictable or less variable than the relationship between M l and GNP. Consequently, these w eighted aggregates do not demonstrate any apparent gain over M l that can be exploited for monetary policy purposes. Controlling Federal Outlays Trends and Proposals Keith M. Carlson I n his February 1985 budget message, President Reagan noted that The past 4years have also seen the beginning of a quiet but profound revolution in the conduct of our Federal Government. We have halted what seemed at the time an inexorable set of trends toward greater and greater Government intrusiveness, more and more regulation, higher and higher taxes, more and more spending, higher and higher inflation, and weaker and weaker defense.1 Yet, federal outlays as a proportion o f GNP w ere still half a percentage point above what they w ere when the administration took office in 1981. The purpose o f this article is to summarize recent trends in federal outlays and assess the administra tion’s future plans by placing them in a historical context.2The focus o f the discussion is on the behavior o f federal outlays as a percent o f GNP — a measure that was used initially by the administration to sum marize the governm ent’s influence on the economy. BUDGET OUTLAYS VS. CURRENT SERVICES Interpreting budget trends requires some reference measure that can be used for comparison. The referKeith M. Carlson is an assistant vice president at the Federal Reserve Bank of St. Louis. Thomas A. Poilmann provided research assistance. ence measure used in the February 1985 budget is the “current services budget.” According to the budget document, "current services” estimates are defined as . . . the estimated budget outlays and proposed budget authority that would be included in the budget for the following fiscal year if programs and activities of the United States Government were carried on during that year at the same level as the current year without a change in policy.3 Current services estimates “provide a base against which budgetary alternatives may be assessed."4 Table 1 summarizes both the administration’s 1985 proposals and the current services estimates for 1985 through 1990.5 A comparison o f the figures indicates that the administration plans to cut federal outlays by $507 billion between 1986 and 1990, with the largest cuts coming in the last three years. W hen converted to percentages, the cuts range from 5 percent in 1986 to 10.7 percent in 1990. The bottom half o f table 1 shows the current ser vices and proposed budget estimates as percentages o f GNP. The proposed estimates represent sizable decreases in the proportion o f federal government outlays to GNP com pared with the current services estimates. Whether such proposed reductions in the propor tion o f federal outlays relative to GNP w ill actually 'Office of Management and Budget (1985a). 3 Office of Management and Budget (1985b), p. A-1. 2Even though the administration’s February 1985 proposals will not be realized, these proposals provide a base for debate by Congress whereby modifications will be made. “Ibid, p. A-2. 5For alternative estimates of both the administration’s program and current services, see Congressional Budget Office (1985). 5 FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 1 Federal Outlays: Budget Estimates vs. Current Services, February 1985 In Billions of Dollars 1985 Current services Budget estimates Difference 1986 1987 1988 1989 1990 $960.4 959.1 $1024.5 973.7 $1109.2 1026.6 $1200.0 1094.8 $1262.8 1137.4 $1332.8 1190.0 $1.3 $50.8 $82.6 $105.2 $125.4 $142.8 | $506.8 Percent difference 0.1% 5.0% 7.5% 8.8% 9.9% 10.7% As a Percent of GNP Current services Budget estimates Difference 24.8% 24.8 24.4% 23.2 24.4% 22.6 24.4% 22.2 23.8% 21.4 23.4% 20.9 0.0% 1.2% 1.8% 2.2% 2.4% 2.5% NOTE: All figures include off-budget outlays. occur depends crucially on both political consider ations and future econom ic conditions — neither of which can be forecast with much reliability.6One way to assess the significance o f the proposed reductions, however, is to compare them with some longer-term trends in federal outlays. In this manner, it is at least possible to see what such reductions w ou ld mean in a historical context. BUDGET OUTLAYS AS A PERCENT OF GNP: A HISTORICAL PERSPECTIVE To examine properly federal outlays relative to GNP from a historical perspective requires adjusting out lays and GNP separately for the direct influence o f the business cycle.7 Since federal outlays generally rise relative to GNP during recessions, the inclusion o f such percentages without adjustment could distort the interpretation o f underlying trends. Total Outlays The historical record o f cyclically adjusted federal outlays as a percent o f adjusted GNP is summarized in chart 1. Even with cyclical adjustment, this measure o f government activity is still quite volatile, especially on a year-to-year basis. Consequently, a trend line for the period 1956-81 has been plotted in the chart. Extending the trend line from 1982 through 1990 indicates that the administration has not been suc cessful in reducing total outlays as a percent o f GNP in the 1981-84 period. Moreover, the proposed 1985 level o f outlays is w ell above the historical trend. Chart 1 does show that the administration is pro posing a path o f outlays after 1985 that differs dramati cally from both the 1956-81 trend and its first four years in office. If the administration’s proposals are enacted, the size o f government w ould be reduced to that prevailing in the mid-1970s. The Composition o f Total Outlays 6See Carlson (1983). 7Federal outlays were adjusted for the cycle using correction factors implicit in the work by de Leeuw and Holloway (1983). This meant adjusting budget outlays in the same proportion as national income accounts federal expenditures are adjusted to derive cyclically adjusted expenditures. Following this procedure captures only the automatic response of federal outlays to the business cycle, mean ing that countercyclical fiscal actions are still reflected in the figures. Trend GNP is middle-expansion trend GNP as defined by de Leeuw and Holloway. See also Holloway (1984). 6 An examination o f total budget outlays relative to GNP masks the contrasting differences taking place between defense and nondefense outlays. Chart 2 summarizes these outlays relative to GNP. Nondefense outlays and GNP are adjusted for the business cycle; defense outlays are not adjusted because they are not systematically related to the business cycle. The de fense portion o f the chart shows the downw ard trend FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 C h a rt 1 Total B u d g e t O u t la y s a s a Pe rcent of G N P Per cent Percent o f defense outlays relative to GNP from 1956 to 1981. Since 1981, the trend has been reversed, with defense spending rising to 6.3 percent o f GNP in 1984. The administration plans for future defense spending to continue to rise relative to GNP; the proposed budget calls for defense outlays to reach 7.5 percent o f GNP by 1990.8 fense outlays to be reduced to 13.4 percent o f GNP by 1990. If realized, the relative size o f the nondefense budget w ould be reduced to levels prevailing in the early 1970s. The nondefense portion o f the chart shows that the growth o f cyclically adjusted nondefense outlays rela tive to trend GNP was extraordinarily rapid from 1956 to 1981. Such spending rose from 6.9 percent o f GNP in 1956 to 18 percent in 1981. Since 1981, however, the ratio o f nondefense outlays to GNP has been reduced relative to its 1956— trend. 81 Chart 3 summarizes nondefense spending by major program category and emphasizes the m ethod o f car rying out government activities. The purpose o f look ing at these categories is to determine where the nondefense budget cuts w ill fall.9 The administration plans for the reduction in non defense outlays relative to GNP to continue; these reductions are quite dramatic relative to the 1956-81 trend. The administration’s proposals call for nonde 8 The administration indicates that its proposed defense outlays will be less than the current services estimates (see the appendix to this article for 1990 estimates). The Congressional Budget Office dis putes this contention, claiming that the administration's defense proposals are greater than current services estimates. See Con gressional Budget Office (1985), p. 22. The Composition o f Nondefense Outlays The largest proportion o f nondefense spending, given this set o f categories, is payments for individuals. This category includes both direct (for example, Social Security benefits) and indirect (via grants to state and local governments, such as M edicaid and assistance payments) transfer payments by the federal govern ment. According to the top tier o f chart 3, this spend ing grew rapidly from 1956 to 1981; its trend has apparently been reversed since 1983. The administra t o r further detail relative to current services estimates, see the appendix. 7 FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Composition of Total Budget Outlays Percent o f G N P Percent N a tio n a l Defense O u tla y s L J _ I _L __ Percent 221 ----------------------------N on d e fe nse O u tla y s 96 5 -8 ^T i i i i T 1956 58 60 iiii 62 64 T 66 Trend iiii 68 T 70 iiii 72 74 T 76 iiii 78 T 80 iiii 82 84 T 86 iiii 88 T,) 1990 N O TE : N o n defense o u tla y s and GNP a rc c y c lic a lly a d ju s te d . N ondefens# o u tla y s in c lu d e u n d is trib u te d o ffse ttin g receipts. tion plans to continue to reduce such payments rela tive to GNP to 9.8 percent by 1990, a dramatic depar ture from its growth over the 1956-81 period. The categoiy labeled "all other grants" includes all grants to state and local governments except transfer payments. Included in this categoiy are grants for wastewater treatment plants, highway construction, community development, education, em ployment and training assistance, and general revenue sharing. The second tier o f chart 3 indicates that this category o f spending has been reduced w ell below the 1956-81 trend line in recent years. The extent o f the cut is dramatic — from a peak o f 2.6 percent o f GNP in 1978 to 1.5 percent in 1984. Furthermore, this categoiy is projected for further cuts in the future, to 0.9 percent o f GNP in 1990. 8 The net interest category has attracted considerable attention in recent years. Once a relatively insignifi cant part o f the budget, it has risen considerably to the point where policymakers n ow view it with major concern.1 The third tier o f chart 3 shows that, after 0 rising from 1.3 percent o f GNP in 1956 to 1.6 percent in 1976, net interest rose steadily to 3.1 percent o f GNP in 1984. Projections o f net interest depend on a number o f factors, the most important o f w hich is the future course o f deficits and the projected level o f interest '“This is because of the cumulative effect of net interest. Higher net interest adds to the current deficit, which carries over to future years in the form of a larger debt that must be financed. See Carlson (1984). RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 ctit Perceat % % 14 % \ \ \ \ \ \ % X \ P a y m e n ts 1o r In d iv id u c i Is 12 10 a 1956-81 Trendy 6 4 0~ 1 1 1 1q ~ l h fceat 1 1 1' ~ 1 1 1 1 ~ ~ 1 1 1 1 = " i i i ~ i r i i ~~ i i i i = i Perceat 4 2 0 Perceat Pi riaat 4 2 0 P< reeat i Perceat 6 A ll O th e r !:e d e ra l O p e rations Actual 4 1956-81 Trend i \ 2 0 1 1 1 1 IfS i 58 60 l i2 1 1 1 64 1 1 1 1 66 6S 70 'JOTE G N P a n d p a y m e n ts f o r in d iv id u a ls a r e c y c lic a lly a d ju s te d . A l l 1 1 1 1 1 1 1 1 72 74 o th e r fe d e r a l 76 7S SO I 1 1 1 82 (4 _ _ ' 1 1 1 1 86 88 1990 o p e r a t io n s d o e s n o t in c lu d e u n d is tr ib u te d o f f s e t t in g r e c e ip ts . 9 FEDERAL RESERVE BANK OF ST. LOUIS rates. Given the administration’s overall plan for re ducing the size o f the deficit and a projected decline in interest rates, net interest outlays as a percent of GNP is projected to continue rising through 1985, level off for two years, then drop sharply to 1990; however, it w ill still remain above the 1956-81 trend as extrapo lated to 1990. The "all other federal operations” categoiy includes outlays for foreign aid, general science research and space technology, energy programs, farm price sup ports, housing credit activities and day-to-day opera tions o f the government. Relative to GNP, as shown in the bottom tier o f chart 3, this category of nondefense outlays displayed a slight upward trend during the 1956-81 period; it has declined in recent years. The jump in the estimate for fiscal year 1985 reflects pri marily the surge in outlays related to the PIK farm program. The administration plans to continue to cut such outlays as a percent o f GNP through 1990. Such proposed cuts are centered on farm price support programs, foreign aid and loan activities o f the government. SUMMARY The federal government in its February 1985 budget announced cuts o f about $507 billion relative to cur rent services estimates for the 1986— period. These 90 proposed cuts w ere com pared with recent trends in federal outlays relative to GNP since 1956; the results o f these comparisons are summarized in table 2. The historical record indicates that, w hile attempts to cut the proportion o f total federal outlays to GNP have been unsuccessful thus far, the administration’s cur rent proposals, if achieved, w ould reduce outlays rela tive to GNP. The historical comparisons show the present administration has altered the mix o f total outlays between defense and nondefense quite dra matically, and a continuation o f this reversal is pro posed for the future. Payments for individuals are scheduled to be cut m oderately relative to GNP for each year after 1985. Net interest as a percent o f GNP, which is currently climb ing w ell above past trends, is projected to continue rising through 1985, level off, then move back toward JUNE/JULY 1985 Table 2 Summary of Trends Relation to 1956-81 trend1 Percent of GNP Total budget outlays National defense Nondefense Payments for individuals All other grants Net interest All other federal operations 1982-84 1985-90 Above Above Below Below Below Above Below Below Above Below Below Below Above Below 'A straight line trend was fitted to the relevant measure of outlays as a percent of GNP. trend after 1987. Budget cuts, as measured by outlays relative to GNP, are concentrated in “all other grants to state and local governm ents’’ and in “all other federal operations.” The governm ent’s program is ambitious: in order to reduce total budget outlays to 20.9 percent o f GNP by 1990, w hile at the same time increasing defense outlays to 7.5 percent o f GNP, nondefense outlays w ill have to be reduced to 13.4 percent o f GNP from the current level o f approximately 18 percent. REFERENCES Carlson, Keith M. “The Critical Role of Economic Assumptions in the Evaluation of Federal Budget Programs,” this Review (October 1983), pp. 5-14. _________ “Money Growth and the Size of the Federal Debt,” this Review (November 1984), pp. 5-16. Congressional Budget Office. An Analysis of the President’s Bud getary Proposals for Fiscal Year 1986 (February 1985). de Leeuw, Frank, and Thomas M. Holloway. "Cyclical Adjustment of the Federal Budget and Federal Debt,” Survey of Current Busi ness (December 1983), pp. 25-40. Holloway, Thomas M. Cyclical Adjustment of the Federal Budget and Federal Debt: Detailed Methodology and Estimates, Bureau of Economic Analysis Staff Paper 40 (U.S. Department of Com merce, June 1984). Office of Management and Budget. Budget of the United States Government: Fiscal Year 1986 (February 1985a), p. M4 -------------- - Special Analysis: Budget of the United States Govern ment: Fiscal Year 1986 (February 1985b). APPENDIX: Composition of Federal Outlays Federal outlays can be classified in terms o f two analytical structures: budget function and major pro 10 gram categoiy. The functional classification presents outlays according to the purposes that federal pro- FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table A1 1990 Federal Outlays: Current Services vs. Administration Proposals (amounts in billions of d o l l a r s ) __________________________ Percent difference Current services Administration proposal Difference $441.7 $428.6 $ -1 3 .1 590.6 48.9 369.2 140.8 31.7 555.0 40.8 355.4 130.6 28.2 -3 5 .6 -8 .1 -1 3 .8 -1 0 .2 -3 .5 - 6 .0 -1 6 .6 - 3 .7 - 7 .2 -1 1 .0 61.2 3.5 22.2 5.7 43.9 2.7 16.8 3.8 -1 7 .3 - 0 .8 - 5 .4 -1 .9 -2 8 .3 -2 2 .9 -2 4 .3 -3 3 .3 22.7 7.1 18.7 1.9 - 4 .0 - 5 .2 -1 7 .6 -7 3 .2 Net interest 164.2 137.7 -2 6 .5 -1 6 .1 Other federal operations' International affairs General service, space and technology Energy Natural resources and environment2 Agriculture Commerce and housing credit Transportation2 Community and regional development2 Education, training, employment and social services2 Administration of justice General government Allowances 117.3 19.5 11.0 6.1 9.3 20.1 3.8 10.0 2.5 69.9 14.5 11.1 2.3 7.3 3.8 - 3 .7 7.6 1.7 -4 7 .4 -5 .0 0.1 - 3 .8 - 2 .0 -1 6 .3 - 7 .5 -2 .4 - 0 .8 -4 0 .4 -2 5 .6 0.9 -6 2 .3 -2 1 .5 -8 1 .1 -1 9 7 .4 -2 4 .0 -3 2 .0 11.6 7.4 6.1 9.9 9.5 6.9 4.9 4.0 -2 .1 - 0 .5 - 1 .2 - 5 .9 -1 8 .1 - 6 .8 -1 9 .7 -5 9 .6 -4 2 .1 -4 5 .0 - 2 .9 — Category .. . National defense Benefit payments for individuals' Health Social Security and Medicare Income security Veteran payments Other grants to state and local governments' National resources and environment2 T ransportation2 Community and regional development2 Education, training, employment and social services2 General purpose fiscal assistance Undistributed offsetting receipts -3 .0 % 'Amounts shown are the sums for the functions listed under them, and differ slightly from the "major program category" amounts shown in the budget. 2 The budget gives current services estimates for the total. Estimates by major program category were estimated by the author. grams are intended to serve. These functions are grouped into 18 broad areas, including, for example, national defense, international affairs, energy p ro grams, agriculture, transportation, health and general government programs. Three additional categories — net interest, allowances and undistributed offsetting receipts — do not address specific functions, but are included to cover the entire budget. Classification o f federal outlays by major program category focuses on the m ethod o f carrying out an activity. The major program categories are national defense, benefit payments to individuals, grants to state and local governments (other than for benefit payments), net interest, other federal operations and undistributed offsetting receipts. National defense, net interest, and undistributed offsetting receipts cor respond to the functional categories o f the same name, but, the remaining major program categories do not correspond to a simple summing o f functional categories. Nonetheless, approximations can be made. The accompanying table groups 1990 outlays by func tion to show the approximate com position o f some o f the major program categories. 11 The Monetary Control Act, Reserve Taxes and the Stock Prices of Commercial Banks G. J. Santoni C L -J lN C E 1980 all depository institutions have been required to hold reserve balances in the form o f Trea sury coins and Federal Reserve notes either in their own vaults or on deposit at their regional Federal Reserve Banks. These reserve balances pay no interest, so the foregone interest earnings on the investments the firm could otherwise have made can be viewed as a tax.1 This tax lowers the firm ’s expected stream o f future incom e net o f taxes which, other things the same, reduces the capital value o f the firm. The tax varies with the general level o f interest rates as well as the spread between bank lending and borrowing rates. Prior to 1980, the tax had differential effects across banks depending on the tax rate (required reserve ratio) faced by these various firms. This was particu larly true with respect to m em ber vs. nonmem ber banks o f the Federal Reserve System.2 These differen tial tax effects are important. Equity considerations aside, they artificially raise the operating costs o f some firms relative to others engaged in essentially the same business activity. This distorts rates o f production and the allocation o f resources among the differentially taxed firms and lowers the value o f output for given costs. The Monetary Control Act o f 1980 im posed uniform reserve requirements on all depository institutions by raising reserve requirements for nonm em ber banks, while lowering them for mem ber banks. The purpose o f this article is to analyze the effect this legislation has had in eliminating the differential tax effect o f interest rate changes on m em ber vs. nonm em ber banks. In particular, the paper examines w hether the act was effective in revising the response o f bank capital values (stock prices) to interest rate changes. Since any revi sion in differences in tax rates between groups gener ally benefits one group over another, the paper pro vides some rough estimates o f this as well. RESERVE REQUIREMENTS: PRE-1980 Prior to the Monetary Control Act, reserve require ments for nonm em ber banks w ere set by the various state banking authorities. These differed across states with respect to the reserve ratio, the form in which the reserves w ere required to be held, the m ethod and frequency o f policing and the penalty im posed for deficiency.3 W hile differences existed, the reserve re quirements o f state banking authorities generally w ere more lenient than those o f the Federal Reserve System. This appears to have been so with respect to the form o f the reserves, policing and penalties for deficiency.4 Specifically, 30 o f the 50 state banking authorities allowed banks to hold at least a portion o f their re serves in interest-earning assets, 36 states did not require periodic reporting o f reserve and deposit bal ances and 22 had no monetary penalty for deficient banks/1In contrast, Fed members had to hold reserves either in their vaults or on deposit at a Federal Reserve Bank. These reserve balances earned no interest. M em ber banks reported their deposit and reserve balances to the Fed on a weekly basis, and a monetary penalty was enforced for deficient banks. The left side o f table 1 gives the reserve require- G.J. Santoni is a senior economist at the Federal Reserve Bank of St. Louis. Thomas A. Pollmann provided research assistance. This paper was produced in conjunction with the Center for Banking Research at Washington University in St. Louis and appears in the working paper series published by that institution. 'For discussions of the effects of differential reserve requirements, see Fama (1985); Cargill and Garcia (1982); Gilbert (1978); Gilbert and Lovati (1978); Prestopino (1976); Goldberg and Rose (1976) and Knight (1974). 2 See Goldberg and Rose (1976) and Prestopino (1976). 12 3 See Gilbert and Lovati (1978), Prestopino (1976) and Knight (1974). "See Gilbert and Lovati (1978), p. 32, and Knight (1974), p. 12-13, for listings of the various state requirements. 5 Seven states imposed reserve requirements that were roughly identical to those of Fed members. These states were Arkansas, California, Kansas, Nevada, New Jersey, Oklahoma and Utah. Nonmember banks in these states are excluded from the data sample in the tests conducted below. FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 1: Depository Institutions’ Reserve Requirements (percent of deposits)1 Type of deposit and deposit interval (millions of dollars) Member bank requirements before implementation of the Monetary Control Act Percent Effective date 7 12/30/76 12/30/76 12/30/76 12/30/76 12/30/76 % 9 1/2 113/4 123/4 16V4 Time and savings234 Savings % 3/16/67 3 % 2Vz 1 3/16/67 1/8/76 10/30/75 6 2Vz 12/12/74 1/8/76 10/30/75 3 Time5 0-5, by maturity 30-179 days 180 days to 4 years 4 years or more Over 5, by maturity 30-179 days 180 days to 4 years 4 years or more Percent Effective date Net transaction accounts7 Net demand2 $ 0 - 2 2 - 10 10-100 100-400 Over 400 Type of deposit, and deposit interval (millions of dollars) Depository institution requirements after implementation of the Monetary Control Act6 $0-$29.8 Over $29.8 3% 12 1/1/85 1/1/85 3% 0 10/6/83 10/6/83 3% 11/13/80 Nonpersonal time deposits s By original maturity Less than 4 years 4 years or more Eurocurrency liabilities 1 All types SOURCE: Federal Reserve Bulletin, November 1980, p. A8. 'For changes in reserve requirements beginning 1963, see Board's Annual Statistical Digest, 1971-1975 and for prior changes, see Board’s Annual Report for 1976, table 13. Under provisions of the Monetary Control Act, depository institutions include commercial banks, mutual savings banks, savings and loan associations, credit unions, agencies and branches of foreign banks, and Edge Act corporations. 2(a) Requirement schedules are graduated, and each deposit interval applies to that part of the deposits of each bank. Demand deposits subject to reserve requirements are gross demand deposits minus cash items in process of collection and demand balances due from domestic banks. (b) The Federal Reserve Act as amended through 1978 specified different ranges of requirements for reserve city banks and for other banks. Reserve cities were designated under a criterion adopted effective November 9, 1972, by which a bank having net demand deposits of more than $400 million was considered to have the character of business of a reserve city bank. The presence of the head office of such a bank constituted designation of that place as a reserve city. Cities in which there were Federal Reserve Banks or branches were also reserve cities. Any banks having net demand deposits of $400 million or less were considered to have the character of business of banks outside of reserve cities and were permitted to maintain reserves at ratios set for banks not in reserve cities. (c) Effective August 24, 1978, the Regulation M reserve requirements on net balances due from domestic banks to their foreign branches and on deposits that foreign branches lend to U.S. residents were reduced to zero from 4 percent and 1 percent, respectively. The Regulation D reserve requirement on borrowings from unrelated banks abroad was also reduced to zero from 4 percent. (d) Effective with the reserve computation period beginning November 16, 1978, domestic deposits of Edge corporations were subject to the same reserve requirements as deposits of member banks. N egotiable order of withdrawal (NOW) accounts and time deposits such as Christmas and vacation club accounts were subject to the same requirements as savings deposits. "The average reserve requirement on savings and other time deposits at that time had to be at least 3 percent, the minimum specified by law. 5Effective November 2, 1978, a supplementary reserve requirement of 2 percent was imposed on large time deposits of $100,000 or more, obligations of affiliates and ineligible acceptances. This supplementary requirement was eliminated with the maintenance period beginning July 24,1980. Effective with the reserve maintenance period beginning October 25, 1979, a marginal reserve requirement of 8 percent was added to managed liabilities in excess of a base amount. This marginal requirement was increased to 10 percent beginning April 3,1980, was decreased to 5 percent beginning June 12,1980, and was reduced to zero beginning July 24,1980. Managed liabilities are defined as large time deposits, Eurodollar borrowings, repurchase agreements against U.S. government and federal agency securities, federal funds borrowings from nonmember institutions and certain other obligations. In general, the base for the marginal reserve requirement was originally the greater of (a) $100 million or (b) the average amount of the managed liabilities held by a member bank, Edge corporation, or family of U.S. branches and agencies of a foreign bank for the two statement weeks ending September 26,1979. For the computation period beginning March 20, 1980, the base was lowered by (a) 7 percent or (b) the decrease in an institution’s U.S. office gross loans to foreigners and gross balances due from foreign offices of other institutions between the base period (September 13-26, 1979) and the week ending March 12, 1980, whichever was greater. From the computation period beginning May 29,1980, the base was increased by 7-1/2 percent above the base used to calculate the marginal reserve in the statement week of May 14-21,1980. In addition, beginning March 19,1980, the base was reduced to the extent that foreign loans and balances declined. 6For existing nonmember banks and thrift institutions, there is a phase-in period ending September 3,1987. For existing member banks the phase-in period is about three years, depending on whether their new reserve requirements are greater or less than the old requirements. For existing agencies and branches of foreign banks, the phase-in ended August 12,1982. All new institutions will have a two-year phase-in, beginning with the date that they open for business. Transaction accounts include all deposits on which the account holder is permitted to make withdrawals by negotiable or transferable instruments, payment orders of withdrawal, telephone and preauthorized transfers (in excess of three per month), for the purpose of making payments to third persons or others. 8ln general, nonpersonal time deposits are time deposits, including savings deposits, that are not transaction accounts and in which the beneficial interest is held by a depositor which is not a natural person. Also included are certain transferable time deposits held by natural persons and certain obligations issued to depository institution offices located outside the United States. For details, see section 204.2 of Regulation D. NOTE: Required reserves must be held in the form of deposits with Federal Reserve Banks or vault cash. After implementation of the Monetary Control Act, nonmembers may maintain reserves on a pass-through basis with certain approved institutions. 13 FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 RELATIVE CAPITAL VALUES AND THE INTEREST RATE Table 2 Member Bank Foregone Interest on Reserve Balances (millions of dollars) Net Demand $ 0 2 10 25 100 400 1,000 2,000 5,000 Required Reserves Foregone Interest When i = .08' $ 0.00 0.14 0.90 2.66 11.48 50.00 147.50 310.00 797.50 $ 0.000 0.011 0.072 0.213 0.918 4.000 11.800 24.800 63.800 'Foregone interest = i x required reserves ments that applied to Federal Reserve m em ber banks before Novem ber 1980.“ These reserve ratios w ere at least as high as those im posed by the various state banking authorities for nonm em ber banks and, in most cases, they w ere higher.7 THE MEMBERSHIP TAX: PRE-1980 Other things the same, the m ore stringent reserve requirements for Fed members raised the cost of maintaining a given level o f deposits relative to the cost experienced by nonmembers. Table 2 uses the data in table 1 to calculate the tax for m em ber banks at various levels o f net dem and deposits.8For example, a member bank with $100 million in net dem and d e posits was required to hold $11.48 m illion in reserves. This resulted in foregone earnings of $918,000 per year if the market rate w ere 8 percent.3 The decline in the expected stream o f earnings was the reserve tax (in this case, $918,000 per year). Since the capital value o f a firm is the present value o f its expected earnings stream, the tax reduced the capital value o f the bank as well. T h e Monetary Control Act was passed in March 1980, but the new reserve requirements did not become effective immediately. The right side of the table indicates the reserve requirements that would have been imposed as of November 13,1980, if there had been no phase-in period. In fact, these new requirements were phased in over a period of years (see table 1, note 6). For the moment, the discussion is focused on pre-November 1980 reserve requirements. Not only does the reserve tax reduce the expected earnings streams and capital values o f m em ber banks (those with higher reserve requirements) relative to nonmem ber banks, but the earnings streams and cap ital values o f m em ber banks change relative to non m ember banks with changes in either the general level o f interest rates or the spread between bank borrow ing and lending rates. A Change in the General Level o f Interest Rates Table 3 illustrates the effect o f a change in the level o f interest rates with the spread held constant. In panel A, the rate at which banks can lend is assumed to be 10 percent, w hile the rate paid on deposits (and other sources o f funds) is 5 percent. The reserve re quirement for m em ber banks is assumed to be 10 percent. For illustrative purposes, the non-interesteamings reserves o f nonmembers are assumed to be zero. The table calculates the amount available for lending, the annual net revenue and the capital value o f the net revenue stream for each $100 o f deposits for both a m em ber and a nonm em ber bank. The reserve requirement lowers the amount that can be loaned, the stream o f net revenue and capital value o f the m em ber relative to the nonm em ber bank. The capital value o f the m em ber’s revenue stream is $40, w hile the nonm em ber’s is $50. The m em ber’s capital value relative to the nonm em ber’s is 80 per cent. Notice that the absolute difference between the two capital values is equal to the required reserves o f members ($50 — $40 = $10). In panel B, both lending and borrowing rates are assumed to increase to 20 percent and 15 percent, respectively, w hile other things remain the same."' The net revenue stream o f the nonm em ber does not change w hile the m em ber’s stream falls. The increase in interest rates causes the capital value o f both banks to decline. M ore importantly, however, the capital value o f the mem ber bank drops from 80 percent to 60 percent in terms o f the capital value o f the nonm em ber bank. Notice that, in this particular case, the absolute 7 See Gilbert and Lovati (1978) for a listing of the reserve ratios imposed by the various state authorities. T h e calculation is intended for illustrative purposes only and ignores the foregone interest on reserves held against time deposits. This represents an upper bound to the tax since the bank would maintain some reserves even if there were no legal requirement to do so. 14 ,0ln the example, the absolute spread is unchanged but the relative spread (iB ) changes. If the ratio of the borrowing to the lending rate /iL remained constant as the general level of interest rates changed, relative capital values would not change. The example is intended for illustrative purposes. A more precise statement of the effect of interest rate changes is given in the appendix. FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 3 A Change in the General Level of Interest Rates and Relative Capital Values1 __________________________________ Members Nonmembers $100 10 $100 $ 90 $100 $ 9 5 $ 10 5 $ 4 $ Relative Capital Value Panel A: Lending rate = 10% Borrowing rate = 5% Deposit Required reserves Available for lending Annual revenue (.10 x Loan) Cost (.05 x Deposit) Net revenue Capital value (Net revenue/,10) $ 40 — 5 $ 50 .80 = Panel B: Lending rate = 20% Borrowing rate = 15% Deposit Required reserves $100 10 $100 — Available for lending $ 90 $100 Annual revenue (.20 x Loan) Cost (.15 x Deposit) $ 18 15 $ 20 15 $ $ Net revenue Capital value (Net revenue/,20) 3 $ 15 =— 5 $ 25 ===: .60 'Conditions of the example are that deposits of both banks are $100, the required reserve ratio for members is 10 percent and zero for nonmembers. difference between the capital values o f the two banks does not change. This is because the banks in this example have the same level o f deposits and, thus, the differential effect caused by m em ber bank reserve re quirements remains constant (see the appendix for a more formal presentation). A Change in the Interest Rate Spread Table 4 is similar to table 3 except that it illustrates the effect on relative earnings streams and capital values o f a change in the spread between the interest rate banks charge on loans and the rate paid on deposits. The top halves o f the two tables are identical. In panel B o f table 4, however, the lending rate is assumed to increase w hile the borrowing rate remains unchanged. The earning streams and capital values of both banks rise with the capital value o f the member rising relative to that o f the nonmember. In this example, the interest rate spread increases as the bank lending rate rises, w hile the borrowing rate remains the same. A qualitatively similar result would occur if the borrowing rate declined, w hile the lending rate remained the same. While changes in the spread are potentially im por tant, two problems arise when testing for this effect. First, prior to 1981, the interest rate banks could pay on deposits was subject to a ceiling. During much o f the earlier portion of the sample period used here, the ceiling was effective. As a result, changes in the spread w ere highly correlated with changes in the general level o f interest rates. Second, the spread between lending and borrowing rates is the compensation 15 FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 4 Changes in the Interest Rate Spread and Relative Capital Values1 Members Nonmembers $100 10 $100 — $ 90 $100 $ 9 5 $ 10 5 $ 4 $ Relative Capital Value Panel A: Lending rate = 10% Borrowing rate = 5% Deposit Required reserves Available for lending Annual revenue (.10 x Loan) Cost (.05 x Deposit) Net revenue Capital value (Net revenue/. 10) 5 $ 40 $ 50 $100 10 $100 — Available for lending $ 90 $100 Annual revenue (.20 x Loan) Cost (.05 x Deposit) $ 18 5 $ 20 5 $ 13 $ 15 $ 65 $ 75 .80 Panel B: Lending rate = 20% Borrowing rate = 5% Deposit Required reserves Net revenue Capital value (Net revenue/.20) .87 'The conditions of the example are that deposits for both banks are $100, the required reserve ratio for member banks is 10 percent, both banks pay 5 percent on deposits, and both extend loans at the same interest rate. In panel A, the loan rate is 10 percent while, in panel B, the loan rate is 20 percent. The borrowing rate does not change. The firm’s cost of capital is assumed to equal the lending rate. banks earn for em ploying their specialized resources to intermediate financial transactions. W hen borrow ing and lending rates are free to move, as was true after 1981, competition among intermediaries assures that the spread is just sufficient to cover costs. Unless there is a change in the technology o f the intermediation process, there is little reason to expect the spread to vaiy significantly. For these reasons, the spread is excluded in the following empirical analysis and at tention is focused on variation in the level o f interest rates." "In regressions not reported here, the product of a dummy variable and various proxies tor the spread were tested. The dummy variable was used to control for the period of deposit rate ceilings that prevailed prior to 1981. The dummy variable assumed a value of one for the period since relaxation of the interest rate ceilings on deposits (1/1981— IV/1983), and zero otherwise. The coefficient of 16 RESERVE REQUIREMENTS AFTER THE MONETARY CONTROL ACT The right side o f table 1 shows the reserve require ments o f depository institutions after the im plementa tion o f the Monetary Control Act.'- These reserve re q u irem e n ts a p p ly to d e p o s ito r y in s titu tio n s regardless o f Fed membership. They substantially re duce the required reserve balances o f m em ber banks this variable did not differ significantly from zero. The proxies for the lending rate used to calculate the spread were the one-month commercial paper rate, the 4 -6 month commercial paper rate and the 90-day bankers acceptance rate. The borrowing rate proxy was the Federal Reserve discount rate. ,2See table 1, note 6, for a discussion of the period over which the new requirements were phased in. In the text, the phase-in period is ignored unless otherwise mentioned. FEDERAL RESERVE BANK OF ST. LOUIS at each level o f net dem and deposits, w hile generally increasing them for nonmem ber banks. Table 1 also presents the pre- and post-reserve requirements on time and savings deposits. Before the Monetary Control Act, required reserve holdings against personal and nonpersonal time deposits ranged from 1 to 6 percent (with a minimum average requirement o f 3 percent), w hile those on savings deposits w ere 3 percent. The act reduced these re quirements to zero for personal time and savings accounts.'3 Since these deposits represent a substan tial portion o f total time and savings deposits, this change results in a significant reduction in member bank required reserves.1 Furthermore, the reserve re 4 quirement on managed liabilities and the supplem en tary reserve requirement on time deposits o f $100,000 or more were reduced to zero in July 1980. W hile the change in the level o f required reserves mandated by the act is clearly important for some issues, what is most important for the purpose o f this paper is that this legislation imposes uniform reserve requirements across m em ber and nonm em ber banks. (See the insert on page 18 for a discussion o f some other provisions o f the act.I SOME IMPLICATIONS AND EVIDENCE The phase-in period for the new reserve require ments, which extended through 1984 for member banks, w ill not be com plete for nonmembers until September 1987. This w ill mitigate the quantitative effect o f the change on the following estimates but the expected qualitative effect should show through.'' The Effect o f Interest Rate Changes on Relative Stock Prices In an effort to evaluate the implications o f the above argument, quarterly data on the share prices and demand deposit liabilities o f 40 publicly traded bank holding companies w ere examined. The holding com panies w ere divided into two categories depending on '3See table 1, note 8, for a definition of personal vs. nonpersonal time and savings deposits. '■•For example, for banks in the Eighth Federal Reserve District, the personal portion of savings deposits was more than five times greater than the nonpersonal portion, while the personal portion of time deposits was more than four times the nonpersonal portion. ,5See Pearce and Roley (1983) and (1985). JUNE/JULY 1985 whether the subsidiary banks making up an individual holding company w ere members or nonmembers o f the Federal Reserve System.1 The stock prices o f each 6 holding company w ere adjusted for stock splits and stock dividends, and simple quarterly averages o f stock prices and dem and deposit liabilities w ere com puted for each o f the two categories o f holding com panies. The sample period runs from 1/1974— IV/1983. The previous arguments im ply that the capital val ues o f mem ber relative to nonm em ber banks w ill be related in a specific w ay to certain other variables. Consequently, the variable to be explained (depen dent variable) in the following regression is the ratio o f the average stock prices o f m em ber to nonm em ber banks. For purposes o f the empirical estimate, the dependent variable is expressed in log form. The following empirical analysis is primarily con cerned with the relationship between the dependent variable and the level o f interest rates. Since an in crease in the level o f interest rates is thought to reduce member bank capital values relative to those o f non member banks, the sign o f the estimated coefficient on the level o f interest rates is expected to be negative. Further, the above arguments indicate that the rela tionship between these variables w ill change in a particular way following implementation o f the M one tary Control Act.'7Consequently, an interaction term is included in the regression as an independent vari able.1 8 The interaction term is included to test for the effect that the M onetary Control Act has had in eliminating the differential response o f the capital values o f m em bers vs. nonmembers to interest rate changes. The interaction term is the product o f a coefficient (to be estimated), a dum m y variable and the .level o f the interest rate. The dum m y variable assumes a value of one for the period subsequent to implementation of the M onetaiy Control Act, w hile its value is zero dur ing the earlier period. Since the hypothesis suggests that the uniform reserve requirements em bodied in the legislation w ill eliminate the adverse conse- 1 The data set includes only state-chartered banks. Nationally char 6 tered banks are required to be members of the Fed, but are ex cluded from this sample mainly because they are much larger on average than state-chartered banks and are subject to different regulatory agencies. "See the appendix for a summary of the theory that underlies the estimating equation. 1 The proxy employed for the general level of interest rates is the 8 corporate Aaa bond rate. A long-term interest rate was selected since it is presumed to represent some average of current and expected future shorter-term interest rates. 17 JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS Chart 1 Ratio of M e m b e r Banks to N onm em ber Banks Ratio Ra tio i Some Other Provisions of the Act The Monetary Control Act contains many other provisions that have important implications for financial firms and markets that are distinct from its impact on required reserves. Most o f these provi sions are not expected to affect m em ber banks any differently than nonm em ber banks. There are two exceptions, however. Before the Monetary Control Act, the Federal Reserve System provided certain services to m em bers that w ere free o f direct charge. In addition, members w ere allowed to borrow from the System at the Federal Reserve discount rate. Neither o f these services w ere available to nonm em ber banks. The Federal Reserve System is now required by the Monetary Control Act to charge for the banking services it provides and to make these services available to any bank that wants to use them. In addition, borrowing from the Fed is no longer the 18 exclusive privilege o f m em ber banks.' The effect o f these tw o changes is to raise the capital values o f nonm em ber banks relative to member banks and to offset the effect o f the reserve requirement changes. It is unlikely, however, that these two provisions o f the act com pletely offset the effect o f the reserve requirement changes on rela tive capital values. Prior to 1980, Federal Reserve membership was declining both absolutely and relative to all comm ercial banks (see chart 1). The most frequently m entioned reason for leaving was the System’s higher reserve requirements. Clearly, for the banks that decided to leave and those new banks that did not join, the System’s reserve re quirements w ere too high a price to pay for “free” services and access to the discount window. 'See Brewer (1980). FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 o f the two types o f banks is significant and positive. Table 5 The Monetary Control Act and the Stock Prices of Member Vs. Nonmember Banks Estimate' Ln(P„/PN = .065 + .089 D„/DN - .055 i + .020 DUM-i ) (.36) (2.89)* (4.68)* (4.56)* R2 = .54 Rho = - .30 (2 .02 )* where: P « /P n = the average stock price of member banks relative to the average for nonmember banks Dm n = the average level of member bank demand /D deposit liabilities relative to the average for nonmember banks i = a proxy for the general level of interest rates. The proxy is the level of the corporate Aaa bond rate. DUM = a dummy variable for the period since implementation of the Monetary Control Act. DUM = 1 for the period 1/1981-IV /1983 and zero otherwise.2 'Significantly different from zero at the 5 percent level 't-values in parentheses. Adjusted for first-order autocorrelation. The regression was checked for second-order autocorrelation with the following result: Rho2 = .07, t-value = .46. 2 The estimate deteriorates if D = 1 for the period 11/1980— IV/1983 and zero otherwise. This definition includes the period between March 1980 when the legislation was passed and November 1980 when it was implemented. quences experienced by m em ber banks w hen the general level o f interest rates rise, the expected sign o f the coefficient on the interaction term is positive. Were it not for the phase-in period, the absolute values o f this coefficient and the coefficient on the level o f interest rates w ou ld be the same, indicating that the elimination o f differential reserve requirements com pletely eliminates the differential response of member bank capital values to the level o f the interest rate. Finally, the ratio o f m em ber to nonm em ber demand deposit liabilities is included as a scale variable. The sign o f the coefficient on this variable is ambiguous. However, variation in the size o f members relative to nonmembers can affect the dependent variable (see appendix) and, if the regression does not control for this variation, it can contaminate estimates o f the other coefficients. For the purposes o f this paper, the coefficients on the interest rate and the interaction term are the most interesting. As expected, the coefficient on the interest rate is negative and significant, indicating that a higher interest rate is associated with a low er value o f the dependent variable. The sensitivity o f the dependent variable to interest rate changes is measured by its interest rate elasticity. An estimate o f the average elasticity during the period prior to the Monetary Control Act is given by the product o f the coefficient o f the interest rate and its average level (8.7 percent). In this case, the interest rate elasticity is estimated to be — ,48( = .055 X 8.7). This indicates that a 1 percent increase in the interest rate reduces the share prices o f m em ber relative to non member banks by about 0.5 percent. Implementation o f the Monetary Control Act ap pears to have mitigated this differential effect. The sign o f the interaction term is positive and significant. The coefficient, however, is less in absolute value than the coefficient o f the interest rate. This is not surprising given that the new reserve requirements w ere phased in and that the phase-in w ill continue through 1987. As o f this point in the phase-in (IV/1983), and with the average level o f interest rates held constant at 8.7 percent, the interest rate elasticity is estimated to be — ,30[ = (.020 — .055) X 8.7]. This represents a decline o f about 40 percent in the interest rate sensitivity of the dependent variable. It is important to recognize that this sensitivity is reduced not only because the sensitivity o f m em ber bank share prices to interest rate changes declines but also because the legislation, by imposing uniform reserve requirements on all banks, increases the interest rate sensitivity o f nonmem ber bank share prices. The average level o f interest rates rose to about 13 percent subsequent to the Monetary Control Act. Had the act not been in place, the share prices o f member relative to nonm em ber banks w ou ld have declined by about 24 percent [ = 100 X -.48(13.0 - 8.71/8.7], The legislation, however, tem pered this to a decline o f only 15 percent [ = 100 X -.30(13.0 - 8.7J/8.7]. CONCLUSION The Evidence The reserve requirements im posed on the deposit liabilities o f financial institutions have the properties o f a tax. This tax varies w ith the interest rate and has Table 5 presents the results o f the regression. The variable included to control for differences in the scale differential effects across banks depending on their reserve requirements. An important change in this tax 19 FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 was made in the Monetary Control Act o f 1980. The act im posed uniform reserve requirements across all financial firms by raising reserve requirements for firms that were not members o f the Federal Reserve System, w hile lowering them for mem ber banks. This paper analyzes the legislation’s effect on the relation ship between the interest rate and the stock prices of member and nonmem ber commercial banks. As ex pected, the legislation has significantly reduced the differential effect o f interest rate changes on the rela tive stock price o f these banks. In the process, it has raised the after-tax earnings streams and stock prices o f m em ber banks, other things the same, while low er ing both for nonmem ber banks. Prestopino, Chris J. “Do Higher Reserve Requirements Discourage Federal Reserve Membership?” Journal of Finance (December 1976), pp. 1471-80. APPENDIX Relative Capital Values Let P, D and r represent capital values, deposits and the required reserve ratio as a function o f deposits, respectively, w hile iL and i„ are the lending rate and borrowing rate that are com m on to all banks. If the subscripts M and N indicate values for m em ber or nonmem ber banks o f the terms in the subscript, then: Pm DM [iL r(DJ*iL Ib il V Dm [1 Ib l ^ r(Dx l)] P„ = DN - i„)/iL = Dx(l - iB (iL /i,l. REFERENCES Board of Governors of the Federal Reserve System. cal Digest, 1941-70 and 1970-79. Annual Statisti Brewer, Elijah, et. al. “The Depository Institutions Deregulation and Monetary Control Act of 1980," Federal Reserve Bank of Chicago Economic Perspectives (September/October 1980), pp. 2-23. Cargill, Thomas F„ and Gillian G. Garcia. Financial Deregulation and Monetary Control (Hoover Institution, 1982). Fama, Eugene F. “What's Different About Banks?,’’ Journal of Mon etary Economics (January 1985), pp. 29-39. Gilbert, R. Alton. "Effectiveness of State Reserve Requirements," this Review (September 1978), pp. 16-28. Gilbert, R. Alton, and Jean M. Lovati. “Bank Reserve Requirements and Their Enforcement: A Comparison Across States," this Re view (March 1978), pp. 22-32. Goldberg, Lawrence G., and John T. Rose. "The Effect on Non member Banks of the Imposition of Member Bank Reserve Re quirements — With and Without Federal Reserve Services,” Jour nal of Finance (December 1976), pp. 1457-69. James, Christopher. "An Analysis of Intra-Industry Differences in the Effect of Regulation: The Case of Deposit Rate Ceilings,” Journal of Monetary Economics (September 1983), pp. 417-32. Knight, Robert E. “Reserve Requirements: Part 1: Comparative Reserve Requirements at Member and Nonmember Banks," Fed eral Reserve Bank of Kansas City Monthly Review (April 1974), pp. 3-20. 1980 Financial Institutions Deregulation and Monetary Control. (Commerce Clearing House, Inc., 1980). Pearce, Douglas K., and V. Vance Roley. “The Reaction of Stock Prices to Unanticipated Changes in Money: A Note,” Journal of Finance (September 1983), pp. 1323-33. “Stock Prices and Economic News,” Journal of Business (January 1985), pp. 49-67. 20 P,/PN = — — [ 1 ---* Dv 1 HDm ) rl iA A Change in Relative Scales: d(Ps,/PJ 3(Dm /DJ rtD„. 1 iA dD., l x - [r'(DM d(Dj/DN ) 1 - iA Dv < A Change in the General Level o f Interest Rates: ^ 3i D, di, = di„ = di iL 2 r(DM < 0 )) A Change in the Member Bank Reserve Schedule: a iP M / P J dr(DJ _ D m Dx . 1 . ^ 0 1 - i„/i, A Change in the Lending Rate Relative to the Borrowing Rate: WPy/P-l d(iA) HDm ) D -, Dn (1 - iA )2 < 0 Note that iB must be less than one. An increase in this /iL ratio is consistent with a decline in the spread be tween lending and borrowing rates. Recent Changes in Handling Bank Failures and Their Effects on the Banking Industry R. Alton Gilbert I n SOME o f its public statements in recent years, the Federal Deposit Insurance Corporation (FDIC) has stressed the objective o f prom oting market discipline o f the risks assumed by banks through the influence of uninsured depositors.1 The FDIC has attempted to accomplish this by allowing the uninsured depositors o f some failed banks to suffer losses. In practice, the cases in w hich uninsured depositors have been ex posed to losses involve relatively small banks. As a consequence, the managers o f some relatively small banks claim that they have lost large-denomination deposit accounts to larger banks as large depositors reduce the risk o f losing part o f their deposits by moving their accounts to relatively large banks.2 This paper investigates w hether the FDIC’s actions in recent years indicate a double standard in the treatment of large depositors at large and small banks. Next, the paper analyzes the effects that such a double standard w ould have on the operation o f the banking system. Finally, it investigates w hether depositors now act as though they perceive an increase in the risk o f holding large-denom ination deposits at small banks over holding them at large banks. R. Alton Gilbert is an assistant vice president at the Federal Reserve Bank of St. Louis. Laura A. Prives provided research assistance. 'Federal Deposit Insurance Corporation (1983) and Isaac (1983). 2Hill and Finn (1984) and King (1984). FDIC ACTIONS IN BANK FAILURE CASES This section presents a brief description o f the FDIC’s procedures in disposing o f the assets and de posit liabilities o f insured banks that fail. A knowledge o f these procedures is necessaiy to understand the effect o f recent FDIC actions on the risks assumed by large depositors at banks o f different size. Deposit Payoff A commercial bank is officially declared a failed bank by its chartering agency — the Com ptroller o f the Currency for a national bank, the state banking au thority for a state-chartered bank. The FDIC becomes the receiver o f a federally insured bank that fails, with authority to dispose o f the assets and to pay off the creditors. One type o f action the FDIC can take as receiver o f a failed bank is called a deposit payoff. The FDIC makes payments to each depositor, up to the insurance limit, as soon as the records o f deposit accounts can be compiled. Depositors with accounts over the insur ance limit becom e general creditors o f the failed bank for the amount o f their deposits in excess o f the insurance limit. They receive payments on the unin sured portions o f their deposits as the FDIC liquidates the assets o f the failed bank. W hether they receive full payment on their uninsured deposits depends on the liquidation value o f these assets. 21 F E D E R A L R E S E R V E B A N K O F ST. LO UIS JUNE/JULY 1985 Purchase and Assumption Transactions Modified Payout Procedure For the FDIC, there are disadvantages to handling the receivership o f a failed bank through a deposit payoff. Banking services are tem porarily disrupted, even for the fully insured depositors, w ho generally must wait a few days to receive their funds. For the uninsured depositors, even if they eventually receive full payment, the delay can throw a wrench into the financing o f their activities. Also, the acquisition o f the failed bank’s assets may be more valuable to another bank than to the FDIC, especially if the other bank could continue to operate the failed bank as a going concern. In choosing between a deposit payoff and a P&A transaction, the FDIC has had to decide which o f the following objectives it w ould give the greatest weight: The FDIC prefers to handle most bank failure cases through what are called purchase and assumption (P&A) transactions. In these transactions, all o f a failed bank’s deposit liabilities are assumed by another bank, which also purchases some o f the failed bank’s assets. The FDIC initiates a P&A transaction by soliciting bids from other banks for the purchase o f assets and the assumption o f deposit liabilities o f a failed bank. The FDIC specifies that an interested bank must assume all deposit liabilities and acquire assets considered to be o f good value (i.e., excluding loans and debt instru ments that are not likely to be paid in full). Additional cash w ill be provided by the FDIC if the value o f the assets o f the failed bank offered for purchase is less than the deposit liabilities to be assumed. Banks that are interested in such a package o f assets and liabili ties bid for it in terms o f a purchase premium. The actual cash payment from the FDIC equals the liabili ties o f the failed bank, minus the value o f the assets o f the failed bank purchased by the bank w ith the w in ning bid, less the purchase premium bid by that bank. In deciding between a deposit payoff or a P&A transaction, the FDIC uses a cost test. It estimates its cost under both a deposit payoff and a P&A transac tion, based on the bid o f the highest purchase p re mium. The FDIC generally w ill accept the highest bid for the P&A transaction if its net cost is low er than the estimated costs o f a deposit payoff. These estimates are not very precise, and the FDIC has tended to use the P&A method except in situations in which: 1. there is virtually no interest by other banks in acquiring the failed bank, or 2. fraud or other circumstances, such as contingent liabilities, make it difficult to estimate losses and, therefore, to apply the cost test.3 3Federal Deposit Insurance Corporation (1984), pp. 83-88. 22 1. to avoid disruption o f banking services, or 2. to prom ote market discipline by uninsured de positors o f the risks assumed by banks. If the FDIC tends to handle bank failure cases through deposit payoffs, uninsured depositors must assume the risk o f losses if their banks fail. In response, the uninsured depositors might put pressure on their banks to limit risk. But deposit payoffs, as w e have seen, disrupt banking services. The use o f P&A transactions prevents disruptions o f banking services. This alternative, however, may give uninsured depositors the impression that they are not exposed to risk o f loss w hen their banks fail. As a consequence, they w ou ld not attempt to restrain the risks assumed by their banks. To avoid the limitations o f both procedures, the FDIC announced, in Decem ber 1983, that it w ou ld use a new procedure for disposing o f assets and deposit liabilities in some bank failure cases. The new “m od ified payout procedure” was adopted to give the FDIC more flexibility in m inim izing disruption o f banking services, w hile exposing uninsured depositors to the risk o f losses on their deposits.4 When a bank failure is handled through the m od ified payout procedure, the FDIC makes full payments to the insured depositors and partial payments to the large depositors on the uninsured portions o f their deposits; the partial payments are based on an FDIC estimate o f the proceeds from the liquidation o f the assets o f the failed bank. If recoveries on the assets eventually exceed the initial estimate, the uninsured depositors receive additional payments; if the pro ceeds from liquidating those assets fall short o f the initial payment, the FDIC absorbs the loss. The partial payment disrupts the activities o f uninsured deposi tors less than the traditional deposit payoff did. In some cases handled under the m odified payout procedure, the insured liabilities o f a failed bank are assumed by another bank. This arrangement prevents a disruption o f banking services for depositors with full federal insurance. The procedures for arranging this deposit assumption are similar to the procedures in a traditional P&A transaction. The FDIC solicits bids “Federal Deposit Insurance Corporation (1983), pp. III-4— III-6. FEDERAL RESERVE BANK OF ST. LOUIS for the purchase o f some o f the assets o f the failed bank and the assumption o f the fully insured deposit liabilities. The FDIC provides cash to cover a gap between the value o f assets purchased and the fully insured deposit liabilities assumed, minus any pur chase premium. The FDIC then receives the remain ing assets and makes a partial payment to the unin sured depositors. This approach to handling bank failure cases has similarities to both the deposit payoff and P&A transaction procedures. IS THERE A DOUBLE STANDARD IN THE FEDERAL INSURANCE OF LARGE DEPOSIT ACCOUNTS? The official limit on deposit insurance coverage, currently the first $100,000 for each depositor at each depository institution, is the same for insured banks of all sizes. There is circumstantial evidence, however, that the FDIC provides large depositors at a few o f the nation’s largest banks greater protection from loss than large depositors at smellier banks. There is no official statement o f this double standard by the FDIC; if it exists, it must be inferred from the FDIC’s actions in bank failure cases. FDIC Actions in Bank Failure Cases Before 1982 Until 1982, every bank failure involving assets greater than $100 million had been handled through P&A transactions, thus protecting the uninsured deposi tors from any losses.’ From 1968, w hen the FDIC adopted its current procedures for P&A transactions, through 1981, only 32 o f the 108 bank failure cases w ere handled through deposit payoffs. These 32 banks, which had average total assets o f $10.4 million, had relatively few deposit accounts in excess o f the insurance limit. The other 76 had average total assets o f $171 million.6These FDIC actions could have con vinced most large depositors that, in effect, they had com plete insurance coverage o f their deposit ac counts, even if their accounts exceeded the officially stated insurance limit. Greater Emphasis on Market Discipline by Large Depositors and the Penn Square Case JUNE/JULY 1985 about a lack o f market discipline im posed by large depositors on the risks assumed by their banks. This concern was stimulated by the view that various forms o f deregulation gave bankers greater freedom to as sume more risk.7 The response o f the FDIC to the failure o f the Penn Square Bank o f Oklahoma City in 1982 reflected, in part, an intention to increase the degree o f market discipline by large depositors. The FDIC closed the Penn Square Bank, w hich had total assets o f $517 million, and paid off each depositor up to the federal insurance limit. A recent history o f the FDIC mentions two reasons for closing the Penn Square Bank and paying off the depositors, rather than protecting the uninsured de positors through a P&A transaction. First, it was not possible at that time for the FDIC to determine the costs o f alternative methods o f handling the case. Second, the FDIC was concerned that, if the large depositors o f the Penn Square Bank w ere protected from losses, market discipline o f the risks assumed by banks through the influence o f large depositors w ould be eroded. The FDIC concluded that paying off the depositors o f the Penn Square Bank, up to the insur ance limit, and allowing the uninsured depositors to suffer losses, w ould cause investors to perceive a greater risk in holding large-denomination deposits.8 The handling o f the Penn Square case indicated that, in order to prom ote market discipline by large depositors, the FDIC was w illing to apply the deposit payoff procedure in the failure o f a much larger bank than it had in the past. This case did not reveal, however, w hether the FDIC w ould put a limit on the size o f a failed bank that w ou ld be handled through a deposit payoff. Thus, the FDIC’s actions in this case did not indicate w hether the risks o f holding large deposit accounts had risen more for those with ac counts at small banks or large banks. Initiation o f the Modified Payout Procedure The next major action by the FDIC to induce unin sured depositors to restrain the risks assumed by their banks was the adoption o f the m odified payout proce dure. As m entioned above, an important objective for adopting this procedure was to expose uninsured In the early 1980s, the FDIC became concerned 5Federal Deposit Insurance Corporation (1984), p. 93. H'his concern about a lack of market discipline is expressed in Federal Deposit Insurance Corporation (1983). 6lbid., table 4-2, p. 65. 8Federa! Deposit Insurance Corporation (1984), pp. 97-98. 23 JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS depositors to some risk o f loss if their banks fail, while m inim izing the disruption to banking services. The first bank failures handled under the m odified payout procedure occurred in March 1984. From March through May 1984, the FDIC used the new m odified payout procedure in nine bank failure cases. In two o f those cases, the banks w ere closed and the FDIC made payments to all depositors. The only dif ference between these two cases and the usual de posit payoff case was that the uninsured depositors received partial payments when their banks were closed, instead o f receiving any payments after the FDIC liquidated the assets. In the other seven cases handled under the m od ified payout procedure, other banks assumed the fully insured deposit liabilities o f the failed banks and pur chased some o f their assets. Since other banks w ere interested in bidding for the assets and fully insured deposit liabilities o f the§e seven banks, it is likely that their uninsured depositors w ould also have been pro tected from losses through P&A transactions if the FDIC had not adopted the m odified payout proce dure. All o f the seven bank failure cases handled under the m odified payout procedure, with assumption o f fully insured deposit liabilities by other banks, in volved relatively small banks. The total deposits o f those seven banks ranged from $16 m illion to $116 million, with a mean o f $54 million. Their uninsured deposits on average w ere $1.6 million. The Continental Illinois Crisis The rapid withdrawal o f foreign deposits from the Continental Illinois National Bank, Chicago, created a financial crisis for that bank in May 1984. The FDIC, the Federal Reserve, and the Com ptroller o f the Cur rency became concerned about the effects that the failure o f Continental w ould have on other depository institutions and econom ic activity in general. These agencies issued a joint news release on May 17,1984, that described a program o f assistance for Continen tal. That joint news release includes the following statement: In view of all the circumstances surrounding Conti nental Illinois Bank, the FDIC provides assurance that, in any arrangements that may be necessary to achieve a permanent solution, all depositors and other general creditors of the bank will be fully protected and ser vices to the bank’s customers will not be interrupted. This statement indicates that, although the FDIC wishes to induce large depositors to restrain the risks 24 assumed by their banks, there is an upper limit on the size o f banks at which large depositors are subject to losses. From June 1984 through May 1985, the FDIC han dled six more bank failure cases by arranging for the assumption o f the fully insured deposit liabilities by other banks, but limiting payments on uninsured de posits to the proceeds from liquidating the assets of the failed banks. Total deposits o f those six banks range between $4 m illion and $46 m illion (a mean of $26 million), with average uninsured deposits of about $400,000/’ A Review o f FDIC Actions in Recent Years The actions o f the FDIC since mid-1982 reveal the following pattern: T o prom ote market discipline by large depositors, the FDIC is willing to close a failing bank with total assets as large as $500 million. In practice, the m odified payout procedure, which was adopted to prom ote market discipline by large deposi tors, has been used in the failure o f a few relatively small banks. Banks as large as Continental Illinois appear to be exempt from this policy. This combina tion of FDIC actions may im ply that the risk o f holding deposits in an account that exceeds the federal insur ance limit has increased in recent years, unless that account is at one o f the largest banks in the nation. IMPLICATIONS FOR THE OPERATION OF THE BANKING SYSTEM If large depositors think they are protected from losses by holding their funds at relatively large banks, they w ill have no incentive to m onitor the risks as sumed by these banks or to put pressure on the management o f these banks to restrain risks. Thus, by protecting depositors at relatively large banks from losses, the FDIC may have reduced the restraints on risks assumed by relatively large banks. Actions that favor uninsured depositors at relatively large banks also may have implications for trends in the nation’s banking structure. The share o f the na tion’s banking assets at a few o f the relatively large banks may rise over time, as large depositors shift their funds to the relatively large banks to reduce risks. 9ln some of these six cases, the large depositors did not receive partial payments when the banks failed, because it was difficult for the FDIC to estimate recovery on the assets it assumed. FE D E R A L R E S E R V E B A N K OF ST. LO UIS HAVE DEPOSITORS RESPONDED TO THE DIFFERENTIAL TREATMENT OF LARGE AND SMALL BANKS? Before concluding that the recent actions o f the FDIC in bank failure cases have the implications for the banking system discussed above, one must deter mine w hether depositors have responded to what appears to be differential treatment o f large and small banks. There are various reasons w hy the events d e scribed above might not affect the behavior o f unin sured depositors. Large depositors may have believed for some time that the FDIC w ould not allow a bank the size o f Continental Illinois to fail. The FDIC had acted in the past to prevent the failure o f relatively large banks with total assets smaller than those of Continental.1 Thus, the announcement o f the deposit 0 guarantee for Continental in May 1984 may have come as no surprise. Alternatively, holders o f uninsured deposits may continue to have confidence in their ow n banks de spite the increased risk o f keeping their accounts at small banks. Or, they might not be aware o f the im pli cations o f FDIC actions in recent bank failure cases. There are two potential pieces o f evidence that w ould support the view that depositors consider the risk o f holding uninsured deposits at small banks to have risen relative to the risk o f holding deposits of similar size at large banks. First, interest rates that small banks offer to attract large-denomination d e posits must rise relative to the interest rates offered by large banks. Second, the share o f total time deposits at all commercial banks in accounts above the insurance limit must rise at relatively large banks and decline at small banks, as depositors shift their large-denomina tion deposit accounts to relatively large banks. Both of these patterns w ould have to begin after mid-May 1984, when the FDIC announced the deposit guaran tee o f Continental Illinois Bank. THE EVIDENCE Data on th e in teres t rates p a id on la rgedenomination time deposits are not available for rela tively small banks. Consequently, the observations are limited to those for the allocation o f large time de posits among large and small banks. 1 See the Federal Deposit Insurance Corporation (1984), pp. 89-97. 0 JUNE/JULY 1985 Data from Weekly Reporting Banks There is no official list o f banks that are too large to fail. As an approximation to the group o f banks that may have such status, this paper uses the 30 largest banks in the nation.” Small banks are identified as those smaller than w eekly reporting banks (which include all comm ercial banks with total assets o f $1.4 billion or more as o f Decem ber 31,1982). Chart 1 does not indicate a sustained pattern o f decline in the share o f large-denomination time de posits at small banks or a rise in the share at relatively large banks after m id-M ay 1984. The share o f largedenomination time deposits at the small banks did decline from almost 40 percent in the first week o f May 1984 to about 37 percent in the last w eek in June o f last year. That change might reflect an initial response by depositors to the handling o f the Continental Illinois situation by the FDIC. In contrast, that decline might reflect a seasonal pattern; the share o f largedenomination time deposits at small banks declined between the same weeks in 1982. Whatever the cause o f that dip, it was more than reversed by October of last year, and the share o f large time deposits at small banks continued to rise through May 1985. The share of large-denomination deposits at small banks de clined in June 1985, as it had in June o f 1982 and 1984.1 2 Data on Small Banks from the Report o f Condition Observations in chart 1 may suffer from several measurement problems. First, the banks with total assets just below $1.4 billion are included together with much smaller banks. A finer breakdown o f banks by asset size may be necessaiy to detect an outflow o f large-denomination time deposits from small banks. Second, some o f the time deposits in denom ina tions o f $100,000 or more are in accounts o f exactly "In September 1984, the Comptroller of the Currency, C.T. Conover, was reported in the press as saying that the federal regulators would not allow the largest 11 banks to fail; later, however, he denied stating any cut-off figure for banks too large to fail. See Trigaux (1984). 1 The patterns in chart 1 reflect differential effects of the authorization 2 of money market deposit accounts (MMDAs) at large and small banks. Large time deposits at all commercial banks declined sharply after the announcement that MMDAs would be available in midDecember 1982. These deposits had peaked in October 1982, but by May 1983, had declined by about $50 billion. During the same period, large time deposits at small banks rose by about $13 billion. Thus, the small banks do not seem to have been affected by the substitution between large-denomination deposits and MMDAs in the same way as the larger banks. 25 JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS Chart 1 Share of L a rg e -D e n o m in a tio n Time Deposits at Large a n d Sm all B a n k s Percent Percent 100 80 60 40 20 1982 1983 1984 1985 N O T E : L a r g e t im e d e p o s i t s a r e in d e n o m i n a t i o n s o f $ 1 0 0 , 0 0 0 o r m o r e . L a r g e b a n k s a r e t h e w e e k l y r e p o r t i n g b a n k s , a n d s m a l l b a n k s a r e b e l o w t he s i z e o f w e e k l y r e p o r t e r s . T h e t o p 3 0 b a n k s a r e t he l a r g e s t c o m m e r c i a l b a n k s , in t ot al a s s e t s , a s o f D e c e m b e r 1984. Late st d a t a plotted: J u n e $100,000, and, therefore, are fully insured. Thus, the percentage o f time deposits in denominations of $100,000 or more overstates the percentage o f time deposits in accounts that are only partially insured. The third possible measurement problem is that many o f the small banks are subsidiaries o f large banking organizations. Uninsured depositors may be less concerned about possible losses o f their deposits at a relatively small bank if it is a subsidiary o f a large banking organization. The relevance of these possible measurement prob lems can be investigated with data from the Report o f Condition o f each commercial bank. Table 1 presents the percentage o f large-denomination time deposits in the banking system at groups o f relatively small banks in various size categories. T o eliminate banks that are subsidiaries o f relatively large banking organi 26 zations, the banks in table 1 are in organizations with total banking assets less than $1 billion. The issue o f w hich large-denomination time d e posits are only partially insured is m ore difficult to settle. In attempting to exclude time deposits in d e nominations o f $100,000 or m ore that are fully insured, the best approach available with the existing data is to exclude from the calculations those banks with brok ered deposits in denominations o f $100,000 or less, which are called "retail brokered deposits.’’ Deposit brokers typically break down the funds they place at an individual bank into units o f $100,000 or less for their individual investors, so that the deposits o f each investor are fully insured. The banks with retail brok ered deposits, therefore, are the ones likely to have the largest proportion o f time deposits in denominations of exactly $100,000. FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 1 Percentage of Large-Denomination Time Deposits in Various Size Banks1 Bank size category (millions of dollars) Number of banks 1985 1984 March June September December March 0.19% 0.14% 0.15% 0.16% 0.17% $10 to 25 1,450 1.13 1.14 1.18 1.21 1.33 $25 to 50 1,464 2.60 2.58 2.64 2.66 2.89 $50 to 100 966 3.94 3.89 3.98 3.93 4.28 $100 to 300 584 5.63 5.63 5.80 5.79 6.11 $300 to 500 93 2.37 2.38 2.46 2.45 2.50 Under $10 $500 to 1,000 Total 422 60 2.33 2.36 2.40 2.53 2.59 5,039 18.14 18.13 18.62 18.74 19.89 'The same banks are included in each size group as of each of the five Report of Condition dates. As of each date, these banks reported no retail brokered deposits. These banks are assigned to the same size group as of each date, based on their total assets as of March 1984. For banks in each group, the sum of their time deposits in denominations of $100,000 or more are calculated as percentages of large-denomination time deposits as of the same date at all commercial banks that reported no retail brokered deposits. Table 2 provides indirect evidence on the extent to which the time deposits in denominations o f $100,000 or more exceed the insurance limit at banks with no retail brokered deposits. The FDIC collected data through June 1981 on various types o f deposit ac counts that exceeded the insurance limit. As o f June 1981, the percentages o f time deposits in denom ina tions o f $100,000 or m ore are only slightly higher than the percentages in accounts that exceeded the insur ance limit. Thus, as o f June 1981, ve iy high percent ages o f time deposits in denominations o f $100,000 or more w ere only partially insured. Also, for banks o f comparable size, the percentages o f time deposits in denominations o f $100,000 or m ore in June 1984 are similar to the percentages in June 1981. These com parisons o f observations in table 2 provide a basis for concluding that high percentages o f the time deposits in denominations o f $100,000 or more, as o f June 1984, w ere only partially insured. Data are not available, however, to provide direct evidence on this issue. Data on retail brokered deposits for all federally insured commercial banks are not available before March 1984. Beginning with the quarterly Report of Condition in March 1984, each bank reports the total dollar amount o f all brokered deposits and o f retail brokered deposits. Data based on the Report o f Condi tion, therefore, are limited to the period since March 1984. The purpose o f the calculations presented in table 1 is to determine w hether banks in various size groups have increased or decreased their share o f largedenomination time deposits in the banking system. Banks in each o f the size groups have the following characteristics: First, each bank filed a Report o f Con dition on all five dates. Second, each bank reported no retail brokered deposits on each date. Third, each bank is assigned to one size class for all five dates, based on its total assets as o f March 1984. Thus, each size group includes the same banks for each o f the Report o f Condition dates. The numerator o f each percentage in table 1 is the sum o f time deposits in denominations o f $100,000 or more for a given group o f banks, as o f a Report of Condition date. The denom inator is the sum o f largedenomination time deposits o f all commercial banks as o f the same date, excluding those banks that re ported retail brokered deposits. As o f March 1984, these 5,039 banks accounted for about 18 percent o f large-denomination time deposits o f the banking system. By March 1985, that percentage rose to almost 20 percent, and the share o f largedenomination time deposits rose for each o f the seven groups o f banks. Thus, the evidence in chart 1 and table 1 are consistent: the share o f large-denomination time deposits at relatively small banks is higher in early 1985 than a year earlier, before the announce ment o f the federal guarantee o f all deposit liabilities at the Continental Illinois bank. 27 F E D E R A L R E S E R V E B A N K OF ST. LO U IS JUNE/JULY 1985 Table 2 Average Percentages of Time Deposits in Large-Denomination Accounts Bank size category (millions of dollars of total deposits) Percentage of time deposits in accounts larger than $ 100,000, June 1981* Percentage of time deposits in accounts of $100,000 or more June 1981 June 19842 Under $10 14.9% 17.2% 21.1% $10 to 25 16.9 18.2 20.8 $25 to 50 20.4 21.4 22.7 $50 to 100 25.6 26.9 26.6 $100 to 500 34.8 37.3 34.1 $500 to 1,000 43.2 46.6 36.7 'The numerators and denominators of these percentages include time deposits of individuals, partnerships, and corporations and public funds invested in time and savings deposits at commercial banks. 2Banks included in the calculations for June 1984 have no retail brokered deposits; they are in banking organizations with total banking assets less than $1 billion. CONCLUSIONS The Federal Deposit Insurance Corporation (FDIC) has been putting greater emphasis in recent years on increasing the incentives for large depositors to re strain the risks assumed by their banks, by allowing the uninsured depositors o f some failed banks to suffer losses. FDIC actions designed to promote mar ket discipline by large depositors, however, have af fected primarily relatively small banks. In contrast, the FDIC guarantee o f all deposit liabilities o f the Conti nental Illinois National Bank in May o f last year indi cates that the large depositors o f a few o f the nation’s largest banks may have no risk o f losses if their banks experience large reductions in the value o f their assets. This contrast in treatment o f large and small banks might be expected to induce large depositors to shift their accounts to a few o f the nation’s largest banks. The data available to test this hypothesis have some limitations. The most appropriate conclusion, given the nature o f the data, however, is that large deposi 28 tors have not shifted their accounts from small to large banks since the announcement o f the FDIC guarantee o f all deposit liabilities o f Continental Illinois. REFERENCES Federal Deposit Insurance Corporation. “Deposit Insurance in a Changing Environment” in A Report to Congress on Federal De posit Insurance (April 1983). ----------------The First Fifty Years: A History of the FDIC, 1933-1983 (Washington, D.C., 1984). Hill, C. Christian and Edwin A. Finn. “Confidence Crisis: Big Deposi tors’ Runs on Beleaguered Banks Speed the Failure Rate,” Wall Street Journal, August 23, 1984. Isaac, William M. Address before the Management Conference of the National Council of Savings Institutions, New York, December 6,1983. King, A. J. “Treat Small Banks, Continental Similarly,” American Banker (May 31,1984), pp. 4, 12. Trigaux, Robert. "Comptroller Trying to Dispel Reports of ‘Guaran tees’ for Top 11 Banks Only,” American Banker (September 28, 1984), p. 3. Are Weighted Monetary Aggregates Better Than Simple-Sum M l? Dallas S. Batten and Daniel L. Thornton T A HE past 10 years have been marked by financial innovation and deregulation, much o f which has blurred the distinction between transaction and sav ings deposits. Traditional non-interest-bearing trans action deposits now pay explicit interest like savings deposits, w hile a number o f savings-type deposits with limited transaction characteristics have been developed. A number o f analysts believe that these financial developments have altered significantly the relation ship between M l growth and the growth o f GNP, rendering the narrow monetary aggregate less useful as an intermediate target for m onetaiy policy.' Others have objected on broader grounds, arguing that these innovations illuminate the problem o f simply adding up various financial assets (currency, dem and de posits, NOW accounts, etc.) to obtain a "simple-sum” monetary aggregate. They argue that various assets have different degrees o f “m oneyness” — that is, the Dallas S. Batten is a research officer and Daniel L Thornton is a senior economist at the Federal Reserve Bank of St. Louis. Paul G. Chris topher and Rosemarie V. Mueller provided research assistance. 'The Federal Open Market Committee was so concerned by these developments that it altered the relative weights given to M1 and the broader monetary aggregates several times during the 1981-82 period in making its policy recommendations and suspended the use of M1 as an intermediate policy target in fall 1982. Furthermore, some analysts have been so concerned that M1 is no longer a useful target of monetary policy that they have suggested a return to the Keynesian system of interest rate targets or a reliance on a broader simple-sum monetary aggregate, like M2, M3 or some measure of credit, as an intermediate target. Still others have suggested that the Fed target directly on nominal GNP (though the procedures for pursuing this target are seldom discussed in detail). See Thornton (1982,1983). Simple-sum M1 was re-introduced as an intermediate policy target in 1984; see Hafer (1985). These other suggestions have been investigated elsewhere. The use of interest rates as an intermediate policy target is predicated on the existence of a liquidity effect, which has been shown to be short lived and weak. See Brown and Santoni (1983) and Melvin (1983). For empirical evidence on M1 and M2, see Batten and Thornton (1983) and on the broader debt measure, see Hafer (1984). monetary services that each asset provides — so that the dollar amount of each asset should be w eighted by its degree of moneyness in obtaining a suitable m one tary aggregate. Such an aggregate presumably should have a closer and more predictable relationship with econom ic activity and may be affected less by financial innovations. The most novel and innovative sugges tions have com e from individuals w ho have con structed weighted m onetaiy aggregates based on al ternative theoretical considerations. Tw o recent and popular innovations along these lines com e from W il liam Barnett (1980) and Paul Spindt 11985).A central issue now is w hether w eighted m onetaiy aggregates are better intermediate policy targets than simple-sum aggregates like M l. A necessary condition for using a m onetaiy aggregate as an intermediate policy target is that there be a close and predictable relationship between the m onetaiy aggregate target and the objectives o f econom ic policy/' Thus, if an aggregate can be found that has a closer and more predictable link to econom ic activity, it could be useful in conducting countercyclical stabilization policy.4 The purpose o f this article is threefold. First, we review briefly the important issues associated with constructing weighted and simple-sum m onetaiy ag gregates and discuss the alternatives suggested by Barnett and Spindt. Second, w e compare and contrast these weighted m onetary aggregates with simple-sum M l. Finally, w e investigate whether there is a more stable and predictable relationship between the alter 2Earlier work along these lines includes Chetty (1967) and Ham burger (1966). 3 The strength of the relationship between the ultimate goals of policy and the intermediate policy target is only one of the criteria for evaluating a monetary target. "This should not be interpreted to imply that monetary policy can be used successfully for short-run economic stabilization. This is merely a necessary condition; it is not sufficient. 29 FEDERAL RESERVE BANK OF ST. LOUIS natives proposed by Barnett and Spindt and GNP, than between simple-sum M l and GNP. W e investigate this by examining the behavior o f the incom e velocity of each o f these aggregates. THE MOTIVATION FOR WEIGHTED AGGREGATES JUNE/JULY 1985 criteria.7 Perhaps the most frequently used criterion was the closeness o f the relationship between a partic ular monetary aggregate and GNP." The Effect o f Financial Innovations The difficulty in distinguishing between m oney and non-money assets has been exacerbated by financial innovation and deregulation. Several savings-type as sets with lim ited transaction characteristics have been d e v e lo p e d (e.g., m o n e y m arket m utu al funds (MMMFs), m oney market deposit accounts (MMDAs) and automatic transfer services (ATS)) and mediumof-exchange assets now pay explicit interest (e.g., NOWs and Super NOWs). Additionally, there have been a number o f other innovations that have in creased the substitutability between medium-ofexchange and non-medium-of-exchange assets, such as overnight repurchase agreements (REPOs) and con tinuous com pounding o f interest on savings-type d e posits.9 Hence, the distinction between transactionand savings-type assets has been blurred even more. Monetary theory has em phasized two different, but not mutually exclusive, functions o f money: a medium o f exchange and a store o f wealth. The medium-ofexchange function was em phasized in the work of Fisher (1911), w hile the store-of-wealth motive was em phasized by Pigou (1917), Marshall (1923) and Keynes (1936). It has been recognized for some time that different financial assets perform these functions to different degrees. For example, currency and de mand deposits are both generally acceptable as media o f exchange, but are not perfect substitutes for this purpose in all transactions. Furthermore, these assets bear no explicit interest and, as a consequence, are poor stores o f wealth relative to interest-bearing sav ings and time deposits o f equal risk. The Role o f Index Numbers Because assets such as time and savings deposits cannot be used directly in exchange, it was comm on to define m oney to include only medium-of-exchange assets. It was not until Friedman (1956), Friedman and Meiselman (1963) and Friedman and Schwartz (1970) em phasized m oney’s role as a “temporary abode of purchasing pow er” (i.e., a temporal bridge between the sale of one item and the purchase o f another), that it became comm on to consider broader monetary aggregates that included non-medium-of-exchange assets/' If different assets have different degrees o f m oney ness, w e may wish to aggregate (add) them with re spect to this hom ogeneous characteristic. This point can be made more clearly with a physical example. A ton o f coal, a kilowatt o f electricity and a barrel o f oil are not hom ogeneous in terms o f their volumes or weights and, hence, cannot be aggregated in terms o f these measures. If, however, w e are concerned with their energy equivalences, measured say by BTUs, they can be thought o f broadly as hom ogeneous and can be aggregated in terms o f their BTU equivalence. The Once the medium-of-exchange line o f demarcation between m oney and non-money assets was breached, however, it became difficult to isolate any other char acteristics that differentiate m oney from non-money assets.'1As a result, many economists defined m oney as that group o f assets that satisfied some empirical 'Although not all of the studies have employed the same empirical criteria, many have focused on the relationship between the pro posed monetary aggregate(s) and economic activity. Furthermore, not all agree that money can be defined empirically, e.g., Mason (1976). frequently, the assets considered had to satisfy an auxiliary condi tion, for example, they must be “gross substitutes.” See Friedman and Schwartz (1970) or Friedman and Meiselman (1963). According to Laidler (1969), the debate about whether non-mediumof-exchange assets are money dates back, at least, to the Napole onic wars. 6 Some characteristics that have been used include liquidity, substi tutability between non-medium-of-exchange and pure medium-ofexchange assets, and the strength and stability of the relationship between a composite of various financial assets and nominal in come. Additionally, Pesek and Saving (1967) have argued that, since money has its primary effect on the economy through a wealth effect, an asset’s moneyness should be determined by the extent to which it is part of society’s net wealth. See Laidler (1969) for a discussion of this point. 30 9 The impact of these innovations on the substitutability between medium-of-exchange and non-medium-of-exchange assets can be made clear via an example. At one time, it was common for deposi tory institutions to compound interest quarterly on savings and time deposits, so that interest was paid only on balances on deposit on the day of compounding. Such practices severely limited the advan tage of these accounts over demand deposits as temporary abodes of purchasing power, since the interest income gain from temporar ily switching from demand deposits to savings deposits could be lost if the transaction had to be made prior to the quarterly compounding date. Other changes that permitted an easier transfer between medium-of-exchange and non-medium-of-exchange assets would have a similar effect. JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS same is true for aggregating financial assets, but, since they are expressed in dollars, it may seem more natu ral simply to add dollar amounts o f assets that have a high degree o f moneyness, how ever defined. This is the rationale for the construction of simple-sum m on etary aggregates. Unfortunately, adding dollar amounts o f assets is not the same as aggregating them by a homogeneous measure o f their moneyness. As the dollar amounts of various components change through time, they may represent different levels or degrees o f moneyness. Conversely, the same dollar value o f the aggregate com posed o f different dollar values o f its various com ponents may not represent the same level o f monetary services. Consequently, the dollar (simple-sum) aggre gate may misrepresent the amount o f such services provided. Index numbers can be used to aggregate assets by a homogeneous characteristic. Conceptually, they en able the construction o f an aggregate based on this characteristic so that changes in the index reflect only changes in some quantitative measure o f this charac teristic. It is not surprising, therefore, that both Barnett and Spindt use index aggregation to construct their alternative w eighted monetary aggregates. (The assets included in simple-sum M l, Barnett’s broadest monetary aggregate (MSI4) and Spindt’s aggregate (MQ) appear in the insert on this page."’) Medium-of-Exchange Assets and the Definition of Monetary Aggregates SimpleSum M1 MQ MSI4 X X X X X X X X X X X X X X X X X X X X X Medium-of-Exchange Assets Currency Travelers checks Demand deposits Other checkable deposits Credit union share draft accounts MMDAs MMMFs Savings deposits subject to telephone transfer Non-Medium-of-Exchange Assets Savings deposits not subject to telephone transfer Small time deposits REPOs Eurodollar deposits Large time deposits U. S. savings bonds Short-term Treasury securities Commercial paper Bankers acceptances X X X X X X X X X Monetary Services Index (MSI) Barnett has developed a number o f monetary aggre gates based on the idea that the essential function of m oney is to bridge the temporal gap between the sale o f one item and the purchase o f another. Assets that serve this purpose must be easily and quickly convert ible into and out o f medium-of-exchange assets. Fol lowing a suggestion o f Friedman and Schwartz (1970) — see Barnett and Spindt (1982) — Barnett extends the approach o f estimating the substitutability between non-medium-of-exchange assets and a pure mediumof-exchange asset em ployed by Chetty (1969), Ham burger (1966) and others. Specifically, he applies index number theory to construct indexes o f financial assets that reflect the total utility, relative to some base pe- 1 Other monetary service indexes (MSI) include the assets in simple0 sum M1, M2 and M3. We ignore these here because MSI4 is the only MSI that has an intuitively appealing rationale, given the asset motive on which it is based. In particular, it attempts to extract the “moneyness” from a broad range of financial assets. In contrast, the narrower MSI are constrained by the assets arbitrarily included in each. riod, attributable to the m onetary services obtained from these assets." This approach can be easily understood by thinking o f assets that provide monetary services as being on a continuum with pure medium-of-exchange assets (currency) at one end and "pure’’ store-of-wealth as sets at the other. The pure medium-of-exchange as sets earn no interest and are useful only as a medium- "The construction of these aggregates need not be based solely on a utility maximization approach. If it is based on other objective functions, however, its interpretation is altered. Originally, Barnett called these aggregates “Divisia monetary aggregates” because a Divisia index was used to construct them. The Federal Reserve Board, under whose auspices these aggre gates were originally constructed and are still maintained, has recently undertaken a substantial revision to correct inconsistencies and errors in the original computer programs and data, and to incorporate new data not readily available at the time these aggre gates were initially constructed; see Farr and Johnson (1985). The Divisia index is no longer used to construct these aggregates. Consequently, they are no longer referred to as Divisia monetary aggregates but are now called "monetary services indexes” (MSI). Since the data reported here reflect these recent changes, this new terminology is adopted here as well. 31 JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS of-exchange.'2 The pure store-of-wealth assets earn £ market interest rate but are not useful as a temporary abode o f purchasing power, although they may be used to transfer purchasing pow er over longer periods o f time. Consequently, the latter group o f assets pro vides no monetary services by this criterion. The as sets that fall between these extremes yield monetary services greater than zero but less than those o f the pure medium-of-exchange assets. contrast, the velocity o f the MSI and simple-sum M l can change even if there is no change in their turnover rates. Hence, w e should expect to see a more stable relationship between M Q and GNP.1 4 Simple-Sum M l The MQ Measure By weighting each com ponent equally, simple-sum aggregates im plicitly assume that each com ponent is a perfect substitute for the others in providing m one tary services. Furthermore, the narrow aggregate, simple-sum M l, excludes both non-m edium -ofexchange assets and some assets with lim ited transac tion characteristics like MMMFs and MMDAs. The br oader simple-sum aggregates, like M2, M3 and the Fed’s broadest measure, total liquidity (L), include larger amounts o f non-medium-of-exchange assets. Consequently, these broader simple-sum aggregates may misrepresent significantly the monetary services provided by including non-medium-of-exchange as sets, which provide relatively low levels o f monetary services, on an equal footin g w ith m edium -ofexchange assets, which provide relatively high levels o f monetary services. Spindt’s w eighted monetary aggregate, MQ, is an index o f transaction assets w hose weights are based on each asset’s turnover, along lines originally sug gested by Fisher 11922). This measure is based on a pure transaction approach to m oney and, thus, marks a clear departure from the MSI o f Barnett. Further more, Spindt’s measure weights each o f its com po nents by a measure o f turnover in purchasing final output (GNP); assets with relatively high turnover rates receive relatively larger weights.1 3 A financial innovation that results in a shift from assets not in simple-sum M l to assets in simple-sum M l w ould cause the same change in measured money, regardless o f the source o f the shift. In con trast, similar innovations w ou ld cause different changes in the MSI or MQ. The extent o f the impact depends on the difference between the asset's own rate o f return and that o f the pure store-of-wealth asset (for the MSI) and on the asset’s relative turnover rate in the purchase o f goods and services (for MQ). Despite the fact that the turnover rates are used in the calculation o f MQ, the m oney stock measure moves only when there is a change in monetary ser vices between periods, so that its velocity changes only when there is a change in the turnover rates. In As a result, these new aggregates mav be affected less by innovations. For example, to the extent that the nationwide introduction o f NOW accounts on January 1,1981, drew deposits out of savings accounts (i.e., idle balances) into NOW accounts, the growth o f simplesum M l w ould be inflated. In contrast, because NOW accounts bear an interest rate closer to the pure storeof-wealth rate, they receive a smaller weight in the MSI. Consequently, if this regulatory change resulted in a significant shift out o f savings-tvpe assets into NOW accounts, the MSI might be affected less bv this regulatory change. The monetaiy-service flow from each asset is based on its "user cost" as measured by the difference be tween the rate o f interest on a pure store-of-wealth asset and the own rate o f return on each asset. Cur rency, which has an own rate o f zero, has the highest user (opportunity) cost. Medium-of-exchange assets like demand deposits (which bear no explicit interest, but bear some im plicit interest, e.g., gifts or no service charges) have a smaller user cost and, hence, receive a smaller weight. Non-medium-of-exchange assets that yield explicit returns closer to those o f the pure storeof-wealth assets receive still smaller weights. 1 Technically, currency, like all financial assets, also acts as a store 2 of wealth; however, the argument is that there exists an asset (fully insured savings deposits) that perform this function better with equal risk. Consequently, no maximizing individual would willingly hold currency purely as a store of wealth given such an alternative. 1 lt is clear from this discussion that two distinct, but related, issues 3 are involved here. The first centers around whether the asset or transactions measure (approach) is preferable. The second is a question of the appropriate weighting scheme. These issues are related in the sense that if the asset approach is preferred, then, by implication, the MSI weighting scheme is preferred as well, since not all of these assets can be used directly in transactions. If the transactions approach is preferred, however, the question of the weighting scheme remains open. The best weighting scheme may still involve the difference between the own rate and the rate on the most liquid non-medium-of-exchange asset. 32 To the extent that NOW accounts are used predom i nantly as a store o f wealth rather than a medium o f exchange, M Q w ould be affected to a lesser degree '"For a discussion of this point, see Spindt (1985). At a more technical level, Spindt (1983) has shown that it is only possible to interpret these aggregates sensibly by using an intertemporal measure. FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Chart 1 Growth Rates of Monetary Aggregates 1971 72 73 74 75 76 77 than simple-sum M l by NOW account growth because NOW accounts initially had a low er turnover rate in transactions than did currency and demand deposits. Also, M Q and broader MSI contain savings-type assets not included in simple-sum M l (e.g., money market mutual funds). Consequently, their growth rates w ould be affected less if the growth in NOW accounts resulted from a shift out o f such deposits. If, however, most o f the growth in NOWs came from demand deposits, then simple-sum M l w ould be relatively unaffected and the growth o f both MQ and the MSI would decrease since dem and deposits had larger weights than NOWs in MQ and the MSI. The advantage these aggregates propose to offer, however, is not without costs. The calculation o f the weights in the MSI and MQ requires more information than that required to construct simple-sum M l. Con sequently, the construction o f these alternative aggre gates may introduce larger measurement and specifi cation errors than those o f omission and inappro priate weighting associated with simple-sum M l (see the insert on the next page).1 5 ,5We say “might be” here for several reasons. What is the appropriate 78 79 80 81 82 83 1984 A COMPARISON OF GROWTH RATES AND WEIGHTS As an initial step in the examination o f alternatives to simple-sum M l, a comparison o f the year-over-vear growth rates o f simple-sum M l (hereafter denoted as M l), M Q and the broadest m onetaiy service index (MSI4) is presented in chart 1. Several interesting points emerge. First, the growth rate o f MSI4 has not conform ed to that o f the other two m onetaiy aggregates anytime during the I/1971-IV/1984 period. Second, up to 1981, the growth rates o f M l and M Q are similar and move together. The mean growth rates for M l and MQ over the I/1971-IV/1980 period are 6.6 and 6.8 percent, re spectively; the standard deviations for the same aggre gates are 1.36 and 1.00 percent, respectively. On the other hand, MSI4 growth during this period is signifi cantly higher and more variable; its average growth was 8.08 percent with a standard deviation o f 3.26 weighting scheme is an open question. Furthermore, if we could decide on the most appropriate scheme from a theoretical point of view, the magnitude of the weights would still be an empirical issue. 33 JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS Information and Estimation Requirements of the MSI and MQ Aggregates The construction o f the monetary service indexes and M Q require more information than is entailed in the construction o f a simple-sum aggregate with identical components. In each case, data on the quantities o f each com ponent are necessary; however, both the MSI and MQ require additional information and, hence, are open to sources o f error not contained in the simple-sum aggregates. The MSI require inform ation that is often incom plete or unavailable. Consequently, certain explicit (or in some cases, implicit) assumptions are m ade that m ay ren d er them less useful as in term ed iate p o lic y targets. First, th ey use information on the ow n rate o f interest on each com ponent. In m any cases, actual data are unavailable so they must be assumed, estimated or set equal to some ceiling rate. For example, the rates on passbook savings deposits at mutual savings banks and savings and loans are assumed to be at their ceiling rates, w hile the rate on demand deposits held by businesses are proxied by the rate on directly placed finance company commercial paper, adjusted for reserve requirements. The own rate o f return on all currency and demand deposits held by households is assumed to be zero. At first, this may seem inappropriate because the ow n rate o f return to holding currency is the negative of the expected rate o f inflation. This assumption is appropriate, however, as long as the interest rate on the pure store-of-wealth asset (M oody’s series o f seasoned Baa bonds) also reflects expectations o f inflation. Nevertheless, changes in inflationary expectations may distort the measure o f m on etary services associated w ith oth er components, because many o f these rates are set at ceiling levels that w ill not respond rapidly to changing expectations o f inflation. Hence, the estimate o f the user cost may erroneously change with changes in expectations o f inflation. This assumption, however, is less appropriate in the 34 case o f demand deposits, because such deposits may yield some explicit return. Furthermore, the theoretical m odel on which these aggregates are based requires that all yields be for an equivalent holding period. As a result, all assets are converted to a one-month holding period y ie ld by a T reasu ry secu rities y ie ld curve adjustment. Moreover, the reference rate that determines the user cost is the maximum o f the Baa corporate bond rate and the rates on the assets contained in the aggregate. Therefore, the user costs are sensitive to changes in the yield curve. In addition to these, a number o f other estimations and assumptions are made (see Farr and Johnson (1985)). Likewise, Spindt’s MQ measure is based on a n u m b e r o f a s s u m p t i o n s n e c e s s i t a t e d by measurement problems. For example, no turnover statistics are available for either currency or travelers checks and the turnover statistics for the other components are gross turnover, not final product (GNP) turnover, as is necessary to be consistent with the underlying theory. As a result, a number of assumptions and estimates are made to generate the final product turnover rates used in the construction o f MQ (see Spindt (1983)). The extent to which these aggregates are affected by the various estimates and implicit assumptions is, o f course, unknown. It could be that there exists a "law o f large numbers” so that, on average, these measurement errors cancel each other. No such law, however, need exist. C onsequently, the potential advantages that these aggregates might o f f e r m us t be d e t e r m i n e d b y s t a t i s t i c a l comparisons like those presented here. Of course, if such comparisons show little or no advantage of these aggregates over simple-sum aggregates, it suggests that w e need to rethink the theory on which they are based or the way in which these aggregates are estimated. FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 1 Weights For Calculating Growth Rates of the Aggregates ( x 100) M1 Year CTC DD 1970 23.0 1971 23.2 1972 MQ MSI4 OCD1 CTC DD OCD1 OQP2 76.9 0.1 45.3 54.7 0.0 0.0 12.7 76.8 0.1 44.8 55.2 0.0 0.0 11.7 23.1 76.8 0.1 43.4 56.6 0.0 0.0 11.7 27.4 1973 23.4 76.5 0.1 40.5 59.5 0.0 0.0 13.4 1974 24.5 75.4 0.1 37.5 62.5 0.1 0.0 13.6 1975 25.7 74.1 0.2 36.8 63.0 0.1 0.1 10.5 21.2 CTC DD OCD1 OCD2 23.4 0.0 0.0 27.0 0.0 0.0 61.3 0.0 0.0 61.0 24.3 0.0 0.0 62.2 23.1 0.0 0.0 63.3 0.1 0.2 68.1 OTHER 63.9 1976 26.7 72.8 0.6 34.5 65.1 0.3 0.1 10.6 21.2 0.1 0.2 68.0 1977 27.1 71.9 1.1 32.8 66.6 0.5 0.1 11.9 21.5 0.2 0.1 66.2 1978 27.5 71.0 1.5 30.9 68.3 0.5 0.2 13.7 20.7 0.4 0.2 65.0 1979 28.1 68.1 3.8 28.6 69.5 1.1 0.8 15.9 20.4 1.3 0.4 62.0 65.4 1980 28.9 65.7 5.4 27.0 68.9 2.1 2.0 13.7 17.7 1.7 1.5 1981 29.0 55.8 15.2 24.3 63.4 9.1 3.3 15.6 15.5 5.8 1.6 61.5 1982 29.2 51.1 19.7 24.6 57.9 12.5 5.1 12.4 12.7 5.6 4.5 64.7 1983 28.8 47.5 23.7 24.9 53.5 16.7 4.8 11.3 11.3 5.4 14.3 57.7 1984 29.2 45.2 25.5 23.4 53.4 17.8 5.4 11.9 10.5 6.1 14.9 56.5 percent. Third, during 1981, the growth rates o f M l and M Q diverge dramatically, reflecting the nation w ide introduction o f NOW accounts. From I/1982-IV/ 1984, the two growth rates exhibit somewhat similar movement, although the growth rate o f M l typically exceeds that o f M Q b y approximately 1.5 to 2 percent age points. An interesting feature o f the growth rates is that each can be expressed as a w eighted average o f the growth rates o f its components. Since weighting is the innovative notion behind these alternative aggregates, an investigation o f these weighting schemes is an instructive way to compare MSI4 and M Q with M l. For M l, the weights are simply each com ponent’s share of M l. The weights for the MSI are each com ponent’s share o f the total expenditure for monetary services. The price o f the monetary services o f each asset is the difference between the yield on a risk-free store of wealth and that asset's own yield. The expenditure on each com ponent’s monetary services is this interest differential times each component's quantity. There fore, each weight is the ratio o f the expenditure on each com ponent’s m onetaiy service to the total ex penditure on m onetaiy services. For MQ, the weights are each com ponent’s total turnover as a percentage o f nominal GNP. In other words, each component's weight is its quantity times its final product turnover rate (i.e., its quantity- w eighted velocity) as a share o f the sum o f these quantity-weighted velocities over the assets in the aggregate, that is, nominal GNP. Annual averages o f these weights for the period 1970-84 are presented in table 1. The weights for the assets in M l are aggregated into three basic groups: those for (a) currency plus travelers checks (CTC), (b) demand deposits (DD), and (c) other checkable de posits (OCD1). The first three columns o f weights for MQ are for the same asset groups as for M l. The fourth column (OCD2) contains the weights for the assets in MQ that are not in M l — m oney market mutual fund shares, m oney market deposit accounts and tele phone transfer savings accounts. The weights for MSI4 are organized similarly. The first three columns con tain the weights for the same asset groups as are in M l; the fourth column (OCD2) contains the weights for the assets in MQ but not in M l. The fifth column (Other) includes the weights o f all the other assets in MSI4. When comparing the weighting schemes, one no tices few similarities. Both the levels, as w ell as the patterns o f movements and the relative magnitudes, are considerably different. Only tw o similarities emerge: The first is the general decline o f the weights o f demand deposits for both M l and MSI4. Alterna tively, the weight for dem and deposits in M Q in creases until 1980, then declines. Even after this de cline, the weight for dem and deposits in M Q currently 35 FEDERAL RESERVE BANK OF ST. LOUIS is about the same as it was at the beginning o f the 1970s, w hile those in M l and MSI4 are approximately 40 percent and 55 percent lower, respectively. Second, the weights o f other checkable deposits in all three aggregates, w hile near zero during most o f the 1970s, have risen dramatically in the 1980s. This rise corre sponds to the increased availability o f new checkable deposits with financial deregulation in the 1980s. The levels and relative magnitudes o f these weights, how ever, differ substantially across aggregates. In particu lar, O C D l’s weight in M l is significantly larger than that in either M Q or MSI4. Moreover, O C D l’s current weight is about 56 percent o f dem and deposits’ weight in M l and 58 percent in MSI4, w hile only about a third o f demand deposits' weight in MQ. The behavior o f currency’s weight across all three aggregates also has been dissimilar. Currency’s weight in M l has risen rather consistently since 1970, while doing just the opposite in MQ. Consequently, changes in the growth rate o f currency now have a larger impact on the growth o f M l and a much smaller impact on the growth o f MQ than earlier. In contrast, currency’s weight in MSI4 has not changed apprecia bly. The decline in demand deposits’ weight, however, has led to a situation in w hich currency growth has a larger impact on MSI4 than does an equivalent change in demand deposit growth, a characteristic not shared by either M l or MQ. By construction, MSI4 contains a large group of assets that, w hile liquid, cannot be exchanged directly for goods and services. It is interesting to note how large the weights o f these non-medium-of-exchange assets are in MSI4. In fact, until the last two years, the weights o f non-medium-of-exchange assets in MSI4 (those classified as "other” in table II have been 1-1/2 to 2 times larger than the weights o f the medium-ofexchange assets (the sum o f the first four MSI4 weights). Only in 1983 and 1984 have the weights o f medium-of-exchange and non-medium-of-exchange assets approached equality. Consequently, until re cently, a one percentage-point change in the rate of growth o f assets that cannot be exchanged directly for goods and services had a substantially larger impact on the growth o f MSI4 than did a one percentagepoint change in the rate o f growth o f transaction balances. JUNE/JULY 1985 predictable relationship with the goals o f policy. Since the growth o f nominal incom e is one o f the principal goals o f monetary policy, it is important that an aggre gate’s incom e velocity be predictable if it is to be used for short-run econom ic stabilization. We begin with a simple comparison o f the levels of the velocities o f M l, MSI4 and MQ. These velocities, norm alized to 1/1970 = 1.0, are presented in chart 2.'“ The velocities o f M l and MQ follow similar patterns. Both appear to increase at a fairly constant rate until 1980, then accelerate through 1981 and decline mark edly after the nationwide introduction o f NOW ac counts. Moreover, both have increased since mid1983. The major difference is that the velocity o f M l was larger than that o f M Q until IV/1980 and has been below it since the introduction o f NOWs.'7While MSI4 velocity has exhibited generally similar movements since the end o f 1980, it grew much more slowly than either M l or M Q velocity up to the beginning o f 1978 and then considerably more rapidly from 1978 to the end o f 1980.1 Moreover, as one w ould expect given the 8 composition o f MSI4, its velocity is significantly low er than that o f the other two aggregates, reflecting the slower turnover rate o f the non-medium-of-exchange assets that are included in it. The quarter-to-quarter growth rates o f the velocities are presented in chart 3. These data indicate that the growth rates of M l and MQ differ little over the period. Indeed, the most significant difference in the growth rates o f M l and M Q occurred in the first two quarters of 1981. The velocities o f both aggregates grew rapidly during the first quarter o f 1981, but the growth in the velocity o f M Q (33.1 percentl was nearly double that of M l (18.2 percent). Furthermore, the velocity o f M l declined in the second quarter o f 1981, w hile that o f MQ increased at a rate o f about 1 percent. In all other cases, the turning points in growth rates o f M l and MQ velocities coincide. In contrast, the growth rate of MSI4 velocity differs from the others, being substan- 1 The velocities for MQ and MSI4 are index numbers and, as such, 6 have no dimension. Hence, they must be normalized to some arbitrarily chosen base period (1/1970 in this case). M1 velocity is normalized similarly to facilitate the comparisons. 1 This is consistent with the earlier observation that simple-sum M1 7 growth has been rapid relative to that of MQ since the nationwide introduction of NOWs. INCOME VELOCITIES OF ALTERNATIVE AGGREGATES For an aggregate to be useful as a short-run interm e diate target o f m onetaiy policy, it must have a stable, 36 '8From 11/1970 to IV/1977, MSI4 velocity grew at a 0.3 percent annual rate while MQ and M1 velocities grew at 2.9 percent and 3.2 percent rates, respectively. MSI4 velocity growth accelerated to a 6.5 per cent rate from 1/1978 to IV/1980 while the growth of MQ and M1 velocities rose only to 3.7 percent and 3.2 percent rates, respectively. JUNE/JULY 1985 FEDERAL RESERVE BANK OF ST. LOUIS Chart 2 Velocities of Monetary Aggregates Parent P«rc»t 1.6 1.6 ___' # 1.4 H I.. 1.2 1.4 r MQ 1.2 MSI4 1.0 1.0 .8 1970 71 72 73 74 75 76 77 78 79 80 81 82 83 1984 C h a rt 3 G r o w t h R a te o f V e lo c itie s o f M o n e t a r y A g g r e g a t e s P trcu t P t r u it 37 FEDERAL RESERVE BANK OF ST. LOUIS JUNE/JULY 1985 Table 2 Tests of the Hypothesis of Zero Autocorrelation: 11/1970— IV/1984 Lag Length Simple-Sum M1 MQ MSI4 Critical \ 2 Value' 6 4.55 3.18 7.80 12.59 12 14.71 14.83 11.44 21.03 18 16.12 15.53 19.46 28.87 24 20.48 19.38 21.49 36.42 'At 5 percent significance level. tially below them until late 1978 and above the others until late 1980. Since 1980 the growth rates o f the velocity o f all these aggregates have behaved similarly. it w ould tend to suggest that it may be no easier to predict MSI4 and M Q velocities than it is for M l velocity. The Predictability o f Velocity Growth To test w hether the growth o f MSI4, M Q or M l velocity contains such regularities, correlation coef ficients between past and current values o f velocity growth are calculated over the period 11/1970 to IV/1984. If these correlations are not statistically sig nificant, then past values o f velocity growth do not contain information helpful in predicting current ve locity growth and, hence, velocity growth cannot be predicted by its ow n past history. The chi-squared statistics for testing w hether the correlations between past and current rates o f velocity growth are different from zero for lag lengths o f 6,12,18 and 24 quarters are presented in table 2. None o f these statistics is statisti cally significant at the 5 percent level. Hence, the hypothesis that each o f these series cannot be pre dicted by its own past cannot be rejected. In other words, the quarterly growth o f the weighted aggre gates’ velocities is no more easily predicted by their own past than is the quarterly growth o f M l velocity2 1 Studies have shown that econom etric forecasts of M l velocity growth tend to produce relatively large forecast errors. This result may be due in part to the fact that velocity growth tends to fluctuate randomly around a fixed mean, so that the expected future growth rate in M l velocity is unrelated to its past growth rates.1 That is to say that M l velocity possesses 3 no regularities that w ill enable it to be predicted on the basis o f its ow n past history. If a series contains such regularities, then its past histoiy provides some basis to predict its future, especially for a short time into the fu tu re .2’ 1 If the growth rates o f M Q and MSI4 velocities also contain no such regularities, then they w ill be just as difficult to predict as M l velocity from their ow n past histories, and may be just as difficult to predict from an econom etric m odel as well. Consequently, it can be argued that a sufficient condition for M Q and MSI4 to be preferable to M l as intermediate policy targets is that the growth rates o f their velocities exhibit regular ities not exhibited by M l velocity. O f course, this finding w ould not preclude the possibility that these velocities could not be predicted on the basis o f infor mation not contained in the past histoiy o f the series itself. Nevertheless, if no such regularities are present, Since the above test indicates that the velocity growth o f each o f these m onetary aggregates varies randomly around its mean, it w ou ld be instructive to examine w hether the velocity growth o f any one aggre gate varies significantly less than that o f the others. The means and standard deviations o f the growth rates given in table 3 indicate that the standard devia tion o f the growth rates o f velocity around their mean levels is not significantly different for any o f the aggre gates 2 Indeed, the standard deviation o f the growth 2 1 Granger (1980) has shown that a series is essentially random if it 9 has no predictable pattern to it. Thus, a time series, X,, is random if the correlation between X, and XH is not significantly different from zero for all j. 2 This result is generally consistent with Spindt’s (1985). 1 “ For example, see Hein and Veugelers (1983) and Nelson and Plosser (1982). ^None of the tests of the hypothesis that the variances are equal could be rejected at the 5 percent level. 38 F E D E R A L R E S E R V E B A N K O F ST. LOUIS JUNE/JULY 1985 Table 3 Means and Standard Deviations of the Growth Rates of Various Velocity Measures: 11/1970-IV/1984 Mean Standard Deviation Simple-Sum M1 2.67 4.99 MQ 3.17 5.85 MSI4 2.21 5.76 Aggregate rates is smallest for M l. Thus, the evidence suggests that the growth rates o f the velocities o f MSI4 and MQ do not appear to be more easily predicted nor any less variable than the growth rate o f M l velocity. Hence, these aggregates may not be better intermediate m on etary targets than M l. While the above analysis indicates that MQ and MSI4 have not been preferable intermediate targets over M l during the 11/1970 to IV/1984 period, it does not preclude that either (or both) o f these aggregates may be better targets during the period o f financial innovation, I/1981-IV/1984. The evidence already pre sented, however, implies that this is not the case. In particular, as seen in charts 2 and 3, both the level and the growth rate o f each velocity behaved similarly from 1/1981 to IV/1984. All three velocities fell in mid-1981 and have rebounded since early 1983. Furthermore, even though the growth o f each velocity is more vari able during this period than it was during the preced ing one, the standard deviations across velocity growth rates are not statistically different. Like the results for the entire period, the growth o f M l velocity is the least variable over the I/1981-IV/1984 period. Consequently, there have not been any substantive changes in the relative performances o f these three aggregates during the past four years."' CONCLUSIONS The introduction of new financial instruments and the recent financial deregulation have confused fur ther the distinction between money and near-money. One response to this confusion has been the construc- tion o f two m onetaiy aggregates as alternatives to the simple-sum measures currently reported by the Fed eral Reserve. These alternatives are the monetary ser vices indexes and MQ. Each o f these new aggregates is a w eighted index o f the same financial assets that constitute the various measures o f m oney as currently defined. The difference between the monetary ser vices indexes and M Q lies primarily in the weighting scheme em ployed to measure the m onetaiy services provided by the assets that com pose each aggregate. The m onetaiy services indexes use opportunity costs o f holding these financial assets to calculate the weights, w hile M Q em ploys the turnover rates o f these assets. When investigated, these weighting schemes differed substantially across the three monetary ag gregates examined. From a policymaking viewpoint, the primary m oti vation for examining different monetary aggregates is to find the one most closely associated with nominal GNP. In this paper, w e com pared the growth and the stability o f the velocity o f these alternative w eighted m onetaiy aggregates w ith the conventional simplesum M l. W e found that the growth rate o f M l velocity was somewhat slower than that o f MQ since the na tionwide introduction o f NOW accounts in 1981; h ow ever, there was little difference in the movements o f these growth rates. Furthermore, the M Q velocity growth was neither less variable nor more predictable than that o f M l. With respect to the broadest monetary services in dex (MSI4), w e found some significant differences in its growth rate and velocity relative to M l and MQ; however, there was no difference in the predictability or the variability o f MSI4 velocity growth. Conse quently, neither MSI4 nor M Q has demonstrated any apparent gain over M l for policy purposes, and both are more difficult to calculate. REFERENCES Barnett, William A. “Economic Monetary Aggregation: An Applica tion of Index Number and Aggregation Theory,” Journal of Econo metrics (September 1980), pp. 11-48. Barnett, William A., and Paul A. Spindt. Divisia Monetary Aggre gates, Staff Study No. 116 (Board of Governors of the Federal Reserve System, May 1982). Barnett, William A., Edward K. Offenbacher and Paul A. Spindt. “The New Divisia Monetary Aggregates,” Journal of Political Econ omy (December 1984), pp. 1049-85. “ This is generally consistent with the results in Batten and Thornton (1985) who found that MQ and MSI4 did not outperform M1 in a St. Louis-type equation during the 1/1981 to 11/1984 period. Batten, Dallas S., and Daniel L. Thornton. “Weighted Monetary Aggregates As Intermediate Targets,” Federal Reserve Bank of St. Louis Research Paper No. 85-010 (1985). 39 FEDERAL RESERVE BANK OF ST. LOUIS P.O. BOX 442 ST. LOUIS, MISSOURI 63166 Subscriber: Please include address label with subscription inquiries or address changes. _________ "M1 or M2: Which Is the Better Monetary Target?” this Review (June/July 1983), pp. 36-42. Brown, W. W., and G. J. Santoni. “Monetary Growth and the Timing of Interest Rate Movements," this Review (August/September 1983), pp. 16-25. Hein, Scott E„ and Paul T. W. M. Veugelers. "Predicting Velocity Growth: A Time Series Perspective,” this Review (October 1983), pp. 34— 43. Keynes, John Maynard. The General Theory of Employment, Interest and Money (Harcourt, Brace and Company, 1936). Chetty, V. K. “On Measuring the Nearness of Near-Moneys,” Amer ican Economic Review (June 1969), pp. 270-81. Laidler, David. “The Definition of Money,” Journal of Money, Credit and Banking (August 1969), pp. 508-25. Farr, Helen T., and Deborah Johnson. “Revisions in the Monetary Services (Divisia) Indexes of Monetary Aggregates,” Mimeo, Board of Governors of the Federal Reserve System (1985). Marshall, Alfred. Money, Credit and Commerce (London: MacMillan and Company, 1923). Fisher, Irving. The Purchasing Power o f Money (New and rev. ed., A. M. Kelley, 1963). _________ The Making of Index Numbers (Houghton-Mifflin Com pany, 1922). Friedman, Milton. “The Quantity Theory of Money — A Restate ment,” in Studies in the Quantity Theory of Money (The University of Chicago Press, 1956), pp. 3-21. Friedman, Milton, and David Meiselman. “The Relative Stability of Monetary Velocity and the Investment Multiplier in the United States, 1897-1958,” in Stabilization Policies (Prentice-Hall, 1963), pp. 165-268. Friedman, Milton, and Anna Jacobson Schwartz. Monetary Statis tics of the United States (National Bureau of Economic Research, 1970). Granger, C. W. J. Press, 1980). Forecasting in Business Economics (Academic Hafer, R. W. “Money, Debt and Economic Activity,” this Review (June/July 1984), pp. 18-25. _________ “The FOMC in 1983-84: Setting Policy in an Uncertain World," this Review (April 1985), pp. 15-37. Hamburger, Michael J. “The Demand for Money by Households, Money Substitutes, and Monetary Policy,” Journal of Political Econ omy (December 1966), pp. 600-23. 40 Mason, Will E. “The Empirical Definition of Money: A Critique,” Economic Inquiry (December 1976), pp. 525-38. Melvin, Michael. “The Vanishing Liquidity Effect of Money on Inter est: Analysis and Implications for Policy,” Economic Inquiry (April 1983), pp. 188-202. Nelson, Charles R., and Charles I. Plosser. “Trends and Random Walks in Macroeconomic Time Series: Some Evidence and Impli cations," Journal of Monetary Economics (September 1982), pp. 139-62. Pesek, Boris P., and Thomas R. Saving. Money, Wealth, and Eco nomic Theory (The MacMillan Company, 1967). Pigou, A. C. “The Value of Money,” Quarterly Journal of Economics (November 1917), pp. 38-65. Spindt, Paul A. “Money Is What Money Does: Monetary Aggrega tion and the Equation of Exchange,” Journal of Political Economy (February 1985), pp. 175-204. _________ “Money is What Money Does: A Revealed Production Approach to Monetary Aggregation,” Special Studies Paper No. 177 (Board of Governors of the Federal Reserve System, June 1983). Thornton, Daniel L. "The FOMC in 1981: Monetary Control in a Changing Financial Environment,” this Review (April 1982), pp. 3 22. _________ “Why Does Velocity Matter?” this Review (December 1983), pp. 5-13.