The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF ST. L O U IS JULY 1972 Will Capital Reflows Induce Domestic Interest Rate C han g es?................................ 2 Income and Expenses of Eighth District Member Banks— 1971 .............................. A Review of Empirical Studies of the Money Supply Mechanism ......................... 11 Vol. 5 4 , No. 7 W ill Capital Reflows Induce Domestic Interest Rate Changes? by ANATOL BALBACH E lV A L U A T IO N of the performance of the U.S. economy in 1971 frequently has placed considerable emphasis on the impact of international capital flows on U.S. interest rates, especially Treasury bill rates. Similarly, prognostications about economic activity in 1972 have involved predictions of interest rate behavior which are heavily influenced by anticipated interna tional capital movements. A typical description of events goes something like this: (1) A large deficit in the U.S. balance of payments in 1971, particularly due to speculative conver sion of dollar assets into assets denominated in foreign currencies, forced huge quantities of dollars on foreign central banks who in turn used them to purchase U.S. Government debt. These purchases bid up the prices of Treasury liabilities and thus depressed their yields. The arbitrage between the Treasury debt market and other credit and equity markets resulted in de pressed yields on all U.S. interest-bearing securities. (2) The devaluation of the dollar eliminated the anticipation of further capital gains accruing to holders of assets denominated in foreign cur rency and, when combined with the narrowing spread between U.S. and foreign interest rates, should cause the “repatriation” of speculative funds that moved abroad in 1971. This will then force foreign central banks to supply dollars which they will acquire by selling U.S. Treas ury securities purchased earlier. Such a massive sale will depress security prices and raise yields, and again, through arbitrage, cause an increase in the yields of all debt and equity instruments in the United States. Therefore, interest rates are projected to rise in 1972 over and above what they would have been without large in ternational capital movements. Page 2 Since monetary and fiscal policymakers consider in terest rate levels important elements of their policy goals, the discussion of interest rate changes induced by international capital movements is not an idle ex ercise. It is a real problem which, through exercise of policy, may have substantial effects on the econ omy’s performance in 1972-73. This note is intended to explain, in a simplified manner, under what circumstances short-term inter national capital movements do or do not affect short term interest rates, and to examine the proposition that “repatriation” of dollar balances in the remainder of 1972 will significantly raise the level of short-term interest rates in the United States. Mechanics of International Capital Transfers International transactions take many forms and vari ous degrees of complexity. Ultimately, however, if there is a net purchase of assets by residents of one country from residents of another, the payment takes place either as a transfer of a reserve asset, such as gold, SDRs, or dollars, or as an accumulation of de mand deposits of the selling country’s residents in the purchasing country’s banks. The sellers are willing to accumulate these deposits only if it is profitable for them to do so. If not, then under a fixed exchange rate system these deposits can be sold to the selling country’s central bank at a fixed price. Suppose that international traders anticipate a de preciation of the dollar vis-a-vis other national cur rencies. This means that they could realize capital gains by converting their dollar assets into assets denominated in a foreign currency and converting FEDERAL. R E S E R V E B A N K O F ST. L O U I S JULY 1 9 7 2 them back into more dollars after the depreciation was realized. Thus, these individuals would use their existing dollar demand deposits, or convert other do mestic assets into dollar demand deposits and proceed to buy foreign assets. These foreign assets may take the form of stocks, bonds, time or demand deposits, or even claims on goods. If the anticipation of dollar depreciation is generally prevalent, foreign recipients of these dollar balances will not be willing to hold them and will sell them to their central banks. Traditionally, and in some cases by statute, central banks have converted these accumulated dollar de posits into short-term Treasury securities. Thus, a net capital outflow from the United States results in the decumulation of dollar denominated assets and de posits by private individuals and corporations, and the accumulation of U.S. Treasury securities by foreign central banks. Conversely, an anticipation of dollar appreciation would induce a demand for dollar de nominated assets. Since dollars can be purchased from foreign central banks at a fixed price, there will be a sale of Treasury securities by foreign central banks and a purchase of securities and deposits by private individuals and corporations. It is these purchases and sales of Treasury securities by foreign central banks that have given rise to allega tions that an outflow of dollars will raise security prices and lower interest rates, and that an inflow of dollars will lower security prices and raise interest rates in the United States. Dollar Outflows and Interest Rates in 1971 The situation in 1971 set into motion events as described above. In addition to the perennial excess of U.S. purchases of goods, services, and long-term capital over sales, yields on foreign short-term securi ties rose relative to those in the United States. This produced a further accumulation of dollar assets by foreign central banks and anticipations that foreign official agencies might stop buying dollars at the prescribed price. The possibility that the dollar might be devalued induced transactors to start converting ever-increasing amounts of dollars into currencies which were expected to appreciate. The rest is his tory. The German mark was floated on May 9, 1971, and similar action was taken soon thereafter by several strong-currency countries. On August 15, con vertibility of the dollar into gold was suspended and, in effect, the dollar was permitted to depreciate according to market forces. On December 18, Smith sonian agreements were reached and the dollar was devalued with respect to gold and most foreign currencies. Y ie ld s on Eu ro d o lla r D e p o sits 1964 1965 1966 1967 1968 1969 1970 1971 1972 Sources: Board of Governors of (he Federal Reserve System and Salomon Brothers and Hutzler D a ta a re lo st Frid a y of the month. 11 Rates on 9 0-d a y deposits. [2 S eco nd ary m arket rotes for n ego tiable 90-day CD*. Latest data plotted: M ay During this period foreign central banks accumu lated $29.8 billion and used them to buy $26.3 billion worth of Treasury securities. There is no doubt that if these massive purchases had been net purchases of Treasury securities, that is, private buyers of foreign assets did not sell domestic securities or equities at the same time, the impact on the prices of these securi ties, and on interest rates in general, would have been significant. It would be tantamount to a sudden shift of $26.3 billion from cash balances to interest-bearing debt. If, on the other hand, the purchasers of foreign assets sold Treasury securities to acquire funds, and foreign central banks ended up buying the same amount of securities, then the impact on interest rates would have been nil. The true picture is somewhat closer to the latter case. Precise data on where and how capital transfers originate are not available, but it is possible to make an educated guess. During 1971 U.S. private and governmental ex penditures on purchases of goods and services and on gifts exceeded receipts from sales and gifts received by $2.8 billion. Long-term capital investment abroad exceeded foreign long-term investments in the United States by $6.4 billion. Thus, the “basic” balance —the amount of dollar balances accumulated by for eigners through these “basic” operations — amounted to $9.2 billion. The total amount of dollars acquired by foreigners during 1971 — the liquidity balance — was $22 billion. Thus, short-term capital flows and Page 3 JULY 1 9 7 2 F E D E R A L R E S E R V E B A N K O F ST. L O U I S N et A cq u isitio n s of U .S. Trea su ry S ecurities b y Foreign O fficial Institutions B illio n s o f D o l l a r s B illio n s o f D o l l a r s ■ n tage of anticipted exchange rate changes, were hold ing dollar denominated assets which were yielding some nominal rate of return. Thus, their purchases of foreign assets and the resulting foreign central bank purchases of U.S. Treasury liabilities were largely off set by a sale of U.S. debt and equity instruments. Such a set of transactions could have had only a minimal influence on short-term interest rates in the United States. “Repatriation” of Dollars in 1972 1 mU 1 1964 1 1965 1 1966 1 1967 1 1968 1 1969 ii n I 1 1970 1971 APRIL 1972 Note: First differences of end-of-period data. errors and omissions made up the difference — $12.8 billion. Official foreign agencies accumulated $29.8 billion, indicating that private foreigners not only sold to their central banks all that they had accumulated during the year ($22 billion), but transferred an addi tional $7.8 billion of liquid U.S. assets to foreign official reserves. If we assume that “basic” transactions are a result of price level differentials, long-term yield considera tions, and various long-term private and govern ment commitments, then the difference between the official transactions balance and the “basic” balance, or $20.6 billion, was a result of short-term interest differentials and anticipated devaluation of the dollar. This estimate is obviously a very rough one, but it gives some insight into the magnitude of the so-called “speculative” outflows, which might reverse under changed conditions. In considering the impact of these outflows on in terest rates and the ease of the Treasury’s financing of its debt, it becomes crucial to know whether these dollar balances were originally held in the form of currency and demand deposits, in some form of inter est-bearing assets, or in equities such as stocks. Al though there is evidence that equity transactions play a minor role in international capital transfers, let us assume that investors shift with ease from one capital market to another and are sensitive to yield differen tials between equities and debt. If this is indeed the case, then it is possible to assert that the individuals and corporations engaged in international interest arbitrage, and who are able and willing to take advan Page 4 Now we can turn to the oft-made statement that the expected return flow of dollars will cause U.S. interest rates to rise even higher in the remainder of 1972 than the levels they would otherwise reach. In the first place, the reflow will occur only if interest rate differentials change sufficiently in favor of the United States to offset anticipations of further dollar devaluation. These interest rate differentials have indeed been narrowing in favor of the United States. However, recent floating of the British pound and the resulting attempts to convert dollars into other European and Japanese currencies indicate rather strikingly that the dollar is still considered a “weak currency.” At this time it appears that any monetary disturbance, irrespective of how far it may be removed from the actual strength of the dollar, raises the possibility of further dollar depreciation. Thus, the expected reflow may not take place. Secondly, if the reflow does materialize, the crucial question is what will these returning dollar balances be converted into? If our rough figures are correct, these flows should amount to at least $20.6 billion, and again, it is reasonable to assume that they would be converted into assets which bear some nom inal return. Thus, the sale of Treasury securities by foreign central banks would be offset by the purchase of other debt and equity instruments by private indi viduals and corporations, and the impact on interest rates would be insignificant. Conclusions Any impact of international short-term capital move ments on U.S. interest rates in 1971 depended on: (1) whether the buyers of foreign currency de nominated assets simultaneously sold dollar de nominated debt and equity, and (2) with particular reference to the yields on Treas ury debt, whether various debt and equity markets were effectively separate in the minds of transactors. F E D E R A L R E S E R V E B A N K O F ST. L O U I S JULY 1 9 7 2 The anticipated effects of a “repatriation” of short term capital in 1972, if it occurs at all, will depend on: ( 1 ) w hat dollar denom inated assets w ill b e bou g h t b y the sellers o f foreign assets, and ( 2 ) the substitutability b etw een different d eb t and equity instruments. It seems reasonable to assume that individuals and corporations who engage in these “speculative” trans actions are knowledgeable investors who do not keep large idle balances (idle in a sense that they do not earn any nominal return) and, therefore, their pur chases of assets denominated in one currency are ac companied by sales of assets denominated in another. Given that the risk factors associated with equities have been rather small in the United States for the past thirty years, it is highly probable that the rates on Treasury debt are highly interdependent with the rates on other debt instruments and yields on equities. Thus, it is unlikely that capital outflows in 1971 were the major cause of the decline in Treasury debt rates or interest rates in general. For these reasons, together with the recent concern that the dollar may still be overvalued, prognosticators of economic activity and policymakers who base their decisions on these predictions should be wary of as sumptions that a “repatriation” of dollars will occur and, if it does, will cause an increase in U.S. interest rates. REVIEW INDEX - 1972 Month o f Issue Jan. Title Government Debt, Mo ne y and Economic Activity A Critical Look at Monetarist Economics Comments on a Monetarist Approach to De mand Management Feb. The Economy in 1972 Operations of the Federal Reserve Bank of St. Louis — 1971 Projecting With the St. Louis Model: A Prog ress Report March The 1972 National Economic Plan: An Experi ment in Fiscal Activism Monetary Expansion and Federal Open Market Committee Operating Strategy m 1971 Has Monetarism Failed? — The Record Ex amined Month o f Issue Title April Recent Monetary Growth U.S. Balance-of-Payments Problems and Policies in 1971 Outlook for Farm Income and Food Prices May Curbing Price Expectations: The Key to Infla tion Control Measurement of the Domestic Money Stock Problems of the International Monetary System and Proposals for Reform — 1944-70 June Recovery Accelerates Eighth District Population Centers The Hunt Commission Report — An Economic View A Look at Ten Months of Price-Wage Controls Page 5 Income and Expenses Of Eighth District Member Banks —1971 by LINDA FAULKNER CONDIT ET INCOME of member banks in the Eighth Federal Reserve District increased 7 percent in 1971 to $151 million, compared with a 15 percent increase in 1970. Chief factors accounting for the slower in come growth in 1971 were a decline in the average rate charged on loans and the increased costs of time and savings deposits. Trends in the sources and dis position of income of all member banks in the nation were generally similar to those in the Eighth Dis trict, with net income increasing 8 percent to $4.1 billion in 1971 following an 11 percent rise in 1970. Operating income of District member banks in creased 8 percent to $944 million in 1971, while op erating expenses rose 11 percent to $748 million. Since expenses rose by a larger dollar amount than earnings, income before taxes and securities gains or losses declined 4 percent, compared with a 12 per cent gain in 1970. Income taxes applicable to operat ing income declined $13.5 million, and net gains on security transactions after taxes rose $5.8 million, re sulting in a net income gain of $9.8 million from the previous year. Operating Income Operating income of commercial banks is primarily determined by the volume and composition of earn ing assets held and their respective rates of return. Revenue from loans and securities is the major source of income; other sources include trust department in come, service charges on deposit accounts, and other fees. The dollar volume of earning assets increased more rapidly in 1971 than in 1970, but the rate of re turn on loans, which comprised over half of the asset portfolio, was less. Thus, operating income rose only 8 percent, compared to a 13 percent rise in the pre vious year. More expansive monetary actions during the first half of 1971 as well as aggressive competition for Page 6 Operating Income, Operating Expenses and Net Income Eighth D istrict M em ber B an k s Ratio S c a le Ratio S cale M illio ns of D o lla rs M illio n s o f D o lla rs 1200 1000 800 600 400 200 100 90 80 70 1966 1967 1968 1969 1970 1971 Note-. Sta rtin g in 1969 O p e ra tin g E xp en ses a n d N et Incom e a re computed on the new inco m e reporting b a sis. Fo r fu rth er e xp la n a tio n o f this revisio n, see the Federal R eserve Bulletin (July 1970), pp 571 and 572. D a ta for 1968 on the new reporting basis a re estimated. savings throughout the year enabled District banks to increase their total assets 13 percent to $17.7 billion, compared to a 10 percent increase in 1970. Time and savings deposits became a more attractive investment for the public because rates of return on most com peting market instruments generally declined rela tive to rates offered by banks. Loans outstanding increased 12 percent to $9.1 bil lion between December 1970 and December 1971. Loans accounted for 52 percent of total assets at JULY 1 9 7 2 F E D E R A L R E S E R V E B A N K O F ST. L O U I S IN C O M E A N D EX PEN SES O F EIGH TH DISTRICT- MEMBER BA N KS Thousands of Dollars 1971 Percent Change 1970-71 1969-70 1970 1969 Total O perating income ................. .............. ................................................................................. $ 9 4 4 ,1 9 4 .2 6 0 4 ,2 5 2 .7 Income from Loans ........... -...................................................... -.................-.............................. $ 8 7 7 ,3 7 7 .3 $ 7 7 9 ,8 4 4 .1 7 .6 % 1 2 .5 % 5 9 1 ,0 7 2 .1 5 2 6 ,3 5 3 .3 2.2 12.3 Income from Securities ................. ...................-..........................................-............................ 2 2 9 ,9 1 8 .4 1 9 2 ,2 7 0 .6 1 7 4 ,1 3 5 .3 19.6 10.4 U .S. Treasury Securities ------ ------ ---- ----- ------------------------- ------ 1 1 3 ,2 0 0 .0 1 0 3 ,5 6 3 .3 9 3 ,8 8 4 .8 9 .3 10.3 O ther ................. —---- ---- ----- ------------------------- ----- ----------- ----- ---- 1 1 6 ,7 1 8 .4 8 8 ,7 0 7 .3 8 0 ,2 5 0 .5 3 1 .6 10.5 Trust Department Income -- ---- ----- ------ ------------------------------------- 2 3 ,7 3 2 .4 2 0 ,7 1 0 .6 1 9,5 0 8 .1 14.6 6.2 Service Charges on Deposit Accounts ................................................ .......................... 2 7 ,5 4 9 .2 2 6 ,0 1 1 .9 2 4 ,7 5 0 .4 5 .9 5.1 O ther O perating Income .............................. ............................... .......................................... 5 8 ,7 4 1 .5 4 7 ,3 1 2 .1 3 5 ,0 9 7 .0 2 4.2 3 4 .8 Total O perating Expenses .................................................................................-........................ 7 4 7 ,5 6 1 .3 6 7 1 ,6 8 2 .5 5 9 5 ,7 1 5 .9 1 1.3 1 2 .8 S a la rie s , W a g e s, and Benefits ...........................................~............... -................ ........... 2 0 0 ,9 9 3 .1 1 8 4 ,4 9 2 .6 1 6 7 ,5 8 6 .2 8 .9 10.1 Interest on Deposits --------- --------------------------------- ---- ---- ----- ----- 3 1 3 ,8 5 4 .2 2 6 1 ,9 9 3 .0 2 2 5 ,7 1 7 .2 19.8 16.1 O ther Interest Expenses _______________________________________ ______________ _ _ 3 7 ,5 7 7 .2 5 0 ,1 1 5 .8 5 0 ,9 8 0 .1 — 2 5 .0 — 1 .7 O ther O perating Expenses ................. ............................... .............. .......... .......................... 1 9 5 ,1 3 6 .8 1 7 5 ,0 8 1 .1 1 5 1 ,4 3 2 .4 11 .5 1 5.6 Income Before Income Taxes and Securities G ain s or Losses ................. -..... 1 9 6 ,6 3 2 .9 2 0 5 ,6 9 4 .8 1 8 4 ,1 2 8 .2 — 4 .4 1 1 .7 Less A pp licable Income T a x e s -- ----------------------------------------------------- 5 2 ,2 3 1 .8 6 5 ,7 6 7 .4 5 8 ,5 3 8 .6 — 2 0 .6 12.4 Income Before Securities G a in s or Losses ........................................... ............................. 1 4 4 ,4 0 1 .1 1 3 9 ,9 2 7 .4 1 2 5 ,5 8 9 .6 3 .2 1 1.4 Net Securities G ain s or Losses after Taxes ..--------------------------- ---------- 6 ,0 2 8 .2 2 6 0 .7 — 2 ,3 7 8 .5 2 ,2 1 2 .3 — Extrao rd in ary Charges or Credits after T a x e s __________________________________ 5 0 3 .4 1 ,0 2 4 .7 117.1 — 5 0 .9 775.1 Less M inority Interest in Consolidated S u bsidiaries _____ .._____ ______________ 2 5.8 1 4.9 2 6 .6 7 3 .2 — 4 4 .0 Net Income ............................................................................................................................................... 1 5 0 ,9 0 6 .9 1 4 1 ,1 9 7 .9 1 2 3 ,3 0 1 .6 6 .9 1 4.5 Cash Dividends Paid .................................... ................................................................................... 6 2 ,5 7 3 .7 4 8 ,1 8 5 .5 4 3 ,8 1 1 .4 2 9 .9 10 .0 0.2 — 1.3 Number of Banks ............. .................—------- ------------------------------------------- 458 4 59 4 65 - all Eighth District member banks, the same as a year earlier. By size of bank, however, the propor tion of loans to total assets varied from an average of 53 percent for those banks with $100 million or more in deposits to 45 percent for those banks with less than $5 million in deposits. In 1971 banks held propor tionately more consumer loans, fewer commercial loans, and the same proportion of real estate loans as in 1970. Outstanding consumer loans rose 14 percent, largely reflecting a 16 percent increase in automobile loans. Real estate loans increased 9 percent, and com mercial loans rose only 5 percent. Included in the total loan data are Federal funds sold and securities purchased under agreements to resell, which com bined increased 47 percent to $9.6 million. Eighth District member banks were net lenders in this mar ket in 1971. Investments at District member banks rose 18 per cent to $5 billion between December 1970 and De cember 1971. Holdings of U.S. Treasury securities, which accounted for about 40 percent of total invest ments, rose 7 percent and other securities, largely tax-exempt state and municipal obligations, rose 26 percent. This large increase in other securities resulted in such holdings rising from 15 to 17 percent of total assets in 1971. U.S. Treasury securities as well as cash Page 7 JULY 1 9 7 2 F EDE RAL . R E S E R V E B A N K O F ST. L O U I S and other noneaming assets declined slightly relative to total assets. Average rates of return on earning assets remained relatively stable from the previous year with the ex ception of the rate on loans, which fell about 60 basis points to 7.1 percent.1 The average rates of return on U.S. Treasury securities and obligations of U.S. Gov ernment agencies remained at 5.8 and 6.3 percent, respectively. The rate on obligations of states and political subdivisions combined increased about 20 basis points to 4.1 percent. Trust income was a more important source of in come for the larger than the smaller banks. For ex ample, 4 percent of total operating income was de rived from trust operations by those banks holding deposits in excess of $100 million. In contrast, only a fraction of a percent of the income of banks with less than $5 million in deposits was derived from trust operations. The growth in income from securities accounted for over one-half of the $66.8 million gain in operating income in 1971. Income from tax-exempt state and municipal bonds rose $16.4 million, representing al most a quarter of the increase. Due to a decline in average lending rates, returns from loans accounted for only one-fifth of the gain, compared to two-thirds in 1970. Trust department income and service charges on deposit accounts, represented 5 and 2 percent, re spectively, of the increase. Miscellaneous income sources, which include safe deposit box rental, prop erty leasing, foreign department operations, and data processing, accounted for 17 percent of the income gain. Operating Expenses Operating expenses of District member banks in creased 11 percent in 1971 to $748 million. Expenses rose at a faster rate and by a larger absolute amount than operating income, and thus income before in come taxes and securities gains or losses fell below the year-earlier level. This growth in expenses can be attributed primarily to a larger volume of time and savings deposits with no change in the average rate paid. Salaries, wages, employee benefits, occupancy and equipment expenses, and provision for loan losses were also higher than in 1970, but interest paid for borrowed money declined. Interest and fees on loans accounted for about twothirds of total operating income but showed little growth in 1971, rising only 2 percent to $604 million. Interest on U.S. Treasury securities increased 9 per cent to $113 million. Returns on all other securities, primarily tax-exempt state and municipal bonds, were the most rapidly rising source of income, increasing 32 percent to $117 million. Income from all other sources, including trust department income and serv ice charges on deposit accounts, totaled $110 million, an increase of 17 percent from 1970. X A11 rates of return are unweighted averages computed from Reports of Condition for December 31, 1970; June 30, 1971; ana December 31, 1971 and Report of Income for 1971. Page 8 Liabilities of member banks increased more rapidly in 1971 than a year earlier. The volume of time and savings deposits increased 20 percent to $6.9 billion, a pace even more rapid than in 1970 when regulatory ceiling rates were suspended on large-denomination CDs having maturities of thirty to eighty-nine days. The growth in 1971 resulted from the decline in short term market interest rates from 1970 levels relative to the average rate paid on time and savings deposits. Demand deposits at District member banks rose 6 per cent to $8.1 billion, and other liabilities increased 30 percent to $1.2 billion. Federal funds purchased and securities sold under agreements to repurchase, which constituted the major portion of “other liabilities,” in creased 42 percent. This sharp increase reflects a sizable decline in the interest rate on Federal funds F E D E R A L R E S E R V E B A N K O F ST. L O U I S relative to the discount rate. In contrast, during most of 1970 the Federal funds rate greatly exceeded the discount rate. The composition of bank liabilities shifted consid erably in 1971 as time and savings deposits continued to grow rapidly. These deposits increased from 37 to 40 percent of total liabilities and capital accounts, while demand deposits fell from 49 to 46 percent. Other liabilities, such as Federal funds purchased and other borrowed funds, increased from 6 to 7 percent, while capital accounts decreased slightly relative to the total. JULY 1 9 7 2 Trends in Operating Expenses Eighth District M em ber Banks Ratio S ca le Ratio Scale M illions of D o llars M illions of D ollars reporting b a sis. For further e xp la n a tio n of this re visio n , see the Fed era l Reserve Bulletin (July 1970), p p 571 a n d 5 7 2 . D a ta fo r 1968 on the new reporting b a sis a re e stim a te d . penses declined 25 percent to $38 million. This drop resulted from a decline in the average rate paid on Federal funds purchased and securities sold under agreements to repurchase. The average rate on such borrowings declined from 6.6 percent in 1970 to 6.2 percent in 1971. The volume of other borrowed funds, consisting largely of Federal Reserve credit, declined 60 percent, reflecting the differential between the discount rate and market interest rates. The average rate paid on time and savings deposits remained at 4.9 percent in 1971. Most banks main tained their rates on regular savings and small-denomination time deposits at or near the regulatory ceilings which have been in effect since early 1970. However, rates on large-denomination CDs fluctuated with other money market rates. Interest on time and savings deposits, the most rap idly growing expense item and the only one to exceed the growth rates of 1970, increased 20 percent to $314 million in 1971. This increase reflected both the rapid growth of time and savings deposits and the rela tively stable average rate paid. Other interest ex Salaries, wages, and employee benefit expenses rose 9 percent to $201 million, reflecting both a rise in average wages paid and an increase of 3 percent in the number of officers and employees. Salaries and wages are a larger percentage of expenses of the very small banks as compared to other banks, reflecting the greater labor intensity of the small banks. Average compensation per person increased only 5 percent in 1971, reflecting in part the price-wage freeze initiated on August 15, 1971. Remaining expenses rose 12 per cent to $195 million. Net Income Income before taxes and securities gains or losses of Eighth District member banks in 1971 was $197 million, a decrease of 4 percent from 1970. Income Page 9 JULY 1 9 7 2 FE DERAL. R E S E R V E B A N K O F ST. L O U I S taxes applicable to operating income declined 21 per cent to $52 million. This reduction in taxes resulted from increased holdings of tax-exempt securities, re moval of the Federal corporate income surtax, a drop in income before taxes and securities gains or losses, and a previous shift in tax accounting techniques from a cash to accrual basis. This considerable de crease in taxes boosted after-tax income 3 percent. Net securities gains after taxes jumped from $0.3 mil lion in 1970 to $6 million in 1971, contributing to further improvement in earnings. After adjusting for the net effects of taxes, securities gains or losses, and extraordinary charges, net income increased 7 percent to $151 million in 1971. Bank Capital Capital accounts of District member banks — equity capital plus capital notes and debentures — totaled $1.3 billion at the end of 1971, an 8 percent increase from the previous year. Equity capital increased 8 percent to $1,271 million, and capital notes and de bentures rose 18 percent to $62 million. Because of a 30 percent increase in cash dividends to $63 million, net retained earnings — the major source of equity capital —declined 5 percent from the previous year to $88 million. Net income as a percentage of equity capital declined from 12 percent in 1970 to 11.9 per cent in 1971. Publications of This Bank Include: Weekly U. S. FINANCIAL DATA Monthly REVIEW MONETARY TRENDS NATIONAL ECONOMIC TRENDS Quarterly QUARTERLY ECONOMIC TRENDS SELECTED ECONOMIC INDICATORS - CENTRAL MISSISSIPPI VALLEY FEDERAL BUDGET TRENDS U. S. BALANCE OF PAYMENTS TRENDS Annually ANNUAL U. S ECONOMIC DATA RATES OF CHANGE IN ECONOMIC DATA FOR TEN INDUSTRIAL COUNTRIES (QUARTERLY SUPPLEMENT) Copies of these publications are available to the public without charge, including bulk mailings to banks, business organizations, educational institutions, and others. For information write: Research Department, Federal Reserve Rank of St. Louis, P. O. Rox 442, St. Louis, Missouri 63166. Page 10 A Review of Empirical Studies of the Money Supply Mechanism by ROBERT H. RASCHE The use of a reserve aggregate as an operating target raises questions about the interest rate effects of policy actions on the ability of the Federal Reserve to achieve a desired growth in the money stock. Open market operations affect both reserve aggregates and interest rates. Changes in interest rates, in turn, influence the portfolio decisions of banks and the public, and are believed by some to be a com plicating factor in the achievement of a desired money growth. This article surveys empirical evidence on the interest rate sensitivity of the money supply, and concludes that this interest elasticity appears to be extremely low. Hence, this factor should exert a negligible effect on the ability of the Federal Reserve to influence money stock growth through control of a reserve aggregate. I N RECENT YEARS there has been considerable discussion concerning techniques for conducting mon etary policy. The traditional practitioners of the art of policymaking have argued for the use of operating procedures which focus on “money market conditions.” At various times this has been construed to mean free reserves, the Treasury bill rate, the federal funds rate, or a combination of these.1 Alternatively, it has been argued that the target of monetary policy actions should be a monetary aggregate, and that this target can be achieved by control of some reserve aggregate concept such as the monetary base.2 This article surveys the accumulated empirical evi dence on the interest sensitivity of some reserve multipliers. If these multipliers are highly sensitive to interest rate changes, then it may be difficult to im plement monetary control through the control of re serve aggregates. The available evidence consistently indicates, however, that the interest sensitivity of vari ous multiplier concepts is extremely low. This suggests that control of monetary aggregates through reserve control should not be very difficult to implement.3 Conditions Inhibiting Control of the Money Stock The issue examined here is the feasibility of control of a monetary aggregate such as the narrowly defined 'Stephen H. Axilrod, “ The FOM C Directive the Late 1960’s: Theory and Appraisal,” Policies and O perating Procedures — Staff ington, D .C.: Board of Governors of the System, 1971), pp. 1-36. as Structured in in O pen M arket Studies (W ash Federal Reserve 2For example, see Albert E. Burger, Lionel Kalish III, and Christopher T. Babb, “ Money Stock Control and Its Implica tions for Monetary Policy,” this Review (O ctober 1971), pp. 6-22. 3If, however, these multipliers are highly sensitive to interest rate changes, then accurate monetary control through a reserve control procedure requires a precise estimate of the impact o f reserve changes on interest rates, in addition to a precise estimate of the interest elasticity of the reserve multiplier. money stock ( M i ) , given control of some reserve ag gregate concept. The problem can be illustrated by the equation M = mil where “M” is the money stock, “R” is some reserve aggregate concept, and “m” is the appropriate reserve multiplier.4 Two sources of difficulty can arise in such a control procedure. First there can be systematic feedbacks on “m” through market forces which tend to offset the ex pected effect of a change in the reserve aggregate on the money stock. This influence of the behavior of reserves on the value of the multiplier can be stated as m = f(R). The sources of feedback from changes in “R” to changes in “m” will vary depending on the choice of a reserve aggregate concept. If the net source base concept is used for “R”, the associated multiplier (m ) is 1 + k m “ (r -b ) (1+t+d) + k where “r” and “b” are the ratios of bank reserves and member bank borrowings to commercial bank deposits, respectively, “t”, “k” and “d”, respectively, are the ratios of time deposits, currency held by the public, and U.S. Government deposits at commercial banks to the demand deposit component of the money supply. Therefore, the important behavioral relationships in fluencing the stability of the multiplier in the presence 4A number o f candidates have been proposed for “ R” in cluding the monetary base, unborrowed reserves plus cur rency, total reserves, unborrowed reserves, and reserves available to support private deposits. For a discussion of the relative virtues o f many o f these, see Richard Davis, “ Short-Run Targets for Open Market Operations,” in Open Market Policies and O perating Procedures — Staff Studies (Washington, D .C .: Board of Governors o f the Federal Re serve System, 1971), pp. 37-45. Page 11 F E D E R A L R E S E R V E B A N K O F ST. L O U I S of reserve changes are the public’s demand for cur rency and time deposits, banks’ demand for excess reserves and borrowings, and the supply of time deposits.5 An example of a feedback effect on “m” would be where there exists a sizable short-run interest elasticity of demand for excess reserves by commercial banks. In order to force additional reserves into the banking sys tem to expand the money stock, the Federal Reserve would have to buy Government securities, thus push ing short-term interest rates down. If the amount of excess reserves demanded by banks is very sensitive to changes in short-term interest rates, this interest rate movement would induce banks to hold larger quan tities of excess reserves This portfolio shift then offsets the policy to increase the money stock. The existence of strong feedback effects on the reserve multiplier does not mean that monetary con trol through reserve aggregates is impossible. The stronger the feedback, the larger the necessary magni tude of the open market operation required to achieve a given change in the money stock and the larger the associated variance in short-term interest rates. The second source of difficulty in this type of mone tary control procedure is that the relationship between the reserve aggregate and the money stock is subject to random fluctuation. Specifically, we can write mt = f( Rt) + £t where “et” is an unknown random disturbance to “mt”. If such fluctuations are truly random, then in the long run policymakers should be able to hit the desired average stock of money quite closely. If this random component is large, then in a short time pe riod, such as one or two months, the average “m” could deviate considerably from the forecast “m” and cause a large average error around the desired path of the money stock. It can be shown that for a given variance of “et”, under a control procedure such as that recently pro posed by Burger, Kalish, and Babb, the variance of the actual path of the money stock around the desired path will depend on the sensitivity of the reserve 5For a detailed discussion o f the functional relationship of the multiplier expression to asset holdings o f the nonbank public, the banking system, and the Treasury, see Albert E. Burger, The Money Supply Process (Belmont, California: Wadsworth, 1971), especially chaps. 4-5, and Karl Brunner and Allan H. Meltzer, “ Liquidity Traps for Money, Bank Credit, and Interest Rates, Journal of Political Economy (January/February 1968), pp. 1-37. Page 12 JULY 1 9 7 2 multiplier (m ) to changes in the reserve aggregate.8 The smaller the sensitivity of the multiplier, the smaller will be the variance of the actual money stock around the desired money stock. The Nature of Available Evidence on Multiplier Sensitivity Over the past decade there has been considerable empirical research directed at measuring the relation ship between the money stock and various reserve ag gregates. This work has evolved primarily from at tempts to construct econometric models of basic finan cial relationships in the U.S. economy. As a by-product, these studies provide information on the interest elas ticities of the behavioral parameters of the reserve multiplier, the existence of which cause feedbacks against policy actions as discussed above. Most of the more detailed studies have worked with quarterly data, which may be too highly ag gregated in time to provide information that policy makers desire if the reactions of the banking system and the public are distributed over time. However, studies using shorter time horizons do exist for some components of the money supply mechanism, and these can be used to obtain information on how the estimated elasticities are likely to change as the horizon becomes shorter. There are several potential sources of feedback which will offset the expected impact of a change in reserve aggregates on the change in the money stock. Some of the feedback, such as a change in the demand for currency and time deposits by the nonbank public as a result of increased economic .activity, has been shown to occur only slowly, and does not cause diffi culties for short-run control.7 The troublesome source of changes in the multiplier relationship is the impact of changes in interest rates on the behavioral parameters in the multiplier. Changes in market interest rates and changes in reserves avail able to the banking system cannot be controlled simul taneously by the Federal Reserve System. When the Federal Reserve follows a reserve aggregate operat ing procedure, interest rates are affected by changes in reserves. Under a money market conditions operat ing strategy, changes in reserve aggregates come about as a result of the attempt to achieve certain 6Burger, Kalish, and Babb, “ Money Stock Control.” 7See David I. Fand, “ Some Implications of Money Supply Analysis,” American Economic Review (M ay 1967), pp. 380-400. F E D E R A L R E S E R V E B A N K O F ST. LO U I S levels of interest rates. Hence, if the goal is to control money through changes in reserve aggregates, the major issue is the interest elasticity of the relationship between the money stock and reserve aggregates. It will be necessary to distinguish between shortrun, or impact, elasticities of the reserve multiplier and long-run, or equilibrium, elasticities. The former include only the impact which comes from the adjust ment of economic units to a change in interest rates within one period of time. Many studies, however, have indicated that economic units respond to such changes with a distributed lag; that is, part of the response takes place in the same period, and the re mainder of the response takes place over several pe riods following a change in interest rates. The impact, or short-run, interest elasticity is the percentage change in the reserve multiplier with respect to a percentage change in interest rates within the time period in which the interest rate changes. The equilibrium, or long-run, elasticity is the total response of the reserve multiplier after economic units have had sufficient time to adjust to a new portfolio equilibrium.8 In the studies cited below, estimates have been ob tained for the interest elasticity of the money stock for given values of various reserve aggregates. Thus, the money stock elasticities computed are the interest elasticities of the reserve multiplier. Interest Elasticity Estimates From Data Prior to 1965 Teigen I An early econometric investigation of the money supply relationship was that of Ronald Teigen.8 His study does not develop the detailed specifications which are characteristic of more recent studies. In particular, the stocks of currency in the hands of the public and demand' deposits at nonmember banks are assumed exogenous.1 In addition, Teigen takes the 0 quantity of time deposits at member banks and gov 8For a discussion o f impact versus long-run responses, see Arthur S. Goldberger, Im pact M ultipliers and Dynamic Prop erties of the Klein-Goldberger M odel (Amsterdam: NorthHolland, 1959). 9Ronald L. Teigen, “ Demand and Supply Functions for Money in the United States: Some Structural Estimates,” Econometrica (October 1964), pp. 476-509. 10It is necessary to distinguish here between the construction of the model from historical data, and the use of the model to determine interest elasticities. In the construction of the model, the ratio o f currency to demand deposits at member banks and the ratio of demand deposits at nonmember banks to those at member banks are in fact exogenous variables which vary from one observation to the next. In JULY 1 9 7 2 ernment deposits at member banks as exogenous variables.1 1 Teigen tests the hypothesis that the banking system takes more than one period to respond to changes in interest rates, but this hypothesis is rejected for the post-war data. Thus his impact and equilibrium in terest elasticities of the money supply relationship are equal. His estimated coefficients of elasticity are 0.1950 for the commercial paper rate and —0.1695 for the discount rate.1 2 DeLeeuw I Frank DeLeeuw attempted to obtain more detailed numerical estimates of behavior in important financial markets than did Teigen.1 In particular, DeLeeuw 3 separates bank borrowing and excess reserve behavior, and explicity estimates functions for currency demand and time deposit demand at commercial banks by the nonbank public. The interest elasticity estimates from this study are summarized in Table I. In all cases the absolute value of the long-run elasticities are less than one, and the short-run elasticities never exceed 0.2 in absolute value. The available data do not permit reconstruction of the interest elasticities of excess reserves. However, DeLeeuw did publish the results of a computation of the implicit interest elasticities of the money-reserve determining the value o f the elasticity of the relationship, these exogenous variables are kept fixed at some point, conventionally their mean value for the sample period. Hence, the computations implicitly assume positive interest rate responses for the public’s currency demand and the supply o f demand deposits by nonmember banks, which are equal to the interest rate response of demand deposits supplied by member banks. For nonmember banks, the as sumption probably does not seriously affect the analysis. On the other hand, the public’s currency demand is usually found to have a zero, or slightly negative, interest elasticity, at least in the long ran. If the true interest elasticity of currency demand is zero, then the bias introduced by the constant ratio of currency to money stock is indeterminate. On the one hand, the direct effect of increased currency in the hands of the public as demand deposits supplied by banks increase biases the interest elasticity o f the money supply upward. On the other hand, the indirect effect that the assumed increase in currency withdraws reserves from the banking system causes the model to understate the de sired amount of deposit expansion. Since the magnitudes involved are small, the net bias should not be substantial. n These variables do not explicitly appear in his model. How ever, the reserve aggregate which he uses is unborrowed reserves available to support private demand deposits. Later studies use more broadly defined aggregates such as un borrowed reserves, or unborrowed reserves plus currency. To make the studies comparable, the model must be re formulated with time deposits at member banks and gov ernment deposits at member banks explicitly appearing as exogenous variables. '-’Teigen, “ Demand and Supply Functions,” p. 502. 13Frank DeLeeuw, “A Model o f Financial Behavior,” in The Brookings Q uarterly Econom etric M odel of the United States, ed. James S. Duesenberry et al. (Chicago: Rand McNally, 1965), chap. 13. Page 13 F E D E R A L R E S E R V E B A N K O F ST. L O U I S JULY 1 9 7 2 Tab le I Interest Elasticities of V ario u s Functions in D eLeeuw 's O rig in a l Brookings M odel Specification Im pact* Equilibriurr Currency Demand Private Securities Rate Time Deposit Rate - .0 3 2 - .0 1 2 — .3 6 4 — .1 3 6 Time Deposit Demand Treasury B ill Rate Time Deposit Rate — .0 3 8 .0 7 0 - .3 7 4 .6 8 3 B ank Borrowings Treasury B ill Rate Discount Rate .1 3 4 — .1 8 6 .5 0 — .7 0 Excess Reserves Treasury B ill Rate Discount Rate n .a . n .a. n .a . n .a . ♦Impact elasticities are calculated by using the respective equilibrium elasticities and regression coefficients o f the lagged dependent vari ables from various tables in the source cited below. The equilibrium elasticities fo r currency and tim e deposit dem and are from p . 493, and fo r bank borrow in gs from p. 512. The respective regression coefficients used in calculating the im pact elasticities can be found in Tables 13.2, 13.4, and 13.12. S ou rce: F rank DeLeeuw, “ A Model o f Financial B ehavior,” in The B rookings Quarterly E con om etric Model o f the U nited States, ed. Jam es S. D uesenberry et al. (C h ica g o: R and M cNally, 1965), chap. 13. relationship derived from the estimated borrowings and excess reserves functions. In this computation, he takes the ratios of currency, time deposits, U.S. Gov ernment deposits, and nonmember bank demand de posits to money stock as constant. Thus the biases which were introduced into Teigen’s computations are again present here. DeLeeuw’s reserve aggregate is nonborrowed reserves plus currency in the hands of the public. His estimated long-run elasticities, valued at the sample means, are 0.172 and — 0.214 for the Treasury bill rate and the discount rate, respectively. When valued in 1962 (the end of his sample period) these elasticities are 0.245 and —0.348, respectively.1 4 These numbers seem quite compatible with those obtained by Teigen for approximately the same sample period. However, DeLeeuw finds that the entire ad justment of banks to portfolio changes takes place only gradually over time, and his impact elasticities for borrowings are only about one-fourth of the equi librium values. This suggests that if the data were available to compute the short-run elasticity for the money supply relationship, the estimates over a onequarter period would be considerably lower than those obtained by Teigen. The currency and time deposit demand equations which DeLeeuw incorporates into the above computa tions are almost completely insensitive to interest rate changes over a one-quarter horizon. This implies that over a one-quarter horizon changes in reserves avail able to support private demand deposits, which are caused by interest induced changes in currency and 14Ibid., p. 518. Page 14 time deposit demand, are negligible. Thus, it is highly probable that the assumptions of constant currency/ money stock or time deposit/money stock ratios result in a net upward bias in the computed interest elastic ity of the money supply relationship. DeLeeuw 11 In a subsequent study for the Brookings model DeLeeuw produced a condensed model of financial behavior in which the excess reserve and borrowings equations were aggregated into a single function to explain free reserves.1 In that study estimates of the 5 interest elasticity of the money supply relationship are not provided. However, using the free reserveinterest rate coefficient estimates and information given in the earlier study it is possible to replicate the computations of the larger model.1 8 The estimated impact elasticities at the sample means are 0.037 for the Treasury bill rate, and — 0.046 for the discount rate.1 The corresponding long-run 7 elasticities are 0.096 and — 0.118 respectively. The ab solute values are lower by a factor of almost fifty percent from the values obtained in the earlier study, even though the data and the sample period have remained essentially unchanged. It is likely that some downward bias has been in troduced into these estimates by aggregating excess reserves and borrowings into free reserves in the esti mation of the model. From the information presented in the first study, it is not possible to aggregate the interest elasticities of these two components. However, the early work suggests that the response of banks to a disequilibrium in borrowings from the Federal Re serve is much faster than the response to a similar situation with respect to excess reserves. Goldfeld The most detailed study of financial markets is found in the work of Stephen Goldfeld.1 In this 8 study, equations are specified for both the demand for excess reserves and the demand for borrowings from the Federal Reserve System. Separate equations are estimated for country banks and city banks. The Goldfeld results suggest very large ( in absolute value) interest elasticities for the borrowings equations 15Frank DeLeeuw, “ A Condensed M odel o f Financial Be havior,” in The Brookings Model: Some F u rth er Results, ed. James S. Duesenberry et al. (Chicago: Rand McNally, 1969), pp. 270-315. 16DeLeeuw, “ A Model of Financial Behavior.” 17Unless otherwise stated, data cited have been computed by this author. 18Stephen M. Goldfeld, Comm ercial Bank Behavior and Eco nomic Activity (Amsterdam: North-Holland, 1966). JULY 1 9 7 2 F E D E R A L R E S E R V E B A N K O F ST. LO U I S relative to those found by DeLeeuw. There do not appear to be large differences in the impact elasticities of borrowings across the bank classes, but the speed of adjustment to interest rate changes is much slower for borrowings by country banks than for city banks. This is reflected in the lower impact elasticities and the higher equilibrium elasticities for the country banks than the corresponding numbers for the city banks. The excess reserve interest elasticities reported by Goldfeld are negligible, particularly when compared with the borrowings elasticities. In addition, he finds that banks respond quite quickly to disequilibrium in excess reserve holdings. Thus, the long-run elasticities for excess reserve demand are not much different from the impact elasticities, particularly for city banks. This result is similar to that of the Teigen study where no evidence of a distributed lag in bank response was found. Goldfeld reports interest elasticities of a money sup ply relationship comparable to that derived by both Teigen and DeLeeuw. The impact elasticities, with respect to the Treasury bill rate and the discount rate in this function, 0.042 and — 0.029, respectively, are derived from the elasticities reported in Table II. The corresponding long-run elasticities are 0.222 and —0.076.1 These results are quite close to the values 9 reported by both Teigen and DeLeeuw for the Treas ury bill rate, but considerably below the estimates for the discount rate in the other studies. The sources of the differences are fairly conspicuous. In Teigen’s study, where there is no disaggregation of excess reserves from borrowings, the Treasury bill rate and the discount rate appear only as the differ ential between the two rates. Hence the regression coefficient of the discount rate is constrained to have the same absolute value, but with the opposite sign from that of the bill rate. Since the mean of the dis count rate for the sample period is slightly larger than that of the bill rate, the computed coefficient of elas ticity of the discount rate is, in effect, constrained to be slightly smaller in absolute value than that of the bill rate. DeLeeuw constrains this excess reserve specification to include only the differential between the bill rate and the discount rate. Given the con straints that are imposed in the estimation of the Teigen and DeLeeuw studies, it would seem reason able to conclude that the Goldfeld estimate of the response of the money supply relationship to discount rate changes is more reliable for this period. 19Ibid., p. 191. Table II Interest Elasticities of V ario u s Functions in the G o ld fe ld M odel Impact* Equilibrium Currency Demand Treasury B ill Rate Time Deposit Rate - .0 0 8 — .0 1 5 — .0 7 - .1 4 Time Deposit Demand Time Deposit Rate Long-Term Governm ent Rate .0 2 8 — .1 2 5 .3 7 — 1 .62 — .9 8 .8 8 — 2 .3 8 2 2 .1 3 4 — .88 .7 9 - 2 .9 2 6 2 .6 2 5 — .3 8 — .3 5 — .15 — .2 5 Specification Bank Borrowings C ity Banks Discount Rate Treasury B ill Rate Country Banks Discount Rate Treasury B ill Rate Excess Reserves C ity Banks Treasury B ill Rate Country Banks Treasury Bill Rate ♦The equilibrium elasticities fo r currency and tim e deposit demand are found on p . 160 o f the source cited below. The corresponding im pact elasticities are calculated by using the equilibrium elastici ties and regression coefficients o f the lagged dependent variables from Tables 5.7 and 5.9, respectively. B oth equilibrium and im pact elasticities fo r bank borrow in gs and excess reserves are found on p p . 150 and 149, respectively. S o u rce : Stephen M. Goldfeld, Com m ercial B ank B ehavior and E conom ic A c tiv ity (A m sterd am : N orth-H olland, 1966). There seem to be no major discrepancies in the estimated long-run responsiveness of the money sup ply to changes in the bill rate, but considerable vari ance exists among the short-run elasticity estimates. The most uncertain issue, on the basis of the evidence reviewed so far, is the source of the interest elasticity. Goldfeld suggests that the source is bank borrowing behavior, DeLeeuw suggests that it is bank behavior with respect to excess reserves, and Teigen does not attempt to discriminate between the two. Goldfeld and Kane There exists an additional study by Goldfeld and Kane which provides some independent information on the question of the interest elasticity of bank bor rowings from the Federal Reserve.2 This study is 0 based on weekly data from the period July 1953 to December 1963 and disaggregates banks into four classes — New York City, Chicago, Other Reserve City, and Country banks. They find that the estimated short-run (one week) Treasury bill rate elasticities range from a high of 0.56 for New York City banks to a low of 0.08 for Chicago banks. When aggregated over all classes of banks, the short-run interest elas 20Stephen M. Goldfeld and Edward J. Kane, “ The Determin ants o f Member-Bank Borrowing: An Econometric Study,” Journal of Finance (September 1966), pp. 499-514, and Stephen M. Goldfeld, “An Extension o f the Monetary Sector,” in The Brookings Model: Some F u rth er Results, ed. James S. Duesenberry et al. (Chicago: Rand McNally, 1969), pp. 317-360. Page 15 F E D E R A L R E S E R V E B A N K O F ST. L O U I S ticity for the banking system as a whole is found to be 0.21. Their reported long-run interest elasticities of borrowings range from 2.8 to 3.9.2 1 These estimates seem consistent with the results of the quarterly study by Goldfeld and tend to add to the uncertainty of the high excess reserve and low bor rowings elasticities reported by DeLeeuw. The only difficulty in reconciling the weekly estimates with the quarterly work of Goldfeld is the implied definition of long run. In the quarterly study, the long run is achieved only after serveral quarters have elapsed. In the weekly study, the implied long run is a period of several weeks. The possibility remains that long run in the two studies has two different meanings. How ever, it seems safe to conclude that borrowing behavior of banks is an important source of interest elasticity of the money supply relationship when the Treasury bill rate changes and the discount rate remains constant. Teigenll A quarterly study which deals with the period of the 1950s through the early 1960s is that of Teigen.2 2 The study contains supply elasticities only for the demand deposit component of the money supply. The results for the elasticity of the discount rate are not very different from those reported by Goldfeld, but the elasticity of the Treasury bill rate is consid erably lower than the results obtained by Goldfeld, DeLeeuw and Teigen’s earlier results. Brunner and Meltzer Karl Brunner and Allan Meltzer have estimated the interest elasticity of the money supply relationship using annual data over a sample period including the interwar and post-war periods.2 In the two stage 3 least-squares estimates of their “nonlinear” money sup ply hypothesis, they find that the elasticity of the money supply function with respect to the adjusted monetary base is insignificantly different from one. Therefore, the interest elasticities of this function can be interpreted as interest elasticities of the reserve multiplier. Their estimate of the Treasury bill rate elasticity is 0.66 and the estimate of the discount rate elasticity is —0.31.2 Since there are no lagged varia 4 2’ Goldfeld and Kane, “ The Determinants of Member-Bank Borrowing: An Econometric Study,” p. 512. 22Ronald L. Teigen, “An Aggregated Quarterly Model of the U.S. Monetary Sector, 1953-1964,” in Targets an d Indicators of M onetary Policy, ed. Karl Brunner (San Francisco: Chandler Publishing Company, 1969), pp. 175-218. 23Karl Brunner and Allan H. Meltzer, “ Some Further In vestigations of Demand and Supply Functions for Money,” Journal of Finance (M ay 1964), pp. 240-283. 24Ibid., p. 277. Page 16 JULY 1 9 7 2 bles in the equation, these estimates can be compared with the equilibrium elasticities derived from the studies which used shorter time intervals. Both elas ticities appear to differ from the implied equilibrium values of the quarterly studies by a factor of over two. Given the many difficulties in estimating distributed lag effects from time series data, such inconsistencies are not surprising. Estimates from Data Including Post-1965 Period The shortcoming of the studies discussed so far is that they are based on data generated in the 1950s and early 1960s. During the 1960s there were many changes in the environment in which the banking system operated which could have significantly altered (and presumably increased) the interest elasticity of the money supply relationship. These changes included the evolution of an active market for large negotiable certificates of deposit, the involvement of large banks in the Eurodollar market through borrowings from ( or lending to) their foreign subsidiaries, and the en trance of banks into the commercial paper market through parent one-bank holding companies. Unfortunately it is difficult to obtain empirical evi dence on many of these innovations since they were effectively legislated out of existence before enough data were generated to assess their effects. The im pact of the CD market can be assessed, along with the responsiveness of the banking system in terms of free reserves in the 1960s, through the quarterly finan cial model in the M.P.S. model.25 In addition, esti mates of the interest elasticity of the money supplyreserve relationship on a monthly basis can be obtained from a financial market model developed by Thomas Thomson and James Pierce.2 6 Evaluation of Quarterly M oney Supply Elasticities The quarterly M.P.S. model contains a financial sector which includes detailed specifications of the commercial loan market and the mortgage market, as well as specifications dealing with bank and nonbank behavior with respect to holdings of currency, time deposits and free reserves. The estimated elasticities for the latter set of functions are tabulated in Table 25This model is the publicly available version which de veloped out o f the Federal Reserve - M .I.T. - Pennsylvania econometric model project. 26Thomas D. Thomson and James L. Pierce, “ A Monthly Econometric Model of the Financial Sector” (A paper pre sented at the May 1971 meeting of the Federal Reserve System Committee on Financial Analysis). F EDERAL. R E S E R V E B A N K O F ST. L O U I S JULY 1 9 7 2 Table III Interest Elasticities of V arious Functions in the M .P.S. M odel Specification Impact Elasticity Currency Demand Treasury B ill Rale .0 0 3 7 Equilibrium Elasticity .0 2 6 Non-CD Time Deposit Demand Time Deposit Rate S&L — Mut. S av. Bk. Rate Treasury B ill Rate .3 — .2 — .1 5 2 .9 - 2 .0 — 1.4 CD Demand [1 9 6 9 V alu es] Treasury Bill Rate Commercial Paper Rate CD Rate — 6 .1 4 — 4 .2 8 1 1 .4 6 — 6 .1 4 — 4 .2 8 1 1 .4 6 - 2 .9 9 — 3 .9 5 - 6 .4 2 - 8 .4 7 3 .2 3 3 .4 8 6 .9 3 7 .4 6 Supply of CDs by Banks £ 1969 Values] CD Rate - 1 .0 6 Treasury B ill Rate .9 8 — 1.06 .9 8 Free Reserves Treasury B ill Rate 1965 Values 1969 Values Discount Rate 1965 Values 19 69 Values III. Both the CD demand and supply functions, which did not exist in the earlier studies, assume that the full response to an interest rate change takes place within one quarter. Thus the impact and equilibrium elasticities are equal. The CD demand function, in particular, indicates a highly sensitive response to interest rate changes, which is consistent with casual impressions of the nature of the CD market. However, these estimates are drawn from a considerably smaller sample than that for the rest of the specifications, and therefore there is less certainty about the stability of the func tions over time. The estimates for the currency demand equation and the demand equation for non-CD time deposits tend to confirm the DeLeeuw and Goldfeld results of extremely low impact elasticities. The time deposit function does suggest higher long-run elasticities than had been previously estimated. This appears attributa ble, in part, to the evolution of special forms of time deposit accounts, such as small consumer-type CDs, during the late 1960s.2 7 The M.P.S. model does not distinguish between excess reserves and borrowings of member banks, but does estimate a relationship between the Treasury bill rate, the discount rate, and free reserves. No con straints are applied to the coefficients of the two rates. In Table III, both the impact and the equilibrium 27The estimated function allows for a change in structure during the early 1960s, which indicates that the interest elasticities in the latter part of the sample period are about fifty percent higher than those estimated for the first part o f the sample period. elasticities of this function are considerably higher than those estimated in the earlier studies. This is partially due to the fact that the estimated function is linear, and therefore the value of the elasticity co efficient is not constant at all points along the func tion. Evaluation of the elasticity coefficient at the very high values of interest rates in 1969 gives estimates of the impact and equilibrium Treasury bill rate elastici ties which are fifty and twenty-five percent higher, respectively, than the values at 1965 interest rate levels. Even after accounting for the higher levels of interest rates in the late 1960s, it appears that differ ences in specifications and/or differences in sample periods have produced higher interest rate elasticity estimates for the free reserve relationship than had previously been found. Simulation experiments were performed with the M.P.S. model which permitted relaxation of restrictions under which interest elasticities of the money supply relationship were computed in the studies discussed above. First, in addition to the impact elasticities, the pattern of response of the money stock over time to a maintained change in the Treasury bill rate was com puted. The simulations were continued for eight quar ters, after which the computed elasticities settled down at close to the equilibrium values. Second, the response of the demand for currency and the demand for time deposits to the changes in interest rates can be included or excluded from the computation of the elasticities. Time deposit demand is split into large negotiable certificates of deposit and other time deposits. The inclusion of the currency and time deposit responses in the simulation is analo gous to a controlled experiment in which the nonbank private sector demand for bank demand deposits is shifted once and maintained in its new position. This shift is allowed to occur without any effect on the demand functions for time deposits or currency. This shift generates an initial change in interest rates. The changes in the money stock, which are observed over time, are the result of the interest rate induced port folio shifts by banks and the nonbank public, and they trace out the interest elasticity of the money sup ply relationship over various time intervals. Finally, elasticities are computed for both demand deposits and the M, money stock concept. The estimated elasticities from three sets of simu lations are presented in Table IV. These computations are generated under the assumption that the Federal Reserve would not impose a Regulation Q constraint which would prevent banks from offering new CDs at competitive rates. Page 17 F E D E R A L R E S E R V E B A N K O F ST. L O U I S If such constraints were effective, increases in the Treasury bill rate would cause a shift in the demand for CDs. At the constrained new issue rate for CDs the public would not renew outstanding certificates as they matured. Over time the stock of CDs would decline, and there could be a sizable reduction in the ratio of time deposits to demand deposits. The change in this ratio would, in turn, cause a fluctuation in the reserve multiplier. The observed result would also be highly sensitive to the initial conditions of the Treas ury bill rate relative to the Regulation Q ceiling, and the historical pattern of Regulation Q restraint. In the first section of Table IV, the interest elastici ties include the interest rate induced reactions in the public’s demand for currency, large certificates of de posit as well as other time deposits, and the interest elasticity of the commercial banking sector’s supply function for large certificates of deposit. The interest elasticity of Mj is consistently smaller than the inter est elasticity of the demand deposit component. This is because the model indicates a small negative re sponse of the demand for currency to changes in inter est rates. Hence, as interest rates increase and the amount of bank deposits available to the economy expands, there is an offsetting movement in currency balances outstanding. The exclusion of the nonbank private sector’s de mand for currency and time deposits other than large certificates of deposit lowers the interest elasticity of the money supply relationship.2 This is because a 8 rise (fall) in interest rates decreases (increases) the quantity demanded of both of these assets. This rela tionship is straightforward in the case of currency. For time deposits the expected equilibrium response would be for a large quantity of time deposits to be de manded with higher levels of all interest rates. The model postulates, however, that the rate which banks offer on non-CD time deposits responds quite slug gishly to changes in market interest rates. Thus the short-run effect is for disintermediation away from commercial bank time deposits. If the elasticity pat terns were computed over a longer time horizon, the elasticities in the first experiment would eventually become smaller than those for the second experiment. In all cases the impact elasticities are essentially the same size. These results can be compared with those from earlier empirical studies which do not include the CD 28This exclusion of currency and time deposit demand allows these demand functions to shift in such a way that the quantity demanded at the new Treasury bill rate is exactly equal to the quantity demanded at the original level of the Treasury bill rate. Page 18 JULY 1 9 7 2 Tab le IV M oney Su pply Elasticity C om putations ---M .P.S. M odel A. Currency, Time Deposits, & CDs Included Q u arter Demand Deposits Mj 1 .1 0 6 .1 7 5 .2 1 4 .2 4 0 .0 8 3 .1 3 7 .1 6 7 .1 8 8 2 3 4 First Y e ar A verag e .1 8 4 Second Y e a r Average B. .1 4 4 .2 7 9 .3 0 9 .3 1 7 ■337 5 6 7 8 .2 1 8 .2 4 0 .2 4 7 .2 5 0 .311 .2 3 9 Currency, CDs Included; Time Deposits Excluded Q uarter Demand Deposits Mi 1 2 3 4 .0 9 9 .1 5 7 .1 8 5 .2 0 4 .0 7 8 .1 2 3 .1 4 5 .1 5 9 First Y e a r A verage 5 6 7 8 Second Y e a r Average .161 .1 2 6 .2 3 3 .2 5 4 .2 5 6 .2 5 9 .181 .1 9 6 .1 9 8 .2 0 0 .251 .1 9 4 C. CDs Included; Time Deposits, Currency Excluded Q u arter Demand Deposits Mi 1 2 3 4 First Y e a r A verage 5 6 7 8 Second Y e ar A verag e .0 9 4 .1 4 9 .1 7 3 ■ 186 .0 7 4 .1 1 8 .1 3 6 .1 4 7 .151 .1 1 9 .2 0 2 .211 .2 2 0 ■ 226 .1 5 9 .1 6 6 .1 7 3 ■ 177 .2 1 5 .1 6 9 market. It appears from section C of Table IV that when the CD market is operating freely, the estimated interest elasticity of the money supply relationship differs little from the results drawn from studies of earlier periods. If anything, the elasticities reported in this section of the Table are generally lower than those discussed above. On the other hand, sections A and B of Table IV suggest that the net bias involved in computing the interest elasticities with a constant currency/deposit ratio and a constant level of time deposits tends toward zero. That is, the estimates ob tained under these assumptions give estimates of elasticities which are too low. Evaluation of Monthly Money Supply Elasticities The same type of analysis of the money stockreserve relationship as that performed with the M.P.S. F E D E R A L R E S E R V E B A N K O F ST. L O U I S quarterly econometric model can be carried out on a monthly basis using the financial market model of Pierce and Thompson. The results over an eighteen month period are presented in Table V. In this model, demand for currency by the public is specified to be completely interest inelastic, so the assumptions un derlying the calculations of sections B and C of Table IV are identical to the assumptions made for the right-hand column of Table V. The analogy to section A of Table IV is presented in the left-hand column of Table V. The implication of the monthly model is that the money stock —reserve relationship is slightly more elastic in the short run than the various quarterly estimates imply. The implied impact elasticity (over a one-month period in this case) is about 0.15. The average elasticity over this first twelve months is estimated at about 0.25, or about one-third more than the estimate over the corresponding four-quarter hori zon from the M.P.S. model. After eighteen months have elapsed the elasticity values reflect the long-run, or equilibrium, values. This horizon agrees reasonably well with the horizon of the M.P.S. model. Conclusions It is difficult to draw a finely defined set of conclu sions from the set of studies which have been ex amined. There exists a range of elasticity estimates among these studies which cannot be reconciled with the information which is readily available at the present time. However, while a single point cannot be estab lished as the most probable value for the interest elasticity of the money supply, it appears that the studies do provide information which can be of value in policy discussions concerning the control of the money stock. A broad, but valuable conclusion is that the interest elasticity of the money supply during the sample period of these studies appears to be ex tremely low. It seems appropriate to conclude with almost complete certainty that the long-run elasticity during this period was less than 0.5 and that the impact elasticity (one quarter) was probably no greater than 0.10 to 0.15. All these elasticities are relevant for policy actions which result in changes in the Treasury bill rate, while leaving the discount rate unchanged. For the class of policy actions which simultaneously alters the Treasury bill rate and the discount rate by the same amount from an initial position where the two are approximately equal, it is the sum of the interest elasticities of the money supply which is rele JULY 1 9 7 2 Table V M oney Supply Elasticity Com putations (M onthly Fin ancial M arket M odel) Time Deposits Included Time Deposits Excluded .1 3 8 .1 9 5 .231 .2 5 0 .2 5 2 .2 5 0 .1 3 7 .1 9 2 .2 2 6 .2 4 3 .2 4 4 .2 4 3 .2 1 9 .2 1 4 .2 6 6 .2 7 2 .2 7 5 .2 7 9 .2 8 4 .2 9 2 .2 5 6 .2 6 2 .2 6 2 .2 6 9 .2 7 2 .281 Second 6-Month A verag e .2 7 8 .2 6 7 Month 1 2 3 4 5 6 First 6-Month A verag e 7 8 9 10 11 12 13 14 15 16 17 18 Third 6-Month A verag e .3 0 3 .311 .2 3 6 .2 2 0 .2 3 3 .2 4 6 .2 5 8 .2 9 0 .2 9 6 .2 2 2 .2 0 2 .2 1 6 .2 3 0 .2 4 3 vant. Two of the studies suggest that the elasticity with respect to the discount rate is slightly smaller than that with respect to the bill rate. The estimation of these relationships involves constraints on parame ters, and hence, is not a valid test of the hypothesis that the two elasticities are significantly different. In the Goldfeld study, where there are no constraints imposed on the estimated parameters, the estimated coefficient of elasticity for the discount rate is con siderably smaller in absolute value than that of the bill rate. Therefore, it would appear that while the interest elasticity of the money supply relationship is likely to be smaller when both rates are changed simultaneously, it is almost certain that the coefficient of elasticity will remain positive. Furthermore, the elasticity under such a policy probably does not exceed one-half to two-thirds of the interest elasticity under a policy of keeping the discount rate fixed. The available evidence suggests quite conclusively that the short-run feedbacks through interest rate changes, which would be generated by policy changes in reserve aggregates, are very weak and should cause little, if any, difficulty for the implementation of policy actions aimed at controlling the money stock through the control of a reserve aggregate. Of course, the size of random fluctuations in the reserve multiplier re mains a major factor in determining the size of devia tions of the money stock from its targeted value. An issue which remains to be investigated is the size of the variance of the multipliers associated with various reserve concepts. Page 19 kjUBSGRIPTIONS to this Bank’s R e v ie w are available to the public without charge, including bulk mailings to banks, business organizations, educational institutions, and others. For information write: Research Department, Federal Reserve Bank of St. Louis, P. O. Box 442, St. Louis, Missouri 63166.