The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
158 MONTHLY REVIEW, JULY 1973 The Business Situation Business activity remains very strong although there are signs that capacity limitations are beginning to slow the advance. Industrial production rose further in May, but at a more moderate pace than that experienced earlier this year and in 1972. Inventory spending slowed in April; however, this slackening may have been largely unintended, as inventory-sales ratios remain inordinately low. The labor market displayed continued strength in June when the un employment rate declined to 4.8 percent, the lowest level in more than three years. Retail sales increased in May but then declined in June. The economy has faced a severe inflationary problem throughout 1973, with exceptionally rapid price increases at both the wholesale and retail levels. While some of these increases reflected a transitory price bulge, there are indications that the strength of demand has contributed substantially to the inflation. Many industries are reported to be operating at or near capacity, and there is evidence of serious delivery delays and shortages of more and more products and materials. On June 13, President Nixon imposed a freeze on virtually all prices, with the exception of prices of raw agricultural commodities sold at the farm level, for a maximum duration of sixty days. The latest price statistics, reflecting developments before the imposition of the price freeze, are very disturbing. Consumer prices increased in May at an annual rate in excess of 7 percent. Wholesale prices surged in that month and rose even more rapidly in June at a 31 percent rate, bringing wholesale price rises for the first half of the year to an extraordinary seasonally adjusted annual rate of 22 percent. While it will be some time before the impact of the freeze is evident in the price statistics, there are reports that serious distortions and increased shortages have al ready arisen in some industries. To avoid further distor tions, it is desirable that the freeze end as quickly as pos sible. The Administration is in the process of devising a controls program more stringent than Phase Three to deal with inflation, but these measures have not yet been an nounced. INDUSTRIAL PRODUCTION, CAPITAL SPENDING, ORDERS, AND INVENTORIES Industrial production increased at a somewhat reduced pace in May. According to the Federal Reserve Board index, production climbed at a 5.9 percent seasonally adjusted annual rate in that month. This marked the smallest expansion in industrial output in ten months and came on the heels of a revised April hike of about 8 percent. Over the first five months of the year, indus trial production has risen at an 8.5 percent pace by comparison with an increase in excess of 10 percent over 1972 as a whole. This recent slowing notwithstanding, production has now advanced steadily on a month-bymonth basis since October 1971. The last time output increased consistently over such an extended period was the interval from November 1964 through October 1966 (see Chart I). Output of business equipment and consumer goods increased considerably in May. Production of business equipment climbed at a 12.3 percent annual rate following an even more rapid expansion in April. Over the first five months of 1973, business equipment output has risen at an annual rate of 16.2 percent, well above the 13.4 percent rate of advance posted in 1972. Production of consumer goods increased at an 8.3 percent rate in May after a very small gain in the previous month. On the other hand, output of industrial materials grew at a much slower annual rate of 2.8 percent. However, this slowing may be a reflection of capacity limitations in some industries which have precluded firms from expanding output to meet demand. Output of defense and space equipment fell off again in May after spurting in April. According to the Commerce Department’s most recent estimates, businesses increased their expenditures on plant and equipment by $4.3 billion in the first quarter to a seasonally adjusted annual rate of $96.2 billion. This gain was about equal to that registered in the fourth quarter of 1972. For 1973 as a whole, the latest Commerce survey FEDERAL RESERVE BANK OF NEW YORK indicates a rise in outlays for plant and equipment of 13.2 percent, slightly less than the increase recorded in their previous survey. Nevertheless, if realized, this would be the largest expansion in capital spending since 1966. Manu facturers’ outlays for plant and equipment are anticipated to increase 18.5 percent over the year, up sharply from the 4.5 percent rise posted last year. Spending by non manufacturing firms is projected to advance 10.3 percent, down slightly from the 11.5 percent gain in 1972. Orders placed with manufacturers of durable goods rose $0.9 billion, or 2.2 percent in May. This gain constitutes some resurgence in orders following an April increase of less than 1 percent. Nevertheless, the past two months represent, on balance, a slowdown in new bookings rela tive to the extraordinarily rapid advances earlier in the year. (It should be noted that the data for new orders have recently been revised to reflect bench-mark and sea sonal factor changes.) About one third of the May increase in orders was attributable to a large rise in bookings for primary metals. Bookings for capital goods industries increased in the month despite a decline in defense orders. The backlog of unfilled orders continued to rise strongly during May as it has for many months. Some of the buildup in the backlog may reflect overbooking, however, as delivery times have lengthened and some products and materials have come into short supply. In April, the book value of total business inventories increased $1.2 billion following gains which averaged more than $1.8 billion per month over the first quarter of the year. The April inventory accumulation was the small est since December. Moreover, it is doubtful whether this advance represents much, if any, increase in the physical volume of stocks, since it probably largely stems from sharply higher prices for replacement goods put into stock. Chart I INDUSTRIAL PRODUCTION Se aso nally adjusted; 1967=100 Percent Percent Note: Shaded areas represent recession periods, according to the National Bureau of Economic Research chronology. The dates of the 1969-70 recession are tentative. Source: Board of Governors of the Federal Reserve System. 159 160 MONTHLY REVIEW, JULY 1973 In any event, the book value of manufacturers’ inven tories rose modestly in April, as did retail and whole sale stocks. The ratio of inventories to sales for all busi nesses remained at a low level, still suggesting that a sub stantial buildup in inventories may be in the offing. Preliminary data indicate that the book value of manu facturers’ inventories rose $0.9 billion on a seasonally adjusted basis in May. Much of this increase in inven tory spending was in durable goods industries. Durable shipments moved ahead by $0.5 billion in May, but nondurables shipments dropped off slightly. The overall inventory-sales ratio for manufacturing was virtually un changed in May. Chart II SALES AND INVENTORIES OF SINGLE-FAMILY HOMES Seasonally adjusted Thousands of units Thousands of units RESIDENTIAL CONSTRUCTION, RETAIL SALES, CONSUMER CREDIT, AND PERSONAL INCOME Housing starts rebounded in May after declining for three consecutive months. The increase, to 2.4 million units at a seasonally adjusted annual rate, reflected a small gain in starts of single-family units and a substantial jump in multifamily starts. Newly issued building permits edged up in May as well, although, except for April, they were at their lowest level since early 1971. Notwithstanding these May increases, there are signs of overbuilding in housing which suggest a further diminution in activity in the months ahead. While inventories of new unsold one-family homes were virtually unchanged in April (the latest data available), the number of unsold units remained well above levels posted in 1972 (see Chart II). Meantime, sales of new one-family homes dropped sharply to a seasonally adjusted annual rate of 667,000 units in April. This represented the lowest sales rate in over one year. Because of the sharp decline in sales, inventories of unsold new homes reached 7.9 months of sales, the highest level since the series began in January 1963. Over the first four months of 1973, the inventory-sales ratio for new homes averaged IVz months, considerably above the already high average of 6 months posted in 1972. The April decline in mobile home shipments may also be an indication of the expected slowdown in housing over the final months of 1973. Recent data provide evidence of some moderation in the growth rate of consumer spending. Of course, season ally adjusted retail sales had grown at a very rapid 26 percent annual rate over the first three months of the year, and this pace was clearly unsustainable. After declining in April, retail sales rose in the next month but then fell again in June to $41.3 billion. Sales in each month of the second quarter remained below their March peak. For the most part, the slowdown in consumer spending has been Note: New-home sales are expressed as an annual rate; the inventory of unsold homes is the stock at the end of the month. Source: United States Department of Commerce, Bureau of the Census. concentrated in durable goods, particularly automotive products. In June, unit sales of new domestic-type auto mobiles moderated from their rapid May pace to a 9.1 million unit seasonally adjusted annual rate. During the first six months of 1973, sales of domestically pro duced automobiles averaged 10 million units. Sales of imported cars were at an annual rate of 1.8 million units in June, compared with 1.9 million units in both April and May. Another hefty advance was recorded in consumer credit in May, when the stock of total consumer debt out standing rose $2.2 billion on a seasonally adjusted basis. By comparison, total consumer credit had increased $1.7 billion in April and on average $2.1 billion over the first three months of the year. In recent months, the apparent leveling-off in the rate of growth of total credit reflected the noticeable pickup in the rate of repayments of instalment debt, which may in turn have been related to the unusually large Federal income tax refunds. Personal income rose by a relatively modest $4.8 billion in May to $1,012 billion at a seasonally adjusted annual rate. Over the first five months of the year, personal income has climbed an average of $5.9 billion per month. FEDERAL RESERVE BANK OF NEW YORK In May, a gain in wage and salary disbursements of $3.1 billion accounted for much of the rise, with the remainder of the increase resulting primarily from gains in interest income and in transfer payments. LABOR MARKET DEVELOPMENTS There are clear indications of continued strength in the labor market. According to the survey of households con ducted by the Department of Labor, civilian employment increased sharply in June, after rising modestly in the two preceding months, and the overall rate of unemployment dropped to a seasonally adjusted 4.8 percent, the lowest it has been in more than three years. Data on the unem ployment rates of major age-sex groups indicate that the unemployment rate for adult men edged down to 3.2 percent in June, while the more volatile teen-age unemploy ment rate fell dramatically (see Chart III). On the other hand, the unemployment rate for adult women rebounded to the level prevailing in February and March of this year. Payroll employment rose by about 200,000 workers in June, a somewhat smaller gain than those recorded on C hart III SELECTED UNEMPLOYMENT RATES Percent Se aso nally adjusted Source: United States Department of Labor, Bureau of Labor Statistics. Percent 161 average in earlier months of the year. Nevertheless, the 4 percent annual rate of growth in payroll employment during the first half of 1973 is slightly above the pace of expansion of the previous year, and it may well be that the strength of the labor market is greater than this comparison sug gests. The data on labor turnover rates in the manufactur ing sector of the economy provide support for this view. The rate at which workers are newly hired climbed to 4.3 per 100 employees in May, the highest level since early 1966. Throughout the first five months of the year, the new hire rate was running more than 23 percent above the average for 1972. At the same time, there was a marked expansion in job vacancies. Over the first five months of 1973, the increase in seasonally adjusted job vacancies amounted to almost 40 percent at an annual rate. PRICES AND WAGES The economy has faced a severe inflationary problem throughout 1973. Wholesale prices soared at a 20.5 percent seasonally adjusted annual rate over the first five months of the year, while consumer prices advanced at a very disturbing 8.2 percent rate. To be sure, some of these increases doubtless reflected a transitory price bulge, following the introduction in January of the largely volun tary Phase Three of the controls program. Moreover, a bunching of price increases probably resulted as some firms raised prices in anticipation of the adoption of more stringent controls. Nevertheless, there are also indications that the recent strength of demand has contributed mea surably to the inflation. Many industries are reported to be operating at or near capacity, and there is evidence of serious delivery delays and shortages of more and more products and materials. In this environment, President Nixon announced on June 13 a freeze on virtually all prices, with the exception of prices of raw agricultural commodities sold at the farm level, for a maximum duration of sixty days. The Admin istration intends to use this period to devise a controls program to replace Phase Three. During the freeze, prices are limited to their highest levels reached between June 1 and June 8. Wages and interest rates were not frozen but remain subject to Phase Three guidelines; in another development, the Committee on Interest and Dividends recently liberalized the rules regarding corporate dividend payments. The impact of the freeze will not be evident in the price statistics until the July data are released sometime in August. The latest data, covering developments before the imposition of the freeze, present a very distressing picture. In June, wholesale prices rose at a disastrous 31.2 percent 162 MONTHLY REVIEW, JULY 1973 seasonally adjusted annual rate. Wholesale prices of indus trial commodities alone climbed at a 12.4 percent rate, about the same rate of advance as that experienced over the preceding five months. Meanwhile, wholesale prices of farm products and processed foods and feeds renewed their unprecedented advance in May and then surged up ward at a seasonally adjusted annual rate in excess of 79 percent in June. These last months have pushed wholesale agricultural prices up at an almost unbelievable 47.5 per cent annual rate over the first six months of the year. Consumer prices continued to spiral upward in May, advancing at a 7.3 percent seasonally adjusted annual rate. Price increases at the consumer level have slowed somewhat over the past few months, leaving the May gain more than 1 percentage point under the rate of increase posted over the first four months of the year. By other standards, of course, this advance is still very rapid. For example, consumer prices rose 3.4 percent over 1972. The May advance in food prices, at a 14 percent season ally adjusted annual rate, is substantially below the 26 percent rate of increase registered during the first four months of 1973. Nevertheless, it is almost three times as great as the increase in food prices in 1972. Nonfood commodity prices, climbing at a 5 percent seasonally adjusted annual rate in May, showed little change from their pace of the previous three months. Higher prices for apparel, used cars, and gasoline accounted for most of the May rise. Prices of consumer services, which are not seasonally adjusted, advanced at a 4.5 percent annual rate in May. Recent hikes in wage rates have continued to be fairly moderate. In June, the average hourly earnings of workers in the private nonfarm sector rose at a 6 percent season ally adjusted annual rate. Adjusted for overtime hours in the manufacturing sector and for shifts in the composition of employment among industries, the rise in average hourly earnings was a more rapid 7.7 percent. So far this year, however, the pattern of monthly advances in hourly earn ings has been rather erratic, perhaps because the timing of pay raises has been affected by the controls program. As a result, it may be preferable to examine the growth in wage rates over periods of several months. Over the five months ended in June, adjusted average hourly earn ings of workers in the private nonfarm economy have increased at a 5.8 percent annual rate, the same rate as that experienced during the preceding twelve-month period. FEDERAL RESERVE BANK OF NEW YORK 163 The Money and Bond Markets in June June was marked by a broadly based advance in interest rates, with rates for the shorter maturities reaching levels that had not been experienced since 1970. Concern about inflation continued to mount over the month. Reports early in June that the Nixon Administration was contem plating stronger price controls temporarily buoyed the securities markets. However, the President’s June 13 announcement of a price freeze had little initial impact on yields. Interest rates resumed their climb shortly thereafter as market participants doubted that the new moves would significantly reduce underlying inflationary pressures. Short-term rates of interest experienced the largest increases in June, and yields on long-term securi ties joined in the general advance in rates in the second half of the month. An increased volume of new issues contributed additional upward pressure on yields in the corporate and municipal bond markets. During June the Federal Reserve took a series of steps to slow the rise in the money and credit aggregates. Around midmonth, the Federal Reserve discount rate was increased by V2 percentage point to 6 V2 percent. Then, at the end of June, the discount rate was raised another Vi percentage point to 7 percent, its highest level in more than fifty years. At this time, reserve requirements on most demand deposits at member banks were raised V2 per centage point. Earlier in the month, reserve requirements had been imposed upon finance bills. For reserve pur poses, these bills were placed on the same footing as large certificates of deposit (CDs) and bank-related commercial paper. Mi—defined as demand deposits adjusted plus currency outside banks—and M2, which also includes time and savings deposits other than large CDs, expanded more rapidly than desired in June, despite efforts by the Federal Reserve to restrain the growth in nonborrowed reserves. Similarly, growth of the adjusted bank credit proxy re mained rapid, although growth was at a somewhat more moderate pace than the expansion experienced earlier in the year. BANK RESERVES AND THE MONEY MARKET Interest rates on money market instruments rose markedly in June, reflecting strong demands for credit and Federal Reserve pressure on bank reserve positions against a background of growing concern over the rate of inflation. The average effective rate on Federal funds reached 8.49 percent in June, compared with 7.84 per cent in May. By the end of the month, the rate was around 9 percent. Commercial paper rates climbed steadily throughout the month. The rate on 90- to 119day commercial paper advanced 1 percentage point over the month, and closed at 8 V2 percent (see Chart I). Rates quoted by dealers in bankers’ acceptances also adjusted sharply higher during June, increasing by lVs percentage points. The commercial bank prime lending rate for large business borrowers was raised to 7% percent in two X A percentage point steps in the first and third weeks in June and was lifted to 8 percent on July 2. However, given the greater increases in commercial paper rates, the prime rate remained relatively attractive and continued to en courage heavy business loan demand. On June 8, the Board of Governors of the Federal Reserve System announced that it had approved increases in the discount rate to 6 V2 percent from 6 percent at ten of the twelve Federal Reserve Banks, including the New York Bank, effective June 11. By the end of that week, the higher discount rate had become uniform throughout the System. The action was taken in recognition of in creases that had already occurred in other short-term rates. The Board also indicated that the decision to approve a rise in the discount rate to a level that had not been equaled since mid-1921 reflected concern over the recent growth in money and bank credit and the continuing rise in the general price level. Borrowings by member banks from the discount window were $1.79 billion on average in June (see Table I), somewhat below the $1.84 billion borrowed in May. Two steps were taken at the end of June by the Federal 164 MONTHLY REVIEW, JULY 1973 Chart I SELECTED INTEREST RATES Percent M O N E Y M A R K E T RA T ES A p ril - June 1973 1973 B O N D M A RK E T YIELD S Percent 1973 Note: Data are shown for business days only, M O N E Y MARKET RATES QUOTED: Bid rates for three-month Euro-dollars in London; offering rates (quoted in terms of rate of discount) on 90- to 119-day prime commercial paper quoted by three of the five dealers that report their rates, or the midpoint of the range quoted if no consensus is available; the effective rate on Federal funds Ithe rate most representative of the transactions executed); closing bid rates (quoted in terms of rate of discount) on newest outstanding three-month Treasury bills. B O N D MARKET YIELDS QUOTED: Yields on new Aoa-rated public utility bonds are bdsed on prices asked by underwriting syndicates, adjusted to make them equiv.alent to a Reserve in a further effort to slow the advance in money and credit. First, the discount rate was raised to 7 percent from 6 V2 percent. This rate equaled that charged in late 1920 and early 1921 and is the highest in Federal Reserve history. In conjunction with the increase in the discount rate, the Board announced an increase in reserve require ments on member bank demand deposits. Reserve require ments were raised by V2 percentage point on all but the first $2 million of deposits at member banks. The new requirements will become effective during the reservecomputation period beginning July 19 and will apply to net demand deposits held in the week ended July 11. The higher requirements will absorb approximately $800 mil standard A a a bond of at least twenty years' maturity; daily averages of yields on seasoned Aoa-rated corporate bonds; daily averages of yields on long* term Government securities (bonds due or callable in ten years or more) and on Government securities due in three to five years, computed on the basis of closing bid prices; Thursday averages of yields on twenty seasoned twenty.year tax-exempt bonds (carrying M oody’s ratings of A a a, Aa, A, and Baa). Sources: Federal Reserve Bank of New York, Board of Governors of the Federal Reserve System/ M oody’* Investors Service, Inc., and The Bond Buyer. lion of reserves.* Earlier, on June 18, the Board of Governors had an nounced that it was carrying out its proposal to impose reserve requirements on funds raised by member banks through the sale of finance bills. The Board amended 511 The new reserve requirement structure is: On net demand deposits of Reserve percentage applicable First $2 million or less .............................. 8 percent (unchanged) IOV2 percent Over $2 million to $10 million ................ Over $10 million to $100 million ............ HV 2 percent Over $100 million to $400 million .......... 13V2 percent Over $400 million ........................................ 18 percent FEDERAL RESERVE BANK OF NEW YORK Regulation D, which governs member bank reserves, to apply a basic 5 percent reserve requirement to finance bills. In addition to the basic requirement, a 3 percent marginal reserve requirement is being applied to the combined total of finance bills, large-denomination CDs, and bank-related commercial paper to the extent that the total exceeds the level outstanding during the week ended May 16 or $10 million, whichever is larger. Finance bills outstanding in the week beginning June 28 were included in reserve calculations, and the banks are re quired to hold the additional reserves in the week begin ning July 12. The monetary aggregates continued to grow at a rapid pace in June. Preliminary estimates indicate that advanced at a seasonally adjusted annual rate of HV 2 percent in that month. This advance brought the growth rate in Mx for the second quarter to about 10V4 percent, and growth for the year ended in June to IV 2 percent (see Chart II). M2 advanced at an estimated 10 percent seasonally adjusted annual rate in June, slightly faster than the pace for the second quarter and for the last twelve months. The adjusted bank credit proxy continued to grow rapidly in June, increasing at an estimated 11 percent seasonally adjusted annual rate. This is, however, a slightly slower pace than that of the preceding four months. Large CDs, which had been expanding at an explosive 103 percent annual rate during the first five months of 1973, grew only modestly in June. Because banks had been restricted to selling short-dated CDs be tween the end of February and mid-May by the Regulation Q ceilings on CDs with initial maturities of ninety days or more, they were faced with an unusually large volume of maturing CDs in June. In addition, corporations relied heavily on CDs to meet their June 15 tax obligations. In attempting to replace the maturing CDs, many banks com peted for short-term funds by raising offering rates on CDs of under six months’ maturity a full percentage point or more. In contrast, after an initial upward adjustment following the removal of interest rate ceilings on May 16, rates on longer dated CDs have advanced very little and are now generally lower than rates on the one-month to six-month maturities. Growth in reserves available to sup port private nonbank deposits (RPD) accelerated to an estimated 15 V2 percent rate in June, bringing growth in this series to 11 percent for the first half of 1973. THE GOVERNMENT SECURITIES MARKET Treasury bill rates rose sharply in June. Advances were concentrated in the shorter maturity ranges during most 165 Table I FACTORS TENDING TO INCREASE OR DECREASE MEMBER BANK RESERVES, JUNE 1973 In millions of dollars; (-f) denotes increase (—) decrease in excess reserves Changes in daily averages— week ended Net Factors chanaes June 6 June 20 June 13 June 27 “ Market” factors Member bank required reserves................. Operating transactions (subtotal) ............. Federal Reserve float .............................. Treasury operations* .............................. Gold and foreign account ...................... Currency outside banks .......................... Other Federal Reserve liabilities and capital ............................................... 243 809 568 548 57 + — 278 + + + + 4 - 220 4-1,455 — — 21 + +1,650 — 4- i i + — 418 — 533 536 186 244 7 365 + 128 —1,013 + 111 —1,399 31 — + 302 85 4 . 232 — 120 + 4 Total “ market" factors ........................ -f-1,052 —1— 1.675 —1,069 - 885 — + + + + + 58 715 844 555 44 759 + 31 + 773 Direct Federal Reserve credit transactions Open market operations (subtotal) ......... Outright holdings: Treasury securities .................................. Bankers' acceptances .............................. Federal agency obligations ..................... Repurchase agreements: Treasury securities .................................. Bankers' acceptances .............................. Federal agency obligations ..................... Member bank borrowings............................ Seasonal borrowings! .............................. Other Federal Reserve assets $ ................. 140 —2,156 + _ _ + + + — + + Total § ........................................................ Excess reservest .................................. + 968 + 951 901 —1,369 + 604 41,060 2 + 8 — 10 _ 1 2 — 6 — 14 + 229 565 50 156 736 18 59 — 565 + — 50 + — 156 + 4 - 33 + 43 + 4 - 30 + 291 268 25 — 23 64 — 48 231 — 79 6 + 23 47 + 59 377 606 19 4* 207 _ + + + 23 2 + + 16 551 50 195 — 817 —2,093 +1.246 + 937 - 727 177 + 52 + 46 235 — 418 + Monthly averages Daily average levels Member bank: Total reserves, including vault c a s h j........ Required reserves ....................................... Excess reserves§ ........................................... Total borrowings ......................................... Seasonal borrowings! .............................. Nonborrowed reserves .................................. Net carry-over, excess or deficit (—)#••• 32,218 31,580 31,817 31,597 401 — 17 1,664 1,697 64 67 30,554 29,883 135 233 32,290 32,130 160 1,928 73 30,362 58 32,214 32,002 212 1,849 96 30,365 95 32,07611 31,88711 18911 1,7851! 7511 30,2911! 13011 Note: Beoause of rounding, figures do not necessarily add to totals. * Includes changes in Treasury currency and cash, t Included in total member bank borrowings. t Includes assets denominated in foreign currencies. § Adjusted to include $172 million of certain reserve deficiencies on which penalties can be waived for a transition period in connection with bank adaptation to Regulation J as amended effective November 9, 1972. The adjustment amounted to $450 million from November 9 through December 27, 1972 and $279 million from December 28, 1972 through March 28, 1973. II Average for four weeks ended June 27. % Not reflected in data above. 166 MONTHLY REVIEW, JULY 1973 of the month, but rates on longer term bills climbed substan tially during the final days. Much of the increase in bill rates reflected the advances in money market interest rates generally. Factors that had previously shielded the bill market from some of the upward pressures in other markets have largely disappeared. Foreign demand for Treasury bills has slackened, as foreign central banks have generally not been accumulating dollars in the foreign exchange markets. In addition, seasonal needs led the Treasury to run down balances at the Federal Reserve during the first half of June. To counteract the expansion ary impact these reductions have on reserves, the Federal Reserve sold sizable quantities of Treasury bills in the market. Furthermore, sharp increases in the cost to Gov ernment securities dealers of financing their inventories have made them reluctant to hold large quantities of bills. Anticipations that the Administration would take steps to counteract inflation gave slight pause to the upward Chart II CHANGES IN MONETARY AND CREDIT AGGREGATES S e a so n a lly adjusted an n u a l rates' Percent •ercent_______________________________________________ ,sf M^l ^ y . / \ ^ / / From 12 V months earlier j 1 1 1 1 1 1 1 1 1 1 1 ij \ M2 _ / p5 From 3 months earlier /"1 0 V 1 1 1 l 1 1 1 1 1 1 _ M i i i 1 From 3 mlonths earlier \ 1 J L __ y r ~ From 12 months earlier ^ 1 1 1 1 1 1 1 L 1 l_ L i i 1 i i .1 m III I L L l.J _ 1972 Note: Data for June 1973 are preliminary. M l = Currency plus adjusted demand deposits held by the public. M 2 = M l plus commercial bank savings and time deposits held by the public', less negotiable certificates of deposit issued in denominations of $100,000 or more. Adjusted bank credit proxy'= Total member bank deposits subject to reserve requirements plus nondeposit sources of funds, such as Euro-dollar borrowings and the proceeds of commercial: paper issued by bank holding companies or other affiliates. Sources: Board of Governors of the Federal Reserve System and the Federal Reserve Bank of New York. Table n AVERAGE ISSUING RATES* AT REGULAR TREASURY BILL AUCTIONS In percent Weekly auction dates— June 1973 Maturities Three-month Six-month .. June 4 June 11 June 18 June 25 7.133 7.210 7.129 7.172 7.263 7.255 7.228 7.299 Monthly auction dates— April-June 1973 Fifty-two weeks ................................ April 24 May 24 June 26 6.598 6.818 7.235 * Interest rates on bills are Quoted in terms of a 360-day year, with the discounts from par as the return on the face amount of the bills payable at maturity. Bond yield equivalents, related to the amount actually invested, would be slightly higher. movement in bill rates early in June. However, the discount rate increase announced June 8 and the feeling that the new controls would not reduce the need for a restrictive monetary policy led quickly to a resumption of the upward trend in bill rates, which continued over the remainder of the month. Advances in the average issuing rates established at the weekly bill auctions were initially greatest for the shorter term issues. At the June 18 auction, the average issuing rate for the three-month bill, at 7.263 percent, actually moved above that for the six-month bill (see Table II). The rate for three-month bills retreated slightly at the next auction but still finished 53 basis points higher than the rate set at the final auction in May. The rate for the sixmonth bill was 44 basis points above its month-earlier level at the June 25 auction, while the yield on 52-week bills in the June 26 auction was 7.235 percent, 42 basis points above the month-earlier level. Rates on all maturities adjusted even higher at the end of the month to levels that had not been experienced since early 1970. Yields on longer term Treasury coupon issues were insulated to some extent from the persistent rise in bill rates. Price increases during the first half of June were reversed later in the month and, on balance, prices on most issues fell slightly. Over the month as a whole, yields on three- to five-year issues rose by an average of 22 basis points while yields on longer term issues increased by about 7 basis points. The steep rise in short-term interest rates relative to FEDERAL RESERVE BANK OF NEW YORK 167 longer rates in the last few months has led to the so-called inverted yield curve in which short-term securities offer higher returns than longer dated issues. The interest rate pattern that normally prevails is one in which yields in crease with the term to maturity. However, short-term interest rates tend to be subject to wider fluctuations than longer rates, and an inverted or “humpback” yield curve is typical in extended periods of robust economic activity and large credit demands. Chart III, which was constructed from dealer bid prices on representative issues, illustrates that an inverted yield curve emerged in the first quarter of 1973 and has subsequently become more pronounced. The market for Federal agency securities remained active in June. On June 6, the Federal Home Loan Bank Board offered $600 million of 35-month bonds priced to yield 7.20 percent. The issue sold out quickly and was soon trading at a premium. Late in the month, the Federal National Mortgage Association raised $500 million in new cash through the sale of 4 Vi-year debentures yielding 7.25 percent. The issue was well received. OTHER SECURITIES MARKETS Prices of corporate and municipal bonds fell in June in the face of continuing increases in short-term interest rates and an expanding new-issue calendar. There was a brief rally during the week and a half preceding the Presi dent’s price freeze announcement, based upon anticipations that a strong program of controls might serve to reduce the need for monetary restraint and thereby reduce interest rates. Following the speech the market initially retained its price gains, but subsequently prices resumed their down ward drift. After several months in which there was only a limited volume of new corporate offerings, the new-issue calendar picked up sharply in June. The rally that began June 5 aided sales of two thirty-year A-rated utility bond issues, marketed June 5 and 6 and yielding 7.88 percent and 7.85 percent, respectively. However, in the succeeding week, several utility offerings carrying either straight Aa ratings or split A a/A ratings (Moody’s/Standard and Poor’s) met with poor receptions at yields ranging from 7.65 percent to 7.70 percent. On June 20, underwriters priced a $250 million issue of Aaa-rated debentures of a Bell Telephone subsidiary to yield 7.75 percent. These forty-year debentures offered a return that was Vs percent age point above the late-May offering of another Bell Telephone subsidiary despite the higher rating on these new debentures. With the more generous yield, the deben tures sold well. This good reception also stimulated sales of older issues but did not prevent yields from increasing further in the final week. On June 25, thirty-year bonds of an A-rated power company sold slowly despite a yield of 7.97 percent, the highest yield on a major A-rated utility issue since September 1971. Prices in the tax-exempt sector fluctuated in response to the outlook for economic controls. On June 20, an offering of $100 million of A-rated 35-year bonds was priced to yield 6.00 percent. This yield proved to be ex ceedingly popular, and these bonds were soon trading at a premium. In the final week, the market was faced with an unusually heavy calendar, and new issues received mixed receptions despite higher yields. Yields on outstand ing tax-exempt issues dipped early in the month but ad vanced later in the period. The Bond Buyer index of twenty tax-exempt bonds climbed from 5.13 percent on June 7 to 5.25 percent on June 28. Dealer inventories continued on the low side but expanded slightly as the month progressed. The Blue List of dealers’ advertised inventories increased by $31 million to $694 million over the month of June. MONTHLY REVIEW, JULY 1973 168 Treasury and Federal Reserve Foreign Exchange Operations Interim Report* By C harles After an unprecedented rush into foreign currencies on March 1, all major exchange markets were officially closed and a series of emergency meetings was called to resolve the crisis. On March 11, six of the members of the European Community (EC) and, later, Norway and Sweden agreed to maintain fixed exchange rate relation ships among themselves within a IVx percent band, which would be permitted to float as a bloc against the dollar. In conjunction with the decision to establish a fixed-rate bloc, the German authorities revalued the mark by 3 per cent. In addition, to guard against large new inflows of funds, most countries participating in the joint float tight ened and extended their existing exchange controls. The Japanese yen, Swiss franc, sterling, and the Italian lira each continued to float independently. The markets were officially reopened on March 19 and, after a brief burst of trading as the backlog of commercial orders was cleared away, activity thinned as dealers paused to assess the radically altered market trading arrange ments. Through the first week in May, the dollar improved hesitantly as earlier adverse leads and lags were partially reversed. The February devaluation and subsequent flight from the dollar had been a shattering blow to confidence, however, and there was no large sustained covering of short dollar positions or reflow of funds. Despite an im proving trend in the United States balance of payments and the frequently voiced view that the dollar was, if any thing, now undervalued, the market became increasingly concerned over the worsening United States inflation, fore casts of vastly higher energy imports, and possible rami * This interim report, covering the period March through May 1973, is the first of a series providing information on Treasury and System foreign exchange operations to supplement the regular series of semiannual reports appearing in this Review. A. Coom bs fications of the Watergate affair. Consequently, the ten dency to shift out of dollars continued. In early May, the dollar began to depreciate once more against most of the European currencies. By midmonth a new speculative attack had broken out in which soaring gold prices, sliding United States equity prices, and a weakening dollar interacted to reinforce each other. Move ments in exchange rates were abnormally large and erratic, with spot rates fluctuating as much as 2 percent during a single day. The German mark advanced the most sharply in response to progressive tightening of Germany’s antiinflationary fiscal and monetary policies. After moving from the bottom to the top of the European band, the mark began by early June to exert upward pressure on the entire band. The dollar was driven down in sporadic bouts of ner vous and, at times, heavy trading to levels unjustified and undesirable on any reasonable assessment of the longer run outlook for the United States payments position. By June 28, the dollar had dropped 16 V2 percent against the mark, some 8 to 13 percent against most other Continental currencies, and the gold price in London had shot up FEDERAL RESERVE SYSTEM DRAWINGS AND REPAYMENTS UNDER RECIPROCAL CURRENCY ARRANGEMENTS In millions of dollars equivalent Transactions with Subsequent System swap drawings ( + ) System swap drawings drawings or repayments outstanding on (— ) through outstanding on March 9,1973 May 31,1973 May 31,1973 National Bank of Belgium ................ 390 -< ^ - 390 Swiss National Bank ......................... 565 - 0— 565 600 - 0- Bank for International Settlements (Swiss francs) ................................ Total ........................................ 1,555 600 1,555 FEDERAL RESERVE BANK OF NEW YORK from $90 to a peak of $127. The dollar also weakened against sterling and the Italian lira, though by much smaller amounts. Sterling advanced along with Continental currencies in May but, as British interest rates eased, held steady during most of June. The Italian lira, which fluctu ated widely in response to domestic political developments, was advancing by the end of June. Dollar rates for the cur rencies of Japan and Canada, two of the United States major trading partners, in contrast, were little affected by developments elsewhere in the exchanges and at the end of 169 June were still at last March levels. Indeed, the Japanese authorities sold considerable amounts of dollars to prevent a depreciation of the yen. During the three months to the end of May, there were no Federal Reserve operations in the exchange markets. Debt outstanding under swap lines remained unchanged at $1,555 million (see table). There were no Treasury exchange market operations in this period, apart from small purchases of foreign exchange to meet scheduled expenditures. 170 MONTHLY REVIEW, JULY 1973 Implementing Open Market Policy with Monetary Aggregate Objectives By R ic h a r d G. D av is Editor’s Note: The author is Vice President, Research and Statistics function, Federal Reserve Bank of New York. This paper was prepared for Second District economics professors attending a central banking seminar at this Bank on April 23. The views expressed are the responsibility of the author alone and do not necessarily reflect the views of the Bank or of the Federal Reserve System. The purpose of this paper is to survey recent research on some technical problems of implementing open market policy at a time when the proper intermediate policy objective is widely believed to be the behavior of the money supply and related monetary aggregates. The mere existence of a widely held preference for monetary aggre gate targets in setting policy is a relatively recent develop ment. Even five years ago, the notion that the money supply should be the primary target of monetary policy was a decidedly minority position. This was true not only of academic and business economists, but of policy mak ers and the interested general public as well. Ten to fifteen years ago, discussions of monetary policy were only rather rarely couched in terms of the money supply. Practical discussions of policy were framed mainly in terms of in terest rates and credit market conditions. Things are now quite different. If one asks the average bank or business economist what they think monetary policy should be over the coming months, most will even tually get around to saying that the money stock should grow at such and such a rate. The ensuing elaboration will often owe more to the familiar equation of exchange than to the Keynesian “IS-LM” analysis.1 Similar com ments could be made of discussions of monetary policy in the Congress and in the business press. There is, to be sure, a danger of overdrawing this pic ture. Views on these matters are not and never have been uniform or monolithic. Yet, it is really striking the extent to which the monetarists have succeeded in shifting the focus of commonly received opinion on the role of money. One could of course ask whether this shift has been justified by an equally clear shift in the weight of the evidence. And one may entertain reservations on this score. However, the subject matter of the present paper is limited to the problems of implementing the monetarist program as regards using monetary policy to control the money supply and related monetary aggregates. MONEY VERSUS INTEREST RATES AS POLICY TARGETS: HOW SHOULD THE CHOICE BE MADE? It is important to stress at the outset that the problem of choosing between the money supply (M), or some other related monetary aggregate as a policy target, on the one hand, and interest rates (r) and credit market conditions, on the other, is really quite distinct from any issue of 1 In this algebraic summary of a simplified version of the Keynes ticular combination of income and interest rates for which both ian system, often used in the cbssroom, the so-called “LM” the supply and demand for money and the supply and demand for currently produced output are equal. Algebraically, the equilibrium equation represents alternative combinations of GNP and the level values of GNP and interest rates are determined by the simul of interest rates at which the supply and demand for money are taneous solution of the two equations, LM and IS, each of which equal. Similarly, the “IS” equation represents alternative combina contains two unknowns, GNP and the level of interest rates (treated tions of GNP and interest rates at which the supply and demand for simplicity as a single, “representative” interest rate). Graph for current output, including consumption goods, capital goods, and ically, the equilibrium values of the unknowns are shown by the Government purchases, are equal. The solution of these two equa intersection of the LM and IS lines (see Figure I). tions, the equilibrium value of the system as a whole, is that par FEDERAL RESERVE BANK OF NEW YORK “monetarism” versus “Keynesianism”, “fiscalism”, or what have you. One could perfectly well believe in the potency of fiscal policy and the importance of market interest rates, and, indeed, in the whole standard neo-Keynesian framework, and yet embrace the money stock rather than interest rates as the proper intermediate target for mone tary policy. The question is, what is the most efficient target for policy makers to aim at in a world of uncer tainties? Recently, several papers have pointed out that even in the context of the standard neo-Keynesian IS-LM analysis, the choice between money stock targets and interest rate targets depends upon the relative importance of the vari ous sources of instability in the economy.2 Thus, for ex ample, a money stock target may work quite badly in a world subject to large and unforeseen fluctuations in 171 Figure I— a The rate o f interest 2 See William Poole, “Optimal Choice of Monetary Policy In struments in a Simple Stochastic Macro Model”, Quarterly Journal of Economics (M ay 1970) and “Rules-of-Thumb for Guiding Monetary Policy” in Open M arket Policies and Operating Pro cedures (Board of Governors o f the Federal Reserve System, July 1971). See also John Kareken, “The Optimal Monetary Instru ment Variable”, Journal of Money, Credit, and Banking (August 1970). The argument made in these papers is as follows: Let the demand for money function be M = b0 + bi Y + b2r + v where Y and r are income and the interest rate and v is a random variable. With M given, the LM schedule becomes Y ^ bt , M. _ X b* b, b, • Let the IS schedule be Y = a0 + air + u where u is a random variable. The effect on income of using a money supply target (M *) is given by the reduced-form equation o f the system, Y — ~ aib" + a"b2 + ___ *1___ M * - ____ —___ v 4 - ____—___ u aibi + b2 aibi -f b2 aibi + b2 aibi + b2 Now assume that the “loss” resulting from deviations of actual income from its target equals the square o f these deviations. If M is used as the instrument, then the expected value of this loss is given by Var Y„, = ai (a,bi + b2) ' 0V ~ - f - (aibi -|- b2) ‘ 2aib2 (aibi -f- b2) 2 Inspection of the model makes it clear that the effect on income of an interest rate target (r*) is simply Y = a„ + ai r* + u, so that the variance of incoifie is Var Y r = <r„2. Comparison of the variances under M and r targets indicates that their relative size depends on the relative variances o f the distur bance terms in IS and LM as well as on the values of the various structural parameters. liquidity preference. If the money stock target is not ad justed for such shifts in the demand for money, the LM curve shifts (as in Figure I-a) and the shifts in the de mand for money are transmitted to interest rates and, ulti mately, to aggregate demand. Conversely, if the major source of unforeseeable disturbances arises in the nonfinancial markets (i.e., from shifts in the IS curve), an interest rate target will work badly. Maintaining interest rates at a predetermined target (r* in Figure I-b) in the face of such shifts in the IS curve will (as shown in Figure I-b) allow these shifts to be transmitted fully into shifts in aggregate demand. A money stock target, in con trast, would limit the effects of shifts in the IS schedule on aggregate demand by allowing interest rates to rise or fall in an offsetting way. In terms of this analysis, therefore, the choice of money versus interest rates depends upon the stochastic properties of the economy (that is, the sources and magnitudes of random disturbances) and not just upon its structural coefficients. Since Milton Friedman has iden tified belief in the stability of the demand for money as the 172 MONTHLY REVIEW, JULY 1973 hallmark of monetarism, monetarists, naturally enough, should prefer money. But, as noted earlier, Keynesians need not prefer interest rates. Indeed, given Keynes’s emphasis on the volatile “animal spirits” of businessmen as a source of economic instability, it is by no means clear he would have thought IS more stable than LM and, therefore, r preferable to M as a policy target. THE CHANGING ROLE OF MONEY SUPPLY TARGETS IN THE UNITED STATES For better or worse, money supply targets have come to have a growing importance in policy making, as already indicated. Before turning to the technical problems raised by the attempt to implement such targets, however, it may be useful to sketch briefly how the role of money supply objectives in policy formulation has evolved in recent years. First, it should be noted that the very concept of a “money supply policy” is open to some ambiguities. The Federal Reserve of course does not control the money stock directly. The actual behavior of the money supply is the joint result of Federal Reserve actions with respect to its own instrument variables—its open market port folio, discount rate, etc.—and the actions of the Treasury, of foreigners, of the banks, and of the nonbank public. Thus both Federal Reserve and non-Reserve influences interact to make the money supply whatever it is at any given time. Under these circumstances, there can really only be a money supply “policy” if the Federal Reserve consciously seeks to achieve a certain path for money by using its instruments to offset the effects of actions taken by others. Prior to at least 1960, while there was much “monetary policy”, there can really not be said to have been much "money supply policy'\ The Federal Reserve, by and large, marched to a different drummer. The actual be havior of the money supply “fell out”, for the most part an endogenous by-product of the System’s actions with respect to whatever targets it was following and the actions of the public and the banks. To be sure, it can be argued —and has been by some—that whatever the System’s conscious targets, the actual behavior of the money stock, or at least its broader and more significant movements, have been dominated all along by the behavior of the Federal Reserve’s policy instruments rather than by the behavior of the public or the banks. But even if this were true, it would still imply only that the Federal Reserve could control the money stock if it chose to, not that it actually did so in any particular historical period. As the 1960’s wore on, the behavior of the money supply seemingly came to have increasing importance in the thinking of the policy makers, roughly paralleling developments in the economics profession and among the public generally. In the first instance, this meant that some individual members of the Federal Open Market Com mittee (FOMC) began to give more weight to money supply behavior in voting on specific policy alternatives. But despite this increased weight, it is probably fair to say that at no time in the 1960’s did the recent and prospective behavior of the money stock become the dominant in fluence in the policy makers’ thinking with regard to open market policy targets. Moreover, the FOMC continued to eschew any agreed-upon, formal money supply target. Actual policy alternatives continued to be stated in terms of money market conditions, as measured, for example, by free reserves and the levels of certain key money market interest rates. Perhaps the earliest operational result, insofar as open market strategies were involved, of the increased con cern over the behavior of the money supply and related monetary aggregates was the use by the FOMC, begin ning in 1966, of the so-called “proviso clause”. This was a clause included in the directive addressed by the FOMC at each of its monthly meetings to the Account Manage ment at the Federal Reserve Bank of New York. It re quired the Account Management1to shift money market targets from the levels initially directed by the Committee in an appropriately offsetting direction whenever growth in bank credit proved to be deviating significantly from the rates projected at the time of the previous meeting. The significance of this “proviso clause” as a step toward direct targeting of money supply and other aggregates was limited, however. First, it stopped short of commit ting the FOMC to an explicit target. Secondly, in practice it involved only quite gingerly and modest adjustments of money market conditions targets in response to unex pectedly rapid or slow growth in the bank credit proxy. A more fundamental change took place in early 1970 when the Committee for the first time adopted explicit goals for the behavior of the narrow and broadly defined money supply (Mi and M2) and the bank credit proxy. At most of its meetings since early 1970 the Committee has continued to adopt explicit goals, covering varying time horizons, for the growth rates of one or more of these aggregates. At the same time, the Committee has experi mented with various operational tactics to achieve these goals. However, this most emphatically does not mean that actual money supply behavior over the period since early 1970 can be interpreted as conforming to the FOMC’s objectives in the short run. The bulk of the remainder of this paper is devoted to reasons why the money supply cannot, and perhaps even should notr FEDERAL RESERVE BANK OF NEW YORK be made to conform exactly to predetermined target values over short periods. Beyond this, however, goals for the growth rates of the monetary aggregates have seldom been the sole immediate objective of the FOMC even in the period since 1970. The Committee has generally re tained concern for avoiding unstable conditions in the money markets and has also retained an interest in the behavior of short-term interest rates and money and capi tal market conditions generally. THE CHOICE OF A TARGET AMONG THE MONETARY AGGREGATES Turning directly to some of the technical problems of implementing monetary policy where the intermediate objectives of policy are framed in terms of the monetary aggregates, several fairly basic questions come to mind immediately. The first might well be which monetary aggregate do you use: M1? M2, some measure of bank credit, total reserves, the monetary base (i.e., what is sometimes called “high-powered money”, or total reserves plus currency in the hands of the nonbank public)? With out defending the point in detail, I would argue that while measures of reserves and the monetary base may be use ful in developing strategies to achieve goals for one of the other aggregates, these measures are not themselves the best choices for framing monetary policy goals. The basic point is that we are interested in influencing the economy at large, not the banking sector per se. Setting targets in terms of reserves would allow random develop ments within the banking sector—which might be sum marized by movements in the reserve-deposit multiplier— to be transmitted to the overall economy, interfering with the achievement of the more basic goals for the gross national product (GNP) and similar variables.3 With regard to the remaining choices, between, say, Mi, M2, and some measure of total bank credit, I would argue that this is essentially a second-order issue. It is, for exam ple, very difficult to differentiate between these three 173 aggregates in terms of the closeness of their relationship to GNP in the postwar period.4 Real questions about which aggregate to use are, however, likely to develop during periods when Regulation Q ceilings are changed or when open market rates are rising above or falling below exist ing ceilings. Such “artificial” distortions in rate spreads induce marked decelerations or accelerations of time deposit growth and therefore distort the “normal” growth rate relationships among M1? M2, and bank credit.5 For example, a rise in market rates above Regulation Q ceilings will cause the public to shift out of time deposits and into open market securities. The resulting slowdown in M2 undoubtedly overstates the restrictiveness of monetary policy in such periods. The moral would seem to be that policy makers can, with reasonable safety, set goals either in terms of Mi, M2, or bank credit during normal times (provided allowance is made for differences in trend growth rates), but careful interpretation of differential growth rates is imperative during periods when Regulation Q (or some other special disturbances that do arise from time to time) is a factor.6 4 Michael Hamburger presents some results for changes in GNP regressed on current and lagged changes in various monetary aggre gates in “Indicators of Monetary Policy: The Arguments and the Evidence”, American Economic Review (May 1970). For the 1953-68 period, the R2s are .39 for Mi and bank credit and .28 for M2. However in the 1961-68 subperiod, M2 does much better than Mi (.43 versus .31) and only a little less well than bank credit (.45). An unpublished paper by Frederick C. Schadrack, “An Empirical Approach to the Definition o f Money” (June 1971) summarizes some previous published work of George Kaufman and Milton Friedman and Anna Schwartz and presents some new results using Almonized distributed lag techniques. For the period 1953-68 there is very little to choose between Mi and M2, both including and excluding large CDs (all adjusted R2s are around .55), though bank credit does a bit better at .61. Schadrack, however, expresses some preference for M2 (excluding large C D s) on the grounds that its coefficients appear more stable over time. 5 The view that Regulation Q provides the main reason for worrying about whether to use Mi or M2 was expressed as long ago as 1959 by Milton Friedman in A Program for Monetary Sta bility, page 91. 6 Milton Friedman has recently expressed a preference for M2 (excluding large C D s) over Mi on the grounds that the income velocity of M2 has shown essentially no trend since the early 1960’s while the income velocity of Mi has continued to show an uptrend 3 In “Improving Monetary Control” (Brookings Papers on Eco of somewhat uncertain dimensions (see “How Much Monetary nomic A ctivity, 2-1972), William Poole extends his analysis cited Growth” in the Morgan Guaranty Survey, February 1973). Fried earlier to examine the situation where the central bank’s options man’s argument is couched in terms of the substantially larger are not M and r, but the monetary base (B ) and r. The additional range of the level of the Mi income velocity relative to the range variance introduced by the banking sector via the supply equation of the level of the M2 income velocity in the 1962-72 period. The for money may make B targets inferior to r targets even where M relevant issue, however, is the variance of the rates of growth of targets would be superior to r targets. The argument against B these two velocity measures— at least as far as setting intermediatetargets is simply that the authorities ought to permit themselves run or countercyclical monetary growth targets is concerned. To maximum flexibility in adjusting as needed to variations in the rela put it differently, what one cares about for these purposes is the tionship between B and M. As Poole points out, arguments for a closeness of fit of equations relating growth in nominal income to steady rate of growth in M simply cannot be extended to a steady growth in money, not the size o f the constant term in such equa rate of growth in B. tions. 174 MONTHLY REVIEW, JULY 1973 HOW LARGE ARE THE ECONOMIC COSTS OF FAILING TO HIT MONETARY TARGETS IN THE SHORT RUN? A second basic question with regard to implementing monetary aggregate targets for monetary policy is how long or short should the time horizon be for achieving these targets: i.e., should you try to set and meet targets for monetary growth over a month, a quarter, six months, a year? The answer to this question seems to depend essentially on two factors: (1) the decreasing feasibility of controlling the aggregates over successively shorter periods and (2) the increasing costs in terms of economic stability of failing to hit them over successively longer periods. This second aspect is examined first. Just how much difference does it make to aggregate demand objectives if the Mi target is missed by 2 percent age points over one month, over three months, etc.? The key to this question lies in the lag structure relating money growth to the behavior of the economy at large. If, for example, the influence of M on GNP were essentially instantaneous, deviations from monetary targets lasting even for very short periods could have a marked impact. On the other hand, if the influence of money operates with a long distributed lag, the impact of deviations from Mx targets may be greatly attenuated. Suppose, for exam ple, the Federal Reserve wants to hit a 6 percent money growth rate target. Suppose, instead, it actually hits 10 percent for two quarters in a row (as illustrated in Figure II) and then drops down to 2 percent for the next two quarters. If the influence of money operates with a dis tributed lag—i.e., the effective impact of money at any point reflects a weighted average of past money supply Figure II Percentage changes (Annual rate] growth rates—the overshoot effect of the two 10 percent quarters will never register its full impact on GNP. Long before this can happen, the overshoot effects will begin to be offset by the undershoot effects of the two 2 percent quarters. If lags are sufficiently long, the course of events in the economy may turn out to differ little from what would have happened had the 6 percent money target been successfully reached in each and every quarter instead of just on average over the whole four-quarter period. One way to arrive at quantitative estimates of the costs of permitting M to deviate from target values over periods of varying lengths is to use econometric model simula tions. Such simulations can be used to compare the results of steady monetary growth at an x percent rate with uneven monetary growth that averages out to the same rate over the longer run. One such simulation has been performed on a version of the well-known Federal Reserve Bank of St. Louis econometric equation.7 In this equation, nominal GNP is determined mainly by the behavior of the money supply in the current and three prior quarters. The control simulation assumes a steady 6 percent rate of growth in Mi in each quarter. Other simulations assume Mi growth rates of 10 percent for one, two, and three quarters, respectively, followed by growth rates of 2 percent for an offsetting number of quarters, with Mx returning to a 6 percent growth rate thereafter. The results of these simulations (see Table I) suggest that a one-quarter deviation of M2 growth from target amount ing to 4 percentage points or less would have essentially negligible effects on GNP. Deviations from target lasting for two quarters would have only moderate effects. In this case, the resulting deviations of GNP from the path implied by steady 6 percent M ± growth path would reach a maximum of only about 1 percent of the level of GNP. Deviations of Mx from its target growth rate amounting to 4 percentage points and lasting for as long as three quarters do have more serious effects, however. Of course these results are only as valid as the lag structures embodied in the underlying model. Probably most large-scale structural models incorporate somewhat longer lags in the money-GNP relationship than do St. Louis-type “reduced-form” equations. As a result, simu lations with these models would no doubt suggest that deviations from target growth rates could occur over 7 See James Pierce and Thomas Thomson, “Some Issues in Con trolling the Stock of Money”, in Controlling M onetary Aggregates II: The Implementation (Federal Reserve Bank of Boston, 1972), pages 115-36. FEDERAL RESERVE BANK OF NEW YORK 175 Table I SIMULATIONS OF GROSS NATIONAL PRODUCT Control simulation (steady 6 % Mi growth) Period % AM Simulation II Level of GNP (billions of dollars) % AM Simulation III GNP minus control simulation (billions of dollars) % AM Simulation IV GNP minus control simulation (billions of dollars) % AM GNP minus control simulation (billions of dollars) 2.4 1972: I ................. 6 1,092.9 10 2.4 10 2.4 10 I I ............... 6 1,108.5 2 3.3 10 8.3 10 8.3 H I .............. . 6 1,125.8 6 2.6 2 12.0 10 17.1 I V .............. 6 1,145.6 6 1.1 2 10.3 2 22.6 1973s I ................. 6 1,166.3 6 - 0 .3 6 5.4 2 21.1 n ..................... 6 1,187.3 6 - 0 .4 6 0.9 2 13.5 i n ................... 6 1,208.4 6 - 0 .4 6 - 0.8 6 5.3 I V .............. 6 1,229.7 6 -0 .4 6 - 0.8 6 0.5 Note: Simulations were performed with the equation described in “A Monetarist Model for Economic Stabilization** by L. Andersen and K. Carlson (Federal Reserve Bank of St. Louis Review, April 1970). The simulations are reported in James Pierce and Thomas Thomson, “Some Issues in Controlling the Stock of Money”. somewhat longer periods without serious consequences for aggregate demand objectives. For what they are worth, however, the available simulation results suggest that the FOMC need not be too concerned about even fairly siz able deviations from Mt target growth rates lasting up to around six months—providing there is some subsequent undershooting. Putting it somewhat differently, the policy makers should perhaps not be too disturbed by sizable intrayearly fluctuations in Mx growth, provided the aver age growth rate for the year as a whole comes out about on target.8 ACHIEVING MONETARY OBJECTIVES: THE NEED FOR SHORT-RUN OPERATING TARGETS Having tried to establish some notion of the costs of failing to hit money supply targets over varying lengths of time, the next question is what operational procedures are available to achieve these M targets and how well can such procedures be expected to work? One begins from the obvious fact, noted earlier, that the Federal Reserve has direct control only over certain instrument variables, most notably the size of its open market portfolio of Gov ernment securities. To hit targets for any monetary vari able—be it the money supply, the monetary base, or even member bank nonborrowed reserves—forecasts of non 8 These conclusions are also supported by the results presented in controlled factors influencing these variables must first be “Income Stabilization and Short-run Variability in Money” by made. Next, the Federal Reserve’s instrument variables E. Gerald Corrigan, Monthly Review (Federal Reserve Bank of New York, April 1973), pages 87-98. In this paper Corrigan first must be adjusted in such a way as to take account of the uses a “money-only” reduced-form equation to compare what GNP movements of these noncontrolled factors. The harder the behavior would have been in 1970-71 if Mi had grown at a steady value equal to its average growth over the entire period relative movements of these noncontrolled factors are to predict, to (a) the actual behavior of GNP over the period and (b) what or the more complex are their interaction with movements the money-only equation would have projected for GNP given the actual behavior of Mi over the period. Corrigan concludes that in the Federal Reserve’s own instrument variables, the events would not have been very different with steady growth. In more difficult will it be to hit any given target. all but two quarters, GNP growth in the steady Mi case would have differed by only 0.7 percent (annual rate) or less relative to the This is a complex matter, but the main points can be GNP growth indicated by the equation given the actual pattern of summarized as follows: to implement a money supply ob Mi growth rates. The 1963-65 period exhibits similar results. Simu lations of the Board-MIT econometric model for 1970-71 also jective, defined, say, in terms of the desired M t growth indicate little difference between the results of smooth Mi growth rate over a month or period of months, an operationally and the quite uneven quarterly pattern of Mi growth rates that actually occurred— the largest difference for any quarter was an meaningful strategy requires that week-to-week open 0.6 percent annual rate of growth in GNP. Corrigan also runs market operations be laid out in terms of target values simulations similar to the Pierce-Thomson simulations cited in Table I and the text, though using a money-only reduced-form for other variables, variables easier to hit than the money equation. The results are essentially the same as the Pierce-Thomson supply itself. The targeted levels of these other variables results. 176 MONTHLY REVIEW, JULY 1973 must then be adjusted so that their achievement maximizes the likelihood of achieving the money supply target itself. In other words, one has to project, for example, the week-by-week levels of nonborrowed reserves that appear to be consistent with the desired money supply growth rate. Once this is done, the day-to-day decisions as to whether to buy or sell in the open market can be made in terms of the nonborrowed reserve objectives.9 As a practical matter, what variables are open to the Federal Reserve as feasible day-to-day and week-to-week operating targets? In practice, the number of available options is really rather small. First, one would have to rule out all total reserve and related measures, such as the total monetary base and the recently developed concept of RPD (total reserves behind private nonbank deposits). In practice, the Federal Reserve does not have the power to fix the levels of any of these measures within a given week. The problem is that changes in borrowings at the discount window, a magnitude over which the System ex erts only the most general influence, will offset the ef fects on total reserves of System actions taken to change nonborrowed reserves.10 However, by the same token, the 9 The problem of laying out short-term tactics for achieving the goals set in a money supply strategy is discussed in Richard G. Davis, “Short-Run Targets for Open Market Operations”, in Open Market Policies and Operating Procedures— Staff Studies (Board of Governors of the Federal Reserve System, July 1971). 10 This problem is examined in more detail in Davis, op. cit., pages 42-44. If the Federal Reserve supplies nonborrowed reserves in excess of required reserves, which are fixed for a given reserve period under lagged reserve accounting, the result is likely to be mostly a paydown of outstanding borrowings and little if any build up of excess reserves— at least up to the point where borrowings are reduced to frictional minima. On the other hand, reductions in the amount of nonborrowed reserves supplied are likely to lead to an offsetting increase in borrowings rather than a reduction in excess reserves. The point is simply that in a period where excess reserves are very low and are probably very insensitive to money market rates in the short run, total reserves are fixed by required reserves (determined on the basis of deposit levels two weeks earlier) plus the frictional minimum level of excess reserves. (A regression of weekly levels of excess reserves on the weekly average Federal funds rate for 1970 and 1971 had a— nonsignificant— R- of only .005; the coefficient indicated that a full 1 percentage point increase in the Federal funds rate would reduce excess reserves by only $16 mil lion. ) Fluctuations in nonborrowed reserves lead to offsetting move ments in the Federal funds rate and in borrowed reserves but not, to any significant extent, to fluctuations in total reserves. Precisely the same argument applies to the total reserve base and total RPD. Note that the situation under lagged reserve accounting, with the resulting fixity of required reserves in any given week, may not be much different from the situation where reserve requirements are determined by deposit levels in the same week if bank asset supplies are quite insensitive to money market interest rates over periods as short as one week. various measures of nonborrowed reserves, including the nonborrowed monetary base and nonborrowed RPD, are feasible weekly operating targets. This is not to say the weekly targets for these nonborrowed reserve measures are easy to hit with accuracy. Quite the contrary. The non-Federal Reserve controlled factors affecting reserves, most notably float, are very difficult to predict accurately on a weekly basis.11 A different sort of weekly operational target that could be used to achieve the more basic money supply objec tives is represented by money market interest rates, per haps most notably the Federal funds rate (the rate on interbank overnight lending). On the one hand, this would be an operationally feasible target since the Trading Desk could feed funds into and out of the market as the actual market rate fell below or rose above the target rate. Thus on a weekly average basis, say, it is possible to operate so that the average Federal funds rate will, most of the time, approximate a target rate. At the same time, the required weekly interest rate objective can be related to the more fundamental money supply target through forecasts of the demand for money at various interest rates. This problem is discussed further below. In summary, to implement a money supply objective, the Federal Reserve must lay out a week-to-week program for operationally feasible short-run targets. It must set values for these targets that are projected to be consistent with the underlying money supply objective. If the pro jections prove wrong, the weekly target values will have to be adjusted. In practice, the Federal Reserve can use either a nonborrowed reserves measure or an interest rate measure, perhaps most especially the Federal funds rate, as its weekly operational target. RELATIONSHIP OF RESERVE AND INTEREST RATE OPERATING TARGETS TO MONEY SUPPLY OBJECTIVES: A SIMPLE MODEL OF THE SUPPLY AND DEMAND FOR MONEY Given the feasibility of either nonborrowed reserves or some measure of short-term interest rates as a week-toweek operating target, what is entailed in using these targets for achieving somewhat longer run objectives for 11 In 1971, the average error in projecting market factors affecting nonborrowed reserves (float, currency in circulation, and the effects of Treasury and international transactions) as of the beginning of statement weeks was $275 million. See “Open Market Operations and the Monetary and Credit Aggregates— 1971”, this M onthly Review (April 1972), pages 79-94. FEDERAL RESERVE BANK OF NEW YORK the money supply? To examine this, it is useful to set up an illustrative skeleton model of the supply and demand for money—or, to simplify matters somewhat, for de posits (D) alone. The demand equation for deposits in this model is the standard liquidity preference schedule. It includes a short-term interest rate (r) and some mea sure of transactions demand (Y). This latter variable can be treated as exogenous, given the short period purposes of the model. The supply of deposits is assumed to depend upon the level of nonborrowed reserves (Ru) and on the short-term interest rate (r). This latter dependency re flects the dependency of the banks’ demand for borrowed reserves on short-term interest rates, the (smaller) de pendency of their demand for excess reserves on rates, and the dependency on interest rates of the time-demand de posit mix. The latter of course has attendant effects on the banks’ demand for reserves as a result of the differ ence in reserve requirement ratios for the two types of deposits. Thus the model (in linear form) consists of (1) D = bxY + b2r + u (Demand) (2) D = CiRu + c2r + e (Supply) where u and e are random terms. Now the choice of an interest rate operating target to achieve the broader money supply objectives is tantamount to treating the interest rate as an exogenous variable. In such a situation, nonborrowed reserves become endoge nous, that is, such reserves are allowed to come out at whatever level proves necessary to achieve the interest rate target. The resulting “reduced-form” equation of the model is the same as the demand equation, i.e., (3) D = b1Y + b2r* + u, where r* is the weekly interest rate target used by the Federal Reserve. On the other hand, if the FOMC decides to work with a nonborrowed reserves operating target, the short-term interest rate becomes endogenous. The relevant reduced form for the reserve-target case is derived from the solu tion of the demand and supply equations as follows: (4) D = c2- b 2 Y- - ^ L . R * + ue. c2- b 2 c2- b 2 c2- b 2 A great deal of work has been done within the Federal 177 Reserve System on estimating structural models of the type represented by equations (1) and (2 )—though of course any realistic model requires far more than two structural equations—and “reduced-form” equations of the type represented by equations (3) and (4). While the estimation of these equations has raised the usual quota of econometric conundrums, some useful insights have been obtained from this work. Three areas in particular should be mentioned: (1) the different sorts of risk one is exposed to in using a nonborrowed reserves operating target as against an interest rate target, (2) the approxi mate limits of our ability to forecast and achieve money supply objectives with, respectively, these two types of operating targets, and (3) the existence of lags and their implication for policy making. SOURCES OF ERROR IN HITTING MONEY SUPPLY OBJECTIVES As equation (3) indicates, the use of a short-term interest rate target entails making a short-term forecast of the demand for deposits. The interest rate target can then be set at the level that is expected to give the desired deposit behavior. (Of course, a formal econometric equa tion need not be used, but a judgmental forecast would require making much the same calculations implicitly.) There are two possible sources of error in picking the inter est rate target needed to achieve the money supply goals: (1) random errors (or shifts in the demand equation) and (2) errors in forecasting Y—which is taken as exogenous for the short-term purposes at hand. Note that both types of error tend to be accommodated when using an interest rate target. That is, if the demand for money is greater than expected, for example, holding to the interest rate target (r* in Figure Ill-a) will mean automatically supplying enough reserves to accommodate the demand. This tendency for interest rate targets—and “money market conditions” targets generally—automatically to accommodate changes in the demand for money has been the chief complaint of monetarists about this type of target over the years. However, the complaint should not be leveled against interest rate operating targets per se. Rather, the complaint should have been, as applied to the procedures in use prior to 1970, (1) that the FOMC did not formulate explicit monetary growth rate objectives and (2) that it would not have been willing to move money market targets often enough, quickly enough, and decisively enough to achieve monetary growth objectives even if it had formulated them. Given the willingness to move interest rate targets as needed to achieve money sup- 178 MONTHLY REVIEW, JULY 1973 ply objectives, such targets are a perfectly feasible way of operating open market policy to achieve money supply objectives. As equation (4) indicates, the use of a nonborrowed reserves operating target can also be expected to lead to errors in controlling the money stock. The new element here is errors stemming from the supply side—errors which might be summarized in terms of unforeseen move ments in the ratio of nonborrowed reserves to private deposits, i.e., unforeseen movements in the “deposit mul tiplier”. Such movements can result, in turn, from unfore seen shifts in the banks’ demand for excess and borrowed reserves and from unforeseen movements in the average required reserve ratio. This ratio is of course affected by shifts among the various categories of deposits and by movements of deposits between banks with different re serve requirement ratios. Another important potential source of error on the supply side can originate from un foreseen movements into and out of Treasury deposits at the commercial banks. These deposits absorb required reserves but are not themselves included in the money supply as usually calculated. Relative to an interest rate target, nonborrowed re serves have some advantages and some disadvantages as an operating target for achieving money supply goals. Nonborrowed reserves are superior to interest rates in the face of a change in the demand for money. Such a change tends to get fully accommodated under an interest rate target, as already noted.12 Under a nonborrowed re serves target, however, an increase in the demand for money will be accommodated only to the extent that the resulting upward pressure on money market rates en genders some elasticity of supply—to the extent, for ex ample, that the banks themselves are induced by rising interest rates to accommodate the increase in demand by increasing borrowings from the Federal Reserve Banks or by drawing down excess reserves. Such offsets may not be too large, however, and will, in any case, not be com plete. Thus the money supply is likely to stay closer to target in the face of a demand shift if the Federal Reserve uses a nonborrowed reserves operating target than with an interest rate target. On the other hand, a nonborrowed reserves target does not perform as well as an interest rate target in the face of an unforeseen shift in the average required reserve ratio —whatever its cause—or a shift in the banks’ demand for excess and borrowed reserves. Such shifts will lead to a change in the actual money supply as long as the supply of nonborrowed reserves is held on target (see Figure Ill-b). With an interest rate operating target, in contrast, the volume of nonborrowed reserves would automatically be adjusted to offset the impact on money of these changes in supply conditions. As a result, their distorting effects on the money supply would be neutralized. ESTIMATING ERRORS IN ACHIEVING MONEY SUPPLY OBJECTIVES IN THE SHORT RUN In principle, then, either nonborrowed reserves or inter est rates can be used as operating handles to achieve money supply goals; each has its own advantages and 12 It should also be noted, however, that short-term, random, and reversible shifts in the demand for money should be accom modated since such accommodation prevents them from having any impact on real activity. FEDERAL RESERVE BANK OF NEW YORK disadvantages.13 Which can be expected to work better in practice, and how well each will work, are empirical questions. A fair amount of statistical work has been done in the Federal Reserve System on the probable size of errors in using these operating targets. Table II is fairly representative of the general thrust of the results—and it also gives some idea of the probable order of magnitude of errors in hitting money supply targets in the short run. Three sets of forecasts are presented in Table II.14 The first two are based on “reduced-form” equations of the types suggested by equations (3) and (4) presented earlier. The first set uses changes in the Federal funds rate as the open market operating target: thus it is essentially a com plex variant of equation (3) cited earlier. The second set of projections uses the nonborrowed monetary base as the operating target—i.e., it is a variant of equation (4). The third set is a method developed at the Federal Reserve 13 As a purely formal matter, the question of whether nonbor rowed reserves or interest rates is the better instrument for con trolling money can be treated with precisely the same analysis used by Poole to examine whether the money supply or the interest rate is the better handle to control GNP (see footnote 2 ). The question turns on the relative instability of demand (analogous to the IS curve in Poole’s analysis) or supply (analogous to L M ). The vari ance o f deposits, using an interest rate target, depends on the vari ance in Y from forecast values and the variance of the error term u in equation (3) and on their covariances as well as on the income elasticity of demand. The variance of deposits, using a non borrowed reserves target, depends on the variance of Y, the vari ances of the error terms in both supply and demand equations, their covariances, and the various income and interest rate elasticities in the supply and demand equations (see Pierce and Thomson, “Some Issues in Controlling the Money Stock” ). Just as Poole’s analysis has to be modified to allow for the possibility that M cannot be precisely controlled, however, the present analysis should really be modified for the possibility that nonborrowed reserves cannot be precisely controlled. In this case, nonborrowed reserves should be replaced in the supply equation with the sum of changes o f Federal Reserve credit, the forecast value of operating factors affecting reserves, and a new random error term reflecting errors in fore casting operating factors. Some specialists object strongly to the proposition that nonborrowed reserves could be hit with any high degree of accuracy. They argue that under a pure nonborrowed re serves target, where the Federal funds rate could be expected to show much wider week-to-week movements than at present, the behavior of the Federal funds rate would no longer serve to assist the Open Market Account Management in warning when operating factors affecting reserves are going seriously off track. They argue that the margin of error in hitting funds rate on average, week to week, would be much smaller. In that case, evidence purporting to compare nonborrowed reserves and the funds rate as competing targets for controlling money is seriously biased in assuming the two operating targets are themselves equally achievable. 14 These results represent an updating of material presented in a paper to be published in a forthcoming issue of the Journal of M oney, Credit and Banking, by Fred J. Levin, “Examination o f the Money Stock Control Approach o f Burger, Kalish, and Babb”. See also A. E. Burger, L. Kalish III, and C. T. Babb, “Money Stock Control and Its Implications for Monetary Policy” (Federal Reserve Bank of St. Louis R eview, October 1971). 179 Table II ERRORS IN FORECASTING Mi GROWTH RATES, 1970-72 Seasonally adjusted annual rates; in percent Measure Using Using Federal nonborrowed FRB St. Louis funds rate* monetary baset method! Mean absolute error of ex post monthly forecasts ........................... 2.91 3.47 6.33 Root mean square error of ex post monthly forecasts............................. 3.60 4.22 8.61 Mean absolute error of ex ante monthly forecasts............................. 3.36 3.61 § Root mean square error of ex ante monthly forecasts............................. 3.93 4.53 § Mean absolute error of ex ante quarterly forecasts ........................... 2.10 2.13 § Root mean square error of ex ante quarterly forecasts........................... 2.34 3.08 § Mean absolute error of ex ante six-month forecasts ......................... 1.19 2.10 § Root mean square error of six-month ex ante forecasts............................... 1.39 2.52 § 7 7 * Based on AMi — 2 bi A Federal funds rate t-i + 2 ci A business sales t-i + i=o i= o d A Treasury deposts + constant. Estimated on 1965-69 data. 7 7 f Based on AMi = 2 bi A nonborrowed monetary base t-i + 2 ci A business i=o i= o sales t-i + d A Treasury deposits + constant. Estimated on 1965-69 data. %See A. Burger, L. Kalish, and C. Babb, “Money Stock Control and Its Impli cations for Monetary Policy’*, Federal Reserve Bank of St. Louis Review (October 1971). § Not available. Bank of St. Louis. It simply estimates the money-reserve base multiplier from a regression using a three-month moving average of past values of the multiplier, an adjust ment for changes in legal reserve requirement ratios, seasonal dummies, and a measure of autocorrelation. The two reduced-form equations were estimated from 1965-69 data, while the St. Louis method calls for updating the re gression equation in each month. The forecasts were made for monthly changes in money supply (expressed as an annual rate of growth) over the 1970-72 period. The forecasts labeled “ex-ante” in the table are ex-ante fore casts in the sense that all inputs were entered as of the estimates or projections that would have been available at the time. Some interesting results emerge from Table II. First, the forecasts using the nonborrowed monetary base and the Federal funds rate, respectively, do roughly equally well. Some other results (not presented in the table) which adjust nonborrowed reserves for required reserves against Treasury and interbank deposits (in other words, nonborrowed RPD) also seem to suggest a roughly com parable performance. This particular evidence, at least, 180 MONTHLY REVIEW, JULY 1973 does not provide a decisive case for or against any one of the potentially available operating targets. A second point is that the two “reduced-form” equa tions forecast markedly better for this period than the St. Louis approach, which is essentially a purely autoregres sive estimate of the money multiplier. Finally, and most important, all three methods of pro jection do rather poorly in forecasting monetary growth rates for individual months. The same conclusion has to be drawn about attempts to forecast short-term growth rates judgmentally. Despite considerable investment of time and talent, all presently available techniques for making short-term projections of the monetary growth rate that would be associated with any particular setting of an operating target are subject to large errors on average and very large errors in many particular instances. It is possible to be more optimistic when somewhat longer time horizons are considered, however. Ex ante forecasts for one quarter ahead have an average absolute error of 2.13 percent for the nonborrowed base equation and 2.10 percent for the Federal funds equation. Forecasts for six months ahead show corresponding average absolute errors of 2.10 percent and 1.19 percent, respectively. This means that for six-month periods, we seem to be able to get at least “ball park” estimates of the conse quences for monetary growth rates of particular settings of open market operating targets. Even for these periods, however, the errors are clearly not negligible. THE LAGS BETWEEN FEDERAL RESERVE ACTIONS AND MONEY SUPPLY RESPONSE Finally, with regard to the lags: All the econometric evidence available to us suggests that there is a lag between a change in operating targets and the full impact of that change on the money supply. Indeed both the Board staff’s monthly money market model—a very com plex version of structural equations (1) and (2 )—and the reduced-form equations developed at this Bank— essentially complex versions of (3) and (4 )—suggest that the effects may take on the order of six to eight months to work themselves out fully.15 This means, for example, that a maintained step-up in the level of nonbor rowed reserves will not have its full effect on the level of the money supply for several months. Similarly, a once and for all increase in a Federal funds rate target will not have its full effect on the level of the money supply for several months. Now it is of course true that the estimation of lag structures is a very uncertain business. The evidence that these lags are as long as six to eight months cannot be con sidered at all firm. One possibility is that the use of monthly time units instead of shorter units may bias estimates of the lag upward; this sort of bias does seem to have turned up in some other areas of econometric work. However, the Board staff has also estimated a weekly model, and its lags, while not as long as those in the monthly model, are still substantial.16 The point is that even if the true lags were, say, only one half the six- to eight-month range indicated by most of the econo metric work, they would still have significant implications for policy. The existence of these lags creates a potential control problem for the Federal Reserve in trying to hold mone tary growth rates to any targeted rate. As indicated, these lags delay the ultimate impact on the money supply of a change in the operating target, be it nonborrowed re serves or short-term interest rates. By the same token, how ever, they also imply that an adjustment in the operating target large enough to produce a desired correction in the monetary growth rate immediately will ultimately over shoot this desired correction and will then have to be reversed (see Figure IV). For example, if Mx is currently growing at 2 percent (January through March in Figure IV) and the Open Market Committee objective is 6 percent, the desired correction could be achieved by lowering a Federal funds rate target (labeled Rff in the figure) or raising a nonborrowed reserves target (labeled Ru' in the figure). However, a reduction large enough to bring M! back to 6 percent within a month (April), would eventually (by June) push the monetary growth rate up to, say, 8 percent. At that point, a near-term correction back to 6 percent (in July) might require a drastic increase in the Federal funds rate target, one that 16 Helen T. Farr, Steven M. Roberts, and Thomas D. Thomson, “A Weekly Money Market Model— A Progress Report” (June 15 See “A Monthly Econometric Model of the Financial Sector” 1972, unpublished). The lags in the major structural equations of this model run from five to thirty-four weeks. Simulations per by Thomas Thomson and James Pierce, unpublished; “Estimating formed by the authors indicate that an increase in nonborrowed Monthly Changes in Deposits with Reduced-Form Equations” by reserves achieves its maximum effect on Mi after about three Richard G. Davis (April 1972, unpublished), and “A Reducedmonths, while a decrease in the Federal funds rate produces its Form Mi Equation” by Frederick C. Schadrack and Susan Skinner maximum impact after about six months. (June 1972, unpublished). FEDERAL RESERVE BANK OF NEW YORK might ultimately (in August and September) push growth once more below the 6 percent target, and so on. Indeed, if the lag structure is sufficiently unfavorable, it would not be too difficult to imagine situations where explosive oscillations in the operational target would be required to hold the Mx target, month by month, to a steady target growth rate.17 There appears to be a fairly clear moral to be derived from these implications of the existence of lags: policy makers should avoid taking too short a view in deciding where to set week-to-week operating targets. Even in a world where the money supply implications of a given target setting were known with certainty, operational targets should be set so as to achieve money supply goals on average over a period of time. Attempts to rejigger operat ing targets so as to hit the long-run money supply objectives in each and every month—assuming this to be possible at all—are likely to involve excessive volatility in the operat ing target. This volatility will mean excessive instability in money market rates. This will result directly, if the Federal 181 funds rate is used as the operating target, and indirectly if some measure of nonborrowed reserves is used. Similarly, when trying to speed up or slow down the monetary growth rate, the policy makers should keep in mind the fact that a movement in operating targets sharp enough to achieve this change rapidly will, because of lags, 17 In principal, at least, this proposition can be tested. First, eventually overshoot. This overshooting will, in turn, reduced-form equations of the sort presented in Table II, having require an eventual further adjustment of the instrumental been estimated statistically, can be solved for the current value of the instrument variable. Thus, for example, equation (1 ) in that target in the opposite direction. For this reason, it may table, where the Federal funds rate is the instrument variable, can well be desirable to take a somewhat gradualist approach be rewritten so that the current change in the Federal funds rate is the dependent variable and is a function of the current change to slowing down or speeding up monetary growth rates, in deposits, lagged changes in the Federal funds rate, and the cur aiming to accomplish the change over a period of months rent and lagged changes in the exogenous variables. If we assume the current change in deposits equal to some given target value in rather than immediately. Of course in any particular situa each and every month, we then have a seventh-order difference tion, the right decision depends on for how long and by equation in the Federal funds rate plus some exogenous variables. This equation tells us the change in the instrument variable— the how much monetary growth has been deviating from what Federal funds rate in this case— that will be required to hold M on the policy makers consider desirable and how serious the target given the current values o f the exogenous variables and the past history of these variables and the funds rate. Some preliminary economic consequences of these deviations seem likely analysis of this equation suggests that under a fairly wide range of to be if not corrected promptly. assumptions about the lag coefficients on the Federal funds rate in the original reduced-form equation, the time path o f monthly In thinking about these matters, it quickly becomes changes in the funds rate needed to maintain changes in M at the apparent that the use of a money supply strategy for targeted rate in each and every month would be oscillatory and explosive. Moreover, using some plausible values for the behavior monetary policy confronts the policy makers with a very of the exogenous variables and initial conditions as of early 1972, interesting and complex problem in control theory—one simulations of the difference equation for constant monthly changes in the money supply did generate explosive oscillations in the Fed that is only just now beginning to be appreciated and ex eral funds rate. Indeed the simulation after just a few months in plored. The policy maker is confronted with two sets of volved a negative funds rate to hold changes in deposits on target, clearly an impossibility. While one would hardly want to jump to distributed lags. One set of lags, the one just discussed, the conclusion that these simulations accurately reflect the way the world is actually constructed, their results are not really so implau relates operationally feasible open market target variables, sible. There may be literally no way of getting M to grow by more such as nonborrowed reserves and the Federal funds rate, than x percent next month. At some point, injections o f nonbor rowed reserves may drive the funds rate and borrowed reserves to to the money supply. The other set, mentioned earlier, zero, with further injections merely having the effect of piling up relates the money supply to the variables that ultimately excess reserves. Of course as these reserves begin to be utilized, with a distributed lag, money supply growth could subsequently matter. Shortening the attempted time horizon of monetary become explosive, forcing the authorities to jump the funds rate up control increases the technical problems of monetary man by amounts that would rock the structure of the money market— and so on. agement and the likelihood of an unacceptable degree of 182 MONTHLY REVIEW, JULY 1973 have been made if hitting money supply objectives had been the sole aim. In trying to improve its ability to achieve money supply goals, the FOMC has experimented with alternative approaches to operating tactics. It has also made a substantial investment of research resources in the prob lems of monetary control. The results to date suggest that attempts to forecast and control the money supply over short periods, what ever operating targets are used, will normally be subject to quite large errors. A rough judgment might be that reasonably close forecasts, and control, can be achieved over periods down to six months if the Committee is prepared to move its operating targets sufficiently vigor ously to achieve the desired results. Fortunately, the tentative evidence also suggests that very short-term control over the money supply is not necessary for satisfactory economic performance. Evi dently, fairly large deviations from target may not do any significant harm, provided they do not last longer than a quarter or two and provided monetary growth rates average out about on target over longer periods of CONCLUSION perhaps a year. None of these estimates can be regarded The main points made in this paper can be summarized as firmly established, however. In controlling money, shorter term operating targets as follows. The Federal Reserve has moved with the shift in the general climate of ideas over the past few years, that are more readily achievable than the monetary targets putting increasing emphasis on the money supply and other themselves are needed as a guide to day-by-day and weekmonetary aggregates as intermediate objectives of monetary by-week decisions. Various measures of nonborrowed policy. The Federal Open Market Committee now sets reserves and short-term interest rates are available for explicit goals for the growth of the money supply and bank this purpose. Each has advantages and disadvantages. credit and issues operating instructions to the Account The available evidence does not establish any clear overall Management in New York that are drawn up largely with superiority for any one of them. a view to achieving these objectives. The qualification Finally, it is clear that a great deal has been learned “largely” is necessary since the behavior of money and about the problems and possibilities of implementing capital market conditions, and international financial de money supply targets in the past few years. Virtually all velopments, have also continued as a source of explicit of the research drawn upon in this survey is less than concern to the Committee. At times, this concern has three or four years old. Many of the topics discussed dictated operating decisions different from those that might would have seemed quite novel only a relatively short time ago. The progress in this area has been rapid. Indeed, one may wonder if diminishing returns may not already have begun to set in in some respects. Perhaps the direc tion in which research efforts will now move is to deal with the implications of the lags and uncertainties—in 18 It might be noted that these problems would not disappear short the “control theory” issues mentioned earlier. What though they might be simplified, if one were to adopt a steady growth of the money supply at some fixed rate a la Milton Fried should be the time horizon for monetary control? How man. In the first place the economy would not “start out” on its should targets be adjusted in response to the past “misses” long-run trend path with full employment and a history o f steady monetary growth, sustainable real growth, and an “acceptable” rate that will inevitably arise? Work in this area will have of inflation. Consequently, one might want to approach the longto be sufficiently grounded in reliable evidence and suf run monetary growth rate target only gradually. Secondly, all the technical problems of short-run control over the money supply ficiently “robust” to be useful to properly skeptical policy would remain. Consequently, the actual growth rate could expect makers. Nevertheless, we can be hopeful that further to go off track much as it does now. Therefore, the problems of an optimum control period, how to compensate for past errors, and progress in this area will be forthcoming over the period how sharply to adjust operating targets to get back on track would ahead. still exist. money market instability. Lengthening the period of control increases the probable deviations of aggregate demand and related variables from desired behavior. No doubt there is an optimum control period here some where. It was mentioned earlier that some calculations seem to suggest that if monetary growth rate targets are hit on average over a year, deviations of GNP from the path it would follow with absolutely steady monetary growth might well remain within acceptable limits. However, if monetary growth rates have drifted off target over the first half of the year, for example, the authorities then have only six months in which to get the yearly average back on track. Thus it seems fairly clear that the authorities will have to be prepared to move their instrument variables with sufficient vigor to control average monetary growth rates over periods shorter than one year. Clearly there are some messy problems here in this whole area. Quite possibly they have not yet even been clearly formulated— let alone solved.18