The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
CO o GO O 03 QQ ^Review March 1980 Vol. 62, No. 3 CD > CD CO CD Qc: a 5 TD CD 2 The F O M C in 1979: Introducing Reserve Targeting 26 The Dynamics and Estimation of Short-Run Money Demand The FOMC in 1979: Introducing Reserve Targeting RICHARD W. LANG M a j o r changes in the implementation of mone tary policy occurred in 1979. In accordance with the Full-Employment and Balanced Growth Act of 1978, the Federal Open Market Committee (FO M C ) was required to announce long-run ranges for monetary growth in 1979 in a new format. In addition, the FOM C had to adjust its one-year range for M l growth to take into account the introduction of two financial innovations in late 1978 — the automatic transfer service (A TS) between savings and checking ac counts at banks nationwide and negotiable orders of withdrawal (N O W ) accounts in New York State. The most important change in the implementation of monetary policy, however, occurred in the FOM C’s short-run operating procedures. On October 6, 1979, the Federal Reserve announced a series of actions to “assure better control over the expansion of money and bank credit, help curb speculative excesses in financial, foreign exchange, and commodity markets, and thereby serve to dampen inflationary forces.”1 The most significant of these actions was the FOM C’s ap proval of a change in the day-to-day procedures used to conduct monetary policy. This action involves placing greater emphasis in day-to-day operations on the supply of bank re serves and less emphasis on confining short-term fluctuations in the federal funds rate.2 This announcement introduced a different strategy for the conduct of open market operations — a strategy of targeting on bank reserves rather than the federal funds rate. Note: Unless specified otherwise, citations throughout this paper are from either the “Record of Policy Actions of the Federal Open Market Committee” or “Statements to Congress,” Federal Reserve Bulletin (February 1979 through March 1980). ^‘Announcements: Monetary Policy Actions,” Federal Reserve Bulletin (October 1979), p. 830. 2Ibid. 2 This article discusses these modifications to the implementation of monetary policy and summarizes the decisions of the FOMC in 1979. The Committee’s domestic policy directives are summarized in table 1. Excerpts from the monthly “Record of Policy Actions of the FOM C,” which provide a more detailed meet ing-by-meeting summary of FOMC discussions, ap pear in a supplement to this article. LONG-RUN RANGES OF MONETARY GROWTH Since 1975 the FOMC has publicly announced oneyear growth ranges for the monetary aggregates (M l, M2, and M 3). At that time, Congress requested in House Concurrent Resolution 133 that the Board of Governors consult with congressional committees on a quarterly basis about the Federal Reserve System’s objectives and plans for the ranges of growth of mone tary and credit aggregates over the next 12 months. Such consultations became a requirement with the Federal Reserve Reform Act of 1977. The format of these one-year ranges was altered in 1979, however, under the requirements of the Full Employment and Balanced Growth Act of 1978 (also called the Humphrey-Hawkins A ct). At the same time, the FOMC had to formulate monetary growth ranges that took into account the introduction of nationwide ATS ac counts and New York NOW accounts. The Humphrey-Hawkins Act Before 1979 the FOM C reviewed and adopted oneyear growth ranges for several monetary and credit aggregates each quarter and presented them to Con gress. The period to which these growth ranges ap plied began with the previous quarter and ended in the same quarter of the following year. For instance, MARCH F E D E R A L R E S E R V E B A N K O F ST. LO U IS 1980 Organization of the Committee in 1979 The Federal Open Market Committee (FO M C ) con sists of the seven members of the Federal Reserve Board of Governors and five of the twelve Federal Re serve Bank presidents. The Chairman of the Board of Governors is also, by tradition, chairman of the Com mittee. The president of the New York Federal Re serve Bank is a permanent member of the Committee and, also by tradition, its vice chairman. All Federal Reserve Bank presidents attend the meetings and present their views, but only those presidents who are members of the Committee may cast votes. Four mem berships rotate among the Bank presidents and are held for one-year terms beginning March 1 of each year. Members of the Board of Governors at the begin ning of 1979 included Chairman G. William Miller, Philip E. Coldwell, J. Charles Partee, Nancy H. Teeters, and Henry C. Wallich. There were two vacancies on the Board as a result of the death of Vice Chairman Stephen S. Gardner and the resigna tion of Governor Phillip C. Jackson in November 1978. In addition to Paul A. Volcker, president of the Fed eral Reserve Bank of New York, the following presi dents served on the Committee during January and February 1979: Ernest T. Baughman (D allas), David P. Eastbum (Philadelphia), Mark H. Willes (M in neapolis), and Willis J. Winn (Cleveland). The Com mittee was reorganized in March and the four rotating positions were filled by: John J. Balles (San Francisco), Robert P. Black (Richmond), Monroe Kimbrel (Atlan ta ), and Robert P. Mayo (Chicago). In July, Emmet J. Rice was appointed to the Board, filling one of the two vacancies, and Frederick H. Schultz was ap pointed to the Board as vice chairman, filling the other vacant position. During August, Chairman Miller resigned and was replaced by Mr. Volcker. The Committee met nine times during 1979 to discuss, among other things, economic trends and to decide upon the future course of open market opera tions.1 As in previous years, however, telephone or telegram consultations were held occasionally between scheduled meetings.2 During each regularly scheduled meeting, a directive was issued to the Federal Reserve Bank of New York. Each directive contained a short review of economic developments, the general economic goals sought by the Committee, and in structions to the Manager of the System Open Market Account at the New York Bank for the conduct of open market operations. These instructions were stated in terms of money market conditions and short-term rates of growth of M l and M2 which were considered to be consistent with desired longer-run growth rates iNo formal meetings were held in January, June, and December 1979. 2Consultations were held on January 12, March 2, April 27, June 15, July 27, August 30, and October 22. of the monetary aggregates. Special factors, such as conditions in domestic financial markets and foreign exchange markets, were also taken into account. The Manager makes all decisions regarding the timing, types, and amount of daily buying and selling of securities in fulfilling the Committee’s directive. Each morning the Manager and his staff plan the open market operations for that day. This plan is developed on the basis of the Committee’s directive and the latest developments affecting money and credit market conditions, monetary aggregate growth, and bank reserve conditions. The Account Manager, in a conference call, then informs staff members of the Board of Governors and one voting president about present market conditions and open market operations that he proposes to execute that day. Other members of the FOMC are informed of the daily plan by wire. The directives issued by the FOMC and a summary of the reasons for FOM C actions are presented to the public in the “Record of Policy Actions of the Federal Open Market Committee.” The “Record” is released a few days after the following FOM C meeting. Soon after its release, the “Record” appears in the F ed eral R eserve Bulletin and, in addition, “Records” for the entire year are published in the Annual R eport o f th e B oard o f Governors. The “Record” for each meeting during 1979 generally included: 1 ) A staff summary of recent economic develop ments — such as changes in prices, employ ment, industrial production, and components of the national income accounts — and projections of general price, output, and employment devel opments for the year ahead; 2 ) A summary of recent international financial de velopments and the U.S. foreign trade balance; 3 ) A summaiy of recent credit market conditions and recent interest rate movements; 4 ) A summary of open market operations, growth of monetary aggregates and bank reserves, and money market conditions since the previous meeting; 5 ) A summary of the Committee’s discussion of current and prospective economic and financial conditions and of current policy considerations, including money market conditions and the move ment of monetary aggregates; 6 ) Conclusions of the FOM C; 7 ) A policy directive issued by the Committee to the Federal Reserve Bank of New York; 8 ) A list of the members’ voting positions and any dissenting comments; 9 ) A description of any actions and consultations that may have occurred between the regularly scheduled meetings. 3 F E D E R A L R E S E R V E B A N K O F ST. LO U IS MARCH 1980 Table 1 FOMC Operating Ranges — 19791 Short-Run Tolerance Ranges Date of meeting December 19,1978 December 29’ January 1 2 ,19792 February 6* Federal funds rate range 9%-10 1/2% 10 (or slightly higher) (no change) 9% -10V2 Initial federal funds rate target Ranges specified Period to which M1 & M2 apply M1 M2 Dec.-Jan. 2- 6% 5-9% Feb.-Mar. 3-7 5-9 10 Mar.-April 4-8 10 April-May 10 % (or slightly higher) Actual growth rates3 M1 -0.8% M2 2.3% 10 (or slightly higher) (no change) 10 0.4 3.3 3y2-7i/2 10.1 9.2 4-8 4-81/2 9.5 9.8 May-June 0-5 4-8V2 7.9 9.8 July-Aug. 21/ 2 - 61/2 6 V2 -IO 1/2 8.6 11.9 Aug.-Sept. 4-8 7-11 9.1 11.6 Sept.-Oct. Oct.-Dec. 3-8 41/2 61/ 2-1 01/2 71/2 7.0 3.1 10.4 6.8 3.3 N.A. 6.0 N.A. (or slightly higher) March 22 March 20b 10 10 (or slightly higher) 9% -10V2 (or slightly higher) (or slightly higher) April 17° April 272 May 22d June 152 July 11 July 272 August 14' August 302 September 18f October 6 October 222 November 20 January 8-9, 1980 9%-10i/2 (no change) 9 % - 101/2 101/4 9% - 101/2 10V2-103/4 10 % - 1 1 1/4 10 % - 111/2 111/4-11% 111/ 2- 151/2 (no change) 111/2-151/2 111/2-15V2 (or slightly higher) (no change) 101/4 101/4 101/4 101/2-10% 11 111/4 111/2 NONE (no change) NONE NONE Nov.-Dee. Jan.-Mar. (no change) (no change) 81/2 5 4-5 7 Longer-Run Ranges Date of meeting February 6s July 11h October 6 Target period IV/78-IV/79 M1 11/ 2 -41/ 2 % 11/ 2 -41/2 3-6 M2 M3 5-8% 5-8 5-8 6-9% 6-9 6-9 Bank credit 71/ 2- 101/ 2 % 71/ 2 - 101/2 71/ 2 -IO 1/2 ‘Short-run ranges were adopted at each of the FOMC’s regularly scheduled meetings. The ranges for the monetary aggregates were specified in terms of two-month or three-month simple annual rates of change from the month prior to each meeting at which ranges were established to the month following the meeting. The ranges for the federal funds rate were specified to cover the period from each meeting at which ranges were adopted to the following regularly scheduled meeting, telephone or telegram consultations were held between scheduled meetings to modify intermeeting ranges for the federal funds rate. 3Data used are revised data as of January 9, 1980. They include benchmark adjustments based on December 1978 and March 1979 Call Reports. the FOM C adopted ranges of monetary growth at its April 1978 meeting that applied to the period from the first quarter of 1978 to the first quarter of 1979.3 3For an example, see Richard W. Lang, “The FOMC in 1978: Clarifying the Role of the Aggregates,” this Review (March 1979), pp. 2-24, especially table 1 and charts 1 and 2. At the July 1978 meeting, it adopted one-year ranges that then applied to the period from the second quarter of 1978 to the second quarter of 1979. Thus, the base period — the quarter from which growth of the monetary aggregates was to be measured — changed each quarter when the FOMC announced MARCH F E D E R A L R E S E R V E B A N K O F ST. L O U IS 1980 Table 1 (continued) Supplementary Footnotes — Dissents to FOMC Actions “Mr. Coldwell dissented from this action because he preferred to direct open market operations early in the coming period toward a slight firming in money market conditions. He felt that the greatest current danger was an intensification of inflationary pressures and that the longer-range prospects for inflation were unacceptable. bMessrs. Volcker, Coldwell, Kimbrel, and Wallich dissented from this action because they favored a somewhat more restric tive policy posture in view of inflationary forces reinforced by pressure on capacity in some industries and in view of the near-term potential for excessive inventory demands. They believed that, despite uncertainty about prospects for economic activity later this year, some additional firming in money market conditions at this time was appropriate to help contain inflationary pressures and maintain renewed confidence in the dollar in foreign exchange markets. 'Messrs. Volcker, Coldwell, and Wallich dissented from this action because they continued to favor a somewhat more restric tive policy posture in view of strong inflationary forces reinforced by pressure on capacity in some industries. They believed that, despite uncertainty about prospects for economic activity later this year, some additional firming in money market con ditions at this time would help in limiting inflationary pressures by curbing inflationary expectations quickly. dMessrs. Balles and Partee dissented from this action in view of indications that a cyclical peak might be near at hand. Thus, they favored a less restrictive policy posture, especially in view of the delayed impact on the economy. In the present un certain environment, they believed that some prompt easing in money market conditions, along with a greater emphasis on the behavior of the monetary aggregates in guiding the conduct of operations, would reduce the risk of a continuing shortfall in monetary growth and would tend to provide needed support to the economy later in the year. Mr. Wallich dissented from this action because, in view of the strong inflationary pressures in the economy, he continued to favor a more restrictive policy posture. Believing that inflationary expectations had increased in recent months while inter est rates had changed little, he thought that additional firming in money market conditions would have a favorable effect on such expectations and would have little effect on the course of real output. eMr. Black dissented from this action because, in view of the rapid monetary growth in recent months, he preferred to specify lower ranges for growth of Ml and M2 over the August-September period in order to increase the probability of holding growth within the Committee’s longer-run ranges. Wl3le he agreed that open market operations should be directed toward attaining a slight increase in the federal funds rate initially in the coming intermeeting period, he believed that the direc tive adopted by the Committee allowed for too rapid monetary growth before a further increase in the funds rate would be triggered. Mr. Rice dissented from this action because he believed that an additional firming in money market conditions at this time to restrict growth of money and credit, in the face of the evidence of weakening in economic activity, would risk deepening the recession. In his view, the effort to balance the goal of reducing the rate of inflation with the objective of minimizing the impact of the recession called for a policy directed toward the maintenance of prevailing money market conditions unless growth of the monetary aggregates over the August-September period appeared to be substantially faster or slower than the rates currently expected. 'Messrs. Balles, Black, and Coldwell agreed with the majority that open market operations should be directed toward attaining a slight increase in the federal funds rate initially in the coming intermeeting period, but they dissented because they be lieved that, given the excessive monetary growth in recent months relative to the Committee’s longer-run ranges, the direc tive adopted by the Committee would allow for too rapid monetary growth before an additional increase in the objective for the funds rate would be triggered. To enhance the prospects for achieving the Committee’s objective of restraining monetary growth, they preferred, moreover, to provide leeway for a rise in the funds rate to an upper limit of 12 percent. Mr. Rice dissented from this action because he believed that an additional firming in money market conditions would inten sify the developing weakness in economic activity and was unlikely to affect the rate of inflation favorably within six to nine months. In his judgment, monetary growth most likely would slow in the months immediately ahead even if current money market conditions were maintained, and growth of the monetary aggregates over the year ending in the fourth quarter of 1979 probably would fall within the Committee’s longer-run ranges. ^Messrs. Wallich and Willes dissented from this action because, with the Committee’s objective of slowing the rate of infla tion in mind, they preferred to specify lower ranges for growth of the monetary aggregates. Mr. Willes believed that the range adopted for Ml, after allowance for the effects of ATS and a possible further downward shift in the public’s demand for money, represented an increase from the ranges that had been adopted during 1978. Mr. Wallich thought that, after allow ance for the expansion in repurchase agreements and Eurodollars in addition to the other forces affecting growth of Ml, the range adopted represented too much of an increase from the ranges set earlier. “Mr. Wallich dissented from this action because, with the Committee’s objective of slowing the rate of inflation in mind, he believed that the range adopted for M l, after allowance for the effects of ATS and NOW accounts, was too high. In his opinion, growth of the money stock, after allowance for the expansion in repurchase agreements and Eurodollars as well as for the effects of ATS and NOW accounts, had been considerably more rapid than indicated by the behavior of Ml. these one-year ranges. This method of establishing one-year monetary growth ranges resulted in “base drift” and allega tions that the FOMC implicitly employed a “forgiveness principle” in adopting these ranges. To illustrate this, suppose that at the beginning of the second quarter the FOMC adopted a one-year range for M l growth of 4 to 6 percent for the period from the first quarter of one year to the first quarter of the next, but that in the second quarter M l actually increased at an 8 percent annual rate. The FOMC would establish a new M l growth range at the be ginning of the third quarter which would be meas ured from the second quarter of the year to the second quarter of the following year. By using the second quarter’s average level of M l as the base from which the new one-year growth rates of M l 5 F E D E R A L R E S E R V E B A N K O F ST. L O U IS MARCH 1980 would be calculated, the level of M l in the base pe riod would “drift” from quarter to quarter. Further more, since the FOM C frequently kept the same numerical one-year ranges (e.g., keeping a 4 to 6 percent range from second quarter to second quarter) despite deviations of M l growth from that range in the previous quarter, deviations of monetary growth from their original ranges were essentially ignored or “forgiven.” centive to earn interest on checkable bank balances, the Federal Beserve expected that many individuals would use ATS savings accounts as a substitute for holding balances in their checking accounts. Thus, during a transition period in which demand deposits were being transferred into ATS accounts, the growth of ATS balances (which were not included in M l) was expected to reduce the growth of M l (demand deposits plus currency).8 The Humphrey-Hawkins Act requires the FOMC to establish calendar-year growth ranges for monetary and credit aggregates in February of each year. The FOM C has chosen to establish these ranges from the fourth quarter of the previous year to the fourth quar ter of the current year. The FOM C must review these ranges in mid-year, although it may reconsider the one-year ranges at any time.4 The Committee, how ever, may not change the period to which the oneyear ranges apply. Thus, the base period (the fourth quarter) remains the same. A similar but smaller effect was expected to result from the introduction of NOW accounts in New York State. NOWs can be used by individuals in the same manner as checks to make transactions, but balances in NOW accounts earn interest as do savings accounts and were not included in M l.6 Upon congressional approval, NOW accounts were extended to New York State in November 1978; they had been introduced much earlier in the New England states. The growth of these accounts in New York also was expected to reduce M l growth during a transition period. By requiring the FOMC to establish calendar-year monetary growth ranges measured from an unchang ing base, the Humphrey-Hawkins Act eliminates “base drift” and the “forgiveness principle” within each calendar year. Nevertheless, these problems still can occur from one year to the next because the Humphrey-Hawkins Act requires the FOMC to establish new one-year ranges each calendar year. Thus, the FOM C would use the most recent fourth quarter average level of the monetary aggregate as the base. Chairman G. William Miller outlined the effect of ATS accounts on the growth of M l and the broader monetary aggregates on July 28, 1978 when he an nounced to Congress the one-year ranges established at the July 1978 FOMC meeting. He noted that, dur ing the transition period in which bank customers adjust to ATS, M l growth would be lowered while M2 and M3 growth would be little affected.7 ATS and New York NOW Accounts Financial innovations in 1978 presented the FOMC in 1979 with the difficult problem of assessing their impact on the growth of various measures of money. Starting November 1, 1978, commercial banks and mutual savings banks were permitted to offer indi vidual customers an automatic transfer service ( ATS) between savings and checking accounts. This seivice provides for the transfer of funds from a savings ac count to cover checks written against an individual’s checking account. Consequently, the new ATS savings accounts enable individuals to earn interest on funds that previously might have been held in their check ing (demand deposit) accounts. Because of the in4At the time of the mid-year review of the one-year ranges, the FOMC also establishes tentative ranges for the monetary aggregates for the next year — measured from the fourth quarter of the current year to the fourth quarter of the follow ing year. 6 Although discussed at the July meeting, the scheduled introduction of ATS accounts in Novem ber 1978 first affected the FOM C’s decision about the one-year monetary growth ranges at the October 17, 1978 meeting. A majority of the Committee voted to retain the existing ranges for M2 and M3 for the pe riod from third quarter 1978 to third quarter 1979 (III/7 8 to 111/79). The Committee also indicated that it expected M l growth to be within a range of 2 to 6 percent during that period, “depending in part on the speed and extent of transfers from demand to savings deposits resulting from the introduction of ATS.”8 This expected range of M l growth was both lower and wider than the one adopted in July. The “For a more detailed analysis of the effect of ATS on the money supply process, see John A. Tatom and Richard W. Lang, “Automatic Transfers and the Money Supply Process,” this Review (February 1979), pp. 2-10. 6NOW balances at commercial banks were included in old M2, while NOW balances at thrift institutions were included in old M3. 7Chairman G. William Miller, “Statements” (August 1978), p. 646. 8“Record” (December 1978), p. 953. F E D E R A L R E S E R V E B A N K O F ST. L O U IS FOMC, however, placed less emphasis on M l than on M2 and M3.9 When the one-year ranges were established at the February 6, 1979 meeting under the requirements of the Humphrey-Hawkins Act, the FOMC felt more confident about the magnitude of the effects of both ATS and New York NOW accounts on M l growth and again gave equal emphasis to M l growth. A staff analysis indicated that the growth of ATS and New York NOW accounts would lower M l growth over 1979 by about 3 percentage points.10 The FOMC used this estimate in formulating its range for M l growth over the period from IV /78 to IV /79. The FOMC’s discussion at the February meeting stressed that monetary growth ranges be chosen that were consistent with a reduction in the rate of inflation.11 The Committee decided at this meeting to lower the monetary growth ranges. The ranges also were widened to 3 percentage points, compared with their previous width of 2.5 percentage points.12 This widening reflected, in part, “the special factors [ATS and New York NOW accounts] expected to influence monetary growth and the uncertainties with respect to the magnitude of their impact.”13 The M2 and M3 ranges were lowered to 5 to 8 percent and 6 to 9 per cent for the period IV /78 to IV /79, respectively, from previous ranges of 6.5 to 9 and 7.5 to 10 percent for the period 111/78 to 111/79 (table 1 ). The FOMC adopted an M l range of 1.5 to 4.5 percent from IV /78 to IV /79 and noted that this “range allowed for the possibility of a significant deceleration of growth from the pace of recent years” even after allowance for the estimated 3 percentage-point slowing in M l growth due to the effects of ATS and New York NOW accounts.14 Two Committee members, however, felt that the 1.5 to 4.5 percent M l range allowed for too rapid M l growth. In fact, they felt that this range repre sented a higher, not a lower, range than the previously adopted M l ranges. Messrs. Wallich and Willes dissented from this action because, with the Committee’s objective of 9Chairman G. William Miller, “Statements” (November 1978), p. 846; and Lang, “The FOMC in 1978,” p. 14. ^ “Record” (April 1979), p. 330. u Ibid. 12In the case of M l, recall that the 2 to 6 percent range from 111/78 to 111/79 was not typical of ranges in earlier periods. The M l range in eifect for most of 1978 was 4 to 6% percent. 13“Record” (April 1979), p. 331. 14Ibid. MARCH 1980 slowing the rate of inflation in mind, they pre ferred to specify lower ranges for growth of the monetary aggregates. Mr. Willes believed that the range adopted for M l, after allowance for the effects of ATS and a possible further downward shift in the public’s demand for money, represented an increase from the ranges that had been adopted during 1978. Mr. Wallich thought that, after allowance for the expansion in repurchase agreements and Eurodollars in addition to the other forces affecting growth of M l, the range adopted represented too much of an increase from the ranges set earlier.15 To understand why the 1979 M l range could be viewed as higher than previous ranges of growth, one must consider what it would have been without the effect of ATS and New York NOW accounts. The FOMC used an estimate that growth of these accounts would lower M l growth by 3 percentage points. This means that 3 percentage points were subtracted from the lower and upper ends of an initial M l range to obtain a lower limit of 1.5 percent and an upper limit of 4.5 percent. The initial range implicit in this calcu lation is 4.5 to 7.5 percent. One can view this 4.5 to 7.5 percent range as the M l range that would have prevailed if ATS and New York NOW accounts had no effect at all on M l growth. The upper end of this range represented almost the same rate of M l growth as actually occurred in the previous two years — M l increased 7.9 percent from IV /76 to IV /77 and 7.2 percent from IV /77 to IV /78. Furthermore, this 4.5 to 7.5 percent range repre sented an increase in the upper limit of the M l range compared with those adopted in 1977 and 1978, when the upper limits were no higher than 6.5 percent. If the FOMC had adjusted the M l range that had been in effect during most of 1978 — 4 to 6.5 percent16 — for the estimated 3-percentage-point effect of ATS and New York NOW account growth, the resulting M l range would have been 1 to 3.5 percent from IV /78 to IV /79 instead of the 1.5 to 4.5 percent M l range actually adopted. Mid-Year Review The Humphrey-Hawkins Act required the FOMC to transmit to Congress in July a review of its 1979 monetary growth ranges as well as a preliminary in dication of its ranges for 1980. At the July 11, 1979 meeting, a staff analysis indicated that “shifts in funds from demand deposits to savings accounts with automatic transfer services and to NOW ac15Ibid., pp. 331-32. 16See footnote 12 and the earlier discussion of the M l range adopted at the October 17, 1978 meeting. 7 F E D E R A L R E S E R V E B A N K O F ST. LO U IS counts had retarded the annual rate of growth of M l by the assumed amount of about 3 percentage points in the first quarter of 1979 but by only about 1.5 per centage points in the second quarter. . . .”17 Uncer tainty about the further growth of ATS accounts was compounded by an April federal court ruling that barred these accounts as of January 1, 1980, unless Congress enacted legislation authorizing them.18 At the July meeting, most Committee members favored ranges that reflected a continuation of the policy stance adopted in February.19 The Committee decided to retain the same ranges for 1979 that it had adopted in February and tentatively indicated that these ranges also would be appropriate for 1980 (table 1 ). Although the FOMC continued to use the estimate that the growth of ATS and New York NOW accounts would reduce M l growth by about 3 percentage points in 1979, it “also agreed that actual growth in M l might vary in relation to its range to the extent of any deviation from that [3 percentagepoint] estimate.”20 In this regard the Committee noted that monetary growth in 1980 “might be within the same ranges” as in 1979, but that this was especially tentative in light of the uncertain legislative and judicial position of ATS and other interest-earning transactions accounts.21 The M l range for 1979 was indeed modified later in the year. At the special meeting held on October 6, 1979, the “Record of Policy Actions” noted that the recent growth of ATS and New York NOW ac counts implied that their effect on M l growth was smaller than the original 3 percentage-point estimate. It now appeared that expansion of such accounts would reduce measured growth of Ml over the year by about 1.5 percentage points. After allowance for the deviation from the earlier estimate, the equiv alent range for M l was 3 to 6 percent.22 The revised 3 to 6 percent range for M l growth from IV /78 to IV /79 (table 1) again translates into an initial range of M l growth of 4.5 to 7.5 percent (obtained by adding 1.5 percentage points to the 17“Record” (September 1979), p. 752. 18Congress temporarily extended authorization of ATS ac counts through March 1980 in order to allow more time to consider permanent legislation. 19“Record” (September 1979), p. 753. 2°Ibid., p. 754. 21The retention of the 1.5 to 4.5 percent range for M l growth again was cause for dissent by Governor Wallich. He still felt that this range was too high given the Committee’s objective of slowing inflation. ( Mr. Willes was not a voting member at this meeting.) Ibid. 22“Record” (December 1979), p. 973. 8 MARCH 1980 FOM C’s revised M l range). In effect, the FOMC in October 1979 continued to use an estimate of the re duction of M l growth due to the growth of ATS and New York NOW accounts to mark down the 1979 M l range from an initial range of 4.5 to 7.5 percent. This latter range was higher and wider than the M l ranges adopted in 1978, and can be interpreted as the range that would have prevailed if the growth of ATS and New York NOW accounts had no effect at all on M l Actual Money Growth and the One-Year Ranges Growth rates of the monetary aggregates were be low their one-year ranges in the first quarter of the year, but were above tbese ranges in the second and third quarters. Thus, after falling below their ranges in the first quarter, the levels of the monetary aggre gates were within their ranges by the end of the sec ond quarter and were above or near the tops of their ranges by the end of the third quarter (charts 1 and 2 ). Rapid growth of M l in the third quarter brought the level of M l above the original 1.5 to 4.5 percent range. After readjusting the M l range for the lower estimate of the effect of ATS and New York NOW accounts, however, the level of M l was near the top of the revised 3 to 6 percent range at the end of the third quarter (chart 1 ). Nevertheless, it was clear to the FOM C in early October that a continuation of monetary growth at third-quarter rates would result in growth rates for the year that would exceed the 1979 ranges.23 The Committee’s desire to hold growth of the monetary aggregates within their 1979 ranges was the principal reason for the October 6 changes in shortrun operating procedures.24 This shift in procedures, discussed below in detail, apparently succeeded in slowing growth of all the monetary aggregates in the fourth quarter. As a result, both M l and M3 in 1979 were kept within the FOM C’s adopted ranges of growth, while M2 was slightly above the upper limit of its range. From IV /78 to IV /79, M l increased about 5.5 percent, M2 about 8.3 percent, and M3 about 8 percent. It must be emphasized, however, that the 5.5 per cent M l growth in 1979 is not comparable to M l 23lbid., p. 974. 2*Ibid. F E D E R A L R E S E R V E B A N K O F ST. LO U IS MARCH 1980 Chart 1 M l R ange for Period I V / 7 8 to I V / 7 9 R a tio Sc a le B illio n s of D o lla rs R a tio Scale M l Range as Revised October 6, 1979 B illio n s of D o lla rs R a tio Scale B illio n s of D o lla rs 1978 N o te : 1979 D a ta used a re s e a s o n a lly a d ju s te d a nd in c o rp o ra te th e b e n c h m a rk adjustm ents released by th e B o a rd o f G o ve rn o rs on Ja n u a ry 10, 1980. growth in previous years. For example, to compare M l growth rates for 1979 and 1978, one must adjust the 1979 growth rate by the estimated effect of ATS and New York NOW accounts. Assuming that M l growth in 1979 was reduced by 1.5 percentage points — the figure used by the FOMC in October to adjust the M l range — one must add this amount to the 5.5 percent rate of growth in 1979. This results in a 7 percent growth rate, about the same as the 7.2 percent rate in 1978. SHORT-RUN OPERATING PROCEDURES Prior to October 6, 1979, FOMC short-run operat ing procedures were carried out as in the previous several years.25 At each meeting, the FOMC set an intermeeting range for the federal funds rate along with two-month tolerance ranges for M l and M2 25Lang, “The FOMC in 1978,” pp. 2-10. For an historical per spective on the FOMC’s short-run operating procedures, see Henry C. Wallich and Peter M. Keir, “The Role of Operating Guides in U.S. Monetary Policy: A Historical Review,” Fed eral Reserve Bulletin (September 1979), pp. 679-91. F E D E R A L R E S E R V E B A N K O F ST. L O U IS MARCH 1980 Chart 2 R anges for M 2 , M 3 , a n d Bank Credit for Period I V / 7 8 to I V / 7 9 R a t io S c a lo R a t i o S c a lo B il l i o a s of D o lla r s l i l l i o i s o f D o lla r s 130 lit lo R a t i o S c a lo S e a l* l i l l i e i s o f D o lla r s B il l i o a s o f D o lla r s l i l l i o a s o f D o lla r s B il l i o a s o f D o lla r s 1971 197* 1980 N ote: D a ta us a d a r e s e a s o n a lly a d ju s te d a n d in c o rp o ra te th e b e n c h m a rk ad justm e nts relea sed b y the B o ard o f G o verno rs J a n u a ry 10, 1980. growth (table 1 and charts 3 and 4 ) . Within the fed eral funds rate range, the Committee specified an initial level of the funds rate that was thought to be consistent with the short-run ranges set for M l and M2. The two-month ranges for M l and M2, in turn, were intended to provide an indicator to determine when changes in the federal funds rate should be made (subject to its own range). These short-run ranges were specified over moving two-month periods. meeting specified ranges for M l and M2 growth over the February-March period. At the March 20 meet ing it then set new ranges for the March-April pe riod (table 1 ). If the two-month growth rates of M l and M2 appeared to deviate from their respective ranges — either from the midpoints of their ranges or from their upper or lower limits — the level of the federal funds rate could be changed within its range, or the Committee could reconsider the range.26 For example, the FOMC at its February 6, 1979 26For more details, see Lang, “The FOMC in 1978,” pp. 2-10. http://fraser.stlouisfed.org/ 10 Federal Reserve Bank of St. Louis F E D E R A L R E S E R V E B A N K O F ST. L O U IS Prior to October 6, the domestic policy directives followed the same general format as in 1978 with regard to short-run operating procedures. Less than three weeks after adopting such a directive at its September 18 meeting,27 this format was aban doned when the FOMC instituted its new short-run operating procedures. The new procedures were adopted at a special FOMC meeting called by Chair man Volcker, “to consider actions that might be taken, in conjunction with actions being contemplated by the Board of Governors, to improve control over the expansion of money and bank credit in the light of developing speculative excesses in financial and commodity markets and additional evidence of strong inflationary forces in the economy.”28 After the September 18, 1979 meeting, incoming data indicated that the economy was stronger than had been expected, while both consumer and producer price indexes continued to record double-digit increases. In the foreign exchange mar kets, the dollar declined substantially despite large purchases of dollars by the United States and by for eign central banks. This decline was attributed to the continuing rapid U.S. inflation rate.29 Furthermore, speculation in gold markets had increased between August and October 1979 and had apparently spread to other metal and commodity markets, as evidenced by a more than 50 percent increase in September in an industrial commodity price index. This was attributed to “a general rise in inflationary expectations that tended to feed on themselves.”30 Against this back ground ( see Supplement, meeting of October 6, 1979), the Federal Reserve announced several policy actions — a 1 percentage-point increase in the discount rate, a new marginal reserve requirement on certain managed liabilities of member banks, and a change in the FOM C’s short-run operating procedures.31 MARCH 1980 C k irl 3 F O M C S h o rt-R u n R a n g e s for M o n e t a r y A g g r e g a t e s 1*79 H I Tolerance Ranges DEC./JAN. JAN/FEB. FEB./MAR MAR./APR. APR /MAY MAY/JUNE JUNE/JUIY JUIY/AUG. AUG./SBT. SEPT./OCT. Note-. After the October 6, 1979 FOMC meeting, the FOMC no longer specified tolerance ranges tor Mi and M2 ir the tame manner. See table I. U. During the January-February and June-July periods, ranges for Ml and M2 wer not specified by the FOMC. \2 Actual growth rate data are revised os of January 10, 1980. \3_The shaded areas represent two-month ranges adopted by the Committee at each regularly scheduled meeting. gates within their 1979 ranges was the catalyst for the change in the FOM C’s short-run operating procedures. The principal reason advanced for shifting to an operating procedure aimed at controlling the supply of bank reserves more directly was that it would provide greater assurance that the Committee’s ob jectives for monetary growth could be achieved.32 30Ibid. The Committee recognized that the relationship be tween bank reserve and monetary growth was not precise; there was no guarantee that the shift in emphasis would be successful in reducing monetary growth over the last few months of 1979. The Com mittee also recognized that this change in short-run procedures could be accompanied initially by sub stantial increases in interest rates and that movements in interest rates would be more uncertain.33 After some debate between members who favored the change in short-run operating procedures and those who preferred continuing to direct open mar ket operations toward controlling the level of the federal funds rate, the Committee unanimously ap proved a modification of the domestic policy direc tive. Greater emphasis was placed on controlling the 31For a more detailed description of the new marginal reserve requirements on managed liabilities, see “Announcements: Monetary Policy Actions,” Federal Reserve Bulletin (October 1979), pp. 830-32. :i2“Record” (December 1979), p. 974. 33Ibid., p. 976. The FOMC was especially concerned with the ex pansion of both the monetary aggregates and bank credit, which had been increasing at rapid rates after the first quarter of 1979 (charts 1 and 2 ). The issue of how to slow monetary growth to keep the aggre27“Record” (November 1979), pp. 912-13. 28“Record” (December 1979), p. 972. 29Govemor Henry C. Wallich, “Statements” (November 1979), p. 899. 11 MARCH F E D E R A L R E S E R V E B A N K O F ST. L O U IS volume of bank reserves and less on controlling fluctuations in the federal funds rate.34 1980 Chart 4 F O M C R a n g e s for the F e d e ra l F u n d s R a te Effect on the Domestic Policy Directive The wording of the domestic policy directive was altered substantially by the FOM C’s new operating procedures. The directive adopted at the October 6 meet ing stated: In the short run, the Committee seeks to restrain expansion of reserve aggregates to a pace consistent with deceleration in growth of Ml, M2, and M3 in the fourth quarter of 1979 to rates that would hold growth of these monetary aggregates over the whole period from the fourth quarter of 1978 to the fourth quarter of 1979 within the Committee’s longer-run ranges, pro vided that in the period before the next regular meeting the weekly aver age federal funds rate remains within 197f a range of 11% to 15% percent. The j y Weekly overages of effective daily rales. 7j At each meeting during 1979. the FOMC established a range for the federal funds rate. These ranges are indicated for the firi Committee will consider the need for full week during whicti they were in effect. supplementary instructions if it appears that operations to restrain expansion of reserve aggregates would maintain the federal November 20, 1979 meeting.37 Again, more specific funds rate near the upper limit of its range.35 short-run objectives for M l and M2 growth were This section of the directive — which deals with the FOM C’s short-run objectives — was not specific about the short-run growth rates of M l and M2 that the FOM C desired to achieve. The “Record of Policy Ac tions” of the October 6 meeting, however, was more precise. Specifically, the Committee instructed the Manager to restrain expansion of bank reserves to a pace consistent with growth from September to December at an annual rate on the order of 4.5 percent in Ml and about 7.5 percent in M2 and M3, provided that in the period before the next regular meeting the federal funds rate remained generally within a range of 11% to 15% percent. Because such rates of expansion would result in growth of the monetary aggregates in the upper part of their ranges for the year, the Committee also agreed that over the three-month period somewhat slower growth would be acceptable.36 stated in the “Record” of that meeting (table l ) . 38 At the January 8-9, 1980 meeting, however, specific short-run objectives for M l and M2 growth were ex plicitly incorporated into the domestic policy directive. In the short run, the Committee seeks expansion of reserve aggregates consistent with growth over the first quarter of 1980 at an annual rate between 4 and 5 percent for Ml and on the order of 7 per cent for M2, provided that in the period before the next regular meeting the weekly average federal funds rate remains within a range of 11% to 15% percent.39 Wording almost identical to that in the October 6 directive appeared in the directive adopted at the Although only three new domestic policy directives had been released by the time this article was writ ten, a few observations can be made about the FOM C’s new operating procedures. First, the FOMC no longer targets on a specific level of the federal funds rate as its short-run operating objective. Second, the 4 percentage-point range of the federal funds rate that has been maintained in the directive since October 6 is substantially larger 3<Ibid., pp. 974-76. ssibid, p. 977. 3«Ibid., pp. 975-76. 37“Record” (January 1980), p. 46. 38Ibid., p. 45. 39“Record” (March 1980), p. 236. 12 F E D E R A L R E S E R V E B A N K O F ST. L O U IS than the one-half to three-quarter percentage-point ranges established previously (table 1 ). The absence of a specific federal funds rate objective and the wider range have been evident in both inter-day and intra-day fluctuations of the federal funds rate, which changed radically after October 6. In the FOM C’s new directives, desired growth rates of M l and M2 over two- or three-month periods are established in terms of specific rates of change, or, in the case of the January 1980 directive, a 1 percentagepoint range. In contrast, short-run ranges of M l and M2 growth typically were specified in terms of 3 percentage-point ranges prior to October 6. Further more, the short-run M l and M2 ranges previously were specified as “tolerance ranges” to be used by the Manager of the System Open Market Account only as an indicator “for determining when changes in the funds rate were appropriate. . . .”40 Under current short-run operating procedures, however, the Manager is directed to achieve growth rates of bank reserves consistent with specified growth rates of M l and M2, provided that the federal funds rate remains within a broad range. The Manager of the System Open Market Account, who is responsible for achieving the FOM C’s objec tives, has had to change the focus of domestic open market operations from attaining specific levels of the funds rate to maintaining growth of “reserve aggre gates” consistent with specified rates of M l and M2 growth. These reserve aggregates, however, are not specified in either the directive or the “Records of Policy Actions.” The FOMC votes on growth rates of the monetary aggregates, not the reserve aggregates. Consequently, it is left to the staffs of the Board of Governors and the New York Federal Reserve’s Open Market Desk to establish guidelines for the growth of these reserve aggregates. Domestic Open Market Operations and Reserve Targeting Until recently, information was sketchy about the Federal Reserve’s new procedures for implementing monetary policy. Although various Federal Reserve officials had mentioned that the Open Market Desk was tracking a “family” of reserve aggregates rather than a single aggregate, it was not clear which meas ures were included in this “family” nor how the Desk related growth of these aggregates to the FOM C’s monetary growth objectives. A staff paper released by the Board of Governors on January 31, 1980 clarified 40“Record” (August 1978), p. 663. MARCH 1980 these matters substantially.41 The staff paper outlined the new operating procedure in eight principal steps, highlights of which are presented in exhibit 1. The process of translating the FOMC’s desired rates of short-run monetary growth into weekly tar get levels of total reserves, monetary base, and non borrowed reserves are described in steps 1 through 4 of exhibit 1. Although the staff paper makes it clear that the “overall” objective of the Open Mar ket Desk is to control the growth of total reserves (step 5, exhibit 1 ), it also makes clear that the im mediate short-run operating objective is to control the growth of nonborrowed reserves (step 6, exhibit 1 ). Thus, the Manager of the Desk directs the purchase and sale of government securities in order to attain some pattern of changes in (or path of) the level of nonborrowed reserves between FOMC meetings. At taining the desired path for nonborrowed reserves will not necessarily yield the desired path for total reserves, however, since the actual path of total re serves also depends on the behavior of member bank borrowing at the discount window. For example, if borrowings are larger than the level assumed in con structing the path for nonborrowed reserves, then total reserves will be above the desired path even though nonborrowed reserves are on their desired path. In the event that total reserves are above the Desk’s desired path, the staff paper indicates that cer tain adjustments could be made to bring the growth of total reserves back to the desired path — either a lowering of the nonborrowed reserve path or an in crease in the discount rate (step 7, exhibit 1 ). The first adjustment may be made by the staff of the Board of Governors in conjunction with the Manager of the System Open Market Account. The second adjustment, however, must be initiated by the Boards of Directors of the district Federal Reserve Banks, subject to the approval of the Board of Governors. According to the staff paper, neither adjustment would be made until the growth of total reserves had been overshooting the desired path for some time. The staff paper’s analysis of the effects of open market operations on reserve and money growth un der the new operating procedures is essentially the same as such analyses under the old procedures. According to this staff paper, control of nonborrowed reserves and the discount rate affect the growth of 41“The New Federal Reserve Technical Procedures for Con trolling Money,” Board of Governors of the Federal Reserve System, January 31, 1980. 13 MARCH FED ER A L. R E S E R V E B A N K O F ST. L O U IS 1980 Exhibit 1 Excerpts from “The New Federal Reserve Technical Procedures for Controlling Money” ( 1 ) The policy process first involves a decision by the FOM C on the rate of increase in money it wishes to achieve . . . . ( 2 ) After the objective for money supply growth is set, reserve paths expected to achieve such growth are established for a family of reserve measures. These measures consist of total reserves, the monetary base (essentially total reserves of member banks plus cur rency in circulation), and nonborrowed reserves. Estab lishment of the paths involves projecting how much of the targeted money growth is likely to take the form of currency, of deposits at nonmember institu tions, and of deposits at member institutions (taking account of differential reserve requirements by size of demand deposits and between the demand and time and savings deposit components of M 2). Moreover, estimates are made of reserves likely to be absorbed by expansion in other bank liabilities subject to reserve requirements, such as large CDs, at a pace that appears consistent with money supply objectives and also takes account of tolerable changes in bank credit. . . . [E stim ates are also made of the amount of excess reserves banks are likely to hold. (3 ) The projected mix of currency and demand deposits, given the reserve requirements for deposits and banks’ excess reserves, yields an estimate of the increase in total reserves and the monetary base con sistent with FOM C monetary targets. The amount of nonborrowed reserves — that is total reserves less member bank borrowing — is obtained by initially assuming a level of borrowing near that prevailing in the most recent period. . . . (4 ) Initial paths established for the family of re serve measures over, say, a 3-month period are then translated into reserve levels covering shorter periods between meetings. . . . (5 ) Total reserves provide the basis for deposits and thereby are more closely related to the aggregates than nonborrowed reserves. Thus total reserves rep re sent the principal overall reserve objective. However, only nonborrowed reserves are directly under control through open market operations, though they can be adjusted in response to changes in bank demand for reserves obtained through borrowing at the discount window. [E m phasis added.] (6 ) Because nonborrowed reserves are more closely under control of the System Account Manager for open market operations (though subject to a small range of error because of the behavior of non-controlled factors affecting reserves, such as float), he would initially aim at a nonborrowed reserve target (seasonally un adjusted for operating puiposes) established for the Digitized for14 FRASER operating period between meetings. To understand how this would lead to control of total reserves and money supply, suppose that the demand for money ran stronger than was being targeted — as it did in early October of last year. The increased demand for money and also for bank reserves to support the money would in the first instance be accompanied by more intensive efforts on the part of banks to obtain reserves in the federal funds market, thereby tending to bid up the federal funds rate, and by increased borrowing at the Federal Reserve discount window. As a result of the latter, total reserves and the mone tary base would for a while run stronger than targeted. Whether total reserves tend to remain above target for any sustained period depends in part on the nature of the bulge in reserve demand — whether or not it was transitory, for example — and in part on the degree to which emerging market conditions reflect or induce adjustments on the part of banks and the public. These responses on the part of banks, for example, could include sales of securities to the public (thereby extinguishing deposits) and changes in lending policies. ( 7 ) Should total reserves be showing sustained strength, closer control over them could be obtained by lowering the nonborrowed reserve path (to at tempt to offset the expansion in member bank bor rowing) and/or by raising the discount rate. A rise in the discount rate would, for any given supply of nonborrowed reserves, initially tend to raise market interest rates, thereby working to speed up the adjust ment process of the public and banks and encouraging a more prompt move back to the path for total reserves and the monetary base. Thus, whether adjust ments are made in the nonborrowed path — the only path that can be controlled directly through open market operations — and/or in the discount rate de pends in part on emerging behavior by banks and the public. Under present circumstances, however, both the timing of market response to a rise in money and reserve demand, and the ability to control total reserves in the short run within close tolerance limits, are influenced by the two-week lag between bank de posits and required reserves behind these deposits. (8 ) Other intermeeting adjustments can be made to the reserve paths as a family. These may be needed when it becomes clear that the multiplier relationship between reserves and money has varied from expecta tions. . . . Given the naturally large week-to-week fluctuations in factors affecting the reserve multiplier, deviation from expectations in one direction over a period of several weeks would be needed before it would be clear that a change in trend has taken place. F E D E R A L R E S E R V E B A N K O F ST. L O U IS total reserves and money by influencing the demand for bank reserves and the demand for money (steps 6 and 7, exhibit 1 ). Open market operations that re duce nonborrowed reserves tend to increase inter est rates, which (after a lag) reduce the demand for money, thereby reducing the demand for bank re serves used to support money. A rise in the discount rate tends to increase other market interest rates, which (after a lag) again reduce the demand for money and the demand for bank reserves. Deviations in the growth of total reserves or money from desired rates are controlled, then, by inducing changes in interest rates which affect the demands for money and bank reserves. This analysis of controlling money growth is comparable to Board staff analyses of the FOM C’s pre-October 6 procedures for controlling money growth. An example of this comparability appears in an article by Governor Wallich and Peter Keir which appeared in the September 1979 Federal Reserve Bul letin.42 Written before the introduction of reserve tar geting, this article discussed control of the monetary aggregates via interest rate changes under the old operating procedures. In particular, the authors ex plain the problems of correcting deviations of money growth from desired rates in terms of the pre-October 6 procedures. When incoming data show a sudden marked ac celeration or slowing in money growth rates, the Committee must decide whether the change is a temporary aberration likely soon to be reversed, or a more fundamental change in money demands that stems from a basic adjustment in the perform ance of the economy. If the Committee acted im mediately to counter an observed change in money growth, and the change then proved to be temporary, the action could be destabilizing and require a sub sequent offsetting adjustment. Since C om m ittee actions affect the pu blic’s willingness to h old m oney w ith a lag through their influence on interest rates, such attempts at fine tuning could produce perverse results. [Em phasis added.] To minimize this risk, the FOM C typically has adopted an intermediate position.43 Thus, the overall framework for analyzing the effects of open market operations on reserve and money growth — that open market operations change inter est rates, which affect the demand for money, thereby influencing the demand for bank reserves used to support money — has not changed. W hat has changed under the new operating procedures is the FOM C’s 42Wallich and Keir, “The Role of Operating Guides in U.S. Monetary Policy,” pp. 679-91. 43Ibid., p. 688. MARCH 1980 emphasis on restricting interest rate fluctuations and, consequently, the Desk’s ability to respond to devia tions of money growth from its desired path. The interest rate constraints placed on the Desk under the old procedures are described by Wallich and Keir: When the performance of the money supply ap pears to be deviating from the Committee’s stated two-month ranges, the Manager of the System Ac count is still constrained in his efforts to offset these deviations by a federal funds rate proviso. He can initiate countering open market purchases or sales only so long as these operations, or other market factors, do not push the weekly average federal funds rate outside its specified range, generally 50 to 100 basis points wide. If growth rates for M l and M2 . . . appear to be remaining outside the Committee’s desired ranges, and the Manager’s actions to counter this deviation have moved the funds rate to the upper or lower limit of its range, he must request new instructions from the Committee.44 Prior to October 6, the Committee generally was un willing to allow large movements of the federal funds rate that would have been sufficient to achieve de sired rates of monetary growth within relatively short periods of time. (Recall that the FOM C’s two-month ranges of M l and M2 growth were not themselves treated as short-run targets, but as indicators for de termining when to change the federal funds rate.) Confronted with an unexpected overshoot or under shoot of its money growth targets, the Committee has taken action that neither fully ignores nor fully responds to the miss, until the underlying growth tendency can be differentiated from the “noise” of aberrations in the data. This approach poses some risk that needed countercyclical policy actions will be less timely than desired. But the Committee be lieves that the accentuated volatility in short-term in terest rates likely to result from efforts at instanta neous fine tuning of the aggregates poses a greater risk.46 Like the old procedures, the new procedures also do not attempt to “fine time” the short-run growth of the monetary aggregates. As pointed out in the staff paper (exhibit 1 ), neither deviations of reserve growth nor deviations of money growth from their desired paths will necessarily be responded to immediately. Adjustments of the nonborrowed reserve path or the discount rate would occur only after overshooting or undershooting of the desired paths persisted. W hat is different about the new procedures is that the Desk no longer operates under as restrictive a federal funds rate constraint. The range for the federal funds rate 44Ibid., p. 686. 45Ibid., p. 688. 15 F E D E R A L R E S E R V E B A N K O F ST. LO U IS MARCH 1980 has been widened to 400 basis points— compared with the typical pre-October 6 range of 50 to 100 basis points — and the FOMC no longer specifies a target level of the funds rate (table l ) . 46 Consequently, the Desk has greater flexibility in offsetting deviations of reserve and money growth from their desired paths. quired by the Humphrey-Hawkins Act, announced new one-year ranges for monetary growth on Febru ary 20, 1980, for the period from IV /79 to IV /80. These new ranges were formulated in terms of the recently redefined monetary aggregates — MIA, M1B, new M2, and new M3.48 In effect, these new procedures allow the Manager of the System Open Market Account to stabilize growth of nonborrowed reserves within narrower limits. The old procedures, in contrast, forced the Manager to keep the federal funds rate within nar row limits, which resulted in wider fluctuations in the growth of non-borrowed reserves. The MIA definition of the money stock is the same as old M l except that it excludes demand deposits held by foreign commercial banks and official institu tions. The M1B definition includes MIA plus other checkable deposits.49 These deposits include NOW and ATS accounts, credit union share drafts, and de mand deposits at thrift institutions. The Desk’s implementation of the FOM C’s policy directives prior to October 6 were described in 1977 by the Manager of the System Open Market Account as short-run accommodation of the public’s demand for money. The construction of M1B as an alternative to MIA was an attempt to avoid the problem of interpreting old M l growth when financial innovations such as ATS and New York NOW accounts occurred. The difference in growth between M1B and MIA is in dicative of the problems the FOMC faced in evaluat ing old M l growth in 1979 relative to previous years. From IV /78 to IV /79, MIA increased 5.5 percent — the same as old M l — which represented a reduction in growth from its 7.4 percent increase in the previous year.50 In contrast, M1B increased in 1979 at about the same rate as in the previous year — 8 percent from IV /78 to IV /79 compared with 8.2 percent from IV /77 to IV /78. The difference in growth be tween MIA and M1B reflected the growth of New York NOW accounts and nationwide ATS accounts, for which the FOMC in 1979 attempted to adjust when adopting a one-year range of growth for old M l.51 As this article went to press, Congress approved proposed legislation to allow nationwide use of NOW accounts. This could cause MIA and M1B again to grow at substantially different rates over a transition period. The possibility of nationwide NOW accounts was the primary reason that the Board of Governors adopted these two measures of M l.62 The FOMC’s instructions to the Manager of the System Open Market Account regarding the manage ment of bank reserves provide — to a considerable extent — for the accommodation of the public’s de mand for money in the short run, while at the same time prescribing a response when growth of money appears inconsistent with the Committee’s long-term objectives.47 For example, upward pressure on the federal funds rate was viewed as reflecting an increase in the de mand for money. The Desk “accommodated” this in creased demand by supplying bank reserves to keep the funds rate constant. If the old short-run operat ing procedures could be described as “accommoda tive,” the current procedures — under which the Desk attempts to stay on some growth path of bank re serves — can be described as “less accommodative” of changes in the demand for money. THE YEAR AHEAD The major changes in the implementation of mone tary policy in 1979 will continue to influence mone tary policy actions in 1980. Chairman Volcker, as re46In fact, on March 7, 1980, the FOMC widened the funds rate range to 650 basis points — HVi to 18 percent. See, “Record of Policy Actions,” Federal Reserve Press Release (March 21, 1980), p. 16; forthcoming in the Federal Re serve Bulletin ( April 1980). 47“The Implementation of Monetary Policy in 1976,” Federal Reserve Bulletin (April 1977), p. 326. This article was adapted from a report submitted to the FOMC by Alan R. Holmes and Peter D. Stemlight, Manager and Deputy Man ager of the System Open Market Account, respectively. John S. Hill and Christopher J. McCurdy were primarily respon sible for its preparation. 16FRASER Digitized for The redefinition of the monetary aggregates also alters the FOM C’s choice of short-run operating tar 48“Monetary Policy Report to Congress,” Federal Reserve Bul letin (March 1980), pp. 177-78; and “The Redefined Mone tary Aggregates,” Federal Reserve Bulletin (February 1980), pp. 97-114. 49For a detailed description of all the new monetary aggregates and their relationship to the old measures, see “The Rede fined Monetary Aggregates;” or R. W. Hafer, “The New Monetary Aggregates,” this Review (February 1980), pp. 25-32. 50Old M l growth in 1978 was also about the same as MIA growth—-7.2 percent from IV/77 to IV/78. 51“The Redefined Monetary Aggregates,” p. 102. 62Ibid., p. 100. MARCH F E D E R A L R E S E R V E B A N K O F ST. L O U IS gets in 1980. Because data on MIA and M1B are available weekly, while data on new M2 and M3 are available only monthly, the FOM C’s short-run ranges of monetary growth are being specified in terms of M IA and M1B.53 Consequently, the Desk must now formulate intermeeting paths of total and nonbor rowed reserves consistent with the FOM C’s short-run ranges of MIA and M1B growth. The introduction of reserve targeting in 1979 marks 53“Record of Policy Actions of the FOMC: Meeting Held on February 5, 1980,’’ Federal Reserve Press Release (March 21, 1980), p. 14, forthcoming in the Federal Reserve Bulle tin (April 1980). 1980 the beginning of a different approach to the imple mentation of monetary policy. The effectiveness of the new short-run operating procedures in controlling monetary growth, and these procedures’ effects (or side-effects) on the economy, can be evaluated ade quately only over a longer period of time. Growth of the monetary aggregates decelerated between October and January of this year, in fine with the FOM C’s ob jectives. Nevertheless, analysts undoubtedly will give greater weight to monetary growth in 1980 — relative to the FOM C’s new one-year ranges — in evaluating the effectiveness of the new operating procedures. Supplement: FOMC Discussions in 1979 This supplement consists of selected excerpts from the “Record of Policy Actions” for each of the FOMC meetings in 1979. Each “Record” includes analyses of current and projected economic developments, discus sions of current policy actions, and long- and shortrun operating instructions issued by the FOM C to the Trading Desk. The complete text of each “Record of Policy Actions” appears in issues of the Federal Re serve Bulletin and “Records” for the entire year are published in the Annual Report of the Board of Governors. Meeting Held on February 6, 1979 By late December, staff projections suggested that growth in M2 over the December-January period would be at an annual rate well below the lower limit of the range of tolerance specified for that aggregate and growth in M l would be in the lower portion of its range of tolerance. These developments pointed to a reduction in the ob jective for the federal funds rate toward the 9% percent lower limit of the specified range. However, on December 29 the Committee voted to modify its directive by calling for open market operations directed at maintaining the weekly average federal funds rate at about 10 percent or slightly above. This action was taken in view of uncer tainties surrounding the interpretation of the behavior of the monetary aggregates and in light of domestic economic conditions and developments in domestic and interna tional financial markets. On January 12 the Committee held a telephone conference to review the situation and to consider whether supplementary instructions were needed, but no change was made in the instruction to the Manager. Most market interest rates declined on balance during the intermeeting period. Factors apparendy contributing to this development included a market sentiment that further tightening in monetary policy had become less likely in light of the behavior of the monetary aggregates and the better performance of the dollar in foreign ex change markets. Another influence appeared to be the re cent modest growth of total business credit demands. W ith respect to the economic situation and outlook, most members of the committee expressed little or no dis agreement with the staff projection of a marked slowing in the expansion by the second quarter of 1979 and of a sustained slow rate of growth over the rest of the year accompanied by some increase in the rate of unemploy ment. However, a few members questioned whether a very slow pace of growth was sustainable and suggested that the onset of a recession before the end of the year, with a larger increase in the unemployment rate, was the more likely development. Other members thought that over the past few months the probabilities of the develop ment of a recession before the end of this year had de clined somewhat. It was also observed that expansion might prove to be stronger than projected by the staff, especially if businessmen believed that effective steps were being taken to moderate the rate of inflation. The members continued to anticipate a relatively rapid rise in average prices. Inflation was viewed as a distortion that could contribute to the development of a recession, and it was noted that forecasters typically had underesti mated the strength of inflationary forces. In this connec 17 F E D E R A L R E S E R V E B A N K O F ST. L O U IS MARCH 1980 tion, it was observed that the economy was operating at a higher rate in relation to its potential than had been thought earlier. At the same time, it was noted, developments since the turn of the year were apparently mixed, contributing to increased uncertainty. In the Committee’s discussion, stress was placed on the importance of adopting [one-year] ranges for mone tary growth over the year ahead that would be consistent with a reduction in the rate of inflation, thereby reinforc ing the governmental actions over recent months in pur suit of that objective. Many members of the Committee thought that the staff was overly optimistic in projecting continued, if sluggish, growth in real GNP throughout the second half of 1979; they believed that the chances of a recession beginning before the end of the year or in early 1980 were fairly high. The recent increase in the price of oil, the accelera tion of the overall rise in prices, and the sluggish growth of the monetary aggregates over the latest five months were cited among the factors that increased the probabil ity of a recession. The observation also was made that if a recession developed, it was likely to be moderate and short-lived. In the discussion of policy for the period immediately ahead, most members of the Committee favored directing operations initially toward maintaining the money market conditions currently prevailing, as indicated by a federal funds rate of 10 percent or slightly higher, but some senti ment was expressed for a slight additional finning in money market conditions. The views of the members dif fered primarily with respect to the influence that the incoming evidence concerning growth of the monetary aggregates should have on the objective for the funds rate later in the period before the next meeting. Meeting Held on March 20, 1979 In January and February growth of total credit at U.S. commercial banks accelerated considerably from its re duced pace during late 1978. Expansion in business loans was unusually strong, and banks also added substantially to their holdings of securities. M l declined in both January and February, M2 changed little, and M3 grew at a relatively slow rate. With interest rates remaining high, the behavior of all three monetary aggregates was affected by unusually large shifts of funds from deposits to money market mutual funds and other liquid assets. The weakness in M l also reflected the effects of continuing movements of funds from demand deposits to savings deposits associated with the recently authorized automatic transfer service (A TS) and negoti able orders of withdrawal (N O W ) accounts in New York State. At the beginning of March, projections suggested that over the February-March period M l would grow at a rate moderately below the lower limit of the range established by the Committee and M2 would grow at a rate just below the lower limit of its range. In a special telephone meeting on March 2, the Committee instructed the Man ager to continue aiming for a weekly average federal funds rate of 10 percent or slightly higher. Most market interest rates rose moderately on balance during the intermeeting period, after having declined in January. In the Committee’s discussion of the current economic situation, attention was drawn to the more rapid expan sion in output of goods and services in the fourth quarter of 1978 than had been anticipated. The Commerce D e partment had just released a second upward revision in its estimate of growth in real gross national product in that quarter, and it was observed that the rate of resource utilization therefore was higher than had been thought earlier, accounting in part for the recent intensification of upward pressures on prices. Digitized for18 FRASER The members expressed some differences of opinion con cerning prospects for prices. A significant easing from the rapid rise of recent months was suggested, to the ex tent that recent increases in prices represented temporary factors or were made in anticipation of possible price and wage controls. Moreover, slackening of economic ac tivity later in the year could be expected to slow the rise in prices generally. The view was also expressed, how ever, that inflation would remain rapid even during a recession. In any case, it was observed, a long lag could be expected in the response of prices to the additional measures of restraint imposed toward the end of 1978. In contemplating policy for the period immediately ahead, the Committee continued to face unusual uncer tainties concerning the forces affecting monetary growth. A staff analysis had suggested that M l was likely to ex pand in March, contributing to a pickup in growth of M2. Nevertheless, M l was expected to register a decline in the first quarter, on a quarterly average basis. It was estimated that shifts of funds from demand deposits to savings accounts with automatic transfer services and to the NOW accounts in New York had depressed growth of M l by about 3 percentage points in the quarter. Moreover, it appeared that growth of both M l and M2 had been affected by a downward shift in the public’s demand for money in relation to income, although the magnitude of that effect was uncertain. In the Committee’s discussion, several members stressed their concern about the shortfall in monetary growth rela tive to the longer-run ranges that the Committee had adopted at its meeting on February 6, 1979, especially in view of the risks that a recession might develop in the period ahead. Supporting the goal of bringing growth of the monetary aggregates up into those ranges over a num ber of months, particularly because of the uncertainty about the outlook for economic activity, they favored di recting operations in the period just after the meeting toward maintaining the money market conditions currentiy prevailing — as indicated by a federal funds rate of 10 percent or slightly higher — or toward a little less firm ness in those conditions. The objective of operations later in the period before the next regular meeting of the Committee would be determined on the basis of the in coming evidence on the behavior of the monetary aggre gates, although it was suggested that the Committee con sult again before any change was made in the operational objective for the funds rate. F E D E R A L R E S E R V E B A N K O F ST. L O U IS Other members of the Committee emphasized the re cent acceleration of the rise in prices, and they believed that action should be taken to demonstrate that inflation represented the greatest risk to economic stability over a period of time. Accordingly, they advocated directing ini tial operations in the period ahead toward a slight firming in money market conditions, represented by an increase in the objective for the federal funds rate to about 10% percent. Their prescription for operations later in the pe riod called for holding the objective for the funds rate within a relatively narrow range. The Manager was instructed to direct open market opera tions initially toward maintaining the federal funds rate at about the current level, represented by a rate of about 10 percent or slightly higher. Meeting Held on April 17, 1979 Total credit at U.S. commercial banks expanded at a much slower pace in March than in January and February, as growth in real estate and business loans moderated considerably and banks reduced their holdings of securi ties. However, commercial paper issued by nonfinancial firms increased sharply, and the overall rate of short-term business borrowing was maintained. For the first quarter as a whole, nonfinancial businesses substantially increased their borrowing in short- and intermediate-term markets. At the same time, they reduced their public offerings of bonds to the smallest quarterly total since 1973. The narrowly defined money supply, M l, grew some what in March after having declined in both January and February. The broader monetary aggregates, M2 and M3, expanded at relatively slow rates during the month, al though growth in both measures picked up somewhat from the pace earlier in the year. In late March and early April staff projections suggested that over the March-April period M l would grow at a rate close to the lower limit of the range established by the Committee and M2 at a rate just below the midpoint of its range. These projections were not viewed as suffi ciently weak in relation to the Committee’s ranges to call for a change in the federal funds rate objective of 10 percent or slightly higher. Short-term interest rates fluctuated over a fairly wide range during the intermeeting period and generally rose a little on balance. In the Committee’s discussion of the current economic situation and outlook, attention was drawn to the indica tions of considerably slower growth in real output of goods and services in the first quarter of 1979 than had appeared likely earlier. It was noted that residential construction and consumer spending for goods had weakened more than had been anticipated, and that such expansion as had occurred in the first quarter apparently reflected a substantial acceleration in the growth of business inventories. The members in general anticipated relatively slow growth in economic activity for the near term, and some believed that growth could remain at a sluggish pace for many quarters. Many continued to believe that the proba bilities of a downturn in activity before the end of 1979 MARCH 1980 were fairly high, especially in view of the unusually long duration of the current business expansion. It was also sug gested by some that a pickup in activity, based in part on a surge in business demands for equipment and for inven tories, might occur and persist for a time before an even tual downturn. As at other recent meetings, great concern was expressed about inflation. It was observed that the rate of increase in prices had tended to accelerate from year to year recently and that there were few if any indications of a near-term reversal in that momentum. In contemplating policy for the period immediately ahead, the Committee continued to face uncertainties con cerning the forces affecting monetary growth. A staff analysis had suggested that M l, after having registered a decline in the first quarter, would expand over the April-May period, reflecting in part rapid growth in nom inal GNP. It was anticipated that shifts of funds from demand deposits to savings accounts with automatic trans fer services and to NOW accounts in New York State . . . would have a somewhat less dampening effect on growth of M l in the period immediately ahead than in the first quarter. Moreover, it was assumed that the public’s demand for money in relation to income would continue to shift downward, but at a sharply slower pace than in recent months. Thus, the rise in the income velocity of M l was expected to be relatively rapid, but less than the unusually rapid rate of the two most recent calendar quarters. In the Committee’s discussion at this meeting, as at the meeting on March 20, 1979, several members stressed their concern about the degree of the shortfall in mone tary growth relative to the longer-run ranges that the Com mittee had adopted at its meeting on February 6. It was observed that restrictive policy actions taken in late 1978 had contributed to the recent slowing of monetary growth (after allowance for the impact of special factors) and apparently also to a moderation of the expansion in eco nomic activity. Now, some easing in money market condi tions might be appropriate, with the objective of raising growth of the monetary aggregates over a number of months into the longer-run ranges and of helping to sup port economic activity later in the year. However, an easing in money market conditions was generally regarded as premature in the current environ ment of rapidly rising prices, although it was felt that monetary policy could have little if any immediate effect on prices of food, energy, and housing items, which had been largely responsible for the recent acceleration of the overall rise. Given the staff expectation of a near-term strengthening of monetary growth, most members advo cated or found acceptable a policy of directing operations early in the period immediately ahead toward maintaining the money market conditions currently prevailing . . . and of having the objective for operations later in the period before the next regular meeting determined on the basis of incoming evidence on rates of growth of the monetary aggregates over the April-May period in relation to the growth rates currently anticipated. A few members advocated an immediate increase in the 19 F E D E R A L R E S E R V E B A N K O F ST. L O U IS MARCH 1980 objective for the federal funds rate to IOV4 percent or IOV2 percent and a range for subsequent operations provid ing for a further increase in the funds rate if incoming evidence suggested relative strength in growth of the mon etary aggregates. They stressed the recent acceleration in the rise in prices and high rates of resource use, and they continued to believe that action should be taken to dem onstrate that inflation represented the greatest risk to eco nomic stability over a period of time. In their view, in flationary expectations had increased over recent months while interest rates on balance had changed little. In the current circumstances, moreover, they attached little sig nificance to the behavior of the monetary aggregates. view of the inflationary pressures that had been accumu lating. It was noted that some reduction in growth of nominal GNP had been an objective of the restrictive pol icy actions taken last autumn, although the reduction had so far been reflected in growth of real GNP rather than in the rate of inflation. Meeting Held on May 22, 1979 In contemplating policy for the period immediately ahead, the Committee took note of a staff analysis sug gesting that over the May-June period growth of M l would be quite slow, in part because of the unwinding of the transitory effects of federal income tax payments and refunds that had contributed to its exceptionally rapid growth in April. It was expected that growth of M2 over the two-month period would be retarded by the slow growth of M l but that growth of the interest-bearing com ponent would remain relatively strong. Total credit outstanding at U.S. commercial banks grew rapidly in April, as it had on balance during the first quarter of the year. The April growth was lead by a re bound in expansion of business loans, which had slack ened in March from rapid rates in January and February. Commercial paper issued by nonfinancial firms increased sharply in April for the second consecutive month. The narrowly defined money supply, M l, expanded sharply in April, after having declined on the average in the first quarter. A substantial part of the April increase was attributable to delays in the Treasury’s processing of checks in payment of federal income taxes and to bunch ing of tax refunds. Reflecting in part the behavior of M l, growth of M2 and M3 accelerated to rapid rates in April from relatively slow rates in the first quarter. In late April projections suggested that over the AprilMay period M l and M 2 would grow at rates that were close to or above the upper limits of their respective ranges. In accordance with the directive issued at the meeting on April 17, operations were directed toward an increase in the federal funds rate to a level of about IOV4 percent. Subsequently, in early May, the two-month rates of growth projected for M l and M2 were somewhat stronger. However, financial markets appeared to be in a sensitive state, and recent developments affecting supplies and distribution of energy were adding to uncertainties about economic prospects. Moreover, it appeared that the rapid pace of monetary growth was attributable in part to transitory forces. In the circumstances, and in view of the directive’s instruction to give due regard to developing conditions in domestic financial markets, the objective for the federal funds rate was maintained at lOVi percent. Short-term interest rates in general changed little on balance during the intermeeting period. In the Committee’s discussion of the economic situa tion and outlook, the members in general agreed that the pace of expansion in economic activity had slowed signifi cantly, apart from the effects of severe weather in the first quarter and of the work stoppage in the trucking industry early in the current quarter. . . . A number of members now thought that a cyclical peak in activity might well be registered in the current quarter. Despite the current risks of recession in activity, the slowing of the expansion from the excessively rapid pace in late 1978 was regarded as a desirable development, in Digitized for20 FRASER Members continued to express great concern about in flation. . . . According to a number of economic projec tions, moreover, deceleration of inflation would be a slow and lengthy process. The observation was made that if the rate of inflation was not sharply reduced in the months immediately ahead, renewed expansion in business activ ity would begin with prices rising at a relatively fast pace. Most members of the Committee believed that, despite increasing signs of weakening in economic activity and the risks of recession, a general easing in monetary re straint at this time would be premature in view of the continuance of strong inflationary pressures. Given the staff expectations of slow growth in M l and M2 over the MayJune period, they favored a policy of directing open mar ket operations early in the period immediately ahead to ward maintaining the money market conditions currently prevailing . . . and of having the objective for operations later in the period before the next meeting determined on the basis of incoming evidence on the behavior of the monetary aggregates in relation to that currently antici pated. In view of uncertainties concerning interpretation of credit conditions and monetary growth in the current environment, they also favored specifying unusually wide ranges for growth of M l and M2 over the May-June period and giving greater weight than usual to money market conditions in the conduct of operations until the next meeting. Subsequent to the meeting, on June 15, incoming data indicated that M l and M2 were growing at exceptionally rapid rates in early June, and projections suggested that for the May-June period both monetary aggregates would grow at annual rates above the upper limits of the ranges that had been specified by the Committee. . . . However, in view of many indications of weakening in economic activity, the difficulties of interpreting the behavior of the aggregates in the light of these circumstances, the condi tion of financial markets, and the general uncertainty about the economic outlook, Chairman Miller recommended that the Manager be instructed to continue to aim for a federal funds rate of about 10 *4 percent. Meeting Held on July 11, 1979 Total credit outstanding at U.S. commercial banks con tinued to expand rapidly in May and June, but the rate F E D E R A L R E S E R V E B A N K O F ST. LO U IS of growth for the two months combined was down some what from the average pace in earlier months of the year. Increases in bank loans during May and June were concentrated in the business and real estate categories. Commercial paper issued by nonfinancial firms rose con siderably further over the two months. The narrowly defined money supply, Ml, increased sharply in June and the broader measures of money, M2 and M3, also grew rapidly.. . . Federal funds traded somewhat above the Committee’s objective in late June and early July, in response to pres sures associated with unusual churning in the money mar ket around the midyear bank statement date and the July 4 holiday. Most interest rates other than the federal funds rate fell substantially on balance during the intermeeting pe riod. The declines appeared to be in response to the grow ing evidence that economic activity had been weakening. With respect to the economic situation and outlook, no member of the Committee expressed disagreement with the staff appraisal that real gross national product had de clined somewhat in the second quarter and that further declines were likely for the remaining two quarters of the year. The suggestion was made that the recession was most likely to be mild and short-lived. However, it could prove to be more severe than currently expected because the recent increases in prices of energy items and inflation generally were reducing disposable income and eroding the financial position of the household sector. Another reason advanced for thinking that the recession could be more severe was the possibility that the down turn in economic activity would become widespread among industrial countries. Members continued to express great concern about in flation. It was suggested that the unexpectedly large in creases in OPEC oil prices in late June had seriously harmed the government’s anti-inflation efforts. Thus, wind ing down the rate of increase in prices might well take considerably longer than had been thought earlier and would be more costly in terms of its impact on output, employment, and real income. In the discussion of policy for the period immediately ahead, members of the Committee in general favored di recting open market operations initially toward maintain ing the money market conditions currently prevailing . . . on the expectation that over the July-August period growth of Ml and M2 would be both moderate and consistent with their longer-run ranges. Some sentiment was ex pressed for a near-term reduction in the federal funds rate because of the downturn in economic activity, but it was agreed that current conditions in foreign exchange markets militated against a prompt reduction. About a week after the meeting, on July 19, projections suggested that over the July-August period Ml would grow at an annual rate moderately above the upper limit of the range of 2% to 6% percent that had been specified by the Committee and that M2 would grow at a rate about equal to the upper limit of its range of 6 V2 to 10% per MARCH 1980 cent; in those circumstances, the Manager began to aim for a weekly average federal funds rate at about the 10% percent upper limit of its range. On July 27, with the pro jections suggesting that growth of both Ml and M2 over the July-August period would exceed the upper limits of their ranges and with the objective for the federal funds rate at the upper limit of its range, the Committee voted to modify the directive adopted at the meeting on July 11. Specifically, the Committee raised the upper limit of the intermeeting range for the federal funds rate to 10% percent and instructed the Manager to aim for a rate within a range of 10% to 10% percent, depending on sub sequent behavior of the monetary aggregates, on conditions in foreign exchange markets, and on the current Treasury financing. Meeting Held on August 14, 1979 Expansion of total credit outstanding at U.S. commercial banks, which had picked up in June, moderated in July to about the April-May pace. Growth in loans also mod erated in July after an acceleration in June. Banks con tinued to add sizable amounts to their holdings of securi ties, especially U.S. government obligations. Growth in commercial paper issued by nonfinancial firms exceeded the strong second-quarter pace, owing in part to large sales by foreign issuers. The monetary aggregates — Ml, M2, and M3 — con tinued to expand rapidly in July. In the Committee’s discussion of the economic situation and oudook, none of the members expressed disagreement with the staff appraisal that real gross national product was continuing to decline in the current quarter. How ever, members expressed considerable uncertainty about the duration and extent of the decline in activity. Members continued to express great concern about in flation. It was observed that for a long period elements in the economic situation had seemed to justify expecta tions of a reduction in the rise in prices. Such expectations had been disappointed. Moreover, little reduction could be expected in the short run because recent increases in energy prices had not yet fully worked through the price structure. It was noted that the decline in the rate of inflation projected for the quarters immediately ahead was small, and much smaller than that associated with the pre vious recession. Thus, inflation might still be at a high rate when economic activity turned up again. Inflationary ex pectations appeared to have worsened in the sense that, more than ever before, consumers and businessmen seemed to take the inflationary environment into account in mak ing spending and investing decisions. In considering policy for the period immediately ahead, Committee members focused on the problems posed by emerging recession and its potential for substantial in creases in unemployment, concurrent with strong mone tary growth, high actual and expected rates of inflation, and an exposed position of the dollar in foreign exchange markets pending anticipated improvement in the U.S. for eign trade and current accounts. There was little disagreement with the proposition that 21 F E D E R A L R E S E R V E B A N K O F ST. LO U IS for the near term modest measures should be taken to direct policy toward slowing growth of the monetary ag gregates. Control of monetary growth was regarded as essential to restore expectations of a decline in the rate of inflation over a period of time. In support of modest measures directed toward restraint, it was suggested that monetary policy recently had not been so restrictive as it might have appeared. Monetary growth since the beginning of the year had been con siderably greater than that indicated by M l, owing to rapid expansion in close substitutes for demand deposits and currency. In addition, the increase in interest rates had been less than that in expected rates of inflation. On the other hand, it was noted that interest rates were close to historic highs. Some doubt was expressed, more over, that further restraint could have a significant effect on inflation, particularly in view of the role of energy in the rapid rate of increase in prices recently. In the face of clear evidence of weakening in economic activity, it was observed, the need to balance the objective of con taining the recession with the goal of moderating infla tion called for a steady policy for the time being. In considering policy specifications for the period im mediately ahead, the Committee took note of a staff analy sis suggesting that the current growth rate of nominal GNP and other influences, including possibly a temporary accumulation of precautionary balances by the public in response to unusual uncertainties, were tending to sup port the demand for money. On the assumption of con tinuance of prevailing money market conditions, therefore, growth of both M l and M2 over the August-September period most likely would be high relative to the Com mittee’s longer-run ranges, although growth could be ex pected to slow substantially from the rapid rates of recent months. At the conclusion of its discussion of policy, the Com mittee decided to instruct the Manager for Domestic Op erations to direct open market operations initially toward an increase in the weekly average federal funds rate to about 11 percent. Subsequent to the meeting, in late August, incoming data indicated that M l and M2 were growing at rapid rates in August. On August 30, projections for the AugustSeptember period suggested that growth of M l would be at an annual rate well above the upper limit of the range that had been specified by the Committee and that growth of M2 would be at about the upper limit of its range. Over the preceding week, the Manager for Domestic Op erations had been aiming for a weekly average federal funds rate approaching the 1 IVi percent upper limit of its specified range, and in the statement week ending August 29, the rate averaged 11.16 percent. In these circum stances, Chairman Volcker recommended that the upper limit of the range for the funds rate be raised to 11% percent, but with the understanding that not all of the additional leeway would be used immediately; use of the leeway would depend on subsequent behavior of the monetary aggregates and on developments in foreign ex change markets. The Committee voted to amend the domestic policy directive in accordance with the Chair man’s recommendation. 22 MARCH 1980 Meeting Held on September 18, 1979 Total credit outstanding at U.S. commercial banks grew more slowly in August than in most earlier months of the year. Banks’ holdings of Treasury obligations declined and growth in their total loans moderated. However, busi ness loans continued to expand rapidly in August and commercial paper issued by nonfinancial firms again in creased sharply. The monetary aggregates — M l, M2, and M3 — con tinued to expand at relatively rapid rates in August and early September, although somewhat less rapidly than in June and July. Short-term interest rates rose substantially during the intermeeting period, in response to strong business de mands for credit as well as to the System’s actions firming money market conditions and to expectations of further monetary restraint. Bond yields also increased somewhat. In the Committee’s discussion of the economic situation and outlook, none of the members expressed disagreement with the staff appraisal of some further contraction in real gross national product after the current quarter’s in terruption of the decline. However, members continued to express uncertainty about the duration and extent of the contraction in activity. Members continued to express great concern about the rapid rise in prices. It was observed that inflation was more persistent now than it had been in earlier periods of some weakening in demands and that there was still a tendency to underestimate its strength. Furthermore, the current and foreseeable rate of inflation could itself lead to additional shocks to the economy. In contemplating policy for the period immediately ahead, Committee members took note of a staff analysis suggesting that growth of M l was likely to taper off dur ing the September-October period in response to the lagged effects of the substantial increase in interest rates during the summer and the prospective weakening of ex pansion in nominal GNP. However, growth over the two months would still be relatively high. Members who favored policy measures directed toward some additional firming in money market conditions stressed the importance of achieving a significant reduc tion in the pace of monetary expansion over the months ahead. Such a reduction was necessary if growth over the year ending in the fourth quarter of 1979 was to be held well within the longer-run ranges that had been reaffirmed by the Committee in July. Additional measures to restrain monetary growth, moreover, would tend to lower expected rates of inflation and, consequently, would have a con structive influence on a range of decisions affecting prices and wages as well as the value of the dollar in foreign exchange markets. It was suggested, in addition, that monetary policy had not been as restrictive as it might have appeared. Despite the level of interest rates, credit demands and credit ex pansion remained strong. Interest rates after allowance for expected rates of inflation were not high. Furthermore, monetary growth this year had been greater than indicated F E D E R A L R E S E R V E B A N K O F ST. L O U IS by M l alone, owing to rapid expansion in close substi tutes for demand deposits and currency. In support of a policy directed toward maintenance for the time being of prevailing money market conditions, members emphasized the substantial rise in interest rates over the past two months and the tendency of changes in rates to affect monetary growth and economic activity only after a considerable lag. In this connection, it was ob served that growth of demand deposits had slowed mark edly in July and August, while expansion of M l had been supported by an unexplained pickup in growth of cur rency in circulation. Growth of the monetary aggregates was likely to taper off in coming months, and additional firming in money market conditions might slow growth to an unwanted degree. In the current circumstances, the Committee should avoid policy actions that might inten sify the developing weakness in economic activity. At the conclusion of its discussion of policy, the Com mittee decided to instruct the Manager for Domestic Op erations to direct open market operations initially toward a slight increase in the weekly average federal funds rate to about 11 Vi percent. Meeting Held on October 6, 1979 This meeting of the Committee was called by the Chair man to consider actions that might be taken, in conjunc tion with actions being contemplated by the Board of Governors, to improve control over the expansion of money and bank credit in the light of developing speculative ex cesses in financial and commodity markets and additional evidence of strong inflationary forces in the economy. Special attention was given to the conduct of open mar ket operations in order to contain growth in the monetary aggregates within the ranges previously adopted by the Committee for the year ending in the fourth quarter of 1979. The information available at the time of the meeting suggested somewhat stronger economic activity in the third quarter than had been indicated at the time of the Com mittee’s meeting on September 18, and real output of goods and services was estimated to have recovered a sig nificant part of the second-quarter decline. According to staff projections, however, a decline in activity in the fourth quarter still appeared probable. Prices on the aver age were continuing to rise somewhat more rapidly than anticipated earlier, in part because of additional large in creases in energy items and renewed upward pressures on foods. Moreover, developments in spot and futures mar kets for a number of commodities were indicative of an intensification of speculative activity and of the possibility of a further singe in prices. In foreign exchange markets the weighted average value of the dollar against major foreign currencies had de clined substantially since the Committee’s meeting in midSeptember, and monetary authorities had purchased, net, a large amount of dollars. Interest rates had remained under considerable upward pressure since mid-September, and most yields had risen to new highs for the year. The monetary aggregates — M l and M2 — continued MARCH 1980 to expand at rapid rates in September, and growth in bank credit appeared to have accelerated appreciably from its pace in the prior two months. In the Committee’s discussion of policy for the period immediately ahead, the members agreed that the current situation called for additional measures to restrain growth of the monetary aggregates over the months ahead. The members felt that growth of the aggregates at rates within the ranges previously established for 1979 remained a reasonable and feasible objective in the light of the avail able information and the business outlook. Given that ob jective, most members strongly supported a shift in the conduct of open market operations to an approach placing emphasis on supplying the volume of bank reserves esti mated to be consistent with the desired rates of growth in monetary aggregates, while permitting much greater fluctuations in the federal funds rate than heretofore. A few members, while urging strong action to restrain mon etary growth, expressed some preference for continuing to direct daily open market operations toward maintenance of levels of the federal funds rate and other short-term in terest rates that appeared to be consistent with the Com mittee’s objectives for growth in the monetary aggregates. The advantages and disadvantages of the different ap proaches were discussed. The principal reason advanced for shifting to an operat ing procedure aimed at controlling the supply of bank reserves more directly was that it would provide greater assurance that the Committee’s objectives for monetary growth could be achieved. In the present environment of rapid inflation, estimates of the relationship among interest rates, monetary growth, and economic activity had become less reliable than before, and monetary growth since the first quarter of 1979 had exceeded the rates expected de spite substantial increases in short-term interest rates. Committee members recognized that for a number of rea sons the relationship between growth of various reserve measures and growth of the monetary aggregates was not precise; thus the shift in emphasis to controlling reserves improved prospects for achievement of the Committee’s objectives for monetary growth over the next few months but did not assure it. Committee members suggested that the shift in operat ing techniques, along with the other actions being con templated by the Board of Governors, would tend to increase confidence at home and abroad in the System’s de termination to achieve its objectives for monetary growth and to avoid further deterioration in the inflationary out look. Partly because it would increase uncertainty about the near-term course of interest rates, the new operating technique should induce banks to exercise greater caution in extending credit and might dampen speculative beha vior by increasing its risks and costs. Altogether, the System’s action would tend to moderate inflationary ex pectations, thereby exerting a constructive influence over time on decisions affecting wages and prices in domestic markets and on the value of the dollar in foreign ex change markets. The observation was made that the new emphasis in open market operations might be accompanied by larger increases in interest rates in the immediate future than 23 F E D E R A L R E S E R V E B A N K O F ST. LO U IS would otherwise occur. On the other hand, the emphasis on reserves also could be expected to produce a shift toward easier conditions in money markets more promptly whenever the demand for money and credit abated sig nificantly in response to a weakening in economic activity. The point was made that an easing in money market con ditions under circumstances in which growth of monetary aggregates was restrained, economic activity was weak ening, and the rise in prices was moderating should not adversely affect inflationary expectations and the value of the dollar in foreign exchange markets. At the conclusion of the discussion and after full consid eration of the advantages and disadvantages of alternative courses of action, the Committee agreed that in the con duct of open market operations over the remainder of 1979 the Manager for Domestic Operations should place primary emphasis on restraining expansion of bank re serves in pursuit of the Committee’s objective of decel erating growth of M l, M2, and M3 to rates that would hold growth of these monetary aggregates over the year from the fourth quarter of 1978 to the fourth quarter of 1979 within the Committee’s ranges for that period. Subsequently, on October 22, 1979, the Committee held a telephone conference to review the situation and to con sider whether supplementary instructions to the Manager were needed. Since October 6, expansion of total reserves had exceeded the pace consistent with the Committee’s objective for growth of the monetary aggregates during the fourth quarter. At the same time, the federal funds rate had begun fluctuating close to the upper limit of the 11% to 15% percent range established by the Committee. It was recognized that the desired restraint in the expan sion of total reserves might involve continued pressure on money market conditions, including higher levels of mem ber bank borrowings from the Federal Reserve than had been anticipated, as banks made orderly adjustments that would in time slow monetary growth. It was not clear, however, that retention of the 15% percent upper limit of the range for the federal funds rate would be inconsistent with the desired restraint on monetary growth. Moreover, unsettled conditions in financial markets also suggested no change in the upper limit of the range for the federal funds rate. Consequently, no change was proposed in the domestic policy directive issued at the meeting on October 6. Meeting Held on November 20, 1979 Over the first half of October, measures of bank reserves in general grew faster than had been anticipated at the time of the meeting on October 6, both because demands for reserves were unexpectedly strong and because System operations provided more reserves than had been expected. Subsequently, System operations were directed more firmly at restraining growth of reserves. As such operations lim ited growth of nonborrowed reserves while demands for reserves remained strong, member bank borrowings rose to a daily average of about $3 billion in the last two statement weeks of October and the federal funds rate rose to an average a little above 15V2 percent in the final week. In the first half of November, demands for reserves eased, and member bank borrowings subsided to a daily 24 MARCH 1980 average of about $2 billion and the federal funds rate declined to an average of about 13% percent. Growth of M l, which had accelerated in September and had been exceptionally rapid in the third quarter as a whole, slowed to an annual rate of 2% percent in Oc tober. Growth of M2 slowed less than that of M l, to a rate of about 8% percent in October, as overall expan sion in the interest-bearing components remained strong. A marked rise of net flows into money market certificates and other time deposits at commercial banks, fostered by substantially higher deposit yields, offset a sharp reduction in savings deposits. Growth in loans and investments at commercial banks moderated appreciably in October. Since early October interest rates had risen sharply in both short- and long-term markets and had been unusually volatile. In foreign exchange markets the downward pressure on the dollar that had developed in September was reversed in early October, and by the end of the month, the tradeweighted value of the dollar against major foreign cur rencies had risen about 3% percent. Around mid-Novem ber, however, the dollar came under renewed downward pressure and lost a portion of its October gain, in part reflecting developments relating to Iran. In the Committee’s discussion of the economic situation and outlook, the members in general agreed with the staff appraisal that the unexpectedly strong rebound in real gross national product in the third quarter would be fol lowed by some contraction in activity and by a rise in unemployment, although uncertainty was expressed about the depth and duration of the anticipated downturn as well as about its precise timing. Some members cited the onset of the heating season with energy prices so much higher than a year earlier, the overall rate of inflation, the recent sharp rise in interest rates, and the developing stringency in some financial markets as influences that might cause the contraction to be relatively severe. Continuation of the rapid rise in prices of goods and services remained a major concern of Committee mem bers, some of whom thought that the risks were on the side of a rise greater than that currently anticipated. The prospects for supplies and prices of oil, which would have a substantial effect on the economy, were regarded as es pecially uncertain, in view of the political situation in Iran and of the meeting of petroleum-exporting countries sched uled to begin on December 17. In contemplating policy for the period immediately ahead, the Committee took note of a staff analysis indi cating that the behavior of the monetary aggregates since September had been reasonably consistent with the policy adopted on October 6 . . . . In the Committee’s discussion of policy for the period immediately ahead, the members indicated that in the present circumstances pursuit of the goal of restraining growth of the monetary aggregates from the fourth quar ter of 1978 to the fourth quarter of 1979 within the ranges previously established for that period remained feasible and desirable; they agreed that in pursuit of that F E D E R A L R E S E R V E B A N K O F ST. L O U IS underlying goal, the broad objectives for monetary growth during the current quarter adopted at the meeting on Oc tober 6 were still appropriate. In contemplating objectives for rates of monetary growth over the weeks through the end of 1979 and into January 1980, the members differed somewhat in their views concerning the extent to which operations should be directed toward promoting accelera tion in growth of M l from the recently reduced rates. A few members favored operations consistent with the Oc tober 6 decision to seek a 4% percent annual rate of growth in M l over the September-December period. A few members favored acceptance of a significantly slower rate of growth for the quarter. Most members, however, advocated a compromise between those two prescriptions. It was recognized that, while the decision affecting such a short period would have quite minor implications for mon etary growth over the year ending in the fourth quarter of 1979, it would affect credit and money market conditions in the weeks ahead and the path of monetary growth en tering the new year. Views with respect to an acceptable range of fluctua tion for the federal funds rate did not vary greatly. It was agreed that the range should continue to be relatively wide, and most members indicated a preference for re taining the range of 11 ¥ 2 to 15% percent adopted at the October 6 meeting. Meeting Held on January 8-9, 1980 Over the first four weeks after the November meeting, both total and nonborrowed reserves grew at about the rates projected at the time of the meeting. Member bank borrowings averaged about $1% billion, compared with an average of slightly less than $2 billion in the preced ing three weeks, and the federal funds rate continued to average around 13% percent. Toward the end of the fourweek period, however, the demand for reserves appeared to be easing relative to the path consistent with desired monetary growth. In the three weeks remaining before this meeting, member bank borrowings declined to a daily average of about $1.1 billion. Despite the decline in bor rowings, the federal funds rate edged up to an average of about 14 percent in late December and early January, at least in part because of exceptionally large demands for excess reserves around the year-end holidays. Expansion in the major monetary aggregates remained at a reduced pace in November and December, after hav ing slowed markedly in October. Growth in total loans and investments at commercial banks slowed sharply in the fourth quarter. Slower expan sion was especially pronounced in business loans. Growth in real estate loans remained close to the pace in the first three quarters of the year. Since the November meeting of the Committee, interest rates had fluctuated over a relatively wide range, although MARCH 1980 they had been somewhat less volatile than in the previous intermeeting period. On balance, most interest rates had declined. Staff projections suggested that growth of nominal gross national product would slow considerably in the current quarter and then pick up gradually over the remainder of 1980. The projections suggested, however, that a con traction in real GNP would develop in the current quarter and would continue later in the year, although at a di minishing pace in the second half, and that the rate of unemployment would increase substantially. The rise in average prices was projected to accelerate slightly during the early part of 1980, mainly because of increases in energy costs, but to subside later. In the Committee’s consideration of the economic out look, several members stressed the elements of uncertainty in the current situation. The observation was made that the relationships of the past appeared to provide less guidance than usual in appraising the current situation and outlook. In the later part of 1979, for example, overall activity had been unexpectedly strong and the widely anticipated re cession had not developed, although automobile produc tion and housing starts had declined. In the judgment of a number of members, a downturn now seemed to be getting under way, but there was also recognition that it could be delayed for another quarter or two. Inflation remained a major concern. In part because of earlier increases in oil prices and in mortgage interest rates, the consumer price indexes to be published in the next few months probably would continue to show excep tionally large advances. In the discussion of policy for the near term, the mem bers in general considered rates of monetary growth for the three months from December to March within the framework of some reduction in ranges for growth over the whole of 1980 from those for 1979 in pursuit of the Committee’s objective of reducing the rate of inflation. The Committee also took note of a staff analysis indi cating that the demand for money could be relatively weak in the first quarter of 1980, if growth of nominal GNP did in fact slow sharply, and could strengthen as the year progressed. Differences in views concerning the particular rates of monetary growth to be specified for the period from D e cember to March were not great. Preferences were ex pressed for growth indexed by expansion in M l at an an nual rate of 4 percent, a rate of 5 percent, and something between the two. With respect to the acceptable range of fluctuation for the federal funds rate, almost all members preferred to retain the range of 11% to 15% percent, originally adopted at the meeting on October 6, 1979, and continued at the meeting on November 20. One member suggested raising the range slightly, to 12 to 16 percent. 25 The Dynamics and Estimation of Short-Run Money Demand R. W. HAFER and SCOTT E. HEIN i \ STABLE money demand function is crucial to the formation and implementation of effective mone tary policy. Consequently, recent findings of temporal instability in this relationship have concerned both policymakers and economists. Previous studies have examined the stability issue by focusing on the “proper” specification of the money demand equation. For the most part, these studies were directed toward discovering which scale variable and interest rates are appropriate. Unfortunately, such attempts to explain the apparent breakdown in the money demand rela tionship in the early 1970s have not been successful. demand function relate the demand for real money balances (m ") to “the” interest rate ( r ) (measured in nominal terms and therefore incorporating inflation ary expectations) on assets that are thought to be relatively close substitutes for money and to some measure of economic activity, such as real GNP (y ), to capture the volume of transactions undertaken in the economy. Real money balances are conventionally measured by M l divided by the price level (GNP deflator). Within this literature there is surprisingly little at tention devoted to the process by which money bal ances are assumed to adjust to the desired level. This paper investigates the importance of the moneydemand adjustment process as well as the technique used to estimate this relationship. Both the specifica tion of the adjustment process and the estimation technique employed are shown to be significant fac tors in determining whether the short-run money de mand function has been temporally stable during recent years. ( 1) BACKGROUND In the transactions view of the demand for real money balances, money is held primarily for two reasons: the lack of synchronization between receipts and expenditures and the existence of positive trans actions costs.1 Formulations of the transactions money 'In contrast to other analyses which place greater emphasis on money’s role as a store-of-value, the transactions approach focuses on the medium of exchange function played by money in the economy. For an introduction to the transactions ap proach, see Thomas M. Havrilesky and John T. Boorman, Monetary Macro-Economics (Culington Heights: AHM Pub lishing Corp., 1978), pp. 96-113. The standard references on this topic are: John Maynard Keynes, The General Theory of Employment, Interest, and Money (London: Harcourt, Brace and World, 1936); William J. Baumol “The Transac tions Demand for Cash: An Inventory Theoretic Approach,” Quarterly Journal of Economics (November 1952), pp. 545 This relationship may be written as: md = f( r, y). This relationship is typically estimated in the loglinear form, (2 ) In m? = a» + aj In rt + a. In yt + £t, where £ is a random error term. Furthermore, the transactions demand for money framework suggests that the following restrictions should hold for the esti mated regression coefficients: 0 > a, > -0.5, and 1 > a= > 0.5.2 Equation 2 often has been estimated directly using annual data.3 Because equation 2 represents a long56; and James Tobin, “The Interest Elasticity of the Trans actions Demand for Cash,” The Review o f Economics and Statistics (August 1956), pp. 241-47. For an example of money viewed as a store-of-value, see Milton Friedman, “The Quantity Theory of Money: A Restatement,” in Milton Fried man, ed., Studies in the Quantity Theory of Money (Chicago: University of Chicago Press, 1956). 2The Baumol-Tobin framework suggests that ai = -0.5 and a= = 0.5. For a generalization, see Robert J. Barro, “Integral Constraints and Aggregation in Inventory Models of Money Demand,” Journal of Finance (March 1976), pp. 77-88. 3See, for example, Allan H. Meltzer, “The Demand for Money: The Evidence from the Time Series,” Journal of Political Economy (June 1963), pp. 219-46; T. J. Courchene and H. T. Shapiro, “The Demand for Money: A Note from the Time Series,” Journal of Political Economy (November 1964), pp. 1205-19; and David E. W. Laidler, “Some Evidence on the Demand for Monev,” Journal of Political Economy (April 1966), pp. 111-31. ' F E D E R A L R E S E R V E B A N K O F ST. L O U IS run equilibrium in which full adjustment between actual and desired real money balances is completed within one year, no adjustment process is specified. When equation 2 is estimated with quarterly data, however, a more flexible specification is needed to characterize the short-term money market disequilibria that may exist. To do this, “desired” money bal ances are posited to depend upon the same variables found in equation 2. Thus, (3 ) In m® = a„ + a> In rt + ai In yt .+ Et> MARCH 1980 mechanism.5 One important implication of this speci fication is that a decline in the real value of last pe riod’s nominal money stock due to rising prices will be fully and immediately offset by an increase in the amount of nominal money balances currently held. In other words, it is implicitly assumed that an in crease in the price level will induce an im m e d ia te increase in nominal money holdings to equate the real value of last period’s nominal money holdings to the currently desired level. where m? represents desired (or long-run) real money balances for period t.4 However, since actual real money balances (m t ) and desired holdings (m?) may not be equal in the contemporaneous period — be cause transaction costs prevent immediate adjustment of actual balances to their desired levels — a specific stock-adjustment process is specified. The real-adjustment mechanism has been criticized on the grounds that the change in money balances due to a price level change will not occur instan taneously because such adjustments are costly — just as they are when interest rates and income change. Goldfeld and White have suggested an alternative adjustment mechanism, commonly referred to as the nominaZ-adjustment mechanism.6 The most commonly used adjustment mechanism can be formalized as, The nominal-adjustment hypothesis can be written as, (4 ) In mt - In mt-i — X (In m? - In m, i); (0 < X < 1), (7 ) In M, - In M,-, = X' (In Mf - to M ,-,); (0 < X' < 1), where X represents the coefficient of adjustment — the speed at which actual money holdings adjust to the gap between last period’s stock and the currently desired level. Substituting equation 3 into equation 4 yields, where M is nominal money balances, that is, Mt = mt(P t ). Transforming equation 3 so that the left-hand side is equal to In Mt and substituting that equation into equation 7 yields, (5 ) In mt - In m,-i = X [(a» + ai In r, + a2 In yt + £t) In m, ,], (8 ) In Mt - to Mi-i = Xf [(a„ + ai In rt + a2 to yt + to P t + Et) - to Mt-i]. Solving equation 8 for In Mt gives, which, upon simplification, gives the following solu tion for In mt: (9 ) In Mt = X^o + X’ai to rt + X'a» to y, + \f to Pt + (6 ) The dependent variable in equation 9 is specified in nominal terms. Equation 9 usually has been esti mated, however, with real money balances as the de pendent variable. To transform the nominal-adjust ment specification so that real money balances are on In mi = Xa„ + Xai In rt + Xa» In yt + (1 - X) In mt-i + Xe i . Equation 6, then, represents a commonly estimated quarterly money demand function. The adjustment coefficient (X ) is derived from the estimated coeffi cient on the lagged dependent variable (In mt_,). If, for example, the estimated coefficient on In m,-, is 0.7, this indicates a 30 percent (1 - 0.7) per quarter adjust ment of actual money balances to the desired level. Also, whereas the estimated coefficients on In r, and In yt represent the short-run elasticities of real money balances with respect to these variables, dividing these coefficients by the adjustment coefficient (X ) yields estimates of these variables’ long-run elasticities. Equation 4 has been labeled the reaZ-adjustment 4Writing equation (3 ) in nominal form yields, In M? = ao + ai In rt + a2 In yt + In Pt + Et, where In Pt is the natural logarithm of the price level in period t, and In M? is the natural logarithm of the desired level of nominal money balances. (1 - X') to Mt-t + X'fit. 5This nomenclature follows that used by Stephen M. Goldfeld, “The Case of the Missing Money,” Brookings Papers on Eco nomic Activity (3:1976), pp. 683-730. A specification very similar to equation (6 ) can be gen erated if one assumes that the appropriate levels of the de pendent variables are formed adaptively. Thus, the dynamics of the adjustment process could be due to expectation forma tion, rather than transactions costs. See David E. W. Laidler, The Demand for Money, 2nd. ed., (New York: Dunn-Donnelley, 1977), pp. 142-43. ®See Stephen M. Goldfeld, “The Demand for Money Revisited,” Brookings Papers on Economic Activity (3: 1973), pp. 577-638, and “The Case of the Missing Money,” Brookings Papers on Economic Activity (3 : 1973), pp. 683-730; William H. White, “Improving the Demand-for-Money Function In Moderate Inflation,” International Monetary Fund Staff Papers (September 1978), pp. 564-607. The nominal-adjustment mechanism discussed here is used in the MPS (MIT-PennSocial Science Research Council) demand deposits equation. See Jared Enzler, Lewis Johnson, and John Paulus, “Some Problems of Money Demand,” Brookings Papers on Economic Activity (1 : 1976), pp. 261-79. 27 F E D E R A L R E S E R V E B A N K O F ST. L O U IS the left-hand side, In Pt must be subtracted from both sides of equation 9: (10) In Mt - In Pt — X'ilo + X'ai In rt + X’a- In yt (1 - X') In P, + (1 - X ' ) In Mt-, + X'e ,. Equation 10 can then be rewritten in the form, (11) In mt = X’ao + X'ai In rt + X’a2 In yt + ( 1 - X ' ) In (M.-./P.) + X'e,. Thus, the only difference between the estimation of the real-adjustment specification (equation 6 ) and the nominal-adjustment specification (equation 11) is the form of the lagged dependent variable. In the real-adjustment version, lagged nominal money bal ances are deflated by lagged prices. In the nominaladjustment version, they are deflated by current prices.7 EMPIRICAL EVIDENCE Cochrane-Orcutt Results Goldfeld found little empirical difference between the coefficient estimates of the real- and nominaladjustment specifications. Based on a superior fit, both in- and out-of-sample, however, he favored the nomi nal adjustment version. Friedman, on the other hand, provides contrasting evidence which suggests that the real-adjustment version provides more stable regres sion coefficients over different sample periods.8 Tables 1 and 2 summarize the empirical evidence on the real- and nominal-adjustment specifications of 7Heller and Khan recently have questioned the applicability of the nominal-adjustment specification. See H. Robert Heller and Mohsin S. Khan, “The Demand for Money and the Term Structure of Interest Rates,” Journal of Political Economy (Februaiy 1979), pp. 109-29. The issue raised by Heller and Khan is essentially an econometric one. Specifically, estima tion of the nominal-adjustment version within a single-equa tion framework will avoid econometric problems associated with simultaneous equations bias only when the dependent variable is viewed as being determined by the exogenous var iables specified on the right-hand side of the equation. Although Heller and Khan suggest that a simultaneous equa tions bias is present when the nominal-adjustment version is estimated, this same criticism applies equally to the real-adjustment specification. Two points should be recognized with respect to the Heller-Khan criticism. First, empirical estimates of the nominal-adjustment specification traditionally define the dependent variable to be real money balances, not nominal money balances as given by equation 7. In a very important sense this variable can be viewed as demand determined — that is, determined by the price level, interest rates, and real income. Second, and perhaps more importantly, the simul taneous equation bias which results from estimating money demand relationships in a single equation framework is quite small. For a recent example of studies addressing the possi bility of simultaneous equation bias, see Goldfeld, “The De mand for Money Revisited” and “The Case of the Missing Money.” 8Benjamin Friedman, “Crowding Out or Crowding In?: Eco nomic Consequences of Financing Government Deficits,” Brookings Papers on Economic Activity (3: 1978), pp. 593641. This evidence is found in his tables, but never discussed. http://fraser.stlouisfed.org/ 28 Federal Reserve Bank of St. Louis MARCH 1980 the money demand relationship.9 The estimated co efficients for the sample period II/1955-IV /1962 are reported first, followed by estimates obtained by lengthening the sample period in increments of four quarters. Relevant summary statistics as well as the static root-mean-squared error (R M SE) for the four quarters immediately following the end of the sample period are also presented. Except for the sample pe riod II/1955-IV /1978, all regressions are estimated using the Cochrane-Orcutt ( C O R C ) serial correlation correction technique — the technique most commonly implemented to estimate money demand when quar terly data are employed.10 The regression results for the real- and nominaladjustment specifications ( tables 1 and 2, respectively) from various sample periods up to and including the II/1955-IV /1973 period are consistent with the results of previous investigations.11 In addition, the coeffi cients on the real income and interest rate variables are similar across adjustment specifications. The nom inal-adjustment specification continually produces, as Goldfeld noted, a slightly slower speed of adjustment. '■•Following Goldfeld, “The Demand for Money Revisited,” these specifications incorporate two interest rates. The com mercial paper rate (CPR) is included as a proxy for market rates of return. The commercial bank passbook rate (RTD) is included also. Banking regulations prevent this latter rate from totally moving with the market rate of return. Small investors, who do not have sufficient funds to invest in market assets, may be sensitive to the yield on passbook rates. 10The last row of table 1 presents the regression results when an alternative serial correlation adjustment procedure is used. This alternative — known as Hildreth-Lu ( HILU) — was employed because of the drastic change in the rho estimate found when adding the observations for 1978 using the CORC procedure. As seen in the table, CORC estimates of rho in crease in value as the sample period is extended. When 1978 observations are added, however, the CORC estimate of rho dropped dramatically to 0.466. The HILU results, however, suggest that the “correct” rho value for the II/1955-IV/1978 sample estimation is 0.980. This finding indicates that the Cochrane-Orcutt technique, when applied to the II/1955-IV/1978 sample, had iterated to a local rather than a global minimum of the sum-ofsquared residuals. This type of problem, although recognized in the econometrics literature, has received little attention in regard to estimating money demand functions. Interestingly enough, while the Cochrane-Orcutt estimates revealed a significant change in the coefficients once the observations for 1978 were added, this deterioration is not evident when the Hildreth-Lu estimation technique is used: The estimated co efficients on the passbook rate and income variables continue to have the anticipated sign and are statistically different from zero. In addition, the coefficient on the lagged depend ent variable is comparable to that found for earlier sample periods. None of these findings were obtained when the Cochrane-Orcutt estimation procedure was employed for the II/1955-IV/1978 sample period. For a discussion of the problems associated with the Cochrane-Orcutt technique, see J. Johnston, Econometric Methods, 2nd ed., (New York: McGraw-Hill, 1972), pp. 262-63. 11See Goldfeld, “The Case of the Missing Money;” Enzler, Johnson, and Paulus, “Some Problems;” and Friedman, “Crowding Out or Crowding In?” F E D E R A L R E S E R V E B A N K O F ST. L O U IS MARCH 1980 Table 1 Real-Adjustment Version (log-level equation) Estimation technique Sample period Coefficients (absolute value of t-statistics in parentheses) -------------------------------------------------------------------In (M t-i/ C In CPRt In RTDt In yt Pt-,) RJ Durbin-h SEE x 10“2 Rho Static 4Q RMSE x102 CORC 11/1955-IV/1962 -0.644 (2.30) 0.140 (2.75) -0.017 (4.11) -0.034 (2.39) 0.698 (7.51) 0.9494 -0.029 0.3808 0.256 0.60 CORC 11/1955-IV/1963 -0.838 (3.13) 0.169 (3.39) -0.018 (4.07) -0.038 (2.61) 0.710 (7.64) 0.9494 0.034 0.3762 0.331 0.55 CORC 11/1955-IV/1964 -0.962 (3.93) 0.184 (3.92) -0.018 (4.08) -0.040 (2.81) 0.742 (8.17) 0.9651 0.181 0.3741 0.371 0.52 CORC 11/1955-IV/1965 -0.954 (4.37) 0.186 (4.28) -0.019 (4.29) -0.041 (3.08) 0.716 (8.25) 0.9774 0.365 0.3786 0.303 1.00 CORC 11/1955-IV/1966 -0.755 (3.40) 0.154 (3.44) -0.021 (4.08) -0.032 (2.32) 0.724 (7.59) 0.9818 0.614 0.4058 0.440 0.42 CORC 11/1955-IV/1967 -0.842 (4.43) 0.170 (4.34) -0.021 (4.48) -0.037 (2.95) 0.708 (8.07) 0.9866 0.569 0.3994 0.406 0.44 CORC 11/1955-IV/1968 -0.893 (4.90) 0.179 (4.72) -0.021 (4.49) -0.039 (3.27) 0.703 (8.27) 0.9905 0.748 0.3956 0.421 0.41 CORC 11/1955-IV/1969 -0.905 (5.04) 0.181 (4.88) -0.023 (5.09) -0.040 (3.36) 0.698 (8.53) 0.9925 0.811 0.3943 0.422 0.31 CORC 11/1955-IV/1970 -0.916 (5.23) 0.184 (5.09) -0.022 (5.13) -0.040 (3.53) 0.688 (8.71) 0.9933 0.932 0.3875 0.422 0.68 CORC 11/1955-IV/1971 -0.916 (4.89) 0.179 (4.84) -0.017 (4.22) -0.040 (3.38) 0.661 (8.12) 0.9933 1.381 0.4074 0.425 0.30 CORC 11/1955-IV/1972 -0.861 (5.18) 0.177 (5.04) -0.016 (4.57) -0.040 (3.51) 0.665 (8.36) 0.9945 1.321 0.4006 0.440 0.51 CORC 11/1955-IV/1973 -0.867 (5.24) 0.180 (5.16) -0.016 (4.70) -0.040 (3.57) 0.649 (8.28) 0.9952 1.460 0.4049 0.440 1.61 CORC 11/1955-IV/1974 -0.703 (3.90) 0.145 (3.87) -0.022 (4.88) -0.032 (2.48) 0.732 (8.54) 0.9940 1.125 0.4596 0.596 1.90 CORC I1/1955-IV/1975 -0.716 (3.69) 0.155 (4.30) -0.014 (3.01) -0.048 (2.70) 0.678 (7.76) 0.9923 0.305 0.5051 0.871 0.65 CORC II/ 1 9 5 5 - IV / 1 9 7 6 -0.702 (3.52) 0.156 (4.45) -0.013 (2.95) -0.050 (2.74) 0.652 (7.70) 0.9922 -0.278 0.4984 0.910 0.62 CORC I I / 1955-IV/1977 -0.688 (3.41) 0.156 (4.50) -0.013 (2.88) -0.050 (2.72) 0.632 (7.51) 0.9912 -0.397 0.4970 0.925 1.17 HILU (-.9, .9, .01) 11/1955-IV/1978 -0.919 (2.43) 0.190 (3.47) -0.014 (3.15) -0.045 (2.33) 0.583 (6.78) 0.9912 -0.386 0.5096 0.980 This specification also yields smaller standard errors of the estimating equation (S E E ) and superior post sample predictions in terms of the static RMSEs. The regression results for sample periods with end points beyond IV/1973, however, provide a different picture. As the end point is advanced, all estimated coefficients for the real-adjustment specification con tinue to have their anticipated sign, are different from zero at traditional levels of significance, and show rel atively small changes in magnitude. There is, however, a significant change in the estimated value of the serial correlation coefficient (0.44 for the period end ing IV /1973 to 0.98 for the period ending IV /1978). While the change in the rho value has no economic significance, it suggests a misspecification problem. Moreover, the results indicate a marked deterioration in both the in- and out-of-sample fit. Altogether, the SEE increases 26 percent and, for the years 1974 and 1975, the RMSE is more than triple that observed for 1973. The increase in the RMSE, however, does not tell the whole story. A major problem with the post 29 MARCH F E D E R A L R E S E R V E B A N K O F ST. L O U IS 1980 Table 2 Nominal-Adjustment Version (Cochrane-Orcutt estimation of log-level equation) Coefficients (absolute value of t-statistics in parentheses) Sample period C In yt In CPR, In RTD, In (M .-./P ,) R2 D.W. SEE x 10 2 Rho Static 4Q RMSE x 10-’ 11/1955-IV/1962 -0.625 (2.20) 0.128 (2.54) -0.017 (3.76) -0.032 (2.35) 0.780 (7.11) 0.9584 1.81 0.3457 0.4922 0.50 11/1955-IV/1963 -0.777 (3.03) 0.148 (3.17) -0.018 (3.98) -0.033 (2.52) 0.811 (7.73) 0.9608 1.83 0.3314 0.5456 0.40 11/1955-IV/1964 -0.841 (3.62) 0.156 (3.53) -0.018 (4.17) -0.035 (2.69) 0.825 (8.14) 0.9723 1.85 0.3337 0.5231 0.38 11/1955-IV/1965 -0.836 (4.10) 0.157 (3.84) -0.018 (4.30) -0.035 (2.90) 0.816 (9.01) 0.9827 1.75 0.3316 0.4917 0.91 11/1955-IV/1966 -0.699 (3.43) 0.136 (3.29) -0.020 (4.14) -0.030 (2.35) 0.808 (8.22) 0.9852 1.65 0.3670 0.5559 0.56 11/1955-IV/1967 -0.810 (4.36) 0.156 (4.02) -0.020 (4.35) -0.035 (2.93) 0.787 (8.28) 0.9882 1.70 0.3746 0.5067 0.50 11/1955-IV/1968 -0.866 (4.87) 0.164 (4.38) -0.021 (4.39) -0.038 (3.24) 0.794 (8.54) 0.9917 1.70 0.3707 0.5307 0.33 11/1955-IV/1969 -0.861 (5.04) 0.164 (4.56) -0.022 (5.00) -0.038 (3.34) 0.792 (9.11) 0.9937 1.69 0.3625 0.5270 0.33 11/1955-IV/1970 -0.855 (5.14) 0.163 (4.65) -0.021 (5.06) -0.037 (3.36) 0.793 (9.42) 0.9944 1.67 0.3534 0.5285 0.72 11/1955-IV/1971 -0.839 (5.05) 0.166 (4.72) -0.016 (4.05) -0.038 (3.42) 0.737 (8.67) 0.9942 1.56 0.3769 0.5340 0.18 11/1955-IV/1972 -0.833 (5.42) 0.164 (4.98) -0.016 (4.57) -0.038 (3.57) 0.743 (9.17) 0.9954 1.69 0.3686 0.5321 0.22 11/1955-IV/1973 -0.809 (5.45) 0.160 (5.01) -0.016 (4.71) -0.037 (3.56) 0.748 (9.60) 0.9961 1.72 0.3644 0.5145 0.66 11/1955-IV/1974 -0.656 (5.15) 0.125 (4.64) -0.017 (5.69) -0.027 (2.97) 0.840 (13.22) 0.9961 1.78 0.3679 0.4967 1.56 11/1955-IV/1975 -0.327 (3.40) 0.054 (2.72) -0.015 (4.68) -0.008 (0.99) 1.003 (21.71) 0.9948 1.79 0.4161 0.4242 0.69 11/1955-IV/1976 -0.233 (3.32) 0.034 (2.40) -0.014 (4.58) -0.002 (0.26) 1.045 (29.39) 0.9945 1.88 0.4183 0.3990 0.16 11/1955-IV/1977 -0.227 (3.97) 0.033 (2.91) -0.014 (4.67) -0.001 (0.23) 1.047 (34.02) 0.9945 1.89 0.4102 0.3929 0.38 11/1955-IV/1978 -0.233 (4.55) 0.034 (3.49) -0.014 (4.98) -0.002 (0.29) 1.046 (38.93) 0.9943 1.88 0.4093 0.3886 sample performance of the real-adjustment equation is that the specification consistently overpredicts money demand. Table 3 provides the mean forecast error for nominal money balances based on both the real- and the nominal-adjustment specifications. The real-adjustment specification, on average, overpredicts money demand for each year following 1973. While the apparent stability of the estimated coefficients pro vides some ad hoc evidence for the belief that the underlying economic relationship for the real-adjust ment specification is stable, the changes in both the rho estimate and the in- and out-of-sample fit question such a conclusion. 30 In contrast, the results for the nominal-adjustment version over the post-1973 era (table 2 ) indicate a marked deterioration in the estimated regression co efficients. This is somewhat surprising since the only difference between these two specifications is whether lagged money is deflated by lagged or contempora neous prices. Interestingly enough, the most trouble some result over this period is the increase in the size of the coefficient on the lagged dependent variable. The coefficient exceeds unity in the longer sample pe riods, a finding that alone obviates any meaningful interpretation of the estimates within the stock-adjustment framework. Based on the dramatic change in the F E D E R A L R E S E R V E B A N K O F ST. L O U IS MARCH 1980 Table 3 Table 4 Mean Static Prediction Error* (billions of nominal money balances) F-statistics for Null Hypothesis that Regression Coefficients Are Equal in Two Sample Periods (CORC) Prediction interval Real-adjustment (Cochrane-Orcutt estimation) Nominal-adjustment (Cochrane-Orcutt estimation) 1/1974-IV/1974 -$2.85 -$1.00 11/1955-IV/1962 vs. 1/1963-IV/1978 5.72* 1.99 I/1975-IV/1975 - 3.33 - 3.01 II/1955-IW1967 vs. 1/1968-IV/1978 5.60* 3.00** 11/1955-IV/1973 vs. I/1974-IV/1978 8.33* 5.31* 1/1976-IV/1976 - 0.80 - 1.28 1/1977-IV/1977 - 0.55 - 0.11 1/1978-IV/1978 - 1.83 + 0.11 “Error is calculated as actual nominal money stock less pre dicted nominal money stock. A negative error thus indicates overprediction. estimated regression coefficients for the nominaladjustment specification, it appears that this economic relationship has indeed broken down. Thus, even though the nominal-adjustment specification continues to have both a better in-sample and out-of-sample fit (see table 3 ), it is difficult to attach any significance to this in light of the coefficient estimates for the post1973 period. Chow tests were employed to ascertain whether either relationship is statistically stable over the full sample period. Three alternative break points were examined: (1 ) IV/1962, a point near which Slovin and Sushka found evidence of a shift in the money de mand relationship;12 (2 ) IV /1967, a point near the middle of the sample period; and (3 ) IV/1973, a point of recent concern and considerable testing. The calcu lated F-statistics are reported in table 4 .13 With the exception of the hypothesized IV /1962 break point for the nominal-adjustment equation, the regression coefficients are all statistically different for the oppos ing sample periods. The finding that the real-adjustment specification is unstable over these sample pe riods contrasts sharply with the apparent stability of 12Myron B. Slovin and Marie Elizabeth Sushka, “The Struc tural Shift in the Demand For Money,” The Journal of Finance (June 1975), pp. 721-31. 13The applicability of the Chow test is complicated in this case by the existence of serial correlation in the disturbances. The F-statistics in table 3 were calculated by estimating the serial coefficient in each alternative sample period separately, using the Cochrane-Orcutt technique. An alternative, “seem ingly unrelated,” procedure was also used (see Franklin M. Fisher, “Test of Equality Between Sets of Coefficients in Two Linear Regressions: An Expository Note,” Econometrica (March 1970), pp. 361-66). This latter procedure constrained the serial correlation coefficient to be the same in each of the respective sample periods. The results for this test did not differ significantly from those reported. Sample periods RealNominaladjustment adjustment * Significant at the 1% level. | * i.u to, , , /Degrees of freedom — 5,84 Significant at the 5% level. ) the regression coefficients in table 1. These results indi cate that cursory examinations of the stability of the regression coefficients can be misleading. First-Difference Results As an alternative to the estimation performed above, both money demand specifications were esti mated in first-difference form using the ordinary leastsquares technique. In other words, instead of estimat ing an equation of the form, (12) In mt = b0 + bj In rt + b2 In yt + k In mt-i + £t; (e, = pet-1 + n<) with the Cochrane-Orcutt technique, the following equation was estimated using ordinary least-squares: (13) In mt - In mt-i = (bo - b0) + bi ( In rt - In rt-i) + b2 (In yt - In yt_i) + h> (In mt-i - In mt-2) + T)t, where the error terms, qt, are assumed to be indepen dent and identically distributed N (0, a 2). The dif ference between these alternative specifications lies in the a priori assumption about the error structures. These two specifications would be empirically equiva lent if rho ( p) were restricted to unity in equation 12. Although equation 12 is more general than equa tion 13, estimation of the latter equation avoids an important econometric problem associated with the estimation of equation 12. Specifically, Theil has shown that, in the presence of a lagged dependent variable, estimation techniques, such as the CochraneOrcutt procedure, will underestimate (in absolute value) the serial coefficient rho.14 This error will ren der the estimated regression coefficients inconsistent and inefficient. If the disturbances in equation 13 are serially independent — a hypothesis that can be exam14Henri Theil, Principles of Econometrics (New York: John Wiley and Sons, 1971), pp. 413-14. 31 F E D E R A L R E S E R V E B A N K O F ST. LO U IS MARCH 1980 Table 5 Real-Adjustment Version: Log Differences (ordinary least squares) Coefficients (absolute value of t-statistics in parentheses) Static ------------------------------------------------------------------------------------------------------------ Iny,In yt-i In CPRt In CPRj.! In RTDt In RTD,-i In M .V P t-iIn Mt-2/P t -2 4Q R! Durbin-h SEE x 10-2 0.536 (3.21) 0.459 -2.94 0.4598 0.32 -0.047 (2.54) 0.571 (3.85) 0.524 -2.05 0.4421 0.35 -0.015 (2.75) -0.050 (2.86) 0.639 (4.85) 0.581 -1.80 0.4295 0.60 0.146 (1.76) -0.015 (2.66) -0.051 (2.81) 0.592 (4.41) 0.539 -1.44 0.4493 0.53 0.0009 .(0.09) 0.151 (1.81) -0.016 (2.85) -0.044 (2.44) 0.590 (4.65) 0.523 -1.06 0.4547 0.55 11/1955-IV/1967 0.0004 (0.40) 0.160 (1.93) -0.018 (3.17) -0.048 (2.64) 0.576 (4.91) 0.530 -0.72 0.4623 0.38 11/1955-IV/1968 0.0004 (0.46) 0.175 (2.19) -0.018 (3.30) -0.050 (2.79) 0.574 (5.24) 0.540 -0.60 0.4557 0.48 11/1955-IV/1969 0.0001 (0.14) 0.194 (2.49) -0.020 (3.72) -0.047 (2.69) 0.571 (5.40) 0.545 -0.61 0.4567 0.31 11/1955-IV/1970 0.00002 (0.03) 0.206 (2.80) -0.019 (3.75) -0.045 (2.69) 0.544 (5.37) 0.541 -0.38 0.4478 0.75 11/1955-IV/1971 0.0003 (0.37) 0.157 (2.20) -0.013 (2.80) -0.049 (2.85) 0.529 (5.16) 0.492 0.60 0.4665 0.40 11/1955-IV/1972 0.0004 (0.53) 0.166 (2.43) -0.013 (2.88) -0.052 (3.10) 0.552 (5.61) 0.527 0.28 0.4614 0.53 11/1955-IV/1973 0.0002 (0.23) 0.171 (2.56) -0.013 (2.90) -0.051 (3.00) 0.548 (5.66) 0.528 0.30 0.4650 0.84 11/1955-IV/1974 -0.0005 (0.64) 0.208 (3.04) -0.015 (3.46) -0.045 (2.56) 0.609 (6.43) 0.587 -1.05 0.4872 0.78 11/1955-IV/1975 -0.001 (1.30) 0.252 (3.93) -0.014 (3.14) -0.044 (2.44) 0.567 (6.40) 0.583 -0.53 0.5041 0.28 11/1955-IV/1976 -0.001 (1.33) 0.253 (4.14) -0.014 (3.21) -0.044 (2.50) 0.564 (6.57) 0.571 -0.97 0.4955 0.46 11/1955-IV/1977 -0.001 (1.31) 0.253 (4.20) -0.013 (3.11) -0.045 (2.56) 0.555 (6.54) 0.571 -1.01 0.4938 0.71 11/1955-IV/1978 -0.001 (1.36) 0.237 (3.96) -0.014 (3.31) -0.042 (2.40) 0.562 (6.63) 0.542 -1.28 0.5048 ---- Period Constant 11/1955-IV/1962 -0.0001 (0.084) 0.147 (1.59) -0.014 (2.31) -0.042 (2.15) II/1955-IW1963 0.0003 (0.22) 0.156 (1.76) -0.014 (2.55) 11/1955-IV/1964 0.0007 (0.73) 0.138 (1.66) 11/1955-IV/1965 0.0001 (0.64) 11/1955-IV/1966 RMSE X10-3 ined empirically — its estimation will avoid the prob lem associated with the Cochrane-Orcutt technique.15 Thus, in this very important sense, estimation of equa tion 13 is preferable. Both the real- and nominal-adjustment versions of the money demand relationship were estimated in first-difference form. The respective findings are re ported in tables 5 and 6.16 15First-di£Ferencing has been suggested as a means of avoiding the econometric problems associated with nonstationary error structures. See C. W. J. Granger and P. Newbold, “Spurious Regressions in Econometrics,” Journal of Econometrics ( June 1974), pp. 111-20; and D. Williams, “Estimating in Levels or First Differences: A Defense of the Method Used for Certain Demand-for-Money Equations,” The Economic Journal ( Sep tember 1978), pp. 564-68. As an additional important matter, Charles I. Plosser and G. William Schwert, “Money, Income, and Sunspots: Meas- Consider first the results for the real-adjustment uring Economic Relationships and the Effects of Differenc ing,’ Journal of Monetary Economics (4: 1978), pp. 637-60, show that the econometric problems of “overdifferencing” an equation are not as severe as those of “underdifferencing.” 16A constant term was included in the specification to ascer tain whether a time trend is evident in money demand. If there is a trend in money demand, the trend rate of change MARCH F E D E R A L R E S E R V E B A N K O F ST. L O U IS 1980 Table 6 Nominal-Adjustment Version: Log Differences (ordinary least squares) D.W. SEE x 10'2 Static 4Q RMSE x10a 0.613 2.15 0.3896 0.21 0.736 (5.86) 0.669 2.15 0.3688 0.41 -0.031 (1.99) 0.730 (6.49) 0.683 2.13 0.3736 0.41 -0.019 (3.90) -0.031 (1.97) 0.742 (6.60) 0.675 2.03 0.3772 0.61 0.184 (2.62) -0.019 (3.80) -0.027 (1.65) 0.701 (6.27) 0.627 1.95 0.4020 0.61 -0.0003 (0.28) 0.193 (2.64) -0.020 (3.90) -0.032 (1.87) 0.683 (6.15) 0.606 1.94 0.4230 0.30 11/1955-IV/1968 -0.0001 (0.17) 0.198 (2.80) -0.020 (4.06) -0.033 (1.99) 0.697 (6.62) 0.620 1.95 0.4141 0.32 11/1955-IV/1969 -0.0004 (0.46) 0.209 (3.10) -0.022 (4.55) -0.031 (1.94) 0.699 (7.11) 0.638 1.95 0.4072 0.17 11/1955-IV/1970 -0.0003 (0.46) 0.205 (3.23) -0.021 (4.71) -0.029 (1.88) 0.692 (7.31) 0.642 1.92 0.3953 0.73 11/1955-IV/1971 -0.0002 (0.22) 0.166 (2.65) -0.016 (3.70) -0.033 (2.06) 0.658 (6.94) 0.591 1.78 0.4185 0.38 11/1955-IV/1972 -0.0001 (0-14) 0.184 (3.10) -0.017 (4.03) -0.035 (2.24) 0.665 (7.38) 0.617 1.94 0.4151 0.38 11/1955-IV/1973 -0.0002 (0.29) 0.183 (3.16) -0.016 (4.05) -0.032 (2.11) 0.669 (7.71) 0.628 1.99 0.4127 0.48 11/1955-IV/1974 -0.001 (0.78) 0.194 (3.45) -0.017 (4.69) -0.027 (1.81) 0.728 (9.21) 0.702 2.12 0.4142 0.81 11/1955-IV/1975 -0.001 (1.67) 0.232 (4.20) -0.015 (3.88) -0.025 (1.59) 0.709 (8.87) 0.683 2.02 0.4394 0.43 11/1955-IV/1976 -0.001 (1.63) 0.230 (4.28) -0.015 (3.88) -0.026 (1.67) 0.700 (8.85) 0.672 2.10 0.4387 0.25 11/1955-IV/1977 -0.001 (1.62) 0.226 (4.33) -0.014 (3.90) -0.026 (1.68) 0.706 (9.10) 0.672 2.12 0.4317 0.45 11/1955-IV/1978 -0.001 (1.71) 0.221 (4.36) -0.016 (4.24) -0.024 (1.55) 0.717 (9.61) 0.665 2.10 0.4319 ---- Coefficients (absolute value of t-statistics in parentheses) In M t-i/P, In Mt-j/Pt-i In CPR, In CPR,- ! In RTD, In RTD,--i 0.170 (2.26) -0.018 (3.44) -0.027 (1.61) 0.702 (4.96) -0.0002 (0.27) 0.175 (2.47) -0.018 (3.78) -0.029 (1.89) 11/1955-IV/1964 -0.0001 (0.09) 0.171 (2.46) -0.018 (3.79) 11/1955-IV/1965 -0.0001 (-0.09) 0.169 (2.54) 11/1955-IV/1966 -0.0006 (0.64) 11/1955-IV/1967 Period Constant 11/1955-IV/1962 -0.0005 (0.51) 11/1955-IV/1963 Iny, In y,-i specification in table 5. The Durbin-h statistics indi cate that, for sample periods ending beyond IV/1964, there is no evidence to reject the hypothesis of serially independent error terms: It is only in the earlier sam ple periods that evidence of first-order autocorrelation exists. The regression coefficients found in table 5, like should be equal to the constant term (see equation 13). Lieberman suggests that such a variable is relevant for money demand. See Charles Lieberman, “Structural and Technolog ical Change in Money Demand,” American Economic Review, Papers and Proceedings (May 1979), pp. 324-29. Rs those reported in table 1, indicate a remarkable de gree of consistency as the sample period is extended.17 17In comparing the results in table 5 with those in table 1, one should be cautioned against using the reported R2 as a basis to judge the respective equations. Granger and Newbold, “Spurious Regressions,” show that when the dependent and independent variables follow a random walk, as they do in our specification, a nonzero R2 will be expected, even if no relationship between the variables actually exists. When the equation is estimated in first-difference form, the variables no longer follow a random walk and thus R! is expected to be zero. Furthermore, the reader is cautioned against solely using the SEE as a basis of comparison. Recall the aforementioned econometric problems associated with the Cochrane-Orcutt estimation results. 33 F E D E R A L R E S E R V E B A N K O F ST. L O U IS Table 7 Mean Static Prediction Error* (billions of nominal money balances) Prediction interval Real-adjustment (First-difference estimation) Nominal-adjustment (First-difference estimation) I/1974-IV/1974 -$1.65 -$0.82 1/1975-IV/1975 - 0.91 - 1.40 I/1976-IV/1976 + 0.10 + 0.14 1/1977-IV/1977 + 0.17 + 0.10 1/1978-IV/1978 - 1.24 - 0.74 "Error is calculated as actual nominal money stock less pre dicted nominal money stock. A negative error thus indicates overprediction. With the single exception of the coefficient on the passbook rate for the sample period ending in IV / 1971, the estimated coefficients all change by less than one standard error. The regression coefficients in table 5 are also simi lar to those found in table 1 in other respects.18 The coefficient on the lagged dependent variable indicates a significant partial adjustment to the desired level of real money balances. The relatively smaller coeffi cient in table 5 indicates, however, a larger coefficient of adjustment — ranging from 0.39 to 0.47. The find ings in table 5 again support the notion of economies to scale in money holdings, with the long-run income elasticity estimated between 0.33 and 0.53. In addi tion, the coefficients on the interest rate variables continue to indicate a greater sensitivity to a propor tional change in the passbook rate than the commer cial paper rate. An important improvement obtained from the firstdifference estimation procedure over the levels results is the post-sample performance. Table 5 indicates a deterioration in post-sample performance over the 1974-75 period, but the deterioration is slight relative to that found in table 1. Not only are the RMSEs consistently lower for the first-difference results, but this specification does not consistently overpredict money demand over the post-1974 period. In fact, 18As stated previously, the estimated constant term has no counterpart in the levels form of the real-adjustment specifi cation. This coefficient, while never significantly different from zero in table 5, does change as the sample period is extended to include the 1974 observation. When post-1974 observations are included in the sample, both the sign and magnitude of this coefficient are in accord with Lieberman’s findings. This suggests a slight, but statistically insignificant negative drift in the relationship, which is unexplained by other variables. 34 MARCH 1980 table 7 indicates that this specification slightly underpredicts money demand on average for 1976 and 1977.19 In contrast to the first-difference estimation of the real-adjustment mechanism, table 6 shows that there is no evidence in the nominal-adjustment specification of any first-order serial correlation in the disturbances. The Durbin-Watson statistics reveal no problems associated with serial dependence in the errors. Many of the previous comparisons between the real- and nominal-adjustment levels estimations (tables 1 and 2 ) continue to hold for the first-difference estimations. The coefficient of adjustment remains smaller for the nominal-adjustment specification than for the real-adjustment specification in table 5. The other regression coefficients continue to be fairly simi lar to those found for the real-adjustment equation. The coefficient on the passbook rate for the nominaladjustment specification is not, however, significantly different from zero over many of the sample periods. Again the SEEs and the RMSEs are consistently smaller for the nominal-adjustment specification, indi cating a better in- and out-of-sample fit. Table 7 also shows that using the first-difference of the nominaladjustment specification not only leads to a smaller forecasting error on average, but more importantly, alleviates the persistent problem of overprediction which plagued the levels estimation.20 Unlike those observed in table 2, the regression coefficients in table 6 do not change dramatically as the sample period is extended beyond IV /1973.21 T h e 19It should also be noted that the mean overprediction that takes place in 1978 is in large part due to overpredicting money demand in the fourth quarter of that year, when ATS accounts were legalized nationwide and NOW accounts were legalized in New York. 20To determine if the inclusion of the constant term seriously biases the post-sample performances of the equations, fore casts based on equations that exclude the constant term were made. For the real-adjustment equation, the most significant effect is to change the sign on the mean static prediction for the interval I/1976-IV/1976 from plus to minus. The mean error, however, for the period is -$0.02 billion, quite small relative to that reported in table 7. For the nominal-adjust ment specification, the positive signs for I/1976-IV/1976 and I/1977-IV/1977 are changed to negative when the constant term is omitted from the forecasting equation. As with the real-adjustment equation, however, the mean errors are very small relative to those reported in table 7: -$0,002 billion for I/1976-IV/1976 and -$0.07 billion for I/1977-IV/1977. These results suggest that it is first-differencing, rather than the inclusion of the trend variable in the specification, that is most responsible for the improved forecasting accuracy of these specifications. 21The possible exception to this is the behavior of the esti mated constant term. This coefficient estimate increases (in absolute value terms) five-fold as the 1974 observations are added to the sample period. While the change in this coeffi cient is noticeable, it is important to bear in mind that this coefficient estimate is never significantly different from zero. F E D E R A L R E S E R V E B A N K O F ST. LO U IS changes that occur, instead, are relatively minor. This is especially true of the coefficients on the real income and lagged dependent variables. Thus, the use of firstdifferences apparently has resulted in a more stable relationship. Chow tests again were used to determine whether either of these first-difference relationships is statis tically stable over the full-sample period. The F-statistics for the same hypothesized break points con sidered previously ( see table 4 ) are provided in table 8.22 These statistics indicate that neither of the specifi cations is statistically different over any of the alter native subperiods considered. This suggests that the previous evidence of breakdowns in these relation ships is the result of the estimation technique em ployed. The first-difference estimation results, which are econometrically preferable, show no evidence of structural breakdown in either of the money-demand specifications considered. SUMMARY AND CONCLUSION This paper has investigated two alternative stockadjustment mechanisms employed to empirically ex plain money demand. In addition, two alternative procedures have been used to estimate these rela tionships. The results indicate that both stock-adjustment relationships are statistically stable when esti22Since the disturbances for the first-difference equation are serially independent, the Chow test results reported in table 8 avoid the problem of serial dependence in the error terms that plagued the previous tests. In an alternative test, the first-difference equations were estimated without a constant term and the Chow test was used to test the stability of these equations. Using the same hypothesized break-points as in table 8, the test results indi cate that stability cannot be rejected at the 5 percent level of significance (e.g., the largest calculated F-value is 0.69). Thus, exclusion of the constant term does not adversely af fect the stability finding. In addition, the type of Chow test described in footnote 13 was specifically used to test if the constant term should be allowed to vary across the various subperiods. For each equation and the different subperiods, the calculated F-statistics were well below standard critical values. Thus, no statistical advantage is gained by allowing the constant term to take on different values in alternative subperiods. MARCH 1980 Table 8 F-statistics for Null Hypothesis that Regressions Are Equal in Two Alternative Sample Periods (first-difference) Sample periods RealNominaladjustment adjustment 11/1955-IV/1962 vs. 1/1963-IV/1978 0.47 11/1955-IV/1967 vs. 1/1968-IV/1978 1.11 1.02 11/1955-IV/1973 vs. I/1974-IV/1978 1.42 1.05 0.44 1% critical value, 3.21 5% critical value, 2.33 mated in first-difference form. This suggests that much of the recent evidence of a breakdown in the money-demand relationship is the result of the esti mation technique employed. To the extent that the nominal-adjustment specification consistently provided a better fit (both in- and out-of-sample), the evidence presented here further suggests that a relaxation of the assumption that the money stock adjusts to a price level shock within the quarter is worthwhile. Furthermore, the results presented in this paper deny the claim that monetary policy is impotent as a result of a shifting money-demand relationship. Those who argue this point recently have suggested that attempts to control inflation through restrictive monetary policy will be unsuccessful since the money demand relationship is unstable. The findings of sta bility presented here seriously question this assertion. It does not appear that the relationship between money demand, real-income, and interest rates has changed significantly over recent periods. The sur prisingly accurate predictions of money demand over the post-1973 period using the first-difference approach buttress the conclusion that the money-demand rela tionship has not suffered from any drastic shifts that would invalidate monetary policy. 35