After Greenspan replaced Volcker, actions became more systematic at least as to choice of target. Greenspan’s FOMC used a federal funds rate target after late 1989, but the choice depended more heavily on Greenspan’s analysis of events than on any systematic economic model. The staff continued to use a Phillips curve to forecast inflation, but both chairmen did not.
The Federal Reserve in the 1980s and beyond accepted and relied on some monetarist propositions. It no longer denied that it was responsible for control of inflation; some accepted that money growth in excess of output growth caused inflation, and it asserted that low inflation was important for getting low and stable unemployment. Also it accepted that high interest rates reflected high expected inflation. It continued to insist that budget deficits caused inflation (memo, Volcker, Board Records, April 30, 1984, 2) and it did not develop successful procedures for controlling money growth.
Volcker expressed doubt about forecast accuracy. He relied on actual data, neglecting to say that revisions reduced data accuracy. At one point, November 1984, he recognized that imprecise forecasts made “fine tuning” impossible. But he did not take the next step—setting a medium-term target. Probably he did not accept that medium-term projections were reliable. The FOMC continued to respond to noisy current data.
The Federal Reserve staff continued to explain many of its forecast errors by claiming that the demand for money shifted. The staff, most members of the FOMC, and large parts of the academic and financial market community concluded that instability of the demand for money made control of money infeasible and undesirable. In 1988, the Federal Reserve staff reported on their research concerning monetary base control. The FOMC concluded that monetary base velocity fluctuated unpredictably in the short-run and required large changes in interest rates
At the time, Robert Rasche (1988) estimated the equations for velocity shown in the appendix to this chapter. His results, using monthly data for the St. Louis base (Appendix Table 1) and the Board’s version (Appendix Table 2) show that the response of base velocity growth to changes in personal income, prices, and interest rates changed very little in the 1980s from experience from the 1950s through the 1970s. The principal and only important change was a decline in trend growth from 2.5 percent a year to zero.
In 1986, the staff reported on its ability to forecast money growth. The results showed that average absolute quarterly errors were large, almost two percentage points at annual rates in 1985–86. Annual forecasts for 1974 to 1986 had an average absolute error of 0.8 percent. Years of policy change have large errors. Table 9.6 shows the forecasts and errors. The data suggest that control of inflation by controlling money growth is feasible if the Federal Reserve decided to pursue a medium-term strategy. And control of the monetary base requires the Federal Reserve to control only the size of its balance sheet. It can continue to set interest rate targets, but it must adjust the target at least quarterly to respond to the base.
It is impossible to date the end of the Great Inflation precisely. I have used 1986. The years 1983 to 1986 saw the unwinding of some main effects of the Great Inflation. Interest rate ceilings ended, and the preferred composition of portfolios changed. One main result was uncertainty about the reasons for growth in monetary aggregates, particularly M 1 . Another was bankruptcy and merger of a large part of the thrift industry.
The anti-inflation policy and the enlarged defense spending that increased aggregate demand caused an appreciation of the dollar against most or all foreign currencies not pegged to the dollar. Pressure from foreigners, domestic exporters, and members of Congress brought a change in policy. By March 1985, the dollar began to depreciate. The new Treasury secretary, James Baker, wanted to demonstrate concern for the rising current account deficit and the strong dollar. He obtained agreement from Germany and Japan in September 1985 on coordinated efforts to depreciate the dollar. The agreement did not specify the policy changes to be made on either side, and it did not decide on appropriate exchange rates or policies. Most importantly, the agreement did not distinguish between real and nominal exchange rates or between sterilized and unsterilized intervention. The dollar depreciated in real and nominal value against the yen. Both the Bundesbank and the Federal Reserve sterilized most or all of their intervention, but the Federal Reserve increased base growth. Faster base growth contributed to depreciation. There was little effect on the current account deficit.
In retrospect, the action appears motivated more by politics than by consistent economic policy. It succeeded in preventing Congress from considering trade restrictions at the time. The dollar depreciated in real terms by the end of the decade.
Perhaps the most enduring lesson for central bankers from the Great Inflation and subsequent disinflation was that the responsibility for stopping inflation fell on them. Although Volcker and others pleaded frequently for reducing government spending and smaller deficits, their pleas did not produce the assistance they wanted. The experience of disinflation was painful for the economy and for them. They learned to avoid a repetition. The weights they gave to inflation and unemployment did not shift back to their earlier values and sustained high inflation did not return in the next two decades.
Experience in the 1970s convinced many that inflation was costly and undesirable. Disinflation in 1979–82 had public support that persisted as a desire for low inflation. The fact that the economy expanded in the 1980s and 1990s, that expansions were long and recessions mild, sustained the policy’s popular support. Unemployment and interest rates eventually returned to the range experienced before the Great Inflation. Pressures to expand more or lower interest rates are always present, but the pressures were muted. The Federal Reserve learned that its policies could count on public, congressional, and administration support for low inflation by providing reasonably stable growth and low inflation. This support remained until the credit crisis that started in 2007.
APPENDIX TO CHAPTER 9
Recent Behavior of Monetary Base Velocity58
Robert H. Rasche
Department of Economics
Michigan State University
East Lansing, Michigan 48824-1038
At the last meeting of the Shadow Committee I reported on research then underway concerning demand functions for the monetary base. In the interim, the staff of the Board of Governors has investigated the question of using the monetary base as target for monetary policy. A summary of that research is published as an appendix to the July, 1988 Monetary Policy Report to Congress [Federal Reserve Bulletin, August, 1988, pp. 530–33].
58. Prepared for meeting of the Shadow Open Market Committee in 1989.
Apparently on the basis of this latter research, the FOMC has dismissed the possibility of any role for the monetary base in the implementation of monetary policy at the present time. “The Committee decided against establishing a range for the monetary base because it seemed unlikely to provide a more reliable guide for policy than the aggregates for which ranges already are established. Although the base has been less variable in relation to economic activity than M1, its velocity nonetheless has fluctuated appreciably and rather unpredictably from year to year.” [Monetary Policy Report to Congress, July 13, 1988; Federal Reserve Bulletin, August, 1988, p. 519].
Unfortunately the staff research that is the basis for this conclusion is classified FOMC material at the present time, so we apparently have to wait at least five years before there is an opportunity to review the studies in detail. At the present time all that is available is the published appendix to the monetary policy report. Approximately 1/3 of that appendix is devoted to describing the differences between the Board measure of the monetary base and the St. Louis Federal Reserve Bank measure of the monetary base (Adjusted Monetary Base). The remainder of the appendix states four major conclusions:
1) [Statistical] technique that allow for a break in behavior (of base velocity) in the early 1980s . . . make somewhat smaller but still large errors in the 1980s and leave unanswe
red questions about the potential for additional shifts in the relationships.
2) The demand for the base has substantial interest sensitivity. . . . The base probably is less interest sensitive than are the other monetary aggregates.
3) Over long periods of time, the demand for the base appears to be fairly predictable, especially compared with M1-A and M1.
4) It is likely that the base, or for that matter any of the broader aggregates, could be controlled reasonably well over a span of several quarters—a period that would be meaningful in terms of the effects of monetary policy. However, the degree of interest rate volatility under base targeting could be quite substantial, especially in the short-tointermediate run.
The remainder of this report will examine these major conclusions, particularly in light of our own research into the demand for the monetary base. The data presented here use both the Board and St. Louis Federal Reserve Bank monetary base concepts, and personal income. Since personal income is available on a monthly basis, this gives a substantial number of observations during the controversial 1980s period. Results available elsewhere [Rasche, 1988] suggest that the conclusions drawn from these data are consistent with those derived from other measures of aggregate economic activity such as GNP or final sales to domestic purchasers, and with other levels of time aggregation.
Any analysis of the demand for the monetary base, or monetary base velocity, has to recognize that while the experience of 1980s is not identical to that of the previous three decades, the similarities far exceed the differences. Emphasizing the similarities is more productive than emphasizing the differences. The primary lesson from the 50s–70s is that base velocity behaves like a random walk. That characterization of base velocity, and its implication for monetary policy, has been discussed many times by this committee. That fundamental property of base velocity has not changed in the 1980s.
Past changes of base velocity are of little, if any use, in predicting future changes in base velocity. The only significant difference in the behavior of base velocity between the 1980s and the previous three decades is the average month-to-month percentage change, or drift. Through 1981 the drift in the random walk of velocity was around 2.5 percent at annual rates; during the 1980s it is zero. After allowing for this break in the drift of base velocity, there is no evidence of increased variability in the 1980s compared with the previous experience.
Thus the first of the conclusions cited above is somewhat misleading. To my knowledge it is correct that no one has a convincing explanation for the shift in the drift of base (and M1) velocity that occurred abruptly in late 1981. This leaves us uncertain as to when, if ever, such a change might occur in the future. It would be nice to live without such uncertainty. Unfortunately this is beyond our present understanding. Yet this does not have to be a matter of major concern to monetary policymakers. First, the fact that over 80 months have passed with no reoccurrence of such a shift suggests that such shifts are not an everyday phenomenon but rather low probability events. Second, even if such shifts occur from time to time, base growth rules that incorporate feedbacks such as proposed by Meltzer (1986) or McCallum (1988), insulate the growth of nominal income from their effects. Thus, the occurrence of infrequent and unpredictable shifts in the drift of base velocity are not a basis for dismissing the monetary base as an operating guide for monetary policy.
The second and third conclusions cited above are consistent with our own research into the demand for the monetary base. As reported at the last Shadow meeting, our preferred specification for the demand for the monetary base is:
where B is the monetary base, Y is nominal personal income, P is the deflator for personal income, RTB is the Treasury bill rate. DINFU is a measure on unexpected inflation and D82 is a dummy variable that is zero through 1981, 12 and 1.0 thereafter. Estimates of the parameters of equation (1) for the St. Louis Adjusted Monetary Base are presented in Table 9A.1 and for the Board Monetary Base are presented in Table 9A.2. Estimates are presented for a full sample period, and for sample periods through and subsequent to December, 1981.
The estimates for the Adjusted Monetary Base in Table 9A.1 indicate that, aside from the shift in the drift at the end of 1981, there is absolutely no difference in the estimated parameters or the standard error of the residuals, regardless of the sample period that is used in the estimation. In particular, the interest sensitivity and short-run real income elasticity parameter estimates from the 1982–88 sample for all practical purposes reproduce the estimates from the 50s–70s.
The results from the estimation for the Monetary Base in Table 9A.2 are quite similar to the results for the Adjusted Monetary Base. In this case there is some slight variation in the estimated parameter values from the pre-82 to the post-81 sample periods, and the standard error of the residuals is somewhat higher in the latter sample period. These differences are far too small to have any significance.
The residual standard errors in both of these tables are considerably smaller than those from the corresponding specifications in terms of M1 or M1-A, which provides support for the conclusion that monetary base velocity is more predictable than that of either measure of transactions money.
At first glance, the long-run interest elasticity of the demand for the monetary base, computed as β times the level of the Treasury bill rate, appears quite small in absolute value. This is an inference that should be treated with great caution. It may not be appropriate to construct an estimate of the long-run interest elasticity with such a calculation from this type of specification. The issues involved in the appropriate measurement of the interest elasticity of the monetary base, given the random walk nature of velocity are complex and highly technical. The highly preliminary results of other research that is currently underway into this question suggest a long-run interest elasticity of the monetary base of the order of −.3 to −.5. The corresponding long-run interest elasticities of M1 demand are somewhat larger in absolute value.
It is not at all clear that the demand for the monetary base is less interest sensitive than the demand for broader monetary aggregates such as M2 or M3. In the case of the broad aggregates, it is not possible to reject the hypothesis that the long-run interest elasticity is zero, computed under the assumption that own rates of return on deposits are fully adjusted to changes in market rates of interest. The size of the short-run interest elasticity of the broader aggregates is critically dependent upon how fast deposit rates are adjusted to changes in market rates of interest. Given regulation Q controls into 1985 on at least some types of deposit rates, there is very little experience from which to infer how unregulated deposit rates adjust to changes in market rates of interest.
When all the dust settles, the ultimate reason for the rejection by the FOMC of either measure of the monetary base as an operating instrument or target for monetary policy is the fourth conclusion above, namely that such an operating instrument would produce intolerable interest rate fluctuations. This is the historical basis of objections by the Federal Reserve to any monetary aggregate that has been proposed as a target or operating instrument for monetary policy. The substantive basis for this position is extremely weak. The experience with the New Operating Procedures in 1979–82 is typically cited as support. However, the experience of 1979–80 is contaminated by (1) the uncertainty of market participants (and perhaps also Federal Reserve officials) in the fall of 1979 about exactly how the New Operating Procedures would be implemented and (2) the credit controls fiasco in the spring of 1980. Analysis of the experience in 1981 and 1982 under the New Operating Procedures suggests that interest rate variability during this period was no greater than prior to 1979 or subsequent to 1982.
TEN
Past Problems and Future Opportunities
Don’t try tricks, don’t try to be too clever; keep steady, keep committed to your mandate, even in exceptional circumstances; say as much as you can about what you are going to do: announce a “strategy”; don’t be dogmatic, but follow a
policy which is always in line with your strategy. (Issing, 2003)
Otmar Issing, a distinguished central banker, showed in his years spent guiding the European Central Bank how to operate successfully in the space between rules and discretion. He rejected both commitment to a fixed rule and no rule at all. He called it rule-like behavior. Central banks’ policy should announce and implement a clear strategy that market participants understand and from which they can predict central bank responses. In the normal course of events, it should follow that strategy.
Central banks operate in an uncertain world. They have lender-of-last-resort responsibility in addition to their macroeconomic responsibility. Serving as lender of last resort requires departing from the normal operating rule or strategy. To prevent liquidity problems from becoming crises, the central bank must not follow a fixed strategy. The rule for crises should be known in advance, as Bagehot (1873) first stated clearly. It then becomes part of rule-like behavior.
The Federal Reserve’s authority is delegated; the U.S. Constitution gives Congress the power over money. In 1913 Congress chose to appoint an agent to carry out these tasks. It granted independence, but it always retained the right to withdraw or restrict it. Members of the Board of Governors hesitate to act in ways that arouse public and congressional ire. This alone makes the Federal Reserve a political as well as an economic institution and weakens independence, as the preceding chapters show.
Political pressures also come, at times, from the administration. The Federal Reserve is not part of the administration, but its actions affect the degree of public support that an administration receives. The modern public holds the administration responsible for the state of the economy and its current and expected well-being. The administration can not sanction the Federal Reserve. Public complaints are usually counterproductive. There is a gap between responsibility and authority. The administration and Congress are held responsible for economic shortcomings; most people are unaware that the shortcomings often result from Federal Reserve actions and decisions.
A History of the Federal Reserve, Volume 2 Page 75