The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FOR RELEASE ON DELIVERY THURSDAY, APRIL 5, 1984 1:15 P.M. CST (2:15 P.M. EST) RECENT TECHNIQUES OF MONETARY POLICY Remarks by Henry C. Wallich Member, Board of Governors of the Federal Reserve System to the Midwest Finance Association Chicago, Illinois April 5, 1984 RECENT TECHNIQUES OF MONETARY POLICY Remarks by Henry C. Wallich Member, Board of Governors of the Federal Reserve System to the Midwest Finance Association Chicago, Illinois April 5, 1984 Federal Reserve policies are subject to widely differing inter pretations. This would probably be the case even if all members of the Federal Open Market Committee shared an identical interpretation, which is hardly plausible. If 12 people are always of the same view, 11 are dispensible. But even at the level of the techniques by which FOMC policy is implemented, there may be different views of "how monetary policy really works." In this paper I provide my own view, which may not be shared by every member of the Committee and the staff, and in all details possibly by none. Today it seems to be widely believed that the Federal Reserve's present technique for controlling the monetary aggregates is the same as that in use prior to October 1979, before the reserve-targeting method was initiated. Observers have noted that the funds rate has moved smoothly, as was the case before October 1979 when the Federal Reserve was controlling the growth of money by influencing the quantity demanded via the funds rate and short-term interest rates generally. The policy record now speaks of "the -2 degree of reserve restraint." Since the record began to speak of the operating instruments in these terms, there have been no sharp, sustained interest-rate movements such as were characteristic of the tight reservetargeting procedure after October 1979. How are these observations to be interpreted? Recent funds-rate movements have indeed differed noticeably from the volatility of the period from October 1979 through the fall of 1982, after which the automatic character of the reserve-targeting method was largely modified. Changes in overall reserve positions of depository institutions since the fall of 1982 largely have reflected deliberate policy judgments rather than an automatic response to deviations of monetary aggregates from pre-set target paths. Nevertheless, the Federal Reserve has not reverted entirely to the old technique. One piece of evidence is the temporary quarter-end statement-date pressures that still affect the funds rate. These pressures were largely absent prior to October 1979. While short-term interest rates, and, among them, the funds rate, have reassumed some of the role they played in controlling the money supply before October 1979, a new layer of indirect control has been added to the pre-1979 procedures, employing a market mechanism. It is not the funds rate that is used as the operational instrument but a level of nonborrowed reserves derived as the difference between estimated total reserves and the desired level of borrowing at the discount window. This can also be viewed as aiming at a particular level of borrowing implemented by means of the nonborrowedreserves path. The resulting funds rate reflecting this level of borrowing, 3 therefore, has some input from very short-term market forces. The procedure amounts to an indirect way of influencing the funds rate and other short-term rates which, in turn, affect the demand for money. Observers may differ as to whether, given the relative frequency of nonborrowed-reserve path adjustment, this procedure is better described as targeting on the nonborrowed path or on the level of borrowing. From the point of view of the market, where I believe these things are well understood, the focus on the level of borrowing is signif icant because it leads to a different interpretation of Desk operations. The funds rate level at which the Desk enters the market does not convey the decisive message that the market tries to unravel, as it did in the days before October 1979. desired by the Desk. It is not indicative of any particular rate It is simply the rate that happens to prevail on a day when the manager believes that reserves should be added or drained in order to achieve the desired level of discount-window borrowing on average for the reserve-maintenance period. The action reflects the Desk's assess ment of reserve availability, rather than a desire to move the funds rate, although the action, of course, may affect the rate. Some aspects that may create a contrary impression are dealt with later in this paper. Direct and Indirect Targeting What is the advantage of pursuing indirectly a target that can also be influenced or controlled directly? greater scope to market forces. Principally, it is to give Direct action runs the risk of introducing - 4- discontinuities and rigidities. It foregoes the opportunity of benefiting from a smoothing effect of the market. Judgment errors in setting the objective of direct actions are less likely to be corrected by the input from the market. This applies primarily when "indirection" implies an inter action between a price and a quantity. It applies also, however, to the relationship of two quantities, such as when borrowed reserves or total reserves are determined by operating on nonborrowed reserves. At the same time, one must keep in mind that indirection, giving room to market forces, can introduce a degree of slippage that may interfere with attainment of the target. The issue whether to address a target variable directly or indirectly is posed at various stages in the monetary-policy transmission mechanism. At each stage, policy confronts, in simplest terms, a price and a quantity. It can determine price directly, by operations in the market, and allow quantity to be determined indirectly. Alternatively, it can determine quantity directly, with varying degrees of precision, and thereby influence price indirectly. In one or two instances, the key relation may be between two quantities, one or both of which are parts of a larger total. For a discussion of some of the alternatives available at each stage in the transmission mechanism, the following stages are relevant, in descending order of closeness to the real sector and ascending order of controllability by the central bank: 1. Intermediate targets — principally long-term rates. the money supply and interest rates, - 5 - 2. Instrumental targets — total reserves and money-market rates. 3. Operating targets -- nonborrowed reserves implied by borrowed reserves intentions and the funds-rate range. These layers could perhaps be structured somewhat differently and even telescoped, but they reflect the hierarchy of markets and instruments as they appear to me. Intermediate Targets At the level of intermediate targets, the policymaker confronts, in simplest terms, the relationship between money and interest rates. can influence either one directly — He money by means of a total reserves technique, relying on the money multiplier, or interest rates by buying and selling at a given rate. Alternatively, he can influence each variable indirectly -- the money supply through short-term interest rates, interest rates through the money supply. It need hardly be said that this two-variable relationship functions within a general-equilibrium model with many variables determined simultaneously. Why should the policymaker prefer one intermediate target or the other, and why, having made his choice, should he prefer the direct or the indirect technique, if he is given the choice only between money supply and interest rates as intermediate targets? As for the choice of intermediate target, this presumably will depend on the policymaker's view of the transmission mechanism of monetary policy. He may believe that expenditure behavior of firms and households is driven by interest rates -- in the broad sense of including all kinds of -6monetary and nonmonetary returns — through a rea1-balance mechanism. or by the money supply, for Instance, If he believes, as I do, that monetary policy works primarily through interest rates, he must choose between implementing his interest-rate policy directly, through market intervention, / or indirectly, through the money supply. In the very short run, setting interest rates directly usually -- not always — is possible for the central bank, through discount-rate and open-market operations. In an extreme sense, it could do so by simply pegging a rate through unlimited purchases and sales of securities at that rate. Naturally, if the interest rate established by this technique is not consistent with a stable rate of inflation, it will have an increasingly disequilibrating effect, causing inflation to accelerate or decelerate. Inability to guess or calculate the equilibrium interest rate gives the policymaker an important reason for not trying to set it directly but instead letting the market do so. To be sure, the policymaker also does not know what rate of money growth will generate equilibrium (constant inflation) interest rates; but his risk of error is smaller. If he sets an inflationary rate of money growth, the long-run result will be stable, not explosive, inflation. Thus, letting the market set the interest rate for a given money-growth target is a safer way of achieving an equilibrium interest rate than trying to set it directly. A secondary reason for choosing a money-supply target is its public information effect. Setting (and adhering to) a target informs the public that an effort is being made to control inflation. Reducing the target over time creates a desirable and persuasive expectation of secularly diminishing inflation* Setting interest rates directly would not clearly convey a sense -7of controlled and diminishing inflation. curbing inflation is widely misunderstood. The role of interest rates in Not a few members of the public apparently believe that because interest enters into many prices, higher interest rates mean more inflation, which is to say that the micro effects outweigh the macro effects. Public support for a money-supply targeting policy is likely to be stronger than for an interest-rate policy, although the experience in recent years of very high interest rates under a moneysupply regime may have changed that perception somewhat. In short, the advantage of influencing interest rates by targeting money is that it gives the market a chance to prevent errors that might occur if interest rates were set directly. Instrumental Targets If it is decided to target on money, whether because the policy maker believes that money drives the economy directly, or because he believes that targeting money is a good way of indirectly targeting interest rates which then drive the economy, again there is both a direct and an indirect technique, this time at the instrumental target level, applying to time horizons of a month or two. The central bank can target on total bank reserves which, together with the money multiplier, determine the money supply. This is a relatively direct approach, giving only limited leeway to market forces via endogenous variation in the multiplier. Slippage, of course, is still possible if control of reserves is less than perfect, or if the multiplier is unstable owing to shifts among deposit categories, changes in excess reserves, and other factors. Even given such slippage, -8 the interaction of a relatively rigid money-supply mechanism with a demand for money that is itself stochastic probably will produce sizable variability of interest rates, at least over the short and intermediate run. One indirect technique of controlling the money supply at the / instrumental target level involves control of short-term interest rates themselves so as to evoke a level of demand for money and a resultant stock equal to the target for the money supply. Given the demand curve for money, a shift in the supply curve changes interest rates along the demand curve, as reserves are added or drained to achieve the desired rate level. The money stock, in this framework, depends on the position and shape of the money-demand curve; it is demand-determined. This approach therefore gives the market greater scope for influencing the money stock. As a result, the money stock is vulnerable to error both in estimating the money-demand function and in predicting the values of arguments in that function, par ticularly income. Moreover, there is a substantial lag in the impact of money-market rates upon the amount of money demanded, with about half of the effect being estimated to occur within two or three months. In any event, in this process, interest rates are likely to be far less variable than under the reserves approach. The danger is that changes in money- market rates will not be made quickly enough when the level consistent with the targeted money supply has been misjudged. Another indirect technique is to target on nonborrowed reserves, which allows both short-term interest rates and the money stock to be deter mined in part by the public's demands for money and by the depository institutions' demands for borrowed reserves. This approach is, in a sense, -9a compromise between total reserves and interest rates as instrumental targets, with the outcome for interest-rate variability likely to fall between these alternative regimes. Operating Targets Finally, at the level of day-to-day or week-to-week operating targets, which are those the Federal Reserve can control most closely (various components of reserves, and the federal funds rate), a choice once more must be made between direct and indirect approaches to targeting reserves or the funds rate, respectively. day-to-day operating target — Using total reserves as the which the Federal Reserve has never done — would be a very direct approach, leaving little scope to the market. kinds of slippage — All especially by means of the discount window, but also through reserve carryovers -- would have to be avoided, or else changes in these magnitudes would have to be compensated by open-market operations. These would have to be massive, since in open-market operations a multiple of the initial increase, for example, in discount window borrowing would be required in order to offset further borrowing as banks sought to make up for further absorption of reserves by open-market operations. Quite possibly, banks would seek to protect themselves by carrying large and variable excess reserves, thereby possibly introducing slippage between total reserves and the money supply. All this severely limits the possibility of targeting on total reserves, to say nothing of the consequences for interest rate variability. Targeting on nonborrowed reserves — which the Federal Reserve did after October 1979 and still does on a day-to-day basis -- is a more indirect - technique. 10- The various elements of slippage in the process — discount-window borrowing, reserve carryover and, until recently, the effect of lagged reserve requirements -- allow the market some leeway. Targeting on nonborrowed reserves also allows for a degree of automaticity. aggregates from target alters required reserves. A deviation of the monetary Given a constant supply of nonborrowed reserves, the deviation changes discount-window borrowing and tends to alter the funds rate and other short-term rates. These rate changes -- downward when the monetary aggregates are undershooting the target, and upward when they are overshooting -- tend to push the money supply back toward target over time. The strength of this automatic control feature, however, is at best moderate. While this technique was in use from October 1979 to fall 1982, it had to be supplemented on occasion by discretionary action in changing the discount rate, or in raising or lowering the nonborrowed-reserves path, thus reducing or increasing the need for borrowing and thereby accentuating the change in short-term rates. A second alternative, also at the day-to-day operating level, is targeting on the funds rate. Once more, there is a choice between relatively direct and indirect techniques. The direct approach, in its extreme form, was represented by the familiar pegging operations practiced during and imme diately after World War II. The Fed fixed certain rates by buying and selling (mostly buying) Treasury obligations throughout the maturity spectrum at fixed prices. A different, much less drastic, approach was that employed before October 1979. A range was set for the funds rate, sometimes as narrow as one-half percent and rarely more than one percent. This range was subject to revision between FOMC meetings if growth in money and/or credit moved outside - spec ified "tolerance" bounds* 11- The Desk bought and sold securities so as to keep the rate within the range, or around a particular area of it, on a weekly average basis and at times on a daily basis. Reserves under this procedure became demand-determined, which made timely adjustment of the funds-rate range very important. The procedure gave some scope to market forces, in the sense that the funds rate was able to move, although only moderately, in response to market forces such as reserve supplies and bank reserve management strategies. It gave further scope to the market in the sense that control of the money supply was relatively indirect. Because demand forces were allowed so much influence on the growth of money, the procedure, in turn, yielded to a nonborrowed-reserve strategy beginning in October 1979. Since the fall of 1982, the nonborrowed-reserve strategy and its automaticity have given way to a technique that allows the funds rate to be determined by the market, through the targeting of discount-window borrowing from one reserve-maintenance period to the next, implemented by allowing a flexible nonborrowed-reserves path. At the FOMC meeting, an intended borrowing level is set, as a policy decision. This level of borrowing is then deducted from the total of required reserves consistent with the target path for the money supply and an assumed level of excess reserves -- in order to derive an initial path for nonborrowed reserves. However, during the intermeeting period, as money and reserve demands deviate from the trajectories set at the time of the FOMC meeting, the intended borrowing level is sought through appropriate adjustments to the initial nonborrowed-reserves path. - 12- The post-fall 1982 procedure differs from the post-October 1979 procedure in that, as anticipated total-reserve demand diverges from initial projections, nonborrowed reserves are adjusted weekly in seeking to achieve a chosen level of borrowed reserves. In contrast, under the October 1979 procedure, borrowing was allowed to change consistent with the attainment of a nonborrowed-reserves path targeted for the entire intermeeting period — although subject to technical adjustments. An assumed level of borrowing under the older procedure was set only initially at the beginning of the inter-FOMC period, but borrowing would subsequently diverge from that initial assumption reflecting unforeseen movements in the demand for money and reserves. This was the automatic feature of the technique which at times was reinforced by discretionary path changes. The relation of the borrowing level to the funds rate, which has been one of the most familiar features of the money market, always has been relatively loose. Since a chosen level of borrowing is consistent with any of a range of values of the funds rate, current operating procedures cannot be regarded as a form of rate-pegging. Demands for discount borrowing by banks no doubt reflect market judgments about present and future deposit flows and likely reserve conditions. Since these considerations play an important role in determing the funds rate, it is clear that the present procedure allows at least one additional degree of freedom with respect to the pre-October 1979 technique. Interpretations of Desk Operations From the point of view of the Fed watcher, the present technique offers problems of interpretation quite different from those of the pre-October -131979 procedure. Under the old procedure, the rate at which the manager entered the market was highly significant. Ordinarily, it meant that he did not want the rate to move substantially beyond that point, or even that he would like the rate to stop somewhat short of the rate at which he had entered. When the market had had an opportunity to explore the upper and lower limits of the range, it had a fairly good understanding of prevailing policy. So long as the market believed that the rate objective remained unchanged, moreover, it would help the manager stabilize the rate, believing that when it had reached one of the limits any move could only go in the other direction. Today, the funds rate range set by the FOMC is much wider than before October 1979, typically 400 basis points. are rarely explored. Its extremes, in fact, So long as the level of borrowing is maintained, there is little reason to expect the funds rate to move strongly, at least for longer than transitory periods. The manager's entry into the market does not signify that one of the limits of the range has been reached, but that, given the borrowing target and the associated nonborrowed-reserves path, reserves need to be added or drained according to Fed projections of reserve availability. In some degree, this is indicated by the fact that entry continues to occur at a set time of day instead of, as during the pre-October 1979 regime, at varying times prompted by intra-day movements in the funds rate. When the reserve objective has been reached, there is no reason why the rate should not move against the intervention if that is the direction of market pressures. -14Uncertainty about the reserve projections available to the Desk sometimes may create the impression that the Desk is indeed working to influence the funds rate directly instead of seeking to influence the borrowing level. In the absence of trustworthy projections, the/ funds rate at times may be a more accurate indicator of reserve availability than the reserves projections. If the manager decides to act on the signal from the funds rate in assessing the volume of reserves needed, he may create the appearance that he is working to influence the rate rather than the supply of nonborrowed reserves consistent with the intended borrowing level. In setting the intended borrowing level, the FOMC must make an assumption about excess reserves. This can be regarded as a technical assumption, however, to be modified later by the staff implementing the directive in accordance with evidence of changes in the demand for excess reserves. Ordinarily such changes are not large and can be reasonably well evaluated. The degree to which the funds rate is determined more reliably by borrowed reserves or by net borrowed reserves (borrowed reserves less excess reserves) is unresolved. net borrowed reserves. answer. There are partisans of both borrowed and Econometric work does not seem to give a decisive It should be noted, however, that when the value of required reserves is known, as under lagged reserve requirements, any nonborrowed-reserves target, rigorously pursued over the reserve-maintenance period, is equivalent to a net-borrowed-reserves target. Under contemporaneous reserve require ments, the same is true to the extent that required reserves can be estimated and that nonborrowed reserves are made to vary with required reserves. A - 15 - word may, therefore, be appropriate at this point about the recently introduced contemporaneous reserve requirements. Contemporaneous Reserve Requirements The shift from lagged to contemporaneous reserve requirements (CRR) reflects a phase in Federal Reserve thinking when it seemed particularly important to tighten and speed up the response of reserve conditions to deviations of Ml from its target path. Lagging required reserves by two weeks implies that, during this period, the expansion of deposits is not directly constrained by reserve availability. Banks theoretically could create deposits without limit, although it strains credulity that they would exploit this opportunity, not knowing where the reserves would come from two weeks later or what they would cost. More plausibly, the response of banks to changes in deposits and the associated changes in short-term interest rates, may be somewhat delayed by the two-week lag in the need to put up reserves. Actually, under its reserve-targeting strategy, the Federal Reserve in effect often cut the two-week lag to one, by recalculating the average level of borrowing implied by a constant intermeeting average level for nonborrowed reserves as soon as incoming weekly deposit data indicated changes in future borrowing needs. This was done by lowering or raising the weekly nonborrowed-reserves path, thereby producing some borrowing response one week earlier than it would have occurred otherwise. The recent move to CRR thus potentially speeds up initial responses by one week rather than two. In any event, CRR seemed a logical complement to the automaticity of the reserve strategy. Their adoption reflected a degree of frustration stemming from the fact that the adverse features of the strategy, in the - 16- form of greater variability of interest rates, were much in evidence, while improved control over the money supply was less so. unlikely to do harm and capable of doing some good. The change seemed It implied an effort to go as far as possible in the direction of making the rigorous reserves strategy effective. Subsequent experience with the behavior of Ml was largely responsible for making this approach less viable. Changes in operating techniques, beginning in the fall of 1982, therefore, downgraded the role of Ml and reduced the degree of automaticity. This seemed to make moot the case for CRR, at least for the duration of this policy approach. On the other hand, concern that CRR would lead to greater volatility of interest rates diminished for the same reason. What remained was a moderate potential improvement in the reserve aggregates to money-supply relation that may help reduce one element of slippage in the mechanism and that expanded the menu of feasible operating procedures for future consideration. Some Comments on the Aggregates A major reason for modifying the automatic reserve-targeting technique has been the erratic behavior of Ml demand relative to its primary determinants. This, in turn, seems to have reflected, at least in part, the transition to a different composition of the aggregate, in the course of the rapid increase in NOW accounts and, subsequently, super-NOWs. Approximately one-fourth of Ml now bears explicit interest. For the $90 billion of regular NOW accounts, this rate is not a market rate, though it is for the $40 billion of super-NOWs. It will become so, for the regular NOWs, as the minimum balance to open super-NOW accounts -- which have no -17interest-rate ceiling — declines to $1,000 in January 1985 from the present level of $2,500 and then is entirely eliminated in January 1986. The ceiling rate on regular NOWs is close enough to the market, however, to allow small changes in market rates to produce large variations in the opportunity cost of holding regular NOW balances, so long as their rate typically remains at the present ceiling levels. For the time being, this may have made Ml more interest elastic than before. However, as the share of super-NOWs grows, and particularly when the minimum-balance requirement for all NOW accounts is removed, rates on the interest-bearing component of Ml increasingly will be market related. This would reduce, perhaps substantially, the interest elasticity of this aggregate. The control of Ml through an interest-rate strategy then would function largely to the extent that interest rates influence GNF and thereby Ml demand. Of course, the possibility of controlling Ml through a total-reserve strategy would remain. But, given a low Ml interest elasticity, the demand for the aggregate would not be much affected by interest-rate variations. Interest-rate volatility resulting from an effort to control Ml through total reserves, therefore, might become even more severe. Instability in the demand function for Ml during 1982 -- which did not occur for the first time in that year — along with the impending introduction of M M M s and maturing of All Saver Certificates — downgrading of the aggregate as a target in 1982. prompted the The demand function seems to have stabilized somewhat in the meantime, but with altered properties. For instance, the large interest-bearing component in Ml is likely to produce more rapid growth of the entire aggregate in the future, relative to nominal -18income and other monetary aggregates. In past years, the difference in the growth rate between Ml on one side, and M2 and M3 on the other, averaged on the order of 3 percentage points, with cyclical variations. difference of 1-2 percentage points now seems more likely. A secular This smaller difference is reflected in the Federal Reserve's 1984 targets of 4-8 percent for Ml and 6-9 percent for M2 and M3. At constant rates of interest, velocity may tend to grow in the 1-2 percent range. Currency also seems to have been experiencing some instability. Until very recently, its average rate of growth had risen to 10 percent or so. This would not by itself be enough to disrupt seriously the rehabilita tion of Ml as a usable target. monetary base. Its implications are more serious for the With currency growing at 10 percent, setting base growth much below its 1983 average rate of almost 9 percent would mean that total reserves, which make up only 20 percent of the base, would have to decline. Reservable deposits would have to do likewise. This, in turn, would, of course, have a severe impact on Ml, the deposit component of which is the principal user of reserves. Accommodating changes in the composition of Ml, on the other hand, i.e., by offsetting fluctuations in the currency/deposits ratio, would be tantamount to targeting on reserves. M2 has also undergone a change that over several years has sub stituted market-related for regulated interest rates. The interest sensitivity of the aggregate accordingly must be presumed to have diminished. M2, in this sense, has already undergone some of the development that may be ahead for Ml. Not enough time has passed, however, to provide adequate data for a test. -19Can We Shed Velocity? Recent vicissitudes of the aggregates, and prospective future changes, raise questions about the time-honored concept of velocity. The notion of a simple velocity relation between nominal income and money is so deeply embedded in the lore of money that it may seem quixotic to try to eradicate it. Nevertheless, in my view, that is what should be done. It is, after all, a primitive concept, clearly inferior to that of a demand function for money. Its calculation leaves out of account the effects of interest rates, wealth, inflation, and other arguments that may play a role in the money-demand function. Its theoretical foundations are weak, unless the demand function is connected to a velocity expression. a luxury good. Secularly, it should decline if money is Historically, since World War II, that has not been its trend, although the upward trend of interest rates and inflation during that period is partly responsible. The most appropriate way of defining velocity, by relating^money to income with a lag, or without, is heuristically rather than theoretically founded. Debates about whether or not there have been shifts in velocity, and how they should be reflected in money-supply targeting, are conducted much more meaningfully in terms of the stability of the demand function for money. Otherwise, changes in velocity that occur along a stable demand function may be confounded with changes associated with a shift in the function. Velocity may even remain stable while offsetting changes occur within the demand function. The principal loss from shedding the simple notion no doubt would be to the reputation of the economic profession, that would probably be accused once more of creating an unnecessary confusion. #