The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
A S eries of O c c a sio n a l Papers in D raft Form P re p ared by M e m b e rs 'o TH E PRO BLEM S OF M O N ETA R Y C O N TR O L U N D ER Q UA SI — C O N TEM PO RA N EO U S RESERVES Robert D. Laurent SM-84-5 March 1985 The Problems of Monetary Control Under Quasi - Contemporaneous Reserves by Robert 0. Laurent Department of Research Federal Reserve Bank of Chicago The views expressed herein are solely those of the author and do not necessarily represent the views of the Federal Reserve Bank of Chicago or the Federal Reserve System. The material contained 1s of a preliminary nature, 1s circulated to stimulate discussion, and not to be quoted without permission. 2 In recent years the reserve accounting system used by the Federal Reserve has received a great deal of attention. In particular, after the Fed announced the adoption of a reserve targeting procedure of monetary control 1n October 1979, there was strong and mounting criticism of lagged reserve accounting. In response, the Fed adopted a "contemporaneous" reserve system 1n February, 1984. This new reserve accounting system appears to have pleased many critics of lagged reserve accounting who apparently see the new system as a solution to the problems posed by lagged reserves for a reserve targeting procedure. This paper examines the motivation behind, and the effects of the move to the new reserve accounting system. The monetary policy history of the last two decades has been replete with changes 1n reserve accounting or operating procedures which have failed to produce the effects Intended. The adoption of lagged reserves Itself 1n 1968 produced exactly the opposite effects from the expected results that motivated the adoption of the system.1 The adoption of reserve targeting 1n 1979 produced an Increase 1n short-run deposit volatility rather than the Intended decrease. 2 Other changes such as operating through RPO's (reserves against private deposits) proved simply Impossible to Implement. This history suggests that changes 1n reserve accounting or operating procedure have often been advocated and/or adopted without a true understanding of either the nature or consequences of the changes. In the vernacular, changes 1n reserve *The author has benefited from the comments of Herb Baer, Anne Marie Gonczy, George Kanatas, Paul Kasrlel, Randy Merrls, Ed Nash, Larry Mote, Mary Rosenbaum, Steve Strongln, and Vefa Tarhan. Any remaining errors are entirely the responsibility of the author. ^See the paper by (Burger) and the article by (Coates). ^Laurent (1982) 3 accounting or operating procedure have often been a "pig 1n a poke." The analysis 1n this paper suggests that the new reserve accounting system 1s but the newest "pig 1n a poke". This paper argues that the reserve accounting system adopted 1n February, 1984 differs 1n Its monetary control properties 1n a substantive way from the contemporaneous reserve system that those who advocated the change suppose 1t to be. The major reason for this difference 1s the existence of a two day lag 1n the new system, making 1t , 1n fact, a quasl-contemporaneous reserve system. The paper argues that the two day lag 1n the quasl-contemporaneous system significantly changes the role of the monetary authority under a reserve targeting procedure and raises both the costs of reserve management for banks and the level and volatility of excess reserves 1n the banking system above that of a true contemporaneous reserve system. The analysis 1n the paper shows quasl-contemporaneous reserves, while Inferior to contemporaneous reserves, to be a potential Improvement over the previous lagged reserve system. However, the potential Improvement 1s likely to be minor and obtained at both Increased adjustment costs to the banks and a Increased complexity, which could easily lead observers to the erroneous conclusion that reserve accounting changes have little potential for Improving monetary control. Perhaps as Interesting as the analysis of the effects of a switch to quasl-contemporaneous reserves 1s the paper's analysis of the reason why so many critics of lagged reserves failed to understand that the new reserve accounting system 1s not a true contemporaneous system, difference 1s Important. and why the The paper argues that this failure stems from a widely held, but Inappropriate model of money stock determination that utilizes a mechanistic multiplier approach where changes 1n reserves lead directly to changes 1n money. The paper argues for a model 1n which the 4 impact of reserves on deposits works through its influence on interest rates. This latter approach reveals a potentially substantive difference in monetary control between quasi-contemporaneous and contemporaneous reserves. The first section of the paper discusses the connection between reserve accounting and monetary control and presents two contrasting views of the money stock determination mechanism. The second section contrasts the monetary control problems of lagged reserves as perceived through these two differing views of money stock determination. Section three describes the difficulties of operating a reserve targeting procedure under the quasi-contemporaneous reserve system and how it differs from the usual conception of a reserve targeting monetary control system. The fourth section contrasts the monetary control properties of quasi-contemporaneous and contemporaneous reserves under a reserve targeting procedure. Conclusions are presented in section five. I Though present discussions of reserve accounting are inextricably intertwined with monetary control, reserve requirements were originally imposed in 1864 for the purposes of increasing bank liquidity. At that time monetary control was not even considered a goal of monetary policy. not become an accepted goal of monetary policy until the 1970s. It did Even after the monetary authority adopted a money stock goal for monetary policy, a controversy remained over whether superior monetary control was obtained by targeting interest rates or by targeting reserves. Through most of the 1970s the monetary authority attempted to control money through an interest rate targeting procedure. As long as monetary control was attempted through interest rate targeting, little attention was paid to the reserve accounting system. It was only with the announcement by the Fed of a reserve targeting 5 procedure of monetary control that the Impact of reserve accounting on monetary control attained Importance. To understand the development of the association between reserve accounting and monetary control it 1s helpful to understand both the common perception and the actual mechanism of money stock determination through reserves. Banks change the money stock by exchanging earning assets with the public. If banks buy earning assets (1.e make loans or buy securities) they Increase the money stock while 1f they sell earning assets (1.e reduce their loans outstanding or sell securities) they reduce the money stock. The analysis of money stock determination then, 1s essentially an analysis of what Induces banks to exchange earning assets with the public. The standard description of the process of deposit expansion or contraction 1n the elementary money and banking textbooks 1s that the level of excess reserves determines money expansion or contraction. In the usual textbook scenario a bank eliminates Its excess reserves or reserve deficiency by purchasing or selling a quantity of earning assets equal to its excess or deficiency 1n reserves. This first bank's adjustment then disturbs another bank's deposits (and excess reserves) by a somewhat smaller amount. This process continues until the banking system comes to an equilibrium 1n which, according to the textbook scenario, required reserves have been equated to the level of reserves. The banking system thus comes to equilibrium through a series of successively smaller adjustments by Individual banks. One widely held view, characterized here as the "mechanistic multiplier" view, of the relationship between reserves and money stock determination seems to approach the Issue much 1n the spirit of the standard textbook approach. In this "mechanistic multiplier" approach, changes 1n the level of reserves are believed to Induce a proportional change 1n deposits through the operation 6 of a money "multiplier" which magnifies and transmits the effect of changes 1n reserves onto money. The typical adherent of the "mechanistic multiplier" view of money stock determination may acknowledge that 1n the very short run the multiplier might be erratic, but the multiplier 1s viewed as sufficiently stable, over the longer periods of time considered Important for monetary policy (e.g. a quarter) to guarantee reasonably accurate monetary control. The typical "mechanistic multiplier" adherent can cite extensive empirical evidence of the stability of the money multiplier over longer periods of time. The Implication 1s that such multipliers can be used to forecast the monetary control properties of a reserve targeting policy. However, the operation of the mechanism that underlies the multiplier 1s seldom detailed, but presumably Involves something akin to the mechanism 1n the standard textbook description of the deposit expansion or contraction process - where banks react to excess reserves and reserve deficiencies. It 1s probably true that the great majority of those who have criticized lagged reserves and advocated the adoption of contemporaneous reserves since the adoption of reserve targeting could properly be classified as adherents of the "mechanistic multiplier" approach to money stock determination. However useful the textbook scenario may be as a pedagogical device for demonstrating the ability of banks to produce a multiple expansion or contraction 1n the money stock from a given change 1n reserves, 1t 1s clearly not an accurate description of the reaction of banks that alters the level of deposits and the money stock. In practice, banks have available a federal funds market where they may buy or sell reserves at the current federal funds rate. Banks, therefore, do not feel obligated to move their earning assets to conform to the level of deposits they have attracted. Rather, the price of reserve credit relative to the return available on bank earning assets determines whether a bank Increases or decreases Its earning assets purchased 7 from the public and therefore Increases or decreases the money stock. For example, even a bank currently deficient 1n reserves will extend additional loans or purchase securities, and thereby Increase the money stock, 1f the rate on current and expected future federal funds 1s low enough. The bank acts as an Intermediary and simply purchases enough federal funds to cover the reserve loss from the purchase of new assets as well as cover Its original deficiency. Conversely, a high federal funds rate will Induce banks to reduce their holdings of securities and loans (thereby reducing deposits) and channel the funds obtained Into the federal funds market. This view of bank behavior 1s consistent with the fact that many large banks continuously purchase, and many small banks continuously sell funds 1n the federal funds market. This pattern reflects a profit maximizing adjustment by banks to a situation where, 1n the absence of federal funds transactions, the rate on the marginal earning asset 1s above the federal funds rate at large banks and below the federal funds rate at small banks. In many cases, large banks consistently purchase more reserves 1n federal funds than their level of required reserves, so that without these funds the banks would not only be deficient, they would actually have negative levels of reserves. If banks mechanistically responded to tne level of reserves associated with their deposits, these banks would long since have reduced their earning asset holdings to cover the potential reserve deficiency. In this "Interest rate" view of money stock determination the Impact of the federal funds rate on bank behavior and the money stock 1s clear. The higher 1s the federal funds rate, the higher will banks expect future federal funds rates to be, and the fewer earning assets purchased from the public a bank will hold and the lower the level of deposits created by the bank. Banks compare the return expected on an earning asset that could be purchased from the public over some period of time with the return expected on federal funds 8 rolled over dally during the same period. At higher expected federal funds rates lending 1n the federal funds market will appear more attractive and lending to the public less attractive. Thus, a higher federal funds rate leads to a lower money stock and a lower federal funds rate leads to a higher money stock. The Important point to emphasize 1s that 1t 1s the federal funds rate and not the level of reserves that 1s the proximate determinant of changes 1n the level of deposits and money. Most Importantly, for the analysis that follows, 1t 1s critical to understand that 1f perchance the reserve accounting system allowed a discrepancy 1n the signals sent to the banking system by reserves and Interest rates, 1t 1s Interest rates, and not reserves, that will determine the actual money stock changes. As noted earlier, there has long been a controversy between advocates of reserve targeting and the previous Fed policy of Interest rate targeting with regard to the better method of monetary control. If as 1n the "Interest rate" view of money stock determination, Interest rates determine changes 1n the money stock, can 1t make any sense to advocate a reserve targeting procedure? It will be argued here that 1t 1s perfectly defensible to argue for a reserve targeting procedure of monetary control even 1n the "Interest rate" view of money stock determination, but 1t must be understood that such an argument 1s actually a call for a specific automatic mechanism for setting the price of reserve credit (i.e. the federal funds rate). Conceptually, the mechanism 1s designed to automatically move the federal funds rate to facilitate monetary control. Under such a mechanism, the monetary authority first sets the level of reserves that 1t believes corresponds to the desired level of the aggregate that 1t wishes to control. In this respect the "Interest rate" view 1s no different than the "mechanistic multiplier" view of a reserve targeting procedure. The difference lies 1n the delineation of the mechanism that underlies the money control mechanism. 9 In the "mechanistic multiplier" view, the setting of reserves seems almost automatically to lead to deposits (and required reserves) moving Into equilibrium with the level of reserves. In the "Interest rate" view the level of reserves tethers an automatic mechanism that moves the federal funds rate to a level that produces a money stock consistent with the level of reserves provided. If the actual level of deposits 1s such that required reserves (which should, with properly set reserve requirements, correspond to the level of the target aggregate) are too low for the level of reserves provided, then the federal funds rate should fall. As 1t falls, it Induces banks to increase their holdings of earning assets purchased from the public and thereby Increase the target aggregate. This process continues until the target aggregate changes enough to move required reserves Into equilibrium with reserves. The monetary authority simply sets the level of reserves. After that the market 1s left to determine the federal funds rate on the basis of the difference between reserves and required reserves. Reserve targeting, as Interpreted by the "Interest rate" view of money stock determination does not deny the Importance of Interest rates, but argues for a system 1n which Interest rates are guided automatically by the market. Having set a predetermined level of reserves, reserve targeting relies on there being no Interference placed on the movements 1n the federal funds rate by the monetary authority. II Even after the monetary authority adopted a monetary target goal 1n the 1970s, 1t continued to operate by setting a federal funds rate. It proved extremely difficult, however, to know what rate was appropriate for the desired level of the money stock. It was frustration with this approach that led to the announcement of a shift to a reserve targeting procedure 1n October, 1979. The shift was Initially greeted with enthusiasm by adherents 10 of money stock control through reserve targeting. Increasingly though, advocates of money stock control through reserve targeting began to complain about the effects of lagged reserves on the Implementation of a reserve targeting procedure. As long as the monetary authority had Implemented policy by setting a federal funds rate, little attention was paid to the reserve accounting system. To one who holds a "mechanistic multiplier" view of money stock determination, an Interest rate targeting procedure means that the monetary authority will essentially move reserves to parallel changes in the level of required reserves. It rules out a policy of hitting some pre-determ1ned target level of total reserves. From the "interest rate" view of money stock determination, the problem of an interest rate targeting procedure 1s that Interest rates are being determined by the monetary authority rather than the automatic mechanism that underlies a reserve targeting procedure. In either view, Interest rate targeting by the monetary authority makes the reserve accounting system irrelevant for monetary control. The Fed announcement 1n October, 1979 that 1t was shifting from an Interest rate targeting procedure to a reserve targeting procedure, brought Increased scrutiny to the lagged reserve accounting system. The problem most often Identified with lagged reserves under a reserve targeting procedure 1s the fact that the monetary authority must supply a level of reserves at least equal to the level of required reserves set by the banking system two weeks earlier. Thus, the monetary authority is constrained from supplying a level of reserves below the level of required reserves set by the banking system two weeks earlier. This constraint 1s effective when the monetary target 1s far enough below the actual level of deposits two weeks earlier to cause required reserves 1n the current week to be above the level of reserves the monetary authority would like to provide. The problem this poses 1n the "mechanistic n multiplier" view of money stock determination, where reserves lead directly to money, 1s that the monetary authority 1s precluded from producing the decline 3 1n reserves necessary to reduce an excessive level of deposits. Indeed, an analysis of a contractionary reserve targeting policy under lagged reserves 1s very Informative 1n contrasting the validity of the "mechanistic multiplier" and "Interest rate" views of the relationship between reserves and money. If required reserves are above the monetary authority's target level of reserves then 1t 1s clear that more reserves must be provided than the monetary authority would wish. When confronted with this constraint, the typical adherent of a "mechanistic multiplier" view might concede that 1t may be necessary to provide more reserves than desired 1n the current settlement period, but that sticking to a slower rate of reserve growth over the longer run would produce slower deposit growth. This observation 1s correct, but 1t begs the Important operational question. That question, under lagged reserves, 1s what the monetary authority can do 1n the current settlement period to reduce deposits 1n the current settlement period so that the level of reserves provided two periods later can be reduced. 1s clear. The answer The monetary authority must provide sufficient reserves to cover the level of required reserves, but 1t can change the price at which 1t provides the reserves. By supplying less of the requisite quantity of reserves through open market operations, the monetary authority forces banks to borrow more reserves at the discount window and thereby pushes up the federal funds rate. The fewer reserves provided through open market operations, the higher will be the level of borrowings and the federal funds rate, and the lower will be deposits 1n the current week. The lesson 1s clear - Interest rates, not reserves, move deposits and the money stock. ^See the articles by Rosenbaum (1984) and Goodfrlend (1984). 12 The problems presented by lagged reserves for short run monetary control are much more pervasive 1n the "interest rate" view than in the "mechanistic multiplier" view of money stock determination. Recall that in the "interest rate" view of money stock determination, a reserve targeting procedure involves having the monetary authority set the target level of reserves, and then the reserve market is supposed to automatically move the federal funds rate, on the basis of the difference between reserves and required reserves, so as to guide required reserves and deposits into equilibrium with the level of reserves. One way of viewing a reserve targeting procedure is that the monetary authority dictates where it wants the banking system to go by setting a level of reserves and the market then automatically adjusts the federal funds rate to move the system on the basis of the difference between where the monetary authority wants the system to go and where the system presently is. Thus, there are two problems with lagged reserves. One is the aforementioned problem, also recognized by the "mechanistic multiplier" view, that on occasion, the monetary authority is constrained from accurately setting reserves as low as they would desire on the basis of where they want the system to go. The second problem, which occurs all the time, is that the mechanism by which the present position of the system feeds back into the determination of the federal funds rate, so as to guide the system to the desired position is faulty. Under a lagged reserve system, the current level of deposits, and therefore any changes in the current level of deposits, have no influence or feedback onto the federal funds rate so as to guide deposits and required reserves into equilibrium with the level of reserves supplied. Indeed, it 1s this second continual problem which explains the fact that not only did interest rates become more volatile, as was expected with the adoption 13 of a reserve targeting procedure under lagged reserves, but short run deposit 4 changes also became more volatile - which was not expected. Ill Monetary control through total reserve targeting under contemporaneous reserves has a somewhat mythical quality. This arrangement underlies the policy procedures recommended by virtually every advocate of steady monetary growth. It 1s also the system Implicit 1n every money and banking textbook description of the banking system's role 1n deposit expansion and contraction. It's somewhat surprising therefore, to observe that total reserve targeting and a contemporaneous reserve system has never characterized monetary policy 1n the U.S. Only 1n October 1979, under a lagged reserve system, did the monetary authority announce for the first time a reserve targeting procedure. Even then, the policy was an unborrowed reserve policy and there 1s a very serious question as to whether the monetary authority 5 actually could have pursued a total reserve operating procedure. The reserve accounting system 1n place before the 1968 switch to lagged reserves, was not a contemporaneous system, but actually a one day lagged reserve accounting system.6 An analysis of monetary control under a combination of contemporaneous reserves (or quasl-contemporaneous reserves) and a total reserve targeting procedure must therefore rely heavily on theoretical analysis since empirical evidence 1s not available. ^See Robert 0. Laurent "A Critique of the Fed's New Operating Procedure" "Staff memoranda 81-3. Federal Reserve Bank of Chicago, 1981. 6Ib1d. 6A bank's required reserves were calculated on the basis of Its deposits at the beginning of the day, so that 1t was effectively based on deposits at the end of the preceding day. 14 The reserve accounting system adopted in February, 1984 differs 1n a number of significant ways from the combination of a total reserve operating procedure and a contemporaneous reserves system Implicit 1n the textbooks. First, the Fed cannot set the level of unborrowed reserves accurately. In practice, such factors as float and Treasury deposits cause random disturbances to the level of unborrowed reserves. Second, banks are able to borrow at the discount window at the sum of the explicit cost of the administratively set discount rate and the Implicit cost of Fed surveillance entailed 1n the discount window. Third, and most significant 1s the fact that the new system retains a two day lag. That 1s, the period from which deposits are taken to compute required reserves runs from the end of business on the Tuesday of the first week to the end of business of the Monday two weeks later. This 1s referred to as the reserve computation period. Banks satisfy reserve requirements with reserves 1n the period from the end of business on Thursday two days after the first Tuesday, to the end of business on the Wednesday two days after the second Monday. reserve maintenance period. This period 1s referred to as the Figure 1 Illustrates the two day lag 1n the quasi-contemporaneous reserve system.7 Figure 1 Reserve Computation and Reserve Maintainence Periods , Tu t W , Th , F , Sa , Su , M Reserve L Th F Sa t Su , Tu , W , Th , F , Sa , Su , ■ Th t F M ( , Sa a Su Computation M ! Tu t Reserve W Maintainence 7For a more complete description of the reserve accounting changes adopted 1n February, 1984 see (Gilbert & Treblng). 15 IV Reserve requirements are designed to help control money under an operating procedure 1n which the monetary authority hits a target level of total reserves. Therefore, a total reserve operating procedure 1s the appropriate background against which to compare the present quasl-contemporaneous reserve g system with the contemporaneous system 1t 1s often thought to be. There are, however, a number of problems Impeding the monetary authority's conduct of a total reserve operating procedure under either contemporaneous or quasl-contemporaneous reserves. Two of the problems, existing under either reserve accounting system - the Inability of the monetary authority to precisely set unborrowed reserves and the availability of the discount window for banks to acquire reserves - are of relatively minor Importance. The major problem presented for the monetary authority in trying to control money through a total reserve operating procedure, occurs under quaslcontemporaneous reserves and arises precisely from the two day lag which differentiates the present quasl-contemporaneous reserve system from the contemporaneous reserve system with which 1t 1s often m1d1dent1fled. This two day lag, together with the fact that a bank may satisfy Its reserve requirements with any pattern of reserves over the settlement period, g produces critical differences for monetary control between quaslcontemporaneous and contemporaneous reserves. ^Essentially, a contemporaneous reserve system 1s one 1n which the Reserve Computation and Reserve Ma1nta1nence period are coterminus, so that 1t 's possible for the banking system to change required reserves after the target level of total reserves has been set. ^Except that a bank cannot have a reserve account which 1s overdrafted (1.e. have negative reserve balances) at the end of any day. 16 One may quickly grasp why the two day lag 1s so Important, by considering an extreme example. Suppose that, under the present reserve accounting system, the monetary authority acted so as to convince the banking system that the federal funds rate would be, say 8 percent on the final two day, lagged portion of the settlement period. Then 1t 1s clear that the federal funds rate will not deviate far from 8 percent on the first twelve days of the settlement period. No bank will sell federal funds at less than 8 percent, or pay more than 8 percent 1f 1t knows that the federal funds rate will be 8 percent sometime later 1n the settlement week. In this extreme situation the monetary authority could literally move reserves anywhere they wanted on the first twelve days of the settlement period and not affect the level of deposits. The major point of such a hypothetical consideration 1s that (because banks can satisfy reserve requirements with any pattern of reserves over the settlement period) the behavior of the vital federal funds rate 1n the twelve day, non-lagged portion of the settlement period depends critically on the expected actions of the monetary authority 1n the two day lagged portion at the end of the settlement period. In presenting the "Interest rate" view of money stock determination earlier, 1t was argued that the degree to which one follows a reserve targeting procedure depends upon the extent to which the monetary authority allows the federal funds rate to be determined by the Interaction of the target level of total reserves and required reserves based on the existing level of deposits. Since the current federal funds rate 1s Influenced by future expected federal funds rates 1n the same settlement period, a true reserve targeting procedure requires that the monetary authority set the target level of reserves and then avoid constraining or further affecting expected federal funds rates 1n the remainder of the settlement period. It will be argued here that the two day lag 1n the quasl-contemporaneous reserve 17 accounting system significantly inhibits the ability of the monetary authority to do this, and thus the possibility of controlling money through a total reserve targeting procedure. A reserve targeting procedure of monetary control requires that the monetary authority set the target level of total reserves. In practice, one minor problem is that the monetary authority is not able to set the level of unborrowed reserves accurately. Fluctuations in Fed float and treasury deposits at the Fed cause the level of unborrowed reserves to fluctuate unpredictably. It is not possible, at present, to predict and offset these fluctuations before they occur, but it is possible to almost completely neutralize disturbances after they occur. For example, the monetary authority could offset daily the disturbances to unborrowed reserves on each preceding day. This means that the level of unborrowed reserves over the settlement period would be off target only by the unexpected disturbance on the final day of the settlement period. This would be a close approximation to a total reserve targeting procedure. A second, somewhat more major, problem afflicting a reserve targeting procedure under quasi-contemporaneous reserves is the existence of the discount window. At the discount window, banks can choose to borrow and move the level of reserves above the target level. Banks would only choose to do this if the cost of reserves at the discount window was attractive relative to the cost of reserves in the federal funds market. The cost of reserves at the discount window has two components - the explicit cost embodied in the discount rate and the implicit cost embodied in the surveillance imposed on borrowers by the Fed. In the "mechanistic multiplier" view, the discount window causes a problem for monetary control through reserve targeting by allowing reserves to deviate, at the initiative of the banks, from the monetary authority's target level. In the "interest rate" view of money stock 18 determination the problem of the discount window 1s that 1t acts to constrain and Influence Interest rates, thereby preventing the automatic operation of the money stock determination model underlying reserve targeting. The discount window Influences the federal funds rate once that rate surpasses the discount rate. As the level of unborrowed reserves falls below the level of required reserves, the federal funds rate rises. As the federal funds rate rises above the discount rate, banks have an Increased Incentive to borrow from the discount window, Increasing the level of reserves and dampening the rise 1n the federal funds rate.^ Perhaps the simplest solution to the discount window problem 1s to Increase the discount rate to a very high level where there would be no Incentive for banks to borrow at the discount window. This proposal, which has often been linked with the adoption of a contemporaneous reserve accounting system, 1s designed to eliminate borrowed reserves and cause total reserves to be Identical to unborrowed reserves. This high discount rate along with the correction of disturbances to unborrowed reserves, would seem to produce a system closely approximating a reserve targeting procedure of monetary control.^ It will be argued here that whether this 1s so depends critically on whether the reserve accounting system retains a lag - and that the existence of the two day lag 1n the quasl-contemporaneous reserve accounting system 1s the major problem Impeding monetary control through reserve targeting.12 T^See the paper by Kasrlel and Merrls. 11 See Pakko p. 66. 12For a more mathematical treatment of quasl-contemporaneous reserve accounting under a very rigid set of assumptions see Kopecky. 19 To demonstrate the Importance of the two day lag consider first an approximate total reserve targeting procedure (1.e. offsetting the previous day's disturbance to unborrowed reserves and setting a high, penalty discount rate) under a true contemporaneous reserve system where 1t 1s possible for the banking system to affect required reserves after the final level of unborrowed reserves has been set. 13 How does such a system come to equilibrium 1f the level of reserves 1s Initially set below the level of required reserves determined by the current level of deposits? causes the federal funds rate to rise. The deficiency of reserves As the federal funds rate rises, banks find the sale of earning assets to be attractive and thereby reduce the level of deposits. As the level of deposits falls, required reserves also fall until they are 1n equilibrium with reserves and the rise 1n the federal funds rate stops. One can easily Imagine a discount rate set so high that the discount window 1s never used and the federal funds rate always equilibrates at a level under the discount rate. So with a high enough discount rate, under true contemporaneous reserves, the monetary authority can, for all practical purposes, run a total reserve targeting procedure. However, the same arrangement does not produce a total reserve targeting procedure under quasl-contemporaneous reserves. The two day lag means that when the target level of total reserves (adjusted for the error on the last day) 1s below the level of required reserves established by the twelfth day of the settlement period, then no matter how high the federal funds rate rises and how much banks reduce the level of deposits, the level of required reserves cannot be reduced below the level of reserves provided by the monetary authority. In this case, the federal funds rate Increases until 1t 1s above 130ne way to view this system 1s that the monetary authority has completed Its open market operations by the morning of the last day and the banking system has until the close of business of the last day to set deposits and required reserves. 20 the discount rate, however high, and banks borrow sufficient reserves to move total reserves above the level of required reserves. Thus, 1n this case, the administratively set discount rate serves to Influence and constrain the federal funds rate and therefore determine the level of deposits. This scenario of federal funds rate and money stock determination 1s contrary to the description of the mechanism underlying a total reserve targeting procedure of monetary control. It may seem that a total reserve targeting procedure could be closely approximated under the two day lag by setting the discount rate so high that the banking system would never enter the last two days with higher required reserves than the level of reserves provided by the monetary authority. That 1s, by setting the discount rate high enough, 1t might seem that one could eliminate all borrowings. If this were possible, 1t would allow the monetary authority to approximate a true total reserve targeting procedure even with a two day lagged quasl-contemporaneous system. To see why this 1s not always possible, consider the behavior of an Individual bank under such a system. A bank, knowing that the monetary authority sets a target level of unborrowed reserves, might Initially look at the high discount rate and decide to avoid any possibility of needing reserves later in the week. All banks would try to build up excess reserves as Insurance earlier 1n the week. If all banks try to do this, then either there are already excess reserves 1n the system or the federal funds rate would have a tendency to rise early in the week and Induce banks to reduce their earning assets (and deposits 1n the system) so that the banking system would enter the last two days of the settlement week with required reserves well below the level of reserves provided, eliminating the possibility of banks being forced to borrow reserves at the discount window. However, 1f banks always act this way then the federal funds rate would always fall on the last two days of the 21 settlement week and the high discount rate would never influence the federal funds rate. 14 Essentially, the situation would be one in which banks always hold insurance (i.e. excess reserves) for an accident that never occurs. This is clearly impossible as a long run solution. If the banking system continuously acts so as to cause the federal funds rate to fall on the last two days of the settlement period, then the individual bank would eventually realize that it has no incentive to try to build up its excess reserve holdings early in the period. On the contrary, the incentive would be for the individual bank to run a deficiency early in the settlement period and then cover the deficiency on the last two days of the settlement period. As long as other banks act so as to cause the federal funds rate to fall on the last two days, banks that run deficiencies on the first twelve days will do well, and this will increase the incentive for more banks to run larger deficiences earlier in the period. Finally, a period must occur when the banking system as a whole enters the final two days with required reserves above the fixed level of reserves provided by the monetary authority. This is a period when the federal funds rate rises until it surpasses the discount rate by an amount sufficient to induce banks to borrow the requisite quantity of reserves to make up the shortfall between required reserves and the target level of reserves. Thus, periods must occur when some banks will be forced to borrow from the discount window under reserve targeting and quasi-contemporaneous reserves. It is important to understand how this occasional occurence affects monetary control under a total reserve operating procedure and quasi-contemporaneous reserves. The monetary authority can come closer to l^This is because the demand for excess reserves becomes very low and very interest inelastic late in the settlement period at any rate high enough to cover transactions costs in the federal funds market. 22 Implementing a total reserve targeting procedure by raising the discount rate. A higher discount rate causes the role of the monetary authority 1n constraining the federal funds rate to diminish 1n the sense that the fedeal funds rate will be less frequently constrained by the administratively set discount rate. procedure. This 1s what 1t means to move toward a reserve targeting However, it 1s also clear that the quasi-contemporaneous system 1s different than a true contemporaneous system in that it 1s possible under a true contemporaneous reserve system (where the banks can affect required reserves after the level of reserves 1s set) for the administratively set discount rate to be set so high that 1t never serves to constrain the federal funds rate. Under a quasi-contemporaneous system, there must be some times when the administratively set discount rate, even under an approximation to a total reserve targeting procedure, importantly Influences the federal funds rate. The problem a low discount rate poses for the conduct of a reserve targeting procedure is that it can hamper the monetary authority's attempts to slow down money growth. In the "mechanistic multiplier" view this problem appears as an inability to constrain the level of reserves while 1t appears in the "Interest rate" view as a constraint on the ability of the federal funds rate to rise and force the banking system to contract deposits Into equilibrium with the level of reserves that the monetary authority would have preferred to set. Raising the discount rate to a very high level can solve these problems under a contemporaneous reserve system (where 1t 1s possible for banks to alter required reserves once reserves are set). The discount rate can be set so high that banks would never borrow, thereby avoiding the problem of having the discount window alter the level of reserves in the "mechanistic multiplier" view or influence the federal funds rate 1n the "interest rate" view. 23 However, under a reserve accounting system that retains a lag there must be occasions when banks will find 1t necessary to borrow at the discount window, thereby Increasing reserves above the target level the monetary authority would like to provide, and causing the federal funds rate to be Influenced by the administratively set discount rate. Raising the discount rate benefits a monetary policy designed to operate through total reserves by reducing the frequency with which banks make use of the discount window. It might appear then, that monetary procedure under a total reserve targeting procedure and a quasl-contemporaneous reserve system can be made to asymptotically approach the same policy under contemporaneous reserves by raising the discount rate. However, while this Improves monetary policy by reducing the occasions when the banking system borrows at the window and the discount rate constrains the federal funds rate, 1t also raises the level and volatility of excess reserves and the cost of reserve management to the Individual bank. Thus, while a higher discount rate has the monetary authority play a role closer to the role 1t 1s meant to play under a reserve targeting procedure (1.e. the monetary authority becomes more Important 1n determlng the level of total reserves), 1t has an additional effect which weakens monetary control by making the relationship between reserves and required reserves less stable. Thus, there is a tradeoff between the monetary authority's ability to Implement a reserve targeting procedure and constrain monetary growth by raising the discount rate and the level and volatility of excess reserves. Figure 2 Illustrates the nature of this tradeoff. Increasing the discount rate Initially Improves, but eventually worsens monetary control. This occurs because the lower the discount rate, the higher will be the average level of borrowed reserves, and the greater will be the average amount by which reserves exceed the target level of total reserves and deposits exceed the target level of deposits. At low discount rates, deposit Figure 2 The Length of the Lag, the Discount Rate, and the Pattern of Errors in Monetary Control Contemporaneous 1 Day Lag 3 Day Lag Discount Rate 0% Errors in Monetary Control 10 Day Lag 24 expansion 1s basically constrained only by the Impact of discount window surveillance costs on the federal funds rate. The response of the federal funds rate to borrowing tends to be sluggish and gradual, working as 1t does solely through the reaction of the discount officers to extended periods and Increased levels of Individual bank borrowing. As the discount rate 1s raised, the response of the federal funds rate to borrowings at the discount window 1s heightened and accelerated. more quickly. This reins 1n excessive monetary growth However, under a quasl-contemporaneous system where there remains some lag, 1t was shown earlier 1n this paper that there will always be some occasions when banks must borrow at the discount rate, no matter how high. As the discount rate 1s raised, the costs of this borrowing Increase. In response banks will Increase their holdings of excess reserves as protection against a possible high federal funds rate late 1n the settlement period. Since banks do not have prior knowledge of the periods 1n which rates will rise, the Increase 1n the level and volatility of excess reserves will have the effect of weakening monetary control. As the discount rate 1s raised, the detrimental effects of an Increasing volatility in excess reserves finally outweighs the beneficial effects of an Increased ability to constrain reserves, monetary control deteriorates and the curves bend back. One way to represent this relationship 1s to express money as the product of reserves and a multiplier, each of which has an error term, e m respectively. K M =(R + • ' J e n )(r n R,v eD, K and Then f e ) nr as the discount rate is raised the level and volatility of the level and volatility of em rises. falls but Thus the optimal arrangement for the discount window under a quasi contemporaneous reserve system 1s more complicated than simply setting a high, penalty discount rate as 1s often implied. 25 The curves 1n Figure 2 also show that the trade-off between monetary control worsens as the length of the lag 1s extended. The reason for this 1s that when the banking system borrows (1.e., the federal funds rate 1s above the discount rate) then the cost to the banking system 1s proportional to both the federal funds rate and the number of days that one has to borrow. Thus when the banking system 1s forced to borrow, costs are higher, the longer 1s the lag. In response, banks will hold higher and more volatile levels of excess reserves and monetary control will worsen as the length of the lag Increases. The figure also shows that for a contemporaneous reserve system the curve does not bend back. Because banks can bring required reserves Into line with reserves under contemporaneous reserves, any Increase 1n the discount rate above the level needed to Induce the largest adjustment 1n required reserves necessary, does not Increase the cost to banks or the holding of excess reserves. Therefore 1t does not weaken monetary control. The operation of a reserve targeting system of monetary control requires that the federal funds rate respond to the difference between the level of reserves and the level of required reserves based on the current level of deposits. Conducive to the operation of this system 1s the transmittal by banks of the pressures from their excess reserve position Into the federal funds market. The more volatile the federal funds rate, the more effectively will banks transmit their reserve position Into the federal funds market. It 1s Informative to consider the adjustment 1n the behavior of the Individual bank as the discount rate 1s raised under an approximate reserve targeting procedure. As the discount rate 1s raised, the bank comes to expect more volatility 1n future federal funds rates over the remainder of the settlement week. This means that the bank will monitor Its reserve position more closely earlier 1n the settlement period. 26 Perhaps, the best way to see why this happens 1s to consider the opposite situation - where a bank knows that the federal funds rate will be a constant fixed rate over the remainder of the settlement week. Under a situation of absolutely constant expected rates, 1t makes no sense for a bank to monitor Its reserve position earlier 1n the week. A bank that monitored Its reserve position earlier 1n the week and offset any unexpected disturbances, would simply be Increasing Its costs. That 1s, 1t might offset an unexpected Inflow by selling federal funds at one time and then have to offset an unexpected outflow by buying federal funds at another time. It could reduce both monitoring and transactions costs by waiting until the end of the settlement period and then offset net unexpected disturbances with one transaction. This 1s a preferable means of operating since the bank knows that the federal funds rate will be the same later 1n the week. The more the federal funds rate 1s expected to fluctuate 1n the rest of the settlement period, the more sense 1t makes for an Individual bank to monitor Its reserve position earlier 1n the week. In this way the bank can reduce the risk of having to sell reserves at a low rate or buy reserves at a high rate later 1n the settlement period. If many banks monitor their reserve position early In the settlement week, then the basic position of the banking system (1.e. reserve surplus or deficiency) will Influence the federal funds rate and cause the rate to move 1n the appropriate direction early 1n the week. This causes banks to adjust their earning assets (and deposits In the banking system) early 1n the week which results 1n better monetary control and a reduced probability of having the banking system enter the last two days with required reserves above the level of reserves. One way of viewing the problem of the lag 1n a quasi-contemporaneous system 1s as an Inefficient pricing mechanism. The federal funds rate 1s supposed to reflect the relative magnitudes of reserves and required reserves 27 (based on the current level of deposits) and thus guide the banks to move deposits and required reserves Into equilibrium with reserves. the lag 1s that 1t throws this pricing system out of whack. The problem of Banks continue to respond to the rate as 1f 1t accurately reflected the discrepancy between reserves and required reserves. In fact, however, the rate 1s responding to the current level of reserves and the level of required reserves 1n the system at the beginning of the lagged portion of the settlement period. Thus, the banking system loses the benefits of Its ability to respond to the pricing system. It causes the monetary authority to Intervene 1n a more obtrusive way than 1t should 1n the federal funds market. This means that the monetary authority can, through movements 1n the discount rate, trade off Its ability to constrain reserve expansion for Increased levels and volatility 1n excess reserves. This trade-off stems from a pricing Impediment 1n the federal funds market produced by the lag. This same analysis of the federal funds market also shows why monetary control under a reserve targeting procedure with quasl-contemporaneous reserves will be easier and more accurate than would be the case under the same procedure with the previous lagged reserve system. The improvement 1n the new reserve accounting system stems from the fact that 1f the monetary authority convinces the banking system that 1t will hold to the target level of reserves whenever possible and that 1t will allow large changes 1n the federal funds rate when 1t 1s not possible to hold to the target level of reserves, then banks can reduce the pressures Imposed on the banking system through appropriate responses 1n the first twelve days of the settlement week under the quasl-contemporaneous system. If the banking system learns that the monetary authority will allow large changes 1n the federal funds rate at the end of the settlement period, then banks will more closely monitor their reserve position early in the settlement period and produce larger changes 1n 28 the federal funds rate early 1n the settlement period. These large movements in the federal funds rate early 1n the settlement week will cause banks to respond through their holdings of earning assets and produce changes 1n deposits under either lagged reserves or quasl-contemporaneous reserves. The advantage of quasl-contemporaneous reserves over the old lagged system 1s that, to the extent that banks respond and change the level of deposits 1n the proper direction 1n the first twelve days of the settlement period, a quasl-contemporaneous system reduces the pressure on the banking system and the cost of reserve management to the Individual bank while simultaneously giving a more accurate and less volatile money stock because the level and volatility of excess reserves are reduced. But finally, 1t 1s extremely Important to note a less desirable effect of quasl-contemporaneous reserves on the over-all progress toward accurate short run monetary control. The costs of adopting the new reserve accounting system have been substantial. It has been argued here that the system adopted retains the same problems, though to a lesser degree, that troubled the previous lagged reserve system. Imposing such substantial costs to adopt a change that falls to remove the pervasive fundamental problem of lagged reserves may well reinforce the feeling of some observers that reserve accounting changes have little potential to Improve monetary control. Nothing could be further from the truth; changes 1n reserves accounting could provide the monetary authority with accurate short run monetary control even while vastly simplifying reserve management for banks.15 However, after the experience with quasl-contemporaneous reserves, 1t 1s not likely that the reserve accounting system will soon again be changed with the goal of Improving monetary control. 15See Laurent (1983) Those advocates of short run monetary control who 29 applauded the move to quasl-contemporaneous reserves have unwittingly hampered the longer run move toward accurate monetary control by basing their view of money stock determination on a "mechanistic multiplier" model of money stock determination and thereby confusing a quasl-contemporaneous reserves with true contemporaneous reserves. V The analysis of quasl-contemporaneous reserve accounting Indicates that while 1t 1s a potential Improvement, quasl-contemporaneous reserves retains the same fundamental problem that plagued the previous lagged reserve system. This analysis, must of necessity be based on theory, since a total reserve operating procedure under a contemporaneous reserve system has never been the monetary control process used by the Federal Reserve. Quasl-contemporaneous reserves has the advantage over lagged reserves that 1f the federal funds rate moves in the first twelve days of the settlement period, the resulting changes 1n deposits produced by these Interest rate changes will affect required reserves and relieve the Interest rate pressure on banks and the banking system. That 1s, the federal funds rate functions as 1t should under a reserve targeting procedure 1n the first twelve days of the settlement period - as a reflection of the discrepancy between reserves and required reserves and a guide to banks on the actions necessary to eliminate the discrepancy. However, to obtain this advantage, the monetary authority must, at least occasionally, determine the federal funds rate through the discount rate on the last two days of the settlement week, Irrespective of the changes produced 1n deposits on those days by the banking system. That 1s, Interest rates do not function as they should under a reserve targeting procedure on the last two days of the settlement period and quasl-contemporaneous reserves presents the same problems, 1n microcosm, as lagged reserves presented for monetary control. Even when the potential Improvement 1n quasi- contemporaneous 30 reserves over lagged reserves 1s realized by having the monetary authority Impose a high discount rate, quasl-contemporaneous reserves remain Inferior to contemporaneous reserves by Imposing a higher level and volatility of excess reserves and Imposing higher reserve management costs on banks and the banking system. Perhaps most Important, the paper Investigates the basis of the analysis which led many critics of lagged reserves to accept quasl-contemporaneous reserves. The paper argues that 1t stems from a widely held but seriously deficient view of the linkage between reserves and money, labeled 1n the paper the "mechanistic multiplier" view of money stock determination. This view perceives changes 1n reserves to lead directly to changes 1n deposits without any description of the mechanism which causes the level of reserves to Influence the level of deposits. As a result, the Importance of the federal funds rate, expected future federal funds rates and the consequences of the fungible nature of reserves within the settlement week are lost 1n the "mechanistic multiplier" view of money stock determination. The paper presents an alternative "Interest rate" view of the linkage between reserves and deposits which gives a decidedly less sanguine view of the adoption of quasl-contemporaneous reserves. This latter view argues that the new reserve accounting system cannot be made to function under a reserve targeting procedure like the contemporaneous reserve system it is widely believed to be. The Implementation of reserve targeting under quasl-contemporaneous reserves can be made to more closely approach the same policy under contemporaneous reserves by targeting unborrowed reserves and raising the discount rate. However, unlike the role the discount rate might play under a true contemporaneous reserve system, the administratively set discount rate under quasl-contemporaneous reserves must, at least occasionally, serve to Influence the price of reserves to the banking system. Not only does this 31 characteristic of the discount rate under quasi-contemporaneous reserves mean that the monetary authority plays a more significant role 1n determining the federal funds rate than 1t should 1n a reserve targeting procedure, 1t also presents the monetary authority with a true dilemma 1n that an Increase 1n the discount rate, while allowing the monetary authority to more closely Implement a total reserves policy, also has the effect of Increasing the level and volatility of excess reserves. Thus, 1t 1s not always true, under a quasi-contemporaneous system, that the higher the discount rate, the better the monetary control. Setting the discount rate presents a difficult optimization problem. Finally, the paper suggests that the shift to quasi-contemporaneous reserves, even though an Improvement over lagged reserves, may actually work to the detriment of the longer run quest of achieving accurate monetary control. By advertising the shift to quasi-contemporaneous reserves as a major Improvement 1n monetary control, advocates of the shift may eventually contribute to the erroneous Idea that reserve accounting 1s not Important for monetary control. Many advocates of short run monetary control, who have been critical of lagged reserves have unwittingly contributed to this outcome by falling to adequately understand the mechanism underlying a reserve targeting procedure of monetary control and accepting quasl-contemporaneous reserves as a true contemporaneous reserve system. 32 LITERATURE Burger, Albert E. "Lagged Reserve Requirements: Their Effects on Reserve Operations, Money Market Stability, Member Banks and the Money Supply Process." Unpublished paper, Federal Reserve Bank of St. Louis, (1971). Coats, Warren L., Jr., "The September 19168, Changes 1n 'Regulation D' and Their Implications for Money Supply Control." Ph.O. thesis University of Chicago, (1972). Goodfrlend, Marvin. "The Promises and Pitfalls of Contemporaneous Reserve Requirements for the Implementation of Montary Policy." Economic Review. Federal Reserve Bank of Richmond. (May/June 1984). Vol. 70/3. Gilbert, R. Alton and Michael E. Treblng. "The New System of Contemporaneous Reserve Requirements." Review. Federal Reserve Bank of St. Louis. December 1982. Vol. 64, No. 10. Kasrlel, Paul and Randall C. Merrls. "Reserve Targeting and Discount Policy, Economic Perspectives. Federal Reserve Bank of Chicago, Fall 1982. Kopecky, Kenneth J. "Interest Rates and the Money Stock Under 'Almost' Contemporaneous Reserve Accounting". Unpublished paper, Board of Governors of the Federal Reserve System. (April, 1983). Laurent, Robert D. "Lagged Reserve Accounting and the Fed's New Operating Procedure". Economic Perspectives. Federal Reserve Bank of Chicago, Midyear 1982. __________________ . "Comparing Alternative Replacements for Lagged Reserves: Why Settle for a Poor Third Best?" Staff Memoranda 83-2. Federal Reserve Bank of Chicago. (1983). Pakko, Michael R. "Lagged and Contemporaneous Reserve Accounting, Money Market Stability and Monetary Control: A Topical History of Recent U.S. Monetary Policy. Federal Reserve Bank of Richmond. (May 27, 1983). Tarhan, Vefa. "Bank Reserve Admustment Process and the Use of Reserve Carryover Provision adn the Implications of the Proposed Accounting Regime." Staff Memoranda 83-6. Federal Reserve Bank of Chicago. (1983)