The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Ohio State Uni.versity, Colmn"b\.ls, Ohio Distinguished Lecture Series March 2, 1972 - Willis J. Winn SOME RANDOM THOUGHTS ON CENTRAL BANKING: THE ART AND SCIENCE OF MONETARY POLICY L.~ TRAI';SITION It is a genuine pleasure to be here this afternoon to participate in the University's Distinguished Lecture Series on Banking and Monetary Policy. 1viy associates in this Series certa:.nly provide any measure of humility I might need. As a former Dean and now as a participant in the monetary pdlicy p!"ocess, I am painfully aware of the explosion of economic knowledge over the last several years, although I have to admit there is a considerable portion of it that I have difficulty reading; some that I have read, but have difficulty understanding; and much of which I haven't even attenlpted. The real coro..fidence I have is the feeling that I approach my topic froUl a position of studied igno::tance and it's been m.ore than a bit' presumptuous of r.n.e to accept this assignUlent. In addition, myacadernic 'c." interests over the years \-vere focused prilnarily on the capital market area, and my strong biases will undoubtedly show through this afternoon, After having served 10 years as a Director and Chairman of the Federal Reserve Ba!lk of Philadelphia and being reasonably familiar with the Systeln, I'd like to share the general obeen-ation that there is considerably more to the job of being President of a Federal Reserve Bank than form.ulation of Ulonetary policy. ·of paper, problerns1 and ideas. I find myself literally drowned in the flow In my varied role, I range tro!n the conscience of the conlUlunity to Rotary Club entertainer, while eJso dealing with 1,400 incli.viduals all with hur{l~"'1. cares ~"'1.d concerns. As a -2- corporation executive, I face nearly 500 stockholders with their individual problems. frightening. Sometimes my problems a.re humorous and sOI'r"$times For example, I was recently called upon in the middle of the night to arrange bail, because a stranger heard the Bank is a public institution with a legal staffo My first ~eaction to the System was that its complex makeup bears a great deal of resemblance to a University structure o Consequently, I felt right at home. The seven member Board of Governors in Washington acts much like a central administrative authority in a University, and the 12 banks and their 24 branches are scattered much like individual schools, each with its own strivings for independence and autonomy. The Open Market Cornrnittee is, of course, a major vehicle for developing !!wnetary policy. The seven Governors and 12 Presidents are talented individuals..,extremely well qualifiecl--and they generate both ideas and sometimes confusion. In many respects it seems a little like being back at faculty meetings. Some come in with predetermined positions; some come ready to vote depending on the persuasiveness of the arguments presented. all committees the differences often dissolve at the voting stage. differences that Like The you will observe from time to time, however, may not mean what they seem--m.embers vote against a recommended action because a move did not go far enough, not that they were opposed to the nature of the action. After my hectic first six months .on .the inside, I appreciate this opportunity to step back and take a look at the econornic framework and -3- linkages through which stabilization policy, particularly monetary policy operates. Specifically lId like to sketch some random thoughts on central banking • . Obviously, Pm the first to recognize that these conclusions may be significantly different in another six months. Overall, though, central banking is exciting. must keep abreast of changing time s. . Operationally, we In fact, we must do more than keep . abreast, we must anticipate and initiate change. daily operations and in our policy activities. This applie s both in our For example, a program of computer entry for Treasury securities is eliminating an enormous volume of paper storage and handling to permit more efficient Fiscal Agent and bank service operations. In a broader context, we are in the process of establishing regional check processing centers--possibJy including On<:> here in Colurrlbus--that will allow one-day clearing of almost all checks written in the United States. These changes in the payments mechanism are one more step in the evolution toward electronic money flows--science fiction is becoming reality. In a different area, decisions made in regulating and supervising member banks have a major influence on nlarket structure and the future character of the banking business. But my major concern this afternoon . is with monetary policy. A simple statement of the complex problem I want to consider is: How can the central bank decide what policy actions to take? One of the complexities involved in this question is a pluralism in our concept of the national economic policy goals of a central bank. On the one hand, the central bank -4- is thought of as a. conservator of orderly markets--meaning financial markets in which prices and quantities are tolerably stable and dependable in the short- run. On the other hand, the central bank is though~ of as the implementor of broader national economic policies seeking price stability, economic growth, full employment, balance of payments equilibrium, etc. These two major roles rarely conflict in the long run, but much of the art of. central b~nking lies in arranging a comfortable combination of the two in the short run. As you are painfully aware, the desired levels of these broader economic variables are not always mutually consistent,and trying to bring them into a common focus can produce a bad case of schizophrenia. A second related complexity involves the mix of judgment and rule, or art and science, or horse sense and quantification, or perhaps just p,ne. luck used in policy decision making. As the science of economics has progressed, so the art and SClence of central banking have These two themes--the plural objectives of central banking changed~ and the transition and the current status of central bank decision making--are the focus of my remarks today. The multiple role of central banking in the United States c an be traced historically from the century preceding creation of the Federal Reserve System through the evolution of the Federal Reserve to ita current position. During the 19th century, the gold standard mechanism. represented an automaticinstrUlnent, which may be thought of as the historical counterpart of modern Federal Reserve Open Market operations. Gold -5- flows had a domir-ant influence on the stock of high- powered money and thereby influenced national economic condi tions, but the flows were conditioned .b y market forces and the happenstance of mineral discoveries rather than by any focus on national goals. Moreover, it was left to the several states to address the problem of creating and rnaintaining orderly financial markets. . After the demise of the 1st and 2nd banks of the United . States, and even after the National · Banking Act, the only available tool was the Treasury's manipulation of its cash balance, which was largely ineffective in preserving orderly financial markets during periods of cash shortages and stringency. After the panic of 1907, concern for financial and economic stability led to establishmer-t of the Federal Reserve. The Systcm was designed to prevent financial panics and economic instability by establishing a discount luechanism that would provide an elastic supply of currency in response to seasonal or erratic needs. That is, the founding of the Federal Reserve gave the country a dependable lender-of-Iast-resort, prepared to lend at the discount rate. The level at which this rate was set was the Fed's principal policy instrument. During the early years of the Federal Reserve and the latter years of the gold standard, the price at which the Fed would lend--the discount rate--was based primarily on international considerations, especially on the flow of gold into and out of the country. However, the collapse of the gold standard system during and after World War I soon made gold flows irrelevant as a guide to policy action. The -6theory behind this approach proved faulty on other scores, as you well know. By 1923, the Federal Reserve recognized the necessity for replacement of this gold standard approach to policy determination, by then the tools included both the level of the discount rate and newly developed Open Market operations. As evidenced in the System's Tenth Annual Report, the new basis was that policy be set with the aim of influencing national economic conditions. Admittedly, Fed policy statements were strongly influenced by the "Real Bills Doctrine" during this period, but the point I wish to make is that policy was to be set on the basis of economic conditions. It was presumed the result would be best if the Fed monitored information from all sectors of the economy and then responded on the basis of the collective judgment of the policy maker s. The possibility of conflict between preserving order in particular fin~ncial markets (the original aim of the Federal Reserve) and pursuing other national economic goals--in particular, an adequate rate of growth of output--became a painful reality in the latter years of the 1920's. The System. was caught on the horns of the dilemma of restraint to dampen speculation in the stock market and stimulus to counteract sluggish business conditions. needs no comment. The outcome for the stock market and for the economy One result of the experience was the later legislation adding margin requirements to the tool kit of the System. In passing, , let me add that other regulatory changes have occurred that affect less visible, -7- but still important, goals of the System. For example, Regulation Q and the extension of reserve requirements to a broader class of liabilities have equity implications for banks and other financial institutions. As a result of the accumulation of regulatory devices since 1913, the Federal Reserve is now in a slightly better position to pursue multiple objectives because it can match tools and goals slightly better than it could by relying on any single mechanism. However, some of the additional regulations may in fact be neutralizing, leading to demands for their elimination and a return to simplicity. The depression era, between 1931 and 1941, was one that at least superficially would seem to have caused the System little difficulty in jointly pursuing the objectives of orderly marketsand other national economic goals. Perhaps we can be criticized for our lack of aggressive- ness in this period. However, by 1937 it appeared--probably mistakenly-- that disorder had a new guise, in the form of sloppy markets - - specifically, too many excess reserves in the banking system. Raising reserve require- ments was the obvious way to remove the problem, and this was probably a contributing factor in the ensuing recession of 1937-1938. During and after World War II, the Treasury's concern for floating debt at low cost was the basis for a ten-year period when the Federal Reserve temporarily forsook any continuing concern for other national goals and concentrated almost exclusively on maintaining stable price s and yields on U. S. Government Securities. With the exception of minor adjustments -8- in the support prices at which the System operated in the market, it was not until the "Accord" in 1951 that the Federal Reserve once more could actively pursue other national economic goals--as by then set forth in the Employment Act of 1946--with only an over-the-shoulder glance to assure itself that there were no disorderly market conditions to be corrected. Since 1952, the Federal Re serve has continued to pur sue a shifting combination of goals by a qualitative, judgmental decision process that has brought movements toward ease or reEitraint as dictated by whatever goals were being pur sued. Now, twenty years after the "Accord", the debate over an appropriate decision-making process still continues, both within the central bank and between central bankers and academicians. However, the debate has taken on a new aspect within the last decade; and I would now like to explore this new aspect in SOlne detail • . Until the 1960 1 s, the central bank had to use a marginalist decision process such as I have just alluded to, a process of Illeading against the wind." That is, if conditions called for stimulus, the System would o ease incrementally, wait, and see whether conditions were responding; if conditions were not judged to be responding adequately, it could then ease some more, if the behavior of related variables allowed further ease. But experience was beginning to point up the problem that the winds might be blowing in several different directions more or less simultaneously. Now, however, as a result of an outpouring of imaginative quantitative macro economic research, the System is being urged to adopt a fundam.entally -9different approach to decision making. Instead of moving incrementally "against the wind" - -and tornadoes require rapid shifts in position- -until the wind dies down and goals are achieved (or overshot), the System is urged to estimate the size of policy action required to get the economy back on 'track' and to move straightaway to take that action. Instead of deciding the direction in which incrernental policy steps should be taken, the System is urged to decide on the direction and Sluantity of action to achieve national economic goals. This urging is not the same as that of the proponents of a simple "rule". A "rule" is the antithesis of discretionary policy and requires a conviction that the economy would be more stable if policy were stable than if discretionary policy atterrJ.pted to offset instauility in uGler secturs, The more recent development does not envision abandoning discretionary policy, but substitutes a quantitative decision rule for the traditional marginal adjustment process. This distinction, between a marginal policy decision rule and a • quantitative policy decision rule, is perhaps less clear-cut than my simplified statement of it implies, primarily because the adjustrnent of financial and real business conditions to policy actions is not instantaneous, nor does it operate through mechanical1inkages. This is one reason why policy sometimes seems to have a tendency to "over shoot" cyclical turning points; by the time the turning point is recognized, a change in direction of policy is overdue--perhaps long overdue. Estilnates of the length -10of lag in the impact of monetary policy are crucial if timing of policy is to be improved. The econometric models that have been developed to provide such estimates for a marginal decision process may be the same models one would use· to estimate the quantity of policy actions needed under a quantitative decision rule o In addition, given that the effects of policy lag behind policy actions, there can be no guarantee that a marginal policy decision process will adjust in large enough increments to catch up with or modify rapidly changing economic conditions. For example, in a situation of rapid economic expansion, marginal upward adjustments in the rate of growth of money and credit may be smaller than required to accommodate the expcUlsion, and prematurely constrain the growth of income and output. Sonle quantitative estim.ate of policy action is . required to avoid this problem; therefore, an inte1Hgent marginal decision process must rely on econometric models. However, a useful distinction can still be made between a marginal decision process that.employs human judgment, inform.ed by models, and a quantitative decision process that employs by some human judgment. models, perhaps informed The thrust of much of the criticism of centr?.! banking--especially in the United States--suggests that the central bank should adopt some quantitative indicator--such as a measure of interest rates, bank reserves, or some other aggregate--and that policy action be designed to achieve the value of the adopted indicator because an econometric model sho",.. s that a particular goal variable can thereby be achieved-- -11frequently, the rate of growth of output. This does not seem to me to be an appropriate policy decision process--not because it is impossible to implement, but because it is inappropriate, for reasons I will spell out in some detail. How could the central bank implement an explicit quantitative decision rule? The modern theory of lI policy sci ence ll provides a useful guide that, with some heroic computational work, might be illustrated in the following steps: First, so we are told, we must specify our goals. Second, we must estimate our model. Third, we must solve the model for the values of instruments required to achieve our goals, subject to constraL'1.ts en the values of instruments and certain other econOlnic variables such as an orclerly market range for rates of change in interest rates. My point is that in attempting to follow this prescription, we encounter critical unresolved problems in the process. we seek? use? What instruments should we employ? What constraints should we impose? What goals should What models should we These problems form the basis of the reasons why I feel we cannot adopt a mechanical quantitative policy making process. Let's consider them one-at-a-time. Goal Specification. Our first problem is that some goals conflict and some are complementary. Occasionally it may be possible to achieve conflicting goals by using policy instruments with differential iInpacts -- .... -12- margin requirements tnight have been a valuable weapon in the late 1920 1 s -but, in general, monetary policy is a single policy weapon that cannot achieve conflicting goals. Hard choices or painful compromises must be made among · conflicting goals requiring sensitivity to the national mood. conflicting goals change over time. Trade-offs among For example, as economic conditions change, financial market participants may react differently to policy actions in the short-run; as fiscal or exchange rate policies change, the economy will respond differently to monetary policy actions. There would be an enormous research and cqmputational burden involved in estimating, and re-estimating every few weeks, the trade-offs among goals, among constraints and goals, among instruments, and I suppose, among models. cation of relationships and UnifOrlTI treatment of diverse events and conditi::ms. We do not know precisely who is helped and hurt by policy actions; what real and financial markets will respond, with what lag, to policy action. Given what we do know, as represented by our data collecting and econometric abilities, we cannot foresee events in sufficient detail to ignore impressions and judgments, unstructured reports and "feel" of current events for me to trust a quantitative decision process, simple or complex, exclusively. ever, Ido not wish to overemphasize this aspect of the complexity. How- Within the Federal Reserve System extensive policy simulation with econolnetric models takes place, and more will take place as time goes on. In addition, the wealth of competing models, goals, instruments, and constraints suggests the -13 fertility of this whole field for research by at:::ademic economists as well as by the central banking staff. I would hope that some of those in the audience will continue to join us in investigating how policy d.ecisions should be made. Model Estimation. The first problem here is that there are competing models from which we are unable to choose. There is probably broad agr eement in theory about the policy proces s: The instrmnents of policy-- Open Market operations, discount rate changes, etc. --influence a broad range of intermediate financial variables--interest rates, money supply, other credit aggregates--and these in turn are of policy--unemployrn.ent, prices, etc. rel~ted to the ultimate goals However, no model can be declared objectively !fright", because reality is incapable of perfect modeling, at least to date. Each model, in effect, calls for a different setting of policy instruments in order to achieve any given goal. Given the current state of statistical testing it is impossible unambiguously to choose the best model--that is, best in terms of its ability to foresee the future--from among a number of competing models. Thus, honorable men disagree, simply on the basis of o their predilection about the ~ priori structure of models, in their choice of a model appropriate to the central bank decision process. some critics urging us to use ~l Therefore, we find as a basis for our decisions; others prefer M2; others prefer interest rates; but none has a monopoly on the truth, and none has the sole responsibility of making the policy decision. A second pl~oblem here is that we know ~that instrument and goal is variable over time. models include error terms. the linkage between In econometric language, the By definition, one cannot predict the error -14- in the forecast of the next observation~ By experience, we know that there are errors in forecastjng, so we must make rational decisions--or at least decisions --under conditions of uncertainty. Decision making under conditions of uncertainty requires the policyrnaker to look at more than one variable--for example, both aggregates and interest rates--in order to build into a judgmental decision making process some feedback that p·rovides information about the current l' eliability of model forecasts. In a somewhat different vein, variability in the linkage between instrurnent and goal arises because models have built-in obsolescence. The structure of the economy and the preferences and behavior of individuals, change over time--not just in an erratic spread around some central tendency, but with a drift from old patterns to new. Models estimated on the basis of historical evidence cannot capture these influences. For example, inflation of the magnitude of the past five years is outside the realm of experience represented by most models of the U. S. economy_. (Hopefully, the new input or experience won't be applicable in the future.) of wage and price controls. The same is true Various statistical "ad hoceries" can be employed to try to adjust model forecasts for these major gaps, but essentially, human judgment must be relied upon to make policy decisions in an altered environment. Beyond these problems of choosing among competing goals and among competing models Witll uncertain linkages betwee~ instrument and goal, there are two further aspects of central bank decision making to be faced. The first recognizes a fund~mental fact of the institutional design of our -15- central banlc The fact is that,in typical American fashion, the Federal Reserve System embodies people of diverse views and diverse values. Even though each of us involved in policy decision making may choose goals, and a rnodel, and an instrument, and constraints, we inevitably disagree and, therefor£) must compl"omise our differences. Just as I do not wish to overemphasize the research and computational problems involved in quantifying policy choices - -lest future researchers be discouraged--so, too, I do not want to underemphasize this point, for whatever gains may be realized in measuring, modeling, simulating, and forecasting, this fact--that decisions are made by a group--will prevent--and rightly so--the implementation of a quantitative decision rule. (You can see that I'm basically opposed to technological change and the introduction of modern techniques if they threaten my position--a familiar theme ?) Finally, even if all participants in central ba.nk decision making could agree on goals, models, and constraints and could in some agreed upon way take account of the uncertainties of policy impact, there is a final matter that requires the exercise of human judgment, essentially in a subjective non-quantifiable framework. Monetary policy implementation is a continuous proces s in time. This is no different from the operation of the gold standard in the 19th century, when gold was mined and shipped, and therefor e injected as high-powered money continuously throughout any given year. Market intelligence and arbitrager s were able to smooth out the effects of most -16irregularities in these flows, although the loss of ships at sea was sOITletiITles a shock to the systeITl. SiITli1ar1y the Federal Reserve operates in the market for securities ' and bank reserves a1ITlost continuously, atteITlpting to offset the seasonal and irregular shocks to the financial systeITl that, prior to 1913, ITlight have generated stringencies and panics. Because of this responsibility for orderly ITlarkets, the Fed ITlust exercise judgITlent in ITlanipulating financial ITlarkets in pursuit of other national econom.ic goals, since these policy actions have the potential for being a ITlajor irregular shock to financial ITlarkets. It ITlay be iITlpoS sible to qucmtify the relation between policy action and ITlarket reaction, at least in the short run. Most existing econoITletric ITlodels are not sufficiently disaggr egated on the financial side even to begin to estimate t..~e relevant relations. I seriously doubt that there are stable relations to be estiITlated anyway, for short-run financial market • psychology--not just in the ITloney ITlarket but in the bond and stock and foreign exchange ITlarkets- -is an elusive factor. run that this factor ha~ It ITlay be true in the long no significance for real econoITlic activity so long I as orderly ITlarkets are ITlaintained. Nevertheless, sharp swings in ITlarket opinion have powerful real consequences on the fortunes of participants and on the long-run efficiency of the financial sector. Central banker s who exercise responsibility for orderly ITlarkets ITlust ITlake judgITlents about the ability of financial ITlarkets to accept policy actions in an orderly way. No ITlatter how thoroughly convinced one ITlay be about the desirability of achieving any particular rate of growth of SOITle monetary aggregate, -17- or level of interest rates, or other intcrlnediate or ultimate target of policy, the central bank must move continuously, and in small or large steps as the conditions of markets here and abroad permit. otherwise would be irresponsible. To do this wisely is often described as the Ilart" of central banking, and requires judgment. II To do Yet an overall grandmotherlytl concern for orderly markets sacrifices control of . . national economic goals; an overly stern disinterest in orderly markets may be destructive of the valuable resource of a highly developed financial system. The horns of a dilemma--a common problem that have had to face over a long period of time. central bankers