FEDERAL ItISERVE HANK OF NF.W YORK ö3

The Role of the In Cycles By RICHARD 0. DAvIs*

Most, if not quite all, arc agreed that the comes virtually impossible to distinguish their views from behavior of the quantity of money makes a significant those of the great majority of professional opinions. In kjjgercnce in the behavior of the —with "money" order to bring a few of the issues into sharper focus, this jsually defined to include in circulation plus article will take a look at some evidence for the "money rivate.demand deposits, but sometimes to include com- supply" view of business fluctuations in one of its more mercial time deposits as well.' Most economists, extreme forms. Without necessarily implying that all the or example, setting Out to forecast next year's gross following positions are held precisely as stated by any national product under the assumption that the money single , an extreme form of the upply would grow by 4 per , would probably want view can perhaps be characterized somewhat as follows: 'to revise their figures if they were to change this assump- The behavior of the rate of change of the money supply is tion to a 2 per cent decrease. the overriding determinant of fluctuations in business ac- In the past five to ten years, however, there has come tivity. spending, taxing policies, fluctuations into increasing prominence a group of economists who in the rate of technological innovation, and similar matters 'ould like to go considerably beyond the simple assertion have a relatively small or even negligible influence on the that the behavior of money is a significant factor inilu- -run course of business activity. Hence, to the extent ncing the behavior of tire economy. It is not easy to that it can control the money supply, a , such haractcrize with any precision the views of this group as the System, can control ups and downs of economists. As is perhaps to he expected where com- in business activity. The influence of money on business plex issues are involved, their statements about the impor- operates with a lung lag, however, and the timing of ance of monetary behavior in determining the course of the influence is highly variable and unpredictable. Thus usiness activity encompass a variety of individual posi- attempts to moderate fluctuations in business activity iOflS, positions which may themselves be undergoing by varying the rate of growth of the monc.y supply hange. Moreover these positions are rarely stated in arc likely to have air uncertain effect after an uncertain Uantitative terms. More frequently, the importance of lag. Thcy may even backfire, producing the very instability money as a determinant of business conditions will be they arc designed to cure. Consequently, the best policy haracterized as "by far the major factor", "the most im- for a central bank to follow is to maintain a steady rate of Ortant factor", "a primary factor", and by similar qualita- growth in the money supply, year in and year out, at a tIve ptrases inescapably open to various interpretations. rate which corresponds roughly to the growth in the Of course as one moves from the stronger phrases to economy's productive capacity. l'ie weaker, one comes closer and closer to the view that The implications of these views are obviously both is simply "a significant factor", at which point it be- highly important and strongly at variance with widely jfl1Oncy held beliefs. Thus they deny the direct importance of (except perhaps in so far as it may influence ), while they attribute to monetary policy a virtually determining role as business fluctua- Assistant Vice Prcsident, Rcsearch and Statistics. Fcctcral rcards eserve Bank New tions. At the same time, deny the usefulness of dis- // of York. they More rarely, other types of liquid such as mutual say- cretionary, countercyclical monetary policy. The issues S bank deposits are also includcd in thc definition of money. involved are highly complex and cannot possibly be ade- r

MONTHLY REVIEW, APRIL 1968

quately treated in their entirety in a single article.2 The provides major support for their views on the caus present article, therefore, confines itself to examining the importance of money in the . For the mo historical relationship between monetary cycles and cycles part, these economists have delineated cycles in the mo in general business.The article concludes that the relation- supply in terms of peaks and troughs in the percentage ship between these two kinds of cycles does not, in fact, rate of change of money (usually including time depOsits) provide any real support for the view that the behavior while cycles in business have been defined in terms of money is the predominant determinant of fluctuations peaks and troughs in the level of business activity in business activity. Moreover, the historical relationship marked off, for instance, by the so-called "reference cycles" between cycles in money and in business cannot be used of the National Bureau of Economic Research (NBER),3 to demonstrate that monetary policy is, in its effects, so They have argued that virtually without exception evefl delayed and so uncertain as to be an unsatisfactory cycle in the level of business activity over the past century countercyclical weapon. of experience can be associated with a cycle The first section shows how proponents of the money in the rate of growth of the money supply. The exceptions supply view have measured cycles in money and exam- that are observed occurred during and just after World ines the persistent tendency of turning points in monetary War 1I—although the events of 1966-67 may also be cycles, so measured, to lead turning points in general interpreted as an exception, since an apparent cyclical business activity. It argues that these leads do not neces- decline in monetary growth was not followed by a reces- sarily point to a predominant causal influence of money sion but only by a very brief slowdown in the rate of on business. A second section suggests that the cyclical business expansion.' The money supply school also finds relationship of money and business activity may be as that cycles in business activity have lagged behind the much a reflection of a reverse influence of business on corresponding cycles in the rate of growth of the money money as it is of a direct causal influence running from supply, with business peaks and troughs thus following money to business. A third section indicates why, for some peaks and troughs in the rate of monetary change. periods at least, the tendency for cycles in money to lead While the evidence supporting these generalizations is cycles in business may reflect nothing more than the im- derived from about a century of United States data, the pact on money of a countercyclicalmonetary policy. Next, nature of the measurements and some of the problems the relative amplitudes of monetary contractions and their associated business contractions are examined. Again it is argued that these relative amplitudes fail to provide any clear evidence for a predominant causal influence of 'Sec. for example, and Anna I. Schwartz, money. A fifth section examines the timing of turning Money and Business Cycles", Review of Econonvcs and Staliuics points in money and in business for evidence that the in- (February 1963, supplemcnt), pages 34-38. While the procedureof fluence with variable these economists in comparing percentage rates of growth of of money operates so long and a lag money with levels of business activity can certainly be defcnded, as to make countercyclical monetary policy ineffective. A it is by no means obvious that this is the most appropriate ap- proach, and thcre are many possible alternatives. Thus, for ex- final section suggests that there may well be better ways ample, cycles in the rate of growth of money could be compared to evaluate the causal influence of money on business than with cycles in the rate of growth, rather than the level, of business activity. For some purposes the choice among these al!ernativcs through the examination of past cyclical patterns. makes a considerable difference, as is noted later in connection with measuring the length of the lags of business-cycle turning points relative to turning points in the monetary cycle. CYCLESIN MONEY AND CYCLES IN 4 BUSINESS ACTIVITY Granting the difficulties of dating specific cycle turning points for series as erratic as the rate of growth of the money supply, a peak (for the definition of money that includes time deposits) As already implied,, proponents of the money supply seems to have occurred in October 1965, with a in October 1966. While there was a slowdown in the rate of growth of bitsi- school have argued that the historical relationship between ness activity in the first half of 1967, there was clearly no business cycles in money and cycles in general business activity cycle peak correspondingto the peak in the money series. Indeed, the current dollar of GNP moved ahead in the first tWO quarters of 1967, although at a reduced rate. The 1965-66 decline m the rate of growth in the money supply Was relatively short (twelve months). In amplitude it was clearly among the milder declines, but it was nevertheless still nearly twice as steep as the mildest of past contractions in thc rats of monetary growth (rio- 'Among the many interesting and relevant issues not discussed vember 1951 to September 1953). In any case, the 1965-66 de- arc the advantages and disadvantagesof the money supply as an cline does appear to represent a specific cycle contraction for the immediate target of monetary policy or as an Indicator of the rate of monetary change underthe standard NBER defInition. SCC effects of policy, the proper definition of the money supply, and Arthur F. Burns and Wesley C. Mitchell,Measuring Business CyCles the nature and stability of the . (National Bureau of Economic Research, 1946). pages 55-66.

L. FEDERAL RESERVEBANK OF NEW YORK 65

CI.,tp CHANGES IN MONEY SUPPLY PLUSTIME DEPOSITS Menth.to-month potcsnag.Change,, compoundoeiiuol rateS p.rcent PCV Cent 20 20

1947 45 49 50 51 52 53 54 SS 56 57 58 59 60 61 62 63 64 65 66 67 H.P.,Pstc.n'q ce.g., or. bo,ed en leesenolyod..Iod a.,.. Shaded or.., ..7r.;enhr.ceS,Onpsrd.. occordng to Notionol5,.teo,ot1inomIc R.,,or,h dtron.Iogy. Soon. leWd .4Gov.rnor, oJtho I.d...I Intone

• of interpretation can be illustrated from the postwar the chart. The erratic nature of the money scries, which experience represented in Chart 1. The chart shows partly reflects short-run shifts of deposits between Trea- monthly percentage changes in the money supply, defined sury and private accounts, does make the precise dating here to include currency in the hands of the public plus of peaks and troughs in the money series somewhat arbi- private demand and time deposits, on a trary. This introduces a corresponding degree of arbitrari- Seasonally adjusted daily average basis.5 The shaded areas ness in measuring timing relationships relative to turning represent periods of business as determined by the points in business activity. Waiving this difficulty, how- NBER. The first point to note is the highly erratic nature ever, peaks and troughs in the money series as dated in of month-to-month movements in the rate of change of one well-known study of the problem are marked on the the money supply. Indeed, the reader might be excused if chart for the 1947-60 period.6 As can be seen, each mone- he found it difficult to see any clear-cut cyclical pattern in

$ 6fle dates used are essentially those presented in Milton Fried- While, as noted, many analysts would prefer to define the man and Anna 3. Schwartz, op. ci:., page 37, Table I. Minor aloney supply to exclude commercialbank time dcosits, such an modifications of the Friedman-Schwartz dates have been made cl510 would not materially affect the general pLCtUrc, at least when these seemed obviously dictated by revisions in the data tot for the period illustrated by the chart. subsequentto publication of their work. 66 MONThLY REVIE\V, APRIL 1968

occurs the tary pcak during expansion phase of the busi- money supply economists to be highly suggestive of SUCh LWSS cycle and thus leads the peak in business. Similarly, an influence. Certainly the consistency with which the there is a monetiry trough marked during three of thc leads show up in cycle after cycle is rather striking four postwar acknowledged by the NBER. A does suggest that cycles in money and cycles in busine fourth monetary trough, however, in February 1960 are related by some mechanism, however loose and u. occurs somewhat before the onset of recession three reliable. Nevertheless, it is important to recognize ti1 months later. this mechanism need consist even not entirely or mainly The leads of the peaks in the money series with respect of a causal influence of money on business. It might, in. to the subsequent peaks in business activity are, it should stead, reflect principally a causal influence of business on be emphasized, quite variable, ranging from twenty money, or it could reflect a complex relationship of mutuaJ months to twenty-nine months for the period covered in interaction. As noted earlier, virtually all economists be. the chart and from six months to twenty-nine months licvc that there is, in fact, at least some causal influen for the entire 1870 to 1961 period. The corresponding of money on business, and it may be that this inhlue range of leads of money troughs relative to subsequent alone is enough to explain the existence of some degree ol • troughs in business cycles varies from three months to consistency, albeit a loose one, in the timing relationships twelve months for the charted period and up to twenty- of peaks and troughs in business and money. However,tht two months for the longer period. existence of a powerful reverse influence of the busine The significance, if any, of these leads in assessing the cycle itself on the monetary cycle would have important • importance of cycles in money in causing cycles in busi- implications. By helping to explain the timing relation. ness is highly problematical. Firstly, chronological leads ships of the money and business cycles, the existence of do not, of course, necessarily imply causation. It is per- such an influencewould certainly tend to question severely fectly possible, for example, to construct models of the any presumption that these timing relationships are them- economy in which money has no influence on business but selves evidence for money as the predominant cause ol which generate a consistent lead of peaks and troughs in businesscycles. the rate of growth of the money supply relative to peaks There are, in fact, a number of important ways in which and troughs in general business activity.7 Secondly, the ex- changing business conditions can affect, and apparently treme variability of the length of the leads would seem to have affected, the rate of growth of the money supply ov suggest, if anything, (he existence of factors other than the 100 years or so covered by the available data. First, thc money that can also exert an important influence on the state of business influences decisions by the monetary timing of business peaks and troughs. Certainly even if a authorities to supply reserves and to take other actions peak or trough in the rate of growth of the money supply likely to affect the money supply—as is discussed in deta1 could be identified around the time it occurred, this would in the next section. Business conditions can also have be of very little, if any, help in predicting the timing of a direct impact on the money supply, however. For example subsequent peak or trough in business activity. Thirdly, they may affect the balance of and the size d there is i real qucstion as to whether anything at all can or . These gold movements, in turn, be inferred from the historical record about the influence may affect the size of the —the sum of curS' of money on business if, as is argued in the next section, rency in the hands of the public and reserves in the banking there is an important reverse influence exerted by the system. Various official policies have tended to reduce businesscycle on the monetary cycle itself. offset this particular influence of business on money, bat at least prior to the creation of the Federal Reserve Systela. THE INFLUENCE OF BUSINESS ON MONEY it may have been of considerable significance. Second, business conditions may influence the jnofl Although the persistent tendency of cycles in monetary through an influenceon the volume of member bank growth rates to lead business activity does not, as noted, borrowings at the Federal Reserve. While the size of sb necessarily imply a predominant causal influence of money borrowings is, of course, importantly conditioned by the on business, this tendency has nevertheless seemed to the terms under which to member arc made, i eluding the level of the discount rate, it may also b significantly affected by the strength of demand and by the yields that banks can obtain on earning These in arc related in to the See James Tobin, "Moncy and Income: Post Hoc Propter matters, turn, clearly part Hoc?", to be published. state of business activity. j FEDERAL RESERVEBANK OF NEW YORK 67

• A: third influence of business on money operates varied over the near 100-year period for which data are rouáh the effects of business on the ratio of the public's available. In part, these variations have reflected the ef- boldin of and currency to its holdings of bank de- fects of the creation and evolution of the Federal Reserve A rise in this ratio, for example, tends to drain System. A detailed examination of the behavior of the serves from banks as the public withdraws coin and monetary base, the currency and reserve ratios, and the jrency. Since one dollar of reserves supports several role of business conditions in fixing their cyclical patterns •doilafS of deposits, the loss of reserves leads to a multiple is beyond the scope of this article. Recently, however, a tjaition of deposits which depresses the total money very thorough analysis of the problem has been done for the supply by more than it is increased through the rise in NBER by Professor Philip Cagan of Columbia Uni- •: public's holdings of . While no one is very sure versity. He finds that "although the cyclical behavior of :astoijust what determines the cyclical pattern of the cur- the three determinants [of the money stock] is not easy rency ratio, a pattern does seem to exist which in some way to interpret, it seems to conclude that most of their • gfleàts shifts in the composition of payments over the short-run variations are closely related to cyclical fluctua- businesS cycle as well as, in the historically important case tions in economic activity. . . . Such effects provide a of banking panics, fluctuations in the public's confidence plausible explanation of recurring cycles in the money I lathebanks themselves.8 stock whether or not the reverse effect occurred."9 A' final avenue of influence of business on money is The fact that the business cycle itself has an important :ojgh the influence of business conditions on the ratio role in determining the course of the monetary cycle se- of bank to deposits. When the ratio of riously undermines the argument that the timing relation- ecess reserves to deposits is relatively high, other things ships of monetary cycles and business cycles point to a ;:l, the money supply will be relatively low since banks dominant influence of money on business. By the same wlll.not be fully utilizing the -creating potential of token, ample room is left for the possibility that many • the supply of reserves available to them. Business condi- other factors, such as fiscal policy, fluctuations in business •.ôns can affect the reserve ratio in various ways. Thus demand, including those related to changes in can influence bank desires to hold excess reserves , fluctuation in exports, and replacement cycles tbroiigh variations in the strength of current and prospec- in consumer durable , may also exert important in- live loan demand, through variations in the yields on the dependent influences on the course of business activity. earning assets of banks, and through variations in peètations. When business is rising, loan demand is apt MONETARY POLICY AND THE libe strengthening, yields on earning assets are apt to be CYCLICAL BEHAVIOR OF MONEY iising, and banker confidence in the future is likely to be iitcreasing. Thus excess reserves are apt to decline, with One important, though perhaps indirect, influence of 4he reserve ratio rising and thereby exerting an upward business on money requires special mention, namely thc of • inflUenceon the money supply. influence it exerts via monetary policy. The relcvance The influence of business on money—acting through monetary policy to the behavior of monetary growth dur- influence on the growth of the monetary base, the cur- ing the business cycle was perhaps especially clear during • ency ratio, and the excess reserve ratio—is extremely the period beginning around 1952 and extending to the niplex and is not necessarily stable over time. The very early 1960's. In this period, policy was more or less Cyclical behavior of the monetary base and the cur- able to concentrate on the requirements of stabilizing the rCucy and reserve ratios have in fact varied from cycle to business cycle relatively (but not entirely) unimpeded by Cycle. Moreover the relative importance of these three considerations of war finance, the , l*Ctors in influencing the cyclical behavior of money has and possible strains on particular sectors of the markets. The ultimate aim of stabilizing the business cycle is, of course, to prevent or moderate recessions and to forestall or limit and structural imbalances during

L811 be noted that while the Federal Reserve has for many might routinely offset the reserve effects of short-term movements COin and currency, such as occur around holidays, for example, ratio of coin and currency in the hands of the public to deposits a Phillip Cagan,Determinants and Effectsof Changes in the Stock tt,apparently continued to show some mild fluctuations of a of Money. 1875-1960 (National Bureau of Economic Research, cheal nature. 1965), page 261. I

MONTHLY REVIEW,APRIL 1968

of advance. The tools available to the Federal ordinarily accelerate the growtn of demand for bank cr periods result such as and and deposits. This, in turn, would normally in up. Reserve, however, open operations on discount rate influence and the ward pressure on the and money markej policy, the in level only through complex and indirect routes. Hence, in rates. Maintaining stability money mark would the short run, policy must be formulated in terms of tone called for by such a policy require, however more variables which respond more directly to the influence of under the assumed circumstances, supplying reser, to offset the the System. Some possibilities include, in addition to the to the banks in order upward pressures with rate of growth of the money supply, the growth of bank money market rates. Thus, unchanged policy, conditions in the market and the behavior of acceleration in the rate of business expansion could gen. , money of short-term interest rates, and the marginal reserve crate an acceleration in the rate of growth reserVe, of banks as measured, for example, by the level of free re- and thence in the money supply. Similarly, a tapering.o in these serves or of member bank borrowings from the Federal in the rate of business expansion could, circu.; Reserve. It is clear that the money supply need not always stances, generate a tapering-off in the rate of monetay be the immediate of policy, and indeed expansion well before an absolute peak in business actjy., objective monetary that it was not by any means always such during the 1950's. ity occurred. It should be emphasized unchangej Given this fact, the behavior of the rate of growth of the monetary policy could be perfectly consistent with counter. under these conditions if the money supply during the period cannot be assumed to be cyclical objectives slowdowu of business advance either simply and directly the result of monetary policy decisions (or speedup) in the rate were. alone. expected to be temporary or were regarded as a healthy Nevertheless, it is clear that the current and prospective development. behavior of business influenced monetary policy The reaction of monetary policy to changing businesi: strongly to decisions, given the primary aim of moderating the cycle, conditions and the reaction of the money supply mona. and that these decisions, in turn, influenced the behavior tary policy undoubtedly help explain the tendency of: of the rate of growth of the money supply. Thus, for peaks and troughs in the rate of growth of the money as recoveries proceeded and threatened to gen- supply to precede peaks and troughs in the level of eco example, The erate inflationary pressures, monetary policy tightened to nomic activity during this period. resulting monctaly counteract these pressures. Regardless of what particular leads, however, cannot then be interpreted as dcmonstrat on in mone variable the System sought to control—whether the money ing a dependence of cycles in business cycles These leads would have existof supply itself, conditions in the money market, or bank mar- tary growth. very likely influence of on business were ginal reserve positions—themovement of any of these vari- even if the money altogether ables in the direction of tightening would, taken by itself, negligible. tend to exert a slowing influence on the rate of monetary expansion. In this way, the firming of monetary policy in SEVERITY OF CYCLICALMOVEMENTS the presence of cumulating expansionary forces would no doubt to the of the rate of monetary Apart from matters of cyclical timing, some proponent3 help explain tendency the growth to peak out well in advance of peaks in the busi- of the money supply school have also regarded rela of movcmdnts th ness cycle. Similarly, the easing of policy to counteract a tionship between the severity cyclical of associated movementl developing recession would help to produce an upturn in money and the severity cyclical the rate of monetary growth in advance of troughs in busi- in business as suggesting a predominant causal role for with some t1m ness activity. money. They argue, perhaps plausibility, In addition to the feedback from business conditions to if the behavior of money were the predominant deteT relative sizes of policy decisions and thence from policy to the money minantof business fluctuations, the cycliori supply, there are circumstances in which developments in movements in business and roughly contemporancoLil the can react on the even with cyclical movements in money should be highly correlatC economy money supply r8 monetary policy unchanged. Consider, for example, a For example, the severity of a cyclical decline in the situation in which the focus of policy is on maintaining of growth of the money supply should be closely relatcd an unchanged money market "tone"—a phrase that has to the severity of the associated business recession 01 been interpreted to imply, among other things, some rough depression. The evidence for such a correlation, howe' stabilization of the average level of certain short-term in- is actually rather mixed. terest rates such as the rate on . Now a Cyclical contractions in the monetary growth rate C8 decline in the rate of mo° speedup in the rate of growth in economic activity would be measured by computing the FEDERAL RESERVEBANK OF i'iEw YIK 69

from its value its value.10 On of the taiy growth peak to trough deposit component of the money supply. e basis of these computations, monetary contractions Developments of this type help to explain the associa- be ranked in order of sevcrity. Similarly, the seventy tion of major monetary contractions with major depres- business contractions can be ranked by choosing some sions but do not seem to account fully for jt.12 Thus it jndCx of business activity and computing its decline dur- may be that catastrophic monetary developmentsare in fact ing each business contraction recognized and dated by a pre-condition for catastrophic declines in business activ- c NBER. If the resulting rankings of monetary contrac- ity. In any case, for more moderate cyclical movements, 'ions are compared with the rankings of their associated the association between the severity of monetary contrac- business declines for eighteen nonwar business contrac- tions and the severity of business contractions breaks down ijons from 1882 to 1961, the size of monetary and completely. There is virtually no correlation whatever be- business contractions proves to be moderately highly tween the relative rankings of the twelve nonmajor con- correIated.1 It turns out, however, that this correlation tractions in the 1882-1961 period and the rankings of depends entirely on the experience of especially severe the associated declines in the rate of monetary growth. cyclical contractions. Among the eighteen business con- Certainly this finding does not support the theory that tractions experienced during the period, six are generally changes in the rate of monetary growth are of predominant recognized as having been particularly deep. They include importance in determining business activity. three pre-Worid War I episodes and the contractions of and 1937-38. In the latter three de- 1920-21, 1929-33, MEASURINGLAGS IN THE INFLUENCE OF dines, the Federal Reserve Board's industrial production MONEY ON BUSINESS fell by 32 per cent, 52 per cent, and 32 per cent, espectiveIy, compared with a decline of only 18 per cent Despite their belief in the crucial role of the money for the next largest contraction covered by the production supply in determining the cyclical course of business activ- index (1923-24). ity, some members of the money supply school neverthe- These six most severe contractions were in fact asso- less argue, as suggested at the beginning of this article, that dated with the six most severe cyclical declines in the rate discretionary monetary policy is aclumsy and even danger- ofgrowth of the money supply, though the rankings within ous countercyclical weapon. The starting point for this the sx do not correspond exactly. As was argued earlier, view is again the fact that peaks and troughs in the level business conditions themselves exert a reverse influence of business activity tend to lag behind peaks and troughs on the money supply, and it seems probable that partic- in the rate of change of the money supply—in particular ularly severe business declines may tend to accentuate the fact that these lags have tended to be quite long on the accompanyingmonetary contractions. Thus, for exam- average and highly variable from one cycle to another. ple, the wholesale of loans and sharp drops in the Thus long average lags of about sixteen months for value of securities that accompanied the 1929-33 depres- peaks and twelve months for troughs have suggested to sIon helped lay the groundwork for the widespread bank these economists that the impact of monetary policy is failures of that period. These failures were in part caused correspondingly delayed, with actions taken to moderate by, but also further encouraged, largc withdrawals of cur- a boom, for example, having their primary impact during rency from the banking system by a frightcned public. By the subsequent recession when precisely the opposite in- Contracting the reserve base of the banking system, in fluence is needed. Moreover, the great variability from tUrn, these withdrawals resulted in multiple contractions cycle to cycle of the lags as measured by the money supply school has suggested that the timing of the impact of monetary policy is similarly variable and unpredictable. For this reason, they argue, it will be impossible for the monetary authorities to gauge when their policy actions

'°Generally, three-month averages centered on the specific cycle t5roing point months have been used to reduce the weight given tOespecially sharp changes in the pcak and trough months them- hcs. 11The Spearman rank correlation, for which satisfactory signifi- flce tests apparently do not exist when medium.sized samples IS Sec Phillip Cagan, op. cli., pages 262-68.

will take effect and therefore whether these actions will The fact that such lags do exist, however, shows only turn out to have been appropriate. that monetary policy cannot be expected to produce ilnnle. it is true, of course, that monetary policy affects the diate results. Like fiscal policy, its effectiveness depe to business trends economy with a lag. The full effects of pur- in part on the ability anticipate so H wili be chases on bank deposits and credit, for example, require that policy actions taken today appropriate t the time to work themselves out. More important, additional tomorrow's conditions. Of course the longer lags j to the further out in time must elapse before businessmen and consumers ad- the effectsof policy prove be, time carried and the is just their spending plans to the resulting changes in the must such anticipations be greater the financial environment. For this reason, the pattern of risk that policy actions will prove to be inappropriate. of the are variable spending at any given time will to some degree reflect Moreover, if the lengths lags highly and the influence of financial conditions as they existed several thus perhaps unpredictable, the risks of inappropriate increased and the need months or quarters earlier. Hence it is certainly possible, policy decisions are obviously for is to arise. for example, that some of the effects of a restrictive mone- continuous adjustments in policy apt could continue to be felt during a recession The timing of cycles in money and cycles in business, tary policy for even the current posture of monetary policy were however, provides absolutely no basis believing that though of are so quite expansionary. the lags in the effects monetary policy tong o

Chari U I. AND IN MONEY SUPPLY PLUSTiME DEPOSITS I. CHANGESIN GROSS NATIONAL PRODUCT Ou,rtci.$o.qoortsrpucinIcga dtanei;cc mpcwrd oieauol rot.. P Per c.nI P., caM 20]

: I

1947 48 49 Nope: Percentage change. or. baud on ..eson&iyadju,l.d date. Sourr•si BeardclOt enroot, o! th,Federal Rotor..Syir.njUeitsd Sta.,D.pcrtr..nr ci Cemmer,.

-j FEDERAL RESERVE RANK OF NEW YORK 71

variable as to vitiate the effectiveness of a counter- The point of these various comparisons is not to prove for that cyclical policy. First, there are many reasons doubting the lag in monetary policy is necessarily either very that the lag in the effects of monetary policy should be long or very short, but rather to illustrate how hard it is measured by comparing the timing relationships between to settle the matter through the kind of evidence that has cyclical turns in money and in business. It has been been offered by the money supply school. Similar difficul- argued, for example, that other variables more directly ties, as well as others, beset attempts to measure the H under the control of policy makcrs, such as member bank variability of the lag in the influence of money on business onborrowcd reserves, or variables more clearly related to by comparrsons of cyclical peaks and troughs in the two. business decisions, such as interest rates, must also be However the turning points are measured, the resulting taken into account. Yet, even if the behavior of the money estimates may seriously overstate the true variability of the supply be accepted as the indicator of policy, there are many lag in the influence of money on business. The reason is alternative ways in which "the lag" between monetary and that observed differences from cycle to cycle in the timing business behavior can be measured, and it makes a great of turning points in money relative to turning points in deal of difference which measure is used. If, for example, business are bound to reflect a number of factors over therate of change in the money supply is replaced by and beyond any variability in the influence of money on deviations in the level of the money supply from its long- business.'7 'These "other" sources of variability include run trend, the average lag between monetary peaks so purely statistical matters such as errors in the data and measured and peaks in general business apparently shrinks the arbitrariness involved in assigning precise dates to from the sixteen months previously cited to a mere five turning points in money and in business. More funda- that I months.14 Alternatively, it can be plausibly argued mentally, the fact that there exists a reverse influence of the appropriate measure is the lag between the rate of business on money, an influence that is probably uneven change in the money supply, and the rate 01 change. from one cycle to the next, imparts a potentially serious • tather than the level, of some measure of business activity source of variability to the observed lags. Moreover, if such as gross national product (GNP) or industrial pro- there arc important influenceson the general level of busi- duction. When peaks and troughs for money and business ness activity other than the bchavior of money, these • arc compared on this basis, the lead of money over busi- factors would also increase the variability of the observed qess appears to be quite short.'5 The near simultaneity, timing relationships between turning points in money and inmost cases, of peaks and troughs in the rates of change in business. Taking all these possibilities into account, it •Ofithe money supply and of GNP during the post-World seems fair to say that whatever the true variability in the War II period can be seen in Chart II. To be sure, move- impact of money on business, its size is ovcrstated when ments in the two series are quite irregular, so that the deci- it is measured in tcrms of the variability of the lags in sion on whether to treat a particular date as a turning cyclical turning points. point is sometimes rather arbitrary. Nevertheless, the lead of peaks and troughs in the rate of growth of money over WAYS IN WHICH MONEY MAY INFLUENCE peaks and troughs in the rate of growth of GNP appears DUSINESS to avcragc about one quarter or less.'° If there is a broad conclusion to be drawn from a study of the historical pattern of relationships between cycles in money and cycles in business, it is that there are distinct limits to what can be learned about the influence of money on business from this kind of statistical 'This estimate is presented by Milton Friedman in "The Lag analysis. Perhaps Effect in MonetaryPolicy", Journal of , Octo- this should not be the business cycle ber surprising. During 196!, page 456. many factors of potential importance to the subsequent 'See John Kareken and , "Lags in Monetary behavior of business activity undergo more or less con- Pokey", Stabilization Policies (Commission on Money and Credit, 1963), pages 21-24. 'When quarterly dollar changes in the money supply are cor- felled with quarterly dollar changes in GNP experimenting with flOus lags, the highest correlation is achieved with GNP lagged tWO quarters behind money. (For the 1947-U to 1967.111 period W5 RI a is .34.) The correlation— with one quarter lag is almost "Other sources of variability are discussed in some detail by 4c1ly as high, however (R .33). When percentage changes in Thomas Mayer in "The Lag in the Effect of Monetary Policy: two series are used instead, the correlation Some Western Economic Journal no virtually disappears, Criticisms", (September 1967), matter what lag is used. pages 335-42. 72 MONTHLY REVIEW,APRIL 1968

tinuous change. At the same time the business cycle itself is directly changed by the transaction. Similarly, wh feeds back on the behavior of these factors. Hence it is banks expand deposits by making loans, the moneta extrcmcly difficult to isolate the importance of any single assets of the borrowers rise, but their liabilities to the factor, such as the behavior of money, and post hoc, banking system rise by an equal amount and their ne propter hoc reasoning becomes especially dangerous. In worth and income are unchanged. these circumstances there appears to be no substitute for Since these substitution effects associated with opce a detailed, and hopefully quantitative, examination of the market operations and with the expansion of bank dc. ways in which changes in the money supply might work posits are by far the most important operations by which through the economy ultimately to affect the various com- the money supply is changed, it seems especially relevant ponents of . Some brief and tentative to study the ways in which these effects may influence sketches aside, the proponents of the monetary school economic activity. The main avenues appear to be through have not attempted such an analysis. changes in interest rates on the various types of assets The possible ways in which an increase, for example, and changes in the availability of credit. When the Federaj in the money supply might stimulate aggregate demand Reserve or the commercial banks buy securities from the can be separated into what arc sometimes called "income nonbank public in exchange for deposits, funds are made effects", " effects", and "substitution effects". In- available for the public to purchase, in turn, a wide va- come effects exist when the same developments that pro- riety of private securities such as mortgages, corporate duce an increase in the quantity of money also add di- bonds, or bankers' acceptances.2° The increased demand rectly to current income. Examples would be increases in for these securities tends to push rates on them down. And and deposits resulting from domestically with borrowing costs down, business firms may be induced mined gold or an surplus. Similarly, a wealth effect to expand outlays on plant and equipment or occurs when a process increasing the money supply also while consumers may increase spending on new homes. In i increases the worth of the private sector of the econ- most cases, the effects of lower interest rates on capital omy. A Treasury deficit financed by a rundown of Trea- spending probably stem from the fact that external financ- sury deposit balances might be regarded as an example of ing has become cheaper. In some cases, however, lowCr such a process, since the resulting buildup of private de- market yields on outstanding government and private posits would represent an increase in private wealth. securities might induce business holders to sell such assets Far more important than the income or wealth effects in order to purchase higher yielding capital goods and in the present-day United States economy are substitution thus, in effect, to make direct substitution of physicl effects such as result when the Federal Reserve engages capital for financial assets in their "portfolios". Finally, in open market operations and banks expand loans and lower interest rates on securities may reduce consumel .'9 When the Federal Reserve buys Govern- incentives to acquire and hold financial assets while tempt- ment securities from the nonbank public, the public of ing them to make more use of consumer credit, thereby course acquires deposits and gives up the securities. There reducing out of current income and increasing con- is no direct change in the public's net worth position,19 or in its income; rather there is a substitution of money for securities in the public's balance sheet. The same is true when the banks expand the money supply by buying se- curities from the nonbank public: the public substitutes money for securities, but neither its wealth nor its income ° The newly creatcd dcposits may of course in principle be used immediately to buy goods rather than financial assets, thus tending directly to stimulate business activity. Even in this caSC, however, the effects of the money-creating operations work through and depend upon reactions to interest rates. When the Federal Reserve or the commercialbanks enter the market to ('UI securities, their bids add to total market demand, making markel for securities higher (and yields lower) than they otherws$C effects would have been. Indeed it is these relatively higher prices (loWCt - 'Tb substitution are sometimes also known as "port- Is folio balance" or "liquidity"effects. ,) that induce the nonbank public to give up securities exchange for deposits. If the deposits are in fact immediatelyusCd ID This statement has to be modifiedto the extent that the Fed- to purchase goods, then the process can be regarded as one eral Reserve's buying activity bids up the market value of the which lower market interest rates on securities stemming troll' public's holdings of Government securities. The significance of bids by the Federal Reserve or the commercial banks have induced this wealth effect is probably minimal and is further limited In Its the public to give up securities in exchange for goods. Thc extdt consequences by the tendency of many holders to value Govern- to which such switching will occur obviously depends upon th ments at original purchase price or at par rather than at current sensitivityto interest rdtes of business and consumer demands f0 market value. goods. FEDERAL RESERVE BANK OF NEW YORK 73

sumption purchases. of interviews of businessmen and consumers with regard With regardto bank lending, open market purchases of to the influence of credit cost and availability conditions Government securities increase bank reserves and may on their spending decisions. Other studies have employed ease the terms on which banks are willing to make loans. modem statistical and computer technology in an attempt Changes in tending terms other than interest rates, which to extract such information from data on past behavior.22 include repayment procedures, compensating balance re- With regard to spending on housing, there has been gen- quirernentS and the maximum amount a bank is willing eral agreement that the cost and availability of credit are to lend to a borrowcr of given credit standing, are often highly important. A number of studies have also found bracketedas changes in "credit availability". Such changes varying degrees of influence on business spending for H arc regarded by many analysts as being more important plant and equipment and for as well as on influences on many types of spending than are changes in consumer spending for durable goods such as autos and interest rates. Morcover, changes in credit availability re- appliances. All these studies, however, have also found lated in part to changes in the money supply are not con- factors other than cost and availability of credit to be flnd to lending by commercial banks, as was dramati- highly important. Moreover, a large degree of disagree- cajly illustrated in 1966 with regard to nonbank mortgage ment exists with regard to the exact quantitative impor- lenders. In any case, an increased availability of funds tance of the financialfactors. permits and encourages potential borrowers to increase Given the serious technical problems that surround these their loan liabilities, thereby providing funds which can be studies, major areas of disagreement arc virtually certain used to build up financial assets (pcrhaps mainly money to exist for some time to come. Nevcrtheless, studies of market instruments) or to purchase physical assets in the the type referred to here appear to offer the hope at least form of business capital goods, inventories, or consumer that firmly grounded and widely accepted conclusions on .durables. Stepped-up purchases of financial assets add to the importance of money in the business cycle may ulti- downward pressures on interest rates, stimulating spend- mately be reached. Of particular interest are large-scale ing through the processes already described, while addi- econometric models which attempt to provide quantitative tional demand for physical assets stimulates business estimates of the timing and magnitude of the effects of activity directly. central bank actions on the money supply and other finan- Studies of the influence of changes in interest rates and cial magnitudes and the subsequent effects, in turn, of these thà availabilityof credit on spending in the various sectors variables on each of the various major components of of.thc economy have appeared with increasing frequency aggregate demand. One such model is currently under in the post-World War 11 period, especially within the construction by members of the Federal Reserve Board past few years. Some of these studies have taken the form staff in cooperation with members of the Economics De- partment of the Massachusetts Institute of Technology.2' Granting the major technical problems still unresolved, projects of this kind appear promising as a means of eventually tracking down the importance of money in ex- •L White there is little general agreement that such direct effects plicit, quantitative terms. on consumption are important, a recentstudy of theproblcm has In fact found a significantinflucnce of interest rateson consumer de- mand for automobiles and other durables. (See Michael J. Ham- burger, "Interest Rates and the Demand for Consumer Durable Goods", American Economic Review, December 1967.) In general, proponents of the monetary school feel that analyses of the role of tfltrt( rates in consumerdemand undertaken to date have ne- For of glected ' a summary some of these studies, see Michael I. Ham- to take irno accountcertain important factors. In particular, burger, "The Imlact of Monetar,y Variables: A Selected Surveyof theythink that the most relevantinterest rates may not be the ones the Recent Empirical Literature (Federal Reserve Bank of New Usuallystudied, namely the rateson financial instruments, but rather the York, July 1967). Copies of this paper are available on request Interest rates "implicit" in the pricesof the durable goods them- from PublicationsServices, Division of Administrative Services, where the the services a consumer selves..Le., a1ue of yielded by Board of Governors of the Federal Reserve System, Washington, durable, such as an auto or a washing , is treated as D.C. 20551. Snalogous to the coupon or dividend yielded by a or stock. The obvious difficulties of defining and measuring the value of 23Some preliminary results of this work are discussed in 'Ilic such services have probably been responsible for the notable Federal Reserve-MITEconometric Model by Frank deLeeuw and dearth of research into this possibility, however, and the issue Edward Gramlich, Federal Reserve Bulletin (January 1968), IlIUM be regarded as completelyunsettled. pages 9.40.