1927150

8/13/2019 1927150

1/30

The Monetary Mechanism and Its Interaction with Real PhenomenaAuthor(s): Franco ModiglianiSource: The Review of Economics and Statistics, Vol. 45, No. 1, Part 2, Supplement (Feb.,1963), pp. 79-107Published by: The MIT PressStable URL: http://www.jstor.org/stable/1927150.

Accessed: 04/01/2014 17:29

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

The MIT Pressis collaborating with JSTOR to digitize, preserve and extend access to The Review of

Economics and Statistics.

http://www.jstor.org
http://www.jstor.org/action/showPublisher?publisherCode=mitpresshttp://www.jstor.org/stable/1927150?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/1927150?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=mitpress

8/13/2019 1927150

2/30

THE MONETARY MECHANISM AND ITS INTERACTIONWITH REAL PHENOMENA

Franco Modigliani, Massachusetts Institute of TechnologyHIS paper races he majordevelopmentsof lasting value in our understanding ofthe monetary mechanism and its role in theeconomy which have occurred since the earlyforties, when the process of digesting the Gen-eral Theory [ii] and integrating it with theearlier streams of thinking had been more orless completed. I consider here primarily thestate of monetary thought as of about themid-fifties, leaving to another time considera-tion of the evolution that has occurred since.The middle of the I950's provides, in myview, a useful landmark in reviewing theevolution of monetary and macroeconomictheory, since the developments that occurredup to that time are, on the whole, rather dif-ferent in purpose and approach from thosethat have occurred later. The development inthe first-mentionedperiod seems to me to con-sist largely of refinements, clarifications, anddevelopments of the basic framework, whichhad already been laid out by the early forties.The most significant contributionsin the more

recent period, on the other hand, have tendedto approach monetary issues in terms of thetheory of asset management and portfolio de-cisions, exploiting concurrent advances in thetheory of saving, in managerial economics and,most importantly,in the theory of choice underuncertainty. These developments, on thewhole, have not led to any major revision orrejection of the positions reached by the mid-fifties, but have rather been concerned withproviding a better understanding of the de-terminants of the demand for money and itsrelation with the demand and supply forvarious other assets, physical as well as fi-nancial, by final transactors and by "financialintermediaries."This paper is, in essence, a summary of alonger and more rigorous statement which Iexpect to complete later, and which, togetherwith an exploration of the developments sincethe I950's, will be published elsewhere whencompleted. Because it is in the nature of a

summary,I have frequently found it necessaryto sacrifice rigor and to omit nearly all of theproofs and many of the references.The material is divided into six major sec-tions. The first is a brief comparison of abasic model of money and the economy whichI believe would have been widely acceptedabout I944, with a correspondingmodel as ofthe middle of the 1950's. The following sec-tions, then, examine in some detail the im-plications of the mid-fifties model for the rela-tionship to crucial variables in the economy,and for major lines of monetary and fiscalpolicy.

The Mid-50'sModeland Its Relationto the 1944ModelI hope I may be permitted the liberty ofusing the macroeconomic model presented inmy I944 article, "Liquidity Preference and theTheory of Interest in Money" [I3], as a rep-resentation of where monetary economiststhought money stood in relation to the econ-omy at about that period. This model is repro-duced in Table i (with very minor modifica-tions of notation), together with a revisedversion labeled the "Mid-so's model." Thisis essentially the model that I would have usedhad I been writing a comparablearticle at thattime (and did actually use in my class lec-tures, cf. [I5]).While no systematic attempt is made hereat justifying in detail the equations of themodel, a task which was substantially per-formed in the I944 paper, in this section wewill review the main differences as well assimilarities between the original and the re-vised model.The two models are basically identical inspirit: they both treat the economy at a highlyaggregative level in terms of four goods andcorrespondingmarkets. Two of the goods arephysical commodities, labor and real output,

[ 791

This content downloaded from 194.154.200.98 on Sat, 4 Jan 2014 17:29:33 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp

8/13/2019 1927150

3/30

8o THE MONETARY MECHANISMand two are money-fixed claims, money andbonds. For the sake of concreteness, real out-put is visualized as consisting of a singlehomogeneous commodity which, in the mid-fifty model, I choose to label the MM (readmum). This output can be either used for cur-rent consumption (C) and thereby disappears,or can be instead devoted to investment (I),thereby becoming part of the capital stock(K). Thus,X, C, I, S are measured n MM/time,and K in MM's. Similarly, labor (N) is meas-ured in man-hours/time. The two money-fixedclaims on the other hand are measuredin unitsof money assumed to be the dollar, and bondsare to be regardedas one-period oans or claimsto future (next-period) money. The amountof bonds held by a transactor may be positiveor negative; in the latter case it representsthetransactor'sdebt.

Corresponding o the four commoditiesthereare three independentprices or terms of tradewith money: the price of output - or generalprice level - P ($/MM); the price of labor,W ($/man-hour); and the price of a dollar,next period, (i/i + r), where r is the rate ofinterest. In addition, both models, at least atthe outset, assume:A. i: certainty

A.2: absence of money illusion (which is animplication of rational behavior)A.3: unit elasticity of price expectations(and independence of interest rate ex-pectations from current prices)The main differencesbetween the two modelscan be summarizedunder five headings:i. Explicit reliance on a general equilibriumformulation. The mid-fifty model is explicitlystructured in terms of markets. one for each

TABLE I.COMPARISON OF I944 WITH Mm-50's MODEL

DEFINITIONS AND CLASSIFICATION OF VARIABLESI. Endogenous variables(a) Flow variables: (i) X, Real income; (2) C, Real consumption; (3) I, Real investment; (4) S, Real saving; (5)

Xd, Real aggregate demand; (6) N, Employment (man-hours per unit of time); (7) N8, Labor supply; (8)Y, Money income(b) Price variables: (g) P, Price of output; (iO) W, Price of labor (wage rate); (ii) r, Rate of interest(c) Money market variables: (I2) Md, Demand for money by private sector; (I3) BP, Demand for bonds by non-bank private sector; (i4) Bb, Demand for bonds by banks; (I5) Mb, Supply of bank moneyII. "Initial conditions": K, Real stock of capital; vi, Net worth of jth household; V = z vJ, Aggregate net worth ofprivate sector; [V] = (vi, v2, . . v?,

III. Other parameters: WO, Rigid money wage rate; N', "Full employment" labor supply; M, Money supply; M*,Supply of government money; M*P, Government money outside banks; M*b = M* - M*P, Governmentmoney held by banks; G, Supply of government bonds.MODEL I (I944) MODEL II (mid-50's)(3) PS = S(r, Y) (I) C = C(X,NW/P,r, [VO/P])

(2) PI= I(r, Y) (2) I= I(r,X,Ko)(6) X = X(N) (3) x-C+I(7) XN(N) = WIP (4a) X = X(N, Ko) or (4b) X = s (P/W,Ko)W = a Wo + (i - a) F(N) X P (5a) XN(N, K = W/P or (5b) N = nd(W/P,Ko)rif N < N6 N_ n>(WP), if nd(WoP,Ko) > n8(WO/P)a o, otherwise I6 = W0 if nd(WolP,Ko)

8/13/2019 1927150

4/30

THE MONETARY MECHANISM 8icommodity,with each market in turn describedby (a) supply conditions, (b) demand condi-tions, and (c) clearing of market or equilibriumconditions, of which one is redundant (Walras'law).

The main advantage of the general equilib-rium framework is that it insures a systematicand, at least initially, symmetricaltreatment olall markets.' For the commodity market, thedemand conditions are described by equations(I) to (3); the supply conditions by (4b) andthe clearing conditions by (7). In the labormarket the supply is given by (6), to be re-viewedmoreclosely below; the demand by (5);and market clearing by (8). The remainingtwo marketsare describedunder the next head-ing, 2.2. Explicit treatmentof the bondmarket andmore precise formulation of the relation be-tween the demand and supply of money andbonds and the banking system. In the I944model, the money market was described verysummarily by a single equation (I.I) and thebond market was omitted altogether. In themid-I95o's model, the money market is agairdescribed by a demand (M.2), a supply (M.4)and a clearing of market equation (M.3)Furthermore,two sources of money supply tothe private sector are recognized: bank money(Mb) and governmentmoney (M*P). The boncmarket is describedby the last three equations(M.5) is a streamlined consolidated balancesheet of the bankingsystem (including the cen-tral bank, if any); the left-hand side is th(liability side or bank money supply, which musibe equal to the asset side consisting of govern-ment money (M*b) and bank credit (B b)which can also be regardedas the banking sys-tem demand for bonds. (M.6) gives the net

t demand for bonds by the p uitd lmbetween lendiniLE97s dmi---)-aordrfw"-n-g'(gross sujpplyj (M7 1ST c&eirfngObf'ara%kelcondition; the left-hand side is the aggregatenet demandby the private sector and the right-

hand side is the net supply of bonds to theprivate sector, which would be zero in the ab-sence of government debt, and is otherwiseequal to this debt, G.While the bond market is thus given explicittreatment, it is still permissible to treat thismarket as the redundantone and we shall findit convenient to do so. Furthermore, for thepurpose of this paper, we shall regard the twocomponents of the money supply and hencetheir sum M as exogenously determinedby thegovernment or through the monetary author-ity's control over the banking system, or both,without concerning ourselves with the manypossible institutional devices through whichthis control can be exercised. Under these con-ditions it is found that the last four equations,(M.4) to (M.7) describing the banking systemand the bond market can be disregarded, solong as one is interested in the functioning ofthe rest of the system. This is because the firsttwelve equations, consisting of the nine "realequations" (i) to (9) and the three monetaryequations (M.I) to (M.3), form a determinedsubsystem in the first twelve variables listed atthe top of Table i, namely the eight flowvariables, the three price variables, and thedemand for money. The analysis that follows,therefore, is primarilyconcernedwith this sub-system.3. Improvements in the consumptionand in-vestment function and in particular more ade-quate recognition of the role of stocks. Sincethe mid-forties the consumption function hasreceived a great deal of attention at both thetheoretical and the empirical levels. Many ofthe contributions,and notably the "permanentincome" hypothesis of Friedman [7] and the"life cycle hypothesis" of Modigliani, Brum-berg and Ando [ I4] [ i6] [ i ], have de-empha-sized the role of currentincome and emphasizedthe role of other variables, particularly long-term incomeexpectationsand wealth. The con-sumption function (i) is sufficientlygeneral tobe consistent both with the original simple-minded Keynesian version and with the morerecent but still controversial formulations. Inparticular, it is consistent with the life-cyclehypothesis according to which aggregate con-sumption can be approximated by a linearhomogeneous function of aggregate labor in-

1Without the benefit of this crutch, in my I944 paper Iwas led in the formal model of section 2, to drop anyexplicit reference to the bond market - implicitly treatedas the redundant one - which, in turn, caused some unfor-tunate and unnecessary misunderstandings [9]. More ser-iously, I forgot altogether the existence of this commodityin the analysis of section I3, which resulted in some outrightwrong statements, as was shown later by Patinkin [20],[2I].


8/13/2019 1927150

5/30

82 THE MONETARY MECHANISMNVWcome p and net worth (V,/P), with coeffi-

cients depending in principle on the rate of re-turn on capital (r) [ i ].Similarly, the investment function, or invest-ment component of aggregate demand (2) canbe shown to be consistent with a wide varietyof approaches and hypotheses, such as thecrude acceleration principle, the so-called "flex-ible accelerator"or gradual adjustment hypoth-esis, or even the more traditional formulationrelying upon the production function and themarginal productivity of capital. Every oneof these approaches implies that investmentdemand must be an increasing function of thelevel of output and a decreasing (or at leastnonincreasing) function of the initial stock ofcapital and of the cost of capital.4. Correction of the faulty, formulation ofthe homogeneityproperties of the consumption,investment, and demand for money functions.The new formulation of equations (I) and (2 )and of (M.2) implies that the real demand forconsumption, nvestment,andmoneyis homoge-neous of zero degree in money income, wealth,and prices,or the correspondingmoney demandis homogeneousof first degree in the same threevariables. This property is an implication ofthe assumption of rational behavior. In theI944 model I had intended to make this sameassumption but in fact did not formulate itproperly; I assumed instead that the threemoney demands were homogeneous of firstdegree in money income alone. This formula-tion leads to the incorrect implication that achange in money income has the same ef-fect on money consumption whether due toa change in real income with prices constant,or to a change in prices with real income con-stant; and similarly for the other two variables.This error in turn led to a peculiar and objec-tionable property of the model, to wit that thefirst four equations of the model form a closedsubsystem in the four variables (PX) (or Y),PS, PI and r, involving in particular M as aparameter. This dichotomization implies thatthe equilibrium values of the rate of interestand the money flow variables are independentof the real variables of the system, and in par-ticular of the form of the production function

and of the level at which the rigid money wageWO s set. This ceases to be true once the con-sumption and investment function are proper-ly 2 formulated as in Model II. In particular,as shown below, for a given production func-tion (including in it the initial stock of capital),and given M, an increase in WOwill normallytend to result in an increase of r, as well as ofP, money income, investmentand consumption,and in a fall in employment, real income andthe other real flows.35. Use of a more convenient and effectivedevice for expressing the hypothesis of wagerigidity. This device embodied in equation (6)relies on the notion of a "potential" supplyfunction ns(W/P), expressing the maximumsupply of labor available at each real wage. Bycombiningthis function with the demand func-tion we can determine the excess supply (ordemand) at any real wage. The hypothesis ofwage rigidity then states that the money wagewill not be bid below the rigid level WO, ven ifthere is an excess supply at this level. Thishypothesis is formalized in equation (6). Infact, this equation together with (sb) and (8)simultaneously implies that W and N are de-termined by the intersection of the demandfunction (sb) and the potential supply func-tion, if this intersection determines a value ofW larger than WO;otherwise W = WOand thelevel of employment is determined by the de-mand function alone. The difference betweenthis level of employment and the potentialsupply at WO s then "involuntary unemploy-ment"in the Keynesian sense. Note finally thatequation (6) can also be used to formalize theclassical assumption of wage flexibility bymerely setting WO qual to zero or, at any rate,

2 By "properly," I do not of course mean empiricallycorrect but merely consistent with the hypothesis I intendedto formalize.3 I might add, in partial defense of my I944 constructionand for solace to those who may have accepted it, that, ifthe production function is given and if W. is treated as anunchangeable parameter, then the system does imply theexistence of a unique relation between money consumptionor investment, and money income; but then it implies alsoa unique relation between these variables and r, so that evenunder the stated conditions equation (II) and (I.2) couldnot be regarded as behavior equations but, at best, as re-duced forms. Furthermore, the functional form of thesereduced forms would shift around with changes in W. or theform of the production function. Thus, even though in manycases my original system leads to "correct" inferences, it isan unreliable tool of analysis.


8/13/2019 1927150

6/30

THE MONETARY MECHANISM 83sufficiently low to insure that the first line willapply in the relevant range.From this brief review it should be apparentthat most of the differencesbetween the I944and the mid-fifties model are matters of ele-gance, clarity, and minorimprovements withthe main exception of the correction of theerror under (4) above. Even the explicit rec-ognition of the role of stocks examined under(3) is not crucial, if we regard the model asfocusing on the determinants of short-runequilibrium,since, in general, stocks are initialconditionswhichcan be treatedas given param-eters. An exception to this statement must bemade with respect to the inclusion of wealth inthe consumption function, for, in the presenceof a government debt fixed in money terms,real wealth, VO/P, cannot be regarded as agiven initial condition. As will presently beshown, the resulting "wealth effect" has inter-esting logical implications, although it is ofpractical relevance only under the assumptionof wage and price flexibility, an assumptionwhich has little, if any, empirical substance.The remaining sections of the paper aredevoted to examining various implications ofthe model, focusingin part on issues which havebeen prominently debated in the period underreview. In the two sections immediately follow-ing, the analysis rests entirely on the mid-fiftiesmodel discussed so far. Then, in sections IVand V, this model will be broadenedand amend-ed to allow explicitly for the role of governmentand for certain imperfections in the capitalmarket. Finally, in the concluding section VI,I endeavor to summarize he resultswith specialreference to the importance of money andmonetary factors as a determinant of the realvariables.

IIImplications f the ModelUnderPriceFlexibility- Homogeneity,Dichotomy,Neutrality,and theRole of the Supplyand Demand for MoneyWe proceed first to a summary of some im-plications of the model within the classicalframework of price and wage flexibility. Itsmain justification is the hope of disposing forgood of a controversy, connected with thenames of Pigou and Patinkin, which has

plagued the profession, draining the resourcesinto what strikes me as a largely barren en-deavor.In essence, the Patinkin controversyrevolvesaround two main issues concerning the prop-erties of an economic system relying on a tokenmoney as medium of payment, namely (i) theneutrality of money and (2) the validity of thedichotomy of monetary and real economics. Itwill be useful to define more precisely thenature of these issues and their implications.Neutrality. Money is said to be neutral ifthe equilibrium value of the real variables ofthe system - in our model the flow variables,X, C, I, N, S, the price ratio P7W, and r - isindependent of the money supply M. If moneyis neutral, then at most M can affect the actuallevel of prices, or terms of trade between cur-rent commodities and money - P and W in ourmodel. A special case of neutrality is where, inaddition, prices are proportional to M, the so-called "quantity theory of money."4Dichotomy. This issue can be defined n manyalternative ways, as is apparent from Patinkin[2 I ] where many possible meanings of theword are considered and carefully divided intogood ones (valid) and bad ones (invalid).The problemI am concernedwith here mightbedefined in terms of certain mathematicalprop-erties of systems of simultaneous relations or,instead, in terms of the substantive economicissues involved.If we take the first approach, we may saythat the dichotomyholds if the system of simul-taneous relations purporting to describe thefunctioning of the economy has the followingproperty: the entire set of relations can be"dichotomized" nto two subsets, one of which,containing the relations describing the func-tioning of all markets except the money andbond market, forms a determinate subsystemof the entire system and is therefore sufficientto determineall the real variables of the system.Obviously, if the dichotomy in the above senseholds, then the equilibrium value of the realvariables is independent of the money supply- i.e., dichotomy implies neutrality. But in ad-

' This definition of the quantity theory of money isobviously quite different from the one adopted by Friedmanin [6] and is, I believe, more consistent with generally ac-cepted terminology.


8/13/2019 1927150

7/30

84 THE MONETARY MECHANISMdition these equilibrium values are also inde-pendent of the form of the demand-for-moneyequation. This last implication is, of course,the significant one from the economic point ofview, for it justifies separating the study of thefunctioning of a free market economy into a"monetary"and a "'real"branch. In particular,when concerned with the real variables we needpay no attention to monetary habits and insti-tutions (although the converse need not betrue).This then suggests an alternative and prob-ably more fruitful definition of dichotomy,namely, as a warrantedseparation of monetaryand real economics (or value theory). Specifi-cally, we can say that the dichotomy holds ifthe equilibrium value of the real variables ofthe system is independent of both the supplyand the demand for money. Note that in thissecond sense it is quite possible for the dichot-omy to hold in the long but not in the shortrun (or, conceivably, vice versa).Of course, to the extent that the equilibriumof the system (whether in the short or in thelong run) can be identifiedwith the solution ofa system of simultaneous relations, dichotomyin the first sense implies dichotomy in the sec-ond. But our second definition is broader andis a proposition about the real world, eventhough in passing judgment about its validitywe may have to rely on the mathematicalproperties of certain formal descriptions there-of. Note finally that, in terms of our seconddefinition, we can meaningfully say that thedichotomy is "approximately" valid, in thesense that it providesa good first approximationto observedphenomena.In terms of these definitions, Patinkin'sbasic contentioncan be summarizedas follows:in an economy relying on a token money as amedium of exchange, the dichotomy does nothold, but under certain conditions money willbe neutral. In particular the following two as-sumptions, in addition to assumptions A.i toA.3 above, are generally sufficient for theneutrality to hold (cf. [2I], especially chaptersIV, VIII, and MathematicalAppendix 4).A.4, Concerningthe functioningof markets:absence of price rigidity in any marketA.s, Concerningtastes: the market demandand supply for each commodity is in-

variant undera redistributionof wealthamong transactors5It is foundthat Patinkin'scontention s basicallyunwarranted,although no attempt at rigorousproof is possible here. Furthermore,while theargument runs in terms of the aggregativemodel of Section I, it holds as well if we recog-nize the existence of many different commodi-ties and kinds of labor. Specifically, the fol-lowing propositions can be established (for aclosed economy).

(i) If the money supply consists entirely ofbank money and there is no governmentdebt,6the dichotomy is valid.7Basically the reason isthat under these conditions, and assumptionsA.i to A.s, all the demand and supply equa-tions for commoditiesare homogeneousof zerodegree in the prices, P and W, and hence canbe expressed as functions of their ratio, W/P.This holds in particular for the consumptionfunction (i). In fact, A.5 implies that wealthaffects consumptiononly through its aggregatevalue VO/P. But VO/P n turn can be replacedby the initial stock of physical capital KO, incethe money componentof aggregatenet worth isexactly offset by the indebtedness to the bank-ing system (cf. Memo iv in Table i). Thus Pappearsin the consumptionfunctiononly in theform iW/P.8 It can be readily verified that

' Patinkin has been especially concerned with what maybe called "neutrality in principle," i.e., neutrality with re-spect to a "neutral" change in the money supply, which isdefined as one that leaves undisturbed the relative distribu-tion of cash balances. For this type of neutrality, Patinkindoes not need assumption A.5. However, as will be apparentfrom (2) and (3) below, when neutrality does not hold withassumption A.5, "neutrality in principle" does not holdeither.6 By government debt is meant here any claim (positiveor negative) of the private sector (including banks) on thegovernment, whether interest bearing or not. Hence, theassumption of no government debt implies that M*P, M*&and G are all zero, which, using (M.4), (M.5) and (M.7),in turn impliesM = Mb = Bb- =-BP. In other words, themoney supply consists entirely of what Gurley and Shawcall "inside money" (cf. [8] p. 73).'This conclusion, which was stated in my I960 Postscriptto [II], has recently been acknowledged by Patinkin [i9]in his very useful review of Gurley and Shaw [8], whichalso contains some of the results summarized under (3)below.8Note that the consumption demand -as well as laborsupply-of individual households will generally not behomogeneous of zero degree in prices. Indeed changes inthe price level affect individual wealth by determining thereal value of money fixed claims and debts. However, sinceaggregate wealth is unaffected, such changes only result in


8/13/2019 1927150

8/30

THE MONETARY MECHANISM 85under these conditions the first nine equationsof the model form a determined subsystem inall the real variables including the price ratioW/P. This subsystem is therefore sufficienttodeterminethese variableswithout any referenceto the supply of money or to the demand formoney equation (M.2). The only role of themonetarypart is to determinethe price level P,and hence finally also W. Furthermore, ifassumptionA.s is extended to the demand formoney, then it also turns out that P is strictlyproportionalto M so that the quantity theoryalso holds.It may be useful to try to state in plain Eng-lish the major forces determining equilibriumin the system describedby Model II under thestated assumptions. This mechanism can besurxmarizedroughly as follows:

L) Given the production function, includingthe initial stock of capital, and hence consumersnet worth, and given the preferences for cur-rent and future consumptionas comparedwithleisure, there is a unique real-wage rate W/Pthat clears the labor market, and leads in turnto a unique level of output X: this correspondsto the simultaneoussolutions of the four equa-tions (4) to (6) and (8) in the four variables,P/W, N, Ns and X.i For given X, preferences for currentver-sus later consumption (including in later con-sumption planned bequests) and the terms oftrade between them, represented by r, deter-mine the size of current consumption C andthe desired carry-over of resources, S = X -C. It is the function of the rate of interest toinsure that investors will be induced to add tothe stock of capital an amount I equal to thedesired carry-over S. These forces are de-scribed by the remainingfive real equations ofthe system, (I) to (3), (7) and (9), and com-plete the determinantsof the real variables.iii. Given the real solution, the price leveltogetherwith certain institutional factors (suchas length of the income period, synchronizationof receipt and expenditure, transaction costs,etc.) determine the demand for money arisingfrom transaction requirements and, possibly,also on asset account. There is then a uniqueprice level that equates the demand to the

exogenous supply M. This same price levelalso induces the private sector to issue, on bal-ance, enough claims against itself to satisfy thebanking system demand for bonds, which bankspay with M.9The determination of the price level can beexhibited graphically as in Figure i, in terms

FIGURE ITHE ROLE OF THE SUPPLY OF MONEY IN THE DETER-MINATION OF THE PRICE LEVEL AND THE RATE OF

INTEREST: THE CASE OF PURE BANK MONEYlog P/Po

r r r r1l ~~~~~~LM-M)I ^ /,/J~~~~~L (M- N/M9)

logP /PO -_ -

r

lLof the coordinates of the point of intersectionof two curves in the (r,P) plane. (Actually,on the ordinate it is convenient to plot thelogarithmof PIP,, where PO,may be thought ofas the previously ruling price.) One of these

Acurves is the graphof the line r = r, perpendic-ular to the r axis, stating that the rate ofAinterest must be at the level r that clears thecommodity market and which is determinedby the real part of the system. The other,denoted by LL, is the graph of the equationthat must be sat'isfied n order for the money

Amarketto be cleared,namely: P X L(r, XI Ko)-M. Treating M as a given parameterAand X as already determinedby the real partof the system, this equation can be lookedupon as a relation between the two variablesr and P. Its graph shows, in effect, the be-

a redistribution of wealth between households, which byA.5 does not affect aggregate demands and supplies.

,It is apparent from the above that even in a systemrelying entirely on "inside money" the determinancy of theprice level can be established without reference, explicit orimplicit, to Gurley and Shaw "portfolio balance" or "diver-sification" mechanism ([8], especially Chap. V). Indeed thatmechanism is intimately related to their very objectionabletreatment of firms as separate and self-contained entities.In our model the net worth of firms is treated as a com-ponent of households' wealth.


8/13/2019 1927150

9/30

86 THE MONETARY MECHANISMhavior of the so-called income velocity of cir-culation (more precisely its logarithm) as afunction of the interest rate, except for a pro-

_Aportionality factor, M/X. It can be expectedto rise from left to right, since a priori con-siderations abundantly supported by empiricalevidence (see, e.g., [I2], [29], [2], [30]) sug-gest that, as r rises, velocity rises, the realdemandfor money falls (since X is given) andhence P must rise to reduce the real supply.The position of LL depends, of course, on themoney supply M. A change in this supplyfrom M to say M' would shift it parallel toitself by a distance log (M'/M).While the general shape of the LL graph isclear within the empirically observed rangeof variation of r, there may be some doubts asto its behavior for extreme values. For in-stance, as r grows indefinitely the graph mayrise indefinitelyor may instead approach someasymptote. Similarly, there is some questionas to whether r can ever be bid down to zero.But whatever views one might hold about thepossibility of a zero interest rate, one thingis certain, namely that in perfect markets, andas long as money has negligible storage costs,the money market can never be cleared witha negative r. For, even if banks were preparedto lend at a negative rate, everybody wouldwish to borrow "infinitely large" amounts andhold the money so borrowed to earn the pre-mium now paid to borrowers. Hence the de-mand for money (as well as the supply ofbonds) must become infinite at rates more thannegligibly negative; and for any finite supplyof money M there would be an excess demandfor money and the money market would notbe cleared. This means that the locus of LLlies entirely in the half plane to the right ofa line perpendicularto the abscissa and coin-ciding with the ordinate axis, or possiblynegligibly to the left of it, on account of thestorage cost of money, if any.As can be seen from the figure, if the moneyauthority is concerned with maintaining aconstant price level, then it should set themoney supply at a level such that the corres-ponding LL curve intersects the other curveon the r axis, where log (P/P,) = o or P = Po.If we visualize the mechanism of price deter-

minationas one in whichprices stay put at somegiven level as long as at that level the marketis cleared, and rise (or fall) if at that levelthere is an excess demand (or an excesssupply), then the task of the money authorityin maintaining a stable price level can be statedin a somewhat different and not uninterestingform. That is, instead of enforcing directlythe money supply M*' of Figure I, it may en-deavor, through appropriate devices, to en-

Aforce the rate of interest r' = r and then let themoney supply seek its own level in responseto the demand generated by the initial pricelevel P and the given value of r. If it succeedsin picking and enforcing a rate r' equal to theAnatural rate r, then the money supply will beprecisely M' as we can see from the figure.Of course, if r' is set too low there will beexcess demandwhich will raise P and cause anexpansion in M, and this process will tend tocontinue as long as r' is enforced-this issimply the familiar Wicksellian cumulativeprocess of inflation. Conversely, if r' is settoo high there will be cumulative deflationand contraction in M. The only way ofstopping the cumulative inflation is eitherto raise r' to the right level or to clamp down

on the money supply letting r seek its equilib-Arium level, at r.The above analysis is subject to one im-portant qualification, namely that the twoequations graphed in the figure may fail topossess a solution. This failure is bound toAoccur if r, the value of r that is consistentwith clearing the commodity market, is nega-tive; and there is nothing impossible in prin-

Aciple about r being negative, whatever mightbe the factual relevance of this possibility.Under these conditions the value of r thatclears the commodity market cannot clear themoney market, and conversely. In graphical

Aterms, if r is negative, then the line r = r liesto the left of the ordinate, and therefore can-not have a point of intersection with LL whichmust lie entirely to its right.This possibility, which is, of course, the"Keynesian case," or liquidity trap, might oc-Acur even with r positive, if sufficiently small,


8/13/2019 1927150

10/30

THE MONETARY MECHANISM 87but is bound to occur if r is negative. The eco-nomic system, then, does not possess a restingpoint except through wage rigidity, or appro-priate interferences with the market process,including devices for making money suf-ficiently expensive to store.In summary, with pure bank money theclassical conclusions about neutrality anddichotomy are valid, and so is the Keynesiancontention that a market system may notpossess a position of equilibrium.

(2) If the money supply consists entirely ofgovernmentmoney and there is no other formof national debt,'0 then money is neutral butthe dichotomy, does not hold, i.e., the equilib-rium value of the real variables is independentof the supply but not of the demand for money.This is the assumption underlying Patinkin'sanalysis and his conclusion stands in this case.The dichotomybreaks down because the vari-able VO/Pnow can be expressed aSKo + M/P,and hence P appears in the consumption func-tion (i) separately and not merely in the formof a price ratio W/P. Thus, the first nine equa-tions of the system now involve io unknownsand cannot be solved separately. However, thissubsystem can be solved for the nine realvariables in terms of M/P, the "real moneysupply." In particular, let r(MIP) andX(M/P) be the solution for r and X. By in-serting this solution into the clearing conditionfor the money market one obtains

p = L(r(MIP), X(M/P), Ko+ )This equation in the single variableM/P yieldsa solution, say (MIP), which in turn can beused to obtain the solution for all the remain-ing realvariables. This solution implies that thereal money su ply anc ate real variablesare ihdependent f the nominal money supply.Inh7-tTerwords, nWoney s neutral in that achange in M merel changes P in proportion(P = ) but leav the equilibriumvalue(M/P)of the real variables unc nged. Yet theseequilibrium values are not invariant under ashift in the demand function f ney ex-

pressed by L, for in general a change in thisfunction would give rise to a differentsolutionfor M/P, and, hence, for the real variables.(3) In the presence of national debt or witha mixed money supply, neither the dichotomynor the neutrality holdsl This conclusion canbe shown to follow frqhi the previous two. Justhow a change in the money supply or in thedemand for money will affect the real variablesof the system depends on which component ofthe supply is changed and on the relative sizeof the components and of the national debt. Asan illustration, consider the empirically mostrelevant case where the money supply consistsentirely of bank money and the national debt ispositive. Then an increase in the money supply,by increasingP and reducing real wealth, willtend to increase saving and reduce the rate ofinterest to the extent necessary to produce amatching increment in investment. But differ-ent conclusions would hold under otherassump-tions, which need not be analyzed in thissummary.The results summarized under (I) to (3)above are helpful in assessing the theoreticaland empirical relevance of explicitly recogniz-ing the dependence of consumption on wealth,to which attention was first called by the well-known contributions of Scitovsky and Pigou[27], [22], [23]. From (i) above it is_ap-parent that this recogfnitionhas no significantimplication unless there exists in the economysome net money fixed claims on (or toj thegoverinent,lwh_ether in the form of moneyor in the form of interest-bearingdebt. How-ever, when such claims do exist, as is usuallythe case, then the wealth effect has significantimplications. In particular,a system satisfyingassumptions A.i to A.5 will generally possessa position of full employment equilibrium,contrary to Keynes' conclusion. This is be-cause V0 P now includes a component whichis inversepyrelated to the price level P. Thus,by making W, and hence P, sufficiently small,it is generally possible to make real wealthand consumption so large and the rate of sav-ing and investment so small that the com-modity market can be cleared with a positiverate of interest, consistent with the clearingof the money market. In terms of Figure i,1In the terminology of Gurley and Shaw, the moneysupply consists entirely of "outside" money (cf. [81 pp. 72-73). " Or foreigners, if we consider an open economy.


8/13/2019 1927150

11/30

88 THE MONETARY MECHANISMthe value of r clearing the commodity marketis no longer a constant r but, instead, a de-creasing funi,' TPi 7.ya curve falling fromlefft''o" igir-A;k t - for suffciently low Pthis curve may be expected to lie to the rightof the ordinate axis and to intersect there theLL curve.This conclusion is not without interest fromthe standpoint of the history of economic con-troversies. But it does not imply by any meansthat one can rely on the Pigou-Scitovsky ef-fect mechanismas a practical stabilization de-vice, and that, if only wages were sufficientlyflexible, a market economy could never beplagued by lack of effective demand. For onething, the conclusion is valid only so long aswe assume that redistributionaleffects can beneglected, and that the elasticity of price ex-pectation is no larger than unity. Furthermore,in depressed situations in which a systemwithout a government-money fixed liabilitywould have no solution under wage flexibility,the size of deflation required to re-establishfull employment might well be such as to pro-duce even more damage than widespread un-employment. Thus, even if the cure could becounted on not to kill the patient -which isdoubtful - there would be a great deal to besaid for a less gruesome remedy such as fiscalpolicy.Another implication of the wealth effect isthat under the conditions of case (3), whichare empirically by far the most relevant, theconvenient classical proposition that "moneyis but a veil" is not warranted. But to keepthe role of the wealth effect in proper per-spective, it should be noted that the same con-clusion would hold even in the absence of thiseffect as soon as we drop any of the heroicassumptions A.i to A.5. In short, while thereis really no groundfor holding that in the "realworld" money is ever strictly neutral, the lackof neutrality is due only in negligible portionto the Pigou-Scitovsky effect. Indeed, withinthe range of variation of prices characterizinga normal healthy economy, that wealth effectis likely to be so negligible that, were it notfor other forces, especially wage-price rigid-ities, money would be very nearly neutral. Inparticular, interest rates would be nearly un-

affected by monetary policy, and prices wouldrespond roughly in proportion to the moneysupply. Since in the long (and even not solong) run, rigidities can be neglected, I wouldconclude that neutrality and the quantitytheory -in the sense of a stable relation be-tween the money supply and the value of out-put at any given interest rate - is a goodlong-run approximation, subject, however, tothe stricture that in the long run monetaryinstitutions may gradually change. In the caseof really rapid inflation, the Pigou effect maybe a little less negligible but still of secondaryimportance in comparisonwith the redistribu-tional effect. Furthermore, given a largechange in M, the proposition that the pricelevel will change roughly in proportion islikely to provide a good approximation, oncethe change in M has come to an end. Thus,the significance of the wealth effect appears tobe primarily technical, but its empirical rel-evance would seem to be small, at least undernormal conditions.

IIIImplications f the ModelUnder Wage Rigidity

A. The Strict Keynesian Version. As al-ready noted, wage rigidity can be said to existif the wage rate W will not be bid below somelevel WOeven though at this wage rate thereis an excess supply of labor; and the rigidityis "effective" if the market clearing value ofW is less than WO. The implications of wagerigidity will be examinedfirst in terms of whatmay be called the "strict Keynesian version"of the model, which assumes pure bank moneyand no national debt (and thus no Pigou ef-fect) and competitive behavior in the com-modity and the capital markets and on thedemand side of the labor market.Under these assumptions in the absence ofwage rigidity the dichotomy holds and thereal system (i) to (9) would possess a solu-

A A Ation for all the real variables. Let X, N, r, andA(P7W) denote this solution for the correspond-ing variables and let us label it the "fullemployment" solution. With wage rigidity,however,the dichotomybreaksdown,forthoughP does not appearseparately in (i), W appears


8/13/2019 1927150

12/30

THE MONETARY MECHANISM 89separately in (6), and thus the nine real equa-tions contain ten unknowns. The equilibriumvalue of the real variables depends then on themoney supply M, as well as on the rigid wageWO, nd more specifically on their ratio M/WO.

This conclusion, and its implications, can beillustrated by the simple graphical apparatusof Figure 2, if we are prepared to make a few

convenient simplifying assumptions. In par-ticular, suppose that, to a first approximation,the demand for money can be treated as ho-mogeneousof first degree in money income andnot significantly affected by wealth, i.e., Md =L(r,Y) = YL*(r). Then in order for themoney market to be cleared we must haveYL*(r) = M, which implies(k.i) Y/W = M/WL* (r)Taking M and W as given parameters, this isa relation between Y/W (income in wageunits) and r, shown in Figure 2 as the MMcurve. (Note that the Keynesian device ofusing labor as the numeraire is a very naturalone since the rigid wage provides a stable unitof measurement.) This curve can again be re-garded as the graph of income velocity as afunction of interest rates, except for a propor-tionality factor, M/W. Thus, a change in theratio M/W will cause the curve to shift up ordown in proportionto the change.Next, from the real part of the system onecan derive a second relation between thesesame variables which must be satisfied for thecommodity and labor markets to be cleared.

For this purpose we first derive from (4b) arelation between P7W and X, which repre-sents in essence the Marshallian short-runsupply function for commodities (short run,because Ko is fixed). By means of this relationand (5b), consumption demand given by (i)can be expressed in terms of X and r only, say,C = C*(X,r). Substituting this result and (2)into (3) and (7) one obtains an expression ofthe form C*(X,r) + I (X,r) = X(the initial condition Ko being subsumed in thefunctional form), which can be solved for X interms of r, say(k.2 ) X _ x(r).For any given value of r this relation yieldsthat level of output for which the sum of thecorresponding consumption and investmentdemands just equal output, and the commoditymarket is cleared. Finally, by multiplying Xby the corresponding supply price P7W, wearrive at the desired relation between Y/W =(P7W) X and r, say(k.3) Y/W = y (r).It is represented in Figure 2 by the yy curveand is shown only for values of Y/W(Y/W), since for larger values there would bean excess demand for labor, i.e., the labormarket would not be cleared (nor could therigidity be effective).Since the assumption of competition in thecommoditymarkets precludes the possibility ofa falling short-runsupply function- i.e., P7Wmust be a nondecreasing function of X - theslope of the function y(r) depends on that ofthe function x(r). This slope referred to the raxis in turn can be shown to be given by theexpression (C* + I) I

I . HereC* and I. denote respectively the marginal pro-pensity to consume and invest with respect tochanges in real income, and C* and I are ther rcorresponding propensities with respect tochanges in r. The first factor (C* + I) can betaken to be negativesince C* is negativeand Irwhatever its sign, is most unlikely to outweighit. The second factor is a generalizationof theconventional multiplier, sometimes called the"supermultiplier,"and is generally assumed to

FIGURE 2WAGE RIGIDITY AND MONETARY POLICY

Y/W

- M

r r


8/13/2019 1927150

13/30

90 THE MONETARY MECHANISMbe positive. If so, X, and hence Y/W will bea declining function of r as shown in the figure.It should be observed, however, that, contraryto a common view, there is really nothing im-possible or unstable about the supermultiplierbeing negative (or zero) in some regions andthe graph of yy rising from left to right (orbeing parallel to the ordinate).The intersection of MM and yy gives the*equilibrium values, say (Y/W) and r*, andthese values clearly depend on M/Wo whichcontrols the position of MM. The equilibriumvalue of real income X*, can then be obtainedby substituting r* into (k.2), (P/W) * fromX* and the supply function and so on; and theymust all be functions of M/Wo.

In our graph the equilibrium position is oneof less than "full employment." Full em-ployment could be reached only with a largermoney supply in wage units which would shiftup the MM curve to a position where it inter-sects the yy curve at the full employment point[r, (YW)] The money supply required toreach this position can be seen from (k.i) to be(MIW)- (YIW) L* (r). Under wage flex-ibility the adjustment would of course occurthrough the market process as unemploymentwould cause W to fall, and hence M/W to rise,until the labor market is cleared. But underwage rigidity it can only come about by an ex-pansion of the nominal money supply to thelevel M' =(M_/W) X WO.This expansion canbe achieved by the monetary authority eitherby directly enforcing the correct money supplyM' or by picking and enforcing the correctrateof interest r, and letting the money supply seekits appropriate evel M'. (Both processes may,of course, go on simultaneously.) Note that inthe presentmodelboth monetary expansionandmoney wage reductions act only through shiftsin the MM curve and lead to the same value ofY/W and r, and to differences only in moneyprices and income. This "Keynesian" resultdepends, however, on our having assumedawaya "Pigou-Scitovsky" effect, by supposing G tobe zero.We need still to considerwhat would happenif the initial money supply were such that theMM curve passes above the full employment

point, as is the case for the curve M"M" inFigure 2. Here, of course, if wages are onlyrigid downward,the classical mechanismtakesover; the rate of interest being initially too low,there is excess demand in the commodity mar-ket which bids up prices and wages; the rise inwages eventually reduces the effective moneysupply M/W, shifting down the MM curve,until it intersects the yy curve at the full em-ployment point. The process just describedcorrespondsto what is usually called "demand-pull" inflation.Oversimplifiedas our model is, it brings intosharp focus the predicament of the moneyauthority in a system with rigid wages, namely,to continuously pick and enforce the correctmonetary policy (the money supply M', or theinterestrate r, orboth) under nadequateknowl-edge of the relevantportion of the MM and yycurves and their shift through time. Also theconsequences of errors are asymmetric. If itfollows a more "loose" policy, it generatesdemand-pull inflation or price rises which,under wage rigidity, are largely irreversible.If it follows a "tighter"policy it engendersun-employment and not merely deflation pre-sumably a somewhat lesser evil as long as con-tained within limits. In reality the problem isfurther complicated by the fact that themonetary authority may be expected to pursuethe double goal of full employment and pricestability. Unfortunately, these two objectiveswill be inconsistent with each other if theexogenously determinedwage rate is such that,when combined with the full employmentprice-wage ratio (P/W), it implies a full employ-ment price level, P (P/W) X WO, higherthan the historically received one. This is ofcourse the "cost push" case. If the monetaryauthority pursues full employment it cannotprevent a rise in prices; while if it refuses to goalong and expand the money supply enough toinsure full employment at the given wage, itwill certainly cause unemployment, while itmay not even succeed in preventing some risein prices.The dilemma is even more dramatic if thewage rate is controlledby a mechanismwhere-by money wages tend to rise at a rate depend-ing on the level of unemployment, and this


8/13/2019 1927150

14/30

THE MONETARY MECHANISM gIrate of change is larger than the "rate of in-crease of productivity" (the rate at whichPIW falls) at rates of unemployment largerthan what might be regarded as unavoidablefrictionalunemployment (see, e.g., [261] ). Ac-cording to some views this is the predicamentof our times, but I don't propose here to assessthis claim or, even less, to propose remedies.In concluding this section it is well to callattention to two types of situation where toreach and maintain full employment is beyondthe power of the monetary authority. One isthe well-knowncase where the full employment

Arate of interest r is negative (or possibly veryclose to zero), represented in Figure 2 by thecurve y'y'. It corresponds to a situationwhere, in an economy with flexible wages,there would be no set of prices and interestrate capable of simultaneously clearing allmarkets. The system can have a solution onlyif wages are rigid and unemployment is al-lowed to develop, permitting the eliminationof the excess supply that would otherwise ariseat any positive rate of interest. In fact, thefall in employmentis accompaniedby a reduc-tion in supply X. To be sure, this fall tendsto reduce also the demand but only to asmallerextent, as consumption,at least, is keptup by initial wealth and the expectation thatthe fall in income is but transitory in nature.(In Keynesian terms, the marginal propensityto consume is less than i.) Thus, at some suf-ficiently low level of employment and output,the excess supply tends to disappear, and thesystem finds a resting point. A very aggressivemonetary policy might at best bring outputclose to the ordinate of the point at which they'y' curve cuts the vertical axis.12 Only fiscalpolicy can get the system back to full em-ployment (cf. section IV).The other case where monetary (as well asfiscal) policy is powerless is the case of "realwage rigidity." Here, because of union pres-sure, legislation, minimum subsistence levels,or otherwise, the real wage cannot be reduced

below some level, say (W/P)m, and this levelexceeds the marginal product of labor cor-responding to the given production function(including the stock of capital) and full em-ployment of labor. (Full employment in thisconnection might be defined in terms of somestandard labor force participation and workweek, or in terms of the potential supplyavailable at the fixed real wage.) There willthen be an effectiveceiling on output, X and onY/W, below the full employment ceiling ex-hibited in Figure 2. Any attempt to expandemployment beyond that ceiling, through ex-pansion of the money supply (or fiscal policy),will only succeed in increasing prices andwages without increasing output and employ-ment. The situation just described frequentlytakes the form of a "balance of paymentsproblem." The attempt to expand employmentand income increases imports, forcing a de-valuation, or equivalent measures, which in-crease the price of imports, and lead finallyto a rise in money wages and prices. This typeof situation, which some hold to be commonin underdevelopedeconomies, can be remediedonly if the wage rigidity can be broken orproductivity increased through technologicalprogress and capital accumulation, providedthis does not immediately result in a commen-surate rise in the rigid real wage.B. Modifications of the Keynesian Model:Imperfections in the Commodity Market andWealth Effects. The assumption of a com-petitive commodity market, which justifies therising supply function of the previous sec-tion, may be replaced by an alternative one,possibly more realistic and certainly moreconvenient. It is the assumption that pricestend to represent a roughly constant markupon unit labor cost, possibly reflecting theprevalence of market imperfections of theoligopolistic type. An alternative formulationof this same hypothesis is that labor incomeWN is a fairly stable share of total incomePX, at least in the neighborhood of full em-ployment, X [3I . Indeed, a constant markupon unit labor cost means:P = (i+ m) (WN/X),where m is the markup.This equation in turn implies:

12 Even the most aggressive monetary policy can at bestforce r to zero, and cannot force the money supply abovethe demand arising at this interest rate and the price levelcorresponding to the given wage rate and the level of outputX = x(o). To make monetary policy more effective wouldrequire very special devices such as making money sufficientlyexpensive to store.


8/13/2019 1927150

15/30

92 THE MONETARY MECHANISMWN= (PX),I + m

where i/(i + m) is the constant share oflabor income in total income. The empiricalevidence seems, on the whole, to supportsuch a stability, at any rate in the mediumrun. To be sure, as output fluctuates below,and up to, full employment the share ofprofits in income tends to fluctuate with out-put. But total property income is the sum ofprofits, interest, and rent income, and the shareof the latter two components moves in the op-posite direction, imparting stability to theshare of total property income and hence alsoof labor income.Suppose further that in the short run N isapproximatelyproportional to X, i.e., that theelasticity of output with respect to labor inputis close to i. This hypothesis does not seemto be grossly inconsistent with the empiricalevidence, at least in the neighborhood of fullemployment. (It is, however, inconsistentwith the assumption that the real wage isequal to the marginalproductof labor, for thenthe elasticity of X with respect to N would beequal to the labor share of income, which iswell below unity.) Under this further assump-tion NIX can be approximated by a constant,say, n, and consequently PIW becomes itselfa constant or(4.i) P/W = (II+m)n = ran equation that replaces the original (4b).The parameter r is of course a constant onlyat a given point of time. It may be expected tofall over time as productivity, X/N, risesthrough technical progress and the accumula-tion of capital, and hence labor input per unitof output, n, falls. If (4.i) holds, then thedemand for labor is no longer given by (5)but directly by solving the production func-tion (4a) for N in terms of X and Ko. Finallywe may wish to recognize that, in the shortrun, the labor supply may be rather inelasticwith respect to the real wage rate, and hencemay be adequately approximated by a con-stant, say Nf. Then equation (6) is replacedby

{NS - Nf if nd(X, K.) - Nf(6.1) W= W if nd(X, Ko) < Nf

These various modifications do not changethe count of equations or unknowns of ModelII. However, the resultingsystem - call it II.i- is somewhat easier to see through. For, solong as W is rigid at WO,P itself can be regard-ed as given exogenously (since it is proportionalto WO).Hence, instead of solving the system forY/W in terms of r, we can solve it for Y/P.But Y/P is X and hence its solution in termsof r is simply (k.2). Similarly, equation (k.i)can be rewritten as Y/P = (M/P)/L(r) andFigure 2 can be redrawnwith the more conven-tional and convenient variable Y/P measuredon the ordinate, and the money supply stated in"real" terms, M/P.The implications of this model are roughlythe same as those discussed in the previoussection, but easier to comprehend and expose.In particular, the price level P may be regardedas determined by the rigid wage (togetherwith labor productivity and the constant mark-up) providedthat the money supply is no largerthan what is required to transact a full employ-ment income at the price level correspondingto WO. But if the money supply is larger, sothat the first line of equation (6.i) holds, thenP is determined by the money supply, P* =M/L(r,Y/P), and the wage rate is determinedby P and hence, indirectly, again by M.The implications of the model are also notappreciably affected if one recognizes theexistence of a money-fixed government debt,G. In terms of the graphical analysis of Figure2, the main effect of G is found to be that equa-tion (k.2) must be changed to(k.2') X = Y/P = x(r,G/P).Accordingly, the graph of this equation- theyy curve - is not independent of the value ofP and hence of WO.In particular,a fall in WO,by reducing P and increasing real wealth andconsumption, would require a smaller rate ofinvestmentand hence a largervalue of r to clearthe commodity market for any given X. Ittends therefore to shift the yy curve to the right.The full employmentlevel of Y/P might also beaffected, via the labor supply, but presumablynot significantly,at least for reasonablechangesin P. Similarly the money market clearingcondition becomes


8/13/2019 1927150

16/30

THE MONETARY MECHANISM 93(k.i') L(r,Y/P,G/P) = M/P.Hence, the position of MM depends not onlyupon the real money supply M/P but also di-rectly on P and hence WO. Given M/P, a fallin WOand P should tend to increase the de-mand for money, i.e., reduce velocity for anygiven r, shifting MM downward.The main implication is that monetary ex-pansion and money wage cuts need not have asymmetrical effect on output and the otherreal variables, since the equilibrium value ofthese variables, while still a function of M1and W, no longer depends merely on theirratio. Pure monetary expansion still affects theequilibriumby shifting up the MM curve. Buta wage cut of the same proportion would shiftup MM somewhat less (because of the wealtheffect on the demand for money), and wouldalso shift yy to the right. If, as seems likely,the wealth effect on the commodity market islarger than that on the demand for money, awage cut will tend to result in a somewhatlarger increase in X than would monetary ex-pansion. But the difference may be expected tobe altogether negligible: a priori considerationsas well as empirical estimates [i] suggest thatthe shift in either curve from the wealth effectis likely to be minor, for reasonable changesin WOand reasonable assumptions about theratio of G to aggregate net worth - say, o to20 per cent. In part, this is due to the factthat the change in wealth resulting from the"deflation" s likely initially to affect consump-tion only moderately,the rest of the effectbeingspread in time.Note, finally, that because wage deflationtends to shift the yy curve to the right - pro-vided it does not generate expectations offurther fall in prices - it could conceivablylead to an expansion of employment even insituations where monetary policy as such ispowerless. Hence, across-the-boardwage cutsmight appear to provide a possible alternativeto fiscal policy. But, as already noted at theend of section II, this view has little practicalmerit. For even if we wave aside the problemof enforcingan over-allwage cut, this approachprovides at best a weak and unreliable tool,inconsistent with the maintenance of a stableprice level.

IVThe Role of GovernmentMonetaryand FiscalOperations

The essential implications of governmentfiscal operationscan be formalizedby the addi-tion of two equations to Model II and certainmodifications in some of the remaining ones.The main modificationsare the introductionof total tax receipts and other fiscal parametersin the consumption function and possibly insome other equations such as (2), (4), and(i); the addition of government purchases ofgoods F to the definition of aggregate demand(3), and the purchases of labor services, Fn, tothe demand for labor on the right-handside of(8); and the addition of the governmentstockof capital, Kg in (2), (4) and (i). The twoadditional equations are: (i) a tax collectionequation, which may be considered as part ofthe real system, of the general form(Io) T = t(X, VO/P,P,[7] Dwhere T denotes tax receipts net of transferpayments measured in real terms (i.e., meas-ured in MM's) and [r] denotes the relevantset of tax parameters; (ii) the governmentbudget identity, to be added to the monetaryset,(M.8) PF + WFn-PT = AG + AM*The left-handside is the budgetdeficit in moneyterms, denoted hereafter by D. On the right-hand side, M* may be usefully defined as "netgovernment money" or the difference betweenthe amount of money issued by the government(if any) and the amount of bank-createdmoneyheld by the government. It is also convenientfor present purposes to replace (M.4) and(M.S) by the single equation(M.4g) M =Bb+M*while droppingthe variable Mb.After these additions and modifications, theset of equations (i) to (io) plus (M.i) to(M.3) turns out again to form a closed sub-system in the original variables plus T, butinvolving now a new set of "fiscal policyparameters,"to wit, F, Fn, and [r]. Hence inexamining the implications of the "governmentmodel," it is permissible to concentrate on thissubset and to disregardagain the bond marketand the other monetary relations. However,


8/13/2019 1927150

17/30

94 THE MONETARY MECHANISMsome of these relations and in particular(M.4g) and (M.8) are useful to clarify therelations between monetary and fiscal policy,which are essentially linked through debt man-agement. In particular, from these two equa-tions we can derive the relationAM = (ABb -AG) + D.This relation serves to make clear that, inprinciple, through appropriate monetary anddebt management, monetary and fiscal poli-cies can be made entirely independent ofeach other. That is, a given deficit (or sur-plus) D can be madeconsistentwith any changein M through appropriate changes in Bb, theamount of credit extended by the banking sys-tem to the rest of the economy, or in G, theamountof governmentdebt held by the privatesector. It is for this reason that, in our model,we can continue to treat the money supply Mas exogenously given even in the presence ofgovernment fiscal operations. And we definemonetary policy as the control of the moneysupply and not merely of the amount of bankcredit.The role of fiscal policy and its relation tomonetarypolicy can again be analyzed througha graphical apparatus analogous to that ofFigure 2 and set out in Figure 3.

The money market clearing conditions arebasically unchanged and so is the nature andinterpretationof the MM curve. Its position isagain controlled by the real money supply

M/W and possibly also to a minor extentdirectly by W, through the wealth effect. Asfor the commodity market, through a series ofsubstitutions of the type described earlier anda few suitable simplifications and approxima-tions -such as neglecting direct governmentpurchases of labor - we can obtain a conditionstating the equality of demand and supply inthis market of the form:(g) C*(X, r, G/W, T, [r])+I*(X,r, [r])+F=X.If we furtheruse (io) to eliminate T, the aboveequation can be solved for X in terms of r andfiscal parameters. From this solution in turnwe can derive the relation(k.3g) Y/W (P/W)X= (P/W)x*(r, F, [v], G,/W), X X.For given values of the fiscal policy "param-eters," F and [7], this equation can be lookedat as a relationbetween Y/W andr, representedin ourgraphby the yy curve. Its position in theplanedepends of course on fiscalpolicy, and thisdependence can be conveniently approximatedby a series of fiscal "multipliers" describingthe upward (or downward) shift of the curvein terms of the change in Y (given r) per unitchange in the indicated parameter. The for-

mulae below, obtained by total differentiationof equation (g), above, or (io) or both, pro-vide a sample of such multipliers'effect on realincomeX. The effect on Y/W can be obtainedby multiplying these formulae by P/W, (ifP/W can be taken as a given parameter). Thesymbol C* denotes here the marginal effect onconsumption of an increase in tax payments(usually assumed to equal the marginalpropen-sity to consume with respect to income, C*with sign reversed) and t, is the marginalchangein tax receiptsper unit changein incomebefore taxes.(i) Effect of unit change in expenditure,F:_ = I/(I - CX - CTt- I*)

(ii) Effect of an increase in tax payments Tor, equivalently, of a decrease in deficit D, ex-penditure constant (under the approximationthat consumption depends only on the taxliability and not on the specific form of taxes):dX costn - I*dT I constant T X Xt

FIGURE 3MONETARY VERSUS FISCAL POLICY

Y/WA /

am

I ,

rm rf


8/13/2019 1927150

18/30

8/13/2019 1927150

19/30

96 THE MONETARY MECHANISMof taxes - disposable incomeat f is higherthanat m. For I to be as high as at m, the tax in-ducement to invest, which increases yieldspermitting the higher interest rate at f, wouldhave to induce larger saving out of a givendisposable income, and sufficiently so to offsetthe increased disposableincome. This is a mostunlikely outcome,especially if one believes thata higherrate of interest is more likely to reducethan to increase saving. Thus, paradoxically,reliance on tax inducementsto invest instead ofon monetary policy to stimulate demand islikely to generate a lower rate of investment.15In addition,it will tend to produce higheryieldson investmentand higher market interest rates.In general, one may conclude that the maindifferential effect of using tax incentives in-stead of monetary policy to stimulate invest-ment, when either method could be effective, isto producehigher yields which are in turn con-sistent with higher market interest rates. Thisdifference might be desirable in the context ofcertain balance of payments problems whichare, however, beyond the scope of the present"closed economy"model.The conclusion to be drawn from this briefanalysis is that, in so far as full employmentcould be maintained by purely monetary de-vices -i.e., where the initial yy curve lies en-tirely sufficiently within the positive quadrantthe choice between monetary policy andvarious types of fiscal policy to achieve the ap-propriate level of aggregate demand must bebased on traditionalconsiderations. These are:the relative merits of private versus public con-sumption in choosing between C and F - Fc;the relative "social" yield of private versusgovernment capital formation in choosing be-tween I and Fc; and finally, on "intergenera-tion" comparisons in choosing between totalcurrent"consumption"C + F - Fc, on the onehand, and total capital formationI + Fc on theother. As I have arguedin some detail in [i 8],

by pushing capital formationat the expense ofconsumption we increase the stock of capitaland real income available to the community inthe future, at the expense of the current genera-tion. I have also argued that a neutral policymight be regarded as one that makes the cur-rent generationpay for the government servicesit is currently receiving, and that such a policyrequires, by and large, collecting currently intaxes an amount equal to F - Fc. Lower taxesmake future generations pay for the servicesenjoyed by the current generation,while highertaxes, in essence, make the current generationspay for services enjoyed by future generations.There remain to consider briefly two cases.The first, when we start out from a position offull employment,requires very little additionalcomment. Here, an increase in governmentex-penditure must clearly be justified on groundsother than maintenanceof full employment. Ifthe increased expenditure is not accompaniedby higher taxes, then consumptionwill be un-changed and hence the whole increase in Fmust come from a reductionin I, as the govern-ment taps private saving that would otherwisehave gone into capital formation. This is thestandard case on which rests the "classical"argument that deficit financing shifts the bur-den to "future generations" (cf. [3] and thereferences cited there, and [I8]). Since thereduction in I and expansion of governmentborrowing will tend to be accompanied byhigher interest rates, an appropriaterestrictivemonetary policy will be called for. In terms ofFigure 3, this situation could be representedas a shift of the commodity market curve fromyy to y'y'. It requires a shift of the moneycurve from MM to M'M', in order to offset thehigher velocity of circulationaccompanyingthehigher interest rate.If the increased expenditure is accompaniedby a matching increase in taxes there will stillbe some shift to the right of the yy curve, asconsumptionwill tend to fall by less than taxes(the multiplier effect of a balanced budget).Hence, again, private capital formation I willhave to be restricted somewhat through ahigher interest rate and a tighter monetarypolicy. The maintenance of private investmentswould, in the short run, require instead an ap-

"5 ome might question the empirical relevance of thisconclusion on the ground that I am ignoring here the cor-porate form, and that tax inducement to corporations mightincrease corporate saving and thus total saving and privateinvestment. Basically, my position on this point is thatcorporate saving, except possibly in the very short run, is asubstitute for, and not an addition to, saving out of con-ventional disposable income. This conclusion follows readilyfrom the M-B-A consumption function in combination withthe argument set forth in [I7].


8/13/2019 1927150

20/30

THE MONETARY MECHANISM 97propriatebudget surplus put at the disposal ofinvestors, to offset the reducedprivate saving.There is, second, the "Keynesian"case wherethe yy curve crosses over to the second quad-rant. Here (aside from the rather impracticalmonetary cures mentioned earlier) fiscal policyis the only remedy, at least to the extent neces-sary to shift the yy curve to a position where itcan make contact with monetary policy. Evenin this situation a case can be made in principlefor favoring government capital formationover otherways of stimulatingdemand,since, asI have arguedin [i 8], there will otherwisebe a"burden" on future generations, although, inany event, the burden will be small in relationto the benefits accruing to the current genera-tion.This analysis of the modusoperandiof mone-tary and fiscal policies and their implicationssuggests, at least to me, that the case for a cur-rently balanced budget, and hence for relyingon monetary rather than on fiscal policy as afirst line of defense in counteracting shifts inthe forces controlling aggregate demand, issomewhat stronger than might have appearedsome time ago. However, the choice of a propermix involves many more aspects than those wecan develop here, including considerations ofreliability of the tools and of feasibility in agiven concrete institutional setting. In partic-ular, exclusive reliance on monetary policy,whenever this policy could, in principle, do thejob, might require swings in the money supplyand interest rates of a size that mightprove un-settling to the working of the economy. (See,however, the argument of the next section.)These considerations support a policy of built-in stabilizerswith reasonablyhigh marginaltaxtake tZ, (which, remember, ncludes transfers).Such stabilizers tend to moderatethe swings intime in the position of the yy curve resultingfrom shifts in the investment or consumptionfunction or both, thus reducing the burden im-posed on monetary policy. On the samegrounds, a good case can be made for somecountercyclical variation in expenditure andtax parameters,although at least from presentevidence one might have reservationsabout thesuitability of tax cuts announced to be buttemporary (cf. [i8] section IX). But therestill remains a prima-facie case for balancing

the budget over a suitable span of time(cyclically balanced budget) in so far as thisis consistent with full employment, unless aconvincingcase can be made for discriminatingbetween generations.V

Imperfectionsn the CapitalMarkets-the AvailabilityDoctrineThe models on which we have relied so farassume, at least implicitly, a well-functioningcompetitive capital market in which invest-ments are limited and brought into line withsaving through the mechanism of the rate ofinterest or cost of capital. In such a modelthere exists a single short-runequilibrium rateof interest which measures both the return tolenders and the cost to borrowers, and alsoequals (or at least is not less than) the internalmarginalrate of return to all units.'6 There isalso no need to give separate treatment to finan-cial intermediaries: all loans may be regardedas extended directly from the lending or sur-plus units to the final borrowersneeding fundsto finance their expenditure.That this assumption is unrealistic probablyno one would have disputed seriously. It was,however, the merit of the availability doctrine,advanced in the postwar period, that it madea convincingcase for the propositionthat disre-gard of certain institutional imperfections ofthe capital market leads to an unsatisfactoryand seriously distortedview of the modus oper-andi of monetary policy and its consequences(see, e.g., [24] [25] [4] [28]). Actually, thepromoters of this doctrine seem to have beenlargely motivated by a specific issue of mon-etary policy: the advisability of abandoningthepolicy of pegging the yield of governmentsecurities, which in turn made it impossible tomaintain close control over the money supply.They were primarily interested in establishingthat, even if one accepted the then-prevailingview that aggregate demand was very inelasticwith resDect to interest rates (i.e.. the elasticity

"ONote that even if we wished to recognize the existenceof a plurality of maturities, under our present assumptionof certainty, the current return - interest plus capital gain- would be the same on all maturities and equal to theshort rate.


8/13/2019 1927150

21/30

8/13/2019 1927150

22/30

THE MONETARY MECHANISM 99in the twelve original endogenous variables(except that r is replaced by rS).The working of this system can again beclarifiedby a graphical analysis of the type ofFigure 2, and exhibited in Figure 4. Specifi-

cally, from the first nine equations in ten en-dogenous variables we can derive again arelationbetweenX and r8, which it is now con-venient to write as r, = R (X). This equationexpresses the relation between the index ofinternal rates and the level of output when thecommodity and labor markets are cleared atthe corresponding evel of output, which meansin particularthat the rate of investment equalsthe rate of saving. Stated differentlyand some-what less precisely, it shows the internal mar-ginal rate of return prevailing when the levelof aggregate demand is X and the flow of re-sources available for investment is equal to therate of saving prevailing at this level of output.Since to each value of X there corresponds avalue of PIW we can also obtain a correspond-ing equilibriumrelation between r, and YIW,which is shown in Figure 4 as the RR curve.This curve is shown as falling from left toright on the same grounds on which the yycurve was drawn with a negative slope inFigures 2 and 3; a largerincomemakespossiblea largerrate of investmentwhich in turn impliesa lower marginalrate of return.

As for the money market, by assuming as aconvenient approximationeither that PIW canbe treated as a constant and G - o, or thatMd is homogeneousof first degree in money in-come and not significantly affected by wealth,we can write the market clearing condition as:L(Y/W, r', r,) -- IW. For given values ofMIW and r', this condition yields a second re-lation between YIW and r,, shown again as theMM curve. The general shape of this curvemust be similar to that of the correspondingMM curve of Figure 2; however, its elasticitywith respect to the variable on the abscissamust be smaller,since r, is only one of the ratesaffecting the velocity of circulation while theother, r', is constant by assumption. The inter-section of the two curves yields the equilibriumvalue of YIW and r,, say (V/W)* and rI,from which the equilibrium value of X andother variables can be inferred. The assumedvalue of the parameterr' is also shown in thefigure. Under certain assumptions, the gap be-tween r' and r8 can be taken as an indicator ofthe size of the rationing gap (or fringe ofunsatisfied borrowers), the size of the demandfor credit unsatisfied at the lending rate r'.It is apparent from our figure that the work-ings of a model with capital rationing of thetype consideredare not radically differentfromthose of the original Model II. In particular,if we start from a position of full employmentequilibrium, an upward or downward shift inthe position of the RR curve, reflecting, e.g.,an improvementor deteriorationof investmentopportunities,would lead, respectively, to "in-flation" or unemployment, unless offset byappropriatechanges in the money supply. Also,in a situation of less than full employmentequilibrium, such as the one assumed in ourfigure, unemployment could be cured by anappropriate monetary expansion. However,under capital rationing this outcome couldcome about without any change in the lendingrate r'.The mechanics of this operation are not dif-ficult to trace out. The expansionof the moneysupply is initially accomplishedby a relaxationof rationing by banks and a consequent expan-sion of lending. However, as income expandsin response to the direct increase in investmentand the induced expansion of consumption,the

FIGURE 4SOME IMPLICATIONS OF THE "AVAILABILITY DOCTRINE"

Y/W

ARW RI R-(Y/W) a~~~~~~m

_ _ _ _ _ _ I_

r rS


8/13/2019 1927150

23/30

ioo THE MONETARY MECHANISMhigher rate of investment can be sustainedwithout further monetary expansion throughthe increased flow of saving, which in part re-sults also in an increased flow of credit avail-able from intermediaries.

Similarly, an increase in W,M constantwould tend to result in higher prices and a fallin the rate of real investment and income,partly moderated by a rise in the velocity ofcirculation under the influence of the increasein r8. In terms of our figure, the MM curveshifts down and (Y/W) * falls as the rise in Y,induced by the higher r8, is proportionallysmaller than the increase in W.While r' may be taken as given in the shortrun, it may be expected to adjust graduallyover time, tending toward some normal rela-tion to r*. But because this adjustment is aslow one, we may infer that even if r* swingssharply and rapidly over time in response tocyclical and other forces, r' will tend to fluctu-ate over a much smaller range. Thus, the capi-tal rationing mechanism provides a plausibleway of reconciling moderate fluctuations inmarketrates with a widely shifting and interest-inelastic investment schedule.We must, however, stop to consider whatmight be expected to happen if, as the result ofa rapid decline in investment opportunities,the RR curve were to shift downward to a po-sition such as R'R' in our figure. This newcurve intersects MM at a, but this intersectioncould not possibly describe a position of equi-librium. It implies in fact an equilibriumvalueof r8 smaller than r' which is impossible since,clearly, for every borrower the internal ratemust be no less than r'. What is involved hereis that at a, with the lending rate r' borrowersare unwilling to borrow all that the interme-diaries would have available to lend and therate of investment is less than the rate of sav-ing, so that income must fall.In orderto exhibit the new equilibriumposi-tion, let us assume at first that the rationinggap can dwindle to zero, intermediariesbeingwilling to lend to anybody prepared to pay r',and that under these conditionsr8can be as lowas r'. Then, as far as the commoditymarket isconcerned, equilibriumcould be reached at b,where the per

1927150

Documents