Capital De Trabajo

A MODEL OF THE DEMAND FOR MONEY BY FIRMS: COMMENT
MARTIN WEITZMAN

of for Recent macroeconomic explanations thedemand cashhave utility" approach.One tendedto focuson the so-called"indirect is this to ofthelatestmodels embody viewpoint thatof Millerand earlier referred as the M-O model. Although to Orr,' hereafter cash of in certain factors theanalysis household key isolated models of thefailedto capture essence theygenerally decisions, inventory Millerand Orrhave endeavored a habitat, voidwhich thebusiness to fill. certain and extend is In thisnotean attempt madeto clarify asbut to partsoftheM-O analysis coveran important neglected Millerand Orrseekto quantify. which situation thebusiness pectof like manyinit thatthe M-O model, speaking, is found Roughly with are results unaltered) is(i.e.,general ventory models, "robust" to be modified. to respect the assumptions by is of structure theM-O model characterized two The policy h parameters, and z. Cash balancesare allowedto wanderfreely h, bound, or reacheither lower the bound, zero, theupper until they cash to transfer undertaken restore is at whichtimesa portfolio balancesto a level of z. The lowerboundis set at zero becauseare as place instantaneously. transfers regarded taking that the fixedcost of transference Miller and Orr postulate from is cost from bonds cash,AB, equal to thefixed oftransference to The prospect thisassumption implausible. y. cashto bonds, I find ata to managerial out ofrunning of cash forces firm appropriate far thanwouldbe the case for tention a scale ofpriority higher on excesscash intobonds. Consequently, the problem converting of plusindirect loss the it is felt, totalopportunity (thesumof direct exceedthe out will costs) of running of cash altogether typically cash loss totalopportunity ofconverting to bonds. is by The return point prescribed Millerand Orrto be indepenor the was dentofwhether lasttransaction a purchase a sale. Such and "natural" thanoneinvolvis a policy,theystate, simpler more a and after purchase a sale,becausefixed points return ingdifferent in to transfer costsare assumed be identical bothdirections.
LXXX 1. "A Model of the Demand forMoney by Firms,"thisJournal, (Aug. 1966),413-35.

162

OF QUARTERLYJOURNAL ECONOMICS

to costsare equal,it suffices or whether nottransfer Actually, becausean optimal policyallowreturn point onlya single considerThis pointstrategy. return is points a single ingfordoublereturn "has no is process Markovian.The system is so becausethe entire is onlywhatstatethesystem in,nothow and memory" it matters to we payment, are free choose it gotthere.Havingmadethefixed depends choice The optimal of (the conditions point return). initial and not on the past in any way-it is thus onlyon the future, will pointof returnin The optimal transfer. from either identical will alwaysbe one pointrather on depend ,8 and y,but it general thantwo.2 of the The path is nowclearforinvestigating effects differing are symbols used: transfer costs. The following from to cost ,8= fixed ofa transfer bonds cash from cost y = fixed ofa transfer cashto bonds on earned theportfolio v = rateofinterest of T = length theplanning period totransfers bonds from of number portfolio E (N1) = expected of T the period length cashduring planning from cash to of transfers number portfolio E (N2) = expected T of the period length bonds during planning a during of transfers totalnumber portfolio E (N) = expected T of period length cashbalance. E (M) = theaverage cash the The expected perday ofmanaging firm's balances cost as of overahorizon T dayscan be expressed (1)
E(c) = (T I) + 8 ET2) + vE(M).

random in Fromnowon,we assumezero-drift the underlying to case it is muchmoredifficult obtainanwalk. Withthegeneral of number and on tractable results, theydepend a larger alytically parameters. withbase is steadystatedistribution triangular The resulting ratesof the occupancy h and modeZ.3 It follows thattherelative slopesof at...