The Portfolio Approach to the Demand for Money and Other Assets

This PDF is a selection from an out-of-print volume from the NationalBureau of Economic Research

Volume Title: The State Of Monetary Economics

Volume Author/Editor: National Bureau Committee for Economic Research

Volume Publisher: NBER

Volume URL: http://www.nber.org/books/conf75-2

Publication Date: 1975

Chapter Title: The Portfolio Approach to the Demand for Money andOther Assets

Chapter Author: James Duesenberry

Chapter URL: http://www.nber.org/chapters/c6638

Chapter pages in book: (p. 9 - 31)

THE PORTFOLIO APPROACH TO THE DEMANDFOR MONEY AND OTHER ASSETS

James S. Duesenberry, Harvard University

THE theory of the demand for financialassets has come in for a good deal of dis-

cussion in the last few years. Undoubtedly thediscussion has been fruitful and has given usmany new insights into the nature of financialprocesses. But it cannot be said that there isany generally agreed upon view as to the way inwhich those processes work. It would be appro-priate at a conference of this kind to reviewthe different hypotheses and give a systematicsummary of the present state of knowledge.Unfortunately, though I have read the litera-ture assiduously I have found it rather indi-gestible. I do not feel prepared to give a fairsummary of other people's views. I must fallback therefore on giving my own.

In this paper I shall deal with the demandfor liquid assets and money by households andcorporations. Those two groups hold over two-thirds of all liquid assets, and the same generalapproach though not the details can probablybe applied to the demands of unincorporatedbusinesses, farmers, state and local govern-ments. In dealing with the demand for liquidassets we must at least implicitly deal with thedemand for other types of assets, but I shallnot, except incidentally, say anything in detailabout the demand for stocks, bonds, or physicalassets. I shall confine myself to the demand forcurrency, demand deposits, commercial banktime deposits, mutual savings bank deposits,savings and loan shares, savings bonds, andshort-term federal securities. There are, ofcourse, other liquid assets, but I shall havelittle to say about them.

I have occasionally used the term money inthe sense of demand deposits and currency buthave usually referred to those assets specificallyto avoid any confusion with other definitionsof money. But though I am happy to try toavoid the semantic confusion involved in argu-ments about whether any particular assetshould be included under the heading money,I do cling to the view that commercial banktime deposits are significantly different from

demand deposits. For that matter, so is cur-rency, and so perhaps we ought to dispensewith the term money in theoretical discussionsand say clearly what we mean.

In the first section of the paper I have dis-cussed very briefly the conditions under whichliquid assets are supplied. There follow in sec-tion 2 a discussion of corporate motives forholding liquid assets and money and a reviewof some empirical evidence on the relative im-portance of various factors influencing cor-porate decisions. In section this theory ofhousehold demand for liquid assets and moneyis discussed together with some empirical evi-dence.

The Supply of Liquid Assets

The major liquid assets today are cash,short-term government securities, savingsbonds, and time deposits at commercial banks,mutual savings banks, and savings and loanassociations. Various other assets, e.g., com-mercial paper, supply some liquidity and, underother circumstances, could supply more, butwe shall confine ourselves here to short gov-ernments, savings bonds, and time deposits.

Savings bonds are, of course, on tap at fixedyields and may be regarded as exogenously de-termined.

The total volume of short-term Treasurysecurities is determined by the Treasury (as-suming that it may affect conversions if itwishes). For theoretical purposes, we may re-gard the composition of the federal debt as anexogenous policy variable. The fact remains,however, that there is some tendency for theproportion of short-term issues outstanding torise in tight money periods and fall in easymoney periods.

If we are concerned with the determinantsof nonbank liquid asset holdings, then thewillingness of banks to switch maturities be-comes important. In general, banks appear tohave been willing to switch out of very short

19:1

io THE PORTFOLIO APPROACH

maturities into somewhat longer ones in re-sponse to changes in the yield curve. As a re-sult, the supply of under-one-year maturitiesto nonbank holders has considerable elasticityin terms of yield spreads, though the supplycurve is certainly not flat.

The supply of savings deposits is a verydifferent matter from the supply of Treasurybills and commercial paper. In the short run,the supply of savings deposits is elastic at theannounced rates. In the long run (except whenthe regulatory authorities interfere) the rateoffered by savings institutions is related tomarket rates and operating costs. For sav-ings and loan associations, the relevant ratesare mainly those on conventional mortgages inthe areas where they operate. For mutual sav-ings banks, yields on mortgages in their ownareas, on guaranteed mortgages in other areas,and on bonds which they are eligible to buy arerelevant. Illegal restrictions on rates did notinterfere, commercial bank time deposit rateswould presumably tend to reflect yields onmortgages and municipal securities.

In the long run the differentials between sav-ings deposit rates and those available on creditmarket instruments depend not only on costfactors in bank operations but also on the rela-tive supplies of debt instruments of differenttypes. A large supply of mortgages tends toraise savings institution yields relative to othersbecause those institutions have a comparativeadvantage in handling mortgages.

If we are content with a partial analysis, theabove comments are perhaps sufficient. Thereare, however, some additional considerationsaffecting the availability of liquid assets. Letus suppose that for some reason the public'stastes change so that households collectivelywish to switch out of other assets into savingsdeposits. Since savings institutions appear tobe willing to take all the deposits they can getat the going rate, there appears to be no reasonwhy the public should not make the switch. But,of course, that is not true. If the yield spreadsexisting when tastes change are to persist, thesavings institutions must not only be willing totake deposits but they must also be willing tobuy the assets from which the public wants toswitch (if we rule out asset expansion by banks).If the public wished to sell a set of assets having

the same relative composition as the existingportfolios of savings institutions, there is noreason why the switch should not take place.But suppose the public wished to sell stocksand take savings deposits. They cannot sell thestocks to savings institutions. In the end thestocks must remain in the hands of holdersother than savings institutions, If the initialholders come to like stocks less, their pricesmust fall until the original or some otherholders other than savings institutions are will-ing to hold them. That change in yields mayset up repercussions which will result in someincrease in savings deposits (and the shiftingof some assets to savings institutions). Buteven when the yields offered by savings institu-tions are a fixed function of yield on credit mar-ket instruments, the public caiinot trade freelybetween savings deposits and other assets atfixed yields.1

We cannot properly analyze the forces in-fluencing the volume of savings deposits inpartial equilibrium supply and demand terms.Those forces work themselves out through thewhole system of interrelated supplies and de-mands for different kinds of assets.

Corporate Holdings of Moneyand Other Liquid Assets

In this section we shall consider the factorsinfluencing the total amount of liquid assetsheld by corporations and the division of thoseassets between cash and interest-bearing formsof liquidity. We start from the assumption thatthe decision-making process involves first adecision about the amount of liquid assets tobe held and then a decision about the propor-tion to be held in cash.

We begin with a brief outline of the rationalefor holding liquid assets and then consider inequally general terms the rationale of thechoice between cash and other liquid assets.We then turn to the empirical problem of ex-plaining observed movements of liquid assets.We first review some work which seems toexplain the short-term cyclical movements of

'Of course, the public's holdings of stocks will In thelong run be affected by a change in tastes because changesin the yields of stocks relative to other assets will influencethe composition of corporate security issues and the com-position of investment.

THE PORTFOLIO APPROACH ix

liquidity positions during the postwar period,and then consider the trend of liquidity posi-tions in the postwar period. Finally there aresome rather sketchy comments on liquiditymovements in the twenties and thirties.

We then turn to the problem of explainingobserved movements in the division of liquiditybetween cash and other forms of liquidity.Once again, we begin with the postwar periodand then proceed to comment briefly on theevents of earlier periods.

Business Demand for Liquidity.Corporate treasurers wish to have liquid as-

sets on hand for a variety of reasons. First,some minimum of liquid assets is required tocover the day-to-day variations in receipts andexpenditures without continually borrowingand repaying bank loans. Second, many cor-porations have wide seasonal variations in theircash inflows and outflows. Some choose to bor-row from banks during seasons when cash out-flows exceed inflows and repay when the bal-ance of the flows turns the other way. Others,however, raise enough from long-term sourcesto cover all or part of their seasonal cash out-flows, and therefore have surplus liquidityduring the seasons of net cash inflow. In doingthat they make an interest sacrifice which de-pends on the relations between the cost of long-term funds, the cost of bank borrowing, andthe yield on liquid assets.2

Third, most, though not all, business firmsseem to feel that it is desirable to fund all ora very large part of their tax liability. A so-phisticated treasurer may feel that he needonly be prepared to meet the net reduction intax liability which will occur when prafits de-cline, but most of them do not appear to takethat view.

Fourth, most firms wish to have liquidity tomeet the problems arising from a decline incash flow from current operations during a de-pression. In a depression a firm must be ableto draw down liquid assets or borrow to meetthe cumulative difference between net cashflow from operations and the sum of (i) debt

'In return for that cost they reduce the risk of beingburdened with short-term debt during a business declineand reduce the cost of long-term borrowing by improvingthe current ratio.

service (2) dividends a minimum level ofinvestment. It should be noted that net cashflow from operations can become negative be-cause, in order to maintain its market share,a firm is willing to continue production whenaverage variable cost exceeds price.

It is not, of course, strictly necessary tomaintain dividends in periods of adversity, butit is clear that many firms are willing to bearsome cost in order to do so.

Investment will, of course, decline to a lowlevel during a depression but, even when thereis a great deal of excess capacity, certain in-vestments can be avoided only at great cost.Those investments include replacements nec-essary to maintain production, and investmentsrequired to adapt to changing market condi-tions. In addition, technical developments orthe possibility of buying out financially weakcompetitors may present investment opportu-nities promising very high returns.

A firm always has the possibility of financingthose cash requirements by borrowing, but ina depression lenders may not be disposed togamble on the firm's prospects. If they arewilling to lend they may require some measureof control or impose restrictions on the bor-rower's action.

To avoid or reduce borrowing under adverseconditions, a firm can raise more long-termcapital than it needs for current outlays, dur-ing periods when it can do so advantageously,and build up a stock of liquidity against ad-versity. It then pays the difference betweenthe cost of long-term funds and return on liquidassets (less the cost of borrowing avoided dur-ing periods of adversity). The cost may besomewhat reduced because the improvement inits current ratio will reduce the cost of itslong-term borrowing.

In all the cases so far mentioned, the firmgains in convenience and reduction of risk bypaying more in interest charges than wouldotherwise be necessary, and the amount ofliquidity maintained will be determined at thepoint where the marginal gains are balancedby the costs of obtaining additional liquidity.That implies, of course, that the amount ofliquidity held by the firm Will (other thingsequal) tend to increase as long-term capital

12 THE PORTFOLIO APPROACH

costs f all relative to the yields on Treasury billsor other short-term investments.

We should, of course, expect business liquid-ity at a given phase of the business cycle tobe roughly scaled to the volume of sales or totalreceipts and payments — with the proportion-ality factor varying with interest rate differ-entials. Since risk is involved, we should ex-pect firms to hold less liquidity as they becomemore confident of the stability of the economyand to increase liquidity ratios with a deterio-ration of confidence. Finally, the level of cor-porate income tax rates and the length of thecollection lag should influence the volume ofliquid assets held.

Before going on to review some empiricalevidence on these propositions, it is necessaryto introduce some other considerations.

The variation in corporate .liquidity over thecycle reflects some "mechanical" aspects ofcash inflows and cash outflows. Firms that donot rely heavily on bank financing will tend tohave certain passive inflows and outflows ofcash (which may be shifted into other formsof liquid assets). To put it another way, firmsmay have certain target levels of liquid assetsbut they do not always eliminate discrepanciesbetween actual and target levels very rapidly.

During recovery years, like 1955 and 1959,corporations tend to gain liquidity becauseprofits rise rapidly while tax payments lag.Fixed investment appropriations rise rapidlybut actual outlays rise more slowly, Inven-tories and receivables also rise but a consider-able part of the increase is bank financed. Theresult is a large net cash inflow. Firms gainingliquidity in this way could pay off long-termdebt, but have little incentive to do so sincethey are committed to higher investment out-lays in the near future and will also have largertax payments to make.

In the later years of the cycle, retained earn-ings and depreciation level off, tax paymentscatch up, and the rate of investment outlayscatches up with new appropriations for in-vestment. Some of the liquidity gained earlieris then disgorged.

The record in slump years is mixed. Cor-porations lost liquidity in and 1960 butgained in 1958. The difference appears to bedue in part to (i) differences in the relative

changes in profits and tax payments arisingfrom changes in the tax law, (2) differences inthe magnitude of inventory reduction, dif-ferences in the magnitude of the change inplant and equipment investment.

Corporate Demand for Demand Deposits.Corporate demand for demand balances has

to be explained in terms of the rationale of thechoice between demand balances and othertypes of liquid assets which earn interest —in this case, mainly Treasury bills.

As in the case of households, a substantialpart of corporate liquidity is held to coverfairly near-term net cash outflows arising fromthe uneven rates of receipts and expenditures.The gross return to be obtained from holdingTreasury bills depends on the amounts in-volved, the interest rate, and the length of theholding period. The gross return is offset bythe transactions cost of going into and out ofbills — dealer spread and any other directcosts. However, dealer spreads on bills arevery low so that the direct cost of transactionsin bills can be earned by holding bills for veryshort periods even at low interest rates.

However, the direct costs of transactions areonly part of the cost of holding Treasury bills.If bills are held only for a few definite, largeout payments — e.g., tax payments — the onlyadditional cost is a little thought on the part ofthe treasurer. But as soon as a firm embarkson a program of trying to earn interest by pre-dicting cash inflows and outflows, and investingtemporary excess funds, it has to bear someoverhead costs to keep track of its cash posi-tion.

In entering on a program of investing short-term surpluses of funds and in deciding how farto carry it, a firm must balance the expectedaverage return from its bill holdings (or anincrement in them) against the overhead costof controlling its cash position. The expectedreturn will vary with the average rate of inter-est on bills expected over a period of years.The cost will vary with the tightness of thecash management. As we pass from simple op-erations such as funding tax liabilities to thevery close cash management practiced by a fewlarge companies, the cost of increasing theaverage amount invested (for a firm on a given

THE PORTFOLIO APPROACH 13

scale) by closer management will rise. Thecost per dollar invested will rise as the scaleof the firm becomes smaller. We should ex-pect, therefore, that at low interest rates onlylarge firms will use Treasury bills for anypurpose other than funding tax liabilities, andeven the large firms may not find very closecash management worthwhile. A rise in inter-est rates should push out both the extensive(size of firm) and intensive (closeness of cashmanagement) margins and cause a shift fromcash to bills. However, since the costs of cashmanagement are mainly overhead costs, we donot expect much shifting between cash andbills in response to short-run fluctuations inbill rate. Instead, we expect shifts in responseto changes in the average rate expected for aperiod of years — probably best measured bya fairly long moving average of past rates.Moreover, some of the costs are costs of gettingstarted and learning the tricks of cash man-agement, so that a rise in bill rates may resultin a shift from cash to bills. A later fall (evenon a long-term basis) may not bring the cash-bill ratio back to its initial position.

Shifting from cash to interest-bearing liquidassets is not the only way to reduce cash bal-ances in relation to activity. Close attentionto cash management may enable a firm to re-duce its operating cash without any investment.For example, measures to speed up processingand collection of checks may reduce the floatof unavailable cash. There are many examplesof this sort. It is not clear whether develop-ments of this sort represent a response tochanging interest rates or whether they aresimply improvements in managerial techniqueswhich, like other changes in techniques, occureven when no factor price has changed. Onewould expect, however, that enthusiasm formeasures to reduce the need for cash wouldincrease with rising interest rates. It shouldbe noted that increased efficiency in the use ofoperating cash is likely to reduce total liquidityand to affect cash directly so that the share ofcash in total liquidity will fall.

Corporate managements also appear to haveachieved some reduction in their cash balancesin relation to activity by sharper negotiationwith banks. It is customary for corporationsholding payroll, dividend, or other accounts in-

volving a large volume of check processing tomaintain an average balance on which the bankcan earn a return (in lieu of a service charge),which covers the costs of servicing the account.In recent years, a good many corporationswhich do not depend on bank loan finance havetried systematically to hold these balances downto a minimum. But, so long as the custom ofremunerating banks by holding balances ratherthan by explicit service charges continues, asubstantial amount of corporate funds will betied up in this way.

Short-Run Movements in Total Liquidity.In discussions of the demand for money and

liquid assets, attention is usually centered onthe elasticity of demand for liquid assets withrespect to interest rates. In fact, however, thevariation in corporate liquid asset holdings overthe business cycle is principally due to varia-tions in the cash flows from operations. Cashflows are generated by retained earnings, plantand equipment outlays, increases or decreasesin inventory and variables, tax accounts and taxpayments. The net balance of those flows —each of which has a large cyclical variation —is first reflected in corporate liquid asset hold-ings. Managements then have to decide whetherthe resulting liquidity position is too high ortoo low with respect to some target. If theliquidity position is out of line with this targetposition, firms must then raise funds from out-side sources or repay debt.

Interest rates may enter this picture in twoways. On the one hand, target levels ofliquidity may be influenced by the cost ofholding liquidity, as indicated above. On theother hand, even if target levels are not affectedby interest costs, the timing of borrowing maybe influenced by cyclical movements in theinterest rate.

Following this reasoning, we may treat liquidasset holdings as an inventory and explainmovements in liquidity in terms of a stockadjustment process of the same type used inexplaining physical inventories.

This has been done by Locke Andersonin a paper presented to the Econometric So-ciety. Briefly, his results can be interpreted asfollows.

i. Target levels of liquid asset holdings ap-


pear to depend on sales and tax liability, withthe desired holdings of liquidity equal to about15 per cent of annual sales and 6o per cent oftax liability.

2. The amount of outside funds raised in aquarter (for a given position in terms of othervariables mentioned below) increases by about2 for every dollar of increase in the differencebetween actual and target liquid asset holdingsat the start of the quarter. That is, firms actas though they tried to close the gap betweenactual and target holdings in about a year.

3. Borrowing in a given quarter also respondsto changes in cash flows from retained earningsand depreciation, plant and equipment expendi-tures, and inventories and receivables. Short-term borrowing declines by about 75 per centof any increase in retained earnings, but theresponse of borrowing to cash flows of the othertypes is much weaker. As a result, an increasein the rate of inventory investment will reduceliquid assets, and vice versa.

4. The rate of borrowing is also significantlyinfluenced by the debt position of corporations.

Finally the rate of long-term borrowing issignificantly influenced by the corporate bondyield. A one percentage point increase in thebond yield appears to reduce corporate bor-rowing by $700 million per quarter. If allother flow variables remained constant whilethe bond yield rose, the result would be a de-cline in borrowing followed by a gradual riseas the liquidity stock was reduced. A one-timerise in the interest rate would result in a one-time reduction in the liquidity stock. However,there are some indications that the relevant in-terest rate variable is not the absolute rate butsome indicator of the cyclical position of therate. That view is supported by the fact thatthe bill rate did not appear to have any signifi-cant influence on borrowing. Since the cost ofliquidity is the difference between bill and bondyield, it seems probable that the interest ratecoefficient represents a timing variable ratherthan a variable affecting the long-term level ofliquidity.

Anderson's results are supported not only byaggregate regressions for all manufacturing butalso by separate analysis of a number of manu-facturing industries. They seem to show, first,that the bulk of the variation in corporate

liquidity is explained by variations in operatingcash flows. Second, they do show that the in-terest rate can have a very substantial short-run effect on corporate borrowing but leave ituncertain whether the upward trend in interestrates plays a significant role in determining thetrend of corporate liquidity.

POSTWAR TREND IN TOTAL LIQUIDITY. Weneed not give any account of movements ofcorporate liquidity in the years immediatelyafter the war. It seems reasonable to take theview that most corporations had made theirbasic adjustment to postwar conditions byabout 1952. Since that time there has been adownward trend in the ratio of corporate liquidasset holdings to GNP and to corporate sales.

Part of the decline is probably due tothe decline in the ratio of tax liabilities tocorporate sales as a result of the decline inprofit margins and the shortening of the lagbetween accrual and payment. The ratio ofliquid assets, less tax liabilities, to salesshows almost no decline. However, it is goingtoo far to offset tax liabilities against liquidity,one for one. We know that some corporationsborrow at tax dates so they cannot have fullyfunded their tax liability. Some corporationshave less total liquidity than their tax liability,so it is obvious that they cannot have done so.Anderson's research suggests about óo per centfunding as a norm. If that figure or one some-where near it is used, then the ratio of corporateliquidity to GNP (adjusted for the tax factor)fell from about .11 in 1952, to .09 in 1960 and1961. The downward drift in the relative hold-ings of cash, governments, and time depositsmay have been offset by an increase in holdingsof various other liquid assets not included inour figures, but those other assets cannot haveincreased enough to offset the downward trendin relative holdings of the ones included.

To what should we attribute the decline?Even if we took Anderson's results with respectto the interest rate at face value, we wouldexpect to get a decline in liquidity of less thani per cent of GNP from the 1952—6I rise inBAA bond yields. Moreover, the net cost ofholding liquidity has increased very little be-cause the average spread between bill and bondyields has changed relatively little. Rising in-

THE PORTFOLIO APPROACH

terest rates may have had some effect on thetrend of liquidity ratios, but it seems doubtfulthat it has been very large.

Three other factors have probably played arole: (i) improved cash management, partlyas a result of higher interest rates but alsosimply as a part of the general and continuingeffort to reduce costs; (2) improved confidencein the stability of the economy may have in-duced some firms to reduce holdings of liquidityfor protective purposes; (3) some firms stillhad "excessive" liquidity in 1952 and haveworked it off since.

I know of no way to measure the influence ofthese factors but it seems unlikely that morethan a fraction of the decline in liquidity ratiosis due to changes in interest costs.

Liquidity in the Twenties Compared with Post-war Liquidity.

The ratio of corporate liquidity to GNP dur-ing the 1920's stood at about .14 until 1928,when it rose sharply to .17, returning to about.15 at the end of 1929. The ruling ratio duringthe twenties was higher than that ruling in 1952(after adjustment for tax liability). Moreoverliquidity ratios showed no downward trendduring the twenties. The rise at the end of thetwenties is probably due to the large volume ofstock issues in and 1929, and the risewould be even larger if corporate holdings ofcall loans were in. cluded in the ratio.

The net cost of holding liquidity has prob-ably not increased significantly between thetwenties and the fifties. Indeed (leaving callloans aside) it has probably decreased becauseof income tax and the availability of Treasurybills. It does not seem reasonable to arguethatpeople have become more confident about in-come stability now than they were in the twen-ties. We can conclude, it seems to me, that thedifference in liquidity ratios between the twoperiods is due to factors other than confidence orthe cost of liquidity. It is hardly surprising thatchanges in the composition of output, inmethods of financing, in the concentration ofindustry, should have produced some substan-tial changes in liquidity ratios.

It seems unlikely that, interest costs aside,practices with respect to liquidity should fol-

low simple and immutable laws over longperiods.

Liquidity Movements in the Great Depression.Corporate liquidity rose relative to GNP

from the already high level in 1929 to 1932 and1933. Thereafter liquidity ratios fell slowly,but even at the beginning of World War II theratio of corporate liquid asset holdings to GNPwas .iô. With some allowance for corporatetax liability, the corporate liquidity ratio in1941 was near the level of the 1920's.

It seems probable that most of the swing inliquidity during the thirties was attributable tothe confidence factor. Firms preferred to holdon to liquid assets in the early thirties ratherthan pay back debt, because the chance thateconomic conditions would deteriorate wastaken very seriously. With recovery, firms per-mitted liquidity ratios to decline gradually, Themovements of liquidity ratios conform muchmore closely to one's guess about the state ofbusiness confidence than to the variations inthe net interest cost of holding liquidity.

Cash Versus Other Forms of Liquid Assets inthe Postwar Period.

In the years since 1952, corporate cash hold-ings have increased slowly and steadily, whileholdings of governments have shown little trendand have varied from year to year in a rangeof about billion. Another set of regressionstudies by Anderson suggests that most of theshort-term variations in liquidity positions arereflected in holdings of government securities,while cash positions respond much more weak-ly to those factors. It is easy to understandthat, if a firm has an improvement in itsliquidity position as a result of a cyclical swingin earnings and finds it inconvenient to repaydebt, it will invest the surplus. When the cashflow picture reverses, it will not have any sur-plus cash because of its previous action and willhave to sell bills if it does not borrow. In spiteof a number of efforts to do so, I believe no onehas found a statistically significant relationshipbetween short-term variations in bill yields andthe distribution of corporate liquidity betweencash and governments. That is to be expectedin view of the considerations with respect tooverhead costs given above.

i6 THE PORTFOLIO APPROACH

It does not follow that the trend in the ratioof cash to other forms of liquidity has not beeninfluenced by interest rates. The share ofliquidity in the form of time deposits (thoughstill small) has risen since 1952, and that canprobably be attributed to interest rates (thoughthe willingness of banks to take corporate timedeposits is also a factor).

The ratio of government security holdingsto cash holdings has fallen since 1952 but, ifwe make allowance for the large proportion ofgovernments held against tax liability, the ratioof cash to total liquidity has fallen from 75 to72 per cent. That is a relatively small declineand does not constitute very impressive evi-dence of a large-scale switch from cash to bills.

The proportion of liquid assets held in theform of cash is of course influenced by otherfactors. An increase in efficiency in the use ofoperating cash works to reduce the ratio of cashto liquid assets. On the other hand, a reductionin protective liquidity resulting from a gain inconfidence should tend to reduce bills relativelymore than cash.

There is room, therefore, for some switchfrom cash to Treasury securities and time de-posits, but there is no evidence that rising inter-est rates have produced a strong swing tointerest-bearing forms of liquidity or that risinginterest rates have accounted for a large partof the reduction in the ratio of corporate cashto sales and GNP.

Cash Versus Other Forms of Liquidity in thePrewar Period.

The ratio of corporate cash to total liquidassets rose slightly in the early years of the de-pression and then after 1933 rose to nearly 8oper cent of total liquidity. The shift must, Ithink, be put down to the very low levels ofinterest rates on time deposits and short-termgovernment securities ruling in the late thirties.

The ratio of corporate cash to total liquidityin the 1920'S was higher than in the postwarperiod but, after adjustment of postwar datafor tax liabilities, the cash liquidity ratio in thetwenties was slightly higher than the one rulingin the postwar period.

The differences are not great enough to callfor extended comment except to note that thedevelopment of Treasury bills does not seem

to have had a great effect on the distribution ofcorporate liquidity.

Household Demand for Liquid Assets and Money

Demand for Liquid Assets.Personal motives for holding liquid assets —

transactions and precautionary motives, liquid-ity preferences, risk avoidance — are all sofamiliar that it is not necessary to discuss themin any detail. Just as in the corporate case, anindividual who holds liquid assets takes a re-duction in expected yield on his portfolio inreturn for a reduction in risk and inconven-ience. The loss in expected yield depends onthe price difference between the expected yieldsof variable assets real estate, stocks, long-term bonds — and the yields on liquid assets.Just as in the corporate case, we expect that,other things equal, the amount of liquid assetsan individual will wish to hold will decline asthe cost of holding them increases. We alsoexpect that increased confidence in the futurestability of income will reduce the demand forliquid assets, and vice versa. An increase in thevariance of the expected performance of vari-able price assets will increase the demand forliquid assets, while a decrease will reduce it.

Some of the reasons for holding liquid assetsare related to uncertainties or unevenness inthe flow of receipts and expenditures. On thataccount we might expect the demand for liquidassets to increase, other things equal, with thelevel of permanent income. But liquid assetsare also required in an optimum portfolio evenwhen there is no problem of income or expendi-ture variation. Other things equal, then, weshould expect the demand for liquid assets togrow with both the level of income and the levelof assets.

In individual portfolio management the sizeof the portfolio has an important influence onthe proportion of assets held in liquid form.That is so for two reasons. Since borrowing iscostly and inconvenient, most people wish tohold enough liquid assets to provide for short-term variations in income and expenditure.Persons whose total financial assets are smallin relation to their incomes will find it advan-tageous to hold all their assets in liquid form.Second, asset management is an activity with

THE PORTFOLIO APPROACH '7decreasing costs to scale. The cost in terms ofcash and effort of choosing assets subject torisk is much smaller per dollar invested for alarge portfolio than for a small one. The netgain to be obtained from buying variable pricesecurities as opposed to savings deposits of onetype or another is not likely to be worth thetrouble for the holder of a relatively smallportfolio.

It seems probable that a substantial volumeof liquid assets is held in connection with assettransfers. Individuals who sell marketablesecurities or real estate may hold funds pend-ing reinvestment for periods ranging from afew days to several months. In some cases theymay go liquid because they are bearish on var-iable price assets generally. But it is verycommon for people to sell a particular assetbecause they consider its net yield prospectsunsatisfactory without having chosen anotherasset. They will hold liquid assets until theyfind a satisfactory alternative investment. Wehave no idea. what volume of assets is tied upin this way but it may be very substantial.

Finally, there appears to be an interchangebetween strictly fixed-price assets — demanddeposits, currency, savings deposits, and sav-ings bonds — and assets with low credit riskand price variability, particularly high gradebonds. Individuals who have sufficient liquidassets to take care of short-term variations inincome and expenditure may wish to have addi-tional low-risk assets in their portfolio whichthey expect to hold for a fairly long time. Thefact that savings deposits of various types canbe converted to cash at any time with no trans-actions cost is of relatively little significance ifone plans to hold an asset for a long time. Pricevariability is of some significance but those whoplan to hold to maturity, anyway, need not giveit a very heavy weight. Savings deposits ofvarious types are therefore close substitutes forhigh quality bonds, particularly those with onlymoderately long maturities.

On that basis we should expect that, duringperiods when market yields on bonds are lowrelative to time-deposit rates, the flow of house-hold funds into bonds would be relatively lowand the flow into time deposits relatively high.Conversely, when bond yields rise relative totime-deposit yields, we should expect household

bond purchases to rise relative to householdtakings of time deposits.

Of course the competitive relationship be-tween time deposits and bonds is not just acyclical phenomenon. The relative levels oftime deposit yields and bond yields over thewhole cycle will influence the division of in-dividual portfolios. But because time deposityields move slowly relative to market yieldsover the cycle (though linked in the long run tomarket yields), the cyclical influence of bond—time deposit substitution is much more appar-ent than any long-run substitution.

LIQUID ASSETS AND EXPENDITURE. A numberof writers have expressed the view that varia-tions in household holdings of liquid assets havea strong influence on the rate of consumer ex-penditure. The rationale of that view has never.been entirely clear to me.

It seems reasonable to expect that an increasein the real net worth of the household sectormight tend to reduce saving and increase con-sumption. If people are saving in order toaccumulate assets for some particular purpose,the desire to save may wane as they approachtheir goal. Of course, they may discover orrecognize new goals for accumulation as theysatisfy old ones, so it remains an empiricalquestion whether an increase in net worth orin the ratio of net worth to income actuallydepresses saving.

But why should the possession of liquid as-sets, as distinguished from other assets, have aspecial effect on saving? One does not comeany closer to any goal for. accumulation byholding a deposit in a savings institution thanby holding an equivalent amount of stocks andbonds.

The only difference seems to be that one canconvert liquid assets into cash more easily andquickly than other kinds of assets. Conse-quently, one can give in to impulses to spendmore easily if one holds liquid assets than ifone holds other kinds of assets. There is someplausibility in that argument, but it obviouslyonly applies to a limited part of the variation inliquid assets. The impulse consideration doesnot apply to persons who save regularly a sub-stantial proportion of income or to persons whoalways have a substantial liquid position. Forreasons which I will indicate below it seems

r8 THE PORTFOLIO APPROACH

likely that — except for the war and earlypostwar periods — most of the variation ittliquid asset holdings is in the holdings of highincome, high asset holders. In that case it isunlikely that cyclical variations in liquid assetholdings have much to do with variations insaving.

Demand for Money.In the last section we discussed the demand

for liquid assets as a group without any distinc-tion between money and other liquid assets.We must now turn to the question why peoplehold part of their liquid assets in noninterest-bearing demand deposits and currency. It shouldbe noted at the outset that, in the literature, thereasons given in the last section for holdingliquid assets are often given as reasons forholding demand deposits and currency. Thatmay have been appropriate in periods whenother forms of virtually riskiess, readily mar-ketable assets were not generally available.But it is not a satisfactory answer nowadays.The demand for currency and demand depositsmust be analyzed, first, in terms of choices be-tween liquid and nonhiquid assets and, second,in terms of choices between currency and de-mand deposits and other forms of liquid assets.

DEMAND FOR CURRENCY. The total amountof currency outstanding since the war hasvaried between $25 and $30 billion. Estimatesmade by the Federal Reserve Board of Gover-nors suggest that about one-third of this amountis in business hands and the rest either in thehands of households, lost, destroyed, or goneabroad. It is fairly obvious that the bulk of thecurrency in nonbusiness hands is not being usedfor pocket money or being carried around bypeople who do not have bank accounts. Theamount of currency not in business hands rep-resents nearly a month's wages for the entirelabor force. Over half of American familieshave checking accounts and some of the re-mainder deposit pay checks in savings accounts•and withdraw currency and registered checksas needed. Moreover, the bulk of the familieswho do not use bank accounts receive wagesweekly. A full week's wages for one part ofthe families and an average of a couple ofhundred dollars for the rest — which seems agenerous estimate even after allowing for

travelers — will not account for as much as $robillion of currency.

Some of the remainder is, no doubt, lost,destroyed, or gone abroad. The rest must be inhoards for some special reason. These wouldinclude currency used in illegal transactions orheld by small businessmen and professionalswho receive currency and hold part of it toevade taxes, and hoarded savings of farmerswho saved during the war and distrusted banks.

The amount of currency outstanding nearlydoubled during the depression and rose by afactor of about four during the war. After thewar it fell slightly until 1950 and has risenslowly since then by nearly billion, It shouldbe noted that the amount of currency outstand-ing showed little trend in years from 1900 to1914, rose sharply during World War I, andthen remained more or less stationary duringthe twenties.

If currency were used only for transactionpurposes, we should expect the amount out-standing to rise with income but at a slowerrate because of the increasing use of checkingaccounts, registered checks, the spread ofcheck-cashing facilities, and the increased .useof credit cards. We might also expect thatcurrency held for tax evasion and illegal activi-ties would grow with the scale of the economy— if the incentives for tax evasion do notchange much.

On the other hand, it is likely that specialfactors connected with war resulted in the gen-eration of abnormally high levels of hoardingrelative to income and tax rate levels. Thegradual liquidation of some of those wartimehoards may be offsetting the other factors tend-ing to make the currency outstanding to grow.That position gains some support from the factthat the currency grew rapidly during WorldWar I and then leveled off during the twenties,even though income grew. The rate of liquida-tion of currency hoards other than those con-nected with tax evasion or illegal activities mayhave been speeded up by the rise in interestrates, but we have no real information on thatpoint.

HOUSEHOLD DEMAND FOR DEMAND DEPOSITS.We can approach the analysis of the demand fordemand deposits by asking why a man, giventhat he has some liquid assets, should hold them

THE PORTFOLIO APPROACH '9in a form which yields no interest. Certainly amajor part of the answer lies in the fact thatchecking accounts are more convenient thanother liquid assets and that funds left withsavings institutions for short periods yield noreturn or a very small one.

Persons who hold liquid assets against a cer-tain or fairly probable excess of payments overreceipts in the near future will not find the returnfrom savings deposits or savings bonds worththe trouble of converting from cash to earn-ing assets and back again unless the amountsinvolved are very large. The income transac-tions demand for money will certainly accountfor some substantial amount of personal de-mand deposits. As the savings deposit interestrate rises, the proportion of "transient" liquidassets held in the form of demand depositsshould decline.

The full theory of an optimum demand de-posit inventory policy is just as complex as anyother kind of inventory policy, but a simpleexample suffices to make the point in question.Suppose we consider only the disposition ofliquid assets held against known lump-sum out-payments to be made at known dates andwhich cannot be financed out of expected netcash inflows in the intervening period. A sumof p dollars held for m months yields the holder

p x X r dollars (when r is the yield on sav-ings deposits neglecting compounding). If anindividual requires a given dollar return tomake worthwhile one round trip from cash tosavings deposit and back to cash, the size ofthe payment p and the length of the interval mrequired to earn that amount of dollars ob-viously becomes shorter as r rises. When in-terest rates are low, savings deposits will beheld only against large distant payments. Asinterest rates rise, people will hold savingsdeposits against smaller nearer-term paymentswhich will produce a shift from demand bal-ances to savings accounts.

It is unlikely, however, that the bulk of per-sonal demand deposits are held for incometransactions purposes. On January 31 (whichis about the low point of the year for indi-vidual deposits), banks' records show that 85per cent of personal demand deposit accountshad less than $x,ooo, but those accounts had

only about 30 per cent of the total amount ofpersonal demand deposits. Since the largerholders have more than one account, it is prob-able that the remaining 14 per cent of the ac-counts were held by no more than zo per centof the persons holding demand deposit ac-counts. And since nearly half of families haveno demand deposit account it must be con-cluded that about per cent of families own70 per cent of personal demand deposits.Moreover, half the personal demand depositsare in accounts of over $5,000.

No doubt some part of the relatively largedeposits is required for income transactions.But it seems probable that a large proportionof the larger personal demand deposit accountsis held in connection with financial transac-tions.

At a rough guess, individual purchases andsales of stocks, bonds, real estate, and otherassets amount to something like $ioo billionper year. An average holding period — be-tween sale of one asset and purchase of an-other — of about four months would tie upover $30 billion. If half that sum were indemand deposits, $x5 billion would be ac-counted for. I have no way of testing whatamounts are tied up in asset float, but it seemsprobable that they are a significant part ofpersonal demand deposits.

Now any individual who sells an asset andplans to reinvest in nonliquid assets, at a timesome distance in the future or at an unspeci-fied time, has the option of keeping his fundsin a demand deposit or obtaining interest froma time deposit. Persons who plan to hold forperiods less than a month cannot get interestfrom time deposits generally, and people inhigh tax brackets, who are interested only incapital gains, may not bother to try to get it.At low interest rates the ptoportion of peoplewho will take the trouble to get time depositinterest in the circumstances under discussionis low. As rates rise, the proportion willing totake the trouble will rise and this will tend toshift funds (in relative terms) from demanddeposits to time deposits.

A SUMMARY AND A MODEL. My conclusionson the relation of liquid assets to income, in-terest rate, and interest rate differentials maybe summarized in the following way.


r. Composition of portfolios. Persons whosetotal financial assets are relatively small willtend to hold them all in liquid form becausethe differential return from other forms offinancial assets is too small to make the addi-tional effort required worthwhile, and becausethe probability of occurrence of a situation re-quiring conversion of a large proportion offinancial assets to cash is high.

Persons with larger portfolios of financialassets will divide them between liquid and non-liquid assets. The proportion held in liquidform will tend to increase if confidence in thestability of income deteriorates, if the differ-ential between the expected net yield on non-liquid assets and that on liquid assets de-creases, if the variance of the expected yieldon nonliquid assets increases, if the ratio oftotal assets to income decreases. In particular,an improvement in confidence will tend toraise the value of stocks in relation to incomeand reduce the need for protective liquidity,thus tending to reduce the ratio of liquid as-sets to total financial assets. A change in thedifferential between mortgage high-gradebond yields will — if it persists long enough tobe reflected in savings institution yieldstend to cause a redistribution between holdingsof savings deposits and near-liquid assets likehigh-grade bonds.

Since savings deposit yields move slowly, thedifferential between savings deposit yields andbond yields will reflect short-term movementsof bond yields. A cyclical increase in bondyields tends to draw funds from liquid assetsto bonds, and vice versa.

2. Liquid assets in relation to income. Fora given state of confidence, relation of totalfinancial assets to income, and given interestrate differentials, we should expect the liquidasset holdings of persons with relatively largefinancial assets to grow from cycle to cycle inrough proportion to income.

For persons with relatively small total fi-nancial assets, our expectation about the liquidasset-income ratio is less clear. If changes inthe liquid asset-income ratio do not influencethe savings ratio then, over a decade in whichthe growth rate of income is above average, theratio of liquid assets to income for small as-

set holders should tend to fall. But, becausethe gross financial savings ratio varies notonly with the total savings ratio but also withthe amount of net investment in housing equityand the net flow of consumer credit, we can-not reach any very clear conclusion on theprobable movements of the ratio of liquid as-sets to income for persons with small portfolios.We cannot, therefore, attach any great signif-icance to observed movements in the ratio ofliquid assets to income.

3. Money holdings versus liquid asset hold-ings. In general, we expect that as the yieldson savings deposits rise, the proportion of liquidassets held in the form of depositswill decline. However, the relationship be-tween interest rates and the distribution ofliquid assets between demand and savings de-posits will differ, as between different classesof people and as between assets held for dif-ferent purposes. Finally, it should be notedthat the yield on savings deposits should beinterpreted to represent not only the rate ofinterest or dividend paid, but also the wholecomplex of advertising and selling effortswhich may induce people to shift from demandto time deposits.

The whole position may be summarized interms of a few very simple equations. Let usfirst divide households into high-asset and low-asset households. Low-asset households holdall financial assets in liquid form. One partof their liquid assets is held for purposes di-rectly related to income and these "transac-tions" holdings are proportional to income; theremainder is a residual. The proportion oftransactions assets held in the form of demanddeposits is a decreasing function of a movingaverage of savings deposit interest rates(strictly speaking, separate rates for differenttypes of institutions and different locationsshould be used — a single rate is used only asa shorthand device). The proportion of theresidual liquid assets held in demand depositform is also a decreasing function of savingsdeposit rates. In general, since the residualbalances are by definition not needed for near-term outlays, a smaller proportion of thosebalances will be held in demand deposit formthan the proportion of transactions balances.

The change in liquid assets of the low-assetgroup over any time period equals the grossfinancial saving of the group during the period

= GFSL.

For the high-asset group, the same considera-tions govern the division of liquid assets be-tween demand deposits and others, except thatwe should add a factor to allow for the assettransactions demand for liquid assets and elim-inate the residual element.

The final term is really another kind of assetfloat which arises from the fact that personswho normally make little use of savings de-posits will take some time to shift from demanddeposits to savings deposits, if they should ac-cumulate liquid funds as a result of a decline inthe attractiveness of securities.

The variable WFH will vary in proportion toincome if the share of property income, valua-tion factors, and the concentration of incomeremain constant.

Gross financial saving for the higher-incomegroups should not be much influenced by varia-tions in consumer credit or net investment in

a The question of permanent income arises here. Whenan individual's income declines and he remains a positivesaver, he may keep his working cash balance unchanged,out of force of habit. If he becomes a negative saver (incash-flow terms) he must draw down liquid assets and Ishould be inclined to think he would draw down his cashbalance because he holds it to absorb fluctuations in ex-penditure relative to income. If aggregate income falls, weshall have three groups of people: (I) those whose incomesare unchanged and who, other things equal, keep cashbalances unchanged; (a) those whose incomes fall but whoremain positive cash savers — if their cash balances do notfall, the ratio of cash balances to current income rises —the permanent income factor; those whose incomes falland who become negative savers, draw down cash balancesabsolutely and relatively to income. The buffer stockfactors 2 and 3 affect the ratio of cash balances to incomein opposite directions. For simplicity I have written theequation as though the two effects cancel out.

residential property but may show some tend-ency to rise when income rises rapidly. How-ever, it would take us too far afield to discussthat point here.

Movements of Household Holdings of LiquidAssets and Money in the Postwar Period.

It is clear that if (i) the ratio of wealth toincome, (2) yields on nonliquid assets and onsavings deposits and the size distribution

M

of wealth are all constant, the ratio of demanddeposits to income will tend to be constant ex-cept for minor fluctuations resulting from var-iations in the ratio of gross financial savings toincome.

If the other conditions are satisfied whilesavings deposit yields have an upward trend,there will be a downward trend in the ratio ofdemand deposits to income.

Now suppose that there are short-run varia-tions in interest rates as a result of changes inmonetary policy and changes in economic ac-tivity. A fall in investment activity will be ac-companied by a decline in corporate retainedearnings and a rise in government deficit. Inmild depressions such as we have had in thepostwar period, there is little decline in grossfinancial saving. Changes in required reserveratios make it possible for banks to bid forsecurities and drive down interest rates to in-duce to reduce their purchases ofsecurities and increase holdings of both de-mand and time deposits. When disposable in-come remains constant, the residual liquid as-sets of those with small portfolios will also riseand some part of this will take the form ofdemand deposits.

Of course, households, who withdraw fromor are pushed out of marketable securities,

Thus

THE PORTFOLIO APPROACH 2!

DDL AyL + (LL — whereDDL = demand deposits of low-asset holdersF = a moving average of savings deposit yieldsyL = the income of the low-asset group

when

DD11 = + (ia) )< D3 (rM — W" + —

= demand balances of high-asset holders= income of high-asset holders

WFH = total financial wealth of high-asset holdersrM = net expected yield on marketable securities.


shift their funds into savings deposits as wellas into demand deposits (indeed, in the post-war period the increase in the flow into timedeposits in recession years has been consider-ably greater than the increase in the flow ofhousehold funds into demand deposits). Sincesavings institutions hold little cash (unless wecount reserve absorption by commercial-banktime deposits as the equivalent of cash), thesefunds come right back into the market anddraw securities away from households.4 How-ever, after a time, savings institutions begintaking mortgages on new houses and the in-crease in economic activity increases the trans-actions demand for cash.

Households go into cash and time depositswhen interest rates fall, partly because bondyields are low relative to savings depositsyields, and partly because they expect a re-covery and higher yields in the future. Somehouseholds, of course, speculate for a capitalgain from a continued fail in interest rates,hoping to get out before the recovery. They,however, are usually bank financed and there-fore merely supplement the demand for bondsgenerated by the expansion of bank reserves.

The process described above for the down-swing works in reverse on the upswing, thoughnot in an entirely symmetrical way. Risinglevels of income will increase transactions bal-ance requirements for liquidity, but a con-tinued upward trend in the moving average ofsavings deposit rates works to lower the pro-portion of such balances held in demand de-posit form. The same considerations apply tothe effects of increasing total wealth. Thus, inthe absence of a change in the level of yieldson marketable securities, demand deposit hold-ings of households are likely to grow at aslower rate than income. A rise in householdpurchases of securities, associated with a risein yields on marketable securities, may reduceboth time and demand balances held as partof the asset float.

It seems to me that the analysis given abovedoes conform fairly well to the actual experi-ence of the postwar period. The ratio of totalliquidity to personal income has shown no

'Secondary market purchases of mortgages from FNMAreduce federal issues of securities and, therefore, reduce theamount available for boustholds.

trend since 1952. Total liquidity (as definedhere) has risen relative to income in recessionyears and fallen in booms. There has been afairly obvious trade-off between time depositsand high-grade bonds — e.g., the "magic fives."The proportion of household liquidity in theform of currency and demand deposits hasfallen steadily since 1952.

Movements of Household Liquidity and De-mand for Money in the Prewar Period.

Total household liquidity remained a fairlyconstant proportion of personal income from1922 to 1927 but fell rapidly during 1928 and1929. The decline may be attributed to thelarge volume of new security issues floated andthe general belief in the prospect of high netyields from investment in common stocks.

The level of liquidity in relation to personalincome was lower (varying about a ratio of.6) than in the years since 1952 when theratio of household liquid asset holdings to per-sonal income has varied about a figure ofSome of the difference may be merely defini-tional since the treatment of high-grade bondsas an element in household liquidity is some-what ambiguous. In addition, changes in in-come distribution have probably increased theshare of financial saving by low-income groupswho tend (for reasons given above) to hold alltheir financial assets in liquid form.

The most interesting and puzzling thingabout the twenties is the steady reduction inthe share of liquid assets held in the form ofdemand deposits and currency. In 1922, 38per cent of household liquid asset holdings tookthe form of currency and demand deposits. By1927 the proportion held in those forms hadfallen to 30 per cent and 'by 1929 a furtherfall to 25 per cent had taken place. The sharpdecline in household holdings of cash from1927 to 1928 may be attributed to the rapidflow of household funds into common stocks.The furious pace of stock market activity re-sulted in a sharp reduction of the "asset float."

The decline in relative cash holdings in theearlier years is more difficult to explain. Itwas not due to rising yields on savings depositsbecause those yields were not rising. There is,however, some reason to believe that at leastpart of the shift resulted from changes in the


competitive position of national banks withrespect to time deposits. The establishment ofdifferential reserve ratios for time deposits in1914 and the widening of national bank mort-gage lending powers, together with the strengthof the demand for mortgages in the early1920's, made time-deposit business attractiveto commercial banks. It was generally believedduring the 1920's that commercial banks en-couraged customers to switch from demandto time deposits and even permitted checkingagainst time deposits. It seems perfectly pos-sible that increased nonprice competition fortime deposits resulted in some redistributionof liquid assets between demand and time de-posits. It is also possible that there was someshift from Liberty bonds to time deposits as aresult of the decline in bond prices in 1920.Since we did not include those bonds in liquidassets, a switch from bonds to time depositswould reduce the ratio of demand deposits tothe liquid assets included in the ratios quotedabove.

Those explanations appear a little ad hoc andthe possibility of other explanations cannot beruled out.

During the decade of the thirties the ratioof household liquid assets to personal incomereached the high figure of 74 per cent Ifl 1932.The ratio then fell almost continuously until,by 1941, it had reached the 6o per cent levelwhich ruled during the middle 1920'S. It seemsreasonable to attribute the variation to changesin confidence particularly in the early part ofthe period.

The share of liquid assets held in the formof demand deposits and currency rose through-out the 1930's, reaching 40 per cent by 1941.Since the yields offered for savings deposits de-clined throughout the period, there seems tobe no special difficulty in explaining the risingshare of demand deposits and currency in totalliquidity.

Conclusions.The household demand for liquid assets and

money is a complex matter which does notseem to have a simple explanation. Withoutrepeating what has been said above we mayconclude that, putting aside short cycle move-ments, the distribution of income and the ex-

tent of confidence in income stability are themajor factors determining the demand forliquid assets in general. The distribution of liq-uid assets between demand and time deposits issignificantly influenced by the efforts of savingsinstitutions — through rate competition andother selling efforts — to obtain time deposits.

In the shorter cyclical movements, the vol-ume of narrowly defined liquid assets held byhouseholds varies considerably with the varia-tion in the difference between rate of return ontime deposits and expected yield on marketablesecurities. Demand deposits holdings are alsosignificantly affected by variations in expectedyields on marketable securities.

Although demand and time deposits are com-petitive with one another, their short-run cy-clical movements often tend to be positivelycorrelated.

Conclusions

In this paper I have tried to do two things:(x) to review the major factors which seemlikely to influence the amount and composi-tion of liquid assets held by households andcorporations; and (2) to examine the data overthe last forty years to see whether they can beexplained in terms of the factors discussedunder (r).

The major conclusions for corporations areas follows:

(r) Short-run movements of corporateliquidity are strongly influenced by variationsin cash flow from operations — most of thesevariations are reflected in holdings of govern-ment securities rather than in holdings of cash.

(2) The movement of total corporateliquidity during the 1930'S suggests a fairlystrong connection between business confidenceand ratio of liquid-asset holdings to activity.

The decline in the ratio of corporateliquidity to sales in the years since 1952 ispartly attributable to the decline in tax li-abilities but also to improved cash manage-ment, and confidence in income stability.

(4) Corporate holdings of liquid assets re-spond fairly strongly to short-run changes ininterest rates.

There is no indication that the divisionof liquid assets between cash and other liquid


assets responds to short-run changes in billrates.

(6) After allowance for the. effect of chang-ing tax liabilities, there has been only a slightdownward trend in the proportion of liquid as-sets held in cash. The rise in interest ratessince 1952 therefore appears to have had littleeffect. However, it may be that improvementsin cash management have been due to risinginterest rates, while reduction in liquidity inother forms was due to improved confidence.In that case, changing interest rates may havehad some influence on the distribution of liquidassets.

(7) During the depression of the thirties theproportion of liquid assets held in cash did riseas interest rates fell. It therefore appears thatthe elasticity of cash holdings to interest rates

is greater at low interest-rate levels than athigh ones.

The conclusions with regard to householdscan be summarized as follows:

Short-run movements of household liquidityshow considerable response to short-run changesin net expected yields in marketable securities.A large part of this response is reflected in time-deposit holdings but demand deposits alsorespond.

Total liquid asset holdings responded tochanges in confidence during the 1930's in thesame way that corporate holdings did.

The division of liquid asset holdings betweencash and time deposits appears to have respond-ed to changes in yields on savings deposits andto changes in the advertising and selling effortsof savings institutions.

COMMENTKENNETH J. ARROW,

Stanford UniversityJames Duesenberry in his encyclopedic cov-

erage of the movements of holdings of cash andof liquid assets has listed many factors, butthree elements appear and reappear: increasingreturns to scale in transactions (whether be-tween cash and other liquid assets or betweenliquid and nonliquid assets); the conveniencesof having cash (and, by derivation, the con-veniences of liquidity in terms of ease of ac-quiring cash as needed); and risk aversion. Iwill state some results in theory of choice underuncertainty which bear on the more preciseinterpretation of these elements and their im-plications for behavior.

Two branches of the theory of choice underuncertainty will be drawn on here: stochasticmultiperiod inventory theory and the theoryof risk aversion.

Implications of Inventory TheoryThere has developed in the last fifteen years

a theory of optimal behavior for the holding ofinventories, where the firm is facing repeateduncertain demands and has repeated opportuni-ties to purchase inventories.' In its simplest

1See P. Masse, Les reserves et regulation de i'avenir,Paris, Herxnann & Cie., 1946; K. J. Arrow, T. E. Harris,

form, we assume discrete time periods. At thebeginning of each period, the firm has a givenstock of inventory. It is then faced with a de-mand, which is a drawing from a probabilitydistribution. The demand is met, at least to theextent possible, and there is a penalty for theshortfall, if any. The firm then places an orderfor as much more stock as it wishes, and the neworder, plus whatever stock may have been leftover after meeting the demand, constitutes thestock on hand at the beginning of the nextperiod. The cost for the period is the sum ofthe penalty and the ordering cost (plus possiblyalso storage cost); this is a random variable,since both the penalty and the amount orderedmay depend on the random demand. The aimof policy is to minimize the sum over time ofdiscounted expected costs.

Two observations may be made on theseassumptions. In the first place, since only theexpected value of returns is considered, we areassuming risk neutrality. This is an assumptionappropriate for a large corporation, where therisk is divided and, in general, small for eachstockholder relative to his total wealth, but notfor an individual.

and J. Marschak, "Optimal Inventory Policy," Econometrica,19 pp. 250—272; K. J. Arrow, S. Karlin, and H. Scarf,Studies in the Mathematical Theory of Inventory and Pro-duction, Stanford University Press, 5958.


In the second place, the whole problem arisesbecause goods cannot be procured instanta-neously, at least not without extra cost. If stock,could be ordered after the demand is known,there would be no costs beyond the irreducibleminimum for ordering the goods. One interpre-tation of the penalty is the price of immediatedelivery of the goods, which is higher than theordering cost for lagged delivery. The differ-ence between the penalty and ordering costs isa form of transaction cost, which we may terma timed transaction cost.

The interpretation of the inventory modelrelevant for the present discussion is that inwhich the good in question is cash or possiblyliquid assets, and the firm is subject to randomdemands for cash, which cannot be replenishedimmediately without additional cost.2 Theoptimal policy depends on the nature of theordering cost function. If cash can be obtainedat a cost proportional to the magnitude de-manded (constant marginal cost), the optimalpolicy is to a target level of cash and then,in each period, after the demand, acquireenough cash to bring the stock up to the targetlevel. The target level depends on' the penaltyfor shortfall, the rate of interest (which is thepenalty for excessive holdings of cash), andthe probability distribution of demand. Thefluctuation in the observed cash holdings,especially if the observation is made just afterthe demand has occurred, will indeed, asDuesenberry notes, reflect primarily the fluc-tuations in demand. However, the target maychange because of changes in the rate of inter-est; it may also change because of changes inthe subjective probability distribution of de-mands, which in turn may be influenced by theobservations. The latter movement will berelatively slow, since conviction of a change inthe probability distribution will only be ob-tained after a series of observations confirm it.

A second case is that in which the orderingcost contains a fixed component as well as aproportional one. However, the ordering costis zero if no order is placed. This corresponds

'The model analyzed by D. Patinkin (in Money, Inter-est, and Prices, Evanston, Illinois, Row, Peterson, 1956,Ch. VII) is similar to a one-period version of the abovemodel; there may be a whole sequence of demands andpayments at random times within a period in which no cashreplenishment at all can be carried out.

to a psychic cost of decision making or to ad-ministrative cost or to some other form ofeconomies of scale in handling cash. This isagain a transaction cost but of a different kindfrom timed transaction costs; we may call it afixed transaction cost, in that it does not dependon the magnitude of the transaction.8

The optimal policy here is of the two-bin orS,s type, to use the terms in the literature.There are two levels, a target, S, and a reorderpoint, s; cash is acquired only if reserves fallbelow s, but when they do, enough cash is ac-quired to bring the stock up to S.4 As can beseen, this implies a stickiness in the response tofalling cash reserves; only beyond a certainlevel is there a response, but the response is aptto be large. The difference, S — s, which isroughly the size of the order (actually the orderis usually somewhat larger), is under certainconditions roughly proportional to the squareroot of mean cash demands.

Risk AversionIn this section, I will assume the absence of

transaction costs and consider the effects on thedemands for cash and for liquid assets of riskaversion. The model is a modified form of thatstudied by Tobin.6 It is basically a study of thechoice between risky and safe assets; in inter-pretation it is perhaps more suited to analysisof the margin between liquid and nonliquidassets than between cash and liquid assets,since transactions motives are more significantin the latter choice.

We use the expected-utility hypothesis ofbehavior under uncertainty; that is, the individ-ual makes choices so as to maximize theexpected value of a suitably chosen utilityfunction for wealth. One implication of theexpected-utility hypothesis is not always under-

'The distinction between transaction costs which dependon the magnitude of the transaction and those which do notappears (in a nonstochastic context) in J. Tobin, "Interest-Elasticity of Transactions Demand for Cash," Review ofEconomics and Statistics, 38, 1936, pp. 241—247.

Although the policy had been discussed a good deal inthe literature, the first proof of the optimality of the S, spolicy under reasonably general conditions is due to mycolleague, H. Scarf, "The Optimality of the (S, s) Policiesin the Dynamic Inventory Problem," in K. J. Arrow, S.Karlin, and P. Suppes, Mathematical Methods in the SocialSciences, Stanford University Press, 1960, pp. 196—302.

'J. Tobin, "Liquidity Preference as Behavior TowardRisk," Review of Economk Studies, 26, 1958, pp.


stood; the utility function must be bounded,for otherwise a version of the St. Petersburgparadox could be found (this point was origi-nally developed by the mathematician, KarlMenger). If X is wealth and U(X). is theutility function, then we suppose, of course,that U(X) is strictly increasing; from theboundedness, it must approach a finite upperlimit as X approaches infinity and a finite lowerlimit as X approaches zero. From the first ofthese, it is clear that on the average there mustbe risk aversion, i.e., U"(X) (the second deriva-tive) must be negative except, at most, forisolated intervals. It will be assumed here thatindividuals are risk averters throughout.

The quantity U"(X) is not itself a suitablemeasure of risk aversion, since it depends on theunits in which utility is measured. Two meas-ures will be used here: (x) relative risk aver-sion, defined as —XU"(X)/U'(X), which isalso the elasticity of the marginal utility of in-come, and (2) absolute risk aversion, definedas —U"(V)/U'(X). The first will be the moreimportant.

The boundedness of the utility function hasthe following consequence; the relative riskaversion must approach a limit greater than ias X approaches infinity, and a limit less thani as X approaches zero. (This may be madeclearer by noting that the logarithmic utilityfunction, which is unbounded at both ends, hasa relative risk aversion constantly equal to i.)If, for simplicity, we assume that the relativerisk aversion is monotonic, then it must bemonotonic increasing. In a sense, safety is aluxury good.

The choice model is that introduced byTobin; the notation differs somewhat. Let Abe the initial wealth of the individual. He caninvest all or some with a random rate of returnR; the remainder, he leaves in cash with acertain rate of return of zero. The model caneasily be extended to the case where there is asecure asset with a positive rate of return; inthat case R is interpreted as the difference be-tween the random and the secure rates. Thisinterpretation applies to the choice betweenliquid and risky assets. Let a be the amountinvested. Then the wealth at the end of theperiod is,

a(x+R)+(A—a)=A+aR,

and the individual seeks to maximize,E[U(A+aR)},

where a must lie between 0 and A. It can beshown that the optimal investment a is neces-sarily positive if and only if E(R) is positive;an individual will always take some part of afavorable risk but, if a risk averter, will nevertake any part of an unfair risk. The optimummight involve investing the entire initial wealth.If it does not, the optimal investment satisfiesthe condition,

E[U'(A + aR) R] = o.

We are interested in the demand for riskyinvestment, a, and its complement, m = A — a,the demand for cash or liquid assets. First, weconsider the effects of initial wealth A. If ab-solute risk aversion is increasing, then it can beshown that a decreases as A increases, thatrisky investment is an inferior good. Since thisresult is certainly empirically implausible, wemust reject the hypothesis of increasing abso-lute risk aversion. It may be noted that thequadratic utility function, often used for itssimplicity, implies increasing absolute riskaversion and so must be rejected. If, on theother hand, we assume decreasing absolute riskaversion, then risky investment becomes anormal good.

Decreasing relative risk aversion, which isa natural assumption, as we have seen, has avery interesting implication; the wealth elas-ticity of the demand for cash or liquid assetsis greater than x, so that money and liquidassets are luxuries. Although the detailedempirical meaning of this implication requiresfurther examination, particularly because ofthe neglect of transaction costs, it correspondsto the empirical work of Friedman and ofSelden.6

These results have referred to wealth effects;one might also ask about price effects. Underuncertainty, the analogue of a price is theprobability distribution of R and, of course,there are many possible ways a distribution canchange. A simple upward shift by a constant

e Friedman, "The Demand for Money: Some Theoreticaland Empirical Results," Journal of Political Economy, 67,1959, pp. 327—351; R. T. Selden, "Monetary Velocity in theUnited States," in Studies in the Quantity Theory of Money,M. Friedman, Ed., University of Chicago Press, 1956, pp.


amount (i.e., the mean increases while thedistribution about the mean remains un-changed) can be shown to increase the demandfor risky investment, as might be expected,provided that risky investment is a normalgood, It may be noted that an increase in therate of interest on the secure asset implies adownward shift in the distribution of thespread, R, and so, again as might be expected,an increase in the demand for liquid assets.

A second type of shift in the distribution ofR is a simple multiplicative shift around zero.Here we have the simple and surprising resultpresented by Tobin; the risky investment isreduced in inverse proportion to the multiplier.Thus an ideal proportional income tax at ratet, which means a multiplier of x — t, will in-crease investment in risky assets in the pro-portion, x/(i — t).

Finally, we can consider a multiplicativeshift around the mean instead of around zero,which might be thought of as a pure change indispersion. This can be regarded as a com-pounding of the two previous shifts, a multi-plication around. zero, followed by an additiveshift to restore the mean to its original value.It follows that an increase in dispersion, in thissense, will reduce the demand for risky invest-ment (and increase that for liquid assets), pro-vided that risky investment is a normal good.

PHILLIP CAGANBrown University

With James Duesenberry's paper as back-ground, I should like to review what we knowabout the interest elasticity of monetary veloc-ity in the short run. Several years ago thewidespread belief that this elasticity might bevery high led to strong misgivings about theeffectiveness of monetary policy. This was oneof the questions the Commission on Money andCredit was to deal with. It did not, at least notclearly. So I appoint myself to a commissionof one to issue a supplementary report on thisquestion.

The issue, you will remember, was whetherchanges in velocity owing to interest-rate move-ments induced by open market operations mightlargely offset the effect of those operations andso prevent the Federal Reserve from restraining

booms and mitigating recessions. The RadcliffeReport seems to suggest that the interest elas-ticity of velocity might be infinite, because ofa ready supply of money substitutes, so thatmonetary measures are trapped in a swamp ofunlimited liquidity. This extreme view washeard much less in this country after our tightmoney period in 1953—56. The view still pre-vails that velocity changes can delay or subduemonetary measures though not offset thementirely.

Some defenders of monetary measures arguedthat such changes in velocity are welcome be-cause they cushion the shock of monetaryrestraint. This is an ingenious counterattackbut very misleading. We do not need this ad-ditional cushion; the economy has others.Those who need to borrow can always do soif they are willing to pay the going rate. To besure, an elastic velocity is in a sense themarket's way of softening the blow of monetaryrestraint. Accordingly, if velocity were com-pletely inelastic, we should not feel deprived ofa needed safety valve; the inelasticity of veloc-ity would indicate that we did not need thevalve. On this argument we might concludethat it doesn't matter how elastic velocity is.

An elastic velocity may cause problems, how-ever. The timing, extent, and duration of in-duced changes in velocity may be difficult topredict and so make the effects of a givenmonetary action uncertain beforehand. Thissurely does not make it easier to stabilize theeconomy by monetary measures and may makeit more difficult. I say "may" because changesin velocity occur frequently for many otherreasons, and this additional source of changemay not make monetary stabilization any moredifficult than it would otherwise be. In statis-tical terms, the standard error of predictions offuture changes in velocity may be larger(though not necessarily) when the interestelasticity of velocity is larger.

What does the evidence show about the sizeof this elasticity? Duesenberry is impreciseabout the effects of interest rates on the de-mand for checking deposits, as he has to be inthe kind of historical survey he presents. If Ido not misread him, however, interest effectsplay a secondary role in his analysis, and thisis so for periods in which the size of their move-


ments was among the sharpest on record. Hefinds little reason for, and little evidence of,interest-rate effects on corporate cash balances,at least in the short run. For households, in-terest rates on savings deposits are important,though they are likely to have their main effectsin the long run, both because households mayadjust savings deposits slowly to changes inthese rates and because these rates respondslowly to changes in bond rates. This all ap-pears plausible to me from the data I haveexamined. He finds the main short-run effectof interest rates on demand deposits to be thatof bond rates on the deposits of high-assethouseholds, and even this effect is subdued be-cause many of the households who move in andout of bonds probably keep their idle balancesin savings accounts.

I come away from Duesenberry's paper,theref ore, with the impression that the interestelasticity of velocity in the short run is not veryhigh, or at least is likely to be much lower thanin the long run. Perhaps he did not intend togive this impression; in any event, I see nothingin the evidence he covers to deny it and muchto support it.

How do such findings square with otherstudies? We have first of all Henry Latané'swell-known study showing an impressive long-run relationship between velocity and interestrates since 1909. This study and others similarto it raised doubts in my mind, because theyseemed to depend heavily on two large move-ments in velocity and interest rates from the1930'S to the 1950'S without any indication thatthe shorter movements within this periodshowed the same relationship. Moreover,Latané's data do not appear to fit so well be-fore World War I. Friedman's results suggestthat long swings in velocity can be explainedjust as well by permanent income, though, ashe indicates, his results may not be inconsistentwith Latané's because of their different treat-ment of time deposits and also because per-manent income may be a proxy for changes ininterest rates.' My colleague Allan Meltzerreports a good fit of real cash balances to in-

1 If y r "the"interest rate, W real national wealth, and p denotes "per-manent" magnitudes), a change in will affect if itreflects a change in productivity of capital, and will affect

if a change in the public's desire to hold wealth.

terest rates for the entire period since 1900using a concept of wealth instead of income asa deflator of money balances. When he usespermanent income as a deflator, he also gets asignificant partial regression coefficient forinterest rates, and does so for the period before1930.

These results suggest that there is an effectof interest rates on velocity in the long run.They are less clear, to me at least, on whetherthere is an effect in the short run. Scatterdiagrams may show a short-run correlation tosome degree, but this may be spurious, reflect-ing the tendency of velocity and interest ratesto respond in a similar way to fluctuations inbusiness activity. These correlation studiesseem to give more consistent results overlonger periods than over individual cycles andfor long-term rather than short-term rates ofinterest, though these points need further clari-fication.

I may cite evidence of quite a different kind.In a study I made of seven hyperinflations,2 Ifound a lagged relation between velocity andthe cost of holding money, where the cost inthis case was the rate of change of prices. Thelag was of the distributed type and had averagelengths of nearly a year or more in mostcountries; moreover, the length appeared todecrease with the extent and duration of hyper-inflation. If this evidence is carried over tonormal times and applied to the effect of inter-est rates on velocity, the implication is that theeffect has an average lag of several years ormore.

My study, Duesenberry's paper, and otherevidence I have seen appear to suggest that theshort-run interest elasticity of velocity issmaller than the long-run elasticity. In Melt-zer's study, the long-run elasticity is approxi-mately unity, so the short-run elasticity may beconsiderably less than unity, depending on thetime span. If so, cyclical fluctuations in veloc-ity are not to be attributed to interest-ratemovements, at least not entirely. Althoughthere may be many serious obstacles to effectivemonetary measures, such as lags, poor fore-casts of business conditions, and so on, off-

11n Studies in the Quantity Theory oJ Money, MiltonFriedman, Ed., University of Chicago Press, z956.


setting changes in velocity may not be one ofthem.8

IRWIN FRIEND,University of Pennsylvania

The description in this interesting paper byDuesenberry of the procedures followed in ar-riving at the conclusions given does not seem tome to be adequate to form an appraisal eitherof the approach followed or of the degree ofsuccess achieved. It is not entirely clear, forexample, how the supply of liquid assets hasbeen handled in his model nor how much varia-tion he has been able to explain in what areapparently presumed to be demand relation-ships for liquid assets by corporations andhouseholds. I assume that this information willbe forthcoming, but until it is, I can commentonly on some of the specifics of the paper ratherthan on the more important general orientation.The paper incidentally would be made moreuseful by presentation of the basic data and keystatistics, so that qualitative conclusions canbe assessed and also comparisons with otherfindings can be effected.

In his discussion of short-run movements incorporate liquidity, Duesenberry finds thatvariations in corporate liquid assets are prin-cipally due to variations in cash flows fromoperations and only to a lesser extent to inter-est rates. There is no indication of any treat-ment of the influence of the level of currentliabilities on liquid-asset holdings except for aconstant percentage tax adjustment. It is notat all clear why tax liabilities are treated inthis unique fashion and why the level as wellas composition of other current liabilities is notconsidered to influence corporate liquidity, par-ticularly since business firms are supposed touse the liquid asset — current liability ratio asone important measure of liquidity. More

a is sometimes also argued that even a 'slow" interestelasticity of velocity is troublesome if the interest elasticity ofinvestment is low. Then a large change in interestrates is required to produce a given change in investmentand, with even a low interest elasticity of velocity, an ex-ceptionally large open market operation is required. It is

•further argued that a large operation may not be feasible for"institutional" reasons, but these reasons have never beenclearly specified. Unless they refer to lags and the conse-quent danger of overshooting, it is hard to take them

• seriously.

generally, of course, the whole current andfixed-asset and liability structure would be ex-pected to be relevant. The answer may be thatDuesenberry has been able to explain virtuallyall variation in liquidity by the variables he hasused, but this seems doubtful and in any casecannot be ascertained from his paper.

While apparently the corporate-bond yieldis considered as one of the determinants ofliquid-asset holdings of corporations, there isno indication that the cost of equity financingor the state of the equity markets has beenconsidered in any systematic form. Thus, forexample, it is not entirely clear why the 1952—6i rise in Baa bond yields might be expected tobe associated with a decline in corporate liquid-ity, since presumably most firms would haveconsidered that that rise in bond yields wasassociated with a sizable decline in the cost ofequity financing. Also, when Duesenberrynotes that "a one percentage point increase inthe bond yield appears to reduce corporateborrowing by $700 million per quarter," wouldnot that rise in bond yields result in a partlyoffsetting increase in equity financing, evenwithout any further adjustment for the fact thatthe cost of equity financing would normally beconsidered by business to be declining wheninterest rates are rising?

Duesenberry finds that the ratio of corporateliquidity to GNP during the twenties was higherthan in the fifties. It is not clear whether he isrelating corporate liquidity to GNP, as hestates, or to income originating in the corporatesector which would appear to be more appro-priate. However, the latter measure of corpo-rate liquidity might even intensify the result heobtained. Duesenberry notes that the changesin the cost of liquidity or in confidence aboutincome stability cannot be used to explain thedifference in liquidity ratios between the twoperiods. I would not be inclined to discountcompletely the possibility that business in re-cent years has been appreciably less concernedabout the danger of major depression than itwas in the twenties. Three other factors whichmight help to explain at least part of the dif-ference in liquidity between the two periods are,first and most important, the much greaterburden of debt in the twenties; second, perhapsgreater access to or willingness to use short-


term bank financing, or both, in the fifties, atleast in manufacturing; and third, perhaps thedifference in the implications of tax liabilities.Since I just received Duesenberry's papershortly before this conference, I have not hadan opportunity to look into the relevant debtstatistics, but I do recall that the ratio of inter-est payments to income before taxes and beforeinterest was about 25 per cent in the latetwenties and is perhaps one-half that currently,which might suggest a situation calling forhigher liquidity in the earlier period. So far asthe tax situation is concerned, it is not clearwhat tax adjustment was made or should havebeen in the twenties as compared with thefifties, but it is possible that the procedure fol-lowed might have distorted the comparison ofliquidity ratios in the two periods. On theother hand, there are several possible reasonsfor expecting a lower conventional liquidityratio in the twenties, including the influence ofcall loans and commercial paper.

Another question relating to the adjustmentfor tax liability relates to the discussion of thetrend in relative importance of cash versusother forms of corporate liquid assets in thepostwar period. Assuming, for want of infor-mation, that a constant percentage adjustmentwas made for tax liability, would this be appro-priate over a period in the earlier part of whichcorporate tax payments due at the end of a yearcould be paid over the following year as awhole, while at the end of the period such taxpayments had to be made in the first half of thefollowing year?

Duesenberry also states in his discussion ofthe composition of corporate liquid assets that"no one has found a statistically significantrelationship between short-term variations inbill yields and the distribution of corporateliquidity between cash and governments."Though this statement may very well be true,I am surprised at such a finding. In a paper,"The Effects of Monetary Policies on Non-monetary Financial Institutions and CapitalMarkets," I prepared for the Commission onMoney and Credit, in which I analyzed thedemand for and supply of claims on differentfinancial institutions, but without generallyseparating corporations from households, Ifound a highly significant effect, in the expected

direction, of quarterly changes in the bill rateon the demand for idle demand deposits by theprivate sector, and also on the demand forfederal securities by this sector. I would haveguessed that this result reflected in good parta corporate response to changes in bill rates.

Turning to the discussion of household de-mand for liquid assets and for demand deposits,Duesenberry notes that such demand might beexpected to increase with permanent incomeand discusses some theoretical possibilities ofdifferent types of response to changes in income.A considerable body of seemingly relevant sur-vey data on the relation of household assetstructure and saving to different types ofchanges in income exists and perhaps mightbe used. To give one such finding: ".families with fluctuating incomes (Over athree-year period) do not seem to show muchdifference from families with constant in-comes in their saving in the form of cash anddeposits; there appears to be some tendencyfor the former to save less in cash and depositsthan the latter at all income levels, though athigh income levels families with fluctuating in-comes might be expected from theoretical con-siderations to accumulate more cash and de-posits than families with constant incomes."'

In his interesting attempt to estimate assetfloat as a significant element in accounting forthe level of personal demand deposits, Duesen-berry notes that a large proportion of the moresubstantial personal demand deposit accountsis held in connection with financial transactionsand approximates at "a rough guess individualpurchases and sales of stocks, bonds, real estate,and other assets amount to something like $ioobillion per year." This seems much too small,which would suggest that asset float is evenlarger than estimated by Duesenberry, thusstrengthening his point, unless the averageholding period has been correspondingly over-stated. In 1961 purchases plus sales of stockalone amounted to billion and while asizable amount of this is broker-dealer or in-stitutional, and 1961 is considerably higherthan the average of preceding years, the house-hold sector alone in recent years (excluding

1 Irwin Friend and Stanley Schor, Consumption andSaving, VoL II, University of Pennsylvania Press, 1960,p. 272.


broker-dealers and institutions) must have hadan average amount of stock purchases plussales of well over $ioo billion in 1961 and per-haps $75 billion annually in the late 1950's. Ihave not had the opportunity to make a roughestimate of the other items, but my guess isthat they amount to very much more than $25billion. I am assuming, incidentally, thoughDuesenberry does not make clear where hetreats the demands of financial institutions,that they are not included in the householdsector; otherwise the apparent understatementof float would be more serious. The figure forgross financial transactions of the householdsector is of course the least unreliable of theestimates required to derive asset float in theform of demand deposits. Duesenberry notesthat the sharp decline in household holdings ofcash in the late twenties may be attributed to asharp reduction of asset float associated with the"furious" pace of the stock market. This seemsquestionable in view of the enormous increasein the volume of stock and other financial trans-actions even if there were an appreciable re-duction in the average holding period betweenthe sale of one asset and the purchase ofanother.

In• the one place in his paper where Duesen-berry does present an explicit model, i.e., wherehe discusses the determinants of money hold-ings versus other liquid assets for the householdsector, it is not at all clear how a key financialasset such as stock is supposed to affect demandbalances. It would appear that the marketvalue of stock enters as an exogenous explana-tory variable for demand behavior of high-assetholders but not at all for small-asset holders.

Liabilities and real assets apparently are pre-sumed to have no effect on demand forany group of asset holders. Insurance isparently assumed to have no effect on demandbehavior at least for low-asset holders. It is notclear whether the yield variables are supposedto be adjusted for expected price level changes,where this is relevant, or how changes in riskevaluation or in uncertainty are handled.Though gross financial saving is introduced asa relevant variable in explaining demand de-posits of high-asset households, Duesenberryconcludes that, other things being equal, therewill only be "minor fluctuations [in the ratio ofdemand deposits to income] resulting fromvariations in the ratio of gross financial savingsto income." This apparently means that thecoefficient of gross financial savings is verysmall since such saving, assuming it is com-prised of currency and deposits, savings sharesand securities, is quite variable in relation toincome.

So far as his general conclusions are con-cerned, it is difficult to assess such statementsas, ". . . the distribution of income and theextent of confidence in income stability are themajor factors determining the [household] de-mand for liquid assets in general," withoutmore background quantitative informationthan is supplied. Thus I was not .able to cotn-pare Duesenberry's interest rate effects on thelevel and composition of liquid assets with re-lated results reported in the CMC paper Ireferred to earlier or with other results. How-ever, qualitatively, his major conclusions forthe corporate and household sectors seem forthe most part to be reasonable.

Date post:	26-Mar-2022
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

The Portfolio Approach to the Demand for Money and Other Assets

Documents