http://rrp.sagepub.com/ Economics Review of Radical Political http://rrp.sagepub.com/content/7/1/20.citation The online version of this article can be found at: DOI: 10.1177/048661347500700102 1975 7: 20 Review of Radical Political Economics Stephen A. Marglin and Stephen A. Marglin What Do Bosses Do? Part II Published by: http://www.sagepublications.com On behalf of: Union for Radical Political Economics can be found at: Review of Radical Political Economics Additional services and information for http://rrp.sagepub.com/cgi/alerts Email Alerts: http://rrp.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions: What is This? - Apr 1, 1975 Version of Record >> at Harvard Libraries on November 13, 2013 rrp.sagepub.com Downloaded from at Harvard Libraries on November 13, 2013 rrp.sagepub.com Downloaded from
Transcript
http://rrp.sagepub.com/Economics
Review of Radical Political
http://rrp.sagepub.com/content/7/1/20.citationThe online version of this article can be found at:
DOI: 10.1177/048661347500700102
1975 7: 20Review of Radical Political EconomicsStephen A. Marglin and Stephen A. Marglin
What Do Bosses Do? Part II
Published by:
http://www.sagepublications.com
On behalf of:
Union for Radical Political Economics
can be found at:Review of Radical Political EconomicsAdditional services and information for
Part I of this study was concerned with the ori-gins of the pyramidal hierarchy that characterizescapitalistic organization of production.[l] Now weshift attention to the function of hierarchy undercapitalism: what role does hierarchy play that canaccount for the acquiescence of workers to forms ofwork organization that are otherwise so destructiveand costly in human terms? It will be argued that itis not efficiency (at least not as this term is used andunderstood by neoclassical economists) but capitalaccumulation that justifies capitalist hierarchy: inbrief, the rate of capital formation remains reason-ably high in capitalist societies because hierarchicalorganization permits a relatively small number ofindividuals to decide how much the rest of us willsave. If, by contrast, savings decisions were left toindividuals - whether capitalists or workers - ac-cumulation of productive capital, at least in plantand equipment, on which workers depend forincreases in wages, would come to a virtual stand-still. And an end to accumulation would spell andend to the present basis of the acceptability of capi-talism - the ever-increasing abundance of mater-ial goods and services.
This argument of course runs counter to allthe canons of neoclassical economics. In neoclassi-cal theory there is no connection between the hier-archical organization of production and the rate ofsaving. Aggregate savings are determined by the
* This study is a continuation of a previous study whichwas published in the Summer, 1974 issue of this review.Copyright by Stephen A. Marglin, 1974.
decisions of millions of households, each making aconscious, deliberate allocation of current incomebetween present and future consumption as anintegral part of a utility-maximizing life-timeconsumption plan chosen in the light of life-timeresources.
The theory would, I submit, hardly commandserious attention if it did not serve so admirably tojustify the historically determined pace of capitalaccumulation and the existence of property in-
come. In the form elaborated by Irving Fisher,[2] itis too general to be contradicted by any conceivableempirical evidence, and therefore too general to beconfirmed. In an uncertain world, &dquo;life-time&dquo;resources must necessarily be a subjective, unmea-surable magnitude subject to continued revision bydecision makers, whose consistency over time withthe axioms of revealed preference is in principle un-verifiable.
The difficulty becomes transparent in thework of Fisher’s latter-day disciples, Franco Modi-gliani (and associates Albert Ando and RichardBrumberg) and Milton Friedman. Modigliani’s&dquo;life-cycle&dquo; hypothesis[3] and Friedman’s &dquo;perma-
1. This paper is self-contained and can be read
independently of Part I.
2. I. Fisher, The Theory of Interest, first published in1930. Republished by Augustus Kelley, New York, 1954.
3. The original published statement is in an articleauthored jointly by Modigliani and Brumberg, "UtilityAnalysis and the Consumption Function: An Interpret-ation of Cross-Section Data," in K. Kurihara (editor)Post-Keynesian Economics, Rutgers University Press,New Brunswick, New Jersey, 1954. Important sub-sequent articles are Ando and Modigliani, "The Life-CycleHypothesis of Saving: Aggregate Implications andTests", American Economic Review, volume 53, March1963, pp. 55-84, and Modigliani and Ando, "The ’Perma-nent Income’ and ’Life-Cycle’ Hypothesis of SavingBehavior: Comparison and Tests" in Irwin Friend (editor)
20
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
nent-income&dquo; hypothesis[4] in fact have little left oftheir lofty Fisherian beginnings when put in a formsuitable for testing against time-series or cross-sec-tion data. For example, the life-cycle hypothesissimply extrapolates current non-property income toestimate future income (corrected for currentunemployment in one variant) and the permanentincome hypothesis makes expected future income afunction of past incomes. As a result, equationsthat are supposed to represent the Fisherian ap-proach to determination of current savings - as aby-product of a grand allocation of life-time re-
sources to life-time consumption - hardly differfrom equations based on myopic decision rules thathave nothing to do with life-time optimization. Thelessons of the life-cycle and permanent-income hy-potheses are at most that current consumption isnot determined by current income alone, a hypo-thesis which (as Friedman points out) no one wouldmaintain in its pure form unless he expected to findorgies of consumption on pay-days and abstinenceat all other times.
In any event, the Fisher-Modigliani-Friedmantheory of household savings leaves open the ques-tion of how household saving, chiefly corporate incharacter, gets determined. The logic of neoclas-sical economics, or rather its ideology of consu-mers’ sovereignty, suggests that, all evidence of theseparation of ownership and control notwithstand-ing, corporate managers allocate profits betweendividends and investment solely in terms of share-holders’ preferences. The corollary is that corpor-ate savings are a substitute for household savings,and that if corporate savings were smaller, house-hold savings would be much larger. The aggregaterate of capital accumulation only appears to be in-fluenced by the institutional separation of owner-ship and control.
However, the identification of the interest ofowners and managers is on the face of it implau-sible, in view of the manager’s multi-dimensionalconcern with tenure, power, and prestige that con-trasts markedly with the shareholder’s single-minded interest in the value of his shares and thestream of his dividends. Moreover, the evidence
offered to buttress the orthodox faith at most sup-ports the proposition that shareholders’ prefer-ences are one of the variables that influence cor-
porate retention policies. For example, John Brit-tain has presented evidence that corporate reten-tion policy responds to changes in individual taxrates that change the relative desirability (from theshareholders’ point of view) of paying out dividendsand retaining earnings.[5] Quite apart from the factthat the periods in which the significant changes intax rates occurred were the ’30’s and ’40’s, years ofdepression and war, so that individual tax ratesmay be acting as proxies for other variables of theeconomic climate, Brittain’s evidence shows onlythat managers pay some attention to shareholders’interests. Equally irrelevant is Wilbur Lewellen’sevidence for the identity of managers’ and share-holders’ interests. Lewellen shows that the mana-
gers of large firms derive much more of their in-come from profits - in the form of dividends, capi-tal gains, and profit-related compensation such asstock options - than from fixed dollar compensa-tion - salaries and the like.[6] Apart from the dif-ficult conceptual problems involved in aggregatingcapital gains and ordinary income, the evidencemisses the point that the manager has a significantinterest in tenure that the ordinary shareholderdoes not have. Not only does his power, prestige, in-deed his very personal identity, depend on his re-maining in control of the corporation, but his
future possibilities for increasing his ownership in-terest and income depend on his continuation inoffice. Lewellen’s evidence may demonstrate thatthe retired corporate executive is at one with theordinary shareholder, but this hardly establishesthe identity of managers’ and shareholders’ inter-ests. Finally, Robert Solow has argued that corpor-ate managers ignore shareholders’ interests only atthe risk of seeing share prices reduced and layingthemselves open thereby to take-over bids.[7] Thismay be true, as Solow suggests, for all but the
largest corporations, but it is a far cry from ignor-ing shareholders to dancing exclusively to the
Proceedings of the Conference on Consumption andSaving, Vol. 2, Wharton School of Finance and Com-merce, Philadelphia, 1960.
4. M. Friedman, A Theory of Consumption Function,Princeton University Press, Princeton, New Jersey,1957.
5. J. Brittain, Corporate Dividend Policy, BrookingsInstitute, Washington D.C., 1966, chapter 4.
6. W. Lewellen, The Oumership Income of Manage-ment, National Bureau of Economic Research, New York,1971.
7. R. Solow, "The New Industrial State or Son ofAffluence," Public Interest, volume 9, Fall 1967.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
shareholders’ tune. The shareholder may posesome constraints on the retention policies that cor-porate managers can pursue in order to increasetheir income, power; prestige, and security of ten-ure. It does not follow that corporate savings deci-sions &dquo;really&dquo; represent the decisions of individualhouseholds in the sense that they reflect consistent,conscious and deliberate allocation of household
resources between present and future consump-tion.[8]
In principle it should be possible to test theproposition that the corporation is an extension ofthe household when it comes to saving decisions.To the extent this proposition is true, cross section-al household data should reveal a systematic substi-tution of corporate for household saving, whichought to be reflected in an inverse relationship be-tween direct household savings and the extent of&dquo;indirect&dquo; saving carried on by corporations &dquo;on
behalf&dquo; of the household. I know of no such
study,[9] but Philip Cagan’s analysis of the effects
of participation in pension plans on household sav-ings tests an analagous proposition, and Cagan’sstudy hardly gives aid and comfort to those whotake Irving Fisher seriously.[10] Adherents to thelife-cycle and permanent-income hypotheses would,I presume, expect to find that savings in the form ofpension funds, like savings carried out by corpora-tions, substitute for other forms of personal sav-ings, at least to the extent that the point of savingsis to provide retirement income. But Cagan’sanalysis of responses of households to a survey ofmembers of a national consumer product-testingorganization suggests a conclusion diametricallyopposed to the predictions of neoclassical ration-ality. In Cagan’s own words
Our analysis of this sample suggests thatwhen households come under a pension plan,offsetting reductions in other savings do notoccur. The net addition to aggregate personalsaving apparently equals the full amount ofemployers’ and employees’ contributions.
To say the very least, the evidence is weak thatsavings rates are independent of capitalist institu-tions, which give direct control over collectivesavings decisions to a relatively small number of in-dividuals. Indeed, I argue more - that under a
capitalist system, hierarchical organization of pro-duction is essential to the maintenance of highrates of capital accumulation. Modern corporatemanagement obliges workers as well as nominalowners of capital to provide for their collective
future, just as compulsory retirement, pension, andsocial security schemes oblige us all to provide forour individual futures. I believe households tend to
spend whatever income they can lay their hands on.Households do not save, by and large and on theaverage, except inadvertently - when their in-comes are rising faster than they can adjust theirspending. And growth-induced disequilibrium
8. The conflict between shareholders, especially therichest group that owns most of the shares, and managerscan easily be overdone—and frequently is. It is only ina relatively few dimensions—of which the rate of accumu-lation is perhaps the most significant—that the procliv-ities of owners and managers diverge. Even here I wouldargue that the class interests (as distinct from short-runproclivities) of owners in the preservation of capitalismare better served by manager’s accumulation-orientedpolicies than by saving policies owners would follow ifindividual control of profits replaced corporate control.The rest of us—with no stake as owners or managers—are the losers insofar as accumulation, or more exactly,the ever-increasing standard of living it produces, blindsus to the destructive features of capitalism.
9. The closest is Modigliani’s cross-sectional studyusing country data, "The Life Cycle Hypothesis of Savingand Intercountry Differences in the Saving Ration" in W.Eltis and others, Induction, Growth and Trade: Essays inHonour of Sir Roy Harrod, Clarendon Press, Oxford,1970, pp. 219-221. Modigliani rightly terms his test incon-clusive but refers to a study by Edward Denison, "A Noteon Private Saving," Review of Economics and Statistics,vol. 40, no. 3, August 1958, pp. 261-267 for confirmation ofthe substitutability of corporate and household saving.Denison bases his argument on very little data (aggre-gate observations for 1929 plus nine post-war years), andit is not clear that even the limited data to which hemakes appeal supports the argument that "individuals’consumption expenditures are not affected by changes inthe proportion of corporate profits paid out as dividends"(p. 264). For one thing, Denison’s argument is based onvariations in the ratios of corporate and household savingto gross national product rather than on variations ofratios of saving to after-tax private income. For another,the growth in the relative importance of the corporatesector is not taken into account. Indeed, the ratios ofcorporate saving to after-tax corporate income and
household saving to disposable income do not tend tomove in opposite directions, as Denison himself shows (p.265). Nevertheless, for all its weaknesses, Denison’s
study is valuable. It suggests tests that might be madeusing aggregate time-series data, for example, tests ofthe effect over time on the ratio of total private saving toprivate income of the growth of the corporate sector.
10. P. Cagan, The Effect of Pension Plans on AggregateSaving: Evidence from a Sample Survey, OccasionalPaper 95, National Bureau of Economic Research, NewYork, 1965.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
hardly provides enough savings to cover the costs ofconstruction of owner-occupied housing: virtuallynothing is left over to finance the acquisition of newplant and equipment. Hierarchical control of pro-duction prevents the spending tendencies of house-holds from putting an end to accumulation becauseit permits those at the top to set aside resources forexpansion of the means of production beforeturning over the value added by producers to
workers and shareholders in the form of wages anddividends.
This theory of saving, however, far from theneoclassical view, is in no sense limited to devoteesof unconventional political persuasions. Indeed,according to Francis Sutton and his collaborators,it is a basic tenet of the faith of modern capitalismthat the corporation bears the primary socialresponsibility for growth of dividends and wages -a responsibility it can discharge only by plowingback a decent portion of corporate revenues intoplant and equipment: &dquo;The dominant business
opinion,&dquo; they say, &dquo;is that earnings have to be re-tained because it is impossible to obtain capitalotherwise.&dquo;[11] As one business leader asked of aCongressional committee:
... if you give it [earnings] all to the stock-holder, by what rhyme or reason do weassume that they are going to save enoughand have it ready for you when you want it?
He quickly provided his own answer:... we have to take calculated risks in busi-
ness, but as an administrator, I would notcare to take that risk.[12]
If shareholders are not to be trusted to save in ade-
quate amounts, it is hardly to be expected thatworkers would be. Latter day Andrew Ures arequick to point out that high profits are the workers’best friend: in the words of the National Associa-tion of Manufacturers,
about half of what industry [earns] goes&dquo;right back into the pot&dquo; to help pay for thedevelopment and expansion that brings moreproducts, more jobs, and greater security forall.[13]
one firm has even defined American capitalism as asystem &dquo;which permits management to assure pro-gress for itself, its employees and its shareholdersby plowing savings back into the business.[14] Theshort-sightedness of workers who would raise wagesat the expense of profits borders on the unpatriotic:what true patriot could have answered negativelywhen asked, as the Cold War was gathering steam,&dquo;Would you sell your country’s future for $140.85[total corporate profits averaged over the popula-tion in 1948]?&dquo;[15]
The acceptability of capitalism to the over-whelming majority of people who have no signifi-cant direct stake in private ownership of the meansof production is intimately bound up with corpor-ate saving, just as the corporation’s defenders aver.This does not mean we need take seriously theclaim that corporate management acts out of itssense of social responsibility in retaining earnings.It is undoubtedly more realistic to interpret corpor-ate saving decisions in terms of managers’ percep-tions of their own self-interest: managerial power,security, prestige, and income are all furthered byplowing back earnings. But in retaining earningsfor investment, corporate managers serve a widersocial function. Without the savings of hierarchi-cally controlled organizations, among which thecorporation is preeminent, the rate of capital ac-cumulation would hardly be adequate to sustainthe proposition that capitalism provides, with thepassage of time, an ever-increasing abundance ofmaterial goods and services for an ever-increasingproportion of the population. And this propositionwell may be essential to the survival of the capitalistsystem. American capitalism has not typicallydepended on forcible repression of the workingclass for its survival. In the large, if not in everydetail, Americans generally - workers as well ascapitalists - accept that what is good for GeneralMotors is good for the country.
Why this easy and general acceptance? Oneessential ingredient has been the creation of a cul-ture that elevates material consumption to an all-embracing value; other basic human needs - suchas the need for community and the need for thesatisfactions of creative work - are either sacri-ficed to the consumption god or sublimated into it.A new automobile not only provides transportation,it compensates the owner for the frustrations of his
11. Francis Sutton, Seymour Harris, and others, TheAmerican Business Creed, Harvard University Press,Cambridge, Massachusetts, 1956, p. 86.
12. Quoted in Sutton, et al., op. cit., p. 86.
13. Ibid.
14. Quoted in Sutton et al., op. cit., p. 85.
15. Quoted in Sutton et al., op. cit., p. 130.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
job and gives him an identity that it otherwise ab-sent from his life. Of course, emphasis of materialconsumption is only necessary, not sufficient forthe survival of capitalism. It is important. Capital-ism must be able to satisfy the material wants thatdefine the consciousnes of capitalist man.
But there is a dilemma here. To the extent thatmaterial consumption fulfills psychological needsof prestige, power, and the like - rather than
physical needs - the utility of consumption is
largely relative. The value of a Cadillac to Jones isenhanced by knowing Smith drives a Ford. And solong as Smith takes his cues from Jones and Joneswatches Smith, only one can be satisfied. Fortun-ately for capitalism, the dilemma can be resolvedby growth insofar as Jones and Smith comparetheir respective levels of consumption not with eachother, but with their own past attainments andfuture aspirations. Provided Jones and Smith looksidewards primarily for guidance in fixing goals forthe future, a growing capitalist economy can satisfymost, if not all, of the people. Thus, paradoxically,the key to social stability is economic growth,[16]and the key to growth is the accumulation of pro-ductive capital. The key to accumulation, in turn, isthe hierarchical structure of the corporation andother organizations that mediate between house-holds and incomes.
A major premise of this argument is that indi-vidual households save relatively little and then
only by accident, not by a conscious and deliberatechoice with respect to a life-time consumption plan.The facts might seem at first glance to refute thisassertion. After all, national income statistics showhousehold saving in the United States equal - yearin, year out - to some six percent of disposable in-come, more if one were to include consumerdurables in saving. How do we reconcile these factswith the assertion that economic growth dependson the saving of corporations and other organiza-tions ?
The resolution of the paradox lies in part inthe growth of household incomes. Household
spending adjusts to increased income only with alag; it takes time to learn to spend. Under condi-tions of virtually continuous growth (characteristicof the postwar American economy), spending never
catches up to income. Thus saving and growth areintimately connected, but as far as households areconcerned, the direction of causality is opposite tothat customarily assumed; household savings arenot a cause of growth, but the result. Moreover, ifwe limit attention to savings that households makeavailable to finance expansion of the means ofproduction, an appropriate limitation for certain ofthe purposes of this paper, the magnitude of house-hold savings, both absolute and relative to corpo-rate savings, becomes quite modest.
That what, pending a more descriptive name, Ishall call the &dquo;disequilibrium hypothesis&dquo; accountsat least as well as the Fisher-inspired competitionfor the observed behavior of hosueholds can, I
think, be demonstrated even to the satisfaction ofthe neoclassically inclined, provided similar stan-dards of rigor are applied here as when the resultsare more congenial to accepted dogma. Indeed, aswe shall see, the real econometric problem lies indistinguishing the disequilibrium hypothesis fromthe explanations advanced by Friedman and Modi-gliani.
The basic idea behind the disequilibrium hy-pothesis is that households tend to spend all theirincome, but that at least some of the response of
changes in consumption to changes in income maybe delayed by a learning process. A simple alge-braic formulation of this hypothesis is a linear
equation relating the rate of change of consump-tion (C) to saving, which, as the gap between in-come (Y) and consumption (C) can be thought of asa reflection of the learning yet to take place, and tothe rate of change of current income (Y):
The parameter 6 measures the immediate adjust-ment of consumption to income, otherwise knownas the short-run marginal propensity to consume;(3 measures the speed of adjustment to an equili-brium in which all income is spent.
However, equation (1) prejudges one of thequestions at issue: namely, is the equilibrium ratioof household consumption to income equal to
one? A more general formulation, which allows usto bring data to bear on this question, is
16. The relationship between the social stability ofcapitalism and economic growth is lucidly articulated byHerbert Gintis in an article entitled "The New WorkingClass and Revolutionary Youth," in Socialist Revolution,May-June 1970, especially pp. 37-43.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
In equation (2), k represents the equilibrium ratioof consumption to income, in other words, the long-run marginal propensity to consume.
The disequilibrium hypothesis turns on theparameter values more than on the degree to whichobserved data fits the general form of equation (2).It is central to the disequilibrium hypothesis that kbe close to 1.0. Moreover, it is implicit in this hypo-thesis that the numerical values of 9 and 6 aretogether sufficiently large that adjustment of con-sumption to income is reasonably complete withina short period, say a year or so, rather than requir-ing anything like a life time.
Before we confront the data, we have to takesome care with the definitions of the variables, par-ticularly with consumption (C) and its complement,savings (S = Y-C). Actually two definitions of con-sumption and saving are appropriate since twoseparate propositions are at issue. One propositionis that households do not make resources availablefor widening and deepening capital in the form ofplant and equipment, the form of saving respon-sible for the secular rise in the real wage and
thereby of the standard of living of the overwhelm-ing majority of people who depend on wages. Thesecond proposition is that all household saving ispurely passive, not at all the result of conscious anddeliberate decisions.
With respect to the first proposition, house-hold saving is appropriately defined in terms ofacquisition of financial assets and liabilities, in-
cluding the investment of the household sector inunincorporated businesses, since financial savingsrepresent the resources that households makeavailable to the rest of the economy. The impor-tance of savings to social stability stems from thefact that by and large people do not own the meansof production they use in their daily work, so whatmatters is not aggregate saving but saving in theform of producers’ goods.[17] In an economy ofsmall proprietors with no market in labor, savingwould be a purely individual matter. Others wouldbe no more concerned with individuals’ savingsdecisions than with their decisions with respect tohair styles. By this token, what a householdaccumulates in the form of consumers’ durables,including owner-occupied housing, is of no rel-evance to the growth of labor productivity andreal wages in general (except to the extent that
acquisition of consumer durables and homes af-fects the rate of saving at some later date). Thus, solong as we focus on the functional relationshipsbetween savings and hierarchy within the capitalistsystem, it is reasonable to define household savingsin terms of the net acquisition of financial assets.
But to the extent the goal is to refute the ideathat household savings are small because corporatesavings are large, the acquisition of physical assetsby households is relevant - so long as these assetsare fungible. That is, insofar as it is the delibera-tion and consciousness of the household saving de-cision that is at issue, investment in physical assetsacquired for personal use is legitimately included insaving, provided these assets can be readily con-verted into money, and therefore into generalizedconsumption. For the acquisition of fungible physi-cal assets, like the acquisition of financial wealth,can conceivably reflect a conscious and deliberatedesire to postpone consumption in order to maxi-mize life-time utility, and it is precisely this hypo-thesis that is here under attack.
Of all the durable physical assets that house-hofds normally purchase for their own uses, hous-ing would appear to be the only one that qualifiesas a surrogate depository for resources allocated tofuture consumption. A home-owner not only hasthe possibility of realizing his nominal equity atalmost any time, but casual empiricism suggeststhat this option is exercised sufficiently often by re-tired couples to give the benefit of the doubt to theview that acquisition of equity in housing repre-sents deliberate abstinence from current consump-tion. I say &dquo;benefit of the doubt&dquo; because of theconsiderable folklore that suggests that the savingthat takes place as mortgages are’paid off is fromthe point of the household inadvertent, or moreaccurately, unavoidable. It is highly likely thatownership reflects a desire to reap the consider-able tax advantages it affords the middle class orsimply the desire for status, and that the savingthat takes place in reducing the principal outstand-ing on the mortgage is nothing but a payment forthese benefits. It is certainly suggested oftenenough that the typical home-owner neither knowsnor is concerned with the division of his monthlypayment between interest and the reduction of
principal. Be this as it may, prudence indicates thatinsofar as the issue is the consciousness or delibera-tion behind household saving decisions, net invest-ment in owner-occupied housing be included in
17. Aggregate saving is of course of crucial importanceto stabilization policy, but that is a question totallyseparate from our present interest.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
saving.[18]This argument distinguishes saving in the
form of owner-occupied housing from saving in theform of consumer durables such as washing ma-chines, automobiles, and television sets. Consumerdurables are in my view rightly excluded fromsaving whether it is the rationality of household
saving decision or their effect on the expansion ofthe means of production that is at issue. Washingmachines and cars are acquired for their direct use-fulness or their prestige value, not as a surrogatefor a savings bank. The fact that these goodsprovide a flow of services over a relatively longperiod, the usual justification for including them insaving, is quite beside the point. Our present in-terest is in the rationality of household savingdecisions and the availability of household savingsfor investment that increases the level of real wages.And relative to these questions, it is wholly imma-terial whether a good provides its services at a
single point in time or over a long interval. Indeed,if the interval of time over which a good providesutility is the touchstone of whether it representscapital or current consumption, then a particularlyexquisite dinner I ate twelve years ago in one ofBoston’s finer restaurants should be considered as
capital; it still brings a warm glow whenever I thinkabout it, a glow that to a neoclassical economistought to represent a current service every bit asmuch as the nightly flow of pleasure from the tele-vision set I bought two years ago.
Thus, to a reasonable first approximation, to-tal household savings is defined as the residual ob-tained by subtracting personal outlays, as mea-
sured by the Department of Commerce,[ 19] fromhousehold incomes. This approximation is im-
proved by excluding imputed expenditures, chieflythe net rental value of owner-occupied housing,from personal outlays.[20] The Department ofCommerce series on disposable incomes serves as a
first approximation to household income. Ofcourse, the same imputations that are excludedfrom consumption must be excluded from income.Moreover the approximation to the income con-trolled by households is improved by excluding netcontributions to pension funds, which are no morehousehold saving than are the retained earnings ofcorporations.[21] Total household saving is thendefined by the difference between disposable in-come (less imputations and pension fund contribu-tions) and personal outlays (less imputations).
Financial saving, including investment in un-incorporated business, is measured by data re-
ported on a quarterly basis by the National Indus-trial Conference Board. On the one hand house-holds accumulate assets in the form of cash,demand deposits, savings accounts, credit andequity market instruments, net investment in un-incorporated business, and life insurance reserves.On the other hand, households incur debt, princi-pally in the form of consumer credit and mortgagecredit. &dquo;Financial saving&dquo; is defined as the differ-ence between the net increment in holdings of theseassets and the net increment in household debt.[22]Household income is defined as before - as dis-
posable income, net of imputations and pension-fund contributions. Household consumption is, forpurposes of measuring households’ contributionsto wage-related capital, the difference between
18. A similar argument suggests that household savingin the form of life-insurance reserves be included in totalhousehold saving, as well as in financial saving, eventhough it can be argued that the acquisition of equity inlife-insurance policies does not reflect in general a savingsde cision.
19. Reported in the Survey of Current Business, U.S.Department of Commerce, Washington D.C., supple-mented by The National Income and Product Accounts ofthe United States, 1929-1965, also published by theDepartment of Commerce.
20. The Survey of Current Business and The NationalIncome and Product Accounts include an annual series of
imputations. The imputations subtracted from both dis-posable income and personal outlays are the following:
1. Space rental value of owner-occupied dwellingsminus associated purchases of goods and servicesminus capital consumption allowances minus in-terest.
2. Space rental value of owner-occupied farm dwell-ings plus food and fuel produced and consumed onfarms minus associated purchases of goods andservices minus capital consumption allowancesminus interest.
3. Services furnished without payment by financialintermediaries.
4. Food furnished employees.5. Standard clothing issued to military personnel.6. Employees’ lodging.
The annual series were converted to quarterly series bylinear interpolation.
21. Pension fund contributions are reported in Flow ofFunds Accounts 1945-1968, published by the Board ofGovernors of the Federal Reserve System, WashingtonD.C. Again the reporting (Table 9, line 27 minus line 5) ison an annual basis, and linear interpolation has beenemployed to convert the annual information to a quarter-ly series.
household income and financial saving.To summarize: for total household saving we
use the following variables
DIS= Disposable income (Department ofCommerce
IMP= Imputations to personal income (Depart-ment of Commerce)
PF= Net contributions to pension funds (FederalReserve)
PO= Personal outlays (Department of Commerce)For financial savings,
,6FA = Change in households’ financial as-sets, including net investment in un-incorporated business, but excludingcontributions to pension funds (Na-tional Industrial Conference Board).
1~FL = Change in financial liabilities (Na-tional Industrial Conference Board)
The historical record from which the para-meters of equation (2) are estimated consists of 61quarterly observations running from the end of theKorean War to mid-1968. The extraordinary condi-tions of the early post-World War II period and thecontrols and scare-buying of the Korean War make1953 appear a sensible beginning point. The lack ofdata beyond 1968 have dictated the end point.
Throughout, ordinary least squares has beenused for estimating purposes. As a preliminary toapplication of regression techniques, equation (2)had to be put into a discrete form. I have followedthe suggestion of Hendrik Houthakker and LesterTaylor[23] in translating differential into differenceequations; this in essence means assuming (1) thatthe actual observation for any quarter is exact forthe mid-point of the quarter, and (2) that variables
follow linear paths between actual observations.[24]Thus integrating
we obtain
where the subscript -1 represents the observationfor the previous quarter. Solving for C, we have
with
a A A
Thus, after obtaining estimates A1 , A 2, and Atfrom ordinary-least-squares regressions, we cansolve for 0, k, and 6 according to the formulas
There is one final and important hurdle to besurmounted before we can profitably examine thedata. The parameters of direct interest - O, k, andd - are nonlinear functions of the parameters A1,A2, and A3. Therefore the standard errors of theestimates A 1, A2, and A 3 are not immediatelytranslatable into confidence intervals. A relativelyelaborate procedure, due to Fieller, had to befollowed to estimate confidence intervals around 0,k, and 6.[25] &dquo;
23. H. Houthakker and L. Taylor, Consumer Demand inthe United States, 1929-1970, Harvard University Press,Cambridge, Massachusetts, 1966, chapter 1.
24. For example, observations for first quarter 1965(1965:I) are assumed to represent exact measurementsfor February 14, 1965, and observations for 1965:II areassumed to reflect the actual state of affairs on May 15,1965. The level of each variable for any date betweenFebruary 14, 1965, and May 15, 1965, is assumed to be aweighted average of the observations for these two dates,with the weights proportional to the length of timeseparating the date in question from February 14 andMay 15, respectively.
25. See Zvi Griliches, "Distributed Lags: A Survey,"Econometrics, January 1967, especially pp. 32-33.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
Consider now the parameter estimates whenhousehold saving is defined in Department of Com-merce terms, to include investment in owner-oc-
cupied housing.[26] The estimates of Al, A2, andA 3 that result from fitting equation (3) to 61 quar-terly observations (1953:11-1968:11) are given im-mediately below, with standard errors in paren-theses:[27]
TotalHousehold
Saving
A A 1The estimates of e, k, and 1, together with 0.90and 0.95 confidence intervals, are presented in thefollowing table.
Parameter Estimates Confidence Intervals
What results emerge from these parameterestimates? First, though the long run marginal pro-pensity to consume is less than unity, k = 1.00 is in-
26. All computations were carried out by MichaelClurman, using a program written by William Raduchel. Iam indebted to Professor Raduchel for writing a specialsub-routine for computing confidence intervals accordingto Fieller’s procedure.
27. This equation was also estimated with a constantterm:
The Durbin-Watson statistic was 2.31.
cluded in both the 0.90 and 0.95 confidence inter-vals. Thus aggregate consumption and income datacannot be used to contradict the hypothesis that thelong-run marginal propensity to consume is unityeven if household saving is defined to include in-vestment in owner-occupied housing. In any event,A
the estimate k = 0.976 is considerably higher thanthe average propensity to consume, which - onpresent definitions of income and consumption -was just under 0.95 over the 15-year period in ques-tion.
Second, the estimate of the short-run marginalpropensity to consume, ~ = 0.533, is significantlydifferent from zero. This estimate suggests a rapidimmediate adjustment of consumption to income,an adjustment hard to reconcile with Friedman’sidea that &dquo;transitory&dquo; income has no impact onconsumption.[28] A
Taken with the estimate G = 0.898, a $1.00increase in household income from an initial posi-tion of equilibrium would, if sustained, produce aconsumption change of $0.80 within a one-yearperiod. The figure below illustrates the response ofconsumption over time to a once-and-for-all in-come change at time t, assuming households are inequilibrium at time t.
FIGURE 1:
The implications of quarterly time-series dataare even more striking when financial savingreplaces total saving. First of all, the accumulationof financial assets net of liabilities, taken togetherwith investment in noncorporate business, was
28. Friedman, op. cit., chapter 3.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
hardly two percent of household income over the 15year period 1953-1968.[29] Second, the estimate ofthe long-run marginal propensity to consume is al-most precisely 1.00. Third, the lag in adjustingspending to income, which (along with the grwothin household incomes) explains the differencebetwen the average and marginal propensities toconsume, is even shorter than on the previous defi-nition of saving.
Specifically, the equation relating consump-tion to income becomes
Financial
Saving
The coefficient of 6,Y is less than half its standarderror, presumably because of multicollinearity.This does not, however, mean that consumptionadjusts slowly to income; if we estimate the rela-
tionship between consumption and income withoutAY, we obtain the equation[30]
, All A
Estimates for O and k are presented below,together with their confidence intervals:
Parameter Estimates Confidence Intervals
, A
The estimate O = 3.537 implies that within onequarter a $1.00 increase in income leads to a $0.60increase in spending. If the $1.00 increase is sus-tained for a year, spending adjusts to $0.95. FigureTwo illustrates the path of adjustment from oneequilbrium to another that accompanies a jump inincome.
29. A simple regression of consumption against incomeyielded the equation
30. Regressions were also run with constant terms,with the following results:
The constant terms are significant at the 0.95 level,though barely so in the second equation. In view of theinherent unreliability of the data and what I understandto be an upward bias in the estimate of the constant termdue both to simultaneous-equations bias and to errors ofmeasurement in the independent variables, the evidencehardly seems to warrant rejecting the hypothesis that theconstant term is in fact zero. Cf. Ando and Modigliani, op.cit., p. 63.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
The estimate of the long run marginal propen-sity to consume, k = 1.005, is in close harmony withthe overall view of household savings propoundedin this paper. Equally impressive, perhaps, is the
narrow band of the confidence interval. These re-
sults, taken together with the relatively rapid ad-justment of spending to income implied by thedata, can give little comfort to anyone who wouldrely on household decisions to provide for the
expansion of the means of production.With a long-run marginal propensity to con-
sume in the neighborhood of 1.00, feasible rates oflong-run growth are nonexistent or miniscule forcombinations of parameter values suggested byhousehold behavior and conventional estimates of
capital productivity. Assuming
the general formula relating steady-state consump-tion and income is
where g represent the rate of growth. That is to say,(4) holds when income, consumption, and savingare all growing continuously at the rate of g.[31]
The problem is that equation (4) takes g asexogenous, which is reasonable only if the rate ofgrowth is determined primarily outside the house-hold sector. If households controlled all income,the equilibrium growth rate would have to takeaccount of the relationship
where d represents the incremental output:capitalratio and s represents the rate of aggregate saving(= aggregate investment) to income; that is, ifhouseholds controlled all income, we should have
Substituting from (5) into (4) gives the equilibriumrelationship
With k approximately 1.00 and 6 equal to 0.0,equation (6) reduces - for nonzero g - to
so that positive steady-state growth is infeasible un-less the output: capital ratio exceeds the rate of ad-justment of spending to income, that is, unlessd > 8. Steady growth pretty clearly requires inter-mediation between households and resource alloca-
tion, since d may reasonably be taken to lie between0.2 and 0.4, and the 0.95 confidence interval puts 0between 1.96 and 5.71.
If we look at total saving rather than financialsaving, steady growth is feasible for all values of Q,[32] but the rates of growth corresponding to thevalues of o between 0.2 and 0.4 are hardly encour-aging.[33] Substituting the maximum-likelihood
~ A A
estimates e = 0.90, k = 0.976, 6 = 0.53, equation(6) becomes
31. Thus, with k = 1.005, δ = 0.0, and &thetas; = 3.54,g = 0.06 is consistent with C/Y = 0.988, the trend ratioover the 15 years of record with saving defined infinancial terms. In other words, with steady growth at arate of 6%, an average propensity to consume of 0.988 isconsistent with a long-run propensity to consume of 1.005and adjustment coefficients of &thetas; = 3.537, δ = 0.0.
32. Plotting the productivity identity and the savingsequilibrium
on a common graph, we have a picture like the followingfor the case k = 1:
Existence of an equilibrium in the interior of the positiveorthant depends on the slopes of the two curves. In thepicture a positive equilibrium exists for δ = σ2 but not ford = σ1.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
With a growing population, these rates of growthare hardly adequate even to maintain a constantstandard of living, much less to increase it at a rateconsistent with the proposition that capitalism pro-vides ever-increasing abundance for all.[34]
Up to a point, these results are consistent withthe Fisherian hypothesis of Friedman and Modi-gliani, as well as with the disequilibrium hypo-thesis. As long as we do not take Fisher so seriouslythat we assume the estimates 6, 3, and k are con-ditioned by the existence of non-household saving- provided, in other words, we do not attempt toexplain the high marginal propensity to consumeand rapid adjustment of spending to disposable in-come by the existence of savings households don’tcontrol - the difference between competing expla-nations of household behavior turn principally onthe values of the parameters estimated from in-
come and spending data. When Friedman and
Modigliani reduce permanent income or life-timeresources to observable magnitude, their consump-tion functions hardly differ in form from the con-sumption function suggested by the disequilibriumhypothesis.[35]
Take Friedman. Consumption is supposed to
depend on permanent income, Y p . In particular, itis hypothesized to be strictly proportional:
so that the rate of change of consumption is pro-portional to the rate of change of permanentincome
But how is permanent income estimated? By arecursive relationship that makes the change in
each household’s permanent income proportionalto the difference between its actual income and itscurrent idea of permanent income, as well as to
current permanent income, the second term sup-posedly reflecting long-run growth in permanentincome: .
Here is the rub: equation (9) is but a restrictedversion of equation (2) in which o is constrained tobe zero. Except for the suppression of Y, the soledifference between (2) and (9) lies in the interpreta-tion of the parameters, which are related to oneanother by the formulas
With k <1, we have a positive intercept for the savings-equilibrium function
and a positive equilibrium exists for every value of σ.
33. Indeed, if output and capital include the rental valueand stock of owner-occupied housing as well as producers’goods, the reasonable range of o might be shifted
downward considerably at both ends.
34. Would zero population growth rescue capitalism?Probably not, for the output:capital ratio would beadversely affected if the opportunity for widening capitalthat population growth presents were to disappear, andall investment became of necessity capital-deepening.
35. It can be shown that the rule
emerges from maximization of a utility function of theform
subject to the life-time wealth constraint
where Ao represents material assets, p the rate ofinterest, and &thetas; the (subjective) rate of discount of utility.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
That is, the discrete analog of (9) can be interpretedin terms of the permanent-income hypothesis or interms of the disequilibrium hypothesis. In bothcases we have
..7 I , ,
Apart from interpretation, the only differencebetween the two approaches is that Friedman’sintroduces an extra parameter a, to reflect anautonomous growth in permanent income that
supposedly exists apart from the pull of currentincome on permanent income. Since two variablespermit us to estimate at most two parameters, oneof Friedman’s three must be specified exogenously.Friedman opts for setting a equal to the long-termrate of growth in actual income, which he puts at0.02. Following him, we have
. ’B
Using the concepts of income and consump-tion elaborated earlier, the regression equationestimated for the period 1953:II-1968:II for totalhousehold saving is
TotalHousehold
Saving:
1B a
If the estimates A = 0.678 and A 2 = 0.138 areinterpreted in terms of the permanent-incomehypothesis, we obtain the following results:
In terms of the disequilibrium hypothesis, theresults are
If saving is defined to exclude owner-occupiedhousing, the relevant equation has already beenreported:
The results are
and
Parameter Estimates Confidence Intervals
If the data are interpreted in terms of the permanent-income hypothesis, the response of perma-nent income to actual income is quite rapid. B =1.556 implies that, if initially permanent incomeand actual income are equal, a change in actualincome of $1.00 sustained for only one year leads toa revision of permanent income of $0.79. B = 3.557implies a virtually complete adjustment within ayear’s time: a $1.00 sustained increase in ac-
tual income leads to a $0.97 response in perman-ent income.
Such rapid adjustment is formally consistentwith the permanent income hypothesis. Any speed
. of adjustment is! But the more rapidly permanent
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
income responds to current income, the less mean-ingful is the concept of permanent income as a dis-tinct basis of spending decisions. If permanent in-come adjusts to actual income as quickly as thedata suggests, it is hard to take seriously the ideathat this unobservable construct represents a fore-cast of the sustainable yield on life-time resources,material and human.[36]
Modigliani’s life-cycle hypothesis leads toequations equally easy as Friedman’s to reconcilewith the equations suggested by the disequilibriumhypothesis. But it is equally hard to reconcile thelife-cycle hypothesis with the parameter estimates.that emerge from the quarterly regressions. Afteraggregating over age cohorts, Ando and Modi-gliani[37] obtain a consumption function of theform
where YL represents current (nonproperty) in-come ; YLe represents what Ando and Modiglianicall the &dquo;average annual expected [nonproperty] in-come,&dquo; defined as the present value of future ex-pected labor earnings averaged over the earningspan; and A represents households’ net worth. YLe ,like Friedman’s Yp , is not observable. Ando andModigliani assume that YL is equal to YL , whichis to say that nonproperty income is expected togrow at a rate equal to the rate at which householdsdiscount future income.[38] Setting YL equal toY L , equation (10) becomes
L
The formal similarity between the consumptionfunctions that emerge from the life-cycle and
disequilbrium hypotheses emerges clearly whenequation (11) is differentiated with respect to time:
Since A, the rate of change of net worth, is thecurrent rate of saving
equation (12) can be written
But this is the same equation as (1) except for thesubstitution of Y for Y. Ignoring this differenceand differences in the treatment of capital gainsand consumer durables,[39] we have
The life cycle hypothesis thus shares with thedisequilibrium hypothesis the prediction that thelong run marginal propensity to consume is 1.00.But the &dquo;long run&dquo; for Modigliani is of the order ofa life time, not the year or two that is implicit in thedisequilibrium hypothesis. The data suggest para-meters more in accord with the disequilibrium viewof saving than with the life-cycle view.
Modigliani and Ando make three assumptionsthat enable them to predict a range of numericalvalues for the coefficients a 1 + a 2 and a 3:(1) households’ utility functions are such that theyallocate resources uniformly over life spans as-
sumed to be 50 years; (2) earnings of various agegroups are the same in any given year, and
expected to remain the same at all times within theearning span assumed to be 40 years overall; and(3) the rate of return on households’ assets is con-stant over time and over households and expectedto remain constant.[40] With these assumptions,the anticipated ranges of a 1 + a 2 and a 3 are
respectively [0.61, 0.73] and [0.08, 0.12] for rates ofreturn and growth rates lying respectively between0 and 5 percent and 0 and 4 percent.[41]
36. Friedman’s own regressions (of. cit., chapter 5)differ in four respects for the present ones. First, theyutilize annual rather than quarterly data. Second, theperiod covered is 1897 to 1949. Third, savings are definedto include consumer durables in addition to the elementsof accumulation we have included under "total householdsaving." Fourth, the integral form of equation (9) is used,so that current consumption is regressed against adistributed-lag function of income. Friedman obtains anestimate B=0.40, hardly one-quarter as large as ourestimate based on total saving and one-ninth the size ofour estimate based on financial saving.
37. Ando and Modigliani, op. cit.
38. Ando and Modigliani also test an alternativespecification, in which current income is corrected by theunemployment rate in forecasting expected income. Asthey note, the analysis is not changed much by thisrelatively minor correction.
39. Ando and Modigliani include both in their definitionof saving.
40. Ando and Modigliani, op. cit., p. 59.
41. Ibid., p. 60.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
is not far from the Ando Modigliani prediction.Moreover the estimate a 1 + a 2 = 0.533 agressrather closely with the Ando-Modigliani estimatebased on annual data from an earlier period and ona definition of saving that includes capital gainsand consumer durables.[42] But the estimate
a 3 = 0.898 is well outside the range plausiblyconsistent with the Ando-Modigliani predictionthat a 3 would fall in the interval [0.08, 0.12]. It ishard to interpret the estimate a 3 = 0.898 in termsof a deliberate and conscious allocation of wealthbetween current and future consumption.[43]
All these tests and comparisons of the disequi-librium hypothesis - comparisons with the perma-nent-income hypothesis as well as comparisonswith the life-cycle hypothesis - are limited by the
form in which the data come to us. Aggregate timeseries based on national income accounts sufferfrom defects too obvious to require extensive dis-cussion. But it should be borne in mind that for our
purposes aggregate time series appear to offer theonly worthwhile basis for comparison of competingexplanations of household behavior, at least untilsuch time as an adequate number of long-term pro-files of income and spending of individual house-holds become available. Certainly cross-sections ofthe kind used by Friedman and Modigliani are be-side the point. These cross-section analyses maydiscriminate between the permanent-income or thelife-cycle hypothesis and the naive hypothesis thatconsumption depends solely on current income,and it may have been important at one time to laythe naive view to rest.[44] But I can detect noresults from these cross-sections that are not asconsistent with the disequilibrium hypothesis as
with the permanent-income or life-cycle hypothesis.For example, both Friedman[44] and Modigli-ani[45] interpret the inconsistency between esti-mates of the propensity to consume based on cross=section relationships between consumption andincome and the estimates based on time-series rela-
tionships in terms of a positive correlation betweentransitory income and observed income in cross-
42. Ando and Modigliani obtain the estimate â1 + â2= 0.52 in equation (6), p. 64, ibid.
43. With k constrained to be equal to 1.00, as both thelife-cycle and disequilibrium hypotheses predict, a re-gression of consumption against income and laggedconsumption variables give the results below. Saving ishere defined to include investment in owner-occupiedhousing (but not consumer durables, capital gains, orpension fund contributions).
44. Obviously many Keynesians believed this naiveview in the years immediately follwing the publication ofThe General Theory of Employment, Interest, andMoney, Harcourt Brace and Co., New York, 1936. ButKeynes himself was more circumspect in limiting thedependence to a relatively short period in which thelagged values of income and consumption might reason-ably be regarded as fixed. Consider how Keynes qualifieshis "fundamental and psychological law ... that men aredisposed, as a rule and on the average, to increase theirconsumption as their income increases, but not by asmuch as the increase in their income" (p. 96):
This is especially the case where we have shortperiods in view, as in the case of the so-calledcyclical fluctuations of employment during whichhabits, as distinct from more permanent psycho-logical propensities, are not given time enough toadapt themselves to changed objective circum-stances. For a man’s habitual standard of lifeusually has the first claim on his income, and he isapt to save the difference which discovers itselfbetween his actual income and the expense of hishabitual standard, or, if he does adjust his expendi-ture to his income, he will over short periods do soimperfectly. Thus a rising income will often beaccompanied by increased saving, on a greaterscale at first then subsequently.
45. Friedman, op. cit., chapter 4.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
sections. But the inconsistency is equally explicablein terms of the disequilibrium hypothesis: in a ran-dom cross section of households one would expectthat the upper-income brackets would include alarger than average proportion of families that re-cently enjoyed income increases to which they havenot yet adjusted their consumption; similarly, it isto be expected that a larger than average propor-tion of low income families recently suffered re-verses to which they have not yet adjusted. Disequi-librium rather than any differences between transi-
tory and permanent incomes would thus accountfor the relatively low propensity to consume esti-mated from cross-section studies.[46]
This is not to say that cross sections are of no
value in assessing the relative merits of the disequi-librium hypothesis and its Fisherian competitors.On the contrary: at the very beginning of this sec-tion, I emphasized the importance of studying theimpact of corporate saving on household saving.And Philip Cagan’s analysis of the effects ofpension plans on household saving was cited insupport of the disequilibrium hypothesis.
Indeed, Cagan’s results may contain more im-plications than have thus far been exploited. In
terms of the disequilibrium hypothesis, the rise ofprivate pension plans and the assumption of publicresponsibility for old-age insurance explicit in thesocial-security system would be explained as socialresponses to a lack of deliberation and planning inhouseholds’ saving decisions. In fact, I know of nosociety in which provision for retirement is left tothe individual or the nuclear family. In much of theworld, the extended family predominates, and themaintenance of the elderly is a common responsi-bility of their able-bodied kin - children, nephews,and so forth. Capitalist societies, it is true, for atime experimented with individual responsibility,only to learn that for the typcial household thebenefits of saving for a distant future were nomatch for the temptations of current spending. Myuniversity - and it is in no sense exceptional in thisregard - never asked me if I wished it to set asideincome for my retirement rather than paying it outto me. Possibly it fears that if I were to control themoney it allocates to my future, I might put offsaving for so long that I would eventually join theranks of the street people in Harvard Square, be-smirching the name of Fair Harvard by importun-ing passers-by for spare change.
Formal models apart, there is one disaggrega-tive test of the competing hypotheses that a fairnumber of you who are reading these words are in aposition to perform, prospectively or retrospective-ly. Graduate students, as a group, have current in-comes far below their &dquo;permanent incomes,&dquo; onany reasonable interpretation of the term. Certainlyif graduate students planned their current con-sumption in terms of their life-time resources, theywould live well beyond their present means. Buthow many in fact exhaust the possibilities for bor-rowing in order to bring their standard of living upto anything like what it might reasonably be on thebasis of their life-time resources? Casual empiri-cism suggests that the overwhelming majority limittheir borrowing to the expenses of their educationand the basic necessities of life.
And the plot thickens. A further test of the dis-equilibrium theory of saving is the effect on savingsbehavior of the transition from graduate student toacademician or bureaucrat. It will be the case, I am
willing to predict, that upon taking your first full-time job and suddenly finding yourself with a salaryin excess of $10,000 you will save incredibleamounts of money - perhaps a quarter of more ofyour disposable income. You will have to search forways to spend sums that far exceed the income onwhich your life style has been patterned the lastseveral years. But don’t worry. If your experienceresembles mine, you will learn quickly enough. Itwon’t be long before you have as much troublemaking ends meet as you did as a graduate stu-dent.[471
Indeed, the Fisherian paradigm is in my viewmuch more notable as an illustration of the intel-lectual sleight of hand that characterizes the teach-ing of the neoclassical theory of choice than as amodel of household behavior. After struggling withand finally mastering the intricacies of the house-
46. Modigliani and Brumberg, op. cit., pp. 406 ff.
47. Thus the disequilibrium hypothesis as well as thepermanent-income hypothesis affords little comfort tothose who argue that income inequality promotes capitalaccumulation. This idea has such obvious appeal for therich, especially for oligarchies that dominate most poorcountries, in providing a social justification for theirwealth and income, that it is not easily shaken byargument or evidence. Insofar as there is truth to theview that inequality promotes accumulation, it is—ac—cording to the argument developed here—inequality ofcontrol, not inequality of income or wealth, that matters.The two kinds of inequality are obviously related incapitalist society. Egalitarian, nonhierarchical societieswell might forge collective decision-making mechanismsfor allocating generous amounts of resources to accumu-lation, but how long it would take is an open question.
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
wife’s choice between canned peaches and cannedpears,the grateful student is one day told that atlast he possesses a general theory equally applic-able to all household decisions. All that is requiredto apply the model in unfamiliar contexts is a
suitable relabeling of the axes of the choice dia-gram : to apply the model to intertemporal choiceproblems, it is merely necessary to substitute &dquo;pre-sent consumption&dquo; for &dquo;canned peaches&dquo; and &dquo;fu-
ture consumption&dquo; for &dquo;canned pears;&dquo; and theinterpret the &dquo;budget line&dquo; in terms of &dquo;life-timeresources&dquo; and the relative &dquo;prices&dquo; of present andfuture consumption rather than in terms of incomeand the prices of peaches and pears. The student’sprofound relief at learning that he has mastered auniversally applicable theory, coupled with theteacher’s reassuring confidence in the theory, typi-cally prevents him from examining the substantivedifferences hidden by the formal similarity with thepeaches-ears paradigm.[48]
These substantive differences are of two kinds.
First, formal statements of the theory of neoclassi-cal choice simply take preferences - endowed withsuch mathematical properties as completeness andconsistency - as given. Presumably a neoclassicaleconomist would argue that this is a legitimatesimplification of the outcome of a process by whichtrial and error determine preferences. And when itcomes to peaches and pears, I suppose I have noserious quarrel with this justification of the as-
sumption of a meaningful preference ordering. Butthere are serious differences between the plausi-bility of a trial-and-error process in the context ofchoosing between peaches and pears and its plausi-bility in the context of intertemporal choice. In thepeches-pears model, it is reasonable to supposethat choices are made repeatedly in the space of ashort time, with little time elapsing betwen any spe-cific decision and the realization of its conse-
quences. The housewife buys canned fruit onMonday, her family eats it during the week, and theprocess is repeated seven days later. By contrast,the choice of an intertemporal consumption plancan reveal its consequences to the decision-maker
only gradually. By the time a twenty-five year oldrealizes the consequences of his choices, he is - letus say - thirty, with new responsibilities and obli-gations. For relevant purposes, he can reasonablybe considered a different person. Certainly he is not
making a similar decision in a similar context, inthe sense that the housewife can be assumed to bein the trial-and-error processes that lie behind the
preferences she exhibits in the supermarket. Forthe housewife the relationship between the choiceof canned fruit and realization of the consequencesof this choice is clear, immediate, and direct. Bycontrast, the lack of such a clear, immediate anddirect relationship between a household’s savingdecisions and its future consumption pattern pre-sents a significant obstacle to basing the assump-tion of a meaningful preference ordering on a trial-and-error process of comparing intertemporal con-sumption plans. And apart from a background oftrial and error, I know of no justification for theneoclassical postulate that complete and consistentpreferences exist.
But whatever one believes about the similarityof intertemporal preferences to the timeless &dquo;indif-ference maps&dquo; on which neoclassical choice theoryis built, there can hardly be any question that thebudget line must play a role in the neoclassicalmodel of intertemporal choice that differs funda-mentally from its role in the timeless model. In theintertermporal context, the budget line must neces-sarily reflect a subjective estimate of one’s futureearnings, as distinct from the income that the bud-get line can plausibly be assumed to represent withcertainty in a model of the housewife’s decisions inthe local supermarket. In my view, the uncertain-ties that surround most households’ estimates offuture earnings, especially insofar as these reflectwages and salaries, reduce the idea of deliberatechoice of a life-time consumption plan based onmaximization of life-time utility subject to a life-time &dquo;budget&dquo; constraint to a parody of the actualchoice process.[49]
This is not to say that literally no one savesdelibrerately. Hardly a month goes by, for example,that a newspaper does not print a story about anold man or woman who, having lived in abjectpoverty, dies with a hoard of carefully accumulatedcash stuffed under the mattress. Obviously thesesavings, like the savings of the Calvinist entrepre-neur of Weberian fame,[50] are not adequately ac-
48. With luck, however, you may save enough over thelearning period to make the down-payment on a house.
49. Of course, the deeper he studies the formalproperties of the paradigm, and the greater the effort heis obliged to make, the greater the psychological invest-ment he has in the theory. The less tempting, therefore,is serious and thorough questioning.
50. In a survey of households conducted by GeorgeKatona and Eva Mueller, only 15 percent of the respon-
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
counted for by the disequilibrium theory. [51] Butneither, I would submit, are such savings acountedfor by the utility calculus of neoclassical theory.Surely the power, prestige, status, and psychologi-cal security that derive from financial or industrialempire are more relevant to the explanation ofatypically large individual saving rates than is a
balancing of the gratifications of future consump-tion against the pleasures of present consumption.Surely it is ludicrous in the extreme to suppose thatthe Rockefeller clan balances an extra yacht todayagainst a yacht and a half ten years hence, as theFisherian model would suggest.
Whatever qualifications may be necessary toexplain deviations from the norm, my hypothesisremains that capitalists as individuals, like workers,
save principally when their incomes rise faster thanthey can manage to spend them. On a statisticalbasis capitalists may save a significantly larger por-tion of their disposable incomes than do workers,and for many purposes there may be adequate jus-tification for reflecting this statistical regularity inthe postulate that saving propensity is a matter ofincome class. But this does not in itself contradictthe disequilbrium hypothesis: successful workersbecome capitalists and failed capitalists becomeworkers; the differences in savings rates are thusaccountable in terms of income changes. In myview the significant dichotomy with respect to
saving rates is not based on differences in indi-vidual incomes, but on the differences betweenindividuals and organizations. It is not the size ofthe capitalists’ incomes or a special set of attitudes,a drive towards accumulation, that matters. It israther capitalist control of the production processthat gives’this class a dominant role in determiningthe rate of saving.
dents claimed to have a "pretty firm idea how much theyought to save." The majority said they saved whateverthey could manage. (Katona and Mueller, ConsumerResponses to Income Increases, Brookings Institution,Washington, D. C. 1968, p. 90).
51. Max Weber, The Protestant Ethic and the Spirit ofCapitalism.
52. But there is no reason to retreat overly hastily. Thesavings of an Arkwright might as well be accounted forby the disequilibrium hypothesis as by the ProtestantEthic. The income of many early entrepreneurs, like theincome of the proverbial Texas oil baron, may simplyhave grown faster than it could be spent, no matter howaristocratic one’s tastes
Stephen A. MarglinDept. of EconomicsHarvard UniversityCambridge, Mass.
02138
at Harvard Libraries on November 13, 2013rrp.sagepub.comDownloaded from
![Review of Radical Political Economics - Threefolding · Review of Radical Political ... schemes of Silvio Gesell, see Preparata and Elliott (2004). 4. See, for instance, Veblen ([1923]](https://static.documents.pub/doc/80x56/5e2ea05e7928fb758346361c/review-of-radical-political-economics-review-of-radical-political-schemes.jpg)
