Stock Ownership Patterns, Stock Market Fluctuations, and ...

JAMES M. POTERBA MassachusettsInstituteTechnology

ANDREW A. SAMWICK Dartmouth College

Stock OwnershipPatterns, Stock MarketFluctuations, and Consumption

THE BULL MARKETof the last yearhas raisedthe totalvalue of corporate stock in the United States by nearly a trillion dollars. While many analystshave triedto explain or interpretthe recentmovementsof the stock market,there has been less attentionto the link between rising stock prices and real economic activity. How are the gains from an increasein sharepricesdistributedacrosshouseholds?Whatfractionof these gains accruesto a small set of wealthy investors?How do rising stock prices affect consumerspending?

Thestandardtextbooktreatmentof aggregateconsumptionholdsthat consumptiondependson laborincome andfinancialwealth.' The marginal propensity to consume out of wealth is typically taken to be approximately0.04 per year. In this framework,the wealtheffect of a stock marketrally shouldhave an importantstimulativeeffect on consumption.Althoughthis view neglects some potentiallyimportantfactorsthatmightalso affect consumptiondirectly(notably,thepossibility thatstock prices may rise as a resultof a decline in real interestrates),

We aregratefulto RochelleAntoniewicz,JohnCampbell,Joel Dickson,JerryHausman, N. Gregory Mankiw, Thomas Morley, Chris Probyn, James Shapiro, Robert Shiller,AndreiShleifer,DavidShulman,ChristopherSims, JustinSmith,MarthaStarrMcCluer,David Wilcox, and participantson the BrookingsPanel and at a seminarat the FederalReserveBoardfor helpful discussions. We also thankAlex Davidfor data on sharerepurchasesand, especially, ArthurKennickellfor assistancewith the 1992 Survey of ConsumerFinances. This researchwas supportedby the NationalScience Foundation.

1. See, for example, DornbuschandFischer(1994).

295

296

BrookingsPapers on EconomicActivity,2:1995

manyeconomic forecastersembracethe textbookview. For example, a recent Wall Street research report explains that "as long as asset prices arerising, the risk of a significantdropin consumerspendingis small," and notes that the rising numberof stockholdershas "made real economic activity moretied to the performanceof financialassets thanever before."2

This paperdescribesthe changingpatternof stock ownershipduring the last three decades, investigates whetherchanging ownershippatterns have, in fact, altered the links between stock values and consumption,andexploresthe "wealtheffect" of stockpricefluctuations. At the outset it is importantto recognize thatan increasein consumer spendingfollowing a rise in shareprices could be attributableto either of two factors. First, stock prices may rise in anticipationof strong economic activity, including consumer spending. The role of share prices as a leading indicatoris well documented.In this case changes in stock marketvalues are not a source of subsequentchanges in consumption, but merely an indicator that subsequentchanges are expected. A second, and not necessarily exclusive, link between stock prices and consumptionis the wealth effect; that is, changes in share values cause changes in consumptionby relaxing the resource constraintsthathouseholdsface. Overlong horizons, theremustbe such a wealth effect; we consider whetherthere is also an importantwealth effect on consumptionat horizonsof one to fourquarters.It is difficult to distinguishbetweenthe leadingindicatorandcausativeviews of the relationshipbetweensharepricesandconsumptionbecausethisrequires identifyingautonomousmovementsin shareprices thatare not attributableto changingexpectationsof futuredividendsor interestrates.

This paperpresentsnew evidence on the associationbetween share price movementsandconsumption.It summarizesthe changesin consumptionthathave typicallyfollowed substantialchangesin stockmarket values, andpresentsseveraltests directedat disentanglingthe leading indicatorand wealth effect views. If the leading indicatorview is correct,the patternof consumptionchangesfollowing stockpricefluctuationsshould be independentof the distributionof stock ownership andthereis no reasonto expect differentconsumptionresponsesfrom householdsthatdo anddo not own corporatestock. This paperpresents

2. Shulman,Usem, and Brown(1995).

JamesM. Poterba and AndrewA. Samwick

297

empiricaltests of both of these propositions.It finds little evidence of an importantwealth effect of shareprices on consumption.The strong positive correlationbetweenconsumptiongrowthandlagged stock market returns,therefore, appearsto be primarilydue to the leading indicatorfeatureof stock price movements.

We begin by placing the 1995 stock marketincreasein context. We reporton the evolution of price-to-dividendand price-to-earningsratios, Tobin's q, andthe ratioof stock marketvalue to GDP duringthe post-World War II period. Some of these measures, notably Tobin's q, suggest thatthe stock marketof 1995 is at a postwarvaluationhigh. Others, notablythe price-to-earningsratio, suggest a less extremesituation.

We next investigate the fractionof stock marketcapital gains that accruedirectlyto individualinvestors, in contrastto gains thataccrue to them indirectly, through financial intermediariessuch as defined benefit pension plans or life insurancecompanies. A range of recent behavioralmodelsof consumptionsuggestthatthe marginalpropensity to spendout of differenttypes of assets dependsnot only on theirrisk andreturncharacteristics,but also on the way in which they areheld.3 Households may exhibit lower marginalpropensitiesto spend out of capitalgains on assets that are held in retirementplans thanon assets thatareheld directly.

Populardiscussions sometimes note that the fraction of corporate stockowned by householdshas declinedduringthe postwarperiodand is currentlyless than 50 percent.4In fact, the principalpostwartrend has been away from direct individualstock ownershipand towardindirectownershipthroughvarious financialintermediaries.This paper reanalyzesthe widely cited Flow of Fundsdataof the FederalReserve Boardthatshow householdsowning less than50 percentof outstanding shares. Combining individual ownership of equities throughmutual funds, definedcontributionpensionplans, andotherfinancialintermediaries, it is apparentthat individualshave direct controlover nearly two-thirdsof outstandingcorporatestock.

To describethe changingincidenceof stock ownershipwe examine

3. Thaler(1994) providesa summaryof this literature. 4. See, for example, "IndividualsLose MarketShare," New YorkTimes,July 18, 1995, p. D21, and "Small InvestorContinuesto Give Up Controlof Stocks," Wall StreetJournal, May 11, 1992, p. C1.

298

BrookingsPapers on EconomicActivity,2:1995

datafromthe 1962 Surveyof FinancialCharacteristicsof Consumers, and the 1983 and 1992 Surveys of ConsumerFinances. While share ownershippatternschanged relatively little between 1962 and 1983, therehas been a substantialincreasein the prevalenceof shareownership duringthe last decade. This growthis the resultof rising ratesof indirect share ownership. The fraction of stock held by the largest stockholders,those in the top 0.5 percentof the distributionof equity investors, has also declined duringthis period.

To develop evidence that can distinguishbetween the leading indicatorand the wealth effect views of how shareprices affect consumption, the paperexplores the correlationbetween stock returnsand the compositionof consumerspending.If thereis a wealtheffect, positive stock returnsshould increase the shareof consumptionaccountedfor by luxurygoods. We consider aggregatedataon several categoriesof consumption that are disproportionatelypurchasedby high-income households,including"upperluxury"vehicles, andfindlittleevidence thatluxuryspendingrises in the wake of rising stock prices.

The paperconsiders whetherchangingpatternsof stock ownership affectthe linkagesbetweenconsumptionandstockmarketfluctuations. The leadingindicatorview suggests thatownershippatternsshouldnot affect this relationship,while the wealtheffect at least admitsthe possibility. We explore the effect of changes in stock prices, as well as changesin thedividend-to-priceandearnings-to-priceratios,on various measuresof consumption.5We recognize that stock prices and consumptionarejointly determined,andsimply tryto describethe typical patternof economic activity following substantialstock price movements. Ourresultssuggest thatchangesin stock priceshave significant predictivepower for futureconsumptionspending. A permanentstock price rise of 17 percent(roughlythe same magnitudeas the price increase in the first six months of 1995) forecasts an increase of about 1.1 percent in consumption in mid-1996, relative to what it would otherwisehave been. Increasesin consumerspendingon new automobiles andotherdurableswould be particularlylarge.

We findlittle evidence to suggestthatthe shiftfromdirectto indirect ownershipof corporatestock has alteredthe link between stock price

5. Fama(1981), FischerandMerton(1984), andBarro(1990) alsoestimatereducedform equationsmeasuringthe predictivepower of stock price movementsfor various macroeconomicaggregates.

James M. Poterba and AndrewA. Samwick

299

fluctuationsandconsumptionspending, andmoregenerally, little evidence of an importantwealth effect on consumption.Since the timeseries variationin the patternof corporatestock ownershipyields tests with low statisticalpower, we also use householdsurveydatafromthe Panel Study of Income Dynamics to comparethe correlationbetween stock marketfluctuationsand growthin consumptionfor stockholders with directand indirectholdings. We findsome evidence thatthe consumptionof individualswho hold stocks throughthriftplans, such as 401(k)s, 403(b)s, and ESOPs, is more sensitive to stock price movements than the consumptionof those who do not hold any stock; but once again, the available tests have low power.

We do not find any evidence that the effect of shareprices on consumption depends on the source of stock price movements. This is somewhatsurprising,given the substantialbodyof researchin financial economics suggesting that price fluctuationsthat change the value of the dividend-to-priceor earnings-to-priceratiosareoften reversedover a periodof several years.

Recent Stock Market Fluctuations in Perspective

The stock markethas climbed to recordheights in the last year. In the six monthsafterthe Dow JonesIndustrialAveragefirstreachedthe historicfour thousandlevel on February23, 1995, it climbed another seven hundredpoints. And beforethe end of 1995, the Dow index had closed at well above five thousand. Between January1 and June 30, 1995, the Standardand Poor's 500-stock index (S&P 500) rose by nearly 17 percent.6Althoughthe news mediahave depictedthe recent bull marketas unprecedented,recent returnsare not extraordinary.In twentyof the sixty-eight years between 1926 and 1993, the real return on stocks of large corporationsexceeded 20 percent. In five of those years the real returnexceeded 40 percent.7

To providebackgroundfor analyzingthe aggregateeffects of stock pricemovements,table 1 presentsseveralsummarystatistics.The first column shows the real value of the S&P 500 in units comparableto its

6. We calculate this as ln(539.4/455.2) = 0. 1697, which we approximateas 17 percent.

7. These statisticsarebasedon IbbotsonAssociates(1994).

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download