A Portfolio View of the U.S. Current Account Deficit Author(s): Jaume Ventura Source: Brookings Papers on Economic Activity, Vol. 2001, No. 1 (2001), pp. 241-253 Published by: The Brookings Institution Stable URL: http://www.jstor.org/stable/1209167 Accessed: 19/08/2010 23:17 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=brookings. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The Brookings Institution is collaborating with JSTOR to digitize, preserve and extend access to Brookings Papers on Economic Activity.

http://www.jstor.org

JAUME VENTURA MassachusettsInstituteof Technology

A Portfolio Viewof the U.S. CurrentAccount Deficit currentaccountbalanceas a shareof U.S. GNP averagedroughlyzero.' Startingin 1983, however,the UnitedStates experiencedincreasinglylarge currentaccount deficits, which reached 3.3 percentand 3.4 percentof GNP in 1986 and 1987, respectively.This tendencytowardlargerdeficits was reversedgraduallyduringthe rest of the decade,andby 1991the currentaccountwas nearzero again.But starting in 1993 the currentaccountagain beganto recordincreasinglylarge deficits, which grew to 3.6 percent of GNP in 1999 and 4.4 percent in 2000. This historyof the currentaccountpromptsseveralquestions:What is the sourceof the large currentaccountdeficits of the 1990s?Are they likely to remainwithus indefinitely?If not, shouldwe expectthemto fade awayslowly as theydid in the 1980s?Orshouldwe expectinsteada sharp reversalin the nearfuture? In this essay I interpretthese trendsin the U.S. currentaccountfrom a perspectivethat focuses on the behaviorof the countryportfolio. The countryportfoliois definedas the sum of all productiveassets locatedin the United States, plus the U.S. net foreign asset position (that is, the sum of all claims on foreign assets held by U.S. residentsminusthe sum of all claims on U.S. assets ownedby foreignresidents).By the composi-

FROM 1971 TO 1982 THE U.S.

I am gratefulto RudigerDornbusch,AartKraay,and KennethRogoff for useful conversations.I am also thankfulto Pol Antrasfor superbresearchassistance. 1. The appendixdescribesthe datasourcesused in this paper. 241

242

BrookingsPapers on EconomicActivity,1:2001

tion of the U.S. portfolioI meanthe shareof the net foreignassetposition in it.2Accordingto the portfolioview, it is useful to separatechangesin the currentaccountinto two components:changesin the size of the country portfolio, which I call portfolio growth effects, and changes in the compositionof the countryportfolio,or portfoliorebalancingeffects. A simple application of this approachreveals a clear picture: the recent currentaccountdeficits are mostly the manifestationof the spectacular increase in U.S. wealth experiencedin the 1990s. Contraryto a widely held belief, these deficitsdo not reflecta rebalancingof portfoliostoward U.S. assets and away fromforeignassets. A naturalquestionfollows: Why did U.S. wealth increaseso much in the 1990s? I explore two alternativehypotheses. The first views the increasein wealthas reflectinga rapidaccumulationof an intangibleform of capital.The secondis based on the notionthatthe 1990s were characterizedby the appearanceand growthof a bubblein the U.S. stock market. Althoughboth explanationsexhibit interestingelements, neitheris fully satisfactory.Ourinabilityto accountfor the growthin wealthmakes the taskof predictingthe futuredirectionof the U.S. currentaccountquite difficult,if not impossible. Nevertheless,each of these stories has a differentending,andI discussthembelow.3

2. A simpleexamplehelps in understanding the implicationsof this definition.Suppose thatDaimlerbuysChryslerandpays Chryslershareholdersin cash, whichthey thenuse to builda new hotel in Las Vegas.This transactiondoes not affect U.S. net worthor the size of the U.S. portfolio. But it does change its composition,accordingto my definition.In particular,thereis an increasein U.S. productiveassets (since Chrysler'sfacilitiesare still in the UnitedStates,and so is the hotel), which is financedby selling claims on U.S. productive assets (since the United Statesmust now pay the Germanownersof Daimlerthe returnto the Chryslerfacilities).In my view,thefactthatthe claimthatthe UnitedStateshas sold is a contingentone does not invalidatethe propositionthatthe UnitedStateshas leverageditself in orderto buymoreU.S. productiveassets.If the Chryslershareholders hadused the cash to buy Germanpublicbondsinsteadof a Las Vegashotel, neitherthe size nor the compositionof the U.S. portfoliowouldhavechanged,in my definition.Therewouldhave been an increase in claims on foreign assets held by the United States (since Germany mustnow use partof its productiveassetsto pay interestto U.S. bondholders),financedby selling claimson U.S. productiveassets. 3. A third story is provided by McGrattanand Prescott (2001), who argue that the increasein wealthis dueto changesin taxesandregulationsand,in particular, to a reduction in the dividendtax.

Jaume Ventura

243

A Portfolio View of the Current Account The point of departurefor a portfolio view of the currentaccount is the celebrated mean-variancetheory of Harry Markowitzand James Tobin.4Accordingto this theory,investorschoose theirportfoliosby optimally tradingoff risk andreturn.The optimalor mean-variance-efficient portfoliocontainsthe risk-freeasset and an optimalcombinationof risky assets (OCRA).A strongresultof the theoryis thatthe OCRAis the same for all investors with access to the same menu of assets, regardlessof their attitudestowardrisk and theirlevel of wealth.Thatis, the shareof each asset in the OCRAdependsonly on the distributionof asset returns. Anotherstrongresultof the theoryis thatthe weightsthatmean-variance investorsassignto the risk-freeasset andthe OCRAdependonly on their risk aversionandthe distributionof asset returns.They do not dependon the investors'wealth.5 Moving from the optimalinvestorportfolioto the averageor country portfoliorequiresan additionalassumption,namely,thatthe averagerisk aversion and the distributionof asset returnsare both independentof wealth.This is a strongassumption,and its validityis an empiricalissue thatis farfromsettled.I shallneverthelessadoptit. This assumptionis useful herebecauseit ensuresthatthe propertiesof individualinvestors'portfolios also apply to the average or countryportfolio. Thus, changes in wealthaffectonly the size of the countryportfoliobutdo not influenceits compositionor asset shares.The latterchangesonly with changesin risk aversionor in the distributionof assetreturns. This portfolio view leads to a sharp and simple rule to predict the responseof the currentaccountto changesin wealth.Define W,andNFA, as the wealthandthe net foreignassetpositionof the country,respectively. Let Xt be the share of net foreign assets in the country portfolio, Xt = SinceXtis not affectedby changesin wealth,the fractionof any NFAt/Wt. changein wealththatis allocatedto net foreignassets equalsthe shareof foreignassets in the countryportfolio: (1)

ANFA, = XAW,.

4. Markowitz(1952);Tobin(1958). 5. Theseresultsapplyin moregeneralmodelsif preferencesarehomotheticandreturns arelog-normallydistributed.See, for instance,Merton(1971).

244

BrookingsPapers on EconomicActivity,1:2001

Or, alternatively,the countryinvests the marginalunit of wealth in the same way as the average unit. This is a simple rule for predicting the effects of changes in wealth on the net foreign asset position, in the absenceof changesin the distributionof assetreturns.Thatis, equation1 measuresthe extentto which a changein the net foreignasset positionis a manifestationof changes in the size of the countryportfolio, or portfolio growth.

A useful approximationto equation 1 applies if asset price revaluations are not too large. Let St and CA,denote gross nationalsaving and the currentaccount,respectively.If asset price revaluations(which are includedin AW,andANFA,but not in St and CA,) arenot too large,gross national saving and the currentaccount are good measures of actual changesin wealthandnet foreignassets.As a result,we can approximate equation1 as follows: (2)

CA, X,S,.

This relationshipshould hold in samples of countrieswhere there is substantialcross-sectionaland time variationin saving ratesbut the distributionof asset returnsis quite stable over time. Perhapssurprisingly, AartKraayandI foundthatthis is the case in a sampleof thirteenindustrialcountriesfrom 1973 to 1995.6 Since the shareof foreignassetsin the portfolios of these countries is typically small, this simple rule is also consistent with the celebratedfinding of MartinFeldstein and Charles Horiokathat saving and investmentmove almost one to one in a cross sectionof countries.7 Althoughthe theoreticalfoundationsof this portfolioview of the currentaccountare quite standard,its implicationsare somewhatsurprising and even counterintuitivecomparedwith those of existing theories. To see this, considerthe effects of an increasein saving due to, say, a productionboom, diminishedexpectationsaboutthe future,a reductionin taxes, or an increasein populationgrowth.The traditionalview is thatat least partof this additionalsaving shouldbe investedabroad,leadingto an increasein the currentaccountsurplus.Instead,the new view embed6. Kraay and Ventura(2000). The sample consists of Australia, Austria, Canada, Finland, France, Germany, Italy, Japan, the Netherlands, Spain, Sweden, the United Kingdom, and the United States. 7. Feldstein and Horioka (1980).

JaumeVentura

245

ded in equation2 suggeststhatthis saving shouldbe investedin the same proportionsas in the existing countryportfolio,leadingto an increasein the currentaccountsurplusin creditorcountries(X,> 0) anda decreasein debtorcountries(X,< 0). Equation2 describesthe responseof the currentaccountto movements in saving, but these are not the only source of variationin the current account. Changesin the distributionof asset returnsconstituteanother importantsourceof currentaccountmovements.To see this, consideran increasein the expectedreturnto domesticcapitaldue to, say, a technological breakthrough,a change in political leadership,or a reductionin capital income taxes. Both the traditionaland the portfolioviews of the currentaccountwould predictthatinvestorswill reactto this changeby rebalancingtheirportfoliostowarddomestic capitaland away from foreign assets. Since equation2 describes the currentaccount surplusthat keeps the share of foreign assets in the countryportfolio constant,this rebalancingcanonly be achievedby runninga smallercurrentaccountsurplus thanpredictedby this equation.Thereforedeviations between the actual currentaccount and the currentaccountpredictedby equation2 can be interpretedas a manifestationof changesin the compositionof the country portfolio, or portfolio rebalancing.

Portfolio Growth or Portfolio Rebalancing? A common view is that the large U.S. currentaccountdeficits of the 1990sreflecteda favorableshiftin the distributionof returnsto U.S. assets relativeto foreign assets. This shift is attributedto variouscauses. Some arguethatincreasedtotal factorproductivity(TFP)growthhas raisedthe expected returnto U.S. capital. Others argue that financial turmoil in emergingmarketshas maderelativelysafe U.S. assets look more attractive. Whetherit is productivitygrowth or the increasedneed for a safe haven,thereis a growingperceptionthatthe recentbehaviorof the U.S. currentaccountreflectsmostlyportfoliorebalancing. A straightforward applicationof equation2 seems to confirmthis perception.Figure1 plots the actualcurrentaccountandthe currentaccount predictedby equation2. Over the last thirty years there have been two episodes in which the predictedcurrentaccountsurplusgrossly overestimates the actualone. The first episode is centeredin the mid-1980s, and

246

BrookingsPapers on EconomicActivity,1:2001

Figure1. ActualandPredictedU.S.CurrentAccountBalance,1970-99 Percent of GNP ~~~~~~Predicted'

1 0

L 0-m

St-

=~~~ ~ -

_--'

~-

-

Actual

-1 -2 -3 -4

1975

1980

1985

1990

1995

Source: Author'scalculationsbased on datafrom Bureauof EconomicAnalysis, NationalIncome and ProductAccounts. a. Using equation2.

the secondepisodestarteda bit beforethe mid-1990s andstill has not concluded.The 1980s episode is not a surpriseat all. We can easily attribute it to the high U.S. interestratesthatresultedfromcombiningtightmonetary policy with large fiscal deficits. The internationaldebt crisis that eruptedin 1982 mustalso have contributedto increaseddemandfor U.S. assetsin this period.One is temptedto concludethatthe 1990s episodeis nothingbut a repetitionof thatof the 1980s. Insteadof high U.S. interest rates,in the 1990s we hadrapidTFP growthin the UnitedStates.Instead of a developingcountrydebt crisis, in the 1990s we had a flurryof currency crises in emerging markets.The parallels are too obvious to be missed. But this sensationof deja vu is just an illusion. Rememberthatequation 2 was shownto be a valid approximationto the theoryif and only if asset price revaluationsare not too large. This is certainly not a good assumptionfor the 1990s episode. From the end of 1992 to the end of 1999, cumulativegross national saving was $8.7 trillion, whereas the increasein the marketvalueof the U.S. capitalstockwas roughly$40 trillion. Thatis, gross nationalsaving capturedslightly over 20 percentof the actualincreasein wealth.It follows thatequation2 is underpredicting

Jaume Ventura

247

the portfolio growth component of the currentaccount by a factor of almostfive. Nothingof the sorthappenedduringthe 1980s.8 A naturalway to correctfor this disconnectbetweensavingandwealth changes is to go back to equation 1. Figure 2 plots the actual change in net foreignassets and the changein net foreignassets predictedby equation 1. The 1980s episodeof portfoliorebalancingis clearlyvisible. From 1980 to 1992 the changein net foreignassets was consistentlybelow the level requiredto keep the shareof foreign assets constant.But the 1990s episode has all but disappeared.In fact, in 1998 and 1999 thereseems to have been a rebalancingof portfolios away from U.S. assets. Figure 3 shows thatthe shareof foreignassets declinedby roughly4.8 percentage points from 1980 to 1992, but declinedby less than0.5 percentagepoint from 1992 to 1999. The picturethatemergesfromthe portfolioview of the currentaccount is slowly cominginto focus. In the 1990sU.S. investorsenjoyedverylarge returnsto theirwealthin the formof asset price revaluation.Ratherthan spendthese returns,U.S. investorslargelydecidedto keep them andbuy domesticandforeignassetsroughlyin the sameproportionsas theiraverage portfolio. Since the averageportfolio is shortin foreign assets, this meansthatthe UnitedStatesleverageditself more, so thatit could invest in domesticassets beyondthe increasein wealth.Hence the largecurrent accountdeficits. At best,thispicturecanbe interpretedas a partialexplanation.At worst, we can thinkof it as posing a set of still-unansweredquestions:Why were the returnsto U.S. wealthin the 1990s so high?Why did thesereturnstake the formof asset price revaluationratherthanincreasedproduction?(Or, ratioincreaseso much?)Why in otherwords,why did the wealth-to-output did U.S. investors save most of these returnsratherthan use them to increasetheirconsumption?Why did they choose not to rebalancetheir portfoliostowardU.S. assets at a time when the latterwere yielding such high returns?I do not claim to have foolproofanswersto these questions. Nobodyreallydoes. But I am willing to speculate.

8. Rememberthattheportfolioview of the currentaccountexplainshow a largeincrease in savingcanleadto a largecurrentaccountdeficitin a debtorcountrylike the UnitedStates. This is a majordifferencewith the traditionalview, whichpredictsthatan increasein saving of this magnitudeshould have generatedlarge currentaccountsurpluses.See Kraay andVentura(2000).

248

BrookingsPapers on EconomicActivity,1:2001

Figure 2. Actual and PredictedChangein U.S. Net ForeignAssets, 1970-99 Percentof GNP 6 4 a Predicted

2 0 -2

Actual

A4

1975

1980

1985

1990

1995

Source: Author'scalculationsbased on data from Bureauof Economic Analysis, National Income and ProductAccounts, and Herman(2000). a. Using equation 1.

Figure 3. Share of U.S. Net ForeignAssets in the CountryPortfolio,1970-99 Percentof nationalwealtha 3 2

0

-2 -3

1975

1980

1985

1990

1995

Source: Author'scalculations based on data from Bureauof Economic Analysis, National Income and ProductAccounts, and Herman(2000). a. National wealth is the value of the domestic capital stock (adjustedfor depreciationand price revaluation)plus the net internationalinvestmentposition.

Jaume Ventura

249

AccumulatingE-Capital A first explanationfor the increasein the wealth-to-outputratiorelies on improvedexpectationsof the futureof the U.S. economy.According to this view, the increase in asset prices measuresupwardrevisions of futureproductionbased on economic fundamentals.One might immediately counterthat the measuredcapital stock (and hence the productive capacityof the U.S. economy) has increasedat roughlythe same rateas productionandat a muchslowerpace thanwealth.Nevertheless,it is also possible that in the 1990s the United States accumulatedintangible or organizationalcapital.This type of capitalmightnot increaseproduction immediatelybut can be expected to raise the productivecapacityof the U.S. economy in the near future.RobertHall has forcefully arguedthis view and has coined the terme-capital to referto this form of intangible asset.9 One difficulty with this story is linked to the behaviorof saving and interestrates.If the futurelooks so rosy,Americansshouldstartconsuming more rightnow. Since the additionalgoods to be consumedwill not come until later, however, attemptsto consume right now should drive interestratesup, loweringasset prices and curbingthe growthin wealth. The accumulationof e-capital can affect wealth if and only if investors are willing to save more at given interestrates.The e-capital story must thereforebe complementedby an explanationof the factorsthatled to an increasein saving duringthe 1990s.10 One could arguethatthe increase in savingis due to the arrivalof baby-boomersat thatstageof the life cycle (latethirtiesandearlyforties)whenpeopletypicallystartsavingfor retirement. One could also point out that much of the increase in wealth has gone into the handsof rich investors,who are likely to have a lower than averagepropensityto consume out of their wealth.11Or it could be that

9. Hall (2000). 10. It is a well-publicized fact that U.S. household saving has declined to the point that it is now negative. This should not, however, obscure the fact that there has been an increase in total saving (by a couple of percentage points of GDP) during the 1990s, as increases in corporate and government saving more than offset the decline in household saving. But these measures of saving do not take into account the income that comes from asset price revaluations. When this is done, one cannot escape the conclusion that saving increased substantially during the 1990s. 11. I thank Kenneth Rogoff for pointing this out to me.

250

BrookingsPapers on EconomicActivity,1:2001

habitpersistenceis quitestrongandthatit is rationalfor investorsto raise theirconsumptionvery slowly.In any case, whethersome combinationof these factorscan explainthe increasein savingin the 1990s remainsto be determined. A seconddifficultywith the e-capitalstoryis how to reconcilethe view thatinvestmentsin e-capitalare highly productivewith the absence of a strongrebalancingof the U.S. countryportfolio.If e-capitalis so productive, the expectedreturnto U.S. capitalshouldhave increased,convincing investorsto rebalancetheir portfolios towardU.S. assets and away from foreign assets. As figure 2 shows, this has not happened,and the e-capitalstorymust come to grips with this observation.Thereare various ways to do this. One possible argumentis thatthereare strongdiminishing returnsto e-capital. In this view the first wave of investmentsin e-capitalyielded rich rewards.The second wave is unlikelyto yield such high returns,and as a result thereis no incentiveto rebalanceportfolios towardU.S. capital. A second possible argumentwould recognize that e-capitalhas indeedraisedthe expectedreturnto U.S. capital,but would then point out that it has also increased the expected returnto foreign capital.This suggests, too, there is no incentive to rebalanceportfolios towardU.S. capital. Subject to these caveats, the e-capital story provides a consistent accountof the main macroeconomicevents of the 1990s. If the story is correct,the futureof the currentaccountis intimatelylinked to the patternof savingin the UnitedStates andthe time it takesfor productionto increase.If the factorsbehindthe increasein savingremainin force until expectationsof increasedfutureproductionarerealized,we shouldexpect a continuationof the currentpatternof high savingandcorrespondingcurrent accountdeficits. If insteadthe factorsbehindthe increasein saving weaken before the expectationsof increasedfutureproductionare realized, we shouldexpectan increasein interestrates,a contractionin wealth, andcurrentaccountsurpluses. At the end of the day, however,the most damagingevidence against the e-capitalstoryis the recentdecline in the U.S. stock market.It is difficult to justify this decline by citing the increase in interest rates of around 1 percentage point (which, in any case, has been recently reversed).It is still more difficultto justify this increasein interestrates by a decline in aggregatesaving. Still, one could argue that the recent decline in stock prices reflects a large negative revision of the value of

Jaume Ventura

251

e-capitalbased on news abouteconomic fundamentals.But I would not want to be in the position of having to explain what this news and these fundamentalsmightbe.

The Dot-Com Bubble A second explanationof the increase in the wealth-to-outputratio is based on the idea thatthe 1990s witnessedthe appearanceof a bubbleor Ponzi schemein the U.S. stockmarket.In such an environment,investors buy stocksbecausetheyexpectto resell themlaterat a higherprice,rather thanbecausethey have revisedupwardtheirexpectationsof firms'profits. Thepriceappreciationmustbe highenoughto compensatefor the possibilityof not findinga buyer.In otherwords,the highertheriskof a crash, the higheris the growthrateof stock prices. Duringthis episode the link between changes in asset prices and those of theirfundamentalvalue is broken. Eventually,buyers are no longer found and the bubble bursts. Since this bubblehas beenmoreevidentfor high-technologyandInternetrelatedfirms,I referto it as the dot-combubble. At first sight, the notion of a bubbleunderlyingthe fast growthof the 1990s might seem counterintuitiveto economists. In existing models,12 bubbles(or unproductiveassets)provideinvestorswith an alternativesavings vehiclethatcompeteswithproductivecapital.Thesemodelstherefore predictthatthe appearanceof a bubbleshouldbe associatedwith a reduction in the stockof capitalandproduction.This descriptionstandsin stark contrastwiththe experienceof the 1990s,whereboththe capitalstockand productionincreasedat a rapidpace. If we wantto attributethe developments of the 1990s to the appearanceof a bubble,we must first explain how this bubblecan fostercapitalaccumulationratherthanhinderit, as it does in existingmodels. A key assumptionof the classic models of bubblesis thatinvestorsare risk-neutraland,consequently,thatbubblesmustoffer the sameexpected returnas productivecapital.Since thereis a probabilitythata bubblewill burst,this returnmustexceed thatof productivecapitalfor as long as the bubbledoes not burst.The returnon the bubblemustalso exceedthereturn on the investor'soverallportfolio(which is a combinationof bubbleand 12. Such as that of Tirole (1985).

252

BrookingsPapers on EconomicActivity,1:2001

productivecapital).Since this returncomes in the formof price appreciation, the bubblegrows continuouslyand crowdsout productivecapitalin the investor'sportfolio.Investorsare willing to acceptthis changein the compositionof theirportfoliosbecausethey arerisk-neutralandperceive the bubbleandproductivecapitalas perfectsubstitutes. Assumeinsteadthatinvestorsarerisk-averseandchoose to hold meanvariance-efficientportfolios.Rememberthata key characteristicof these portfoliosis thatasset sharesare independentof wealth.This has importantimplicationsfor the relationshipbetweenthe bubbleandproductive capital. As the bubble grows, so does the wealth of the investor.This inducesinvestorsto buymoreproductivecapitalso as to keepthe sharesof their portfolios constant. This means more productiveinvestmentand higher growth in the stock of capital and output. In a world of meanvarianceinvestors,bubblesandthe stockof capitalarecomplementsrather thansubstitutes.In such a world,the appearanceof a bubblecan generate a boom in productiveinvestmentandoutput."3 This dot-com bubble story easily gets aroundthe problems of the e-capital story. It can explain why U.S. investors saved most of their increasein wealth,andwhy they decidednot to rebalancetheirportfolios towardU.S. assets. Both choices are nothingbut naturalreactionsto the increasedriskgeneratedby thepossibilityof the bubblebursting.Highrisk encouragesinvestorsto save as a precautionarymeasureand can therefore explainthe shift in savingbehavior.High risk also inducesinvestors to requirehigherratesof returnon U.S. assets and can thereforeexplain why therehas not been a rebalancingof investorportfoliostowardU.S. assets. The dot-combubblestoryis harderto rule out thanthe e-capitalstory. But I do not regardthis as a merit.To the contrary,it mostly reflectshow vague the theory still is regardingits implications.For instance, should we expect the whole bubbleto burstin a single installment,or to deflate gradually?As the bubblebursts,whatwill happento the marketvalue of productivefirms?Whatsortsof eventswouldtriggerthebirthandthe death 13. I have recently formalized this model of "expansionary"bubbles; see Ventura(2001). Caballero and Hammour (2001) have simultaneously developed two alternative models of "expansionary" bubbles. In their first model, the bubble arises in the stock of capital itself, and hence productive investment is pulled upward by the bubble. In their second model, the complementarity between the bubble and conventional capital arises from externalities in production.

Jaume Ventura

253

of a bubble?To be honest, we simply know very little aboutthe answers to these questions. to predictthe Despitethis ignorance,it is still relativelystraightforward effects of a burstingof the bubbleon the currentaccount.Rememberthat this event would generate a reductionin aggregatewealth and saving. Since the United Statesis a debtorcountry,this would in turngeneratea reductionin the currentaccountdeficit. If the burstis quick andviolent, the United Statesmightexperiencea sharpreversal,in which the current accountgoes into surplus.If the burstis slow andprotracted,the current accountdeficit will simply decline andremainclose to zero. In the aftermathI wouldexpectthe currentaccountto registermoderatedeficitsas the growthrateof wealthreturnsto prebubblelevels.

APPENDIX

Data Sources I OBTAINED DATA for the U.S. currentaccount,internationalinvestment position (or net foreign asset position), gross nationalproduct,andgross domesticinvestmentfromthe WorldWideWebsite of the Bureauof Economic Analysis (BEA). I computedgross nationalsaving as the current accountplus gross domesticinvestment. To obtainestimatesof U.S. wealth,I took an initialvalue for the capital stockin 1950.14This sourceis also availableon the BEA website.This initial stock is dividedinto threecomponents:privatenonresidential,private residential,and government.I then cumulatedflows of investment, assuminga rateof depreciationof 4 percentandrevaluatingexistingstocks using the appropriateinvestmentdeflatorfor privateresidentialandgovernmentcomponents,butusing a sharepriceindex for the privatenonresidentialcomponent.Thisproceduregenerateda seriesfor the U.S. stockof capital.ThenU.S. wealthis obtainedas the sum of the domesticstock of capitalandthe internationalinvestmentposition.

14. The sourceof this initialvalueis Herman(2000).