American Economic Association

Foreign Direct Investment and the Domestic Capital Stock Author(s): Mihir A. Desai, C. Fritz Foley and James R. Hines Jr. Reviewed work(s): Source: The American Economic Review, Vol. 95, No. 2, Papers and Proceedings of the One Hundred Seventeenth Annual Meeting of the American Economic Association, Philadelphia, PA, January 7-9, 2005 (May, 2005), pp. 33-38 Published by: American Economic Association Stable URL: http://www.jstor.org/stable/4132786 . Accessed: 20/12/2012 14:08
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
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 support@jstor.org.

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review.

http://www.jstor.org

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions

ForeignDirectInvestmentand the Domestic CapitalStock

A.

DESAI, C. FRITZ FOLEY, AND JAMES R. HINES JR.*

Rising levels of foreign direct investment (FDI) concerngrowingnumbersof policymakers and members of the American public. These concerns stem from the perception that foreign activities of American multinational corporations reduce employment and other economic activities within the United States. While investment flows within the United States go largely unnoticed,in an internationalsetting the lexicon of "winners"and "losers" can be inescouncapable. Curiously,both capital-exporting tries and capital-importingcountries have at times expressed concern over the consequences of internationalcapital flows. Capital-exporting countriesworry that too much of their capital goes abroad, while capital-importingcountries fear foreign control of domestic assets and the possible macroeconomic instability associated with rapidchanges in foreign investment levels. The concerns of capital-exporting countries, while diffuse, often are based on conceptions of outbound FDI as diverting economic activity. Unsurprisingly, growing overseas activities of multinational firms have become a source of economic insecurityfor workers,managers,and tax collectors (see e.g., Kenneth F. Scheve and Matthew J. Slaughter, 2001). Concerns over the economic impact of rising FDI have limited analytic and empirical support. The paucity of analysis reflects the nascent nature of FDI theories and the difficulty, until recently, of analyzing the internal dynamics of multinational firms whose activities span borders. Given the rapidly rising scope of multinational activity, the absence of a readily available framework with which to analyze FDI is increasingly costly.1 While the ratio of outbound
* Desai and Foley: HarvardBusiness School, Soldiers Field Road, Boston, MA 02163 (e-mail: mdesai@hbs.edu, ffoley@hbs.edu); Hines: Office of Tax Policy Research, Ross School of Business, University of Michigan, 701 Tappan Street, Ann Arbor, MI 48109 (e-mail: jrhines@umich. edu). Desai and Foley thank the Division of Research at HarvardBusiness School for financial support. 1 David L. Carret al. (2001), Gene Grossman Elhanan and a (2003) represent new, and Helpman(2002), and Pol Antrais firms. promising,generationof theoriesof multinational 33

FDI to private nonresidential fixed investment in the United States averaged 6 percent through the 1960s and 1970s, this ratio has risen to 15 percent by the 2000s. This dramaticincrease in outbound FDI is matched by a similarly dramatic increase in FDI flows into the United States. This paperevaluates evidence of the relationship between outbound FDI and levels of domestic capital formation. Earlier findings reported by Martin S. Feldstein (1995), using data from the 1970s and 1980s, indicate that outbound FDI reduces total domestic investment in the United States roughly dollar-fordollar, whereas inboundFDI contributesto total domestic investment by the same magnitude. Evidence in this paper from a much broader sample of countries for the 1980s and 1990s confirms that the aggregate patterns identified by Feldstein persist in this larger, and more recent, sample. However, an entirely different picture emerges when attention is confined to multinationalfirms. Higher levels of capital expenditures by their own foreign affiliates are associated with greater levels of domestic investment by American multinational firms, suggesting that foreign and domestic investment are complements rather than substitutes. This pattern is consistent with recent firmlevel evidence reportedin Desai et al. (2004b), which uses foreign economic growth as an instrument for foreign investment by multinational firms. Therefore, a growing body of evidence offers no support for the simple, and common, perception that foreign investment diverts resources from domestic investment. This evidence also raises several new questions for further research. Firms I. ForeignInvestmentand Multinational The common intuition that outboundFDI reduces domestic investment is a special case of a broaderset of possible effects of FDI on domestic economic activity. Consider, for example, a multinationalfirm producing worldwide output with the function Q(K, K*, 0), in which K is

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions

34

AEA PAPERS AND PROCEEDINGS

MAY2005

domestic capital, K* is foreign capital, and 0 is a vector including prices and other marketconditions relevant to output. The first-ordercondition corresponding to profit-maximizing levels of domestic investment is &Q(K, K*, 0) (1) = A

in which A is the firm's cost of capital.2 From equation (1) it is clear that foreign and domestic investment can be related either through the production process, if a2Q(K,K*, O)/aKaK*is nonzero, or through the cost of capital, if A is somehow a function of K*. The common intuition of diversion corresponds to a setting in which firm resources are fixed, so that a dollar invested abroad corresponds to one less dollar that can be invested domestically. In the notation of equation (1), this corresponds to a A(K + K*) function that exhibits a discreteupwardjump at the point that investment resources are exhausted. However, multinationals finance investment projects on world markets and make extensive use of their internal capital markets. Desai et al. (2004d) presents evidence that affiliates borrow extensively from local sources and opportunistically structure their internal capital markets in response to varying costs of external finance and tax factors. Additionally, Desai et al. (2004a) find that parents provide affiliates with additional equity to finance investment in the wake of severe currency depreciations. If financial resources are not fixed, then the primarysource of interaction between foreign and domestic investment comes from the productionprocess. Existing theories offer alternative, and contradictory, intuitions for the likely relationshipbetween home and foreign capital in the firm's derived demand function (i.e., the sign of
a2Q/aKaK*).

So-called horizontal FDI is investment that replicatesbusiness activities in foreign countries in responseto tradecosts or otherfrictions.To the degree that domestic exports are substitutesfor FDI, such FDI can outputproducedby horizontal be viewed as representing diversionof domestic a activity. Once horizontalinvestmentshave been
2 Equation (1) does not include tax effects on investment; for an analysis, see Desai et al. (2004c).

between foreign investmade, complementarity ment and domestic investment may emerge as foreign operations make use of functions performed by headquarters. Alternatively, foreign investments might be vertical in nature, whereby production processes are fragmentedinto different stages and optimized globally. Vertical investments might substitute foreign activity for domestic activity if firms are shifting the location of activities that have been performed domestically. However, once the productionprocess has been split up, foreign and domestic activities are likely to complement one another. Vertical foreign investments can raise the demand for domestic capital by permittinggreaterexploitation of intangible assets producedby domestic activity or by increasing the profitabilityof domestic production that can be combined with foreign output. A voluminous literature examines the relative ability of vertical or horizontaltheories of FDI to explain investment and tradepatterns and finds evidence of both types of activity.3 Since substitutionand complementaritycan be operative for different firms at different times, their relative importance remains a matter for empirical resolution. In addition to these conceptual issues, substantial measurement issues arise when investigating foreign investment. The common intuition is that FDI consists of investment or capital expenditures by multinational firms abroad. In fact, the common measurement of FDI in balance-of-payment accounts reflects the flow of financing for that investmentacross borders. Specifically, FDI flows in balance-ofpaymentaccountsequal the sum of equity flows from home to abroad,intercompany debt flows fromparentsto subsidiaries, retained and earnings reinvested in those subsidiaries(see Alicia M. Quijano, 1990; OECD, 1999). These methodsof measuringforeign investmentmake FDI flows a bettermeasureof the financingof overseasoperations throughthe use of internalcapitalmarkets ratherthan the actualcapitalexpenditures forof As eign subsidiaries. discussedbelow, the distinction between investmentand financingmay help resolve some of the apparently conflicting patterns that appearin the data.

3 See, for example, RichardE. Caves (1996) and James R. Markusen (2002) for surveys of this literature.

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions

VOL.95 NO. 2

INTRAFIRM INTERNATIONAL PRODUCTIONDECISIONS

35 OECD

II. Does OutboundFDI Stimulateor Reduce DomesticInvestment? Table 1 presents regressions constructed to resemble those of Feldstein (1995). While Feldstein's analysis was restricted to a relatively small number of countries in the 1970s and 1980s, the regressionspresentedin Table 1 cover a broadersample of countriesfor the 1980s and 1990s. Observations representdecade-longaverage valuesfor each of 20 (in the case of the 1980s) or 26 (for the 1990s) OECD countries.The dependent variable in the regressions reportedin Table 1 is the ratio of nationalgross capitalformationto GDP, whichis equivalent the variable to thatFeldsteinuses, andit is regressedon measures of savings, outwardFDI flows and inwardFDI flows.4 Column (i) reports estimated coefficients from regressions for the 1980s sample. The 0.7801 coefficient on the gross saving/GDP variable is consistent with the finding of Feldstein and Charles Y. Horioka (1980) that national saving and investment rates exhibit close to a one-to-one correlation.The - 1.3357 coefficient on the outwardFDI flow/GDP variable, while only marginally significant, is consistent with Feldstein's finding that FDI outflows reduce domestic capital formation almost dollarfor-dollar. Similarly, the 1.1869 coefficient on inward FDI flows/GDP, while insignificant, is consistent with the effect of FDI inflows estimated by Feldstein. Column (ii) reports estimated coefficients from regressions run on observations from 26 countries in the 1990s. The estimated 0.6255 coefficient on gross savings/GDP remains large and significant, though significantly different from unity. The -1.0767 coefficient on out-

FORMATION/GDP)

Independentvariable Constant
Outbound FDI flow/GDP

(i) 0.0575 (0.0302)

(ii) 0.0921 (0.0327)

(iii) 0.0846 (0.0205)

(0.7136) InboundFDI flow/GDP Gross savings/GDP Period 1.1869 (1.1465) 0.7801 (0.1005) 1980s

(0.2349) 0.3220 (0.2402) 0.6255 (0.1383) 1990s

(0.2160) 0.3009 (0.2084) 0.6718 (0.0798) 1980s and 1990s 46 0.6885

No. of observations:
R2:

20 0.7226

26 0.6652

Notes: The dependent variable is the average ratio of gross capital formation (investment in fixed assets and inventories) to GDP, for OECD countries over particulardecades. Column (i) restrictsattentionto the 1980s, column (ii) to the 1990s, and column (iii) covers both decades. OutboundFDI flow/GDP is the decade-averageratio of total FDI outflows to GDP, and inbound FDI flow/GDP is the decade-average ratio of total FDI inflows to GDP. Gross savings/GDP measures the decade average ratio of GDP less total consumption to GDP. Heteroscedasticity-consistent standard errors appearin parentheses.

4 Gross capital formation is the measure of investment employed, and this variableincludes outlays on additionsto fixed assets as well as net changes in inventories. Measures of gross savings are computed as GDP less final consumption expenditures. Aggregate national measures of gross capital formationas a share of GDP, and gross savings as a share of GDP, are drawn from the World Bank's World Development Indicators. Data on total outwardand inward FDI flows come from the OECD's International Direct Investment Statistics database. These flows are scaled by GDP as measured in the World Bank data. All of these variables are averagedover the 1980-1989 and 1990-1999 periods, yielding observations for individual countries by decade.

ward FDI flows/GDP is significant and again very similar to the equivalent coefficient estimated by Feldstein for the 1970s and 1980s, whereas the 0.3220 coefficient on inward FDI flow/GDP is statistically indistinguishablefrom zero. Pooling datafor the 1980s and 1990s, as in the regressionreportedin column (iii), produces results that closely resemble those for the 1990s. The patternthat Feldstein identified in a smaller sample of countries in the 1970s and 1980s persists in a broadersample of countries in the 1980s and 1990s: higher outboundFDI is associated with lower domestic investment. The cross-country evidence presented in Table 1 describes the determinants of aggregate investment patterns for entire countries; the analysis is not restricted to estimating investment demandby multinationalfirms.In orderto obtain estimates of the effects of foreign investment on domestic investment by multinational firms, the analysis in Table 2 uses time-series data on the domestic and foreign capital expenditures of U.S. multinationals,while controlling

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions

36

AEA PAPERSAND PROCEEDINGS

MAY2005

TABLE AND 2-DOMESTIC FOREIGN CAPITAL EXPENDITURES ital expenditureis associated with 3.5 dollars of = OFU.S. MULTINATIONALS VARIABLE (DEPENDENT domestic capital expenditures by the same DOMESTIC CAPITAL EXPENDITURES OFU.S. group of multinationalfirms, strongly suggestMULTINATIONALS/U.S. GDP) Independentvariable Constant Foreign capital expendituresof U.S. multinationals/U.S.GDP U.S. capital expendituresof foreign multinationals/U.S.GDP U.S. gross savings rate Time trend? No. of observations:
R2:

(i) 1.5082 (0.2957) 3.5059 (0.6311)

(ii) 0.0402 (0.4203) 3.8796 (0.6262) -1.8550 (0.5340) 0.2565 (0.0535)

yes 21 0.6857

yes 21 0.8624

Notes: The dependent variable is the domestic capital expendituresof U.S. multinationalfirms scaled by U.S. GDP. Foreign capital expenditures of U.S. multinational firms measure the host-countrycapital expendituresof majorityowned nonbankaffiliatesof U.S. multinationals. U.S. capital of expenditures foreignfirmsmeasurethe capitalexpenditures of all nonbankU.S. affiliatesof foreign multinationals. Both specifications include a time trend, and heteroscedasticityconsistentstandard errorsappearin parentheses.

ing a complementaryrelationshipbetween foreign and domestic investment. The regression reported in column (ii) of Table 2 includes the domestic saving rate and U.S. capital expenditures by foreign-owned firms in the United States as additionalcontrol variables. The inclusion of these variablesdoes little to affect the estimated complementaryrelationship between foreign and domestic capital spending by American multinationalfirms: the estimated 3.8796 coefficient implies that an additional dollar of foreign capital expendituresis associated with 3.9 dollars of domestic capital expenditures. The estimated - 1.8550 coefficient implies that an additionaldollar of capital spending by foreign-owned firms in the United States reduces domestic expenditures by U.S. multinational firms by 1.9 dollars, while the estimated 0.2565 coefficient implies that U.S. capital spending by American multinational firms increases by 26 cents for each additional dollar of domestic savings.6 III. Reconcilingthe Evidence

for domestic savings and the capital expenditures of the U.S. affiliates of multinationals based outside the United States.5 The dependentvariablein these regressionsis the ratio of aggregate annual domestic capital expendituresof Americanmultinationalfirmsto U.S. GDP, so it differs from the dependent variable in the regressions in Table 1 by focusing on a particulartype of capital formationby a particularclass of investor. Both regressions include time trends. The independent variable of interest in the regression reportedin the first column of Table 2 is the ratio of U.S. multinational firms' foreign capital expenditures to U.S. GDP. The estimated 3.5059 coefficient implies that an additionaldollar of foreign cap5 The annual Survey of U.S. Direct InvestmentAbroad, published by the Bureau of Economic Analysis, provides the U.S. capital expendituresof nonbank American multinationals as well as the foreign capital expendituresof the majority-ownednonbank affiliates of these firms. The annual Survey of Foreign Direct Investment in the United States provides the U.S. capital expenditures of nonbank multinationals based in other countries, and the National Income and Product Accounts provide annual measures of U.S. gross savings and U.S. GDP.

Why are the implications of the time-series evidence on investment by American multinational firms so contradictoryto the implications of the cross-sectional evidence examined by Feldstein and updatedto the 1980s and 1990s? There are a numberof potentialexplanationsfor the distinct results. First, and most obviously, the regressionspresentedin Table 2 exclusively consider the United States, while Table 1 employs data from a large sample of OECD countries. It is possible that foreign and domestic investment are complements in the American economy, whereas they are substitutesin other OECD economies. Given the relatively limited available evidence of the behaviorof non-U.S.based multinationalfirms, it is difficult to dismiss or accept this explanation. Second, the two analyses also differ in their scope, as the cross-sectionalevidence considers
6 The Durbin-Watsonstatistic for this specificationis in the range that requiresfurtherexplorationgiven uncertainty in the distributionof the d statistic. Estimatingthis specification using the Prais-Winstonor Cochrane-Orcutt procedures yields qualitatively similar results.

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions

VOL.95 NO. 2

INTRAFIRM INTERNATIONAL PRODUCTIONDECISIONS

37

economy-wide investment while the time-series evidence considers only the activities of U.S. multinational firms. Higher levels of foreign investmentmight be associated with higher levels of domestic investmentby parentsbut lower levels of investmentby otherfirms in the source country. In short, higher foreign investment by multinationalfirms may representthe decision to internalize activities abroad that previously were undertaken an arms-length on basis domesWhile there is evidence on the growing tically. firmsto internalizeactendencyof multinational tivity (as in Desai et al. [2004e]), there has been little, if any, analysisof this channelof diversion of domesticactivityby FDI. A third, and much more likely, possibility is that either the cross-sectional or time-series equations (or both!) are seriously biased by the omission of importantvariables. For example, FDI flows at the aggregate level are defined to include financingflows, while the multinational firm analysis restrictsattentionto capital expenditures. As a consequence, high FDI outflows might indicate that domestic investment opportunities are poor, and these poor opportunities could be the force behind lower domestic investment and the reallocation of funds to more profitableforeign opportunities.The analysis of the capital expenditures of foreign affiliates, which does not consider their associated financing, is less subject (but not immune) to this concern. In Desai et al. (2004b), we address the problem of omitted variables by applying instruments for changes in foreign investment in a panel of American multinationalfirms. The instrument is the growth rate of foreign economies in which a firm invests in the base period, which is a strong predictor of subsequent changes in foreign investment by American multinational firms. Thus, a firm investing in the United Kingdom in 1982 is more likely to exhibit rapid subsequenteconomic growth than is a firm that invested in France in 1982, since the British economy subsequently grew much more rapidly than did the French economy. Using this instrument,we find that rising foreign investment (at the firm level) is associated with growing domestic investment, which is indicative of complementarity(Desai et al., 2004b). It is noteworthy that the estimated effect of foreign investment on domestic investment has a larger (positive) magnitudein the instrumental-

variables equation than does the corresponding estimated coefficient in the (uninstrumented) ordinary least-squares equation, implying that, in firm-level U.S. data, omitted variables have the effect of making foreign and domestic investment look more like substitutes than they really are. While it is difficult to find suitable instruments for aggregate cross-sectional or timeseries estimation of the effects of FDI on domestic investment, the ability to restrict attention to the activities of multinational firms makes the U.S. time-series evidence likely to be more reliable than the OECD cross section. The evidence from the time series of U.S. data implies that FDI encourages greater domestic investment. The firm-level panel estimates reported in Desai et al. (2004b) are consistent with this finding, and they are reassuring in that they suggest that biases introduced by omitted variables have a tendency to mitigate against a finding of complementarity between foreign and domestic investment. IV. Conclusion It has been natural to assume that foreign investment comes at the expense of domestic investment. New evidence from analyses of American multinational firms suggests instead that greater foreign investment is associated with higher levels of domestic investment. This estimated complementarity implies that firms combine home productionwith foreign production to generate final output at lower cost than would be possible with production in just one country, making each stage of the production process more profitable, and therefore, in equilibrium, more abundant.It is clear that the simple story, in which the world has a fixed stock of investmentcapitalthat can eithergo to one place or another,is due for rethinking.The growing firms, their reliance prominenceof multinational on theirinternalproductand capitalmarkets,and this evidence on the complementarity theirinof vestment worldwide suggest the importanceand possible contoursof such a reconsideration. REFERENCES Antris, Pol. "Firms,Contracts,and Trade Structure."QuarterlyJournal of Economics, 2003, 118(4), pp. 1375-1418.

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions

38

AEA PAPERSAND PROCEEDINGS

MAY2005

JamesR. and Maskus, Carr,DavidL.; Markusen, Keith E. "Estimatingthe Knowledge-Capital Model of the Multinational Enterprise." AmericanEconomic Review, 2001, 91(3), pp. 693-708. Caves, RichardE. Multinational enterprise and economic analysis, 2nd Ed. Cambridge,UK: CambridgeUniversity Press, 1996. Desai, MihirA.; Foley, C. Fritz and Forbes,Kristin J. "Financial Constraints and Growth: Multinational and Local Firm Responses to Currency Depreciations." Mimeo, Harvard University, 2004a. Desai, MihirA.; Foley, C. Fritz and Hines,James R., Jr. "Foreign Investment and Domestic Economic Activity." Mimeo, Harvard University, 2004b. Desai, MihirA.; Foley, C. Fritz and Hines,James R., Jr. "ForeignDirect Investmentin a World of Multiple Taxes." Journal of Public Economics, 2004c, 88(12), pp. 2727-44. Desai, MihirA.; Foley, C. Fritz and Hines,James R., Jr. "A MultinationalPerspective on Capital Structure Choice and Internal Capital Markets."Journal of Finance, 2004d, 59(6), pp. 2451-88. Desai, MihirA.; Foley, C. Fritz and Hines,James R., Jr. "The Costs of SharedOwnership:Evidence from International Joint Ventures." Journal of Financial Economics, 2004e, 73(2), pp. 323-74.

Feldstein,Martin S. "The Effects of Outbound Foreign Direct Investment on the Domestic Capital Stock," in MartinFeldstein, JamesR. Hines, Jr., and R. Glenn Hubbard,eds., The effects of taxation on multinationalcorporations. Chicago: University of Chicago Press, 1995, pp. 43-66. Feldstein, Martin S. and Horioka, Charles Y. "Domestic Savings and International Capital Flows: The 1980 W. A. MackintoshLecture at Queen's University." Economic Journal, 1980, 90(358), pp. 314-29. Grossman,Gene M. and Helpman,Elhanan."Integration versus Outsourcing in Industry Equilibrium."QuarterlyJournal of Economics, 2002, 117(1), pp. 85-120. Markusen,James R. Multinational firms and the theory of international trade. Cambridge, MA: MIT Press, 2002. OECD (Organisation EconomicCooperation for and Development).OECD benchmarkdefinition of foreign direct investment. Paris, France: OECD, 1999. Quijano,Alicia M. "A Guide to BEA Statistics on Foreign Direct Investment in the United States." Survey of Current Business, 1990, pp. 29-37. Scheve, Kenneth F. and Slaughter,MatthewJ. Globalization and the perception of American workers. Washington, DC: Institute for InternationalEconomics, 2001.

This content downloaded on Thu, 20 Dec 2012 14:08:11 PM All use subject to JSTOR Terms and Conditions