W hic h Coun tries Export FD I, and H o w M uc h?∗Assaf Razin†, Yona R ubinstein‡and E fraim S adka§November, 2003AbstractThe paper provides a reconciliation of Lucas’ paradox, based on fixed setup costsof new investments. With such costs, it does not pay a firm to make a “small”inv estment, even though suc h an investment is called for by marginal productivityconditions.Using a sample of 45 developed and developing countries we estimate jointlythe participation equation (the decision whether to invest at all) and the FDI flowequation (the decision how much to invest).We find that countries which a re more likely to serve as source for FDI exportsthan their characteristics project export lower flow of FDI than is predicted bytheir characteristics. This negative correlation suggests that the source countrieswith relatively low setup costs are also those with high marginal productivity ofcapital.∗We wish to thank Anil Kashyap, Elhanan Helpman, Gita Gopinath and John Romalis for manyuseful comment s and suggestions. This paper was completed when Rubinstein and Sadka visited Razinat Cornell University.†Tel-Aviv University, Cornell University, CEPR, NBER and CESifo.‡Tel-Aviv Univ ersity§Tel-Aviv Univ ersity1Table A2: Data Source Variables: Source: Import of Goods Direction of Trade Statistics, IMF FDI Inflows International Direct Investment Database, OECD Unit Value of Manufactured Exports World Economic Outlook, IMF Population International Financial Statistics, IMF GDP World Development Indicators, World Bank Distance Shang Jin Wei’s Website: www.nber.org/~wei Bilateral Telephone Traffic Direction of Traffic: Trends in International Telephone Tariffs, International Telecommunications Union Education Attainment Burro-Lee Dataset: www.nber.org/N... Roads ….. Language ….. Longitude and Altitude ….Table A1: List of Countries, by Observed Source/Host Status Observed as Country Source1 Host Argentina + + Australia + + Austria + + Belgium + Brazil + Canada + + Chile + China + Columbia + Denmark + Ecuador + Egypt + Finland + France + + Germany + + Greece + Hong Kong + India + Ireland + Israel + Italy + + Japan + + Korea + Kuwait + Malaysia + Mexico + Netherlands + + New Zealand + Nigeria + Norway + + Peru + Philippines + Portugal + Saudi Arabia + Singapore + South Africa + Spain + Sweden + + Switzerland + Taiwan + Thailand + Turkey +United Kingdom + + United States + + Venezuala + 1We have information on whether a country is a source country only for OECD countries.1 IntroductionIn an influential paper, Lucas (1990) asks: “Why doesn’t capital flow from rich to poorcountries? ” Indeed, the law of diminishing returns implies that the margin al product ofcapital is high in poor coun tries and low in ric h countries. Therefore, capital should flowfrom ric h to the poor countries.W ith standard constan t-return s-to-sca le production functions, when the w a ge (perefficiency units of labor) is higher in a ric h countr y than in a poor country, then thereturn to capital must be lower in the ric h coun try than in the poor country. Therefore,the existence of “h u ge” wag e gap bet we en ric h an d poor coun tries mu st be associatedwith an opposite gap in the rates o f return to capital. Given that labor is not allo wedto freely migrate from poor to rich countries, it follow s that cap ital wou ld flow in theopposite direction, thereb y equaling the returns to capital and concomitantly wages too.The eq u alization of w ag es (in directly through internatio na l capit al flo ws) w ould eliminatethe need to con tro l m igr ation. In practice, ho wever, this is hardly the case. Ev en thoughbarriers to in ternational capital mobility are by and large being eliminated, the wage gapis still in force, and migration quotas from poor to rich countries have to be enforced.1Lucas reconciled this paradox by appealing to a hum an capital externality that gen-erates a Hicks-neu tral productivity ad vantage for rich countries o ver poor countries. Thea verag e lev el of educational attainm ent is an externa l factor into the production function,that raises the productivit y of labor, and mor e importan tly, capital. As a result, capitalflows, though equalize the rates of return to capital, fall short of equalizing the capita-labor ratio. Therefore, wages (per efficiency un its) are not equalized. Labor of all skilllevels has still strong incentives to migrat e from poor to rich countries.In this paper we provide ye t another reconcilia tion of Lucas’ parad ox , based on fixedsetup costs of new investm ents. W ith such costs, it does not pay a firm to make a“small” investm ent, even though suc h an investmen t is called for by marginal productivityconditions (that is, the stand ard first-order conditions for p rofit m aximization). Put it1Note also that despite the expansion of international trade in goods, still the Stolper-Samuelson(1941) factor price equalization theorem does not manage to eliminate the wage gap.2differently, the firm ’s in vestmen t decision is twofold now : mar ginal productivit y conditionsdetermine how m uc h to in v est, whereas a “participation” condition determines whetherto invest at all. In suc h a framework, the Lucas paradox can be reconciled: ra tes ofreturn to capital are equalized and concomitantly the wage gap remains in force.When one looks at data on gross in ternational capital flo w s of foreign direct investmen t(FDI ), one is imm ediately struck by the lack of flows from man y rich countries to manyhost coun tries. We look ed at data on bilateral FDI flows in a sample of 45 cou ntries,both developing and developed, over the period of years from 1981 to 1998. Out of 45 x44 = 1980 source-host pairs of countries with potential bilateral FDI flo w s, w e found thatthe number of pairs with actual flows is only 334! There were only 12 countries that mad eany FDI export over that period and m ost of th ese countries exported FDI to on ly oneother coun try. These crude findings provide a prima facia suggestion for the existenceof fixed setup costs of investm ent that nullify the poten tial of “small” capital flows thatmay have been called for by margin al productivity conditions.We empha size again that whether a cell of s−h pair becomes active or inactive and howmuch flo w is