T HE relationship between direct investment by U.S. firms and the decline in U.S. export trade shares has been a subject of bitter controversy for at least the last twenty years. These changes over time in trade flows have undoubtedly been influenced by other factors such as productivity and monetary and fiscal policy in the United States and elsewhere. All of these were presumably reflected in price and income changes, the effects of which on trade probably have swamped any that might have stemmed from changes in the level of direct investment. One way to disentangle some of these influences might be to disaggregate by country, industry, and even better, by firm. However, we have not yet developed enough disaggregated time series data for this purpose and have therefore chosen to work first with cross-sections by country and industry. The method used here can be summarized by saying that we examine exports to a crosssection of 44 foreign destinations in 1970 by the United States and by 13 other major exporting countries. We relate these exports to characteristics of the destinations and to production in them by affiliates of companies based in the United States and those based in other countries. The destination country characteristics are their size (measured by their Gross Domestic Product (GDP), converted to dollars by exchange rates, or their total imports of manufactures), their membership in the European Economic Community (EEC), and their distance from the United States and from Germany (representing distance from most of the other major exporters). The affiliate activity variables are measures of the output of U.S.-owned manufacturing and nonmanufacturing affiliates and of the number of foreign-owned manufacturing affiliates in each country.' The measures for U.S.-owned affiliates were net sales (total sales less imports from the United States), and estimated net local sales (net sales multiplied by the ratio of local sales to total sales of the affiliate). In effect we have taken the elements of a fairly crude standard trade model (country size, distance, and membership in a trade bloc)2 and added to that some variables describing direct investment by the United States and other countries, to ask whether the latter have any impact on trade beyond that of the country characteristics. The basic assumption implied by this analysis is that the answer to the question What would have happened if U.S. or foreign companies had not invested in a country or had invested less? is provided by trade with those countries in which U.S. or foreign firms did not invest, or invested less than in others. The danger in this assumption is obvious: that there are factors which simultaneously affect investment and trade, giving a spurious appearance of a relationship between them. Such spurious relationships could work in the direction of showing that investment either increases or decreases trade. Size of host country, if it were omitted, would work in the former direction of suggesting complementarity between investment and trade, Received for publication October 2, 1980. Accepted for publication October 21, 1980. * National Bureau of Economic Research and Queens College, City University of New York, and National Bureau of Economic Research and Temple University, respectively. The research reported on in this paper was financed by grants to the National Bureau of Economic Research from the National Science Foundation and the Ford Foundation. However, it is not an official National Bureau publication. In particular, it has not been submitted to the Board of Directors for approval. Any opinions, findings, conclusions or recommendations expressed herein are those of the authors and do not necessarily reflect the views of the sponsoring or financing organizations. Mary Boger, Linda Quandt, and Marianne Rey were responsible for data collection and organization and Muriel Moeller for preparation of the manuscript. We are indebted to the Bureau of Economic Analysis of the U.S. Department of Commerce for the use of their data and particularly to Arnold Gilbert and Michael Liliestedt of the BEA for programming and other assistance with these data. Earlier versions of this paper appeared as NBER Working Papers 87 and 131. I No data were available on foreign-owned nonmanufacturing affiliates or on the size of foreign-owned manufacturing affiliates. 2 See, for example, Leamer and Stern (1970), Linneman (1966), Taplin (1967), and Tinbergen (1962).