Publication Type

The Relationship Between Trade and Foreign Investment: Empirical Results for Taiwan and South Korea

Working Papers 98-7



This is an issue that has long concerned policymakers in the large industrial nations, who have worried about possible negative effects of outward FDI upon the nation's balance of payments and employment of its work force. Thus, a number of empirical studies have been published regarding this issue for these countries, but not for developing or newly industrializing countries.

In recent years, however, relative costs of labor in Taiwan and Korea have risen and, in response, Taiwanese and Korean firms have attempted to move up the "production ladder" into more capital intensive (including human-capital intensive) operations and have moved certain of their production activities overseas. Thus, the effect of FDI on trade has also become a concern of policymakers in Taiwan and Korea. For this reason, in this paper we investigate this relationship empirically for Taiwan and South Korea. We begin with a general discussion of this relationship and a review of previously published studies of the relationship for industrialized nations.

In principle, either relationship between FDI and exports-complementarity or substitutability-could hold. FDI takes place when investors, usually multinational firms, based in one nation (the "home" nation) establish operations under their managerial control in some other nation (the "host" nation). Often, the motivation is to produce locally in the host nation products that had previously been exported from the home nation, and to the extent that this happens, FDI and home nation exports are substitutes. But the home nation operations of a multinational firm can be vertically linked with host nation operations, such that an increase in the activity in the latter generates increased demand for intermediate products (including capital goods) from the former. Also, marketing and distribution capabilities created by FDI might enable the home nation operations to export final goods and services to customers that would not be reached in the absence of FDI. To the extent that either of these happens, home country FDI and exports will be complements.

Because the value of intermediate products is a component of the value added to the final goods, it could be argued that FDI and exports must be net substitutes in some long run sense. If exports of final goods from home nation are displaced by local production, there will be a net loss of export value even if the gross loss is offset in part by export of capital and intermediate goods. This is true in a trivial sense because the value of final goods must be greater than or equal to the value of all inputs used to produce those goods. However, this line of argument supposes that host nation demand for a particular good will always be fulfilled by exports from the home country, which might not be the case. Changes in the relative cost of production might imply that, with the passage of time, home nation exports will be displaced by local production irrespective of whether the displacement is done by multinational firms shifting production from the home to the host nation or by local firms operating entirely within the host nation.

Indeed, with the passage of time, the relationship between FDI and exports could very well change. Suppose, for example, that the host nation were to become over time relatively more efficient in the production of a particular class of final goods and the home nation were to become relatively more efficient in the production of intermediate goods used to produce these final goods. If multinational firms were to hold specialized skills enabling the realization of internal economies associated with vertically linking the production of the two sets of goods, the relationship between additional FDI and exports by these firms could become increasingly complementary even if at some earlier point in history an initial FDI served to displace home country exports.

More complex relationships between FDI and international trade have been noted. Urata (1995) has examined the growth of the electronics industry in East Asia, and finds that direct investment and trade in electronics goods have grown hand-in-hand in the region. The electronics industry worldwide has been marked by rapid overall growth and by rapid rates of new product development and cost reduction. Urata thus finds that outward FDI by Japanese firms in the East Asian region has been driven both by growth of host nation demand and by complex patterns of shifting relative costs. These cause firms to seek new production sites and to create complex patterns of cross hauling of both final goods and intermediate products. He notes that, as these Japanese MNEs have over time placed new direct investments in countries where they were previously absent (for example China), these firms have not stopped nor even curtailed production in countries with older-vintage FDI.

A further reason for complementarity between international trade and activity of multinational firms is explored by Brainard (1995a, 1998). She notes that multinational firms typically hold intellectual property advantages (e.g., technologies and trademarks) that might enable larger market shares and hence increase both trade and investment in markets where these firms operate. Brainard hypothesizes that the share of trade in total sales by a firm to a particular market will be negatively affected by transport costs and trade barriers, but positively affected by investment barriers and firm-level scale economies. Using US Commerce Department data for US direct investment abroad and foreign direct investment in the United States, she finds that trade and FDI barriers and scale economies are robust explanatory variables, while transportation costs are not. In a related work, Brainard shows that relative factor proportions are not robust explanators of multinational firm activity (Brainard 1995b).

Whether FDI and exports are net substitutes or net complements thus is indeterminate on the basis of principles and, as a practical matter, the issue becomes an empirical one. Most studies of this relationship in fact tend to indicate that the relationship is complementary, that more FDI is associated with more, rather than less, exports.

In both the United States and in the United Kingdom during the late 1960s, for example, there was official concern over the effects of outward FDI on the overall balance of payments on a current account basis. Central to this concern was the question of the impact of outward FDI on trade flows. In response, two studies of these effects were carried out under official auspices (Reddaway et al. 1967 and Hufbauer and Adler 1968).

Using somewhat different methodologies and coverage, both studies arrived at roughly similar conclusions: If future cash flows are not discounted, the overall long term effects of outward FDI on the balance of payments are positive. That the effects of financial flows alone are positive should not be a surprise to anyone. This is because a firm undertakes an investment undertaking of any sort on the expectation that the investment will yield a positive return for the firm's shareholders, and ultimately that return must be reflected in dividend payments by the parent organization to those shareholders. Thus, to the extent that the shareholders of the firm are nationals of the home country, the returns accruing to the foreign affiliates of a firm must ultimately accrue to home country nationals funded through the parent organization.

However, both studies also indicated that outward FDI tended to stimulate exports (mostly of capital goods and intermediate goods) without stimulating imports in equal magnitude.

Later studies yielded results generally consistent with these findings. Bergsten, Horst, and Moran (1978), for example, found that the growth of US affiliates abroad had a significantly positive effect on the growth of exports of the US parent firms. Lipsey and Weiss (1981) also found that US outward FDI was associated with increased US exports, even after controlling for other effects (firm size, expenditures on R&D and marketing, etc.) but that the production of US affiliates abroad substituted for exports to the host country of third countries. In a later study, the same authors (Lipsey and Weiss 1984) analyzed unpublished US Commerce Department data at the level of the individual firm to examine foreign production and US exports in 14 industries in the manufacturing sector. They reported positive and significant relationships in 11 of these industries.

Similar relations have been found for the impact of offshore production of Swedish-owned firms upon the exports manufactured goods of the home country, Sweden (Blomström, Lipsey, and Kulchyck 1988). Sweden is an advanced industrial economy located in close proximity to other advanced economies, and most of Sweden's direct investment is located either elsewhere in Western Europe or in North America. Blomström et al. found that increases in the production of affiliates of Swedish firms are positively related to increases in exports for the seven industrial categories studied. Also, they showed that there was no propensity for this positive relationship to change as the foreign production grew.

Pearce (1990), following an approach similar to that of Blomström et al., examined the exports and foreign production of 458 of the world's largest industrial MNEs for the year 1982. His findings are that increases in foreign production are generally positively related to increases in exports. This was found to be especially true for intrafirm (as opposed to interfirm) exports, underscoring the importance of vertical relationships among the various international affiliates of this sample of MNE's.

Buigues and Jacquemin (1996) examine the issue of complementarity versus substitution between FDI and exports with respect to both US and Japanese direct investment in the European Union. The basic assumption is that if share of the total exports from each of these countries going to the EU is positively related to the share of FDI going to the EU after controlling for three additional variables, the relationship is complementary. The three additional variables are: intra-EC nontariff trade barriers rate of growth of final demand, and the EC's sectoral specialization, all of which are assumed to be positively related to FDI. Buigues and Jacquemin's sample is pooled cross sectionally across seven industries (six for the United States) and ten years. They find the relationship between FDI and exports to be complementary for both the United States and Japan.

Industry Canada (1994) found that FDI from Canada is associated both with increases of Canada's exports and imports. The same finding is reported with respect to foreign direct investment in Canada. The findings are aggregate and (apparently) based on time series analysis. Estimates are made of the elasticities of exports and imports with respect to Canada's outward investment and the latter are higher than the former. These estimated elasticities of trade with respect to investment stocks (see Industry Canada 1994, Table 7) are not, however, controlled for the influence of factors such as economic activity, comparative costs, or other variables that could affect the outcomes.

Graham (1998), using methodology similar to that reported later in this paper, found that complementary relationships exist between outward FDI and exports and outward FDI and imports for the United States and for Japan.

Thus, all these studies involving developed countries have concluded that the relationship between FDI and exports is complementary. As is described in the next section, the results of our empirical investigations for two newly industrialized countries point to a consistent result, but with some twists.

The Empirical Methodology

The current studies on FDI/trade relationship suffer two shortcomings. One is that they only examine this relationship for industrialized countries. This relationship for the newly industrialized nations remains to be examined. Second is that most of the studies cited above could be criticized for ignoring the possible effects of simultaneous determination of FDI and exports which could be causing a spurious correlation between these two and hence lead to an erroneous interpretation of complementarity.2 This would be the case if both FDI and exports were responding to a common, unspecified causal element. For example, suppose that income or size of market alone determined both direct investment abroad and exports-that is, both exporters and direct investors put their energies into developing large markets and/or those with high per capita incomes but ignored small and/or low per capita income markets.3 Then simply to show that large share of exports was associated with markets where the share of direct investment was also large would not be sufficient to show that exports and direct investment abroad were complementary. They could still be substitutes once the effects of market size were taken into account.4 Likewise, elements of simultaneous determination could distort results of studies based on differences across industries.

Thus, the effort is made in the results reported here to control for these factors when examining the relationships between FDI and trade for Taiwan and South Korea. In particular, we remove factors that might simultaneously determine exports, imports, and FDI and then examine the residual relationship between the first two and the lattermost of these variables with the source of the simultaneity bias removed. Specifically, a gravity model was used first to estimate the effects of three variables deemed to be very important determinants of both FDI and exports. The three variables chosen were (1) per capita income in each host nation market (for which GDP per capita was used), (2) total size of this market (for which total population was used), and (3) distance from the host to the home country.5 The model was used to test determinants of FDI and exports for two home countries, Taiwan and South Korea. The "distance" from the home country to the host country was from the home countries' capitals, i.e., Taipei and Seoul, to the host nation capital. The gravity specification was multiplicative, i.e., the assumed relationship was


where y is the logarithm of the dependent variable (FDI or exports), the xi are the three independent variables, and Î is an error term (assumed, as usual, to be log-normally distributed with mean 1). The expected signs of ß1 and ß2 are positive (both home nation exports and FDI would be expected to positive functions of per capita income and market size). The expected sign of ß3 is negative for exports (the further the market is from the home nation, the higher transport costs, and hence the less likely that firms would export from the home nation) but indeterminate for direct investment. For example, if direct investment were to be a substitute for exports, then arguably the substitution would be most likely in those markets for which transactions costs associated with exports were high, and the expected sign of ß3 would be positive; but one can envisage circumstances where direct investment would occur in geographically proximate markets {see, e.g., Graham 1995}). The residuals from each of the two estimations (exports and FDI as a function of the three variables) were then regressed upon one another. The presumption was that if the gravity models have succeeded in removing simultaneity bias, then any correlation of the residuals would reflect some other causal relationship between FDI and exports-such as that due to sourcing substitution or to complementarities in production or distribution and marketing. A positive correlation coefficient would suggest complementarity and a negative coefficient substitutability. Also performed were similar two stage analyses between imports and direct investment abroad.

Results for Taiwan

For Taiwan, the sample included 54 individual countries that were destinations of both Taiwan's exports and its direct investment. These fifty-four countries accounted in 1994 for almost the entire stock of Taiwan's direct investment abroad and most of its manufactured goods exports. Separate analyses were performed using (1) the data for all countries with and without adjusting for language linkage, development status (OECD), and geographical location (Asia); and (2) the data for all countries except mainland China (which, for reasons to be explained, might act as a distorting "outlier" in the sample) with and without adjusting for language linkage, development status (OECD), and geographical location (Asia).

Summary results of the gravity analyses are given in Table 1 below. As can be seen, absent the dummy variables, as an explanator of FDI, only market size (population) appears significant. As determinants of exports and imports, however, all three explanatory variables appear significant. When using dummy variables to take into account language, development status, and geographical location, per capita income becomes positively and significantly related to Taiwan's outward FDI. Also, although distance continues to be insignificant statistically, it continues to have negative sign, i.e., the farther away the host country is from Taiwan, the less likely Taiwan is going to invest there.

The language dummy does not appear to be important as a determinant of direct investment. This might be due to the fact that Chinese is rarely spoken in the world other than mainland China and Hong Kong. By contrast, the OECD dummy is negatively and significantly related to FDI and Asia dummy is positively and significantly related to FDI. This result seems to indicate that in terms of the distribution of geographical location, Taiwan's FDI is concentrated in non-OECD Asian countries.

As determinants of exports and imports, all explanatory variables are of the expected sign and significant except for two dummy variables, language and OECD, both of which are not significant. In contrast to FDI, distance is a significant determinant of both exports and imports of Taiwan. These exports and imports are not, however, affected by whether the destination country is a developed economy or not. As with FDI, language does not appear to be a determinant of either exports or imports.

Table 2 presents the second stage regressions. As can be seen, the relationship between the remaining unexplained variation in Taiwan's outward direct investment in the manufacturing sector and the remaining unexplained variation in Taiwan's exports of manufactured goods for regressions with and without dummy variables is positive and significant at 99 percent significant level. This relationship indicates a strong complementary relationship between Taiwan FDI and exports.

The results of second stage regressions of the relationship between Taiwan's outward direct investment in the manufacturing sector and Taiwan's imports of manufactured goods are also indicated in Table 2. The coefficients, although positive, are not statistically significant for these regressions, either with or without dummy variables. Therefore, the complementary relationship between FDI and imports appears weak if indeed there is one at all.

Given Taiwan's recent large outward investment activities in Mainland China, one may wonder whether the results presented above could be biased because China of China's outlier status in the sample. This status arises because China has a huge population with a very low per capita income and is close geographically to Taiwan. To examine whether the outlier characteristics of China bias our results, we repeat the regressions dropping China from the sample. The gravity model regression results are presented in Table 3. As can be seen, the results do not change much at all, except for the size of some coefficients and t-statistics. Indeed, the complementary relationship between Taiwan's FDI and exports becomes slightly more significant, as the coefficients and t-statistics in Table 4 indicate.

These results may also lead one to wonder whether the strong complementary relationship between outward FDI and exports, as found above, is unique to Taiwan. To answer this question, in next section, we present econometric results for South Korea, a newly industrialized country which also has experienced a rapid outward FDI in recent years.

Results for South Korea

Gravity model results for South Korea are presented in Table 5. Without adjusting for development status (OECD) and geographical location (Asia) of host nations6, South Korea's outward FDI is positively and statistically related to per capita income level (unlike the case for Taiwan) as well as market size. Although the coefficient of the distance variable is positive-the opposite of the result for Taiwan-it is not significant. Similar to Taiwan, South Korea exports and imports are all significantly correlated with income, market size, and distance, the coefficients of all of these variables being of the expected sign.

However, after adjusting for development status and geographical location, FDI equation now is only marginally significantly correlated with per capita income and market size (the coefficients are significant at only the 12 percent level). And, surprisingly, the coefficient on the distance variable now becomes significant (and remains positive). Although the coefficient of the Asia dummy is still statistically significant, the coefficient of the OECD dummy is not significant, but it is of positive sign. This finding seems to indicate that although the FDI made by South Korea companies are mainly in Asia, their FDI location decisions do not hinge heavily on host country market size and per capita income. But, even though Asia remains the dominant region for Korean FDI, compared to firms from Taiwan, South Korea companies have also made relatively a large number of investments in OECD countries.

As for exports and imports, the results for South Korea are quite similar to those for Taiwan. All coefficients of explanatory variables are of the expected sign and all are significant except for that of the OECD dummy, which implies that the development status of a country is not an important factor in determining exports and imports to and from Korea.

Table 6 presents the second stage regressions. As indicated, the relationship between the remaining unexplained variation in South Korea's outward direct investment in the manufacturing sector and in South Korea's exports of manufactured goods for regressions with and without dummy variables is positive and also significant at 99 percent significant level. The relationship thus is strongly complementary. However, The results of the second stage regressions of the relationship between South Korea's FDI and its imports do not yield clear results. Although this relationship appears to be complementary and statistically significant for the residuals from the first stage regressions performed without dummy variables, it is no longer significant once dummy variables are used in this first stage analysis. We are not entirely sure how to interpret this last result but, given that the second specification (with dummy variables) is more complete than the first, we are inclined to conclude that FDI and imports for Korea, as for Taiwan, do not appear to be significantly related.

Conclusions and Policy Implications

The empirical evidence presented in the previous section is generally consistent with that of earlier studies on developed countries' FDI pattern reviewed in the introduction. The evidence tends to support that, as is the case for the industrialized countries, Taiwan's outward direct investment and Taiwan's exports in manufacturing are complements and not substitutes. This same complementarity appears in the South Korean data. However, for both countries, outward direct investment does not appear to be significantly related to imports.

These results have some important policy implications. To the extent that direct investment and exports indeed are complements, this result is not supportive of the claim that direct investment abroad is associated with loss of jobs or deindustrialization in these newly industrialized countries. This conclusion is consistent with those reported for the industrialized nations.

More importantly, however, the complementarity between FDI and exports suggests that outward direct investment from neither country is associated with "hollowing out" or "deindustrialization", as has been claimed. Rather the opposite would appear to be true: that as direct investment abroad expands, the affiliates of Taiwanese and Korean multinationals created by this investment acquire large appetites for goods produced in the home economies, and thus that expansion abroad is associated with increased, rather than decreased, export possibilities.

It is, however, also true that, even though the results pertaining to FDI and imports are not signficant statistically, these results are consistent with the notion that expansion of output is associated weakly with increased imports of manufactured goods into the home economies. But are these expanded imports associated with job loss or deindustrialization?

This last issue cannot be answered on the basis of the evidence provided here. A reasonable (but, on the basis of the evidence here, untestable) hypothesis would be that the imports associated with multinational activity embody a higher percentage of unskilled or semiskilled labor, and a lower percentage of higher skilled labor, than do the associated exports. If this hypothesis is correct, the implication would be that expansion of multinational activity does put wage or unemployment pressure on low skilled labor in the home countries but creates additional demand for high skilled labor. This in turn would cause the wages of the latter class of workers to rise relative to the former, and thus it is not out of the question that multinational activity has contributed to the growing disparities in income distribution in home countries. However, this possibility is conjectural and is not the only possible interpretation of the empirical results presented here. As is so often the case, it would appear that more research is necessary to test these propositions.

Table 1: Gravity Model Results (Mainland China Included, 1994)

Dependent Variables


Income/cap .32(1.1) .83(8.7)a 1.23(9.26)a .77(1.97)c .90(6.3)a 1.43(7.1)a
Population .95(4.1)a .72(8.9)a .97(8.7)a .67(2.94)b .62(7.3)a .86(7.3)a
Distance -.21(-.79) -.36(-4.0)a -.35(-7.0) -.49(1.52) -.39(-3.3)a -1.0(-6.0)a
Language       .87(.43) .40(.52) -1.32(-1.2)
OECD       -2.39(2.01)b -.42(-.98) -.79(-1.3)
Asia       2.19(2.31)b .84(2.4)b 1.1(2.25)b
Adjusted R-Square .22 .70 .75 .44 .76 .79

Note: superscripts a,b,c indicate 1, 5, 10 percent statistical significance level. Numbers in parenthesis are t-statistics.

Table 2: Regressions of Residuals on Residuals of Gravity Equations (Mainland China Included)

Taiwan FDI on Taiwan Exports Coefficient

 Regression without Dummy 1.81(5.99)a
 Regression with Dummy 1.47(4.82)a
Taiwan FDI and Taiwan Imports  

 Regression without Dummy .42(1.44)
 Regression with Dummy .10(.36)

Note: superscripts a, b, c indicate 1, 5, 10 percent statistical significance level. Numbers in parenthesis are t-statistics.

Table 3: Gravity Model Results (Mainland China Excluded, 1994)

Dependent Variables


Income/cap .30(1.1) .83(8.6)a 1.23(9.26)a .78(1.96)c .87(6.3)a 1.38(7.3)a
Population .83(3.3)a .74(8.2)a .97(8.7)a .64(2.49)b .68(7.5)a 1.0(8.0)a
Distance -.16(-.59) -.36(-4.0)a -.35(-7.0) -.50(1.50) -.40(-3.4)a -1.0(-6.4)a
Language       .37(.43) 1.69(1.63) 1.15(.82)
OECD       -2.40(2.0)b -.38(-.92) -.72(-1.3)
Asia       2.23(2.30)b .74(2.1)b .9(1.93)b
Adjusted R-Square .15 .70 .77 .38 .77 .81

Note: superscripts a,b,c indicate 1, 5, 10 percent statistical significance level. Numbers in parenthesis are t-statistics.

Table 4: Regressions of Residuals on Residuals of Gravity Equations (Mainland China Excluded)

Taiwan FDI on Taiwan Exports Coefficient

 Regression without Dummy 1.84(6.25)a
 Regression with Dummy 1.61(5.18)
Taiwan FDI and Taiwan Imports  

 Regression without Dummy .56(1.87)c
 Regression with Dummy .13(.47)

Note: superscripts a, b, c indicate 1, 5, 10 percent statistical significance level. Numbers in parenthesis are t-statistics.

Table 5: Gravity Model Results for South Korea (1993)

Dependent Variables


Income/cap .63(3.4)a .80(9.6)a 1.16(11.9)a .38(1.55) .82(7.9)a 1.22(8.5)a
Population .65(3.5)a .81(9.1)a .91(8.1)a .29(1.56) .59(7.2)a .79(6.5)a
Distance .16(.93) -.37(-4.6)a -.83(-8.2)a .39(1.87)c -.34(-3.8)a -.84(-6.8)a
OECD       1.1(1.2) -.30(-.81) -.41(-.77)
Asia       3.04(4.23)b 1.6(4.7) a .75(1.64)
Adjusted R-Square .12 .67 .76 .34 .80 .78

Note: superscripts a,b,c indicate 1, 5, 10 percent statistical significance level. Numbers in parenthesis are t-statistics.

Table 6: Regressions of Residuals on Residuals of Gravity Equations
for South Korea

Taiwan FDI on Taiwan Exports Coefficient

 Regression without Dummy 1.25(5.45)a
 Regression with Dummy 1.09(3.89)a
Taiwan FDI and Taiwan Imports  

 Regression without Dummy .59(2.31)b
 Regression with Dummy .36(1.60)

Note: superscripts a, b, c indicate 1, 5, 10 percent statistical significance level. Numbers in parenthesis are t-statistics.


1978 Bergsten, C. Fred, Thomas Horst, and Theodore H. Moran, American Multinationals and American Interests (Washington, DC: Brookings Institution).

1988 Blomström, M., R. E. Lipsey, and K. Kulchyck, "US and Swedish Direct Investment and Exports", in Robert E. Baldwin, editor, Trade Policy Issues and Empirical Analysis (Chicago: University of Chicago Press, for the National Bureau of Economic Research).

1995a Brainard, S. Lael, An Empirical Assessment of the Proximity-Concentration Tradeoff between Multinational Sales and Trade", National Bureau of Economic Research Working Paper No. 4580.

1995b , "An Empirical Assessment of the Factor Proportions Explanation of Multinational Sales", National Bureau of Economic Research Working Paper No. 4583.

1998 , An Empirical Assessment of the Proximity-Concentration Tradeoff between Multinational Sales and Trade", American Economic Review, Vol. 87, No.4. pp 520-544.

1994 Buigues, Pierre, and Alexis Jacquemin, "Foreign Direct Investment and Exports to the European Community", in Mark Mason and Dennis Encarnation, editors, Does Ownership Matter: Japanese Multinationals in Europe (Oxford and New York: The Oxford University Press).

1995 Graham, Edward M., "Canadian Direct Investment Abroad and the Canadian Economy: Some Theoretical and Empirical Considerations", in Steven Globerman, editor, Canadian-Based Multinationals (Calgary: University of Calgary Press for Industry Canada).

1998 , "On the Relationships Among Direct Investment and International Trade in the Manu-facturing Sector: Empirical Results for the United States and Japan", to appear in Dennis Encarnation, editor, Does Ownership Matter: Japanese Multinationals in East Asia ,Oxford and London: The Oxford University Press.

1968 Hufbauer, Gary C., and F. M. Adler, Overseas Manufacturing Investment and the Balance of Payments, US Treasury Department Tax Policy Research Study No. 1, (Washington, DC, US Government Printing Office.

1994 Industry Canada (Micro-Economic Policy Branch), "Canadian-Based Multinational Enterprises: An Analysis of Activities and Performance", in Steven Globerman, editor, Canadian Direct Investment Abroad (Calgary, Alberta, Canada: The University of Calgary Press).

1981 Lipsey, R. E., and M. Y. Weiss, "Foreign Production and Exports in Manufacturing Industries", Review of Economics and Statistics 63, pp. 488-494.

1984 , "Foreign Production and Exports of Individual Firms", Review of Economics and Statistics 66, pp. 304-308.

1990 Pearce, R. D., "Overseas Production and Exporting Performance: Some Further Investigations", University of Reading Discussion Papers in International Investment and Business Studies, No. 135.

1967 Reddaway, W. B., J. O. N. Perkins, S. J. Potter, and C. T. Potter, Effects of U.K. Direct Investment Overseas, London, HMSO.

1995 Urata, Shujiro, "Emerging Patterns of Production and Foreign Trade in Electronics Products in East Asia: An Examination of a Role Played by Foreign Direct Investment", paper presented to the conference "Competing Production Networks in Asia: Host Country Perspectives", Asia Foundation, San Francisco, California, April 27-28.


1. The introduction of this paper is based on Graham 1998. Results for Taiwan and Korea are entirely new and do not appear in this earlier article.

2. The major exceptions are Brainard (1995a, 1998), which are not efforts directly to test the complementarity/substitutability issue.

3. Brainard 1995a in fact shows that high income levels in countries are associated with both increased multinational sales and increased trade.

4. That is, in any market, an increase in FDI could at the margin reduce home nation exports.

5. In later regression, we also use dummy variables such as language, OECD (whether a country belongs to OECD), and Asia (whether a country belongs to Asia).

6. Since no other country uses the Korean Language other than North and South Korea, and in addition, South Korea has not invested directly in North Korea yet, we drop the language dummy in the South Korea regression.

More From