This paper assembles evidence on the economic impact and the geographic dispersion of European Foreign Direct Investment in the United States. Using 1987 data, Shaver (1998) finds that FDI differs in its location pattern relative to domestic investment. This paper, relying on 1992 data, finds similarities and differences in location patterns by source country. It also finds a puzzling performance difference between European and Japanese FDI in the U.S. Initial econometric evidence experiments points to country-specific responses of FDI to local labour costs. The paper also raises some unresolved puzzles, which should stimulate further research.
*The authors are, respectively, Professor and Assistant Professor, Faculty of Business, University of Alberta, Edmonton, Alberta, T6G 2R6. They are grateful to the Centre for International Business Studies for its support, and to Tat Fan and Bettina Witlneben for their assistance with the database.
Introduction
As the OECD observes, FDI by firms into and from OECD countries has been setting new records recently, driven in particular by a number of major cross-country mergers and acquisitions, especially between American and European firms. The forces driving the strong growth in FDI are diverse, involving technological change, privatization, liberalization of trade and investment and a host of industry- and firm-specific factors that make modeling the FDI process an extremely complex enterprise.
Given the importance of FDI, in terms of its effects on growth and technology transfer, researchers and policymakers are equally interested in a better understanding of this phenomenon. While the release, in 1997, of a major new data set on FDI in the U.S. was a response to the perceived threat of foreign takeovers, these data also opened up opportunities to study FDI in more detail. An important paper by Shaver (1998), for example, found the location pattern of foreign investors to differ significantly from that chosen by U.S. firms and advanced some reasons for why this should be the case. In this paper we use the same data source to assemble evidence on the economic impact, in terms of sales, employment and payroll for French, German and British FDI in the U.S. We further describe the location patterns of manufacturing FDI from these countries and compute a simple measure to gauge European firms’ performance against that of Japanese and Canadian manufacturing outlets in the U.S. Finally, some initial econometric evidence is presented to explain the location pattern of European FDI. Thus the common thread of this paper is to pursue how and why European manufacturing firms differ in their American operations. Of necessity, our conclusions are based on a rather high level of aggregation, namely, the SIC-classifications for manufacturing (SIC 201-399), with the result that the findings should be taken as puzzles in search of solutions.
The first section of this paper provides a backdrop by showing the levels of overall manufacturing FDI in the U.S. and the shares of major European countries. The emphasis is on presenting the relative overall economic impact of major European sources of FDI. The second section focuses on the manufacturing sector and pursues the pattern of FDI location by state of destination, both in terms of employment impact and value of shipments. The third and fourth sections focus on differences in performance and location patterns of FDI in the U.S. from the various source countries. Thus this paper follows up Shaver’s (1998) finding that FDI is different from domestic investment by asking whether or not foreign investors from different source countries are different from each other.
The data set defines FDI as at least 10 percent ultimate beneficial ownership by a foreign resident. Linking Bureau of Economic Analysis data on FDI at the company or enterprise level with the Census Bureau data at the plant or establishment level for all U.S. companies resulted in a comprehensive data base on FDI by country of origin, state of destination and by SIC classification at the three and four digit levels. Available for 1987 and with more detail for 1992, it gives the most detailed snapshot to-date of the geographic dispersion and industry characteristics of FDI in the U.S. Previously hidden in aggregate figures on the U.S, International Investment Position, the new information allows a much more detailed description of the economic impact of FDI in general, and European FDI in particular, on the U.S. economy.
Table 1 provides the backdrop for our investigation, the overall impact of FDI on the U.S. in terms of the number of establishments, the number of employees, their payroll, and their shipments (or sales) – though the latter may be to other establishments in the same corporate entity.
Number of Establishments |
Number of Employees |
Payroll In Mill. $ |
Sales Mill. $ |
|
Mining |
1,604 |
120,782 |
5,407.4 |
30,508.9 |
Constructions |
1,220 |
93,322 |
3,508.8 |
19,870.3 |
Manufacturing |
12,781 |
2,004,947 |
69,628.0 |
435,809.3 |
Transportation and Utilities |
3,905 |
231,638 |
7,208.4 |
23,578.7 |
Wholesale |
18,791 |
513,012 |
21,084.2 |
513,277.5 |
Retail |
37,892 |
853,190 |
12,283.5 |
91,657.1 |
Finance, Insurance and Real Estate |
11,541 |
401,018 |
18,467.1 |
190,690.7 |
Services |
15,194 |
722,775 |
17,359.8 |
49,647.1 |
Total |
102,958 |
4,944,157 |
155,032.3 |
1,354,724.3 |
*The source for all the tables and data quoted in the text is "Foreign Direct Investment in the United States: Establishment Data for 1992," U.S. Department of Commerce, Bureau of Economic Analysis and Bureau of the Census, Release Date June 1997. This source also contains data for 1987, albeit less detailed.
With a payroll of $155 billion, involving nearly 5 million employees and $1.35 trillion in output, FDI clearly emerges as a major contributor to the US economy. Of course, more than half of 103,000 establishments are in retail and wholesale operations, averaging only slightly more than 20 employees per establishment. We therefore focus our further efforts on the 12,781 manufacturing establishments: while still covering diverse firms, this focus on SIC classification at the two digit level gives a sufficiently large number of observations in the industry classification which figures prominently in international trade. Manufacturing establishments are also the object of considerable efforts at the state and local level to attract FDI. Table 2 shows that three major European countries accounted for 41% of manufacturing FDI in the U.S. in 1992 in terms of number of employees and 41.4% in terms of payroll.
(% of total FDI establishments)
Country |
% of establishments1,4 |
% of employees2 |
% of payroll3 |
France |
9.4 (8.5) |
8.6 |
8.8 |
Germany |
9.9 (7.4) |
11.7 |
12.3 |
U.K. |
25.9 (21.8) |
20.7 |
20.0 |
Switzerland |
5.1 (5.1) |
7.1 |
7.8 |
Netherlands |
4.5 (6.5) |
5.6 |
5.9 |
Japan |
13.8 (16.7) |
17.1 |
17.0 |
Canada |
11.6 (13.6) |
12.5 |
13.1 |
The bracketed numbers in the Table imply that the major European countries are numerically relatively strongly concentrated in manufacturing, their comparative advantage sector, whereas Japan and Canada are less manufacturing-oriented in their FDI than the average foreign investor.
Table 3 reports the average number of manufacturing employees for FDI from the various source countries, as well as the average employment over all FDI establishments. It shows German establishments on average twice as large as Japanese, whereas French and British establishments are smaller than the average foreign plant in terms of their number of employees.
Country |
Overall – FDI |
Manufacturing – FDI |
France |
- |
144 |
Germany |
70 |
185 |
U.K. |
44 |
125 |
Japan |
46 |
91 |
Canada |
43 |
169 |
Overall Average |
48 |
157 |
These simple facts suggest that there are significant differences in the size and/or labour-intensity of foreign-owned establishments in the U.S. As Shaver found FDI in general to be different in its location pattern from U.S. investment, the new data set shows that, furthermore, the size of FDI, as proxied by employment, differs by source country. The surprising fact here is the small size of Japanese establishments relative to German and the average of all foreign investors. Future work will have to address whether the industry-specific differences lead to this result or whether different entry-scales are chosen in the same industry by source country firms from different countries.
Shaver (1998) has shown that FDI follows a distinctly different location pattern from domestic investment: FDI is disproportionately strong in coastal states. We add to this finding by breaking out country-specific destination states to determine whether all manufacturing FDI is similar or whether there are again distinct differences among establishments from different source countries. Our motivation is that if there are differences, these may lead to a fuller understanding of the complexity of the FDI process.
The five states attracting the largest number of foreign-controlled establishments were, in order, California, Texas, New York, Florida and Ohio. In terms of employment California ranked ahead of New York, Texas, Illinois and New Jersey; and with respect to the total payroll, California led New York, Texas, New Jersey and Illinois. This confirms Shaver’s findings. Table 4 gives the rankings for manufacturing FDI by employment and value of shipments. It shows that the interior states of Ohio, Illinois and Pennsylvania are significant attractors of FDI in manufacturing, pushing New York and New Jersey into lower rankings.
Rank |
Employment (# of employees) |
State |
Rank |
Value of Shipments ($ billions) |
1 |
189,710 |
California |
1 |
39.2 |
2 |
122,716 |
Ohio |
3 |
30.1 |
3 |
120,917 |
Illinois |
4 |
27.0 |
4 |
118,187 |
Pennsylvania |
7 |
19.6 |
5 |
114,330 |
North Carolina |
5 |
25.0 |
6 |
105,159 |
Texas |
2 |
33.2 |
7 |
99,779 |
New Jersey |
- |
- |
8 |
94,378 |
New York |
- |
- |
- |
- |
Indiana |
8 |
17.6 |
A further breakdown of Manufacturing FDI into destination states in terms of value of shipments and by source country yields Table 5.
$ billions in Manufacturing Shipments |
|||||
State |
France |
Germany |
U.K. |
Japan |
Canada |
North Carolina |
1.6 |
5.0 |
8.0 |
2.2 |
4.9 |
Texas |
2.0 |
3.7 |
* |
2.9 |
5.3 |
South Carolina |
2.1 |
3.3 |
* |
* |
* |
New Jersey |
* |
2.9 |
3.3 |
* |
2.6 |
Illinois |
1.5 |
2.6 |
4.5 |
5.2 |
2.6 |
Michigan |
1.5 |
2.1 |
* |
6.0 |
* |
California |
2.4 |
2.1 |
5.7 |
12.9 |
2.0 |
Ohio |
* |
1.9 |
6.7 |
13.3 |
* |
Pennsylvania |
2.5 |
1.9 |
6.4 |
* |
2.7 |
New York |
1.3 |
1.6 |
4.3 |
* |
3.0 |
Georgia |
* |
* |
5.4 |
2.7 |
1.7 |
Indiana |
2.7 |
* |
* |
6.9 |
* |
Tennessee |
* |
* |
* |
6.3 |
3.5 |
Washington |
* |
* |
1.9 |
* |
|
Alabama |
1.7 |
* |
* |
* |
|
Virginia |
* |
* |
* |
2.2 |
*Not in top 10 locations for country in question.
Table 5 attempts a ranking of states according to the value of output produced by source country manufacturing FDI. Germany’s FDI output is distributed quite differently from Japan’s, and the U.K.’s FDI output is distributed over the same states as Germany’s but with very different values, resulting in a very different rankings of these states. Canadian and German manufacturing FDI shipments exhibit a similar pattern, but the German and French FDI shipments are again differently distributed over the recipient states. In short, while there are some similarities there are also significant differences in terms of the location pattern of FDI shipments by source country establishments. Not only are foreign firms as a group locating their establishments differently from U.S. firms, there are also differences among the location patterns of FDI establishments by source country.
In light of the diverse distribution pattern observed above, it is appropriate to attempt an explanation in terms of FDI theory by resorting to a country-by-country approach. The OLI framework (Caves 1982, Dunning 1979, Rugman 1981) suggests a combination of state characteristics, industry characteristics and firm characteristics will determine where in the U.S. FDI from each source country will occur. Having narrowed the focus to manufacturing, which state characteristics are likely to attract FDI? Gravity models would include distance to major industrial concentrations and population centres but spatial concepts become difficult to apply in the context of a sample of U.S. states where small states with small populations can have proximity to both small and large states with and without concentrations of industries and consumers. Access to special skills and important natural resources is also difficult to make operational at the state level. And informational barriers in the U.S. are believed to be low, generally.
Shaver (1998) found that as a group foreign firms were more responsive to unionization, right to work laws and coastal locations than U.S. firms. For FDI from some countries unionization and attendant high wages may increase the adjustment costs in the U.S. Alternatively, firms from particular countries may see their advantage in the U.S. productivity based on lower wage costs coupled with extensive training. Given the need for firm-specific training foreign investors may view American employees as uniform in skill across states, so that location, at the margin, is chosen where wages are lowest. In addition, the tax burden has been hypothesized as one of the environmental variables that has a bearing on the state chosen for FDI. For these reasons, analysis at the state lebel is the appropriate choice.
Accordingly, for manufacturing FDI in the U.S. by firms from a particular country, we hypothesize that
FDIij = fj(tax burdeni, unionizationi, wage costsi) , (1)
with all three independent variables hypothesized to be negatively related to the level of FDI in state i. In operationalizing these theoretical expectations for empirical testing, a number of data constraints were encountered starting with the proxy variable for the level of FDI in the various states.
The data set does not provide the market or book value of the foreign controlled establishments, so shipments/sales or number of employees are the possible proxy variables. As FDI occurs over time, the current (1992) level is, of course, the result of decisions over time. Explaining this stock by current environmental conditions is, therefore, making a strong assumption about the speed of adjustment of FDI to a changing environment.
To lessen heteroscedasticity the first of these variables was scaled by Gross State Product. The assumption, then, is that, on average, the value of shipments, as a percentage of Gross State Product, reflects the value of FDI assets, just as Gross State Product reflects the total capital employed in each state.
The second proxy chosen for country-j FDI at the state level is employment by j country establishments relative to total employment in each state. Using employment as a proxy for investment value is more problematic because capital can be substituted for labour, especially when the latter is too expensive. To the extent that foreign firms are more responsive than U.S. firms, the hypothesized negative relationship between FDI and wage costs should prevail.
The independent variables for the regressions reported in Table 6 are, respectively, net corporation income tax collected relative to Gross State Product, the percentage of employees covered by unions in each state, and the average hourly earnings of production workers in each state relative to the national average.
The unionization and average hourly wage variables, not unexpectedly, are highly correlated (0.76), so that unionization is omitted as a separate independent variable. The results are shown in Table. 6.
(t-values in parentheses)
Independent Variables |
||||
Dependent Variables |
Constant |
Tax Burden |
Wage Cost |
R2/F |
France: shipments/GSP |
0.01 (1.34) |
0.74 (1.11) |
-0.01 (-0.82) |
0.04 0.72 |
France: employees/total state employment |
0.01 (1.47) |
0.18 (1.18) |
-0.01 (-0.94) |
0.04 0.83 |
Germany: shipments/GSP |
0.02 (1.81) |
0.40 (0.54) |
-0.01 (-1.22) |
0.04 0.76 |
Germany: employees/total state employment |
0.01 (1.04) |
0.18 (1.11) |
-0.01 (-0.47) |
0.03 0.63 |
U.K.: shipments/GSP |
0.03 (2.58) |
0.65 (0.77) |
-0.02 (-1.71) |
0.07 1.51 |
U.K.: employees/total state employment |
0.01 (2.52) |
0.18 (1.06) |
-0.01 (-1.43) |
0.06 1.23 |
Japan: shipments/GSP |
-0.00 (-0.13) |
1.70 (1.17) |
0.01 (0.42) |
0.05 0.38 |
Japan: employees/total state employment |
0.01 (0.07) |
0.48 (2.12) |
0.00 (0.24) |
0.12 2.71 |
Canada: shipments/GSP |
0.03 (3.93) |
0.54 (0.83) |
-0.03 (-2.98) |
0.20 4.45 |
Canada: employees/total state employment |
0.05 (3.14) |
0.15 (1.10) |
-0.01 (2.06) |
0.10 2.16 |
The results, by and large, show that, with the exception of Canada and possibly the U.K., investors were not sensitive to the relative wage costs in the various states, and the tax variable is consistently insignificant.
Canadian FDI appears different from European and Japanese FDI in its sensitivity to relative wage costs. UK FDI appears more similar to Canadian FDI on this dimension, though the wage cost coefficient is, at best, marginally significant. The results suggest that Canadian and UK FDI are similar, while continental European and Japanese FDI are not cost sensitive.
The R-square and F-values are indicative of the fact that the FDI process is a much more complex issue than is reflected in out relatively simple model, which only attempts to link country firms to environmental conditions. For example, incentives provided by states and local administrations for larger investments cannot be considered in this context, and other shortcomings of the variables were pointed out earlier.
Thus, if anything can be inferred, it is that Canadians are different from continental European and Japanese investors in their location of FDI in the U.S., and the reason for their seeming sensitivity may stimulate further inquiry. One possible implication is that FDI attraction policies may have to be targeted in a differentiated manner if FDI from different countries shows different sensitivity to environmental conditions, though differences in technology or product structure may ultimately turn out to be responsible for the observed phenomenon.
The data enabled an additional line of inquiry that led to some puzzling results. If country of origin of manufacturing FDI matters for location – and there were suggestions in our previous findings that this is the case - are there also differences in the performance of FDI from different (groups of) countries?
A proxy for the labour-productivity of FDI from different countries was constructed by dividing the total manufacturing shipments of each source country’s U.S. establishments by the total number of their employees. Table 7 summarizes the findings.
Country |
Labour Productivity of FDI in the U.S. ($/employee) |
France |
18,911 |
Germany |
19,105 |
U.K. |
19,707 |
Japan |
24,238 |
Canada |
21,861 |
Average for all foreign establishments |
21,737 |
It appears that Japanese-controlled establishments in the U.S. concentrate on the production of more capital-intensive or more human-capital intensive products, especially when compared to the three major European investors in the U.S. Our final finding, then, points out a possible future research direction: What are the reasons underlying the relatively low labour productivity of European FDI? Is European FDI concentrated in medium and low-tech manufacturing industries? Are there performance differences revealed in these numbers?
One possible explanation is provided by the growth of FDI in the U.S. between 1987 and 1992 (Table 8).
Year |
All Countries |
Germany |
U.K. |
Japan |
Canada |
1987 |
66,878 |
5,916 |
16,542 |
5,284 |
12,251 |
1992 |
102,958 |
7,652 |
22,400 |
17,178 |
14,133 |
1992/87 |
1.54 |
1.29 |
1.35 |
3.25 |
1.15 |
While European FDI increased moderately and Canadian FDI increased modestly (despite or because of Canada-U.S. Free Trade), Japanese FDI soared, and much new capacity and technology were brought on stream (though some FDI is the result of mergers, as well). It is this recent substantial additional FDI from Japan that could explain the comparatively less impressive performance of European establishments. Still, the observed phenomenon is in need of further explanation.
V. Summary and Conclusions
This study uses a new and comprehensive data base to explore country characteristics of manufacturing FDI in the U.S. Based on a 10% ownership threshold, in 1992 FDI in the U.S. comprised over 100,000 establishments, 12,781 of them in manufacturing. Their total employment was nearly 5 million, with more than 2 million in manufacturing earning nearly $70 billion. Thus, the impact of FDI on the US economy is significant.
German, French and U.K. FDI, in terms of payroll and number of employees comprised just over 40% of total FDI, with Britain edging out Japan as the biggest investor in the U.S. As for size of the average foreign establishment, German investments tended to be larger than French and British, while the overall Japanese establishment was less than half the size of the average German one.
We noted that European establishments appear clustered distinctly from each other, a finding that – while not rigorously documented – sheds further light on Shaver’s (1998) finding of FDI being distinct from domestic investment in its location pattern. In an attempt to further explore the location of FDI over the U.S., the regression results yielded a strong hint of sensitivity to wage costs in the case of Canada and a weak hint in the case of U.K. FDI. In general, however, the stock of FDI is not well explained by the attracting state’s business tax burden and wage costs. More work at a lower level of aggregation within manufacturing is necessary to shed more light on the question of whether all foreign firms are equal or whether national or continental differences can be ascertained. Some further experimentation with dummy variables for states with coastal locations and unitary tax is already underway.
Finally, we found distinct differences in labour productivity of foreign controlled establishments that suggest a closer look at their causes. Given the complexity of the FDI phenomenon, it should not be surprising that our explanations are tentative and merely suggestive. The release of the 1997 survey of FDI in the U.S. will make a fellow-up possible and will open the opportunity to pursue the changes that have occurred since 1992, thus adding a time-series dimension to the bird’s eye view presented in this paper.
Caves, Richard E., Multinational Enterprise and Economic Analysis, Cambridge University Press, Cambridge, 1982.
Dunning, John H., "Explaining Changing Patterns of International Production," Oxford Bulletin of Economics and Statistics, 41, 1979, pp. 261-95.
OECD, "Recent Trends in Foreign Direct Investment," Financial Market Trends, No. 73, June 1999, pp. 109-126.
Rugman, Alan, Inside the Multinationals: The Economics of Internal Markets, Croom Helm, London, 1981.
Shaver, J. Myles, "Do Foreign-Owned and U.S.-Owned Establishments Exhibit the Same Location Pattern in the U.S. Manufacturing Industries?." Journal of International Business Studies, 29, 1, 1998, pp. 469-91.