Foreign Direct Investments and Domestic Employment of German SMEs: The Moderating Effect of Owner Management

Abstract

Despite considerable research, the relationship between foreign direct investment (FDI) and domestic employment is still inconclusive. The present paper contributes to extant research by providing differentiated findings on the FDI and domestic employment relationship. We examine how horizontal and vertical FDI impact domestic employment and how the effects are contingent on the ownership structure of the firm. Testing hypotheses on a dataset of 1,079 erman small and medium‐sized enterprises (SMEs) shows that horizontal FDI has a positive influence, whereas vertical FDI has a negative influence on domestic employment. Furthermore, domestic downsizing due to the establishment of foreign production subsidiaries is less severe in owner‐managed SMEs.

Notes

13. Consistent with previous literature that analyzed the moderating influence of family ownership and family management on the likelihood of downsizing (e.g., Block 2010; Stavrou, Kassinis, and Filotheou 2007), we expect owner management only to attenuate negative employment effects in the course of vertical FDI—not to amplify positive employment effects in the course of horizontal FDI. The theoretical rationale is the long‐term orientation of owner‐managed firms, which on the one hand reduces the likelihood of short‐term downsizing but on the other hand also avoids extensive hiring. In fact, previous research suggests that firms with long‐term orientation are rather choosy in the process of hiring new employees because their intention is to hire for a career, not a temporarily job (Miller and Le Breton‐Miller 2006).

14. Because a definition including firms with up to 500 employees is certainly arbitrary, we conducted robustness checks with varying SME definitions. We applied the SME definitions according to the European Union (employees <250 and annual revenues ≤50 Mio. €) and the German Institute of SMEs (1997) (employees <500 and annual revenues ≤50 Mio. €). Our findings remained robust regardless of the chosen SME definition.

15. To additionally assess whether the correlations between “number of employees (log)” and “annual sales” (= 0.531*) as well as “share of foreign sales to total sales” and “international growth orientation” (= 0.585*) were problematic, we conducted robustness checks. We reran our models as illustrated in Table 4, excluding these variables (consecutively and simultaneously). Our statistical results remained robust in all models. Hence, we are confident that our findings are not biased due to the relatively high correlations between these sets of variables.

16. We further checked the robustness of our results by applying a binary logistic regression model with extension (coded “1”) versus reduction (coded “0”) of firms' domestic employment as dependant variable. The findings were robust compared with the results obtained from the ordered logit. However, applying binary logistic regression, the sample size was considerably reduced to n = 733 because 339 SMEs had indicated that their domestic employment remained unchanged in the course of FDI and hence were excluded from the binary logistic regression model.

17. Although the interaction between “foreign market mode: production subsidiary” and “owner management” is statistically significant and adds a significant amount of χ² value to the model fit, it only leads to a slight increase of roughly 1 percent in R² to Models 5 and 6 compared with the direct‐effects model (Model 2). However, Champoux and Peters (1987) report that interaction terms typically do not account for more than 1–3 percent of the variance. Moreover, Evans (1985) and Chaplin (1991) assent that moderator effects are so difficult to detect that even interaction terms accounting for not more than 1 percent of the total variance should be considered as important as long as they are discovered in the context of a well‐articulated theory. The statistical reasoning for the usually limited explanatory power of interaction terms is outlined in detail in Aiken and West (1991), McClelland and Judd (1993), and Whisman and McClelland (2005).

18. We thank an anonymous reviewer for reminding us to check whether our hypothesized moderator variable “owner management” was instead a mediator variable intervening the influence of the three subdimensions of vertical FDI on domestic employment. We applied the procedure suggested by Sobel (1982) to test the significance of the intervening variable effect by dividing the estimate of the intervening variable effect by its standard error and comparing this value with a standard normal distribution. Because this test yielded only nonsignificant results, we are confident that “owner management” is indeed a moderator variable, as originally suggested.

Additional information

Notes on contributors

Florian B. Zapkau

Florian B. Zapkau is a research assistant at the University of Düsseldorf.

Christian Schwens

Christian Schwens is a professor of management at the University of Düsseldorf.

Rüdiger Kabst

Rüdiger Kabst is a professor of international business at the University of Paderborn.