![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
This article investigates the effect of foreign direct investment (FDI) on the productivity of parent firms for multinational enterprises in Taiwan. The current research specifically examines the potential differences in productivity effect between FDI toward developing (vertical FDI) and developed countries (horizontal FDI) and between electronics and non-electronics firms. Using panel data on Taiwan firms from 2000 to 2005, results obtained using propensity score matching (PSM) show thatmultinational firms experience a higher productivity following their FDI in developing countries. A time lag exists in productivity gain of investment to developed countries, and is relevant only to electronics firms. Employing the generalized method of moment of the panel fixed model to control for problems of endogeneity and unobservable heterogeneity, the empirical finding suggests that productivity effect caused by investing in developing countries remains significantly positive. A lagged productivity-enhancing effect is also found after FDI in developed countries for both electronics and non electronics firms.
Acknowledgements
The authors would like to thank two anonymous referees, Toshiyuki Matsuura and participants of the Western Economic Association (WEA) 8th Biennial Pacific Rim Conference in Kyoto in 2009 for their helpful comments and suggestions on earlier versions of this article.
Notes
1. For example, Blomstrom, Fors, and Lipsey (1997), and Agosin and Machado (2005).
2. Based on aggregate data, Van Pottelsberghe de la Potterie and Lichtenberg (2001) found that outward FDI to R&D-intensive countries can transfer technology abroad and then raise domestic industry productivity.
3. The unusual amounts of FDI reported in some specific years mainly arise from the government's restrictions on China FDI. For a comprehensive review of Taiwan's outward FDI regulation toward China, see Yang, Lin, and Wu (2010)
4. For example, OEM production accounted for approximately 60% of total output in the Taiwanese PC and PC components industry in the mid-1990s. See Chung (1997).
5. FDI also serves as an important channel of knowledge spillover for advanced countries. See Branstetter (2006).
6. Some Taiwanese MNEs invest in both developing and developed countries (mixed FDI). Because the ratio of mixed FDI is minor for total manufacturing firms, this study does not attempt to include this FDI mode into establishing testable hypotheses.
7. There were 20 industry classifications and 682 firms in 2005, and the number of electronics firms accounts for 44%.
8. China is the main destination for Taiwan's outward FDI. However, a clear-cut distinction between vertical and HFDI is hardly possible because China offers low production costs (wages) and a large, expanding market. Owing to the specific operation modes of OEM/ODM and merchanting trade, the main purpose of Taiwan investments in China should concentrate on utilizing low production costs.
9. In the non-electronics sample, pharmaceutical firms are generally thought to be high R&D-intensive. However, Taiwanese pharmaceutical firms produce generic drugs rather than patent drugs, and their R&D intensity reaches only 2.467% that is lower than the sample mean.
10. The TEJ is a commercial company with a fine reputation for collecting and summarizing information for companies listed on the TSE. The TEJ databank is reliable and widely adopted by most universities in Taiwan and in financial sector firms. This databank contains comprehensive information for balance sheets, financial statements, annual reports, etc.
11. It is also possible to compare decisions between investing developing countries and developed countries. However, it is not the focus of this study.
12. To estimate the propensity score and test the balancing hypothesis, Becker and Ichino (2002) developed a seven-step algorithm.
13. For an extensive discussion of matching methods, see Heckman, Ichimura, and Todd (1998).
14. For a detailed derivation of the semi-parametric estimates of TFP, please see the appendix. The STATA codes for the semi-parametric estimate of TFP are also available in Petrin, Poi, and Levinsohn (2004).
15. Most FDI is classified as vertical, which may lead to a more significant effect of VFDI, whereas HFDI plays a minor role.
16. The cost of this flexibility is that GMM estimators are asymptotically efficient in a large class, but are rarely efficient in finite samples; see Hansen (1982).
17. For the construction of the TFP Index, please see the Appendix.
18. Correlation analysis is undertaken to ensure no high multicollinearity among variables. The result is available on request from the authors.
19. To save space, we do not display the results of endogenous tests between FDI modes and productivity. Those results are available upon request.
20. The detailed derivation on the semi-parametric estimates of TFP, please see Levinsohn and Petrin (2003).