The Impact of Regional Integration on Insider and Outsider FDI

Note that there are different degrees of regional integration (e.g., Baldwin and Venables 1995) and that the term is used differently in the literature (e.g., Altomonte 2007; Blomström and Kokko 1997; Fratianni and Oh 2009; te Velde and Bezemer 2006). In our study, RIAs refer to those areas of adjacent countries that are characterized not only by trade liberalization but also include investment liberalization and/or other institutional changes (Blomström and Kokko 1997). The degree of investment liberalization and the type of additional institutional changes are unique to each RIA. The major provisions of the three RIAs in our sample are summarized in the Appendix. Our selection of RIAs is consistent with Ethier’s (1998) characterization of the New Regionalism trend. He explicitly discusses the EU and NAFTA as two of the regions that fit his description; ANZCERTA also fulfills his criteria for a ‘home’ region. Note that in all regions some or all of the new member countries had already some free trade agreements with each other. In this paper, we do not test the effect a free trade agreement on FDI but rather that of regional integration in the sense proposed by Ethier (1998), Rugman and Verbeke (2004, 2005), and Wolf et al. (2008).

Note, the literature also classifies downstream FDI as market-seeking, and upstream FDI as efficiency-seeking and resource-seeking (e.g., Dunning 1993).

Changes in FDI stock may be used as an alternative dependent variable. Stephan and Pfaffmann (2001) provide a discussion of data limitations for both FDI stock and flow data. Some recent studies that use the FDI flow data are Buckley et al. (2007b), Thomas and Grosse (2001), and Wu (2006).

We dropped Belgium and Luxembourg from the sample because the information on these two countries is often combined and the individual country effects cannot be determined. Information on the member countries of the Central European Free Trade Agreement was often missing. Therefore we dropped these countries as well from the final sample. Due to missing observations the panels are not balanced.

Exports and imports are highly correlated. Therefore, we use the sum of exports and imports to measure the bilateral trade intensity.

The world governance indicators and the world competitiveness report are not available for the entire time horizon of our sample period. We extrapolated the missing observations for the legal and regulatory environment and we extra- and intrapolated the missing observations for the world governance indicators.

We used EU 1992 to take into account the deepening of the European Union after 1992; we thank an anonymous referee for the suggestion.

Although we did not posit a sign for Geographic Distance, Daude and Stein (2007) and Balasubramanyam et al. (2002) found a negative and significant coefficient for it.