Contracting with Whom? The Differential Effects of Investment Treaties on FDI

ABSTRACT

Under what conditions can governments use international commitments such as Bilateral Investment Treaties (BITs) to attract foreign direct investment (FDI)? Although numerous studies have attempted to answer this question, none considers how investment treaties may have heterogeneous affects across industry. I argue BIT effect is strongest when the obsolescing bargaining problem between firms and governments is most protracted, namely, when FDI relies on strong contracts between firms and states. Using a time series cross-sectional data set of 114 developing countries from 1985 to 2011, I find BITs are associated with increases in infrastructure investment, an industry particularly reliant on the sanctity of government contracts, but not with total FDI inflows. Moreover, BITs with strong arbitration provisions display the strongest statistical effect on infrastructure investment, while BITs that do not provide investors with such protections are not associated with increased investment. My results have implications for both scholarship on the relationship between governments and multinational firms as well as for the study of international institutions more broadly. To properly ascertain the effects of international treaties and institutions, scholars should consider not just whether institutions constrain or inform—or matter at all—but also the extent to which the targets of institutions have heterogeneous responses to them.

KEYWORDS:

• Bilateral investment treaties
• foreign direct investment
• infrastructure
• international political economy

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ These arguments stand in contrast to an earlier emphasis on autocratic governments’ alliances with foreign capital (Evans and Gereffi 1982; O’Donnell 1988; Oneal 1994).

² See Elkins, Guzman, and Simmons (2006) for an extended history of BIT diffusion.

³ Performance requirements specify the percentage of local sources a firm must use in production.

⁴ Many BITs include national security carve-outs that allow host governments to retain a significant degree of policy flexibility during economic crises. For example, Argentina was able to successfully defend several of the discriminatory actions it took against US-based firms in the aftermath of its 1999 financial crisis. The legal basis of its defense rested on a security and public order carve-out in its BIT with the US (Blake 2013:804).

⁵ Emma Lindsay, personal interview, September 15, 2015.

⁶ Included in my cases are OECD countries for which data are available as well: Chile, Mexico, and Turkey.

⁷ The database limits coverage to sectors characterized by monopoly or oligopoly and therefore does not include more competitive sectors such as airlines and gas production.

⁸ This measure comes from UNCTAD’s FDI database.

⁹ These capital exporters include: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, Sweden, Switzerland, the United Kingdom, and the United States. Source: Yackee (2008).

¹⁰ Source: Graham (2015).

¹¹ All control variables come from the World Development Indicators database unless noted otherwise.

¹² Others proxy for skilled labor with Tertiary Enrollment. Including this variable constricts my sample size considerably, so I run models without this control. In the online appendix, I show that my central findings are robust to including this measure.

¹³ These include: Population, GDP per capita, Trade Openness, and Inflation.

¹⁴ Unlike some other empirical studies of BIT effects, I do not control for Political Risk, as measured by the Political Risk Services’ International Country Risk Guide (ICRG). I choose to omit Political Risk in my analysis for logistical as well as theoretical concerns. First, the variable displays approximately 50% missingness in my sample. This creates estimation problems and unduly restricts my sample. Second, the ICRG’s Political Risk index is a composite of many political factors, and one component of the index explicitly accounts for the presence of BITs. (Personal correspondence with ICRG staff, Thomas L. Gerken, December 4, 2012.) Thus, using this measure as a control unduly attributes correlation to Political Risk instead of BITs.

¹⁵ Results provided in the online appendix.

¹⁶ These ranges are calculated from coefficient estimates of both lagged dependent variable with fixed-effects models and GLS models.

¹⁷ Results reported in the online appendix.