Rights, governance, and foreign direct investment: an industry-level assessment

Abstract

Theories of multinational enterprises emphasize that foreign direct investment (FDI) is undertaken in different industries for different reasons, yet studies of the effects of rights and governance on FDI generally rely on aggregate-level FDI data. This paper evaluates US FDI outflows to 15 industries (eight manufacturing, seven non-manufacturing) in 54 countries in a linear dynamic panel data gravity FDI model using a ‘system’ generalized method of moments estimator and several widely used rights and governance indexes. At the aggregate level, we estimate that stronger rights and governance have a positive effect on FDI, consistent with most prior studies. At the industry level, we estimate larger positive effects of rights and governance on FDI for service than manufacturing industries, particularly for the information and the finance and insurance industries.

Keywords:

• Foreign direct investment
• democracy
• human rights
• governance
• gravity model
• GMM

JEL Classifications:

• D72
• F23
• J51

Notes

1. The sole study surveyed by Asiedu and Lien (2011) finding a negative effect of democracy on FDI is Li and Resnick (2003), but Jakobsen and de Soysa (2006) replicate Li and Resnick’s analysis and find that the relationship reverses from negative to positive upon adding additional countries to the sample and logging FDI inflows. Note that although Asiedu and Lien (2011) report that Adam and Filippaios (2007) find a positive effect of democracy on FDI, the findings of the latter are more nuanced. That is, Adam and Filippaios find coefficient estimates of opposite sign on the Freedom House (2011) civil liberties and political rights indexes for some samples of countries, though this is based on the simultaneous inclusion in their regressions of these two highly correlated indexes.

2. While Blanton and Blanton (2009, 2012) are based on industry-level data, they only present aggregate-level results. Kucera and Principi (2014) apply the same approach as the present paper using the Freedom House (2011) civil liberties and political rights indexes.

3. Democracies do indeed tend to pay higher wages, accounting for differences in labor productivity (Rodrik 1999).

4. Supporting this view, Blanton and Blanton (2012) show that there is a somewhat stronger positive correlation between education and democracy than between education and human rights, based on the measures used in their study.

5. Blanton and Blanton (2012) also argue that considerations of ‘dynamic inconsistency’ are especially important for capital-intensive industries with a risk of expropriation, specifically the mining and oil and services industries (the last including finance and wholesale trade as well as other services). As such, MNEs in these industries have a particularly high stake in whether democracies or autocracies are better able to provide policy stability.

6. Based on country case studies of democracy and foreign investment in the oil industry, Bayulgen (2010) addresses why both democracies (Norway) and autocracies (Azerbaijan) have been successful in attracting large amounts of foreign investment. She argues that this results because each is able to provide different things that are attractive to foreign investors: policy stability in the case of democracies and preferential treatment in the case of autocracies. In contrast, what Bayulgen describes as ‘hybrid’ countries, like Russia, are unable to effectively provide either.

7. In a more recent paper, Kleinert and Toubal (2013) also provide theoretical foundations for the distinction between production-oriented and distribution-oriented FDI, though not in a gravity model context.

8. See Kennedy (1998, 225, 226) for a fuller explanation of such use of dummy variables.

9. We do not use earlier years because of a break in the series, a result of which data from 1994 on are not directly comparable with data from earlier years, as described by Bach (1998).

10. Note that other industries includes a very large and diverse set of activities: agriculture, forestry, fishing, and hunting; construction; utilities; retail trade; transportation and warehousing; real estate rental and leasing; administration, support, waste management, and remediation services; health care and social assistance; accommodation and food services; and miscellaneous services.

11. That is, other manufacturing (not elsewhere classified) includes: tobacco products; textile products and apparel; lumber, wood furniture, and fixtures; paper and allied products; printing and publishing; rubber products; miscellaneous plastic products; glass products; and stone, clay, and non-metallic mineral products.

12. Freedom House’s political rights and civil liberties indexes have also been widely used in this literature (Freedom House 2011). See Kucera and Principi (2014) in this regard.

13. Cf. Bollen (1993), Poe and Tate (1994), and Munck and Verkuilen (2002) for valuable assessments of a range of democracy indexes.

14. For further information, see: https://services.fsd.uta.fi/catalog/FSD1289?lang=en&study_language=en. Accessed 3 September 2014.

15. For further information, see: http://www.prsgroup.com/wp-content/uploads/2012/11/icrgmethodology.pdf. Accessed 3 September 2014.

16. Note that our use of GMM satisfies all the conditions specified by Roodman (2006). Following Baum, Schaffer, and Stillman (2003), the presence of heteroscedasticity was tested in each of our specifications with a Pagan–Hall test, and the null hypothesis of homoscedasticity was consistently rejected.

17. These high levels of statistical significance are typical of gravity FDI models with large numbers of observations, for example Busse (2003) and Asiedu and Lien (2011).

18. This is calculated as 100 × (e^β1 – 1) multiplied by the difference between the index values for two countries.

19. Standard errors for industry-level coefficient estimates are available from the authors on request.

20. Regional dummy variables are based on the nine regions described in the Appendix, with the dummy variable for developed countries excluded. Note that dropping developed countries from the sample results in high p-values for the Hansen J test, about which Roodman cautions (2006).

21. For literature on the effects of FDI and global market liberalization more broadly on democracy and human rights, see Li and Reuveny (2003) and Kim and Trumbore (2010).

22. Aggregate-level results are also fairly robust when time-variant data are constructed as two-year (rather than three-year) averages. For example, coefficient estimates are positive and statistically significant at the level of 1% for the CIRI empowerment rights index, 5% for the Polity2 index and CIRI physical integrity index, and 10% for the Vanhanen democracy index. For all three regressions using two-year averages, however, we reject the null hypothesis for the Hansen J test, and so we do not regard them as particularly reliable. (We do not have enough years of data in our sample to construct comparable estimates based on four-year averages.) We also used the ‘collapse’ option in the xtabond2 program, which reduces the number of instruments by collapsing the matrix of GMM-type instruments into a column. But for both the full sample of countries and the sample without developed countries, this results in unrealistically high p-values for the Hansen J test (>0.9).