Is Foreign Direct Investment “Gender Blind”? Women's Rights as a Determinant of US FDI

ABSTRACT

The impact of women's rights on a country's competitiveness in the global economy is a source of contention. While educational opportunities for women, as well as political empowerment, are linked to a variety of positive outcomes, the impact of economic rights is mixed. Toward better understanding these issues, we focus on the role of women's rights in attracting foreign direct investment (FDI). Though foreign capital plays a key role in the development strategies of many countries, and many of the growth areas in FDI rely heavily on women's labor, extant literature on the determinants of FDI largely ignores gender. To gain insight into these issues, we examine the impact of women's political, economic, and educational rights across four different types of US FDI into the developing world. We find a mixed relationship between women's rights and FDI that varies across industrial sectors.

KEYWORDS:

• Political economy
• human capital
• women and the economy
• trade liberalization

JEL Codes:

• F23
• J16
• P50

NOTES ON CONTRIBUTORS

Robert G. Blanton is Professor of Political Science at the University of Alabama at Birmingham. His main research focus is international political economy, particularly international trade and foreign direct investment. His work has appeared in such journals as Journal of Politics, Journal of Peace Research, International Studies Quarterly, International Interactions and Political Research Quarterly.

Shannon Lindsey Blanton is Professor of Political Science and Dean of the Honors College at the University of Alabama at Birmingham. Her research is in the areas of human rights and US foreign policy, and her work has appeared in the American Journal of Political Science, Journal of Politics, International Studies Quarterly, Journal of Peace Research, International Interactions, and elsewhere. She is also a co-author (with Charles W. Kegley) of World Politics: Trend and Transformation (Cengage Publications, 2015).

ACKNOWLEDGMENTS

The authors would like to thank the editors and anonymous reviewers, as well as Dursen Peksen and Nicole Detraz, for their comments on earlier versions of this manuscript, as well as Roxanne Buckman, Clinton Thompson, and Drew Wagstaff for their research assistance. Any errors are the sole responsibility of the authors.

Notes

1 Data are obtained from the “interactive tables” for Balance of Payments and Direct Investment Position Data (http://www.bea.gov/international/reference_interactivesystem_01.htmhttp://www.bea.gov/international/reference_interactivesystem_01.htm).

2 As FDI stocks are valuated in historical cost terms – that is, the value of these assets when they are purchased – it is not possible to construct flow measures by subtracting the stock values across years (Blanton and Blanton 2009).

3 Negative or zero values are reset to one before logging. While this truncates the actual range of variation, there are very few instances of negative FDI stock. In most sectors it was less than five percent.

4 Our sample includes any country that was not an OECD member throughout the period covered here. The countries analyzed here are Argentina, Brazil, Chile, China, Colombia, Costa Rica, Czech Republic, Dominican Republic, Ecuador, Egypt, Guatemala, Honduras, Hungary, India, Indonesia, Israel, Jamaica, Malaysia, Mexico, Nigeria, Panama, Peru, Philippines, Poland, Russia, Saudi Arabia, Singapore, South Africa, South Korea, Thailand, Trinidad and Tobago, United Arab Emirates, and Venezuela. A few developing countries (Bahamas, Barbados, Bermuda, Hong Kong, Netherlands Antilles, and Taiwan) are not included due to missing data on other variables.

5 While the BEA provides data for 1982–2007, the way in which sectors were classified changed in 1999 from the SIC (Standard Industrial Classification) to the NAICS (North American Industry Classification System), with the latter providing more refined categories. The categories remained unchanged for eight of the sectors included here. We also include three sectors for which data are only available for some of the years, namely Petroleum (1982–1998), Mining (1999–2007), and Services, which was listed as a single sector under the SIC system (1982–1998). The number of observations varied across each sector, ranging from 740 for Fabricated Metal Manufacturing to 843 for Finance (out of a possible 1092 observations). Missing data in any particular year could not be interpreted as a zero value. However, as the overall stock of FDI tends to change slowly across any given year, we interpolated FDI values in cases where data were missing for only a single year, so long as data were available for the years before and after.

6 Briefly, the petroleum and mining sectors cover extraction as well as field services and exploration. As for the lower-skilled sectors, food manufacturing includes food as well as processes related to food production such as canning or slaughtering. Fabricated metals include the production of such goods as cans, cutlery, and plumbing equipment. Among the higher-skilled sectors, chemical manufacturing includes such products as industrial chemicals, paint, and plastics, while electrical manufacturing includes electrical equipment as well as electrical goods and components. Industrial machinery covers a variety of non-electrical machinery, including farm and construction equipment. Transportation manufacturing includes various types of vehicles as well as parts production. Three sectors are in the services category: wholesale trade, which involves the trading of both durable and nondurable goods, while the finance sector includes banking institutions as well as such services as brokering services and insurance. For the period before 1999 there is also a separate “services” category that covers a broad variety of services including tourism and retail-related services. For more information, see product classification guides by the US Census Bureau (http://www.census.gov/eos/www/naics/http://www.census.gov/eos/www/naics/) or the Bureau of Economic Analysis (http://www.bea.gov/surveys/iftcmat.htm).

7 The index is an ordinal scale ranging from 0 to 3. Substantively, a score of 0 would indicate a country where women enjoy none of the aforementioned political rights, such as Saudi Arabia in 2005. A one-unit improvement – that is, a score of 1 – would occur if women had some basic democratic rights (suffrage and the right to seek office) but virtually no women held major elected office (specifically, if women made up less than 5 percent of the national legislature or major cabinet positions). An example of this score would be Libya in 2005, during which only one of the 760 members of their parliament was a woman. While countries that deny all political rights to both men and women would score a 0, it is not assumed that the lack of formal democratic institutions (as is the case with Libya) means that women lack all political rights. Complete explanations of the coding rules, as well as country examples, can be found in the CIRI Codebook (David L. Cingranelli and David L. Richards 2014).

8 The CIRI project also includes a measure of women's social rights. Unfortunately, missing data was a problem with this index – inclusion of this variable reduced our sample size by 12 percent – and therefore it was not incorporated into our model.

9 Correlations among women's rights and the various sociopolitical variables are not high, with the largest being –.428 between democracy and educational gap. A correlation matrix for all independent variables is in Appendix Table 2.

10 Data on WTO membership are from the WTO website (wto.org), while data on PTAs and BITs with the United States are from the US Trade Representative's listing of trade agreements (http://www.ustr.gov/trade-agreements/free-trade-agreementshttp://www.ustr.gov/trade-agreements/free-trade-agreements).

11 Income is logged GDP per capita, adjusted for purchasing power parity, while economic growth is the yearly change in total GDP. Data for these variables, as well as unemployment, are from the World Bank (2013).

12 The measure of currency fluctuation is drawn from Globerman and Shapiro (2003: 25). It is the ratio of the value of the exchange rate in a given year (relative to the US dollar) to its mean value over the two previous years. A higher value indicates a relatively “cheaper” currency. Data on exchange rates are from the Penn World Tables 6.2 (Alan Heston, Robert Summers, and Bettina Aten 2006).

13 This represents the maximum span for which data on all of the variables are available. Missing data precludes the examination of every country across every possible year, so we used an unbalanced dataset to maximize the number of observations (Globerman and Shapiro 2003). The average time span for each of the countries we examined was 17.1 years. Descriptive statistics for all variables used in this analysis, as well as the correlations between all independent variables, are in Appendix 1.

14 Though the dummy variables themselves have no real substantive interpretation (and thus are not reported in the findings), we follow the recommended guidelines for such models by incorporating them in our analysis (Thomas Brambor, William Roberts Clark, and Matt Golder 2006).

15 Unit root tests, specifically the augmented Dickey–Fuller test, revealed that there were unit roots in our measure of FDI, which further reinforces the need for a trend variable.

16 We used the following instruments for the models: the women's rights variables were instrumented with dummy variables for CEDAW (Convention on Elimination of Discrimination Against Women) treaty ratification and Islamic religion, two variables that previous research has found to be related to the level of women's rights (Gray, Kittleson, and Sandholtz 2006). Unemployment and economic growth are both instrumented by yearly growth in the labor force, which is a proxy measure of both demographic trends and the supply of labor. Trade levels are instrumented with logged population, and income is instrumented with educational attainment for the total population. Due to collinearity concerns, the women's educational attainment variable was dropped from the main equation for the model that tested income.

17 The Arellano–Bond test for second-order autocorrelation is insignificant – that is, the null hypothesis that there is no autocorrelation is not rejected – as is the Hansen's J test, which tests the exogeneity of the independent variables not treated as endogenous.