Sectoral Foreign Investment and Nitrous Oxide Emissions: A Quantitative Investigation

Abstract Studies of the anthropogenic contributors to greenhouse gas emissions have increased markedly in recent years. However, not all greenhouse gases have received their deserved attention. This study tests the theoretically derived assertion that foreign investment dependence increases the emissions of one such neglected greenhouse gas, nitrous oxide. Findings of the cross-national ordinary least squares (OLS) regression analysis of 44 countries show that foreign investment dependence is associated with higher levels of nitrous oxide emissions. However, the findings are contingent upon the sector of investment. That is, primary-sector investment is associated with higher levels of emissions, whereas secondary-sector investment is not. The findings support the theory of foreign investment dependence while also underscoring the need to study these sorts of structural relationships at the sector level. Keywords: agricultureenvironmental sociologyforeign investmentglobal warmingglobalizationmanufacturingnitrous oxide emissions Acknowledgments We thank Eugene Rosa and Kennon Kuykendall for their input on earlier versions of this article. We also thank the editors and anonymous reviewers of Society & Natural Resources for all of their comments. An earlier version of this article was presented at the 2007 annual meetings of the American Sociological Association in New York. Notes Note. Statistical significance indicated by *p < .05, **p < .01, ***p < .001. Unstandardized coefficients in parentheses, standard errors in brackets, VIFs in italics. The use of nitrogen-rich fertilizers adds to the available amount of nitrogen for denitrification, which is the process by which bacteria break down nitrates in the soil to give nitrogen gas, which returns to the atmosphere (with N2O being one of the by-products). Due to this finding the argument could be made that we should control for fertilizer use. However, the World Resources Institute (2004) reports that fertilizer use is one of the key components of the measurement of nitrous oxide emissions. This would, in effect put the same variable (or parts thereof) on both sides of the regression equation. Data for three time points (1990, 1995, 2000) are available from World Resources Institute (2004). However, a sizeable portion of the data for the year 2000 is estimated based on 1995 totals. We do not feel that these data are adequate for our purposes here since the WRI claims that these data are only "rough approximations." We have also conducted analyses with per capita nitrous oxide emissions as the dependent variable. The findings are substantively similar, and these findings are available on request from the authors. Ideally, we would prefer measures of domestic investment for different economic sectors. Nevertheless, those types of data were unavailable at the time of this study. It is important to note that residualizing a variable on GDP per capita will give the shared variation from the two variables to GDP per capita. Thus, if we reversed things, the shared variation would go to the other variables. We also used five other control variables (democratization, state environmentalism Dietz and Kalof Citation1992, environmental international non-governmental organization INGO presence, services as a percentage of GDP, and environmental INGOs per capita) in other models, but do not report these analyses for sake of space because the effects of these additional controls are all nonsignificant and their inclusion does not change the substantive findings of interest. Also, per the request of a reviewer, we include a measure of total foreign investment dependence in the primary and secondary sectors. This aggregate measure does not reach statistical significance. These analyses are available from the authors upon request. Elsewhere we include slope–dummy interactions to assess whether the effects of primary-sector and secondary-sector investment differ for the developed and less-developed countries in the sample. The effects of all the interactions are nonsignificant and the findings of these additional analyses are available from the authors upon request.