![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
The growing literature on public opinion toward foreign direct investment (FDI) focuses primarily on the FDI-related job prospects of individuals. However, different types of FDI have varying economic and environmental impacts; we argue that individuals consider the trade-offs between these effects when forming attitudes about FDI. Utilizing a province-level dataset of recent FDI projects (categorized by their level of pollution and job creation) matched with public opinion data in seven Latin American countries, we find that exposure to investment that creates larger numbers of jobs improves attitudes toward FDI, regardless of environmental impact. However, environmental impact is a significant component of how some individuals form perspectives about FDI: respondents that prioritize the environment have more positive attitudes toward FDI when they are exposed to a higher share of FDI with relatively less environmental impact. Among all respondents, exposure to more economically beneficial FDI has a stronger positive impact when a higher percentage of the FDI is in lower-polluting industries.
Acknowledgments
The authors wish to thank the editors of Review of International Political Economy, the anonymous reviewers, Sonal Pandya and Edmund Malesky for their valuable comments on earlier drafts.
Disclosure statement
The authors declare no conflict of interest.
Data availability statement
Replication materials (excluding proprietary data from FDIMarkets, accessed through Princeton University’s data collections. Per the terms of that access agreement, the data cannot be shared publicly) are available at https://doi.org/10.7910/DVN/H40Z1I.
Notes
1 By contrast, our article uses real-world data in lieu of a survey experiment, and focuses on a range of countries in a different region with a different cultural and socioeconomic context. Our results are conditional on existing attitudes, and we also categorize a wider variety of projects.
2 Authors’ calculations using 344 protests listed as beginning prior to 2008 (i.e. originating prior to the collection of public opinion survey data used in this study). The 56% includes protests that target MNCS alone or target both domestic firms and MNCs simultaneously.
3 An exception is Schwab (Citation2018), which uses aggregate national FDI stock in some analyses. However, he does not find this measure to be significant and does not include it in models that also contain individual-level variables.
4 By looking at FDI projects over time preceding the measurement of attitudes, we minimize (though do not completely eliminate) concerns that FDI goes where it is popular (i.e. that public opinion causes the FDI flows).
5 This is consistent with our coding of environmental impact, which is based on industry-level abatement costs and emissions of air pollutants, water pollutants and heavy metal, rather than, for example, carbon emissions. While we recognize that this does not capture all possible environmental concerns, it follows the research on FDI and pollution havens and allows individuals to more directly link investment to its local environmental impact. This is consistent with findings from the literature on awareness of pollution, which finds that individuals are more cognizant of visible pollution and pollution relative to their baseline environmental quality.
6 The 2007 Pew Global data is the most recently available region-wide data that includes both questions about investment and a question about the tradeoff between the environment and development. The Pew survey asked the same questions in 2010, but only in Brazil. While the 2018 and 2020 Latinobarometer surveys include questions about attitudes toward foreign investment, they do not include questions on the environment/development trade-off. The LAPop survey only asks questions about Chinese investment, and only in select surveys.
7 While those within a province may vary considerably in their direct exposure to a given FDI project, their overall sense of FDI and its impacts are likely to be influenced by reading and hearing about FDI projects which are likely to be discussed in a broader geographic area. This should be important due to findings that attitudes about globalization’s economic flows are based more strongly on sociotropic considerations than personal ones (Mansfield & Mutz, Citation2009). Ideal data to test exactly how exposure influences people would include data geo-coded for both the respondent and the project, allowing us to calculate absolute distance and more directly measure exposure to FDI. However, such data is not currently available. The identification of the location of survey respondents restricts us to examining effects at the province level. An additional strategy would be to collect data at the city level in future research; we have partial, but incomplete, data on city-level FDI projects, but survey measures do not identify respondents as precisely. Future data collection should focus on further increasing the granularity of data about both respondents and projects. In its absence, an examination of mostly indirect exposure to FDI helps to establish that FDI projects do have a broader impact on attitudes, rather than only mattering to people whose employment and ‘backyards’ are directly affected. Some projects in the fDi Markets database listed the investment destination province as ‘not specified’. Of these 739 projects with unspecified sub-national locations, we were able to use outside research to identify provinces for 441. Only 298 remain with an unknown location and had to be excluded from the aggregated province-level data.
8 T-tests comparing the average number of jobs produced per project in ‘clean’ industries does not significantly differ from the average jobs per project in ‘dirty’ industries.
9 See appendix Tables A4, A5, A7, A8, A10 and A11 for robustness checks with these alternative measures. Our preferred measure is based on project count because data on the capital or jobs involved in each FDI project is estimated and may represent intended investment or job creation over a number of years (making these measures less reliable and noisier). Additionally, we recognize that measuring FDI exposure to clean or dirty FDI as a percentage of total FDI makes certain assumptions about how exposure works. This measure reflects an assumption that individuals take an overall, holistic approach to assessing investment, weighing the positive and negative impacts of foreign direct investment, which should mean that not only will ‘clean’ investment be weighed more heavily, but that the negative impact of a ‘dirty’ project will be mitigated when ‘clean’ projects are also present. This is consistent with our theoretical arguments that economic benefits can mitigate the negative impact of polluting FDI projects, rather than influencing attitudes toward FDI independently. However, we acknowledge that our data do not allow us to test whether this assumption is correct, versus exposure operating primarily through the impressions made by the largest or most prominent recent FDI project. While we cannot resolve the exact mechanisms by which real-world exposure impacts attitudes towards FDI, as a robustness check for our findings we alternatively have coded the largest FDI project (by capital investment and alternatively by job creation) in each province according to its environmental impact. We repeat the analyses testing H2A and H2B using these alternative measures and find consistent results, where clean FDI is viewed more positively, but dirty FDI is not penalized (see Appendix, Tables A12–A15).
10 We selected 100 jobs as an important threshold at which the economic impact of an FDI project would have more public notice, media coverage, and political attention brought to the general public. The preponderance of micro- and small enterprises in Latin America (nearly 90% of firms are micro-enterprises (OECD, 2019) also makes it more likely that a project providing 100 jobs is notable. Provinces that have no FDI projects are assigned zero values for these and all of our other calculated percentages. See appendix for descriptive statistics for each province. However, we acknowledge that 100 jobs is a somewhat arbitrary threshold. Accordingly, we also show results for alternative distinctions between economically important and less important projects using 200 jobs, 250 jobs or varying the size of the project considered economically important based on the population of the province as cut points (see Appendix, Tables A17–A21). Key results are consistent with the predictions summarized in across measures. For H2A, the only departure from the results in (Table A6) is that the percentage of economically unimportant, polluting projects becomes insignificant at higher thresholds. This is consistent with the argument that projects creating at least 100 jobs are considered economically important and their inclusion in the low economic impact category dilutes the negative impact.
11 In a few cases, industry cluster was further used to distinguish between different categories of investment.
12 Since people in provinces without exposure to any recent FDI projects do not have exposure to either clean or dirty (or more job-creating versus less job-creating) projects, attitudes may reflect a high degree of variance. Provinces without recent FDI projects may contain more people who have no opinion about FDI, for example, so their survey responses may be noisier.
13 This is the same dependent variable used in other studies by Ahmed et al. (Citation2016) and Lee and Shin (Citation2019). While this question is about the firms, rather than the investment flows, MNCs and FDI are highly interconnected concepts. FDI is defined as ‘an investment involving a long-term relationship and reflecting a lasting interest and control by a resident entity in one economy (foreign direct investor or parent enterprise) in an enterprise resident in an economy other than that of the foreign direct investor (FDI enterprise or affiliate enterprise or foreign affiliate)’ (United Nations Conference on Trade & Development, 2007, p. 245). MNCs become ‘multinational’ by engaging in the practice of FDI. While it is possible to have FDI without an MNC, in practice, most FDI takes place through MNCs. By definition, all MNCs involve FDI (OECD, 2015, p. 5). Since multinational corporations are, by definition, the actors engaged in foreign direct investment, we hold that perceptions of multinational corporations are directly connected to FDI. Unlike other survey questions in studies of FDI preferences (Pandya, Citation2010; Schwab, Citation2018), this question asks about the effects of the investment generally and not merely about on the economic development of the country. This wording provides greater scope for the respondents to consider both the economic and non-economic influences of investment in answering the question. For our purposes, this more holistic measure of opinion about investment is desirable but may not be entirely comparable to measures from other studies that focus on economic or developmental impact.
14 One potential concern with using environmental attitudes may be that more educated people are more environmentally conscious. We show that education does not produce the same effect in our models as environmentalist attitudes in the Appendix (Table A12). Accordingly, we control for education in the models below. A second concern may be that it assumes a trade-off between economics and the environment, which is not always present. While green economic growth is possible, some economic projects will be environmentally polluting. In order to assess the extent to which respondents prioritize environmental concerns, the measure must consider these concerns in comparison with other types of concerns. Thus, we use this question appropriate for our purposes.
15 The education categories vary slightly by country, but this coding can be applied across all seven countries.
16 Since the survey data containing all of these control variables is collected at the end of the period for which we measure FDI projects, there may be concern that these measures are influenced by the FDI. FDI projects could have altered some respondents’ employment or economic satisfaction, or even conceivably the level of education they chose to attain (though this is less likely over the five-year period in question). While we cannot obtain pre-treatment measures for the individuals in the dataset, we mitigate the post-treatment concern by presenting models without post-treatment controls, keeping only age and gender as they, conceptually, are not affected by FDI (see Appendix, Tables A22–A24).
17 The percentage of projects that create fewer than 100 jobs is the omitted category. Because we test for alternative cut points as measures of economic impact, in the appendix we show models where the omitted categories are projects with fewer than 200 jobs, fewer than 250 jobs or where the cut point differs depending on the population of the province (see Tables A16–A21).
18 Tables A4 and A5 repeat the analyses using measures of FDI projects’ jobs and capital, rather than count.
19 The percentage of FDI in service industries (regardless of the number of jobs created) is the omitted category.
20 Full results for this figure are shown in the appendix, Table A6. Robustness checks using jobs or capital in place of FDI count are shown in Tables A7 and A8. Using alternative cut points for economic impact (200 or 250 jobs, or varying the number of jobs needed for economic impact by the population of the province) shows that the percentage of clean, higher impact projects is significant and positive, while the effects of the other percentages are not consistently significant (see Tables A16–A21).
21 The share of projects in service industries is the omitted category.
22 Full results used to produce are shown in the Appendix, Table A9, with robustness checks using jobs or capital in place of FDI projects count in Tables A10 and A11. An analysis of exposure to FDI disaggregated by environmental impact, but without accounting for the weight that respondents place on environmental concerns, shows no significant results.
23 While reverse causality may be a concern, we find that there is no correlation between attitudes toward FDI and total FDI. Since FDI is not more prevalent in places where perceptions of FDI are more positive (and our FDI measures are from the five-year period prior to the measure of FDI perceptions), it is unlikely that FDI flows are responding to public opinion. In the appendix (Figures A3 and A4), we include graphs of province average environmental attitudes and clean and dirty percentages in order to address any potential worries that environmental attitudes are caused by FDI exposure or that clean FDI is being directed to areas because of public opinion on the environment.
Additional information
Notes on contributors
Celeste Beesley
Celeste Beesley is an assistant professor of Political Science at Brigham Young University. She earned her Ph.D. from the University of California, San Diego. Her research focuses on the myriad political effects of economic globalization and insecurity.
Alexander Slaski
Alexander Slaski is a visiting professor in Georgetown’s department of Government. He earned his Ph.D. from Princeton University in 2018. Dr. Slaski’s research lies in comparative and international political economy, with a focus on the political economy of foreign direct investment, investment incentives, and currency flows in the Global South, particularly Latin America.