The spectacular growth in FDI flows during the 1990s, along with the increasing importance of developing economies as host countries, has raised concerns about the potential effect of FDI on the natural environment (Zarsky 1999). A lot of research effort has been devoted to testing the so-called pollution haven hypothesis, which refers to the possibility that multinational firms, particularly those engaged in highly polluting activities, relocate to countries with weaker environmental standards.
This literature encompasses three strands. The first strand considers the patterns of outward FDI from industrialized countries. In an early study, Eskeland and AE Harrison (2003) test whether the pattern of outbound US investment during the 1980s and early 1990s can be explained by variations in pollution abatement costs across different sectors of the US economy. They conclude that sector-specific abatement costs are not significantly associated with outbound US FDI, though they do point out that their sample may not have included enough time series variation. In contrast, Kellenberg (2009) finds support for a pollution haven effect when accounting for strategically determined environment, trade, and intellectual property rights policies. His results suggest that for the top 20th percentile of countries in terms of growth in US foreign affiliate value-added, as much as 8.6% of that growth between 1999 and 2003 can be attributed to declining relative stringency and enforcement of environmental policy. This effect is even more pronounced in developing and transition economies, where approximately 32% of US foreign affiliate value-added growth can be attributed to falling relative levels of environmental stringency and enforcement.
Simple summary statistics presented by Hanna (2010) show that between 1982 and 1999 (the last year considered), foreign assets of US multinationals in clean industries grew at a relatively faster rate than foreign assets in pollution-intensive industries. This is true when all destination countries are considered, as well as when non-OECD destinations are considered. However, this pattern cannot be taken as evidence against the pollution haven hypothesis, as it has been suggested that by nature of their technologies, industries with the largest pollution abatement costs also happen to be the least footloose (Ederington, Levinson, and Minier 2005). In response to this concern, Hanna (2010) does not use industry-level measures of environmental stringency but rather exploits the plausibly exogenous variation in firm-level regulation created by the Clean Air Act Amendments (CAAA). Following the passage of the CAAA in 1970, the Environmental Protection Agency (EPA) established separate national ambient air quality standards, i.e., a minimum quality level that all US counties were required to meet, for four criteria pollutants. Each year, counties where air pollution concentrations exceeded federal standards for a specific pollutant received a non-attainment designation for that pollutant, while counties that were in attainment of federal standards received an attainment designation. Manufacturing plants that emit a criteria pollutant and were located in a county that was designated as non-attainment were subject to relatively tougher regulatory oversight than emitting plants in attainment counties. The nature of the CAAA regulatory program allowed the author to employ a differences-in-differences-style approach to test whether firms were more likely to expand their overseas manufacturing operations when the US counties in which they operated fell into non-attainment and were, thereby, subject to tougher environmental oversight. The results suggest that the CAAA legislation increased the outbound FDI of US-based multinational firms, inducing multinationals to increase their foreign assets in polluting industries by 5.3% and their foreign output by 9%. Larger multinational firms accounted for much of the increase in FDI, but contrary to the pollution haven hypothesis, heavily regulated firms did not disproportionately increase production in developing nations relative to other countries.
The second strand of the literature considers inflows of FDI to developing countries. In an early study, Javorcik and Wei (2004) focus on firm-level investment flows from multiple industrialized countries to 25 economies in Eastern Europe and the former Soviet Union, which have a large variation in environmental standards. The authors account for the effect of host country corruption and include information on both the polluting-intensity of the potential investor and the environmental stringency in the potential host country, which allows them to test whether dirty industries are relatively more attracted to locations with weak standards. They find no systematic evidence that FDI from "dirtier" industries is more likely to go to countries with weak environmental regulations. Similarly, Dean, Lovely, and H. Wang (2009) focus on manufacturing equity joint ventures (EJV) projects in China across 28 industries during 1993-1996. They use data on actual collected water pollution levies to construct a measure of provincial environmental stringency, drawing on annual Chinese environmental and economic censuses. Their results show that EJVs in highly-polluting industries funded through Hong Kong, Macao, and Taiwan are attracted by weak provincial environmental standards. In contrast, EJVs funded from non-ethnically Chinese sources are not significantly attracted by weak standards, regardless of the pollution intensity of the industry. These findings are consistent with pollution haven behaviour, but not by investors from high-income countries. Cai et al. (2016) examine the effects of China’s 1998 Two Control Zones (TCZ) policy, in which tougher environmental regulations were imposed by the central government in TCZ cities but not others. Their estimation exploits differences across three dimensions: (i) city (TCZ versus non-TCZ cities), (ii) industry (more polluting versus less polluting), and (iii) year (before and after the policy change). They find that multinationals from countries with better environmental protections than China are insensitive to the toughening environmental regulation, while those from countries with worse environmental protections than China show strong negative responses[1].
In sum, the evidence produced by the first two strands of the literature does not suggest that relocation of dirty production from industrialized countries to developing countries with weak environmental standards is taking place to an extent that raises concerns, if at all. The third strand of the literature compares the performance of foreign affiliates in terms of pollution emissions or (more frequently) energy intensity to that of local firms. The focus on energy intensity is necessitated by the lack of data on firm-level emissions, particularly in developing countries, but the two are highly correlated (Chung 2016). Most existing studies examine the correlations between environmental impact and foreign ownership, but are unable to establish a causal relationship. Using data collected by the Indonesian Environment Ministry’s PROKASIH (Clean Rivers) program for 1989-90, and controlling for plant scale, age, and efficiency, Pargal and Wheeler (1996) find that foreign ownership is not significantly correlated with water pollution emissions in Indonesia. Eskeland and AE Harrison (2003) examine plant-level data from Cote d’Ivoire, Mexico, and Venezuela and find that the energy share (the cost of energy use divided by the total value of the plant’s output) is negatively correlated with foreign ownership. Cole, Elliott, and Strobl (2008) use plant-level data from Ghana and find no strong evidence that foreign ownership influences total energy use, though plants with foreign-trained managers are found to have lower energy intensity. In a cross-sectional study of around 1,200 firms, Albornoz et al. (2009) find a positive correlation between foreign ownership and implementation of environmental management systems. A recent paper by Brucal, Javorcik, and Love (2019) accounts for selection into foreign ownership, i.e. the possibility that foreign investors choose to acquire local plants with better environmental performance, and thus comes much closer to capturing the causal effect of foreign ownership. These authors work with panel data and hence are able to account for unobserved plant-specific characteristics and examine the stability of the estimated effects over time. Thanks to very detailed plant-level information on energy inputs, they are also able to measure energy use in physical units and provide a more detailed analysis on the types of fuel used. Their analysis of 210 acquisition cases in Indonesia over the period 1983-2001 suggests that while foreign ownership increases the overall energy usage due to the expansion of output, it decreases the plant’s energy intensity and carbon emission intensity. Specifically, acquired plants reduce energy intensity by about 30% two years after acquisition, relative to the control group of non-acquired plants. The authors also find that foreign divestments lead to an increase in energy intensity. Finally, they find that at the aggregate level, entry of foreign-owned plants is associated with an industry-wide reduction in energy intensity.
The overall message emerging from this literature is that, contrary to the fears expressed in the 1990s, FDI may actually benefit the environmental performance of developing countries.
[1] Interestingly, two studies focusing on FDI inflows into the US find that stronger environmental regulations deter FDI (Keller and Levinson 2002, Millimet and Roy 2016).
