Global foreign direct investment (FDI) flows rebounded strongly in 2021, surpassing pre-pandemic levels, with developed economies receiving the lion's share. FDI's impact on both developed and developing economies is significant, influencing economic growth and environmental quality. Existing literature presents contradictory views: the pollution paradise hypothesis posits that FDI leads to environmental degradation in countries with lax environmental regulations, while the pollution halo hypothesis suggests that FDI brings technological advancements and improved environmental quality. Empirical studies have yielded conflicting results, partly due to variations in research samples, particularly concerning income levels, institutional quality, and environmental regulations. Income level is a crucial factor as it impacts the stringency of environmental policies and public preference for environmental quality. This study addresses three research questions: 1) Is there a non-linear relationship between FDI and environmental quality? 2) If so, is this linked to income level? 3) Is there heterogeneity in results across income groups? Using panel data from 67 countries (1990-2019), this paper examines these questions using a nonlinear approach to capture structural breaks and heterogeneity.

The theoretical debate surrounding FDI and environmental quality centers on the pollution haven hypothesis (PHH) and the pollution halo hypothesis (PHL). PHH suggests firms relocate pollution-intensive activities to countries with weaker environmental regulations. PHL argues that FDI leads to technological advancements and cleaner production, thus improving environmental quality. Empirical studies testing these hypotheses have produced mixed results, with some supporting PHH (e.g., studies on Kenya, Zimbabwe, India, Pakistan), others supporting PHL (e.g., studies on Nigeria, ASEAN countries), and still others finding no significant relationship or mixed effects. The inconsistencies highlight the complex and multi-faceted nature of the FDI-environment relationship, with factors like income level, institutional quality, and environmental regulations likely playing crucial roles. Previous studies have often used linear models, failing to capture potential nonlinear effects and structural breaks.

The study employs a panel dataset of 67 countries from 1990 to 2019, categorized into high, upper-middle, and lower-middle income groups. The dependent variable is per capita CO2 emissions, the main explanatory variable is FDI (net inflows as a percentage of GDP), and the threshold variable is GDP per capita. Control variables include financial development, population size, industrial structure, and trade openness. Initially, a baseline model is estimated using Fully Modified Ordinary Least Squares (FMOLS) to address endogeneity and serial correlation issues. This model includes a quadratic term for GDP per capita to explore the Environmental Kuznets Curve (EKC) hypothesis. To investigate the nonlinear relationship between FDI and carbon emissions, a panel threshold regression (PTR) model is implemented. The PTR model allows for different relationships between FDI and emissions based on different GDP per capita thresholds, identifying potential structural breaks in the relationship. Unit root and cointegration tests are performed to ensure data stationarity and the validity of long-run relationships before estimation. Bootstrap methods are used to determine the number of thresholds and their significance. Robustness tests are conducted through sub-sample regressions based on income groups.

FMOLS estimation confirms an inverted U-shaped EKC curve across all 67 countries. FDI shows an overall positive effect on carbon emissions (0.02% increase per 1% increase in FDI). Financial development also contributes significantly to increased emissions. Panel threshold regression reveals a significant double threshold effect of GDP per capita on the FDI-emission relationship. Below a GDP per capita of $541.87, FDI positively impacts emissions (0.11% increase per 1% increase in FDI). Between $541.87 and $46515, the effect is minimal. Above $46515, FDI has a significant negative impact on emissions (-0.08% decrease per 1% increase in FDI). Analysis of the spatial and temporal distribution of countries across these threshold intervals reveals that the number of countries in the intermediate range is consistently high, while the number of countries above the higher threshold grows over time. Robustness tests using sub-sample regressions for each income group confirm the presence of a single threshold effect in each group, with varying threshold values and magnitudes of effects. For high-income countries, the threshold is around $31,820, with FDI increasing emissions below and having a non-significant effect above. For lower-middle-income countries, the threshold is around $518, with FDI significantly increasing emissions below and having a non-significant effect above. For upper-middle-income countries, the threshold is around $1148, with FDI significantly decreasing emissions below and having a non-significant effect above.

The findings demonstrate a nonlinear relationship between FDI and carbon emissions, contingent on income level. The positive effect of FDI on emissions in low-income countries is consistent with the pollution haven hypothesis (PHH). This suggests that these countries attract pollution-intensive industries due to weak environmental regulations and a focus on economic growth. The negative effect of FDI on emissions in high-income countries aligns with the pollution halo hypothesis (PHL), indicating the transfer of cleaner technologies and higher environmental standards. The study reconciles these seemingly contradictory hypotheses by showing that both can exist simultaneously, depending on the level of economic development. The threshold effects highlight the importance of considering economic context when assessing FDI's environmental impact. The results underscore the need for targeted policies based on a country's income level, promoting both economic growth and environmental sustainability.

This study reveals a significant nonlinear relationship between FDI and carbon emissions, mediated by GDP per capita. Low-income countries experience a positive relationship, while high-income countries experience a negative relationship. This challenges the simplistic dichotomy between PHH and PHL, highlighting the importance of income levels in determining the environmental effects of FDI. Future research could explore other factors influencing this relationship, expand the range of countries and environmental indicators, and examine the effectiveness of policies designed to mitigate the negative environmental consequences of FDI in low-income economies.

The study's limitations include the use of a specific time period (1990-2019) and a limited number of countries (67). The focus on CO2 emissions as the sole environmental indicator limits the scope of the findings. Future studies should expand the data, incorporate more environmental variables, and investigate potential country-specific factors. The use of GDP per capita as the sole threshold variable may also be considered a limitation.