The Role of Carbon Emissions on Inward Foreign Direct Investment: A Nonlinear Dynamic Panel Data Analysis

Abstract

An increase in carbon emissions (CO₂) may increase inward foreign direct investment (FDI) in developing countries since they are seen as pollution havens because of lax environmental regulations (pollution haven hypothesis). Developed countries may also attract FDI since stringent environment regulations in these countries working to reduce emissions might be more attractive to foreign investors concerned with their repute from a green perspective. A rise in CO₂ emissions in developed countries therefore deters inward FDI (green haven hypothesis). The existing empirical studies investigate the empirical validity of these hypotheses by focusing on the impacts of environmental policies and regulations on FDI and have yet to produce conclusive results. We examined the effect of CO₂ emissions on FDI and provide a more accurate and novel way of investigating the empirical validity of the pollution haven hypothesis against the green haven hypothesis. Specifically, we examined the non-linear effects of CO₂ emissions on inward FDI in a sample of 124 countries over the period 1997–2022. The results indicate that CO₂ emissions have an inverted-U-shaped relationship with FDI, confirming our hypotheses that higher CO₂ emissions in countries with lax environmental standards attract FDI while environmental degradation in countries with stringent environmental standards deter FDI.

1. Introduction

One of the main causes of climate change and environmental degradation is carbon (CO₂) emissions. The increasing risk of climate change has led to uncommon and prolonged periods of extreme heat in the Northern Hemisphere in 2022 and has significantly affected both economic progress and the well-being of individuals in the region [1].

The Millennium Development Goals (MDGs) and the Sustainable Development Goals (SDGs) have prioritized emission reduction measures in environmental action plans; since the Rio de Janeiro Earth Summit, policymakers and practitioners have placed significant emphasis on CO₂ reduction due to its role as the primary catalyst of climate change, stemming from its substantial contribution to overall greenhouse gas emissions [2].

To reduce the consumption of conventional fossil fuels in production, it is imperative to decarbonize the economy and mitigate the pollution of the environment [3].

These impacts may be even bigger as higher levels of CO₂ emissions are also associated with changing the patterns and nature of foreign direct investment (FDI) inflows, which is considered as an engine to growth due to its ability to transfer knowledge, bridge the savings gap, and boost domestic investment, increasing productivity and economic development [4]. Consequently, investors, policy makers, and governments worldwide are increasingly conscious of the association between environmental quality and FDI.

The relationship between environmental pollution and FDI stems primarily from two alternative perspectives. The “pollution haven hypothesis” (PHH) suggests that low-income countries having lax environmental standards and a dirty environment attract foreign investors eager to shift their dirty production to these countries to avoid environment-related production costs in their environmentally stringent home countries. Having lax environmental regulations in comparison to its trading partners, a highly polluted country could attract more foreign investment to its polluting industries [5] whereas stringent environmental regulations in a host country may discourage polluting manufacturing investments [6,7].

The second perspective, the “green haven hypothesis” (GHH), defines the mechanism whereby countries having stringent environmental regulations with good environmental quality become attractive to foreign investors. There has been a global trend toward sustainable investing and Environmental, Social, and Governance (ESG) considerations. Investors are increasingly considering environmental factors, including CO₂ emissions, in their investment decisions. Global companies that value their reputation for sustainable management and corporate social responsibility (CSR) can decide to make investments in nations with stronger and more stringent environmental laws, according to the GHH [8]. It is now more difficult for businesses to exploit loose environmental restrictions as doing so would damage their brand as environmentally conscious global players [9]. Developed countries with stringent environmental regulations working to reduce emissions might be more attractive to investors who prefer to work in safe environments.

Although these theoretical arguments are evident, the empirical studies designed to test the pollution haven hypothesis and green haven hypothesis have yet to produce conclusive results. Importantly, a host of these studies analyze the impacts of environmental policies and regulations on FDI (e.g., [8,9,10,11,12,13,14,15]) in testing the validity of these hypotheses. An implicit assumption behind this could be that more stringent environmental laws and regulations leads to higher environmental quality. However, stringent environmental standards do not ensure low CO₂ emissions. It is evident that the US and 28 EU countries were among the top three CO₂ emitters in the world in 2018 [16] despite adopting stringent environmental regulations. According to [17], the domestic component of consumption-based CO₂ emissions of developed countries shows a declining trend after reaching its peak, while the foreign component of emissions of some developed countries continues to show an upward trend [17]. Emissions could rise consistently with income as new pollutants arise over time.

In the current study, therefore, we explicitly focused on the impacts of CO₂ emissions on inward FDI to test the empirical validity of the PHH against the GHH. In the existing literature, only a few studies have examined the two-way causation between FDI and CO₂ emissions (e.g., [18,19,20]) and some of these studies have found a two-way relationship between them. The relationship between CO₂ emissions and FDI is complex and multifaceted. As we discussed in the theoretical perspective in detail, several factors can influence this relationship, and the impact may vary depending on the context and specific circumstances of countries involved. We anticipated that CO₂ emissions have both a pollution haven effect and a green haven effect on inward FDI simultaneously. Specifically, we hypothesized that in developing countries the pollution haven effect dominates; that is, a rise in CO₂ attracts inward FDI. However, in countries with stringent environment laws and regulations, the developed ones, the green haven effect of CO₂ emissions will become prevalent and the increase in CO₂ may cause a deterrent effect on inward FDI.

The current study contributes to the literature in the following ways. First, this study plans to access whether inward FDI responds differently to an increase in CO₂ emissions in host countries and provides a more accurate and novel way of investigating the validity of the pollution haven hypothesis against the green haven hypothesis. To that goal, we innovate by examining an inverse-U-shaped relationship between CO₂ emissions and inward FDI. Second, to solve the possible endogeneity problem, we use a dynamic Blundell–Bond System GMM estimator, looking at the dataset’s cross-section and time-series dimensions. Third, the dataset used in the study consists of a substantial sample of 124 economies, spanning the period of 1997–2021. The results demonstrate that CO₂ emissions exhibit an inverse-U-shaped relationship with inward FDI, confirming that the initial positive effect of CO₂ emissions on FDI eventually transforms to a discouraging effect. The dominance of the pollution haven effect in developing countries with loose environmental rules and the green haven effect in developed countries with strict restrictions is the reason for the inverse-U-shaped relationship.

This paper is structured as follows: The theoretical perspective is presented in Section 2, the estimation technique and data are explained in Section 3, and the empirical findings are discussed in Section 4. Furthermore, the conclusion is drawn in Section 5.

2. Theoretical Perspective and Relevant Literature

The positive association between environmental quality and FDI is based on the classical trade theory of comparative advantage. This view stipulates that stringent environment policies can increase production costs, implying countries with lax environmental standards have relatively low production costs [21] and therefore enjoy a comparative advantage in polluting production. Based on this perspective, the pollution haven effect predicts that countries with lax environmental standards attract multinational firms and tend to decrease costs due to the lack of pollution control activities [22], while strict environmental rules in host countries discourage MNCs [6,7].

Generally, developed countries take part in international environmental treaties and have stringent environmental standards. Developing countries, on the other hand, intentionally impose lower energy and environmental regulations, thereby offering a comparative advantage to polluting FDI and transforming into global hubs of intensified pollution [23,24]. The pollution haven hypothesis posits that environmentally taxing production activities from developed nations tend to relocate to developing nations where adhering to environmental standards in the form of FDI is more economically viable. This process enables the developed world to focus on specializing in environment friendly production activities [25], while developing countries gain a comparative advantage in producing goods that results in high levels of pollutants. Polluting industries demonstrate a stronger inclination to set up their new foreign affiliates in countries with less stringent environmental standards compared to cleaner industries [6].

Countries at their initial level of development experience higher levels of environmental degradation. Theories explaining how development level and environmental degradation are related regard production-related as particularly rapid-industrialization and energy-intensive production activities, responsible for higher environmental degradation at the initial development level. A well-known hypothesis, the Environmental Kuznets Curve (EKC), states that the initial negative effect of GDP per capita on environmental quality eventually transforms to a positive effect as per capita income increases. The structural change perspective postulates that early industrial development is characterized by increased environmental degradation because scale effects outweigh composition effects and technological effects [26,27]. In this context, the ecologically unequal exchange theory posits that, through trade and FDI, high-income countries transfer to low-income countries the negative environmental effects of their consumption [28]. Asıcı and Acar [29] report that countries tend to transfer the ecological cost of their consumption to less affluent economies as they experience economic growth. Consequently, low-income countries experience environmental degradation as shown in production-related environmental quality indicators [30,31,32]. This discussion leads us to draw following hypothesis:

Hypothesis (H1): 

An increase in CO₂ increases FDI inflows for countries with less stringent environmental regulations (pollution haven hypothesis).

In situations where factor endowment holds significance, several forecasts made by the PHH can be reversed [22]. Despite stringent environmental regulations in capital-abundant nations, capital-intensive industries that generate pollution might opt to establish themselves in these advanced countries to take advantage of their wealth in capital resources [25,33] and consequently transform them into potential pollution havens. This idea is based upon the theories explaining the horizontal (market seeking) and vertical (lower production cost seeking) motives of FDI where capital intensity is identified as the source of comparative advantage explaining the vertical flow of FDI in developed countries.

The existing literature identifies another motive of FDI, known as the “green haven hypothesis” or “race to the top”, whereby multinational firms prioritize their reputations for sustainable management and corporate social responsibility (CSR), often leaning towards operating in clean environments and investing in countries with more stringent environmental standards [8,9]. Foreign firms concerned about long-term sustainability and environmental risks may avoid investing in regions or industries associated with significant carbon emissions to avoid negative repercussions from a green perspective. Consequently, countries with stringent environmental regulations and policies and good environmental quality are more likely to attract clean FDI. Testing the PHH against the GHH at a sector level, Poelhekke and Ploeg [8] report that countries having stricter environmental laws that are better upheld deter the pollution-intensive industries of the Netherlands’ FDI (pollution haven effect), whilst CSR minded-industries are attracted to such countries (green haven effect).

However, high-income countries that have stricter environmental regulations do not ensure improved environmental quality. International agreements involving environmental diplomacy, which includes diplomatic strategies, negotiations, and international collaboration to address trans-boundary environmental challenges and promote sustainable solutions, can significantly reduce CO₂ emissions. However, the existing studies highlight a distinct gap between the commitments made in signed treaties and the actual implementation of these environmental agreements, resulting in increased CO₂ emissions (e.g., [34,35]).

New evidence reveals a positive significant relationship between environmental pollution and development level in developed high-income countries [16,36,37] and a negative significant relationship in developing low- and middle-income countries [16] and casts doubts in the EKC pattern. High-income nations not only have higher production levels but also emit greater amounts of CO₂ compared to middle-income emerging countries [16]. The explanations come from consumption-related causes driven by affordability to use more energy-intensive and industrial products.

From a global perspective, the globalization of production that allows high-income countries to shift adverse environmental impacts to their low-income counterparts without changing unsustainable consumption patterns causes consumption-related environmental pollution in these countries [28]. Asıcı and Acar [29] show that rich countries exhibit progress in production-related environmental quality indicators, however experience environmental degradation when they are assessed through consumption-related environmental quality indicators. Onofrei et al. [37] argue that, while European Union countries have yielded some environmental benefits with their implemented environmental policies, they have not prevented the escalation of environmental problems. They highlight that the trajectory of economic growth persists in exerting pressure on the environment and giving rise to novel challenges and vulnerabilities.

It is argued that stringent environmental policies might not effectively curb CO₂ emissions stemming from consumption activities unless they are incorporated with significant costs on emitting–consumption activities. In a study of 15 large greenhouse gas emitter countries, Demiral et al. [16] report that stringent environmental standards, proxied by the Environmental Policy Stringency (EPS) Index, are positively associated with CO₂ emissions in middle-income countries; however, in a sub-sample of high-income countries, this impact is statistically insignificant.

Existing studies examining the correlation between FDI inflows and CO₂ emissions (although very few) find a robust two-way causal relationship between FDI and emissions (e.g., [18,19,20,38]).

The above discussion leads us to our second hypothesis:

Hypothesis (H2): 

An increase in CO₂ decreases FDI in countries with more stringent environmental regulations (green haven hypothesis).

There is investor inertia which involves a continuation of foreign direct investment activities, despite changing circumstances, economic crises, or economic instabilities, due to the persistent adherence to profit maximization through the animal spirit of investors, market dynamics, globalization, psychological factors, and external influences.

Hypothesis (H3): 

Investor inertia exists.

• Main Hypothesis: Figure 1 shows a graphical representation of our main hypotheses. Below a certain threshold level of CO₂, an increase in CO₂ increases FDI in countries with less stringent environmental standards (pollution haven hypothesis) and, above a certain threshold level of CO₂, an increase in CO₂ decreases FDI in countries with more stringent environmental standards (green haven hypothesis).

3. Empirical Analysis

3.1. Data and Variables

Appendix A presents the list of countries that were analyzed. The rationale was to include as many countries as possible to show the nonlinear effect of differing levels of CO₂ emissions on FDI inflows.

The summary statistics, cross-country dependency test, and unit root analyses cover the period of 1996–2022, and the System GMM analyses and the robustness analysis cover the period of 1997–2022; since some variables are I (1), we took the first difference between them to make them stationary.

Appendix B presents the source of the variables. Less than 5 percent of the variables are missing values. The Stata command of ipolate was used for interpolation.

Appendix C presents descriptive statistics. The variables of governance (gov), education (educ), and infrastructure (infra) take negative values since we used principal component analysis (PCA) to create these indices. FDI inflows take negative values since they are net FDI inflows.

We employed a cross-sectional dependency test in Appendix D to decide which unit root test is appropriate for each variable. Every variable in Appendix D has cross-sectional dependency, with the exception of governance (gov), based on the findings of cross-sectional dependency tests. We employed the first-generation panel unit root test of [39] for governance that has no cross-sectional dependency and the second-generation unit root test of [40] for the variables that have cross-sectional dependencies.

According to the results of unit root tests in Appendix E, FDI inflows (fdi) and financial development (findev) are I(0), and the rest of the variables are I(1).

Correlation matrices of the variables were added as Appendix G. Taking the square of CO₂ emissions only produced 20.71% covariance. The highest covariance is less than 26%. Since we used System GMM, the multicollinearity issue was severely decreased due to the instruments lagging.

3.2. Methodology

We employed System GMM analysis for the estimation primarily for three reasons. First, the concept of FDI inflows has the potential to reinforce itself, as previous levels of FDI inflows could significantly influence current levels. Second is the endogeneity issue depending on the reverse causality from FDI to the main independent variable of CO₂ emissions and its square, and to control variables. Last, due to their potential correlation with both past and presumably present realizations of the error term, the independent variables may be endogenous. Also, System GMM addresses the issues of fixed effects, heteroskedasticity, and reduces autocorrelation due to the instruments lagging.

According to [41], System GMM estimates the following model:

$$y_{it}=\mathsf{\alpha }{y}_{i,t-1}+X_{it}^{\prime }\mathsf{\beta }+{\mathsf{\epsilon }}_{it}$$

(1)

$${\mathsf{\epsilon }}_{it}={\mathsf{\mu }}_i+{\mathsf{\nu }}_{it}$$

(2)

$$E\left({\mathsf{\mu }}_i\right)=E\left({\mathsf{\nu }}_{it}\right)=E\left({\mathsf{\mu }}_i{\mathsf{\nu }}_{it}\right)=0$$

(3)

The disturbance term ${\mathsf{\epsilon }}_{it}$ has two components; ${\mathsf{\mu }}_i$ as fixed effects and ${\mathsf{\nu }}_{it}$ as idiosyncratic shocks, which are orthogonal to each other.

$$\Delta y_{it}=(\mathsf{\alpha }-1)\Delta y_{i,t-1}+\Delta X_{it}^{\prime }\mathsf{\beta }+\Delta {\mathsf{\nu }}_{it}$$

(4)

$$E\left[\Delta w_{it}{\mathsf{\mu }}_i\right]=0$$

(5)

where $w_{it}$ is the instrumenting variable, which is uncorrelated with the fixed effects.

The Difference GMM estimator created by [42] is enhanced by the System GMM estimator created by [43,44]. Equation (4) is added to the original Equation (1) in the System GMM estimator, increasing the estimate’s efficiency. While the Difference GMM instrument differences with levels, the System GMM instrument levels with differences [41]. To allow the model to include more instruments, the System GMM estimator additionally assumes that there is no association between the initial differences of instrumental variables in Equation (7) and fixed effects. v_it should not be serially correlated. However, if it should, for instance, be serially correlated of order 1, then, for instance, y_i,t−2 is endogenous to the v_i,t−1 in the error term in differences, ∆ε_it = v_it − v_i,t−1, making it a potentially invalid instrument. The researcher would need to restrict the instrument set starting with a third lag or longer [41].

Finally, the probability of the F statistic should be less than 0.1, which indicates the significance of the estimated model.

For each of the three country clusters, we estimated the effect of CO₂ emissions and its square in the main model and we made a robustness check in another 4 models. We used the structure for the determinants of FDI inflows as in [45,46].

The following are the reduced form equations:

Model 1: Equation for the Main Variables

$${fdi}_{it}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{fdi}_{i,t-1}+{{\mathsf{\alpha }}_{2}d\_carbon}_{it}+{{\mathsf{\alpha }}_{3}d\_carbon2}_{it}+{\mathsf{\epsilon }}_{it}$$

(6)

Model 2: Equation for only the Base Factors

$$\begin{array}{ll}{fdi}_{it}=& {\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{fdi}_{i,t-1}+{{\mathsf{\alpha }}_{2}d\_carbon}_{it}+{{\mathsf{\alpha }}_{3}d\_carbon2}_{it}+{{\mathsf{\alpha }}_{4}d\_gov}_{it}\\ & +{{\mathsf{\alpha }}_{5}d\_humcap}_{it}+{{\mathsf{\alpha }}_{6}d\_infra}_{it}+{{\mathsf{\alpha }}_{7}findev}_{it}+{\mathsf{\epsilon }}_{it}\end{array}$$

(7)

Model 3: Equation for the Motivation Factor I

$$\begin{array}{ll}{fdi}_{it}=& {\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{fdi}_{i,t-1}+{{\mathsf{\alpha }}_{2}d\_carbon}_{it}+{{\mathsf{\alpha }}_{3}d\_carbon2}_{it}+{\mathsf{\alpha }}_4{d\_gov}_{it}\\ & +{{\mathsf{\alpha }}_{5}d\_humcap}_{it}+{{\mathsf{\alpha }}_{6}d\_infra}_{it}+{{\mathsf{\alpha }}_{7}findev}_{it}+{{\mathsf{\alpha }}_{8}d\_lngdp}_{it}+{\mathsf{\epsilon }}_{it}\end{array}$$

(8)

Model 4: Equation for the Motivation Factor II

$$\begin{array}{ll}{fdi}_{it}=& {\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{fdi}_{i,t-1}+{{\mathsf{\alpha }}_{2}d\_carbon}_{it}+{{\mathsf{\alpha }}_{3}d\_carbon2}_{it}+{{\mathsf{\alpha }}_{4}d\_gov}_{it}\\ & +{{\mathsf{\alpha }}_{5}d\_humcap}_{it}+{{\mathsf{\alpha }}_{6}d\_infra}_{it}+{{\mathsf{\alpha }}_{7}findev}_{it}+{{\mathsf{\alpha }}_{8}d\_trade}_{it}+{\mathsf{\epsilon }}_{it}\end{array}$$

(9)

Model 5: Equation for the Motivation Factor III

$$\begin{array}{ll}{fdi}_{it}=& {\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{fdi}_{i,t-1}+{{\mathsf{\alpha }}_{2}d\_carbon}_{it}+{{\mathsf{\alpha }}_{3}d\_carbon2}_{it}+{{\mathsf{\alpha }}_{4}d\_gov}_{it}\\ & +{{\mathsf{\alpha }}_{5}d\_humcap}_{it}+{{\mathsf{\alpha }}_{6}d\_infra}_{it}+{{\mathsf{\alpha }}_{7}findev}_{it}+{{\mathsf{\alpha }}_{8}d\_phycap}_{it}+{\mathsf{\epsilon }}_{it}\end{array}$$

(10)

4. Discussion

According to

Table 1. System GMM results.

Dependent Variable: fdi
Variable	Model 1	Model 2	Model 3	Model 4	Model 5
l.fdi	0.6031 ***,	0.5687 ***	0.5031 ***	0.5614 ***	0.5680 ***
	(0.0097)	(0.0050)	(0.0032)	(0.0037)	(0.0036)
d_carbon	9.8546 ***	3.2420 ***	0.6761 ***	1.8000 ***	3.0997 ***
	(1.5597)	(0.6799)	(0.2544)	(0.5099)	(0.4997)
d_carbon2	−3.4758 ***	−1.5738 ***	−0.7560 ***	−1.1273 ***	−1.4200 ***
	(0.4818)	(0.2137)	(0.1255)	(0.1668)	(0.1584)
d_gov		17.7298 ***	31.2761 ***	13.8973 ***	23.8498 ***
		(3.1407)	(2.3965)	(2.5655)	(2.0782)
d_humcap		−7.6422 **	−11.5443 ***	−11.0929 ***	−7.8155 ***
		(3.2251)	(2.2418)	(2.4783)	(2.0046)
d_infra		14.6817 ***	13.5195 ***	13.3436 ***	15.3577 ***
		(1.1713)	(1.4774)	(1.0162)	(1.0635)
findev		−5.5755 **	−7.5022 ***	−5.4411 ***	−6.7631 ***
		(2.2911)	(2.0281)	(1.8191)	(1.9846)
d_lngdp			7.3511 *
			(3.8431)
d_trade				0.1199 ***
				(0.0193)
d_phycap					0.0610 **
					(0.0299)
constant	1.8725 ***	2.4348 ***	3.3178 ***	2.6481 ***	2.8119 ***
	(0.1800)	(0.8482)	(0.7757)	(0.6757)	(0.7293)
Prob > F	0.000	0.000	0.000	0.000	0.000
Observations	2967	2967	2967	2967	2967
Countries	129	129	129	129	129
Instruments	19	64	89	73	77
Hansen	0.310	0.539	0.920	0.340	0.524
AR(2)	0.628	0.500	0.452	0.481	0.494

Notes: The values in brackets are two-step robust standard errors. ***, **, and * denote significance levels at % 1, % 5, and % 10 respectively. The Hansen test of over-identification is under the null that all instruments are valid. All variables are treated endogenously according to Appendix F. The p-values are reported for the Hansen and AR(2) tests.

, the estimated models are meaningful since their Prob > F is less than 0.1, Hansen p-values are over 0.1, and AR(2) p-values are over 0.1 and not equal to 1.0, which indicate an appropriate number of lags are instrumented.

Table 1 indicates that the lag of FDI inflows (l.fdi) has a positive significant effect on the current level of FDI inflows (fdi) in all specifications. This result supports Hypothesis 3 that an investor inertia exists for MNCs and foreign investors since past FDI inflows become significant determinants of current FDI inflows due to strategic complementarity between past and current levels of FDI inflows. Thus, there is a trend in FDI inflows in the sense that countries exposed to a high number of FDI inflows are exposed to an even higher number of FDI inflows in the following year. It also implies that inward FDI is a rooted concept shaped by base and motivation factors in the country, hence it cannot be reversed in the short term. Since we used annual data, we can infer that the countries exposed to FDI inflows converge slowly in the short run concerning exposed FDI inflows.

Model 1 in Table 1 reports the results just for the main variables. CO₂ emissions (d_carbon) has a positive significant and its square (d_carbon2) has a negative effect on FDI inflows (fdi). This result supports our hypothesis that there is an inverted-U curve relationship between CO₂ emissions and FDI inflows. The results are robust with respect to the other four specifications.

Table 2. Testing the presence of an inverse-U-shaped curve.

Specification: f(x) = x2
Extreme Point: 1.4175
	Lower bound	Upper bound
Interval	−4.3067	6.5626
Slope	39.79	−35.77
t-value	7.3133	−6.85
p > |t|	1.26 × 10−11	1.39 × 10−10
Overall test of presence of an inverse-U shape
t-value		6.85
p > |t|		1.39 × 10−10

Notes: The null hypothesis is the monotone or U-shaped relationship, and the alternative hypothesis is the inverse-U-shaped relationship.

presents the results of an inverse U-shaped relationship test.

Model 2 in Table 1 reports the results for just the base variables together with the main variables.

Governance (d_gov) has a positive significant effect on FDI inflows (fdi). Because they reduce opportunism, foster transactional trust in financial transactions, and eventually persuade international players to engage in cross-border transactions, formal institutions (constitutions, prudential laws, regulations, government policies, etc.) and informal institutions (traditions, habits, customs, etc.) have an impact on how investors perceive a nation and their willingness to invest there. By defining the choice set, governance establishes the transaction and production costs, which in turn establish the viability and profitability of conducting business, whether domestically or internationally. Thus, an improvement in governance increases FDI inflows. This result supports [47,48,49,50,51,52]. The result is robust.

Human capital (d_humcap) has a negative significant effect on FDI inflows (fdi). Mainly, there are three reasons behind this negative effect. First, by allocating resources to enhance its human capital via education and skill enhancement, a nation may increase the proficiency and productivity of its labor force. Consequently, the increase in labor expenses may diminish the country’s appeal to international investors who are looking for affordable labor. Second, as the host country’s human capital grows, it may reduce its reliance on foreign knowledge and technology, which might diminish the motivation for foreign enterprises to invest in that host country since they may see less prospects to use their technology or knowledge for specific advantages. Last, as the level of human capital develops, host countries have the potential to elevate their regulatory and quality requirements, which might be advantageous for local businesses and customers, but it may also impose supplementary expenses and obligations on international investors, so reducing the appeal of inward FDI. This result rejects the empirical findings of [53,54,55,56]. Hence, we can conclude that human capital for this sample of countries is a motivation factor, not a base factor, that a lower level of human capital attracts MNCs and foreign investors, not a higher level of it. The result is robust.

Physical infrastructure (d_infra) has a positive significant impact on FDI inflows (fdi). Enhancing physical infrastructure is linked to increased accessibility and lower transportation expenses. Without public physical infrastructure, local and multinational businesses would have to construct their own infrastructure, which would result in resource waste and duplication. As a result, these businesses would operate less efficiently [45]. The investment climate for FDI is enhanced by better physical infrastructure since foreign investors can receive subsidies for their whole investment, increasing the rate of return [45]. This result supports the empirical findings of [57,58,59,60,61]. Hence, physical infrastructure is a base factor for this sample of countries. The result is robust.

Financial development (findev) has a negative significant effect on FDI inflows (fdi). First, if a highly developed financial market is overly complex or if there are high transaction costs associated with financial activities, it could discourage foreign companies from investing in the host country. Second, if a highly developed financial market led to greater exchange rate volatility, excessive volatility may pose risks for investors, especially if there is uncertainty about the stability of the local currency. This result supports the empirical findings of [46,62]. Hence, we can conclude that financial development for this sample of countries is a motivation factor, not a base factor, that a lower level of financial development attracts MNCs and foreign investors, not a higher level of it. The result is robust.

Market size (d_lngdp) has a positive significant impact on FDI inflows (fdi). A growing GDP, which is reflective of a larger market, is an attractive factor for foreign investors seeking to expand their operations overseas [63]. The appeal of growing economies and their potential for market expansion can be attractive to foreign investors [64]. This result supports the empirical findings of [63,64,65]. Hence, market size is a base factor for this sample of countries.

International trade or trade openness (d_trade) has a positive significant impact on FDI inflows (fdi). First, because it indicates a nation’s readiness to welcome foreign investment, economies that are more open to international trade get larger inflows of foreign direct investment. Foreign investors typically focus on nations that are seen as being open to international trade, especially when making efficiency-seeking FDI [66]. Second, when a company engages in international trade and receives increasing demand for its products in a foreign market, it may decide to establish a physical presence in that market through FDI, which allows the company to be closer to its customers, understand local preferences, and respond more effectively to market dynamics. Last, as countries engage in international trade, they often work to reduce trade barriers such as tariffs and quotas, which makes it easier for businesses to operate across borders since foreign companies are more inclined to invest directly in foreign markets to take advantage of the improved business environment. This result supports the empirical findings of [67,68,69]. Hence, market size is a base factor for this sample of countries.

Physical capital (d_phycap) has a positive significant impact on FDI inflows (fdi). First, host countries endowed with improved physical capital, such as modern manufacturing facilities and technology, may enhance the efficiency of production processes and attract foreign investors to benefit from state-of-the-art facilities, advanced technology, and streamlined production processes, leading to cost savings and increased competitiveness. Second, host countries endowed with abundant and easily accessible resources may attract foreign investors seeking a stable and secure supply of inputs for their production processes. This result supports [70]. Hence, market size is a base factor for this sample of countries.

5. Conclusions

In this study, we augmented the current literature on environmental economics through the following contributions. First, we tested two alternative hypotheses: the pollution haven hypothesis and green haven hypothesis, and differentiated this from previous studies by focusing on the impact of CO₂ emissions on inward FDI. Previous empirical studies investigate the validity of these hypotheses by focusing on the effects of environmental policies and regulations on FDI and have not provided conclusive results. We hypothesized that developing countries having lax environmental standards attract more FDI with an increase in CO₂ emissions (pollution haven effect) while an increase in CO₂ emissions in countries with stricter environmental regulations deters FDI (green haven effect). Second, to that goal, we innovated by incorporating an inverse-U-shaped relationship between CO₂ emissions and inward FDI. Third, we applied a dynamic Blundell–Bond System GMM methodology on a substantial sample of 124 economies from 1997 to 2021. The estimation results indicate that CO₂ emissions have an inverted-U-shaped relationship with inward FDI, leading us to conclude that the pollution haven effect tends to outweigh the green haven effect in countries having lax environmental standards and the dominance of the green haven effect in countries having stringent environmental regulations.

These findings have important policy implications. Governments should strive to detect and maintain carbon dioxide emissions at levels that maximize inward foreign direct investment. This entails achieving a delicate equilibrium where emissions are sufficiently minimized to guarantee a sustainable environment, while avoiding excessive restrictions that impede industrial and economic operations.

The negative effect of CO₂ emissions on inward FDI suggests that high-income economies should not only adopt more stringent environment regulations but also enforce them strictly to reduce CO₂ emissions and attract FDI. This finding also supports the evidence that existing environmental policies might be unable to cater to the foreign component of CO₂ emissions. Policy makers should offer incentives for research and development activities focused on sustainable and low-carbon technologies, which stimulates innovation and positions the country as a hub for green technology development.

At the same time, the existence of pollution havens in low-income countries stresses that policy makers should strengthen environmental regulations to mitigate the negative impact of increased CO₂ emissions by implementing strict emission standards and penalties for non-compliance. At the same time, they should introduce incentives for companies that adopt eco-friendly technologies and practices, which include tax breaks, subsidies, or other financial rewards. They should encourage technology transfer and collaboration between foreign and domestic companies to facilitate the adoption of cleaner technologies. They should invest in research and development to promote innovation in sustainable practices and green technologies. Hence, their main target should be transforming from dirty FDI to clean FDI in order to benefit in terms of improved environmental quality from clean inward FDI. Low-income countries have a window of opportunity with respect to their low levels of CO₂ emissions. They should act fast to actualize this transition, otherwise their future emission levels will be higher than high-income countries with the pollution haven effect increasing it further. All countries should prioritize investments in renewable energy infrastructure, such as solar, wind, and hydroelectric power, which not only attracts green investors but also contributes to a sustainable and low-carbon energy sector.

It is essential to have strong mechanisms for monitoring and enforcing adherence to environmental norms. This may encompass routine emissions reporting, audits, and sanctions for failure to comply. Promoting and endorsing CSR efforts may enhance the attractiveness of attracting FDI by appealing to investors who are seeking to enhance their environmental impact.

There are the following limitations in this study: We proxied environmental quality with only the level of CO₂ emissions due to data availability. We had to start the analysis from 1996 since the components of governance data start in 1996. One future direction is that we found financial development has a negative significant effect on FDI inflows. Hence, it is the case that foreign investors prefer to finance their investment from financial intermediaries rather than in the host country. This negative significant effect seems worthy of studying. This study did not investigate the relationship between the sectoral composition of CO₂ emissions and FDI due to insufficient data for the countries included. There are plenty of sources of CO₂ emissions (see e.g., [71,72,73]), and their impacts on FDI could be divergent. Future research could test the hypotheses focusing on different sectors of CO₂ emissions (e.g., transport sector, manufacturing sector, etc.).

Thanks to one of the reviewers, we have two possible future studies. The first direction is that the relationship between carbon intensity (CO₂ emissions/GDP) and FDI may be analyzed. The second direction is that a spatial–temporal analysis may be carried out for the relationship between CO₂ emissions and FDI across continents and time. Also, outward FDI may be analyzed instead of inward FDI.