Carbon Border Adjustment Mechanism as a Catalyst for Greenfield Investment: Evidence from Chinese Listed Firms Using a Difference-in-Differences Model

Abstract

Research on the EU’s Carbon Border Adjustment Mechanism (CBAM) has predominantly examined its implications for climate governance and export trade yet overlooked how enterprises adapt their foreign investment strategies. Using panel data from Chinese listed companies between 2011 and 2022, this study employs the CBAM as a quasi-natural experiment and applies a difference-in-differences (DID) model for analysis. Our findings indicate that the CBAM has a significant positive impact on outward greenfield investments, as robustly validated through a series of rigorous robustness checks. Mechanism analysis reveals two operational channels: trade restructuring effect (reduced export shares) and innovation-driven demand effect (enhanced R&D intensity). Heterogeneity tests further indicate more substantial CBAM responsiveness among eastern coastal firms, non-state-owned enterprises, and those pursuing horizontal production-oriented expansions. This study contributes to the literature on CBAM’s effects and offers practical recommendations for enterprises to mitigate CBAM’s impact via greenfield investments.

1. Introduction

Global climate change is intensifying, with carbon emission mitigation now constituting international consensus [1]. The United Nations has specified in the Paris Agreement that the rise in global average temperature must remain below 2 °C [2]. However, fragmented national climate commitments risk carbon leakage—where one nation’s emission reductions are offset by others’ increased outputs, diminishing global mitigation efficacy [3]. Since climate change entered global governance agendas in the late 1980s, the EU has actively shaped climate leadership, pioneering emission control through its 2005 Emissions Trading System (EU ETS). Yet, reduced carbon quotas under EU ETS highlighted carbon leakage and industrial competitiveness challenges [4]. The 2019 European Green Deal introduced CBAM (Carbon Border Adjustment Mechanism) as a carbon leakage countermeasure. Initially targeting six sectors—steel, cement, aluminum, fertilizers, electricity, and hydrogen—CBAM entered legislative implementation through three key phases: draft release (July 2021), parliamentary approval (April 2023), and operational timelines (transition period October 2023—formal implementation January 2026).

China remains the world’s largest carbon emitter [5], as documented in IEA’s 2023 Carbon Emissions Report with 12.6 billion metric tons of CO₂ equivalent emissions. The EU constitutes a pivotal market for Chinese steel and aluminum exports, representing 98.8% of carbon-intensive Sino-European trade during 2017–2023, with steel representing 75.1% and aluminum 23.7%. This concentration in CBAM-covered sectors exposes China’s export profile to substantial regulatory risks. With five of the six CBAM-targeted industries (excluding electricity) actively exporting to the EU, Chinese manufacturers face unavoidable CBAM impacts. Therefore, strategic countermeasures have become imperative for China to mitigate CBAM’s trade impacts within global decarbonization trends.

China’s “Going Global” strategy implementation since 2000 has propelled its emergence as a leading OFDI source [6]. According to the Ministry of Commerce, China’s non-financial outward foreign direct investment surged from 60.07 billion (2011) to 130.13 billion (2023), representing a growth rate of 116.6% [7]. OFDI serves dual strategic purposes for Chinese enterprises: facilitating integration into global economic circuits and addressing carbon tariff mechanisms alongside other green trade barriers. Greenfield investment and cross-border M&As constitute principal implementation channels. Specifically, greenfield investment refers to the inception of new companies overseas—including production facilities, R&D centers, and distribution networks—to build fresh production capacity [8]. In contrast, cross-border M&As involve acquiring stakes in existing foreign firms to gain operational control through asset or equity transfers. Two critical distinctions emerge: On one hand, greenfield investments exhibit heightened sensitivity to the business, tax, and other policy environments of the host country, particularly tax regimes [9,10]. The lower tax rates in the host country tend to enhance the attractiveness of greenfield investment for enterprises. On the other hand, we compare the greenfield investment and cross-border merger and acquisition data from FDI Markets and Zephyr databases. As shown in Figure 1, it is found that the number and amount of greenfield investments by enterprises are significantly higher than those of cross-border mergers and acquisitions. Cross-border mergers and acquisitions may exhibit occasional and random patterns, while greenfield investment behavior by enterprises may be relatively more continuous and stable. Given greenfield investment’s superior reliability in reflecting authentic OFDI behaviors and data availability advantages, this study focuses on greenfield investment as the primary OFDI analytical framework, with subsequent empirical validation.

This study makes three principal contributions to the extant literature: First, prior studies have predominantly examined the effects of CBAM on international trade and carbon leakage, with limited attention to corporate outward foreign direct investment (OFDI) behavior. Given that OFDI, especially greenfield investment, may be an effective strategy for Chinese enterprises to mitigate the trade shocks of CBAM, this paper extends the existing research on carbon tariffs to the realm of corporate greenfield investment. Second, we innovatively dissect CBAM’s operational mechanisms driving greenfield investments via dual channels: trade restructuring effects and innovation demand-driven effects. Third, the current body of research is primarily conceptual, with scarce empirical support, and is largely based on general equilibrium approaches. Our empirical strategy employs difference-in-differences (DID) methodology to establish causal inferences.

The paper proceeds as follows: Section 2 synthesizes a literature review on CBAM and greenfield investment. Section 3 develops theoretical hypotheses grounded in trade restructuring and innovation demand frameworks. Section 4 outlines the research framework, including the DID formulation, variable definition, and data compilation. Section 5 presents baseline estimates, robustness checks, and mechanism analyses. Section 6 explores heterogeneous treatment effects. Section 7 culminates with a discussion of conclusions, corresponding policy recommendations, limitations, and future research directions.

2. Literature Review

2.1. Existing Research on CBAM

The EU Carbon Border Adjustment Mechanism is a focal issue in current global governance. The existing literature has explored its overall impacts on the international community. Regarding climate governance, carbon tariffs can effectively reduce global carbon emissions [11]. However, focusing solely on direct emissions fails to curb carbon leakage, making embedded carbon emissions critical for policy design [12]. In socioeconomic terms, developing countries face welfare losses and economic setbacks due to high carbon tariffs [13]. Gu et al. further demonstrated negative impacts on economic development, household welfare, trade volumes, and energy-intensive industry outputs in targeted countries (China, India, Brazil, and South Africa) [14].

As China is a key EU import partner, numerous studies have specifically analyzed CBAM’s effects on China. Some scholars argue that CBAM severely disrupts China’s foreign trade and investment. Zhu et al. employed the GTAP-E model to simulate CBAM’s impact on China-EU exports [15], finding that additional tariffs reduce Chinese export prices to the EU. Rising export costs trigger diversion effects, ultimately decreasing export volumes [16]. In mid-to-long-term analyses, Wang et al. identified challenges such as reverse supply chain investments and heightened overseas investment uncertainties in related industries [17]. Conversely, other studies have highlighted potential opportunities. Fu et al. suggested that carbon barriers could foster green trade advantages, elevating China’s role in global climate governance [18]. Chang posits that declining international market shares may incentivize China to strengthen low-carbon practices to enhance export competitiveness [19]. The most directly relevant study is Sun et al. [20], which examined Chinese FDI in the EU. Their findings indicate that CBAM increases investments in the EU, shifts investment patterns (toward non-EU neighboring countries), and boosts investments in renewable energy sectors.

Additionally, industry-level analyses reveal that excessive carbon tariffs surpass corporate profit margins, suppressing exports in high-carbon industries [21]. Competitive advantages in developed countries further reduce energy-intensive enterprises, thereby promoting industrial decarbonization [22,23].

In summary, existing scholarship has extensively analyzed the EU Carbon Border Adjustment Mechanism (CBAM)’s multifaceted impacts on global governance and China’s economic development from diverse perspectives. Nevertheless, three critical research gaps persist. First, existing studies predominantly remain theoretical, lacking empirical validation of CBAM’s actual impacts. Second, policy effect analyses typically operate at national levels, with scant attention to enterprise-level adaptation strategies. Third, the current recommendations are deficient in robust empirical substantiation, thereby rendering the assessment of their practical applicability and feasibility challenging. Moreover, while the prevailing literature emphasizes CBAM’s negative economic consequences for China, potential developmental opportunities remain underexplored.

2.2. Influencing Factors of Greenfield Investment

Greenfield investment serves as a crucial instrument for China’s participation in global economic integration and supply chain development. Research in this field has matured increasingly. As a primary form of outward foreign direct investment (OFDI), greenfield investment is shaped by three key determinants: home-country policy environments [24], host-country market characteristics [25], and firm-specific factors [26]. Regarding home-country policies, Qi and Guan developed the “OFDI-S” model, categorizing institutional influences into three distinct effects: guidance, crowding-out, and weakening [27]. In host-country markets, political risks negatively affect OFDI decisions. Notably, coastal nations demonstrate greater sensitivity to political uncertainties than their landlocked counterparts [28]. Chinese multinationals primarily target countries offering high institutional quality, abundant natural resources, and lower economic development levels—motivated by resource accessibility and cost advantages [29]. Firm-level analyses reveal three principal motivations: resource acquisition, market expansion, and sustainable development [30]. Higher productivity correlates with OFDI preferences over exports [31], while tax avoidance incentives strengthen OFDI commitments [32].

The evolving global trade landscape amplifies trade policy’s role in OFDI decisions [33,34]. Yan et al. demonstrate that anti-dumping measures initially stimulate OFDI by impeding trade exports and forcing productivity improvements but ultimately undermine this effect through increased product concentration [35]. Lv et al. illustrate that the Belt and Road Initiative (BRI) has prompted a substantial increase in greenfield OFDI by Chinese firms, particularly in core infrastructure sectors including energy, transportation, and telecommunications [36].

Existing studies extensively examine OFDI determinants and their trade barrier mitigation effects. However, limited empirical research addresses EU CBAM’s specific impacts and causes on Chinese OFDI patterns. Cardamone and Scoppola confirm differential tariff responses between developed and developing economies [37]. This study contributes by analyzing EU carbon tariffs’ influence on Chinese OFDI, offering both theoretical insights and practical strategies for navigating evolving trade dynamics.

3. Theoretical Analysis and Research Hypothesis

3.1. The Impact of CBAM on Enterprises’ Greenfield Investment

The Carbon Border Adjustment Mechanism (CBAM) exerts dual impacts on enterprise export trade and carbon emissions [11]. While formally structured as a tariff-based mechanism, CBAM essentially functions as a non-tariff barrier. First, its additional levies on carbon-intensive industries immediately escalate export costs, directly suppressing EU-bound exports from China’s five key sectors and potentially eroding their global market share. Second, as an environmental instrument targeting carbon leakage reduction [38], CBAM creates sustained indirect pressures for enterprises to fundamentally transform operations through three pathways: lowering carbon emissions, adopting green production technologies, and enhancing environmental accountability.

Greenfield investment emerges as a strategic response to trade barriers, enabling enterprises to offset export losses and upgrade technological capabilities. Under this framework, enterprises make greenfield investments partly for the purpose of seeking new markets and new technologies. On the one hand, under CBAM’s stringent trade controls, firms may pursue resource-intensive greenfield investments to mitigate trade risks and sustain overseas market presence [39]. On the other hand, China’s reverse gradient outward foreign direct investment in developed countries facilitates dual objectives: overseas market penetration and absorption of advanced technologies, representing typical market-technology-seeking behavior [40,41]. With CBAM’s anticipated sectoral expansion and potential global policy emulation [18], enterprises face increasing urgency to conduct technology-seeking OFDI. This strategic shift aims to enhance internal green technological capacities and adapt to evolving green trade barriers.

From the perspective of the impact of CBAM on Chinese enterprises, enterprises need to take corresponding measures to replace exports and reduce carbon emissions, so as to overcome trade barriers. In terms of the motivations driving OFDI, both market opportunities and technological advancements play crucial roles in encouraging enterprises to engage in greenfield investments. To sum up, greenfield investment may be the first choice and most effective measure for enterprises to deal with CBAM. We therefore propose the following:

H1. 

CBAM increases enterprise greenfield investment.

3.2. Pathway Analysis of CBAM’s Impact on Corporate Greenfield Investments

3.2.1. Trade Restructuring Effect

The implementation of carbon tariffs imposes dual pressures on Chinese enterprises: escalating export costs and eroding competitive advantages in EU markets. Specifically, this erosion manifests through diminishing market share, constrained international trade participation, and potential deceleration in globalization progress for five key industries.

Empirical evidence reveals an inverse relationship between trade openness and outward FDI [42]. Horizontal investment theory posits that heightened EU market entry thresholds motivate firms to establish host-country production facilities, circumventing trade barriers through OFDI’s “barrier-jumping effect” [43]. When barrier-avoidance benefits outweigh multinational operational costs, overseas investment becomes strategically preferable [31].

“Trade restructuring” refers to the interaction between OFDI and export trade, comprising “trade substitution” and “trade creation”. The substitution effect occurs when Chinese enterprises replace direct exports with EU-based production, preserving market share through localized operations [35]. Host-country production enables the adoption of EU-compliant low-carbon technologies, effectively mitigating the additional costs brought by carbon taxes. Given greenfield investments’ heightened sensitivity to host nations’ emission policies and environmental standards [44], under the impact of CBAM, enterprises may choose more greenfield investment in EU countries with abundant clean energy and more developed green technology to increase the total amount of enterprise green field investment.

The “Trade creation effect” usually refers to the fact that an enterprise invests in the host country and produces related products but does not sell them in the host country or home country, rather exporting them to third countries, thus “creating” trade [45]. Chinese enterprises may establish manufacturing facilities in non-EU jurisdictions with lower CBAM exposure and trade barriers, enabling indirect EU exports that circumvent carbon tariff constraints. Both substitution and creation pathways consequently drive greenfield investment growth. We therefore hypothesize that:

H2a. 

CBAM implementation reduces product export share while increasing greenfield investment through trade substitution and creation effects.

3.2.2. Innovation-Driven Demand Effect

Functioning as a carbon leakage remedy [46], the CBAM operates through tariff imposition on high-carbon imports to align with EU carbon pricing mechanisms and prevent “carbon leakage” [21]. The Porter Hypothesis suggests that stringent environmental regulations compel firms to pursue technological innovations. As a market-driven environmental instrument, CBAM simultaneously pressures Chinese industries toward low-carbon transitions and stimulates corporate innovation demands [47]. Green technologies constitute critical pathways for emission reduction and sustainable development [48], enabling enterprises to alleviate CBAM-induced cost pressures through multiple pathways including carbon intensity reduction, resource efficiency optimization, and green product development. Consequently, EU-exporting enterprises in five key industries exhibit heightened motivations for green innovation and environmental performance enhancement.

OFDI demonstrates reverse green technology spillovers that enhance home-country innovation capacity [49]. First, greenfield investments’ operational control facilitates scale effects, particularly post-2012, in driving China’s green innovation surge [47]. Overseas subsidiaries access host-country green technology incentives (subsidies and tax benefits) and assimilate cutting-edge sustainable technologies. Second, the knowledge spillover mechanism [50] renders greenfield investments more effective in technology transfer than cross-border M&A. Greenfield investments ensure better control over acquired technical knowledge [51], allowing the reverse diffusion of environmental management expertise to parent companies and subsequent emission reductions [52]. These factors motivate enterprises to pursue greenfield investments for EU market competitiveness through low-carbon product development.

Enterprises navigate dual pressures from external dynamic trade reconfiguration and internal innovation constraints—capital inadequacy, technological deficits, and skilled labor shortages. Greenfield investments serve as pivotal mechanisms for technological advancement, enabling the assimilation of global frontier technologies and best practices to address these compounded challenges. We therefore hypothesize that:

H2b. 

CBAM implementation stimulates corporate innovation demand (manifested through R&D intensity increases), thereby promoting greenfield investments.

4. Research Design

4.1. Empirical Framework

To effectively eliminate the confounding effects of temporal trends and inherent firm-level heterogeneity on the results, thereby enabling more precise identification of the causal impact of the Carbon Border Adjustment Mechanism on Chinese enterprises’ greenfield investment, this study employs a difference-in-differences (DID) model. Within a quasi-natural experiment framework, we utilize the European Green Deal, enacted by the European Union in 2019, as both the temporal cutoff and grouping criterion. This approach quantifies the net effect of CBAM on greenfield investment by comparing pre- and post-policy differences in outcomes between treated and control groups. Crucially, the counterfactual framework inherent in the DID methodology mitigates selection bias stemming from the non-random implementation of the carbon tariff policy. All regression analyses apply cluster-robust standard errors at the firm level to account for potential serial correlation across time periods within individual entities. The baseline specification is as follows:

$${\mathrm{L}\mathrm{n}\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1\mathrm{d}\mathrm{i}\mathrm{d}+{\mathsf{\gamma }}_1{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\xi }}_{\mathrm{i}\mathrm{t}}$$

(1)

The dependent variable ${\mathrm{L}\mathrm{n}\mathrm{G}\mathrm{I}}_{\mathrm{i}\mathrm{t}}$ denotes the natural logarithm of enterprise $\mathrm{i}$’s greenfield investment value (plus one) in year $\mathrm{t}$. The key explanatory variable $\mathrm{d}\mathrm{i}\mathrm{d}$ constitutes the interaction term between treatment group designation and policy timeline, operationalized as a binary indicator (1 for the post-CBAM period, 0 otherwise). Coefficient ${\mathsf{\beta }}_1$ captures CBAM’s net treatment effect on greenfield investment. The vector ${\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i}\mathrm{t}}$ encompasses firm-, industry-, and region-level covariates, with ${\mathsf{\mu }}_{\mathrm{i}}$ and ${\mathsf{\lambda }}_{\mathrm{t}}$ representing enterprise-fixed effects and year-fixed effects, respectively. The stochastic error term ${\mathsf{\xi }}_{\mathrm{i}\mathrm{t}}$ accounts for unexplained variance.

In addition, in order to further examine CBAM’s effects on aggregate outward foreign direct investment and cross-border M&As, we develop supplemental regression specifications. Formally, we let ${\mathrm{O}\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}$ denote firm $\mathrm{i}$’s total overseas investment transactions in year $\mathrm{t}$, and ${\mathrm{C}\mathrm{M}\mathrm{A}}_{\mathrm{i}\mathrm{t}}$ captures its corresponding M&A expenditures during the same period. Other model components maintain consistency with Equation (1).

$${\mathrm{O}\mathrm{F}\mathrm{D}\mathrm{I}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1\mathrm{d}\mathrm{i}\mathrm{d}+{\mathsf{\gamma }}_1+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\xi }}_{\mathrm{i}\mathrm{t}}$$

(2)

$${\mathrm{C}\mathrm{M}\mathrm{A}}_{\mathrm{i}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\beta }}_1\mathrm{d}\mathrm{i}\mathrm{d}+{\mathsf{\gamma }}_1+\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\xi }}_{\mathrm{i}\mathrm{t}}$$

(3)

4.2. Variable Selection

4.2.1. Explained Variables

Greenfield Investment (LnGI): The annual greenfield investment transaction values of enterprises are sourced from the FDI Markets database and Zephyr database. Following the methodology of Zheng and Chen [53], we apply a unity-adjusted logarithmic transformation to these transaction values to mitigate the influence of extreme observations and heteroscedasticity. This transformed metric serves as our measure of greenfield investment activity for listed firms. The study further conducts sensitivity tests on parameter specifications, with the results confirming the robustness of our dependent variable operationalization.

4.2.2. Core Explanatory Variables

EU Carbon Border Adjustment Mechanism (DID): $\mathrm{d}\mathrm{i}\mathrm{d}=\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\times \mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}$. The treatment group dummy variable ($\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}$) identifies firms in CBAM-regulated sectors—steel, aluminum, cement, fertilizers, and hydrogen—based on the EU’s 2019 European Green Deal framework. Notably, China’s power sector is excluded from the treatment group due to absent EU exports. These exposed firms receive a value of 1, whereas non-affected enterprises (control group) are coded as 0. The temporal dimension ($\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}$) adopts a value of 1 for observation years since policy implementation (2019 onward) and 0 otherwise.

4.2.3. Control Variables

First, building on established methodologies, we incorporate the following firm-level control variables based on prior studies:

Enterprise Size (LnSize): Measured by the natural logarithm of listed firms’ operating revenue [7]. Larger firms typically possess greater financial capacity and operational breadth, thereby exhibiting a higher propensity for greenfield investments.

Financing constraint measurement: Frequently employed metrics encompass the Kaplan–Zingales (KZ), Whited–Wu (WW), and Size–Age (SA) indicators. The KZ and WW indices incorporate endogenous financial variables (e.g., cash flow and leverage ratios), potentially introducing estimation bias. To mitigate endogeneity concerns, we employ the SA index developed by Hadlock and Pierce as our primary financing constraint measure [54].

Management Overseas Background (MOB): Coded as 1 for firms with overseas-experienced directors/supervisors and 0 otherwise. International exposure among executives reduces information asymmetry risks in cross-border investments, justifying this variable’s inclusion.

Top Five Suppliers’ Purchase Amount (AMONT): Calculated as the log-transformed sum of procurement expenditures from a firm’s five largest suppliers, as disclosed in annual reports. This metric reflects supply chain concentration risk—higher values indicate greater dependence on key suppliers. By controlling for AMONT, we isolate the influence of inherent supply chain structures from carbon tariff policy effects, thereby enhancing causal identification accuracy.

Second, we control for industry-specific factors influencing corporate investment decisions:

Market Concentration (HHI): Measured by the four-digit Herfindahl–Hirschman Index, where lower HHI values indicate higher market competition [55]. The index is computed as: $\mathrm{H}\mathrm{H}\mathrm{I}=\sum {\left({\mathrm{X}}_{\mathrm{i}}/\mathrm{X}\right)}^2$, where Xi denotes firm-level operating revenue and X represents total industry operating revenue. This metric quantifies market share concentration within an industry.

Finally, we incorporate prefecture-level city controls:

Financial Development (FIND): Measured by the ratio of year-end financial institution deposits/loans to regional GDP [56]. Data sourced from the China City Statistical Yearbook (deposit–loan balances) and prefecture-level city yearbooks (nominal GDP).

Government Intervention (GI): Calculated as the share of local government general fiscal expenditure in regional GDP [57], this metric quantifies direct government involvement in economic activities.

Government self-sufficiency (GC): Defined by the fiscal revenue-to-expenditure ratio [58]. Higher ratios indicate greater fiscal autonomy, fostering more stable policy environments that influence corporate investment decisions.

Science and Technology level (SOA): Measured through S&T expenditure’s proportion of fiscal budgets, reflecting regional commitments to innovation. Regions with abundant technology resources may strengthen the incentive of technology-seeking investment.

4.3. Sources of Data

Our study draws on a panel dataset of Shanghai and Shenzhen A-share listed firms (2011–2022). Greenfield investment data were primarily sourced from FDI Markets and Zephyr databases. Firm- and industry-level financial indicators originate from the China Stock Market & Accounting Research (CSMAR) Database. City-level economic metrics were compiled from the China City Statistical Yearbook and prefecture-level municipal yearbooks. All data-cleaning procedures and variable computations were implemented using Stata 15.0. Before empirical analysis, we performed the following procedures on raw data: ① eliminating observations with missing values, ② excluding financially distressed (ST/*ST) and financial sector firms, and ③ applying 1% winsorization to continuous variables to mitigate outlier effects.

Table 1. Descriptive statistics of variables.

Variables	Count	Mean	Sd	Min	Max
LnGI	39,534	0.147	0.759	0.000	5.380
did	39,173	0.069	0.253	0.000	1.000
LnSize	41,763	21.868	1.499	9.333	26.855
SA	39,971	−3.811	0.267	−4.447	−2.808
MOB	38,278	0.575	0.494	0.000	1.000
AMONT	45,520	19.552	1.675	7.931	23.902
HI	40,403	0.200	0.178	0.041	1.000
FIND	42,213	4.081	1.612	1.272	7.409
GI1	42,213	0.160	0.081	−2.823	2.349
GC	35,343	0.729	0.204	0.051	1.541
SOA1	42,213	0.044	0.028	0.002	0.134

presents the descriptive statistics for the full sample. The dependent variable (LnGI) exhibits a standard deviation of 0.768 with substantial dispersion between its maximum (5.380) and minimum values, indicating significant heterogeneity in greenfield investment across firms. The core explanatory variable (did) demonstrates pronounced variation, evidenced by a standard deviation (0.251) exceeding its mean value (0.067). This dispersion pattern confirms distinct treatment exposure across observations. Control variables display statistical characteristics consistent with prior empirical studies.

5. Empirical Results and Analysis

5.1. Baseline Regression

Table 2. Results of baseline regression.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
	LnGI	LnGI	LnGI	LnGI	LnGI	LnGI	LnGI	LnGI	LnGI	LnGI
did	0.0418 **	0.0455 **	0.0452 **	0.0458 **	0.0433 **	0.0451 **	0.0535 **	0.0538 **	0.0524 **	0.0516 **
	(2.17)	(2.29)	(2.29)	(2.29)	(2.11)	(2.21)	(2.46)	(2.47)	(2.38)	(2.35)
LnSize		0.0252 ***	0.0207 **	0.0206 **	0.0203 **	0.0201 **	0.0230 **	0.0230 **	0.0221 **	0.0226 **
		(2.94)	(2.33)	(2.28)	(2.11)	(2.09)	(2.21)	(2.21)	(2.11)	(2.16)
SA			−0.343 ***	−0.341 ***	−0.388 ***	−0.393 ***	−0.373 ***	−0.373 ***	−0.374 ***	−0.375 ***
			(−3.25)	(−3.22)	(−3.74)	(−3.78)	(−3.45)	(−3.45)	(−3.44)	(−3.45)
MOB				0.00590	0.00375	0.00372	0.00769	0.00774	0.00708	0.00689
				(0.59)	(0.37)	(0.37)	(0.74)	(0.74)	(0.68)	(0.66)
AMONT					−0.000560	−0.000737	−0.000297	−0.000323	−0.000401	0.0765 *
					(−0.10)	(−0.13)	(−0.05)	(−0.06)	(−0.07)	(1.83)
HI						0.0845 **	0.0786 *	0.0785 *	0.0769 *	−0.00834
						(2.16)	(1.88)	(1.88)	(1.84)	(−0.80)
FIND							−0.00811	−0.00727	−0.00838	−0.0179
							(−0.80)	(−0.71)	(−0.81)	(−0.25)
GI1								−0.0371	−0.0196	0.0593
								(−0.53)	(−0.27)	(1.02)
GC									0.0555	−0.482
									(0.95)	(−1.57)
SOA1										−0.000181
										(−0.03)
_cons	0.123 ***	−0.429 **	−1.637 ***	−1.634 ***	−1.798 ***	−1.824 ***	−1.785 ***	−1.783 ***	−1.804 ***	−1.806 ***
	(92.72)	(−2.27)	(−4.53)	(−4.47)	(−5.01)	(−5.06)	(−4.78)	(−4.78)	(−4.80)	(−4.80)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Code FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	38,812	37,892	37,890	37,620	36,525	36,454	32,933	32,933	32,806	32,806
F	5 **	6 ***	10 ***	7 ***	7 ***	6 ***	5 ***	4 ***	4 ***	4 ***
r2	0.285	0.283	0.283	0.283	0.280	0.280	0.279	0.279	0.280	0.280

Note: Robust t-statistics are reported in parentheses. * p < 0.10; ** p < 0.05; *** p < 0.01.

displays the baseline regression estimates. Column (1) includes only firm and year-fixed effects, while columns (2)–(10) progressively incorporate control variables. The results of columns (1)–(10) show that did coefficients are all positive and pass the significance test at the 5% level. This pattern confirms CBAM’s stimulative effect on corporate greenfield investments. Simultaneously, as control variables were incrementally incorporated, the coefficient of the primary variable stabilized within the range of 0.0418 to 0.0538, with the significance level remaining largely unchanged (p < 0.05). These results validate the model’s robustness and empirically support Hypothesis H1.

Our analysis further investigates the heterogeneous effects of CBAM on China’s outward FDI patterns.

Table 3. Results of OFDI and CMA baseline regression.

Variables	(1)	(2)	(3)	(4)
	OFDI	OFDI	CMA	CMA
did	0.0559	0.0355	−0.384	−0.478
	(0.28)	(0.17)	(−0.68)	(−0.78)
Controls	Yes	Yes	Yes	Yes
_cons	3.590 ***	−29.26 ***	4.522 ***	2.682
	(260.82)	(−6.87)	(121.77)	(0.22)
Year FE	Yes	Yes	Yes	Yes
Code FE	Yes	Yes	Yes	Yes
N	38,812	32,806	701	616
F	0	9 ***	0	3 ***
r2	0.624	0.634	0.507	0.532

Note: Robust t-statistics are reported in parentheses. *** p < 0.01.

presents baseline estimates for total OFDI and cross-border M&A activities. The CBAM treatment indicator shows statistically insignificant impacts on aggregate OFDI (β = 0.056, t-stat = 0.28). Disaggregated analysis reveals significant policy effect heterogeneity: greenfield investment (LnGI) demonstrates positive responsiveness contrasting with negative yet insignificant M&A coefficients (β = −0.478). This structural divergence suggests that CBAM’s influence operates through investment mode selection rather than aggregate FDI channels, with firms systematically favoring greenfield investments over M&As to address the environmental regulatory pressure of carbon tariffs.

The baseline results confirm CBAM’s structural transmission mechanism, where policy effects manifest predominantly through greenfield investment pathways. These findings validate our theoretical framework and underscore the practical relevance of examining the impact of carbon tariffs on enterprises’ greenfield investment.

5.2. Parallel Trends Test

The difference-in-differences (DID) estimator’s validity hinges on the parallel trend assumption—requiring comparable pre-treatment trajectories between treatment and control groups. We implement an event study framework with policy year (t = 0) dummies spanning pre_4 to post_3 periods to assess dynamic effects (Figure 2).

The empirical results show that the coefficients for pre-policy periods (pre_4 to pre_2) are all statistically insignificant, indicating no systematic differences in greenfield investment trends between treatment and control groups before the policy. This supports the validity of the parallel trend assumption. The findings indirectly suggest that firms did not adjust their investment behavior in advance due to policy anticipation. A plausible explanation lies in the high uncertainty surrounding carbon tariff policy details during the transition period, combined with the long-term nature of greenfield investments, which contributes to delayed responses to policy signals. The short-term impact during the initial phase of policy execution is not readily apparent due to the time lag; however, the coefficients for post_1 through post_3 following policy enactment are notably positive. Specifically, the coefficient for post_3 maintains stability at a higher level of significance, with its absolute value exceeding that of post_2, which suggests that the policy’s influence is likely to grow over the extended period. Therefore, it can be observed that the carbon tariff policy effects only became formally evident in 2022. Notably, while the coefficient for the post_2 period is marginally significant at the 10% level (0.051, p = 0.078), the lower bound of its confidence interval slightly intersects the horizontal axis, indicating phased fluctuations in the medium-term policy effects. Integrated with the literature and global contextual analysis, this phenomenon may stem from overlapping factors: First, the 2021 global economic recovery escalated carbon emissions [59], prompting host countries to accelerate restrictions on foreign high-carbon projects [60,61] to address domestic decarbonization pressures. This intensified environmental scrutiny and market entry barriers for Chinese enterprises in overseas greenfield investments, thereby curbing their growth rate. Second, the 26th United Nations Climate Change Conference in 2021 conference amplified uncertainty regarding policy expectations, leading firms to adopt a wait-and-see approach toward long-term strategic adjustments and fragmented supply chain restructuring. These factors collectively exacerbated the asynchronicity and volatility of policy effects during the medium term. Meanwhile, the direction and significance of control variables (e.g., firm size and financing constraints) align with baseline regression results, further confirming model robustness. Simultaneously, the orientation and importance of the coefficients associated with the control variables (like firm size and financing limitations) align with the benchmark regression, thereby reinforcing the model’s robustness.

5.3. Robustness Test

5.3.1. Placebo Test

We conducted 500 random permutations of the treatment–control interactions. Figure 3 presents two key findings: (1) the permuted coefficients follow a normal distribution centered around zero and (2) the majority of simulated p-values exceed 0.1. These results confirm that our baseline estimate cannot be attributed to random chance, demonstrating strong empirical robustness.

5.3.2. Exclusion of Policy Implementation Year

To ensure robustness, we exclude observations from the policy enactment year (2019) to mitigate potential distortions from policy announcements, anticipatory behaviors, and transitional strategies (e.g., temporary capacity relocation). This approach isolates sustained policy effects from transient implementation-phase disruptions. By excluding policy years, we can more accurately assess the long-term impact of policies. As shown in Column (2) of

Table 4. Robustness test results.

Variables	(1)	(2)	(3)	(4)	(5)
	LnGI	Excl. 2019	Non-DCM	PSM-DID	Lag1.did
did	0.0516 **	0.057 ***	0.056 **	0.0586 **
	(2.35)	(2.63)	(2.33)	(2.3921)
L.did					0.0574 ***
					(2.69)
Controls	Yes	Yes	Yes	Yes	Yes
Cons	−1.806 ***	−1.974 ***	−1.815 ***	−1.0976 **	−1.528 ***
	(−4.80)	(−5.14)	(−4.86)	(−2.2343)	(−2.85)
Observations	32,806	29,676	25,502	18,224	26,666
R-squared	0.0280	0.282	0.272	0.013	0.298
Year FE	Yes	Yes	Yes	Yes	Yes
Code FE	Yes	Yes	Yes	Yes	Yes

Note: Robust t-statistics are reported in parentheses. ** p < 0.05; *** p < 0.01.

, the did coefficient remains positive (0.057) with enhanced statistical precision (1% significance level). This consistency in effect direction and magnitude improvement confirms the policy’s enduring impact unaffected by initial adjustment shocks.

5.3.3. Exclusion of Direct-Controlled Municipalities

Given the distinct institutional characteristics of provincial-level municipalities (Beijing, Shanghai, Tianjin, and Chongqing) in governance structures and economic ecosystems, we exclude these regions from the sample. Column 3 of Table 4 reveals a nearly identical did coefficient (β = 0.056, p < 0.05) compared to baseline estimates. This finding demonstrates the policy’s generalizability across non-municipal regions, with treatment effects remaining robust to DCM-specific institutional configurations.

5.3.4. PSM-DID

To address endogeneity concerns from sample selection bias, we employ propensity score matching (PSM) combined with difference-in-differences (DID). Specifically, we designate LnGI as the treatment indicator and match firms using covariates including enterprise size. Nearest-neighbor matching is implemented based on estimated propensity scores to identify comparable control group firms.

Table 5. Results of the balance test.

Variables	Unmatched Matched	Mean		%bias	t	p
		Treated	Control
LnSize	U	22.22	21.99	15.80	10.83	0
	M	22.22	22.21	0.300	0.130	0.894
SA	U	−3.8141	−3.820	2.200	1.440	0.150
	M	−3.8142	−3.810	−1.400	−0.760	0.450
HI	U	0.125	0.211	−61.40	−33.21	0
	M	0.125	0.126	−0.600	−0.630	0.529
MOB	U	0.527	0.597	−14.30	−9.610	0
	M	0.527	0.538	−2.200	−1.120	0.263
FIND	U	3.533	4.237	−43.80	−29.34	0
	M	3.533	3.562	−1.800	−0.940	0.347
GI1	U	0.156	0.161	−5.700	−3.560	0
	M	0.156	0.157	−0.400	−0.240	0.808
GC	U	0.666	0.741	−35.60	−24.56	0
	M	0.666	0.670	−1.700	−0.830	0.407
SOA1	U	0.0358	0.0457	−37.70	−24.86	0
	M	0.0358	0.0363	−1.700	−0.920	0.360
AMONT	U	20.34	19.62	44.30	29.15	0
	M	20.33	20.37	−2.200	−1.130	0.258

confirms successful matching: post-matching covariate mean differences are significantly reduced, with standardized biases and variance ratios falling within acceptable thresholds. As shown in Column (4) of Table 4, the did coefficient remains positive and statistically significant (0.0586, p < 0.05), reaffirming CBAM’s robust positive effect on greenfield investments unaffected by selection bias.

5.3.5. One-Period Lagged Explanatory Variables

The baseline regression fails to address two critical issues: first, potential bidirectional causality between policy implementation intensity and corporate low-carbon transition pathways; second, simultaneous influences on core variables from unobservable strategic firm adjustments or host-country policy shifts. To mitigate these concerns, we incorporate one-period lagged explanatory variables in our empirical framework to test model robustness. As presented in Column (5) of Table 4, the coefficient for the lagged policy variable (L. did) measures 0.0574, statistically significant at the 1% level. This demonstrates the persistent nature of CBAM’s policy effects, with its stimulative impact on corporate greenfield investments remaining statistically significant one year post-implementation. Notably, the lagged effect’s coefficient magnitude exceeds that of the contemporaneous effect, suggesting cumulative policy impacts over time. The persistence of dynamic effects indicates that CBAM not only elicits short-term investment responses but also establishes medium-to-long-term influence mechanisms through institutional adaptation and technological sedimentation. Crucially, since current-period greenfield investments cannot retroactively affect prior-period CBAM policy determinations, these findings partially alleviate reverse causality concerns, thereby strengthening regression robustness.

5.4. Mechanism Analysis

5.4.1. Export Share

This study employs export share (LnPLP) as a mechanism variable to assess trade-restructuring effects. Measured by the ratio of export value to operating profit, LnPLP captures enterprises’ reliance on international markets. As shown in column (2) of

Table 6. Mechanism analysis results.

Variables	(1)	(2)	(3)
	LnGI	LnPLP	RD
did	0.0516 **	−0.0599 **	0.0324 *
	(2.35)	(−2.09)	(1.72)
Controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
N	32,806	31,499	27,232
F	4 ***	5 ***	17 *
R2	0.280	0.694	0.945

Note: Robust t-statistics are reported in parentheses. * p < 0.10; ** p < 0.05; *** p < 0.01.

, the carbon tariff policy demonstrates a statistically significant negative coefficient of −0.0599 (p < 0.05) on the export share, supporting hypothesis H2a. This finding implies that carbon tariffs reduce product price competitiveness by raising costs associated with conventional export models, thereby suppressing export activities. Furthermore, diminished export shares drive increased foreign direct investment (FDI) through trade-investment substitution effects. Specifically, when trade barriers escalate or export marginal returns decline, enterprises strategically substitute exports with cross-border investments to circumvent tariff burdens and access terminal markets. These results collectively validate hypothesis H2a.

5.4.2. R&D Intensity

We utilize research and development intensity (RD), defined as the proportion of R&D spending to operating income, to reflect companies’ demand for innovation. This metric reflects strategic resource allocation for technological advancement, where higher values indicate stronger innovation incentives and greater responsiveness to policy-driven technological upgrading. Table 6 Column (3) demonstrates a statistically significant positive association between CBAM exposure and R&D intensity (β₁ = 0.0324, p < 0.10). This 3.24% marginal effect confirms that carbon border adjustments effectively stimulate corporate R&D investments, validating Hypothesis H2b. The findings support the Porter Hypothesis by demonstrating environmental regulation’s innovation compensation effect, wherein firms optimize R&D resource allocation under carbon constraints. The green technology spillover inherent in outward FDI further amplifies this mechanism. Enhanced innovation capabilities through R&D investments motivate greenfield investments that solidify technological advantages, creating endogenous pathways to circumvent carbon restrictions.

6. Heterogeneity Analysis

6.1. Regional Heterogeneity

To examine spatial variations in carbon tariff impacts, we group the sample into eastern, central, and western regions based on corporate registration locations.

Table 7. Heterogeneity tests by geographic area and ownership type.

Variables	(1)	(2)	(3)	(4)	(5)
	Region			Ownership
	East	Central	West	Non-SOEs	Non-SOEs
did	0.0540 **	0.0302	0.0302	0.0502	0.0588 *
	(1.94)	(0.56)	(0.57)	(1.42)	(1.90)
Controls	Yes	Yes	Yes	Yes	Yes
N	23,383	5184	4227	11,084	20,961
R2	0.303	0.234	0.216	0.308	0.265
Year FE	Yes	Yes	Yes	Yes	Yes
Code FE	Yes	Yes	Yes	Yes	Yes

Note: Robust t-statistics are reported in parentheses. * p < 0.10; ** p < 0.05.

columns (1)–(3) reveal distinct regional effects: Eastern enterprises demonstrate significant positive responsiveness (β₁ = 0.0540, p < 0.05), whereas central and Western counterparts show statistically insignificant results. This divergence stems from three institutional advantages in eastern China. First, the concentration of export-oriented enterprises and advanced manufacturing industries in Eastern China equips the region with relatively abundant foreign investment expertise and resources, enabling the swift identification of host countries’ carbon regulatory variations and strategic mitigation of carbon costs through greenfield investments. Second, the Eastern region’s industrial chains demonstrate greater integration into global value chains, meaning that carbon tariff-induced trade cost escalations propagate more directly and intensely through interconnected supply networks. This heightened exposure compels Eastern enterprises to optimize carbon efficiency in overseas production capacities via greenfield investments. Third, the region’s superior green financial infrastructure provides critical financial backing for corporate low-carbon technology R&D and facilitates the deployment of environmentally sustainable production capabilities abroad. Conversely, central and Western enterprises exhibit lower responsiveness to carbon policy adjustments due to constrained geographical advantages and underdeveloped industrial foundations.

6.2. Ownership Heterogeneity

Ownership heterogeneity analysis reveals divergent responses to CBAM between state-owned enterprises (SOEs) and non-SOEs, attributable to differential resource endowments, policy responsiveness, and risk mitigation capacities. As shown in Columns (4)–(5) of Table 7, non-SOEs exhibit statistically significant positive responses (β₁ = 0.0588, p < 0.10), whereas SOEs show statistically insignificant reactions. This divergence primarily stems from three structural distinctions: First, in governance mechanisms, non-SOEs can rapidly respond to policy signals through market-oriented decision-making frameworks and risk-aversion strategies, whereas SOEs experience response delays due to administrative interventions and multi-objective performance evaluations. Second, regarding market constraints, non-SOEs face intensified international competitive pressures, where carbon tariff costs directly compel them to reconfigure global value chains through greenfield investments. In contrast, SOEs’ transformation incentives are weakened by their resource monopolies. Third, in financing incentives, non-SOEs are subject to stringent international ESG investment requirements, driving them to proactively cultivate environmental reputation assets, while SOEs’ environmental governance predominantly relies on administrative mandates.

6.3. Investment Motive Heterogeneity

Our analysis reveals significant heterogeneity in investment motives when examining CBAM’s effects. Specifically, building on Li’s taxonomy categorizing firms into five strategic types—trade-service, horizontal production, vertical production, R&D-intensive, and resource-seeking [62]—we identify distinct responses to carbon border adjustments. As shown in

Table 8. Heterogeneity tests by investment motivation.

Variables	(1)	(2)	(3)	(4)	(5)
	Trade-Service	Horizontal Production	Vertical Production	R&D-Intensive	Resource-Seeking
did	0.251	0.246 *	0.0601	−0.120	−0.0288
	(0.64)	(1.70)	(0.84)	(−0.55)	(−0.17)
Controls	Yes	Yes	Yes	Yes	Yes
N	3470	8250	9868	4086	2408
R2	0.342	0.424	0.306	0.372	0.399
Year FE	Yes	Yes	Yes	Yes	Yes
Code FE	Yes	Yes	Yes	Yes	Yes

Note: Robust t-statistics are reported in parentheses. * p < 0.10.

, horizontal production-oriented firms exhibit the strongest positive response (β₁ = 0.246, p < 0.10). Notably, these firms typically adopt cost-driven strategies to expand overseas production capacities. This investment pattern predominantly clusters in high trade-intensity industries, with their foreign capacity deployment exhibiting significantly greater responsiveness elasticity to marginal cost fluctuations compared to other enterprise types. The export barriers induced by carbon tariffs may escalate operational costs, thereby accelerating firms’ overseas plant establishment for localized manufacturing. In contrast, the coefficients for the remaining four enterprise types remain statistically insignificant, potentially because non-horizontal production investments’ host country selection is primarily determined by host countries’ resource endowments and technology spillover effects. Their decision-making frameworks appear fundamentally decoupled from carbon cost transmission mechanisms through the following pathways: First, trade-service and resource-seeking investments typically involve minimal production-phase emissions, thereby resulting in lower carbon policy sensitivity. Second, vertical production strategies prioritize supply chain integration and synergies [63,64], enabling them to achieve centralized carbon control and optimization through internal management systems, thereby exhibiting lower vulnerability to CBAM impacts. Third, R&D-focused investments show delayed responsiveness due to extended technology commercialization cycles, while their knowledge-intensive nature reduces immediate carbon exposure. Collectively, these motive-specific variations underscore the importance of strategic positioning in determining environmental regulation impacts on global investment patterns.

7. Conclusions and Recommendations

7.1. Conclusions

We utilized a difference-in-differences (DID) model to examine the effect of the European Union’s Carbon Border Adjustment Mechanism (CBAM), announced in 2019, on greenfield investment of Chinese firms. This analysis is based on panel data from Chinese A-share listed companies in the Shanghai and Shenzhen stock markets over the period from 2011 to 2022. The analysis yields three key insights:

First, CBAM exerts a statistically significant positive effect on greenfield investment (β = 0.0516, p < 0.05). Nevertheless, its effects on aggregate outward FDI and cross-border M&A remain insignificant. This discrepancy may be attributed to the sporadic and stochastic nature of M&A transactions. Concurrently, it indicates that outward foreign direct investment (OFDI) responds to CBAM through selective restructuring rather than through broad capital flow adjustments.

Second, mechanism analysis reveals CBAM’s dual operational pathways: (i) trade restructuring—increased export costs reduce dependence on EU markets, promoting corporate greenfield investment by bypassing trade barriers (substitution effect) and expanding non-EU market penetration (creation effect); (ii) innovation-driven adaptation—the imperative for low-carbon development has driven corporate demand for green technology innovation (as reflected in increased R&D intensity), thereby fostering greenfield investment by enterprises. These pathways collectively validate the Theory of Overcoming Trade Barriers and the Reverse Green Technology Spillover Theory, while expanding the regulatory determinants framework of OFDI.

Third, heterogeneity tests show that CBAM’s impact on greenfield investment differs by firm location, ownership, and investment motives. Specifically, Eastern-region enterprises with internationalization expertise, non-state-owned firms (heightened risk adaptability), and market-seeking horizontal producers (cost-driven relocation patterns) are more susceptible to CBAM’s influence.

This study advances empirical support for the “trans-trade barrier effect” theory of outward foreign direct investment by demonstrating how carbon tariffs stimulate greenfield investments among Chinese firms. It extends theoretical and empirical investigations of the Porter Hypothesis into the domain of environment–trade composite policies. In contrast to conventional perspectives that treat trade barriers as pure investment deterrents, our findings reveal that firms strategically intensify greenfield investments to circumvent compliance costs associated with the Carbon Border Adjustment Mechanism.

Compared with existing studies, our findings enrich theoretical reasoning and empirical analysis through three distinct dimensions. First, in contrast to the existing literature that predominantly focuses on CBAM’s negative policy effects, we systematically investigate and empirically confirm its positive promotional effects on greenfield investment—a sustainable investment modality. Second, while prior research has predominantly concentrated on simulating the comprehensive impacts of the Carbon Border Adjustment Mechanism on international relations and national economies, few studies have microscopically analyzed its formative mechanisms shaping enterprises’ long-term investment decisions. This study breaks through conventional research paradigms by extending the economic effects of carbon tariffs to corporate response behaviors, thereby addressing the research gap in CBAM’s micro-level consequences. Third, whereas existing scholarship predominantly relies on general equilibrium methods, we employ the Difference-in-Differences methodology. This approach overcomes previous limitations in dynamic causal effect evaluation and reduces excessive dependence on model assumptions, thereby facilitating more robust assessments of CBAM’s short- to medium-term impacts while effectively mitigating endogeneity concerns.

7.2. Recommendations

First, enterprises in CBAM-affected industries should strategically leverage greenfield investment to circumvent trade barriers. By establishing production facilities in host countries, firms can not only bypass carbon tariff-induced export costs but also reduce emission intensity through localized supply chains, achieving an “investment-over-export” strategic shift. Location selection should prioritize (i) EU member states—utilizing their advanced technologies and stringent emission standards to produce competitive goods; and (ii) non-EU regions abundant in clean energy and green innovation—thereby enabling indirect EU market access while avoiding direct trade constraints.

Second, enterprises should optimally utilize greenfield investments to acquire overseas green technologies, resources, and knowledge, thereby mitigating CBAM’s long-term negative impacts while enhancing adaptability to evolving trade regimes. By capitalizing on host countries’ environmental subsidies and tax incentives, firms can actively develop clean energy solutions and low-carbon technologies. The repatriation of acquired technologies and equipment enables domestic enterprises to strengthen CBAM compliance capabilities. Furthermore, deploying technical personnel to assimilate advanced environmental management systems and production techniques facilitates knowledge transfer, ultimately elevating the home country’s innovative capacity through applied expertise integration.

Third, to enhance the level and efficiency of greenfield investment, enterprises must improve three key capabilities: risk sensitivity, investment flexibility, and market acuity. By leveraging these capabilities, they can nimbly circumvent the trade barriers posed by CBAM and maintain the competitiveness of their products in the EU market. Given the capital-intensive nature and extended development timelines inherent in greenfield projects, firms should conduct comprehensive due diligence on host countries’ carbon regulatory frameworks and market entry requirements. This necessitates developing robust intelligence-gathering systems to systematically analyze evolving market dynamics and consumer preferences, enabling strategic selection of optimal investment locations, timing, and operational models aligned with CBAM compliance objectives.

Fourth, strategic synergy between governments and enterprises is essential for enhancing outward foreign direct investment capabilities and trade barrier resilience. Eastern provincial governments should intensify fiscal support for OFDI initiatives and low-carbon technology R&D. Eastern enterprises ought to pioneer investment excellence, guiding central/western counterparts in executing strategic greenfield investments to integrate into global value chains. Concurrently, central/western governments should deepen collaboration with Belt and Road Initiative (BRI) partners, leveraging geoeconomic and natural advantages to cultivate alternative export markets and production networks. This dual approach not only mitigates EU market dependence but also activates trade creation effects through localized horizontal production capacity building, thereby optimizing the utilization of resources in BRI economies.

7.3. Limitations and Future Research Directions

This study has several limitations. First, constrained by data timeliness, the findings primarily reflect corporate responses during CBAM’s initial phase. As carbon tariffs constitute a progressive policy tool, their long-term effects and dynamic adjustment mechanisms remain unclear and warrant continuous investigation. Second, while our sample consists of Chinese listed companies, the generalizability of results to other economies may be limited—we strongly encourage international scholars to extend this research framework. Methodologically, although the difference-in-differences approach effectively identifies short-term causal effects of CBAM, the parallel trend assumption may be violated in long-term policy evaluations, necessitating further dynamic effect analyses. Finally, potential omitted variables—such as interactions with other climate policies, green technology reserves, and environmental regulation pressures—could bias the results and require systematic examination.

Future research could be extended through the following directions: First, a dynamic analytical framework could be developed to characterize the lifecycle effects of carbon tariffs by integrating longitudinal data tracking after full policy implementation. Second, cross-national comparative studies should be conducted to reveal how differences in CBAM policy designs across jurisdictions induce heterogeneous impacts on greenfield investment location choices. Third, while this study primarily examines market mechanisms, subsequent investigations could examine the role of governmental factors and policy synergies between CBAM implementation and China’s domestic carbon regulatory framework. Fourth, researchers should deepen the analysis of intra-group heterogeneity sources through subgroup differentiation studies to elucidate the underlying causes of carbon tariffs’ varied impacts across entities.