Has the Free Trade Zone Enhanced the Regional Economic Resilience? Evidence from China

Abstract

This study examines the impact of free trade zone (FTZ) establishment on regional economic resilience (RER) in China, using provincial-level panel data spanning from 2010 to 2022 and a multi-period difference-in-differences (DID) approach. The empirical results indicate that FTZ implementation significantly enhances regional economic resilience by 3.46%, with the development of green finance acting as a key moderating mechanism that amplifies this positive effect. Heterogeneity analysis uncovers notable disparities across policy cohorts and geographical regions: the first wave of FTZs demonstrates the most pronounced resilience-enhancing impact, whereas later cohorts exhibit weaker or even adverse effects. Coastal regions experience substantial benefits from FTZ policies, in contrast to statistically insignificant outcomes observed in inland areas. These findings suggest that strategically expanding the FTZ network, when paired with tailored implementation mechanisms and the integration of green finance, could serve as a powerful policy tool for post-COVID economic recovery. Importantly, by strengthening economic resilience through institutional openness and green investment, this study offers valuable insights into balancing economic growth with environmental sustainability. It provides empirical evidence to support the optimization of FTZ spatial governance and institutional innovation pathways, thereby contributing to the pursuit of sustainable regional development.

1. Introduction

Since the 2008 global financial crisis, global trade has progressively weakened, profoundly affecting export-oriented economies such as China. Currently, the insufficient resilience of economic recovery has increasingly become a critical issue that hinders improvements in economic efficiency and national growth. This challenge represents a key area of concern for China as it faces slowing global growth and strives to overcome the middle-income trap [1]. Green finance plays a significant role in enhancing the economic resilience of BRICS economies, particularly under targeted policy interventions [2]. Similarly, institutional integration can amplify the positive effects of green finance, acting as a catalyst for strengthening regional recovery resilience [3]. This trend is further supported by scholars who highlight the favorable impact of trade integration on China’s green economic development [4]. Notably, since 2013, China has established free trade zones (FTZs) with a focus on institutional reform of the trading environment, thereby further promoting green development and economic recovery.

To date, China has established 21 FTZs. In August 2013, the China (Shanghai) FTZ was officially launched. Subsequently, pilot zones in Tianjin, Fujian, and Guangdong were established in 2015; those in Hubei, Henan, Chongqing, Sichuan, Liaoning, Zhejiang, and Shaanxi in 2017; Hainan in 2018; Shandong, Hebei, Jiangsu, Heilongjiang, Guangxi, and Yunnan in 2019; and finally, Beijing, Anhui, and Hunan in 2020. Altogether, China’s FTZs have undergone six rounds of expansion, covering 21 provinces and forming a new “flying geese” development pattern that coordinates growth across eastern, western, southern, northern, and central regions, while integrating maritime and inland economies [5].

These free trade zones have been developing rapidly. For instance, the two expansions of the Shanghai FTZ and the doubling in size of the Zhejiang FTZ in 2020 have markedly enhanced regional openness. The Guangdong FTZ has strengthened economic cooperation with Hong Kong and Macau, achieving breakthroughs in cross-border RMB business and trade liberalization, while the Tianjin FTZ has explored new approaches for coordinated development in the Beijing–Tianjin–Hebei region, attracting clusters of high-end manufacturing and modern service industries. In terms of trade facilitation, FTZs have significantly reduced customs clearance times and business costs through measures such as paperless processing, intelligent supervision, and risk early warning systems. In the realm of foreign investment, the adoption of negative lists, filing systems, and “one-stop” foreign investment service centers has improved market access and service efficiency, thereby increasing foreign capital attraction [6].

Green development is widely regarded as a key driver of regional economic recovery [7]. In the face of global challenges such as climate change, resource scarcity, and environmental degradation, green development has emerged as a crucial strategy for achieving sustainable economic growth [8]. In recent years, China has made notable strides in green development. For example, since the launch of green bond trading in 2016, the scale of green bond issuance has steadily expanded—from CNY 205.23 billion in 2016 to CNY 854.85 billion in 2023. Similarly, the balance of green loans has surged from CNY 8.23 trillion at the end of 2018 to CNY 34.76 trillion in 2024, representing a more than fourfold increase. Moreover, the number of member institutions in the Green Finance Committee has grown to 324, collectively managing over CNY 320 trillion in financial assets, with green credit and green bond services accounting for the majority of national totals. Nevertheless, achieving green development goals still requires the optimization of institutional environments to reduce uncertainties [9]. As a key practice promoting green development, FTZ establishment reflects China’s efforts to adapt to evolving global economic trends and deepen international economic integration, thereby providing strong support for regional recovery and sustainable development [10].

China’s FTZs have evolved from initial pilot projects to a stage of comprehensive implementation, offering a valuable opportunity to assess their impact on regional economic resilience. Based on panel data from 2010 to 2022, this study includes all FTZs established before 2022. The treatment group comprises 21 provinces (those that had established FTZs by 2020), while the remaining 9 provinces form the control group. The sample period (2010–2022) is selected for three key reasons: (1) China’s first FTZ was launched in 2013, and 2010 serves as the pre-policy baseline, ensuring a sufficient number of pre-treatment observations for DID analysis; (2) 2022 represents the latest year with complete provincial economic data at the time of study; and (3) this period captures major events such as FTZ expansions (2013–2020), U.S.–China trade tensions (2018–2019), and the post-COVID recovery (2020–2022), providing a comprehensive backdrop for analyzing RER dynamics during both normal and shock periods [11].

This paper constructs a comprehensive evaluation index system for regional economic resilience and conducts baseline regressions using a multi-period DID model. In the heterogeneity analysis, the 21 FTZs are categorized by establishment batch (first, second, and third) and geographic location (inland vs. coastal). Furthermore, by introducing green finance as an intermediary variable, this study explores the transmission mechanism from FTZ policy to regional resilience. A series of robustness checks—including parallel trends testing, alternative RER measures, placebo tests, and propensity score matching DID estimation—are also conducted.

The existing literature on FTZs has primarily focused on their impacts on international trade flows [12], green total factor productivity [13], and regional financial development [14]. In parallel, research on regional economic resilience (RER) has examined its links to innovation and green finance, but rarely within the FTZ context [2,15]. Although some studies touch on FTZs’ role in economic stability [16], they lack a systematic analysis of how specific FTZ policies—such as streamlined foreign exchange hedging, 50% reductions in customs clearance times via intelligent supervision, and tax rebates for high-tech imports—translate into stronger RER. Moreover, research on green finance as a moderating factor between FTZs and RER remains underdeveloped. Existing works often treat green finance as an independent driver of resilience or isolate its environmental impacts from FTZ reforms [17]. This reveals a key gap: empirical evidence remains limited on how FTZ policies, through institutional innovations, affect RER, and whether green finance plays an amplifying role.

Against this backdrop, this study addresses the following core question: Do China’s FTZ policies enhance regional economic resilience, and how do policy implementation batches, geographic differences, and green finance development influence this relationship?

This research makes three key contributions to the literature. First, it systematically investigates the impact of FTZs on RER, addressing the current lack of attention to the link between institutional opening-up policies and economic resilience. By connecting FTZs to specific implementation mechanisms—such as negative list foreign investment management and paperless customs processing—this study clarifies how FTZs may shape economic outcomes. Second, it introduces green finance as a moderating variable, revealing how green capital allocation interacts with FTZ policies to reinforce regional resilience, thereby bridging two previously fragmented strands of literature. Third, it examines heterogeneity by policy batch and regional location, offering new insights into why early batches and coastal FTZs show stronger effects, thus enriching our understanding of policy effectiveness in uneven development contexts. From a sustainability perspective, this study highlights how institutional innovation and green finance can work synergistically to promote economic resilience, offering a viable framework for balancing growth and environmental sustainability in emerging economies.

The rest of this article is organized as follows. The next section reviews the relevant literature. Section 3 presents empirical models and data. The results and analysis are presented in Section 4. Robustness tests are reported in Section 5. Finally, Section 6 provides policy recommendations.

2. Literature Review and Theoretical Assumptions

2.1. Pilot Free Trade Zones and Regional Economic Resilience

Globalization has intensified market competition, particularly in technology-intensive industries, thereby driving significant improvements in innovation and productivity [18]. While earlier studies primarily focused on the relationship between FTZs and innovation activities, recent research has increasingly turned to the impact of FTZs on RER.

From the perspective of evolutionary economics, RER refers to a region’s multidimensional capability: it encompasses not only the ability to withstand external shocks (e.g., economic downturns, pandemics) and recover to a stable growth trajectory (resistance and recovery capacity), but also the capacity to maintain growth and adapt to structural changes under normal conditions (adaptability and robustness) [11,19]. This dual nature allows RER to be observed both during crises (e.g., post-COVID recovery) and during periods of economic stability (e.g., industrial upgrading and sustained GDP growth).

The regional economic system is shaped by interactions among public and private actors, relying on the generation, dissemination, and application of knowledge to maintain economic stability and adaptability [20,21,22]. One study constructs an urban resilience index based on economic, environmental, social, and infrastructure dimensions, revealing that the economic component is the primary determinant of resilience disparities [23]. Another shows that each sub-dimension—ecological, social, infrastructure, and economic—positively influences outcomes such as residents’ well-being [24].

This study focuses specifically on the economic dimension of resilience (GDP-related measures), for three reasons: First, the primary objective is to examine how FTZs—an economic policy intervention—influence regional economic performance and recovery. Second, economic indicators such as GDP growth and income are more directly responsive to FTZ initiatives (e.g., trade liberalization, investment facilitation) than social or institutional variables. Third, data on social capital or institutional adaptability at the provincial level in China are scarce and lack comparability [25]. While this narrower approach differs from broader indices, it aligns with the literature that treats economic resilience as a central pillar of overall regional resilience [11].

China’s provinces enjoy a degree of administrative and economic autonomy, and their diverse socio-cultural contexts and resource endowments render them relatively independent economic entities [26]. Within this framework, FTZ policies have emerged as strategic pilot reforms, underscoring China’s commitment to global economic integration and regional competitiveness [10].

Agglomeration economics theory provides a robust foundation for Hypothesis 1. FTZs reduce trade barriers and regulatory burdens, thereby attracting multinational enterprises, skilled labor, and advanced technologies [27]. This agglomeration fosters knowledge spillovers, economies of scale, and industrial diversification—all of which contribute to a region’s ability to resist shocks and adapt to change [11,28]. For instance, FTZs’ “one-stop” investment services facilitate rapid capital inflows, sustaining investment momentum even during external shocks [6].

Studies on China’s FTZs have shown their wide-ranging economic impacts. Some highlight the significant improvement in green total factor productivity due to technological and innovative advancements [13], while others confirm the Shanghai FTZ’s positive influence on regional financial development [14]. These results underscore FTZs’ role in strengthening regional economies through innovation, capital attraction, and efficient resource allocation. For example, the Guangdong FTZ has advanced service industry upgrading through innovation and business clustering, thereby improving regional economic structure [29].

At the national level, FTZs have reshaped China’s internal trade patterns, a process further amplified by the “Belt and Road” initiative [12]. In addition, studies show that FTZs enhance listed companies’ risk resilience by improving operational efficiency [16]. While some note that FTZs initially caused environmental deterioration, these issues have been partially mitigated through policy refinement and sustainable initiatives [30].

A resilient economic system depends on infrastructure, human capital, financial institutions, knowledge networks, and supportive policies [31]. As platforms for institutional experimentation, FTZs promote economic stability and adaptability through reforms in trade, investment, and customs practices. Studies using the SBM model confirm FTZs’ positive influence on infrastructure and innovation efficiency—two key elements of RER [32]. Furthermore, FTZs enhance the mobility of capital and talent, strengthening regional economic robustness [33]. Other evidence shows FTZs boost innovation output, such as patent filings, which further contributes to economic resilience [15].

Agglomeration economics also supports these observations by highlighting the benefits of industrial clustering. By attracting global enterprises, facilitating knowledge exchange, and improving access to markets, FTZs build dynamic economic ecosystems capable of cushioning shocks and ensuring long-term growth. Based on the above, we propose the following hypothesis:

Hypothesis 1 (H1).

The establishment of FTZs significantly enhances RER.

2.2. The Role of Green Finance in the Transmission

As the global economy transitions toward sustainability, green finance has become a pivotal policy instrument for driving both ecological development and economic resilience. Green finance aims to channel capital into green technologies, sustainable investment, and environmental governance [34].

Theoretically, green finance reinforces the FTZ–RER linkage through two key mechanisms, supporting Hypothesis 2. First, it helps to bridge the “green investment gap” in FTZs: their institutional advantages (e.g., cross-border capital facilitation) attract green projects, while green finance tools (e.g., green bonds, credit) supply long-term financing [35]. This shift reduces dependence on polluting industries, thereby improving structural stability. Second, green finance enhances risk diversification by supporting low-carbon sectors, thus reducing vulnerability to shocks like carbon pricing volatility or tightening environmental regulation [2,11].

In China, FTZs serve not only as gateways for global trade, but also as platforms for sustainable economic experimentation. As a supply side policy, green finance augments RER by aligning capital allocation with green goals, making it a key intermediary between FTZs and RER.

RER depends on innovation, industrial diversification, environmental sustainability, and adaptive resource allocation [36]. Green finance drives these processes by supporting clean industries, incentivizing R&D in green technologies, and facilitating the growth of the green bond market [37]. Evidence shows that green finance significantly enhances the green efficiency of regional infrastructure, thereby promoting more rational resource use [38].

In the FTZ context, green finance and institutional reform can yield synergistic policy effects. On one hand, FTZs offer the regulatory environment and openness necessary for green finance to thrive. On the other hand, green finance promotes green transformation within FTZs, bringing in both green capital and enterprises, and enhancing regional adaptability and recovery capacity. Empirical studies show that FTZs stimulate green patenting, and green finance amplifies this effect through financial incentives and capital support [17]. Other research indicates that green finance policies in FTZs accelerate the development of green technologies, improving both innovation and resilience [35].

Green finance also contributes to environmental governance. FTZs support this by improving capital mobility and resource allocation, while green finance helps reduce regional exposure to environmental and economic risks [39]. Together, they foster innovative ecosystems and disseminate green technologies more effectively.

From the perspective of regional innovation systems, integrating green finance with FTZ policy optimizes capital allocation and promotes sustainable growth. Green finance lowers the barriers for SMEs to engage in green tech development, thereby supporting economic diversification and resilience. For example, the green bond market helps SMEs upgrade technologies and expand market access [40], reducing R&D costs and accelerating green industrialization. A multi-period spatial DID study of 265 Chinese cities finds that FTZ-driven liberalization improves urban energy efficiency, with positive spillovers to nearby areas [41]. These findings are consistent with our emphasis on the joint environmental and economic benefits of FTZs. Therefore, we propose the following hypothesis:

Hypothesis 2 (H2).

Green finance plays a moderating role in enhancing the positive effects of FTZs on RER.

3. Methodology and Data

3.1. Identification Strategy

3.1.1. Difference-in-Differences

To estimate the impact of FTZs on RER, we construct a multi-period DID model, specified as follows:

$${RER}_{it}={\beta }_0+{\beta }_1{FTZ}_{it}+\alpha X_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

where ${RER}_{it}$ is the RER of province i in year t (measured by the entropy-based index, see Section 3.2). ${FTZ}_{it}$ is the core explanatory variable (policy dummy), equal to 1 if province i has established an FTZ by year t, otherwise 0. It is the interaction of ${Treat}_i$ (1 if province i ever established an FTZ by 2022) and ${Post}_{it}$ (1 for years after FTZ establishment in province i). $X_{it}$ is the control variables. $\epsilon$ denotes the random error. ${\mu }_i$ and ${\lambda }_t$ control for the province and time effects (to control for unobserved regional characteristics and time trends). ${\beta }_1$is the key coefficient, capturing the average treatment effect of FTZs on RER.

Green finance in a province plays a crucial role in enhancing RER by driving environmental effects such as supporting low-carbon industries, generating policy synergies, and contributing to regional economic environmental governance and the optimization of resource allocation. Following previous studies [2], this study uses the interaction term G to investigate the moderating role of green finance, and we include an interaction term as shown in Equation (2):

$${RER}_{it}={\beta }_0+{\beta }_1{FTZ}_{it}+{\beta }_2{GFin}_{it}\times {FTZ}_{it}+{\beta }_3{GFin}_{it}+\alpha X_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

where ${GFin}_{it}$ is the green finance index of province i in year t. ${\beta }_2$is the coefficient of the interaction term, measuring whether green finance moderates the FTZ–RER relationship (positive ${\beta }_2$ indicates amplification).

3.1.2. Propensity Score Matching-Difference-in-Differences

To address potential endogeneity and selection bias, we adopt a PSM-DID approach. The propensity score matching step estimates the likelihood that a province becomes an FTZ, based on observable characteristics:

$$P_i=P_r\left(region=T|X_{it}\right)$$

(3)

where $region =\left\{T,C\right\}$, including the T (treatment group, provinces with FTZs) and C (control group, provinces without FTZs), $X_{it}$ are the covariates affecting FTZ selection (e.g., FDI, infrastructure) measured pre-policy (2010–2012) to avoid post-treatment bias [42].

The nearest-neighbors matching is implemented for PSM, with a 1-to-4 nearest-neighbor approach. This approach is chosen because: (1) Our treatment group (21 provinces) is small relative to the control group (9 provinces), and 1-to-4 matching maximizes the use of control observations without excessive information loss. (2) It balances statistical efficiency and bias reduction, as shown in Lu et al. (2019) [27] for similar policy evaluation settings.

The PSM-DID model is constructed as follows:

$${RER}_{it}^{PSM}={\beta }_0+{\beta }_1{FTZ}_{it}+\alpha X_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(4)

3.2. Variables Description and Data

The dependent variable is regional economic resilience (RER), constructed as a composite index focusing on economic performance. Based on the existing literature [23,24], RER incorporates components such as economic growth, stability, and structure, as these are the most influential dimensions of regional resilience.

To test robustness, we adopt two alternative measures: (1) Employment stability, measured as the five-year rolling standard deviation of urban employment growth [11], captures a region’s labor market resilience during downturns. (2) Industrial diversification, proxied by the Herfindahl index of industrial output, reflects the region’s adaptability to structural shocks [43].

Following the evolutionary economics framework, our RER index isolates the economic dimension of resilience. This emphasis is justified by our research goal—to measure FTZs’ contribution to post-crisis economic recovery. Institutional and social factors are intentionally excluded due to both conceptual clarity and data availability constraints.

Various methods have been proposed in the existing literature to measure RER, including case studies [43], sensitivity indices [19], statistical time series models [44], and causal structure models [45]. However, these approaches are not entirely suitable for China, where regional development is uneven. To more accurately and effectively assess the resilience of regional economies in China, we follow previous studies in defining regional economic resilience (RER) based on economic growth capacity and stability [25,46]. To accommodate China’s regional heterogeneity, we adopt the entropy method for constructing the index (see

Table 1. Indicator system of regional economic resilience (RER).

Primary Indicator	Secondary Indicators	Measurement Method	Indicator Attributes
Regional economic resilience (RER)	Economic growth capacity	Per capita disposable income	+
		GDP growth rate	+
		Proportion of deposit balances in financial institutions	+
	Economic stability capacity	The five-year rolling standard deviation of GDP growth	−
		Proportion of value added of tertiary industry	+
		Proportion of added value of secondary production	+
		Proportion of loans outstanding by financial institutions	−

). This objective weighting method avoids subjectivity and ensures more reliable indicator aggregation. Indicators include both growth-oriented variables (e.g., GDP growth, income) and stability-related measures (e.g., industry structure, credit exposure).

Before measuring regional economic resilience, it is necessary to determine the weights of the aforementioned indicators. This study employs the entropy method for the calculation. The entropy method is an objective weighting approach based on information entropy, which avoids the shortcomings of subjective weighting methods that are often influenced by individual biases. The calculation steps are as follows:

The first step is to standardize the data:

$${\mathrm{X}}_{\mathrm{i}\mathrm{j}}^{\prime }\left\{\begin{array}{c}{\displaystyle \frac{X_{ij}-{minX}_{ij}}{{maxX}_{ij}-min{X}_{ij}}}\\ {\displaystyle \frac{max{X}_{ij}-X_{ij}}{max{X}_{ij}-min{X}_{ij}}}\end{array}\right.$$

(5)

Among them, j is a positive indicator, and i is a negative indicator. $X_{ij}$ is the value of indicator i in region j. $X_{ij}^{\prime }$ is its standardized value. $max{\mathrm{X}}_{\mathrm{i}\mathrm{j}}$ and $min{\mathrm{X}}_{\mathrm{i}\mathrm{j}}$ are the maximum and minimum values of each indicator, respectively, where $0<i\le \mathrm{n}$, $0<j\le \mathrm{m}$.

The second step is to calculate the entropy value $e_j$ of each indicator:

$$e_j=-1/lnS \times {\sum }_{i=1}^n{y}_{ij}ln{y}_{ij}$$

(6)

In the Formula (6), $y_{ij}=X_{ij}^{\prime }/\sum _{i=1}^n{X}_{ij}^{\prime }$, $S$ is the number of observations in the sample.

Secondly, calculate the weight $W_j$ of each indicator:

$$W_j={\displaystyle \frac{1-e_j}{\sum _{j=1}^m\left(1-e_j\right)}}$$

(7)

Finally, determine the index of i province in t year:

$${RER}_{it}={\sum }_{j=1}^m{W}_j{X}_{ij}^{\prime }$$

(8)

The core explanatory variables include the province-level dummy (${Treat}_i$) indicating whether a region has established an FTZ, the time dummy (${Post}_{it}$) marking the post-establishment period, and their interaction term (${FTZ}_{it}={Treat}_i\times {Post}_{it}$). The province dummy captures baseline differences in regional economic resilience (RER) between FTZ and non-FTZ provinces, while the time dummy controls for general time trends affecting both groups. The interaction term isolates the differential effect of FTZ establishment on RER, representing the average treatment effect of the policy intervention.

To investigate the underlying mechanism, this study introduces a green finance index as a moderating variable. Green finance refers to financial activities that support environmental sustainability, including instruments such as green bonds, green credit, green insurance, and green investment, along with fiscal support mechanisms and market-based tools like green funds and environmental rights trading [34,47]. While these tools are implemented nationwide, FTZs often serve as pilot zones for advanced applications—such as cross-border green bond issuance or streamlined approval for eco-friendly projects-aligning with their institutional innovation agendas [35]. Given the regional variation in green finance intensity, driven by differences in governance and industrial structure, the model controls for observable heterogeneity. However, potential bias from unobservable regional characteristics remains a limitation and is duly acknowledged.

The green finance index is constructed across eight dimensions, extending traditional indicators (e.g., green credit, green bonds, carbon finance) by incorporating green support, green funds, and green equities for a more comprehensive evaluation (see

Table 2. Indicator system of green finance index.

Primary Indicator	Secondary Indicators	Indicators Description	References
Green finance	Green securities	Proportion of green bonds in total bonds issued	[34,47]
	Green credit	Proportion of environmental protection credit in total credit	[37,47]
	Green investment	Environmental pollution control investment/GDP	[8,38]
	Green insurance	Environmental liability insurance income/total premium income	[47]
	Green support	Fiscal environmental expenditure/general fiscal expenditure	[7,17]
	Green funds	Green funds’ market cap/total funds’ market cap	[17]
	Green rights and interests	Carbon/energy trading volume/total equity trading volume	[34,41]

) [47].

A principal component analysis (PCA) is used to extract the index. The results (Appendix A 

Table A1. PCA results for green finance index.

Secondary Indicators	Factor Loading (PC1)	Factor Loading (PC2)	Communality
Green securities	0.79	0.25	0.70
Green credit	0.82	0.21	0.72
Green investment	0.76	0.30	0.68
Green insurance	0.65	0.42	0.61
Green support	0.33	0.73	0.64
Green funds	0.29	0.68	0.54
Green rights and interests	0.41	0.59	0.51
Variance explained	48.1%	24.2%	/
Cumulative variance explained	48.1%	72.3%	/

Notes: (1) Factor loadings represent the correlation between each indicator and the principal component (PC1/PC2). Values with absolute values > 0.7 are considered strong correlations. (2) Communality measures the proportion of variance in each indicator explained by the two principal components (threshold ≥ 0.5 for retention). The symbol “/” in the table indicates that no data need to be added for the corresponding entries, the same below.

) show the following: (1) Green credit (0.82), green bonds (0.79), and green investment (0.76) load heavily on the first principal component (PC1), which captures direct environmental investment and explains 48.1% of the variance. (2) Green support (0.73) and green funds (0.68) dominate PC2, accounting for 24.2% of variance and reflecting indirect policy and market-driven mechanisms. (3) All indicators are retained, as their communalities exceed 0.6, confirming their meaningful contribution to the composite index [48]. Furthermore, variance inflation factor (VIF) analysis reveals acceptable multicollinearity (mean $\mathrm{V}\mathrm{I}\mathrm{F}=1.8<5$), indicating that no single variable overly influences the index construction.

Based on previous studies, foreign direct investment (FDI), economic agglomeration (agg), business environment (BE), unemployment rate (une), R&D investment (RD), infrastructure (infra), medical and health level (medic) were selected for estimation [4,6,49]. These variables are chosen because they are theoretically linked to RER: FDI and R&D reflect capital and innovation capacity; infrastructure and business environment affect trade costs; unemployment and medical level capture social stability.

Economic agglomeration is measured by GDP per square kilometer of each province. The business environment indicator is sourced from <China Provincial Business Environment Index 2023 Report> [50]. R&D intensity is measured as the ratio of R&D expenditure to provincial GDP, capturing the region’s innovation capacity. Infrastructure development is proxied by the density of transportation networks, specifically the total operating mileage of railways and roads per square kilometer. Healthcare provision is represented by the number of medical institutions per 10,000 residents, reflecting access to public services. All raw data are sourced from the China Statistical Yearbook. Summary statistics for these variables are presented in

Table 3. Descriptive statistics.

VarName	Obs	Mean	SD	Max	Min
RER	390	0.3308	0.0893	0.1948	0.7158
FTZ	390	0.2923	0.4554	0.0000	1.0000
FDI	390	0.7905	3.9944	0.0472	57.3213
agg	390	4.5159	2.3024	0.7667	12.1578
BE	390	3.4165	0.2919	2.8000	4.0875
une	390	3.1978	0.7450	0.6187	5.1659
RD	390	0.0110	0.0060	0.0017	0.0324
infra	390	40.3370	32.1279	1.4199	132.5447
medic	390	7.1862	2.4145	2.0119	12.1555

.

To facilitate clarity and transparency, Appendix A 

Table A2. Summary of variables (integrated table).

Variable Type	Name	Measurement Method	Data Source and References	Expected Direction
Dependent	RER	Entropy-based index combining economic growth capacity and stability capacity	China Statistical Yearbook	/
Independent	FTZ	Dummy = 1 if province has an FTZ in year t	Ministry of Commerce official documents	+
Moderator	GFin	PCA-based index using 7 indicators	China Financial Statistic Yearbook, Provincial Financial Reports	+
Controls	FDI	FDI/GDP	China Statistical Yearbook	+
	agg	GDP per square kilometer	China Statistical Yearbook	+
	BE	China Provincial Business Environment Index	Local Bureau of Commerce public data	+
	une	Registered unemployment rate (%)	China Statistical Yearbook	−
	RD	R&D expenditure/GDP	China Statistical Yearbook	+
	infra	Railway + road mileage per square kilometer	China Statistical Yearbook	+
	medic	Medical institutions per 10,000 people	China Statistical Yearbook	+

presents a comprehensive summary of all variables, including their definitions, data sources, and expected relationships with RER. This enables readers to better understand the operationalization of key constructs and their empirical role.

4. Results and Analysis

4.1. Basic Regression Analysis

The baseline regression results are presented in

Table 4. Baseline regression results.

VarName	(1)	(2)
	RER	RER
FTZ	0.0423 ***	0.0346 **
	(2.8956)	(2.5812)
FDI		0.0000
		(0.0097)
agg		−0.0115 ***
		(−4.4573)
BE		0.1171 **
		(2.3944)
une		−0.0215 ***
		(−3.7463)
RD		1.1256
		(1.1309)
infra		−0.0000
		(−0.1378)
medic		−0.0070 ***
		(−3.2853)
Fixed time	YES	YES
Fixed city	YES	YES
N	390	390
R2	0.1041	0.3207

Note: ** p < 0.05, *** p < 0.01. Standard errors are reported in parentheses.

. All specifications include year and province fixed effects—with year effects controlling for national time trends and province effects accounting for unobserved regional heterogeneity. As shown, regardless of whether control variables are included, the estimated coefficient on FTZ remains consistently positive and statistically significant, suggesting that the establishment of FTZs significantly improves regional economic resilience. Specifically, Column (2) indicates that, at the 1% significance level, provinces with FTZs exhibit a 3.46% higher RER than those without, thereby confirming H1.

4.2. Heterogeneity Analysis

4.2.1. Batch Heterogeneity

The baseline regression results suggest that FTZ establishment positively contributes to RER. However, the magnitude and direction of this impact may vary across different policy implementation periods. Accordingly, the 21 FTZs are categorized into three cohorts based on their launch years: the first batch (2013–2015), the second batch (2017–2018), and the third batch (2019–2020). A batch heterogeneity analysis is conducted to assess differential effects across these phases. As shown in

Table 5. Regression results of heterogeneity analysis.

VarName	Different Batches			Different Regions
	(1)	(2)	(3)	(4)	(5)
	RER	RER	RER	RER	RER
FTZ × first	0.1059 ***
	(5.6014)
FTZ × second		−0.0046
		(−0.3549)
FTZ × third			−0.0373 *
			(−1.7864)
Coastal × FTZ				0.0524 ***
				(3.5587)
Inland × FTZ					0.0086
					(0.5329)
Control variables	YES	YES	YES	YES	YES
Fixed time	YES	YES	YES	YES	YES
Fixed city	YES	YES	YES	YES	YES
N	390	390	390	390	390
R2	0.3849	0.3849	0.3849	0.3349	0.3349

Note: * p < 0.1, *** p < 0.01. Standard errors are reported in parentheses. Columns (1)–(3) show the results of the interaction between batch dummy variables and FTZs, while columns (4)–(5) present the results of the interaction between regional dummy variables and FTZs. For ease of analysis, these results are classified into different groups in this paper.

, columns (1)–(3), the first batch of FTZs exhibits a significant positive impact on RER at the 1% level, confirming their strong resilience-enhancing effect. In contrast, the second batch shows no statistically significant effect, while the third batch demonstrates a weak negative effect on RER, significant at the 10% level, although the magnitude of the coefficient is relatively small.

This finding deviates from conventional expectations. The author argues that the first batch of FTZs (2013–2015), as national-level strategic pilots, rapidly generated scale and demonstration effects due to preferential policy treatment and concentrated governmental support, thereby substantially boosting regional economic resilience. Specifically, leading FTZs such as Shanghai and Guangdong benefited from “policy novelty”—they were the first to implement core institutional innovations, including negative lists for foreign investment and cross-border RMB settlement, which significantly lowered transaction costs through strong marginal effects [10]. However, these innovations lost effectiveness in later phases due to policy diffusion and saturation—by 2019, most key reforms had already been extended nationwide, diminishing the distinctiveness and competitiveness of newly established FTZs [5].

In contrast, the second (2017–2018) and third (2019–2020) FTZ cohorts may have faced constraints from policy convergence and adverse global conditions, which hindered the replication of the initial success. Notably, these two rounds coincided with major external shocks: the U.S.—China trade war (2018–2019), which disrupted export-oriented industries central to FTZ development, and the COVID-19 pandemic (2020–2022), which severely restricted cross-border flows, thereby undermining the FTZs’ roles in trade and investment facilitation [11]. Moreover, many later-stage FTZs were located in inland provinces (e.g., Heilongjiang, Yunnan) with relatively weaker infrastructure and lower openness, making it more difficult for them to capitalize on FTZ policy advantages compared to the coastal regions in the first batch.

4.2.2. Region Heterogeneity Analysis

Given the significant disparities in economic development between coastal and inland regions, the impact of FTZs on RER is likely to be heterogeneous. This analysis adopts a broad regional classification consistent with the existing literature on Chinese regional policy [23], while more granular regional breakdowns are left for future research.

According to the results in columns (4)–(5) of Table 5, the establishment of FTZs has a significant positive impact on the RER of coastal provinces, while the impact on the RER of inland provinces is not statistically significant.

This regional disparity can be attributed to three key factors, consistent with prior findings on market accessibility and industrial structure [11]: (1) Logistics Infrastructure: Coastal provinces (e.g., Guangdong, Zhejiang) benefit from well-developed ports and integrated rail networks, allowing them to lower trade costs through FTZ policies such as paperless customs clearance. In contrast, inland regions, burdened by higher transportation costs, are less able to capitalize on these trade facilitation measures. (2) Governance Capacity: Coastal areas possess stronger administrative capacity and institutional innovation experience (e.g., Shanghai’s leadership in adopting international trade rules), enabling more efficient implementation of FTZ reforms. Inland provinces often face bureaucratic inertia and delayed policy execution. (3) Industrial Structure: Coastal regions feature diversified and innovation-driven industries (e.g., high-tech manufacturing, advanced services) that are well-positioned to respond to FTZ-driven openness. In contrast, inland provinces tend to rely on resource-intensive and traditional sectors, which are less responsive to trade liberalization and green finance incentives.

In summary, coastal provinces—backed by superior logistics, advanced financial ecosystems, and higher openness—are better equipped to exploit FTZ policy advantages. They are more successful in attracting foreign investment, fostering technological spillovers, and promoting industrial upgrading, all of which contribute to a marked enhancement in regional economic resilience. In contrast, inland regions, constrained by geographical disadvantages, underdeveloped infrastructure, and weaker integration into global markets, face greater challenges in translating FTZ policies into tangible resilience gains.

4.3. Transmission Mechanism Analysis

Table 6. The transmission mechanism regression.

VarName	(1) GFin	(2) RER
FTZ	0.1439 **	0.0096
	(0.0604)	(0.0084)
GFin		0.1734 ***
		(0.0072)
Control variables	YES	YES
Fixed time	YES	YES
Fixed city	YES	YES
N	390	390
R2	0.4469	0.7349
Bootstrap	95% Conf. Interval
_bs_1	0.0018	0.0224
_bs_2	0.0110	0.0586
Sobel Z	2.405 **

Note: ** p < 0.05, *** p < 0.01. Standard errors are reported in parentheses.

presents the regression results of the transmission mechanism. We follow the standard mediation procedures of [51]. Column (1) in Table 6 reports the following:

$$GFin={\beta }_0+{\beta }_{1}FTZ+Controls$$

(9)

Column (2) reports the following:

$$RER={\gamma }_0+{\gamma }_{1}FTZ+{\gamma }_{2}GFin+Controls$$

(10)

A significant ${\gamma }_2$ alongside a reduced${ \gamma }_1$ indicates the existence of a mediating effect.

To enhance the robustness of this conclusion, this study further employs both the Bootstrap test and the Sobel test to examine the mediating effect of green finance. The confidence intervals of _bs_1 and _bs_2 do not contain zero, confirming the partial mediating effect through the Bootstrap test. In addition, the Sobel test rejects the null hypothesis of no mediation at the 5% significance level, providing further support for the significance of green finance as a mediating mechanism. This finding validates H2. However, the interaction between FTZ development and green finance may be subject to endogeneity concerns. For instance, FTZs with greater institutional openness may attract more green finance, introducing reverse causality. Similarly, unobserved factors, such as local environmental awareness or governance quality, may jointly influence both FTZ effectiveness and green finance growth. We mitigate this by (1) using lagged green finance indices and (2) controlling for provincial fixed effects, but residual endogeneity remains a limitation [2].

5. Robustness Tests

5.1. Parallel Trend Test

The multi-period DID method can be used to evaluate the impact of China’s FTZ on RER. This method rests on a critical assumption: the control group must provide a valid counterfactual for the treated group. Therefore, the parallel trend assumption is essential to ensure the credibility of the estimation results.

If the treatment and control groups exhibit divergent trends prior to policy implementation, the DID method would fail to produce reliable causal inferences. To assess the validity of this assumption, we construct interaction terms between the time dummy variables (before and after FTZ establishment) and the treated subsample indicator. This allows us to perform a formal test of the parallel trend hypothesis, structured as follows:

$${RER}_{it}={\omega }_0+{\sum }_{j=-M}^N{\omega }_{j}Treat\times {Post}_{i,t-j_1}+\gamma X_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(11)

For them, ${Treat}_i \times {Post}_{i,t-j}$ represents a dummy variable. If Province i established an FTZ within period t − j, the interaction dummy variable takes the value of 1; otherwise, it is 0. M represents the number of cycles (years) before the strategy experiment, and N represents the number of cycles after the strategy experiment. This approach quantifies the policy effects on Province i over M periods before and N periods after the establishment of the FTZ.

Table 7. Parallel trend test results (pre-treatment periods).

Relative Year (t-T)	Coefficient	Std. Error	p-Value	95% Conf. Interval
−6	0.002	0.011	0.857	[−0.019, 0.023]
−5	−0.005	0.010	0.631	[−0.025, 0.015]
−4	0.008	0.009	0.372	[−0.009, 0.025]
−3	−0.003	0.008	0.715	[−0.019, 0.013]
−2	0.001	0.007	0.886	[−0.013, 0.015]
−1	0.004	0.006	0.521	[−0.008, 0.016]

Note: Relative year = 0 is the first year of FTZ establishment. All pre-treatment coefficients are statistically insignificant (p > 0.1), confirming parallel trends.

reports the statistical results of the parallel trend test for the pre-treatment period, where “Relative Year (t-T)” represents the number of years prior to the establishment of the FTZ (e.g., −6 indicates six years before policy implementation). The coefficients for all pre-treatment periods are statistically insignificant (p-values > 0.1), and their 95% confidence intervals all include zero. These results confirm that the treatment and control groups followed parallel trends in regional economic resilience prior to the FTZ policy intervention, thereby validating the key identification assumption of the multi-period DID framework.

This statistical evidence is further visually supported by Figure 1, before the policy was implemented (on the left side of the black vertical line), the estimated coefficient fluctuated around zero. However, after the establishment of the FTZs (to the right of the black vertical line), the estimated coefficients become statistically significant and positive, exhibiting a steady upward trend. These findings suggest that FTZ policies have the potential to enhance regional economic resilience. Furthermore, the positive effect is positively correlated with the duration of the policy implementation.

5.2. Placebo Test

While the benchmark model controls for regional characteristics, the impact of FTZs on RER may still be subject to missing variables or other policy disruptions. To this end, the robustness of the results was verified with a placebo test.

Following the relevant approach, we randomly select 3 provinces from the 30 provinces as the treatment group, while the remaining provinces serve as the control group [4]. We then generate 500 random samples to re-estimate the baseline model specified in Equation (1). Figure 2 presents the kernel density distribution and p-values of the FTZ coefficients. The results indicate that most estimated coefficients are statistically insignificant and approximately follow a normal distribution. This suggests that baseline regression results are unlikely to be driven by unobservable factors or other policies, confirming that our conclusions are reliable rather than coincidental.

5.3. PSM-DID

To address sample selection bias and enhance comparability between the treatment group and the national baseline, thereby improving the accuracy of the DID estimates, this study adopts the PSM-DID approach as a robustness check. In the absence of additional identification strategies, combining PSM with DID helps mitigate potential selection bias.

Following best practices, the propensity score model includes only covariates measured prior to FTZ implementation, in order to avoid post-treatment bias [42]. Variables that may have been influenced by FTZ policies are either excluded or introduced in lagged form. Given the relatively small sample size of the treatment group and the large number of matching covariates, a 1-to-4 nearest neighbor matching strategy is adopted to minimize information loss due to excessive unmatched observations.

Table 8. Balance diagnostics before and after PSM.

Covariate	Mean (T)	Mean (C)	Std. Bias (%)	t-Test (p)	After Matching: Std. Bias (%)	After Matching: t-Test (p)
FDI	1.24	0.68	23.5	0.032	4.1	0.781
agg	5.12	3.89	18.7	0.045	3.2	0.812
BE	3.62	3.21	16.9	0.058	2.8	0.835
RD	0.015	0.009	21.3	0.039	3.5	0.796

Note: Std. bias < 10% and p > 0.1 after matching indicate successful balance.

presents the balance diagnostics for key covariates before and after matching. Before matching, several covariates—such as foreign direct investment (FDI) and economic agglomeration—exhibited substantial standardized biases (exceeding 10%) between the treatment group (provinces with FTZs) and the control group (provinces without FTZs), with some differences being statistically significant (p < 0.1). After applying 1-to-4 nearest neighbor matching, the standardized biases across all covariates drop below 5%, and none of the differences remain statistically significant (p > 0.7). These results suggest that the observable characteristics between the two groups are effectively balanced, thereby ensuring that the subsequent PSM-DID estimates are less likely to be biased by selection on observables.

Regression analysis is conducted using the matched sample, as shown in column (1) of

Table 9. Robustness tests.

VarName	PSM-DID	Winsorize Data	Change Period	Replace Variable	Replace Variable
	(1)	(2)	(3)	(4)	(5)
	RER	RER	RER	RER_1	RER_2
FTZ	0.0374 **	0.0325 ***	0.0378 ***	0.1349 *	0.0298 *
	(2.0583)	(2.8213)	(2.7258)	(1.7726)	(2.0073)
Control variables	YES	YES	YES	YES	YES
Fixed time	YES	YES	YES	YES	YES
Fixed city	YES	YES	YES	YES	YES
N	172	390	330	390	390
R2	0.4090	0.3250	0.3190	0.1229	0.3015

Note: * p < 0.1, ** p < 0.05, *** p < 0.01. Standard errors are reported in parentheses.

. The results indicate that the coefficient for FTZ remains significantly positive at the 5% level, which is consistent with the baseline regression, further confirming that FTZ policies contribute to enhancing RER.

5.4. Winsorization of Data

Extreme values in the dataset may introduce estimation bias or result in instability. To address this, a Winsorization technique is applied, trimming the top and bottom 5% of the data distribution. The regression results using the Winsorized dataset, reported in column (2) of Table 9, show that the FTZ coefficient remains significantly positive at the 1% level. This finding confirms that outliers do not distort the validity of the baseline estimation results.

5.5. Shorten the Data Period

The primary analysis utilizes data spanning from 2010 to 2022. However, external shocks or overlapping policies may bias estimation outcomes. Notably, the COVID-19 pandemic, which began in 2020, presents a major external disruption. To minimize pandemic-related confounding effects, the dataset is truncated by excluding post-2020 observations, and the model is re-estimated. As shown in Column (3) of Table 9, the FTZ coefficient remains significantly positive, indicating that the core conclusions are robust to changes in the time window.

5.6. Replace the Explained Variable

Given the regional disparities in economic performance, alternative measures of economic resilience are introduced as a sensitivity analysis. First, economic resilience is proxied by employment changes, capturing labor market responses [52]. Second, an extended RER index is constructed by incorporating industrial diversification using the Herfindahl index of the secondary and tertiary sectors [43]. The regression results in columns (4) and (5) of Table 9 indicate that FTZ establishment continues to exert a significant positive impact on RER, thereby reinforcing the robustness and reliability of the main findings.

6. Conclusions and Discussion

This study utilizes provincial panel data from 2010 to 2022 to first estimate China’s RER and then applies a multi-period DID method to evaluate the impact of FTZ establishment. Our findings confirm that FTZ policies promote RER, with provinces hosting FTZ exhibiting 3.46% higher economic resilience than those without FTZ. Even after a series of robustness checks, this conclusion remains robust. Furthermore, we find that green finance development plays a moderating role in enhancing the impact of FTZ on RER. Finally, the effect of FTZ on RER varies significantly across different batches and regions. Regarding batch heterogeneity, our results indicate that the first batch of FTZs had a significant positive impact on RER, whereas the subsequent two batches had weaker or even negative effects. In terms of regional heterogeneity, FTZs significantly enhance RER in coastal provinces, while their impact on inland provinces is not statistically significant.

These findings align with China’s institutional context: as pilots of institutional openness, FTZs reduce trade and investment frictions through policies like negative lists and paperless customs clearance [10], crucial for an export-driven economy. For instance, early reforms in the Shanghai FTZ helped stabilize foreign trade during the 2018–2019 US—China trade war, directly boosting RER. In contrast, later FTZs in inland regions (e.g., Heilongjiang) lack such infrastructure and global trade connections, explaining their weaker effects.

Internationally, similar patterns emerge: FTZs in Vietnam and Malaysia also show stronger resilience effects in coastal, export-oriented regions [27]. However, China’s top-down policy design, concentrating resources on early batches, results in larger batch disparities than more market-driven FTZs like Singapore. The moderating role of green finance aligns with EU findings [34], where green bonds and credit improve regional stability; China’s FTZ-specific green finance innovations further amplify this effect.

Our DID analysis assumes no spillover effects, yet neighboring provinces may be indirectly affected. For example, city-level FTZs have modest positive spillovers on neighboring cities’ energy efficiency [41], and spatial resilience research shows economic shocks propagate via supply chains [50]. Coastal FTZs might thus indirectly enhance inland provinces’ RER through trade links, suggesting our estimates of direct FTZ effects may be downwardly biased or partially capture indirect impacts. Due to data and method limitations, spatial spillovers are not modeled here, representing a caveat and an area for future research using spatial econometric or synthetic control methods.

This study has several limitations. First, it does not model spatial spillovers between provinces, potentially underestimating total FTZ impacts or conflating direct and indirect effects. Second, residual endogeneity remains between FTZs and green finance despite using lagged indices and fixed effects, as unobserved factors like local environmental governance may influence both. Third, analysis at the provincial level limits insights into micro-level mechanisms, such as how FTZ firms leverage green finance to build resilience.

Future research could address these by incorporating spatial econometric models to separate direct from spillover effects, using instrumental variables to resolve endogeneity in the FTZ–green finance relationship, and employing firm-level data to explore heterogeneous responses. Comparative studies with other developing countries implementing FTZs (e.g., India, Brazil) could reveal broader lessons for leveraging institutional openness to enhance economic resilience.

These findings have important policy implications. First, optimizing FTZ strategic layouts requires regionally differentiated development: prioritize expanding FTZs in coastal provinces to strengthen institutional innovation and open hubs, while supporting inland provinces with targeted policies such as cross-border logistics subsidies and industry development. Second, a dynamic policy iteration mechanism is needed to improve the effectiveness of later FTZ batches, involving continuous pilot–evaluation–optimization and focus on emerging areas like digital trade and green supply chains. Third, integrate FTZ policies with green finance through coordinated frameworks, including pilot zones for green finance innovation, mandatory environmental disclosure, cross-border green products, tax incentives, and fast-track approvals for green projects. Fourth, establish a cross-batch experience-sharing platform and risk warning system using big data and adaptive assessment models, with dynamic exit mechanisms for ineffective FTZ batches to mitigate policy risks. Finally, considering potential spatial spillovers, policymakers should promote regional coordination linking inland FTZs to coastal supply chains rather than isolated development.

Given each FTZ’s unique local economic base, studying localized characteristics and drivers is vital for accelerating FTZ development. The insignificant FTZ impact in inland regions underscores the need to consider these local factors carefully.