Recent empirical research confirms an inverted U-shaped relationship between judicial development (JD) and economic growth in China. Using a comprehensive evaluation index system and advanced econometric models, the study analyzed data from 30 provinces over two decades to uncover the complex dynamics between legal institutional development and regional economic performance.
The research constructed a JD evaluation index based on four dimensions—legal framework, judicial services, case volume, and case processing—using 11 specific indicators. The Analytic Hierarchy Process (AHP) and Entropy Method (EM) were integrated to determine indicator weights, enhancing the objectivity and robustness of the evaluation model. This approach allowed for a more accurate assessment of JD levels across provinces from 2000 to 2019.
The baseline regression model revealed a significant non-linear relationship: in the early stages of JD development, improvements in the judicial system enhance economic growth by improving contract enforcement, dispute resolution, and property rights protection. However, beyond a certain threshold, excessive judicial development begins to constrain economic activity through increased procedural complexity, higher compliance costs, and reduced business flexibility.
The threshold analysis identified a critical turning point at a JD level of 0.645. Below this threshold, JD significantly stimulates economic growth, with a coefficient of 1.027 (significant at the 1% level). Above this level, the relationship reverses, with JD imposing net economic costs, as indicated by a coefficient of -0.637 (also significant at the 1% level).
Further analysis using the System Generalized Method of Moments (SYS-GMM) confirmed the robustness of these findings, addressing potential endogeneity issues arising from bidirectional causality between JD and economic growth. The SYS-GMM model showed consistent results, with JD exhibiting a positive effect at early stages and a diminishing, eventually negative, impact beyond the threshold.
Regional heterogeneity was also observed. The Eastern region, which includes more economically developed provinces like Shanghai, Beijing, and Jiangsu, showed the strongest inverted U-shaped relationship. In contrast, the Central and Western regions exhibited similar patterns but with weaker marginal effects, likely due to less mature market mechanisms and institutional environments.
The study also examined the impact of control variables. Employment expansion (NEP) and capital formation (GCF) were consistently positive and significant, reinforcing the importance of labor and capital inputs in driving economic growth. Marketization (MI) also showed a positive effect, indicating that a more open and competitive market environment enhances resource allocation efficiency. However, technological input (FERDP) had a negative and statistically insignificant coefficient in some models, suggesting inefficiencies in converting innovation investments into economic output.
Case studies of provinces like Jiangsu illustrated the practical implications of the findings. In 2000, Jiangsu had a JD level of 0.414 and a GDP growth rate of 10.6%. As JD improved, GDP growth rose, peaking at 14.9% when JD reached 0.556 in 2008. Beyond this point, further increases in JD were associated with declining GDP growth, dropping to 5.9% by 2019 when JD reached 0.726.
The decline in growth beyond the threshold can be attributed to several factors. First, excessive judicialization introduces institutional friction, increasing the complexity and cost of legal procedures. Second, rising legal service costs disproportionately affect small and medium enterprises (SMEs), limiting their operational flexibility. Third, inefficiencies in judicial resource allocation—such as a shortage of grassroots judges and poor interdepartmental coordination—lead to slower case processing and redundant workloads, creating a paradox of high-quality but low-efficiency judicial systems.
The findings suggest that while judicial development is crucial for economic growth in its early stages, policymakers must be cautious about over-institutionalization. Balancing legal reforms with economic dynamism requires moderation, coordination, and efficiency in the future development of China’s rule-of-law environment.
— news from Nature Human Behaviour
— News Original —
The inverted U-shaped relationship between judicial development and economic growth: evidence from China
China’s JD evaluation
China’s JD evaluation index system
Assessing the level of China’s JD is a key aspect of quantitatively studying the impact of JD on economic growth. The evaluation of regional JD includes assessments of the regional legal system, access to judicial services, civic legal awareness, case processing efficiency, and other various aspects (Marciano et al., 2019). However, there is currently no widely accepted set of indicators and publicly available authoritative statistical data for measuring the level of JD. Therefore, most scholars use related indicators as proxies for the level of JD (Kaufmann et al., 1999). Drawing on Voigt’s research, this paper selects 11 indicators from four dimensions—legal framework, judicial services, case volume, and case processing—to construct the evaluation index system for China’s JD, as detailed in Table 1 (Voigt, 2012; Voigt, 2013; Voigt, 2016).
Evaluation method for China’s JD
The aforementioned four dimensions and 11 specific indicators reflect different facets and levels of China’s JD. To enhance the objectivity, accuracy, and comprehensiveness in gauging the level of JD in China, this paper integrates the Analytic Hierarchy Process (AHP) and Entropy Method (EM). The AHP is employed to determine the weights of the criteria at the top level, while the EM is utilized to ascertain the weights of the indicators at the lower level (Xiao et al., 2022).
The comprehensive use of AHP and the EM leverages the strengths of each method, enhancing the comprehensiveness and robustness of the evaluation model (Shen and Liao, 2022). AHP, characterized by a clear structure and subjective involvement, is more suitable for small-scale problems. The EM proves more efficient in handling large-scale problems by considering information entropy to mitigate subjective influences. AHP considers interdependencies between indicators through pairwise comparisons but is sensitive to correlation. In contrast, the EM is relatively independent of indicator relationships. Additionally, AHP introduces decision-makers through the hierarchy structure and comparisons, while the EM reduces reliance on pairwise comparisons through computation. Therefore, integrating both methods maintains clarity in problem structure, improves computational efficiency, balances subjective and objective factors, and enhances the robustness of the evaluation model.
Model construction
This research explores whether an inverted U-shaped relationship exists between JD and economic growth in China. To achieve this objective, the study focuses on the 30 provinces in China from 2000 to 2019. This timeframe is chosen considering the impact of the COVID-19 pandemic on China’s economy in 2020 and 2021, as well as data gaps in the Tibet Autonomous Region. Furthermore, by carefully selecting the study subjects and considering external factors, this research establishes non-linear panel regression models and threshold regression models to unveil the non-linear impact of JD on China’s economic growth (Chen and Lee, 2005).
Variables selection
The selection of variables in an econometric model requires a thorough consideration of their complex relationships (Glaeser et al., 2004). Consequently, we systematically reviewed existing research findings on the impact of JD on economic growth. Combining these findings with the results of the earlier analysis regarding the pathways through which JD influences regional economic growth, we ultimately determined six model variables from key dimensions: economic growth, JD, labor force, capital, technology, and marketization, as illustrated in Table 2. These six variables span critical aspects of the economic growth research domain, aiming to accurately reflect the impact of JD on China’s economic growth.
Dependent variable
The GDP growth rate is selected to represent the level of economic growth across China’s provinces.
Core explanatory variable: JD levels for the 30 provinces in China are calculated from 2000 to 2019 based on the JD evaluation index system constructed in the previous section.
Control variables
According to the Cobb-Douglas production function, labor, capital, and technology are important variables that affect economic growth (F Reynès, 2019). Sufficient high-quality labor enhances human capital, driving enterprise productivity, innovation, and regional economic growth (Wijaya et al., 2021). Capital investment stimulates business expansion and infrastructure development, elevating regional output while boosting investment attractiveness (Pelinescu, 2015). Technological adoption accelerates production efficiency and industrial upgrading, strengthening global competitiveness and economic resilience (Asongu & Odhiambo, 2020). Collectively, these factors constitute fundamental drivers of regional economic advancement. Hence, this paper selects employment, total capital formation, and R&D personnel full-time equivalent as indicators for labor, capital, and technology. Currently, there is no authoritative data on capital stock indicators for various provinces in China. Existing research often uses the perpetual inventory method to estimate physical capital stock. However, different studies use different base periods for capital stock and depreciation rates, leading to significant differences in data (Kataoka and Mitsuhiko, 2013). Therefore, this paper uses total capital formation as the indicator for capital stock. On the other hand, when studying the specific situation in China, it is necessary to consider the impact of China’s market-oriented reforms on economic growth. The improvement in the level of marketization is usually accompanied by a freer, more open, and competitive market environment, promoting enhanced resource allocation efficiency, stimulating entrepreneurial and innovative vitality, enhancing enterprise competitiveness, and promoting regional economic growth (Chen et al., 2021). This paper draws on Xin’s research and selects the marketization index as the indicator to measure the level of marketization in China (Xin and Xin, 2017).
Baseline regression model construction
To empirically test whether JD has a non-linear impact on China’s economic growth, we first constructed a baseline regression model, as shown in Formula (1).
$${{GDP}{GR}}_{{it}}= alpha +{ beta }_{1}{JDL}_{{it}}+{ beta }_{2}{{JD}L}_{{it}}^{2}+{ beta }_{3}{{NEP}}_{{it}}+{ beta }_{4}{{GCF}}_{{it}}+{{ beta }_{5}{{FERDP}}_{{it}}+{ beta }_{6}{{MI}}_{{it}}+ mu }_{i}+{ lambda }_{t}+{ varepsilon }_{{it}}$$
(1)
In this model, ({{ rm{GDP}}{ rm{GR}}}_{{ rm{it}}} ) denotes the GDP growth rate of province i in year t, serving as the dependent variable. ({{ rm{J}}{ rm{D}}{ rm{L}}}_{{ rm{it}}} ) is the core explanatory variable, capturing the level of JD in each region. ({{ rm{NEP}}}_{{ rm{it}}} ), ({{ rm{GCF}}}_{{ rm{it}}} ), ({{ rm{FERDF}}}_{{ rm{it}}} ) and ({{ rm{MI}}}_{{ rm{it}}} ) are control variables representing the number of employed persons, gross capital formation, full-time equivalent of R&D personnel, and the marketization index, respectively. The constant term is denoted by ( alpha ). To account for unobservable time-invariant factors across provinces (e.g., institutional foundations, geographic conditions) and common shocks in specific years (e.g., national policies, macroeconomic fluctuations), we incorporate both province fixed effects (({ mu }_{i}) ) and year fixed effects (({ lambda }_{t}) ). The term ({ varepsilon }_{{it}} ) represents the idiosyncratic error. This model aims to identify the fundamental direction of the impact of JD on China’s economic growth, as well as to explore the potential existence of an inverted U-shaped relationship.
SYS-GMM model construction
Considering the potential bidirectional causal relationship between JD and economic growth—where robust judicial systems stimulate growth, while economic advancement enables greater investment in judicial capacity—the baseline empirical model likely suffers from endogeneity issues. Estimating this relationship using OLS or fixed effects methods may lead to biased coefficient estimates. To address this problem, we adopt the System Generalized Method of Moments (SYS-GMM) approach, which was proposed by Blundell and Bond. This method extends the Difference GMM by combining the level equation with the first-differenced equation, using lagged levels and lagged differences of the variables as instruments. SYS-GMM is particularly suitable for short panel data with dynamic structures, as it effectively mitigates the correlation between explanatory variables and the error term (Vitenu-Sackey and Acheampong, 2022). By introducing a lagged dependent variable, the SYS-GMM model is set as shown in Formula (2).
$${{GDP}{GR}}_{{it}}= alpha +{ beta }_{1}{{GDP}{GR}}_{{it}-1}+{ beta }_{2}{JDL}_{{it}}+{ beta }_{3}{{JD}L}_{{it}}^{2}+{ beta }_{4}{{NEP}}_{{it}}+{ beta }_{5}{{GCF}}_{{it}}+{{ beta }_{6}{{FERDP}}_{{it}}+{ beta }_{7}{{MI}}_{{it}}+ mu }_{i}+{ lambda }_{t}+{ varepsilon }_{{it}}$$
(2)
Here, ({{ rm{GDP}}{ rm{GR}}}_{{ rm{it}}} ) represents the one-period lag of the GDP growth rate, which captures the dynamic persistence of economic growth. The other variables retain the same definitions as in Eq. (1). The model continues to include province fixed effects and year fixed effects to control for unobserved heterogeneity and time-specific shocks. The introduction of the SYS-GMM approach enhances the causal inference of the regression analysis and serves as a robustness check against potential endogeneity, thereby improving the credibility and reliability of the estimation results.
Threshold regression model construction
To further examine whether JD exerts a significant non-linear threshold effect on economic growth—and to validate the robustness of the conclusions derived from the baseline and SYS-GMM models—this study introduces a threshold regression model proposed by Hansen, as shown in Formula (3) and (4) (Huang et al., 2018) .
$${{GDPGR}}_{{it}}= alpha +{ beta }_{1}{JDL}_{{it}}+{ beta }_{2}{{NEP}}_{{it}}+{ beta }_{3}{{GCF}}_{{it}}+{{ beta }_{4}{{FERDP}}_{{it}}+{ beta }_{5}{{MI}}_{{it}}+ mu }_{i}+{ lambda }_{t}+{ varepsilon }_{{it}} ,({JDL} le eta )$$
(3)
$${{GDPGR}}_{{it}}= alpha +{ beta }_{1}{{JDL}}_{{it}}+{ beta }_{2}{{NEP}}_{{it}}+{ beta }_{3}{{GCF}}_{{it}}+{{ beta }_{4}{{FERDP}}_{{it}}+{ beta }_{5}{{MI}}_{{it}}+ mu }_{i}+{ lambda }_{t}+{ varepsilon }_{{it}} ,({JDL} > eta )$$
(4)
In these formulas, ({ rm{JDL}} ) represents the threshold variable, and ({ rm{ eta }} ) denotes the threshold value. The threshold regression model is a statistical method used to study non-linear relationships between variables. The flexibility of the threshold regression model allows it to adapt to complex data structures. This model permits researchers to explore whether there is a threshold effect between variables. This is crucial for understanding situations where the relationship may change within a specific range of values. To ensure the robustness of threshold estimation, this study applies a non-parametric bootstrap method, repeating the sampling procedure 10,000 times to obtain an empirical distribution of the threshold and to test the statistical significance of the threshold effect. Through the construction and estimation of the threshold regression model, we can rigorously examine whether the relationship between JD and economic growth varies by stage, enhancing the credibility of the empirical findings and strengthening the study’s policy implications.
Descriptive statistical analysis
This paper conducted descriptive statistical analysis on the six variables involved in the model, and the results are shown in Table 3. The difference between the maximum and minimum values of the six variables is relatively small. The skewness value is small, while the kurtosis value is large. This suggests that the distribution of the sampled data is relatively symmetric but has heavy tails. Additionally, the p-values for the JB (Jarque-Bera) statistics are all less than 0.05, leading to the rejection of the null hypothesis. Thus, it can be concluded that none of the six variables follows a normal distribution.
Results of baseline regression model
Before conducting the baseline regression analysis, it is essential to determine whether a fixed effects model (FE) or a random effects model (RE) is more appropriate. The two models make different assumptions regarding individual heterogeneity in panel data; hence, a formal statistical test is necessary for model selection. This study applies the Hausman test to make this determination. As shown in Table 4, the Chi-square statistic is 67.32 with a corresponding p-value of 0.000, which strongly rejects the null hypothesis that the random effects estimator is consistent and efficient. Therefore, the fixed effects model is deemed superior, and all subsequent regressions are estimated accordingly.
The estimation results based on the fixed effects model are presented in Table 5. The R-squared value of 0.885 suggests that the selected explanatory variables explain a substantial proportion of the variation in economic growth rate, indicating a good overall fit.
From the perspective of the core explanatory variables, the coefficient of ({{ rm{J}}{ rm{D}}{ rm{L}}}_{{ rm{it}}} ) is positive, while that of ({{ rm{JD}}{ rm{L}}}_{{ rm{it}}}^{2} ) is negative, both significant at 5%. This indicates a significant inverted U-shaped relationship between the level of JD and economic growth. In the early stages of JD, the gradual improvement of the judicial system leads to more efficient resolution of commercial disputes, more effective enforcement of contracts, and stronger protection of property rights. These enhancements safeguard the legitimate interests of enterprises and residents, improving economic efficiency, boosting business confidence, and encouraging increased investment, innovation, and production expansion, which together enhance regional economic dynamism. Moreover, improvements in JD strengthen China’s competitiveness in the global economic landscape by attracting more foreign investors, thereby injecting new momentum into the country’s economic growth. In addition, a more developed judicial system contributes to social stability by reducing conflict and dissatisfaction, thus fostering a more productive environment for both businesses and individuals. However, when the level of JD surpasses a certain threshold, its positive effect on economic growth begins to diminish and may even become negative. This turning point can be attributed to several factors. First, an excessively developed judicial system may place undue emphasis on legal procedures, creating rigidities that constrain business flexibility and innovation. Second, the costs associated with legal services may rise sharply, increasing the financial burden on firms and households. Lastly, as judicial institutions mature, the government may shift its focus toward sustainable development, which can lead to the curtailment of investments and capital flows that do not align with long-term sustainability goals, thereby reducing the growth momentum of the regional economy.
Further analysis reveals that the turning point in the relationship between JDL and GDPGR occurs when JDL reaches 0.684. That is, when JDL exceeds 0.684, the marginal impact of JD on economic growth turns negative—indicating diminishing returns or even inhibitory effects of excessive JD. A spatiotemporal analysis of 30 Chinese provinces shows that the top five provinces in terms of JDL are Shanghai, Beijing, Tianjin, Guangdong, and Jiangsu. These regions are also among the most economically developed areas in China, having achieved rapid growth by leveraging policy advantages, geographical positioning, and early-stage reform initiatives. However, with the side effects of early growth—such as environmental degradation and emerging property rights disputes—these provinces have shifted their judicial focus toward protecting regional environments, intellectual property rights, and personal privacy. As JDL exceeds the threshold of 0.684, the GDPGR in these regions has shown a notable decline. Taking Jiangsu Province as a case study, in 2000, Jiangsu’s JDL stood at 0.414 with a GDPGR of 0.106. As JD improved, GDPGR also rose, peaking in 2008 when JDL reached 0.556 and GDPGR hit 0.149. However, beyond this point, further increases in JDL were associated with declining GDPGR. By 2019, JDL had risen to 0.726, while GDPGR dropped to 0.059. This declining trend may stem from institutional friction and rising governance costs associated with excessive judicialization. On one hand, as the rule of law is strengthened, legal procedures have become increasingly complex, and compliance requirements have become more stringent. This has imposed greater institutional constraints on firms’ financing, expansion, and innovation activities. In Jiangsu, for instance, enterprises seeking to engage in land transfers, environmental approvals, or IP-related expansions often face multiple layers of judicial review and burdensome document verification processes, which significantly lengthen investment cycles and raise transaction costs. According to the China SME Development Report (2022), about 62% of manufacturing SMEs in Jiangsu cited legal and regulatory requirements as a major obstacle to their expansion. On the other hand, the market mismatch between the supply and demand of legal services has become increasingly pronounced. High-quality legal resources tend to be concentrated in large enterprises, while SMEs face high service costs and access barriers. In cities like Nanjing and Suzhou, the rising fees of well-established law firms have compelled ordinary enterprises to allocate substantial financial and human resources to legal compliance, contracts, and arbitration, undermining their operational flexibility and capital efficiency. Furthermore, there is evidence of diminishing marginal returns on judicial resource allocation. That is, increased judicial investment has not always translated into proportional improvements in judicial performance. Although Jiangsu has advanced initiatives such as smart courts and one-stop diversified dispute resolution platforms, persistent challenges—such as a shortage of grassroots judges and poor interdepartmental system interoperability—have led to slower case processing and redundant workloads, creating a paradox of high quality but low efficiency. In regions with advanced judicial systems, these institutional frictions are particularly evident. They not only raise the cost of institutional operation but also erode the marginal momentum of economic growth, signaling the need for greater moderation, coordination, and efficiency in the future development of the rule-of-law environment.
Regarding the control variables, the coefficient of NEP is significantly positive, indicating that an expansion in employment scale contributes to economic growth. The increase in labor supply directly enhances regional production efficiency and output levels. Similarly, GCF exhibits a significantly positive coefficient, suggesting that capital accumulation remains a key driver of China’s economic growth, as it provides sufficient production funds for enterprises, improves infrastructure, and enhances investment attractiveness. The coefficient of MI is also significantly positive, implying that a higher degree of marketization facilitates more efficient resource allocation, promotes the rational flow of production factors, and thus stimulates economic growth. In contrast, the coefficient of FERDP is negative and significant at the 10% level, suggesting that technological input in some regions has not yet translated effectively into economic output, indicating a mismatch between input and output. This reflects the fact that the effectiveness of innovation-driven growth still needs further improvement.
In summary, the baseline regression model not only confirms a significant non-linear inverted U-shaped relationship between JD and economic growth, but also highlights multiple structural drivers of economic growth. These findings provide a robust empirical foundation for the subsequent use of the SYS-GMM model and threshold regression to conduct robustness and heterogeneity checks.
Endogeneity test
Considering the potential bidirectional causality between JD and economic growth—where economic growth may also drive judicial resource allocation and institutional reforms—this study addresses endogeneity concerns by employing the SYS-GMM based on the baseline model. The regression results are presented in Table 6.
From the perspective of model validity, firstly, the coefficient of the dynamic term ({{ rm{GDPGR}}}_{{ rm{it}}-1} ) is 0.163 and statistically significant at the 1% level, indicating the presence of inertia in economic growth and thus justifying the inclusion of the lagged dependent variable in the model. Secondly, the SYS-GMM estimation passes relevant diagnostic tests: the AR(1) test yields a p-value less than 0.05, indicating first-order serial correlation in the residuals, which is a common and acceptable feature in dynamic panel models. The AR(2) test has a p-value greater than 0.1, suggesting no evidence of second-order serial correlation, thereby satisfying a key assumption for the validity of the SYS-GMM approach. Additionally, the Hansen test for overidentifying restrictions yields a p-value greater than 0.1, indicating that the instruments used are valid and there is no issue of overidentification, thus supporting the reliability of the SYS-GMM estimates.
Regarding the regression results, the coefficient on JDL is positive while its squared term is negative, both significant at the 5% level. This again confirms the significant non-linear inverted U-shaped relationship between JD and economic growth. It suggests that JD has a positive and significant promoting effect on economic growth at early stages, but when JD surpasses a certain threshold, its marginal effect on growth diminishes and eventually turns negative, consistent with the conclusions of the baseline regression model. Among the control variables, the coefficients of NEP and GCF are positive and significant at the 5% and 10% levels, respectively, further affirming the critical roles of labor and capital inputs in driving economic growth. The coefficient on MI is also positive and highly significant at the 1% level, indicating that improved market mechanisms substantially enhance resource allocation efficiency and economic dynamism. Notably, the coefficient on FERDP is negative but statistically insignificant, reflecting potential inefficiencies in the input-output conversion of technological investments in some regions of China. This highlights the need for further policy measures aimed at optimizing innovation resource allocation and improving technology commercialization efficiency.
In summary, the SYS-GMM estimation results maintain the robustness of the core findings after addressing endogeneity, fixed effects, and dynamic characteristics, strengthening the causal inference of the empirical analysis in this study.
Robustness test
To further verify the non-linear effects of JD on economic growth in China and to examine the robustness of the inverted U-shaped relationship obtained from the baseline regression and SYS-GMM models, this study introduces a threshold regression model for extended analysis. The threshold regression model effectively captures non-linear mechanisms by identifying structural breaks in the explanatory variable at specific threshold values, serving as an important supplement and validation to the previous models. A critical step in the threshold model setup is the selection of the optimal number of thresholds, which directly affects the model’s fit and explanatory power. Following Hansen’s bootstrap methodology, this study tests single, double, and triple-threshold specifications for the threshold variable JDL, conducting significance tests for each case. The bootstrap procedure is repeated 10,000 times to ensure the robustness of the test results. The relevant test outcomes are summarized in Table 7.
As shown in Table 7, the single-threshold model for JDL yields a highly significant F-statistic of 92.34 (p = 0.000), confirming a structural break in JD’s economic impact at a critical threshold. Conversely, double- and triple-threshold specifications show statistically insignificant results (p = 0.269 and p = 0.563, respectively), strongly rejecting the need for additional thresholds. Therefore, this study adopts the single-threshold model as the final threshold specification. The threshold value is further estimated using the bootstrap method, and the results are presented in Table 8.
According to the estimation results, the threshold value for JDL is 0.645, with a relatively narrow confidence interval, indicating a precise estimate and good statistical properties. The detailed regression results of the threshold model are shown in Table 9.
According to the regression results, when the JDL is below the threshold value of 0.645, the coefficient is 1.027 and statistically significant at the 1% level. This indicates that JD significantly stimulates economic growth during its early stages. However, once the JDL exceeds the threshold of 0.645, the relationship reverses—the coefficient becomes −0.637 (significant at 1%), suggesting excessive institutionalization imposes net economic costs. This inflection point confirms an inverted U-shaped relationship between JD and economic growth. These findings provide further evidence of the inverted U-shaped relationship between JD and economic growth. As for the control variables, both NEP and GCF continue to exert positive effects on economic growth, reaffirming the critical role of labor and capital in supporting China’s economic expansion. MI, however, exhibits a negative coefficient, which may imply a potential tension between market mechanisms and institutional constraints in regions with higher levels of the rule of law. Although the coefficient for FERDP is negative and not statistically significant, it may reflect the limited short-term economic returns of technological investment. In summary, the results of the threshold regression are highly consistent with those of the baseline and SYS-GMM models, further confirming the non-linear inverted U-shaped impact of JD on economic growth. This consistency strengthens the robustness and credibility of the empirical findings presented in this study.
Heterogeneity analysis
To further investigate whether the impact of JD on economic growth varies across regions, this study considers China’s significant regional disparities. Based on the classification criteria of the National Bureau of Statistics, the 30 provinces in mainland China (excluding Hong Kong, Macao, and Taiwan) are divided into three major regions: Eastern, Central, and Western. Fixed effects models are constructed separately for each region to examine the heterogeneity in the relationship between JD and economic growth. Building upon the baseline model, we estimate the effects of both the linear and quadratic terms of JDL on regional economic growth within each region, comparing the degree of nonlinearity, marginal effects, and statistical significance. The regression results are presented in Table 10.
Based on the regression results, an inverted U-shaped relationship between JD and economic growth is observed across all three regions, though the shape of the curve and the level of statistical significance differ notably.
First, in the Eastern region of China, the coefficient of JDL is 1.103 and significant at the 1% level, while the coefficient of its squared term is −0.841, significant at the 5% level. This indicates that the non-linear effect of JD on economic growth is most pronounced in the East, forming a typical and highly significant inverted U-shaped relationship. This suggests that in the early stages, judicial reform and institutional development in the East provided strong legal support for economic activities, significantly improving the business environment and the efficiency of resource allocation. However, as the judicial system reaches a higher level of development, excessive judicial intervention and rising enforcement costs may begin to suppress enterprise innovation and investment enthusiasm, leading to a decline in marginal benefits. Second, in the Central region, the coefficients of JDL and its squared term are significant at the 5% and 10% levels, respectively, also showing a clear inverted U-shaped structure. This implies that the JD in the Central region similarly exhibits a non-linear mechanism in promoting economic growth. However, the marginal effects are relatively weaker compared to the East. This may be due to the less mature market mechanisms and relatively lagging institutional environments in the Central region, which limit the full realization of institutional dividends from JD. Lastly, in the Western region, the coefficient of JDL is 0.592 and significant at the 10