The impact of informal environmental regulation on total-factor carbon emission performance

Abstract

Growing public concern over environmental issues has heightened the significance of examining the impact of informal environmental regulation (IER) on total-factor carbon emission performance (TFCEP), offering a novel pathway to advance sustainable development goals (SDGs). This study establishes a theoretical framework elucidating the influence of IER on TFCEP, followed by empirical analysis utilizing China’s provincial panel data from 2004 to 2019 through fixed-effects models, threshold regression, and the spatial Durbin model. The results demonstrate that: (1) IER exerts a significant positive influence on TFCEP. This significant impact persists even after rigorous endogeneity and robustness checks are conducted. Moreover, this impact is particularly pronounced during the 2012–2019 period and in western regions. (2) While the short-term impact of IER on TFCEP is statistically insignificant, it exhibits a significant positive effect once the intensity of IER or the level of green technology innovation surpasses specific thresholds. (3) IER not only improves local TFCEP but also generates substantial spatial spillover effects, elevating TFCEP in neighboring regions. By investigating the low-carbon transition effects of public participation in environmental governance, this study provides policymakers with fresh theoretical insights and empirical evidence to support SDGs.

Introduction

Climate change stands as one of the most severe and urgent environmental challenges confronting nations worldwide¹. It not only exacerbates ecological degradation and biodiversity loss but also jeopardizes human health and well-being^2,3,4, thereby hindering progress toward Sustainable Development Goals (SDGs), particularly SDG 13 (Climate Action). China surpassed the United States to become the world’s largest carbon emitter as early as 2006⁵. In 2019, China contributes to 26% of global carbon emissions⁶. Notably, China is the first developing country to establish binding national targets for energy conservation and emission reduction, pledging to peak carbon emissions by 2030 and achieve carbon neutrality by 2060 ^{5; 6}. Total-factor carbon emission performance (TFCEP), a pivotal metric for assessing low-carbon development quality⁷, reflects the optimization of labor, capital, and other production factors to maximize economic output while minimizing carbon emissions^7,8,9. Unlike single-factor carbon emission performance (SFCEP) indicators—such as carbon intensity or carbon productivity, which focus narrowly on outputs^10,11,12—TFCEP holistically captures substitution effects among multiple production factors^13,14. This comprehensive approach better reflects the complexity and systemic nature of low-carbon transitions while aligning with broader SDGs, including sustainable economic growth^13,15. Given TFCEP’s critical role in climate mitigation and sustainable development^16,17, identifying its drivers and formulating optimization strategies is essential not only for achieving China’s “dual carbon” goals but also for informing global climate governance frameworks.

Environmental regulation (ER) is widely recognized as a pivotal institutional instrument for advancing low-carbon transitions^10,11,12. Based on institutional forms, ER can be categorized into formal environmental regulation (FER) and informal environmental regulation (IER)^18,19. FER constitutes a regulatory framework implemented by governments through mandatory instruments such as environmental legislation, carbon market mechanisms (e.g., emissions trading systems), and fiscal policies (e.g., carbon taxes)^18,20,21. Although FER has demonstrated positive effects on low-carbon transformation²², its limitations—including high regulatory costs and information asymmetry—cannot be overlooked^18,23. Furthermore, under this “bottom-up” decentralized governance model, local governments facing interregional economic competition may relax environmental standards to attract investment (a “race to the bottom”) or even establish collusive relationships with polluting enterprises (regulatory capture)^24,25,26, significantly impeding low-carbon progress. In contrast, IER serves as a complementary constraint mechanism through public petitions, environmental litigation, and community engagement, addressing environmental violations inadequately regulated by FER^18,27. Specifically, IER compels firms to improve environmental practices via enhanced disclosure mechanisms¹⁸, simultaneously reducing governmental oversight costs and mitigating information asymmetry^28,29. This positions IER as an effective catalyst for low-carbon transitions. Notably, with rising public environmental awareness and improved disclosure systems, IER has emerged as a critical force in global decarbonization efforts^10,24,27.

Existing studies primarily employ data envelopment analysis (DEA) and stochastic frontier analysis (SFA) to measure TFCEP^30,31,32, subsequently analyzing its driving factors (e.g., industrial structure, tourism development, agglomeration economy)^15,33,34. Compared with SFA, DEA has been more widely adopted in prior literature due to its avoidance of pre-specified functional forms and subjective weight allocation³⁵, coupled with its derived models (e.g., Super-SBM) that better address slack variable issues^15,36. Notably, ER has been empirically demonstrated as a critical driver of low-carbon development, garnering significant scholarly attention^10,11,12. Specifically, existing research has predominantly focused on examining the low-carbon transition effects of FER, including its impacts on SFCEP (e.g., carbon emission intensity, carbon productivity)¹⁰ and TFCEP^29,37. Concurrently, some scholars have systematically investigated the role of IER in facilitating low-carbon transitions, mainly analyzing its influence on aggregate carbon emissions¹⁸,^19,24,38 or carbon intensity¹⁰. Although limited literature has explored IER’s positive effect on enhancing TFCEP, these studies have primarily concentrated on specific sectors such as transportation, food production, and construction industries^29,36,39. Furthermore, these analyses typically adopt DEA-based approaches (e.g., SBM, Super-SBM) for TFCEP measurement^36,39.

In summary, while the low-carbon transition effects of IER have been preliminarily verified²⁴, its potential to enhance TFCEP remains substantially underexplored. Consequently, investigating the IER-TFCEP relationship carries both theoretical urgency and practical necessity. Simultaneously, the Porter Hypothesis and its extended studies demonstrate that environmental regulation’s impact on low-carbon transitions exhibits dynamic nonlinear characteristics, which may be influenced by regulatory intensity or green technology innovation levels^11,19. However, existing literature has yet to systematically reveal how these two factors influence the IER-TFCEP relationship. Furthermore, the synergistic nature of regional environmental governance and the transboundary diffusion of public environmental concerns endow environmental regulations with inherent spatial spillover effects^11,39. Nevertheless, the spatial spillover effects of IER on TFCEP remain insufficiently examined. This study therefore systematically investigates IER’s impact on TFCEP, focusing on three core research questions: (1) Can IER effectively enhance TFCEP? (2) How do informal environmental regulation intensity and green technology innovation levels impact the IER-TFCEP relationship? (3) Does IER generate spatial spillover effects on TFCEP?

The marginal contributions of this study are threefold. First, while existing literature has employed SBM or Super-SBM methods to measure TFCEP and examined IER’s impact on TFCEP, such studies predominantly focus on specific sectors like transportation and construction^29,39 moreover, their reliance on cross-sectional data comparisons fails to capture the dynamic and long-term nature of low-carbon transitions. In contrast, this study adopts the Super-SBM-GML model with panel data to construct economy-wide TFCEP metrics⁴⁰, thereby overcoming these limitations. These methodological advancements enrich current understanding of the relationship between IER and TFCEP by providing novel dual perspectives—dynamic and holistic. Second, recognizing that IER’s low-carbon transition effects may be contingent upon both IER intensity and green technology innovation levels^11,19;, we develop an analytical framework integrating these variables into the IER-TFCEP relationship analysis. This framework not only bridges critical knowledge gaps in quantitative research but also contributes to the Porter Hypothesis debate regarding environmental regulation. Furthermore, it yields fresh insights into how informal regulation and green innovation synergistically drive decarbonization. Third, accounting for the transboundary nature of public environmental demands and regional governance synergies, we pioneer spatial spillover effect analysis of IER’s impact on TFCEP, significantly expanding existing research^24,29;. Most importantly, this study offers policymakers new lenses for enhancing TFCEP through regional environmental cooperation while bolstering global confidence in IER-driven low-carbon transitions.

The remainder of this paper is structured as follows: Sect. 2 presents the theoretical hypotheses; Sect. 3 details the research design; Sect. 4 reports the empirical results; Sect. 5 provides discussion; and Sect. 6 concludes the study.

Theoretical hypothesis

Impact of IER on TFCEP

As a core component of environmental regulation instruments, IER plays a pivotal role in rectifying environmental violations and facilitating low-carbon transitions¹⁸²⁴;. Specifically, existing studies demonstrate that IER directly enhances TFCEP through three fundamental mechanisms: expanding regulatory channels, driving cost savings, and stimulating demand effects^18,19,39. First, IER broadens regulatory oversight through public complaints, social scrutiny, and media exposure, thereby intensifying corporate pressure for low-carbon transformation and consequently improving TFCEP^24,39. When FER proves inadequate or ineffective, these alternative regulatory channels become crucial for low-carbon governance by strengthening environmental constraints, optimizing resource allocation, and accelerating decarbonization¹⁸. Second, IER achieves cost savings through reduced information acquisition and policy implementation expenditures^18,28, thereby optimizing environmental governance funding. This reduction in compliance costs enhances the innovation compensation effect while improving the regulatory environment through mitigated information asymmetry²⁸, ultimately accelerating low-carbon transitions. Third, IER intrinsically reflects public demand for green products and environmental awareness^24,39. As consumer preference for sustainable goods grows, market pressures compel polluting firms to adopt low-carbon production methods³⁹, thereby promoting sustainable development. Furthermore, this demand-driven effect facilitates resource reallocation toward eco-friendly enterprises while phasing out carbon-intensive firms, ultimately enhancing TFCEP through industrial restructuring¹⁹. Therefore, this paper proposes the following hypothesis.

• H1: IER significantly improves TFCEP.

Nonlinear effect of IER on TFCEP

The Porter Hypothesis and existing research indicate that IER’s impact on low-carbon transitions exhibits significant threshold characteristics, with this nonlinear effect being influenced by key factors such as IER intensity and green technology innovation levels^11,19. First, regarding IER intensity: As a non-mandatory regulatory instrument, IER may fail to exert immediate pressure for low-carbon transformation in the short term¹¹, potentially increasing corporate compliance costs and environmental governance burdens²⁸, thereby yielding insignificant or even negative effects on TFCEP. However, over the long term, sustained social pressure from growing public environmental awareness will force polluting firms to adopt low-carbon retrofits and increase clean energy usage¹¹, ultimately enhancing TFCEP. Second, concerning green technology innovation: Studies directly and indirectly demonstrate its critical role in IER-driven low-carbon transitions¹⁹. Specifically, given the extended development cycles and systemic complexity of green technologies, their short-term application to low-carbon governance remains challenging^11,41, potentially raising compliance costs and resulting in insignificant or temporarily negative TFCEP effects. However, the cumulative advancement and diffusion of green innovations will eventually optimize resource allocation and accelerate corporate decarbonization³⁹, thereby significantly improving TFCEP. Therefore, this paper proposes the following hypothesis.

• H2a: The TFCEP-enhancing effect of IER becomes significant when informal environmental regulation exceeds certain thresholds.

• H2b: The TFCEP-enhancing effect of IER becomes significant when green technology innovation reaches sufficient levels.

Spatial spillover effects of IER on TFCEP

Existing studies demonstrate that IER typically exhibits significant spatial spillover effects due to the cross-regional dissemination of public environmental demands and the synergistic nature of regional environmental governance^11,39. Specifically, these spatial spillovers operate through three primary channels: information diffusion, factor mobility, and market linkage mechanisms. First, IER activities such as public supervision and media exposure in a given region generate demonstration effects that influence low-carbon governance practices in neighboring areas through information transmission²⁸. This cross-regional impact simultaneously exerts direct decarbonization pressure and enhances resource allocation efficiency in adjacent regions through institutional imitation mechanisms^28,42, ultimately improving their TFCEP. Second, facing pressure from social scrutiny and media exposure under IER, polluting industries often relocate to neighboring regions with weaker IER intensity^19,39. This industrial spatial restructuring, accompanied by factor mobility (including technology spillovers, capital flows, and skilled labor migration)²⁸, creates a “pollution halo effect” that elevates TFCEP in recipient regions. Third, IER fosters inter-regional collaboration between firms and stakeholders, facilitating the establishment of regional green governance synergies and industrial symbiosis networks^18,19. Such mechanisms enhance coordinated environmental governance, simultaneously reducing local energy consumption/emissions while amplifying spatial spillovers, thereby boosting TFCEP in neighboring regions. Therefore, this paper proposes the following hypothesis.

• H3: IER generates significant positive spatial spillover effects on TFCEP.

Figure 1 presents the theoretical framework of IER’s impact on TFCEP.

Fig. 1

The impact of IER on the TFCEP.

Research design

Model

Fixed effects model

Fixed-effects models are widely employed in empirical analyses of causal relationships between variables due to their capacity to mitigate omitted variable bias and endogeneity concerns^31,43. This study adopts this approach to test H1. The specification is as follows:

$$TFCE{P_{it}}={\alpha _1}IE{R_{it}}+{\alpha _2}Contro{l_{it}}+{\mu _i}+{\varphi _t}+{\varepsilon _{it}}$$

(1)

where: $TFCE{P_{it}}$ represents total-factor carbon emission performance; $IE{R_{it}}$ denotes informal environmental regulation; i indicates province; t signifies year; $Contro{l_{jit}}$stands for control variables; ${\mu _i}$ and ${\varphi _t}$ represent individual and time fixed effects, respectively; ${\varepsilon _{it}}$ is the random disturbance term.

Threshold regression model

Threshold regression models are widely employed to examine dynamic effects and nonlinear relationships in socioeconomic factors^29,37. This study adopts this approach to test H2. The specification is as follows:

$$TFCEPit=\alpha 0+\beta 1IERit*I\left( {\delta it \leqslant \eta } \right)+\beta 2IERit*I\left( {\delta it{\text{>}}\eta } \right)+{\alpha _2}Contro{l_{it}}+\varepsilon it$$

(2)

where: I(·) denotes the indicator function, with $\delta i,t$ and $\eta$ representing the threshold variable and threshold value, respectively.

Spatial regression model

In contrast to conventional econometric models, spatial econometric models incorporate spatial interdependence among variables¹⁸, thereby yielding empirical results that better reflect the geographical context and real-world dynamics of socioeconomic development¹¹. We therefore employ this methodology to test H3. The model specification follows:

$$TFCE{P_{it}}=\rho \sum\limits_{{j=1}}^{n} {WijTFCEPjt} +{\alpha _1}IERit+{\varphi _1}\sum\limits_{{j=1}}^{n} {WijIERjt+{\alpha _2}Contro{l_{it}}+{\varphi _2}\sum\limits_{{j=1}}^{n} {WijContro{l_{it}}+} \eta t+\mu i+} \varepsilon it$$

(3)

where: W_ij denotes the spatial weights matrix; $WijTFCEPjt$ and $WijIERjt$ represent the spatial lags of TFCEP and IER respectively; ${\beta _0}$ signifies the temporal lag effect; ${\rho _1}$, ${\rho _2}$, ${\varphi _1}$, and ${\varphi _2}$ capture spatial lag terms; ${\alpha _1}$ and ${\alpha _2}$ denote estimated coefficients; $\mu i$ and ${\eta _t}$ indicate spatial and time effects correspondingly; $\varepsilon it$ constitutes the stochastic error term.

Spatial kernel density estimation

Compared with conventional kernel density estimation, spatial kernel density estimation effectively incorporates spatial attributes, thereby better revealing intrinsic connections and interactions among geographical units⁴⁴. Specifically, spatial static kernel density estimation captures the mutual influences between local and neighboring socioeconomic phenomena in year t while accounting for spatial characteristics⁴⁴. We therefore employ this method to analyze the spatial dependence patterns of TFCEP. The formal specification is as follows:

$$f(x,y)=\frac{1}{{Nhxhy}}\sum\limits_{{i=1}}^{N} {Kx\left( {\frac{{{x_i} - {x_a}}}{{hx}}} \right)} Ky\left( {\frac{{{y_i} - {y_b}}}{{hy}}} \right)$$

(4)

$$g(y|x)=\frac{{f(x,y)}}{{f(x)}}$$

(5)

where: f(x,y)denotes the joint kernel density function of x and y; $g(y\mid x)$ represents the conditional distribution of random variable y given x; x_i and x_a indicate the random variable and its mean along the X-axis, respectively; y_i and y_a denote the random variable and its mean along the Y-axis, respectively; N is the number of provinces; h signifies the bandwidth; K_x and K_y are the kernel functions for x and y, respectively.

Variables

Dependent variable

The dependent variable in this study is TFCEP. As TFCEP fundamentally reflects relative changes in input-output efficiency, prior literature has predominantly measured it from an input-output perspective³⁰. Building on this consensus and existing research^13,30,33, we construct a comprehensive TFCEP indicator system (Table 1). Specifically, inputs include: fixed capital stock (capital input), employed population (labor input), and energy consumption (energy input); outputs encompass GDP (desirable output) and carbon emissions (undesirable output). While existing studies often employ SBM or Super-SBM models for measurement^36,39, these cross-sectional-data-dependent approaches fail to capture the dynamic, long-term nature of low-carbon transitions. In contrast, the Super-SBM-GML model, utilizing panel data, overcomes these limitations and better reflects real-world low-carbon economic growth⁴⁰. Following the TFCEP indicator system, we adopt the Super-SBM-GML model to calculate TFCEP indices, with formal specifications referencing the study of Wang and Luo⁴⁰. Notably, as these indices represent growth rates, they cannot directly indicate absolute TFCEP levels across years. Thus, following established methodology⁴⁵, we set the base year value at 1 and derive subsequent TFCEP indices through cumulative multiplication of annual change rates, ultimately serving as our dependent variable.

Independent variable

The independent variable in this study is IER. IER essentially embodies public environmental awareness^18,24. Existing studies demonstrate that key determinants of public environmental consciousness—including age structure, population density, income levels, and education attainment—are commonly adopted for IER measurement^18,27,39. Aligned with this consensus and constrained by data availability, we develop a comprehensive IER indicator system (Table 1). Building upon this system, we employ the entropy method coupled with linear weighted summation to quantify IER. This approach simultaneously mitigates subjective weighting bias and effectively captures temporal dynamics of variables^3,20, thereby yielding more objective and reliable measurements. The specific computational procedures follow the methodology of Wang et al.²⁰.

Table 1 Indicator system for TFCEP and IER.

Threshold variables

The threshold variables in this study include informal environmental regulation (IER) and green technology innovation (GTI). As the measurement method for IER aligns with that of the independent variable, we omit redundant elaboration here. Existing research establishes patent counts as direct manifestations of technological innovation⁴⁶, with green patents playing particularly pivotal roles in facilitating low-carbon transitions under environmental regulations^19,47. Following established methodology^40,47, we therefore adopt the number of green patent applications as the proxy variable for GTI.

Control variables

Building on existing research^31,48, this study incorporates key control variables to account for potential confounding factors: (1) Government regulation (GR), measured as the ratio of fiscal expenditure to GDP; (2) Opening up (OP), represented by total import and export volume; (3) Green level (GL), quantified as per capita public green space area; (4) Urbanization level (UL), measured by the proportion of urban population; (5) Human capital (HC), proxied by the number of college students per 10,000 population; and (6) Industrial structure (IS), assessed through industrial transformation indices.

Data

This study examines 30 Chinese provinces from 2004 to 2019 (excluding Tibet, Taiwan, Hong Kong, and Macao due to data limitations). Data were primarily collected from the China Statistical Yearbook, China Energy Statistical Yearbook, and national economic/social development bulletins, with minor missing values addressed through interpolation. To eliminate price fluctuations, all monetary values were converted to constant 2004 prices. Table 2 presents descriptive statistics, while Fig. 2 displays correlation coefficients. Variance inflation factors (VIF < 7.5) and correlation coefficients (< 0.8) for all independent variables meet established thresholds⁴³, confirming no severe multicollinearity issues.

Table 2 Descriptive statistical analysis of variables.

Fig. 2

Correlation coefficient matrix.

Results

Baseline regression

Table 3 presents the estimated effects of IER on TFCEP. Results from five regression models consistently demonstrate that IER exerts a statistically significant positive influence on TFCEP (p < 0.05), thereby supporting H1. This finding primarily stems from IER’s capacity to compel polluting firms toward low-carbon development, thereby optimizing resource allocation and reducing carbon emissions³⁹. These results provide novel empirical evidence reinforcing existing conclusions about IER’s role in facilitating low-carbon transitions^18,24.

Table 3 The baseline regression results.

Endogeneity test

While this study incorporates key control variables and employs multi-dimensional fixed-effects models to mitigate omitted variable bias, potential reverse causality may exist as low-carbon transitions could simultaneously influence public environmental awareness and subsequent IER intensity. To address this endogeneity concern, we adopt an instrumental variable (IV) approach. Following established methodology^43,49, we select air mobility coefficients and 5-period lagged IER as instruments. Higher air mobility coefficients indicate greater regional air dispersion capacity, which correlates with relaxed IER enforcement yet remains exogenous as it solely depends on geographical conditions⁴⁹. The 5-period lagged IER maintains strong correlation with current values while theoretically not directly affecting contemporaneous TFCEP - a specification further justified by China’s distinctive 5-year environmental governance cycles. Table 4 presents IV regression results. First-stage estimates show both instruments are statistically significant (F-statistic >10, p < 0.01), effectively ruling out weak instrument concerns. Second-stage diagnostics confirm robustness: Cragg-Donald Wald F-statistic substantially exceeds Stock-Yogo critical values; Kleibergen-Paap rk LM statistic reaches 38.836 (p < 0.01); and Hansen’s J-test (p >0.1) validates instrument exogeneity. After addressing endogeneity, IER retains statistically significant positive effects on TFCEP, confirming H1’s robustness.

Table 4 Endogeneity test.

Robustness check

Table 5 presents robustness check results through three methodological approaches. First, alternative dependent variables were constructed using the ratio method, SBM, and Super-SBM to measure carbon emission intensity/performance (Columns 1–3). Regression results consistently demonstrate IER’s significant effects in either reducing emission intensity or enhancing performance metrics. Second, replacing the independent variable with IER measured via the coefficient of variation method (Column 4) confirms IER’s robust positive impact on TFCEP. Third, simultaneous replacement of both independent and dependent variables using the aforementioned approaches (Columns 5–7) yields consistent findings - IER significantly mitigates emission intensity or improves performance metrics. Collectively, these tests provide robust verification for H1.

Table 5 Robustness check.

Heterogeneity analysis

Temporal heterogeneity

Given the dynamic and long-term nature of environmental governance, regulatory impacts may exhibit phased variations. We therefore partition the sample into two periods (2004–2011 and 2012–2019) to examine temporal heterogeneity in IER’s effects on TFCEP (Table 6, Columns 1–2). Results show positive coefficients for both periods, but statistical significance emerges only in the latter phase, indicating stronger TFCEP enhancement effects post-2012. This pattern likely reflects the time-lagged characteristics of environmental governance outcomes, providing novel empirical support for the Porter Hypothesis’ proposition that environmental regulation ultimately fosters low-carbon transitions through long-term mechanisms (innovation compensation effect)⁵⁰.

Regional heterogeneity

Significant regional disparities in economic development may lead to geographical heterogeneity in environmental regulation effects. Particularly, the eastern and central regions’ economic advancement far exceeds that of western China, potentially creating divergent low-carbon transition potentials and challenges. We therefore stratify the sample into eastern/central (economically developed) and western (less developed) regions to analyze spatial heterogeneity (Table 6, Columns 3–4). Results demonstrate IER’s TFCEP-enhancing effects across both zones, with notably stronger impacts in western areas. While eastern/central regions possess economic advantages, they face greater obstacles including higher carbon emissions and mitigation costs^51,52. Conversely, western China’s ecological endowments and lower historical emissions baseline amplify its transition potential, rendering IER more effective there.

Table 6 Heterogeneity analysis.

Nonlinear effect

Threshold model test

Table 7 presents the threshold model test results, revealing statistically significant single-threshold effects (p < 0.01) for both informal environmental regulation (IER) intensity and green technology innovation levels. This evidence indicates that IER’s impact on TFCEP is constrained by either its own regulatory intensity or the prevailing green innovation capacity, thereby exhibiting distinct nonlinear characteristics.

Table 7 Threshold model test.

Threshold regression results

Table 8 column (1) presents regression results with IER intensity as the threshold variable. When IER intensity falls below 0.450, the estimated coefficient (−0.716) lacks statistical significance; however, beyond this threshold, the coefficient becomes 1.092 with strong significance at the 5% level. These findings confirm that while IER’s short-term effects remain negligible, its long-term capacity to enhance TFCEP is substantial, thereby validating Hypothesis 2a. This result provides new evidence supporting existing conclusions¹¹.

Table 8 column (2) displays results using green technology innovation as the threshold variable. Below the innovation threshold of 9.185, IER’s coefficient (0.194) is statistically insignificant, whereas above this cutoff, it rises to 0.928 with significance at the 10% level. This demonstrates that green technology innovation exhibits a threshold effect—only beyond a critical level does it amplify IER’s positive impact on TFCEP, thus supporting Hypothesis 2b. These results provide new evidence for the findings of Jiang et al.⁵³.

Table 8 Threshold regression results.

Spatial spillover effects

Spatial autocorrelation

Figure 3 presents the spatial kernel density distribution and contour maps of TFCEP, revealing distinct spatial dependence patterns. When TFCEP values are below 1.5, observations cluster predominantly along the positive 45° line, indicating significant spatial positive autocorrelation. Conversely, at TFCEP levels exceeding 1.5, the distribution shifts toward the negative 45° line or aligns parallel to the X-axis, demonstrating spatial negative correlation or independence.

Fig. 3

Kernel density distribution (a) and density contour line (b) of TFCEP.

Spatial model selection

Table 9 Spatial model selection results.

The Table 9 reveal significantly positive global spatial autocorrelation indices, reaffirming the necessity and appropriateness of spatial regression modeling. Specifically, both Lagrange Multiplier (LM) and Robust-LM tests demonstrate statistical significance at the 1% level, supporting the adoption of a Spatial Durbin Model (SDM). Likelihood Ratio (LR) and Wald tests decisively reject (p < 0.01) the null hypotheses of SDM degenerating into either Spatial Error Model (SEM) or Spatial Lag Model (SLM). Furthermore, combined LR and Hausman test results confirm that the spatial-temporal fixed effects specification significantly outperforms alternative configurations. Consequently, this study ultimately employs a spatial-temporal fixed effects SDM for empirical analysis.

Spatial effects decomposition

Given that point estimates in Spatial Durbin Models cannot directly represent marginal effects and may be biased, this study employs partial differentiation^54,55 to decompose IER’s direct, indirect (spillover), and total effects (Table 10). Results demonstrate statistically significant total effects (p < 0.01) across three spatial weight matrices: contiguity-based (W_n), geographical distance (W_g), and nested contiguity-distance (W_ng), confirming IER’s robust TFCEP-enhancing capacity even after accounting for spatial dependence. Notably, both direct and indirect effects remain positive and statistically significant (p < 0.01), indicating IER not only accelerates local low-carbon transitions but also generates significant spatial spillovers to neighboring regions, thereby validating H3.

Table 10 Spatial effects decomposition.

Discussion

The role of environmental regulation in facilitating low-carbon transitions remains a contested scholarly discourse with unresolved consensus. While growing public environmental awareness and improved disclosure systems have established IER as a critical pathway for low-carbon growth^18,24, its impact on TFCEP remains underexplored. TFCEP embodies the SDGs’ coordinated development framework between economic and environmental objectives (particularly SDG 13), serving as a vital metric of low-carbon development quality^7,15. This study therefore investigates the IER-TFCEP relationship to address existing knowledge gaps. Our findings make a valuable theoretical contribution to the understanding of the low-carbon effects induced by IER and the Porter Hypothesis, while proposing novel insights worthy of discussion.

First, this study establishes that IER significantly enhances TFCEP, providing novel empirical evidence and fresh theoretical perspectives that reinforce existing conclusions about IER’s role in facilitating low-carbon transitions (e.g., reducing aggregate emissions, emission intensity)^10,18,24. Importantly, our findings extend prior sector-specific evidence (originally limited to industries like transportation, food production, and construction industries^29,36,39) to the broader economy-wide context, thereby substantially advancing understanding of the IER-TFCEP relationship. Furthermore, we identify significant temporal and spatial heterogeneity in these effects: This effect is particularly pronounced during the 2012–2019 period and in western regions. The temporal pattern provides novel empirical support for the Porter Hypothesis’ proposition that environmental regulation ultimately fosters low-carbon transitions through long-term mechanisms (innovation compensation effect)^29,50. In addition, compared with the eastern and central regions, the western region demonstrates significantly lower levels of carbon emissions and emission reduction costs^51,52. This regional disparity may lead to a more pronounced effect of IER on TFCEP in the western region.

Second, this study empirically validates that the positive effect of IER on TFCEP intensifies with increasing regulatory stringency. This finding corroborates existing conclusions regarding the more pronounced long-term impacts of environmental regulation on low-carbon transitions¹¹, while simultaneously providing novel empirical evidence supporting the Porter Hypothesis^29,50. The Porter Hypothesis postulates a nonlinear relationship between environmental regulation and sustainable economic growth, with its beneficial effects typically becoming more substantial over extended periods. This temporal pattern primarily stems from the immediate compliance costs imposed on firms in the short run, coupled with the inherent time lags in pollution control and technological progress^11,41, resulting in delayed but ultimately more significant policy outcomes. Furthermore, recognizing the pivotal role of green technology innovation in facilitating low-carbon transitions⁴⁷ and its influence on regulation-driven decarbonization processes¹⁹, this study systematically incorporates this factor into the IER-TFCEP analytical framework. The results reveal that IER’s TFCEP-enhancing effects become markedly stronger at higher levels of green technology innovation. This key finding not only reinforces the Porter Hypothesis but also substantially enriches the current body of literature¹⁸,^24,53.

Third, the results demonstrate that IER maintains a statistically significant positive effect on TFCEP even when controlling for spatial interactions, thereby providing additional empirical support for existing conclusions regarding IER’s role in promoting low-carbon transitions^10,18,24. Importantly, this study reveals that IER not only enhances local TFCEP but also generates significant spatial spillover effects that facilitate low-carbon development in adjacent regions. These effects can be primarily attributed to the synergistic nature of regional environmental governance and the cross-boundary dissemination of public environmental concerns, which collectively contribute to IER’s inherent spatial spillover characteristics^11,39. By extending the investigation of IER’s spatial spillover effects from SFCEP indicators (e.g., carbon emissions and carbon intensity) to the more comprehensive TFCEP framework, these findings substantially enrich the existing literature^11,21. Moreover, the results provide empirical evidence supporting the necessity of regional collaborative governance for low-carbon development. As shown in previous studies^56,57,58, the implementation of coordinated regional carbon emission control mechanisms can effectively accelerate low-carbon transitions due to the pronounced spatial interdependencies inherent in environmental governance systems.

Conclusions

Main conclusions

This study establishes a theoretical framework examining the relationship between IER and TFCEP, with empirical validation using China as the research context. The key findings are as follows: (1) IER significantly enhances TFCEP, a result that remains robust after addressing endogeneity concerns and conducting comprehensive sensitivity analyses. Notably, this effect proves particularly pronounced during the 2012–2019 period and in western regions of China. (2) IER’s impact on TFCEP exhibits dynamic nonlinear characteristics: while short-term effects are statistically insignificant, IER demonstrates significant TFCEP-enhancing effects once regulatory intensity or green technology innovation surpasses critical thresholds. (3) Beyond driving local low-carbon transitions, IER generates substantial spatial spillover effects that amplify TFCEP in neighboring regions.

Practical implications

These findings provide critical policy implications for leveraging IER to facilitate low-carbon transitions.

Establishing institutionalized and diversified public participation mechanisms is essential to enhance the governance efficacy of informal environmental regulation. Against the backdrop of continuously deepening global environmental awareness, the public has become a critical force in advancing green and low-carbon transformation. Studies indicate that informal environmental regulation significantly promotes total-factor carbon emission performance, with particularly pronounced effects observed during the period 2012–2019 and in the western regions of China. It is recommended to systematically advance efforts in the following three aspects: first, implement a mandatory environmental information disclosure system that specifies disclosure requirements and frequency for key corporate indicators such as carbon emissions, establish a unified digital platform, and introduce third-party verification to enhance the accessibility, comparability, and credibility of information; second, improve the environmental public interest litigation system by expanding the scope of litigation subjects, streamlining case filing and evidence procedures, and exploring mechanisms for cost-sharing and legal aid to substantially lower the barriers for public litigation; third, design diversified incentive measures, such as linking environmental behavior scores with access to public services and integrating green credit ratings with preferential loan policies, to effectively increase public participation motivation. In light of the notable effects observed in the western regions, efforts should be directed toward systematically enhancing residents’ environmental awareness and monitoring capabilities through community-based environmental education, fostering non-governmental monitoring initiatives, and implementing localized communication strategies.

Constructing a green technology innovation policy system is crucial to overcome the threshold limitations inherent in informal environmental regulation. Research indicates that the emission reduction effects of informal environmental regulation can only be fully realized when public participation and green technology reach a critical level. Policy design should be developed along the following three dimensions: first, enhance the standardization and binding force of corporate environmental information disclosure, establish a closed-loop management mechanism for public complaints, feedback, and corrective actions, and set up routine supervision channels and public opinion response procedures to improve the implementation intensity and responsiveness of informal environmental regulation; second, create special funds for low-carbon technologies in high-carbon industries (such as steel and chemicals), clarify key support areas and technology roadmaps, and attract diverse entities in tackling critical emission reduction technology research and development, thereby breaking through common technical bottlenecks; third, improve the industry-university-research collaborative green innovation mechanism by establishing a full-chain support system including pilot testing platforms, technology trading markets, and achievement transformation funds to accelerate the industrial application of green innovation outcomes. Meanwhile, performance-oriented differentiated incentive policies should be formulated, linking benefits such as green credit incentives and super-deduction of R&D expenses to quantifiable indicators including corporate carbon emission intensity reduction rates and technological innovation outputs, thereby forming an operable and assessable policy toolkit.

Promoting cross-regional collaborative environmental governance is essential to optimize the spatial spillover effects of informal environmental regulation. Informal environmental regulation not only facilitates local emission reduction but also generates regional synergistic benefits through technology diffusion and joint pollution prevention mechanisms. Institutional innovation should focus on the following three aspects: first, establish a cross-administrative “Regional Environmental Governance Alliance” to develop unified data-sharing protocols, mutually recognized standard lists, and joint enforcement procedures, thereby constructing a regularized coordination and deliberation mechanism; second, design a scientifically-grounded ecological compensation mechanism that determines compensation standards and flow directions based on observed pollution transfer, enabling beneficiary regions to provide financial, technological, and capacity-building support to source treatment areas, which in turn funds local environmental projects and public incentives; third, develop a cross-regional digital supervision platform integrating satellite remote sensing, big data, and Internet of Things (IoT) technologies to achieve real-time pollution source tracking, dynamic assessment of emission reduction outcomes, and efficient public access to environmental information, significantly reducing the cost of public cross-regional oversight. Through these innovative pathways, the overall environmental benefits of informal environmental regulation can be maximized, advancing coordinated low-carbon transition across regions.

Research limitations

While this study makes substantive contributions to the existing literature, several limitations warrant acknowledgment. First, constrained by data availability at the municipal/county level and the impact of COVID-19, our analysis relies on provincial panel data from 2004 to 2019. Although this approach validates established findings and yields novel insights, future research incorporating finer-grained sub-provincial data could significantly deepen understanding. Second, given the complexity of contemporary socioeconomic systems, subsequent studies should investigate potential threshold variables such as digital economy development and green finance in moderating the IER-TFCEP relationship. Third, despite extending current knowledge through nonlinear and spatial spillover perspectives, the temporal-spatial heterogeneity analysis remains incomplete, presenting critical directions for future inquiry.