Impact of air pollution perception on environmental governance satisfaction

The study investigates how public perceptions of air pollution affect satisfaction with environmental governance in China within an ESG framework. It posits that air pollution is a salient, widely perceived environmental issue and can serve as a gauge for the effectiveness of government environmental management. The research explores two key mechanisms: (1) environmental tax as a mediating factor that could mitigate dissatisfaction stemming from pollution perceptions; and (2) government trust as a moderating factor that conditions how perceptions translate into satisfaction. Using CSS 2019 survey data matched with provincial environmental tax data, the authors aim to identify the direct effect of perceived air pollution on environmental governance satisfaction, test heterogeneity across regions (east vs. central-west), urban/rural residence, and age groups, and quantify mediation by environmental tax and moderation by trust at different government levels (central, district/county, township). The study addresses gaps in understanding the interplay between perception, policy instruments, and institutional trust, with implications for ESG performance and public policy.

The literature indicates a consistent negative association between perceived pollution and satisfaction with environmental governance. Prior studies (e.g., Zhu et al., 2023; Sun and Zhu, 2023; Sun et al., 2021; Ruan et al., 2022) show that higher perceived pollution correlates with lower satisfaction, supporting H1. Heterogeneity evidence suggests stronger negativity in eastern regions, urban areas, and among older cohorts, motivating H2a–H2c. Research on environmental taxes shows they can reduce pollution, spur green innovation, enhance productivity and well-being, and potentially increase satisfaction with governance, leading to H3 that environmental tax mitigates dissatisfaction tied to pollution perception. On government trust, theories of political trust and local governance (Hetherington, 2005; Nabatchi, 2010) and resource dependence (Pfeffer & Salancik, 1978) suggest nuanced moderation: higher central trust may increase expectation mismatches (H4a), lower county/district trust may exacerbate dissatisfaction under pollution (H4b), and township trust may have limited influence due to resource constraints (H4c). The research logic (Fig. 1) links air pollution perception to satisfaction (H1), with mediation by environmental tax (H3), moderation by multilevel government trust (H4a–c), and heterogeneity by region, urban/rural, and age (H2a–c).

Data: The study uses the 2019 Chinese Social Survey (CSS), a nationally representative survey covering 30 provinces (excluding Xinjiang, Hong Kong, Macao, Taiwan), yielding 10,283 valid responses. Analytical samples for regressions comprise 4,663 observations. Provincial environmental tax data are matched from the China Statistical Yearbook. Measures:

• Dependent variable: Environmental governance satisfaction (envg_sat), based on the item assessing government performance in protecting the environment and addressing pollution; responses originally 1 (very good) to 4 (very poor), inverted so higher values reflect greater satisfaction.
• Independent variable: Air pollution perception (airppp), from 1 (very serious) to 4 (no such phenomenon), inverted so higher values indicate greater perceived severity.
• Mediator: Environmental protection tax (envtax), a composite of six types of levies (urban land usage, resource tax, environmental protection tax, arable land occupation, vehicle/boat tax, vehicle purchase tax).
• Moderators: Trust in central (trust_cent), district/county (trust_dc), and township (trust_town) governments, coded from 1 (total distrust) to 4 (total trust).
• Controls: Education, income (lninc), gender, age, marital status, party membership, household registration (urban/rural), employment status, and region (east/central/west). Descriptive statistics indicate mean envg_sat 2.94 (SD 0.78), mean airppp 2.07 (SD 0.92), mean envtax 432.98 (SD 218.5). Models:
• Baseline OLS: envg_sat = α0 + α1 airppp + Z + ε.
• Mediation (Baron & Kenny): envtax = β0 + β1 airppp + Z + ε; envg_sat = γ0 + γ1 airppp + γ2 envtax + Z + ε; total effect includes indirect path β1γ2. KHB method is also used to decompose total, direct, and indirect effects.
• Moderation: envg_sat = A0 + A1 airppp + A2 mod + A3 (airppp × mod) + Z + ε, for each trust level. Estimation and diagnostics:
• Baseline regressions by OLS; ordered probit (Oprobit) used for robustness given ordinal outcome; VIFs indicate no multicollinearity (mean VIF 1.46).
• Robustness: (i) Oprobit instead of OLS; (ii) alternative subjective pollution measure (water pollution perception); (iii) objective pollution proxy (provincial AQI); (iv) province-level fixed effects.
• Heterogeneity: Subsamples by urban/rural, region (east vs. central-west), and age (old vs. young), with Bdiff tests for coefficient differences.
• Marginal effects: Oprobit marginal effects computed for probability changes across outcome categories.
• Endogeneity: 2SLS IV approach using the cubic deviation from the mean of air pollution perception (airpppgap3) as an instrument; first-stage strength assessed by F-statistic; exclusion restriction discussed via semi-reduced-form checks. All models include standard controls; standard errors reported; significance at conventional levels.

• Baseline effect (H1): Air pollution perception significantly and negatively predicts environmental governance satisfaction. OLS without controls: coefficient -0.2650 (SE 0.0117, p<0.01). With controls: -0.2605 (SE 0.0117, p<0.01).
• Robustness: Results remain negative and significant using Oprobit; when replacing airppp with water pollution perception; when using AQI as objective proxy (AQI coefficient -0.2271, p<0.01); and with province effects.
• Heterogeneity (H2):
  • Urban vs. rural: negative in both subsamples (urban -0.256; rural -0.260; both p<0.01) but difference not significant (Bdiff p=0.44), so H2b not supported.
  • Region: East -0.280 vs. Central-West -0.223 (both p<0.01); difference significant (Bdiff p=0.02), supporting H2a.
  • Age: Old -0.337 vs. Young -0.235 (both p<0.01); difference significant (Bdiff p<0.001), supporting H2c.
• Marginal effects: At means, a one-unit increase in perceived air pollution is associated with a 10.8 percentage-point decrease in the probability of rating environmental governance as “very good,” with corresponding increases in lower categories.
• Mediation by environmental tax (H3):
  • airppp → envtax: 8.6427 (SE 3.1116, p<0.01).
  • envtax → envg_sat: 0.0002 (SE 0.0001, p<0.01).
  • In the full model with envtax, airppp remains negative (-0.2627, p<0.01).
  • KHB decomposition: total effect -0.261; direct effect -0.263; indirect effect via envtax 0.00213 (p<0.05), indicating a small but significant mitigating (masking) effect of environmental tax on dissatisfaction.
• Moderation by government trust (H4):
  • Central trust: interaction airppp × trust_cent = -0.0359 (SE 0.0176, p<0.05), supporting H4a—higher central trust is associated with a more pronounced decline in satisfaction under severe pollution.
  • District/county trust: interaction airppp × trust_dc = 0.0348 (SE 0.0120, p<0.01), supporting H4b—lower trust exacerbates dissatisfaction; higher trust buffers it.
  • Township trust: interaction airppp × trust_town = 0.0154 (SE 0.0106, n.s.), supporting H4c—no significant moderating effect.
• Endogeneity (2SLS): First stage shows strong instrument strength (F=1818.85); second stage confirms a significant inverse relationship between air pollution perception and satisfaction (e.g., -0.268, SE 0.0257, p<0.01), supporting causal interpretation.

The findings confirm that higher perceived air pollution reduces satisfaction with environmental governance, addressing the core research question on perception-satisfaction dynamics within the ESG context. The negative association persists across model specifications and alternative measures, indicating robustness and suggesting that citizens’ subjective experiences align with objective pollution indicators (AQI). Heterogeneity patterns imply that expectations and exposure profiles vary: residents in economically advanced eastern provinces and older individuals exhibit stronger dissatisfaction responses, consistent with greater pollution exposure and health vulnerability. The environmental tax’s small but significant mediating effect suggests that fiscal instruments can partially counteract dissatisfaction by signaling proactive governance, incentivizing pollution control, and potentially delivering visible improvements. Government trust moderates the perception-satisfaction link in nuanced ways: high central trust amplifies dissatisfaction under high pollution (expectation gap), while higher district/county trust attenuates dissatisfaction (perceived local efficacy). Township-level trust shows limited moderating influence, consistent with resource constraints and narrower mandates. Collectively, these results highlight the importance of integrating policy tools (taxation) with trust-building strategies to improve governance satisfaction amid environmental challenges, offering practical insights for ESG-oriented governance.

This study demonstrates a robust inverse relationship between air pollution perception and satisfaction with environmental governance in China. It identifies environmental tax as a mitigating mediator and shows that government trust differentially moderates the perception-satisfaction linkage across administrative levels: central trust can heighten dissatisfaction when pollution is severe, while district/county trust can buffer it; township trust has no significant moderating effect. The research contributes by quantifying these mechanisms using a large national survey matched with fiscal data, employing multiple robustness checks and IV estimation to address endogeneity. Future research should extend to longitudinal designs to assess dynamic policy impacts (e.g., post-2018 Environmental Protection Tax Law), examine roles of information transparency, government reputation, and public participation, and explore how internet use and environmental awareness shape perceptions and satisfaction. Comparative panel studies across regions and time can further elucidate evolving governance effectiveness. Policy implications include strengthening ecological policy design and implementation, enhancing governance capacity and transparency, refining environmental tax structures (e.g., differential rates and dedicated funds), and expanding public engagement in environmental decision-making to improve trust and satisfaction.

• Cross-sectional analysis (CSS 2019) limits causal inference over time; dynamic effects of policies like the 2018 Environmental Protection Tax Law are not captured.
• Potential measurement inconsistencies in subjective scales (e.g., trust item ranges) and reliance on self-reported perceptions may introduce bias.
• Environmental tax is measured at the provincial level, which may not fully align with individual perceptions at finer spatial scales.
• The moderating role of trust is assessed with limited institutional dimensions; broader constructs (government reputation, transparency, participation) were not directly modeled.
• External validity beyond China may be constrained by institutional and cultural contexts.