📄 medicine

Mismatch between Urban Eco-sanitation Improvement and Health Service Supply Capacity: Evidence from Chinese City Panel Data and Interpretable Machine Learning

Using panel data from Chinese cities and interpretable machine learning, this study reveals that while urban eco-sanitation conditions have improved, they do not automatically translate into enhanced health service supply capacity due to a mismatch in resource allocation and fiscal mechanisms, necessitating an integrated governance framework to bridge the gap.

Original authors: Zhen Peng, Lin Wang

Published 2026-07-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Zhen Peng, Lin Wang

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a city as a giant, living house. For a long time, researchers have been asking: "If we make the house cleaner, greener, and better ventilated, will the people living inside automatically get healthier and have better access to doctors?"

This paper, titled "Mismatch between Urban Eco-sanitation Improvement and Health Service Supply Capacity," asks a slightly different, more practical question: "If we fix the house's plumbing and plant more trees, does that automatically mean we will have more doctors, more hospital beds, and more clinics ready to treat people?"

The authors, Zhen Peng and Lin Wang from Qingdao University of Technology, looked at data from hundreds of Chinese cities over 22 years (2002–2024) to find the answer. They used a mix of traditional statistics and advanced computer learning (machine learning) to see how these two things connect.

Here is the breakdown of their findings in simple terms:

1. The "House" Got Cleaner, But the "Clinic" Didn't Grow Fast Enough

The study found that Chinese cities have done a great job improving their "eco-sanitation." Think of this as fixing the pipes, cleaning the trash, adding parks, and making sure the air and water are cleaner. These things improved steadily.

However, the "health service supply"—the actual number of doctors, hospital beds, and clinics per person—did not keep up. It grew, but much slower.

The Analogy: Imagine you renovate a kitchen to be spotless and add a beautiful garden. You expect the house to be healthier. But the study found that just because the garden is beautiful, it doesn't automatically mean you hired more chefs or bought more pots and pans to cook the meals. The "clean house" didn't automatically create more "medical resources."

2. The "Cleanliness" Doesn't Automatically Turn into "Doctors"

The researchers tested if better green spaces or cleaner water directly led to more doctors. The answer was no.

The Finding: Improving the environment is not a magic button that instantly creates more healthcare capacity.
The "Pollution" Surprise: Interestingly, they found that cities with more pollution pressure actually had more doctors and beds.
The Explanation: This isn't because pollution is good for health! It's because cities with heavy industry and pollution usually have bigger populations, more money (fiscal capacity), and more urgent needs. So, they naturally hire more doctors. It's a "response to pressure," not a benefit of the pollution itself.

3. The Money Trail is Broken

The authors wondered: "Maybe cities spend money on cleaning the environment, and that same money eventually helps build hospitals?"

The Finding: They couldn't find a clear path. The money spent on landscaping or sewage treatment didn't seem to flow directly into building more hospitals or hiring more staff.
The Analogy: It's like the city has two separate bank accounts. One account pays for the garden and trash collection, and the other pays for the hospital. Even if the garden account gets a big deposit, the hospital account doesn't automatically get a transfer. They are managed by different departments with different goals.

4. It Depends on the City's "Stage of Life"

The relationship between a clean environment and good healthcare isn't the same for every city.

Big, Rich Cities: In large, wealthy cities (mostly in the east), the link is weak. These cities already have lots of doctors and lots of parks. Adding more parks doesn't suddenly create more doctors because the system is already full or complex.
Smaller or Developing Cities: In smaller or less wealthy cities, just planting trees or cleaning water isn't enough to fix the lack of doctors. They need direct investment in healthcare first.
The Machine Learning Insight: The computer models showed that the most important things for having a good healthcare system are actually how many people live there, how much money the city makes, and what kind of jobs they have—not just how clean the air is.

5. The Big Takeaway

The paper concludes that a clean environment is necessary, but it is not enough.

You cannot assume that if you build a beautiful, green, clean city, the healthcare system will magically fix itself.

The Metaphor: Think of the city as a car. The "eco-sanitation" is the shiny paint and the clean engine oil. The "health service" is the driver and the spare tires. You can have the shiniest car in the world, but if you don't have a driver or spare tires, you still can't go on a long trip.
The Recommendation: City planners need to stop treating these as separate projects. They need to plan the "cleaning" and the "doctor hiring" together. If you build a new park, you should also plan where the nearest clinic is and make sure there are enough nurses to go there.

In short: Making a city greener and cleaner is great, but it doesn't automatically mean you will have better healthcare. To get better healthcare, you have to specifically plan for it, hire the staff, and build the hospitals, regardless of how pretty the parks are.

Technical Summary: Mismatch between Urban Eco-sanitation Improvement and Health Service Supply Capacity

Problem Statement
Existing literature predominantly examines the effects of environmental exposure on individual health outcomes (e.g., mortality, disease burden). However, there is a significant gap in understanding whether improvements in the urban eco-sanitation environment (green infrastructure, municipal sanitation, pollution control) can be effectively converted into enhanced urban health service supply capacity. This study addresses whether environmental improvements automatically translate into the ability of a city to allocate, support, and continuously provide basic health service resources. It challenges the assumption that ecological governance and health service expansion are synchronous, proposing instead that structural mismatches may exist between environmental upgrades and the capacity of the health system to respond.

Methodology
The study employs a mixed-methods approach combining panel econometrics with interpretable machine learning, utilizing a dataset of Chinese prefecture-level and above cities from 2002 to 2024.

Data Sources: The analysis integrates four data streams: comprehensive urban panel data (GDP, fiscal, health resources, pollution), the China Urban Construction Statistical Yearbook (green space, water/gas supply, waste treatment), prefecture-level energy consumption data, and the CEADs city-level sectoral carbon-emission inventory.
Variable Construction:
- Dependent Variable: Urban Health Service Supply Capacity Index (HSC), constructed via entropy weighting using hospital beds, licensed physicians, and medical institutions per 10,000 persons.
- Core Explanatory Variable: One-year lagged Eco-sanitation Environment Index (ESE), composed of green infrastructure, municipal sanitation facilities, and pollution pressure indicators.
- Dimensions: The ESE is decomposed into Green Infrastructure Index (GII), Municipal Sanitation Facility Index (MSI), and Pollution Pressure Index (PPI).
Econometric Models:
- Baseline: Two-way fixed effects models (city and year) to estimate the average effect of lagged ESE on HSC.
- Mechanism Tests: Analysis of transmission channels via urban construction investment and public fiscal capacity.
- Moderation & Heterogeneity: Interaction models to test the modifying roles of energy intensity, carbon emissions, and pollution pressure, alongside subgroup analyses by region, city size, and economic development.
- Robustness: Includes alternative variable definitions, lag structures, sample adjustments, and placebo tests (future ESE) to address endogeneity.
Interpretable Machine Learning: Random Forest and Gradient Boosting models are used as supplementary tools to predict HSC, identify nonlinear relationships, and assess variable importance using SHAP (SHapley Additive exPlanations) values. The training set covers 2002–2018, with 2019–2024 as the test set to prevent temporal leakage.

Key Results

Conversion Mismatch: The baseline two-way fixed effects models reveal that improvements in the eco-sanitation environment do not stably convert into stronger health service supply capacity. In fact, the lagged ESE coefficient is significantly negative in most specifications, suggesting that environmental improvements are not automatically followed by proportional increases in health resource supply.
Dimensional Differentiation:
- Green Infrastructure: Shows a significantly negative association with HSC in dimension-specific models.
- Municipal Sanitation: Effects are statistically insignificant.
- Pollution Pressure: Shows a significantly positive coefficient. The authors interpret this not as a health benefit of pollution, but as a "pressure-resource allocation response": cities with high pollution often have larger populations, stronger fiscal capacity, and higher healthcare demand, leading to greater resource allocation.
Weak Transmission Mechanisms: Mechanism tests fail to identify a stable transmission chain where eco-sanitation improvements lead to increased urban construction investment that subsequently boosts health service capacity. Investment in landscaping and sanitation does not significantly correlate with expanded health service supply in the short term.
Moderation and Heterogeneity:
- Carbon Emissions: The interaction between ESE and city-level carbon emissions is positive and significant, suggesting that in high-emission (industrialized) cities, environmental improvements coincide more closely with public service expansion due to stronger underlying economic and fiscal bases.
- City Types: The relationship is significant and negative in eastern, central, high-GDP, and large-population cities (mature urban systems), but insignificant in western, low-GDP, and small-population cities. This indicates that the mismatch is most pronounced in developed urban centers where structural adjustments, rather than simple expansion, are occurring.
Machine Learning Insights: SHAP analysis confirms that economic factors (per capita GDP, population size, industrial structure) are the primary drivers of HSC variation, with eco-sanitation variables playing a secondary, supplementary role. Nonlinear dependence plots show that the marginal effect of eco-sanitation improvements flattens at high levels, while pollution pressure exhibits a nonlinear increasing relationship with supply capacity.

Key Contributions

Theoretical Shift: The study moves the research focus from "environmental exposure $\to$ health outcomes" to "environmental improvement $\to$ health service supply capacity." It reframes the issue from a positive-effect assumption to an investigation of conversion mechanisms and structural mismatches.
Integrated Framework: It constructs a unified analytical framework linking green infrastructure, municipal sanitation, pollution pressure, energy intensity, and carbon emissions to health service governance, revealing that these factors play differentiated and often non-linear roles.
Methodological Integration: By combining panel econometrics (for causal direction and average effects) with interpretable machine learning (for nonlinear patterns and heterogeneity), the study provides a more nuanced understanding of the boundary conditions under which environmental improvements fail to translate into health service capacity.

Significance and Claims
The paper claims that eco-sanitation improvement should not be regarded as a sufficient condition for enhancing health service supply capacity. The core finding is the existence of temporal, spatial, and institutional mismatches between environmental governance and health resource allocation.

Policy Implication: Healthy city development requires a unified governance framework that integrates eco-sanitation, municipal infrastructure, primary care planning, and fiscal arrangements. Environmental improvements alone cannot substitute for the strategic allocation of medical resources, workforce planning, and network density.
Modest Causal Claims: The authors explicitly state that their results represent "conditional within-city associations and directional evidence" rather than strict causal effects derived from a natural experiment. They acknowledge limitations regarding unobserved spatial spillovers and the inability to capture micro-level accessibility or service quality.
Governance Insight: The study suggests that in mature cities, the relationship between environment and health capacity reflects structural adjustment and diminishing marginal returns, whereas in developing cities, the lack of conversion is due to constraints in fiscal capacity and resource endowment.

In conclusion, the paper argues that to translate environmental gains into real health service capacity, policymakers must move beyond siloed infrastructure projects and adopt coordinated strategies that align ecological governance with health service network planning and resource allocation.