Dataset Development of Industrial Green Total Factor Productivity and Water Environment Drag Index for Core Riverside Cities in the Yangtze River Economic Belt (2000–2022)

LIU Rong^1,2  WANG Zhenbo^1*

1. Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China;

2. University of Chinese Academy of Sciences, Beijing 100049, China

Abstract: We constructed a dataset of industrial Green Total Factor Productivity (GTFP) and environmental constraint drag indices for 25 core riverside cities along the Yangtze River Economic Belt (YREB), covering the period from 2000 to 2022. Based on the directional distance function (DDF) and the global Malmquist-Luenberger (GML) index, we employ a three-tier progressive measurement system to isolate environmental constraint effects: baseline TFP (excluding water environment variables), water-constrained GTFP^w (incorporating industrial water inputs and water pollution emissions), and water-energy dual-constrained GTFP^we (incorporating industrial electricity inputs). All three tiers share identical directional vectors, global production technology frontiers, and returns-to-scale specifications, differing only in terms of their input-output variable sets. Consequently, the differences between cumulative productivity indices exclusively reflect changes in these variable sets. Based on this, we constructed 2 constraint drag indices: the Water Environment Drag Index (WEDI), which quantifies the deviations in productivity growth trajectory attributable to water-environment constraints, and the Energy Consumption Drag Index (ECDI), which measures the drag effect associated with energy consumption beyond water constraints. The dataset comprises 575 records (25 cities×23 years) encompassing 6 core variables: GTFP and its decomposition indices (Technical Change index (TC); Efficiency Change index (EC)), cumulative productivity indices, WEDI, and ECDI. Empirical calculations reveal that basin-wide GTFP^we increased at an average annual rate of 1.80%, with technical progress as the primary driver of industrial green transformation. WEDI exhibited a distinct inverted U-shaped evolution, peaking at 0.074 in 2012 before fluctuating downward. Its 3-year rolling average became negative in the 2018–2020 window, with the “green crossing” firmly established at the turn of 2020–2021, indicating a shift in regional water governance from a “drag period” to a “dividend period”. Urban heterogeneity was prominent: during 2020–2022, WEDI turned negative in 17 cities and remained positive in 8. The dataset comprises: (1) industrial GTFP and drag index data under baseline, water-constrained, and water-energy dual-constrained specifications; (2) spatial information for the 25 cities; and (3) the input-output panel data used for the calculations. The dataset is archived in .xlsx format, and consists of one file with data size of 214 KB. This dataset supported the first author’s doctoral thesis in Science.

Keywords: industrial Green Total Factor Productivity; Yangtze River Economic Belt; Global Malmquist- Luenberger index; Water Environment Drag Index; Energy Consumption Drag Index; doctoral thesis in Science

DOI: https://doi.org/10.3974/geodp.2026.05.03

Dataset Availability Statement:

The dataset supporting this paper was published and is accessible through the Digital Journal of Global Change Data Repository at: https://doi.org/10.3974/geodb.2026.04.08.V1.

1 Introduction

Industrial Green Total Factor Productivity (GTFP) is a key indicator for evaluating the efficiency of industrial green transformation. By incorporating resource consumption and environmental pollution into the production function, the GTFP effectively distinguishes “genuine economic growth” from “environmental overdraft”^[1,2]. Since the implementation of the strategy of “stepping up conservation of the Yangtze River and stopping its over development” in 2016, core riverside cities within the Yangtze River Economic Belt (YREB) have faced increasingly stringent environmental regulations. Consequently, the scientific measurement of industrial green transformation efficiency has become a critical requirement for policy evaluation and path optimization^[3,4]. Existing GTFP measurement methodologies can be categorized primarily into the data envelopment analysis (DEA) and stochastic frontier analysis (SFA) paradigms^[5–8], and have produced several regional or sectoral GTFP datasets. Research using these approaches has established an important foundation for understanding industrial green growth.

However, from the perspective of supporting the strategic evaluation of “stepping up conservation” along the YREB, existing datasets present 3 main limitations.

First, constraint effects are insufficiently isolated. Most studies concurrently incorporate multiple environmental constraints (e.g., water pollution, air pollution, and carbon emissions) into a single composite model^[7,8]. The resulting GTFP thus captures an aggregated effect of all constraints, making it impossible to isolate the marginal contribution of individual constraints. Consequently, core policy questions remain unaddressed, such as “How large is the productivity drag exclusively caused by water pollution constraints?” or “Which cities have successfully converted water governance costs into efficiency dividends?”

Second, basin characteristics are inadequately characterized. The YREB is a representative water-sensitive basin where water pollution control is the priority task under the overarching protection strategy. However, existing datasets primarily focus on carbon emissions or air pollutants as undesirable outputs and lack measurements and index formulations tailored to water-environment constraints.

Third, the coverage of policy cycles is incomplete. Most datasets only cover post-2010 periods, failing to capture the full policy evolution cycle since 2000, which ranges from the extensive growth during the 10th Five-Year Plan, to energy conservation during the 11th and 12th Plans, and finally to comprehensive basin protection during the 13th and 14th Plans. This limitation precludes the identification of long-term trends, structural inflection points, and policy lag dynamics.

To address these challenges, the present study constructed the Dataset of industrial Green Total Factor Productivity and Water Environment Drag Index for core riverside cities in the Yangtze River Economic Belt (2000–2022)^[9]. Utilizing a directional distance function (DDF) and global Malmquist-Luenberger (GML) index framework, we employed a three-tier progressive measurement system to isolate environmental constraint effects: Tier 1 baseline TFP excludes water environment variables; Tier 2 water-constrained GTFP^w incorporates industrial water inputs and water pollution emissions; and Tier 3 water-energy dual-constrained GTFP^we further incorporates industrial energy inputs. By calculating cumulative index differencing across these tiers, the dataset provides 2 drag indices: the Water Environment Drag Index (WEDI), which quantifies the net productivity effect of water environment constraint, and the Energy Consumption Drag Index (ECDI), which measures the additional effect of energy inputs superimposed on water constraints.

The key innovations of this dataset include: isolating constraint effects via model structural differencing rather than regression controls, thereby avoiding multicollinearity issues; proposing the novel WEDI metric to quantify the independent productivity drag of water pollution and map city-level transitions from “compliance cost periods” to “regulatory dividend periods”; capturing complete policy dynamics across a 23-year longitudinal series, revealing an inverted U-shaped trajectory of WEDI and identifying the turn of 2020–2021 as a critical inflection point for green industrial transformation in the YREB.

2 Metadata of the Dataset

The metadata of the Dataset of industrial Green Total Factor Productivity and Water Environment Drag Index for core riverside cities in the Yangtze River Economic Belt (2000–2022)^[9] including the dataset name, authors, geographic region, year of the dataset, data format, data size, data files, data publisher, are summarized in Table 1.

Table 1  Metadata summary of the Dataset of industrial Green Total Factor Productivity and Water Environment Drag Index for core riverside cities in the Yangtze River Economic Belt (2000–2022)

3 Methods

3.1 Data Sources and Preprocessing

This study used a panel dataset of 25 core riverside cities along the YREB, spanning from 2000–2022, comprising 575 items (25 cities×23 years). Raw data were primarily extracted from the China City Statistical Yearbook (2001–2023) and local municipal statistical bulletins. The key indicators encompass 3 dimensions.

Basic input variables (included in Tiers 1–3): Labor input, year-end number of industrial employees (10⁴ persons); capital input, net value of industrial fixed assets (10⁸ CNY), calculated at 2000 constant prices using the perpetual inventory method; intermediate input, operating costs (10⁸ CNY), deflated to 2000 constant prices using provincial industrial producer price indices (PPIs).

Resource input variables (progressively incorporated): Water resource input, industrial water consumption (10⁴ tons), incorporated in Tier 2 (GTFP^w) and Tier 3 (GTFP^we). Energy input, industrial electricity consumption (10⁸ kWh), incorporated only in Tier 3 (GTFP^we). Industrial electricity accounts for approximately 30% of the total terminal industrial energy consumption in the YREB. Backed by abundant hydropower generation (e.g., the Three Gorges Dam and the Jinsha River cascade hydropower stations), electricity is the predominant form of clean energy in the regional industrial sector. A comparison of GTFP^w and GTFP^we reveals whether cities incurred excessive energy costs when complying with water pollution regulations.

Output variables: desirable output, gross industrial output value (10⁸ CNY), calculated at 2000 constant prices; undesirable outputs, industrial wastewater discharge (10⁴ tons) and industrial chemical oxygen demand (COD) emissions (tons), incorporated in Tiers 2 and 3.

The collected data were preprocessed, including missing value imputation: linear interpolation was used for rare missing values (<2% of total sample); outlier screening: the 3σ rule was applied to identify extreme outliers, which were replaced by linear interpolation from adjacent years (<1% of total sample); price deflation: all nominal monetary values were converted to real terms using provincial PPI series, with 2000 as the base year.

3.2 Algorithm

3.2.1 Three-Tier Progressive Measurement Framework

A three-tier progressive measurement framework was established to isolate environmental constraint effects within a unified DDF framework. All three tiers share a global production technology frontier constructed from the pooled 575 observations. Cumulative index differencing was employed to isolate the marginal effects of each constraint (Table 2).

All three tiers share identical command configurations, directional vectors, global

Table 2  Structure of the three-tier progressive measurement system

Note: Bold text highlights variables newly added or substituted at each tier. Tier 1 excludes water environment variables; however, the DDF specification requires non-empty dimension definitions. To maintain computational homogeneity across all tiers, a placeholder constant was introduced into the undesirable output set, identical across all decision-making units (DMUs) and thereby conveying zero cross-sectional variation.

production technology specifications, and constant returns to scale (CRS) assumptions, and differ exclusively in their variable sets. Tier 1 (baseline TFP) serves as a benchmark containing no environmental information. Tier 2 (GTFP^w) incorporates industrial water consumption and water pollutant emissions to measure green efficiency under water constraints. Tier 3 (GTFP^we) adds energy inputs to evaluate efficiency under dual water- energy constraints. Accordingly, moving from Tier 1 to Tier 2 adds industrial water to the inputs and replaces the placeholder constant in the undesirable outputs with 2 actual emissions, whereas moving from Tier 2 to Tier 3 is a pure addition on the input side. The marginal effect of each constraint is thus identified by differencing the cumulative indices: WEDI is the difference between the cumulative TFP and GTFP^w indices, and ECDI is the difference between the cumulative GTFP^w and GTFP^we indices. Sequential decomposition depends on the order of inclusion; water is prioritized because water pollution governance represents the primary constraint on industrial transformation along the Yangtze River.

3.2.2 Global Malmquist-Luenberger (GML) Index Model

To facilitate dynamic cross-period comparisons under desirable and undesirable outputs, this study adopted the DDF and Malmquist-Luenberger (ML) index framework proposed by Chung, et al.^[11], integrated with Oh’s^[12] global production technology approach. Computations were performed using the gtfpch command in Stata^[13].

Let the input vector be , the desirable output vector be , and the undesirable output vector be . The global production possibility set P^G (x) is defined as:

               (1)

This set strictly satisfies the weak disposability assumption of undesirable outputs (that reducing pollution requires sacrificing desirable outputs) and the null-jointness assumption (that pollution is an inevitable byproduct of production). The global frontier utilizes data from all sample periods as a unified reference, avoiding linear programming infeasibility and “spurious technical regress”.

The global directional output distance function is formulated as:

                                                                               (2)

where g=(g_y​, g_b​) is the directional vector, and  represents the maximum proportional expansion of desirable outputs and the contraction of undesirable outputs along this vector. In this study, the directional vector is specified as . With factor inputs held constant, desirable outputs expand proportionally according to their own scale, while undesirable outputs contract proportionally. This formulation inherently aligns with the core imperative of China’s industrial green transformation: to synergistically advance economic efficiency growth and pollution reduction without incurring additional factor resource consumption. The corresponding linear programming model is formulated as follows:

where  is the intensity variable, and x₀, y₀, and b₀ denote the inputs and outputs of the DMU under evaluation. Undesirable output constraints are subject to strict equality constraints to reflect weak disposability. No summation constraint () is imposed, conforming to constant returns to scale (CRS). The GML index from period t to t+1 is defined as:

                                                                                                                         (3)

GML>1 denotes GTFP growth, while GML<1 denotes decline. The GML index is multiplicatively decomposed into technical EC and TC:

                                                                                                                      (4)

                                                          (5)

                                                                                                                                                                                             (6)

where denotes the DDF benchmarked against contemporaneous production technology in period t. An index value of EC>1 indicates that actual production is approaching the contemporaneous frontier, representing an “efficiency catch-up effect” driven by managerial and allocative optimization; TC>1 signifies an outward shift of the contemporaneous technology frontier toward the global frontier, reflecting a “frontier expansion effect” driven by green technological innovation. Because the decomposition of EC and TC relies on contemporaneous technology as a reference—in which each annual reference set contains only 25 Decision- Making Units (DMUs)—its discriminatory resolution is inherently lower than that of the global technology framework benchmarked across all pooled observations. Consequently, these decomposed components are primarily suitable for evaluating longitudinal trends.

3.2.3 Construction of Derived Drag Indices

(1) Cumulative Productivity Index

Taking 2000 as the base year (=1), cumulative indices are constructed by sequential multiplication of the GML index (where GML_i,t−1 is the GML index of city i from period t−1 to t):

                                                                                                                (7)

Because all three cumulative indices are normalized to 1 in the base year, their baseline difference is identically 0. Differences between cumulative indices therefore track deviations in growth trajectories across accounting frameworks rather than static distances between frontiers.

(2) Water Environment Drag Index (WEDI)

                              WEDI_i,t=TFP Index_i,t −Index_i,t                                     (8)

WEDI quantifies the deviation in growth trajectory caused by switching the water environment variables from non-informative to informative. WEDI > 0 indicates that water pollution constraints impose a net productivity drag. WEDI < 0 suggests that water pollution costs have been absorbed or converted into green innovation dividends. The year WEDI transitions from positive to negative marks the structural shift from a “drag period” to a “dividend period”.

(3) Energy Consumption Drag Index (ECDI)

                                       ECDI_i,t=Index_i,t −Index_i,t                             (9)

ECDI quantifies the additional effect of further accounting for the energy constraint on top of the water environment constraint. ECDI>0 indicates that energy inputs impose a productivity drag in addition to water constraints. ECDI<0 indicates that energy efficiency gains yield net positive dividends, signaling synergistic water-energy governance.

It should be noted that both indices are directional diagnostics of the net effect of environmental constraints; they are not equivalent to the shadow prices of pollutants, and their magnitudes should not be interpreted on a cardinal basis.

4 Data Results

4.1 Dataset Composition

The dataset comprises 3 main data structures: (1) A panel dataset containing calculated baseline TFP, GTFP^w, GTFP^we, efficiency decomposition components (EC, TC), cumulative productivity series, WEDI, and ECDI for 25 riverside cities for the period 2000–2022; (2) Spatial attribute data that define urban geographical locations and basin regionalization (upstream, midstream, downstream); (3) Raw input-output economic and environmental panel data used during computation.

4.2 Basin-Wide Productivity Dynamics

Calculation results from 575 observations indicate that all three cumulative productivity indices showed upward trends over the 2000–2022 period. Average annual growth rates were 1.80% for baseline TFP, 1.84% for water-constrained GTFP^w, and 1.80% for water-energy dual-constrained GTFP^we. Growth measured after accounting for the water environment constraint was slightly higher than the baseline, whereas further accounting for the energy constraint brought it back to roughly the baseline level (cumulative index of 1.479 in 2022 vs. 1.480 for the baseline), the latter reflecting the offsetting of water environment dividends by energy costs. From a driver perspective (Figure 1), green technological progress (TC) served as the primary engine of industrial green transformation (efficiency catch-up dominated in the upstream cities). Over 2000–2022, cumulative GTFP^we increased by 47.9%, with both technical change and efficiency change contributing positively, aligning with the Porter Hypothesis^[14], which holds that moderate environmental regulation can stimulate technological innovation and managerial improvement in firms, shifting the production frontier outward.

Figure 1  Temporal evolution of GTFP^we in core riverside cities along the YREB (2000–2022)

Across the full sample, the mean WEDI value was 0.0198, indicating that water environment constraints exerted a slight overall drag on industrial productivity. However, its substantial standard deviation (0.1703) and wide range (1.5159) underscore pronounced urban heterogeneity. The mean ECDI value was −0.0073 with a standard deviation of 0.0552, suggesting that the additional drag effect of energy constraints was relatively modest and, on average, net negative (yielding productivity dividends from energy-efficiency improve­ments), although city-level divergence remained evident.

4.3 Spatial-Temporal Evolution of the Water Environment Drag Index (WEDI)

The WEDI exhibited a representative inverted U-shaped trajectory (Figure 2), tracing a complete cycle of industrial water pollution (from accumulation and surge to substantial abatement) along the YREB.

Notes: The left axis shows cumulative productivity indices with 2000 as the base year (=1); the right axis shows WEDI=cumulative TFP index–cumulative GTFP^w index. Positive WEDI values indicate that the water-environment constraint drags on industrial productivity, and negative values indicate that it has turned into a green dividend. The three-year moving average is centered, with truncated windows at the end points.

Figure 2  Dynamic trajectory of industrial productivity and the WEDI

Phase 1 (2000–2012): The drag phase. WEDI fluctuated slightly around 0 during 2000–2005, rose steadily from 2006 onward, and peaked at 0.0736 in 2012, reflecting an escalating productivity drag imposed by water pollution constraints. This period coincided with accelerated industrialization across riverside cities, characterized by the swiftexpansion of heavy chemical industries and underinvestment in water pollution control. Consequently, the upward trend in WEDI quantifies the economic penalty imposed by the region’s legacy “pollute first, treat later” development model.

Phase 2 (2013–2019): The transition phase. WEDI fluctuated downward from 0.0736 in 2012 to 0.0023 in 2019, approaching 0. During this phase, landmark policies, including the Action Plan for Prevention and Control of Water Pollution (“Water Ten Plan”, 2015) and the Yangtze River Conservation Strategy (2016), were successively promulgated, considerably intensifying environmental regulatory pressure. Industrial enterprises responded by deploying cleaner production technologies and closed-loop water recirculation systems, shifting water governance paradigms from end-of-pipe treatment to source reduction. This decline was non-monotonic: after dropping to 0.0259 in 2015, WEDI temporarily rebounded to 0.0529 in 2016 and 0.0521 in 2017, then resumed a downward trajectory from 2018 onward. This two-stage “rebound-then-decline” pattern reflects the temporal lag in innovation offsets, with immediate compliance costs preceding full technological adaptation.

Phase 3 (2020–2022): The dividend phase. The official implementation of the Yangtze River Protection Law (2021) signaled a deepening phase of the basin-wide protection strategy. Annual WEDI metrics contained noise from single-year data fluctuations, turning negative in 2020 (−0.027) for the first time since the peak, rebounding to near 0 in 2021 (+0.002), and then dropping back into negative territory in 2022 (−0.014). To reduce transient noise, this study employed a 3-year rolling average as the analytical benchmark: the rolling WEDI turned negative in the 2018–2020 window (−0.0002), declined to −0.008 in 2019–2021, and further fell to −0.013 in 2020–2022. The “green crossing” was firmly established at the turn of 2020–2021. A negative WEDI indicates that cumulative productivity evaluated under water constraints exceeds that measured under baseline conditions, confirming that the dividends of green technological innovation have surpassed environmental compliance costs.

4.4 Spatial-Temporal Evolution of the Energy Consumption Drag Index (ECDI)

ECDI exhibited a U-shaped trajectory (Figure 3), displaying a clear mirror-image symmetry to the inverted U-shaped WEDI curve. During 2000–2007, ECDI fluctuated between 0 and 0.0113 (peaking in 2002 and standing at 0.0071 in 2007), indicating an additional productivity drag from energy inputs. From 2008 to 2013, fueled by stringent energy-saving and emissions-reduction targets under the 11th and 12th Five-Year Plans, widespread industrial technological retrofitting drove ECDI to a trough of −0.0398, thereby converting energy-efficiency gains into net productivity dividends. From 2014 to 2018, this dividend narrowed as the ECDI increased back to −0.0050. In 2019, ECDI turned positive (+0.0034) and rose to 0.0155 in 2021 before reaching 0.0147 in 2022, indicating diminishing marginal returns from conventional energy-saving techniques and the emergence of energy penalties associated with advanced water treatment.

The inverse symmetry between the inverted U-shaped WEDI and the U-shaped ECDI— both featuring structural inflection points around 2013—highlights the systemic complexity of industrial green transformation. Water pollution control inherently demands negative entropy input, which requires free energy consumption (predominantly industrial electricity). While the fluctuating decline of WEDI (2013–2022) confirms the efficacy of basin water gov­ernance, the delayed rebound of ECDI (2019–2022) reveals that when water governance relies heavily on energy-intensive treatment technologies, the green dividends from water environment improvements will be offset by energy-consumption costs. The mirror-image relationship between WEDI and ECDI shows that green transformation cannot be pursued as a single-objective optimization but must be managed as a multi-constraint systemic equilibrium.

4.5 Urban Heterogeneity: Three Transformation Paths based on WEDI

To smooth out single-year fluctuations, the 25 core cities were categorized into 3 transformation pathways based on their 3-year average WEDI over the 2020–2022 period (Figure 4). Notably, the cross-sectional dispersion of ECDI was much lower than that of WEDI (with |ECDI| < 0.01 across most cities, representing approximately 1/7 of WEDI’s average magnitude), rendering ECDI insufficient as an independent classification dimension. Urban paths were thus strictly defined by WEDI, with ECDI serving as an auxiliary

Figure 4  Mean WEDI of core riverside cities in the YREB (2020–2022)

indicator of energy penalties.

(1) Dividend type (WEDI<−0.1; 8 cities)

Nanjing, Wuhan, Wuxi, Chongqing, Luzhou, Yibin, Shanghai, and Yangzhou all successfully converted water environment governance into net productivity dividends. With a group mean WEDI of −0.167, water constraints in these cities no longer acted as a drag but instead drove productivity growth by forcing technological innovation. Divergent transformation mechanisms can be observed among these cities: Shanghai and Wuhan leveraged forward-looking high-tech industrial layouts to drive productivity via frontier technological innovation, whereas interior cities such as Luzhou and Yibin may have benefited from latecomer advantages and the effective absorption of advanced technologies and management practices.

Notably, Wuxi (ECDI=0.0775) and Yangzhou (ECDI=0.0566) incurred high energy penalties while realizing water governance dividends, indicating their potential reliance on energy-intensive wastewater treatment technologies such as membrane filtration and deep denitrification.

(2) In-transition type (−0.1≤WEDI<0; 9 cities)

Yueyang, Anqing, Taizhou, Huangshi, Huanggang, Zhenjiang, Tongling, Wuhu, and Jingzhou all exhibited WEDI values between −0.1 and 0, indicating the dissipation of the water pollution drag. Their group mean WEDI of −0.054 suggests that these cities are undergoing a structural transition from the “compliance cost period” to the “regulatory dividend period”.

Among them, Taizhou (ECDI=0.0726) and Jingzhou (ECDI=0.2073) displayed prono­unced energy penalties. Jingzhou, in particular, recorded the highest ECDI in the entire sample (0.2073), suggesting possible overreliance on energy-intensive water treatment technologies and highlighting the need for technological pathway optimization.

(3) Drag type (WEDI≥0; 8 cities)

Ma’anshan, Ezhou, Nantong, Suzhou, Jiujiang, Yichang, Xianning, and Chizhou remain subject to a productivity drag due to water pollution constraints. Their group mean WEDI was 0.187, indicating that water governance investments have not yet fully translated into green productivity gains. Xianning (WEDI=0.5022) and Chizhou (WEDI=0.5374) experienced severe drag, with WEDI values exceeding 0.5.

For major manufacturing hubs such as Suzhou and Nantong, this situation may reflect the duality of the “scale effect”: while large-scale production enhances energy efficiency through economies of scale, the substantial volume of industrial activity generates consi­derable absolute water consumption and pollutant discharges, exerting persistent pressure on the local water environment.

5 Discussion and Conclusion

This study constructed a dataset of industrial GTFP and the water environment drag index for core riverside cities along the YREB (2000–2022). By employing a three-tier progressive measurement framework, this study advances the existing literature in 2 key ways. First, it transitions from single-constraint efficiency evaluation to mul­tilayer constraint decoupling. By progressively incorporating water pollution and energy consumption constraints, the framework quantifies the marginal productivity effects of distinct environmental factors. Second, it shifts from static productivity measurement to dynamic mechanism identification. By proposing the WEDI and the ECDI, the dataset isolates the independent drag effects of water pollution and energy consumption and maps the basin-wide transition from the “compliance cost period” to the “regulatory dividend period”.

The empirical results reveal that urban heterogeneity is the most prominent spatial feature. During 2020–2022, 17 cities turned WEDI-negative (dividend and in-transition types), while 8 cities remained positive (drag type), and the cumulative GTFP^we index differed 2.11-fold between the highest (Luzhou, 1.840) and lowest (Chizhou, 0.873) cities. Green technological progress served as the primary engine driving industrial green transformation. WEDI’s inverted U-shaped evolution (peaking in 2012 before fluctuating downward) marks the shift in regional water governance from a drag period to a dividend period. Concurrently, the U-shaped trajectory of ECDI (reaching a trough in 2013 before rising) reveals the emerging energy penalty associated with advanced water treatment.

This dataset provides robust empirical support for 3 major research domains.

(1) Environmental policy evaluation: tracing the longitudinal trajectory of WEDI enables the evaluation of the structural impacts and lag effects of landmark policies, such as the Yangtze River Conservation Strategy (2016) and the Yangtze River Protection Law (2021).

(2) Regional green pathway optimization: Spatial classification based on WEDI provides empirical evidence for tailored policy design across dividend, in-transition, and drag cities.

(3) Water-energy nexus research: The inverse symmetry between WEDI and ECDI quantifies the energy tradeoff of water governance and identifies cities with high energy costs (Wuxi, Yangzhou, Taizhou, and Jingzhou), establishing a solid empirical foundation for water-energy nexus studies.

Although the dataset was carefully designed and processed, it has several limitations. First, the dataset is currently aggregated at the overall industrial level without sub-sectoral breakdown (e.g., chemical, metallurgical, or electronic manufacturing), which limits the ability to generate fine-grained sectoral insights. Second, to characterize constraint penalties, this dataset uses index differencing within a unified DDF framework (WEDI and ECDI) to evaluate the impacts on macro-efficiency. While well-suited for analyzing macro-level longitudinal trends and spatial patterns, future research can leverage the dual relationship of the DDF to derive shadow prices for specific pollutants, thereby deepening investigations of micro-mechanisms. Third, the spatial scope covers 25 core riverside cities rather than all 110 cities at the prefecture level and above in the YREB, constraining basin-wide comprehensiveness.

This dataset supported the discovery of the first author’s doctoral thesis in Science.

Author Contributions

Liu, R. designed the overall research, collected and processed the data, performed data visualization, and drafted the initial manuscript. Wang, Z. B. provided conceptual guidance, reviewed and edited the manuscript, and secured project funding.

Acknowledgments

The authors express their sincere gratitude to Dr. Zhou, Y. K. for continuous and selfless assistance throughout the manuscript submission process. Gratitude is also extended to Dr. Zhao, L. F. for providing formatting references and valuable constructive comments on earlier drafts.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]     Chen, S. Y. Green industrial revolution in China: a perspective from the change of Environmental Total Factor Productivity (1980–2008) [J]. Economic Research Journal, 2010, 45(11): 21–34. 

[2]     Lu, L. W., Song, D. Y., Huang, C. Measurement of urban green total factor productivity in the Yangtze River Economic Belt: a case study of 108 cities in the Yangtze River Economic Belt [J]. Urban Problems, 2017(1): 61–67. 

[3]     Outline of the development plan for the Yangtze River Economic Belt [Z]. 2016. 

[4]     Huang, L., Wu, C. Q. Industrial green development efficiency and spatial driven mechanism in cities of the Yangtze River Economic Belt [J]. China Population, Resources and Environment, 2019, 29(8): 40–49. 

[5]     Chen, C. F. China’s industrial Green Total Factor Productivity and its determinants: an empirical study based on ML index and dynamic panel data model [J]. Statistical Research, 2016, 33(3): 53–62. 

[6]     Yin, L. H., Meng, X. Q., Wu, C. Q. The impact of environmental regulation on the Green Total Factor Productivity of manufacturing in the Yangtze River Economic Belt [J]. Reform, 2022(3): 101–103. 

[7]     Yuan, Y. J., Xie, R. H. Research on the effect of environmental regulation to industrial restructuring— empirical test based on provincial panel data of China [J]. China Industrial Economics, 2014(8): 57–69. 

[8]     Wu, Z. J., Wu, N. J. Research on the measurement, decomposition and influencing factors of Green Economic Efficiency in the Yangtze River Economic Belt: based on super-efficiency SBM-ML-Tobit model [J]. Urban Problems, 2021(1): 52–62, 89.

[9]     Liu, R., Wang, Z. B. Dataset of industrial Green Total Factor Productivity and Water Environment Drag Index for core riverside cities in the Yangtze River Economic Belt (2000–2022) [J/DB/OL]. Digital Journal of Global Change Data Repository, 2026. https://doi.org/10.3974/geodb.2026.04.08.V1. 

[10]   GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. https://doi.org/10.3974/dp.policy.2014.05 (Updated 2017).

[11]   Chung, Y. H., Färe, R., Grosskopf, S. Productivity and undesirable outputs: a directional distance function approach [J]. Journal of Environmental Management, 1997, 51(3): 229–240.

[12]   Oh, D. H. A global Malmquist-Luenberger productivity index [J]. Journal of Productivity Analysis, 2010, 34(3): 183–197.

[13]   Wang, D. P., Du, K. R., Zhang, N. Measuring technical efficiency and total factor productivity change with undesirable outputs in Stata [J]. The Stata Journal, 2022, 22(1): 103–124.

[14]   Porter, M. E., van der Linde, C. Toward a new conception of the environment-competitiveness relationship [J]. Journal of Economic Perspectives, 1995, 9(4): 97–118.