Public Data Trading Platforms and Sustainable Urban Productivity: Evidence from China

1. Introduction

Sustainable urban development depends not only on the expansion of economic output but also on the ability of cities to use productive resources efficiently, sustain innovation, and strengthen the institutional conditions for long-term development. As urban economies become increasingly knowledge-intensive, productivity growth depends on how effectively capital, labor, information, and technological capabilities are combined. Improving total factor productivity (TFP) is therefore an important economic dimension of urban sustainability because it reflects improvements in production efficiency beyond the accumulation of conventional inputs. However, achieving sustained productivity growth remains challenging when information is fragmented, resources are imperfectly allocated, and firms face barriers to accessing and applying external knowledge.

Data have become an increasingly important resource in addressing these challenges. Unlike many conventional inputs, data can be reused across multiple applications and shared among different organizations, creating opportunities for more informed decision-making, knowledge diffusion, and production coordination (Jones & Tonetti, 2020; Goldfarb & Tucker, 2019). Nevertheless, the availability of data does not automatically translate into productivity gains. Data may remain underutilized when ownership arrangements are unclear, transaction costs are high, data quality is difficult to assess, or potential users cannot identify suitable data products. The economic value of data consequently depends not only on their availability but also on the institutional arrangements that facilitate their exchange, governance, and productive use. Understanding how these arrangements affect urban productivity can help clarify the role of digital-market institutions in supporting innovation-driven and economically sustainable urban development.

Public data trading platforms provide one institutional arrangement through which data products can be registered, matched, exchanged, and used under more standardized procedures. By improving data discoverability and facilitating interactions between data suppliers and users, these platforms may reduce information and transaction frictions and create better conditions for combining data with capital, labor, and technological capabilities. Their potential contribution to urban productivity extends beyond individual data transactions. Better access to relevant information may support technological innovation, while changes in the demand for specialized skills and the connections among firms, universities, and research institutions may influence the spatial distribution of human and scientific resources. These processes are relevant to sustainable urban development because they concern the efficiency with which existing resources are used and the capacity of local economies to maintain innovation and productivity over time. At the same time, the establishment of a trading platform does not necessarily guarantee effective data use; its economic consequences may depend on local capabilities and complementary institutional and technological conditions.

Although existing research has examined the economics of data, data sharing, and market-oriented data allocation, several questions remain unresolved. First, much of the literature considers data availability, digitalization, or data-factor market development in broad terms. Comparatively less attention has been paid to the establishment of dedicated trading platforms as a specific institutional change. Examining these platforms makes it possible to investigate whether the creation of formal infrastructure for data exchange is associated with measurable productivity improvements. Second, existing evidence often focuses on firm-level performance, innovation, or digital economic activity. These outcomes are important, but they do not fully establish whether data-market institutions are associated with changes in aggregate urban production efficiency. Because cities bring together firms, workers, universities, research organizations, and infrastructure, the productivity consequences of data exchange may extend beyond individual firms through local knowledge flows and factor reallocation. Third, the potential pathways connecting data trading platforms with urban productivity warrant further empirical examination. Technological innovation may provide one channel, while the agglomeration of high-level human capital and the spatial concentration of scientific and technological resources may provide complementary channels.

This study addresses these gaps by examining the relationship between public data trading platform establishment and urban TFP in China. Using prefecture-level city panel data covering 2010–2024, we exploit the staggered establishment of public data trading platforms as a quasi-natural experiment and employ a staggered difference-in-differences (DID) framework. We first estimate the relationship between platform establishment and urban TFP and then investigate technological innovation, high-level human-capital agglomeration, and the spatial agglomeration of scientific and technological resources as potential transmission channels. We also conduct robustness checks and examine whether the estimated relationships vary across cities with different levels of digital infrastructure and across geographic regions.

The empirical results indicate that the establishment of public data trading platforms is associated with a statistically significant increase in urban TFP. The findings are robust to a range of alternative specifications and sample adjustments. Additional analyses show a positive relationship between platform establishment and technological innovation. The mechanism tests are also consistent with potential roles for high-level human-capital agglomeration and the spatial agglomeration of scientific and technological resources. The heterogeneity analysis reveals differences in point estimates across digital-infrastructure and regional subsamples, although these differences should not be interpreted as statistically significant cross-group effects without formal coefficient-difference tests.

This study makes three contributions. First, it extends research on data markets by shifting attention from the general availability of data to the institutional infrastructure that facilitates data exchange and use. This perspective helps clarify why the economic consequences of data depend on the arrangements governing their circulation and application. Second, by focusing on urban TFP rather than only firm-level outcomes or digital economic activity, the study provides evidence on the potential productivity implications of data-market institutions at the city level. Third, by examining technological innovation and the agglomeration of human and scientific resources, it develops an integrated account of the potential pathways connecting data trading platforms with urban productivity.

The study also contributes to discussions of sustainability by examining an institutional factor that may support the economic and innovation-related dimensions of sustainable urban development. More efficient use of productive resources, stronger innovation capacity, and improved connections among skilled workers and scientific organizations can contribute to the foundations of longer-term urban productivity. These implications should, however, be interpreted within the scope of the empirical analysis: the study estimates productivity-related outcomes and does not directly measure environmental performance, carbon emissions, or other ecological indicators. Its relevance to sustainability therefore lies primarily in the economic and institutional dimensions of urban development, rather than in demonstrating direct environmental benefits.

The remainder of the paper is organized as follows. Section 2 reviews the relevant literature. Section 3 develops the theoretical framework and research hypotheses. Section 4 describes the research design, variables, and data. Section 5 presents the baseline results, robustness checks, and heterogeneity analysis. Section 6 examines the potential transmission mechanisms. Section 7 concludes and discusses the policy implications and limitations of the study.

2. Literature Review

2.1. Data Markets, Data Allocation, and Productivity

Research on the economics of data emphasizes that data differ from conventional productive inputs because they are non-rival, reusable, and capable of generating externalities across users and applications (Jones & Tonetti, 2020; Goldfarb & Tucker, 2019). These characteristics create the potential for data to raise productivity, but they also make conventional market arrangements less directly applicable. Questions of data quality, appropriability, governance, and transaction design can affect whether data are exchanged and combined with complementary inputs effectively (Koutroumpis et al., 2020; Gregory et al., 2021).

A related literature studies how data sharing and market-oriented allocation affect resource allocation. The common mechanism is a reduction in information frictions: better access to data can lower search costs, reduce information asymmetries, improve matching, and facilitate the coordination of capital, labor, and technology (Graef & Prüfer, 2021). Data can therefore affect productivity not only as a direct input but also by improving the allocation and combination of other inputs. Evidence at the firm and regional levels is broadly consistent with this view, showing links between data-driven decision making, innovation, and productivity (Brynjolfsson et al., 2011; Xie et al., 2020).

Another strand of research focuses on innovation. Data can expand access to external knowledge, reduce information and experimentation costs, and facilitate the recombination of knowledge and technologies (Goldfarb & Tucker, 2019). Recent empirical studies likewise report positive associations between market-oriented data allocation and innovation outcomes at the firm and city levels (Chen et al., 2024; Zhao et al., 2024; Xu et al., 2025). These findings suggest that data-market institutions may influence productivity through innovation as well as through more efficient allocation of existing resources.

2.2. Public Data Trading Platforms as Institutional Infrastructure

Public data trading platforms represent a more specific institutional arrangement within the broader development of data markets. Rather than treating data availability itself as the treatment, this perspective focuses on the infrastructure that makes data discoverable, tradable, and usable. Platform-based arrangements can standardize transaction procedures, improve information disclosure and matching, and provide governance mechanisms that reduce frictions associated with data exchange (Demchenko et al., 2020; Parker et al., 2016; Cusumano et al., 2019). The resulting effects may extend beyond the direct participants in data transactions when improved data access changes firms’ production and innovation decisions.

2.3. Research Gap and Contribution

Existing empirical evidence provides support for this possibility. Studies of data trading platforms report effects on firm productivity, innovation, and regional economic outcomes, while research on public data openness points to improvements in firms’ information environments and operating efficiency (Dai et al., 2023; Wu et al., 2024). Other studies find that market-oriented data allocation can generate effects beyond individual firms through factor reallocation and spatial spillovers (Chen & Liu, 2024; Zhao et al., 2025). Taken together, this literature suggests that the economic consequences of data-market institutions can propagate through local production and innovation systems.

3. Theoretical Framework and Research Hypotheses

This section develops a theoretical framework linking public data trading platforms to urban total factor productivity (TFP) and proposes four research hypotheses.

3.1. Data Trading Platforms and Urban Total Factor Productivity

The central mechanism linking data trading platforms to urban TFP is the reduction of information and coordination frictions. Because data can be reused across organizations, their economic value depends on whether potential users can identify relevant data, evaluate their quality, and combine them with complementary inputs. A trading platform can improve these conditions by making data products more discoverable, standardizing transaction procedures, and facilitating supply–demand matching.

Improved data access can affect the allocation of conventional factors. Firms and other economic agents that obtain more relevant information may adjust investment, labor demand, and production decisions with less uncertainty. At the urban level, these adjustments can reduce resource idleness and misallocation and improve the productivity of existing capital and labor. The effect therefore does not require data to operate as an additional input in isolation; it can arise because data improve the way other inputs are combined.

Platform establishment may also strengthen connections among firms, data suppliers, technical service providers, universities, and research organizations. As the number of participants and potential applications expands, the platform can facilitate repeated interactions and broaden the range of feasible data uses. These network effects can reinforce information diffusion and resource matching, providing an additional route through which institutional improvements in data exchange may translate into higher urban TFP.

We therefore hypothesize:

H1: The establishment of public data trading platforms increases urban total factor productivity.

3.2. Data Trading Platforms and Technological Innovation

Technological innovation provides a second pathway through which data-market institutions may affect productivity (Romer, 1990; Aghion & Howitt, 1992). R&D activities depend on access to information about technologies, markets, and production processes (Cohen & Levinthal, 1990). When relevant data are fragmented across organizations or difficult to locate and evaluate, innovators face higher search and experimentation costs. A trading platform can lower these costs by improving data discoverability, transaction transparency, and matching between data supply and innovation demand.

Data circulation can also facilitate knowledge recombination. Because data are reusable, the same information can be incorporated into multiple research and production processes. Wider circulation increases opportunities for firms, universities, and research organizations to combine external information with their existing knowledge and technologies. This can expand the scope of innovation activities and improve the efficiency with which innovation resources are deployed.

Thus, the establishment of a data trading platform may increase urban innovation output by easing information constraints and strengthening knowledge flows. We therefore hypothesize:

H2: The establishment of public data trading platforms promotes urban technological innovation.

3.3. High-Level Human-Capital Agglomeration

The productivity consequences of data-market institutions may depend on the availability of workers capable of interpreting, processing, and applying data. Better access to data and a more developed digital environment can increase the demand for specialized skills, while new data-intensive applications can raise the returns to human capital. Cities in which these opportunities expand may become more attractive to highly skilled workers and firms that employ them.

Human-capital agglomeration can, in turn, affect productivity through matching economies and knowledge spillovers (Ma et al., 2019; Shi & Shen, 2021; Su et al., 2021). A larger pool of specialized workers allows firms to match skills to tasks more efficiently and can facilitate the diffusion of knowledge across organizations. If platform establishment strengthens these local agglomeration forces, part of its association with higher urban TFP should operate through high-level human capital.

We therefore hypothesize:

H3: Public data trading platform establishment increases urban TFP partly by promoting the agglomeration of high-level human capital.

3.4. Spatial Agglomeration of Scientific and Technological Resources

A second complementary agglomeration channel concerns the spatial concentration of scientific and technological resources. Data exchange can increase the value of proximity among firms, universities, research institutions, and technology-oriented service providers by creating more opportunities for collaboration and information sharing. Platform development may therefore reinforce the concentration of innovation-related resources within cities.

Spatial concentration can facilitate knowledge diffusion and reduce the costs of collaboration (Guo et al., 2022; Qin et al., 2023). When research organizations and technology-oriented firms are more closely connected, ideas and technical expertise can circulate more rapidly, and innovation resources can be allocated toward applications with greater potential. These effects can raise the efficiency of local innovation activity and ultimately contribute to higher urban TFP.

We therefore hypothesize:

H4: Public data trading platform establishment increases urban TFP partly by promoting the spatial agglomeration of scientific and technological resources.

4. Research Design

4.1. Sample Selection, Variable Definitions, and Data Sources

This study uses prefecture-level city panel data for China from 2010 to 2024 to examine the effect of public data trading platform construction on urban economic performance. Information on the establishment of public data trading platforms is compiled primarily from publicly available government information and records of relevant platform construction. Other data on urban economic conditions, industrial structure, infrastructure, and related characteristics are mainly obtained from the China City Statistical Yearbook and statistical yearbooks of individual regions.

The dependent variables are defined as follows. (1) Urban economic performance is measured by urban total factor productivity (TFP). A stochastic frontier analysis (SFA) is used to estimate the urban production function. The output measure (Y) is real GDP at the prefecture-city level; taking 2010 as the base year, nominal GDP is deflated using provincial GDP deflators. Capital input (K) is measured as the city’s capital stock, constructed from total fixed-asset investment using the perpetual inventory method with a fixed depreciation rate. Labor input (L) is measured by the total number of employed persons at the end of each year. The estimation results are reported in Table 2. After estimating the production function, the annual TFP of each city is calculated and used in the subsequent DID regressions. (2) Urban technological innovation is measured as the logarithm of the number of digital-technology-related or total invention patents granted per 10,000 people in each city.

Core identification variable: staggered DID. This study treats the opening of public data trading platforms as a “quasi-natural experiment.” The core explanatory variable is the platform-treatment indicator: if city i establishes a public data trading platform in year t, the indicator equals 1 in that year and thereafter, and 0 otherwise.

The mechanism variables include the agglomeration of high-level human capital and the spatial agglomeration of scientific and technological resources. High-level human-capital agglomeration is measured by the number of students enrolled in regular higher-education institutions, while spatial agglomeration of scientific and technological resources is measured by the density of research institutions. In addition, the study controls for urban characteristics including the level of economic development, industrial structure, degree of government intervention, degree of openness, infrastructure level, digital infrastructure level, human-capital level, and population density. All continuous variables are winsorized to reduce the influence of extreme observations on the estimates.

The variables used in the analysis are summarized in the descriptive statistics in Table 1.

4.2. Staggered Difference-in-Differences Model Specification

To estimate the relationship between platform establishment and urban TFP, we employ a staggered difference-in-differences framework with city and year fixed effects. City fixed effects absorb time-invariant differences across cities, while year fixed effects absorb common time shocks. The baseline specification is:

$$\mathit{TF}{P}_{\mathit{it}}={\alpha }_0+{\alpha }_1\mathit{DI}{D}_{\mathit{it}}+{\displaystyle \sum \gamma }\mathit{Control}{s}_{\mathit{it}}+{\mu }_i+{\delta }_t+{\epsilon }_{\mathit{it}}$$

where TFP_it denotes the total factor productivity of city i in year t; DID_it is the core policy variable, equal to 1 if city i establishes a public data trading platform in year t and thereafter, and 0 otherwise. Controls_it denotes the set of control variables; μ_i denotes city fixed effects, which absorb time-invariant city characteristics; δ_t denotes year fixed effects, which control for time-varying macroeconomic shocks; andε_it denotes the random disturbance term.

5. Empirical Results and Analysis

5.1. Baseline Regression Results

Table 3 reports the baseline regression results for the effects of public data trading platform establishment on urban TFP and technological innovation. As shown in Table 3, the estimated coefficients of the public data trading platform variable (DID) are positive across the relevant specifications, indicating that data trading platform construction is generally associated with improved urban economic performance.

First, for urban TFP, the DID coefficient in column (1), without control variables or fixed effects, is 0.289 and statistically significant at the 1% level, providing initial evidence that public data trading platform construction can improve urban production efficiency. After control variables are gradually introduced, the DID coefficient in column (2) falls to −0.056 and is statistically insignificant, possibly because unobserved regional factors and time trends interfere with the estimates when city and year fixed effects are not controlled for. After further controlling for city and year fixed effects, columns (3) and (4) report DID coefficients of 0.089 and 0.079, respectively, both significant at the 1% level. This indicates that, after controlling for city-specific characteristics and macroeconomic time trends, the establishment of public data trading platforms continues to significantly increase urban TFP.

In the two-way fixed-effects specification in column (4), the coefficient on platform establishment is 0.079 and statistically significant at the 1% level. The estimate is consistent with the hypothesis that improved data circulation and lower transaction frictions are associated with higher production efficiency.

Columns (5)–(8) examine technological innovation. In the specifications with city and year fixed effects, the coefficients on platform establishment are 2.575 and 2.542, respectively, and are statistically significant at the 1% level. The estimates are consistent with the proposed innovation channel.

This result indicates that, after excluding time-invariant city characteristics and macroeconomic fluctuations, public data trading platforms continue to significantly promote urban technological innovation. One possible explanation is that data trading platforms increase the availability of data resources, enabling firms, universities, and research institutions to obtain data needed for innovation more conveniently while also promoting data sharing and knowledge exchange among innovation actors. These mechanisms can reduce information-search and trial-and-error costs during innovation and increase urban innovation output. This finding is consistent with Hypothesis H2.

Overall, the baseline estimates are consistent with a positive association between platform establishment, urban TFP, and technological innovation. These results motivate the subsequent robustness, heterogeneity, and mechanism analyses.

5.2. Robustness Checks

To ensure that the estimated effects of public data trading platforms are not driven by specific indicator choices, sample bias, or omitted variables, a series of robustness checks is conducted from multiple dimensions, including alternative dependent variables, sample adjustments, treatment of extreme values, model specifications, and correction for selection bias. The regression results are reported collectively in Table 4.

(1) Alternative dependent variables. Measurement of core indicators inevitably depends on statistical definitions and the choice of proxy variables. To ensure that the conclusions do not depend on a single measurement approach, the two dependent variables are replaced separately: per-capita GDP replaces TFP as a measure of urban development, and patent grants replace the patent-based measure of technological innovation. The results in column (1) show that the coefficient on DID remains positive for both outcomes, with the effect on technological innovation statistically significant at the 1% level. This indicates that the enabling effect of data trading platforms is not limited to a particular measurement choice but is reflected in substantive improvements in aggregate urban output and innovation.

(2) Excluding the influence of administrative status. Direct-controlled municipalities, provincial capitals, and sub-provincial cities often have structural advantages in digital infrastructure construction, preferential allocation of policy resources, and talent attraction. These higher-administrative-level cities are therefore excluded from the sample and the regressions are re-estimated. The results in column (2) show that the DID coefficient remains statistically significant in the subsample of ordinary prefecture-level cities, indicating that the estimated productivity effect is not entirely attributable to the endowments of special-status cities.

(3) Adjusting the research sample to exclude major exogenous shocks. The COVID-19 pandemic from 2020 to 2022 generated substantial exogenous shocks to macroeconomic activity, firm production, and supply chains, and these large fluctuations may interfere with the regression results. The observations for this period are excluded and the model is re-estimated. Column (3) shows that the DID coefficients remain statistically significant, suggesting that the baseline finding is not driven by a short-term major public event.

(4) More stringent winsorization. To prevent abnormal economic observations from a small number of cities from dominating the regression results, the winsorization threshold for all continuous variables is increased from the upper and lower 1% to the upper and lower 5%. The coefficients in column (4) remain highly statistically significant, providing additional evidence that the findings are not statistical artifacts caused by extreme observations.

(5) Propensity-score-matched DID. The location and establishment of public data trading platforms are not completely random and may be influenced by factors such as urban economic fundamentals and informatization, potentially generating sample self-selection bias. Accordingly, propensity score matching (PSM) is employed. City propensity scores are calculated using a Logit model, and nearest-neighbor matching at ratios of 1:2 and 1:7 is used to identify control-group “counterfactual” observations with similar characteristics. After ensuring covariate balance between the treatment and control groups, the DID regressions are re-estimated. The results in columns (5) and (6) show that the DID coefficients remain positive and statistically significant, indicating that the positive association survives in the matched samples.

5.3. Heterogeneity Analysis

The effect of public data trading platforms on urban macroeconomic efficiency may differ structurally according to cities’ digital infrastructure and geographic location. To further examine the asymmetric features of platform effects and the theoretical propositions developed above, this section conducts heterogeneity tests along two dimensions—digital infrastructure level and geographic location—with regression results reported in Table 5 and Table 6.

Table 5 reports positive and statistically significant coefficients for the high-digital-infrastructure group and the eastern and central/western regional subsamples, while the coefficient for the low-digital-infrastructure group is positive but imprecisely estimated. The point estimate is larger in the low-infrastructure group and in the central and western regions; however, differences between subgroup coefficients should not be interpreted as statistically different without formal cross-group tests. Table 6 indicates a positive and statistically significant innovation coefficient in the high-infrastructure group and in the central and western regions, whereas the estimates for the other two groups are not statistically significant.

6. Mechanism Tests

This section examines the transmission channels through which public data trading platforms may affect urban TFP. Technological innovation was examined in Section 5; this section focuses on the two agglomeration channels—high-level human-capital agglomeration and the spatial agglomeration of scientific and technological resources.

6.1. Mechanism of High-Level Human-Capital Agglomeration

The establishment of data trading platforms may strengthen cities’ attractiveness to high-level human capital by improving data access and expanding opportunities for data-intensive activities. A larger pool of specialized workers can then improve matching and knowledge spillovers, providing a plausible link to urban productivity.

Table 7. Mechanism Test I: Transmission Path through High-Level Human-Capital Agglomeration.

	First-stage regression	Main regression with mediator
	(1)	(2)	(3)
	High-level human capital (ln_TechHR)	Technological innovation	Total factor productivity
DID	0.093***	2.392**	0.066***
	(0.030)	(1.047)	(0.018)
ln_TechHR		1.610**	0.142***
		(2.503)	(0.023)
City fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Sample size	7,400	7,400	7,389
Adjusted R²	0.879	0.999	0.952

Table 7. Mechanism Test I: Transmission Path through High-Level Human-Capital Agglomeration.

	First-stage regression	Main regression with mediator
	(1)	(2)	(3)
	High-level human capital (ln_TechHR)	Technological innovation	Total factor productivity
DID	0.093***	2.392**	0.066***
	(0.030)	(1.047)	(0.018)
ln_TechHR		1.610**	0.142***
		(2.503)	(0.023)
City fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Sample size	7,400	7,400	7,389
Adjusted R²	0.879	0.999	0.952

6.2. Mechanism of Spatial Agglomeration of Scientific and Technological Resources

Spatial agglomeration of scientific and technological resources may facilitate information exchange, knowledge diffusion, and coordination among innovation actors. If platform establishment strengthens such agglomeration, it may provide an additional route to higher urban productivity.

Table 8. Mechanism Test II: Transmission Path through Spatial Agglomeration of Scientific and Technological Resources.

	First-stage regression	Main regression with mediator
	(1)	(2)	(3)
	Spatial agglomeration	Technological innovation	Total factor productivity
DID	0.015*	2.645**	0.077***
	(0.012)	(1.022)	(0.020)
ln_Agglo		6.840*	0.184**
		(3.789)	(0.082)
City fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Sample size	7,400	7,400	7,389
Adjusted R²	0.991	0.999	0.950

Table 8. Mechanism Test II: Transmission Path through Spatial Agglomeration of Scientific and Technological Resources.

	First-stage regression	Main regression with mediator
	(1)	(2)	(3)
	Spatial agglomeration	Technological innovation	Total factor productivity
DID	0.015*	2.645**	0.077***
	(0.012)	(1.022)	(0.020)
ln_Agglo		6.840*	0.184**
		(3.789)	(0.082)
City fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Sample size	7,400	7,400	7,389
Adjusted R²	0.991	0.999	0.950

7. Conclusions and Policy Implications

7.1. Conclusions

This study examines whether the establishment of public data trading platforms is associated with improvements in urban total factor productivity (TFP), using prefecture-level city panel data from China covering 2010–2024. By exploiting the staggered establishment of these platforms and employing a staggered difference-in-differences (DID) framework, the analysis investigates the productivity implications of institutional infrastructure for data exchange. The study further examines technological innovation, high-level human-capital agglomeration, and the spatial agglomeration of scientific and technological resources as potential transmission channels.

The baseline results indicate that public data trading platform establishment is associated with a statistically significant increase in urban TFP. This finding remains positive and statistically significant across a series of robustness checks, including alternative outcome measures, the exclusion of cities with special administrative status, the exclusion of observations during the COVID-19 period, more stringent winsorization, and propensity-score-matched DID specifications. Taken together, these results provide evidence that the productivity relationship is not driven solely by a single measurement choice, a particular sample composition, or a small number of extreme observations.

The additional analyses provide insight into the possible pathways through which platform establishment relates to urban productivity. First, platform establishment is positively associated with technological innovation, consistent with the view that improved arrangements for data exchange may facilitate knowledge access and innovation-related activities. Second, the mechanism tests are consistent with potential roles for the agglomeration of high-level human capital and the spatial agglomeration of scientific and technological resources. These findings suggest that the productivity implications of data-market institutions may depend not only on the exchange of data itself but also on the availability and spatial organization of complementary skills, knowledge, and research resources. Because mechanism regressions do not by themselves establish definitive causal mediation, these channels should be interpreted as evidence consistent with the proposed explanations rather than conclusive proof of each mechanism.

The heterogeneity analysis further indicates that estimated relationships vary across cities with different levels of digital infrastructure and across geographic regions. The baseline estimates are statistically significant in the high-digital-infrastructure subsample and in the eastern and central/western regional subsamples, whereas the estimate for the low-digital-infrastructure subsample is positive but imprecisely estimated. These patterns suggest that the relationship between platform establishment and urban productivity may vary with local conditions. However, without formal tests of differences between subgroup coefficients, the observed differences in point estimates cannot establish that the treatment effects are statistically different across groups.

These findings have implications for sustainable urban development. Productivity growth, innovation, and the effective use of human and scientific resources are relevant to the economic and institutional dimensions of sustainability because they concern how cities maintain productive capacity over time. Public data trading platforms may form part of the institutional infrastructure supporting these objectives by facilitating data exchange and strengthening connections between data resources and complementary productive capabilities. Nevertheless, the present analysis does not directly evaluate environmental sustainability or establish that platform establishment reduces emissions, energy use, or ecological pressures. The sustainability implications of the results should therefore be understood primarily in terms of economic efficiency, innovation capacity, and the institutional foundations of long-term urban development.

7.2. Policy Implications

Several policy implications follow from the findings.

First, policymakers should focus on the effective functioning of public data trading platforms rather than platform establishment alone. The existence of a platform does not guarantee that data will be accessible, trustworthy, interoperable, or productively used. Clearer rules for data ownership and authorized use, standardized data descriptions, transparent transaction procedures, appropriate benefit-sharing arrangements, and effective privacy and security safeguards can help reduce uncertainty and facilitate data exchange. The objective should be to create reliable institutional conditions under which data resources can be identified, accessed, and used for legitimate productive and innovative purposes. Such arrangements may strengthen the contribution of data-market institutions to urban productivity while supporting responsible data governance.

Second, platform development should be accompanied by complementary digital infrastructure and application capabilities. The heterogeneous estimates indicate that the productivity relationship may vary across cities, and the estimated coefficient in the low-digital-infrastructure subsample is not statistically precise. This finding cautions against assuming that the establishment of a platform will produce uniform gains in every location. Investment in data connectivity, storage, interoperability, technical services, and digital skills may help cities address constraints that limit the practical use of traded data. Policy evaluation should therefore consider not only whether a platform has been established but also whether local firms and institutions have the capacity to participate in and benefit from it.

Third, policies should strengthen the links between data-market institutions and local innovation systems. The positive association between platform establishment and technological innovation, together with the mechanism results, points to the potential importance of connections among firms, universities, research institutions, skilled workers, and technology-service providers. Measures that facilitate lawful data sharing, collaborative research, knowledge exchange, and the application of research outcomes may help translate improved data access into productive uses. Human-capital development and the ability to attract and retain specialized workers should be considered complementary components of data-market development rather than separate policy concerns.

Fourth, the design and evaluation of data-market policies should reflect differences in local conditions. Cities differ in digital infrastructure, human capital, industrial composition, and scientific resources, all of which may shape their ability to use data effectively. A uniform platform-development approach may therefore be insufficient. Policymakers should assess local constraints, improve the institutional and technical conditions that are weakest, and monitor whether platforms generate practical improvements in data use, innovation, and productivity. Because the subgroup coefficient differences in this study have not been formally tested, differentiated policy should be based on local needs and further evaluation rather than on an assumption that the estimated effects are definitively larger in one group than another.

Finally, the sustainability performance of data-market institutions should be evaluated through a broader and more explicit set of indicators. The present study provides evidence on urban TFP and examines innovation and factor agglomeration, but it does not directly measure environmental outcomes or distributional consequences. Future policy assessments could combine productivity indicators with measures of energy efficiency, carbon emissions, resource consumption, data accessibility, and the distribution of benefits across firms and communities. Such evaluation would help determine whether productivity gains are accompanied by wider environmental and social improvements rather than assuming that higher productivity automatically translates into every dimension of sustainability.