1 Introduction
With continuous advancements in Internet technology, big data, cloud computing, and related fields, the processes of data collection, storage, processing, and dissemination have become increasingly efficient and accessible (Bose et al., 2023; Igual & Seguí, 2024; Lutfi et al., 2023). These technological developments have significantly facilitated the opening of government data. In recent years, government data open platforms have proliferated on a global scale (de Juana-Espinosa & Lujan-Mora, 2019; de Magalhães Santos, 2024; Lnenicka et al., 2024). These platforms offer a comprehensive array of information covering areas such as geography, energy, resource exploration, population demographics, consumption patterns, industrial activities, production materials, and product markets (Golub & Lund, 2021; Yan et al., 2023; Zeleti et al., 2016). Enterprises have the opportunity to access high-quality data at no cost, which allows them to extract relevant information. This capability enhances the ease of acquiring information and reduces the costs associated with searching for, acquiring, and verifying it (Leviäkangas & Molarius, 2020). Additionally, the availability of open government data facilitates the rapid flow of information, breaking down informational silos between enterprises, significantly reducing market information asymmetry, and providing substantial informational support for enterprise decision-making (Hardy & Maurushat, 2017; McBride et al., 2019). Concurrently, transparency and openness fostered by accessible government data can improve governance processes, decrease governmental intervention, reduce administrative barriers, and promote a favorable business environment (Attard et al., 2016; Janssen et al., 2012).
We identified two case studies that demonstrate how enterprises can enhance corporate investment through the utilization of public data. One notable example is the Shanghai Xinyidai platform developed by the Shanghai Public Credit Information Service Center. This platform consolidates diverse sources of credit information, including tax payments, social security contributions, housing provident funds, import and export activities, and utility consumption data such as water, electricity, and gas usage. It provides comprehensive credit assessment data for financial institutions. To date, the platform has registered over 4.3 million users and attracted 43 financial institutions, with a cumulative credit line approaching 920 billion yuan. Utilizing this data, banks can more accurately evaluate the creditworthiness of small, medium, and micro enterprises, enabling them to set appropriate credit limits and interest rates. This targeted allocation of funds supports enterprises with growth potential, thereby bolstering the real economy while maintaining the quality of banks’ credit assets. An additional illustration is the approach adopted by the Qingdao City Investment Group, which synthesizes public and social data to facilitate a thorough evaluation of market risks and opportunities, thereby allowing strategic planning of investment projects. Specifically, in the context of urban renewal project investments, a company employs precise enterprise information verification data to identify and select high-quality partners, thereby ensuring the successful advancement of investment initiatives. Consequently, an intriguing and significant research question pertains to the extent to which the availability of public data can facilitate corporate investments. More specifically, does the accessibility of public data enhance the scale of corporate investment and improve investment efficiency?
Corporate investment serves as a fundamental mechanism for business operations and growth, playing a crucial role in enabling enterprises to generate shareholder wealth through strategic resource allocation. It is vital for promoting future cash flow growth and wealth accumulation (Roychowdhury et al., 2019; Shen et al., 2022). From a microeconomic perspective, a company’s capacity to efficiently utilize limited resources and convert investment opportunities into tangible outcomes can substantially enhance its competitiveness and developmental potential (Jiang et al., 2018; Rajkovic, 2020). On a macroeconomic level, improving the investment efficiency of enterprises facilitates optimal resource allocation across society, thereby making a significant contribution to the high-quality development of the economy (Wang et al., 2014). The extant body of literature suggests that the financing constraints faced by enterprises constitute a significant impediment to investment efficiency (Li et al., 2021; Whited & Wu, 2006). Insufficient external financing channels may lead to a paucity of funds for enterprises, resulting in suboptimal investment levels (Chit & Rizov, 2023; Deng & Liu, 2024; Lamont et al., 2001). Concurrently, information asymmetry can give rise to moral hazard and adverse selection, potentially leading to the misallocation of funds and subsequent overinvestment (Ahmad et al., 2023; Ferguson & Lam, 2023; Yao et al., 2024). This prompts a critical inquiry: Can open government data mitigate information asymmetry, alleviate the financial constraints faced by enterprises, and consequently enhance both the scale and efficiency of investment? Specifically, this study seeks to explore the following research questions: Does the establishment of open government data platforms by local governments affect corporate investment? If so, what mechanisms drive this influence? What factors either constrain or facilitate the impact of government data openness on corporate investment? Furthermore, does the openness of government data affect the efficiency of corporate investments?
To address these research questions, this study employs the establishment of a government data open platform to construct a quasi-natural experiment. By utilizing Chinese A-share listed companies from 2007 to 2022 as the research sample, we examine the effects of open government data on corporate investment through a staggered Difference-in-Differences model. Our analysis reveals that following the implementation of government open data platforms, enterprises located in cities with access to open government data demonstrate significantly higher levels of investment compared to those situated in cities without open government data. This observation underscores the pivotal role that open government data plays in stimulating corporate investment. The robustness of our regression results remains intact after conducting a comprehensive suite of robustness checks, which includes parallel trend tests, placebo tests, propensity score matching, entropy balance matching, and alternative measures of corporate investment. The underlying mechanism suggests that open government data mitigate financing constraints, lower external financing costs, increase the scale of financing, and consequently promote corporate investment. Heterogeneity analysis indicates that this effect is particularly pronounced among state-owned enterprises, high-tech firms, companies facing elevated levels of macroeconomic uncertainty, and those located in the eastern region. Further examination reveals that open government data improves the responsiveness of enterprises to investment opportunities, thereby augmenting their investment activities.
This study contributes significantly to the existing body of literature in four key areas: (1) Enhancing the microeconomic analysis of open government data. Previous research has seldom addressed the microeconomic impacts of open government data on corporate investment. This study addresses this gap by providing causal evidence on the relationship between the openness of government data and corporate investment behavior. We specifically demonstrate that open government data not only stimulate the scale of corporate investment but also enhance investment efficiency by improving firms’ ability to identify and respond to investment opportunities. This dual-dimensional analysis enriches the literature by offering a more comprehensive understanding of how open government data influence corporate decision-making at the micro level. (2) Elucidating the underlying mechanisms of data-driven investments. Existing studies frequently overlook the transmission channels through which open government data affect corporate investment. Our research investigates financing-related mechanisms, revealing that open government data reduces external financing costs and expands the scale of corporate financing. By empirically validating these pathways, we offer a more nuanced explanation of how data openness facilitates corporate investment, thereby enriching the literature on transmission mechanisms from policy to firm behavior. (3) Expanding the scope of corporate investment determinants, most research in the context of the digital economy has concentrated on digital transformation and technological factors, often overlooking the significance of data elements. This paper pioneers an empirical analysis of open government data as a critical determinant of corporate investment. By integrating data elements into the traditional framework of investment determinants, we provide a novel perspective for future research on the digital economy’s impact on firm-level investment decisions. (4) Informing policymaking with empirical evidence, our study offers practical guidance for local governments. We present empirical evidence supporting the effectiveness of open government data policies in promoting corporate investment and enhancing investment efficiency. These findings have the potential to inform government strategies aimed at utilizing data openness to stimulate economic growth, thereby bridging the gap between academic research and policy implementation.
The remainder of this paper is organized as follows. In Section 2, we introduce the policy background, literature review, and hypothesis development. The research design is given in Section 3. Section 4 describes the results of the empirical analysis. Section 5 presents the underlying mechanism and a heterogeneity analysis. Section 6 presents the impact of open government data on corporate investment efficiency. Finally, Section 7 concludes the paper and provides implications for practice.
2 Policy background, literature review, and hypothesis development
2.1 Policy background
In recent years, governments globally have faced mounting pressure to disclose government data in order to optimize its value creation potential. In 2015, the Chinese government issued the Action Plan for Promoting the Development of Big Data and underscored the critical role of government data openness in enhancing governance capabilities and facilitating economic transformation and development. In 2022, China released a significant document entitled Overall Plan for Comprehensive Reform Pilot of Market-Oriented Allocation of Factors, which identified the enhancement of government data openness and sharing mechanisms as a critical component in the exploration of market-oriented data factor allocation. Guided by a range of policy documents, local governments are actively investigating the implementation of open government data initiatives and advancing the development of online platforms for this purpose. Notably, Shanghai and Beijing pioneered the establishment of government data open platforms in 2012. Subsequently, similar platforms were launched in cities such as Wuxi, Foshan, Zhanjiang, and Wuhan in 2014. By the end of 2023, the number of government data open platforms in China had increased to 204, representing 60.53% of the total. Cities exhibiting government data openness are predominantly located in the eastern coastal regions and southern China, while the central and western regions demonstrate comparatively slower progress. Consequently, the implementation of the government data open platform serves as a quasi-natural experiment, offering valuable insights to address endogeneity issues in economic models. Furthermore, it facilitates the evaluation of policy impacts, thereby accumulating essential experience and providing strategic guidance for the future application of government data.
2.2 Literature review
2.2.1 Literature related to the economic impact of open government data
With the progressive advancement of government data openness, scholarly attention has transitioned from evaluating the performance and determining the influencing factors of government data openness to examining its economic implications. The initial body of literature predominantly addressed the macroeconomic value of government data openness. The majority of these studies have concluded that government data openness positively contributes to local economic growth (Altayar, 2018; Attard et al., 2016; de Magalhães Santos, 2024; Hardy & Maurushat, 2017; Janssen et al., 2012; Leviäkangas & Molarius, 2020; McBride et al., 2019; Zeleti et al., 2016; Zuo et al., 2023). Nevertheless, the other literature indicates that the influence of government data openness on regional economic growth is heterogeneous across different regions (Zhao et al., 2022). Additionally, some other studies have demonstrated that government data openness positively contributes to entrepreneurial vitality (Hardy & Maurushat, 2017) and plays a significant role in advancing the reform of administrative streamlining and power delegation, as well as in optimizing the business environment (Park & Gil-Garcia, 2022).
The second category of literature investigates the microeconomic value derived from government data openness, specifically analyzing its impact on micro enterprises. The government data open platform not only diminishes the costs associated with the search, acquisition, and verification of data for market entities (Chen et al., 2023; Goldfarb & Tucker, 2019), but also facilitates enterprises in extracting valuable information from government data. This capability supports a range of economic activities that necessitate precise identification of business opportunities, such as site selection, innovation, and timely sales (Brynjolfsson & McElheran, 2016; Ghasemaghaei & Calic, 2019; Müller et al., 2018; Yan et al., 2023), and facilitates enterprises in estimating market demand to mitigate the supply-demand discrepancies arising from market misjudgments (Caputo et al., 2019; Cong et al., 2021). Additionally, the openness of government data can reduce information asymmetry between creditors and enterprises, thereby lowering the debt financing costs for enterprises (Xing et al., 2024). Several other papers have indicated that the openness of government data can enhance the quality of enterprise operations and total factor productivity by lowering operating costs, optimizing the business environment, revitalizing asset circulation, and improving the efficiency of resource allocation (Jetzek et al., 2019; Magalhaes & Roseira, 2020; Nikiforova & Lnenicka, 2021). Concurrently, the accessibility of government data is also found to facilitate the digital transformation of enterprises (Begany & Gil-Garcia, 2024).
According to the literature review above, the existing literature addresses the macroeconomic and microeconomic value of government data openness. However, few studies have explored the economic value of open government data from the perspective of corporate investment. Our study fills this gap by investigating the relationship between open government data and corporate investments.
2.2.2 Literature related to corporate investment
Corporate investment decision making is a multifaceted process shaped by a variety of factors, including internal business management practices and external macroeconomic policies. The existing literature examines the determinants of corporate investment from both micro and macro perspectives. At the micro level, factors influencing corporate investment can be further categorized into (a) management-level factors, such as managerial discretion (Shen et al., 2022), management discussion and analysis (Durnev & Mangen, 2020); (b) human factors, such as CEO overconfidence (Malmendierm & Tate, 2005; Malmendierm et al., 2011), the role of lead independent director (Rajkovic, 2020), the governance role of multiple large shareholders(Jiang et al., 2018), motivated monitoring by institutional investors (Ward et al., 2020), (c) technical factor, such as digitalization (Chen & Jiang, 2024; Xu et al., 2023), analysts’ forecast quality(Chen et al., 2017); (d) information factors, such as financial reporting quality (Biddle et al., 2009; Chen et al., 2011; Roychowdhury et al., 2019), mandatory corporate social responsibility disclosure (Liu & Tian, 2021), carbon information (Lewandowski, 2017; Phan et al., 2022; Trinks et al., 2020). At the macro level, the existing literature has focused on the impact of digital finance (Tang & Geng, 2024), micro-environment uncertainty (Li et al., 2021), political uncertainty (Brandon & Youndgsuk, 2012), and policy uncertainty (Gulen & Ion, 2016; Wang et al., 2014) on corporate investment.
As mentioned above, in the investigation of the factors influencing corporate investment, the existing literature has predominantly focused on micro and macro factors. However, few studies have addressed the role of data elements in corporate investment. In the context of the digital economy, where data elements play a crucial role in enterprise decision-making, it has become imperative to analyze the impact of government data openness on corporate decision-making processes. Therefore, we empirically examine the impact of open government data on corporate investment in this paper. It will not only augment the existing body of literature concerning the economic value of open government data but also further explore the factors influencing corporate investment.
2.3 Hypothesis development
2.3.1 Open government data and corporate investment
Corporate investment projects typically require substantial capital. Relying exclusively on internal funding may hinder the ability to sustain the normal operation of these projects, thereby necessitating the need for external financing. However, a prevalent issue of information asymmetry exists between external investors, such as financial institutions and enterprises, resulting in significant financing constraints for investment initiatives undertaken by enterprises (Roychowdhury et al., 2019; Yang et al., 2023). The implementation of open government data platforms, which provide large-scale public data free of charge, has significantly enhanced the information environment and facilitated the dismantling of inter-regional information barriers (Zeleti et al., 2016; Zhao et al., 2022). This initiative alleviates the temporal and spatial limitations on communication between enterprises and various stakeholders, thereby accelerating the transmission and sharing of information and ensuring more timely disclosure of corporate information (Xing et al., 2024). Corporate stakeholders leverage government data open platforms to enhance their capacity to access, process, and utilize information, thus mitigating the traditional issues of information asymmetry (Ahmad et al., 2023; Bellucci et al., 2023). The accessibility of government data facilitates a more comprehensive and precise understanding of the various dimensions of a city, including its economic, social, demographic, and market characteristics. This availability allows enterprises to mitigate the risks associated with inadequate or erroneous information during the investment decision-making process, thereby enhancing their confidence in investment and promoting increased investment activity (Bello, 2024; Gao et al., 2023). Simultaneously, the enhancement of government data transparency has augmented market transparency, allowing financing entities to gain a deeper understanding of industry development trends and competitive dynamics (Xing et al., 2024; Wang et al., 2024). Additionally, this also facilitates a more accurate assessment of the positioning and competitive advantages of financed enterprises, thereby bolstering investment confidence and mitigating risk perceptions. Consequently, it attracts a greater number of financing parties and assists enterprises in transforming information flows into external capital flows and alleviating corporate financing constraints (Zeleti et al., 2016; Zhao et al., 2022). Based on these considerations, the following hypothesis is proposed:
H1: Open government data can enhance corporate investment.
2.3.2 The influencing mechanism of open government data on corporate investment
Previous research has demonstrated that financial constraints negatively impact corporate investments (Albuquerque, 2024; Howes, 2023; Jumah et al., 2023; Xu et al., 2023). Conversely, the availability of government data has been shown to mitigate information asymmetry and reduce financial constraints (Xing et al., 2024). Government agencies possess primary information regarding enterprises’ operational conditions through their regulatory activities. Utilizing the credit information-sharing mechanism, financial institutions can construct precise profiles of enterprise credit. This enables a more comprehensive understanding of the operational conditions and credit levels of enterprises, thereby reducing their financing costs and thresholds through innovative financial service models (deHaan et al., 2023; Luo et al., 2023). Additionally, it provides diversified financing options for enterprises and supports credit financing for enterprises in the real economy, particularly small and medium-sized enterprises. Given adequate funding, enterprises are more inclined to augment investment, particularly in innovative projects (Shaturaev, 2024). Additionally, the availability of government data facilitates enterprises in estimating market demand, mitigates the supply-demand imbalances resulting from market misjudgments, enhances managerial decision-making capabilities, strengthens risk assessment and predictive abilities, and improves enterprise profitability, thus relieving financing constraints.
Furthermore, to delve deeper into the mechanisms through which open government data impact corporate investment, this study deconstructs the concept of financing constraints. According to the theory of financing constraints, enterprises experiencing such constraints typically encounter higher financing costs and reduced financing availability (Kaplan & Zingales, 1997). This indicates that financing constraints encompass two dimensions: the scale of financing that can be obtained and the cost of financing that needs to be borne. In addition, the transparency of government data should be leveraged to improve financial outcomes, specifically by optimizing debt financing mechanisms. Consequently, this study hypothesizes that the openness of government data may influence corporate investments by affecting both the scale and cost of debt financing. This implies a logical progression wherein open government data results in either an increase in the scale of financing or a reduction in financing costs, thereby impacting corporate investment.
The availability of open government data can significantly enhance the scale of debt financing, lower associated costs, and improve enterprises’ access to investment funds via debt financing channels. This impact is manifested through three primary mechanisms. First, financial institutions leveraging big data to create digital inclusive finance can make more efficient and accurate assessments of enterprise operations, thereby reducing financing costs and barriers for enterprises through innovative financial solutions (Tomar & Periyasamy, 2024). In 2019, Shanghai introduced inclusive finance applications on an open data platform, consolidating data from eight governmental departments and making over 300 data items accessible to enhance bank credit mechanisms (Zhu & Guo, 2024). Therefore, the openness of government data has augmented the supply of credit funds in the financial system.
Second, classical financial theory posits that information asymmetry is intricately linked to corporate borrowing costs (Bellucci et al., 2023; Kong, 2023). The transparency of government data facilitates external funding providers in obtaining timely and precise insights into the operational dynamics of enterprises, thereby enabling them to effectively identify potential investment targets, mitigate information frictions, and decrease information costs. This bolsters the confidence of external funding providers and diminishes perceived risks, thus promoting increased investment activity (Wang et al., 2024).
Third, the increased transparency and availability of government data facilitated the entry of foreign banks. This promotes business competition among financial institutions and forces domestic banks to adjust their credit strategies. On the one hand, banks are anticipated to adopt a more proactive approach in information mining and screening, thereby broadening their business scope and customer base, reducing credit approval requirements, and enhancing approval efficiency (Lyu et al., 2023). On the other hand, in response to the competitive pressures induced by the entry of foreign banks, domestic financial institutions are likely to moderately lower their loan interest rates to maintain or expand their market share, consequently reducing the borrowing costs for enterprises (Jigeer & Koroleva, 2023; Yao & Song, 2023).
Therefore, we posit the following hypotheses:
H2: Open government data facilitate corporate investment by mitigating the overall financing constraints faced by enterprises.
H2a: Open government data facilitate corporate investment by expanding the scale of corporate financing.
H2b: Open government data facilitate corporate investment by lowering financing costs for enterprises.
3 Research design
3.1 Data source and sample selection
Given that Shanghai pioneered the establishment of government data open platforms in 2012, this study establishes a time frame from 2007 to 2022 in order to examine the differences between the treatment and control groups. This timeframe included observations from the five years preceding 2012. The initial sample comprises all A-share listed companies in the Shanghai and Shenzhen stock markets of China. The data processing for this study involves several steps: excluding financial and insurance firms; omitting companies designated as ST, *ST, and PT; removing companies that are temporarily suspended from trading; and excluding enterprises with missing key data. To eliminate the influence of outliers, we applied a 1% truncation to all continuous variables. We obtain data on firm-year structure, comprising 31,284 observations detailed in Table 1. The launch dates for government data open platforms were primarily gathered from the China Local Government Data Open Report, published by the Digital and Mobile Governance Laboratory at Fudan University. The initiation date of the government data open platform was verified and supplemented with information obtained from local government data open platforms and Baidu searches. The remaining data were sourced from the China Stock Market & Accounting Research (CSMAR) database.
Table 1. Sample description. |
| Panel A: Sample selection process. | |||||
|---|---|---|---|---|---|
| Initial observations of all Chinese A-share listed firms from 2007 to 2022 | 49,194 | ||||
| Less: observations pertaining to the financial industry | (1,071) | ||||
| Less: observations designated as ST, *ST, and PT firms | (1,933) | ||||
| Less: observations with firms temporarily suspended | (5) | ||||
| Less: observations with missing information on control variables | (14,901) | ||||
| Available observations | 31,284 | ||||
| Unique firms | 3,355 | ||||
| Panel B: Distribution by year. | |||||
| Year | Freq. | Percent(%) | Open | Cum.(%) | |
| 0 | 1 | ||||
| 2007 | 1,004 | 0 | 1,004 | 0 | 0.00 |
| 2008 | 1,065 | 3.4 | 1,065 | 0 | 0.00 |
| 2009 | 1,171 | 3.74 | 1,171 | 0 | 0.00 |
| 2010 | 1,233 | 3.94 | 1,233 | 0 | 0.00 |
| 2011 | 1,349 | 4.31 | 1,349 | 0 | 0.00 |
| 2012 | 1,674 | 5.35 | 1,369 | 305 | 18.22 |
| 2013 | 1,958 | 6.26 | 1,603 | 355 | 18.13 |
| 2014 | 2,037 | 6.51 | 1,653 | 384 | 18.85 |
| 2015 | 1,950 | 6.23 | 1,498 | 452 | 23.18 |
| 2016 | 2,012 | 6.43 | 1,303 | 709 | 35.24 |
| 2017 | 2,200 | 7.03 | 1,335 | 865 | 39.32 |
| 2018 | 2,342 | 7.49 | 1,038 | 1,304 | 55.68 |
| 2019 | 2,839 | 9.07 | 939 | 1,900 | 66.92 |
| 2020 | 2,809 | 8.98 | 589 | 2,220 | 79.03 |
| 2021 | 2,953 | 9.44 | 454 | 2,499 | 84.63 |
| 2022 | 2,688 | 8.59 | 340 | 2,348 | 87.35 |
| Total | 31,284 | 100 | 17,943 | 13,341 | 42.64 |
Note: This table presents the sample selection process and sample distribution. Panel A presents the sample selection process. Panel B presents the distribution of the sample by year. |
3.2 Variable definition
3.2.1 Dependent variable
The dependent variable is corporate investment, denoted by Invest. In alignment with previous research (Cheng et al., 2024; Crawford & Markarian, 2024; Deng et al., 2020; Li et al., 2021), we normalized corporate investment by total assets. Specifically, we quantify corporate investment expenditure as the ratio of net capital expenditure (cash expended on the acquisition and construction of fixed assets, intangible assets, and other long-term assets minus the net cash received from the disposal of fixed assets, intangible assets, and other long-term assets) to total assets.
3.2.2 Independent variable
The data open platform established by local governments serves as a crucial mechanism for facilitating the sharing of government data. The introduction of this platform offers an optimal quasi-natural experimental setting for examining the economic value of open government data. Consequently, the independent variable in this context is open government data, denoted by the variable open. We employ a staggered Difference-in-Differences model for empirical analysis, and then the variable open is operationalized as a dummy variable. It takes the value of 1 if the enterprise’s office is situated in a city that has implemented a government data open platform and 0 if it is not.
3.2.3 Control variables
By referring to Deng et al. (2020) and Duchin et al. (2010), we identify the following control variables: firm size (Size), financial leverage (Lev), return on total assets (Roa), cash flow from operations (Cfo), firm age (Age), sales growth (Growth), equity balance (Balance), board independence (Independent), board size (Board), and CEO duality (Dual). Additionally, to mitigate the impact of macroeconomic development levels, we incorporate macroeconomic control variables at the city level, specifically including economic development level (lngdp) and government revenue (lnfinance). The variable definitions are listed in Table 2.
Table 2. Variable definitions. |
| Property of variable | Variable name | Variable symbol | Variable description |
|---|---|---|---|
| Dependent variable | corporate investment | Invest | (cash expended on the acquisition and construction of fixed assets, intangible assets, and other long-term assets - the net cash received from the disposal of fixed assets, intangible assets, and other long-term assets) / by the total assets at the beginning of the period. |
| Independent variable | Open government data | Open | Whether the city where the enterprise is located has a government data open platform |
| Control variables | Firm size | Size | The natural logarithm of total assets |
| Debt-to-asset ratio | Lev | Total liabilities/total assets | |
| Return on total assets | Roa | Net profit/total assets | |
| Operating cash flow | Cfo | Net cash flow from operating activities/total assets | |
| Age of the company | Age | The natural logarithm of listing year +1 | |
| Company growth | Growth | Revenue growth rate | |
| Equity balance degree | Balance | Shareholding ratio of the 2nd to 5th largest shareholders/shareholding ratio of the first largest shareholders | |
| independent director ratio | Independent | Number of independent directors/Number of directors | |
| Board size | Board | the natural logarithm of (number of directors + 1) | |
| Two duties in one | Dual | Whether the chairman and the If the It takes the value of 1 if chairman and the general manager are the same person, and 0 if it is not. | |
| economic development level | Lngdp | Natural logarithm of per capita GDP in cities | |
| Government revenue | lnfinance | Natural logarithm of local fiscal budget revenue |
3.3 Model design
This study leverages the implementation of government data open platforms by local governments as a quasi-natural experiment. We explore the impact of government data openness on corporate investment by employing a staggered Difference-in-Differences model. The treatment group comprises enterprises situated in cities where the government data openness platform is operational, whereas the control group consists of enterprises located in cities without such platforms. The specific baseline difference-in-differences model is formulated as follows:
$\text { Invest }_{i t}=\alpha_{0}+\alpha_{1} \text { Open }_{i t}+\alpha_{2} \text { control }_{i t}+\lambda_{i}+\mu_{t}+\varepsilon_{i t}$
In this context, i and t denote the enterprise and year, respectively. Investit signifies the level of corporate investment, while Openit represents the implementation of government data open platforms. The coefficient α1 captures the DID estimate for the causal effect of the open government data on corporate investment. Controlit encapsulates financial characteristics, corporate governance characteristics, and regional economic factors that may influence corporate investment decisions. λi and μt denote firm fixed effects and year fixed effects, respectively. εit represents the regression residual. Given that there are multiple observations per firm and industry, we use standard errors clustered at the firm and industry double clustering level.
3.4 Descriptive statistics
The descriptive statistical results for all variables are presented in Table 3. Specifically, the mean value of the investment variable (Invest) is 0.0541, with a median of 0.0343 and a maximum value of 0.362, suggesting considerable variability in investment levels among the sampled companies. The mean value for the openness variable (Open) is 0.426, indicating that approximately 42.6% of the sample was influenced by government data openness, thereby reflecting the strong representativeness of the sample. The median value for Open is 0, implying that nearly half of the sample entities adopted the government data open platform at a relatively later stage.
Table 3. Descriptive statistics. |
| Variable | N | Mean | P25 | Median | P75 | Min | SD | Max |
|---|---|---|---|---|---|---|---|---|
| Invest | 31,284 | 0.0541 | 0.0123 | 0.0343 | 0.0734 | -0.0259 | 0.0634 | 0.362 |
| Open | 31,284 | 0.426 | 0 | 0 | 1 | 0 | 0.495 | 1 |
| Size | 31,284 | 22.27 | 21.36 | 22.09 | 23.03 | 19.17 | 1.312 | 26.11 |
| Lev | 31,284 | 0.452 | 0.294 | 0.450 | 0.605 | 0.0529 | 0.203 | 0.899 |
| Roe | 31,284 | 0.0481 | 0.0234 | 0.0650 | 0.113 | -0.870 | 0.158 | 0.364 |
| Cfo | 31,284 | 0.0463 | 0.0070 | 0.0453 | 0.0869 | -0.177 | 0.0713 | 0.260 |
| Age | 31,284 | 2.335 | 1.792 | 2.398 | 2.890 | 1.099 | 0.638 | 3.367 |
| Growth | 31,284 | 0.165 | -0.0343 | 0.103 | 0.266 | -0.585 | 0.414 | 2.499 |
| Balance | 31,284 | 0.697 | 0.224 | 0.524 | 1.011 | 0.0269 | 0.601 | 2.784 |
| Independ | 31,284 | 0.374 | 0.333 | 0.333 | 0.429 | 0.300 | 0.0532 | 0.571 |
| Board | 31,284 | 2.251 | 2.079 | 2.303 | 2.303 | 1.792 | 0.181 | 2.773 |
| Dual | 31,284 | 0.246 | 0 | 0 | 0 | 0 | 0.431 | 1 |
| lngdp | 31,284 | 8.878 | 8.049 | 8.969 | 9.839 | 6.005 | 1.174 | 10.67 |
| lnfinance | 31,284 | 16.16 | 15.23 | 16.07 | 17.21 | 13.42 | 1.249 | 18.25 |
4 Empirical results and analysis
4.1 Benchmark regression analysis
Table 4 presents the regression analysis results for the relationship between the open government data and corporate investment. Column (1) presents the results of the univariate regression analysis, incorporating only year fixed effects and firm fixed effects. The regression coefficient for Open is 0.0028, which is significantly positive at the 5% level. This finding suggests that the investment levels of firms increase significantly following the implementation of government data open platforms. These results provide preliminary evidence to support the positive impact of government data openness on corporate investment levels. This also implies that the theoretical proposition that government data openness can help alleviate financing constraints is substantiated.
Table 4. The impact of the establishment of government data open platforms on corporate investment. |
| Variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Invest | Invest | Invest | Invest | |
| Open | 0.0028** | 0.0032** | 0.0032*** | 0.0030*** |
| (2.02) | (2.43) | (3.41) | (3.19) | |
| Size | 0.0126*** | 0.0126*** | 0.0128*** | |
| (10.84) | (6.30) | (5.82) | ||
| Lev | 0.0028 | 0.0036 | 0.0036 | |
| (0.58) | (0.71) | (0.63) | ||
| Roe | 0.0231*** | 0.0236*** | 0.0237*** | |
| (9.39) | (5.68) | (5.68) | ||
| Cfo | -0.0027 | -0.0035 | -0.0033 | |
| (-0.49) | (-0.41) | (-0.39) | ||
| Age | -0.0412*** | -0.0412*** | -0.0420*** | |
| (-15.06) | (-9.34) | (-9.39) | ||
| Growth | 0.0214*** | 0.0214*** | 0.0212*** | |
| (15.60) | (6.16) | (6.17) | ||
| Balance | 0.0018 | 0.0017 | 0.0020 | |
| (1.16) | (1.43) | (1.54) | ||
| Independ | -0.0189 | -0.0186* | -0.0179 | |
| (-1.48) | (-1.86) | (-1.65) | ||
| Board | -0.0029 | -0.0032 | -0.0028 | |
| (-0.58) | (-0.94) | (-0.76) | ||
| Dual | 0.0029** | 0.0029* | 0.0029* | |
| (2.16) | (1.83) | (1.76) | ||
| lngdp | 0.0070 | 0.0069 | 0.0090 | |
| (1.51) | (1.53) | (1.49) | ||
| lnfinance | -0.0076* | -0.0072 | -0.0021 | |
| (-1.95) | (-1.50) | (-0.47) | ||
| Cons | 0.0529*** | -0.0681 | -0.0720 | -0.1761*** |
| (89.95) | (-1.47) | (-1.55) | (-4.06) | |
| Firm Fixed Effects | YES | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES | YES |
| Industry Fixed Effects | NO | NO | YES | YES |
| City Fixed Effects | NO | NO | NO | YES |
| Observations | 31,284 | 31,284 | 31,284 | 31,284 |
| Adj. R2 | 0.3715 | 0.4196 | 0.4207 | 0.4215 |
Note: This table presents the regression results for the impact of open government data on corporate investment. Column (1) presents the results of the univariate regression analysis. Column (2) reports the regression results with all control variables added. Column (3) incorporates industry trend controls, in addition to the comprehensive variable regression analysis. Column (4) incorporates regional trend controls in addition to a comprehensive variable regression analysis. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
Taking into account the potential influence of financial conditions, corporate governance characteristics, and regional economic features on the investment levels of enterprises, this study incorporates the corresponding control variables into the univariate regression model previously discussed. The results of the comprehensive multivariate regression analysis (Column (2)) indicate that the regression coefficient for Open is 0.0032, which is statistically significant at the 5% level. Therefore, the hypothesis that government data openness can significantly enhance the investment levels of enterprises is well-supported by empirical evidence. Economically, following the establishment of the government data open platform, the average investment level of enterprises experienced an increase of 5.92%. This figure is derived from the calculation of economic significance, which is defined as the regression coefficient of the independent variable divided by the average value of the dependent variable (i.e. 0.0032/0.0541 * 100% = 5.92%).
Furthermore, variations in the developmental trajectories of different industries, such as those induced by the industry lifecycle, may exacerbate the inherent disparities in investment levels across groups. Consequently, this study incorporates industry trend controls in addition to comprehensive variable controls. The results presented in Column (3) indicate that the regression coefficient for Open is significantly positive at the 1% level, suggesting that the primary regression conclusion is not attributable to differences in industry trends.
In addition, to mitigate the potential confounding effects of local policy variations on the research conclusions, this study incorporated regional trend controls, in addition to a comprehensive variable regression analysis. The results presented in Column (4) demonstrate that the regression coefficient between Open and Invest is significantly positive at the 1% level, suggesting that regional trend differences do not account for the primary regression findings. These results robustly indicate that the observed positive impact of open government data on corporate investment is not attributable to industry-specific or regional trend variations. This finding strongly supports the reliability of the benchmark regression conclusion. Therefore, hypothesis 1 is empirically supported.
Furthermore, the dataset may exhibit an intrinsic group structure that can influence the distribution of the response variable. To address the potential heterogeneity across companies, regions, and industries, we implemented a clustering approach prior to the regression analysis. By clustering at the firm, regional, and industry levels, we partitioned the dataset into distinct subsets, enabling the construction of more streamlined models for each subset. This approach mitigates the risk of overfitting and enhances the model’s generalization and predictive capabilities. The results are detailed in the subsequent Table 5: Column (1) corresponds to clustering at the firm and time levels, Column (2) at the city level, column (3) at the city and time levels, Column (4) at the city and industry levels, and Column (5) at the city, firm, and industry levels. Our findings indicate that the research results demonstrate robustness across various clustering scenarios.
Table 5. The regression results after clustering at the firm, regional, and industry levels. |
| Variable | (1) | (2) | (3) | (4) | (5) |
|---|---|---|---|---|---|
| Invest | Invest | Invest | Invest | Invest | |
| Open | 0.0030** | 0.0030** | 0.0030** | 0.0030*** | 0.0030*** |
| (2.14) | (2.43) | (2.23) | (3.25) | (3.44) | |
| Size | 0.0128*** | 0.0128*** | 0.0128*** | 0.0128*** | 0.0128*** |
| (8.83) | (10.74) | (8.69) | (5.56) | (5.89) | |
| Lev | 0.0036 | 0.0036 | 0.0036 | 0.0036 | 0.0036 |
| (0.63) | (0.67) | (0.59) | (0.57) | (0.60) | |
| Roe | 0.0237*** | 0.0237*** | 0.0237*** | 0.0237*** | 0.0237*** |
| (7.36) | (7.55) | (6.20) | (4.81) | (5.09) | |
| Cfo | -0.0033 | -0.0033 | -0.0033 | -0.0033 | -0.0033 |
| (-0.58) | (-0.54) | (-0.52) | (-0.35) | (-0.37) | |
| Age | -0.0420*** | -0.0420*** | -0.0420*** | -0.0420*** | -0.0420*** |
| (-11.17) | (-12.90) | (-9.92) | (-8.70) | (-9.20) | |
| Growth | 0.0212*** | 0.0212*** | 0.0212*** | 0.0212*** | 0.0212*** |
| (13.75) | (10.21) | (9.51) | (5.51) | (5.83) | |
| Balance | 0.0020 | 0.0020 | 0.0020 | 0.0020 | 0.0020* |
| (1.22) | (1.41) | (1.29) | (1.67) | (1.78) | |
| Independ | -0.0179 | -0.0179 | -0.0179 | -0.0179 | -0.0179 |
| (-1.36) | (-1.29) | (-1.25) | (-1.41) | (-1.49) | |
| Board | -0.0028 | -0.0028 | -0.0028 | -0.0028 | -0.0028 |
| (-0.53) | (-0.47) | (-0.47) | (-0.60) | (-0.63) | |
| Dual | 0.0029* | 0.0029** | 0.0029** | 0.0029 | 0.0029* |
| (2.12) | (2.35) | (2.33) | (1.72) | (1.82) | |
| Lngdp | 0.0090 | 0.0090* | 0.0090 | 0.0090 | 0.0090 |
| (1.68) | (1.70) | (1.70) | (1.42) | (1.50) | |
| lnfinance | -0.0021 | -0.0021 | -0.0021 | -0.0021 | -0.0021 |
| (-0.49) | (-0.62) | (-0.59) | (-0.51) | (-0.53) | |
| Cons | -0.1761** | -0.1761*** | -0.1761*** | -0.1761*** | -0.1761*** |
| (-2.84) | (-3.27) | (-3.08) | (-5.06) | (-5.35) | |
| Firm Fixed Effects | Yes | Yes | Yes | Yes | Yes |
| Year Fixed Effects | Yes | Yes | Yes | Yes | Yes |
| Industry Fixed Effects | Yes | Yes | Yes | Yes | Yes |
| City Fixed Effects | Yes | Yes | Yes | Yes | Yes |
| Firm and year cluster | Yes | ||||
| City cluster | Yes | ||||
| City and year cluster | Yes | ||||
| City and industry cluster | Yes | ||||
| City,firm and industry cluster | Yes | ||||
| Observations | 31,284 | 31,284 | 31,284 | 31,284 | 31,284 |
| Adj. R2 | 0.4215 | 0.4215 | 0.4215 | 0.4215 | 0.4215 |
Note: This table presents the regression results after clustering at the firm, regional, and industrial levels. Column (1) pertains to clustering at both the firm and temporal levels, Column (2) pertains to clustering at the city level, Column (3) pertains to clustering at both the city and temporal levels, Column (4) pertains to clustering at both the city and industrial levels, and Column (5) pertains to clustering at the city, firm, and industrial levels. |
4.2 Robustness checks
4.2.1 Parallel trend test
In accordance with the parallel trend assumption, it is posited that in the absence of the government data openness initiative, the investment trajectories of enterprises in both open and closed regions would exhibit similar patterns of change. However, the government’s data openness initiative has not been rescinded before, precluding direct verification through counterfactual analysis. To address this limitation, we employ a dynamic difference-in-differences model to examine the inter-group variation trends in corporate investment levels prior to government data openness. The analysis aims to determine whether there is a significant difference in the trends of investment level changes between enterprises located in cities with government data openness and those in cities without such openness before policy implementation. If no significant difference is observed, it indicates that the parallel trend assumption is satisfied. The dynamic difference-in-differences model is specified as follows:
$\text { Invest }_{i t}=\beta_{0}+\beta_{k} \sum_{k=-13}^{10} \text { Event }_{i t}^{k}+\gamma \text { Controls }_{i t}+\delta_{i}+\theta_{t}+\mu_{c}+\vartheta_{j}+\varepsilon_{i t}.$
In this context, represents the temporal distance variable, indicating the number of years relative to the year of government data openness. When k equals 0, it denotes the year of data openness. Conversely, a negative value of k signifies years prior to openness, whereas a positive value indicates years following openness. The definitions of the remaining variables align with those in the basic difference-in-differences model. Specifically, Event0 denotes the year the government data open platform was introduced, EventN1 corresponds to the year preceding the platform’s launch, and EventP1 refers to the year following the platform’s launch. This nomenclature extends similarly to other variables. To mitigate multicollinearity among temporal distance variables, we designated the year preceding the initiation of government data openness as the reference period, thereby excluding EventN1 from the regression analysis. We performed parallel trend analyses by adjusting for control variables and fixed effects, utilizing data processed through the propensity score matching method. The results presented in Table 6 demonstrate that the coefficients for EventN2 through EventN13 are not statistically significant across all the columns. The analysis indicates that there is no statistically significant difference in investment level trends between enterprises located in cities with government data disclosure and those in cities without such disclosure prior to the implementation of data openness. This observation fulfills the parallel trend assumption inherent in the Difference-in-Differences (DID) model. As a result, the findings of this study are robust against the potential concerns of omitted variable bias and reverse causality, thereby supporting the notion that government data openness can effectively promote corporate investment.
Table 6. Parallel trend test. |
| Variable | (1) | (2) | (3) |
|---|---|---|---|
| Invest | Invest | Invest | |
| EventN13 | -0.0090 | -0.0083 | -0.0090 |
| (-1.23) | (-1.46) | (-1.63) | |
| EventN12 | -0.0063 | -0.0053 | -0.0063 |
| (-1.18) | (-1.07) | (-1.31) | |
| EventN11 | -0.0044 | -0.0041 | -0.0044 |
| (-0.67) | (-1.04) | (-1.17) | |
| EventN10 | -0.0059 | -0.0055 | -0.0059 |
| (-0.95) | (-1.45) | (-1.62) | |
| EventN9 | -0.0013 | -0.0010 | -0.0013 |
| (-0.28) | (-0.26) | (-0.35) | |
| EventN8 | -0.0002 | 0.0003 | -0.0002 |
| (-0.05) | (0.11) | (-0.06) | |
| EventN7 | -0.0032 | -0.0026 | -0.0032 |
| (-0.95) | (-0.89) | (-1.14) | |
| EventN6 | -0.0038 | -0.0035 | -0.0038 |
| (-1.61) | (-1.46) | (-1.64) | |
| EventN5 | -0.0022 | -0.0020 | -0.0022 |
| (-0.95) | (-0.97) | (-1.11) | |
| EventN4 | -0.0021 | -0.0021 | -0.0021 |
| (-0.89) | (-1.06) | (-1.09) | |
| EventN3 | 0.0013 | 0.0015 | 0.0013 |
| (0.71) | (0.80) | (0.68) | |
| EventN2 | 0.0014 | 0.0013 | 0.0014 |
| (0.84) | (0.92) | (0.98) | |
| Event0 | 0.0028** | 0.0026** | 0.0028** |
| (2.43) | (2.35) | (2.54) | |
| EventP1 | 0.0031** | 0.0030** | 0.0031** |
| (2.25) | (2.02) | (2.19) | |
| EventP2 | 0.0044** | 0.0043** | 0.0044** |
| (2.37) | (2.36) | (2.53) | |
| EventP3 | 0.0044* | 0.0040* | 0.0044* |
| (1.89) | (1.67) | (1.87) | |
| EventP4 | 0.0052* | 0.0050 | 0.0052* |
| (1.86) | (1.64) | (1.70) | |
| EventP5 | 0.0061* | 0.0052* | 0.0061** |
| (1.86) | (1.93) | (2.26) | |
| EventP6 | 0.0066* | 0.0059* | 0.0066** |
| (1.81) | (1.83) | (2.05) | |
| EventP7 | 0.0068 | 0.0067** | 0.0068** |
| (1.46) | (2.28) | (2.43) | |
| EventP8 | 0.0087 | 0.0086** | 0.0087** |
| (1.59) | (2.38) | (2.34) | |
| EventP9 | 0.0060 | 0.0056 | 0.0060 |
| (1.09) | (1.16) | (1.33) | |
| EventP10 | 0.0081 | 0.0077* | 0.0081** |
| (1.44) | (1.85) | (2.10) | |
| Size | 0.0070*** | 0.0069*** | 0.0070*** |
| (7.29) | (11.40) | (10.89) | |
| Lev | 0.0123** | -0.0014 | 0.0123*** |
| (2.13) | (-0.32) | (2.87) | |
| Roe | 0.0226*** | 0.0167*** | 0.0226*** |
| (3.95) | (5.39) | (7.25) | |
| Cfo | 0.0640*** | 0.0956*** | 0.0640*** |
| (7.24) | (13.35) | (9.90) | |
| Age | -0.0226*** | -0.0239*** | -0.0226*** |
| (-8.03) | (-18.81) | (-17.42) | |
| Growth | 0.0249*** | 0.0244*** | 0.0249*** |
| (5.36) | (12.62) | (13.66) | |
| Balance | 0.0012 | 0.0007 | 0.0012 |
| (1.38) | (0.66) | (1.14) | |
| Independ | -0.0035 | -0.0063 | -0.0035 |
| (-0.28) | (-0.45) | (-0.27) | |
| Board | -0.0036 | 0.0029 | -0.0036 |
| (-0.74) | (0.61) | (-0.87) | |
| Dual | 0.0037** | 0.0042*** | 0.0037*** |
| (2.23) | (2.99) | (2.97) | |
| lngdp | 0.0072 | 0.0055 | 0.0072 |
| (1.05) | (0.96) | (1.30) | |
| lnfinance | -0.0012 | -0.0004 | -0.0012 |
| (-0.25) | (-0.09) | (-0.30) | |
| Cons | -0.1022** | -0.1014* | -0.1022* |
| (-2.71) | (-1.90) | (-1.96) | |
| Firm Fixed Effects | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES |
| Industry Fixed Effects | NO | YES | YES |
| City Fixed Effects | NO | NO | YES |
| Observations | 31,064 | 31,064 | 31,064 |
| Adj. R2 | 0.2191 | 0.1778 | 0.2191 |
Note: This table presents the regression results of the parallel trend test. Column (1) presents the results without considering industry fixed effects and city fixed effects. Column (2) presents the results with consideration of industry fixed effects and without consideration of city fixed effects. Column (3) presents the results with consideration of industry fixed effects and city fixed effects. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
Furthermore, we provide Figure 1, which depicts the findings within a 95% confidence interval. The data indicate that prior to the implementation of government data openness, there was no statistically significant difference in the investment behavior trends between affected and unaffected enterprises. However, subsequent to the introduction of open government data, a significant increase in the investment behavior of affected enterprises was observed, thereby providing additional support for the parallel trend hypothesis.
Figure 1. Parallel trend test. This figure plots the parallel trend test based on the results from Column 3 of Table 5. |
While Figure 1 initially confirms the parallel trend assumption, it is important to acknowledge that a growing body of recent research has highlighted potential biases in parameter estimation when using the traditional two-way fixed effects model, particularly due to heterogeneous treatment effects in the multi-period difference-in-differences framework (Goodman-Bacon, 2021). To address this issue and mitigate the influence of heterogeneous treatment effects on parameter estimation, this study adopts the methodology proposed by Borusyak et al. (2021) and utilizes the imputation estimator of heterogeneous treatment effects to reassess the dynamic impacts of introducing local public data open platforms on enterprise investment. As illustrated in Figure 2, the estimated coefficient of the primary explanatory variable exhibits a change in magnitude; however, it remains statistically insignificant at the 10% level prior to the implementation of government data open platforms and becomes significantly positive after the implementation of government data open platforms. Compared to Figure 1, no fundamental alterations are evident. This suggests that even after accounting for heterogeneous treatment effects, the parallel trend assumption of the difference-in-differences model remains valid.
Figure 2. Parallel trend test with consideration of heterogeneous treatment effects. |
4.2.2 Placebo test
While the model specifications for benchmark regression and parallel trend testing have partially mitigated concerns regarding endogeneity, there may still exist additional stochastic factors that influence corporate investment and contribute to estimation bias. To further isolate the impact of unobservable variables on corporate investment levels and to ensure that the findings of this study are attributable to the implementation of government data open platforms rather than extraneous factors, this study conducts additional verification through placebo testing. Given the challenges associated with directly measuring numerous factors at the city and temporal levels, a placebo test was implemented by randomly selecting cities and years with available open government data. Considering that 204 cities in the final sample had established open government data platforms, this study initially randomized the sample regions and designated the top 204 regions as pseudo-open government data regions. Subsequently, we randomly sorted the city-year samples and selected the first city-year sample at each city level. We merge the randomly selected cities and city-year samples, retaining only those city-year samples that correspond to the pseudo-government data open cities. Construct a pseudo-government data openness variable (_b[open]) and conduct a regression analysis. Finally, we document the regression coefficients for the _b[open] variable and the kernel density distribution of P-values. The placebo test conducted in this study was repeated 500 times. If the conclusions drawn from the benchmark regression are not influenced by omitted variables at the city and time levels, the estimated coefficient kernel density of the regression should approximate a normal distribution with a mean of zero. Figure 3 presents the kernel density distribution of the estimation coefficients and P-values for the explanatory variables in the placebo test. The kernel density of the estimated coefficients from the sample regression exhibited an approximately normal distribution, with a mean of zero. The majority of the estimated coefficients possess P-values exceeding the 10% statistical significance threshold of 0.1. Furthermore, all estimated coefficients fall below the benchmark regression coefficient of 0.003, as indicated by the red solid line on the right. These findings suggest that the assertion that government data openness enhances enterprise investment levels is not attributable to random factors, thereby indicating that the conclusions of this study have successfully passed the placebo test.
Figure 3. Placebo test. |
4.2.3 Propensity score matching method and entropy balance matching method
To mitigate the interference of selection bias, this study employed two matching methodologies to match each treatment group observation with a control group, followed by regression analysis on the matched samples. Initially, propensity score matching was utilized to address inter-group feature differences within samples. The nearest neighbor matching, radius matching and kernel matching are employed. Subsequently, a staggered Difference-in-Differences model is applied to estimate the impact of government data openness on corporate investment levels. The regression results after matching are shown in columns (1) to (3) of Table 7. It is evident that, irrespective of the nearest neighbor matching, radius matching and kernel matching method employed, the regression coefficient of government data openness remains significantly positive. This finding suggests that the benchmark regression results are not attributable to selection bias and demonstrate robustness.
Table 7. Results of propensity score matching method and entropy balance matching method. |
| Variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Nearest neighbor matching (1:1) | Radius matching | Kernel matching | Entropy balance matching | |
| Open | 0.0032** | 0.0032*** | 0.0032*** | 0.0020* |
| (2.57) | (3.45) | (3.40) | (1.75) | |
| Size | 0.0129*** | 0.0125*** | 0.0125*** | 0.0108*** |
| (6.28) | (6.54) | (6.54) | (4.11) | |
| Lev | 0.0003 | 0.0031 | 0.0031 | 0.0088 |
| (0.07) | (0.63) | (0.63) | (1.54) | |
| Roe | 0.0200*** | 0.0234*** | 0.0234*** | 0.0226*** |
| (4.57) | (5.55) | (5.58) | (5.47) | |
| Cfo | -0.0027 | -0.0026 | -0.0026 | -0.0019 |
| (-0.35) | (-0.31) | (-0.31) | (-0.18) | |
| Age | -0.0434*** | -0.0412*** | -0.0411*** | -0.0355*** |
| (-10.42) | (-9.40) | (-9.35) | (-6.80) | |
| Growth | 0.0209*** | 0.0210*** | 0.0210*** | 0.0173*** |
| (5.99) | (6.06) | (6.06) | (3.77) | |
| Balance | 0.0010 | 0.0016 | 0.0016 | 0.0033 |
| (0.65) | (1.40) | (1.38) | (1.38) | |
| Independ | -0.0151 | -0.0177* | -0.0181* | -0.0031 |
| (-1.50) | (-1.86) | (-1.90) | (-0.20) | |
| Board | -0.0021 | -0.0031 | -0.0032 | -0.0064 |
| (-0.54) | (-0.87) | (-0.89) | (-1.43) | |
| Dual | 0.0021 | 0.0030* | 0.0029* | 0.0013 |
| (1.13) | (1.97) | (1.95) | (1.12) | |
| lngdp | 0.0072 | 0.0075 | 0.0075 | -0.0068 |
| (1.40) | (1.52) | (1.52) | (-0.87) | |
| lnfinance | -0.0083 | -0.0079 | -0.0078 | -0.0004 |
| (-1.65) | (-1.53) | (-1.54) | (-0.05) | |
| Cons | -0.0607 | -0.0654 | -0.0651 | -0.0344 |
| (-1.43) | (-1.36) | (-1.36) | (-0.58) | |
| Firm Fixed Effects | YES | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES | YES |
| Industry Fixed Effects | YES | YES | YES | YES |
| City Fixed Effects | YES | YES | YES | YES |
| Observations | 26,661 | 31,021 | 31,063 | 31,284 |
| Adj. R2 | 0.4299 | 0.4222 | 0.4224 | 0.5301 |
Note: This table presents the regression results of propensity score matching method and entropy balance matching method. Column (1) presents the results of nearest neighbor matching with 1:1. Column (2) presents the results of radius matching. Column (3) presents the results of kernel matching. Column (4) presents the results of entropy balance matching. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
Given that the propensity score matching (PSM) method is susceptible to sample loss, we adopt Hainmueller’s (2012) entropy balance method to mitigate the limitations associated with PSM and ensure the randomness and exogeneity of model testing. Consequently, the regression results obtained from the entropy balance matching method are presented in Column (4) of Table 7. The results indicate that the regression coefficient for government data openness remains significantly positive, thereby reaffirming the validity of the benchmark regression results.
4.2.4 Alternative measures of corporate investment
Given the potential sensitivity of our conclusions to the measurement of corporate investments, we introduce three novel metrics: Invest2, Invest3, and Invest4. Specifically, Invest2 is defined as the ratio of cash expended by a company for the acquisition and construction of fixed assets, intangible assets, and other long-term assets to total assets at the end of the period. Invest3 is defined as the ratio of cash expended by a company for the acquisition and construction of fixed assets, intangible assets, and other long-term assets to total assets at the beginning of the period. Invest4 is defined as the quotient obtained by dividing the difference between the cash expenditures for the acquisition and construction of fixed assets, intangible assets, and other long-term assets, and the net cash inflows from the disposal of these assets by the total assets at the end of the reporting period. Table 8 presents the regression results using Invest2, Invest3, and Invest4 as the dependent variables. The findings indicate that irrespective of whether Invest2, Invest3, or Invest4 is employed to gauge the level of corporate investment, the coefficient for government data openness remains significantly positive. This suggests that government data openness consistently enhances corporate investment across different measures, thereby affirming the robustness and reliability of the research conclusions presented in this paper.
Table 8. The regression results with alternative measures of corporate investment. |
| Variable | (1) | (2) | (3) |
|---|---|---|---|
| Invest2 | Invest3 | Invest4 | |
| Open | 0.0017** | 0.0027*** | 0.0019*** |
| (2.63) | (2.87) | (3.04) | |
| Size | 0.0054*** | 0.0115*** | 0.0068*** |
| (3.93) | (5.76) | (4.44) | |
| Lev | 0.0047 | 0.0068 | 0.0014 |
| (1.07) | (1.07) | (0.36) | |
| Roe | 0.0156*** | 0.0257*** | 0.0146*** |
| (4.83) | (5.67) | (4.92) | |
| Cfo | 0.0056 | -0.0088 | 0.0102* |
| (0.96) | (-0.99) | (1.89) | |
| Age | -0.0305*** | -0.0398*** | -0.0328*** |
| (-8.82) | (-8.57) | (-9.93) | |
| Growth | 0.0019*** | 0.0217*** | 0.0021*** |
| (3.98) | (6.19) | (4.17) | |
| Balance | 0.0006 | 0.0022* | 0.0005 |
| (0.63) | (1.83) | (0.47) | |
| Independ | -0.0048 | -0.0146 | -0.0078 |
| (-0.65) | (-1.30) | (-1.12) | |
| Board | -0.0001 | -0.0032 | 0.0002 |
| (-0.03) | (-0.83) | (0.10) | |
| Dual | 0.0016 | 0.0027 | 0.0018 |
| (1.47) | (1.59) | (1.69) | |
| lngdp | 0.0058** | 0.0061 | 0.0054** |
| (2.30) | (1.39) | (2.21) | |
| lnfinance | 0.0032 | 0.0030 | 0.0026 |
| (1.15) | (0.73) | (0.98) | |
| Cons | -0.1074** | -0.2074*** | -0.1225** |
| (-2.14) | (-2.88) | (-2.33) | |
| Firm Fixed Effects | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES |
| Industry Fixed Effects | YES | YES | YES |
| City Fixed Effects | YES | YES | YES |
| Observations | 31,284 | 31,284 | 31,284 |
| Adj. R2 | 0.4553 | 0.4264 | 0.4456 |
Note: This table presents the regression results with alternative measures of corporate investment. Column (1) displays the results for the scenario in which corporate investment is quantified as the ratio of cash expended by a company for the acquisition and construction of fixed assets, intangible assets, and other long-term assets to total assets at the end of the period. Column (2) displays the results for the scenario in which corporate investment is quantified as the ratio of cash expended by a company for the acquisition and construction of fixed assets, intangible assets, and other long-term assets to total assets at the beginning of the period. Column (3) displays the results for the scenario in which corporate investment is quantified as the quotient obtained by dividing the difference between the cash expenditures for the acquisition and construction of fixed assets, intangible assets, and other long-term assets, and the net cash inflows from the disposal of these assets by the total assets at the end of the reporting period. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
4.2.5 The impact of provincial government data open platform
Under the hierarchical administrative management system in China, provincial governments exert a substantial influence on the development of cities within their jurisdiction. The initiation timelines of municipal government data open platforms often coincide with those of provincial government data open platforms. Consequently, provincial-level platforms may impact the online activities of municipal platforms within their jurisdiction, potentially resulting in biased identification and regression outcomes. To evaluate the robustness of the results, first, the fixed effects of provinces were controlled for based on the benchmark model. This approach accounts for the impact of provincial government data open platforms and other relevant factors on regression outcomes. Column (1) of Table 9 presents the findings of this analysis. The results indicate that even after controlling for the influence of provincial platforms, the investment promotion effect of government data openness remains significantly evident. Second, to assess the impact of the municipal government’s open data platform on a sample of prefecture-level cities, municipalities directly under the central government were excluded from the analysis. Column (2) of Table 9 presents the results of this analysis, which indicate that the regression coefficient for government data openness remains significantly positive. Finally, it is important to consider that the establishment timelines of provincial public data platforms overlap with those of prefecture-level city public data platforms, potentially leading to biased identification and estimation results. To validate the robustness of the findings, this study employs the establishment of a provincial public data platform as a quasi-natural experiment and performs a difference-in-differences analysis akin to a benchmark regression. The corresponding results are presented in Column (3) of Table 9. The regression coefficient indicates that the effect of government data openness remained significantly positive, thereby affirming the robustness of the study’s conclusions.
Table 9. Impact of provincial government data open platform. |
| Variable | (1) | (2) | (3) |
|---|---|---|---|
| Invest | Invest | Invest | |
| Open | 0.0030** | 0.0027* | 0.0043*** |
| (2.30) | (1.82) | (4.38) | |
| Size | 0.0128*** | 0.0140*** | 0.0128*** |
| (10.91) | (5.66) | (5.83) | |
| Lev | 0.0036 | 0.0020 | 0.0034 |
| (0.74) | (0.33) | (0.60) | |
| Roe | 0.0237*** | 0.0259*** | 0.0237*** |
| (9.80) | (6.27) | (5.64) | |
| Cfo | -0.0033 | -0.0065 | -0.0034 |
| (-0.60) | (-0.74) | (-0.40) | |
| Age | -0.0420*** | -0.0432*** | -0.0421*** |
| (-15.30) | (-8.49) | (-9.41) | |
| Growth | 0.0212*** | 0.0241*** | 0.0212*** |
| (15.49) | (8.56) | (6.17) | |
| Balance | 0.0020 | 0.0016 | 0.0021 |
| (1.32) | (1.08) | (1.56) | |
| Independ | -0.0179 | -0.0258* | -0.0177 |
| (-1.39) | (-1.95) | (-1.63) | |
| Board | -0.0028 | -0.0051 | -0.0029 |
| (-0.54) | (-1.26) | (-0.80) | |
| Dual | 0.0029** | 0.0033* | 0.0029* |
| (2.15) | (1.83) | (1.76) | |
| lngdp | 0.0090* | 0.0098 | 0.0084 |
| (1.69) | (1.51) | (1.41) | |
| lnfinance | -0.0021 | -0.0020 | -0.0023 |
| (-0.50) | (-0.39) | (-0.51) | |
| Cons | -0.1761*** | -0.1959*** | -0.1685*** |
| (-2.91) | (-4.63) | (-3.82) | |
| Firm Fixed Effects | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES |
| Industry Fixed Effects | YES | YES | YES |
| City Fixed Effects | YES | YES | NO |
| Province Fixed Effects | YES | NO | YES |
| Observations | 31,284 | 24,416 | 31,284 |
| Adj. R2 | 0.4215 | 0.4119 | 0.4216 |
Note: This table presents the regression results on Impact of provincial government data open platform. Column (1) presents the findings with consideration of province fixed effects. Column (2) presents the findings for the scenario wherein municipalities directly under the central government were excluded from the analysis. Column (3) presents the findings for the scenario wherein the establishment of a provincial public data platform was taken as a quasi-natural experiment. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
4.2.6 The impact of other contemporaneous policies
To isolate the effects of other concurrent policies, we examined three representative initiatives: the Broadband China Pilot Policy, the National Big Data Comprehensive Experimental Zone Policy, and the Free Trade Zone Policy. The detailed outcomes are presented in Table 10. Column (1) illustrates the results after accounting for the Broadband China Pilot Policy. Column (2) demonstrates the findings after controlling for the National Big Data Comprehensive Experimental Zone Policy. Column (3) depicts the outcomes after considering the Free Trade Zone Policy. Column (4) presents the results after simultaneously controlling for all three policies. The findings suggest that the results remain robust even after accounting for the influence of these various policies.
Table 10. The impact of other contemporaneous policies. |
| Variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Invest | Invest | Invest | Invest | |
| Open | 0.0028** | 0.0031*** | 0.0030*** | 0.0026** |
| (2.83) | (3.24) | (3.04) | (2.70) | |
| Size | 0.0127*** | 0.0127*** | 0.0127*** | 0.0127*** |
| (5.71) | (5.66) | (5.65) | (5.66) | |
| Lev | 0.0038 | 0.0038 | 0.0034 | 0.0034 |
| (0.65) | (0.63) | (0.57) | (0.57) | |
| Roe | 0.0237*** | 0.0236*** | 0.0235*** | 0.0237*** |
| (5.76) | (5.73) | (5.68) | (5.74) | |
| Cfo | -0.0024 | -0.0025 | -0.0024 | -0.0023 |
| (-0.28) | (-0.31) | (-0.29) | (-0.27) | |
| Age | -0.0419*** | -0.0420*** | -0.0415*** | -0.0415*** |
| (-9.12) | (-9.11) | (-8.88) | (-8.85) | |
| Growth | 0.0211*** | 0.0211*** | 0.0211*** | 0.0211*** |
| (6.07) | (6.10) | (6.11) | (6.09) | |
| Balance | 0.0019 | 0.0019 | 0.0019 | 0.0019 |
| (1.39) | (1.41) | (1.39) | (1.38) | |
| Independ | -0.0188 | -0.0191 | -0.0193 | -0.0192 |
| (-1.60) | (-1.64) | (-1.67) | (-1.64) | |
| Board | -0.0026 | -0.0027 | -0.0029 | -0.0027 |
| (-0.66) | (-0.72) | (-0.76) | (-0.70) | |
| Dual | 0.0029* | 0.0030* | 0.0029* | 0.0029* |
| (1.85) | (1.89) | (1.85) | (1.82) | |
| lngdp | 0.0092 | 0.0095 | 0.0093 | 0.0091 |
| (1.16) | (1.18) | (1.19) | (1.16) | |
| lnfinance | -0.0019 | -0.0018 | -0.0012 | -0.0014 |
| (-0.40) | (-0.37) | (-0.26) | (-0.30) | |
| Kdzgdid | 0.0039** | 0.0036** | ||
| (2.30) | (2.15) | |||
| Datadid | 0.0005 | 0.0002 | ||
| (0.29) | (0.11) | |||
| Zmqdid | 0.0048** | 0.0043* | ||
| (2.34) | (2.02) | |||
| Cons | -0.1810*** | -0.1826*** | -0.1950*** | -0.1922*** |
| (-4.34) | (-4.34) | (-4.54) | (-4.46) | |
| Firm Fixed Effects | Yes | Yes | Yes | Yes |
| Year Fixed Effects | Yes | Yes | Yes | Yes |
| Industry Fixed Effects | Yes | Yes | Yes | Yes |
| City Fixed Effects | Yes | Yes | Yes | Yes |
| Observations | 31,098 | 31,098 | 31,098 | 31,098 |
| Adj. R2 | 0.4228 | 0.4226 | 0.4228 | 0.4229 |
Note: This table presents the impact of other contemporaneous policies. Column (1) illustrates the results after accounting for the Broadband China Pilot Policy. Column (2) demonstrates the findings after controlling for the National Big Data Comprehensive Experimental Zone Policy. Column (3) depicts the outcomes after considering the Free Trade Zone Policy. Column (4) presents the results after simultaneously controlling for all three policies. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
5 The underlying mechanism and heterogeneity analysis
In Section 4, we demonstrate that open government data significantly contribute to fostering corporate investment. The regression results maintain their significance following an extensive series of robustness checks. This raises several pertinent questions. First, what mechanisms underlie this influence? Secondly, what factors either constrain or facilitate the effect of government data openness on corporate investment? Sections 5.1 and 5.2 address the first question, while Section 5.3 addresses the second.
5.1 The impact of open government data on overall financing constraint
Financing constraints constitute a significant determinant influencing corporate investment activities. As previously discussed, the increased transparency and availability of government data have augmented the supply of funds within the domestic financial system, thereby alleviating the external financing challenges encountered by corporate investments. This study employs investment-cash flow sensitivity as a metric to assess the overall level of financing constraints experienced by enterprises and formulates the following Model (3) to empirically test research Hypothesis 2.
$\text { Invest }_{i t}=\alpha_{0}+\alpha_{1} \text { Open }_{i t}+\alpha_{2} \text { Open }_{i t} * \text { Cfo }_{i t}+\alpha_{3} \text { control }_{i t}+\lambda_{i}+\mu_{t}+\varepsilon_{i t}$
Investit, Openit, and Cfoit are defined in the Table 2. The test results are presented in Table 11. An analysis of these results reveals that the regression coefficient for the interaction term Open×Cfo is significantly negative at the 1% significance level. This finding suggests that the implementation of government data openness has effectively reduced the investment-cash flow sensitivity of enterprises. Consequently, when using investment-cash flow sensitivity as a proxy for the overall level of financing constraints faced by enterprises, it can be inferred that the openness of government data significantly alleviates these financing constraints. This conclusion suggests that open government data substantially mitigate the financial constraints faced by enterprises, thereby promoting an increase in corporate investment.
Table 11. The impact of open government data on overall financing constraint. |
| Variable | Invest |
|---|---|
| Open | 0.0046*** |
| (4.25) | |
| Open×Cfo | -0.0330*** |
| (-4.12) | |
| Size | 0.0129*** |
| (5.81) | |
| Lev | 0.0033 |
| (0.57) | |
| Roe | 0.0239*** |
| (5.75) | |
| Cfo | 0.0091 |
| (0.95) | |
| Age | -0.0419*** |
| (-9.39) | |
| Growth | 0.0212*** |
| (6.17) | |
| Balance | 0.0020 |
| (1.54) | |
| Independ | -0.0175 |
| (-1.62) | |
| Board | -0.0027 |
| (-0.74) | |
| Dual | 0.0028* |
| (1.75) | |
| lngdp | 0.0090 |
| (1.49) | |
| lnfinance | -0.0021 |
| (-0.47) | |
| Cons | -0.1790*** |
| (-4.13) | |
| Firm Fixed Effects | YES |
| Year Fixed Effects | YES |
| Industry Fixed Effects | YES |
| City Fixed Effects | YES |
| Observations | 31,284 |
| Adj. R2 | 0.4217 |
Note: This table presents the regression results of the impact of open government data on overall financing constraints. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
To rigorously assess the robustness of the impact of government data openness on overall financing constraints, we draw upon the methodologies of Lamont et al. (2001) and Whited and Wu (2006), employing the KZ index as a metric for the degree of financing constraints experienced by enterprises. Higher values of the KZ index indicate greater financing constraints, whereas lower values suggest reduced constraints. The variables representing government data openness (Open), financing constraints (KZ), and their interaction terms were incorporated into Model (4) for empirical testing.
$\text { Invest }_{i t}=\alpha_{0}+\alpha_{1} \text { Open }_{i t}+\alpha_{2} K Z_{i t}+\alpha_{3} \text { Open }_{i t} * K Z_{i t}+\alpha_{3} \text { control }_{i t}+\lambda_{i}+\mu_{t}+\varepsilon_{i t}$
The regression results in Table 12 indicate that the coefficient of the interaction term is significantly positive. This finding suggests that for enterprises facing greater financing constraints, government data openness exerts a more pronounced positive effect on corporate investment. Consequently, this outcome further substantiates the assertion that government data openness can enhance corporate investment levels by mitigating financing constraints. Thus, Hypothesis 2 was empirically validated.
Table 12. The impact of open government data on overall financing constraint with KZ index. |
| Variable | Invest |
|---|---|
| Open×KZ | 0.0018*** |
| (5.11) | |
| KZ | -0.0049*** |
| (-8.29) | |
| Open | 0.0004 |
| (0.38) | |
| Size | 0.0106*** |
| (5.01) | |
| Lev | 0.0298*** |
| (4.73) | |
| Roe | 0.0222*** |
| (5.50) | |
| Cfo | -0.0505*** |
| (-4.02) | |
| Age | -0.0367*** |
| (-7.63) | |
| Growth | 0.0204*** |
| (6.24) | |
| Balance | 0.0021 |
| (1.59) | |
| Independ | -0.0190* |
| (-1.84) | |
| Board | -0.0032 |
| (-0.87) | |
| Dual | 0.0028 |
| (1.63) | |
| lngdp | 0.0102 |
| (1.72) | |
| lnfinance | -0.0023 |
| (-0.53) | |
| Cons | -0.1463*** |
| (-3.40) | |
| Firm Fixed Effects | YES |
| Year Fixed Effects | YES |
| Industry Fixed Effects | YES |
| City Fixed Effects | YES |
| Observations | 30,775 |
| Adj. R2 | 0.4269 |
Note: This table presents the regression results of the impact of open government data on overall financing constraints by employing the KZ index as a metric for the degree of financing constraints. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
5.2 The impact of open government data on financing cost and financing scale
This study further disaggregates corporate financing constraints into issues related to financing scale and financing costs and investigates the influence of government data openness on both corporate financing scale and financing costs, thereby affecting corporate investment. Drawing on the framework established by Baron and Kenny (1986), a mediation effect test model is developed to analyze the pathways through which increased financing scale and reduced financing costs impact corporate investment and is specified as follows:
$\text { Mediators }_{i t}=\beta_{0}+\beta_{1} \text { Open }_{i t}+\beta_{2} \text { control }_{i t}+\lambda_{i}+\mu_{t}+\varepsilon_{i t}$
$\text { Invest }_{i t}=\gamma_{0}+\gamma_{1} \text { Open }_{i t}+\gamma_{2} \text { Mediators }_{i t}+\gamma_{3} \text { control }_{i t}+\lambda_{i}+\mu_{t}+\varepsilon_{i t}$
Wherein, Mediatorsit is the mechanism variable. Model (5) tests the correlation between the mechanism variable and independent variable, and Model (6) tests the total effect of the mechanism variable and independent variable on corporate investment.
5.2.1 The impact of open government data on financing cost
To investigate whether the reduction in external financing costs serves as a mechanism through which government data openness influences corporate investment, we operationalize the mechanism variable financing cost (OutFin) as the ratio of interest payable to total liabilities. This hypothesis was tested using models (5) and (6). The regression results for the mechanism variable relative to the independent variable are presented in Column (1) of Table 13. The coefficient of the independent variable (Open) is significantly negative at the 5% level, suggesting that government data openness mitigates financial resource mismatch and reduces external financing costs. In addition, the regression analysis results examining the relationship between corporate investment levels and the mechanism variables are presented in Column (2) of Table 13. The coefficient of the independent variable (Open) is significantly positive at the 1% level, whereas the coefficient of corporate financing cost (OutFin) is significantly negative at the 1% level. This indicates that a reduction in financing costs can enhance corporate investment levels. Consequently, it can be inferred that the openness of government data can be substantiated by its effect of lowering enterprise financing costs and subsequently improving investment levels. Thus, Hypothesis 2a was empirically validated.
Table 13. The impact of open government data on financing cost and financing scale. |
| Variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| OutFin | Invest | Loan | Invest | |
| Open | -0.0006** | 0.0027*** | 0.0025* | 0.0029*** |
| (-2.34) | (2.88) | (1.93) | (3.05) | |
| OutFin | -0.6592*** | |||
| (-7.88) | ||||
| Loan | 0.0490*** | |||
| (5.11) | ||||
| Size | -0.0002 | 0.0126*** | 0.0101*** | 0.0123*** |
| (-1.31) | (5.56) | (11.29) | (5.81) | |
| Lev | 0.0063*** | 0.0078 | 0.4269*** | -0.0174** |
| (8.71) | (1.27) | (112.49) | (-2.62) | |
| Roe | -0.0052*** | 0.0203*** | 0.0043 | 0.0235*** |
| (-10.34) | (4.91) | (1.51) | (5.68) | |
| Cfo | 0.0076*** | 0.0016 | -0.1947*** | 0.0063 |
| (9.04) | (0.18) | (-31.19) | (0.79) | |
| Age | 0.0002 | -0.0418*** | 0.0064*** | -0.0423*** |
| (0.45) | (-9.69) | (2.76) | (-9.36) | |
| Growth | -0.0008*** | 0.0206*** | -0.0048*** | 0.0214*** |
| (-6.78) | (6.21) | (-5.09) | (5.99) | |
| Balance | 0.0009*** | 0.0025* | -0.0054*** | 0.0023* |
| (3.60) | (1.97) | (-4.41) | (1.76) | |
| Independ | -0.0010 | -0.0188* | -0.0210* | -0.0168 |
| (-0.50) | (-1.75) | (-1.74) | (-1.56) | |
| Board | -0.0005 | -0.0031 | -0.0096** | -0.0023 |
| (-0.63) | (-0.86) | (-2.17) | (-0.62) | |
| Dual | 0.0001 | 0.0029* | -0.0028** | 0.0030* |
| (0.23) | (1.81) | (-2.15) | (1.85) | |
| lngdp | -0.0027*** | 0.0067 | 0.0013 | 0.0090 |
| (-3.20) | (1.09) | (0.30) | (1.54) | |
| lnfinance | 0.0018*** | -0.0018 | -0.0049 | -0.0019 |
| (2.73) | (-0.42) | (-1.37) | (-0.43) | |
| Cons | 0.0058 | -0.1527*** | -0.1759*** | -0.1674*** |
| (0.82) | (-3.54) | (-3.63) | (-4.01) | |
| Firm Fixed Effects | YES | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES | YES |
| Industry Fixed Effects | YES | YES | YES | YES |
| City Fixed Effects | YES | YES | YES | YES |
| Observations | 31,284 | 31,284 | 31,284 | 31,284 |
| Adj. R2 | 0.6732 | 0.4261 | 0.7926 | 0.4237 |
Note: This table presents the regression results for the impact of open government data on financing cost and financing scale. Column (1) presents the findings for the scenario wherein the relationship between the mechanism variable (OutFin) and the independent variable (Open) is examined. Column (2) presents the findings for the scenario wherein the relationship between corporate investment levels and mechanism variables (OutFin) is examined. Column (3) presents the findings for the scenario wherein the relationship between the mechanism variable (Loan) and the independent variable (Open) is examined. Column (4) presents the findings for the scenario wherein the relationship between the corporate investment levels and mechanism variables (Loan) is examined. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
5.2.2 The impact of open government data on financing scale
To examine whether the expansion of the financing scale serves as a mechanism through which government data openness influences corporate investment, this study primarily focuses on the increase in debt financing. It employs the volume of bank credit acquired as a metric for assessing the scale of corporate financing. Specifically, corporate bank credit is defined as follows: the volume of bank credit acquired (Loan) is calculated as the sum of long-term loans and short-term loans divided by total assets. The regression results pertaining to the mechanism variable (the volume of bank credit acquired) in relation to the explanatory variable (Open) are presented in Column (3) of Table 13. The coefficient for the explanatory variable is 0.0025, which is statistically significant at the 10% level, suggesting that increased government data openness is associated with a rise in the amount of credit accessible to enterprises. Additionally, the regression results concerning the enterprise investment level in relation to the mechanism variables are displayed in Column (4) of Table 13. We find that the coefficient of the explanatory variable (Open) is significantly positive at the 1% level, and the coefficient of financing scale (Loan) is likewise significantly positive at the 1% level. This suggests that an increase in credit acquisition can partially satisfy the financing needs of enterprises, provide financial support for enhancing investment, and elevate the level of corporate investment. Consequently, the mechanism of government data openness can be validated through its role in expanding the financing scale of enterprises and enhancing their investment levels. Thus, Hypothesis 2b was empirically validated.
5.3 Heterogeneity analysis
5.3.1 State-owned enterprises VS Non-state-owned enterprises
Given that enterprises of varying ownership structures may exhibit different levels of data utilization on government open data platforms, it is observed that state-owned enterprises (SOEs) are more likely to find alignment between the fields and directions of government data openness and their development strategies. This alignment facilitates the rapid transformation of open data into foundational elements for investment decisions and the identification of new investment opportunities, in contrast to non-state-owned enterprises. Simultaneously, state-owned enterprises typically maintain a close relationship and cooperative dynamic with government departments, which confers inherent advantages in accessing government data, obtaining authorization for data use, and facilitating data access. This symbiotic relationship enables enterprises to swiftly and efficiently acquire and utilize government data for informed investment decisions and strategic business expansion. Furthermore, state-owned enterprises exhibit a comparatively higher tolerance for risk in their investment decisions and demonstrate greater capacity and willingness to invest in emerging sectors and projects that leverage government data openness. This propensity facilitates the transformation and application of achievements derived from government data openness. Consequently, it can be inferred that the influence of government data openness on enterprise investment levels is more pronounced in state-owned enterprises.
To examine whether the influence of government data openness on corporate investment levels is moderated by varying ownership structures, we segregated the sample into two subgroups: state-owned enterprises (SOEs) and non-state-owned enterprises (Non-SOEs), based on the nature of the actual controllers of listed companies. Model (1) was employed for group testing, and the results are presented in columns (1) and (2) of Table 14. The findings indicate that the significance of government data openness is not significant within the category of non-state-owned enterprises. Conversely, in the context of state-owned enterprises, the coefficient associated with government data openness is significantly positive, suggesting that the influence of government data openness on the investment levels of state-owned enterprises is more pronounced than that observed in non-state-owned enterprises.
Table 14. The regression outcomes derived from the heterogeneity analysis. |
| Variable | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) |
|---|---|---|---|---|---|---|---|---|
| Non-SOEs | SOEs | High-Tech | Non-High-Tech | EU_low | EU_high | Eastern | Central-western | |
| Invest | Invest | Invest | Invest | Invest | Invest | Invest | Invest | |
| Open | 0.0001 | 0.0056*** | 0.0024** | 0.0037 | 0.0015 | 0.0032* | 0.0023** | 0.0016 |
| (0.11) | (2.98) | (2.59) | (1.59) | (0.72) | (1.79) | (2.22) | (0.63) | |
| Size | 0.0140*** | 0.0106*** | 0.0165*** | 0.0071** | 0.0150*** | 0.0116*** | 0.0132*** | 0.0121*** |
| (5.77) | (4.30) | (14.47) | (2.75) | (4.02) | (7.19) | (5.28) | (5.56) | |
| Lev | 0.0102 | 0.0015 | 0.0070 | 0.0072 | -0.0020 | 0.0066 | 0.0090 | -0.0057 |
| (1.72) | (0.12) | (0.90) | (0.75) | (-0.16) | (1.40) | (1.33) | (-0.77) | |
| Roe | 0.0216*** | 0.0282*** | 0.0246** | 0.0229*** | 0.0299*** | 0.0195*** | 0.0212*** | 0.0298*** |
| (4.49) | (4.32) | (3.98) | (3.20) | (4.24) | (5.00) | (5.19) | (5.56) | |
| Cfo | -0.0069 | -0.0073 | -0.0102 | 0.0040 | 0.0048 | -0.0085 | -0.0045 | 0.0002 |
| (-0.80) | (-0.68) | (-1.26) | (0.34) | (0.57) | (-0.72) | (-0.55) | (0.02) | |
| Age | -0.0473*** | -0.0394*** | -0.0408*** | -0.0457*** | -0.0448*** | -0.0446*** | -0.0455*** | -0.0356*** |
| (-8.28) | (-5.81) | (-8.71) | (-6.34) | (-8.74) | (-9.05) | (-10.75) | (-4.06) | |
| Growth | 0.0178*** | 0.0246*** | 0.0252*** | 0.0164*** | 0.0162*** | 0.0227*** | 0.0174*** | 0.0277*** |
| (5.95) | (5.33) | (11.82) | (3.65) | (4.50) | (6.90) | (4.72) | (9.11) | |
| Balance | 0.0010 | 0.0035* | -0.0002 | 0.0043 | 0.0031 | 0.0016 | 0.0017 | 0.0023 |
| (0.63) | (1.95) | (-0.12) | (1.63) | (1.41) | (1.00) | (1.14) | (0.95) | |
| Independ | -0.0230* | -0.0262 | -0.0108 | -0.0324 | -0.0410* | -0.0097 | -0.0092 | -0.0431** |
| (-2.06) | (-1.61) | (-1.86) | (-1.34) | (-2.09) | (-0.75) | (-0.71) | (-2.31) | |
| Board | -0.0072 | -0.0029 | -0.0015 | -0.0020 | -0.0000 | -0.0024 | -0.0042 | -0.0012 |
| (-1.51) | (-0.46) | (-0.35) | (-0.27) | (-0.01) | (-0.46) | (-0.73) | (-0.19) | |
| Dual | 0.0021 | 0.0015 | 0.0015 | 0.0036 | 0.0055** | 0.0011 | 0.0042** | -0.0016 |
| (1.17) | (0.72) | (0.93) | (1.66) | (2.66) | (0.45) | (2.72) | (-0.51) | |
| lngdp | 0.0115 | 0.0151** | 0.0078 | 0.0128 | 0.0091 | 0.0104** | 0.0171** | 0.0061 |
| (1.30) | (2.16) | (0.97) | (1.36) | (0.55) | (2.22) | (2.18) | (0.75) | |
| lnfinance | -0.0055 | -0.0049 | -0.0032 | -0.0057 | 0.0033 | -0.0057 | -0.0041 | -0.0020 |
| (-0.80) | (-1.14) | (-0.50) | (-1.06) | (0.49) | (-1.30) | (-0.65) | (-0.33) | |
| Cons | -0.1503* | -0.1361* | -0.2355*** | -0.0160 | -0.2953** | -0.1046** | -0.2271*** | -0.1303 |
| (-2.01) | (-2.04) | (-5.89) | (-0.32) | (-2.68) | (-2.43) | (-3.40) | (-1.72) | |
| Firm Fixed Effects | YES | YES | YES | YES | YES | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES | YES | YES | YES | YES | YES |
| Industry Fixed Effects | YES | YES | YES | YES | YES | YES | YES | YES |
| City Fixed Effects | YES | YES | YES | YES | YES | YES | YES | YES |
| Observations | 17,364 | 13,227 | 18,980 | 12,266 | 10,871 | 19,952 | 21,520 | 9,745 |
| Adj. R2 | 0.4501 | 0.3947 | 0.4389 | 0.4179 | 0.4175 | 0.4444 | 0.4516 | 0.3726 |
| Empirical p-value | 0.000 | 0.001 | 0.060 | 0.008 | ||||
Note: This table delineates the regression outcomes derived from the heterogeneity analysis. Column (1) specifically illustrates the results pertaining to the sample of non-state-owned enterprises (Non-SOEs). Column (2) specifically illustrates the results pertaining to the sample of state-owned enterprises (SOEs). Column (3) specifically illustrates the results pertaining to the sample of high-tech enterprises. Column (4) specifically illustrates the results pertaining to the sample of non-high-tech enterprises. Column (5) specifically illustrates the results pertaining to the sample of low uncertainty groups. Column (6) specifically illustrates the results pertaining to the sample of high uncertainty groups. Column (7) specifically illustrates the results pertaining to the sample of eastern enterprises. Column (8) specifically illustrates the results pertaining to the sample of central and western enterprises. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
5.3.2 High-tech enterprise VS Non-high-tech enterprise
The influence of government data openness on corporate investment across various industries may exhibit significant variability. High-technology enterprises, in contrast to their non-high-technology counterparts, demonstrate a heightened demand for information and knowledge. The availability of government data can provide these enterprises with valuable insights into technological development trends, market demand dynamics, and the status of competitors. Enterprises utilize government data to comprehend the latest technological trends and shifts in market demand within their industry. This enables them to promptly adjust their investment strategies and development priorities, thereby allocating resources to projects with greater market potential and technological feasibility. Consequently, this accelerates the transformation of innovative achievements into new products and fosters enterprise investment. It can be inferred that the impact of open government data on corporate investment is likely to be more pronounced in high-tech enterprises. To evaluate this inference, we categorized the sample into two subsamples: high-tech enterprises and non-high-tech enterprises, subsequently employing Model (1) for group testing. As evidenced by columns (3) and (4) of Table 14, the coefficient for government data openness is significantly positive at the 5% level within high-tech enterprises, whereas its influence is not statistically significant in the non-high-tech enterprise. This finding suggests that the effect of government data openness on corporate investment is more pronounced in high-tech enterprises.
5.3.3 The impact of macroeconomic uncertainty
Economic policy uncertainty not only heightens the unpredictability of a company’s future cash flows but also amplifies the risk of investment failure. Consequently, as macroeconomic uncertainty escalates, the demand for information among enterprises intensifies. On one hand, the transparency of government data can furnish enterprises with critical insights into macroeconomic trends, industry dynamics, and relevant policies and regulations, thereby enhancing their capacity to identify and assess potential risks and opportunities more effectively. On the other hand, government data openness can provide enterprises with critical information concerning industrial development, facilitating a deeper understanding of development trends, market saturation, and technological innovation trajectories. By synthesizing this data with their internal resources and capabilities, enterprises can strategically allocate investment capital to industries exhibiting substantial development potential and strong risk resilience, thereby reducing the probability of investment failure.
It can be inferred that the influence of government data openness on corporate investment is more pronounced under conditions of elevated macroeconomic uncertainty. To empirically examine this hypothesis, we employ the Economic Policy Uncertainty Index (EU) as a proxy for economic policy uncertainty, following the methodology outlined by Baker et al. (2016). The sample is stratified into high (EU-high) and low (EU-low) uncertainty groups based on the mean value of the macroeconomic uncertainty index. Subsequently, Model (1) was utilized to conduct the grouped analysis. The test results are presented in columns (5) and (6) of Table 14. The regression coefficient for the variable Open is not statistically significant in the samples characterized by low macroeconomic uncertainty. However, it is significantly positive in samples exhibiting high macroeconomic uncertainty. This finding suggests that the influence of open government data on corporate investment is more pronounced in enterprises experiencing high levels of macroeconomic uncertainty.
5.3.4 The impact of location heterogeneity
There exist notable disparities in data infrastructure, levels of economic development, data integration capabilities, and market demand across various regions, which may result in divergent effects of government data openness on corporate investment levels. Generally, the eastern region exhibits a relatively advanced data infrastructure and technological environment, facilitating easier access to and utilization of government data by enterprises. This enhanced accessibility is likely to improve production efficiency and competitiveness among enterprises, thereby fostering increased investment. In addition, the eastern region exhibits a comparatively advanced level of economic development, characterized by robust corporate investment and intense market competition. The openness of government data is likely to further stimulate enterprise innovation and enhance market competition, thereby improving investment efficiency and return on investment. Conversely, the central and western regions experience relatively slower economic development, with potentially weaker corporate investment demand. Consequently, we can infer that the influence of government data openness on corporate investment is more pronounced in the eastern region. We categorized cities into two groups, the East and the Midwest, and subsequently applied Model (1) for group testing. The regression results for the Eastern group are presented in Column (7) of Table 14, where the coefficient of the independent variable (Open) is significantly positive. In contrast, Column (8) displays the regression results for the central and western groups, where the coefficient of the independent variable (Open) is not significant. These findings suggest that the effect of government data openness on corporate investment is more pronounced in the eastern cities.
6 Additional tests
6.1 The quality of government data openness and corporate investment
This study employs the China Open Data Forest Index, developed by the Digital and Mobile Governance Laboratory at Fudan University, to assess the quality of public data openness. This index offers a comprehensive and objective evaluation of the data openness platforms across various local governments. In addition to the overall index, it comprises sub-indices, such as the data layer index, platform layer index, and readiness index. The data layer index evaluates the quality of open data, including the breadth of openness, data standards, and data volume. The platform layer index assesses the quality of platform construction by focusing on the user experience, interactive feedback, and data discovery and acquisition. Finally, the readiness index examines the robustness of policy support, including the effectiveness and content of regulations and policies, establishment of standards and norms, and organization and implementation processes. In this study, the natural logarithm of the comprehensive index was employed to assess the overall quality of government data openness (Lnquality). Concurrently, the natural logarithms of the data layer index, platform layer index, and readiness index are utilized to evaluate the quality of open data (Lndata), quality of platform construction (Lnplat), and strength of policy support (Lnpolicy), respectively. It is anticipated that an enhancement in the quality of government data openness, open data quality, platform construction quality, and policy support strength will correspond to increased enterprise investment.
The detailed results of the regression analysis are presented in Table 15. In Column (1), the regression coefficient for the relationship between Lnquality and Invest is 0.0029, which is statistically significant at the 1% level. This finding implies that an increase in the quality of government data openness correlates with higher levels of enterprise investment, thereby reinforcing the conclusions drawn from the benchmark regression. Upon disaggregating the index dimensions, Column (2) reveals that the regression coefficient for Lndata and Invest is 0.0024, which is also significant at the 1% level, indicating that the enhanced quality of open data is associated with increased enterprise investment. Similarly, Column (3) shows a regression coefficient of 0.0036 for Lnplat and Invest, significant at the 1% level, suggesting that improvements in the quality of platform system construction lead to greater enterprise investment. Finally, Column (4) presents a regression coefficient of 0.0025 for Lnpolicy and Invest, which is significantly positive at the 1% level, suggesting that stronger policy guarantees are linked to higher enterprise investment. The findings suggest that the significance of both the quality of open data and the construction of platforms for government data transparency is paramount. Furthermore, robust policy frameworks and institutional safeguards are essential to ensure effective government data openness.
Table 15. The impact of the quality of government data openness on corporate investment. |
| Variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Invest | Invest | Invest | Invest | |
| Lnquality | 0.0029*** | |||
| (3.33) | ||||
| Lndata | 0.0024*** | |||
| (3.62) | ||||
| Lnplat | 0.0036*** | |||
| (2.93) | ||||
| Lnpolicy | 0.0025*** | |||
| (3.40) | ||||
| size | 0.0066*** | 0.0066*** | 0.0066*** | 0.0069*** |
| (8.87) | (8.85) | (8.92) | (9.12) | |
| lev | 0.0183*** | 0.0183*** | 0.0177** | 0.0166*** |
| (2.85) | (2.87) | (2.72) | (3.01) | |
| roe | 0.0210*** | 0.0211*** | 0.0211*** | 0.0205*** |
| (7.22) | (7.37) | (7.10) | (7.73) | |
| cf | 0.0536*** | 0.0538*** | 0.0530*** | 0.0481*** |
| (4.67) | (4.66) | (4.50) | (4.52) | |
| age | -0.0232*** | -0.0233*** | -0.0231*** | -0.0232*** |
| (-18.66) | (-18.87) | (-18.14) | (-17.65) | |
| growth | 0.0254*** | 0.0253*** | 0.0256*** | 0.0275*** |
| (9.59) | (9.61) | (9.28) | (9.46) | |
| sharesbalance | -0.0011 | -0.0011 | -0.0011 | -0.0011 |
| (-1.22) | (-1.21) | (-1.20) | (-1.09) | |
| independ | 0.0083 | 0.0080 | 0.0090 | 0.0127 |
| (0.58) | (0.56) | (0.62) | (0.82) | |
| board | -0.0062 | -0.0064 | -0.0062 | -0.0065 |
| (-1.63) | (-1.68) | (-1.57) | (-1.59) | |
| dual | 0.0045** | 0.0045** | 0.0046** | 0.0042* |
| (2.15) | (2.16) | (2.08) | (1.71) | |
| lngdp | 0.0053 | 0.0054 | 0.0045 | 0.0049 |
| (1.69) | (1.66) | (1.45) | (1.52) | |
| lnfinance | -0.0068** | -0.0067** | -0.0061** | -0.0062** |
| (-2.54) | (-2.43) | (-2.37) | (-2.31) | |
| _cons | 0.0018 | 0.0049 | -0.0017 | -0.0052 |
| (0.07) | (0.18) | (-0.07) | (-0.18) | |
| Firm Fixed Effects | YES | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES | YES |
| Industry Fixed Effects | YES | YES | YES | YES |
| City Fixed Effects | YES | YES | YES | YES |
| Observations | 11,145 | 11,145 | 10,937 | 10,123 |
| Adj. R2 | 0.2040 | 0.2036 | 0.2041 | 0.2066 |
Note: This table presents the impact of the quality of government data openness on enterprise investment. Column (1) pertains to the assessment of the overall quality of government data openness (Lnquality); Column (2) pertains to the evaluation of the quality of open data (Lndata); Column (3) pertains to the evaluation of the quality of platform construction (Lnplat); and Column (4) pertains to the evaluation of the strength of policy support (Lnpolicy). |
6.2 The impact of open government data on corporate investment efficiency
Following the release of government data resources, the costs associated with information acquisition for enterprise management and external stakeholders are significantly reduced. This reduction in information acquisition costs facilitates the increased utilization of government data resources by both internal and external stakeholders, thereby transitioning investment decisions from human judgment to data-driven methodologies and enhancing the overall quality of decision-making. Simultaneously, the transparency of government data addresses the issue of information dissemination through the integration of big data analytics and digital operations. This enhances the quality of accounting reports available to enterprises, provides more comprehensive credit data, and facilitates more precise investment risk assessments. Consequently, it promotes the optimization of corporate strategies, enhances the precision of investments, mitigates opportunity costs, and improves overall investment efficiency. Furthermore, the transparency of government data can substantially mitigate the principal-agent problem within enterprises by enhancing oversight, providing incentives, and informing decision-making processes. This reduction in agency costs not only minimizes potential harm to shareholders but also optimizes managerial investment decisions. External stakeholders are better equipped to monitor corporate activities through the analysis of government data resources, including registries of abnormal business operations, lists of dishonest entities, and directories of entities recognized for trustworthy behavior.
Therefore, we deduce that the transparency of government data can facilitate companies in promptly addressing potential investment opportunities, thereby enhancing their investment efficiency. Drawing upon the studies by Chen et al. (2011) and McLean et al. (2012), we utilize the sensitivity of investment expenditure to investment opportunities (Tobin’s Q) as our metric for assessing investment efficiency, as delineated in Model (7).
$\text { Invest }_{i t}=\alpha_{0}+\alpha_{1} \text { Open }^{*} \text { Tobin's } Q_{i t}+\alpha_{2} \text { Tobin's } Q_{i t}+\alpha_{3} \text { Open }_{i t}+\alpha_{4} \text { control }_{i t}+\lambda_{i}+\mu_{t}+\varepsilon_{i t}.$
Among these variables in Model (7), with the exception of Tobin’s Q, all are defined consistently in Model (1). As outlined by Chen et al. (2011), Tobin’s Q is calculated using the following formula: (year-end market value of tradable shares + book value of non-tradable shares + total long-term liabilities + total short-term liabilities) divided by the year-end book value of total assets. This metric serves as a proxy for corporate investment opportunities. This study employs the interaction term (Open * Tobin’s Q) between government data openness (Open) and investment opportunities (Tobin’s Q) as the explanatory variable. A significantly positive coefficient α1 for the interaction term would suggest that government data openness enhances the corporate investment efficiency. The empirical results presented in Table 16 reveal that the coefficient α1 is always significantly positive from columns (1) to (4). This finding suggests that the openness of government data enhances the responsiveness of enterprises to investment opportunities, thereby improving the efficiency of corporate investment. In summary, the openness of government data not only stimulates corporate investment but also facilitates the allocation of scarce capital to investment projects, thereby converting potential investment opportunities into actual investments and enhancing overall corporate investment efficiency.
Table 16. The impact of government data opening on corporate investment efficiency. |
| Variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Invest | Invest | Invest | Invest | |
| Open*Tobin’s Q | 0.0026*** | 0.0024*** | 0.0024*** | 0.0022*** |
| (3.48) | (3.76) | (3.63) | (3.67) | |
| Tobin’s Q | -0.0011 | 0.0014** | 0.0014** | 0.0013** |
| (-1.47) | (2.45) | (2.48) | (2.30) | |
| Open | -0.0022 | -0.0016 | -0.0015 | -0.0015 |
| (-0.96) | (-1.02) | (-0.94) | (-0.86) | |
| Size | 0.0140*** | 0.0140*** | 0.0142*** | |
| (7.90) | (7.55) | (6.89) | ||
| Lev | 0.0019 | 0.0027 | 0.0028 | |
| (0.36) | (0.55) | (0.49) | ||
| Roe | 0.0218*** | 0.0222*** | 0.0224*** | |
| (5.25) | (5.47) | (5.49) | ||
| Cfo | -0.0043 | -0.0051 | -0.0048 | |
| (-0.50) | (-0.61) | (-0.58) | ||
| Age | -0.0430*** | -0.0430*** | -0.0437*** | |
| (-9.31) | (-9.48) | (-9.48) | ||
| Growth | 0.0211*** | 0.0211*** | 0.0209*** | |
| (6.12) | (6.15) | (6.15) | ||
| Balance | 0.0016 | 0.0016 | 0.0019 | |
| (1.36) | (1.31) | (1.41) | ||
| Independ | -0.0204* | -0.0201* | -0.0193* | |
| (-1.98) | (-1.97) | (-1.76) | ||
| Board | -0.0032 | -0.0034 | -0.0030 | |
| (-0.91) | (-1.00) | (-0.80) | ||
| Dual | 0.0030* | 0.0030* | 0.0030* | |
| (1.80) | (1.85) | (1.78) | ||
| lngdp | 0.0068 | 0.0067 | 0.0087 | |
| (1.54) | (1.49) | (1.46) | ||
| lnfinance | -0.0077 | -0.0073 | -0.0023 | |
| (-1.68) | (-1.55) | (-0.52) | ||
| _cons | 0.0550*** | -0.0933** | -0.0973* | -0.1973*** |
| (30.40) | (-2.11) | (-2.09) | (-4.68) | |
| Firm Fixed Effects | YES | YES | YES | YES |
| Year Fixed Effects | YES | YES | YES | YES |
| Industry Fixed Effects | NO | NO | YES | YES |
| City Fixed Effects | NO | NO | NO | YES |
| Observations | 31,284 | 31,284 | 31,284 | 31,284 |
| Adj. R2 | 0.3743 | 0.4210 | 0.4221 | 0.4227 |
Note: This table presents the regression results for the impact of government data opening on corporate investment efficiency. Column (1) presents the results of the regression analysis without the control variables. Column (2) reports the regression results with all control variables added. Column (3) incorporates industry trend controls, in addition to the comprehensive variable regression analysis. Column (4) incorporates regional trend controls in addition to a comprehensive variable regression analysis. ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively. |
7 Conclusion and implication
7.1 Conclusions
Given the established observation that the application of big data has substantially influenced contemporary economic and social development in China, this study examines the impact of data elements on corporate investment, with a particular emphasis on the role of government open data platforms as pivotal facilitators of data sharing. By analyzing data from A-share listed companies from 2007 to 2022, we employ a staggered Difference-in-Differences model to empirically investigate the effects and underlying mechanisms through which data elements influence corporate investment. The principal findings of this research can be summarized as follows: The dissemination of data elements facilitated by the establishment of government data open platforms has markedly enhanced the investment levels of enterprises, underscoring the critical importance of data elements. Mechanistic analysis indicates that the alleviation of financing constraints, reduction of financing costs, and expansion of financing scale constitute the primary mechanisms driving corporate investment. Moreover, we find that the introduction of the government data open platform has heightened the sensitivity of corporate investment to available investment opportunities, thereby enhancing the efficiency of corporate investment. In addition, the findings of this study revealed considerable heterogeneity. Specifically, the impact of government data openness on improving investment levels is notably more pronounced among state-owned enterprises, high-tech firms, enterprises operating in environments characterized by high macroeconomic uncertainty, and those located in the eastern region.
7.2 Implications
This study examines the value of data resources from the perspective of enterprise investment, thereby enriching the existing literature on the economic implications of government data resources. The findings have several practical implications. First, we demonstrate that government data openness enhances both the level and efficiency of corporate investment. This suggests that government data openness has, to some extent, succeeded in achieving its objective of “guiding economic development and serving public enterprises.” Consequently, policymakers should actively advocate for the implementation of government data openness policies, aiming to both increase the volume of publicly accessible data and enhance its quality. Additionally, it is essential to bolster public awareness and support initiatives that empower enterprises to leverage public data resources effectively.
Second, a mechanistic analysis indicates that government data openness can mitigate financing constraints, lower financing costs, and expand the scale of financing, thereby fostering corporate investment. Therefore, it is imperative for enterprises to comprehensively acknowledge the importance of data elements. They must expedite digital transformation by utilizing advanced technologies to seamlessly integrate data into their business operations. Furthermore, the ongoing exploration of the potential value inherent in public data resources can unveil new opportunities for growth. By implementing these strategies, companies can enhance their competitiveness and facilitate accelerated development in the digital era.
Third, heterogeneity testing indicates that the openness of government data exerts a more significant impact on enhancing investment levels in state-owned enterprises, high-tech enterprises, enterprises experiencing substantial macroeconomic uncertainty, and those located in the eastern region. This finding has two implications. First, compared to other types of enterprises, state-owned enterprises, high-tech enterprises, those confronting considerable macroeconomic uncertainty, and enterprises situated in the eastern region should prioritize the integration of data elements into their investment strategies. Second, while enterprises in the eastern region demonstrate proficiency in utilizing data resources to boost investment levels and operational efficiency, it is crucial for local governments in the central and western regions to intensify their efforts in developing data infrastructure and enhancing data integration capabilities. This will enable the optimal utilization and realization of the value of data resources in the central and western regions.
7.3 Research deficiency and prospect
In this study, the relationship between government data openness and corporate investment was explored; however, there are opportunities for improvement in the problem analysis process. The research predominantly investigated the impact of government data openness on corporate investment without considering the effects of macroeconomic fluctuations, internal corporate governance structures, and industry regulatory policies. Additionally, this study focuses exclusively on the influence of government data openness on overall corporate investment levels, neglecting an analysis of the investment structure. It is crucial to acknowledge that various types of corporate investment, such as research and development (R&D) investment, may be differentially influenced by the openness of government data. The lack of a comprehensive examination of the investment structure constrains the ability of the research findings to provide targeted recommendations for optimizing corporate investment strategies.
Funding information
This research was supported by the following grants: National Natural Science Foundation of China (Grant No. 72271147), Natural Science Foundation of Guangdong Province (Grant No. 2022A1515010573), STU Scientific Research Initiation (Grant No. STF21005; STF21004), the Philosophy and Social Science Planning Project of Guangdong Province (Grant No. GD21CGL12; GD23XYJ76).
Author contributions
Yumei Fu (Email: fuyumei@stu.edu.cn): Conceptualization, Data curation, Methodology, Writing - original draft.
Baixun Li (Email: baixunli@stu.edu.cn; ORCID: 0000-0002-4731-0976): Data curation, Formal analysis, Funding acquisition, Methodology, Writing - review & editing.