2.1. Theoretical analysis

Since the mid-1970s, policy science researchers have increasingly turned their attention to the micro-level processes of policy implementation, developing a variety of theoretical models to explain differences in implementation performance. These include process models, interaction models, cyclical models, systems models, and integrated models. Among them, Van Meter and Van Horn are representative figures of the systems model. Drawing on studies of organizational control, judicial decisions, and inter-organizational relationships, they proposed a policy implementation system model. Van Meter and Van Horn identified six key factors that influence implementation performance following policy enactment: standards and objectives—that is, the clarity and consensus of policy objectives; policy resources; characteristics of the implementing agencies; interorganizational communication and enforcement activities; economic, social, and political conditions; and the disposition of implementors—that is, their cognition, reactions, and responsiveness [4].Compared to the widely discussed policy delivery system models of that time, the policy implementation system model offered a more comprehensive framework for explaining the linkage between policy content and implementation outcomes.

In China’s PPP development model, the relationship between providers and users is characterized by a top-down vertical authority structure [9]. As a classic example of a top-down policy implementation approach, Van Meter and Van Horn’s model aligns well with the study of PPP project landing in the context of digital society development. Generally speaking, the external environment—comprising economic, social, and political conditions—tends to remain relatively stable over time and is therefore unlikely to serve as a practical lever for accelerating the landing of digital society PPP projects. As such, we exclude this fifth dimension and concentrate on the dynamic, agent-centered components of the policy implementation system model: internal goal consensus, project resources, actor characteristics, inter-organizational relationships, and actor dispositions. External factors like the economic, social, and political context are instead incorporated as control variables in the econometric analysis.

2.2. Research hypothesis

Van Meter and Van Horn argued that during the policy formulation stage, policymakers define implementation standards, procedures, and instruments. If internal actors within the system diverge in their goals, it may hinder the allocation of resources and the development of strategies essential for successful implementation [4]. Digital society PPP projects, which are core to advancing China’s digital transformation strategy, reflect policy goals such as technology-driven development and innovation-led growth [10]. However, government agencies at various levels are often tasked with numerous administrative responsibilities across sectors, and conflicting priorities may compete for attention and resources. Executing agencies may seek to secure additional resources for their own performance or incentives, thereby crowding out support for innovation-oriented goals [11]. Therefore, broad internal support within the government for the innovation-driven development agenda becomes a key endogenous force for securing the necessary resources for digital society PPP projects. Accordingly, the following hypothesis is proposed:

3.1 Event history analysis

To examine the impact of policy implementation systems on the speed at which Public-Private Partnership (PPP) projects in digital society development are launched, we adopt a continuous-time event history analysis (also known as survival analysis). This method focuses on the hazard probability ,which is the probability that a subject who has survived up to time t experiences a specific event at time that a subject, having survived until time t, experiences a specific event at time . This hazard function can be expressed as the ratio of the failure density function to the survival function , as shown in Equation (1). The survival function represents the probability that a case survives beyond time , while the failure density reflects the instantaneous risk of failure at time . Based on this relationship, a higher failure density and shorter survival time indicate a higher hazard probability. Unlike discrete-time event history analysis, the continuous-time method captures not only the likelihood of an event occurring but also the timing of its occurrence. In the context our study, this method allows for an investigation into both (1) how the characteristics of the policy implementation system influence the likelihood of project landing, and (2) how these characteristics affect the speed at which digital society PPP projects are launched.

(1)

In selecting observational units, we first employed web-scraping techniques to extract project data from the Project Management Library and the Project Reserve Library of the China PPP Integrated Information Platform. Through this method, a total of 14,100 projects were collected, each with information on 12 key attributes, including project name, location, sector, approval date, development stage, investment size, return mechanism, project overview, scope of cooperation, cooperation duration, operational model, and procurement method. Based on this dataset, the research team manually screened and identified 319 projects relevant to four key domains: digital public services, smart cities, digital rural initiatives, and digital literacy enhancement programs. After further evaluating the completeness of project data and the availability of independent variables, 300 projects were ultimately included in the study. In accordance with the principles of survival analysis, right-censored data were applied to form an unbalanced panel dataset. This process yielded a total of 736 usable observations. Among these, 141 projects had reached landing (i.e., “landed” projects), with provincial-level projects accounting for 21.30%, municipal-level projects 73.15%, and county-level projects 61.11%.

3.2 Baseline regression model

To test the hypotheses proposed, the Cox proportional hazards model—a standard method in continuous-time event history analysis—is employed as the baseline regression model. As a semi-parametric model, the Cox model offers several advantages over traditional regression techniques, including not requiring a pre-specified baseline hazard distribution and providing robust coefficient estimates, making it well-suited for this analysis. The baseline regression model is specified as follows:

(2)

In this equation, represents the hazard rate, or the probability that a project is implemented at time t. The baseline hazard function depends only on time. The terms and denote the lagged explanatory and control variables, respectively.

3.2. Variables and data

3.2.3. Control variables.

We control for the influence of the project environment on the landing speed of PPP projects from two dimensions: the market technological environment and the external knowledge environment. The market technological environment is measured by the natural logarithm of the number of invention patents granted to high-tech enterprises, while the external knowledge environment is proxied by the number of universities in the region. In addition, drawing on existing studies on PPP project landing, this paper also includes project characteristics as control variables—specifically, project level, demonstration status, investment size, return mechanism, and cooperation duration. Year fixed effects based on the project approval year are also controlled for. Detailed variable definitions are provided in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Variable measurements.

https://doi.org/10.1371/journal.pone.0329180.t001

4.1. Baseline results

Table 2 presents the descriptive statistics and multicollinearity diagnostics for the main variables. As shown, all variables have variance inflation factors (VIFs) below 10, indicating that multicollinearity is not a serious concern. The descriptive statistics on project duration reveal that among all digital society PPP projects, the maximum duration for projects that remained unimplemented reached 1,856 days. In contrast, the longest duration for implemented projects was 896 days, with an average duration of 369.37 days. These figures highlight the existence of challenges in project landing, including significant delays and difficulties in achieving completion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics and diagnosis of multicollinearity.

https://doi.org/10.1371/journal.pone.0329180.t002

To illustrate the distribution of project landing risk over time, we employ a weighted kernel density estimate to construct a smoothed hazard function, as shown in Fig 1. The hazard rate represents the instantaneous probability that a surviving project will be implemented in the next infinitesimally small-time interval. The curve in Fig 1 exhibits an inverted U-shape, with the peak occurring around 300 days, indicating that the probability of project landing is highest at approximately that time. After this point, the likelihood of landing declines, suggesting increasing difficulty as time progresses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Estimated probability of project landing risk.

https://doi.org/10.1371/journal.pone.0329180.g001

4.2. Hypothesis testing

Model 1 in Table 3 reports the results of the baseline regression. Regarding internal goal consensus, the coefficient for internal goal consensus on innovation is significantly positive at the 1% level. A 1% increase in goal alignment leads to a 4% increase in the probability of project landing. This finding supports Hypothesis 1, indicating that higher consistency in innovation goals among stakeholders accelerates project landing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Results of hypothesis testing.

https://doi.org/10.1371/journal.pone.0329180.t003

As for project resources, while the coefficients for government fiscal resources and regional financial resources show the expected direction, their statistical significance does not reach conventional levels. This suggests that in the context of digital society PPP projects, the availability of resources does not significantly influence landing speed, thus failing to support Hypotheses 2 and 3.

In terms of actor characteristics, the level of internal government coordination is positively significant at the 1% level. Specifically, each additional department present in a city’s administrative approval center increases the probability of project landing by 1.7%. This confirms Hypothesis 4, indicating that improved internal government coordination facilitates faster project landing. Unexpectedly, the level of information enterprise development is significantly and negatively associated with project landing at the 5% level. A 1% increase in the information enterprise development is associated with a 57.7% decrease in landing probability, contradicting Hypothesis 5. This suggests that more developed information enterprises may paradoxically hinder the landing of digital society PPP projects.

In the dimension of inter-organizational relationships, the coefficient for government-business relationship quality is significantly positive at the 5% level. Specifically, for each one-point increase in the “government-business relations” score within the marketization index, the probability of a digital society PPP project being implemented increases by 29%. This suggests that stronger and more favorable government-business relations are more likely to facilitate the successful landing of digital society PPP projects— a finding that is broadly consistent with Hypothesis 6.

In terms of actor dispositions, the coefficient for leadership performance demand is significantly positive at the 10% level. That is, with each additional year of a local administrative leader’s tenure, the likelihood of a digital society PPP project being landed in that region increases by 9.8%, a result generally aligned with Hypothesis 7. This implies that the longer a local official remains in office, the stronger the performance-based promotion incentive becomes, thereby increasing the likelihood of promoting project landing. With regard to corporate social responsibility, the coefficient is significantly positive at the 5% level. For each one-point increase in the average CSR index of listed companies within a region, the probability of implementing a digital society PPP project rises by 7.0%. This indicates that stronger corporate social responsibility enhances the likelihood of collaboration with the government in providing public services, thereby increasing the probability of project landing. We conducted Schoenfeld residuals tests and found that several variables – particularly external knowledge (χ² = 8.27, p = 0.004) – violated the proportional hazards assumption, which could bias the estimation results [5]. To address this violation while maintaining modeling precision, we replaced the Cox model with a Weibull parametric survival model. This alternative approach provides three key advantages: (1) it does not require proportional hazards assumptions; (2) enables direct estimation of duration dependence through the shape parameter; and (3) permits straightforward prediction of survival probabilities. The results are presented in Model 2 of Table 3. Importantly, despite this methodological adjustment, our core findings regarding project investment effects remained statistically significant and directionally consistent, demonstrating result robustness.

Hypotheses 2, 3, and 5 were not supported by empirical evidence. Drawing upon existing research, we offer the following explanations. First, in terms of project resources, their role is primarily foundational, and we may have overestimated their impact in the hypotheses. Systems theory suggests that for any given system, the quantity of resource input determines the potential for performance output, while the composition and relationships among actors in the system are key efficiency factors that determine whether that potential can be realized [30]. Therefore, in the context of digital society PPP project landing, inputs such as fiscal and financial resources determine only the potential speed of landing, whereas other organizational conditions are the key variables that drive actual progress. This explains their lack of statistical significance in the hypothesis tests. Second, while the development of information enterprises can objectively facilitate project landing, our hypothesis overlooked the subjective factor of firms’ willingness to participate. Specifically, the development of information enterprises is associated with their competitiveness and investment opportunities in the market, both of which influence firms’ expected returns [31]. However, previous research has pointed out that the current PPP model still suffers from high entry barriers and low returns [32]. Consequently, when expected market returns are high, firms may be less inclined to engage in PPP projects. As a result, the development of information enterprises may be negatively correlated with the speed of digital society PPP project landing. This perspective may provide an alternative explanation for the negative relationship between market development and project landing rates observed in previous PPP studies [33].

4.3. Robustness checks

To further enhance the credibility of our findings, additional tests are conducted to examine the potential adverse effects of information enterprise development. To rule out the alternative explanation that the observed negative correlation between information enterprise development and project landing is due to inappropriate measurement indicators, we re-evaluate the relationship using the Information Economy Index, a sub-index from the China Information Society Development Index published by the National Information Center. Furthermore, to verify that the negative relationship between information enterprise development and the speed of project landing is driven by market mechanisms, we substitute information enterprise development with a more general indicator of factor market development and conduct an additional robustness test.

Model 3 in Table 4 presents the results using the alternative measure for information enterprise development. The coefficient remains significantly negative at the 5% level, suggesting that the negative correlation is not the result of measurement bias. Model 4 reports the results using factor market development instead. The coefficient is significantly negative at the 1% level, indicating that the negative relationship between information enterprise development and project landing speed is indeed driven by market dynamics. These findings confirm the robustness of our empirical results and reinforce the validity of the study’s central hypothesis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Results of robustness checks.

https://doi.org/10.1371/journal.pone.0329180.t004

5.1. Conclusions

The PPP model has increasingly become a substitute for, or complement to, traditional construction approaches in the realm of digital society development. However, PPP projects in this sector are currently facing significant landing challenges. In the Chinese context, PPP projects are typically initiated by the government and managed through administrative directives issued by national or local authorities. As such, the speed of PPP project landing essentially reflects the effectiveness of policy execution. The policy execution system in which a project is embedded is theoretically a critical factor influencing its progress. Drawing on the theoretical framework proposed by Van Meter and Van Horn, we identify the key components of the policy execution system in digital society PPP projects and constructs a causal mechanism linking this system to project landing speed. Using data from 300 projects, the study employs continuous-time event history analysis to empirically validate the proposed mechanism. The results yield the following findings. (1) Compared with organizational and human factors, material resources such as government fiscal resources and regional financial resources have a relatively limited effect on project landing speed. (2) Among various organizational and actor factors, internal goal consensus on innovation, internal government coordination, government-business relationship quality, leadership performance demand, and corporate social responsibility all exhibit a significant positive correlation with landing speed. (3) However, due to low project profitability and obstacles to mobilizing private sector capital, information enterprise development shows a significant negative correlation with project landing speed.

This research not only identifies the key components of the policy implementation system in digital society PPP projects based on the Van Meter-Van Horn model and establishes their causal relationship with landing speed, but also makes broader theoretical contributions to the field of policy implementation. The study finds that in governance contexts where material resources are relatively abundant, inter-organizational relationships and institutional embeddedness—such as government coordination, government–enterprise relations, and leadership incentives—have a more pronounced impact on execution performance than material inputs. This suggests a need to move beyond the traditional overemphasis on resource provision in policy implementation research and instead focus more on institutional coordination and actor motivation. The study highlights the role of corporate social responsibility as an informal institutional factor in policy execution, offering new theoretical insights into understanding enterprise behavior within execution systems. Finally, the application of PPPs in digital society development—as an emerging policy tool—reflects the growing complexity and technical nature of policy goals. We provide an analytical framework for understanding how such complex goals can be achieved through coordination mechanisms, offering theoretical relevance for the implementation of complex policies in other domains.

5.2. Policy implications

The above findings offer the following implications for promoting the successful landing of PPP projects in digital society development.

On one hand, there is a need to move away from traditional resource-intensive development models and focus on cultivating an enabling organizational environment. While resource investment has the potential to accelerate project landing, realizing this potential depends heavily on the development of organizational and human factors. At this stage, local governments should prioritize internal capacity building and the integration of relevant stakeholders, with the aim of constructing a more supportive policy execution system that enables timely, appropriate, and efficient resource allocation. Specifically, local governments should reinforce the principle of “innovation-driven development“ as a key criterion for evaluating the performance and political credibility of subordinate governments and officials. Institutional arrangements should promote stable coordination mechanisms among functional departments to improve administrative efficiency. Cross-sector collaboration among government, enterprises, and academic institutions should be strengthened to form enduring partnerships. Furthermore, governments should advocate for and foster a culture of corporate social responsibility within society and the market.

On the other hand, policy instruments should be leveraged to enhance the intangible (soft) returns for enterprises participating in digital society PPP projects. The findings suggest that the tangible (hard) financial returns available to firms are currently limited, and external market forces may hinder the mobilization of necessary capital. In this context, local governments should implement supportive policies to enhance the public image of private enterprises involved in these projects. By improving their reputation and social standing, such soft returns can help compensate for the lack of hard returns, thereby making PPP projects more attractive to private capital.