2016年-世界发展银行全球_Effects_of_Multilateral_Support_on_Infrastructure_PPP_Contract_Cancellation_18页_1mb

Summary of "Effects of Multilateral Support on Infrastructure PPP Contract Cancellation"

Core Content

This working paper investigates the impact of multilateral support on the cancellation rates of long-term infrastructure public-private partnership (PPP) contracts. It employs a quasi-experimental approach using a large dataset of PPP projects from the World Bank's PPI Database to estimate the effect of multilateral support on contract survival.

Main Findings

• Multilateral Support and Contract Survival: The study finds that multilateral support significantly reduces the likelihood of PPP contract cancellation. Specifically, the cancellation rate for projects with multilateral support (MLS) is estimated to be 48% lower than it would have been without such support.
• Cancellation Rates: The observed cancellation rates between MLS and non-MLS projects are not statistically different (6.1% vs. 5.8%), but this is due to selection bias. A properly matched comparison group reveals that non-MLS projects have a higher cancellation rate (8.6%) compared to MLS projects (5.8%).
• Propensity Score Matching: The study uses a multilevel econometric model and propensity score matching to control for selection bias. The results indicate that the average probability of receiving MLS is significantly higher for projects with MLS (10.7%) than for those without (2.8%).
• Project Characteristics: Larger PPP projects are more likely to receive MLS, and certain sectors (e.g., power projects) are more favored than others (e.g., transport projects). Projects in Africa, Eastern Europe and Central Asia, and Latin America and the Caribbean are more likely to receive MLS than those in South Asia.
• Project Type: Brownfield concessions and management and lease contracts are less likely to involve multilateral participation compared to Greenfield projects.
• Financial Closure Year: Projects that reached financial closure earlier were more likely to receive MLS.
• Sensitivity Analysis: Additional matching methods (Nearest-Neighbor and Radius matching) confirm the kernel-based results, showing that non-MLS projects have a higher chance of cancellation (9.3% and 8.8% respectively) compared to MLS projects.

Key Information

• Methodology: The paper uses a multilevel mixed effects probit regression to estimate the probability of receiving MLS. It then applies propensity score matching to identify a counterfactual group of non-MLS projects with similar characteristics to the MLS group.
• Data: The analysis is based on 5,054 PPP projects from the PPI Database (1990–2010), of which 637 received MLS.
• Cancellation Definition: A PPP is considered canceled if the private sector exits before fulfilling the contract terms, removes all management, or ceases operation for 15% or more of the license period.
• Financial Impact: The total committed investments of canceled MLS projects between 1990 and 2010 were US$19.3 billion, which could have increased to US$28.6 billion without MLS, representing 19% more in investments.
• Country-Level Correlation: PPP projects are correlated at the country level, with 28.5% of the variation in MLS probability attributed to country-level factors.
• Policy Implications: The findings suggest that multilateral support plays a crucial role in ensuring the survival of long-term infrastructure PPP contracts. The support includes policy advice, capacity building, oversight, and risk mitigation, which may contribute to the success of PPPs beyond direct financial assistance.

Structure of the Paper

• Introduction: Sets the context of PPPs and the importance of multilateral support in infrastructure projects.
• Multilateral Support for Infrastructure PPPs: Defines MLS and discusses factors influencing the likelihood of receiving it.
• Data: Describes the dataset and criteria for defining PPPs and cancellation.
• Methodology: Explains the use of propensity score matching and multilevel modeling to estimate the effect of MLS on cancellation.
• Results: Presents the findings of the probabilistic model, matching process, and comparison of cancellation rates.
• Sensitivity Analysis: Applies alternative matching methods to confirm the robustness of the results.
• Conclusions and Next Steps: Highlights the positive impact of multilateral support on contract survival and suggests areas for further research.

Key Viewpoints

• Quasi-Experimental Design: The study uses a quasi-experimental approach to isolate the effect of multilateral support from other confounding variables.
• Selection Bias: Direct comparison of MLS and non-MLS projects is misleading due to systematic differences in project characteristics.
• Role of Multilateral Support: Beyond financial assistance, multilateral support may improve PPP performance through institutional capacity building, risk mitigation, and policy alignment.
• Regional and Sectoral Preferences: Multilateral development banks show preferences for certain regions and sectors, which may influence the likelihood of receiving MLS.
• Project Size: Larger PPP projects are more likely to receive multilateral support, indicating that scale is a key determinant.

Conclusion

The study concludes that multilateral support has a positive impact on the survival of long-term infrastructure PPP contracts. The use of propensity score matching and multilevel modeling helps to isolate the true effect of MLS, revealing that its absence would have led to a significantly higher cancellation rate. This highlights the importance of multilateral involvement in enhancing PPP performance and sustainability.