2017年-世界发展银行全球_Do_Countries_Learn_from_Experience_in_Infrastructure_PPP____PPP_Practice_and_Contract_Cancellation_24页_1mb

报告摘要

Summary of "Do Countries Learn from Experience in Infrastructure PPP?"

Core Content

This working paper investigates whether countries learn from their experience with public-private partnerships (PPPs) to reduce the likelihood of contract cancellation in future infrastructure projects. The study focuses on the impact of country-level PPP experience on the probability of premature contract cancellation, which represents the most extreme form of PPP failure.

Main Findings

Learning from Experience: The paper confirms that country experience with PPPs reduces the probability of contract cancellation. However, the benefits of learning are typically concentrated in the initial stages of a country's PPP journey.
Sectoral Differences: The probability of contract cancellation varies across sectors. The energy sector has the lowest cancellation rate at 3.2%, while the ICT sector has the highest at 15.4%. Water and sewerage projects have a higher cancellation rate than energy and transport projects.
Project Type: Greenfield projects (new infrastructure) have a cancellation rate of 4.3%, which is lower than brownfield concession projects (7.1%) and management and lease contracts (13.7%).
Regional Variations: Africa has the highest cancellation rate (9.6%), which is 70% higher than in East Asia and the Pacific (EAP) and seven times higher than in South Asia (SAR).
Economic Impact: An estimated $1.5 billion per year could have been saved through interventions aimed at reducing cancellations in less experienced countries (those with fewer than 23 prior PPPs).

Methodology

Data Source: The study uses data from the World Bank's Private Participation in Infrastructure (PPI) Project Database, which includes 3,400 PPP projects that reached financial closure before 2011 across 139 countries.
Statistical Approach: Mixed-effects probit regression models are used, combined with linear and cubic spline models to capture the non-linear relationship between PPP experience and cancellation rates.
Key Variables:
- PPP Country Experience: Defined as the number of PPPs that reached financial closure in the same country within the past 10 years.
- Project Status: Whether the project is active, canceled, or concluded.
- Sector and Region: Categorization of projects based on infrastructure sectors and geographical regions.

Key Insights

Asymptotic Relationship: The relationship between PPP experience and cancellation rates is asymptotic, meaning that the rate of reduction in cancellation is most significant in the early stages of PPP implementation.
Learning Curve: The benefits of learning are concentrated in the first few PPP deals. For instance, the cancellation rate drops from 22% to nearly 8% when a country has closed at least 50 PPP deals.
Importance of Demonstration Projects: The first PPPs, or demonstration projects, are crucial for learning and building a supportive PPP framework. These projects help governments develop the necessary skills and institutional mechanisms to manage future PPPs effectively.
Multilevel Models: These models are essential for capturing the clustered nature of PPP outcomes at the country level, which can otherwise lead to biased statistical inferences.

Conclusion

The study underscores the importance of experience in improving the success of PPPs, particularly in reducing the risk of premature contract cancellation. While the benefits of learning are most pronounced in the initial stages, the paper highlights the need for continued support and institutional development to sustain improvements in PPP performance over time. The findings also emphasize the value of sector-specific and regional analyses in understanding the complexities of PPP implementation and failure.

Key Terms

PPP: Public-Private Partnership
Mixed-effect probit model: A statistical model used to analyze the probability of binary outcomes (e.g., cancellation or not) while accounting for clustering at the country level.
Linear spline: A piecewise linear function used to model the relationship between PPP experience and cancellation rates.
Cubic spline: A more flexible model that allows for smooth, curved relationships between PPP experience and cancellation rates.
Thick learning: A concept that emphasizes learning across program, process, and political dimensions in complex organizational arrangements.