大规模学习：基础设施、援助有效性和世界银行绩效

Policy Research Working Paper Learning at Scale Infrastructure, Aid Effectiveness, and World Bank Performance Somik LallXinyi SuMaria Vagliasindi Policy Research Working Paper11412 Abstract This paper reexamines aid effectiveness using ex-post eco-nomic rates of return from 2,500 World Bank–financedinfrastructure projects over six decades. Three facts emerge.First, average value creation is high: the mean economicrate of return is 24 percent (median 18.5 percent). Second,returns rise over time: after relative stability through themid-1980s, economic rates of return increased from the In such countries, high-quality World Bank performanceis associated with about a 12-point increase in the averageeconomic rate of return. A plausible mechanism is talentdeployment. Higher-quality task team leaders are dispro-portionately assigned to the most challenging contexts andlinked to faster time-to-effectiveness and higher economicrates of return, despite longer, more intensive implementa- This paper is a product of the Strategy Group, Development Economics. It is part of a larger effort by the World Bank toprovide open access to its research and make a contribution to development policy discussions around the world. PolicyResearch Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contactedat slall1@worldbank.org, xs308@cam.ac.uk, and mvagliasindi@worldbank.org. A verified reproducibility package for this The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about developmentissues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry thenames of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely thoseof the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and Learning at Scale Infrastructure, Aid Effectiveness, and World Bank Performance* Somik Lall, Xinyi Su and Maria Vagliasindi JEL Classification:D02 D61 F35 H43 L5 O22 Keywords:Economic rates of return,Infrastructure, aid effectiveness, FCV, low-income countries, World Bank, *The authors are respectively Director of Strategy for Development Economics,World Bank Group(slall1@worldbank.org); doctoral student at Cambridge University (xs308@cam.ac.uk). and Lead Economist in theInfrastructure Chief Economist office, World Bank Group (mvagliasindi@worldbank.org). This paper is a product ofa broader WBG-Hoover Project “Learning from Lending”, complementing recent analysis of the IFC portfolio, Chariet al., 2025. This paper has benefited from discussions with Indermit Gill, Peter Henry, Dan Rogger, Arianna Legovini, 1.Introduction This paper reassesses aid effectiveness using six decades of World Bank infrastructure lending throughroughly 2,500 projects across nearly 160 developing countries, evaluated through their ex post economicrates of return (ERRs). Moving beyond aggregate, country-level metrics and qualitative assessments, we The central theme is staff allocation. High-quality Task Team Leaders (TTLs) and managers—especiallythose experienced in low-capacity settings—are associated with higher ERRs and better outcomes inweaker-governance countries. Over time, the World Bank has improved portfolio performance in part bydeploying its strongest staff to the most challenging contexts, though career management for TTLs has not Our work builds on Denizer et al. (2013) and Bulman et al. (2017), who find that country characteristicsexplain only 10–25 percent of project performance variation. Faster growth and stronger policies correlatewith better outcomes at the country level, while shorter project duration and TTL track record matter at theproject level. Thus, while country selectivity “pays”, most actionable leverage is at the project level. We contribute along three dimensions. First, we expand the dataset in time (evaluations through June 2025, versus 2011) and scope (TTL CVs anddisbursement histories), focusing on the more homogeneous sample of infrastructure projects. Second, we center the analysis on ERRs rather than the qualitative ratings of the World Bank's IndependentEvaluation Group (IEG), mitigating concerns about measurement artifacts and collinearity between IEG Third, the ERR lens yields new findings: satisfactory World Bank performance is associated with materiallyhigher returns in countries with weak governance (low CPIA); targeting aid to such settings can pay offprecisely because the World Bank deploys its most capable TTLs—both newly hired and provenperformers—to these contexts. Yet incentive systems remain imperfect: many high-performing TTLs This study is also the first to apply an instrumental variable strategy to isolate the causal effect of ERRs onIEG implementation ratings, produci