Published online by Cambridge University Press: 02 October 2017
Development projects like schools and latrines are popular with politicians and voters alike, yet many developing countries are littered with half-finished projects that were abandoned mid-construction. Using an original database of over 14,000 small development projects in Ghana, I estimate that one-third of projects that start are never completed, consuming nearly one-fifth of all local government investment. I develop a theory of project noncompletion as the outcome of a dynamically inconsistent collective choice process among political actors facing commitment problems in contexts of limited resources. I find evidence consistent with key predictions of this theory, but inconsistent with alternative explanations based on corruption or clientelism. I show that fiscal institutions can increase completion rates by mitigating the operational consequences of these collective choice failures. These findings have theoretical and methodological implications for distributive politics, the design of intergovernmental transfers and aid, and the development of state capacity.
This project has benefited from data and discussions with individuals from numerous institutions in Ghana, including the National Development Planning Commission, Ministry of Local Government and Rural Development, Local Government Service Secretariat, Ministry of Finance, District Assemblies Common Fund Administrator, Ghana Audit Service, World Bank, Ministry of Education, and GETFund Secretariat. However, the findings and opinions herein should not be attributed to any of these institutions. I am grateful for comments from Nana Agyekum-Dwamena, Ben Ansell, Daniel Berger, Catherine Boone, Christian Breunig, Stefano Caria, Ali Cirone, Darin Christensen, Patrick Dunleavy, Douglas Gollin, Nahomi Ichino, Ryan Jablonski, Julien Labonne, Noah Nathan, Pia Raffler, Imran Rasul, Dan Rogger, Anisha Sharma, Joachim Wehner, Erik Wibbels, Stephane Wolton, three anonymous reviewers, and workshop participants at the LSE, Ghana CSTC, ODI, DFID, EPSA, APSA, PacDev, Princeton, Georgetown, IMF, ABCDE, MPSA, WGAPE, and the Ghana Ministry of Finance, and to Nahomi Ichino and Erik Wibbels for sharing data. This research was funded by International Growth Centre grant 1-VRG-VGHA-VXXXX-89105. Allan Kasapa, Joseph Napen, and Abdul-Kadir Mumuni provided excellent research assistance. Any remaining errors are my own.