This project analyzes development finance at the project level to answer unresolved questions in the literatures on international relations, comparative political economy, foreign aid, and multilateral lending. The investigators apply a variety of principal-agent (PA) models on both the donor and recipient sides to explain the amount, type, and distribution of overseas assistance. They explore how changing preferences in the donor countries have altered the priorities of development agencies and multilateral institutions. These underlying changes in principal preferences are mediated by collective-action problems among the principals, information asymmetries vis-a-vis their bureaucratic agents, and delegation problems in recipient countries. The researchers test these hypotheses alongside conventional alternatives. The scholarly debates raise three overarching questions: (1) why and how do donors provide aid to particular countries? (2) What drives recipient countries in their pursuit and consumption of the assistance? And, (3) what are the effects of development finance - both intended and unintended? The completion of this project-level aid (PLAID) database will help answer these questions by allowing systematic tests of a large number of core hypotheses from the international relations and development-finance literatures. To date, such inclusive specifications and general tests have been lacking. Theory in this literature has developed in the context of plausibility probes and qualitative case studies in small-n samples. Quantitative work, conversely, has been largely descriptive and employs incomplete and biased data. Theoretically informed hypotheses are rare in extant quantitative work. Hence, accumulation of knowledge in this field has been limited. Hypotheses have not been systematically tested with data gathered at the appropriate level of analysis: at the level of development projects. Instead, scholars have aggregated incorrect and biased sums of aid and loans at the sectoral or country level. The reason: most quantitative studies are based on OECD data, for which the investigators have identified persistent and systematic errors. Broader Value: The PLAID dataset will provide detailed coding of key variables for over 500,000 development projects by bilateral and multilateral foreign aid agencies and concessional loans by international financial institutions. It will thus allow scholars (and policymakers), for the first time, to examine the universe of development finance data at the level of each project. With the data we explore the effects of international and domestic delegation on development finance. To evaluate the PA hypotheses and their alternatives, the researchers employ a variety of limited dependent variable regression techniques, including Tobit, Cragg, and Heckman estimation. With these tools they can test the effects of core independent variables-principals' preferences, collective-action problems among donors, information asymmetries with aid agencies, and recipient credibility and demands-simultaneously on both the receipt of a project and the project's amount. These quantitative tools are supplemented by ongoing qualitative research within both recipient and donor countries.
