International organizations like development banks implement development programs in the world?s poorest places. Evaluation information about past programs can assist development banks in choosing the types of projects that will achieve the best results given scarce development financing. However, despite the commitment that development banks have made to evaluate their programs, there is very little evidence available about when and why evaluation information is actually used to improve decision-making. This research examines the factors that promote or impede the use of evaluation information in resource allocation decisions. Environmental programming has been a priority for development banks since the early-1990s and the environmental outcome of projects has been one of the most consistently measured types of evaluation information across many different programming sectors. As a consequence, the research focuses on environmentally-targeted and environmentally-risky projects. Project evaluations completed by four development banks since 1990 will be coded for several measures of environmental performance. This data will be used to model whether development banks decrease environmentally-risky financing to recipient countries and project types that failed to achieve environmental goals in the past and increase environmentally-targeted financing in cases where past efforts have been successful. These models will be combined with data from development bank staff interviews to produce one of the first rigorous tests of the factors that promote or impede the use of evaluation information in development bank decision-making. This research adds to existing knowledge about effective evaluation use in large organizations by going beyond self-reporting methods and actually verifying when and why evaluation impacts decision outcomes. Given the large sums of money being spent on both development assistance and evaluation, it is vital to understand what external political factors and internal organizational procedures support decision-making that fully utilizes evaluation results.
This research investigated when and why multilateral development banks respond to their environmental performance as measured in evaluations. The multilateral development banks, which include the World Bank, Asian Development Bank, African Development Bank, and Inter-American Development Bank, manage the majority of environmentally focused development assistance and have been leading organizations in establishing policies to mitigate the negative environmental damages of infrastructure projects. In recent years, development specialists have emphasized the need to allocate development assistance according to past recipient performance. By doing so, development organizations ensure that scarce resources are used efficiently and allocated to achieve a maximum impact. However, it is unknown whether evaluation can support this type of performance-based decision-making at the multilateral development banks as intended. To address this question, this research collected environmental performance information from every available evaluation completed by four multilateral development banks between 1990-2008. This included 960 project evaluations and 174 country program evaluations. The resulting database contains the most comprehensive information currently available about the environmental performance of international organizations and will be publicly released in the coming months. This information was used to model the extent to which environmentally focused development assistance is allocated to borrowing countries with good performance records, thereby maximizing impact. In addition, the information was used to test whether development assistance with negative environmental risks was restricted to borrowing countries with good records implementing environmental mitigation activities. The results of these models show that when borrowers drive their own lending portfolios, they tend to select project types where they will be successful. In contrast, for "donor-driven" investment areas like climate change, evaluation does not appear to support performance-based decision-making. For infrastructure projects, past performance mitigating environmental risks is a strong determinant of future lending. Taken together, these results indicate that evaluation itself does not support performance-based decision-making. Instead, evaluation must be coupled with operational policies that incentivize the use of past performance information within organizations. To investigate how these incentives can be created within large organizations, I conducted 54 interviews at the headquarters of the development banks considered. The results of this research will be written up in several scholarly journal articles. In addition, the co-PI presented several seminars at the multilateral development banks about designing evaluation to improve performance.