A new EBRD Working Paper (number 253)
This paper uses a novel dataset on investments considered by the European Bank for Reconstruction and Development to examine project selection and project design at a multilateral development bank which pursues a combination of financial and environmental, social or governance (ESG) objectives. The analysis exploits the fact that details of the projects, including assessments of their expected ESG impact, are recorded at least twice: when a concept is first reviewed by the investment committee and when the final particulars of a project are approved, with around 55 per cent of concepts translating into signed deals. We show that projects are simultaneously selected on the quality of credit and ESG impact, with ESG characteristics of a project having a greater impact on the probability of a project being signed in the case of commercially riskier investments. At the median, a weakening of risk profile of a project by 0.4 of a standard deviation is offset by a strengthening of the expected ESG impact by 0.6 a standard deviation, with unchanged probability of a project being implemented. We further use machine-learning-based analysis of project review documents to show that ESG impact of some projects is strengthened between approval of the project concept and project signing.