It was in 2016 that my co-author Tobias Schmidt and I reflected about changes in power generation financing in developing countries. We build on earlier research which showed that in developed countries, state investment banks like the German KfW played an important role to “mainstream” the financing of new renewables. In developing countries, though, such actors hardly exist. At the same time, we knew from our practical work on infrastructure in developing countries how multilateral development banks (MDBs) are instrumental in financing capital investments. So could it be that MDBs play a decisive role in choosing which type of power plants gets financed, and thereby deciding if countries develop on a high- or low-carbon trajectory?
As finance is a highly data-driven business, we expected to find answers to our questions in data, to be revealed through regression analyses. Surprisingly, though, we noticed that data with the breadth and depths of information that would be needed for a proper assessment did not exist: For instance, the MDB’s climate finance reporting is to aggregate, and commercial databases such as Bloomberg New Energy Finance were not complete enough (plus, they only cover renewables, not fossil fuel-based power plants). Hence, we took the challenge to compile a new database ourselves. A group of graduate students hand-coded project-level information from the MDBs, and on the way learned how to deal with incomplete or inconsistent information.
In parallel, we started discussing the project selection process with MDB officials, both in formal expert interviews and in informal discussions at various events. Importantly, we noticed, the role taken by MDB experts and the influence they exert on technology decisions differs strongly between the public and private sector branch of a bank, much more between branches than between banks it seemed. This insight motivated us to differentiate also in the quantitative analysis between public- and private sector branches, which highlighted how much the renewables share differs between the branches. It was this detailed breakup that led to the policy implications as discussed in the paper.
Eventually, the article we wrote turned out quite different than initially planned, not as a regression analysis but instead as a descriptive analysis of new and enriched data. The work shows how social scientists can contribute to address societal challenges in a different way than identifying causal relationships (which we normally spend much of our time on, and also might do in future research with the MDB data). Presenting new data in a meaningful way, informed by subject matter expertise from qualitative research, is how here we aim to inform policymakers. In doing so, the article shows how how complementing data work with parallel qualitative interviews can help to focus quantitative analyses on aspects that otherwise would be missed.