My research fields are in Political Economy, Development Economics, Conflict and Machine Learning. I am particularly interested in how instability arises and how it affects long run economic development. Most of my research uses simple difference-in-difference estimates but I am increasingly interested in adapting machine learning methods for social science. This has two elements: the use for feature of extraction from text and images and the use of supervised learning for forecasting and nowcasting.
My research has been published in top journals in political science and economics like the American Economic Review, the American Review of Political Science, the Journal of the European Economic Association, the American Journal of Economics: Macro and the Journal of Public Economics.
My research has no particular geographic focus but I am particularly interested in countries in Latin America, Northern Africa and the Middle Easter because of their cultural and geographic proximity to Spain.
Much of my academic research uses simple theory to make sense of the data. I see this as no contradiction to using machine learning as I believe that the priors we put into the statistical models used in the empirical analysis carry an element of theory which, in fact, that help in social science. The perfect example is the Latent Dirichlet Allocation which we use in our analysis of civil war and which is literally a model of writing text which reveals underlying semantic structures.
I strongly believe in the idea that structural estimation, theory, natural experiments, case studies, RCTs and reduced form work should complement each other to help us understand society. But it is not always clear whether we focus on the right mechanisms and channels. This is where I think forecasting has a role to play even if the ultimate goal is to identify causal mechanisms.