Last week we had in the Department of Applied Economics of the Autonomous University of Barcelona our PhD workshop where our doctoral students presented their work in progress. For the first time we decided that the students themselves would be the discussants of their colleagues. I feared that there would be some degree of collusion, as the presenters themselves were allowed to propose the name of their discussants. However, they proved me wrong, and both presenters and discussants did a great job and we learned a lot from their ongoing research and the ensuing discussions. Most of the theses are empirical, and as always many issues revolve around the issue of the potential endogeneity of the explanatory variables. In the absence of pure experiments with human beings, ideally, one would like either to find exogenous instruments that explain the explanatory variable and are uncorrelated with the dependent variable, or perform natural or random field experiments. Along these lines, I warned (not only in this workshop but also on other evaluation sessions at the Master or PhD level) that I find an excessive tendency to call some policy reforms "natural experiments" when they are not such. Natural experiments should be exogenous events that are uncorrelated with the dependent variable but are correlated with the "problematic" explanatory variable. Policy reforms are hardly exogenous and are usually the result of long public debates and interest group battles. Natural experiments are policy changes that are uncorrelated with the phenomenon that one wants to explain, or natural phenomena such as climate or catastrophes. In other cases, I suggested caution with simplistic institutional explanations, such as those based on legal origins or other formal features. But these were only suggestions. At the PhD level, the hyerarchies between faculty and students fortunately start to vanish and we have the opportunity to learn together in what should be a cooperative profession.