DSE Curriculum

There is a strong symbiotic relationship between computational economics and structural econometrics due to the fact that many if not most structural econometric models do not have convenient analytic close-form expressions or solutions, and thus require substantial amount of programming skills to implement. A number of significant advancements in econometric methods and computational methods in econometrics have been inspired and come from researchers seeking to overcome computational roadblocks encountered in empirical work, including ways to circumvent the well known “curse of dimensionality” of dynamic programming, and the multiplicity of equilibria in dynamic games.

The DSE summer schools are designed to provide instruction in the state of the art methods by identifying leading scholars who are experts and innovators in some of the leading new methods and techniques in this rapidly evolving area. This degree of specialized instruction is difficult to provide in most economics departments due to the rapid pace of new developments in multiple areas.

Because of the rapidly evolving state of knowledge, the curriculum for the DSE is designed to evolve over time and specific topics will change from year to year depending on the host and theme of each meeting. But there will be a core curriculum that consists in instruction in fundamental skills in computer programming and computational economics as well as covering fundamentals of dynamic programming, computational game theory, and econometrics that are pre-requisites for most work in structural econometrics. The idea is to make the summer schools reasonably self-contained, but to build on the fact that most graduate students will have had prior exposure to dynamic programming and econometrics.

Our definition of “structural econometrics” is intentionally flexible to accommodate a range of different perspectives: we view as “structural” any empirical work that attempts to develop, empirically implement, and test economic theories. We are quite friendly to faculty and students who have a “reduced-form” orientation and have an interesting topic or data but are seeking the skills to develop economic models to improve their understanding of phenomena. We are as interested in the analysis of observational/survey data such as we are in laboratory and field experiments, and we are not wedded to particular fields of economics such as industrial organization. In particular, we do not see structural econometrics as being synonymous with rational models of economic behavior, but instead are entirely welcoming of alternative viewpoints such as embodied in a growing literature on “structural behavioral economics” (the title of a chapter in the Handbook of Behavioral Economics by Stefano DellaVigna.

The DSE summer schools are designed to promote an appreciation for the value of estimation methods justified by rigorous econometric theory, but the focus is primarily on helping students develop the skills for processing data, programming estimators, and providing hands-on instruction of how to apply contemporary numerical methods to the types of problems and topics they are interested in. We do value the role of econometric theory in helping to motivate and design improved estimators and DSE workshops do involve leading econometric theorists to help provide this guidance: examples include lectures by Victor Chernozhukov and Sanjog Misra on machine learing in the first DSE summer school in Copenhagen. Wherever possible we try to expose students to the best economic and econometric theory, but also to show them how these theories and methods can be applied in practice in empirical applications.