Does it makes sense for a scientist to restrict her hypotheses in such a way that she obtains closed form solutions for her models, even if these hypotheses are clearly at odds with the basic facts the models are supposed to explain? Natural scientists have answered this question several decades ago with a clear “no”, gaining in exchange for the opportunity to build a very sophisticated theory with unparalleled explanatory and predictive power of highly complex phenomena. In order to go beyond the limits of analytical tractability, it is possible to resort to a vast kit of computational methods: numerical root finding, simulated non linear dynamic systems, agent based simulations, etc. Although many of these have already become of common use in econometrics, the theoretical implications of computational methods for economics still have to be widely accepted, especially when they involve topics such as non linearity, agent heterogeneity, bounded rationality, learning and interaction. The purpose of the course is to provide an introduction to scientific programming and computational methods in economics, combining a “hands on” approach with a theoretic oriented perspective.
A precise reading list will be available on http://e-l.unifi.it/ at the beginning of the course. It will contain selected chapters from:
Cars Hommes (2013) Behavioral Rationality and Heterogeneous Expectations in Complex Economic Systems, Cambridge.
AA VV. (2016) Economics with Heterogeneous Interacting Agents, Springer.
Thomas J. Sargent, John Stachursky (2017), Quantitative Economic, https://lectures.quantecon.org/index.html.