Seminars at the Faculty of Informatics

The Faculty of Informatics is pleased to announce a seminar given by Ken Judd

DATE: Thursday, November 6th, 2014
PLACE: USI Lugano Campus, room A22, Red building (Via G. Buffi 13)
TIME: 10.30

ABSTRACT:
We develop a computational model combining both economic risk and risky climate change. We solve the model for a large range of parameter vectors consistent with the latest empirical estimates of long-run consumption risk and utility function parameters, and consensus opinions of how warming will affect economic productivity. We find that economic risk substantially increases the expected social costs of carbon. We also find that the log of social costs roughly follows a random walk with high variance, plausibly rising to $1000 per ton of carbon within this century. We use massive parallelism (10K - 70K cores) to solve nine-dimensional dynamic programming problems at yearly time periods over 200 years, followed by ten-year time periods for the next
400 years. The continuous-time version of this problem would be a Hamilton-Jacobi-Bellman problem. We use verification methods to demonstrate the high accuracy of our solutions for decision rules.
Moreover, the tools used to solve this problem can be directly applied to a vast range of economic problems. This work shows that it is now possible to solve problems that are orders of magnitude larger than those solved today in economics.

BIO:
Prof. Kenneth L. Judd, Ph.D., is a computational economist at Stanford University and the Paul H. Bauer Senior Fellow at the Hoover Institution in Stanford. He received his PhD in economics from the University of Wisconsin in 1980. He is perhaps best known as the author of Numerical Methods in Economics, and he is also among the editors of the Handbook of Computational Economics and of the Journal of Economic Dynamics and Control. He is currently a visiting professor at the University of Zurich. He has published papers in mathematical economics, economic theory, public finance, and currently focuses on developing numerical methods for applications in economics.

HOST: Prof. Olaf Schenk