webinar register page

Webinar banner
Christopher Sims - Large Parameter Spaces and Weighted Data: A Bayesian Perspective
Bayesian analysis can suggest ignoring sampling weights, even in contexts where popular estimation methods like Horvitz-Thompson or “doubly robust” estimates do use the weights. Weights are sometimes treated in frequentist theory as “parameters”, i.e. as non-random, and thus often lead to inference in large parameter spaces. Bayesian analysis can produce paradoxes in large parameter spaces. So is Horvitz-Thompson or doubly robust estimation a mistake? Are Bayesian approaches a dead end in large parameter spaces? The answers to these questions are not simple yes or no, as we will see in some examples.

Feb 28, 2022 05:00 PM in Paris

Webinar logo
* Required information
Loading

Speakers

Christopher Sims
Professor of Economics @Princeton University
Christopher Sims is Professor of Economics, emeritus, at Princeton University, where he has been on the faculty since 1999. He received his PhD from Harvard University in 1968 and was on the economics faculty at Harvard, the University of Minnesota and Yale. He is a member of the National Academy of Sciences and has served as a visiting scholar at several US Federal Reserve Banks and at the Board of Governors of the Federal Reserve. With Thomas J. Sargent he won the 2011 Economics Nobel Prize. His research has dealt with econometric time series methods, estimation of monetary policy behavior and the effects of monetary policy, and the theory of price level determination. He is known for promoting the usefulness of loosely structured models (VAR's and SVAR's), for advocating a Bayesian perspective on econometric inference, for emphasizing the importance of fiscal policy in determining the path of inflation, and suggesting the application of information theory to economics.