Sponsoring Section/Society: ASA-SBSS
Session Slot: 8:30-10:20 Monday
Estimated Audience Size: 100
AudioVisual Request: Overhead Projector, Slide Projector
Session Title: Bayesian Analysis and Public and Private Policy Making
Theme Session: Yes
Applied Session: No
Session Organizer: Zellner, Arnold University of Chicago
Address: Arnold Zellner Grad. Sch. of Business U. of Chicago 1101 E. 58 St. Chicago, IL 60637
Session Timing: 110 minutes total (Sorry about format):
Opening Remarks by Chair - 0 minutes First Speaker - 30 minutes Second Speaker - 30 minutes Third Speaker - 30 minutes First Discussant - 10 minutes Second Discussant - 10 minutes Floor Discusion - 0 minutes
Session Chair: Zellner, Arnold University of Chicago
Address: Arnold Zellner Graduate School of Business U. of Chicago 1101 E. 58 St. Chicago, IL 60637
1. Mutual and Pension Funds Management: Beating the Markets using a Global Bayesian Investment Strategy
Quintana, Jose Mario, CDC, Investments, New York
Address: Jose Mario Quintana CDC Investment Management Corp. 9 West 57th St., New York, NY 10019
Putnam, Bluford H., CDC, Investments, New York
Abstract: There is a trend in mutual and pension funds investments toward indexing (i.e., to buy and hold "the market"). This tendency is fueled by the apparent confirmation of the efficient market hypothesis given the inability of the majority of active U.S. equity managers to beat the S&P 500 index in recent years. However, the concept of an efficient market is a relative one. In this presentation we will show a global Bayesian strategy that can generate attractive risk-adjusted excess returns by exploiting inefficiencies in the broad global markets. Furthermore, these excess returns are transportable, allowing the strategy to outperform any index, including top-performing indices such as the S&P 500.
2. What to Do When the Crystal Ball is Cloudy: Conditional and Unconditional Forecasting in Iowa
Whiteman, Charles, University of Iowa
Address: Charles H. Whiteman Pioneer Hi-Bred Professor of Financial Economics Chair, Department of Economics W210 PBAB The University of Iowa Iowa City, IA 52242
Otrok, Christopher, University of Iowa
Abstract: Since 1990, the predictions of state tax revenues in Iowa which provide the foundation for official forecasts have been made using Bayesian methods under asymmetric linear loss. The procedures utilize vector autoregressions and uninformative priors; predictive distributions are calculated using the Monte Carlo method. Unconditional forecasts have been remarkably accurate. During policy discussions surrounding the 1997 income tax cut, conditional forecasts utilizing the methods were used to assess the usefulness of outside consultants' ``structural'' model estimates of the effects of the cut. This effort suggested that the uncertainty associated with the structural, behavioral effects of the tax cut was small relative to the inherent "reduced form" uncertainty in revenue forecasting.
3. Why use hierarchical models to assess medical profiles?
Morris, Carl, Harvard University
Address: Carl Morris Dept. of Statistics Harvard U. Cambridge, MA 02138
Christiansen, Cindy L., Harvard University
Abstract: Hierarchical models are being used for quality assessment more and more frequently, including for profiling medical units. In what sense are they better than simpler methods, and can we decide before fitting them whether they will produce noticeably different results? What pitfalls must be watched for, and how? Do we need to describe multi-level models and present their results more simply in order to widen their acceptance for regulatory use? The talk addresses these questions, partly through examples involving hospital comparisons.
Discussant: Polson, Nicholas University of Chicago
Address: Nicholas Polson Grad. Sch. of Business Univ. of Chicago 1101 E. 58th St. Chicago, IL 60637
Discussant: Dorfman, Jeffrey University of Georgia
Address: Jeffrey H. Dorfman Associate Professor of Ag. & Applied Economics 315 Conner Hall University of Georgia Athens, GA 30602-7509
List of speakers who are nonmembers: None