Probability Forecasting, Quantitative Decision Modeling, and the Economic Value of Forecasts (Invited Presentation)

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 5 February 2014: 11:30 AM
Room C205 (The Georgia World Congress Center )
Daniel S. Wilks, Cornell University, Ithaca, NY
Manuscript (1.2 MB)

In the course of inviting me to contribute to this symposium honoring Ed Epstein, Bob Glahn wrote: "Using probability forecasts in decision making is what its all about, right? Of course, Ed knew this . . ." Ultimately, the economic justification for both research and operations in the atmospheric sciences rests on the suitability and use of forecasts to support decision making by forecast users. Probability forecasts impart more value than do their nonprobabilistic counterparts, and the probabilistic format is essential for separating the role of the forecaster from that of the decision makers who use the forecasts. Both of these assertions can be demonstrated using quantitative decision-making models, which have roots in Bayesian statistics, and can be used to compute potential (assuming optimal use) economic value of forecasts in particular settings. This talk will review the structure of quantitative decision models in relation to forecasts, the calculation of economic value of forecasts using these models, and aspects of the relationship between forecast value and measures of forecast quality.