Prediction Technologies and Decision-Making: A Model for Forecast-based Financing for Risk Mitigation

Abstract

Advances in prediction technologies provide opportunities for making informed decisions based on updated forecasts. However, forecast-based changes in actions may cause significant disruption within an organization when multiple decisions are interrelated and coordination is costly. This is particularly true for disaster risk management, which often involves multiple decisionmakers. A pre-agreed rule-based mechanism can allow forecast-contingent decisions with better coordination. In this paper, I examine one such mechanism-the Forecast Based Financing (FbF) program that provides trigger-based financing for early-actions before a disaster. I build a model to incorporate the existing structure of the FbF mechanism and provide comparative statics for the optimal financing and forecast trigger. My analyses provide three key insights. First, when forecast skill is low, more risk aversion leads to higher financing and lower forecast trigger levels. However, as forecast skill improves, the effect of risk aversion on the optimal financing becomes small and can even flip direction. Second, when forecast skill, likelihood of risky event, or the marginal benefit of actions increases, it is optimal to choose higher levels of financing as well as higher levels of trigger, and vice versa. Third, the benefits of a trigger-based FbF plan are comparable to those of a fully-contingent scenario.

Keywords: forecasts, predictions, risk management, adaptation, natural disasters, anticipatory action