UC Berkeley

This dissertation evaluates risk management for disasters where the losses unfold over time, with two key applications: environmental accidents and exceptional losses in crop production. Both applications evaluate policy against goals of equity and efficiency, but the environmental policy application is a normative analysis, while the production risk application is a positive analysis.

Environmental accidents are stochastic externalities - they impose a social cost not accounted for by whichever business constitutes their source. In many cases, adequate regulation does not exist. We show that standard pollution regulations must be adjusted for accidents, because random triggers and unobservable actions lead to a moral hazard problem. We identify three policies that lead to the optimal solution when both care and cleanup are considered: strict liability, a stochastic subsidy, and a mandatory mutual insurance scheme.

The subsidy policy may be very costly to taxpayers, especially when prevention affects the probability of accident occurrence, and strict liability may be excessively draconian; polluters are also victims and liabilities must exist regardless of adherence to professional standards of care. Thus, we propose a new policy of liability risk-pooling, which demonstrates a role for insurance policy among accidentally polluting firms, even when such firms are profit-maximizers (that is, they are risk neutral). The new policy also generates, in expectation, the most equitable distribution of resources among polluting firms while preserving efficiency - in this sense it is the stochastic equivalent of a system of tradable pollution permits.

Our second application addresses production risk in US crop production and the impact of the SURE disaster support program in the 2008 Farm Act. Supplemental disaster insurance is nested insurance, an insurance policy on top of another insurance policy, which may actually increase riskiness in the distribution of outcomes. Thus, we evaluate whether, and under what circumstances, nested insurance actually provides risk management. We develop a comprehensive economic theory of nested insurance, and provide new insight into the concept of targeted subsidies, which use kinked insurance pricing to limit variation in farmers' coverage purchasing decisions.

The theoretical evaluation is supported by an in-depth simulation analysis, which simulates the joint price-yield distribution for dramatically different risk profiles of Illinois corn and South Dakota wheat. Using a time series of county- and national-level yields and expected and realized commodity prices, we construct a simulated revenue distribution over which a representative farmer can maximize expected utility. We show that disaster policies may distort acreage and insurance choices, but that these distortions are likely small. Distortions are largest for the primary beneficiaries of the SURE program, the most risk-neutral farmers, who are least in need of risk management.

Both applications take a classical, welfare economic approach to policy. In the environmental case, considerations of equity play a larger role as a result of uncertainty, whereas in the crop insurance case, nested insurance is shown to behave more like a stochastic subsidy than actual risk management. Overall, we have shown that managing the risk from disasters across varying economic agents can lead to dramatic distributional implications. When more than one efficient policy is available, then the distributional characteristics of policies will be the deciding factor. However, when equity is the objective, poorly designed disaster policies can backfire and be of little use to those who need them most.