This paper presents my thinking and concerns about formation of COVID-19 policy. Policy formation must cope with substantial uncertainties about the nature of the disease, the dynamics of transmission, and behavioral responses. Data uncertainties limit our knowledge of the past trajectory and current state of the pandemic. Data and modeling uncertainties limit our ability to predict the impacts of alternative policies. I explain why current epidemiological and macroeconomic modeling cannot deliver realistically optimal policy. I describe my recent work quantifying basic data uncertainties that make policy analysis difficult. I discuss approaches for policy choice under uncertainty and suggest adaptive policy diversification.

This article presents the first meta-analysis documenting the extent of publication selection biases in stated preference estimates of the value of a statistical life (VSL). Stated preference studies fail to overcome the publication biases that affect much of the VSL literature. Such biases account for approximately 90% of the mean value of published VSL estimates in this subset of the literature. The bias is greatest for the largest estimates, possibly because the high-income labor market and stated preference estimates from the USA serve as an anchor for the VSL in other higher income countries. Estimates from lower-income countries exhibit less bias but remain unreliable for benefit-cost analysis. Unlike labor market estimates of the VSL, there is no evidence that any subsample of VSL estimates is free of significant publication selection biases. Although stated preference studies often provide the most readily accessible country-specific VSL estimates, a preferable approach to monetizing mortality risk benefits is to draw on income-adjusted estimates from labor market studies in the USA that use Census of Fatal Occupational Injuries risk data. These estimates lack publication selection effects as well as the limitations that are endemic to stated preference methods.

We analyze the economic costs and benefits of “community-led total sanitation” (CLTS), a sanitation intervention that relies on community-level behavioral change, in a hypothetical rural region in sub-Saharan Africa with 200 villages and 100,000 people. The analysis incorporates data on the effectiveness of CLTS from recent randomized controlled trials and other evaluations. The net benefits of this intervention are estimated both with and without the inclusion of a positive health externality, that is, the additional reduction in diarrhea for an individual when a sufficient proportion of other individuals in the community construct and use latrines and thereby decrease the overall load of waterborne pathogens and fecal bacteria in the environment. We find that CLTS interventions would pass a benefit–cost test in many situations, but that outcomes are not as favorable as some previous studies suggest. The model results are sensitive to baseline conditions, including the value of time, income level used to calculate the value of a statistical life, discount rate, case fatality rate, diarrhea incidence, and time spent traveling to defecation sites. We conclude that many communities likely have economic investment opportunities that are more attractive than CLTS, and recommend careful economic analysis of CLTS in specific locations.

What is the benefit from obtaining more precise values of environmental or other public goods through surveys or other information gathering? In the value of information (VOI) problem studied here, a buyer who wishes to preserve a resource sets a price to offer a seller without knowing precisely its protection value, B, nor its value to the seller, V. The VOI from more precise information about B is important for environmental and natural resource valuation, but is typically not quantified nor compared to valuation costs. More precise environmental values reduce the frequency of two types of mistakes (protecting the resource when it should not be; and not protecting it when it should), and increases ex ante welfare. We apply our analysis to Amazon rainforest protection, focusing on the “value of perfect information,” VOPI, which, we show through simulations, typically exceeds realistic valuation costs, justifying significant valuation expenditures. VOPI also depends on the nature of buyer–seller interactions, and takes its highest value when the buyer has full concern for the seller’s outcome. Our paper proposes and prepares the base for a new, needed, field in applied welfare economics, the “benefit–cost analysis of public-good valuation studies.”

Since its introduction to the field of environmental and natural resource economics in the late 1960s, existence value has faced several critiques from economists, psychologists, and philosophers. Critics have taken aim at the notion’s conceptual ambiguity and lack of connection to observable behavior, its incompatibility with cognitive processes and its sensitivity to cognitive biases, and ethical shortcomings in applying existence values to environmental decisionmaking. Unlike some critiques of existence value that draw on cognitive and ethical frameworks for decisionmaking fundamentally at odds with stated preference methods and benefit–cost analysis (BCA), this paper takes as given the use and adequacy of both. It focuses on challenges to existence value per se, with respect to the ability of existence value estimates to contribute to benefit–cost analyses in a way that is consistent with qualities of BCA that its proponents value: the objectivity, commensurability, and moral salience of the values analyzed. In light of the challenges, inclusion of existence value in benefit–cost analyses is found to inevitably compromise the quality of the BCA with respect to each criterion.

Some issues in the application of benefit–cost analysis (BCA) remain contentious. Although a strong conceptual case can be made for taking account of the marginal excess tax burden (METB) in conducting BCAs, it is usually excluded. Although a strong conceptual case can be made that BCA should not include distributional values, some analysts continue to advocate doing so. We discuss the cases for inclusion of the METB and the exclusion of distributional weights from what we refer to as “core” BCA, which we argue should be preserved as a protocol for assessing allocative efficiency. These issues are topical because a recent article in this journal recommends ignoring the METB on the grounds that desirable distributional effects offset its cost. We challenge the logic of this article and explain why it may encourage inefficient policies.

Research and development (R&D) planners in homeland security agencies would like to be able to prioritize investments in projects based on costs versus future safety and security benefits. While costs are often readily available, estimates of safety and security benefits are fraught with uncertainty. To address these challenges, a benefit–cost model of technological change is adapted to the homeland security context. Data are sparse; therefore, estimation is facilitated by developing a familiar linear welfare model using derivatives of cost and risk reduction functions to estimate areas of costs and benefits. The theoretical model is applied to two homeland security projects involving airport patrols and the assignment of U.S. federal air marshals to international flights. Retrospective data are available for most periods. Welfare-based rates of return are reported for the two cases, each of which is estimated to return large present value net benefits. Extensive sensitivity and Monte Carlo simulation explores uncertainties. Two important findings are that (i) given the rationality assumption, relative increases in security levels can be valued, even if the absolute level of security is not known; and (ii) large uncertainties about risk reduction exist but can be bounded by parametric sensitivity and uncertainty analysis.

Nepal has suffered from the worst electricity shortages in South Asia. This study is an attempt to measure the willingness to pay for an improved service using a model of revealed preference. Respondents are asked about the actions they are taking to reduce the impact on their household or business of scheduled and unscheduled outages and more stable voltage. We estimate the averting expenditures that were being incurred to compensate for the lack of reliability of the electricity service. The estimated cost of the averting actions as a percentage of the electricity bills is 53 % for households, 47 % for small businesses, 46 % for medium businesses, and 35 % for large businesses. Based on the estimations, we find that in 2017 the annual benefit from improving the reliability of the electricity service would be approximately US$ 188 million with a present value over 20 years of US$ 1.6 billion.