UW Aquatic & Fishery Sciences Quantitative Seminar
Suresh Sethi
School of Aquatic & Fishery Sciences, UW
Measuring risk in Alaskan fisheries
Abstract
A risk is the chance that something bad will happen. As fisheries management shifts focus to dual socio-ecological success, measures of catch and revenue risk provide a way to distill complex system dynamics into simple metrics that are meaningful for resource users. In this talk, I will introduce several quantitative measures of risk that can be calculated using commonly recorded catch and price data. These include semideviation, the probability of ruin, and conditional value at risk. Using a dataset of 90 Alaskan fisheries, nearly a complete census, I will: i) examine how risks vary across species, gear type, and region, ii) examine whether price supply and demand dynamics buffer revenue risks, and finally iii) highlight fishery characteristics associated with risks by estimating the relationship between environmental, biological, and socioeconomic attributes and risks. This estimation is challenging because risk data are bounded on the (0,1) interval and there may be nonlinear relationships between predictor variables and response data. To tackle this analysis, I will compare three different statistical models including linear regression with a logit transformation, beta regression, and random forests.
