Accounting for variation in exogenous shocks in economic impact modeling

To increase confidence in results from a computable general equilibrium (CGE) model, modelers often conduct sensitivity analysis to better understand the impact of uncertainty or variations in underlying assumptions and parameter values on results. However, CGE modelers have paid little attention to variations in input (i.e., policy changes or exogenous shocks). Recently, a few input–output studies consider sources of input variations, but focus on addressing a single source of variation and, therefore, ignore the interactions (and cumulative effects) among them. In this study, we propose methods for calculating the range (confidence intervals) of impacts from input variations within a CGE framework, explicitly accounting for multiple sources of input variations. To illustrate our approach, we use, as an example, recreational fisheries in Alaska, and estimate confidence intervals of economic impacts from regulatory changes. In doing so, we consider two important sources of input variation driving the impacts: (i) sample variation in recreational fishing-related expenditures, which is estimated using bootstrapping approach and (ii) stochastic variation from the parameters in the recreation demand model, which is estimated using a simulation-based approach. Results show that confidence bounds on total economic impacts calculated while only accounting for the first type of variation (sample variation) are much narrower than the confidence bounds on the impacts when we account for both sample and stochastic variation in model inputs.