A Dynamic Model of Endogenous Fishing Duration

Production functions are often estimated using temporally aggregated data to evaluate productivity in natural resource industry. However, a direct estimation can be misleading if the variable inputs are dynamically determined within a data time period because the variation within a period is not taken into account. In this study, we explicitly model fishery harvesters’ trip duration choices as a dynamic discrete choice problem, revealing that trip length is not generally exogenous as required by standard production models. The model is estimated by a conditional choice probability estimator using daily logbook data on a Japanese longline fishery. The results reveal the freshness deterioration function to which the harvesters respond, hence showing that the harvesters terminate trips to avoid having old fish before they maximize catch quantity. A simulation based on the behavioral mechanism we reveal shows that the estimated production frontier could be biased and that inefficiency is systematically misevaluated.