Production from natural resources with latent abundance information: The case of the fishery

Ecosystem-based management of natural resources is widely reported by ecologists and has attracted attention from economists. For this management tool to be practical, it requires knowledge about dynamic human-natural interactions. However, the resource abundance, which is the most important state variable in natural resource production, is not directly observable to researchers in most cases. This dissertation is dedicated to addressing this challenge in empirical studies. The underlying rationale is that: if a common pool resource is jointly exploited by heterogeneous agents, the abundance information can be inferred from the economic behavior given structural information. The empirical study is the reef-fish fishery in the Gulf of Mexico. The comprehensive data set provides individual and spatial-temporal variations to deal with the latent stock problem. In addition, a rich set of fisheries management policies enable me to evaluate policy impacts and effectiveness. The dissertation consists of three parts.;The first part tries to infer biological parameters from individual economic behavior, which is realized by a proposed two-stage method. In the first stage, a generalized Schaefer production function with a latent stock is estimated by the within-period estimator. A stock index is constructed and used to estimate the logistic growth model in the second stage. To correct bias caused by using the estimated stock information, the simulation-extrapolation (SIMEX) method is adopted. All parameters can be identified by the two-stage approach. The empirical results show that the traditional method, which uses catch-per-unit-effort as a stock proxy and ignores measurement error, significantly overstates the prescribed harvest level.;The second part models fishermen's short-run labor supply behavior (resource extraction effort). Since fishermen do not earn a wage, a shadow wage is constructed from fishing profit, which is function of the latent stock. To deal with the latency problem, a three-step estimation method is proposed. First, count data models are adopted to estimate the fishing trip frequency, in which stock and other individual-invariant variables are represented by a time effect. Second, a stock index is estimated using the structural harvest-stock relationship. Last, the time effect is decomposed using the estimated stock and observable temporal variables. With three steps, the fishing effort model can be identified without knowing the biological stock. The empirical results show that fishermen are profit maximizing individuals and the labor supply curve is not backward bending.;The last part discusses the latent stock in fishermen's spatial choice behavior. The fishery resources are heterogeneously distributed and fishermen tend to choose the location that maximizes their utilities. Fishery resource abundance is the major driving factor for locational choice, however, this information is again unobservable. In the discrete choice model describing fishermen's spatial behavior, the latent spatially-, and species-explicit stock information is controlled by alternative-specific constant. Besides the unobservable stock, another challenge is to account for fleet heterogeneity. Such heterogeneity can manifest in a wide range of both observable and unobservable characteristics of fishing vessels and individual fishermen. In this paper, we draw on the literature on spatial sorting to estimate sorting models of observable heterogeneity. The models are used to explore spatial and inter-temporal species effort substitution in response to two marine reserves.;This research is expected to contribute to the literature in empirical resource economics and micro-econometrics. While this application is the fishery, econometric modeling of latent state variables is applicable to a broad range of natural resource and agricultural production settings.