Modeling in natural resource economics: Exploring three techniques

The three chapters of this dissertation consider three distinct modeling techniques, with applications in natural resource and environmental economics. Chapter I makes use of ordinary least-squares (OLS) regression analysis to investigate the effects of hypothesized climate warming on electricity demand. OLS techniques are applied in a general-to-specific model selection procedure to determine the most appropriate specification from a set of related hourly electricity demand models. The models selected are used to develop temperature elasticities of electric load, which are in turn used to quantify demand response to a warming simulation, under the scenario of a uniform 2°F increase in temperature.; Exogenous variables for the hourly electricity demand models include three classes of variables: (1) deterministic elements (Day, Month, Holiday), (2) autoregressive components (electricity demand from earlier hours), and (3) hourly dry-bulb temperature readings. Hourly weather and demand data were gathered for the U.S. mid-Atlantic region served by ten electric utility companies of the PJM independent system operator. Using these data, the model forecasts show a 4.6% increase in electricity demand due to the hypothesized 2°F warming. This outcome is consistent with results reported in the literature.; Chapter II offers an introduction to modeling with artificial neural networks (ANNs) for the researcher familiar with OLS techniques. The general form of an ANN is shown to be equivalent to a nested series of OLS models. This interpretation of the technique reveals that the flexible ANN specification is more readily understood and utilized than many researchers may expect. An example of applying ANNs to economic studies is provided by reworking the forecasting exercise from Chapter I using the ANN framework. The results of the simulation using the ANN specification confirm the results obtained using OLS techniques for simulating electricity demand under a warming scenario. These results are subjected to further analysis, including consideration of the forecast residuals.; Chapter III presents a model of forest-landowner behavior in the context of managing timberland that is subject to damage by fire. The model is used to examine differences between privately and socially optimal outcomes arising from externalities associated with fire suppression and fire-risk management. Variables under landowner control are (1) timber planting density, (2) length of a timber rotation, (3) the level of fire prevention undertaken, and (4) the timing of fire prevention application.; A numerical simulation is developed to examine fire-related externalities that may lead to sub-optimal outcomes. The externalities considered arise because (1) adjacent stands are linked by fire risk, such that fire prevention undertaken on one stand benefits both landowners, and (2) government provision of fire suppression services induces landowners to deviate from socially optimal management practices. The simulation considers several management scenarios for an infinite series of timber rotations, first for a lone stand, and then assuming two adjacent stands. In the game-theoretical framework of the two-stand simulation, landowners interact through the effects of their fire prevention activities on the common risk of fire. The final portion of the chapter considers two policy options and their effect on improving social welfare. The first policy is a subsidy on intermediate treatment measures that landowners may use to reduce fuel load (and thus, fire risk) on their stand. The second policy is a cost-sharing program, by which a landowner must bear a portion of the costs for controlling fires on his stand.