Three essays on applied economics: Rural electric cooperative call center demand, fertilizer price risk, and estimating efficiency with data aggregation

Scope and Method of Study. This study consists of three sections. The purpose of first section is to forecast peak call volume to allow a centralized after-hours call center for rural electric cooperatives to estimate staffing levels. A Gaussian copula is used to capture the dependence among nonnormal distributions. The purpose of second section is to examine the effectiveness of systematic cash purchase strategies in reducing fertilizer price risk for Oklahoma fertilizer dealers. The historical effectiveness of hedging with the fertilizer future market contracts (which have been discontinued) is also analyzed to provide a benchmark for comparison. The purpose of last section is to determine the effects of data aggregation on estimation of a stochastic frontier cost function using a Monte Carlo study.;Findings and Conclusions. For the first section, ignoring the dependence that the copula includes, would have resulted in an underestimation of peak values. The centralized call center resulted in cost savings of approximately 75% relative to individual centers at each cooperative. Adding cooperatives to the centralized call center is projected to further decrease costs per member. The magnitude of additional cost savings depends on the regional location of the new call center member. For the second section, cash purchase strategies were shown to be slightly effective in reducing average price and moderately effective in reducing risk. The reduction in price variance through cash purchase strategies was comparable to the historical effectiveness of hedging. For the last section, when the translog form of a stochastic frontier cost function with aggregated data is estimated, the variations of total cost decrease as output increases. If the variations of explanatory variables are small, then heteroscedasticity on the inefficiency error might be negligible. Stochastic frontier functions hold up rather well in the presence of data aggregation, but efficiency measurement from DEA diverges from true efficiency measurement.