Publisher's Synopsis
This thesis both evaluates, and presents improvements to, the current method of forecasting flying costs of Air Force aircraft. It uses depot level repairable (DRL) and consumable (CONS) data for the Air FOrce's bomber platforms: B-1B, B-2, and B-52H. The current forecasting method assumes a proportional relationship between costs and flying hours such that 1) when no hours are flown costs are zero, and 2) a 1% increase in flying hours will increase costs by 1%. The findings of this research indicate that applying log-linear ordinary least squares regression techniques may be an improved fit of flying cost data over the current proportional model; the actual data indicate a non-zero intercept and a less than proportional relationship between costs and flying hours. This research also found that models including factors other than fklying hours as independent variables, such as sorties, lagged costs, and fiscal trends, may be more useful than models based solely on flying hours. Finally, this research found that estimating quarterly costs at the base-level may yield more accurate estimates than estimating at the monthly level, or mission design series level.
