Prediction models for validating development phase cost estimates for missile development programs

Cost estimates for missile development phase programs are almost always significantly underestimated. The purpose of this research was to determine if missile program development cost might be estimated using regression models at the uppermost Cost Estimating Structure (CES) level. Using this process, it would be possible to use this same model to validate development phase cost estimates made by summing estimates made at lower CES levels. A regression model developed at a higher CES level has the inherent advantage of including more unknown-unknowns (embedded in the historical data) than models created at more detailed CES levels. Unknown-unknowns are defined as cost related activities that are not known to the cost estimator at the time the estimate must be prepared. The results of this research indicate that statistically significant models are achievable, with a coefficient of determination ( r2) value greater than 90%. The resulting cost estimating models may be used to validate future development phase missile program cost estimates. Thus, missile developers, program managers, and acquisition authorities may use this information to make better, more confident decisions concerning the billions of dollars annually spent on missile development programs.