Using Earned Value Information to Predict Program Cancellation

Abstract

Since 2001, 12 major defense acquisition programs (MDAPs) have been cancelled. Although each of these programs had problems with cost or schedule overruns (or both), there were other MDAPs that had similar problems and were not cancelled. Is it possible that program managers had information that might help determine which program was likely to survive and which was more likely to be cancelled? We employ a unique and rigorous statistical methodology to help program managers and their overseers understand and quantify the risk to their programs based on key earned value metrics. We compare programs that were cancelled to programs that had significant cost overruns but were not cancelled. We use survival analysis to investigate whether differences in key EV metrics reported for cancelled programs and troubled but not cancelled programs can be used to model the probability of cancellation for MDAPs. Our most significant finding across models is that when there is high cost growth in the EAC reported by the contractor, programs run far larger risks of cancellation. We find less robust evidence that increases in PM estimates and high cost variance also can drive risk of program cancellation.