A Classification Model for Early Estimation and Validation of US Government R&D Program Costs
Open Access DepositedDuring the initial stages of planning, US government research and development programs tend to underestimate program costs, resulting in program costs 20-120% higher than estimated and inefficient use of funds. This results from subjective cognitive biases and the absence of scope changes in the estimation process. This research introduces a novel metric, Total Financial Reserve (TFR), to the field, providing sponsors, portfolio managers, and the Department of Defense (DoD) with a comprehensive perspective on the anticipated costs associated with a program. This advancement facilitates improved planning and resource allocation, thereby minimizing the incidence of funding reallocations and program cancellations.Utilizing publicly available data from the Government Accountability Office (GAO), a comprehensive analysis of 20 years of Major Defense Acquisition Programs (MDAP) was conducted, allowing for a comparison between 133 initial baselines and their corresponding final costs. Reference Class Forecasting (RCF) was employed for the first time in the context of cost estimation for MDAP. A model was constructed utilizing a fitted Johnson SU distribution, thereby facilitating the establishment of quartiles for ten reference classes. Through cross-validation, it was demonstrated that this model accurately categorized the 50th and 75th quantiles for all the test data within 10% of the expected distribution. Furthermore, a comparison with established reserve estimation heuristics revealed a substantial improvement in accuracy.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.