Improving the Goodness-of-Fits Associated with the Current and Proposed Combat Active Replacement Factors (CARF) Methodology

Abstract

The U.S. Marine Corps developed the Combat Active Replacement Factor (CARF) methodology as a way to obtain reliable logistics planning factors to aid in the estimation of equipment losses in future conflicts. The continuous evaluation and validation of these types of methodologies is considered of critical importance, since its effects directly impact combat effectiveness, supply chain management, logistics, acquisitions, and overall budgeting. This thesis analyzes a proposed methodology for use in calculating Explicitly Calculated CARFs (ECCs), making use of real-world Master Data Repository (MDR) data from previous low- and medium-intensity conflicts. As well, this thesis analyzes proposed regression models used in calculating Federal Supply Code (FSC) and Federal Supply Group (FSG) CARFs. We employ bootstrapping techniques in order to analyze the sensitivity of ECCs and find that as many of 70% may exhibit extreme sensitivity to reasonable changes in usage data. We employ Ordinary Least Squares regression models to estimate CARFs by FSC and FSG and obtain dramatically more CARFs relative to the draft methodology. Finally, a cross validation of a sample of the regression models reveals that CARFs generated from such models tend to vary substantially from their actual values.