Using Budgeted Cost of Work Performed to Predict Estimates at Completion for Mid-Acquisition Space Programs

Abstract

Conventional techniques to predict estimates at completion using earned value management data generally underpredict the end costs for space acquisition programs. This research assesses the accuracies of earned value management analysis methods to forecast estimates at completion for on-going space system acquisitions. The identification of more accurate “mid-stream” earned value management analysis methods improves managers’ abilities to assess the reasonableness of cost estimates. This research finds the linear relationship between budgeted cost of work performed and time yields a more accurate earned value management-based prediction for space program estimates at completion than existing methods. This result offers an empirically-supported method for earned value management data that may be useful to cost analysts and managers alike.

Notes

1. The Cost Performance Index (CPI) and the Schedule Performance Index (SPI) are two common measures of a program’s cost and schedule efficiencies; they measure performance by measuring actual versus planned costs and schedules, respectively.

2. Milestone A approves entry into the technology development phase. For this reason, contracts pre-Milestone A are debatably too conceptual to measure performance fairly. On the other hand, Milestone B approves entry into the engineering and manufacturing phase, a phase at which the program is defined and its performance can be assessed.

3. The consequence of conducting this study with WBS level 1 data is that one may only make observations from the perspective of the total program. A study at WBS levels 2 or 3 may reveal the subsystems, which drive an increase in the EAC for the total program.

4. Moreover, a negatively signed CPI is not meaningful because it implies the government is paid for the work performed on its behalf.

5. Coleman et al. (2003) used over 350 contracts from OSD CAIG’s Contract Analysis System. Druker et al. (2009) claimed to use a representative dataset. Although the latter claim is unverifiable, the results of their analysis is consistent with prior research (Christensen, 1994).

6. That is, parity between the parameter and the cumulative CPI term.

7. The research uses an R² threshold of 0.90. This threshold is chosen in lieu of a more restrictive threshold (e.g., 0.95) because of the small number of contracts this analysis uses.

8. Absent a final EAC, the SCP provides a high-fidelity estimate to which the EAC methods can be compared.

Additional information

Notes on contributors

C. Grant Keaton

C. Grant Keaton is an officer in the United States Air Force.