Earned Schedule Forecasting Method Selection



By Walt Lipke

Oklahoma, USA



Recent research indicates Earned Schedule (ES) forecasting from Earned Value Management data is generally improved when the performance factor, PF=1, is used. However, the use of the ES schedule performance index, SPI(t), remains as accepted practice and, at times, provides the better deterministic forecast. It is postulated that there may be recognizable conditions for which one method is preferable. This paper is an investigation to discern those conditions.


The authors, Batselier and Vanhoucke (B&V) examined several methods of forecasting in their paper, “Empirical Evaluation of Earned Value Management Forecasting Accuracy for Time and Cost” [Batselier et al, 2015]. Their research comprehensively evaluated forecasting using Earned Value Management (EVM) data from 51 projects, predominantly construction. The results of the research demonstrated the use of performance factor, PF=1, with the Earned Schedule (ES) method often provides the more accurate deterministic forecast.

Independently, the B&V finding was corroborated, albeit with a smaller amount of data [Lipke, 2017]. Of the 16 projects examined, ES forecasting using PF=1 was shown to provide better results for 12. Beyond affirming the B&V research, the application of PF=1 to statistical forecasting was examined. Additionally, an observation was made that when performance variation is small, forecasting with the ES schedule performance index, SPI(t), is usually better.

The good showing by PF=1, in both cited papers, was surprising and unexpected. Although PF=1 more often provides the better forecast, curiosity was created as to why it is not always so; there are several instances when SPI(t) is preferred. Because the examinations involved real data, there was some concern as to whether the data sets represented a very localized set of conditions, or if there were anomalies. That is, do peculiarities within the data examined cause the forecasting performance of PF=1 to be enhanced? For instance, management actions, such as re-plans, can obscure performance and cause the SPI(t) value to be close to 1.0. Of course, when the index value is close to 1.0, it is very likely that PF=1 provides the better forecast.

Certainly management actions can perturb the forecast and assuredly there are disturbances imbedded in the data. Nevertheless, examination of the data from the second paper did not reveal significant issues causing PF=1 to be preferred. It is now believed that the results from the two studies are reasonably broad-based and representative.

Nevertheless, there is reason to explore; there may be performance characteristics useful for identifying which method is more likely to provide the better forecast. We know by inspection that when performance is close to planned, PF=1 should provide the better forecast. As well, it is easily shown that when performance has little variation, the forecast from SPI(t) is better.

From these observations, the research proposition is made:

Two characteristics determine the selection of the more accurate forecasting method, either PF=1 or SPI(t):

1) Performance deviation from SPI(t) = 1.0

2) Periodic variation in ES

This paper examines simulated and real data to seek and establish the selection criteria for choosing between PF=1 and SPI(t) forecasting methods. Presuming selection criteria can be established, it is reasoned ES deterministic forecasting will yield increased accuracy and thereby improve decision-making by project managers.


To read entire paper, click here


How to cite this article:  Lipke, W. (2019). Earned Schedule Forecasting Method Selection; PM World Journal, Volume VIII, Issue I (January); Available online at https://pmworldjournal.net/wp-content/uploads/2018/12/pmwj78-Jan2019-Lipke-earned-schedule-forecasting-method-selection.pdf


About the Author

Walt Lipke

Oklahoma City, USA




Walt Lipke retired in 2005 as deputy chief of the Software Division at Tinker Air Force Base, where he led the organization to the 1999 SEI/IEEE award for Software Process Achievement. He is the creator of the Earned Schedule technique, which extracts schedule information from earned value data.

Credentials & Honors:

Master of Science Physics
Licensed Professional Engineer
Graduate of DOD Program Management Course
Physics honor society – Sigma Pi Sigma (SPS)
Academic honors – Phi Kappa Phi (FKF)
PMI Metrics SIG Scholar Award (2007)
PMI Eric Jenett Award (2007)
Who’s Who in the World (2010)
EVM Europe Award (2013)
CPM Driessnack Award (2014)
Australian Project Governance and Control Symposium established the annual Walt Lipke Project Governance and Control Excellence Award (2017)
Albert Nelson Marquis Lifetime Achievement Award 2018