Project Managers

The Project Management Social Network | Project Manager Jobs

There is undoubtedly the equivalent of an ‘urban myth’ still in circulation within the general project management community, arising from US Defense based research from the early 1990s, that the Earned Value (EV), Cost Performance Index (CPI) always stabilizes at the 20% completion and the final outcome will be within 10% of this value and usually worse. This myth has been extended by some authors to all projects in all industries; and I would suggest that this is demonstrably false in at least some circumstances. If CPI stability was an incontrovertible ‘fact’ for all projects, there would be no need for active management of the project after 20% completion! Newer research suggests CPI stability is not automatic.
Earned Value is a very useful project management control tool mandated by many Government agencies in the USA, UK and Australia. However, the migration of the EV toolset from carefully controlled major defense projects into the general PM business community is definitely creating issues.

The DoD research established ‘CPI stability’ on a large number of major US military projects. Newer research has found CPI stability is not a ‘given’ and it rarely exists on small commercial projects. Rather than arguing over research findings, I would have hoped in the intervening years, there would have been some practical research to identify what underlying factors cause stability in the CPI measure as evidenced by the DoD research, determine if the factors are desirable and then find ways to improve project management practice in other industries so that the desirable factors are encouraged.

My feeling is that when CPI stability is shown to be established, the ‘CPI Stability’ is a strong indicator of other important (but much harder to measure) factors such as stable management, stable requirements, an efficient management system, effective project culture, etc (many of which are likely to be present in major Defense projects as evidenced by the research undertaken by Christensen). Conversely, where CPI is unstable, significant changes in the underlying project can be reasonably assumed to be occurring, either at the management level or at the requirements/scope level. These changes may be beneficial or detrimental but are undoubtedly a risk that warrants the attention of senior management.

If these feelings are correct, it would also be useful to develop an understanding of the usefulness of CPI instability as a risk indicator (ie, what level of instability indicates a ‘project at risk’).

I’m wondering if anyone is undertaking this type of analysis?

Views: 128

Tags: CPI, EV

Comment

You need to be a member of Project Managers to add comments!

Join Project Managers

Comment by DAVID ESPINA, PMP on October 20, 2011 at 4:50pm
No, I think I understood exactly what you wrote and was not disagreeing.  I was merely trying to convey that not everyone over generalized this finding as some sort of physical law.  I consider when assessing a PM's EAC, but I understand the limitations of the study that found this correlation and base my findings accordingly.  If studies now show that this 20% rule is completely useless, then that's new information for me and I would like to see those studies.
Comment by Patrick Weaver on October 20, 2011 at 4:37pm

David, You have completely missed the point of the post.  Crystal did excellent work and demonstrated a strong correlation in a very specific type of project; large US DoD defense programs – these represent 2 or 3% of the worlds programs.

 

The error made by most authors and commentators is generalizing this very specific set of finding relating to one type of project in one very specific environment to ‘all projects’.  This is simply not possible statistically and has created a very damaging ‘urban myth’.  Until a lot more work is done there is no established correlation on projects of different types to one included in the work done by Crystal and studies by Henderson, Zwikael and others suggest the 20% rule is not useful outside of the USA DoD.

 

In these ‘other’ projects CPI and TCPI are very useful indicators of completion outcomes – it’s the high level unreliability of the 20% rule that was the focus of the post (this is an established fact) and how the difference in findings may be used. 

David, You have completely missed the point of the post.  Crystal did excellent work and demonstrated a strong correlation in a very specific type of project; large US DoD defense programs – these represent 2 or 3% of the worlds programs.

 

The error made by most authors and commentators is generalizing this very specific set of finding relating to one type of project in one very specific environment to ‘all projects’.  This is simply not possible statistically and has created a very damaging ‘urban myth’.  Until a lot more work is done there is no established correlation on projects of different types to one included in the work done by Crystal and studies by Henderson, Zwikael and others suggest the 20% rule is not useful outside of the USA DoD.

 

In these ‘other’ projects CPI and TCPI are very useful indicators of completion outcomes – it’s the high level unreliability of the 20% rule that was the focus of the post (this is an established fact) and how the difference in findings may be used.
Comment by DAVID ESPINA, PMP on October 20, 2011 at 8:57am
I did not interpret this finding as an immutable physical law, but rather a strong correlation between at completion CPI and 20% CPI.  This finding contributes into the analysis as to whether that EAC is credible given where you are in time and current cumulative CPI.  It is the equivalent as assessing the TCPI given the current CPIcum.  As with any retrospective survey or study, cause and effect is not established; however, an established correlation is nevertheless useful.

FEATURED

FOLLOW US
@projectnetwork

© 2014   Created by Mike.

Badges  |  Report an Issue  |  Terms of Service