PKMC Module III study materials
PKMC Module III study resources

Blended CPI That Hides a Labour Overrun

PKMC Module III · Put your knowledge to work in context.

All case studies

Blended CPI That Hides a Labour Overrun

Your project has a budget at completion of $2,400,000 over 20 weeks, planned linear except for a $400,000 perpetual licence scheduled for acceptance in period 4. At the end of week 10 the figures are: PV $1,200,000, EV $960,000, AC $1,320,000. The $400,000 licence has been delivered, accepted, and paid. The remaining labour and services budget is $2,000,000. Root-cause review shows a subcontractor rework loop caused by ambiguous interface specifications that were baselined and never resolved; the same interfaces are still being built. Separately, a change request worth $180,000 is sitting with the change control board and is not in the baseline, and a known integration risk of $250,000 is on the register with no response plan. The steering committee meets Friday and the project manager's draft slide says 'CPI 0.73, recoverable, on track for Q3'.

1. Calculate and interpret the earned value position. Why is the headline CPI misleading?

Read the worked answer

Headline figures.

  • CV = EV − AC = 960,000 − 1,320,000 = −$360,000
  • SV = EV − PV = 960,000 − 1,200,000 = −$240,000
  • CPI = 960,000 ÷ 1,320,000 = 0.73
  • SPI = 960,000 ÷ 1,200,000 = 0.80

Decomposition. The $400,000 licence is a discrete, fully earned, non-recurring element. Strip it out:

  • Labour PV = 1,200,000 − 400,000 = $800,000
  • Labour EV = 960,000 − 400,000 = $560,000
  • Labour AC = 1,320,000 − 400,000 = $920,000
  • Labour CPI = 560,000 ÷ 920,000 = 0.61; labour SPI = 0.70

Interpretation. The blended CPI of 0.73 flatters performance because a fixed, non-labour cost that cannot overrun is being averaged with work that is badly overrunning. The actual delivery engine is running at 61 cents of earned value per dollar - a 64% cost overrun on labour - and is 30% behind on schedule. 'Recoverable' is not supported: the root cause is a baselined specification defect that is still present in the work being built, so the variance is systemic and will continue.

Action. Report both views, explain the decomposition, and replace the 'on track' language with a forecast and a recovery decision.

2. Which EAC method do you recommend, and what range would you put in front of the steering committee?

Read the worked answer

The choice of formula is a judgement about cause, not arithmetic preference.

Atypical (one-off cause): EAC = AC + BAC − EV = 1,320,000 + 2,400,000 − 960,000 = $2,760,000. This assumes the remaining work will proceed at planned efficiency. It is only valid if the rework was a discrete event now closed. It is not: the ambiguous interfaces are still being built.

Typical, labour-adjusted (recommended): remaining labour at the observed labour efficiency.

  • ETC = remaining labour BAC ÷ labour CPI = (2,000,000 − 560,000) ÷ 0.61 = 1,440,000 ÷ 0.61 ≈ $2,365,000
  • EAC = AC + ETC = 1,320,000 + 2,365,000 ≈ $3,685,000

Sanity check with TCPI: (BAC − EV) ÷ (BAC − AC) = 1,440,000 ÷ 1,080,000 = 1.33. The team would have to deliver 33% more efficiently than plan for the rest of the project. That is not credible on unchanged specifications.

What I would present. A range of roughly $2.8M to $3.7M, with $3.7M as the working forecast and a commitment to replace it with a bottom-up ETC within three weeks. The low end is the best case if the interface defect is closed immediately and no further rework occurs; the high end is what the current trend implies. Presenting one number without the cause analysis is what invites the committee to disbelieve the next forecast too.

3. How should the pending change request and the integration risk be treated in the forecast and in the report to the committee?

Read the worked answer

Keep three things separate: the approved baseline, known pending changes, and unallocated reserve. Blending them produces a number nobody can act on.

Pending change request ($180,000). It is not in the baseline and must not be booked into EAC as if approved. Show it as a separate line: 'EAC $3.69M excluding pending CR-xx; $3.87M if approved.' Push the board for a decision date. In parallel, ensure no one is performing the changed work before approval - if work has already started, that is an unauthorised commitment and a control failure to report, not to hide.

Integration risk ($250,000). A known risk with a cost estimate belongs in the contingency reserve, calculated from the risk register (probability-weighted or expected-value), not in management reserve. The absence of a response plan is itself the finding: either fund a mitigation (for example a dedicated interface owner and a specification freeze) or accept the exposure explicitly.

Management reserve is for unknown-unknowns and requires sponsor approval to release; do not use it to absorb a known, identified risk.

Reporting format. Present baseline BAC, approved changes, current EAC, pending change, contingency draw, and management reserve as distinct columns, alongside the labour CPI/SPI and the TCPI. State clearly that the current forecast breaches the approved budget by roughly $1.3M and requires a decision: fund, descope, or re-plan. That is the conversation the committee actually needs.