Most project reports answer two questions: how much have we spent, and are we on schedule? Neither tells the board what it really needs to know. A project that has spent 40% of its budget may be in excellent shape or in serious trouble, depending on how much work that money actually bought. Spending is an input; the asset being built is the output.
Earned value management (EVM) closes that gap with one idea: put a price on the work completed, using the same budget that priced the plan. Once planned work, completed work and actual cost are expressed in the same currency, cost and schedule performance can be compared, trended and projected, and the forecast at completion stops being a matter of opinion.
Three numbers, one currency
- Planned value (PV): the budgeted cost of the work scheduled to date. It is the baseline S-curve.
- Earned value (EV): the budgeted cost of the work actually completed. If a 2 million foundation is 50% done, it has earned 1 million, whatever it cost.
- Actual cost (AC): what the completed work really cost, from the ledger, on the same cut-off date.
Consider a plant expansion with a budget at completion (BAC) of 48 million and an 18-month schedule. At month 8 the baseline expected 20.0 million of work to be done. The site has completed work worth 18.2 million at budget rates and booked 20.9 million of cost.
The traditional report would say the project has spent 20.9 million against a plan of 20.0, a modest 4% over. Earned value tells a different story. The site has produced less than planned and paid more for it: every million of budgeted work has cost 1.15 million. Spending looks close to plan only because one problem is hiding the other.
The indices that drive the forecast
| Measure | Formula | At month 8 |
|---|---|---|
| Cost variance | CV = EV − AC | −2.7 million |
| Schedule variance | SV = EV − PV | −1.8 million |
| Cost performance index | CPI = EV / AC | 0.87 |
| Schedule performance index | SPI = EV / PV | 0.91 |
| Estimate at completion | EAC = BAC / CPI | 55.2 million |
| To-complete performance index | TCPI = (BAC − EV) / (BAC − AC) | 1.10 |
The estimate at completion assumes that the remaining work will be done at the efficiency seen so far. That gives 55.2 million, a 7.2 million (15%) overrun, and a finish around month 20. The to-complete index is the reality check for any recovery plan: to stay within 48 million, every remaining million of work would have to cost 0.91 million, against 1.15 million today: a 26% jump in efficiency from a team that has not yet met its plan once. Recovery plans that require a TCPI above 1.05 rarely deliver.
Why the early signal is reliable
The power of earned value is not the arithmetic; it is how early the arithmetic becomes stable. Studies of US defence contracts by David Christensen and colleagues in the 1990s found that once a contract is about 20% complete, its cumulative CPI rarely improves by more than ten percent, and when it moves, it usually gets worse. Cost overruns are not discovered at the end; they are visible in the first fifth of the work to anyone who measures earned value.
What finance should ask for
- A frozen baseline. Planned value is only meaningful if changes to it go through formal change control. A baseline that is re-planned every quarter hides every variance.
- Objective progress rules. Earn value on measurable milestones (0/100, 50/50 or physical quantities installed), not on the percentage that the site manager feels.
- Accruals in actual cost. Work done but not invoiced must be in AC, or the CPI looks better than reality for months.
- The EAC in the financial forecast. If the project forecast still shows the budget while the CPI is 0.87, the company is reporting hope, not a forecast.
Bartor S.
Founder
