Earned value management: what EV, PV and AC actually measure

Planned value is the budget for the work your baseline said would be finished by today. Earned value is the budget for the work that actually is finished. Actual cost is what finishing it has cost so far. Three figures, one date, all in money, so that being late and being expensive stop looking like the same problem.

The three quantities, one status date

A project reporting a single number, spend to date, tells you almost nothing. Six hundred thousand pounds against a nine hundred thousand pound budget might be a project two thirds of the way through and exactly on plan, or a project a third of the way through burning money at twice the rate it should. The spend figure is identical in both cases, and no amount of staring at it separates them.

Earned value management separates them by adding a third quantity and insisting all three are read at the same moment. That moment is the status date, and it is the part people skip. Planned value comes off the baseline. Actual cost comes off the ledger. Earned value is the one that has to be produced deliberately, because nothing in a finance system knows how much work is done.

Quantity	What it measures, and how	Question it answers
Planned value (PV)	Budgeted cost of work the baseline set for now. PV = BAC × planned percent complete.	How much should be done by now?
Earned value (EV)	Budgeted cost of the work genuinely done. EV = BAC × actual percent complete.	How much is actually done?
Actual cost (AC)	Costs incurred for the work EV counts. AC = costs booked to date.	What has that work cost?

Every figure is cumulative to the status date and every figure is in currency. The currency is the whole trick. Percent complete cannot be compared with pounds spent, and days late cannot be compared with either, so a method that converts progress into money is a method that can hold all three in one sentence.

Planned value is the plan priced out

Take the budget at completion, take the proportion of it the baseline says should have been earned by the status date, and multiply. A £500,000 project running twelve months, four months in, with the baseline scheduling 35 percent of the work complete by then, has a planned value of £175,000.

That multiplication is a shortcut, and it is worth knowing what it hides. The rigorous version uses no project-level percentage at all: it sums the budgets of the individual work packages the baseline scheduled to finish by that date, plus the earned portion of any package straddling it. NASA’s earned value tutorial describes the baseline as work planned, budgeted and scheduled in time-phased planned value increments, which is the sum talking. The single percentage only reproduces that sum where your spend profile really is the smooth curve the top-level number implies, and on anything with a heavy mobilisation or a long commissioning tail it is not.

Which makes rebaselining a governance act rather than a scheduling one. Move the baseline and every planned value figure you have ever reported quietly changes meaning, along with every variance derived from it. Projects that rebaseline twice a year have no performance history, only a series of fresh starts, each one on plan by construction.

Earned value is progress priced at plan rates

Earned value is the same multiplication with a different percentage: budget at completion times the proportion actually complete. The £500,000 project, 30 percent finished at month four, has earned £150,000.

The convention that matters is the pricing. Earned value prices finished work at what the baseline said it would cost, never at what it did cost. NASA’s tutorial puts it in five words: work is earned on the same basis as it was planned. That is not a simplification for teaching, it is the entire mechanism. Because earned value is denominated in baseline pounds it can be set against planned value, which is also baseline pounds, and against actual cost, which is real pounds. One quantity, two conversations. Price earned value at what the work actually cost and it becomes actual cost, the two collapse into each other, and you are back to reading a spend report.

Which leaves exactly one soft input in the whole method: who decides that the project is 30 percent done.

Actual cost is the bill, and the bill runs late

Actual cost is the only one of the three that already exists somewhere. It comes out of the ledger: labour booked, invoices received, materials consumed, for work up to the status date. Our project has spent £160,000.

Two things break it in practice, and both are timing rather than arithmetic. The first is that ledgers lag. A subcontractor invoicing 45 days in arrears makes actual cost look flattering for six weeks, and a cost variance struck against an incomplete bill reads as good news right up until it does not. Accrue the known commitments, or say plainly in the report that you have not.

The second is scope matching. Actual cost has to cover the work that earned value counts, and nothing else. Costs booked against work not yet credited, or credit taken for work whose invoices land next month, put the two figures on different scopes and the difference between them stops meaning anything at all. That mismatch is what produces the cost variance that swings positive one month and negative the next while nothing on the project has changed.

The two gaps that turn three figures into a verdict

Three numbers on their own are description. Subtract them and you get a verdict.

Schedule variance is earned value minus planned value: £150,000 minus £175,000, so minus £25,000. Cost variance is earned value minus actual cost: £150,000 minus £160,000, so minus £10,000. Both negative, and the two negatives say different things. The project has produced £25,000 less work than the baseline scheduled, and it has paid £10,000 more than the work it did produce was budgeted to cost. Divide rather than subtract and the same facts arrive as ratios, a schedule performance index of 0.86 and a cost performance index of 0.94, which travel better into a report and lose the sense of scale that the money forms keep.

Two cautions on schedule variance in particular. It is denominated in money, so minus £25,000 is not twenty five thousand pounds of delay and cannot be turned into days without going back to the schedule and asking which activities are missing. And it converges on zero at the end of every project, because earned value climbs to the budget at completion whether the finish was early, on time or a year late. Schedule variance is a mid-flight instrument. On the last approach it stops reading, which is precisely when people quote it hardest.

What the three figures still cannot tell you

Earned value management is rigorous about cost and quietly weak about time, and the weakness is structural rather than a defect in somebody’s implementation. Three limits are worth saying out loud before the numbers reach a steering committee.

Earned value credits work, not the right work. A project can hold a schedule performance index of 1.0 while finishing everything except the one activity on the critical path, because a pound earned on a task with six weeks of float counts exactly the same as a pound earned on the task that decides the end date. The index is blind to sequence. The schedule is not, which is why the two get read together or neither is worth reading.

Level of effort packages earn on the calendar. Project management, technical support, anything credited by time elapsed rather than by output, earns its budget whether or not anything happened. Every pound of level of effort in the baseline is a pound of schedule variance the method cannot see, so keep the proportion small and keep it declared.

And none of it survives a poor decomposition. Earned value is only as granular as the work breakdown structure beneath it. Credit a £200,000 work package as a single item and the only honest answers available are zero and £200,000; everything between them is opinion wearing a decimal point. The US Government Accountability Office publishes earned value management inside its cost estimating and assessment guide rather than alongside it, which is the right filing: the method assumes a credible baseline and has no way to manufacture one.

None of which is an argument against the three quantities. It is an argument for knowing what they are for. Planned value, earned value and actual cost do not manage a project. They make one particular lie impossible to keep telling, which is that a project still inside its budget must be a project still on track.