Every project can say what it has spent and what it planned to spend. Neither figure says how much work the money bought, and that missing third number is what earned value supplies.
The three inputs
Planned value is the budgeted cost of the work the baseline said would be complete by today. It comes from the plan alone and advances whether or not anybody works.
Earned value is the budgeted cost of the work that actually is complete. The word budgeted is doing the heavy lifting. Finished work is priced at the rate the baseline set rather than the rate it turned out to cost, and that is the only reason it can be compared with planned value at all.
Actual cost is what has been spent producing that work, and it is the one input that comes from the finance system rather than from a judgement about progress.
Four months into a ten month project
The budget at completion is 200,000, spent evenly over ten months. By the end of month four the plan expected 40 per cent of the work finished, measurement says 34 per cent genuinely is, and the ledger says 85,000 has gone.
| Figure | Arithmetic | Result |
|---|---|---|
| Planned value | 40 per cent of 200,000 | 80,000 |
| Earned value | 34 per cent of 200,000 | 68,000 |
| Actual cost | from the ledger | 85,000 |
| Schedule variance | 68,000 less 80,000 | 12,000 short |
| Cost variance | 68,000 less 85,000 | 17,000 over |
| Schedule performance index | 68,000 over 80,000 | 0.85 |
| Cost performance index | 68,000 over 85,000 | 0.80 |
| Estimate at completion, if efficiency persists | 200,000 over 0.80 | 250,000 |
| Estimate at completion, if the rest runs to plan | 85,000 plus 132,000 | 217,000 |
| Estimate at completion, if lateness keeps costing | 85,000 plus 132,000 over 0.68 | about 279,000 |
| Estimate to complete | 250,000 less 85,000 | 165,000 |
| Variance at completion | 200,000 less 250,000 | 50,000 over |
| To complete performance index | 132,000 over 115,000 | 1.15 |
Plotted, the same four months look like this. All three curves start together at zero, and by month four they have separated in the two directions the method exists to measure.
Planned value climbs in a straight line because the baseline spread 200,000 evenly across ten months. At the end of month four the vertical gap between planned value and earned value is the schedule variance, and the gap between actual cost and earned value is the cost variance. Both are measured in currency, which is why 12,000 short says nothing on its own about how many weeks late the project has become.
Reading the two variances
Schedule variance is earned value less planned value, here 12,000 short, and it says work the baseline valued at 12,000 was due by now and has not appeared. It is a currency figure rather than a number of weeks, and converting it needs the schedule, because 12,000 of budget might be one week in a heavy month and four in a light one.
Cost variance is earned value less actual cost. Here it is 17,000 over, and it says the finished work was budgeted at 68,000 and consumed 85,000. Somebody who has only memorised the formulas sees two negative numbers. Somebody who can read them sees a project that is behind and paying above plan for what little it has managed, a combination that deepens on its own.
The two indices
Dividing instead of subtracting gives a rate that survives comparison across periods and projects. The schedule performance index of 0.85 says the project is producing planned work at 85 per cent of the planned rate. The cost performance index of 0.80 says every unit of currency spent has bought 80 per cent of a unit of budgeted work.
Below one is unfavourable and above one is favourable, and the discipline that matters is saying what a figure means before deciding what it implies. A cost performance index of 0.80 does not mean a 20 per cent overrun. It means 25 per cent so far, because one over 0.80 is 1.25.
Forecasting the rest
Estimate at completion depends entirely on the assumption made about the work still to come. If current cost efficiency persists, the forecast is the budget divided by the cost performance index, giving 250,000. If the overrun was a one off and the remainder runs to plan, it is actual cost plus the remaining budgeted work, so 217,000. If schedule pressure keeps costing money, the remaining work is divided by both indices and the forecast moves towards 279,000. Three defensible answers for one project, and the assumption is the whole of the argument.
Two smaller figures follow from whichever forecast was chosen. Estimate to complete is the money still to be spent, which is the forecast less what has already gone, so 165,000 against a forecast of 250,000. Variance at completion sets the same forecast against the original budget, here 50,000 over, and it is usually the first figure a sponsor asks for, because it is stated in the currency the funding decision was made in.
The to complete performance index turns the question round and asks what efficiency the remaining work must reach to land on a chosen target. Against the original budget that is 132,000 of work left over 115,000 of money left, or 1.15. A team demonstrating 0.80 asked to deliver 1.15 is the plainest signal the method produces.
Every formula and what its answer says
The abbreviations below are the ones an exam question will use. PV, EV and AC are the three inputs, BAC is the budget at completion, and EAC, ETC, VAC and TCPI are the forecasts built on top of them. The middle column is the arithmetic and the right hand column is the sentence the figure produces, which is the half that decides a scenario question.
| Term | Calculation | What the answer says |
|---|---|---|
| Planned value | PV = BAC × planned per cent complete | How much of the budget the calendar said would have become finished work by now. 80,000 in the worked example. |
| Earned value | EV = BAC × actual per cent complete | What the work genuinely finished is worth at baseline rates. 68,000. |
| Actual cost | AC, read off the ledger | What has been spent producing that work. 85,000. |
| Schedule variance | SV = EV - PV | A negative figure says work the baseline valued at that amount was due and has not appeared. Minus 12,000, stated in currency rather than in weeks. |
| Cost variance | CV = EV - AC | A negative figure says the finished work cost that much more than it was budgeted to cost. Minus 17,000. |
| Schedule performance index | SPI = EV / PV | Planned work is arriving at that fraction of the planned rate. 0.85 is 85 per cent of the assumed pace. |
| Cost performance index | CPI = EV / AC | Each unit of currency spent has bought that much budgeted work. 0.80 buys eight tenths of a unit, so what has been delivered cost 25 per cent more than it was meant to. |
| Budget at completion | BAC, the cost baseline total | The whole authorised spend the project is measured against. 200,000. |
| Estimate at completion, efficiency persists | EAC = BAC / CPI | What the whole job costs if the cost efficiency achieved so far holds for the rest. 250,000, and the assumption a question means unless it says otherwise. |
| Estimate at completion, variance was a one off | EAC = AC + (BAC - EV) | What the whole job costs if the remaining work runs exactly to budget. 217,000. |
| Estimate at completion, lateness keeps costing | EAC = AC + (BAC - EV) / (CPI × SPI) | What the whole job costs if schedule pressure carries on being paid for. About 279,000. |
| Estimate to complete | ETC = EAC - AC | The money still to be spent under whichever estimate at completion was chosen. 165,000 against the 250,000 forecast. |
| Variance at completion | VAC = BAC - EAC | How far the forecast lands from the original budget. Minus 50,000 against the 250,000 forecast, which is a 50,000 overrun. |
| To complete performance index | TCPI = (BAC - EV) / (BAC - AC) | The cost efficiency the remaining work has to achieve to finish inside the original budget. 1.15 here, against the 0.80 actually being demonstrated. Put EAC in place of BAC in the denominator to test a revised target instead. |
Two habits make that table usable under time pressure. Every variance is a subtraction and every index is a division, and earned value leads both, which is why a positive variance and an index above one always read as favourable. Every forecast is an assumption about the future written as arithmetic, so the first question to ask about an estimate at completion is which assumption produced it.
Why schedule variance in currency goes quiet
The weakness is structural. At completion, earned value equals planned value equals the budget at completion, because all the planned work eventually gets done. Schedule variance therefore returns to zero and the index returns to one on a project delivered a year late, exactly as on one delivered to the day.
The schedule side of earned value degrades as the project approaches its finish, which is when people stare at it hardest. Earned schedule and ordinary critical path analysis answer the question in units of time instead. The cost side has no equivalent flaw, because actual cost never converges on the budget by construction.