Earned value management, or EVM, has a reputation for being a government reporting requirement invented to torture project managers. That reputation is undeserved. Strip away the acronyms and EVM is the answer to a question every contractor asks on every job: am I getting paid for the work I am doing, and am I doing it fast enough?
The trick is that you cannot answer either half of that question with two numbers. You need three.
The problem with comparing budget to actual
Most job cost reports compare two numbers: what you budgeted and what you spent. Suppose you budgeted $260,000 for an excavation item and you have spent $97,000. You are 37 percent through the budget. Is that good?
You have no idea. It is excellent if you have finished half the work and a disaster if you have finished a quarter of it. The two-number comparison cannot tell the difference, because neither number says anything about how much work is actually done.
Earned value adds that third number. Once you have it, you can separate a spending problem from a productivity problem from a schedule problem, which is the entire point.
The three core numbers
These have modern names and older acronyms from the original defense-contracting standard. Both are still in use, so it is worth knowing both.
| Term | Old acronym | What it means |
|---|---|---|
| PV — Planned Value | BCWS | Budgeted Cost of Work Scheduled. The budgeted value of the work you planned to have completed by today. |
| EV — Earned Value | BCWP | Budgeted Cost of Work Performed. The budgeted value of the work you have actually completed, regardless of what it cost. |
| AC — Actual Cost | ACWP | Actual Cost of Work Performed. What you have actually spent on that completed work. |
The one that trips people up is EV. Earned value is measured in budget dollars, not actual dollars. If you budgeted $6.50 per cubic yard and you have moved 13,000 yards, you have earned 13,000 × $6.50 = $84,500, whether that dirt cost you $60,000 or $120,000 to move. EV measures work performed on a common yardstick. Cost is compared to it separately. Keeping those two things apart is what makes the whole system work.
Pick one basis and stay on it
You can run EVM on your internal cost budget or on the sell value of your pay items. Both are valid, and they answer slightly different questions: cost basis tells you about execution, sell basis tells you about the job's financial outcome. What you cannot do is mix them. The worked example below uses a cost budget throughout.
A worked example with dirt
Take a single pay item: 40,000 cubic yards of unclassified excavation, budgeted at $6.50 per cubic yard. That makes the Budget at Completion, or BAC, equal to $260,000. The item is planned to run 10 weeks at 4,000 CY per week.
It is now the end of week 4. Here is where you stand:
| Input | Calculation | Value |
|---|---|---|
| Planned Value (PV) | 4 weeks × 4,000 CY × $6.50 | $104,000 |
| Earned Value (EV) | 13,000 CY installed × $6.50 | $84,500 |
| Actual Cost (AC) | Labor, equipment, and materials to date | $97,000 |
Three numbers. Everything else is arithmetic on those three.
Variances: the dollar answers
SV = EV − PV = $84,500 − $104,000 = −$19,500
Cost Variance (CV) is negative, so the work performed cost $12,500 more than it was budgeted to cost. Schedule Variance (SV) is negative, so you are $19,500 of budgeted work behind where the plan said you would be. Note that SV is expressed in dollars, not days. That confuses people at first, but it is useful precisely because it lets you add schedule slippage across items with different units.
Indices: the ratio answers
SPI = EV ÷ PV = $84,500 ÷ $104,000 = 0.81
The Cost Performance Index says you are getting 87 cents of budgeted work for every dollar you spend. The Schedule Performance Index says you are progressing at about 81 percent of the planned rate. Anything below 1.0 is unfavorable; above 1.0 is favorable.
Sanity-check the SPI against physical production: you planned 4,000 CY per week and you are averaging 13,000 ÷ 4 = 3,250 CY per week. And 3,250 ÷ 4,000 = 0.8125. The index and the dirt agree, which is exactly what should happen when the underlying data is honest.
Forecast: where this ends up
VAC = BAC − EAC = $260,000 − $298,500 = −$38,500
% complete = EV ÷ BAC = $84,500 ÷ $260,000 = 32.5%
Estimate at Completion projects the finish cost if current performance continues. On this item, that is roughly $38,500 over budget. Note the assumption baked into that formula: it presumes the rest of the item runs at the same CPI as the part you have done. That is a reasonable default and a poor certainty. If the overrun came from rock in the first 13,000 yards and the remaining 27,000 are sand, your real EAC is better than the formula says. Use the formula as a warning, then apply what you know about the ground.
You can do the same for duration: at an SPI of 0.8125, a 10-week item finishes in roughly 10 ÷ 0.8125 = 12.3 weeks. Cross-check against production: 40,000 CY ÷ 3,250 CY per week = 12.3 weeks. Again, agreement.
Reading the result
Both indices are below 1.0, so this item is over cost and behind schedule. That combination usually means one of three things, and the difference matters enormously:
- Productivity is genuinely low. The crew is not moving the yards the estimate assumed. Fix: crew, equipment, or method.
- The estimate was wrong. The ground is harder, wetter, or farther from the stockpile than bid. Fix: notice, and possibly a claim.
- The data is wrong. Hours are landing on this code that belong somewhere else, or quantities are undercounted. Fix: the capture process.
Number three is by far the most common in civil construction, and it is the reason many contractors have quietly given up on EVM. The math is not the hard part. Getting trustworthy quantity and cost data, dated to the day the work happened, is the hard part.
EVM does not fail in civil because the formulas are difficult. It fails because the inputs are reconstructed from memory a week later.
Common mistakes
- Using cost spent as percent complete. "We have spent 40 percent of the budget so we are 40 percent done" defeats the entire purpose. Percent complete must come from measured work.
- Estimating percent complete by eye. A foreman's "about 60 percent" is not a measurement. Count the units.
- Rolling up too far. Whole-job CPI hides everything. A job at 0.98 overall can contain an item at 0.55 that is eating the whole margin.
- Mixing cost basis and sell basis between the numerator and denominator, which produces indices that mean nothing.
- Letting AC lag EV. If quantities are current and invoices are three weeks behind, CPI will look artificially good and then crash. Compare like periods.
- Treating 100 percent as automatic. An item is not complete until it has passed the tests and the punch. Earning full value at physical placement overstates progress on inspection-heavy work.
Where to go next
For a deeper read on the two indices and how to act on them, see CPI and SPI explained. For applying this specifically to excavation and embankment, where "percent complete" is genuinely ambiguous, see earned value for dirt work. For the data discipline that makes all of it possible, see the field data capture guide.
PM Axsus is built around the input problem described above: it captures installed quantities, hours, and cost events at the point of installation so that EV and AC describe the same day. If that is the part you are stuck on, the earned value management page covers how it works.