"The project is 16% over budget" is a sentence that ends a steering committee and starts an argument. It is a fact without a cause. It does not say whether the company asked for more, whether the market moved, whether the design grew or whether the work is simply being done badly. Each of those has a different owner and a different fix, and some of them are not problems at all.
Project cost variance analysis does for a capital project what price-volume-mix analysis does for a P&L: it breaks one number into causes that someone can act on. Done well, it turns the forecast at completion into a bridge that every reader can follow from the approved budget to the latest estimate.
From budget to forecast, one cause at a time
Take a 100 million infrastructure project, of which 8 million is contingency. Twelve months in, the estimate at completion is 116.5 million. The bridge separates what the owner decided from what happened to the project.
- Approved changes (+6.0): scope the owner added on purpose. It is not an overrun; it is a new budget, and it should come with its own business case.
- Rates and prices (+3.5): steel, cable and labour rates above the estimate. Largely external, but it tests the escalation assumptions and the procurement strategy.
- Quantities (+4.0): more concrete, more piping, more earthworks than the design estimated. This is design maturity, and it is usually the largest controllable item in early phases.
- Productivity (+2.5): more hours per unit installed than planned. This is execution, and it is the item that earned value tracks month by month.
- Schedule-related (+2.0): the time-related costs of a three-month slip: site overheads, equipment hire, supervision.
- Buyout savings (−1.5): packages awarded below their estimate. They are real, but they should be banked, not quietly used to absorb overruns elsewhere.
Separating price, quantity and productivity
The three execution variances come from the same decomposition used in manufacturing cost accounting, applied to each work package. For the concrete package, budgeted at 10,000 m³ and 150 per m³, now forecast at 11,200 m³ and 162 per m³:
| Variance | Formula | Concrete package |
|---|---|---|
| Quantity | (forecast quantity − budget quantity) × budget rate | 1,200 × 150 = 180,000 |
| Price | (forecast rate − budget rate) × forecast quantity | 12 × 11,200 = 134,400 |
| Total | forecast cost − budget cost | 1,814,400 − 1,500,000 = 314,400 |
Labour is split the same way, with hours per unit as the productivity driver: the extra units cost the budgeted hours (quantity), the extra hours per unit cost the budgeted rate (productivity), and the higher rate applies to every hour worked (price). The order of the calculation matters, so it should be fixed once and used for every package, every month, or the bridge will move for reasons that have nothing to do with the project.
Contingency is a risk budget, not a slush fund
The second half of the story is whether the remaining contingency still covers the remaining risk. Contingency should be drawn when identified risks occur, and it should fall at roughly the pace at which risk is retired. When the money left falls below the exposure that remains, the project is no longer funded at the confidence level the board approved, long before the budget is formally exceeded.
A variance report that leads to decisions
- Report the bridge, not the total. Every month, the movement of the EAC split into the same categories, with an owner for each.
- Keep changes out of performance. Approved scope changes adjust the budget; they never appear as overruns, and overruns are never relabelled as changes.
- Show trends, not just positions. A quantity variance that grows every month signals an immature design that will keep growing.
- Track contingency against risk. Contingency drawn, contingency left and risk exposure left belong on the same page.
Bartor S.
Founder
