Most estimates carry escalation as a single percentage per year applied to everything. On a four-year project in a moving market, that assumption can be worth more than the whole contingency, and it gets a fraction of the attention.
The imbalance is odd when you stand back from it. Teams will argue for a day about a crane rate worth two hundred thousand dollars, then wave through an escalation line worth five million because it arrived as one tidy percentage from somewhere upstream. The market does not care that the number was tidy.
What escalation is, and what it is not
Escalation is price movement between the estimate’s pricing date and the date the money is actually spent. It is not contingency, which prices identified risks, and it is not scope. The two get blended constantly, and the blend ruins both numbers: when costs rise, nobody can say whether the market moved or a risk landed, so nobody learns anything. The clean stack is four lines, always in the same order: base estimate at the pricing date, escalation to the midpoint of spend, contingency, management reserve. What belongs in the third line is its own subject: see Contingency is not padding.
The method has a home in the AACE International literature: Recommended Practice 58R-10, Escalation Estimating Principles and Methods Using Indices (RP 58R-10), sets out the principles of building escalation from published indices rather than from habit. The practice is available to AACE members at web.aacei.org. What follows is how it looks applied.
Price it by commodity
A single percentage assumes the whole estimate rides one market. It does not. The estimate is a basket of different markets moving at different speeds, and the escalation has to be built the way the basket is built.
- Structural steel and rebar: index-linked, volatile, and usually bought early, which shortens their exposure but sharpens their timing risk.
- Copper and cable: index-linked, often bought late, carrying the longest market exposure in the estimate.
- Labour: agreement-driven and comparatively predictable in Canada, where collective agreements publish their increases years ahead; less predictable elsewhere.
- Engineered equipment: vendor-specific, usually fixed at purchase order, which converts its escalation risk into a scheduling question: how soon can it be ordered, and how long is the quote valid?
Cash-flow it against the buy dates
Here is the mechanic that a flat percentage cannot perform: escalation applies to money at the moment it is committed, not evenly across the project. Equipment fixed at order in month six stops escalating in month six. Rebar bought through the first construction year escalates to the middle of that year. Labour escalates along the whole duration, agreement by agreement. So each commodity gets cash-flowed to its own midpoint of spend, and the escalation is computed there. Escalate everything to the end of the project and the estimate is fat; escalate nothing past the pricing date and it is starving; both errors hide inside “three percent per year”.
A worked example
Take a base estimate of 100 million dollars at today’s pricing date, spending over three years. Split it three ways to keep the arithmetic visible. Engineered equipment, 40 million, ordered around month six: half a year of exposure at, call it, four percent per year, about 800,000 dollars, and then it is locked. Materials, 25 million, bought across the first construction year with a midpoint around month fourteen: the index outlook says four percent likely, but the credible range runs two to nine, so the likely escalation is about 1.2 million with a high case near 2.6. Labour and everything time-spread, 35 million, midpoint around month twenty: the published agreement increases average three and a half percent, about two million, and this one barely has a range because the increases are already printed.
The likely total is roughly four million dollars, four percent of base, and the honest range around it runs from about three to six and a half. A flat three percent would have booked three million: the bottom of the range, presented as the answer. The spread between likely and high does not get buried in contingency; it goes to the risk model as a correlated market risk, visible, with the index assumptions written down.
A flat three percent is a guess with a decimal point.
Where this goes wrong
Escalating committed money. The flat percentage runs across the whole estimate, including the equipment fixed at order in month six. The estimate quietly carries market exposure on money the market can no longer touch.
Escalating to the wrong date. Everything escalated to project completion, or nothing past the pricing date. The midpoint-of-spend discipline exists precisely because both shortcuts fail, in opposite directions, by material amounts.
The consumer price index. General CPI describes groceries and rent, not rebar and electricians. Construction input indices exist, are published, and behave differently. Using CPI because it is famous is choosing the wrong instrument because it is nearby.
Escalation hiding in contingency. One pool absorbs both market movement and risk events. When the pool drains, the post-mortem cannot say which force drained it, and the next estimate inherits the same blindness.
The lapsed quote. The estimate carries a vendor price whose validity expired months ago, with no escalation on it because “we have a quote”. A lapsed quote is not a price; it is a memory. What quote validity does to an estimate is its own subject: see Reading a vendor quote like an estimator.
Show it separately
Base estimate at pricing date, escalation to the midpoint of spend, contingency, management reserve. Four lines, always in that order, so nobody can say later that escalation was hidden in the number, and so each line can be repriced alone when its driver moves. When the steel index jumps eight percent, an estimate built this way updates one line in one afternoon and the board sees exactly what happened and why. That is the entire reward: a market event becomes an update instead of a crisis.
Emerald Group builds commodity-based escalation into estimates and reviews escalation assumptions on estimates prepared by others, for owners and engineering firms with multi-year capital projects. If your escalation line is one percentage on one row, get in touch.