Bid Dispersion on Unfamiliar Work
Bid dispersion, the spread across the prices contractors return for one set of drawings, is widest on exactly the kind of project Michael Stern writes about, the building with no close relative, and the question this page asks is why. Five firms read the same documents and return five numbers; on a conventional building the numbers cluster, and on an unfamiliar one they scatter. A Los Angeles owner’s representative reports the ordinary case as a project that pencils on paper and then draws first bids 8 to 12 percent over budget, not because anyone estimated badly but because input prices have climbed more than 40 percent since early 2020 and keep moving.5 That is the market repricing a building everyone understands. The spread on a building nobody has built is made of something else.
The something else is judgment about risk, and judgment varies by who holds it. A software vendor’s comparison of estimating techniques notes that analogous estimates, the kind a bidder reaches for when there is no close precedent, are subjective, vary with whoever performs them, and do not adjust well when the scope or the technology sits far from the precedent.6 Each bidder, lacking a comparable, substitutes a different one. One prices a slender concrete core off the last core it poured, another off the worst one. One carries a large sum for a curtain wall it has never detailed; another assumes a supplier will carry it. The documents are identical and the mental comparables are not, and the dispersion in the envelope is the dispersion in those private substitutions.
An architecture practice in Iowa gives further reasons bids come in above an architect’s estimate: material prices that moved at the last minute, other projects bidding in the same weeks and thinning competition, and an error by the estimator. Its answer is a bidding contingency, usually 5 percent of the construction figure, held against bid day anomalies.4 A contingency absorbs a spread; it does not narrow one. The devices that narrow it work on the documents and on the bidders’ information, and they are the subject of the rest of this page.
Three ways a bid prices what is not yet known
- Add alternateScope beyond the base bid, priced separately so the owner can buy it if the numbers allow.
- Deduct or substitution alternateScope removed or a different system offered, priced so the owner can trade down or sideways.
- AllowanceA placeholder figure for a product the owner has not yet chosen, with the labor to install it declared in or out.
- Unit priceA rate agreed in advance for work that may or may not happen, used when quantities are uncertain.
What narrows the spread
The first device is to stop asking bidders to guess. An estimating guide for mechanical and electrical bidders describes three instruments for pricing what is not yet known. Alternates let the owner ask what X costs instead of Y, or what it costs to add Z, as separately priced add, deduct or substitution items. Allowances carry a declared figure for a product the owner has yet to choose. Unit prices fix a rate in advance for work whose quantity is uncertain.1 Each moves a private guess into a public line. The guide’s own warning shows why that matters for dispersion: estimators who pad add alternates or shave deduct alternates are caught because owners compare alternate pricing across bidders, and when the alternates make no sense the base bid is doubted too.1 It also describes the trap in a loosely worded allowance, a stated sum for fixtures that the owner later reads as covering installation as well, and the fix, which is to declare in the bid what the allowance includes and how an overage will be processed.1
The second device is visibility. A consultancy that writes estimates for contractors lays out a bid summary with contingency on its own line, allowances listed individually with their basis, and alternates priced separately below the base total, so that an owner can see exactly what is firm and what is still open. Risk money buried in a unit rate, it argues, is money an owner cannot inspect, a bidder cannot justify and a budget review removes without discussion.2 When every bidder shows its risk money in the same place, the owner can compare how each read the risk rather than guessing at it from the totals.
The third device is restraint in how many options are asked for. A provincial construction association’s guidance on alternative prices recommends against using them at all where scope is clearly defined and the budget has room, limits them to meeting target budgets and to no more than 10 percent of the tender’s value, and discourages their use as a shopping list.3 Its reasoning is about time. Bidding often comes down to minutes, bidders cannot price many options accurately in that window, and alternative prices are often prepared before the final bid so a true value is not obtained anyway. The association prefers choosing the low bidder on the base price alone, with alternative prices supplied within 48 hours by the lowest bidder, as the most transparent of the two procedures it describes.3
Two ways to read alternative prices
| Alternatives count toward the low bid | Low bid on the base price only | |
|---|---|---|
| How the winner is picked | Base bid plus or minus whichever alternatives are accepted | Base bid alone; alternatives are priced afterward |
| Risk to the process | The winner can depend on which combination is accepted, inviting accusations of manipulation | Transparent; the association calls it the most desirable method |
| Timing of the prices | Submitted with the bid, often prepared before the final number | Provided within 48 hours by the lowest bidder only |
| Recommended use | Restricted to meeting target budgets | Used where alternatives are genuinely needed |
The fourth device shifts a risk nobody can price into a clause. A standard contract publisher describes price escalation clauses that adjust the contract price when a named material index moves beyond a threshold, up or down, and notes that under a lump sum agreement the constructor and its suppliers ordinarily carry the risk of price increases even when those increases are not their fault.7 A bidder asked to hold a fixed price on steel or copper for two years prices the worst case; one given a clause prices the material. The publisher lists the other devices that do the same work: limiting how long a bid can be relied on, cost of the work agreements rather than lump sums, phasing, allowances or alternates for volatile materials, early procurement and storage, prefabrication, and higher contingency.7 Where the indexes behind such clauses come from is covered in the entry on cost indices.
The fifth device is evidence that the unfamiliar part works. A glazing industry magazine explains that most standard wall systems have been tested and certified and need no further testing for a particular job, but that a custom design carrying features never tried before is another matter, and a mockup test run well ahead of fabrication is the one way to prove it performs, insurance the magazine calls relatively inexpensive.8 A bidder pricing an envelope nobody has built carries the chance that it leaks or fails under wind. A bidder pricing an envelope that has already passed its mockup test carries less, and so does every other bidder, which is the point. The mockup entry describes the test itself.
So the reply to the question runs in reverse. Bids converge when the documents replace private comparables with public lines, when risk money is shown in the same place on every summary, when the number of options is small enough to price in the time available, when the risks no bidder can forecast are moved into an indexed clause, and when the one part of the building nobody has built has been built once already as a mockup and tested at full size. The spread that remains is the honest difference between contractors, and that is the only spread an owner should be paying to resolve.