Seventeen points. That's how far the gap between low bids and engineers' estimates moved overnight — from bidders coming in 4.7% under what agencies expected to landing 31.7% over it. Same 1,000-result sample, one cycle apart.
Now, before anyone panics: this is a rolling dataset, not two different weeks of bidding. A sample this size doesn't flip its underlying market in a day. What moved is the mix — older, leaner-priced results rolled out of the 1,000-result window and got replaced by newer ones where contractors priced well above what the agency engineers wrote down.
“The published number on a bid document is worth less than it used to be as a planning tool. Build your own number.”
Here's why that matters even if it's a sampling artifact: it tells you engineers' estimates are running stale across a big chunk of this board. Average bidder count barely budged, 9.1 to 9.4, so it's not that fewer firms are competing it down. The same crowd is showing up and still pricing jobs well north of the pre-bid number.
If you're building a budget off an agency's own estimate right now, treat it as a floor, not a target. A job advertised at $2 million that's tracking toward this pattern could land at $2.6 million once real bids hit the table, and that's before change orders. For estimators on the other side, chasing jobs where you can beat the published range by double digits isn't unusual right now — it's the market.
Spread behavior backs this up in its own confusing way: the average gap between low and second bid is reported at $7.78 million in dollar terms this cycle, way out of line with a 16.1% spread the cycle before. That's not two bidders $7.78 million apart on every job — it's a handful of very large contracts (MTA, NJDOT, big DASNY work) pulling the average by sheer scale. Don't read it as typical job-to-job behavior; read it as proof the board's current mix skews toward bigger-ticket work than it did a day ago.
So the real story isn't that pricing suddenly jumped 36 points. It's that agency estimates and contractor pricing are drifting apart badly enough, across enough of this sample, that the published number on a bid document is worth less than it used to be as a planning tool. Build your own number. Use the engineer's estimate as a conversation starter, not a ceiling.