How the Trade Demand Index works

The Trade Demand Index (TDI) is a 0–100 score for how tight the labor market is for a given trade in a given state, right now, relative to every other trade and state we track. A 90 means “among the tightest markets in the country this month.” A 20 means “comparatively slack.” It is not a prediction and it is not a wage estimate — it’s a snapshot, recomputed every month, of demand pressure.

Alongside the TDI, every trade × state pair also gets a Shortage Signal — a plain-language demand-minus-supply gap. Positive means job postings and an aging workforce are outrunning the pipeline of new workers coming in behind them.

The weights above are version-locked — if we ever change them, we’ll bump the methodology version and say so, and every historical number stays interpretable against the version it was computed under.

Through July 2026 the monthly input was per-trade job-posting counts from a commercial job board. We removed it. That board’s terms permit displaying its listings but bar publishing counts derived from them without written permission we don’t have — so continuing to publish an index built on those counts wasn’t a option we were willing to take. Everything in the score is now public government data or openly-licensed data anyone can check.

The honest cost of that swap: the replacement signal is measured for the whole state, not per trade. It blends three things that move monthly — construction employment (BLS), housing units authorized by building permit (Census), and overall job-postings momentum (Indeed Hiring Lab) — into one year-over-year read on whether a state’s building economy is expanding or contracting. Electricians and roofers in the same state therefore get the same momentum number. The rest of the score, which is where trades separate from each other, stays per-trade.

We weighted it accordingly: a signal that can’t tell one trade from another shouldn’t carry as much of a trade’s score, so it dropped from 25% to 20%, and the freed weight went to projected growth — the strongest per-trade signal we have. A per-trade posting number returns when our application to the federal CareerOneStop/NLx job feed is approved; that data is public domain and can be published without any of this.

Two of the three momentum inputs are missing for a handful of states — BLS publishes no seasonally-adjusted construction-employment series for Delaware, DC, Hawaii, Maryland or Nebraska, and Tennessee’s series was discontinued in 2007. Those states are scored on the inputs that do exist rather than being given a substitute number.

A few inputs we’d like to have aren’t reliably available at the state level for every trade — most notably a precise breakdown of workforce age by state and trade, and apprenticeship-completion counts by state. Where we don’t have a real number, we leave that input blank for that row rather than estimate one. A blank input doesn’t zero out a trade’s score — it’s excluded and the remaining inputs are reweighted so a missing annual figure doesn’t unfairly drag a score down.

Because those two inputs are blank for every row today, the live score is effectively driven by four of the six: market momentum, projected growth, cost-of-living-adjusted wages, and automation durability. We’d rather say that plainly than present six weights and let you assume all six are doing work.

Before publishing, we checked the index against outside data we didn’t build. These checks were first run against methodology v0.1; the structural components they cover are unchanged in v0.2, but the monthly signal is new and has not yet been through the same external re-anchoring — that is scheduled with the next scoring revision.

• Growth rankings. The index’s projected-growth component ranks trades in the same order as the federal government’s own occupational growth forecasts — solar installation projected to grow fastest, then electrical work, then HVAC, then roofing, then plumbing. Six for six, same order.
• Workforce size. Summing our state-by-state employment figures for electricians lands within the range you’d expect once you account for the fact that our figures are drawn from a slightly older baseline than the most current national count — and the gap between the two is almost exactly explained by the growth the government itself has been projecting for the trade since then.
• The broader hiring climate. National data on employer job openings in construction shows a market that’s recovering, not collapsing, through 2026 — the same picture our own monthly numbers show across the trades we track.

None of this makes the index perfect. It means that when the index says a trade is tightening in a state, that’s not an artifact of how we built it — it shows up in independent government data too.

• The TDI compares trades and states to each other in the same month. It doesn’t tell you a trade is objectively “good” or “bad” — only how it stacks up right now.
• A single month’s number can be noisy, especially for smaller trades in smaller states. Look at the trend, not one snapshot.
• The Shortage Signal is a gap indicator, not a job-count forecast. A large positive number means the pressure is building — it doesn’t promise a specific number of openings next year.