Market Intelligence13 min read

Leading Indicators: Liquor Licenses, Hiring Gaps, and Commuter Flows

Three under-used public signals that predict commercial demand 12–18 months before broker data confirms it

Axiom Intelligence2026-03-24

Foot traffic data is the standard CRE leading indicator. It's also expensive, gated, and lags by 30–60 days. Public records contain three signals that beat it on cost, freshness, and (in our backtests) accuracy. None of the three is sufficient alone. The intersection is.

Act 1 — Liquor licenses as the 18-month signal

Restaurant operators don't open until permits are in hand and liquor licenses are filed. The state ABC application is one of the earliest public commitments. Between filing and opening: 9–14 months typically. Between filing and observed foot-traffic uplift: 12–18 months. By the time foot-traffic data shows the surge, the operator already knew.

Locus ingests 72K liquor license filings across 38 states. The signal we extract is clustering: hex cells where 3+ new on-premise licenses are filed within a 6-month window. That clustering correlates 0.68 with subsequent F&B-driven foot-traffic increases in the same hex.

New on-premise license filings (per 1k residents, top metros, 2025)
Nashville
4.2
Greenville
3.6
Tampa
3.1
Pittsburgh
2.4
Cleveland
1.9
Phoenix
1.4
Austin
0.7

Pittsburgh and Cleveland in the top five is not what the consensus map says. That's the point.

Act 2 — The job-posting salary gap

61K job postings scraped weekly from the major aggregators. The signal we extract is not posting volume — it's the spread between offered salary and the metro median for the same occupation. When the spread widens above 12%, employers are signaling that local supply is exhausted and they're paying a premium to import labor.

That premium has two CRE consequences. The first is downstream housing demand from the imported workers. The second — and more interesting — is the office siting question: companies hiring expensively in metro A will eventually look for cheaper office space in metro B, B+ submarkets, or fully-remote arbitrage that pulls capital out of the primary.

Metro / submarketAvg salary gap vs medianPosting volume YoYRead
Raleigh — RTP+17.3%+22%Acute talent shortage; office demand rising
Denver — DTC+13.8%+11%Moderate shortage; selective hiring
Austin — Domain+4.1%-8%Cooling; some pullback
Boise downtown+14.6%+19%Tertiary catching up to primary
Greenville — ICAR+15.9%+24%Manufacturing + tech overlap
Chattanooga — Innovation District+12.1%+16%Building toward inflection

Act 3 — LEHD commuter flows: where workers actually go

The Census Bureau's LEHD program publishes 454K origin-destination records describing where US workers live versus where they work. It is the single most under-used CRE dataset in public records. The reason: it has a 2-year lag, and most platforms refuse to use anything that lagged. That's a mistake — structural commute geography doesn't change quickly, and the patterns the data exposes are stable enough to act on.

Specifically: LEHD tells you the modal commute distance and earnings distribution for each work-cell. Low average commute distance plus high earnings means the work-cell is sitting on top of a wealthy walkable workforce — exactly the demographic that supports premium F&B, fitness, and personal-services retail. High commute distance plus high earnings means the same wealth is driving past your storefront, not stopping at it.

Retail siting based on residential demographics misses ~40% of the actual purchasing power. LEHD captures it — the wealthy lunch crowd that lives 12 miles away.

The three-signal intersection

Each signal beats the consensus narrative on its own. The interesting result is in the intersection. Hex cells that fire on all three — new on-premise license cluster, positive salary-gap differential, top-quartile LEHD work density — outperform the metro baseline by 4.8× in observed commercial rent change at t+12.

Signal count → commercial rent Δ at t+12 months
All 3 signals firing
+11.5%
2 of 3
+6.7%
1 of 3
+3.1%
0 of 3
+2.0%

Why the consensus misses this

Each of these datasets is mid-quality on its own. Liquor data is messy (state-by-state schemas); job postings have de-duplication problems; LEHD is lagged. Most platforms either skip them or use one in isolation. The intersection product — three weak signals jointly confirmed — is meaningfully stronger than any of the three and meaningfully cheaper than buying private foot-traffic feeds. That gap is the opportunity.