Labor Market Momentum by Metro: A Q2 2026 Scorecard
Across the 2018–2024 training window, a simple four-factor metro labor momentum score correlates at 0.52 with realized 12-month HPI, second only to lagged price momentum itself. We describe the score — total nonfarm employment growth, job-posting intensity, wage growth, and unemployment rate change — and its construction from BLS CES and QCEW series. We then identify 23 metros where the labor score is materially out of line with the re-invest v0.2 price forecast: 12 where labor is stronger than price would suggest (bullish read-through) and 11 where it's weaker (bearish). Finally, we show how the labor factor will enter the v1.0 model as one of four canonical features.
1. Why labor?
Housing is a derived demand for a labor market. Nobody buys a house in a place they can't earn; nobody sells one to leave a place growing payrolls. This statement is so obvious that it's easy to skip over — but the hard problem is quantifying labor "momentum" at the MSA level with data that is (a) released frequently, (b) revised meaningfully rarely, and (c) comparable across metros with different industry mixes.
We use four components:
- Nonfarm payroll growth (BLS CES) — 12-month change in total employment, MSA level. Monthly, 3-month lag, subject to annual benchmark revision.
- Job posting intensity — Indeed/LinkedIn aggregate postings per capita, indexed to 2019=100. Weekly, no revision, partial geographic coverage (we impute the remaining 10% using QCEW establishment counts).
- Wage growth (BLS QCEW) — 12-month change in average weekly wages, nominal, aggregated across industries with a cross-metro fixed-basket weighting. Quarterly, 6-month lag.
- Unemployment rate change (BLS LAUS) — 12-month change in the metro unemployment rate (sign inverted). Monthly, 1-month lag, lightly smoothed.
Each component is z-scored cross-sectionally against the 410- metro universe, winsorized at ±3σ, and the four z-scores are averaged. The result is a labor momentum score centered at zero with standard deviation 1: positive means out-performing the national median, negative means underperforming.
2. Calibration
We calibrated the equal-weight specification against a six-year backtest (2018–2024), measuring the correlation of the labor score with 12-month-forward HPI. Equal weighting outperformed both naive single-factor approaches and an OLS-fitted weighting that overfit the specific 2020–2022 dislocation.
| Specification | Corr w/ 12m HPI | Hit rate (top decile) |
|---|---|---|
| Payroll growth only | 0.43 | 54% |
| Job posting intensity only | 0.38 | 52% |
| Wage growth only | 0.29 | 51% |
| Unemployment Δ only | 0.41 | 55% |
| Equal-weight 4-factor | 0.52 | 59% |
| OLS-fit 4-factor (2018–22) | 0.48 (OOS) | 54% |
| OLS-fit 4-factor (2018–24) | 0.50 (OOS) | 56% |
| Lagged HPI (v0.1 baseline) | 0.61 | 62% |
3. 2026 Q2 scorecard — top and bottom
Rankings below are as-of April 2026 data. Payroll and unemployment components are current through March; job postings through the week of April 11; wage growth through 2025 Q4.
| Rank | Metro | Labor score | Payroll 12m | Postings index | Wage 12m |
|---|---|---|---|---|---|
| 1 | Provo-Orem, UT | +2.41 | +4.1% | 148 | +5.8% |
| 2 | Austin-Round Rock, TX | +2.28 | +3.8% | 141 | +5.2% |
| 3 | Raleigh-Cary, NC | +2.18 | +3.4% | 138 | +4.9% |
| 4 | Nashville-Davidson, TN | +2.04 | +3.2% | 132 | +5.1% |
| 5 | Boise City, ID | +1.98 | +3.5% | 129 | +4.6% |
| 6 | Huntsville, AL | +1.95 | +3.3% | 126 | +5.4% |
| 7 | Colorado Springs, CO | +1.91 | +3.0% | 128 | +4.7% |
| 8 | Salt Lake City, UT | +1.88 | +2.9% | 131 | +4.8% |
| 9 | Fayetteville-Springdale, AR | +1.81 | +3.2% | 122 | +5.0% |
| 10 | Charlotte-Concord, NC | +1.74 | +2.8% | 125 | +4.4% |
| — | — | — | — | — | — |
| 400 | Beaumont-Port Arthur, TX | -1.81 | -0.8% | 84 | +1.8% |
| 401 | Toledo, OH | -1.86 | -0.9% | 82 | +1.9% |
| 402 | Peoria, IL | -1.89 | -1.0% | 81 | +2.0% |
| 403 | Flint, MI | -1.94 | -1.1% | 78 | +1.7% |
| 404 | Youngstown-Warren, OH-PA | -2.01 | -1.2% | 76 | +2.1% |
| 405 | Binghamton, NY | -2.08 | -1.0% | 75 | +1.8% |
| 406 | Erie, PA | -2.12 | -1.3% | 77 | +1.9% |
| 407 | Rockford, IL | -2.18 | -1.4% | 72 | +2.0% |
| 408 | Decatur, IL | -2.27 | -1.6% | 71 | +1.5% |
| 409 | Kokomo, IN | -2.34 | -1.8% | 69 | +1.7% |
| 410 | Danville, IL | -2.41 | -1.9% | 68 | +1.6% |
The top-10 should not surprise anyone who has been watching the migration data: Mountain West and Southeast growth metros, with Utah outperforming Texas on a per-capita basis. The bottom-10 is almost entirely legacy industrial Midwest plus a couple of Texas energy-dominated metros that haven't recovered from the 2015–16 oil cycle.
4. Out-of-consensus metros
The interesting question isn't the top or bottom but the divergence: metros where the labor score and the forecast model disagree. We define "out of consensus" as metros in the top or bottom quintile of labor score but in the opposite quintile of 12-month HPI forecast. Twenty-three metros meet the criterion as of this print.
4.1 Strong labor, weak price forecast (12 metros)
These are markets where the labor signal is running ahead of the price model. Either labor is a leading indicator that hasn't hit prices yet, or price-specific drags (affordability, insurance, inventory) are masking the labor support.
| Metro | Labor z-score | v0.2 HPI 12m | Likely reason |
|---|---|---|---|
| Huntsville, AL | +1.95 | +0.8% | Catching up from 2023 base |
| Colorado Springs, CO | +1.91 | +0.4% | Inventory still absorbing 2021–22 build |
| Fayetteville-Springdale, AR | +1.81 | +0.9% | Supply elasticity intact |
| Boise City, ID | +1.98 | +0.2% | Priced in + affordability drag |
| Provo-Orem, UT | +2.41 | +1.1% | Affordability ceiling |
| Salt Lake City, UT | +1.88 | +1.4% | Labor still accelerating |
| Austin-Round Rock, TX | +2.28 | +0.9% | Office + supply overhang |
| Nashville-Davidson, TN | +2.04 | +3.1% | Model on-consensus (borderline) |
| Raleigh-Cary, NC | +2.18 | +2.4% | Moderate agreement |
| Charlotte-Concord, NC | +1.74 | +2.4% | Moderate agreement |
| Des Moines-West Des Moines, IA | +1.66 | +1.0% | Labor out-running housing permits |
| Omaha-Council Bluffs, NE | +1.52 | +1.1% | Stable wage growth, price lag |
4.2 Weak labor, strong price forecast (11 metros)
These are markets where the price model is calling for above-trend appreciation even though the labor fundamentals are weak. Almost all of them are supply-constrained coastal or legacy-desirable metros where momentum from 2023–24 is still carrying the forecast.
| Metro | Labor z-score | v0.2 HPI 12m | Likely reason |
|---|---|---|---|
| Boston-Cambridge, MA | -1.24 | +3.2% | Severe supply constraint |
| Hartford-E. Hartford, CT | -1.39 | +3.4% | Low inventory, stable wages |
| Providence-Warwick, RI | -1.41 | +3.1% | Commuter spillover from BOS |
| New Haven-Milford, CT | -1.18 | +3.0% | BOS/NYC spillover |
| Rochester, NY | -1.55 | +2.8% | Affordability floor |
| Albany-Schenectady, NY | -1.32 | +2.7% | State employment stability |
| Syracuse, NY | -1.68 | +2.6% | Similar profile to Rochester |
| Scranton-Wilkes-Barre, PA | -1.47 | +2.5% | NYC-adjacent migration |
| Reading, PA | -1.29 | +2.4% | Philly spillover |
| Springfield, MA | -1.35 | +2.3% | Below-trend labor, above-trend supply constraint |
| Worcester, MA | -1.22 | +2.3% | BOS spillover |
A sensible read is to be more skeptical of the bearish label (category 4.1) than the bullish one (4.2). Category 4.1 metros are mostly experiencing supply-driven price drags that will dissipate; once they do, the labor momentum will assert itself. Category 4.2 metros are relying on supply constraint to deliver returns in a weak labor environment — a more fragile setup.
5. Integration with the re-invest forecast (v1.0 path)
The current v0.2 model is a closed-form mean-reverting momentum specification that does not ingest labor features directly. The v1.0 model, targeted for 2026 Q3 release, replaces the closed form with a gradient-boosted tree regressor trained on four canonical feature blocks:
Labor features combined to ~28% of model output in the preliminary run. Of those, payroll growth and unemployment change are the two largest — consistent with the single-factor correlations reported in Table 2.
6. For the practitioner
Two takeaways for anyone using the re-invest forecast today:
- Cross-check the v0.2 forecast against the labor score. When labor is materially stronger than the HPI forecast, lean long the metro on a 24-month horizon even if the 12-month number looks uninspiring. Category 4.1 metros are roughly this.
- Discount supply-constraint-driven forecasts in weak labor markets. Category 4.2 metros are benefitting from the model's momentum term; if labor stays weak, expect the forecast to ratchet down quarter-over-quarter.
References
- [1]BLS Current Employment Statistics (CES), Quarterly Census of Employment and Wages (QCEW), and Local Area Unemployment Statistics (LAUS). Metro-level not seasonally adjusted.
- [2]Job-posting intensity is our own aggregation combining Indeed Hiring Lab weekly posts-per-capita (82% of metros) with a QCEW- establishment-based nowcast for the remainder. 2019=100 normalization.
- [3]Cross-metro wage aggregation uses 2019 national employment shares as the fixed basket (held constant across years) to separate wage inflation from industry-mix drift.
- [4]Correlation and hit-rate figures in Table 2 are from a leave-one-year- out cross-validation on the 2018–2024 panel, with winsorized residuals at ±4σ.
- [5]The v1.0 forecasting model uses XGBoost with the four labor features above plus lagged HPI, inventory, permits, mortgage rate, and CBSA as a categorical embedding. Preliminary backtest MAE 1.62pp vs v0.2's 1.87pp on a 2024 holdout.