Forward 6-month and 12-month Atlanta Fed Wage Growth Tracker nowcast from a 7-input composite of leading wage indicators (JOLTS quits, hires, unemployment gap, prime-age employment, WEI, inflation breakevens, manufacturing employment) calibrated by 60-month rolling-origin OLS. Same architecture as the Inflation Nowcast — the composite is regime-stable across pre/post 2020 splits at h=12m (corr +0.81 / +0.24 — caveat in source notes). Updated .
Current reading: composite at σ (well above long-run mean = tight labor market). Atlanta Fed Wage Growth Tracker currently % as of ; rolling-origin OLS calibrated on the most recent 60 months projects forward wage growth at % (6m ahead) and % (12m ahead).
The composite z (cyan) is the equal-weighted z-mean of 7 wage-leading indicators. The WGT YoY series (gold) is the Atlanta Fed Wage Growth Tracker — the realized target. Composite peaks lead WGT peaks; troughs lead troughs. The 2020 COVID shock is visible as a sharp dip in composite (slack indicators briefly spiked) followed by the 2021-2022 wage surge as labor markets tightened. The 2023-2025 "soft landing" period shows composite gradually normalizing while wages persistently above pre-COVID levels.
Rolling-origin OOS at h=6m. Aggregate metrics: corr , R² , MAE pp, bias pp. Pre-2020 corr ; post-2020 corr . Important caveat: the post-2020 era has been genuinely harder to predict — COVID labor shocks distorted the relationship between traditional slack measures and wage outcomes. The rolling refit adapts, but the conditional confidence interval is wider in the post-2020 period than the headline aggregate suggests.
Each lens contributes one z-scored input. JOLTS quits rate is the Fed's preferred wage leading indicator — workers leaving for higher pay is a clean signal of wage pressure. JOLTS hires rate measures absolute labor demand. Unemployment gap (UNRATE − 4) is flipped so HIGH = tight market. Prime-age employment filters out demographic noise from headline employment. WEI adds the broader real-economy pulse. 5Y inflation breakeven captures the inflation-expectations channel through which workers anchor wage demands. Manufacturing employment YoY adds a sector-specific demand lens.
Composite construction. z_full = mean of 7 expanding-window z-scores. Each input is z-scored using only data available at each month (no full-sample peeking). Sign convention is set so HIGH = wage-acceleration signal (slack indicators are flipped where appropriate).
Calibration. 60-month rolling-origin OLS — same architecture as the Inflation Nowcast and CCMI v2. At each historical month, the model is fit on the prior 60 months of (z, forward_wgt) pairs and used to predict wage growth at t+h. The "current" nowcast at the page top uses OLS fit on the most recent 60 months for the latest composite reading.
Why z_full and not z_slack. The Phase 0 probe (scripts/wage_nowcast_phase0.py) tested four composites — z_full (7 inputs), z_slack (3-input slack subset), z_demand (3-input demand subset), z_expect (just inflation breakeven). z_slack had the highest in-sample correlation (+0.80 at h=6m) but post-2020 dropped to +0.58 — regime instability driven by COVID's distortion of slack measures. z_full kept correlation at +0.79 post-2020 (h=6m) and +0.75 (h=12m) — broader composite is more regime-robust. Same Occam principle as the inflation z_pipeline pick: simplest model that maximizes OOS skill, not in-sample fit.
Validation summary (from scripts/wage_nowcast_phase0.py):
- Rolling-origin OOS corr (h=6m): , n=
- Rolling-origin OOS corr (h=12m): , n=
- MAE ~0.5-0.6 pp at both horizons; bias near zero
- Pre/post 2020 stability is REAL but weaker than inflation: post-COVID labor dynamics are harder to predict from traditional indicators
Sources.
Live track record. Every prediction this page emits is recorded in the Calibration Ledger at the moment it's made and graded against the realized Atlanta Fed Wage Growth Tracker when it lands.