Based on current macro regime conditions and unemployment rate (u3)'s historical behaviour in similar regimes, the model projects 4.29% by 2026-12-31 ( -0.1% from 4.30% today). The 68% confidence range is 2.46% to 6.13%; the wider 95% range is 0.70% to 7.89%. Methodology below the headline.
Unemployment Rate (U3) Forecast 2026
Quantitative analysis from 298 observations of Unemployment Rate (U3) history, joined to four universal macro regime classifications. Numbers are computed, not narrated.
Performance by Window[02]
| WINDOW | N | ANN RET | ANN VOL | RET/VOL | HIT % | TOTAL |
|---|---|---|---|---|---|---|
| 1Y | 12 | 2.38% | 9.90% | 0.24 | 45.5% | 2.38% |
| 3Y | 35 | 6.28% | 10.29% | 0.61 | 38.2% | 19.44% |
| 5Y | 60 | -6.76% | 12.40% | -0.55 | 32.2% | -29.51% |
| 10Y | 120 | -1.69% | 77.12% | -0.02 | 26.9% | -15.69% |
| All | 298 | -0.18% | 49.25% | -0.00 | 31.3% | -4.44% |
Annualized total return = (1 + total)^(1/years) - 1. Ret/Vol is the annualized return divided by annualized volatility (Sharpe-equivalent without risk-free subtraction). Hit % = pct of single periods that were positive.
Where We Are Now[03]
Forward Returns by Macro Regime[04]
How Unemployment Rate (U3) has performed historically conditional on the prevailing macro regime. The current bucket is highlighted; +1Y averages drive the headline signal above.
| REGIME BUCKET | N | +30D | +90D | +1Y AVG | +1Y MED | HIT % |
|---|---|---|---|---|---|---|
| Low (<15) | 65 | -1.06% | -1.87% | 7.77% | -8.00% | 27.7% |
| Normal (15-25) | 90 | -0.35% | -1.39% | 2.28% | -7.41% | 33.7% |
| Elevated (25-40) | 32 | -1.23% | -3.55% | -7.65% | -3.06% | 28.1% |
| Extreme (>40) | 3 | n/a | n/a | n/a | n/a | n/a |
| REGIME BUCKET | N | +30D | +90D | +1Y AVG | +1Y MED | HIT % |
|---|---|---|---|---|---|---|
| Inverted (<0bps) | 27 | -0.20% | 1.60% | 7.18% | 7.89% | 88.9% |
| Flat (0-100bps) | 62 | -1.62% | -4.46% | 10.59% | -6.98% | 23.2% |
| Steep (>100bps) | 100 | -0.37% | -1.43% | -3.50% | -8.33% | 20.0% |
| REGIME BUCKET | N | +30D | +90D | +1Y AVG | +1Y MED | HIT % |
|---|---|---|---|---|---|---|
| Tight (<350bps) | 24 | -1.60% | -3.98% | -8.79% | -5.26% | 36.8% |
| Normal (350-500bps) | 45 | -1.47% | -2.33% | 19.73% | -5.00% | 44.2% |
| Stressed (>500bps) | 18 | -2.96% | -10.75% | -18.59% | -10.64% | 5.6% |
| REGIME BUCKET | N | +30D | +90D | +1Y AVG | +1Y MED | HIT % |
|---|---|---|---|---|---|---|
| Weak (bottom tercile) | 34 | 0.21% | 1.82% | 11.17% | -6.06% | 44.1% |
| Neutral (middle) | 38 | -1.13% | -3.54% | -10.33% | -9.09% | 2.9% |
| Strong (top tercile) | 77 | -1.25% | -3.05% | 8.52% | -5.41% | 40.0% |
Forward returns are forward-looking from each historical observation in the bucket; +252d corresponds to one trading year. Buckets with fewer than 5 forward-return observations are reported as n/a. These are conditional historical averages, not forecasts.
Lead-Lag Relationships[05]
For each universally-recognised leading indicator, the lag at which the daily-return correlation peaks. Positive lag means the anchor leads Unemployment Rate (U3); negative means it lags.
| ANCHOR | ROLE | PEAK LAG | PEAK CORR | ZERO-LAG | RELATIONSHIP |
|---|---|---|---|---|---|
| Initial Jobless Claims | Labor leader | 0d | 0.980 | 0.980 | coincident |
| HY OAS Spread | Credit risk leader | 0d | 0.581 | 0.581 | coincident |
| Baa-10Y Spread | Credit risk (slow) | 0d | 0.495 | 0.495 | coincident |
| 10Y-2Y Yield Spread | Recession leader | -28d | -0.489 | 0.027 | lags target by 28d |
| VIX | Volatility leader | +55d | 0.390 | 0.114 | leads target by 55d |
| 10Y Treasury Yield | Discount-rate driver | 0d | -0.269 | -0.269 | coincident |
| U-Mich Consumer Sentiment | Survey leader | 0d | -0.225 | -0.225 | coincident |
| Trade-Weighted Dollar | FX driver | 0d | 0.193 | 0.193 | coincident |
| Copper | Global growth proxy | +40d | 0.140 | -0.117 | weak |
| NFCI | Financial conditions | +5d | 0.031 | 0.029 | weak |
Pearson correlation of daily returns over up to 25 years of overlapping history, searched across a ±60-day lag grid. Indicators classified as “weak” don't have meaningful predictive power at daily resolution; many of these (yield curve, NFCI, sentiment) lead at monthly/quarterly horizons instead.
Historical Analogs[06]
Periods where Unemployment Rate (U3) sat at a similar percentile rank to today, with what happened over the next 30 / 90 / 252 trading days. Analogs are clustered to avoid double-counting nearby dates.
| DATE | VALUE | +30D | +90D | +1Y |
|---|---|---|---|---|
| Mar 1, 2025 | 4.2000 | 0.00% | 2.38% | 2.38% |
| Dec 1, 2024 | 4.1000 | -2.44% | 2.44% | 4.88% |
| Sep 1, 2024 | 4.1000 | 0.00% | -2.44% | 7.32% |
| Jun 1, 2024 | 4.1000 | 2.44% | 0.00% | 0.00% |
| Sep 1, 2021 | 4.7000 | -4.26% | -14.89% | -25.53% |
Worst Historical Drawdown[07]
Largest Single-Period Moves[09]
- Apr 1, 2020236.36%
- Mar 1, 202025.71%
- May 1, 20088.00%
- Dec 1, 20087.35%
- Jan 1, 20096.85%
- Aug 1, 2020-17.65%
- Jun 1, 2020-16.67%
- Oct 1, 2020-11.54%
- May 1, 2020-10.81%
- Nov 1, 2021-8.89%
Calendar-Month Seasonality[10]
Average single-period return aggregated by the calendar month in which the period ended.
| MONTH | AVG RETURN | HIT % | N |
|---|---|---|---|
| January | -0.60% | 24.0% | 25 |
| February | 0.16% | 44.0% | 25 |
| March | 0.36% | 24.0% | 25 |
| April | 9.33% | 32.0% | 25 |
| May | -0.74% | 29.2% | 24 |
| June | -0.34% | 37.5% | 24 |
| July | -0.62% | 32.0% | 25 |
| August | -0.58% | 32.0% | 25 |
| September | -1.13% | 24.0% | 25 |
| October | -0.02% | 33.3% | 24 |
| November | -0.32% | 36.0% | 25 |
| December | -0.33% | 28.0% | 25 |
N = 298 OBS · GENERATED 2026-05-18 10:00Z
Forecast Approach
regime implied: The current macro regime classification (Goldilocks, Reflation, Stagflation, or Deflation) dictates the expected direction and magnitude of movement, calibrated against historical regime performance.
Consensus source: Bloomberg survey consensus
Key Drivers & Risks
- •Economic growth
- •Monetary policy
- •Fiscal spending
- •Immigration
- •Productivity
Historical Volatility
Low: labor market is a lagging indicator with slow-moving trends
Scenarios That Affect This Forecast
How Unemployment Forecasts Have Held Up Historically
Unemployment forecasts have a strong track record on the trend (the rate is highly persistent) but a poor track record on turning points. The 2009 peak (10.0%), 2020 spike (14.7%), and the 2024 bottom (3.4%) were all missed by consensus on timing and magnitude. Recessions are typically labeled in retrospect; the regime model reads back from unemployment rather than predicting it.
Regime-conditional models on unemployment achieve approximately 75% directional accuracy on quarterly windows because the rate moves slowly. The Sahm Rule (3-month average UR rises 0.5pp above the prior 12-month low) has been the most-reliable real-time recession indicator: it has a 100% true-positive rate over 1949-2025 with one false-positive scare in 2024.
Regime Sensitivity for Unemployment
Unemployment is itself a regime variable. The Recession regime classifier reads back from unemployment plus payrolls plus claims. Goldilocks regimes anchor unemployment at 3.5-4.5%; stagflation regimes push it above 4.5%; deflation regimes spike it above 5%.
The April 2026 setup has unemployment at approximately 4.1%, up from the 3.4% 2024 low. The Sahm Rule was triggered briefly in summer 2024 (3-month average rose 0.5pp above the prior 12-month low) but the recession didn't materialize; the false positive was attributed to a labor-supply expansion (immigration) rather than demand destruction. The regime conditional reads as moderately negative on direction but well below recessionary thresholds.
What Drives Unemployment Forecast Errors
Three structural issues drive unemployment forecast errors. First, the labor-supply versus labor-demand decomposition is unstable. The 2024 Sahm-Rule false positive came from supply expansion (immigration adding to the labor force) rather than demand destruction (firms laying off). The regime classifier doesn't decompose this in real time.
Second, JOLTS quits and hires data lead unemployment by 1-2 quarters but are themselves noisy. Quit rate above 3% signals labor strength; below 2% signals labor weakness. The 2024-2025 quit rate has been declining toward 2%, suggesting weakening hiring without yet pushing UR materially higher.
Third, BLS revisions to payrolls data have been unusually large in 2024-2025. The annual benchmark revision in February 2025 cut prior payroll estimates by roughly 800k jobs, raising questions about real-time signal reliability.
How to Use This Forecast in Practice
For unemployment, watch the Sahm Rule indicator alongside JOLTS quits and continuing claims. The Sahm Rule is the highest-precision recession indicator but with the 2024 false-positive caveat; quits and claims provide higher-frequency leading signals.
The cleanest cross-check for unemployment is the relationship between UR and the broader U-6 underemployment rate. When U-6 widens versus U-3, labor market weakness is concentrating in part-time and discouraged workers; when both rise together, broad labor demand is contracting. The 68% band on unemployment forecasts is the tightest of any macro variable because of the rate's persistence.
Frequently Asked Questions
What factors could push Unemployment Rate (U3) higher?▾
The primary drivers that tend to lift Unemployment Rate (U3) depend on the current macro regime. The labor market is the backbone of the consumer economy. Rising jobless claims and a climbing unemployment rate are classic late-cycle signals that precede recessions and rate cuts. The Fed has a dual mandate, maximum employment and stable prices, so labor data directly influences the path of monetary policy. Convex tracks these drivers live across the Labor Market category and flags when multiple forces align in the same direction. See the "Key Drivers & Risks" section on this page for the current list, and check the regime dashboard for how the macro backdrop is currently tilted.
What factors could push Unemployment Rate (U3) lower?▾
The same transmission channels that drive Unemployment Rate (U3) higher operate in reverse when conditions flip. The risk drivers listed above map directly to scenarios that, if triggered, would pull this metric in the opposite direction. Convex aggregates these into a scenario-weighted probability distribution rather than a point forecast, so the magnitude depends on which scenarios activate.
Where does consensus see Unemployment Rate (U3) heading?▾
Rather than publish a point target that goes stale the day after release, Convex assembles consensus from the macro regime classification, active scenario probabilities, and historical base rates. Point forecasts from banks and strategists are worth reading for context, but they typically cluster around the consensus and miss the tail events that actually move markets. The scenario-weighted approach here captures that tail risk explicitly.
What is the historical range for Unemployment Rate (U3)?▾
Historical ranges for Unemployment Rate (U3) vary dramatically by regime. A level that is extreme in Goldilocks can be routine in Stagflation, and vice versa. The Historical Volatility section on this page describes the typical range and regime-specific behavior. For the full multi-decade history, visit the Unemployment Rate (U3) chart page, which includes selectable time ranges up to five years and downloadable data.
How often is the Unemployment Rate (U3) forecast updated?▾
This forecast page recalculates whenever the underlying data or regime classification changes, typically within hours of new data releases. The scenario probabilities refresh daily as the macro state is regenerated. Specific drivers listed on this page reflect the current state of the Convex regime engine, not static historical assumptions.
Is this forecast actionable for trading?▾
Convex forecasts are informational and educational. They describe probability distributions and regime-conditional paths rather than specific entry and exit levels. Traders and portfolio managers use them alongside other inputs including position sizing rules, risk management, and their own conviction calibration. They are not investment advice.
Get forecast updates for Unemployment Rate (U3) and related indicators.
Forecasts are model-based projections derived from current regime classification, scenario probabilities, and historical patterns. They are not investment advice. All investments involve risk.