Forecast · BLS data through Sep 2026 · engine pfx-forecast-1.1
1 month · Oct 2026
+0.6%
80% range +0.1% to +1.6%
PFX $6.81
3 months · Dec 2026
+1.5%
80% range +0.4% to +4.6%
PFX $6.87
6 months · Mar 2027
+2.3%
80% range −0.3% to +10.6%
PFX $6.92
12 months · Sep 2027
+3.3%
80% range −1.4% to +22.3%
PFX $6.99
BLS wood pallet price index, last 4 years, and the PFX forecast with 50, 80 and 90% ranges
Show forecast table
| Month | Median | Change | 5% | 10% | 25% | 75% | 90% | 95% | PFX $ median |
|---|---|---|---|---|---|---|---|---|---|
| Oct 2026 | 364.1 | +0.6% | 361.6 | 362.3 | 363.3 | 365.2 | 367.7 | 370.1 | $6.81 |
| Nov 2026 | 365.9 | +1.1% | 362.4 | 363.6 | 364.8 | 368.0 | 372.6 | 375.9 | $6.85 |
| Dec 2026 | 367.3 | +1.5% | 361.7 | 363.3 | 365.4 | 371.3 | 378.3 | 382.6 | $6.87 |
| Jan 2027 | 368.2 | +1.8% | 359.8 | 362.4 | 365.0 | 376.1 | 385.6 | 394.8 | $6.89 |
| Feb 2027 | 369.1 | +2.0% | 359.1 | 361.1 | 364.5 | 380.6 | 393.0 | 405.5 | $6.91 |
| Mar 2027 | 370.1 | +2.3% | 359.1 | 360.9 | 365.2 | 384.4 | 400.2 | 416.9 | $6.92 |
| Apr 2027 | 371.0 | +2.5% | 359.7 | 361.2 | 365.8 | 387.6 | 411.6 | 422.6 | $6.94 |
| May 2027 | 371.9 | +2.8% | 359.2 | 361.0 | 366.4 | 390.3 | 413.1 | 429.4 | $6.96 |
| Jun 2027 | 372.7 | +3.0% | 359.1 | 361.1 | 366.6 | 393.6 | 414.3 | 431.7 | $6.97 |
| Jul 2027 | 373.2 | +3.2% | 357.8 | 360.0 | 365.8 | 397.8 | 423.1 | 445.9 | $6.98 |
| Aug 2027 | 373.6 | +3.2% | 357.0 | 359.0 | 364.9 | 403.8 | 432.4 | 462.4 | $6.99 |
| Sep 2027 | 373.7 | +3.3% | 354.6 | 356.9 | 363.3 | 408.1 | 442.3 | 477.2 | $6.99 |
Scenarios
What if lumber or freight moves?
Move the drivers and watch the forecast respond. Built on the driver regression's estimated sensitivities.
3 months · Dec 2026
+1.5%
$6.87 PFX
6 months · Mar 2027
+2.3%
$6.92 PFX
12 months · Sep 2027
+3.3%
$6.99 PFX
Bars: scenario change from today (warm = up, cool = down). White ticks: the base forecast. Sensitivities come from the driver regression, assuming the shock happens now and holds. Lumber declines use a tenth of the upward slope, because historically falls have barely reached pallet prices. A planning tool, not a promise.
What's driving it
Driver attribution
How far each group of drivers pushes the driver regression away from its average forecast. Warm bars push prices up, cool bars pull them down.
3-month horizon
- Momentum+0.32%
- Lumber+0.13%
- Freight+0.87%
- Demand−0.31%
- Labor+0.01%
- Valuation+0.33%
6-month horizon
- Momentum−0.03%
- Lumber+0.15%
- Freight+1.90%
- Demand−0.84%
- Labor+0.03%
- Valuation+0.69%
12-month horizon
- Momentum+0.34%
- Lumber+0.58%
- Freight+0.83%
- Demand−1.56%
- Labor+0.07%
- Valuation+1.17%
The models
Five models, weighted by their record
The published forecast is a weighted combination. Weights come from each model's accuracy over the last three years of out-of-sample forecasts, so models that have been right lately count for more.
Random walk
benchmarkNo change from the latest value. The benchmark every model has to beat.
Random walk with drift
benchmarkExtends the average monthly change of the last five years.
Autoregressive AR(p)
statisticalMonthly changes explained by their own recent history; lag order chosen by AIC (up to 6).
Damped-trend exponential smoothing
statisticalHolt's linear method with a damped trend on log prices; parameters fit by one-step error.
Driver regression
driverDirect multi-horizon ridge regression on pallet momentum, hardwood and softwood lumber, sawmill prices, diesel, truckload freight, manufacturing and warehouse jobs, wood-products wages, and an error-correction term on the pallet-to-inputs price ratio.
Scorecard
How accurate is it?
| Model | 1 mo · error / vs RW / direction | 3 mo · error / vs RW / direction | 6 mo · error / vs RW / direction | 12 mo · error / vs RW / direction |
|---|---|---|---|---|
| PFX ensemble (published) | 1.2%0.7384% | 3.6%0.8580% | 6.8%0.9476% | 11.3%0.9278% |
| Driver regression | 1.0%0.6486% | 3.4%0.7976% | 6.1%0.8370% | 10.9%0.8973% |
| Damped-trend exponential smoothing | 1.3%0.8081% | 4.0%0.9482% | 7.6%1.0482% | 12.7%1.0473% |
| Autoregressive AR(p) | 1.2%0.7579% | 4.0%0.9377% | 8.3%1.1464% | 13.4%1.1067% |
| Random walk with drift | 1.6%1.0045% | 4.3%1.0260% | 7.5%1.0464% | 13.3%1.0869% |
| Random walk | 1.6%1.00n/a | 4.3%1.00n/a | 7.3%1.00n/a | 12.3%1.00n/a |
| 80% band coverage (target 80%) | 75% | 76% | 82% | 85% |
Out-of-sample rolling-origin backtest, 2016-01 to 2026-08 (128 monthly forecast origins), each forecast made with only the data available at the time. Error is root-mean-square error of the % change. “vs RW” is Theil's U against a no-change forecast: below 1.00 beats it. Direction counts months where prices moved more than 0.5%.
3-month-ahead forecasts vs. what happened
Each point on the forecast line was made three months earlier, using only the data available then (out-of-sample backtest).
- BLS pallet index
- Forecast made 3 months before
Show data table
| Month | BLS pallet index | Forecast made 3 months before |
|---|---|---|
| 2026-11 | 359.5 | |
| 2026-10 | 361.8 | |
| 2026-09 | 361.8 | 356.2 |
| 2026-08 | 356.4 | 351.0 |
| 2026-07 | 356.4 | 353.7 |
| 2026-06 | 351.7 | 343.3 |
| 2026-05 | 347.7 | 341.1 |
| 2026-04 | 347.2 | 341.2 |
| 2026-03 | 340.7 | 338.3 |
| 2026-02 | 339.2 | 338.6 |
| 2026-01 | 338.9 | 338.3 |
| 2025-12 | 337.0 | 337.6 |
| 2025-11 | 337.0 | 338.0 |
Live forecasts are archived every month as published, starting with this one. Full forecast methodology.