In our estimates, today’s AHE reading likely overstates the “true” pace of wage growth. In November 2022 (see here our Dec 7th 2022 note), AHE grew 55bps MoM sa. In that note, we argued that the published number was overstating the “true” pace. In retrospective, it was the correct call (Nov ’22 got revised down significantly in subsequent revisions). We estimate that today’s AHE reading is overstating the underlying pace of wage growth for the same reasons we described in Dec ’22: Average Weekly Earnings (AHE numerator) came in very weak (in fact, negative), and our AHE model confirms the impression. Overall, we continue to estimate that (AHE) wage growth is around 3.5%-4.0% at annual rate.
(A PPT containing all US wage charts has been updated and posted here)
The facts
Average Hourly Earnings rose 55bps in January. The reading is driven by a large drop in average weekly hours (the AHE denominator). In January, there are no visible distortions from seasonal adjustment issues as in previous months (i.e. here). Rather, the super strong January reading is driven by a large drop in hours, possibly related to weather, unlikely to repeat going forward. The effect of residual calendarity is negative in January, while the contribution of employment shifts across sectors is positive (essentially because the share of employment in “Education & Health Services” increased while it dropped in “Leisure and Hospitality” which has lower AHE level and MoM change than the aggregate average).
Note: please note that the YoY of AWE is at 3.0% in January vs 4.5% of AHE. The gap is large and quite unusual, likely in our view to resolve in favor of AWE going forward.
Note2: if the reader is unfamiliar with the notion of “residual calendarity” in AHE space, (s)he can refer to this BLS page, and to our previous notes here, here, and here.
The AHE model
The “true” AHE pace is around 3.5%-4% (ar). Figure 1 shows an update of our AHE monthly model. As a reminder, this model explains monthly AHE growth rates using recent past, a variable capturing “residual seasonality”, a variable capturing economic “slack”, and sectoral employment shares. In January (latest point on the chart), the fitted value of the model is way below the published AHE growth rate: 31bps vs 55bps. In other words, the model does not understand today’s reading, partially because it does not control for hours, and partially because AHE does not control for employment share shifts.
This is not the first time there is a large gap between published data and the model predicted values, especially post-Covid given the volatility of hours and employment shares. We continue to take the blue line as the “true” signal, given it has provided good guidance so far. The overall message has not changed: the fitted values (blue line) are going roughly sideways, as argued in recent months.
(For the record, the recent predicted values of the models are: Sep (39bps), Oct (34bps), Nov (35bps), Dec (44bps), and Jan (31bps)).
Figure 1. AHE MoM (sa) and UnderlyingInflation AHE model fitted values.
Note: the figure shows published MoM (sa) AHE growth rates (red line) and UnderlyingInflation AHE model fitted values (blue line). The figure excludes the first 6 months of Covid for scaling issues (although they are included in the model estimation).
Why sectoral shifts are crucial. We often get the following question: how important are sectoral employment shifts for AHE growth? Figure 2 clarifies why it is very important to control, somehow, for sectoral employment shifts. Figure 2 shows the published AHE YoY growth rate (the yellow line), and the share of employment in industries with AHE level above the aggregate mean. Figure 2 is not meant to say that industry shifts can explain everything. Rather, it is just a reminder not to take at face value the MoM (or YoY) of AHE when it pops up on your Bloomberg terminal.
Figure 2. YoY of AHE and Sectoral Employment Share.
What to expect in February
Soft AHE in February. According to the model, AHE MoM (sa) growth in February is expected at 0.2508%. Please, remember that the model assumes no shift in employment shares across sectors and constant aggregate hours (we cannot forecast the weather..).
Conclusion and implications for the Fed
Unlikely to change the Fed staff view. A very solid report, no doubt. However, as shown, the AHE figure is probably distorted. In our estimate (and we suspect in those of the Fed staff as well), the strong January reading is expected to fully reverse next month. For this reason, we think that the Fed staff and the FOMC members will be happy today but unlikely to change their views based on today’s report only.