After a turbulent week, the big picture remains the same. The Fed staff Taylor rule (and FRB-US) continues to suggest 5%-ish terminal rate, kept at that level in 2023. (The same is true for the ECB with a terminal rate of at least 3.5%). In this note, we show that Friday’s Average Hourly Earnings (AHE) reading was far from friendly. We assess the risks around the inflation forecast.
The evidence
Average Weekly Earnings (AWE) in January grew 15% MoM at annual rate. According to the BLS, in January AWE -the numerator of AHE- grew at the astonishing annualized rate of 15.1% MoM (broad-based across sectors with low chance for a composition bias). The reading is the strongest ever excluding April 2020 (Figure 1). On December 7th when commenting the November AHE reading (see here), we cautioned against a straight read of the data (which was a 0.6% MoM in real time) due to possible (i) revisions, (ii) small sample biases, and (iii) the movement of the denominator (hours). In retrospective, it was the right thing to do. For the same reason (although this time in the opposite direction), today we caution against a straight read of the 0.3% MoM of AHE in January, which in our view is erroneously taken as “benign” or even worse a sign that “the Phillips curve is dead”. The truth is that AWE suggests not only a very tight labor market but also possible second round effects, as (a share of) nominal wages were re-set at a rate well above inflation.
Figure 1. Average Weekly Earnings (MoM, ar)
Where are wages?
We project wages to accelerate (slightly) in Q1. We have run our models (results available upon request) to nowcast 2023:Q1. Figure 2 shows our results. The gray lines in Figure 2 show the QoQ (ar) of all measures of wages and total compensation (8 in total), as well as the mean across measures (the blue line). The takeaway of this exercise is that in 2023:Q1 the models expect wages and total compensation to tick up to 4.5% (ar) and remain close to their post-Covid mean (5.0%, the dotted horizontal line in Figure 2). Therefore, unless the labor market slows down significantly, wage growth is projected persistent and can continue to put some upper pressure on consumers’ prices.
Figure 2. Measures of wages and total compensation (QoQ, ar)
Note: the measures of wages and total compensation (gray lines) in the figure are: Average Hourly Earnings (AHE), Average Weekly Earnings (AWE), the Atlanta Fed Wage Tracker (AFWT), the Employer Cost Index (ECI), the ECI wages and salaries, Compensation Per Hour (CPH), and aggregate compensation (the CPH numerator). All measures are at quarterly frequency. The blue line shows the average across measures. The dotted horizontal lines are the pre-Covid and post-Covid means.
Implications for price inflation
Wages can put upper pressure on core services excluding rents. We have used our “main” Phillips curve model to assess the risks around some labor-intensive sectors. The idea is to estimate if the inclusion of wages (measured by the Employment Cost Index (ECI)) helps improving the in-sample fit and RMSFE of the model when run on sectoral data (when run on core PCE itself the marginal gain is small, if any). Overall, we have identified 5 sub-sectors (all in the services sector) where the ECI has some predictive power: food away from home, intercity transport, intracity transport, medical services, and the “other services” aggregate. The conclusion is that if one looks at the disaggregate data, wages can put upper pressure on consumers prices going forward if they continue at 5% annual rate. As an example, Figure 3 shows the model-based forecast for “food away from home” (which is part of “core” inflation in PCE space). Absent of a large recession, given the estimated persistency of the process and the forecasted wage dynamics, the model projects food away from home inflation to remain well above pre-Covid throughout the entire medium-term. While the evidence in Figure 3 is for illustrative purposes only, it shows well the potential upside risks from wage dynamics (for the record, in January AWE in “leisure and hospitality” grew 27.2% (!) MoM (ar)).
Figure 3. Food away from home (YoY), history and forecast
Note: the confidence bands are calculated via a quasi-out of sample exercise, that is stopping the estimation at time “t”, and computing the forecast error at horizon “t+h” recursively for each “t” and “h”.
Implications for the Fed staff
While the Friday AHE was in line with expectations, in our view it is giving a headache to the Fed staff. In the last few months, we got mixed news and it is not easy to extrapolate a signal. Having said so, the models forecasts are largely unchanged and the estimated terminal rate remains around 5% (kept at that level throughout 2023). The signal from AWE is a reminder that we are still on a mind field. For this reason, in our experience the Fed staff will be careful before changing its message to the FOMC.
(Post scriptum: in our view (and estimates) what happened with NFP and AWE can happen also with the January CPI print, as the risks are to the upside given residual seasonality and the high inflation environment)