A Gentle Disinflation
ECI came in as expected. The Employment Cost Index (ECI) released today came in as we expected (YoY at 4.5%, as our Phillips curve model expected). The labor market is taking its time, but it is disinflating. In this note, we update our ECI model forecast and discuss the evidence from the “heterodox” ECI Phillips curve. Overall, if the labor market will continue its gentle disinflation, sequential ECI growth could be around the level consistent with the Fed target in a few quarters.
(Note: we did not circulate a preview to limit the number of emails but in private conversations we suggested a +1.0% reading for ECI in Q2 based on our models. We continue to be very puzzled by those arguing that “there is no Phillips curve”, given the performance of the models)
The incoming data
According to the BLS, over the three months ending in June, total compensation for private industry workers increased 4.1% QoQ (a.r.) and 4.5 percent YoY. Wages and salaries increased 4.6 percent for the 12-month period ending in June 2023 and increased 5.3 percent for the 12-month period ending in June 2022. Benefit costs increased 4.2 percent over the year and increased 4.8 percent for the 12month period ending in June 2022.
Our model-based forecast
The inclusion of the second quarter in the sample resulted in a forecast little changed (1 tenth lower at the end of the medium-term). Figure 1 (right panel) shows a comparison between the current forecast, the published figures and the previous forecast (our previous ECI notes are here and here). The published YoY at 4.5% in 2023:Q2 is in line with the model forecast in March (in this sense we are very puzzled by those arguing that “there is no Phillips curve”, given that the model forecasted the YoY at 4.5% in 2023:Q2 already last October).
The YoY of ECI total compensation for private industry workers is now projected at 4.2 percent in 2023:Q4, 3.6 percent in 2024:Q4, and 3.1 percent in 2025:Q4 – see Figure 1. (For the record: our ECI model uses the unemployment gap as a measure of slack. We have run the model using other measures of “slack” and we got similar or even more pronounced results)
Figure 1. Phillips curve model-based ECI forecast
Current forecast, model-based
Comparison: current vs. previous forecast (QoQ and YoY)
Note: the figure shows the ECI Phillips curve model-based forecast (YoY). The chart refers to the year-over-year percent changes of compensation costs for private industry workers. The confidence intervals are based on the estimated parameters distributions. The table on the right shows a comparison between the current forecast (2023:Q2 last quarter in-sample) and the previous forecast (real-time in 2023:Q1).
An heterodox Phillips curve
An heterodox Phillips curve continues to suggest there is no free lunch: wage growth disinflates when labor market cools off. Figure 2 shows what can be labeled as “an heterodox Phillips curve”. The figure shows the traditional relation between ECI total compensation growth (3m changes ar) and a measure of slack. Each point is a month calculated by linearly interpolating the quarterly level of the ECI. As a measure of “slack” we plot the difference between the unemployment rate and the jobs opening rate. The idea is that we remain agnostic about the current slope of the Beveridge curve, although we continue to think that the assumptions behind Figura and Waller (2022) were too optimistic (see here our August 2022 note on the point). While it is unclear what side of the Beveridge curve will dominate going forward, a convincing disinflation that brings ECI growth to about 3% requires our measure of “slack” to be in between -1% and 0. This means that we have at least another 1 to 1.5pp of disinflation to achieve, either side of the Beveridge curve, before wage growth can return to a level consistent with the Fed target. The good news is that there is no mystery through the lens of this “heterodox” Phillips curve: ECI growth is precisely where it should be.
Figure 2. An heterodox ECI Phillips curve (2010-2023 monthly data)
Conclusion
Underlying inflation at 3%. Today’s data imply that “underlying inflation” as defined by Fed staff is around 3%, in line with the evidence of our models (see for instance the trend models in our latest pre-FOMC meeting package). Everything seems to be in place for more good news in the next few months. Risks are now skewed to the downside around the SEP forecast. We act consequently.