On Wages and Inflation Expectations
Wage Growth Remains Out-of-Whack Compared to the Fed's Inflation Target
Wage growth accelerated sharply throughout 2022, but still less so than actual inflation. This is often seen as a sign that despite high inflation, the U.S. economy at least has not been plagued by a wage-price spiral. However, as wage contracts are not reset every month but rather remain fixed for a year or so, it might be better to compare current wage growth with a measure of expected inflation over the course of the contract. Furthermore, if the average worker produces more per hour than before, this could potentially absorb wage growth that outpaces inflation expectations resulting in stable inflation rates. Any deviation from these trends, however, could drive an acceleration or deceleration of inflation due to wages. In this note I lay out a simple framework that quantifies this and can be used to inform us about the U.S. economic outlook.
More specifically, for a given month I construct
where the first term on the right hand side above is an estimate of inflation expectations in a month and the second term is an estimate of longer term (“trend”) labor productivity growth. Intuitively, this W* measure reflects a simple textbook model where in the end demand and supply in the labor market are in balance when expected real wages equal trend labor productivity growth. In the current environment the Fed would like to see wage growth slowing below this measure, suggesting a decreasing impact from wages on future inflation, and ideally to have it slow to a W* level in line with the Fed's 2% inflation target.
Inflation Expectations
In approximating inflation expectations, I focus solely on inflation expectation surveys drawn from “Main Street” entities, i.e., households and firms. Specifically, I use the following 1-year ahead inflation expectations from the following surveys:
University of Michigan Survey of Consumer Sentiment, with monthly year-ahead inflation expectations since 1978.
The Conference Board Survey of Consumer Confidence, with monthly year-ahead inflation expectations since 1987.
The NY Fed Survey of Consumer Expectations, with monthly year-ahead inflation expectations since 2013.
The Atlanta Fed Survey of Business Inflation Expectations, with monthly year-ahead inflation expectations since 2011.
The Cleveland Fed Survey of Firms Inflation Expectations, with quarterly year-ahead inflation expectations since 2018.
As these surveys measure different aspects of household and firms expectations in different ways, my goal is to aggregate these expectations in a single metric that reflects this uncertainty about the actual level of year-ahead “Main Street” inflation expectations. The above described data contains series that are different in sample size as well as data frequency. A convenient way to compute an aggregate of these survey-based expectations is to estimate a single dynamic common factor that depends on its own lag using the approach of Banbura and Modugno (JAE, 2014).
The individual expectations series are plotted in the chart below alongside the estimated common factor across these series. There is a great deal of heterogeneity across the different surveys, especially since 2020. However, most series accelerated rapidly since early 2021 and peaked around mid-2022. Since then the expectations measures have eased but remain elevated relative to their historical averages. It is the common factor depicted in the chart below that I use to construct the expectations based W* later, and to do that I scale it in terms of the Fed’s favorite inflation gauge, year-over-year PCE inflation.
Trend Labor Productivity Growth
Data on labor productivity for the nonfarm business sector from the BLS’s quarterly “Productivity and Costs” publication is quite volatile, as is clear from the chart below. For my purpose I want use a smoother series to proxy longer term trends in this productivity measure. Without wanting to resort to some structural model, I instead use a number of different purely statistical approaches as outlined in the Box below, and use an average of the 6 estimates as the trend labor productivity estimate (orange line in the chart below). The chart suggests that trend labor productivity growth slowed meaningfully since early 2020, limiting its ability to isolate inflation from any large accelerations in wage growth.
Box: Trend Estimation Methods
I use 6 purely statistical approaches to proxy trend labor productivity: (i) A quadratic trend that implies a linear decrease in labor productivity growth over time; (ii) Successively applying the Hodrick-Prescott filter until unit root tests are satisfied, as suggest by Phillips and Shi (IER, 2021); (iii) The same as in (ii) but until a BIC criterion is satisfied; (iv) Applying the Quast and Wolters (JBES, 2022) version of the Hamilton filter (where the trend results from regressing the current level on many lags); (v) Transforming into quarterly changes and then locally averaging these changes using sliding windows of 33 quarterly observations as in Stock and Watson (BPEA, 2012); (vi) Same as in (v) but now using sliding windows of 25 quarterly observations. There is a large variation across the respective estimates, so I use the average of the 6 estimates to proxy trend labor productivity, which plotted in the chart above. As such this average reflects the uncertainty with respect to 'true' trend level of labor productivity.
Putting Everything Together
I now can combine the trend productivity growth estimate described earlier (linearly interpolated from the quarterly to the monthly frequency), with either the above-mentioned year-ahead inflation expectation proxy based on the common factor from “Main Street” surveys or the 2% inflation target to get alternative W* measures. These are compared in the chart below with Atlanta Fed wage tracker wage series that measure average hourly wages of workers that are robust to compositional changes.
The chart suggests that in the second half of 2022 wages have caught up with one-year inflation expectations corrected for trend labor productivity growth. Heading into 2023, however, wages started to overshoot the expectations based W* metric, suggesting the potential for continued upward inflationary pressures to persist. Note, though, that both actual wage growth as well as the expectations based W* measure still remain far above a level that is consistent with the Fed’s 2% inflation target, and are more consistent with 3.5%-4% PCE inflation rates.
Implications
The 2023 overshoot of wages relative to expectations based W* in the chart above is partly driven by the decline in this W* as inflation expectations eased further throughout the year. This is a good trend for the Fed, as the downward move in this W* suggests that further easing in wage growth could be in store for the remainder of the year.
However, the longer the current wage overshoot lasts as labor demand outstrips supply, the more likely it is that it will feed back into inflation expectations, putting a floor on how far the expectations based W* can ease in the remainder of the year. This overshoot reflects an acceleration in expected real wage growth into positive territory, which boosts expected real wage incomes of household. This will keep consumption at robust levels, and with the ongoing rotation of consumption back to services this could in particular result in upward pressures in core services prices to persist. To avoid this the Fed will remain especially sensitive to continued easing in labor market tightness and a lack of progress on this over the next 2 to 3 months could increase the likelihood of another rate hike at the September or November FOMC meetings.