admin | February 21, 2020
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
The Fed recently has argued that the average US paycheck is tracking productivity growth and inflation, so wages look consistent with a balanced labor market instead of a tight one. But using the Fed’s own framework, a closer look at wage growth over the last 25 years or so suggests that labor markets are, in fact, pretty tight.
Inputs to wage growth
In theory, a firm should willingly pay its workers based on some function of real output from the workers along with some allowance for inflation. Wage growth should track productivity growth, a gauge of the additional increment that each hour of work provides in real output from one period to the next, along with inflation.
There are two complications with this approach. First, the data, especially for productivity, are notoriously noisy. When productivity growth gyrates from year to year or inflation swings higher or lower driven by oil prices, wage gains would not necessarily follow. Second, expectations are probably as important, if not more important, than backward-looking data. For example, as workers negotiate a wage package, especially a multi-year deal, they should want compensation for inflation expected over the term of the contract rather than for inflation in the past. This is one key reason why low and stable inflation helps the economy; it eliminates a key source of uncertainty for business planning.
A survey from the Philadelphia Fed offers a way around some of the noise. A Philly Fed survey in place for decades asks for 10-year projections of productivity as well as inflation. Using these 10-year expectations results offers an escape from the short-term noise in the actual data as well as offering an insight into all-important forward expectations, as opposed to backward-looking data. I use the 10-year productivity projections as well as the 10-year projections for the Fed’s preferred gauge of inflation, the PCE deflator.
We can then look at the sum of long-term productivity and inflation projections as a benchmark for wage gains (Exhibit 1). Going back to the 1990s, the projected trend in productivity growth has decelerated significantly, which is why it is unfair to expect wage gains to be as fast today as they were 20 years ago.
Exhibit 1: Projections of productivity growth plus inflation
New Gauge of Labor Market Slack
The next step in the calculation is to compare actual wage growth to the proxy of trend wage growth. One good metric is the average hourly earnings for production and nonsupervisory workers. Subtracting the proxy from actual wage gains–using the December-to-December increase for each year to construct an annual series—provides a measure of wages relative to one concept of “equilibrium” (Exhibit 2).
Exhibit 2: Wages compared to a proxy of projected wages
Among other things, the series is almost always in negative territory. This could reflect any number of issues, but my first guess would be that benefit costs such as health insurance have generally gone up faster than wages over this period, so workers may be settling for slower increases in base pay to achieve their desired mix of fringe benefits. Alternatively, workers could have simply lost bargaining power over time.
Additionally, and more importantly for our purposes, the series appears to offer a surprisingly reliable gauge of the state of the labor market. During expansions, when labor markets are growing tighter, the series is almost always rising, and then it plunges during recessions, when slack builds quickly. Looking more precisely, the series suggests that wages may respond to the economy with a slight lag, as the big drop in the gap comes a year or two after the recession begins, which is consistent with the bulk of economic research showing that movements in wage growth do tend to lag somewhat. This is a phenomenon that makes sense, as many workers only negotiate wages once a year, or, in the case of unionized workers, even less frequently, so it takes a bit of time for wages to adjust to an economic downturn.
Exhibit 3 compares the gap portrayed in Exhibit 2 with the unemployment rate using an inverted scale. The correlation is amazingly tight, which confirms that the proxy constructed here is indeed a solid indicator of the state of the labor market.
Exhibit 3: Wage gap proxy versus the unemployment rate
Source: BLS, Philadelphia Fed Survey of Professional Forecasters.
Where do we stand?
During an era when productivity growth has varied widely, making it difficult to compare nominal wage growth across business cycles, the proxy created here offers a much better way to compare the state of the labor market over different time periods. The gap registered just below zero in 2018 and 2019, slightly higher than the peak readings seen in the expansion of the 2000s, but just below the best observations of the late-1990s, the last time that the unemployment rate slid below 4%. The gap series jumped into positive territory in 1997, which appears to be an anomaly, but settled at -0.1% in 1998 and -0.05% in 1999. By comparison, the 2018 reading was consistent with the 1998/1999 figures at -0.1%, but the modest deceleration in wages last year took the gap proxy down to -0.35% last year. In any case, that late-1990s period has been widely viewed as the tightest labor market since the 1960s, and the key question that I have been grappling with in recent months is how to compare the current labor market environment with that time. This gauge confirms my sense that the current labor market is either on par with the late-1990s or slightly less tight.
It is worth noting that the deceleration in average hourly earnings growth in 2019 relative to 2018 seems at odds with most anecdotal and survey evidence that the labor market has continued to tighten. Indeed, in the January FOMC minutes, from which I lifted the quote in the introduction, several participants noted that their business contacts indicated that “observed wage growth might currently understate the degree of tightness in the labor market.” If wage gains re-accelerate this year, as I expect, then the gap proxy depicted in Exhibits 2 and 3 will presumably move back to around the zero level, consistent with the 1998/1999 period.
The wage proxy developed in this piece offers support for the view that the labor market is in fact quite tight, though possibly not quite to the extremes reached in the late-1990s – yet. However, it appears to be a far cry from the view of many doves on the FOMC, who have argued that the “absence of stronger broad-based wage pressures” may indicate that “the labor market was not as tight as the historically low unemployment rate would suggest.” After adjusting for the changes in the projected trend of productivity growth, even 3% average hourly earnings growth appears to be consistent with a labor market operating well beyond its long-run equilibrium.