As the stock market works out its manic episode of the past few days, let’s get into a question of great importance: if wage growth is really accelerating, what will that mean for price growth?
This relationship is at the heart of the market selloff that’s got everyone pretty freaked right about now. As I wrote in the WaPo this AM:
The wage pop [last Friday’s 2.9% growth in hourly wages] spooked the markets because investors, already skittish as valuations were a bit steep (though not as bad as people have been saying, given strong current and expected corporate earnings), envisioned this sequence: wage growth gooses price growth (i.e., inflation), which raises both market and Federal Reserve interest rates, which slows growth and shaves corporate profit margins.
I then wrote: “But there are many links in that chain. First, the correlation between wage and price growth has been low in recent years.”
The figure below gets at what I mean by running a simple regression of price growth, using the core PCE, on wage growth, using the ECI wage measure for private sector workers (annualized quarterly growth rates). With data back to 1980, I ran a rolling regression with 10 year samples, so the first coefficient plotted below is for 1980q2-1990q1, the second is for 1980q3-1990q2, etc. to track the movement in this correlation, with a 95% confidence interval around the estimates. As you see, it goes up and down, was essentially zero for a while, and has been sliding down in recent years.
Source: BLS, BEA, my calculations
Next, assuming continued, gradual acceleration of wage growth–I allow wages to get up to a pretty hefty 3.5% by 2019Q4–we can use this simple model to predict future price growth. But given the variation shown above in the coefficients that map wage growth onto price growth, which coefficient should we choose? The table below shows the sensitivity to these choices. If I use the elasticity (price gains with respect to wage growth) from the full sample, the model predicts inflation hitting 2.8% by the end of 2019; if I limit the sample to the 1980s, when the elasticity was at its highest, prices hit 3.7% at the end of 2019, before which point the Fed would surely slam on the brakes.
But if I use the elasticity of the last 10 years, prices grow to a mere 2%, and the Fed is all like, “3.5% wage growth…no problem!” That elasticity choice makes sense, but views will legitimately differ.
The punchline is the same one I’ve been touting for a long time: these key macro-relationships move all over the place, and you can’t assume you know where things are headed based on past results. Recent correlations, at least from my simplistic approach (though more elaborate models find the same thing), suggest that faster wage growth will only very slowly bleed into price growth. So, at least until it’s clear what’s happening with both wage growth and price growth, the markets should probably find something else to freak out about.
Share the post "Will wage growth boost price growth? Sure, but how much depends on whether the near past or the distant past is prologue."
Please explain to stupid me, and I’ve read this somewhere else by someone else why would wages which constitute 15% to 20% of costs rise at all even if we gave all workers a real 5% raise above inflation which would increase prices 1% so you’d have 3% percent inflation instead of 2%, or with just a very low productivity increase of 1% you’d have zero inflation. What am I missing?
*it makes better sense to read it as “why would inflation ….rise at all…” obvious in the latter half of the text
The whole Phillips curve story is a conflation of causation with correlation. It makes sense only if the supply of last-hired, first-fired workers is the constraining factor on output. There is no theoretical reason to assume that full-employment is the same thing as potential output (It’s just one of many possible candidates). The constraining factor on output could easily be some other factor of production or structural constraint (e.g. market power over prices). Too many lazy economists using a formula that they learned (but didn’t understand) in graduate school.