Regarding quality, the Obama administration took a strong step in the right direction last week with its new overtime rule making millions more low- and mid-wage workers eligible for time-and-a-half when they work over 40 hours per week. In this post, I’d like to talk about another institution that’s absolutely critical for job creation and the reaching our destination of full employment: the Federal Reserve.
The Fed’s fingerprints were all over the government’s response to the Great Recession and recovery that followed. Economists Alan Blinder and Mark Zandi estimate that the financial policy response to the downturn, of which Fed monetary policy was a key component, raised employment by more than 12 million jobs compared to a scenario without such aggressive interventions.
Though there will always be critics, the broad consensus among economists is not only did the Fed perform its countercyclical function well in the recession, but since around 2010, when the tea-party-addled Congress checked out on economic (and most other) policy, the Fed’s also been the only game in town.
Lately, however, its campaign to “normalize” the interest rate it controls — i.e., raise it back up to where it usually is at this point in an economic expansion — has been criticized by some Fed watchers, especially those of us who a) worry that Fed rate hikes will block to the path to full employment, which we still haven’t reached, b) want to see less-advantaged workers benefit from growth, and c) see little in the way of inflationary threats.
For Fed officials who’ve spoken out on this point, like Jeffrey Lacker in this recent interview, the rationale for hiking seems to come from a somewhat vague sense that rates have been low long enough and the U.S. economy, if not many others, is finally firming up.
Meh. I’m not a central banker, but some this sounds like what I think of as MONETary policy, more impressionism (“getting behind the curve,” “reducing uncertainty,” rates are below where they “ought to be”) than hard-core analysis of the key relations between the variables in question.
If I wanted to justify a rate hike, I’d examine two fairly clear, observable trends: slightly faster wage growth and slower productivity growth.
The former, which I discuss/show here, potentially drives inflation in a pretty intuitive way. As employers’ labor costs rise, to maintain their profit margins they either have to raise prices or productivity (produce more output per hour). But given slower productivity growth — it’s been notoriously sluggish in recent years — that latter absorption mechanism is constrained (of course, the productivity slowdown represents an average across firms and sectors, so some firms can likely pursue this offset while other cannot).
Back in the latter 1990s, when the job market was tightening and wages were starting to catch a real buzz, many economists were pushing Fed chair Alan Greenspan to raise interest rates. But he recognized that faster productivity growth could allow for noninflationary wage growth and, to his credit, ignored their warnings. The unemployment rate eventually fell to 4 percent, inflation was “well-behaved,” and a lot of low-wage workers finally saw some serious real wage gains (e.g., the 20th percentile real wage rose 11 percent, 1995-2000; over the past five years, it’s down 4 percent).
So, does slightly faster wage growth amid slower productivity growth really warrant a rate hike?
Here’s where you need to look at some correlations, and the extent to which these variables move inflation these days. What I find when I do so is … wait for it … bupkiss, a Yiddish word derived from — you should excuse the expression — goat droppings, i.e., nothing.
The figures below tell the story. Each one uses a statistical tool (see data note for details) to associate inflation with the different variables in question. I simulate an increase in the wage variable and estimate the impact on price growth. This is an admittedly simple approach and there are other, maybe better, ways to get at this, but I suspect they’d yield similar findings (plus, I really like “simple”).
For example, the first figure uses data since the early 1980s to show the impact of a boost in compensation to the Fed’s favored inflation measure, the core PCE. The dotted lines measure the confidence interval around the estimate of the impact over 2½ years (10 quarters). Inflation hardly budges and to the extent that it does, the confidence interval includes zero, meaning the estimates are statistically insignificant. If you plug in wages — compensation minus benefits — you get the same result.
I believe that both globalization and the Fed itself have had profound impacts on inflation dynamics in more recent years, suggesting that these exercises make more sense when you restrict the sample to start in the mid-1990s. As the next figure shows, that dampens the comp/inflation relationship even more. It starts out negative (faster wage growth implying less inflation) and is even less significant.
What about productivity? In the next figure I replace compensation growth with the growth in compensation relative to productivity (a.k.a., “unit labor costs,” or ULCs), sticking with the post-1995 sample. There’s tiny bit of a significant impulse around quarter 7, but these are pretty much more goat droppings.
You’re welcome to try this at home (I provide a link to the data in the note below) but end of the day, if not by the Fed meeting next month, I expect you’ll find that while wage and price growth are slowly firming up a bit, even amid low productivity growth, it’s still hard to see a rationale for hiking rates. In fact, while there’s still slack in the job market — e.g., 6 million involuntary part-timers — we’re getting closer to full employment and that’s finally giving wages a nice nudge without creating inflationary pressures, and least none I can tease out of the data.
If such a trend is truly emerging, we should celebrate and prolong it, not stop it.
[Data note: The statistical analysis uses quarterly, year-over-year percent changes in the employment cost index, compensation and wage measures, the core PCE deflator, and ULCs (non-farm business sector). These are run through a VAR (vector autoregression, with 4 lags of the price and wage variables, and one control variable: the annual change in the broad trade-weighted index of the dollar. The figures show impulse-response functions of a one standard deviation shock in the wage or ULC variables, with confidence bands of two standard errors. You can download the data here.]