The Overall Employment to Population Ratio: Not the Best Summary Indicator, But Not That Misleading, Either

A blog post by researchers at the Federal Reserve Bank of New York has been making the rounds today, arguing that much of the decline in the employment to population ratio (EPOP) since the Great Recession began is actually reflecting changing demographics of the workforce. The researchers (Samuel Kapon and Joseph Tracy) argue that the overall U.S. EPOP in recent years should have been expected to fall relatively rapidly simply because so many U.S. workers were reaching typical retirement ages and were voluntarily leaving paid work. Given this, they claim that the overall EPOP is a misleading labor market indicator if it’s large fall and subsequent non-recovery is taken as evidence that a demand shortfall continues to keep aggregate demand further beneath the economy’s productive potential than other labor market measures—like the overall unemployment—are currently indicating.

I don’t think this is right.

First, as Matthew Klein has noted on Twitter, the same reasoning doesn’t apply if we just look at the EPOP for prime-age working adults—those between 25 and 54. It seems hard to imagine why lots of these workers would be voluntarily retiring starting right at the beginning of the Great Recession. And yet this measure shows the same sharp decline and subsequent very slow recovery as the overall measure.

Second, a key piece of the method used by Kapon and Tracy to estimate the “trend” EPOP actually answers the question that should be answered by the data. This is how they describe one aspect of their method for constructing the trend:

“… we adopt the normalization that over the thirty-one years in our data sample any business-cycle deviations between the actual and the adjusted E/P ratios will average to zero.”

There are two problems with this. First is the assumption that over long periods of time there is no cumulative deviation between actual and trend EPOPs. This essentially means that macroeconomic policymakers are largely successful at keeping aggregate demand growth right in line with productive potential. But there’s clear evidence that the Fed has been much more likely to miss even its own too-conservative NAIRU targets on the high-side for decades before the Great Recession, even when including the rapid growth of the late 1990s—see the figure below and concentrate on the post-1979 segments.


Second, by applying this method to an economy with a large output gap, the normalization will mechanically re-evaluate earlier periods as having higher rates of labor utilization (ie, higher EPOPs) relative to trend. And indeed, they write:

“In addition, based on our normalization, the E/P ratio was 1.6 percentage points above the demographically adjusted E/P ratio just prior to the onset of the recession. This suggests that the labor market was relatively tight prior to the recession”.

Essentially, their normalization defines earlier periods that previously looked on-trend as now being ones of clear over-employment. So in comparison, today’s economy does not look to be operating that far below potential, it’s just that the pre-Great Recession economy was operating quite a bit above potential. Key evidence against this view, however, is that real wage growth for the bottom nine deciles of workers in 2006 and 2007 averaged 0.5 percent. This hardly seems like an overheating labor market.

All in all, I think Kapon and Tracy are right that one should be awfully careful in looking at the overall EPOP and not thinking about the impact of demographics. But I think looking at the prime-age EPOP solves this problem, and this measure does indeed suggest that the other labor market indicators—like the overall unemployment rate—are indeed understating the full degree of labor market slack.