As we pointed out previously, the growing convergence between BLS-reported initial jobless claims (at 42 year lows) and reported job cuts (highest since 2009) suggests someone is lying. It appears we have found the cuplrit as Goldman Sachs confirms that changes in gross labor market flows (e.g. gross hires and quits), as well as changes in the unemployment insurance benefit take up rate, affect the relationship between jobless claims and employment growth over the cycle. For this reason, today’s low level of jobless claims should probably not be taken as a sign of a booming labor market.
Something changed when QE3 ended...
And now Goldman Sachs confirms...
Although payroll employment growth has slowed in recent months, initial claims for unemployment insurance benefits remain very low. The four-week moving average of initial claims has trended lower again this year—despite meaningful layoffs in energy-producing states—and is currently at the lowest level since early 2000 (Exhibit 1). Does this mean that the current rate of nonfarm payroll growth understates the strength of the labor market?
Not necessarily. As we have noted in prior research, the structural relationship between jobless claims and employment growth changes over the business cycle. Unemployment insurance claims are an observable proxy for one type of labor market flow: the number of persons laid-off each month. However, employment growth is a function of other flows as well—specifically, the number of persons hired, the number who quit voluntarily, and those who separate from employment for other reasons. These other types of labor market flows—other components of Fed Chair Yellen’s labor market “dashboard”—can affect the relationship between layoffs and employment growth over time.
Moreover, initial jobless claims are an imperfect measure of layoffs because the propensity to file a claim—often called the “filing rate” or the “take up rate”—also changes over time. During the financial crisis, for example, the benefit take up rate increased significantly. Exhibit 2 shows the level of jobless claims alongside the measure of total layoffs from the Job Openings and Labor Turnover Survey (JOLTS) (claims here are expressed as a monthly rate by multiplying the average weekly rate by the number of weeks per months). Before 2007, approximately 70-80% of layoffs resulted in an unemployment insurance benefit filing. During the recession, claims increased more rapidly than reported layoffs, implying an increase in the claims filing rate. In the years since, claims have fallen much faster than layoffs, implying a decline in the benefit take up rate.
Because of these confounding effects, it can be helpful to think of a “breakeven rate” for jobless claims—or the level of claims consistent with zero employment growth. We can arrive at this figure by using the identity that relates gross labor market flows to changes in employment:
Change in Employment = Hires – Quits – Layoffs – Other Separations
If we then express layoffs as jobless claims multiplied by the inverse of the take up rate, we can write the breakeven level of claims as:
Breakeven Claims = (Hires – Quits – Other Separations)*[Take Up Rate]
This definition says that the breakeven level of jobless claims rises with the benefit take up rate and with the number of new hires, and declines with increases in quitting and other separations.
Exhibit 3 shows our calculated breakeven rate for jobless claims, which we derived using data from the monthly JOLTS reports through July (we smoothed the breakeven rate for the purposes of this graph). Over the last few years, the breakeven level of jobless claims has steadily declined, reflecting an increase in job market separations (including quits) and a lower benefit take up rate. These trends were partly offset by an increase in gross hires. While it’s encouraging that unemployment insurance claims remain very low, they are not a sufficient indicator of labor market conditions. And at the moment, the raw level likely overstates the underlying pace of job growth. The gap between claims and their estimated breakeven rate—likely a better indicator of job market health than claims alone—points to payroll growth of about 160-200k per month, or only slightly above the run rate over the last two months.
We think this is a more realistic signal to take from the current level of jobless claims.