Looking Beyond Tomorrow's Non-Farm Payroll Number To Spot A Negative Shift In Structural Unemployment
Goldman chief economist Jan Hatzius has created a useful preview of tomorrow's NFP number (consensus +90,000 private, -65,000 overall), explaining why Goldman has a more negative outlook on the number than most (+75k and -75K, respectively). Jan's conclusion on tomorrow's, and recent trending data :"Our view remains that the primary job market problem is a shortfall in labor demand." More relevantly, Hatzius does an extended analysis of the Beveridge curve (i.e., the relationship between unemployment and job vacancies) to determine if there has been a shift in the overall level of structural unemployment
, as opposed to the more simple seasonal variety. Hatzius' modestly negative conclusion: "The answer is that the vacancy rate has not picked up by enough to push gross hiring sufficiently far above gross separations—i.e., layoffs plus quits—to create large numbers of net new jobs... Structural unemployment may well increase over time if large numbers of people remain without a job for long periods of time, and thus lose their skills and attachment to the labor force. But it is not clear that this process has started yet."
Full note from Jan Hatzius:
Jobs Preview, and Some Thoughts on Structural Unemployment
We estimate that private payrolls in July rose 75,000, close to the average pace of the past two months but more slowly than earlier in the year. Overall payrolls probably declined by about 75,000 as a result of Census layoffs, and the unemployment rate probably edged up to 9.6%. On balance, our forecasts imply that job market conditions have not changed much in the last month.
Today’s comment takes a look at the supply side of the labor market. Some have argued recently that the failure of the unemployment rate to decline significantly over the past 6-9 months despite a notable rise in job vacancies shows that the “matching efficiency” of the US labor market—its ability to turn open positions into jobs—has started to deteriorate. Such a development would suggest that the structural unemployment rate has risen and that there is perhaps not as much cyclical “slack” in the labor market as one might think at first glance.
An increase in structural unemployment is possible, but at this point the conclusion seems premature. Indeed, the increase in job vacancies has coincided with a significant increase in gross hiring over the past year. However, neither vacancies nor hiring have risen by enough to create large numbers of net new jobs. Our view remains that the primary job market problem is a shortfall in labor demand.
We expect Friday’s employment report for July to come in slightly weaker than the current market consensus. We estimate a 75,000 gain in private payrolls (consensus +90,000) and a 75,000 drop in overall payrolls (consensus -65,000), with most of the 150,000 difference accounted for by a drop in Census employment. We also expect the unemployment rate to edge up to 9.6% (consensus 9.6%) and average hourly earnings to show a 0.1% gain (consensus 0.1%). This would mean that payrolls will show a gain well below the average of March/April (200,000) but somewhat above the average of May/June (58,000).
Our forecast this month is not far from the consensus, so instead of dwelling on the strong and weak points in the recent labor market data—short version: claims inconclusive, ISM employment indexes a bit stronger, consumer job perceptions weaker, online hiring indexes firmer—we will devote today’s comment to a longer-term labor market issue. This is whether we are seeing the first signs in the labor market data of an increase in the “structural” rate of unemployment. There are many different angles on this issue, and we only focus on one today, namely what to make of the recent shift in the relationship between unemployment and job vacancies. (Economists refer to this relationship as the “Beveridge curve”, named after William Beveridge, a British economist and civil servant in the first half of the Twentieth Century.)
The chart below shows the “breakdown” of the Beveridge curve, plotting the unemployment rate on the horizontal axis and the job vacancy rate on the vertical axis. After tracing out a near-perfect downward-sloping curve for most of the past decade, the observations over the past few months have been notably above and to the right of the previous relationship. Such a move is often viewed as a sign that the “matching process”—the efficiency of the labor market in bringing together jobs and job-seekers—is deteriorating (for more on this issue, see the blog post by David Altig, head of research at the Atlanta Fed: http://macroblog.typepad.com/macroblog/2010/07/a-curious-unemployment-picture-gets-more-curious.html).
However, the evidence for a breakdown in matching efficiency is far from clear-cut. For starters, reliable job vacancy data (from the Labor Department’s Job Openings and Labor Turnover Survey, or JOLTS) are only available back to 2000, so the plot of the Beveridge curve above is only based on one-and-a-half business cycles. This means that the “breakdown” of the Beveridge curve relative to prior patterns is effectively based on a very small number of observations
Moreover, we would note that while the unemployment rate remains very high, gross hiring has actually picked up significantly. As shown in the chart below, the increase in hiring has been roughly what one would have expected relative to the increase in vacancies, again based on the experience of the past decade as reflected in the JOLTS data. Admittedly, one might argue that hiring should have picked up even more than it has, since there are now a lot more job-seekers per vacant job than in the past. But historically, the number of gross hires has been more closely related to the number of vacancies than to the number of unemployed. In our view, this makes sense, because gross hiring can come not just from currently unemployed workers, but also from the currently employed as well as the economically inactive. The fact that the link between vacancies and hiring still seems broadly intact therefore casts doubt on the notion that the efficiency of the US labor market in filling job vacancies has deteriorated significantly—or at minimum, renders such a verdict premature.
So why is the labor market still stuck in the doldrums? The answer is that the vacancy rate has not picked up by enough to push gross hiring sufficiently far above gross separations—i.e., layoffs plus quits—to create large numbers of net new jobs. But that is a statement about the demand for labor, not about the efficiency of the matching process and the structural rate of unemployment. To be sure, structural unemployment may well increase over time if large numbers of people remain without a job for long periods of time, and thus lose their skills and attachment to the labor force. But it is not clear that this process has started yet.