Jobs Report, First Impressions: A Tale of Two Surveys

May 2nd, 2014 at 9:24 am

Payroll growth was strong and broad based last month and the jobless rate fell sharply, but that seemingly impressive decline in unemployment was wholly due to the shrinking labor force.  In other words, a confusing month of labor market data as the two surveys reported on today tell very different stories.

The survey of establishments recorded strong payroll gains in April of 288,000, above the expected addition of 215,000.  The household survey reported that the unemployment rate fell by 0.4 tenths, to 6.3%.  That’s the largest monthly decline since December 2010 and the lowest jobless rate since September 2008.

So, as my Grandma might have asked, what’s not to like?

It’s this: the decline in unemployment is entirely due not to job creation, but to labor force decline (employment actually fell slightly in the household survey).  This important and closely watched indicator—the labor force participation rate—also fell 0.4 tenths, reversing recent gains and returning the lfpr to its low where it stood at the end of last year, commensurate with levels we haven’t seen since the late 1970s.  Though part of the recent decline in the participation rate reflects our aging demographics, more than half in my judgment is due to weak demand.

The BLS noted that the large decline in the labor force—about -800,000—was likely due to fewer entrants as opposed to more leavers.  And this is a volatile number, as I stress below.  But neither can it be dismissed out of hand: it has been essentially stuck at historically low levels for a while now.

On the other hand, the payroll report shows pretty decent labor demand/job creation.  As noted, the 288,000 jobs beat expectations, and gains for the prior two months were revised up by 36,000.  Job gains occurred across most industries, with 67% of private industries expanding employment, the largest share in over two years (government employment was also up 15,000, almost all due to local government; federal employment was down slightly).

As I noted yesterday, you’ve really got to smooth these payroll gains out to get a more reliable read on the monthly employment changes.

Average change in jobs over the past:
3 months: 238,000
6 months: 203,000
12 months: 197,000

So by this metric we’re wiggling around a trend of 200,000 per month, a decent and solid, if not stellar trend, though one that should be strong enough to lower the unemployment rate for the right reasons: not by shrinking the labor force but by offering more employment opportunities to all comers.

A few other notable indicators, and more to come later in the day:

–Long-term unemployment as a share of the labor force fell from 2.4% to 2.2%, the lowest it’s been since early 2009, though still highly elevated in historical terms.

–Both hourly wage growth and weekly hours were flat over the month, and these measures come from the more positive establishment survey.  Over the past year, average hourly pay is up 1.9%, slightly ahead of recent inflation readings, which have tracked around 1.5%.

Summing up, when I see a positive payroll report and negative household report, I tend to give more weight to the payroll survey, which has a much larger sample, smaller confidence intervals around its monthly estimates, and less volatile monthly jumps.  So I’d call this a positive report, but with a big, nervous-making asterisk.

Print Friendly, PDF & Email

7 comments in reply to "Jobs Report, First Impressions: A Tale of Two Surveys"

  1. Jim Neal says:


    Cursory review on my part at what sorts of jobs were reporting gains. Appeared somewhat split between lower-wage retail + food/beverage services offset by professional/business services. Construction was also up; I don’t know off the top of my head the wage profile for that cohort.

    Notwithstanding, the 800,000 person decline in the labor force (as you indicated) is sobering.

    Can you comment on both the gains in employment (the 280K stat) coupled with the sorts of wages that those workers were earning? I’m curious to know if these are generally higher-paying or lower-wage jobs.


    Jim Neal

    • Jared Bernstein says:

      I’ve been meaning to get to that–job quality–but it will take some crunching so will be a few days.

      • Jim Neal says:

        Thanks. Just spent an hour on the phone with the BLS. Numbers as you well know pose a quagmire coming from two different surveys. I’m learning. You have to dig deeply into the weeds before making any grand pronouncements (which of course every Administration does, and suspect the WH will spin later today).

        Looking forward to your insights. jim

  2. peter thom says:

    I think you haven’t done the math here yet Jared.
    The relatively huge size of the baby boom generation is having an outsize effect on the LFPR via the 10,000 retirees from that cohort each day. This Nov. 2013 Philly Fed study (linked below) concluded: “the decline in the participation rate in the past one-and-a-half years (when the unemployment rate declined faster than expected) is mostly due to retirement.” And according to the authors the second largest factor was “nonparticipation due to enrollment in school.” I read recently that there has been a large increase in summer school enrollment among the 16-19 cohort since 2000 and many more in the 20-24 cohort are attending colleges. Whether the latter has been due to weak job prospects is irrelevant. Moreover, the 300,000 retiring each month for almost the next two decades is going to keep the LFPR at historical lows for at least the next decade. The latest conspiracy theory bandied about today (not by you Jared) goes something like this: vast hordes have simply given up looking for work and that liar Obama is masking much higher unemployment. Just look at how the LFPR hasn’t budged.
    However, this conspiracy theory is based on faulty math due to a thorough ignorance of changing demographics.

    • Jared Bernstein says:

      Here’s my take on this question. I think at least half of the decline of the lfpr from its peak is demand driven.

      • Peter Thom says:

        You may be right Jared. But as Plumer points out Fujita’s take is quite nuanced: per Plumer, “Demographics, he argued, didn’t play a huge role in the labor-force drop between 2007 and 2011. But since then, retirements are responsible for basically the ‘entire’ fall of the participation rate. One possible reason is that many older Americans postponed retirement immediately after the financial crisis to rebuild their battered 401(k)s. By 2012 or so, they began retiring en masse.” This makes more sense to me than a static view of causation for the LFPR decline.

  3. jay says:

    Someone really needs to number crunch the participation rate because it’s a game changer. You can’t have high GDP growth with such a weak participation rate.