Orlando's Outlook: Wonky August payrolls miss yet again


Bottom line This morning’s labor-market headlines for August were a disappointment across the board, as nonfarm payrolls missed consensus estimates by 24,000 jobs last month, with downward revisions totaling 41,000 jobs in June and July. Moreover, household and government hiring both fell last month, the unemployment rate ticked up, hours worked declined and temporary hiring was unchanged for the first time in eight months.

This weakness, however, is completely inconsistent with broad-based strength we have seen recently across the economy, such as the stronger-than-expected GDP revision for the second quarter to 3% (highest in two years), positive net trade trends for the U.S. and powerful consumer and manufacturing spending in July. Even within the labor market, Wednesday’s ADP report for private hiring was the strongest in five months, while the initial weekly jobless claims for the August survey week remain just off a 44-year cycle low, which means that Hurricane Harvey had no meaningful impact on today’s report.

So what’s the problem? We reminded investors a month ago that August tends to be the statistically quirkiest month of the year for the labor market, with sizable underperformance when the initial report is released. But this initial miss usually gets revised away during the month’s two subsequent revisions. As a result, we continue to believe that economic activity is accelerating here in the U.S. at present, and that today’s disappointing labor-market report for August will be revised sharply higher over the next two months.

Why is August historically so quirky? According to our research friends at Berenberg Capital Markets, since 2011, March is the most volatile payroll month due to weather problems, with August a close second due to vacations:

  • Many of the government’s data-collection employees at the Bureau of Labor Statistics (BLS) take their summer vacations in August, leaving important calculation functions to prepare the report temporarily short-staffed.
  • Teachers, administrators and staff return from their summer vacations for the new school year.
  • High school and college students quit their summer jobs and paid internships to return back to school.
  • Manufacturers—such as the auto industry—close their factories for a few weeks in August and have employees take temporary furloughs to retool and repair equipment and attempt to right-size inventories relative to demand.

All of these timing issues are resolved by the end of October, and we will know the correct August labor-market information by the second revision on Nov. 3. But temporary quirks abound for now:

  • Today marks the seventh consecutive August in which the month missed consensus nonfarm payroll expectations to the downside.
  • Since 2011 (inclusive), the last six Augusts (excluding today) averaged a flash miss of 45,000 jobs.
  • Since the Great Recession ended in June 2009, the difference between the flash report and the second revision has been an average gain of 65,000 jobs.

If this historical pattern holds, today’s disappointing flash report of only 156,000 jobs added could be revised to a strong gain of 221,000 jobs in two months’ time, close to our own 220,000 estimate for August and well above today’s consensus forecast of 180,000 jobs.

Powerful August ADP The ADP National Employment Report, a forward-looking proxy for private payroll growth, reported an eye-opening gain of 237,000 jobs in August – its strongest reading in five months – which blew away consensus expectations for a gain of 185,000 jobs. In addition, July was revised sharply higher to a gain of 201,000 private jobs. Small firms with fewer than 50 employees added 48,000 jobs in August (20% of total, compared with July’s downwardly revised 49,000, or 24% of its total); midsized companies with 50 to 500 employees added 74,000 (31%, compared with July’s upwardly revised 85,000, or 42%); and larger companies with more than 500 employees hired 115,000 (49%, compared with July’s upwardly revised 68,000, or 34%).

Jobless claims remain low, survey was too early for Harvey Initial weekly unemployment claims (a leading economic and employment indicator) for the survey week that ended Aug. 12 was only 232,000, down slightly from July’s survey week reading of 234,000 claims. That’s how we know that Hurricane Harvey’s devastation in Houston did not have any meaningful impact on today’s labor report. The employment survey for August was completed on August 12, but Harvey hit Houston only a week ago, on August 25. However, we’re bracing for a temporary labor-market spike in claims later this month, as newly displaced workers in Texas and Louisiana begin to file unemployment claims due to Harvey.

Disappointing August nonfarm payrolls added a much weaker-than-expected gain of only 156,000 jobs, significantly below the Bloomberg consensus of 180,000 and our own more optimistic 220,000 estimate here at Federated. The BLS revised June and July lower by a combined 41,000 jobs. June’s preliminary gain of 222,000 jobs, which was revised up to an increase of 231,000 last month, was revised back down to a final gain of 210,000 this morning. July’s preliminary gain of 209,000 jobs was revised down to a gain of 189,000 this morning.

Private payrolls soft, but firmer than government hiring, adding a weaker-than-expected 165,000 jobs in August, below the Bloomberg consensus of 172,000. The BLS revised June and July up by a combined 10,000 jobs. June’s preliminary gain of 187,000, which was revised up to an increase of 194,000 last month, was revised up again this morning to a final gain of 207,000. July’s preliminary gain of 205,000 jobs was revised down slightly this morning to a gain of 202,000.

Government hiring fell for the second consecutive month, as the difference between private and nonfarm payrolls collectively subtracted 9,000 federal, state and local government jobs in August. This data series has been very volatile and prone to downward revisions, as July was revised down from a preliminary gain of 4,000 jobs to a loss of 13,000, while a strong increase of 37,000 jobs in June was revised down to a paltry final gain of only 3,000. We also lost 8,000 jobs in May, gained 13,000 in April and lost 9,000 in March. It was government job losses across the board in August, as the feds lost 1,000 jobs, the states slashed 5,000 and local governments cut 3,000.

Household jobs fall, as the admittedly volatile household survey lost 74,000 jobs in August, in sharp contrast to the powerful gains of 345,000 jobs in July and 245,000 in June. The household survey had lost 233,000 positions in May, but had added 156,000 jobs in April, 472,000 in March and 447,000 in February. The household survey is a leading indicator for nonfarm and private payrolls, so August’s decline is disconcerting, particularly as we await future temporary payroll losses due to Harvey. It also serves as the basis for the unemployment rate (U-3).

Unemployment rate up, while labor-impairment and participation rates unchanged The civilian labor force rose by only 77,000 workers in August, a much softer pace than the robust addition of 349,000 workers in July and 361,000 workers in June. May had suffered a sharp loss of 429,000 workers. That contributed to an increase in the unemployment rate to 4.4% in August, up from 4.3% in July, which matched its lowest level since February 2001. The labor-impairment rate (U-6)—also known as the “total” rate of unemployment (or the underemployment rate) because it more broadly includes discouraged workers and the underemployed—was unchanged for the third consecutive month at 8.6% in August, slightly above May’s 8.4%, its lowest level since November 2007. But that’s still down sharply from 9.4% in January, so the economy is doing much better thus far this year. Finally, the labor-force participation rate (the share of working-age people in the labor force) was unchanged at 62.9% in August, still just below its two-year high of 63% in March, but above its 38-year cycle low of 62.4% set in October 2014.

Wage growth flat, but hours worked declines Hourly wages ticked up on a month-over-month basis, compared with a stronger 0.3% gain in July. On a year-over-year (y/y) basis, wage growth was unchanged for the third consecutive month at an increase of 2.5% in August, compared with December 2016’s 8-year, cycle-high gain of 2.9%. The average private work week for all employees ticked down in August to 34.4 hours. But that’s still marginally better than the 2-year low of 34.3 hours in both March and February. This is discouraging, as a change of 0.1 hour worked is the equivalent of adding or subtracting an estimated 350,000 or more jobs to or from the economy.

Construction and manufacturing both soar In what was the most encouraging part of today’s labor update, manufacturing added 36,000 new jobs in August (a new 3-year high), up from a positively revised gain of 26,000 in July (originally reported as a gain of 16,000 jobs) and an upwardly revised gain of 21,000 jobs in June (up from a gain of 12,000 jobs, which was originally reported as a paltry increase of only1,000). Coming from breakeven in May, manufacturing is clearly gathering momentum, perhaps buoyed by the weak dollar and strong net trade trends. Construction hiring rose by 28,000 new jobs in August, sharply higher than the downwardly-revised loss of 3,000 in July (originally reported as a gain of 6,000), versus gains of 15,000 in June and 7,000 in May, compared with breakeven levels in each of April and March. But February had gained 54,000 jobs, the largest gain since March 2007.

Retail may be bottoming, amid a good Back-to-School (BTS) retail shopping season, adding 1,000 jobs in August after losing jobs for six consecutive months: 2,000 jobs in July, 4,000 in June, 10,000 in May, 4,000 in April, 40,000 in March and 29,000 in February.

Temps break even after gains for seven consecutive months: 10,000 in July, 11,000 in June, 15,000 in May, 2,000 in April, 13,000 in March, 10,000 in February and 15,000 in January. Temps are an important leading indicator of employment growth, so this downtick is discouraging for September.

Leisure slumps This economically sensitive category added only 4,000 new jobs in August, after adding 58,000 in July, 38,000 in June, 33,000 in May and 60,000 in April. So we’ll be watching this disappointing slump closely.

Education & health moderated in August, adding only 25,000 new workers, compared with 54,000 jobs in July, 40,000 in June, 37,000 in May, 45,000 in April, 16,000 in March, and a strong 66,000 in February. We had expected a stronger bump with the start of school.

JOLTS soar to record high The Job Openings and Labor Turnover Survey (JOLTS) measures labor-market dynamics such as resignations, help-wanted ads and the pace of hiring (with a one-month lag) to provide some context to general employment trends. In the most recent data available, job openings in June rose 8% on a month-over-month (m/m) basis to a record 6.163 million. Hires fell nearly 2% m/m to 5.356 million in June, the job openings rate rose from 3.7% to 4%, the hires rate remained at 3.7%, and the quits rate slipped from 2.2% to 2.1%. The sharp jump in openings suggests that companies continue to struggle finding workers with the right skills. In the past, this has led to upward wage pressure, as employers must pay more to attract qualified applicants, but this has not yet developed.

Challenger cuts rise Challenger, Gray & Christmas reported 33,825 layoffs in August, a nearly 20% month-over-month increase from July and 5% higher than a year ago. But July represented an 8-month low, and only three times in the last decade has this monthly measure dipped below 30,000 job cuts. So it appears that the August reading represents a return to more typical layoff levels, with retail and construction accounting for the bulk of the cuts.

Connect with Phil Orlando on LinkedIn