With a 9 standard deviation range between the highest and lowest excuse for a forecast from the 81 "qualified" economists on Bloomberg's survey, there is plenty of room for noise to dominate signal with tomorrow's payrolls data. Goldman forecasts a softer-than-consensus 210k increase in non-farm-payrolls as May employment data flow looks more mixed, and they expect that the unemployment rate rose two-tenths to 6.5% in May (vs. consensus 6.4%). Average hourly earnings (AHE) are likely to be in focus again following several months of heightened attention to wage growth and labor market slack; Goldman expects an increase of 0.2% in May (vs. consensus 0.2%).
Recent employment data has been mixed... (and today's dismal Challenger Layoffs data)...
[10]
And Goldman is less exuberant than many on the street...
We forecast a 210k increase in nonfarm payrolls in May, a touch below the consensus estimate of 215k. We expect that private payrolls increased 210k (vs. consensus 210k), with government a neutral contributor. This would mark a substantial decline from April's better-than-expected 288k gain. The reason for our more modest expectations for May is that the employment data have been considerably more mixed this month after pointing almost uniformly to a stronger number in April. As a result, we expect this month's gain to fall somewhere in between the 6-month (203k) or 12-month (197k) moving average and the roughly 225k trend rate we expect as growth accelerates this year. Based on May data from recent years, we also see some risk of a net downward revision to the prior two months.
Arguing for a stronger report:
- The employment components of business surveys sent a mixed-to-positive overall message this month. Among manufacturing surveys, the ISM (-1.9 to 52.8) and Dallas Fed (-16.8pt to 2.9) surveys declined, but the Philly (+0.9pt to 7.8), Empire (+12.7pt to 20.9), Richmond (+6pt to 10), and Kansas City (+7pt to 10) Fed surveys improved. Among service sector surveys, the employment components of the ISM non-manufacturing (+1.1pt to 52.4) and New York Fed (+10.3pt to 16.4) surveys improved, while the Richmond (-2pt to 4) and Dallas (-2.6pt to 13.8) surveys declined. The Beige Book also reported "steady to stronger" hiring in almost all districts.
- The labor differential?the difference in the percentage of respondents in the Conference Board's consumer confidence survey describing jobs as plentiful vs. hard to get?improved 1.6pt to -18.2 in May. The index has shown a fairly steady recovery since late 2011, but was a misleading indicator last month, when it fell 2.2pt.
Arguing for a weaker report
- Private job gains reported by ADP disappointed consensus expectations this week at 179k in May, down from an initially-reported April gain of 220k. That said, we attach only limited weight to the ADP report because its initial print has yet to prove itself as a reliable indicator of payroll job growth as measured by the Labor Department.
- Announced layoffs rose about 25% in May on a seasonally-adjusted basis, according to Challenger, Gray, and Christmas, with the largest cuts coming in the computer industry. As in April, job cuts in the health care sector remained at a normal level, suggesting little impact from layoffs of temporary workers after the end of the sign-up period for health insurance under the Affordable Care Act.
Neutral indicators
- The four-week moving average of initial claims for unemployment benefits increased slightly from the April to the May reference week. However, the April weekly data might have been distorted by the Easter holiday and spring break from schools, and the difference between the two months was relatively small.
- New and total online job ads fell a touch in May. Both series have been fairly stable over the last few months relative to their usual volatility, and we therefore view this as a neutral indicator. The Help Wanted OnLine report noted weaker demand for workers in sales, management, and computing, but increases in transportation, production, and construction job ads.
- Weather conditions returned to seasonal norms in April and were a bit more supportive than usual in May.
We expect that the unemployment rate rose two-tenths to 6.5% in May (vs. consensus 6.4%), following an unexpected four-tenths decline to an unrounded 6.28% in April. The reason is that we expect the participation rate will partially reverse last month's unusually large four-tenths decline. Building on our decomposition of non-participators into the retired, disabled and other categories shown in the household survey micro data, we find that the 'other' category contributed most of the decline in participation in April.
At the margin, this might point to a larger retracement of last month's decline because--as researchers who use the household survey often note--survey respondents sometimes confuse unemployment with being out of the labor force for other reasons (the marginally attached are a subset of this category). To test this possibility, we model the change in the share of non-participators with a constant and last month's change. We find a tendency for about one-quarter of the change to reverse, but when we replace last month's total change with the changes in the three components of non-participation, we find that the effect is driven by changes in the 'other' category.
Average hourly earnings (AHE) are likely to be in focus again on Friday following several months of heightened attention to wage growth and labor market slack. We expect an increase of 0.2% in May (vs. consensus 0.2%). Wage data released over the last few months have remained soft, and qualitative evidence from yesterday's Beige Book also noted subdued wage growth. AHE for all workers grew just 0.1% in March and were flat in April. The Employment Cost Index showed compensation growth of 1.3% (annualized) and wage & salary growth of 1% (annualized) in Q1, while the national accounts data showed 2.3% (annualized) growth of compensation per hour in the nonfarm business sector in Q1. Our wage tracker--a smoother aggregate of all three wage measures--grew 1.8% over the past year.
Here is the full distribution - with Lavorgna actually slightly below consensus also...
[11]