Payrolls Preview: Goldman Warns Of “Seasonal Tendency To Disappoint Consensus”

Against a consensus expectation of 215,000, Goldman forecast a 230,000 increase in September nonfarm payroll employment and a one-tenth decline in the unemployment rate to 6.0% (vs. consensus 6.1%). Overall they think the available employment indicators for September point to a solid report, despite slightly weaker data this week. In addition, the reversal of a couple of special factors from August should be a positive. If history is any guide, however, regarding downside risks, there has been some seasonal tendency for September payrolls to disappoint consensus.

Recent Data is more negatively biased… (via Bloomberg)

Via Goldman Sachs,

We forecast a 230,000 increase in September nonfarm payroll employment (vs. consensus 215,000). This would represent a substantial pickup from August’s 142,000 gain, pushing us back towards our view of the underlying trend and placing the payroll numbers back in line with the broader dataflow.

Arguing for a stronger report:

Solid business surveys. The employment components of all business surveys we track remained in expansionary territory in September, with most at fairly strong levels. Relative to August, six of twelve moved up, while the other six moved down. With respect to national (rather than regional) surveys specifically, ISM manufacturing employment moved down in September, but Markit manufacturing employment and Markit services employment advanced.

Claims continued to move down. The four-week moving average of initial jobless claims moved down by 6k from the August reference week to 295k.

Bounce-back from grocery store work stoppage. Although not technically classified as a “strike” by the Labor Department, work stoppages at Market Basket grocery stores resulted in a drop in food and beverage retail employment in August, which underperformed the twelve-month trend by about 21k. We would expect all of this effect to reverse in September.

End of auto shutdown distortions. Last month, durable manufacturing employment fell short of its twelve-month trend by about 10k, as seasonal adjustment volatility associated with smaller-than-normal July auto plant shutdowns distorted the August gain. We expect a more normal contribution from manufacturing in September. Although the ADP report was broadly in line with expectations for September—and has generally been of marginal use in predicting payroll employment more broadly—it did show a strong gain in manufacturing employment.

Arguing for a weaker report:

Seasonal bias. Historically, there has been a tendency for September payrolls to fall short of consensus expectations. This occurred in twelve of the past seventeen years. This could potentially be due to systematic differences between early and late survey responses by industry, which would not be captured by the seasonal adjustment methodology. While the average historical miss has been about 40k to the downside, over the past three years the average miss has been about zero.

Labor differential worsens. The Conference Board’s labor differential—the net percent of survey respondents reporting jobs are plentiful vs. hard to get—worsened by 2.6pt in September to -15.0.

Casinos run out of luck. The recent closure of three large casinos in Atlantic City (two of which occurred before the September reference week) might be expected to reduce payroll employment by around 5,000 jobs.

On top of the September numbers themselves, we think a positive upward revision to the past two months of data is likely. The September report has had a strong seasonal tendency to include positive back-revisions, with the average revision +42,000 over the past ten years and +65,000 over the past three years. As a result, we would not be surprised to see last month’s 142,000 revised up to something closer to 200,000, which would provide reassurance that the trend in job growth has not deteriorated significantly.

We also update our payrolls forecast “distributions” for September. The median model forecast is slightly above consensus at around 225,000 (vs. about 240,000 in August). The width of the distribution widened somewhat from August, likely as models which place more emphasis on the recent slower payrolls trend increasingly disagreed with models placing more weight on the generally solid September dataflow. However, downside skew moderated slightly. Purely model-based forecasts are also unlikely to capture the full extent of bounce-back from the end of grocery store work stoppages that we would judgmentally expect.

Exhibit 1. Distribution of Payroll Forecasts

With respect to the unemployment rate, we anticipate a decline of one-tenth to 6.0% on a rounded basis (vs. consensus +6.1%). We do see a substantial risk of the unemployment rate remaining unchanged, in light of the “firm 6.1%” starting point for August (6.1497% on an unrounded basis). Nonetheless, the risk on the participation rate is probably still slightly to the downside, and employment growth according to the household survey was somewhat soft in July and August, affording the possibility of a strong gain in September.

On average hourly earnings, we expect a trend-like gain of 0.2%, in line with consensus. This would push the year-on-year rate up slightly, to a still-subdued 2.2%.

* * *

Finally we offer this ‘humor’ from CNBC… winning in the new nominal vs real normal…