No Manipulation of US Jobs Data, but the Numbers Are Noisy

By Joseph G. Carson October 09, 2012

In the heat of a US election season, the sharp drop in September’s unemployment rate raised some eyebrows. I think it is blatantly wrong to argue that government statisticians manipulated the data. But the jump in household jobs that triggered the drop in the unemployment  rate was indeed extraordinary and requires further scrutiny.

According to the household employment survey, 873,000 people found jobs in September and the unemployment rate fell by 0.3 percentage points to 7.8%. The outsized gains in household employment were far above the payroll survey gain of 114,000, and one of the largest gains on record.

There’s no doubt that the huge jump in the number of people finding jobs last month does look odd, as it conflicts with all other labor market indicators as well as assessments of labor market conditions by individuals and businesses. This led some people to question the accuracy of the results and to suggest that the data had been engineered for political purposes.

But it’s simply wrong to argue that government statisticians at the Bureau of Labor Statistics cooked the data. Monthly household jobs data are always very volatile. And this time, the figures may have been influenced by a one-time event that led to an overstatement of the results for the period.

For example, in September, there was a very unusual move in the US workforce for younger workers between the ages of 20 to 24. For the first time since 1961, the number of people in this age group actually working (before seasonal adjustments) increased by 101,000. Typically we would expect a decline of at least 300,000 in September. This age group usually experiences a drop in actual employment levels in September, as many young people who worked in the summer return to college. Although this age cohort only comprises 10% of the workforce, it accounted for 42% of the near-record employment gains in September

What might account for this unusual pattern?  One possibility is the timing of the political conventions. For the first time ever, both the Republican and Democratic conventions were held in early September, and since many young people do part-time work for the campaigns, it could have influenced the household employment results. According to my estimates it would take a shift of only several hundred people in the 20 to 24 age group—of the approximately 6,000 surveyed—to trigger an unusual movement in the September jobs data.

Perhaps it’s no coincidence that part-time employment rose sharply in the household employment report. In addition, these jobs did not appear in the payroll survey because they were not part of a traditional business establishment.

None of this suggests that the household employment data for September were wrong. But, given that they may have been influenced by a one-time event, I think a reversal of the 368,000 employment gains in the 20 to 24 age cohort seems likely in October.

The views expressed herein do not constitute research, investment advice or trade recommendations and do not necessarily represent the views of all AllianceBernstein portfolio-management teams.

No Manipulation of US Jobs Data, but the Numbers Are Noisy
Back to a top