(Phys.org)—The hullabaloo surrounding last week's release of the nation's employment numbers was a bit overblown, says a University of Michigan economist.
While U.S. employers posted modest job gains in September, the unemployment rate fell sharply from 8.1 percent to 7.8 percent—the lowest point in nearly four years, according to the Bureau of Labor Statistics.
"There is some controversy over the legitimacy of the BLS household employment measure and the unemployment rate that is derived from the same survey, but the numbers seem statistically reasonable," said Donald Grimes of the U-M Institute for Research on Labor, Employment and the Economy.
Grimes looked at the historical difference in month-to-month employment movement since 1970 as reported by the payroll survey, which collects data from business and government, and the household survey, which asks individuals about their work situation and is used to calculate the nation's official unemployment rate.
The payroll survey showed job gains of 114,000 during September, but the household survey revealed much higher employment growth of 873,000, which was the major reason for the sudden drop in the unemployment rate.
Using statistical analysis, Grimes said that we should expect a difference of this magnitude between the two sets of numbers once every 35 months of data, given the randomness of measurement errors.
And, in fact, there were 20 months since January 1970 where the difference between the two surveys was larger than the one reported in September (measured in percentage terms).
"So the results were unusual, but not unprecedented," Grimes said. "One word of caution, however—there is probably about a one-third chance, based solely on the self-correction of sample-based measurement errors as seen in the historical data, that the household survey will show a substantial decline in employment in October causing the unemployment rate to jump up again.
"We'll know soon enough, since that data will be released on the Friday just before the election."
Explore further: Why companies don't learn from their mistakes