Friday, March 5, 2010

What Happened to the Snowstorm Impact on Unemployment Numbers?

Nonfarm payroll employment was little changed (-36,000) in February, and the unemployment rate held at 9.7 percent, the Bureau of Labor Statistics reported today.

This is after, as we have been reporting, everyone from NYT columnist Floyd Norris to the President's top economic adviser, Larry Summers, warned of a dire numbers because of the severe snowstorms. What happened to the anticipated unemployment surge?

A source at the BLS that I spoke to this morning questions the data used by Norris in his column. Norris compared the snowstorms in February of this year to those that occurred in 1996, and claimed they would result in a "horrid" uptick in unemployment in the current number. This person states that there was not as dramatic a swing in numbers in 1996 as Norris stated. The revision was only within the range of 60,000. Norris has the 1996 number going from 206,000 unemployed to 19,000. BLS says the revision was only down to approximately 140,000.

Asked about Summers warning, the response was a chuckle, followed by a no comment.

So why wasn't there a bigger impact on the numbers because of the snowstorms?

A few reasons, first the BLS counts snow shovelers who come into work to shovel snow as employed for the entire period, even if they shovel snow for a couple of hours and then head home.

In order for severe weather conditions to reduce the estimate of payroll employment, employees have to be off work for an entire pay period and not be paid for the time missed. About half of all workers in the payroll survey have a 2-week, semi-monthly,or monthly pay period. If workers received pay for any part of the pay period, even one hour, they are counted in the February payroll employment figures.

Bottom line, the BLS doesn't even know how to interpret the numbers for February. They write:
Severe winter weather in parts of the country may have affected payroll employment and hours; however, it is not possible to quantify precisely the net impact of the winter storms on these measures.
Also of note, the Great Census surge in employment has started as the BLS added the hiring of 15,000 temporary workers for Census 2010 as part of the employment picture.


  1. How many times has the BLS changed the method of calculation (operational definition) since 1996?

  2. Does anyone know the answer to PR/SOD's question? Does anyone care? After all, Government data are worse than useless.