Natonal Jobs data both up and down

Jobs data both up and down

Sam Zuckerman, Chronicle Staff Writer

Saturday, May 3, 2008

The Labor Department's April employment report was a head-scratcher. The nation shed 20,000 jobs during the month, but the number of employed people rose by 362,000 and the unemployment rate fell to 5 percent from 5.1 percent in March.

Such seemingly contradictory data are actually not unusual. That's because the jobs numbers and the employment numbers are derived from two different surveys.

The number of jobs created or lost comes from a survey of about 160,000 businesses and government agencies. The number of employed people is based on a separate survey of 60,000 households. The two polls can produce significantly different results.

The employer survey counts only jobs on government, business or nonprofit payrolls, excluding farm workers, the self-employed, unpaid family workers and household workers. By contrast, the household survey counts a person as employed if he or she did any paid work at all during the week of the survey, or worked without pay at least 15 weeks in a family business or farm. People are also considered employed if they're on temporary leave.

In addition, both surveys have significant margins of error. For example, the household survey has what's called a "confidence interval" of a whopping plus or minus 430,000. That means there's a 90 percent chance that the actual change in the number of employed people is within 430,000 of the published figure.

Both survey results are revised as more accurate data come in. The department originally estimated that February's job loss was 63,000, but later revised the number to 83,000.

For the Labor Department's explanation, go to links.sfgate.com/ZDGE.

E-mail Sam Zuckerman at szuckerman@sfchronicle.com.