WASHINGTON - February 2 - The most important source of data on the U.S. labor market may be systematically overstating employment, according to a new report by the Center for Economic and Policy Research. The report, "Missing Inaction: Evidence of Undercounting of Non-Workers in the Current Population Survey (CPS)," found that the CPS appears to be overstating the share of American adults who are working by about 1.4 percentage points.
The report by economists John Schmitt and Dean Baker noted that a large and growing portion of the population does not respond to the CPS, and that the non-responders appear less likely to be employed than people who take the survey. This overstatement is significant because the CPS is the source for the government's most important statistics on the labor market, including the unemployment rate, poverty rate and health-insurance coverage.
"Current labor market estimates appear to be overstating the share of working Americans by 1.4 percentage points. This corresponds to roughly 3 million fewer people working - almost as big a drop in employment as in a typical recession," said John Schmitt, CEPR economist and lead author of the report.
The study assessed employment rates among non-responders by comparing employment rates in the CPS with employment rates in the 2000 Census. In 2000, 8 percent of the population did not respond to the CPS. In contrast, only 2 percent did not respond to the 2000 Census. After adjusting for the errors in reported employment in the Census data (and excluding the prison population), the study found that employment rates were 1.4 percent lower overall in the Census than in the CPS.
The study also found that the CPS overstates employment rates for blacks by about 2 percentage points, with the gap for younger black men as high as 8 percentage points. The CPS also appears to be overstating employment rates of younger Hispanic women by about the same margin, and younger Hispanic men by 3 to 6 percentage points.
Since the CPS is also the source of official statistics on poverty rates and health-insurance coverage, the report warns that these widely reported numbers could also be overly optimistic. Non-working adults are more likely to be in poverty and less likely to be covered by health insurance. Therefore, if non-working adults are disproportionately excluded from the CPS, then the survey is understating the true poverty rate and overstating the share of the population covered by health insurance.
The Current Population Survey is a monthly survey of households conducted by the Bureau of Census for the U.S. Bureau of Labor Statistics.
To read the report, see: http://cepr.net/publications/undercounting_cps_2006_01.pdf