Feds Take Over Employment Surveys From The States
In the midst of the economic recession in 2008 and 2009, federal officials gathered economic data showing a sobering slide in employment across the country.
But the Bureau of Labor Statistics wasn’t the only agency collecting such data.
And when it compared its numbers with those of state labor and workforce development agencies, federal officials realized that states were painting an overall picture that was far less severe, said Deborah Brown, who is the bureau’s assistant regional commissioner for federal and state cooperative activities.
“The decline was less than half of what we had seen,” Brown said. “So we knew we had problems.”
The BLS concluded that some individual state agencies were using techniques that led to statistical bias, especially in the face of big employment losses that could look like errors to some analysts.
Federal Authority
For years, federal officials who track economic data have tried to get more accurate numbers for job gains and losses in each state, especially from those with smaller populations, which carry a greater potential for statistical error, Brown said.
The latest development in that effort happened in March, when the BLS took over responsibility from state-level labor and workforce development agencies of the employment survey, which produces the monthly job numbers.
Brown said that the BLS believes that analysts in some states had too much influence over the final data incorporated into the survey. The result was a “statistical bias” that led to understatements of job losses, she said.
In Massachusetts, the Executive Office of Labor and Workforce Development said in its recent jobs release that the methods employed by the BLS carry the potential for increased month-to-month variability in the jobs numbers.
Brown said that is true in the short term, but in the long run, through revisions that incorporate new data into initial estimates, the BLS data should be more accurate than the workforce agency data.
“We’re trying to make the best possible estimates we can based on the information we have at the time we put the estimate out,” Brown said.
Brown said that the changes will save the BLS approximately $5 million per year, a portion of which will be invested in improving the number of survey responses the agency receives.
For those that pay closer attention to the jobs numbers, the effect of the change is yet to be seen.
Maureen Dunne, co-director of the MetroWest Economic Research Center (MERC) at Framingham State University, said that Massachusetts may have a large enough population where the revisions won’t be much more significant than they have been in the past.
Dunne said that at a May conference of the New England Economic Partnership held in Boston, economists were discussing how the new methods might have a greater effect on Vermont, which has a population one-tenth the size of Massachusetts.
Statistical Swings
A recent example of a Massachusetts job numbers release illustrates a frequent occurrence in the uncertain world of creating estimates based on samples of a population.
The good news in March was that Massachusetts added 3,200 jobs. The bad news came a few months later from the Executive Office of Labor and Workforce Development in the form of a revised estimate stating that Massachusetts actually shed 1,300 jobs in March.
Brown, of the BLS, said that such revisions are the result of the BLS receiving more information from employers and households. The window of time the BLS has to collect data — one week — is not enough to get survey responses from all of participants, Brown said.
Despite the fact that monthly job numbers sometimes don’t paint the most accurate picture, they are often the first glimpse that economists and the public get into how the economy is performing.
And compared to the lagging data of the U.S. Census Bureau and other sources, they are by far the fastest, said Andre Mayer, senior vice president of communications and research at the Associated Industries of Massachusetts.
“They get some usable numbers out pretty quickly,” Mayer said. “If you can get something within six weeks, that’s pretty good.”
WBJ Web Partners
Most Recent
Most Recent
0 Comments