So Did Nonfarm Payroll Really Increase by 290,000 Jobs?

Americans deeply distrust government these days, and sure enough, multiple bloggers have challenged the accuracy of the Bureau of Labor Statistics's claim that 290,000 jobs were added in April. So, were that many new jobs really created last month? Maybe. Does the reported number convey useful information about the job market? Yes.

The bloggers focus on one adjustment the BLS made: Net jobs from business "births" minus business "deaths." New jobs from new businesses can't be captured in the survey because the new businesses take a few months to show up in the database that allows them to be contacted as part of the survey. The blogs focus on the +188,000 job birth/death adjustment used in calculating April's 290,000 number to claim that the report is wildly inflated.

So How Fuzzy Is the Math?

The BLS complies the change in nonfarm payroll estimate by surveying businesses and then adjusting the result with statistical methods. The biggest adjustment is for seasonal hiring patterns, hence the labeling of numbers as "seasonally adjusted" or not. The business birth/death adjustment is made prior to the seasonal adjustment, so you can't just subtract 188,000 from the 290,000 to get a sense of what the hiring was without it, although at least one website critical of the official report did just that.

The non-seasonally adjusted number for April 2010 was 1.158 million new jobs (April 2010 minus March 2010); seasonal adjustment turned that into 290,000. Without the business birth/death number, the unadjusted total would have been 970,000. What would that be, seasonally adjusted? Not sure, because I have no idea what the seasonal adjustment formula is. A crass calculation shows that the reported 290,000 is about a quarter of the non-seasonally adjusted 1.158 million. The same percentage of 970,000 would be about 243,000.

These Aren't Normal Economic Times

In short, if we don't adjust for net jobs from business births and deaths, the economy added something more like 240,000 jobs last month. So is 240,000 more accurate than the reported 290,000? In normal economic times, the answer would be clearly no. The BLS website has nifty charts showing how the business birth/death adjustment substantially reduces the job estimate's error rate, a fact the BLS can determine every year after the full universe of job data, as opposed to a sample, becomes available through tax data.

Unfortunately, the most recent data point on the charts is the March 06-07 correction, prior to the Great Recession. That limitation is important, because even the BLS cautions that the birth/death adjustment is based on historical data. If historical business birth/death trends don't apply, the adjustment will be less accurate. Obviously, these aren't normal economic times.

Even if not as accurate an estimate as it would be in normal times, the 290,000 jobs number has validity and usefulness as a data point in a trend. The Great Recession officially began in December 2007. Starting in January 2008 and continuing through April's number, the nonfarm payroll jobs report, adjusted for business births/deaths and seasonality, looks like this:

Since all of these numbers were generated in these unusual times using the same methodology, the trend line they depict should have basic validity even if the absolute values are a bit off. And it's clear that March and April of this year reflect sharply more hiring than we've seen previously during these turbulent times. Still, it's worth taking a closer look at the numbers reported during the Great Recession to see if we can make any more sense of April's 290,000 number.

Maybe the True Number Is Better than 290,000

Looking at Tables 2 and 3 below, it's clear that there's a substantial difference between the numbers the BLS initially reports, like April's 290,000, and the number it finally decides is accurate. Early on, the bureau generally overestimated job losses and then really underestimated them. However, from August 2009 through March 2010, the corrected number was better than the initial report six out of eight times, suggesting that the adjustments the BLS does currently underestimate job gains.

So, contrary to the critical bloggers railing about the birth/death distortion of the jobs report, it may turn out that April added even more jobs than reported. Does that mean the U.S. economy is out of the woods? Not by a long shot. It doesn't even mean that current gains are sustainable. But if you read or hear people tearing down the 290,000 number by citing the birth/death adjustment, ignore them.