Last October, when the government released its monthly tally of how many people had jobs, there was a collective groan. The September report, which came out the first Friday in October, said the number of employed people in the U.S. had dropped by 95,000, worse than the 57,000 job drop the month before. After looking like we had finally hit an economic rebound, the jobs market was again slipping back again perhaps toward the dreaded double-dip recession. Or was it?
A month later the Bureau of Labor Statistics, which tracks and releases the employment numbers for the government, revised its jobs count for September. In fact, the economy, it turned out, had lost only 41,000 jobs that month. Is that right? Actually no. A month later, the BLS revised the number again. The final tally: In September, the number of people working in America fell by just 24,000. So the economy was improving? Not quite. Remember that month before figure of 57,000 jobs lost. Yeah, well, that was wrong, too, off by nearly all of the drop, or 56,000 jobs.
In the past year, the hype around the government’s jobs report has become intense. None of the dozens of economic reports the government or others put out get nearly as much attention. Forget inflation. Forget retail sales. The jobs report is the report to watch these days.
On the first Friday of the month, when the government releases the number, business channel CNBC has a countdown clock, and a roundtable of experts waiting to comment. The New York Times has recently been live-blogging the “event.” Experts at Washington think-tanks say they are now flooded with questions about what the numbers mean on the day of the jobs report. “Unemployment” and “jobs number” is a top Google trend on the day of the release. “We tweet out what we think about the jobs number,” says Heather Boushey, a senior economist at liberal think tank Center for American Progress. “We never did that before.”
Boushey says the difference is in part because the media landscape is different from when she started following the report a decade ago. Nonetheless, she says the main reason for the increased interest is the poor economy. “With the unemployment rate at 9%, it makes sense that there would be more interest in the report,” says Boushey. “The purpose of the economy is to help us eat and provide for ourselves. So the jobs report is the best indicator of whether the economy is fulfilling its function.”
And yet, the report that is supposed to tell us that, seems flawed. Here’s why:
Wall Street strategists and economists contend that the jobs report has always been important. But in the past two years, as financial crisis abated leaving no obvious rebound in its wake, the focus on the jobs report has grown. That’s because jobs are the true sign that the economy has recovered. The number of employed people has indeed been growing lately, but the long wait for that to happen, nearly two years after the economy officially left the recession, has drawn more focus to the jobs tally – a number this is almost always, at least initially, wrong.
Revisions are of course the norm with government reports, but most of those reports don’t get the attention that the jobs report gets. Worse, the size of the revision seem to matter more at times when the economy is weak. A difference of 50,000 jobs, which is well within what the government deems as an acceptable error, can really change the employment picture at a time when the economy is producing less than 100,000 jobs a month.
“It gets a lot of hoopla,” says James Paulson, the chief investment strategist at Wells Capital Management. “And all this over one of the most volatile, revised numbers we have. All it creates is a lot of VIX [volatility], and not a lot of value.”
For a number of decades the size of the monthly revisions was dropping. The government had expanded the number of companies it surveys making the data more accurate. But during the Bush years, according to Boushey, the BLS’s budget was cut forcing it to drop some of its sampling data. The government’s data has proven to be more faulty than it was just a few years ago. In 2008, the government’s initial count and finally tally differed by about 40,000 jobs. This year the revisions have ticked up an average of just over 50,000 jobs.
What’s going on? The problem is in part that it is hard to measure things like jobs at times when the economy is turning, which is ironic because that’s the time when people care about economic data the most. But the jobs report has some significant problems. First of all, as blogger Felix Salmon has pointed out, self-employed people are left out of the jobs number. And there has been a general trend in this economy toward consultants and more people being self-employed. The jobs report misses that trend. Second, as the BLS recognizes, a big problem with the report is how to account for new companies or companies that go bankrupt. And since new companies create the vast majority of new jobs, that’s an issue. The BLS does it’s best to adjust for the fact that it doesn’t get data from new companies, or companies that no longer exist, and that is in part what leads to the revisions.
Making matters worse, the monthly numbers we get on the jobs market come from not one, but two surveys. The number of jobs comes from something called the establishment survey, which is done by asking 140,000 businesses and 440,000 workplaces to fill out a form in the middle of the month detailing how their payrolls have changed. The government takes that data and makes an estimate of how many people are employed in the entire country. The unemployment number, which is also released on the first Friday of the month, comes from a survey of 60,000 households. Despite the smaller sample, the unemployment rate often gets more attention than the jobs tally. Worse, at times the two surveys can say opposite things, which is exactly what happened last month when the unemployment rate rose, but so did the number of people with a job. The differing data can leave economists scratching their heads.
So what should we do about it? Paulsen suggests that perhaps the government should think about only releasing the jobs number once every three months. That’s how often the gross domestic product comes out. If the jobs report came out less often, it would perhaps be more accurate, and it wouldn’t lead to as much constant back and forth as to whether the economy is recovering or not – discussions that in the end may not be all that useful, other than to confuse the business leaders and policy makers who are trying to judge where the economy is headed. “Capital decisions are made based on the report,” says Paulsen. “So releasing a misleading number can affect real economic activity in a potentially harmful way.”