The Bureau of Labor Statistics (BLS) released its May jobs report this morning and the world was shocked to learn that the U.S. economy only produced a net additional 38,000 jobs last month. That wasn’t the only bad news. Total nonfarm payroll employment for March was reduced down from 208,000 to 186,000 and April was revised down from 160,000 to 123,000 for a combined 59,000 jobs less than previously reported. Over the past three months, job gains have averaged only 116,000 per month.
Those numbers are from the Establishment Survey of employers. The Household Survey was just as bad, with only 26,000 jobs added in May, bringing the total employed to 151.03 million, up just 1.5% year-over-year. How’s that for what the Fed calls a strong labor market!
The number of unemployed tumbled from 7.920 million to 7.436 million, dropping the unemployment rate to just 4.7%, the lowest since August, 2007. That looks impressive until you learn that it was driven by a massive surge of 664,000 people who left the labor force in May, bringing the total to a record 94.7 million Americans not working and not looking for a job. Does this sound like the stuff of a June Fed rate hike to you?
How can this happen you ask? It’s not easy. You probably naively think that someone actually counts the number of jobs each month. They do not. The BLS runs two surveys which sample a relatively small number of employers and households.
The survey results are run through some very complex mathematical algorithms for interpretation. These algorithms reflect something called ‘trend cycle analysis’. This means that the BLS makes assumptions about where the U.S. economy is in the economic cycle. The BLS relies on notoriously rosy Federal Reserve projections and indicators like the weekly jobless claims to determine that the economy is growing. Having assumed that, the data is massaged to reflect a growth bias in new business start-ups and failures and an expansion of industries that have grown with the economy in past recoveries. Of course, this recovery is not like past ones.
What am I saying here? The BLS assumes the numbers are going to be good and then produces numbers to prove it. That’s what happens when you get a bunch of Ph.D. statisticians to estimate employment.
The problem now is that the algorithms are gradually being revised to reflect the emerging economic slowdown, the continual downward revisions in employment and the weakness in other labor data such as hours worked, wage rates and labor participation.
My conclusion? Expect further weakness in future employment reports. If the Fed doesn’t raise rates soon, it never will. Eventually, we will find out well after the fact that we were in a recession all this time.