Our Elusive Employment Statistics

The Texas central time zone serves me well at night, especially on New Year's Eve when New York drops the ball at 11 PM. Mornings are another matter.  I have to rise before the rooster crows for my East Coast entertainment, which involves crowing and dropping of balls of a different kind.

The first Friday of the month (sometimes the second), however, is worth getting up for. That's when the talking heads on the financial news programs explain the newly-released employment numbers for the previous month. I always smile when they tell me that what matters isn't how much employment grew, but how much it grew relative to expectations.

Those expectations are like the yellow first-down line on TV football, presumably visible to everyone in TV land, but not to the players. Except that the amount needed for a first down in employment growth varies from month to month and often seems based on nothing in particular.

Just before last December's employment numbers were reported on January 5, CNBC put the all-important consensus expectation at 115,000 while the Wall Street Journal had it between 110,000 and 115,000.  The official estimate came in above those expectations, at 167,000, which made it a "strong" number. That strength pushed the stock market down for a few hours, presumably because investors and traders thought it made the Fed less likely to lower interest rates, and they apparently like the Fed's medicine for a weak economy more than they like a strong economy.

The December employment-growth estimate of 167,000 remained the official number until it was revised up to 206,000 on February 2, and on up to 226,000 on March 9. March 9 was also when the initial estimate for February was announced and when the January number, as well as the December number, was revised up, from the initial estimate of 110,000 to 146,000. If recent experience is a guide, it will be revised up once again next month.  Do you detect a pattern here?

For some time now, official employment estimates have been revised substantially upward, often more than once. The first estimates, which get most of the press and TV attention, have been rather lackluster, while the final numbers have been much stronger.  This must be a source of great frustration to economic policymakers and others in the Administration.  It contributes to the Rodney Dangerfield economy.

The February employment increase of 97,000 announced on March 9 was the smallest in some time, but the commentators didn't make much of it since it was close to their consensus estimate of 100,000. The companion announcement of the February unemployment rate was a decline from 4.6 percent to 4.5 percent.

How did the smallest employment gain in some time cause the unemployment rate to decline?

The general answer is that the employment number and the unemployment number that are reported together and discussed together aren't directly related.  The employment number comes from an establishment survey of payrolls while the unemployment rate is calculated from a separate household survey.  This has become a pet peeve of mine-not that they are separate and unrelated, but because they are reported together and discussed together as if they are related.

Usually when a decline in the unemployment rate accompanies only a small increase, or even a decrease in reported employment, it's because there was a larger increase in employment in  the household survey.  In the particular case of February, however, employment actually declined slightly in the household survey and the decline in the unemployment rate resulted from an even larger decline in the labor force, i.e., by more people who are not employed failing to look for a job during the survey period. 

Since I was going to be interviewed on the employment report later in the day, I listened carefully to see if anyone on financial TV or radio mentioned the household survey as the source of the apparent discrepancy between the reported employment and unemployment numbers.  If anyone mentioned the household survey at all, I never heard it.

The public would be better served if the Labor Department included the household data more prominently in its announcement and if the financial reporters gave it more attention.  As it is, I'm afraid that the apparent discrepancy between the reported headline numbers will cause the public to lose faith in the accuracy of our official statistics.

The discrepancy between the establishment and household survey results has become significant in recent years, both because if its size and its recent tendency not to converge quickly if at all. Conflicting signals are sent regarding the strength of the economy, which, I know for a fact, even policymakers aren't immune to. Misleading data can lead to bad policy.  

That was certainly the case earlier in the current expansion, when I wrote an op ed on it for the New York Sun. The favored payroll survey was persistently and substantially weaker than the household survey which exaggerated the impression of weakness in the economy. I kept asking my research staff at the Dallas Fed if the stronger household survey didn't better reflect the economy at the time, but they wouldn't humor me, largely because of the payroll survey's larger sample size. On their side of the argument was elite academic opinion, which I used to define as opinion coming from universities without good football teams.   

The answer to the "so what" question about that continues to be important.  During the past two years, net employment growth as measured by the household survey has been substantially greater than the establishment measure.  We aren't talking about monthly diversions here.  This one has lasted two years.

Whatever the merits of the two measures in absolute terms, the relative merits of the household survey are surely rising as our entrepreneurial job churn scatters workers across the landscape.  More and more of us are dropping off large establishment payrolls and working from home, as I am now. My favorite Texas picker-poet, Billy Joe Shaver, says, "I've got it down to one moving part, and that moving part is me."  You won't find Billy Joe on a payroll.

When I recently retired from a large university system, I followed the lead of another Texas poet and philosopher, Kinky Friedman.  Kinky said he wanted a profession that didn't require his actual presence, which he found as an author. I couldn't have said it better myself. I now work at home, thinking part-time for a think tank.  Along with Billy Joe and Kinky, if they're going to count us as employed, they're going to have to find us first.

Print