US payrolls: don't believe the hype

With US payroll data out later today, it is worth reading John Authers's piece in today's Financial Times, which bemoans "absurdities such as the regular wild overreactions to US employment data":

It normally falls to columns like this to inveigh against absurdities such as the regular wild overreactions to US employment data. But in this case, Tim Bond of Barclays Capital in London has done a fine job himself in a research note.

According to Mr Bond: “This writer has always suspected that the bond market’s concentration on the payrolls number had very little to do with the economic import and a good deal to do with an endemic gambling dependency. Rather like roulette, the payrolls number is more or less random, but within a reasonably well-defined range. The economics profession sets the range for the roulette wheel and the random-number generation techniques of the Bureau of Labor Statistics do the rest.”

He suggests that last month’s astonishing upward revision in the ranks of the employed by 810,000, should prompt the market to “abandon its monthly genuflection before the payrolls data”. Or, it can “abandon any pretence to being an efficient discounting mechanism for economic conditions”.

So far, so intemperate. The criticism is justified. Payrolls are treated as a huge market event, but the statistics are frequently incorrect when first reported. With the Federal Reserve on pause, the bond market has made extreme reactions to several other data releases this year, only to correct itself later. This matters, as treasury yields set the baseline for the world financial system.

Today’s data look important. The bet is back on that the economy is slowing, after some disquieting economic releases this week, and that hence the Fed will soon cut rates. Mr Bond disagrees, saying the unemployment rate, at 4.6 per cent, is too low for inflation to head back to the Fed’s “comfort zone”. This unemployment rate, he says, is more reliable than the payroll data, as it is compiled by surveys of households, rather than of companies.

If the market still trades off the payroll number, he says, “then bussing in grannies for a monthly Bingo-Friday would be the more optimal way to allocate capital in the global bond market”.

Comments

[Lafayette]

"Rather like roulette, the payrolls number is more or less random, but within a reasonably well-defined range. The economics profession sets the range for the roulette wheel and the random-number generation techniques of the Bureau of Labor Statistics do the rest."

This is an exagerrating of the facts to suit journalistic license. The payrolls number may be inexact from month to month, but year upon year they indicate trends worth noting.

Once again, it is a matter of sensationalist journalism that MUST respond to data announcement to fit a daily deadline (or even hourly, on the Internet). What must one expect them to say? What is there to say, except announce the number, which is rarely enough to fill a submission of at least 300 words.

It would be sufficient to report either year-to-year based changes to the day or, even, half-year changes. Economic data, whether macro - or micro- is best interpreted, if at all interpreted, as a trend series for public consumption. All other manipulation of the data is a matter of research and, therefore, peer review before publishing.

Most (important) economic conditions take a considerable time to evolve. The problem is that stock markets are very volatile at present and they react (stupidly) to the input of statistical "facts".

One need only look at the daily reporting of almost any European index, between 1400 and 1600 hours daily. What happens in New York invariably affects European indexes. Why the hell should that happen? The capitalisation of European indexes should contain only European components and, on a daily basis, what happens in New York should have no bearing on European indices. But, this is clearly not the case. Emotion prevails over common sense.

It is a case of the tail wagging the tiger, given that the European indices have outpaced American indices over the past three years.

[Scott M. Koser, CFA]

The assertion that the unemployment number, based on the household survey, is the correct number to follow is my opinion.

The self employed entrepreneur in the USA reflects the marginal gains in employment at the present time. These numbers are not reflected in the establishment survey because we are talking about fairly new enterprises with five or fewer employees. The collapse in the prices of forming a LLC, buying a computer with an office suite and accounting software, and inexpensive telecom services has lead to a cottage industry of the self employed who are 1099 employees of their own firm and not w2 employees of a larger corporation. Yours truly is a member of this unmeasured group.

I have been searching for a private market (read not labor union) economist who has both access to aggregate IRS data and the ability to proof this thesis out. If this trend exists it must be reflected in actual tax returns filed.

Why does it matter? Because labor conditions in the USA are much tighter than the bond market perceives via the establishment survey. If this is indeed correct, the FED could be on the sidelines for a long, long time.