“Essentially, all models are wrong, but some are useful.”
-George E. P. Box



OK kids, gather round for a quick debunking of the usual monthly idiocy. The May Employment Situation report seems to have taken on a special urgency from the carnival barkets, testosterone-poisoned traders, and other ne’er-do-wells working their hardest to separate you from your hard earned Do-Re-Mi.

As we have discussed so very frequently, the monthly change in net jobs is the most over-emphasized, least important highly questionable data point you will find. NFP is a minor monthly rounding error a series subject to the oscillations of both the economy and the data assembly work performed by legions of BLS economists and statistical wonks. It is subject to corrections, revisions and re-benchmarking. It is produced by a model, as so astutely observed by Professor Box, that is wrong. However, it would be foolish not to recognize that the BLS model can be to the intelligent observer who understand context, quite useful.

To achieve that insight, you must acknowledge your inner cognitive idiot, which often creates huge over-emphasis on the most recent data point (“Recency effect”), despite the simple reality that this is one in an ever growing series. Your inner idiot tends to de-emphasize the longer term trend, ignore internal specifics that have often contradictory implications, while failing to understand the difference between a variation above and below trend versus actual signs of a reversal of that trend.

Many (if not most) traders may be wholly unaware of the statistical wisdom the lovely prose above contains, but rest assured that one Ben Bernanke and his deep bench of statistical wizards are not unaware. They know that anyone monthly data point does not a series make.

Those of you who are expecting a wholesale policy shift based on this (or any other) month’s NFP are deeply, disturbingly delusional. You need to put the crack pipe down, stop drinking so early in the morning, and check yourself into rehab if you harbor even the slightest illusion that QE hangs in the balance of the data released at 8:30am today.

Instead, you should be looking at the context in which this report arrives:

1) This is a post-credit crisis recovery which is progressing via a gradual improvement in balance sheets (a “beautiful de-leveraging” to quote Dalio), sub-par GDP growth, anemic job creation.

2) Secular changes in employment — huge improvements in productivity, baby boomers retiring, globalization — have all impacted the overall employment trends, mostly with a negative bias.

3) The US equity markets have had an extraordinary run, tacking on 146% from the lows, based on the natural snapback that occurs when markets get cut in half, recovering profits, government bailouts, and Fed action;

4) Year to date, major US indices have run up 16% over the first five months of the year. This is simply too far too fast and needs to be digested.

With that context, we can look at the soft ADP numbers or some of the other inputs which suggest this may be a so-so report.

Will it be  ‘The Most Important Payroll Release in Years’? If you have been paying attention, you may have guessed not. But then again, I don’t have 24 hours of broadcast air to fill or a quota of blog pages to publish or a daily dead tree to spill ink on and sell to a gullible public.

This joint is called the Big Picture for a reason — my goal is to get you to focus on the 30,000 foot view, think in terms of long arcs of time, preferably years and decades (the astrophysicists and geologists and evolutionary biologists in the audience should feel free to chuckle to amongst yourselves at this). Please pull your foolish primate perspective out of your puny second-to-second existence that unfortunately blinds you to the world passing you by.


BLS report is out at 8:30 am


An Unusually Unusual NFP Payroll Day! (June 3rd, 2011)



* George Edward Pelham Box FRS (18 October 1919 – 28 March 2013) was a statistician, who worked in the areas of quality control, time-series analysis, design of experiments, and Bayesian inference.



Print Friendly, PDF & Email

What's been said:

Discussions found on the web:

Posted Under