You Can Ignore Non Farm Payroll Day
Over the past few years, we have admonished traders to ignore the gyrations of any given NFP report. A single monthly number in a series is meaningless, and the overall trend is what matters more, as do certain specific components.
We should generally ignore a single month for many reasons. First, they are the output of a model — increasingly counted less, and hypothesized more. Second, they are highly subject to revisions, as a real time snapshot of a very large and complex economy. Lastly, the monthly changes in a 135 million person labor pool are practically a rounding error. 135k change is one tenth of one percent — 0.1%. And as we have seen on some recent months, a change of ~10k or so is a 0.01% of the full pool, well within the margin of error.
That’s why we emphasize trend, or a 3 month moving average, versus getting too bent out of shape over any given month. And we have done this in good times and bad, when we were bullish or bearish.
Given the potential impact of the snowstorms across the United States during the BLS survey week, the potential range of NFP for this month is half a million people wide. Any jobs report between +100k and -400k would not surprise us.
Here’s a real world example of how easily economists can be led astray: Robert Barbera, chief economist of ITG, noted that January 1996 saw a -200k due to snow, versus the previously run rate of +200k. Economists freaked, and two forecast recession BASED ON THIS SINGLE DATA POINT. The following month? The next month’s NFP was +700k.
That’s why you never rely on a single, potentially aberrational data point oin a long series.
Employment Situation out in one hour at 8:30 am
Employment Situation Summary
http://www.bls.gov/news.release/empsit.nr0.htm
Tweet
Facebook
Reddit
Digg this!
March 5th, 2010 at 7:36 am
CA revision for 2009 looks to increase job losses by 50% compared to their original numbers, 870K jobs lost vs. 570K original. How could they be off by 50%, employers croaked, their death index shot up faster than births. Guess Trimtabs tax receipts based forecast is right again
http://www.contracostatimes.com/ci_14493140?source=most_viewed&nclick_check=1
March 5th, 2010 at 8:04 am
Not only is a single data point indicative of nothing, the veracity of the data point itself must be questioned. As rktbrkr points out, what good is an inaccurate data point? Maybe we should set automobile speedometers and airplane altimeters so that they are accurate within +- 50% of the actual speed or altitude of the vehicle. What could go wrong?
March 5th, 2010 at 8:13 am
Barry–
I don’t disagree with your comments.–just an added thought. Joanie McC looked at the definitions of employment used in both the establishment and household surveys the other day, in the wake of Summers’ comments. The ability of snow to impact the numbers is tiny, close to non-existent.
March 5th, 2010 at 8:16 am
What good is an inaccurate data point? Lets rephrase this slightly,”Why would the government provide inaccurate employment data to the public”? Secondly,is there any pattern to subsequent revisions of government provided data points? Is all this coincidental?
March 5th, 2010 at 8:18 am
Looking at the BIG PICTURE we can ask, “would the government lie to us”?
Treasury official denies “too big to fail”
http://www.businessinsider.com/treasury-official-says-with-a-straight-face-too-big-to-fail-doesnt-apply-to-big-firms-2010-3
March 5th, 2010 at 9:10 am
[...] true, as Barry Ritholtz points out, that 36,000 in a work force of about 135M is little more than a rounding error. What matters is [...]
March 5th, 2010 at 9:54 am
That’s why you never rely on a single, potentially aberrational data point oin a long series
Happens to correspond precisely with the business of sighting a rifle. ALWAYS get a 3- or 5-shot group before adjusting your sights. NEVER use a single shot to adjust.
March 5th, 2010 at 11:15 am
[...] be positively in the plus category by now. Not exactly. Economies and recoveries never move in straight lines. And most of the other economic data out there is pointing to a recovery. Retail sales are rising [...]
March 5th, 2010 at 1:00 pm
[...] Feel free to ignore any single month of the Non-Farm Payroll numbers. (Big Picture) [...]
March 5th, 2010 at 4:34 pm
[...] – Investors would be better off ignoring today’s jobs report, FusionIQ CEO Barry Ritholtz advises, as the overall labor market trend matters much more than one month of data. A monthly number is [...]