Job Report: Much Ado About Nothing

The usual suspects are freaking out about the bad Friday jobs report, which showed unusually weak hiring (38,000 jobs created vs. estimates of 150k). However, as you can see below, every year for reasons unknown there are a couple unusually low prints, and then it rebounds.

In Jun 5, 2013, the number missed badly, yet it proved temporary. Same for early 2012.

Not much to lose sleep over.

Seeing the massive consecutive losses in 2008 helps put this recent miss in perspective.

For selfish reasons, I kinda hope it does get worse because I am long treasury bonds (which benefit from low interest rates) and it increases the odds of Trump winning.

Odds are the July number will be above 100,000 again, and no one will stop to ask why just a month ago everyone was losing their minds.

I decided to run a statistical analysis to see how abnormal or unexpected the bad job report really was. I downloaded BLS data dating back to Jan 2010 (78 months), which is close to when the recession ended, from from stlouisfed.org and uploaded it to plot.ly, an online spreadsheet program. I calculated a mean of 136,000 jobs created per month and a standard deviation of 44,000. Calculating the z-score for 38,000 yields a probability of 1.4%.

The result is a binomial distribution of a population size of 78 with probability of 1.4% of a worse job number and 98.6% of it being better than 38,000. Now the probability that there will be zero occurrences within the 78 months of worse number than 38,000 is only .986^78 ~ 33%, so we were due for a bad number.