Friday, December 2, 2016

A May 2017 recession?

There's been some discussion out there that we're due for a recession. For example the US has never gone more than 10 years without one. I thought I'd try this simplistic model based on the structure of the unemployment rate (more here). It's still uncertain, but it does say we should expect  a higher unemployment rate rather than a lower one in the next few years:

The vertical line is the estimate of the recession date (per the link above) of the latter half of May 2017 (plus or minus 4 months, technically 2017.4 ± 0.4).

Note that this model is very simplistic ‒ it allows for a zero or even negative unemployment rate. That means that adding the prior that the unemployment rate will typically be above 3-4% would increase the expected unemployment rate, so our estimate is a conservative one. Here are the probability distributions for the unemployment rate for 2017 Q1 through 2018 Q1 (by quarter):

Here are the results in table form (end of quarter)

2017 Q1       5.0 ± 0.3 %
2017 Q2       5.4 ± 0.8 %
2017 Q3       6.2 ± 2.0 %

The rest of the uncertainties are too high to be useful.

Anyway, this is more for fun. This "model" is independent of the information equilibrium model, and is based entirely on an assumption of an "dynamic equilibrium" in the unemployment rate (such that it drops by about 0.05 percentage points per month (0.6 percentage points per year).

It is interesting that a possible information equilibrium (IE) indicator of recession (see here and here) has been showing that a recession is possible. In the past, short interest rates above the IE model are a precursor to recessions, and we seem to be entering another period of "high" interest rates:


  1. "A May 2017 recession? "

    Given you believe in the magnificent predictive ability of your model, presumably, you have shorted the shit out of the Dow Jones with everything you have?

    1. I assume you quoted the only part of the article that you read?