Saturday, September 14, 2019

Odds and ends from the first half of September

I been really busy these past few weeks, so haven't made many updates to the blog — mostly posting half-thoughts and forecast tracking on twitter. One thing I did post about was a fluctuation in the JOLTS data around the dynamic equilibrium appeared correlated with the S&P 500. I updated it today to emphasize that this is a 2nd order effect — on the order of a few percent deviation from a dynamic equilibrium. I did try out a scheduled tweet that came out just before the unemployment data was released at 8am ET on Friday 6 September 2019 (click to enlarge):

The DIEM forecast got the data exactly right. I also noted in the thread that the DIEM forecast outperforms linear extrapolation — even if you try to choose the domain of data you extrapolate from (the different lines in the second graph show all the different starting points for the extrapolation):

This means that the DIEM is conveying real information about the system.

CPI data came out this week and the DIEM continues to do well there too (continuously compounded and year over year inflation):

One thing to note is that the DIEM model is extremely close to a fit to the pre-forecast and post-forecast data (black dashed) and the non-linearity in the DIEM model (red) actually improves the relative performance:

This means that for a function that is this smooth over time, no other model could be anything more than a marginal improvement. The only possibility of doing better is if the fluctuations around the DIEM path are not noise — and in fact the "cyclic" fluctuations around the DIEM path might be related to the fluctuations around the JOLTS log-linear path:

If you squint, the inflation fluctuations might be in sync with the JOLTS fluctuations:

However, this is fairly uncertain — it's not a robust conclusion at this point.


In addition to looking at macro time series, I also took a look at some demographic data about childhood mortality using a new data set. We can see the effect of sanitation in the UK, as well as a potential effect of the more general legalization of abortion:

The data for Japan doesn't go back as far, but shows data consistent with a similar "sanitation transition" (when extrapolated) as well as the effect of WWII:

The US data doesn't go back far enough to make any conclusions (and the shocks are somewhat ambiguous):

Thursday, September 12, 2019

Market-correlated fluctuations in employment data

Something I noticed in the JOLTS data was that if you subtracted out a "dynamic equilibrium" (log-linear path) [1], the residual data was almost sinusoidal (click to enlarge):

I wouldn't expect this sinusoidal fluctuation with a constant frequency to continue simply because there aren't a lot of true waves in economics — the sine wave is more of a curiosity [ETA: the period is about 3.5 years].

But then I noticed that this pattern matched the same residual S&P 500 fairly well, but with a lag of about a year (meaning the S&P predicts the fluctuation):

In fact, all of the other JOLTS data series appear to show this correlation as well (separations = TSR, hires = HIR, quits = QUR):

Even the unemployment rate shows the same fluctuation — but with a lead of about 16 months:

That is to say this would indicate the fluctuations in the unemployment rate predict the S&P 500. As the unemployment data goes back farther, I can look at whether the correlation holds up over time. Eyeballing the data, it actually looks more correlated at zero lag/lead. The run-up in the market before the 1987 crash (might be interesting viewed in the context of the post-80s recession step response), the dip in the mid-90s, and the dip in the 2002-2003 time period line up with fluctuations of the unemployment rate data away from a local log-linear fit. It's also just much noisier in general. 

In any case, I will look more closely at this as well as "help wanted" index data from Barnichon (2010) (see e.g. here).


Update 14 September 2019

One thing I would like to point out is that this fluctuation is on the order of a few percent on top of the dynamic equilibrium path —  a second order effect.



[1] Take the data series JTSJOR (job openings on FRED) take the log and fit a line y = a t + b to the log data and subtract log JTSJOR − (a t + b) ≡ Δlog JTSJOR.

Monday, September 2, 2019

Under-employment case studies: US and Australia

Happy labor day (in the US)!

John Quiggin was surprised at the steady increase in Australian under-employment [1] — I was, too. In the US, most of these labor metrics all follow the unemployment rate ... it's as if there's a prototype labor market time series and all the other metrics are just minor log-linear transformations. And that's true for US under-employment (employed part time for economic reasons) as well:

There was either a change in the way the data was recorded, or a major policy success in January 1994 (I'll look into it more, and would be grateful for anyone who might have any suggestions). In any case, it's not continuously rising, but rising in recessions and falling in their aftermath like most other labor market metrics in the US.

The article Quiggin discusses also talks about youth unemployment (ages 15-24), which in the US shows exactly the expected behavior — following the US unemployment rate:

The youth employment rate also shows the same basic structure:

As a side note, the previously observed dip in the youth employment rate appears to have faded (link shows both employment and unemployment).

The one major difference in behavior is in the fraction of long term unemployed ("long term" is 27 weeks or longer), which in the US seems to have undergone a change since the 1990s:

The basic structure is again similar to the US unemployment rate, but each subsequent recession since the 90s has increased the fraction of long term unemployed without a commensurate drop in the recovery. This results in an increasing fraction of long term unemployed over time relative to the unemployment rate.

Australia on the other hand has just seen a relatively steady increase (about 2.5% per 25 years, or 0.1% per year) in under-employment since the late 70s with only a big surge in the (last) Australian recession in the 90s. Note: no major surge occurs with the 80s recession in Australia ... only a tiny blip. This is primarily driven by people aged 15-24 [2].

At this rate, 10% of the population will be under-employed in 15 years — whereas the US will, in the absence of a recession, get down below 2%.



[1] The nut "graph" in the article on Australia (click to enlarge):

[2] Here's the breakdown by age showing it's driven by youth under-employment (click to enlarge):