Recession Model Update (May 2021)

Friends,

Surprisingly, model values spiked up this month. But I believe this is caused by an arithmetic glitch rather than a deeper underlying signal. Explanation is below.

Current Values (as of May 8, 2021)

The Short Story

I think this spike will be short-lived, since it is likely caused by the model’s input variables having 12-month lookback windows. As last spring’s numbers leave the 12-month lookback windows, the model output values should resume their decline.

The Long Story

Last Friday, April nonfarm payrolls showed a gain of +266,000. The problem is this model focuses on the difference between the 3-month change and 12-month change in nonfarm payrolls, instead of the month-by-month change. The reasoning here is that the difference between the 3-month change and 12-month change in nonfarm payrolls captures acceleration or deceleration in nonfarm payrolls, which may foreshadow peaks and valleys in nonfarm payrolls.

The issue is that last spring (April in particular) was the valley for nonfarm payrolls (see below), causing the 12-month change in nonfarm payrolls to be incredibly large.

This causes the difference between the 3-month change and 12-month change in nonfarm payrolls to turn deeply negative this month, which typically indicates the beginning of a recession (see below). But in today’s case, this is just caused by the timing of the 12-month lookback window, which still includes last spring’s extreme values.

So as last spring’s numbers leave the 12-month lookback windows used by several model variables, I expect the model output values to resume declining.

---

Until Next Time

Terrence | terrencez.com

View previous updates in the archive.

I’m always eager to hear your comments or questions. Just hit reply to this email, and I will respond. You can also fill out the contact form here. Either way, I will respond from my personal email.

All opinions are my own and not the views of my employer.