All Models Are Wrong, Some Are Useful

Summary

Let the Index of Leading Economic Indicators' failure to predict the post-COVID economy be a constant reminder that this is no longer the pre-COVID economy.

Assuming it is as too many do, including Powell, is how mistakes are made.

Comment

The August Index of Leading Economic Indicators (LEI) was released yesterday. It is a model of ten indicators that predict the economy.

The ten indicators are combined into an index.

As this chart shows, it is at its lowest level since 2016.

This Index has now gone 30 consecutive months without a positive reading. This is far and away the longest string in the 64 years of data shown.

Typically, a recession occurs after six to eight negative months (shaded areas above). The number of negative months is now at 30 (top panel below), and the economy is nowhere near a recession, as shown by quarterly GDP growth (bottom panel).

All models are wrong, some are useful - George Box (UK Statistician)

The LEI is a model. Its construction and indicators mimic the thinking of many economists and market types. Many have a similar conclusion to the LEI that the US economy is a disaster. This is why so many of them pined for a 50 basis cut, which they got, and continue to look for more cuts at this pace.

But as the chart above shows, this model is flat-out not working.

Even though this model is wrong, it is useful. It helps explain why so many are so pessimistic about the US economy.

Change

We have argued that recessions and financial crises change an economy for years. We had both in 2020. This changed the economy.

Change does not mean worse or dystopian; it means different. This economy is different from the 2019 (pre-COVID) version.

Remote work, mass immigration, deglobalization, changing spending attitudes, and the psychological impact of 40-year high inflation have all contributed to this change.

How long will these changes to the economy last? Until the next recession and/or financial crisis? Most likely, that change, like 2020, will not be good or bad; it will just be different.

Data Dependent

Who does not think the economy has changed? Jay Powell, as he noted in his July Press Conference. Highlights

And, as I said, we’re prepared to respond if we see that it’s—that it’s not what we wanted to see, which was, you know, a gradual normalization of conditions ... I think you’re back to conditions that are close to 2019 conditions, and that was not an inflationary economy—broadly similar labor markets then.

We are not going back to 2019. That was the "pre-COVID" cycle. We are now in the "post-COVID" cycle.

This is why data dependence does not work. You have to assume a set of relationships between the economy and measures of it, like payrolls, inflation, retail sales, etc. However, if the economy is shocked by a change, like in 2020, these relationships will not work in the new cycle.

So, let LEI's failure to predict the post-COVID economy be a constant reminder that Powell, many economists, and several market pundits spend their days describing how things worked in 2019, like LEI is doing, and not how they work now (post-COVID). This is how mistakes are made.

Inflation

As we have argued in previous posts, 2020 has also turned the inflation cycle for almost four years now. We are not returning to pre-2020 inflation (red) but to a new post-COVID inflation cycle (blue).

Cutting rates under the assumption that this is still the 2019 economy risks a big policy mistake.