If you want to predict how this tariff story is going to shake out, the best way to do so is to invoke Regression To The Mean.

We really don't have much (any?) information about what's going to happen next, so everything is a guess.

When you're predicting without information, Regression To The Mean should be your go-to.

Here's how it works intuitively:
* A guy who has been poor all his life, but wins the lottery or something you can expect will become poor again.
* A guy who has been wealthy all his life, but from some freak market condition or lawsuit, ends up bankrupt, you can expect he will end up becoming wealthy again.

Why do the poor tend to go back to being poor and the rich tend to go back to being rich? There are a lot of reasons which, for prediction purposes, we can safely ignore.

It's actually BETTER to ignore it, because when you lack reliable information, you start trying to pull signal out of noise and you can end up telling yourself whatever story you want.

Regression To The Mean just means that things tend to go back to their normal state of being. Does it always work that way? NO. If it did, we'd still be speaking Latin. Things do change, but far far far less often than how often people THINK they will change. Refer to figure 1.

China has had a really good run over the past 30 years. Like a guy going to the casino and doubling his money in one night. Unfortunately for him, the odds are he won't be able to pull it off twice.

Using Regression To The Mean and looking at countries over the past 100 years, we can make some predictions:

Rich countries:
* USA
* Western Europe
* Japan
* Canada / Australia / NZ
* Argentina
* Venezuela

Poor countries:
* China
* Russia
* Brazil
* Eastern Europe
* India
* Africa

Argentina and Venezuela are special countries to watch because they are historically rich countries, but have been beaten down very badly. These are potentially big (albeit risky) opportunities.

@cjd
> there are lots of reasons, which, for the prediction purposes, we can safely ignore

I think it's more like: since there are many reasons we don't know, and most of them will be unaffected by the disruptive event, therefore things will likely return to normal.

This isn't ignoring those reasons.
We're relying on them being many and independent to predict that nothing will change.

@cjd It's the same reason that makes most random-looking processes in nature have gaussian distribution: that's what you get by summing up many independent factors that you don't know (so can't correlate to)