Three Common Mistakes on Mean Reversion

Ryan Leggio: Finally one of the other things that we have talked about it is chapter eight of your book "Reversion to the Mean" and how those principles could be applied or misapplied sometimes in the investment world. Can you talk a little bit more about that concept?

Michael Mauboussin: I appreciate that question, because chapter eight is one of the more difficult chapters to read but actually for investors, I think is some of the most richest material. I think investors get this wrong for a few reasons. One is people tend to ignore mean reversion. It's not that people don't understand it, but they tend to ignore it. Second is they misunderstand what it means. Third thing is they often miss-specify it.

What happens when they ignore it? So again you talk to people and say "The reversion mean habits of markets," everybody goes yes, yes, we understand it. But, we see studying after study that show that people are really good at buying high and selling low. And this is not just mom and pop. This is a relatively sophisticated institution.

<TRANSCRIPT>

There is an article in the Journal of Finance in August of 2008 that studied 3,400 plan sponsors buy and sell decisions. Yeah, you know this one right?

Leggio: Yeah.

Mauboussin: So the number one reason to buy something is typically that it is done well. There are a lot of reasons to fire a manager, but typically the top of the list is they have underperformed and what you also see is from the time of hire/fire, subsequent 24 months the fired managers do better than the hired managers. So classic reversion to the mean. So again people know this but they are behaving in a way that suggests that they don't.

The second thing is to misinterpret it. And there is this very interesting idea of reversion to the mean. People tend to think you of...you run quartiles of anything, like say return on capital, the highest to lowest and of course over time those cohorts were moved towards the middle.

So you get a sense that things are moving towards mediocrity, returns. But, there is another way of looking at this that is somewhat differen--looking at the distribution's return on capital. So to be specific in our book, we showed that the 1997 return on capital pattern. It looks something like a normal distribution. Then we show the mean reversion from 1997 to 2007 and you get that pattern. But, then you look at the 2007 distribution, and it looks almost exactly the same as 1997.

It feels like there are two things going on. One is things are getting more towards the average. But, the other thing is things are not changing. So what is it? Is it change or no change?

I think the best way to think about that is what happens is luck basically reshuffles the companies on the distribution. So when luck reshuffles it, it shows up as mean reversion, but the overall distribution doesn't change at all.

So another way of saying that, which seems kind of perverse, is not only is there reversion to the mean, but there is also repulsion from the mean, as well, within the system.

So it is just to be very mindful that that's not the only phenomenon.

The third thing is mis-pecifying it, which is reverting to the mean. The question is: What mean? What is the actual number? In some series it is very stationary, very predictable, so the meaning is something that is very reliable. And other series that mean may be justifiably moving around.

You mentioned the P/E multiple before. Great example, right? What comprises a PE multiple? The answer would be things like interest rates, equity risk premium expectations, inflation expectations, tax rates and what do those series look like? They move all over the place over time.

So saying the multiple today should be the same as 1965 or 1935 really with those conditions being so radically different, it's hard to say what the mean should be.

So in that case I would say it is very circumstantial. Pay attention to the proper inputs that will generate, maybe not a perfect number, but a range of numbers that is much more appropriate given today's conditions and that would be a much better guide than saying, "Oh, well the last 200 years have been this."

Leggio: And you mentioned in institutional consultants and it maybe kind of predictable because of the career risk for them. It's pretty hard for an institutional consultant with a large money manager to stick with a manager who has maybe not done well for a few years. But the other troubling aspect of that is we see the same thing with mutual fund holders.

Mauboussin: Sure.

Leggio: You see tons of flows into funds that have done well recently only for them to revert to the mean and go back and you see outflows of funds that have temporary setbacks and then of course, those are the funds to rebound. How do investors really protect themselves from making those same types of problems?

Mauboussin: It is such a hard problem and there is a new book by Burt Malkiel and Charlie Ellis, which you may have seen. I was just reading this over the weekend, they were talking at about...this is over 20 years ended 2008. The market was up a little over 8 percent that period. The average mutual fund was up 7 so the difference being primarily fees, but the more disturbing thing is exactly that, is that the average investor earns substantially less than the average fund precisely because of that bad timing.

Again, everyone understands reversion to mean it's just this idea of being really disciplined to say when things are good for a fund or an asset class is typically time to take money off the table, when things are bad is time to put money in. It is easy to say and almost impossible to do.

There is another thing that you can do which is typically just go for really long periods of time so you ride the cycles out. I know there is a lot of controversy about buy and hold, but in a sense it would say buy and hold makes sense. You are just going to ride out these cycles and that would be another way to deal with it. That may not be optimal. But you have to have some mechanism to cope with that problem for sure.

Leggio: Well great, Michael. Thanks so much for joining us today.

Mauboussin: My pleasure, Ryan. Thanks.

Leggio: And thank you for joining us. This is Ryan Leggio for Morningstar.