Herd Theory

[jducoeur]

As we watch the volatility continue to downright goofy levels, I find myself idly wondering whether economists have started formalizing the madness of crowds yet.

I mean, even if you believe that homo economicus is rational (which I don't really), it's very clear that people en masse are not. Herd mentality is a very powerful force, and tends to overwhelm rational behaviour. Indeed, it's not even 100% irrational: if you expect everyone *else* to be nuts, following along can be the most appropriate course.

Consider: I didn't get involved with the 1998 tech bubble at all, because it was so clearly ridiculous. Which meant that, while I didn't lose anything in the crash, I also didn't gain anything in the runup. (A few people *did* manage to profit quite nicely from the bubble.) And bank panics are very sensible, in a way: the more people pull their money out of a bank, the greater the chance that it will go under, which means that you had probably better pull your money out of the bank.

I'd bet that there are some fairly consistent curves that can be interpolated into the standard economic ones, though: even if you assume human behaviour is irrational, that doesn't mean it is inconsistent. There are almost certainly chaotic feedback effects -- it's a dynamic system, and those are always complicated -- and it's very hard to predict on the small scale, but I'd bet that it is susceptible to smart modeling that can at least make some statistical sense out of the madness...

Comments

Bank panic

[dlevey.livejournal.com]

Does this hold anymore - at least for FDIC/FSLIC-insured banks?

Re: Bank panic

[alexx-kay.livejournal.com]

It moves up a level. Now it 'makes sense' to panic if you believe that there's a substantial risk that the FDIC system itself will fail. Now, that's close to saying "the US government collapses", but the risk of that has certainly gone up recently. Not enough to panic *me*, but enough to move some quantity of people at the margin.

Re: Bank panic

[jducoeur]

As Alexx says, there is still *some* risk, although it's more diffuse.

More to the point, though, FDIC is a fine illustration of what I'm talking about. Bank-panic behaviour is so consistent -- almost predictable -- that the FDIC was created as a specific mechanism to tamp down the effect. But I don't hear much about these sorts of tools at the theoretical level, despite how clearly effective it is, and I suspect that's because it *assumes* herd behaviour that traditional economics doesn't talk about very much...

Re: Bank panic

[dlevey.livejournal.com]

Well, it's not only predictable, but the insurance agencies were formed with hindsight. This sort of thing happened almost 80 years ago. And as far as the more diffuse risk is concerned, if the FDIC/FSLIC go under, we've got a LOT more to worry about. Chances are that US currency won't help us much by then.

But can you speak Chinese?

[qarylla.livejournal.com]

Unrelated in a way to the current economic forces, you might find Bellwether by Connie Willis an interesting read. It is about herd mentality in both sheep and people. It's a non-economic look at what you're talking about right here.

[serakit.livejournal.com]

So what starts it? If herd mentality is based on what everyone else is doing, there has to be, somewhere, someone who decided that what the herd is doing was logical, because someone had to do it before it was a herd mentality to create the herd mentality.

If that even made sense...

[jducoeur]

Yes and no. The relevant phrase here is "system dynamics" -- you might find that a fun area to dig into, next time you have some free time. It's one of the most interesting formalisms around, very broadly applicable but not all that well-known.

Keep in mind that most of these herd situations *start* with an entirely reasonable basis. For instance, in the most recent case, with concepts like "home ownership is good" and "we should do more to make it easier for people to own homes" and "if we spread the risk around, we reduce the odds than any one financial institution will fail". It was only when these ideas very, very gradually got carried to extremes that the situation started to get dangerous.

But the thing about herds is that they have a lot of inertia. Even though the situation is rationally kind of weird and scary (and a few people were predicting more or less this mess several years ago), the herd is very good at convincing itself that there is nothing dangerous going on, because nothing bad is currently happening. So the underlying situation just gets more and more dangerous, but on the surface it all seems to be okay.

So it all just keeps toddling along until something -- often something quite minor -- rocks the structure. This is called a "tipping point" in system dynamics: the point at which the system is so fragile that it really doesn't take much to change it. The longer things have continued unsustainably, and the more fragile the edifice, the less it takes to knock it over.

*Eventually*, something happens, though. In this particular case, it was the start of the ARM disaster -- adjustable-rate mortgages starting to "reset" to higher interest rates, which set off an initially small but growing wave of foreclosures. That caused house prices to fall, which caused banks to weaken, which caused credit to tighten, and before long the herd has gone from "It's a little problem" to "this is kind of worrying" to "OMIGODTHESKYISFALLING!!!!!!!".

Note that the reaction *was* in the correct direction: it's just that the herd has run right on past the rational new state and kept on moving, with the panic building on itself. This is particularly bad in market economics, because the market is built entirely on trust, and nobody trusts the market or each other at all at the moment. This is why government actions are so heavily focused on trying to inject *stability* right now, because stability breeds trust.

But really: it's useful to think of it as a herd of buffalo trying to follow the moving target of What Makes Sense. When the herd is stopped, it stays grazing in one place for a long time, even while What Makes Sense is moving further and further away. Eventually, a few of them start to move in the right direction again, and ever-so-slowly the group starts to notice and run in that direction. But once the whole thing is moving, it has such unstoppable momentum that it tends to move way past the current position of What Makes Sense before it begins to slow down and graze again, often in a position that is just as irrational as the one it started in...

[meiczyslaw.livejournal.com]

I ran across an article about such a study. Unfortunately, I didn't think to dump it onto my "sparky" tag on my journal because I'd posted something similar on September 30th.

But there are smart people who've figured out that traditional economic theory has a flaw, in that people are expected to act rationally. The new field includes trying to predict irrational behavior.

[dryfoo.livejournal.com]

Fella named Dan Ariely (I think that's the last name) a/k/a dandan@mit.edu -- formerly at MIT now at Duke -- wrote a well-received book about irrationality in economic actors. It might have some of what you're looking for.

Regarding predicting the market: since it really is chaotic, those characteristic shapes you're looking for will be there, but they're going to appear at every scale, so you won't know if you're seeing the peak of the day, the week, the year, or just the previous 5 minutes.

[jducoeur]

It might have some of what you're looking for.

Interesting -- I'll have to keep an eye open for this. Thanks!