PsyFi Search

Saturday 1 May 2010

Herd of Investors

Bovine and Asinine

It seems sort of obvious that investors, a generally bovine group when not in asinine mood, will tend to congregate in herds and then charge about randomly, often over the edge of the nearest cliff. If this is true, however, it poses a set of puzzles that it’s not clear that any of the current approaches to understanding mass market behaviour can properly explain.

Certainly this behaviour isn’t efficient in the sense of obtaining the right price for a security, because deciding what to do based on the person running about in ever-decreasing circles next to you doesn’t come close to propagating useful information. However, if investors are irrational in herds then this implies that somehow their behaviour is synchronised and it’s certainly not clear that the simple set of isolated psychological biases that analysts currently work with is anywhere near sufficient to explain this.

Flocking and Herding

A simple model of flocking in computer models has been shown to be sufficient to generate surprisingly realistic herding behaviour – the sort of thing we looked at in Boids, Bacteria and Market Behaviour. Being able to show that this type of model can produce something that looks like the synchronised behaviour we see in the real-world is a long way from explaining how it works, however, because humans demonstrate intentionality – that is, we don’t normally work on some pre-programmed autopilot but have internal motivations that cause us to do things based on personal preferences. Or do we?

One of the other noticeable attributes of humanity is its strong tendency for mimicry. From a very young age human children copy adults and the urge to imitate is found in many higher animals. Indeed it’s been suggested that the large human brain evolved in response to selective pressures to copy others. This idea would be more credible if there weren’t dozens of alternative evolutionary theories for the size of our crania – this one-size-fits-all theorising is one of the reasons why many people are sceptical of adaptationist claims: natural selection can be adduced to explain everything, so in the end may explain nothing.

Information Cascades

This aside, it certainly seems that mimicry lies behind a lot of observable animal behaviour, from mate selection through to dietary choices. Especially in situations of uncertainty it makes perfect sense to copy what everyone else is doing. You just hope that when the fire alarm goes off the confident person everyone else is following knows where they’re going.

What psychology suggests is that sometimes these mimicry effects can set off information cascades – a kind of chain reaction where everyone copies everyone else thinking that everyone else knows what they’re doing. At this point we’re no longer doing our own rational analysis but are simply copy the nearest available role models, whether these are neighbours, internet buddies or just the bloke in the bar last night.

Information cascades have a number of interesting qualities: they can emerge quickly based on the behaviour of a few dominant individuals but they’re very fragile, because the receipt of new and conclusive information can cause people to suddenly change their minds. They’re also associated with sudden stampedes as everyone charges off in the same direction, seemingly all at once. Basically, an information cascade looks an awful lot like what happens when a herd of stockmarket investors all go mad at exactly the same time.

Availability, Again

Quite how the individual psychology works in order to induce these wide social effects isn’t understood. Simple behavioural biases like the availability heuristic may be enough under some situations to trigger this type of herding – if the press and the cab drivers are doing nothing but broadcasting the latest set of stock tips then such information is easily available, although of highly dubious (yet approximately equal) quality. Similarly the mere exposure effect suggests that people start to like things more merely by being exposed to it: expose people to enough information about a certain stock and you’ll increase the likelihood of wanting to own it, regardless of the fundamentals. Ever had that “must own” feeling after reading about a stock?

Kuran and Sunstein looked at a model of availability cascades which links together social interaction and the availability heuristic to create waves of self-replicating and self-reinforcing behaviour. The more information is available to investors the more likely they are to take notice of it, to transmit the information to others and so build up the cascade of information which starts to swamp more rational interpretations of events.

It’s Good To Talk

This is potentially quite a radical reinterpretation of the way securities valuations move: it’s not purely because investors observe price movements and use this as information about whether to trade or not – a classical momentum trading approach. No, this rather amazing idea suggests that people actually talk to each other. This stunning discovery, along with the realisation that humans communicate via bulletin boards, social networking sites and broadcast media, makes it possible to conceive that people may trade in securities on types of information other than that gleaned from securities analysis – you know, like that hot-tip from the shoe shine boy.

Sophie Shive has done some interesting research using a model of epidemics based on what she calls “socially motivated trades” – basically trading based on the spread of information through social networks – and has found that these can predict future market returns:
"...socially motivated trading can predict stock returns. Individuals themselves over time have become less susceptible to social in°uence but they are more subject to it, and thus the number of trades caused by social in°uence increases slightly over the sample period"
Moreover these effects don’t disappear: this is real market moving information spreading amongst market participants. Behind this there’s some evidence that the asymmetries in market movements we see from time to time, may emanate from a couple of simple human psychological errors.

Conversational Biases

Firstly we have a tendency to present ourselves in the best possible light. These conversational biases usually mean we talk about things which have gone right and tend to avoid discussing those which have gone wrong. This is a perfectly natural bias but has the effect of reducing the amount of bad trading anecdotes which get relayed around social networks. As any brief perusal of stock trading bulletin boards will show we’re exposed to way more positive information than we are negative.

Secondly we don’t discount for the way that information gets relayed around or social networks. Although the empirical evidence on this is limited there is some suggestion, as outlined in DeMarzo, Vayanos, Zweibel, that we’re fooled by something they label as persuasion bias:
"We assume that agents treat any information they receive as new, independent information, thus failing to adjust for possible repetitions. We refer to this heuristic as persuasion bias because it implies the phenomenon of persuasion. Suppose, for example, that one forms an opinion on a political candidate by reading a newspaper column. Then, by treating the opinions expressed by the journalist each day as new, indep endent information,one would gradually agree more with the journalist’s opinions."
Essentially we may receive information from multiple people that may actually originate with a single person. In a network single critical nodes – influential and well connected people – may have disproportionately more influence than we realise.

Avoid the Stampede

Herding seems to occur in many aspects of the market from mutual fund stock purchases to stock analyst recommendations through to private investor trading and waves of behaviour in corporations who all suddenly decide to do rights issues or make expensive acquisitions or whatever with perfect synchronicity. That this social behaviour is somehow underpinned by individual psychological biases is a certainty, but that’s nowhere near enough to help us make money. In fact, sometimes following the herd can be a very profitable activity if you have the personal courage to peel off before everyone else goes over the edge.

Not many people can do this and generally ignoring the crowd’s signal is the safest route for most investors. How people choose to do this is down to the individual but, as usual, it probably comes down to either focussing on the fundamentals or simply isolating yourself from events. Generally, though, if you have to join a stampede do it early – and leave before the round-up.


Related Articles: What's the Shape of Your Recession?, Anchoring, the Mother of Behavioral Biases, Boids, Bacteria and Market Instability

7 comments:

  1. Interesting post. It is absolutely clear that a lot of "investing" is done without any semblance of analysis - we hate to see our neighbor get rich. The question - to me at least - is how can we tell when herding is near a peak. When herding peaks happen they are probably followed by the biggest downturns. It seems to me that herding happens in sports as well in the "jumping on the bandwagon" effect. People seem to jump on at the last minute. When the Redskins went to the Super Bowl the first time, people who had never watched a football game became avid fans as they started to win playoff games.
    Maybe the best indicator of herding in the markets (although crude in my opinion) is the P/E ratio as proposed by one of your frequent commentators.
    It has always amused me that someone who had always criticized the capitalist system and the stock market in general got on board in late 1999 and actually called everyone he knew touting a couple of companies that subsequently became essentially worthless - instead of "...heading to $100/share" , today they are penny stocks. We know of course that the same behavior took place in the late 1920s.

    ReplyDelete
  2. ignoring the crowd’s signal is the safest route for most investors. How people choose to do this is down to the individual

    I disagree, Tim.

    Bull markets destroy the society in which they take place. The primary reason why we are living through an economic crisis today is that as a society we permitted stock prices to rise to insanely dangerous levels in the late 1990s.

    Here are a few words from Irrational Exuberance:

    "It is a serious mistake for public figures to acquiesce in the stock market valuations we have seen recently, to remain silent about the implications of such high valuations and to leave all commentary to the market analysts who specialize in the nearly impossible task of forecasting the market over the short term and who share interests with investment banks or brokerage firms. The valuation of the stock market is an important national -- indeed international -- issue. All of our plans for the future, as individuals and as a society, hinge on our perceived wealth, and plans can be throw into disarray if much of that wealth evaporates tomorrow."

    We either act collectively to stop bull markets before they begin or we all suffer the damage done to our economic and political system as the result of failing to do so. By caring a bit for others we ultimately greatly enrich ourselves.

    Rob

    ReplyDelete
  3. Interesting and sensible post as usual. a few questions:
    First: how much effect do small speculators have on share prices (I imagine that it's small fry who make bubbles).
    Second: assuming the answer to that question is 'a big effect', are there studies comparing bulletin board activity to share performance, and the effect of certain favoured posters on bulletin boards. I look at iii.co.uk quite a bit. Some boards operate in a state of mania and abuse, some are quiet and measured. I tend to think that a board with maybe 5 well-spelled, sober posts each day, and including one of the more-favourited posters indicates a better investment prospect. Do I have any justification for this?

    ReplyDelete
  4. Hi x

    Good questions, worth researching which I'll do, so no easy answers. As a starter I did write an article about this a while ago, although it doesn't directly address these questions: Bulletin Boards Are Bad For Your Wealth

    FWIW I'm not convinced that bubbles are entirely about small speculators - it may be that it's the small fry that get things going but once stuff is bubbling the professional investment classes are generally predisposed to keep things going. But there's no doubt that bb's are often hotbeds of confirmation bias: it's essential to continually check your beliefs against the numbers.

    ReplyDelete
  5. Thanks. One deadly effect I've noticed is the result of actually making posts on a bb, never mind just reading them. A negative reaction, especially, causes extra commitment to the share as I try to overcome the drop in social rank (among people I haven't met) by justifing my holding; and a positive reaction causes a feeling of guru-like confidence. Disaster.

    I could and should do the bb research myself, though it might be a bit complex: what % of the float is ouside the possession of institutions; quantity/length/originality of posts; spelling; abuse; tolerance of dissent; ranking of posters; number of active posters; perhaps one could make a 3d space in which to plot and characterise a bb..

    ReplyDelete
  6. Interesting article and posts, as always.

    I've always had a rule of thumb that whenever a advfn poster (which i consider to be a bb with more than usual levels of froth) starts posting on the Motley Fool (which I consider to have less froth than most) about a particular stock or sector or market, it's time to start thinking about peeling away from the herd.

    I certainly agree that BB's in particular are hot-bed's of confirmation bias - in fact they are structurally designed that way with individuals joining particular boards covered shares/sectors that interest them, which suggests some sort of personal commitment to that share/sector.

    Mind you, it's exactly the same sort of self-selection and confirmation bias that governs those who comment on a blog such as this and agree with each other how much cleverer we are than the average punter...

    :)

    ReplyDelete
  7. The non-billionaire status of the owner of iii probably rules out the existence of a board's predictive power... however, neither is he a pauper, so I live in hope.

    ReplyDelete