Numbers and Stories
We’ve seen before that investors are generally attracted to a good story and tend to shy away from the hard problems associated with analysing numbers. Worse, even if people do look at the numbers they tend to be swamped by information to the extent of not knowing what’s important and what’s not. Although generally this is only obvious in retrospect, anyway.
However, there are strong suggestions that our inclination to follow a good and particularly interesting story isn’t simply stuff that happens. It looks as though this is built into our processing centres and is a driving force behind a lot of what we do on an everyday basis. We’re simply misapplying the lessons of life.
Under-estimating Bad Stuff
As we saw in Investor Decisions – Experience Is Not Enough investors are, in defiance of the basic tenants of behavioural finance, liable to under-estimate the likelihood of rare events in particular situations where they’re repeatedly exposed to feedback from small decisions. People only get a small sample of possible results and update their internal models slowly, so the relative rarity of a rare but extreme event fools them into under-estimating the dangers.
For example, if this idea is correct then casino slot machines that pay out huge amounts but only on rare occasions would end up dying of rust in cobwebbed corners – people would decide that the machines weren’t likely to pay out and give up. To prevent this happening the machines cough up regularly, just enough to confuse our return estimating algorithms. People seem to make the same calculation with minor traffic transgressions, like running traffic lights in the dilemma zone – nothing bad happens nearly all of the time, so they keep on doing it.
To Jay-Walk or Not
Although this looks a lot like Prospect Theory – one of the mainstays of behavioural finance – is wrong in fact it’s more that this type of decision making from the experience of multiple blind trials with feedback is exactly the opposite of the scenarios that a lot of behavioural finance researchers have looked at. Much of this research has focused on so-called one shot, decision making from description; big decisions like taking out a mortgage or getting married.
These standard experiments provide information in a written and structured form and format that’s quite different from what we experience in real-life. The situation is more akin to a formal written exam than figuring out when to jay-walk safely. Both of these are real-life situations, but we decide to jay-walk or not frequently while we only occasionally take an exam.
Meta-analysis Madness
In 2004 Weber and colleagues did a so-called meta-analysis of the experiments carried out on decision making of the kind where a choice has to be made between a sure thing and risky prospect. A meta-analysis is simply a piece of research which analyses the results of lots of similar studies – which sounds like a bit of a cheat but is actually incredibly important because such studies can sometimes generate enough data to extract important findings where none appear otherwise. For example, the finding that steriods significantly improve the chances of a premature baby surviving was pulled out of a meta-analysis, even though the underlying studies weren't anywhere near conclusive.
Anyway the study found 204 studies of two-condition decision making and found that every single one of them was based on a descriptive questionnaire. Now I don’t know about you, but when I’m about to cross the road in traffic I don’t expect a friendly researcher to walk up to me with a piece of paper explaining the probabilities of being flattened by an eight-wheeler in the middle lane. In everyday reality we need to figure these things out for ourselves by messy trial and error, so it rather looks as though a lot of these studies are looking at the wrong thing.
As we saw previously this doesn’t utterly undermine the findings of behavioural psychology, although there’s certainly a feeling that someone, somewhere is trying to light a fuse. Generally we can identify two differences between decision making from description and decision making from experience – firstly the different form of the question and secondly that the former is generally one-shot with limited feedback and the latter happens multiple times with pretty much instant feedback.
Sampling and Feedback
When Herwig and colleagues investigated this they figured out that it was the experience of making decisions and receiving feedback on the results that was the key. The design of the study allowed them to look at why this might be. Participants were allowed to do as many trials on each question as they wanted in order to figure out the correct result in any particular study yet the actual number of trials performed were relatively small – 15 on average – and because of this relative small sample set people generally under-estimated the likelihood of the rare event.
The reason for this is pretty obvious – if the rare event is, well, rare, then the fewer times you sample the information then the less likely you are to come across the rare event. The research showed clearly that those people who did experience the rare event were more likely to either avoid that option, if it elicited a negative outcome, or select that option, if it elicited a positive one. We have simple learning strategies underneath all of the complex social baggage.
Recency Redux
Unfortunately this gets worse, because of our old friend recency. Even those people who do experience the rare event get confused by more recent information. Again the research shows that those people who experienced the rare event late in the experiment seem to update their mental models to include it while those people who experience it early in the experiment, and not later on, seem to perform a similar update to ignore it.
All of this complex statistical stuff seems to boil down to some very simple and pretty much recognisable outcomes. In real-life situations where extreme events happen rarely people are likely to not experience the event very often and will therefore ignore the probability, even when it’s relatively high. In addition, even if they experience an extreme event early on recency effects may see this get overwritten by later information which doesn’t include it.
Investors in the School of Hard-Knocks
The relevance of this to stockmarket investors is immediately obvious. The markets are real-life learning experience in which most people tend to learn by trial and error. Generally extreme events, like companies failing altogether or markets collapsing like a pack of cards, aren’t experienced very often so the opportunity to learn from them is relatively rare. Unsurprisingly then, people’s expectations of the relative probability of these really bad things happening are generally much lower than the real probabilities.
All of which essentially means that if you’ve never experienced really bad stuff on the markets you’re almost certainly just very, very lucky. It also implies that the best way of learning about investing is from books rather than experience. Sadly books can’t teach you how to deal with the emotional rollercoaster of seeing your personal wealth ebbing and flowing with disconcerting regularity, over time.
The answer, such as it is, is that learning from experience and learning from description are both necessary to be a good investor. A few hard knocks along the way are no bad thing, as long as you don’t consign them to the bad luck bin. Stuff happens, frequently: give the markets enough chances and they’ll make sure it happens to you.
Related Articles: Fairy Tales for Investors, Investor Decisions - Experience is Not Enough, Recency: Hot Hands and the Gambler's Fallacy
We’ve seen before that investors are generally attracted to a good story and tend to shy away from the hard problems associated with analysing numbers. Worse, even if people do look at the numbers they tend to be swamped by information to the extent of not knowing what’s important and what’s not. Although generally this is only obvious in retrospect, anyway.
However, there are strong suggestions that our inclination to follow a good and particularly interesting story isn’t simply stuff that happens. It looks as though this is built into our processing centres and is a driving force behind a lot of what we do on an everyday basis. We’re simply misapplying the lessons of life.
Under-estimating Bad Stuff
As we saw in Investor Decisions – Experience Is Not Enough investors are, in defiance of the basic tenants of behavioural finance, liable to under-estimate the likelihood of rare events in particular situations where they’re repeatedly exposed to feedback from small decisions. People only get a small sample of possible results and update their internal models slowly, so the relative rarity of a rare but extreme event fools them into under-estimating the dangers.
For example, if this idea is correct then casino slot machines that pay out huge amounts but only on rare occasions would end up dying of rust in cobwebbed corners – people would decide that the machines weren’t likely to pay out and give up. To prevent this happening the machines cough up regularly, just enough to confuse our return estimating algorithms. People seem to make the same calculation with minor traffic transgressions, like running traffic lights in the dilemma zone – nothing bad happens nearly all of the time, so they keep on doing it.
To Jay-Walk or Not
Although this looks a lot like Prospect Theory – one of the mainstays of behavioural finance – is wrong in fact it’s more that this type of decision making from the experience of multiple blind trials with feedback is exactly the opposite of the scenarios that a lot of behavioural finance researchers have looked at. Much of this research has focused on so-called one shot, decision making from description; big decisions like taking out a mortgage or getting married.
These standard experiments provide information in a written and structured form and format that’s quite different from what we experience in real-life. The situation is more akin to a formal written exam than figuring out when to jay-walk safely. Both of these are real-life situations, but we decide to jay-walk or not frequently while we only occasionally take an exam.
Meta-analysis Madness
In 2004 Weber and colleagues did a so-called meta-analysis of the experiments carried out on decision making of the kind where a choice has to be made between a sure thing and risky prospect. A meta-analysis is simply a piece of research which analyses the results of lots of similar studies – which sounds like a bit of a cheat but is actually incredibly important because such studies can sometimes generate enough data to extract important findings where none appear otherwise. For example, the finding that steriods significantly improve the chances of a premature baby surviving was pulled out of a meta-analysis, even though the underlying studies weren't anywhere near conclusive.
Anyway the study found 204 studies of two-condition decision making and found that every single one of them was based on a descriptive questionnaire. Now I don’t know about you, but when I’m about to cross the road in traffic I don’t expect a friendly researcher to walk up to me with a piece of paper explaining the probabilities of being flattened by an eight-wheeler in the middle lane. In everyday reality we need to figure these things out for ourselves by messy trial and error, so it rather looks as though a lot of these studies are looking at the wrong thing.
As we saw previously this doesn’t utterly undermine the findings of behavioural psychology, although there’s certainly a feeling that someone, somewhere is trying to light a fuse. Generally we can identify two differences between decision making from description and decision making from experience – firstly the different form of the question and secondly that the former is generally one-shot with limited feedback and the latter happens multiple times with pretty much instant feedback.
Sampling and Feedback
When Herwig and colleagues investigated this they figured out that it was the experience of making decisions and receiving feedback on the results that was the key. The design of the study allowed them to look at why this might be. Participants were allowed to do as many trials on each question as they wanted in order to figure out the correct result in any particular study yet the actual number of trials performed were relatively small – 15 on average – and because of this relative small sample set people generally under-estimated the likelihood of the rare event.
The reason for this is pretty obvious – if the rare event is, well, rare, then the fewer times you sample the information then the less likely you are to come across the rare event. The research showed clearly that those people who did experience the rare event were more likely to either avoid that option, if it elicited a negative outcome, or select that option, if it elicited a positive one. We have simple learning strategies underneath all of the complex social baggage.
Recency Redux
Unfortunately this gets worse, because of our old friend recency. Even those people who do experience the rare event get confused by more recent information. Again the research shows that those people who experienced the rare event late in the experiment seem to update their mental models to include it while those people who experience it early in the experiment, and not later on, seem to perform a similar update to ignore it.
All of this complex statistical stuff seems to boil down to some very simple and pretty much recognisable outcomes. In real-life situations where extreme events happen rarely people are likely to not experience the event very often and will therefore ignore the probability, even when it’s relatively high. In addition, even if they experience an extreme event early on recency effects may see this get overwritten by later information which doesn’t include it.
Investors in the School of Hard-Knocks
The relevance of this to stockmarket investors is immediately obvious. The markets are real-life learning experience in which most people tend to learn by trial and error. Generally extreme events, like companies failing altogether or markets collapsing like a pack of cards, aren’t experienced very often so the opportunity to learn from them is relatively rare. Unsurprisingly then, people’s expectations of the relative probability of these really bad things happening are generally much lower than the real probabilities.
All of which essentially means that if you’ve never experienced really bad stuff on the markets you’re almost certainly just very, very lucky. It also implies that the best way of learning about investing is from books rather than experience. Sadly books can’t teach you how to deal with the emotional rollercoaster of seeing your personal wealth ebbing and flowing with disconcerting regularity, over time.
The answer, such as it is, is that learning from experience and learning from description are both necessary to be a good investor. A few hard knocks along the way are no bad thing, as long as you don’t consign them to the bad luck bin. Stuff happens, frequently: give the markets enough chances and they’ll make sure it happens to you.
Related Articles: Fairy Tales for Investors, Investor Decisions - Experience is Not Enough, Recency: Hot Hands and the Gambler's Fallacy
It also implies that the best way of learning about investing is from books rather than experience.
ReplyDeleteThat's a great point. In most life endeavors, it's the other way around. So the realities of stock investing are always going to be counter-intuitive.
I think the answer is to make the teaching of these counter-intuitive realities the focus of all investment education. Just keep drilling them and the message over time sinks in. We have to accept that the realities really are counter-intuitive and it won't do just to mention them now and again.
Rob
"Sadly books can’t teach you how to deal with the emotional rollercoaster of seeing your personal wealth ebbing and flowing with disconcerting regularity, over time"
ReplyDeleteTaleb's "Fooled by Randomness" suggests planning your entry, planning your exit and not looking at the results until afterwards. Then you can modify your plans without emotional baggage.