The latest Journal of Prediction Markets has a lit review on overconfidence, with a to-me surprising result: financial overconfidence increases with age, experience, and success. Here are three investment experiments:
Although higher ability may lead to higher overconfidence, people with higher overconfidence are not necessarily have higher ability because there are factors other than ability that affect level of overconfidence.
After thinking about this for a while, I realized that it's completely obvious. Everyone is overconfident, and of course as people's abilities increase, their confidence increases even faster, and so they become even more overconfident. Look at Eliezer: he has a lot more ability than most people, and an immensely higher degree of overconfidence than most people.
Pablo: It seems to me that an implicit assumption in most economic thinking is that everyone already implicitly knows the one true economics except we nerds who don't know anything until we have verbalized it. This is obviously a gross exaggeration of reality, but there is some truth to it.
If overconfidence is positively correlated with ability, then observers can rationally take overconfidence as a signal of ability, and people can want to appear more overconfident to appear more able.
Robin, why do you assume that people know what you ignored until you read the study? Even if overconfidence is positively correlated with ability, observers can rationally take overconfidence as a signal of ability only if they are aware of this correlation. We should not impute to people beliefs about facts which took experts long time to uncover, unless we have independent evidence that people do know such facts, or can think of a mechanism which credibly explains how such facts are known by people.
"If overconfidence is positively correlated with ability, then observers can rationally take overconfidence as a signal of ability, and people can want to appear more overconfident to appear more able."
Wouldn't that be just "confidence"? I thought "overconfidence" meant confidence that wasn't justified.
@Marius: Maybe there is no skill to trading stocks, only luck. If that were the case, then the Dunning-Kruger effect wouldn't be allowed to operate, because there would be no difficult concepts that only an experienced trader could understand. Just a thought.
This reminds me of Nassim Taleb's "Fooled by Randomness", where he argues a lot of the stock market industry is based purely on luck, but that naive people, both participants and observers, imagine success or failure was based on the ability of the trader or analyst.
If that's true, then these results could be an example of survivorship bias. Successful, old, experienced traders and analysts are the ones who have spent many years without going bankrupt or getting fired - the ones who have gotten lucky. They probably attribute their success entirely to their own ability, and are confident in being able to continue having that success. But if their success was really caused by good luck, that confidence will be overconfidence.
I am under the impression that analogous results are well known in a variety of other contexts. For example, the first time you drive you are very cautious. As you drive a few times without any accidents, you become more confident that you will not get into an accident and so drive more recklessly. Success, or maybe more accurately lack or failure, gives a false impression of the robustness of continued success.
I remember a study of police officers, where they guessed whether suspects they had arrested were guilty or not. Confidence and inaccuracy (presumably measured against convictions) both increased with the number of years of experience.
When you're young, and careful, and successful, that gives you more confidence. The additional confidence stays with you, and as you get a little older, you're still careful, and successful. The ones who aren't successful, of course, even if they are careful, leave the field.
So what you're left with is the successful people who have seen others fail, giving them a sense of importance and power. And we know from a previous blog post that a sense of power leads to less time spent considering issues, overconfidence,etc.
It sounds like the best bet is to find someone who is not yet powerful, but has a track record of success up to this point.
"But to make this story work, somehow it should on average be easier to get away with being more overconfident when one is more experienced and successful. How can this be?"
I'm with Brian on this one. The callow and brash might feign or genuinely signal overconfidence, but people are likely to spot/call them out, either as posers or blow hards. The experienced and successful can buttress their overconfidence with the trappings and signals of their experience and success, giving most others reason to believe their confidence is well-founded (and maybe fanning the flames of the callow/brash who tell themselves they are innately as intelligent and capable as these experienced/successful types and thus justified in their own strutting).
Who are our financial experts? Rich people who made money in finance. But generally their experience has zero relevance for a current investors, because they made their fortunes playing games that are now extinct, just as Henry Ford's experience would be of almost no use today in designing cars. Soros played a currency game that was for a long time highly political and and corrupt, when finance ministers would deal in private information, and their bankers were trading on private information. Charles Schwab is a broker, a marketer, and knows little about markets. His small analogue, your wealthy local broker at Merril (or whatever) is also mainly a good at gab, and has no audited list of stock performance. You have deal makers like Felix Rohatyn who could make a lot of money because he knew lots of rich, powerful people (eg, Rahm Emmanuel made about $18MM in three years as an investment banker when he left the Clinton administration). Ken Fisher or Marty Zweig built mutual fund empires because they have high Q scores from prominent columns or TV appearances. In sum, the do not have any evidence of being able to predict the market, or relative performance of sectors, or individual stocks. They are smart, that's about all you can say.
So, overconfident because they have lots of information (info increases overconfidence), they have done well, and as successful marketers are good at selling. Add to that their success is probably orthogonal to any issue or idea they are currently opining upon, and it truly implies caveat emptor.
"But to make this story work, somehow it should on average be easier to get away with being more overconfident when one is more experienced and successful. How can this be?"
How could it not be? Experience and success already make it easier to get away with being wrong. Of course they make it easier to get away with being overconfident.
Although higher ability may lead to higher overconfidence, people with higher overconfidence are not necessarily have higher ability because there are factors other than ability that affect level of overconfidence.
After thinking about this for a while, I realized that it's completely obvious. Everyone is overconfident, and of course as people's abilities increase, their confidence increases even faster, and so they become even more overconfident. Look at Eliezer: he has a lot more ability than most people, and an immensely higher degree of overconfidence than most people.
Pablo: It seems to me that an implicit assumption in most economic thinking is that everyone already implicitly knows the one true economics except we nerds who don't know anything until we have verbalized it. This is obviously a gross exaggeration of reality, but there is some truth to it.
If overconfidence is positively correlated with ability, then observers can rationally take overconfidence as a signal of ability, and people can want to appear more overconfident to appear more able.
Robin, why do you assume that people know what you ignored until you read the study? Even if overconfidence is positively correlated with ability, observers can rationally take overconfidence as a signal of ability only if they are aware of this correlation. We should not impute to people beliefs about facts which took experts long time to uncover, unless we have independent evidence that people do know such facts, or can think of a mechanism which credibly explains how such facts are known by people.
Sorry, I was a little confused with this bit:
"If overconfidence is positively correlated with ability, then observers can rationally take overconfidence as a signal of ability, and people can want to appear more overconfident to appear more able."
Wouldn't that be just "confidence"? I thought "overconfidence" meant confidence that wasn't justified.
@Marius: Maybe there is no skill to trading stocks, only luck. If that were the case, then the Dunning-Kruger effect wouldn't be allowed to operate, because there would be no difficult concepts that only an experienced trader could understand. Just a thought.
Experince and success increase confidence, and confidence increases overconfidence? Or is that too intuitive?
This reminds me of Nassim Taleb's "Fooled by Randomness", where he argues a lot of the stock market industry is based purely on luck, but that naive people, both participants and observers, imagine success or failure was based on the ability of the trader or analyst.
If that's true, then these results could be an example of survivorship bias. Successful, old, experienced traders and analysts are the ones who have spent many years without going bankrupt or getting fired - the ones who have gotten lucky. They probably attribute their success entirely to their own ability, and are confident in being able to continue having that success. But if their success was really caused by good luck, that confidence will be overconfidence.
I am under the impression that analogous results are well known in a variety of other contexts. For example, the first time you drive you are very cautious. As you drive a few times without any accidents, you become more confident that you will not get into an accident and so drive more recklessly. Success, or maybe more accurately lack or failure, gives a false impression of the robustness of continued success.
I remember a study of police officers, where they guessed whether suspects they had arrested were guilty or not. Confidence and inaccuracy (presumably measured against convictions) both increased with the number of years of experience.
When you're young, and careful, and successful, that gives you more confidence. The additional confidence stays with you, and as you get a little older, you're still careful, and successful. The ones who aren't successful, of course, even if they are careful, leave the field.
So what you're left with is the successful people who have seen others fail, giving them a sense of importance and power. And we know from a previous blog post that a sense of power leads to less time spent considering issues, overconfidence,etc.
It sounds like the best bet is to find someone who is not yet powerful, but has a track record of success up to this point.
Doesn't this contradict the studies that established the Dunning-Kruger effect?
"But to make this story work, somehow it should on average be easier to get away with being more overconfident when one is more experienced and successful. How can this be?"
I'm with Brian on this one. The callow and brash might feign or genuinely signal overconfidence, but people are likely to spot/call them out, either as posers or blow hards. The experienced and successful can buttress their overconfidence with the trappings and signals of their experience and success, giving most others reason to believe their confidence is well-founded (and maybe fanning the flames of the callow/brash who tell themselves they are innately as intelligent and capable as these experienced/successful types and thus justified in their own strutting).
Who are our financial experts? Rich people who made money in finance. But generally their experience has zero relevance for a current investors, because they made their fortunes playing games that are now extinct, just as Henry Ford's experience would be of almost no use today in designing cars. Soros played a currency game that was for a long time highly political and and corrupt, when finance ministers would deal in private information, and their bankers were trading on private information. Charles Schwab is a broker, a marketer, and knows little about markets. His small analogue, your wealthy local broker at Merril (or whatever) is also mainly a good at gab, and has no audited list of stock performance. You have deal makers like Felix Rohatyn who could make a lot of money because he knew lots of rich, powerful people (eg, Rahm Emmanuel made about $18MM in three years as an investment banker when he left the Clinton administration). Ken Fisher or Marty Zweig built mutual fund empires because they have high Q scores from prominent columns or TV appearances. In sum, the do not have any evidence of being able to predict the market, or relative performance of sectors, or individual stocks. They are smart, that's about all you can say.
So, overconfident because they have lots of information (info increases overconfidence), they have done well, and as successful marketers are good at selling. Add to that their success is probably orthogonal to any issue or idea they are currently opining upon, and it truly implies caveat emptor.
"But to make this story work, somehow it should on average be easier to get away with being more overconfident when one is more experienced and successful. How can this be?"
How could it not be? Experience and success already make it easier to get away with being wrong. Of course they make it easier to get away with being overconfident.