Behavioural Finance
Summary of the Lectures
University of Amsterdam
MSc Finance
2022
,Week 1
Part 1: Introduction to Behavioral Finance
Behavioral Finance in a Nutshell
Classical Finance Theory
● assumes that economic agents are fully rational (in preferences and beliefs)
● implies that markets are “efficient”, i.e. assets are always correctly priced
Behavioral Finance Theory
● acknowledges that people can be irrational (have inconsistent preferences and irrational
beliefs)
● implies that assets can be mispriced: existence of bubbles possible
The Rational Model
In words, classical economics assumes that individuals
● have rational (=objective=correct) beliefs (p(st) part of equation)
● are able to process all available information immediately
● are Bayesian information processors
● have standard (CARA) preferences over final *levels* of outcomes (u part of equation)
● maximize their expected utility (max part of equation)
● are self-interested, narrowly defined (u part of equation)
What is Behavioral Finance?
The above model is too complicated and unrealistic. Behavioral Finance seeks to improve realism by
acknowledging deviations from these assumptions, in particular:
● biased beliefs (optimism and overconfidence), permanent
● biased updating of beliefs (overreaction, underreaction), transitory
● limits to (the speed of) information processing.
● non-standard preferences (like loss aversion and reference dependence).
(The above formula in words: In classic economic theory, individuals are assumed to make decisions
today (and at any point in the future) so as to maximize the present value of their expected lifetime
utility.)
1
,Why Should We Care?
It is now widely acknowledged that the rational model fails to capture many important economic
phenomena. Examples are:
● Bubbles and crashes in asset markets
● Asset pricing anomalies (momentum, value)
● High trading volume, speculative trading
● High volatility of prices (relative to fundamentals)
● Overpricing in IPOs (eg Facebook, Rivian, etc.)
● Failure of many mergers to deliver promised benefits
Why Should We Care?
● Understanding mispricing is key for investment managers
● Important macroeconomic consequences of large price distortions:
○ misallocation of investment
○ high volatility of GDP, reversal of the “Great Moderation”
○ distributional issues: winners and losers
● Policy implications: Should we
○ tax or restrict speculative trading?
○ ban short selling?
○ restrict levered investments? Cf. Dutch housing market?
○ outlaw certain assets (e.g. bitcoin)?
○ regulate banks more tightly?
U.S. equities: index level vs earnings
Above you can see a disconnect between the real index level and real earnings, which is not rational.
2
,U.S. equities: price-earnings ratio
Above you can see an increase in PE in recent years.
Bond yields
Bond prices are high in the last few years, meaning that the yields are low.
U.S. House Price Index
This concludes that in recent years all assets have extremely high prices, making it harder to find a
good investment.
3
,Historic Bubbles in Comparison
This graph is based on the Bitcoin peak of December 2017. In 2021, Bitcoin reached more than three
times the 2017 peak price. Shows how important, but also how hard it is, to call bubbles!
Step 1: Non-Standard Preferences
● We will first discuss“standard” preferences and then “non-standard” preferences
● Keyword: Prospect Theory (Kahneman and Tversky, 1979, 1992)
(Classic) Utility
● Utility: measure of (relative) satisfaction
● Bernoulli (1738): Utility is not linear in monetary payoffs “The same amount of additional
money is less useful to an already-wealthy person than it would be to a poor person”
● You can formalize this idea with a concave utility function:
(Classic) Expected Utility
Individuals are assumed to make decisions by maximizing expected utility:
4
,Expected Utility and Risk Aversion
A concave (left) utility function means people are risk averse, convex (right) means risk seeking
Classic or Standard Utility: Summary
● Individuals have preferences (utility functions) over levels of wealth or consumption
● They are typically risk-averse
● They have stable preferences (relatively stable over time, and not dependent on the framing of
the situation).
Prospect Theory (Kahneman and Tversky)
In reality, when presented with risky choices, people do not behave as if they maximized classic
expected utility…
Differences, Not Levels, Matter
People typically respond to changes of circumstances much more than to levels.
Sensory:
● Brightness
● Loudness
● Temperature
Non-sensory:
● Health
● Salary
● Wealth
● Course grades
Kahneman & Tversky: Why should financial outcomes not be evaluated based on changes relative to
some reference point?
5
,Preference Reversals
People typically have different preferences over (perceived) gains than over losses:
● Risk averse over gains
● Risk seeking over losses
Loss Aversion
“Losses loom larger than gains”. Loss aversion: displeasure associated with losing a sum of money is
greater than pleasure of winning the same amount.
Above you can see that the value function “steeper” over losses than over gains.
6
,Implication: Disposition Effect
● Hypothesis: Investors tend to hold losing investments too long and sell winning investments
too soon.
● Study: Odean (1998) “Are Investors Reluctant to Realize Their Losses?”
● Data:
○ Discount retail brokerage data set (1987-’93, 10,000 accounts)
○ Results replicated on more recent, and larger, datasets
● They calculated realized gains and paper gains (unrealized gains). They calculated how much
of the stocks in the green were sold and how much of the stocks in the red have been sold.
● If the disposition effect were true, then the stocks in the green would have been sold more
than the stocks in the red. This would mean asymmetry.
● So the PGR would be higher than the PLR, as people hold onto losing investments longer.
You can see that this is the case.
What is the “Right” Reference Point?
Often, the reference point is the price at which you bought something. You don’t want to sell at a loss
● your stocks
● your house
● your firm
But sometimes reference points are even more random
● past peak prices
● year-beginning, year-end prices
7
,Step 2: Non-Standard Beliefs
Recall the standard model (simplified)
Evidence from Psychology
Psychology: Evidence on overconfidence/overoptimism
● Svenson (1981): 93 percent of subjects rated their driving skill as above median, compared to
the other subjects in the experiment
● Weinstein (1980): Most individuals underestimate the probability of negative events such as
hospitalization
● Buehler-Griffin-Ross (1994): Underestimate time needed to finish a project
Evidence in Finance
Ben-David, Graham and Harvey (2013). Ask CFOs to give their estimate for the S&P500 1-year
● 10% return percentile
● 90% return percentile
the 80% confidence interval can be used to measure overconfidence:
● Compute individual perceived volatility:
CFO Miscalibration: Predicted vs Actual Volatility
● 58% of imputed individual volatilities are lower than 5.1%, the lowest value realized in the
past 60 years.
● Only 3% of imputed individual volatilities are greater than 20%, a value reached in more than
30% of all quarters during 2001-2011.
8
, So CFOs predicted a lot less volatility than they should do, because of overconfidence/overoptimism.
Implication: Excessive Trading
You can see this in the illustration below:
9
Summary of the Lectures
University of Amsterdam
MSc Finance
2022
,Week 1
Part 1: Introduction to Behavioral Finance
Behavioral Finance in a Nutshell
Classical Finance Theory
● assumes that economic agents are fully rational (in preferences and beliefs)
● implies that markets are “efficient”, i.e. assets are always correctly priced
Behavioral Finance Theory
● acknowledges that people can be irrational (have inconsistent preferences and irrational
beliefs)
● implies that assets can be mispriced: existence of bubbles possible
The Rational Model
In words, classical economics assumes that individuals
● have rational (=objective=correct) beliefs (p(st) part of equation)
● are able to process all available information immediately
● are Bayesian information processors
● have standard (CARA) preferences over final *levels* of outcomes (u part of equation)
● maximize their expected utility (max part of equation)
● are self-interested, narrowly defined (u part of equation)
What is Behavioral Finance?
The above model is too complicated and unrealistic. Behavioral Finance seeks to improve realism by
acknowledging deviations from these assumptions, in particular:
● biased beliefs (optimism and overconfidence), permanent
● biased updating of beliefs (overreaction, underreaction), transitory
● limits to (the speed of) information processing.
● non-standard preferences (like loss aversion and reference dependence).
(The above formula in words: In classic economic theory, individuals are assumed to make decisions
today (and at any point in the future) so as to maximize the present value of their expected lifetime
utility.)
1
,Why Should We Care?
It is now widely acknowledged that the rational model fails to capture many important economic
phenomena. Examples are:
● Bubbles and crashes in asset markets
● Asset pricing anomalies (momentum, value)
● High trading volume, speculative trading
● High volatility of prices (relative to fundamentals)
● Overpricing in IPOs (eg Facebook, Rivian, etc.)
● Failure of many mergers to deliver promised benefits
Why Should We Care?
● Understanding mispricing is key for investment managers
● Important macroeconomic consequences of large price distortions:
○ misallocation of investment
○ high volatility of GDP, reversal of the “Great Moderation”
○ distributional issues: winners and losers
● Policy implications: Should we
○ tax or restrict speculative trading?
○ ban short selling?
○ restrict levered investments? Cf. Dutch housing market?
○ outlaw certain assets (e.g. bitcoin)?
○ regulate banks more tightly?
U.S. equities: index level vs earnings
Above you can see a disconnect between the real index level and real earnings, which is not rational.
2
,U.S. equities: price-earnings ratio
Above you can see an increase in PE in recent years.
Bond yields
Bond prices are high in the last few years, meaning that the yields are low.
U.S. House Price Index
This concludes that in recent years all assets have extremely high prices, making it harder to find a
good investment.
3
,Historic Bubbles in Comparison
This graph is based on the Bitcoin peak of December 2017. In 2021, Bitcoin reached more than three
times the 2017 peak price. Shows how important, but also how hard it is, to call bubbles!
Step 1: Non-Standard Preferences
● We will first discuss“standard” preferences and then “non-standard” preferences
● Keyword: Prospect Theory (Kahneman and Tversky, 1979, 1992)
(Classic) Utility
● Utility: measure of (relative) satisfaction
● Bernoulli (1738): Utility is not linear in monetary payoffs “The same amount of additional
money is less useful to an already-wealthy person than it would be to a poor person”
● You can formalize this idea with a concave utility function:
(Classic) Expected Utility
Individuals are assumed to make decisions by maximizing expected utility:
4
,Expected Utility and Risk Aversion
A concave (left) utility function means people are risk averse, convex (right) means risk seeking
Classic or Standard Utility: Summary
● Individuals have preferences (utility functions) over levels of wealth or consumption
● They are typically risk-averse
● They have stable preferences (relatively stable over time, and not dependent on the framing of
the situation).
Prospect Theory (Kahneman and Tversky)
In reality, when presented with risky choices, people do not behave as if they maximized classic
expected utility…
Differences, Not Levels, Matter
People typically respond to changes of circumstances much more than to levels.
Sensory:
● Brightness
● Loudness
● Temperature
Non-sensory:
● Health
● Salary
● Wealth
● Course grades
Kahneman & Tversky: Why should financial outcomes not be evaluated based on changes relative to
some reference point?
5
,Preference Reversals
People typically have different preferences over (perceived) gains than over losses:
● Risk averse over gains
● Risk seeking over losses
Loss Aversion
“Losses loom larger than gains”. Loss aversion: displeasure associated with losing a sum of money is
greater than pleasure of winning the same amount.
Above you can see that the value function “steeper” over losses than over gains.
6
,Implication: Disposition Effect
● Hypothesis: Investors tend to hold losing investments too long and sell winning investments
too soon.
● Study: Odean (1998) “Are Investors Reluctant to Realize Their Losses?”
● Data:
○ Discount retail brokerage data set (1987-’93, 10,000 accounts)
○ Results replicated on more recent, and larger, datasets
● They calculated realized gains and paper gains (unrealized gains). They calculated how much
of the stocks in the green were sold and how much of the stocks in the red have been sold.
● If the disposition effect were true, then the stocks in the green would have been sold more
than the stocks in the red. This would mean asymmetry.
● So the PGR would be higher than the PLR, as people hold onto losing investments longer.
You can see that this is the case.
What is the “Right” Reference Point?
Often, the reference point is the price at which you bought something. You don’t want to sell at a loss
● your stocks
● your house
● your firm
But sometimes reference points are even more random
● past peak prices
● year-beginning, year-end prices
7
,Step 2: Non-Standard Beliefs
Recall the standard model (simplified)
Evidence from Psychology
Psychology: Evidence on overconfidence/overoptimism
● Svenson (1981): 93 percent of subjects rated their driving skill as above median, compared to
the other subjects in the experiment
● Weinstein (1980): Most individuals underestimate the probability of negative events such as
hospitalization
● Buehler-Griffin-Ross (1994): Underestimate time needed to finish a project
Evidence in Finance
Ben-David, Graham and Harvey (2013). Ask CFOs to give their estimate for the S&P500 1-year
● 10% return percentile
● 90% return percentile
the 80% confidence interval can be used to measure overconfidence:
● Compute individual perceived volatility:
CFO Miscalibration: Predicted vs Actual Volatility
● 58% of imputed individual volatilities are lower than 5.1%, the lowest value realized in the
past 60 years.
● Only 3% of imputed individual volatilities are greater than 20%, a value reached in more than
30% of all quarters during 2001-2011.
8
, So CFOs predicted a lot less volatility than they should do, because of overconfidence/overoptimism.
Implication: Excessive Trading
You can see this in the illustration below:
9
