Nature, Nurture, and Financial
Decision-Making
Why Do Individuals Exhibit
Investment Biases?
Henrik Cronqvist
Claremont McKenna College
Stephan Siegel
University of Washington
Arizona State University
University of
Michigan-Dearborn
Betty F. Elliott Initiative for
Academic Excellence:
Financial Literacy
April 9, 2012
Nature, Nurture, and Financial DecisionMaking

The Origins and Remediation of Human Inequality
(James Heckman)

This research is rooted in economics but goes well outside of traditional
analyses to integrate research in psychology, demography, neuroscience
and biology.
“Economics is a branch of biology broadly interpreted.”

A complete theory of human behavior (Andy Lo (2010))
Alfred
(1920)

Can Marshall
we develop
a complete
theory of human behavior that is predictive in all
contexts?

Effectiveness of policy initiatives (Doug Bernheim (2009))
“The discovery of a patience gene could shed light on the extent to which
correlations between the wealth of parents and their children reflect
predispositions rather than environmental factors that are presumably
more amenable to policy intervention”
Genoeconomics*
• Study the sources of variation in economic behaviors and
outcomes
• Understand how institutions or environments moderate or
amplify genetic differences
• Education lowers genetic variation of health outcomes
• Teacher quality increases genetic variation of reading skills
• Identify specific genes that predict behaviors/outcomes.
Design interventions for those at “genetic” risk
• Reduce omitted variable bias by including genetic markers
* Benjamin, Chabris, Glaeser, Gudnason, Harris, Laibson, Launer, and Purcell (2007)
Research Methods
Sample Studies
• Twin and Adoption Studies
•
•
•
•
•
Behrman and Taubman (1976)
Ashenfelter and Krueger (1994)
Sacerdote (2002)
Bjoerklund et al. (2006)
Cesarini et al. (2009, 2010)
• Molecular Genetics Studies
• Candidate gene studies
• Genome-wide association
studies (GWAS)
• Rotterdam Study
• Framingham Heart Study
• AGES/Reykjavik Study
• Kuhnen and Chiao (2009)
• Dreber et al. (2009)
• Beauchamp et al. (2011)
• Glaeser, Laibson, et al.
(ongoing)
Our Contributions, so far . . .
Large Scale Twin Studies using Swedish Data (1999 – 2007)
• Risk Aversion and Financial Risk Taking
(Barnea, Cronqvist, and Siegel (2010))
• Discount Rates, Impatience, and Wealth Accumulation
(Cronqvist and Siegel (2011))
• Preferences over Homeownership and Home Location
(Cronqvist, Muenkel, and Siegel (2012))
Investment Biases
The investor’s chief problem and even
his worst enemy is likely to be himself.
Benjamin Graham
Investment Biases
Long list of investment behaviors that cannot be
explained by standard preferences and belief formation






Underdiversify
Prefer local securities
Avoid realizing losses
Chase past performance
Trade a lot
Prefer lottery-type stocks
Behaviors have been shown to be:



Wide-spread, even present among professional traders/investors
Potentially costly
Generally linked to fundamental psychology construct
But, degrees vary across investors
Why Do Individuals Exhibit Investment
Biases?
 Born
with biases
 Preferences and belief formation as outcome of natural selection
 Jack Hirshleifer (1977), Becker (1976)
 Robson (1996, 2001), Netzer (2009), Robson and Samuelson (2009)
 Nature
selects behaviors that maximize fitness
 Depending on environment, biases can emerge



Loss Aversion: McDermott, Fowler, Smirnov (2008) [in biology: e.g. Caraco (1980)]
Over-confidence: Johnson and Fowler (2011)
Probability Matching: Brennan and Lo (2009)
 Environmental
conditions
 Parenting
 Information
 Institutions
 Incentives
and education
Objectives
 Quantify
 Models
the importance of different sources
of natural selection require some genetic variation
 Understand
whether education, experience, or incentives
affect the importance of different sources
 In
particular, what conditions moderate genetic predispositions
Improve policy design
Invest in gene and genome wide association studies
Existing Evidence

Capuchin monkeys exhibit loss aversion

Capuchin monkeys prefer gambles with good outcomes framed as bonuses over
identical pay-off gambles with bad outcomes framed as losses
 Loss aversion is part of decision-making process that evolved before humans and
capuchins separated (Chen et al. (2006), Lakshminarayanan et al. (2011))

Experimental and survey evidence
Twin Studies
 Overconfidence: Cesarini et al. (2009)
 Conflicting evidence for several biases: Cesarini et al. (2011) and Simonson and Sela
(2011)
Gene Association Studies
 Different genes associated with risk attitudes over gains and losses (Zhong et al. (2010))
Neuro-scientific Studies
 Different brain activity for realized vs. unrealized gains (Frydman, Barberis, Camerer,
Bossaerts, and Rangel (2011))

No empirical evidence based on “real world” financial decisions
Our Research Methodology
Identical Twins
Fraternal Twins
Elin and Josefin
Nordegren
Mary Kate and Ashley
Olson
Intuition of Methodology
Use identical & fraternal twins to decompose variation:

Identical twins have 100% of their genes in common

Fraternal twins on average have 50% of their genes in common

Twins who grew up in same family have a common environment

Each twin has his or her individual-specific environment
If genes matter, then identical twins should be more
similar than fraternal twins in terms of their behavior.
Methodology

Random effect model with genetic effect a, common effect
c and individual-specific effect e:

Covariance structure implied by genetic theory:
MZ
DZ
Methodology, cont’d

Estimate parameters σ2a, σ2c, and σ2e via maximum likelihood
estimation (MLE) with bootstrapped standard errors

Derive the variance components:
A-share – genetic component:
 a2
 a2   c2   2
C-share – common environment
(parenting):
 c2
 a2   c2   2
E-share – individual environment
& measurement error:
 2
 a2   c2   2
Data

Twins from the Swedish Twin Registry.

Matched with annual financial data (including holdings of assets and
sales transactions) and socioeconomic data from Statistics Sweden
(1999 – 2007, no transactions in 2001/02)

Require:

At least 18 years old
 Both twins hold some equities (directly or indirectly) in one year
 Average all variables over the years that individual is in sample
All Twins
Identical Twins
Male
Female
Total
Same
Sex:
Male
Fraternal Twins
Same
Sex:
Opposite
Female
Sex
Total
Number of twins (N )
30,416
4,066
5,206
9,272
4,522
5,326
11,296
21,144
Fraction (%)
100%
13%
17%
30%
15%
18%
37%
70%
Socioeconomic Characteristics and
Equity Portfolio Characteristics
DETAIL
Variable
Age
Less than High School
High School
College or more
No Education Data available
Years of Education
Married
Disposable Income (USD)
Financial Assets (USD)
Total Assets (USD)
Total Debt (USD)
Net Worth (USD)
Number of Stocks and Equity Mutual
Value of Stocks and Equity Mutual
Number of Stocks
Value of Stocks (USD)
Number of Equity Mutual Funds
Value of Equity Mutual Funds (USD)
All Twins
N
Mean
30,416
30,416
30,416
30,416
30,416
17,395
30,416
30,416
30,416
30,416
30,416
30,416
30,416
30,416
12,378
12,378
23,870
23,870
47.08
0.15
0.22
0.58
0.06
11.22
0.46
31,379
40,759
124,351
31,802
92,549
3.56
16,841
3.32
22,558
2.41
7,018
Identical Twins
Std. Dev.
Median
48.00
0.00
0.00
1.00
0.00
11.00
0.00
25,476
14,537
71,883
16,020
42,961
2.33
3,662
1.89
2,825
1.89
2,059
17.64
0.35
0.41
0.49
0.23
3.26
0.50
27,592
155,296
252,478
68,330
223,277
3.80
109,292
3.91
163,360
1.84
20,160
Mean
53.06
0.20
0.26
0.47
0.06
11.11
0.54
35,203
48,062
142,603
30,396
112,207
3.62
24,815
3.42
29,218
2.34
7,788
Fraternal Twins
Std. Dev.
Median
55.00
0.00
0.00
0.00
0.00
11.00
1.00
27,678
17,342
83,504
13,759
56,417
2.25
4,159
1.89
2,819
1.80
2,292
15.51
0.40
0.44
0.50
0.24
3.29
0.50
35,449
442,298
576,198
149,778
516,665
3.97
663,773
4.15
543,596
1.86
17,304
Measuring Investment Biases

Home Bias
Proportion of equity portfolio held in Swedish equity

Disposition Effect
Conceptually: PGR – PLR (Odean (1998), Dhar and Zhu (2006), Campbell et al. (2009))
Based raw return in years with at least one sales transaction

Turnover
Annual sales volume (SEK) divided by value of portfolio at beginning of year.

Performance Chasing
Fractions of stocks acquired with raw returns in top two deciles

Skewness Preference
Fraction of lottery stocks in portfolio (Kumar (2009))
Investment Biases: Summary Statistics
All Twins
N
Stocks
Home Bias
Turnover
Disposition Effect
Performance Chasing
Skewness Preference
12,378
11,508
2,268
6,672
12,378
Identical Twins
Mean
Median
Std. Dev.
0.94
0.20
0.05
0.15
0.04
1.00
0.03
0.03
0.00
0.00
0.16
0.35
0.41
0.22
0.10
Fraternal Twins
Mean
Median
Std. Dev.
0.94
0.17
0.07
0.14
0.03
1.00
0.02
0.03
0.00
0.00
0.15
0.33
0.41
0.22
0.10
Evidence from Correlations (Stocks)
Figure 1
Correlations by Genetic Similarity
Identical Twins
Fraternal Twins - Opposite Sex
Fraternal Twins
Random Match
Fraternal Twins - Same Sex
0.55
0.45
0.35
0.25
0.15
0.05
-0.05
-0.15
Home
Bias
Turnover
Disposition
Effect
Performance
Chasing
Skewness
Preference
Variance Decomposition: Stocks
Home
Bias
Turnover
Intercept
Male
Age
Age - squared
High School
College or More
No Education Data Available
Married
Second Net Worth Quartile Indicator
Third Net Worth Quartile Indicator
Highest Net Worth Quartile Indicator
Log of Disposable Income
Number of Trades (Sales)
Number of Holdings
0.955
0.004
0.004
0.000
-0.001
-0.012
-0.026
-0.001
-0.001
0.001
-0.010
-0.001
0.134
0.062
0.031
-0.005
0.000
0.022
0.037
-0.001
-0.005
-0.011
-0.025
-0.002
0.132
-0.007
0.011
0.000
-0.010
-0.032
-0.025
-0.054
-0.056
-0.006
-0.007
-0.004
0.003
-0.003
2.313
0.062
0.092
-0.008
-0.117
-0.156
-0.002
-0.051
0.122
0.200
0.294
0.117
0.004
0.008
0.015
-0.002
0.001
0.005
0.010
0.002
0.003
-0.002
-0.004
0.000
A Share
0.453
0.257
0.297
0.311
0.281
0.052
0.029
0.077
0.090
0.051
0.000
0.000
0.000
0.096
0.000
0.027
0.008
0.041
0.065
0.028
0.547
0.743
0.703
0.593
0.719
0.037
0.027
0.052
0.038
0.034
R2
0.010
0.014
0.020
0.009
0.000
N
12,378
11,508
2,268
6,672
12,378
C Share
E Share
Disposition Performance Skewness
Effect
Chasing
Preference
Mutual Fund and Large Investors
Repeat analysis combining direct stockholdings and mutual
fund investments
 Results are essentially the same
 Diversification measure (mutual fund / all risky financial assets)
has A component of about 39%
Repeat analysis for investors that hold at least 20% of assets
in risky financial assets
 Genetic component increases by typically 10 to 20% points
Robustness

Same sex twin only

Model Misspecification



Allowing for negative variance components
Nonlinear models
Communication between twins





Identical twins communicate more with one another
Financial decisions are influenced by communication (e.g. Shiller and
Pound (1998), Hong, Kubik, and Stein (2004))
Sort pairs into 10 communication intensity bins and randomly drop
identical/fraternal pairs until both types are equally often present per
bin.
Estimate model across all 10 bins
A component slightly reduced, but generally robust
Moderating Genetic Predisposition

Evidence that experience, education, and wealth affect
investment biases and trading behavior (e.g. Vissing-Jorgensen (2003),
Dhar and Zhu (2006), Calvet et al. (2009), Graham et al. (2009))

Environment can enhance or constrain genetic predisposition

Heritability of reading ability increases with quality of teacher
(Taylor et al. (2010))

Education seems to reduce genetic variation of health status
(Johnson et al. (2009))

Examine (for a sub sample) how years of education interact with
different sources of variation

We find no significant effect of years of education on size of
genetic variance
G x E Interaction in Presence of G x E
Correlation
Figure 2
Gene-Environment Interaction
AM
CM
EM
cc+ cc M
AU
ec+ c M
aU+ aU M
CU
cU+ cU M
aM cM eM
M
EU
eU+ U M
ac+ ac M
B
Years of Education
Bias/Behavior
Moderator: Years of Education
Turnover
Home Bias
0.03
0.12
0.025
0.1
0.02
0.08
Var(A)
0.015
Var(C)
Var(E)
0.01
0.005
Var(A)
0.06
Var(C)
Var(E)
0.04
0.02
0
0
8
10
12
14
16
8
Loss Aversion
10
12
14
16
Performance Chasing
0.06
0.25
0.05
0.2
0.04
0.15
Var(A)
Var(C)
0.1
Var(E)
0.05
Var(A)
0.03
Var(C)
Var(E)
0.02
0.01
0
0
8
10
12
14
16
8
10
12
14
16
Twins with Financial Experience in their Jobs
Variance Components
Model
Diversification
Home Bias
Turnover
Performance Chasing
Skewness Preference
N
A - Share
C - Share
E - Share
622
0.000
0.222
0.778
0.104
0.090
0.069
0.000
0.206
0.794
0.088
0.082
0.073
0.000
0.110
0.890
0.106
0.067
0.088
0.026
0.106
0.868
0.102
0.068
0.078
0.187
0.000
0.813
0.091
0.042
0.079
622
582
562
622
Sources of Behavioral Consistency

Behavior across different domains is often correlated
If genetic factors matter, source of the correlation should be genetic

Correlate Home Bias with


Distance to birthplace
 Indicator whether spouse is from same home state
Model I
A - Share
C - Share
E - Share
Model II
Home
Bias
Distance to
Birthplace
Home
Bias
Spouse from
Home Region
0.455
0.400
0.364
0.146
0.059
0.085
0.116
0.092
0.000
0.210
0.000
0.192
0.039
0.061
0.066
0.067
0.545
0.389
0.636
0.662
0.031
0.036
0.081
0.041
Sources of Behavioral Consistency

Behavior across different domains is often correlated
If genetic factors matter, source of the correlation should be genetic

Correlate Home Bias with


Distance to birthplace
 Indicator whether spouse is from same home state
Model I
Home
Bias
Correlation
Distance to
Birthplace
Model II
Home
Bias
Spouse from
Home Region
-0.031
0.010
0.009
0.022
-0.106
0.240
0.036
0.239
Correlation of Common Environment
0.000
0.000
Correlation of Individual Environment
0.031
-0.069
0.021
0.035
12,180
2,566
Genetic Correlation
N
Conclusions

Why do investor exhibit investment biases? We show that to a
large existent biases, such as Home Bias, Disposition Effect,
Turnover, Performance Chasing, as well as Skewness Preference
are innate

Our findings are consistent with recent theoretical models that argue
that biases are the outcome of natural selection

While genetic effects are important, they are not destiny:

A large part of the cross-sectional variation appears to be related to
individual experiences and circumstances
 General educational achievement does not seem to moderate genetic
predispositions
 For twins with occupational experience in finance genetic factors seem to
matter less
END