Portfolio Selection in Utero Anna Dreber† and Moshe Hoffman‡ July 20th 2007 †

Department of Economics, Stockholm School of Economics, and Program for Evolutionary Dynamics, Harvard University ‡ University of Chicago, Graduate School of Business

Abstract Most decisions are made under conditions of uncertainty. Thus, risk preferences are important to decisions, ranging from career choice to stock picking (Barsky et al. 1997). Consequently, risk preferences are one of the most crucial parameters in economics and every day life. Nevertheless, little is known about the determinants. Using a laboratory measure previously validated by economists (Gneezy and Potters 1997), we show that risk aversion positively correlates with 2D:4D – the ratio between the length of the 2nd (index) finger and the 4th (ring) finger. This is a biological measure that is thought to positively correlate with prenatal estrogen and negatively correlate with prenatal testosterone and is fixed early in life (Manning et al. 1998). Our findings are the first to indicate that an important economic preference is partially predetermined. These results also suggest that prenatal hormones influence risk preferences 20 odd years later. Thus, before the language one speaks is determined, one’s stock portfolio is already taking shape.

Intro

2D:4D is the ratio between the length of the 2nd (index) finger and the 4th (ring) finger. 2D:4D positively correlates with prenatal estrogen and negatively correlates with prenatal testosterone exposure (Manning et al. 1998).There is suggestive but not conclusive evidence for the role of 2D:4D as a biomarker for the organizational i.e. permanent effects of prenatal androgens on the brain and behavior. 2D:4D has been shown to be largely genetically determined (Paul et al. 2006). 2D:4D is presumed to be affected by the environment, while the fetus is in the womb, although, again, this has not been fully demonstrated. Critically for our interpretation, 2D:4D is constant over an individuals’ life (McIntyre 2006), and doesn’t appear to be related to any observable characteristics such as height (Manning et al. 1998).

1

Men tend to have lower ratios than women. Additionally, within each gender, lower 2D:4D predicts a number of fitness-related factors that tend to show pronounced gender differences, such as more competitiveness in sports and better performance in a mental rotation task (Manning and Taylor 2001). Risk preferences have also been shown to differ by gender, with women being more risk averse (Croson and Gneezy 2004, Byrnes et al. 1999). We add to the existing literature on 2D:4D and to the existing literature on risk preferences by testing the hypothesis that 2D:4D is positively correlated with risk aversion, even while controlling for gender. We are the first, as far as we know, to show that a direct measure of risk preferences, and a direct measure of preferences more generally, are partially predetermined and correlate with prenatal hormones.

Two previous studies relate 2D:4D to economics. Van den Bergh and Dewitte, (2006) find that men with lower digit ratios are more likely to reject an unfair split in neutral contexts, but are more likely to accept unfair offers in sex-related contexts, in an ultimatum game. Millet and Dewitte (2006) find that men and women with lower digit ratios gave exactly their fair share (a fair share is the provision point of the public good divided by the number of players), whereas those with higher digit ratios contributed either more or less than this in a modified version of the public good game. Several other studies relate biological variables and economics. Exogenously introduced oxytocin has been shown to increase the level of giving in the trust game (Kosfeld et al. 2005). Menstruation was shown to increase bids in a first price auction but not a second price auction (Chen et al. 2005). Wallace et al. (2007) study monozygotic

2

and dizygotic twins and find that additive genetic effects explain over 40% of the variation in subjects' threshold value as receivers in the ultimatum game. Burnham (2007) finds that high and low testosterone men give similar offers in the ultimatum game, but that high testosterone men are more likely to reject selfish offers.

We measure subject’s 2D: 4D (Fig. 1) for both the right and the left hand after they have made a decision involving a risky investment, with real monetary payoffs1. The participant is given a balance of SEK 1700 (approx. $250) and is asked to choose an amount X between 0 and 1700 that they wish to allocate to the risky investment. They are told that a coin will be tossed to determine the success or failure of the investment. In case of failure, the money invested is lost, and they earn SEK 1700-X. In case of success, the money invested is multiplied by 2.5, and they earn SEK 1700+1.5X. One of the subjects is randomly drawn and is paid according to the choices he or she made. Because investing is risky but offers higher returns, subjects must weigh the expected returns against the risk involved. We thus use X as our measure of risk preference.

Method A total of 147 students from the Stockholm School of Economics (55 women and 92 men) participated in the study. 125 of these participants were Caucasian (45 women, 80 men). Since there are ethnic differences in risk preferences (Manning et al. 2004), and we only have few non-Caucasian subjects, the results we report are based on the

1

Previous literature often find one hand to be more significant than the other (McIntyre 2006); why, however, has not been established.

3

Caucasian sample only. However, if we include all subjects and control for ethnicity the results don’t change.

Participants were asked to put both their hands on a scanner, and one scan was taken per participant. Digits were then measured from the scanned images with the Adobe Photoshop tool. However, 15 digit ratios could not be used due to incomplete scans or unclear creases. All subjects had both their hands measured. In addition to the chance of winning their balance, subjects were given SEK 30 for participating.

Results X can take a value between SEK 0 and SEK 1700. In our sample, the mean of X is 1063 (n=125) and the standard deviation is 520. Left hand 2D:4D (n=120) has a mean of 0.962 and a standard deviation of 0.034. Right hand 2D:4D (n=116) has a mean of 0.956 and a standard deviation 0.036. All results are analyzed using OLS.

In line with previous research on 2D:4D (McIntyre 2006) we find that women have significantly higher 2D:4D than men (left hand: t = 2.07, P = 0.041). Also in line with previous research on risk preferences (Croson and Gneezy 2004, Byrnes et al. 1999), we find that women are significantly more risk averse than men (t = -3.79, P < 0.001).

4

We hypothesized that lower 2D:4D would correspond with higher X. Our data confirm this (Table 1). Higher left hand 2D:4D significantly predicts more risk aversion (t = -2.75, P = 0.007), as does right hand 2D:4D (t = -2.15, P = 0.034)2. When controlling for gender, higher left hand 2D:4D still significantly predicts more risk aversion (t = -2.27, P = 0.025). Right hand 2D:4D is marginally significant (t = -1.71, P = 0.091). This effect is also economically significant; a person with a left hand 2D:4D one standard deviation above average invests SEK 91 less than a person with average 2D:4D, holding gender constant. This amount is sizeable; it is about a third the magnitude of the gender effect which, in turn, is large. There is no evidence that the effect of 2D:4D is different for men and women (interaction variable gender and 2D:4D: t = 0.28, P = 0.781).

Conclusion We conclude that prenatal hormones and 2D:4D are correlated with risk preferences. This conclusion has large practical implications for behavioral scientists; 2D:4D can be used as an instrument to study the effect of risk preferences on other choices, where endogeneity might be an issue, such as marriage and career choices; since 2D:4D is fixed early in life, it seems like the ideal instrument. Additionally, this conclusion offers a potential mediator for many of the previous findings of behaviors related to 2D:4D, as the dependent variable is often associated with risk preferences.

2

One should keep in mind that our results are likely underestimated by the fact that we are using a proxy for prenatal hormones and not prenatal hormones themselves.

5

Finally, we conclude that risk preferences, and preferences more generally, are partially determined early in life. Thus, your mother may have had a bigger impact on your career choice than you want to admit; maybe not through her nagging advice, but through wielding her influence in uterus.

We note that it remains unknown whether the prenatal environment plays a role or not. For this reason we are not making a claim about nature vs. nurture. It is possible that our correlation is driven fully by environmental factors that affect androgen levels in the womb. It is also possible that our correlation is driven fully by the common cause of genetics. Nevertheless, our study suggests the possibility that the prenatal environment influences risk preferences. We speculate that this may be adaptive; the fetus might shape its preferences according to the world it is about to be born into. Of course, such a claim requires future investigation.

References

Barsky, Robert B., F. Thomas Juster, Miles S. Kimball, and Matthew D. Shapiro. 1997. “Preference Parameters and Behavioral Heterogeneity: An Experimental Approach in the Health and Retirement Study.” Quarterly Journal of Economics, 112(2): 537-579.

Burnham, Terence B. 2007. “High-testosterone men reject low ultimatum game offers.” Proceedings of the Royal Society B, published online: doi 10.1098/rspb.2007.0546.

6

Byrnes, James, David C. Miller, and William D. Schafer. 1999. ”Gender differences in risk taking: a meta analysis.” Psychological Bulletin, 125(3): 367-383.

Chen, Yan, Katuscak, Peter, and Emre Ozdenoren. 2005. “Why Can't a Woman Bid More Like a Man?” Working Paper, University of Michigan.

Croson, Rachel and Uri Gneezy. 2004. “Gender differences in preferences.” Submitted to Journal of Economic Literature.

Gneezy, Uri, and Jan Potters. 1997. “An Experiment on Risk Taking and Evaluation Periods.” Quarterly Journal of Economics, 112(2): 631-645.

Kosfeld, Micheal, Heinrichs, Markus, Zak, Paul J., Fischbacher, Urs and Ernst Fehr. 2005. Oxytocin increases trust in humans. Nature 435: 673-676.

Manning, John T., Diane Scutt, James Wilson, and D. Iwan Lewis-Jones. 1998. “The ratio of 2nd to 4th digit length: a predictor of sperm numbers and concentrations of testosterone, luteinizing hormones and oestrogen.” Human Reproduction, 13(11), 3000-4.

Manning, John T., and Rogan P. Taylor. 2001. “Second to fourth digit ratio and male ability in sport: implications for sexual selection in humans.” Evolution and Human Behavior, 22(1): 61-69.

7

Manning, John T., A. Stewart, Peter E. Bundred, and Robert L. Trivers. 2004. “Sex and ethnic differences in 2nd to 4th digit ratio of children.” Early Human Development, 80(2): 161-8.

McIntyre, Matthew H. 2006. “The use of digit ratios as markers for perinatal androgen action.” Reproductive Biology and Endocrinology, 4, doi: 10.1186/1477-7827-4-10.

Millet, Kobe, and Siegfried Dewitte. 2006. “Second to fourth digit ratio and cooperative behavior.” Biological Psychology, 71: 111-115.

Paul, Simon N., Bernet.S. Kato, Lynn F. Cherkas, Toby Andrew, and Tim D. Spector. 2006. “Heritability of the Second to Fourth Digit Ratio (2d:4d): A Twin Study.” Twin Research and Human Genetics, 9(2): 215-219.

Van den Bergh, Bram, and Siegfried Dewitte. 2006. ”Digit ratio (2D:4D) moderates the impact of sexual cues on men's decisions in ultimatum games.” Proceedings of the Royal Society B, 273: 2091-5.

Wallace, Bjorn F.N., Cesarini, David, Lichtenstein, Paul, and Magnus Johannesson. 2007. “Nature, Nurture and the Ultimatum Bargaining Game.” Mimeo, Stockholm School of Economics.

8

Acknowledgements We thank Johan Almenberg, David Cesarini, Uri Gneezy, Emir Kamenica, Magnus Johannesson, Matthew McIntyre, Karen Norberg, Martin Nowak, Jessica Pan, Thomas Pfeiffer, Al Roth and Robert Trivers for great comments. Anna Dreber thanks the Jan Wallander Foundation for financial support. The Program for Evolutionary Dynamics at Harvard University is sponsored by Jeffrey Epstein.

Author Information The authors declare no competing financial interests. Correspondence should be addressed to Anna Dreber ([email protected]) or Moshe Hoffman ([email protected]).

9

2….to the top of the fingertip

ring

index

1. Measure from the center of the crease closest to the palm….

Figure 1. 2D:4D 2D:4D is the ratio between the length of the 2nd (index) finger and the 4th (ring) finger. Fingers are measured from the flexion crease proximal to the palm to the top of the digit.

10

(1)

(2)

Female

Left 2D:4D

(3)

(4)

(5)

-333.125 (-3.79)**

-298.9681 (-3.16)*

-306.4422 (-3.27)*

-3364.464

-2619.291

(-2.75)**

(-2.27)*

Right 2D:4D

-2477.64 (-2.15)*

-1916.778 (-1.71)

Constant

4306.69 (3.66)**

3426.337 (3.09)**

1183.125 (20.10)**

3692.333 (3.35)**

3003.66 (2.80)**

Observations

120

116

125

120

116

R-squared

0.05

0.03

0.10

0.12

0.11

Robust t statistics in parentheses * significant at 5%; ** significant at 1% Table 1. Determinants of Risk (X, in SEK), OLS. The dependent variable is X, our measure of risk preference. Lower X indicates greater risk aversion. Right and Left 2D:4D indicate the digit ratios on each hand, our proxy for prenatal androgens.

11