Risk Attitudes, Gender, and Risk Behavior: Evidence from Two Laboratory Experiments

Abstract

This study analyzes whether self-reported attitudes in economic risk taking predict experimentally revealed risk behavior, and how gender moderates the relationship between both measures. Prior research often finds women reporting higher risk averse attitudes than men and showing more risk averse behavior in observational or experimental studies. This study analyzes observations from 369 students who participated in two laboratory experiments and answered a survey about their risk preferences. The findings show that risk attitudes are not likely to predict risk behavior directly, but being female predicts risk averse behavior robustly. Most interestingly, the analyses show that in the experiments, women behave consistently to their self-reported risk attitudes, but men do not. Methodological and practical implications are briefly discussed.

I am very grateful for the support and valuable comments on this paper from Markus Tepe and Michael Jankowski. I wish to thank the participants of the 2018 Annual Meeting of the DVPW working group on decision theory for the fruitful discussion on a prior version of this paper, as well as the two anonymous reviewers for their thorough comments, which further helped to improve this work.

Appendix

Displayed instructions—Auction

In the following auction you will bet against a computerized bidder over 30 rounds. Your goal is to outbid the computer. If your bid is higher than the computer’s bid you have won the respective round. In the case of ties your bid does not win. In each round, you will have a new randomly chosen endowment with an amount between 1 and 100 points. The computer always bids between 1 and 50 points. All amounts are distributed equally.

Gain = endowment − bid

All the points you win are added, and converted into Euro with an exchange rate of

1 point = 0.01 €.

For each bid helpful information is provided (see Fig. 2 of the paper instructions). The following data can be displayed as often as you wish:

Expected payoff: Outcome in points in the case of winning

Expected payoff = endowment − bid

Winning probability: Probability of winning the auction with the actual bid. The value ranges from 0 to 1. The higher the value, the higher is the probability to win. 0 = sure no win, 0.5 = same probability of winning or not winning, respectively, 1 = sure win

Expected value = expected payoff * winning probability

Example:

With an endowment of 59 points you bid 27 points, and have a winning probability of 0.5 (i.e. 50%). The computer’s bid is 13 points. Yor bid is higher than the bid of the computer, thus you are the winner of the auction. In this round you are winning

59 − 27 = 32 points * 0.01€ = 0.32 €.

As soon as you have decided on your bid, please select the appropriate row and click the button “submit bid”.

Displayed instructions—Lottery

Structure

In this game you choose between alternative lotteries.

You will decide 10 times. Each time you choose between option A and option B.

Each option can lead to two events. The probability and the gain in points of each event will be displayed in a table.

Please notice that only one of your decisions will determine your earnings.

Previously you will not know which lottery will be played.

When the time for input expired the computer will calculate two random numbers.

The first random number (between 1 and 10) determines which lottery will be played.

The second random number (between 1 and 10) determines your gain in points depending on your choice for option A or B.

Depending on the option you chose (A or B) the respective amount in points will be added to your account.

Example

Notice the example on the following screen. Please click “Forward”.

Lottery number	Option A	Option B	Your decision (A or B)
1	40 of 10% or 32 of 90%	77 of 10% or 2 of 90%
2	40 of 20% or 32 of 80%	77 of 20% or 2 of 80%
3	40 of 30% or 32 of 70%	77 of 30% or 2 of 70%	A
4	40 of 40% or 32 of 60%	77 of 40% or 2 of 60%
5	40 of 50% or 32 of 50%	77 of 50% or 2 of 50%

Example

If the first random number is 3, the lottery number 3 will be played. We assume you chose Option A in lottery number 3. Option A in lottery number 3 provides “40 of 30% or 32 of 70%”. In other words Option A in lottery number 3 offers a 30% chance to win 40 points and a 70% chance to win 32 points. If you would have been chosen Option B in lottery number 3 you had have a 30% chance to win 77 points and a 70% chance to win 2 points. The second random number determines which event actually occurs. Imagine a 10-sided die. If the throw of the 10-sided die in Option A of lottery number 3 is smaller than or equal 3 you win 40 points. If the throw of the 10-sided die in Option A of lottery number 3 is greater than 3 you win 32 points. Choose one of the options in each of the 10 lotteries. There are two runs of this game each with different amounts of points.

Appendix (Tab. 4, 5, 6, 7, 8 und 9; Fig. 4)

Table 4 Lottery choice experiment by Holt and Laury (2002)

Table 5 Variables

Table 6 Descriptive statistics by experimental vehicle and bivariate analysis for the unbalanced and balanced data

Table 7 Risk attitudes and risk behavior by gender

Table 8 Descriptive statistics by experimental vehicle and bivariate analysis for the unbalanced and balanced data, all risk attitudes

Table 9 Robustness check—Risk behavior

Fig. 4

(Note: Based on full OLS regression model, 90% confidence intervals, from left to right: Predicted values of CRRA by gender; AME of gender on the relationship between risk attitudes and gender; AME sizes of stated risk preferences on risk behavior by gender)

Gender effect on the attitudes-behavior relationship, all risk attitudes.