First Order Risk Averse and Risk Seeking Behavior: Evidence from the Laboratory

Objectives
This paper uses a laboratory experiment to study choice under uncertainty when
people are offered small stake gambles.
Data/Methods
This paper uses a laboratory experiment to study choice under uncertainty when
people are offered small stake gambles. Arrow (1971) shows that an expected utility
maximizer with a differentiable utility function accepts small stake positive-expectedvalue
bets. Rabin (2000) shows that first-order risk aversion for small stake gambles
implies unrealistic behavior over modest and large stake gambles. Empirical evidence
shows that many people are first-order risk averse, that is, they reject small stake gainloss
positive-expected-value bets. First-order risk aversion might be due to loss
aversion, probability distortion, or both. In this experiment we ask individuals to
choose between accepting or rejecting a series of small stake gain-loss gambles that
vary in terms of expected value and skewness.
Results/Expected Results
We find that most individuals rejectpositive expected value gambles with negative skew but accept negative expected
value gambles with positive skew. Thus, most people are first-order risk averse for
small stake gambles with negative skew but first-order risk seeking for small stake
gambles with positive skew. Additionally, we find that 48% of individuals accept
fifty-fifty gambles with positive expected value and reject fifty-fifty gambles with
negative expected value, 32% reject fifty-fifty gambles with positive expected value
and 20% accept fifty-fifty gambles with negative expected value. We show that firstorder
risk averse and risk seeking behavior are mostly driven by probability distortion
but that loss aversion also plays a role.