Risk aversion and religious behaviour: Analysis using a sample of Danish twins.

Area of Science:

• Behavioral Economics
• Psychology of Religion
• Behavioral Genetics

Background:

• Economics, particularly Expected Utility Theory, provides a framework for understanding religious behavior as a form of insurance for life and afterlife.
• Previous empirical studies suggest a correlation between risk aversion and religious activities like church attendance and prayer.
• Concerns exist regarding spurious correlations due to unobserved factors influencing both risk attitudes and religious behavior.

Purpose of the Study:

• To empirically examine the association between risk attitudes and religious behavior (church attendance, prayer frequency).
• To control for unobservable variables using data from Danish same-sex twin pairs.
• To investigate whether the observed correlation is driven by individuals believing in an afterlife.

Main Methods:

• Cross-sectional analyses to verify the correlation between risk preferences and religion.
• Panel data analyses of same-sex twin pairs to control for shared genetic and environmental factors.
• Analysis focusing on the subgroup of individuals who believe in an afterlife.

Main Results:

• Cross-sectional analyses confirmed the previously found correlation between risk aversion and religious behavior.
• The association between risk attitudes and religious behavior was significant only among individuals who believe in an afterlife.
• Panel data analyses, which control for unobservable factors within twin pairs, revealed that the correlation between risk aversion and religious behavior is no longer significant.

Conclusions:

• The initial correlation between risk aversion and religious behavior may be explained by unobserved factors, as it becomes insignificant when controlling for shared influences in twin pairs.
• Belief in an afterlife appears to be a key factor in the observed association between risk attitudes and religious behavior.
• Caution is advised in interpreting the results; the lack of significance in panel data could stem from measurement error or lack of within-pair variation, rather than a true absence of association.