Access Type

Open Access Dissertation

Date of Award

January 2016

Degree Type


Degree Name




First Advisor

Allen Goodman


Theoretical models of insurance suggest that when individuals have private information about their risk type alone, insurance coverage will be adversely selected by those riskier individuals due to asymmetric information. In this study we investigate whether individuals have private information of their risk type and riskier individuals are indeed more likely to choose health insurance in the context of Medigap insurance market in the United States where government intervention reinforces information asymmetry. Medigap is supplemental private insurance that optional to those who have Medicare Part A and Part B and it is used to cover some of the cost sharing required by Medicare

We find out that conditional on risk classification assessed by insurance companies, self-rated health status is positively correlated with total health expenditure and negative correlated with Medigap coverage. It demonstrates that people have private information about their risk type; however, those with worse self-rated health status less likely to choose Medigap coverage. This finding provides evidence that on average there is advantageous selection rather than adverse selection in Medigap insurance market. Meanwhile, we find out coexistence of moral hazard.

The only factor that could be potentially identified as source of advantageous selection in this study is cognitive limitation. Cognitive limitation is positively and significantly correlated with logarithm of total health expenditure in 2009, 2010 and 2011; at the same time, it is negatively and significantly correlated with Medigap purchase regardless of controlling only premium variables or premium variables plus risk aversion, education and poverty categories in Seemingly Unrelated Regression model in 2009 and 2010.

The factors that have been investigated in several previous articles such as education, risk aversion, poverty categories which reflect economic status seem to have little effect if there is any in our study.

Included in

Economics Commons