How Different is a College Degree From a Spouse?

More educated individuals face substantially lower mortality rates than less educated ones. In our recent paper Pijoan-Mas and Ríos-Rull (2014), we use data from the Health and Retirement Study (HRS) to compute expected longevity at age 50 for white males and white females of different education levels in the US (the focus on these age and race groups is because of sample sizes.) We find that the difference in expected longevity between college graduates and individuals without a high school degree is large: 6.6 years for males and 5.8 years for females.

In an attempt to understand where these differences come from, we also use information on individual self-rated health to decompose the longevity advantage of education into three parts: (1) differences in health at age 50, (2) differences in health evolution after age 50, and (3) differences in mortality rates conditional on health. Our results show that 1/3 of the gradient comes from differences in health at age 50, while the remaining 2/3 comes from the health-protecting role of education after age 50 (see Chart 1). Strikingly, mortality rates conditional on health are not different across education groups. That is to say, once a person states that her health is bad or very bad her chances of survival to the next wave are independent of her educational level.


To some extent these results are not surprising: we know that education is health protecting and that, if well measured, health should be a sufficient statistic for survival. However, our results reveal two important findings. The first finding is that self-rated health is indeed a very good summary of the health condition of an individual, at least regarding the chances of survival. This is important because information on self-rated health is cheap to gather in survey data, and it is widely available today in many household level surveys. The second finding is the equality of individuals in their final hours, or more precisely, in their final years. Our estimated biennial survival probabilities are independent of education once self-rated health is taken into account, which means that there is no socio-economic advantage for individuals in bad health: they die as often as their low-educated counterparts. Higher-educated individuals earn more income, own more wealth, and are more likely to have expensive private health insurance; yet, once health has deteriorated enough socio-economic status does not make any difference in the biennial survival rates. These results are preserved if we measure socio-economic status by wealth or income instead of education.

However, we found different results when performing the same exercise with marital status. White males who are married at age 50 live on average 2.6 more years than those who are not. The gap for women is 1.4 years. We also decompose these differences in differences in initial health, differences in health evolution, and differences in survival rates conditional on health. In contrast to education, wealth, or income, we find that the largest component in the differences in expected longevity between marital status types is the difference in mortality rates conditional on health status (see Chart 2). That is to say, when individuals respond that their health is bad or very bad, those that are married are more likely to survive. This begs the question of whether spouses are doing something for you that money cannot buy.



Josep Pijoan-Mas and Víctor Ríos-Rull (2014), “Heterogeneity in Expected Longevities”, Demography, 2014, 51(6), 2075-2102.

This entry was posted in Education, Health, Mortality by Josep Pijoan-Mas. Bookmark the permalink.
Josep Pijoan-Mas

About Josep Pijoan-Mas

Josep Pijoan-Mas is associate professor of economics at CEMFI and research fellow at CEPR. He got his PhD in economics at University College London in 2002. His research interests are related to household decisions under uncertainty, with special interest on the implications for income and wealth inequality. He has recently started to do research on inequality of health outcomes.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>