The Causal Effect of Age on Subjective Well-Being

Clemens Noelke, Harvard University
Marcia Jimenez, Brown University
Corsi Daniel, Ottawa Hospital Research Institute

Empirical research has thus far not clarified whether and how subjective well-being changes with age. We replicate, clarify and reconcile evidence from three widely-cited studies that use the same US dataset to arrive at very different conclusions about how well-being changes with age. We explore three sources of bias to explain the discrepant findings: bias due to period or cohort effects, collider bias, and functional form misspecification. Our results indicate that estimated age effects are highly sensitive to the inclusion of collider variables, such as marital status or health, which lie on the causal path from age to happiness. We suspect that inappropriate conditioning on collider variables when estimating age effects is more widespread, which is why our results illustrate a potentially important form of misspecification in research on aging.