by Aureo de Paula

My research project utilizes the Health and Retirement Study to analyze joint retirement decisions using a novel methodology proposed by Honoré and de Paula (2008). Joint retirement of husband and wife is not an uncommon phenomenon and has been documented by various researchers. The concurrent retirement of husband and wife can be explained either by correlated factors leading retirement (observable and non-observable to the researcher) and by taste interaction between spouses. The distinction is similar to the differentiation between correlated and endogenous effects in Manski (1993) and has implications for policy prescription and estimation purposes. As recognized previously in the retirement literature, overlooking the `simultaneous equation' nature of the problem can bias estimation and mislead inference. To study the question, I rely on the model delineated by Honoré and de Paula (2008) which extends the familiar duration analysis to handle strategic interactions among multiple agents. That paper shows point identification and suggests estimation strategies for various parameters related to correlated and endogenous effects even in the presence of multiple equilibria. The model can be seen as a non-unitary model of the household which does not impose constraints (equilibrium selection rules) that may lead to erroneous inferences if imposed. This project will allow me to quantify the relative importance of common shocks versus direct interaction reasons for joint retirement and in doing so it has the potential of enlightening policymakers and researchers interested in retirement issues.