The association between health problems and retirement decisions

This paper studies the association between health problems and retirement decisions by distinguishing between physical- and mental health and also distinguishing between fulland partial retirement. To account for the endogeneity issue between health and retirement, a two-stage IV approach is taken. In the first-stage, self-perceived health and retirement variables are instrumented. In the second-stage, predicted values of these variables are used to model the association between health and retirement. Consistent with previous studies, the results show that full retirement is associated with worse physical health and partial retirement is associated with better physical health. In terms of magnitude, the marginal effects of full retirement on self-perceived health are larger (difference of 0.1 to 3.4 percentage points in probability) than those of partial retirement in all health categories. Contrary to previous studies, the initial results show that full retirement improves cognitive abilities. In terms of magnitude, partial retirement has larger marginal effects (difference of 0.002 to 0.050 in coefficient) on fluency and numeracy scores than full retirement but smaller effects (difference of 0.020 to 0.068 in coefficient) on memory scores. Fortunately, after including an interaction term between age and retirement, this research finds evidence (contrary to the initial results) that full retirement and partial retirement are associated with declining cognitive measures for people of different ages.