Three Essays on Elderly Migration and Local Fiscal Policy

The first essay develops a retirement migration model using a nested logit (NL) specification. The dichotomous nest structure is constructed based on States' average temperature; Sunbelt nest and Non-sunbelt nest. The estimation results from the census 2000 reveal two interesting facts. First, the Multinomial-logit model and the NL retirement migration model agree on which state tax policy is an important determinant of interstate migration of retirees in terms of the magnitude and statistical significance of the coefficient. Second, the migration patterns predicted by the MNL-based model tend to be under-estimated, compared to the NL retirement migration model. Second essay develops a retirement migration model based on the random coefficient specification to investigate the effect of individual heterogeneity on migration decisions of retirees. The estimation results from the census 2000 reveal three interesting facts regarding the effect of individual heterogeneity on migration decisions. First, income and housing value of retirees interact negatively with income tax and property tax, respectively. Second, warm winter interacts positively with health condition and education level, and negatively with home ownership. Third, retirees are heterogeneous regarding the degree to which they appreciate warm winter. Third essay develops a stochastic dynamic programming discrete choice model which explains the sequence of migration and retirement decisions of older workers in the United States. The structural parameters of the model are estimated using the panel data from the Health and Retirement Study for the households whose heads worked in 1992. This paper contributes to the elderly migration literature in the following three aspects. First, it examines the migration decisions of older workers approaching retirement, an age group whose migration behavior has rarely been investigated. Second, including retirement decision in the model allows me to investigate the interaction between retirement and migration decisions. Third, this is the first study which estimates a stochastic dynamic programming discrete choice model for elderly migration. I find that the differentials in local fiscal policy across U.S. states affect the decision of where-to-move conditioning on moving, but does not affect the decision of whether-to-move to any significant extent.

Fiscal policy

17740