This paper develops a state choice model and estimates it using the
1990 census in order to determine if state and local governments can
attract retirees to their states using fiscal policy. In order to control for
the sunbelt-state effects, a nested logit structure which is less restrictive
than the multinomial logit model is embedded in the model. The empirical
results show that the nested logit model catches a significant role of state
fiscal policy on retirement migration. More specifically, they reveal that
the young old of age 60 to 65 are significantly influenced by income tax and
property tax, and the normal old of age 66 to 75 are affected by death tax,
as well as income tax and property tax. In addition, the migration decision
of healthy retirees turns out to be significantly influenced by fiscal policy
whereas the migration decision of unhealthy retirees is not. A specification
test of the state choice model rejects the competing multinomial logit
specification. The estimated model predicts that the average migration
rate will be close to the observed migration rates, but over-predicts return
migration because of the presence of high birthplace attachment