Changes in the Stochastic Planning Horizon and Its Effect upon the Elderly

When individuals are making decisions regarding the purchase of a durable commodity, the choice of job, or any other decision that has a time duration element to it, these decisions critically depend upon the lifetime or the time horizon of the object being decided upon. The reason that the time horizon is so critical comes from the fact that the worth of each option depends not only upon its initial value, but also its value over time. For an elderly person, the durable or useful life of an object is often greater than the remaining lifetime of the person. In situations where this is true, the relevant planning horizon is the individual's time until death. Since the date of an individual's death is unknown in advance, this creates a stochastic planning horizon. To see how the stochastic planning horizon of the elderly changed with time, the first chapter of this thesis examines elderly mortality and how it is changing over time. The method used in Chapter I to describe the mortality experience of the elderly is the time to failure or hazard model. Using this approach, a time varying version of Gompertz's Law is developed and estimated. The estimated model is then used both as a way to quantify exactly how mortality for elderly has been changing and as a basis for estimating life expectancies in the subsequent chapters of this thesis. The elderly's housing decision, a decision where the effective time horizon is the person's remaining lifetime, is the focus of the second chapter. The econometric model of the decision to move or not to move, developed and estimated in this chapter, explicitly incorporates the changing and stochastic nature of the elderly's planning horizon in this type of situation. Another important feature of the model is that unobserved differences (unobserved heterogeneity) between households are accounted for in the decision process. The third chapter turns again to mortality. This chapter develops a method to use both observed and unobserved characteristics of elderly males to estimate the effect that the various observed characteristics have upon the survival probability of elderly males. These estimates are then used to look at some specific implications of the changing elderly male mortality. Thesis Supervisor: Daniel McFadden Title: Professor of Economics