AN INVESTIGATION OF LIFE CYCLE SAVING BEHAVIOR IN THE UNITED STATES

This dissertation is an attempt to utilize as much available detailed micro data as we can put our hands on in order to describe the behavior of household savings as realistically as the data allow. The aim was to build up a sufficiently detailed and well-formulated, empirically-based, description of micro savings behavior with particular emphasis on pattern of savings over life of individuals and households, so that eventually we could pin down in much more detail aggregate consumption function characterizing aggregate time series data. To this end, I have assembled a number of cross-section data, some of which are familiar in the literature as well as others that have not been utilized, and constructed a coherent story based on them. In the process, I have encountered several important features of the data which apparently biased many earlier results. For example, for the first time my investigation gave us some quantitative estimates of how badly our estimate of the life pattern of savings can be biased when we work with data generated by surveys of independent households because of the fact that, as individuals become older, some of them cease to be independent households and are not observed in such survey data. Since the probability of individuals remaining as independent households depends on their income and wealth, this can be a source of serious bias