Saving behavior of U.S. households a prospect theory approach /

by 1979- Fisher, Patricia Jo

The main purpose of this dissertation is to explore household saving using a prospect theory approach through the use of the loss aversion model and behavioral portfolio theory. The research begins by investigating the effect of having expected perperiod income above or below the reference level as well as the effect of uncertainty on the likelihood of saving based on the loss aversion model. The focus then moves to saving motives based on the ideas of behavioral portfolio theory. The direct measure of saving available in the dataset is saving over the previous year. Saving horizon is also investigated since the saving measure is a short-term measure and some regular savers may not have saved during the past year. The dataset used is the 2004 Survey of Consumer Finances. The sample excludes retired U.S. households for a final number of 3,694 households. Having expected perperiod income above the reference level increases the likelihood of saving. Having expected per-period income below the reference level is significantly and negatively related to the likelihood of saving, and has a greater effect on the likelihood of saving than having expected per-period income above the reference. The group of uncertainty variables is significant in explaining the likelihood of saving. In contrast to the theories ii reviewed, most of the uncertainty variables are not found to increase the likelihood of saving. Saving motives and saving horizon are significant in explaining the likelihood of saving. Saving for a foreseeable expense significantly increases the likelihood of saving in both the models with and without interaction terms. Having a motive to save for the education of children or grandchildren significantly decreases the likelihood of saving in the model without interactions, while this variable is not significant when interactions are added. Inclusion of interactions of saving horizon variables with the saving motive variables is found to be significant in explaining the likelihood of saving, indicating that saving motives do differ by saving horizon. iii