Test of a Hierarchical Model of Auto Choice on Data from the National Transportation Survey

In this study a hierarchical decision model of the choice of different size cars is tested and used to make aggregate forecasts of the U.S. automobile market. The model, originally developed in a study of southern California automobile buyers, was tested using two types of national data. First, it was used to predict choices made by the new car buyers who were respondents in the National Transportation Survey done in 1978. Second, an aggregation procedure was developed to enable the model to forecast market shares for different size cars for the period 1980-1990. The predictions of the model were then compared with the actual market shares. Both tests confirm the principal prediction of the model: that there is a constant demand for cars with larger interior capacity. The model answers an important question in transportation policy: what array of fuel- efficient cars will best meet the transportation needs of U.S. families? The aggregation technique developed for the study is a straightforward method for making macro predictions from micro decision models.