Determinants of subway travel in the New York City Metropolitan Area: An empirical research and econometric application of discrete choice and time series models to urban travel demand

This dissertation deals with the modeling of transportation decisions in the New York City Area. Three different data sets are used: cross sectional data, micro sample data and time series data. The main model estimated is a logistic model developed earlier by McFadden and the independent variables are socio-economic variables, mode characteristic variables, and dummies. Each data set addresses a particular question. By using the first data set we find out that the estimation of the subway ridership is sensitive to spatial area. By using the second data set we estimate the behavior of the riders and show that their optimal decision based on optimizing their utility depends on the mode characteristics. We also derive the value of walking time and the value of auto in-vehicle time and we estimate aggregate elasticities for auto and bus. By using the third data set we estimate an aggregate elasticity for the demand for subway trips and estimate the impact of public policies such as increase in fare.; In addition, in this dissertation we attempt to develop an econometric approach that unifies time series and prediction from static models such as the logistic regression. By using one canonical model and varying the underlying assumptions and the distribution of the dependent variable we show that forecasting results can be obtained in ways not so different; only the optimizing algorithm is different. It appears that the two approaches underlined above are useful because they can be used to explain the dynamics of human behavior. This is one of the main contributions to the field of transportation economics.