MAXIMUM LIKELIHOOD ESTIMATION OF AN INTEGRATED LAND USE AND TRANSPORT SYSTEM

Discrete choice models are widely used in modeling residents' location and travel behaviors. An example is the application of logit based stochastic bid rent model to capture urban land use decisions and the resultant land prices. On the other hand, maximum likelihood estimation (MLE) is recognized as a valid and efficient approach in estimating parameters defined in these models in practice. While abundant empirical analyses have been done in estimating bid rent functions or on the hedonic pricing of housing sale values, there are, as far as we know, few studies that investigate the relationship between housing rents and residents' location and complicated travel choices. This paper develops a full information MLE formulation to estimate the parameters defined in a nested multinomial logit model combined with the bid-rent process within a general equilibrium formulation, using data from a transit oriented development (TOD) area in Hong Kong. The approach derives the first-order derivatives of the log-likelihood function with respect to the unknown parameters in order to speed up the estimation process. The results for the calibration of the study area will be provided to illustrate the approach.