On the cost structure, efficiency and productivity of mass transit systems

The purpose of the present study was to investigate the cost and production structure of mass transit systems using both panel and time series data. Accepting cost minimization (subject to a set of constraints that reflect demand, regulatory, and urban setting conditions) as the economic objective of transit systems, empirical analysis of the principal cost and production properties of transit operations was performed. The present research will help in understanding the cost properties of transit services production, and as a result it can foster better utilization of available resources and improved fleet management. Moreover, the study of the economic conditions under regulation can help identify what potential benefits can be expected from policies of privatization and deregulation. The present study introduced a fixed-effects type model for the panel data. This model captures differences in economies of density and scale between large, medium, and small systems by introducing a separate coefficient of output for the different system sizes. The results indicate that, in the short-run and at mean production level, large and medium systems exhibit economies of density while small systems exhibit diseconomies of density. It is very important to note that the coefficients of output yielding the above results cannot be statistically restricted to being equal across system sizes. The models introduced for the time series data indicate that error term serial correlation, a time series phenomenon disregarded in most previous studies, can have serious effects on the results pertaining to scale economies and overcapitalization in the transit sector. Transit systems exhibit slightly decreasing returns to density when the models correct for autocorrelation, while they exhibit economies of density when the models do not correct. Furthermore, the present study found that the widely accepted notion of overcapitalization in the transit industry might be, at least in part, the result of modeling mis-specification