The rapid and continued escalation of transit company deficits as well as the emphasis on improving transit system management, has led to an increasing demand for transit performance evaluations. Critical to carrying out such a performance audit is the development of a system of performance measures against which a transit company's performance can be evaluated. As a diagnostic tool for improving transit performance, evaluation must go beyond a review of performance measures and determine why the measures have responded in a particular way. With this in mind, this dissertation develops a system of performance measures, explains the variation in these performance measures, and draws conclusions regarding the policy and managerial implications for transit companies. A framework which evaluates the process of transit service provision, and incorporates the transit system objectives of multiple actors (transit managers, transit users, and government) is developed. Based on this framework, a system of performance measures is proposed: efficiency, effectiveness, user's cost effectiveness, subsidy effectiveness, and level of service measures or quality measures. The analysis utilizes Section 15 1979-1982 Urban Mass Transportation Administration (UMTA) data, focusing on bus systems within large companies (250 revenue vehicles or more); 41 companies over a four-year period (130 units of observations) are included. Due to data limitations quality indicators are represented by proxy variables. A modified hedonic price model was employed for selecting variables to be incorporated in the quality measurement model. Data envelopment analysis (DEA), a technique which enables the use of multiple-inputs and multiple-outputs, is used to compute the performance measures. This dissertation develops explanatory models to explain variation in the performance measures. A simultaneous system is specified which incorporates the interrelationships among the performance measures (endogenous variables), and other local environmental factors (exogenous variables) that may affect the measures. Two stage least squares (2SLS) is used to estimate these simultaneous systems. Based on the performance measures and the explanatory results, a set of policy and managerial recommendations is presented which cover labor, capital, services, fares, and subsidies.
