Abstract:
Short-run variable cost functions are estimated with the use of a sample of 68 bus systems in the U.S. Midwest operating in the 1996 to 2002 period. Variable costs are specified as functions of outputs, input prices, bus stocks, and network size. Various output, input price, network, capacity, and disturbance specifications are examined. The cost functions are estimated with the use of ordinary least squares, switching regression, seemingly unrelated regression, and stochastic frontier cost models. The results indicate constant returns with vehicle outputs, increasing returns with passenger outputs, inelastic input demands, and substantial technical efficiencies that vary by firm and by state.
