De Borger (1984) claims that literature results on scale economies vary because
the samples include agencies operating in different environments, and the studies disregard
differences in these environments. Miller (1970) states: “If all cities had identical
distributions of population, economic activity and identical geography, the outputs of bus
firms in different cities could be meaningfully compared by use of the single output
variable used in most previous studies: the vehicle-mile” (p. 24). Clearly, most of the
previous cost studies are of little help to understand the role of site-specific geographical
and physical cost factors. Only a few studies consider site-specific variables. Lee and
Steedman (1970) suggest that traffic and terrain conditions are cost factors. They use
population density as a proxy for traffic conditions, but terrain conditions are ignored
because of difficulty in measuring them. Lee and Steedman’s (1970) findings indicate
higher power, traffic operation, and total costs (per bus-mile) for higher levels of
population density; and lower traffic operation costs for lower average bus speeds. Koshal
(1970) reports higher marginal costs for agencies operating in mountainous routes.
Jorgensen et al. (1995) show that bus agencies operating in the coastal areas of Norway
have higher costs than the ones operating in the inner regions. Total route length is used to
approximate network characteristics in Matas and Raymond (1998) and Karlaftiz and
McCarthy (2002).