In a review of the cost structure of the railway industry,
Kessides and Willig (1995) establish that the industry as an
aggregate displays scale economies: An equi-proportionate
change in the levels of all services provided in the firms’
complete network would require a less than proportionate
change in the level of costs within relevant ranges of traffic
production. This means that activities can be described by
way of a long-run average cost curve that declines as the
quantity of the firms’ output of a given collection of services
increases. While we may hypothesize that possible scale,
scope and density economies have a natural basis in the
huge and lumpy investments in track-and-structures, and
less in the costs for operating the system, previously
available data has not lent itself to understand these aspects.
Using disaggregate data of a different nature than in
previous studies, the present paper demonstrates that the
costs for maintaining tracks constitute a complementary
reason for the industry’s cost reduction with usage, at least
for the levels of track use reported here; the cost elasticity
for marginal variations in traffic levels is far below one. One
policy implication is that if traffic is priced at marginal costs
for track wear, revenue from these charges would be
inadequate for recovering the total costs for the maintenance
of the railway infrastructure.10
We have also discussed the necessity of making reinvestment
costs part of the charging structure. Lack of data
(Sweden) and a short time series (Finland) have not made it
feasible to draw any conclusions about the appropriate size of
the marginal cost with respect to renewal spending. Policy
makers would therefore have to rely on engineering data and
rules of thumb to come up with a number, much in the same
way as the corresponding cost is estimated for roads.
Implicitly, this also points to the sort of data collection that
has to be initiated as a consequence of Europe’s institutional
reforms. In order to estimate reliable marginal costs,
appropriate data bases have to be compiled andmade available
for others than the railway administrations themselves. If not,
it will be difficult to convince operators that charges have been
calculated along the lines established in the directives.
Like all empirical, non-experimental analyses, the data
sets that have been analyzed reflect the way in which activities have been handled in historic time. With hindsight,
this may be flawed policies, for instance since
decision makers had incomplete information when allocating
resources or in view of external budget constraints that
are seen to be inappropriate. Spending in 1 year may also
reflect the accumulated consequences of traffic loads over
previous years. With the data at hand, we have poor
possibilities of controlling for any of these circumstances.
The strength of the analysis, on the other hand, is that we
base the results on very few assumptions and simply look at
the evidence at hand.