Franchising and competitive tendering has become an important method for introducing competition for the market where competition in the market is considered to be undesirable. In its simplest form, subject to certain assumptions, theory predicts that franchising should result in the tender being awarded to the most efficient operator. The real world is more complex,of course, but in general franchising should at least put pressure on allbidders to be efficient. The empirical evidence suggests that competitivetendering / franchising has produced substantial cost reductions in a range of contexts, for example, refuse collection and NHS ancillary services,as well as, in the transport sector, London buses. By contrast, despite some reported Train Operating Company (TOC) cost savings during the early years after privatisation, TOC costs have increased very sharply since the Hatfield accident. TOC own costs (that is, excluding rolling stock costs and access charges) increased by almost 50% between 1999/00 and 2003/04 (ornearly 40% on a cost per train km basis). As a result, subsidies to passenger train operators increased sharply over this period. This paper aims to explore why franchising failed to bring costs down in the passenger railsector in Britain over the period 1996-2004. State-of-the-art efficiency methods were used to attribute cost rises to inefficiency effects between TOCs, shifts in the production possibilities frontier over time and changes resulting from scale effects. While a range of methods have been employed across a limited number of previous studies, it is unclear whether differences in results are true differences or simply due to the statistical method employed. For example it is acknowledged that utilising the fixed effects approach may over state the true efficiency effects since the dummy variables capture all time invariant characteristics. Several models estimated using different parametric efficiency techniques utilising the same data set. This allows us to investigate the extent to which the econometric method impacts on the results of the analysis. Panel models will be estimated including fixed and random effects, as well as stochastic frontier models (SFA). SFA methods decompose the error term of the cost function into a symmetric random error term capturing uncontrollable factors and a one sided error which captures firm specific inefficiency and these are estimated using the maximum likelihood principle. Data are used from 1996/97 to 2005/06 for the 25 TOCs in Great Britain. Studies to date have only considered data up to 1999/2000. There is great policy interest in extending the analysis forward given the cost explosion in railways and specifically in train operations. Such are explored as can the cost of improved quality over the period be measured? Is the push towards fewer, larger franchises likely to reduce costs given the cost structure? To what extent have cost rises been industry-wide versus TOC specific? Finally, the data are improvedby utilising not just data from TOC accounts as in previous studies, but also access and penalty payments to/from Network Rail/Railtrack which, once subtracted from TOC costs give us a much more reliable measure of the costs directly under control of the TOCs. Identifying TOC own costs is essential to understanding the true drivers of TOC cost increases versus those which have been imposed on the TOCs by other areas of the industry. For the covering abstract see ITRD E137145.
Abstract