Ownership and competition in European transit: assessing efficiency

Abstract

Ownership and competition and their effect on public transit system efficiency is a topic of considerable interest. In this article, we use a stochastic frontier error component model specification that incorporates a dynamic ARCH error structure to assess the effects of ownership and competition on European public transport systems using data from 15 European systems for a 10-year time period (1990–2000). Results indicate: (i) a clear rejection of the standard frontier model and existence of ARCH-related heteroskedastic technical inefficiency; (ii) transport operators in cities where competition has been introduced to demonstrate increased efficiency and (iii) efficiency findings vary between different error formulations.

Keywords:

• ownership
• competition
• public transport
• efficiency

Acknowledgements

The author would like to thank two anonymous reviewers and the Editor-in-chief, S.C. Wong, for their very helpful comments. All remaining errors are the responsibility of the author.

Notes

Notes

1. A detailed and comprehensive analysis of regulatory schemes including rationale and findings from a large number of research studies can be found in Karlaftis (2008).

2. The review in this article concentrates on the EU experience with Public Transport privatisation and deregulation. International experience with a variety of regulatory schemes can be found in Karlaftis (2008); comprehensive assessment of the US experience are offered in Iseki (2003, 2006), while very interesting European country-specific comparisons can be found in Bekken et al. (2006), Boitani and Cambini (2006), Farsi et al. (2006) and Roy and Yvrande-Billon (2007).

3. Theoretically at least, it is possible to show that private production of transit services has the potential of being both profitable and welfare-improving (Viton 1982, Dodgson and Katsoulakos 1988).

4. Variables included in Tables 1 and 2 are intended to give a general idea of the area and of some of the operating characteristics for the systems examined; they are certainly not an exhaustive list of all system operating parameters or city characteristics.

5. One of the very important issues dealt with in econometric modeling concerns the analysis of units (cities, firms, systems and so on) of dissimilar sizes and, as a result, heterogeneous. Although the issue of heterogeneity has been an important aspect of recent research in econometrics, it is almost impossible to fully account for it. A first step in recognising the potential problems is to cluster systems into, smaller, more homogeneous groups. Although we do not follow the statistical clustering approach used in other research (see, e.g. Karlaftis and McCarthy 2002, Karlaftis 2003, 2004, Iacono 2009), our grouping is based on an intuitive combination of city population, area (as a proxy for density) and system operating characteristics.

6. We note here that estimating a cost function rather than a production function takes advantage of the well-established duality between firm technology and costs which imply that a firm's cost function summarises all of the economically relevant information embodied in its production function (Varian 1992).

7. As much of the literature on transit performance have repeatedly indicated (see, e.g. Karlaftis and McCarthy (1997) for an extensive discussion and analysis on the subject), services should not be solely judged on financial success (efficiency) but also to the degree to which they achieve their social role (effectiveness). To this end, measures of output in transit analyses include vehicle-miles (or kilometres) or seat-miles when efficiency is concerned and passenger boardings or passenger-miles when effectiveness is concerned (Karlaftis and McCarthy 1999, Karlaftis et al. 1999). In this article, we restrict ourselves to measuring output as vehicle-kilometres because the primary goal is to assess efficiency differentials between various operational and ownership schemes. Readers interested in multi-output cost function specifications where potentially, efficiency and effectiveness can be jointly considered can refer to Lang and Welzel (1998) and Tsionas and Greene (2003).

8. The essential observation is that the mean of each normalised variable is equal to one. To obtain this result, consider the variable X, with mean X and sample size n. Normalising the variable X amounts to subtracting the mean of the variable ( X ) from each of the observations (X _i ). Since the variable is in logarithmic form, the normalization yields: Finally, we can show that the mean (μ) of the normalised variable is:

9. Statistical estimation details can be found in Washington et al. (2003), while tests for linearity are found in Vlahogianni et al. (2006).

10. The discussion regarding developments in efficiency modeling and its various facets, in the context of this article, is very brief. There are many sources of extensive and detailed analyses of recent developments, but Dorfman and Koop (2005) and Greene (2008) are excellent references.

11. This model is an extension of Kumbhakar's (1993) and Caudill et al.'s (1995) heteroskedasticity specifications.

12. One difference between the ARCH-related specification in Equation (3) and the well-known ARCH time-series model is that the variance h _it (of the technical inefficiency error component u _it ) is specified as a first-order function of previous values of the error component term ϵ _it and not of lags of u _it itself as in the ARCH model.

13. The loglikelihood function for this model is a straightforward modification of the function for the original Aigner et al.'s (1977) model and can be found in Lothgren (1998).

14. The log-likelihood function for the specification with ARCH correction is given in Lothgren (1998).

15. Monotonicity requires that the cost function be non-decreasing in input prices and is satisfied if the fitted factor shares are positive at each observation. Concavity of the cost function in input prices is satisfied if the Hessian matrix based on the fitted factor shares is negative semidefinite.

16. This, although it may at first appear as a minor finding, has far reaching practical implications and sheds some light at a long standing debate in the privatisation literature. Some authors, in-part as a justification for some counter-intuitive findings, have argued that frequently–in practice–‘good’ (i.e. financially healthy) systems are privatised and subpar performers are nationalised; in these cases, findings cannot per se be an indication of the success or failure of privatisation but rather a result of insufficient data. Additionally, and along these lines, some authors have argued that many of the previous studies have failed in correctly assessing the ‘true’ effects of privatisation (or nationalisation) since short periods of post-privatisation data have been used; this leads to insufficient data and questionable statistical analyses. This argument is largely supported by our findings which suggest that, since ‘shocks’ carry over to future time periods, longer time frames–i.e. more data–are necessary before correctly assessing results and basing policy decisions on them.

17. It is important to note, however, that we cannot reject the hypothesis of constant returns to density. The t-statistic for the null hypothesis of constant returns to density is −0.81, well below the 1.67 critical value.

18. As Farsi et al. (2006) discuss, stochastic frontier models assume that the stochastic error term is uncorrelated with the exogenous factors included in the equation, leading to results where firm specific unobserved effects are mistaken for efficiency differences and, further, that efficiency is constant over time. However, since excess costs (inefficiency) may be due to factors outside the transit system's control (such as contract type and type of ownership), we correct this by including the ownership-type dummy variables. The latter problem, that of constant efficiency over time, is addressed by the ARCH-type specification.

19. The issue of contract exogeneity is another of the important, yet unresolved, issues related to the privatisation debate (and to transit efficiency studies in general). Although Mathisen and Solvoll (2008) and Odeck (2008) indicate that mergers depend to a large degree on the regulatory scheme already in-place, there is no clear evidence in the literature whether the decision to privatise a transit system depends on its efficiency (also see footnote 16 on a related issue). Iseki (2003) attempted to capture this endogeneity problem using a 2SLS model, but in a simpler regression framework, and his findings were inconclusive.

20. Similar findings were also reported by Roy and Yvrande-Billon (2007).

21. Testing for homotheticity is equivalent to testing the null hypothesis that γ_iy = 0 (i = labour, maintenance) versus the alternative hypothesis that at least one of these parameters is non-zero.

22. A Cobb–Douglas technology characterises the production of transit trips if we can accept the null hypothesis that .

23. The literature provides mixed evidence on homotheticity. Berechman and Giuliano (1984), De Borger (1984) and De Rus (1989) also found a non-homothetic production structure. However, Williams and Dalal (1981), Williams and Hall (1981) and Berechman (1993) could not reject the null hypothesis of a homothetic production structure.