Quality of service, efficiency and scale in network industries: an analysis of European electricity distribution

Abstract

Quality of service (QoS) is of major economic significance in natural monopoly infrastructure industries. In this article, we present an efficiency analysis of electricity distribution networks from seven European countries. We apply the stochastic frontier analysis method to multi-output translog input distance function models to estimate cost efficiency and scale economies. We show that introducing the quality dimension into the analysis affects estimated efficiency significantly, especially that smaller utilities’ efficiency seems to decrease. Our results emphasize that QoS should be an integrated part of efficiency and economic analysis of regulated natural monopolies.

Acknowledgements

We acknowledge the support of the ESRC Electricity Policy Research Group. Authors are grateful for constructive comments from Catherine Morrison-Paul, Tim Coelli, Toru Hattori, Michael Kuenzle and anonymous referees. All remaining errors remain the responsibility of the authors.

Notes

¹For an overview of mergers and acquisitions activity in the European electricity supply industry, cp. Codognet et al. (2002).

²For example, Gilsdorf (1994, 1995), Filippini (1998), Salvanes and Tjøtta (1998) and Yatchew (2000).

³For a detailed description of outputs in electricity distribution, see Section IV.

⁴Sharkey (1982) presents numerous examples of subadditive cost functions to illustrate that neither economies of scale nor concavity are necessary conditions for subadditivity. As the existence of natural monopoly conditions in electricity distribution is an established notion, we are going to concentrate on more straightforward calculations rather than hypothetical exceptions.

⁵See Baumol (1977) and Baumol et al. (1982) for a detailed description of these concepts and sufficient conditions for natural monopolies.

⁶For an introduction, see Coelli et al. (1998).

⁷However, it should be mentioned that some degree of self-selection may exist as small firms with low QoS may have gradually been acquired by larger and more efficient firms.

⁸Estache et al. (2004) refer to a violation of standard assumptions (for example profit maximization or efficient production) in production economics due to regulation and public ownership.

⁹See Ray (1999, 2003), Balk (2001) and Orea (2002) for more detailed discussions.

¹⁰Jamasb and Pollitt (2003) in a DEA of efficiency of electricity distribution networks (in Norway, United Kingdom, Netherlands, Italy, Spain, Portugal) found a correlation coefficient of 0.99 between the efficiency scores when using PPP vs. using exchange rates. Furthermore, four out of seven countries share a common currency, bearing difficulties accounting for relative price level differences when using currency rates.

¹¹Applying the procedure suggested dropping all observations from Spain due to inconsistency of quality of service data with other countries. To avoid systematic biases, we dropped the Spanish utilities for the cost-only model as well.

¹²We suppressed the environmental variables constant to obtain better interpretable results.

¹³This analysis is based on the gross technical efficiency scores (Coelli et al., 1999). Calculating net technical efficiency scores as suggested by Coelli et al. reduces the efficiency differences between small and large firms but does not alter the relation in general.

¹⁴To estimate the confidence intervals of the technical efficiency scores, we applied the procedure suggested in Horrace and Schmidt (1996) and Kumbhakar and Lovell (2000). In Kumbhakar and Lovell (2000), formula (3.2.34) contains a typo. We thank William C. Horrace for bringing this to our attention. The interested reader may be referred to Horrace and Schmidt (2000).

¹⁵Applying the procedure of testing for economies of joint production (Denny and Pinto, 1978) indicates that this result of constant scale economies throughout the sample cannot be explained by scope economies. This seems to be counter-intuitive, as a comparable RTS measure for Model 2 can clearly be described by increasing economies of scope with firm size (see below).

¹⁶Again, we suppressed the environmental variables constant to obtain better interpretable results.

¹⁷The net technical efficiency scores are less different between small and large firms; but again, this relation still holds in general.

¹⁸We obtained this result by estimating a logged trend function in a scatter plot diagram of RTS and number of customers, setting the trend equation to one and solving for the number of customers (R ²= 0.2317).

¹⁹This result was obtained by estimating a logged trend function in a scatter plot diagram of economies of scope and number of customers, setting the trend equation to one and solving for the number of customers (R ²= 0.2477).