Doctoral thesis

Cost efficiency and regulation of Slovenian water distribution utilities : an application of stochastic frontier methods



Thèse de doctorat: Università della Svizzera italiana, 2006 (jury note: Summa cum laude)

English Over the last two decades it has become increasingly important to promote the efficiency and improve the performance of natural monopolies operating within network industries. In line with this, different regulatory approaches have been designed aiming at preventing the abuse of monopoly power and at the same time enhancing the performance of regulated firms. The most widely adopted incentive-based regulatory schemes involve price cap (RPI-X), revenue cap, and yardstick regulation models. These schemes aim to give firms an incentive for efficient production and cost reduction. However, due to the imperfect information available to the regulator there are some drawbacks related to the use of price-cap regulation since the regulator does not know a firm’s true costs. High costs may be due to a firm’s particular production situation or merely to its inefficiency. Thus, in setting the initial price level and the yearly efficiency factor X in the price-cap formula the regulator can use some form of cost-based benchmarking analysis. In this case, benchmarking is used to establish a larger information basis for more effective regulation that reduces the informational asymmetry between firms and the regulator. Hence, there is a close link between efficiency measurements and incentive-based price regulation. Today’s price regulation of Slovenian water distribution utilities resembles the rate-of-return regulation scheme. Nevertheless, the current Rules on Price Determination of Obligatory Local Public Utilities for Environmental Protection (2004) envisage the use of benchmarking in the price-regulation process and defining of the best-practice performance. However, the Rules have not yet been put into practice since the benchmarking method has still not been determined. In the thesis we consider the use of parametric frontier benchmarking methods and suggest how the results could be used in the price-regulation process. The main method employed is Stochastic Frontier Analysis (SFA), while the Corrected Ordinary Least Squares (COLS) method is used to cross-check the results. A translog frontier cost function is estimated based on an unbalanced panel data set of 52 utilities over the 1997-2003 period. The cost inefficiency estimates of Slovenian water distribution utilities are obtained by several different parametric frontier methods. The employed models differ in their assumptions, method of estimation and in their ability to account for firm-specific effects and distinguish between firm heterogeneity and inefficiency. The pooled model does not take into account the panel structure of the data and is therefore unable to separate unobserved heterogeneity from inefficiency. While conventional fixed and random effects panel data models take firm-specific effects into account in the estimation of inefficiency, they treat any time-invariant unobserved heterogeneity as inefficiency. They, too, are found to fail when it comes to separating heterogeneity from inefficiency. This problem is tackled by ‘true’ fixed and random effects models by adding in an additional term in the model which captures time-invariant and firm-specific effects and therefore separates these effects from inefficiency (Greene, 2002a, b). However, it remains debatable whether the time-invariant firm-specific effect should, in fact, be attributed to unobserved heterogeneity or to inefficiency. Mundlak’s (1978) formulation of the random effects model is also considered since it allows controlling for any correlation between unobserved heterogeneity and regressors. In our study it is found that the estimation results based on the conventional random effects models tend to highly overestimate cost inefficiency, while the true fixed effects (TFE) model seems to slightly underestimate it. Nevertheless, since the inefficiency estimates obtained by the TFE model closely correspond to the pooled model it is believed that these two models provide a better approximation of the actual cost inefficiency of Slovenian water distribution utilities, which is found to be close to or slightly above 20% on average. The TFE model is also found to perform the best with respect to the expected signs and significance of the regression coefficients. The inefficiency results indicate that significant cost inefficiency is present in Slovenian water distribution companies and that the utilities would have to considerably cut their costs in order to become efficient. This may be facilitated by a properly designed price regulation that introduces incentives for efficiency improvements. The inefficiency scores obtained from the different methods are, however, not found to be consistent in their levels and rankings of the utilities. A possible explanation of these inconsistent results can be found in the different ability of stochastic frontier methods to account for unobservable heterogeneity. However, since the regulator needs reliable estimates of the efficiency potential of a regulated firm this finding is particularly unwelcome. It is thus recommended to use the benchmarking results obtained by the SFA methods only as a starting point for providing information about the range in which the inefficiency score can be located. Alternatively, the estimated cost function can also be used to predict utilities’ costs, with this approach being in line with yardstick competition. Besides achieving cost efficiency, i.e. operating at minimum cost at a given size, important cost savings may result from achieving scale efficiency, i.e. operating at the size that minimises average production costs. The results from the different models in the latter case prove to be fairly consistent. Based on the obtained results, the presence of economies of output density and customer density in Slovenian water distribution utilities is confirmed. Therefore, it would be beneficial for the utilities if they managed to distribute larger volumes of output to their existing customers as well as to acquire new customers. With respect to economies of scale, medium-sized utilities are found to closely correspond to the optimal size of water distribution utilities in Slovenia. Economies of scale prevail in smaller utilities, implying they should consider expanding the scale of their operations through mergers. Conversely, large utilities are found to operate at levels where economies of scale are already exhausted. Overall, based on the results obtained it can be concluded that there is large potential for cost savings in Slovenian water distribution utilities. However, no evidence of any notable improvements being made can be found so far. The total factor productivity growth over the examined period is found to be around zero where technical progress is established but, on the other hand, no significant improvements in cost efficiency are found. In order to facilitate these improvements, a new regulatory framework is needed, where the choice should be made among incentive-based price regulation schemes. Rate of return regulation combined with benchmarking as proposed by the Rules on Price Determination of Obligatory Local Public Utilities for Environment Protection (2004) would be one of the appropriate alternatives.
  • English
License undefined
Persistent URL

Document views: 25 File downloads:
  • 2006ECO003.pdf: 5