Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Utilities' provision does not excite too many people. Nevertheless, the effectiveness of the supply of public transport (trains and busses), electricity, gas, water, waste, sports and recreation, medical and social services and urban development influences the shape of an economy and the well-being of its citizens. Frequently, the utilities are arranged by the municipalities which have organised themselves in intermunicipal utilities. As intermunicipal utilities are initially established by the municipal budget, they still constitute the municipal balance sheet. As such, municipalities could sell their share in the intermunicipal utilities to counterbalance financial setbacks (if e.g. budget surpluses or unused tax levels are depleted). Knowledge of the value of the utility seems essential. Still, 94% of the Dutch municipalities could not provide full insight in the value of its participations. Among the explanations, they attribute wrong (not up-to-date) information and wrong bookings of assets (Gerritsen, 2002). By using the net assets, De Witte and Moesen (2006) estimated the value of the Flemish intermunicipal utilities at 13 billion euros in 2003 (about 9% of Flemish GDP). Not a neglectible sector. Besides the lack of knowledge of the value, (public) utilities often face a poor corporate governance structure (De Witte and Moesen, 2007). For example in Flanders, by law mandatory in (public) utilities are required to be politicians. This raises doubts about the objectivity of the appointment (i.e. are they appointed because of intrinsic qualities or because of a political agenda?). In addition, the board of directors (which according to corporate governance rules is preferably small in order to be effective) is often too large and acts as a consolidation price for unelected politicians. Large boards provide incentives for free riding, complexity and low effectiveness. This in turn harms the efficiency of the utility. The quiet life Basically, the two problems are related to a third problem which is intrinsic to the nature of the goods. The utilities often have a natural monopoly characteristic in that the provision of the good is cheaper at a larger scale. As such, the public utilities are monopolists, or at least regional monopolists. Without facing competition and with captive customers, the monopolists lack incentives to provide the goods as effective and as efficient as possible. Hicks (1935) illustrated this by the quiet life in which the workers of a monopolistic company are unmotivated (e.g. longer coffee breaks, surfing on the internet, etc.). Leibenstein (1966) showed the problem by pointing to the X-inefficiencies in which the firms' managers overinvest, acquire unnecessary resources to expand (and show) their wealth or in which labour unions bargain for above-market wages. Moreover, a (regional) monopolist can easily make excessive profits by increasing the prices for the captive customers. To increase the performances of the regional monopolies, society (the principal) should create incentives for the utility (the agent). The incentives could firstly originate from the utilities which anticipate changes in society (when e.g. thanks to increased technological possibilities and to changed ideological preferences a threat of privatization arises). Secondly, regulators could provide incentives by setting maximum prices or maximum revenues (i.e. yardstick competition). Thirdly, a concerned public opinion could stimulate efficiency if performances are published (i.e. sunshine regulation). In a similar sunshine regulatory framework, the performance measure should be comprehensible (e.g. one composite indicator instead of a set of (contradictory) values), correct (e.g. account for diverging exogenous influences among the utilities) and robust to the model assumptions (e.g. they should not rely on a particular functional form which heavily influences the results). Especially the composite indicators could yield clear advantages for both the utilities, the regulator and the public opinion: (1) knowing at a single glance the performance, (2) comparing the overall performance of entities, (3) easiness to explain to outsiders and stakeholders, and (4) setting targets and guiding long term objectives easily. Methodological choice and issues In this thesis, we focus on composite indicators as a tool to provide incentives to regional monopolists. To develop composite indicators for the performance of the utilities, we start from the non-parametric efficiency measurement literature. Relying on only a few assumptions, the non-parametric models `let the data speak for themselves' as they do not require an a priori specified functional form on the data. As such, efficiency is defined as a relative concept in which the input usage and the output production of an evaluated observation is compared to best practice observations. The excessive input use, or output shortfall, relative to the best practice observation is considered as inefficiency. As efficiency is defined in terms of production possibilities, efficiency is technical. The focus on non-parametric models is not unchallanged. The parametric literature is well developed (e.g. Greene, 2007). It is for example relatively easy to explore stastical inference, capture endogeneity and allow for noisy data. In addition, parametric models have in comparison to non-parametric models a high rate of convergence and a low curse of dimensionality. Efficiency evaluations are possible by parametric (e.g. Corrected Ordinarily Least Squares) and semi-parametric (e.g. Stochastic Frontier Analysis) models. Nevertheless, the parametric literature has one major drawback on the non-parametric models. In particular, the models require the a priori specification of the production function. This is extremely problematic in many real life studies as the researcher often has no idea on this specification. A well specified parametric model (thus the same estimated model specification as the true underlying model) delivers superior results (in terms of rate of convergence) than a non-parametric model. However, a wrongly specified parametric model delivers inconsistent results. As particularly this true underlying model specification is typically unknown, the main focus of the thesis lies on non-parametric models. Nevertheless, in Chapter 5 we employ parametric models as a robustness test for the non-parametric estimations. There are three intricate issues in the non-parametric models. Firstly, by construction the relative performance evaluation is sensitive to outlying and atypical observations (as these best practice observations could heavily distort the efficiency scores). We deal with these observations by the robust order-m efficiency scores (see Chapter 1). Secondly, the utilities should only be compared to `comparable' utilities, in particular utilities with similar exogenous variables. For this purpose, we use the conditional efficiency estimates (see Chapter 1). Finally, non-parametric efficiency estimates depend on the observed data so that a bias between a true (unobserved) production process and the observed production process could arise. Bootstrapping algorithms provide a solution to estimate the bias (see Chapter 3). Structure and contributions By exploring and developing methodological advances in efficiency measurement, and by carefully examining particular issues in a regional monopolistic sector (i.e. the drinking water sector), this thesis could be of interest to both academics and practitioners. The academics will focus on the first part of the thesis, whereas the second part will potentially attract the practitioners. This first part explores the theoretical foundations of the non-parametric efficiency estimation. Alternatively, it can be considered as a general toolbox for regulators. In Chapter 1, we define the traditional Free Disposal Hull and Data Envelopment Analysis model. As has been outlined before, this traditional model suffers from some inconveniences. We show how to reduce the deterministic nature by considering partial frontiers and how to introduce heterogeneity incorporation by using conditional efficiency estimates. Besides these main topics, we examine panel data applications and non-oriented models. In addition, we propose a procedure to tackle sample size bias by using the robust order-m efficiencies. To make these contributions and the state-of-the-art survey comprehensible, we illustrated the procedures by the use of a real life data set. As is pointed out in Chapter 1, the traditional selection in non-parametric models of heterogenous inputs, outputs and exogenous variables influences the results. In general, the selection of inputs and outputs is rather straightforward and less vulnerable to discussion. However, the more important is the selection of exogenous variables. Evaluated entities operating in a favorable environment will obtain higher efficiency scores. To avoid similar discussions, Chapter 2 develops a model which estimates efficiency without the a priori selection of a set of exogenous variables. In particular, by reinterpreting the metafrontier literature, we propose a model which assesses efficiency while `fully' accounting for the operational environment. Outlying observations are particularly disturbing in non-parametric estimations. In Chapter 3, we suggest a model which detects and analyses a more extensive range of outlying observations. Indeed, the existing outlier detection models are only able to detect a specific range of outlying observations (e.g. super-efficient observations or units with a large leverage). The suggested procedure is able to detect a broad range of outlying observations. The detected observations could be deleted from the sample or could further investigated (as they are potentially the most interesting observations in that they present the best or worst practices).   The second part of the thesis applies the non-parametric approaches of the first part to the regional monopolistic drinking water sector (although the toolbox of Part 1 contains more tools than we apply in this thesis). In Part 2, we explain and develop some additional methodologies to tackle the specific research questions in this part. The drinking water sector is an extremely interesting and promising sector to study as (1) the debate is complicated by ideological opinions (e.g. private versus public water provision), (2) these ideological opinions created various regulatory models with a diverging effectiveness, (3) as supra-national bodies (e.g. European Union) do not seem to succeed in creating homogenous regulation (in contrast to other network sectors as railways or electricity), (4) as the responsibility of the water provision (mostly) belongs to the municipalities which often lack the resources to create a strong principal-agent setting, and finally (5) as in the future the importance of a (clean, cheap and efficient) water provision is expected to increase in importance. Studying technical efficiency in the water sector is attractive as we focus on the ability to avoid waste. In Chapter 4, by comparing several regulatory models as implemented in 5 countries, we evaluate the effectiveness of incentive mechanisms. By the use of a double bootstrap, we estimate the impact of several exogenous variables and explore some of them more in detail in Chapter 5 and 6 (in particular the scale, scope and incentive regulation). Chapter 4 develops also an alternative and bootstrapping inspired procedure to tackle sample size bias. In Chapter 5, we examine the influence of potential scale economies for both large and small utilities. To investigate the large utilities we consider the Dutch drinking water utilities. However, this complicates the analysis as we have to distinguish scale (or merger) economies from the structural efficiency gains in the sector. The outcome of the specially tailored non-parametric model is compared to parametric specifications. The small utilities are analyzed by considering Portuguese drinking water data. We propose a systematic framework to analyze the existence of scale and scope economies and to detect the optimal scale of operations. Chapter 6 analyses for the Dutch drinking water sector the influence of the public debate. In other words, could a public debate provide an effective alternative to (strict) regulation? If public opinion makers denounce the performance, profits or prices of the utilities, do the utilities change their behavior? We analyze this research question by developing a robust and conditional extension of a profit decomposition model. We relate the outcomes of the profit decomposition to the changes in the regulatory framework. Obviously, the procedures and results of this thesis have a broader application than the drinking water sector. The procedures can easily be adapted to sectors which allow defining input, output and contextual variables. The results could easily be translated to other network sectors (airports, ports, solid waste, etc.) or to sectors which face similar incentive issues (e.g. education, health sector, etc.). This in turn harms the efficiency of the utility. The quiet life Basically, the two problems are related to a third problem which is intrinsic to the nature of the goods. The utilities often have a natural monopoly characteristic in that the provision of the good is cheaper at a larger scale. As such, the public utilities are monopolists, or at least regional monopolists. Without facing competition and with captive customers, the monopolists lack incentives to provide the goods as effective and as efficient as possible. Hicks (1935) illustrated this by the quiet life in which the workers of a monopolistic company are unmotivated (e.g. longer coffee breaks, surfing on the internet, etc.). Leibenstein (1966) showed the problem by pointing to the X-inefficiencies in which the firms' managers overinvest, acquire unnecessary resources to expand (and show) their wealth or in which labour unions bargain for above-market wages. Moreover, a (regional) monopolist can easily make excessive profits by increasing the prices for the captive customers. To increase the performances of the regional monopolies, society (the principal) should create incentives for the utility (the agent). The incentives could firstly originate from the utilities which anticipate changes in society (when e.g. thanks to increased technological possibilities and to changed ideological preferences a threat of privatization arises). Secondly, regulators could provide incentives by setting maximum prices or maximum revenues (i.e. yardstick competition). Thirdly, a concerned public opinion could stimulate efficiency if performances are published (i.e. sunshine regulation). In a similar sunshine regulatory framework, the performance measure should be comprehensible (e.g. one composite indicator instead of a set of (contradictory) values), correct (e.g. account for diverging exogenous influences among the utilities)