at SciVerse ScienceDirect

Utilities Policy 21 (2012) 40e49

Contents lists available

Utilities Policy

journal homepage: www.elsevier .com/locate/ jup

A meta-regression analysis of benchmarking studies on water utilitiesmarket structure

Pedro Carvalho a, Rui Cunha Marques a,*, Sanford Berg b

aCenter for Management Studies (CEG-IST), Technical University of Lisbon, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisbon, Portugalb Public Utility Research Center (PURC), University of Florida, PO Box 117142, Gainesville, FL, USA

a r t i c l e i n f o

Article history:Received 28 September 2011Received in revised form27 December 2011Accepted 28 December 2011

Keywords:Scale economiesScope economiesMeta-regression analysisWater utilities

* Corresponding author. Tel.: þ351 218417981; fax:E-mail addresses: pedro.c.carvalho@ist.utl.pt (

ist.utl.pt (R.C. Marques), sanford.berg@warrington.ufl

0957-1787/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.jup.2011.12.005

a b s t r a c t

This paper updates the literature on water utility benchmarking studies developed worldwide, focusingon scale and scope economies. Using meta-regression analysis, the study investigates which variablesfrom published studies influence these economies. Our analysis yields several conclusions. The resultsindicate that there is a higher probability of finding diseconomies of scale and scope in large utilities;however, only the results for scale economies are significant. Diseconomies of scale and scope are morelikely to be found in publicly-owned utilities than when the ownership is private; as would be expected,multi-utilities are more likely to have scale and scope economies.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

In recent years, quantitative water sector studies have focusedon measuring performance and identifying factors affecting costs.Until the 1980s only a handful of benchmarking studies could befound in the literature (papers in academic journals, PhD disser-tations, working papers, chapters of books and books). In the 90sabout three dozen studies were published. By the end of 2010, morethan 250 studies were available compared to 2009, when Berg andMarques (2011) were only able to identify 190 studies. Past studiesuse cost or production functions to evaluate the performance ofwater utilities with several aims, such as examining the scale, scopeor density economies of a particular country or region, determiningthe influence of ownership and other exogenous variables on effi-ciency (e.g. Renzetti and Dupont, 2009), investigating the extentand impact of incentive systems and alternative governancemodels, and assessing performance and identifying best practices.Most of these water utility benchmarking studies use parametricmethods for efficiency estimation, although some apply non-parametric methods. Quantitative studies have proven to beextremely useful in introducing yardstick competition, promotingcost containment, identifying efficient prices, and encouragingquality of service improvements in water utilities (Rassier andEarnhart, 2011).

þ351 218417979.P. Carvalho), rui.marques@.edu (S. Berg).

All rights reserved.

The present study should be useful for academics as well as fora variety of stakeholders, including government officials respon-sible for water sector policy, water/wastewater regulators, andutility managers. By integrating the results of the growing numberof quantitative studies, the current study provides scholars witha rigorous evaluation of past research. Market structure andoptimal scale and scope have been investigated using a wide rangeof models and sample sizes. Understanding patterns of findings canhelp scholars utilize appropriate models in the future. Thoseresponsible for developing and implementing public policy willfind the results of the meta-analysis presented here to be a usefulcompilation of factors that need to be taken into account whenassessing empirical studies. For example, policy-makers who, in thepast, have been directly involved in proposals to merge publicwater utilities, with the aim of trying to achieve economies of scale,will be very interested in the sensitivity of results to the inclusion(or exclusion) of relevant variables.

The purpose of this study is to rigorously analyze the results ofpast studies examining scale and scope economies in the watersector. Inappropriate market structure (size distribution of utilitiesrelative to the market size) jeopardizes the attainment of organi-zational efficiency and reduces the value for money inmonopolizedmarkets (Bel and Warner, 2008). Although governments intervenein infrastructure to curb the exercise of market power and correctmarket failures (through regulation or competition policy), thepresence of scale and scope economies suggests that customersrequire protection from monopoly pricing in developed countries.

P. Carvalho et al. / Utilities Policy 21 (2012) 40e49 41

In developing nations, where water service is likely to be under-priced (and coverage low), inappropriate market structures canresult in higher costs than necessarydfurther exacerbating thefinancial problems of operators. In transition economies, identi-fying the appropriate scale for production can help policy-makersdevelop incentives for consolidation or decentralization.

Scale (or size) economies exist when an expansion in an outputcan be achieved with less than a proportionate increase in costs,whereas scope economies are present when the production costs oftwo or more products jointly produced are lower than when theyare produced separately by two specialized entities (Baumol et al.,1988). Thus, scale economies are related to the scale of productionand scope economies to the savings arising from the jointproduction of goods or services. Joint production could involvemeeting residential and industrial/commercial demands, operatingin different stages of the production process (bulk water anddistribution), or providing both water and wastewater services.

There is no consensus in the literature regarding the (1) optimalscale of water utilities, (2) existence of scope economies betweendifferent types of services (e.g. water and wastewater services), (3)extent of economies of vertical integration (scope economiesbetween the various stages of the production chaindas with bulkwater andwater supply to final consumers). This should come as nosurprise: most empirical studies only cover a single country, so theoperating conditions, legal context, income levels, and other factorswill differ across studies. Thus, controlling for these factors isessential if the results of past studies are to be systematicallyassessed. Some empirical studies find scale and scope economies;the existence of high (short run) fixed costs in the water sector mayreflect sharing of administrative and procurement costs. However,other studies find diseconomies and explain the higher costs asstemming from network complexity, and the bureaucratic proce-dures associated with large utilities (Abbott and Cohen, 2009).

Perhaps because of these ambiguities (and unique circum-stances related to water resource access, national income

Table 1Market structure in the Europe.

Country Water utilities(no.)

Average population(no./utility)

Austria 5000 1640Belgium 28 375,000Czech Republic 1211 8505Denmark 2622 2059

England and Wales 25 2,148,000Finland 1400 3786France 19,300 3337Germany 6000 13,667

Greece 1000 11,000Holland 10 (water) 1,650,000

443 (collection) 66025 (sewage treatment) 37,246

Ireland 3051 1409Italy 91 648,352Luxembourg 106 4528Northern Ireland 1 1,700,000Norway 1616 2908

Portugal 300 31,278

Romania 2000 7700Scotland 1 5,100,000Spain 8100 5556Sweden 294 30,612

Switzerland 3000 2467

differences, and different legal systems), quite distinct water utilityarrangements and different results between studies can be foundaround the world. Even in a single nation, suppliers can range fromvery small water utilities providing services to small villages tolarge utilities providing services to many customers in a largemunicipality or a region. Moreover, in some countries the drinkingwater supply is provided as a single service (e.g. the Netherlands);in others it is provided together with wastewater (e.g. France) andalso with other services, such as urbanwaste, electricity or gas (e.g.Germany). There are also vertically integrated water utilities thatare responsible for the wholesale and retail (distribution) segments(e.g. Spain), and others that deal with the wholesale or retailsegments separately (e.g. Portugal). Table 1 presents the marketstructure in Europe. No clear patterns are observed in terms of sizeof scale and scope services. Developing countries have a similarrange of institutional arrangements, with relatively low coverageby wastewater networks.

After briefly surveying previously published benchmarkingstudies addressing scale and scope economies, the current researchinvestigates, by means of a meta-regression analysis, which vari-ables or characteristics of samples from published studies have thegreatest influence on estimated size of scale and scope economies,based on thirty-five and thirteen studies. The sample encompassesall the benchmarking studies (published and unpublished) dealingwith performance scores based on production or cost estimates.The sample comprises all the articles published in academic jour-nals in the following fields: Economics, Public Policy, PublicAdministration, Political Science and Environmental and WaterPolicy. The sample also includes articles in edited books, chapters ofbooks, PhD dissertations and in working paper series, includingthose in Policy Research Working Paper Series or Social ScienceResearch Network. Section 2 presents lessons reported in theliterature, followed by Section 3 which introduces the meta-regression analysis, the key concepts (on economies of scale andscope), and explains the variables utilized in the study. Section 4

Scope of services Vertical integration

Water and wastewater IntegratedWater and wastewater Mostly unbundledWater and wastewater IntegratedWater and wastewater(two multi-utilities)

Integrated

Water and wastewater IntegratedWater and wastewater IntegratedWater and wastewater IntegratedMulti-utilities and waterand wastewater separated

Integrated

Water and wastewater Mostly integratedWater UnbundledWastewaterWastewaterWater and wastewater Mostly integratedWater and wastewater IntegratedWater and wastewater Mostly integratedWater and wastewater IntegratedWater and wastewaterand other municipal activities

Integrated

Water and wastewaterand sometimes other activities

Mostly unbundled

Water, wastewater and urban waste UnbundledWater and wastewater Mostly integratedWater and wastewater Mostly integratedWater and wastewater. There are 5multi-utilities

Integrated

Multi-utilities Integrated

P. Carvalho et al. / Utilities Policy 21 (2012) 40e4942

presents and discusses the results. Concluding observations areoutlined in Section 5.

2. Lessons from literature: a short survey

Ford and Warford (1969) were among the first authors toinvestigate scale economies in the water sector. They tried toidentify an appropriate specification of the cost function for waterutilities in the UK and argued that amalgamation of firmswould notnecessarily lead to a lowering of average costs. In the following twodecades (70s and 80s), the literature focused primarily on issues ofownership, researching whether the type of ownership (public orprivate) influenced performance. Past research is inconclusiveregarding this issue. While some papers provided empiricalevidence that publicly-owned water utilities have better perfor-mance (e.g. Bruggink, 1982), other studies concluded that privatewater utilities outperform the public ones (e.g. Crain andZardkoohi, 1978); others did not find conclusive evidence thatone ownership regime outperformed the other (e.g. Byrnes et al.,1986). Of course, model specification differed across these earlystudies; in addition, controlling for different governance (andincentive) systems was difficult due to inadequate model specifi-cation and data availability.

Some studies of scale and scope economies in the water sectorwere found in our census of studies, but all refer to the US and theUK. For example, Knapp (1978) reported strong economies of scalein the wastewater treatment in England and Wales. Fox and Hofler(1986) observed economies of scale in water distribution anddiseconomies of scale in the production of water in the US. Kim(1987), in a study of US utilities, did not find significant econo-mies of scale in drinking water supply, but found some economiesof scale for non-residential water supply and some diseconomiesof scale for residential water supply. During this period (70s and80s), the first studies examining scope economies appeared. Forinstance, Hayes (1987) observed economies of scope in the US forsmall, vertically integrated companies, but not for the large ones.In another US study, Kim and Clarke (1988) detected scope econ-omies for joint production of residential and non-residential watersupply.

In the two decades that followed (the 1990s and 2000s), thenumber of benchmarking studies published about water sectorcost and production functions more than doubled. Fig. 1 showsthe growth of studies examining scale and scope economies, withthe number published each year shown on the vertical axis.However, there was still no consensus regarding the optimal sizeof water utilities or the extent of scope economies (Abbott andCohen, 2009). For example, Ashton (1999), among others, found

Fig. 1. Studies of scale and sc

economies of scale in the UK, but Saal and Parker (2000) andother authors obtained opposite results. Subsequently, a studyfrom Stone & Webster Consultants (2004) concluded that thebiggest companies in England and Wales (water and wastewatercompanies) presented diseconomies of scale and the remainingsmall water-only companies displayed economies of scale. LaterSaal and Parker (2004) found constant returns to scale and Saalet al. (2007) revisited the issue, finding that large water andsewerage utilities are characterized by diseconomies of scale inEngland and Wales.

For the US, as in the two previous decades, strong economiesof scale were found (Bhattacharyya et al., 1994; Shih et al., 2006).However, Torres and Morrison (2006) observed economies ofscale for small utilities but diseconomies of scale for the largestutilities. Similar results were obtained for other countries,including Italy (Fabbri and Fraquelli, 2000; Fraquelli andGiandrone, 2003; Fraquelli and Moiso, 2005), France (Garcia andThomas, 2001), Japan (Mizutani and Urakami, 2001), SouthKorea (Kim and Lee, 1998) and South Africa (Tsegai et al., 2009).Recently, Nauges and Berg (2008) found economies of scale inColombia, Moldova and Vietnam for small and medium utilitiesbut not in Brazil. In other countries, such as Portugal (Correia andMarques, 2010; Martins et al., 2006), Germany (Sauer, 2005), inLatin America (Ferro et al., 2010), Peru (Corton, 2010), Spain(Prieto et al., 2009) and Canada (Renzetti, 1999), most studiesfound economies of scale in providing water services. However,for Switzerland, Baranzini and Faust (2009) found diseconomiesof scale for multi-utilities. The Appendix lists thirty-five studiesestimating economies of scale that made up the sample utilized inthe current study.

Regarding research on economies of scope, again results in theliterature are mixed. Of the studies examining scope economiesbetween water and wastewater services, a large proportion findsscope economies (e.g. Fraquelli et al., 2004; Link, 1993; Hunt andLynk, 1995; Fraquelli and Moiso, 2005; Martins et al., 2006),although some analysts conclude that the savings are greater forsmall companies than for large ones. However, other studiesconcluded just the opposite (Saal and Parker, 2000; Stone &Webster Consultants, 2004). For example, the study by Stone andWebster Consultants reports that in England and Wales no scopeeconomies exist between water and wastewater. Also, Correia andMarques (2010) observed decreasing economies of scope inPortugal. Similarly, Prieto et al. (2009) found no evidence ofstatistically significant scope economies in Spain.

Apart from the activities of water supply and wastewatercollection, the comprehensive sample identified studies quanti-fying economies of scope between other activities in multi-

ope economies per year.

P. Carvalho et al. / Utilities Policy 21 (2012) 40e49 43

utilities, such as electricity, urban waste or gas services. Forexample, Fraquelli et al. (2004) and Piacenza and Vanonni (2004)found significant scope economies for multi-utilities in Italy.Other studies investigated scope economies associated withvertical integration and suggest the existence of scope economiesin joint retail and wholesale segments, especially for the smallestwater utilities (Hayes, 1987). For example, Stone & WebsterConsultants (2004) observed benefits in vertical integration inthe water supply in England and Wales; Torres and Morrison(2006) reached the same conclusions for the water sector in theUS. Urakami (2007) examined the Japanese water supply industryand noted the existence of economies of vertical integrationbetween water intake-purification and the water distributionstage. In contrast, Garcia et al. (2007), using a sample of USutilities, concluded that separation maybe advantageous in somecircumstances, and that economies of vertical integration are notsignificant except for the smallest utilities. Appendix 2 lists thethirteen studies found in the literature that empirically examinedeconomies of scope.

Although there is no consensus in the literature regarding theoptimal size of water utilities or the existence of economies ofscope among various activities, it is generally agreed that smallwater utilities providing only one service or that are not verticallyintegrated have significant scale and scope economies that seldomare achieved due to relatively low levels of output (reflectingunderlying demand within the service territory); large or verticallyintegrated utilities seem to have scale and scope diseconomies atthe levels of output they produce. The literature reports a widerange of a maximum number of connections where economies ofscale were not exhausted. For example, Fraquelli and Giandrone(2003) found values of 100,000 in Italy, while Mizutani andUrakami (2001) found 766,000 (in Japan) and Fraquelli and Moiso(2005) estimated the leveling off point at one million for Italy.There is less divergence of views in relation to wastewater activi-ties, in part because these activities have been subject to lessresearch or have been examined in the context of businessesundertaking both water supply and wastewater activities (e.g.Ashton, 2000). Both Knapp (1978) and Renzetti (1999), who areexceptions, found that economies of scale exist in the wastewatersector. The results reported in the literature are more consistentwith regard to economies of density and, in general, the studiespoint to both their existence and to their importance (Antonioli andFilippini, 2001; Zoric, 2006).

These findings are not surprising, given differences betweenthe geographic, hydrologic, topographic circumstances in differentcountries or regions or to differences in ownership, degree ofcorporatization, and to the legal framework for providing theactivities of the water sector (including the existence, or absence,of regulatory or ministerial oversight). Also density, customerincome levels, distance from water sources, and a variety of otherfactors may (or may not) have been adequately controlled for inparticular studies. However, to a certain extent it is clear thatthere are savings from the use of shared resources and fromreduced administrative and procurement costs per customer; theissue is whether such economies outweigh the greater costsassociated with network complexity, and possibly the bureau-cracy accompanying the increase in the size of the utility. Theliterature is also inconclusive as far as the issue of ownership isconcerned. Recently Bel et al. (2010) conducted a meta-regressionanalysis examining privately-owned and publically-owned waterdistribution services with a sample of twenty-seven studies; theyfound no systematic support for lower costs with privateproduction. They did not investigate the extent of economies ofscale and scope, so it is useful to apply the technique to thisimportant issue.

3. Meta-regression analysis

Narrative literature reviews generally identify all the availablestudies, listing conclusions for each one. However, such anapproach does not consider model specification, number ofobservations, and other factors affecting the quality of the studies,so the conclusions from such surveys tend to be impressionisticrather than definitive. This problem can be mitigated by applyinga meta-regression analysis. This technique combines the results ofseveral studies that address a particular research topic and, throughstatistical methods, tries to find the true relationship betweendifferent variables and the results reported in the different studies(Stanley and Jarrell, 1989). The present study examines the influ-ence of several variables on the results found in benchmarkingstudies that address the issue of scale and scope economies in thewater services sector.

According to Stanley and Jarrell (1989) a meta-regression anal-ysis can be performed using the following model:

bj ¼ bþXK

k¼1

akZjk þ ej ðj ¼ 1;2;.; LÞ (1)

where bj is the reported estimate of b of the jth study of the liter-ature consisting of L studies, b is the “true” value of the parameterunder examination, Zjk are the meta-independent variables thatmeasure relevant characteristics of an empirical study, and ak arethe coefficients related to these independent variables whichreflect the systematic biasing effect of particular characteristicsstudied, and ej is the meta-regression disturbance term.

In this study we apply a random effects meta-analysis thatestimates the extent to which several covariates explain hetero-geneity in the dependent variables (in this case, scale and scopeeconomies) between studies. For this purpose, a weighted normal-error regression model was performed in which an additivebetween-study variance component s2 is estimated. First themaximum-likelihood estimates of the ak parameters assumings2 ¼ 0 are obtained by weighted regression, and then a momentestimator of s2 is calculated using the residual sum of squares(Sharp, 1998).

For situations where the number of studies is particularly small,it is advisable to apply an additional test (permutation test), toobtain better results with respect to the p-values associated withindependent variables. Permutation tests provide a non-parametricway of simulating data (by Monte Carlo simulation) under the nullhypothesis (Higgins and Thompson, 2004).

The statistical significance of the coefficients associatedwith theindependent variables will allow us to examine the influence of thisvariable on the dependent variable. This meta-regression analysisconsiders as dependent variable bj, the corresponding values ofestimated overall (or aggregate) scale economies (SL) and thedegree of economies of scope (SC) obtained by the studies in thesample (here, a sample that approaches a census).

According to Baumol (1976) and Panzar and Willig (1977) thedegree of scale economies (SL) for the multi-product water utility isdefined as:

SLðyÞ ¼ CðyÞPiyiMCi

¼ 1Pi

3Cyi(2)

where C is the total cost of producing outputs y,MCi is the marginalcost with respect to the ith output and 3Cyi ¼ vlnC=vlnyi is the costelasticity of the ith output. The existence of (dis) economies of scaleis assessed according to whether SL(y) is greater than, equal to, orless than unity. If SL > 1 the water utility operates with economiesof scale (or size); if SL < 1 the water utility operates with

P. Carvalho et al. / Utilities Policy 21 (2012) 40e4944

diseconomies of scale. If, on the other hand, SL ¼ 1, that means thatthe water utility operates at constant returns to scale.

Moreover, economies of scope are related to the fact that jointproduction costs are lower than the sum of the production costs forseparate specialized water utilities, and the degree of economies ofscope (SC) is defined by:

SC ¼Pn

i¼1 Cðyn�iÞ � CðyÞCðyÞ (3)

where C (yn�i) ¼ C (y1,., yi�1, 0, yiþ1,., yn)Here, if SC > 0 the water utility faces economies of scope and if

SC < 0 the water utility presents diseconomies of scope.The meta-regression analysis utilized Stata software. The model

included the following variables for each study: sample size(number of water utilities analyzed), year of publication, number ofyears studied, the GDP of the country, the continent where theutilities are located, the estimation method used in the study, thetype of utilities involved (whether or not they are multi-utilities),publication type of the study, the average size of utilities, theprimary ownership-type for firms in the study sample and theexistence (or nonexistence) of regulation. A positive sign of thecoefficient ak indicates that the corresponding explanatory variabletends to be larger, thus increasing the reported scale or the scopeeconomies; if the coefficient presents a negative sign, it means thatthe explanatory variable tends to negatively affect estimated scaleor scope economies.

The variable GDP of the country took on the values of GrossDomestic Product (GDP) reported for 2009. This variable reflectsthe country’s standard of living. The variable Continent is a dummyvariable with assigned values from 1 to 5: 1 corresponds to theContinent with the highest GDP (United States/Canada) and 5 to theContinent with lowest GDP (Africa). The value 2 was assigned toEurope, 3 to Asia, and 4 to South America/Central America. Themeta-regression included the estimation method used in thestudies, the type of utilities involved (whether they were multi-utilities, value 1, or not, value 0) and the publication type of thestudy. Each of the variables captures differences in the studies. Forthe estimation method, the dummy variable takes on the value 1for studies that apply several estimation methods and takes on thevalue 0 for studies that apply only one estimation method. Theestimation methods observed in the literature included OrdinaryLeast Squares (OLS), Seemingly Unrelated Regression (SUR),

Table 2Meta-regression estimates.

Variables Dependent variable: estimates of scale

Coefficient Standaerrors

Sample size (No. of water utilities analyzed) 0.000030 0.0001Year of Publication 0.002 0.012No. of years studied 0.0012 0.0084GDP of country �3.46e�8a 1.41e�

Continent �0.096 0.065Estimation method �0.20b 0.10Type of utilities 0.031 0.098Publication Type �0.095 0.093Average size of utilities �1.12e�9b 8.55e�

Ownership �0.018 0.13Regulation 0.07 0.17Intercept 1.65 0.25R2 0.177Joint test for all covariates With

Knapp-Hartung modificationF(9.25) ¼ 1.65Prob > F ¼ 0.1543

N 35

a Significant at 5 percent level.b Significant at 10 percent level.

Nonlinear Seemingly Unrelated Regression (NLSUR), MaximumLikelihood Estimation (MLE), Generalized Method of Moments(GMM) and Restricted Least Squares (RLS). When each variable wastested separately, the coefficients lacked statistical significance, sothe final model does not consider them individually.

The meta-analysis took other factors into account as well. Forexample, publication venues have different standards, with journalreviewers and dissertation committees requiring more rigor instudies on which they pass judgment. The dummy variable forpublication type takes the value 1 for Theses and Journals and thevalue 0 for Reports, Chapters and Working Papers. The sign on thiscoefficient provides an indication of whether economies are morelikely to be found in more technically-refined empirical analyses. Inaddition, we examined the influence of the average size of utilities(represented by the amount of water billed in cubic meters), theinfluence of ownership (if mostly public, value 1, or mostly private,value 0) and the existence or nonexistence of regulation (if there isa sector specific regulator, value 1, or if not, value 0). Althoughmonopoly utilities generally face some type of oversightdeven ifby locally elected officials or a water ministry, this variable isintended to capture the impact of more formal national regulation.

4. Results and discussion

Since sample sizes in this study are relatively small, we appliedthe permutation test, which supports the results presented inTable 2. Table 3 provides the p-values associated with each inde-pendent variable. The first column shows the permutation p-valueswithout an adjustment for the multiplicity of variables, which aresimilar to those obtained without using the permutation test, andthe second columnpresents the p-values adjusted formultiplicity ofvariables. This adjustment consists in comparing the observed tstatistic for every covariate with the largest t statistic for any cova-riate in each random permutation (Harbord and Higgins, 2008).

Regarding economies of scale, the results indicate that thecoefficient of the variable sample size (number of utilities analyzed)is positive. This means that when the sample size is greater,economies of scale are more likely to be found. Although the valueof this coefficient is not statistically significant to a confidence levelof 95%, we would expect a relationship since, in general, thesamples are composed of many utilities which are generallysmaller; thus, there is greater potential for scale economies. After

economies (SL) Dependent variable: estimates of scope economies (SC)

rd Coefficient Standarderrors

0 �0.00011 0.000120.018 0.0080.03 0.047

8 3.53e�9 1.73e�8

0.0037 0.13�0.29 0.220.030 0.18�0.23 0.22

10 �2.69e�9 1.92e�9

�0.068 0.240.49 0.440.39 0.400.170F(9.3) ¼ 1.19Prob > F ¼ 0.493213

Table 3p-values associated with each independent variable using the permutation test.

Dependent variable: estimates of scale economies (SL) Dependent variable: estimates of scope economies (SC)

Unadjusted Adjusted Unadjusted Adjusted

Sample size (No. of water utilities analyzed) 0.698 1.000 0.260 0.657Year of Publication 0.927 1.000 0.130 0.446No. of years studied 0.940 1.000 0.731 0.990GDP of country 0.024 0.203 0.646 0.970Continent 0.143 0.692 0.552 0.928Estimation method 0.063 0.418 0.152 0.262Type of utilities 0.930 1.000 0.541 0.921Publication Type 0.379 0.971 0.980 1.000Average size of utilities 0.083 0.779 0.396 0.812Ownership 0.901 1.000 0.956 1.000Regulation 0.671 1.000 0.562 0.934

P. Carvalho et al. / Utilities Policy 21 (2012) 40e49 45

adjusting for permutation testing, there remains some weakevidence that the number of utilities analyzed has influence on thedegree of economies of scale. The adjusted p-value of 1.000 givesthe probability under the null hypothesis (that all regressioncoefficients are zero) of a T statistic for any of the covariates being asextreme as the one observed for the covariate “number of utilitiesanalyzed”. However, the opposite result is obtained for scopeeconomies (although without statistical significance to a confi-dence level of 95%), which suggests that the samples with greaternumber of utilities are more likely to have diseconomies of scope.

A positive signal was determined for the year of publication bothfor scale and scope economies. It means that the most recentstudies are more likely to find economies of scale and economies ofscope. This can be related to some consolidation of the water sectormarket structure overtime or to the fact that early studiespresumably used less sophisticated estimation techniques andsimpler models (reflecting limited data availability). However, thestatistical significance of the results obtained is weak at a confi-dence level of 95%.

The coefficient for the variable “number of years studied” hasa positive sign for both cases. This means that studies that analyzea greater number of years are more likely to find scale and scopeeconomies. However, again, these coefficients are not statisticallysignificant at a confidence level of 95%. The coefficient for countryGDP has a negative sign and it is statistically significant for the caseof scale economies at a confidence level of 95%. This result suggeststhat countries with higher GDP are more likely to exhibitdiseconomies of scale. This finding also indicates that in countrieswhere the standard of living is lower, there are probably manyutilities with significant economies of scale to be exploited as theyobtain the financial resources required for network expansion andincreased available hours. That point is corroborated by the nega-tive value of the coefficient of the variable Continent, correspond-ing to the higher value of GDP to the lowest value of thisexplanatory variable. Yet, for the case of economies of scope, thecoefficients of the variables GDP and Continent have a positivevalue. This result indicates that despite not being statisticallysignificant, there could be a greater probability of finding scopeeconomies in more developed countries. Since few of these studiesinclude service quality as an output, the role of GDP is mixed.Developed countries tend to meet World Health Organizationstandards for water quality and the treatment levels for wastewatermeet strict standards. Neither is the case for low income nations, sothat interpreting the role of income levels on estimated economiesof scale and scope is not straightforward.

Clearly, the application of more sophisticated estimation tech-niques leads to improved modeling. How do results from studiesusing single estimation techniques differ from those employingmultiple approaches? For the estimation method adopted, thecorresponding coefficient is negative for both cases, but only

statistically significant to a confidence level of 90% for the case ofscale economies. This means that studies that apply multiple esti-mation methods have a higher probability of finding scale andscope diseconomies.

Regarding the type of utilities studied, it seems that the corre-sponding coefficient is positive for both cases, which means thatmulti-utilities are more likely to have scale and scope economies.Finding economies of scope is more likely in multi-utilitiesbetween different types of services (e.g. water and wastewaterservices) than finding economies of vertical integration in theutilities that provide only one service. This reflects typical resultsfound in the literature, although, again, without statistical signifi-cance at a 95% confidence level.

As for the type of publication, the corresponding coefficient isnegative for both cases, but (again) without statistical significanceat a confidence level of 95%. It seems that scale and scopediseconomies are more likely to be found in studies published inTheses and Journals. In their examination of the impact of priva-tized production, Bel et al. (2010) concluded that there is a publi-cation bias related to privatization studies: “papers obtainingsignificant cost savings are more likely to be published” (p. 573).However, one could argue that unpublished papers have not passedthe hurdle of critical external reviews, so the differences in resultscould be a reflection of reality rather than some presumed inves-tigator bias. In the case of the results reported here, there is noreason to think that there is a systematic journal acceptance bias infavor of studies that find scale and scope diseconomies.

The variable average size of utilities has a negative sign for bothcases, but it is only statistically significant at 90% confidence levelfor the case of scale economies. This supports typical findings in theliterature: there is a higher probability of finding diseconomies ofscale and scope when the sample involves large utilities.

As to the ownership variable ownership, the coefficient isnegative for both cases. This suggests that publicly-owned utilitiesare more likely to display scale and scope diseconomies. Althoughthe value of this coefficient is not statistically significant ata confidence level of 95% for either case, this finding is consistentwith private utilities enjoying greater scale and scope economies.This result contrasts with the meta-regression study by Bel et al.(2010) which finds no empirical evidence of cost savings fromprivate provision of water distribution. One explanation may bethat given the prevalence of state-owned enterprises in water(including municipal ownership), there is some endogeneity in thedecision to privatize: utilities with scale and scope economies aremore likely to be privatized.

Finally, regarding the regulation variable, the coefficient ispositive for both cases but not statistically significant at a confi-dence level of 95%. This result provides some support for regulatedenvironments being associated with utilities having scale andscope economies.

P. Carvalho et al. / Utilities Policy 21 (2012) 40e4946

5. Conclusions

The above analysis yields some interesting conclusions that havenot been emphasized in the quantitative literature on water sectoreconomies of scale and scope. The results show that there is a higherprobability of finding diseconomies of scale and scope in large utili-ties, although statistical significance was only found for scalediseconomies. In addition, countrieswith higherGDP aremore likelyto have utilities that exhibit (statistically significant) diseconomies ofscale, while the opposite pattern appears for economies of scope.There are several possible reasons for the multi-product results. Inthe water (and wastewater) sector the transportation cost is quitehigh (when comparing with other infrastructure services) not onlybecause of the pipes but also due to theweight ofwater that needs tobe pumped. So a municipal utility created under a local governmentumbrella may (for political reasons) decide to incur higher produc-tion costs, while meeting service demands within governmentalboundaries. Network expansion is one factor affecting cityboundariesda politically potent issue.

In addition, in the water sector, annualized capital costs are verysignificant compared to the operation and maintenance costs(Marques, 2008). Finally, water networks become very complexwith increasing size (as topology and conditions of existing infra-structure affect network operations). For example, network main-tenance and operation costs in a huge urban center are muchhigher when compared with rural systemsdgiven the costs ofmaintaining networks that have been constructed under highdensity transportation corridors. Therefore, when two water utili-ties consider a merger they face a higher cost of transportation andoperation costs (which may not be balanced by savings throughlarger treatment plants). Given the vintages of pipes, energyexpenses, the cost of the adapting each utility’s networks intoa more comprehensive system, consolidation can be expensive. Inaddition, networks have small diameters in the extremities(precisely where the network inter-connections will tend to bedeveloped), affecting both managerial and maintenance costs. Ofcourse, the potential roles played by these factors depend on theparticular features of each water utility: for example, the location ofwater sources and access to bulk water are quite relevant.Furthermore, having a core group of talented engineers andfinancial experts can enhance performance of utility sub-divisions.

Although not statistically significant, publicly-owned utilities aremore likely to have scale and scope diseconomies than when theownership is mostly private. In addition, multi-utilities are morelikely to exhibit scale and scope economies. These patterns suggestthat, based on empirical results reported in the literature, economiesof scope in multi-utilities are more pronounced between differenttypesof services (e.g.waterandwastewater services) thaneconomiesof vertical integration in the utilities that provide only one service.

Finding definitive results in meta-regression analysis requiresthat all the factors characterizing individual studies be included inthe model. Furthermore, the linearity of the model (assumed inmost analyses of multiple regression-based econometric studies)ignores possible non-linearity in the relationships among theexplanatory variables. Nevertheless, periodic examination ofempirical studies provides a systematic check on results to date. Asnew databases become available, and additional variables areincluded in cost and production function models, our under-standing of causal links between inputs and outputs will improve.

Here, the focus has been on economies of scale and scope, topicsthat are central to decisions to consolidate current operators or todecentralize operations. However, the decision-relevance of thescholarly literature is still an openquestion. Is it obvious that a utilityoperating ina regionof diseconomiesof scale shouldbe split up, as inthe case of Manila?Were the resulting performance gains inManila

due to the availability of yardstick comparisons used by regulators,privatization, or to achieving an “optimal” size? A similar case arisesin Kampala, Uganda,where themanagement of the systemhas beendivided into fourdistribution territories. The rationalewasnotbasedon a technical econometric study identifying optimal scale, but thelikelihood that performance monitoring and high powered incen-tive plans could dramatically improve water utility performance inthe city: and the evidence supports the strategy (Mugisha and Silver,2011). The strategy itself was based on having a comprehensivebenchmarking system for the parent utility.

A similar set of policy issues arisewhenconsideringeconomies ofscope. Should two separate organizations be maintained for waterand wastewater when a coefficient of a study suggests that econo-mies of scope are being missed? Although fewwould argue against“evidence-based” decision-making, the weight given to facts isunlikely to be the deciding factor for those responsible for watersector policies. A strong case can be made that quantitative studiesare necessary, but not sufficient, for making strategic decisionsregarding consolidation/disaggregation or adding other productlines to a water utility’s portfolio. Issues of financial sustainability,political feasibility, governance, managerial incentives, and profes-sional capacity are more likely to drive operational decisions.

Nevertheless, empirical studies represent an important tool thatcan be utilized by both policy-makers and those implementingpolicy. The present study documents the roles of different factors inthe coefficients obtained in past studies of economies of scale andscope. Thus, empirical studies should be drawn upon only to theextent that underlyingmodels arewell-specified and the results arerobust and fully understood.Oneof the ethical principles adoptedbyphysicians is “Do no harm”. The same principle applies to thoseconducting empirical analyses. In a classic article, Leamer (1983)calls for taking the “con” out of econometrics: quantitative resultscan be fragileddepending on assumptions regarding model speci-fication. Here, we have demonstrated that across a wide range ofstudies, particular factors affect the findings related to economies ofscale and scope. Aware of the strengths and limitations of differentmethodologies, scholars maybe in a better position to ensure thattheir empirical results are not misinterpreted or misused.

An interview with John Briscoe (2011) identifies the discon-nection between national policy-makers, advocates, internationalcivil servants and researchers as creating lost opportunities. He seesthe need for water policy to recognize that a broader set of socialchallenges warrant our attention and that the timing of dramaticinitiatives seldom depends on empirical studies. Briscoe concludeshis interview with some words of wisdom:

So I see a great challenge in building a new water intelligence,one which has learned that there is a big difference betweenidea and practice, but one which seeks to bring new ideas e butideas which will work e to address this generation of chal-lenges. And one which communicates far better than the prac-tical water communities today manage to do! . We have tolearn that political capital is limited, and challenges are many.The great advances are made in those special moments whenpolitical possibility aligns with our agenda. We have to learn tostrike when that iron is hot (and not keep thinking that we arethe ones who will heat the iron!).

Thus, quantitative studies of water and wastewater utilities canidentify opportunities for consolidation and for disaggregation. Tothe extent that scholarly studies document relatively weakperformance, they can help establish a policy agenda. However, weprobably need to exhibit a greater sense of humility when wederive implications from estimated coefficients. Ultimately, polit-ical factors will play a dominant role in policy-decisions related toeconomies of scale and scope.

Appendix 1. Literature on scale economies

Study Waterutilities(no.)

Years (no.) Country Estimationmethod

Scopeof services

Publicationtype

Average size(106 m3/year)

Ownership Scale economies

Fox and Hofler (1986) 176 1981 (1) US OLS Water Journal 35.8 Mostly public 0.886Kim (1987) 60 1973 (1) US SUR Water Journal 1.9 Mostly public 0.992Kim and Clarke (1988) 60 1973 (1) US MLE Water Journal 1.9 Mostly public 0.992Bhattacharyya et al. (1995) 221 1992 (1) US Several Water Journal 60.7 Mostly public YesKim and Lee (1998) 42 1989e1995 (7) Korea SUR Water Journal 1.0 Public 1.25Ashton (1999) 20 1991e1996 (6) UK Several Water Working Paper 57.5 Private 0.963Renzetti (1999) 77 1991 (1) Canada Several Water

and wastewaterJournal 8.1 Mostly public 1.24

Ashton (2000) 10 1989e1997 (9) UK SUR Waterand wastewater

Journal 57.5 Private 1.475

Fabbri and Fraquelli (2000) 173 1991 (1) Italy SUR Waterand wastewater

Journal 18.9 Mostly public 0.99

Antonioli and Filippini (2001) 32 1991e1995 (5) Italy OLS Water Journal 6.8 Mostly public 0.95Garcia and Thomas (2001) 55 1995e1997 (3) France GMM Water Journal 0.6 Mostly private 1.002Parker and Saal (2001) 39 1985e1999 (15) UK (E&W) SUR Water

and wastewaterJournal 60.0 Private 0.853

Mizutani and Urakami (2001) 112 1994 (1) Japan SUR Water Journal 66.6 Public 0.911Fraquelli and Giandrone (2003) 103 1996 (1) Italy RLS Wastewater Journal 1.0 Mostly public 1.225Sauer (2003) 59 2001 (1) Germany

(East and West)Several Water Working Paper 1.3 Mostly public 1.289

Fraquelli et al. (2004) 90 1994e1996 (3) Italy NLSUR Waterand wastewater

Journal 18.9 Mostly public 1.103

Piacenza and Vanonni (2004) 90 1994e1996 (3) Italy NLSUR Waterand wastewater

Journal 11.0 Mostly public 1.094

Stone and WebsterConsultants (2004)

38 1992/93e2002/03. (2) UK Several Waterand wastewater

Report 60.0 Private 1.255

Fraquelli and Moiso (2005) 18 "30 year unbalancedpanel data" (30)

Italy MLE Waterand wastewater

Working Paper 59.2 Mostly public 1.4

Saal and Parker (2006) 30 (1993) 1993e2003 (11) UK (E&W) e Waterand wastewater

Chapter 373.0 Private 0.9822 (2003) 63.0 1

Sauer (2005) 47 2001 (1) Germany(East and West)

SUR Water Journal 1.2 Mostly public 2.087

Martins et al. (2006) 282 2002 (1) Portugal OLS Water Working Paper 2.5 Mostly public 1.747Torres and Morrison (2006) 255 1996 (1) US MLE Water Journal 0.5 Public 0.813Zoric (2006) 52 1997e2003 (7) Slovenia Several Water

and wastewaterThesis 1.2 Mostly public 1.04

Garcia et al. (2007) 211 1997e2000 (4) US GMM Water Journal 1.6 Mostly public 1.12Nauges and Berg (2008) 27 1996e2004 (9) Brazil Several Water

and wastewaterWorking Paper 395.0 Mostly public 0.99

48 2003e2004 (2) Colombia Waterand wastewater

22.0 1.11

41 1996e2004 (9) Moldova Waterand wastewater

4.0 1.26

67 1997e2000 (4) Vietnam Water 13.0 1.16Saal et al. (2007) 25 1985e2000 (16) UK (E&W) Several Water

and wastewaterJournal 60.0 Private 0.861

Bottasso and Conti (2008) 144 1995e2004 (10) UK Several Water Journal 67.8 Private 1.113Correia and Marques (2010) 70 2005 (1) Portugal Several Water Working Paper 6.3 Mostly public 1.113Farsi et al. (2008) 87 1997e2005 (9) Switzerland Several Water

and wastewaterJournal 3.4 Public 1.09

(continued on next page)

P.Carvalhoet

al./Utilities

Policy21

(2012)40

e49

47

(continued )

Study Waterutilities(no.)

Years (no.) Country Estimationmethod

Scopeof services

Publicationtype

Average size(106 m3/year)

Ownership Scale economies

Farsi and Filippini (2009) 34 1997e2005 (9) Switzerland Several Waterand wastewater

Journal 5.8 Public 1.103

Fernández and Londoño (2008) 126 2003e2005 (3) Colombia SUR Waterand wastewater

Working Paper 18.9 Mostly public 1.28

Filippini et al. (2008) 52 1997e2003 (7) Slovenia Several Water Journal 2.3 Public 1.05Martins et al. (2008) 265 2002 (1) Portugal OLS Water

and wastewaterWorking Paper 1.7 Mostly public 1.487

Urakami and Parker (2008) 1330 1999e2006 (8) Japan SUR Water Working Paper 9.4 Public 1.08Baranzini and Faust (2009) 330 2000e2006 (7) Switzerland SUR Water Working Paper 1.2 Public 1.1Prieto et al. (2009) 2248 2000 (1) Spain SUR Water

and wastewaterJournal 0.06 Mostly private 1.481

Tsegai et al. (2009) 50 2004 and 2006 (2) South Africa SUR Water Working Paper 3.1 Public 1.16Corton (2010) 43 1996e2005 (10) Peru MLE Water Journal 9.9 Public 1.25Ferro et al. (2010) 90 2005 (1) Latin America -

Argentina, Bolivia,Chile, Colombia,Costa Rica, Ecuador,Honduras, México,Nicaragua, Panamá,Paraguay,Peru y Uruguay

SUR Water and wastewater Working Paper 0.07 Mostly public 1.11

Zschille and Walter (2010) 72 1998e2007 (10) Germany Several Water Working Paper 3.9 Mostly public 1.146

Appendix 2. Literature on scope economies

Study Water utilities (no.) Years (no.) Country Estimationmethod

Scope of services Publication type Average size(106 m3/year)

Ownership Scope economies

Hayes (1987) 475 1960, 1970 and 1976 (3) US OLS Water Journal 1.9 Mostly public 0.0687Kim and Clarke (1988) 60 1973 (1) US MLE Water Journal 1.9 Mostly public 0.1663Hunt and Lynk (1995) Not defined

(90 observations)1980e1988 (9) UK (E&W) OLS Water and wastewater Journal 60.0 Private Yes

Garcia and Thomas (2001) 55 1995e1997 (3) France GMM Water Journal 0.6 Mostly private 0.2367Parker and Saal (2001) 39 1985e1999 (15) UK (E&W) SUR Water and wastewater Journal 50.0 Private �0.23Fraquelli and Giandrone (2003) 103 1996 (1) Italy RLS Wastewater Journal 1.0 Mostly public YesFraquelli et al. (2004) 90 1994e1996 (3) Italy NLSUR Water and wastewater Journal 18.9 Mostly public 0.124Piacenza and Vanonni (2004) 90 1994e1996 (3) Italy NLSUR Water and wastewater Journal 11.0 Mostly public 0.1867Stone and Webster

Consultants (2004)38 1992/93e2002/03 (2) UK Several Water and wastewater Report 50.0 Private No (for water

and sewerage)Yes (for verticalintegration of water)

Martins et al. (2006) 282 2002 (1) Portugal OLS Water Working Paper 2.5 Mostly public 0.455Torres and Morrison (2006) 255 1996 (1) US MLE Water Journal 0.5 Mostly public 0.45Garcia et al. (2007) 211 1997e2000 (4) US GMM Water Journal 1.6 Mostly public NoCorreia and Marques (2010) 70 2005 (1) Portugal Several Water Working Paper 6.3 Mostly public 0.054Farsi et al. (2008) 87 1997e2005 (9) Switzerland Several Water and wastewater Journal 3.4 Public 0.145Fernández and Londoño (2008) 126 2003e2005 (3) Colombia SUR Water and wastewater Working Paper 18.9 Mostly public 0.38Martins et al. (2008) 265 2002 (1) Portugal OLS Water and wastewater Working Paper 1.7 Mostly public 0.327Prieto et al. (2009) 2248 2000 (1) Spain SUR Water and wastewater Journal 0.06 Mostly private �0.084

P.Carvalhoet

al./Utilities

Policy21

(2012)40

e49

48

P. Carvalho et al. / Utilities Policy 21 (2012) 40e49 49

References

Abbott, M., Cohen, B., 2009. Productivity and efficiency in the water industry.Utilities Policy 17, 233e244.

Antonioli, B., Filippini, M., 2001. The use of a variable cost function in the regulationof the Italian water industry. Utilities Policy 10, 181e187.

Ashton, J., 1999. Economies of Scale, Economies of Capital Utilisation and CapitalUtilisation in the English and Welsh Water Industry. Working Paper Series.Bournemouth University, Bournemouth.

Ashton, J., 2000. Total factor productivity growth and technical change in the waterand sewerage industry. Serv. Ind. J. 20, 121e130.

Baranzini, A., Faust, A., 2009. The Cost Structure of Water Utilities in SwitzerlandWorking Paper Cahier de recherche, Haute école de gestion de Genève, Geneve.

Baumol, W., 1976. Scale economies, average cost, and the profitability of marginalcost pricing. In: Grieson, R.F. (Ed.), Public and Urban Economics: Essays inHonor of William S. Vickrey Lexington, D.C.

Baumol, W., Panzar, J., Willig, R., 1988. Contestable Markets and the Theory ofIndustry Structure. Harcourt Brace Jovanovich, New York.

Bel, G., Warner, M., 2008. Does privatization of solid waste and water servicesreduce costs? A review of empirical studies. Resour. Conserv. Recy. 52,1337e1348.

Bel, G., Fageda, X., Warner, M., 2010. Is private production of public services cheaperthan public production? A meta-regression analysis of solid waste and waterservices. J. Pol. Anal. Manag. 29, 553e577.

Berg, S., Marques, R., 2011. Quantitative studies of water and sanitation utilities:a benchmarking literature survey. Water Policy 13, 591e606.

Bhattacharyya, A., Parker, E., Raffiee, K., 1994. An examination of the effect ofownership on the relative efficiency of public and private water utilities. LandEcon. 70, 197e209.

Bhattacharyya, A., Harris, T., Narayanan, R., Raffiee, K., 1995. Specification andestimation of the effect of ownership on the economic efficiency of the waterutilities. Reg. Sci. Urban Econ. 25, 759e784.

Bottasso, A., Conti, M., 2008. Scale economies, technology and technical change inthe water industry: evidence from the English water only sector. Reg. Sci. UrbanEcon. 39, 138e147.

Briscoe, J., 2011. Invited opinion interview: two decades at the center of world waterpolicy. Water Policy 12, 147e160.

Bruggink, T., 1982. Public versus regulated private enterprise in the municipal waterindustry: a comparison of operating costs. Q. Rev. Econ. Bus. 22, 111e125.

Byrnes, P., Grosskopf, S., Hayes, K., 1986. Efficiency and ownership: further evidence.Rev. Econ. Stat. 68, 337e341.

Correia, T., Marques, R., 2010. Performance of Portuguese water utilities: how doownership, size, diversification and vertical integration relate to efficiency?Water Policy 13, 343e361.

Corton, L., 2010. Sector fragmentation and aggregation of service provision in thewater industry. J. Prod. Anal. 2, 1e11.

Crain, W., Zardkoohi, A., 1978. A test of the property-rights theory of the firm: waterutilities in the United States. J. Law Econ. 21, 395e408.

Fabbri, P., Fraquelli, G., 2000. Costs and structure of technology in the Italian waterindustry. Empirica 27, 65e82.

Farsi, M., Fetz, A., Filippini, M., 2008. Economies of scale and scope in Swiss multi-Utilities. Energy J. 29, 123e145.

Farsi, M., Filippini, M., 2009. An analysis of cost efficiency in Swiss multi-utilities.Energ. Econ. 31, 306e315.

Fernández, D., Londoño, G., 2008. Análisis de economías de escala y alcance en losservicios de acueducto y alcantarillado en Colombia. Comisión de Regulación deAgua Potable y Saneamiento Básico (CRA), Bogotá.

Ferro, G., Lentini, E., Mercadier, A., Romero, C., 2010. Retornos a escala en agua ysaneamiento: estimaciones para América Latina. MPRA Paper No. 20324.Munich University Library.

Filippini, M., Hrovatin, N., Zori�c, J., 2008. Cost efficiency of Slovenianwater distributionutilities: an application of stochastic frontier methods. J. Prod. Anal. 29, 169e182.

Ford, J.,Warford, J.,1969. Cost functions for thewater industry. J. Ind. Econ.18, 53e63.Fox, W., Hofler, R., 1986. Using homothetic composed error frontiers to measure

water utility efficiency. Southern Econ. J. 53, 461e477.Fraquelli, G., Giandrone, R., 2003. Reforming the wastewater treatment sector in

Italy: implications of plant size, structure and scale economies. Water Resour.Res. 39, 1293.

Fraquelli, G., Piacenza, M., Vannoni, D., 2004. Scope and scale economies in multi-utilities: evidence from gas, water and electricity combinations. Appl. Econ. 36,2045e2057.

Fraquelli, G., Moiso, V., 2005. Cost Efficiency and Economies of Scale in the ItalianWater Industry Working Paper, Pavia.

Garcia, S., Thomas, A., 2001. The structure of municipal water supply costs: appli-cation to a panel of French local communities. J. Prod. Anal. 16, 5e29.

Garcia, S., Moreaux, M., Reynaud, A., 2007. Measuring economies of vertical inte-gration in network industries: an application to the water sector. Int. J. Ind.Organ. 25, 791e820.

Harbord, R., Higgins, J., 2008. Meta-regression in Stata. Stata J. 8, 493e519.Hayes, K., 1987. Cost structure of the water utility industry. Appl. Econ. 19, 417e425.Higgins, J., Thompson, S., 2004. Controlling the risk of spurious findings from meta-

regression. Stat. Med. 23, 1663e1682.Hunt, L., Lynk, E., 1995. Privatization and efficiency in the UK water industry: an

empirical analysis. Oxford B. Econ. Stat. 57, 371e388.

Kim, H., 1987. Economies of scale in multiproduct firms: an empirical analysis.Economica (new series) 54, 185e206.

Kim, H., Clarke, R., 1988. Economies of scale and scope in water supply. Reg. Sci.Urban Econ. 18, 479e502.

Kim, E., Lee, H., 1998. Spatial integration of urban water services and economies ofscale. Rev. Urban Reg. Devel. Stud. 10, 3e18.

Knapp, M., 1978. Economies of scale in sewerage purification and disposal. J. Ind.Econ. 27, 163e183.

Leamer, Edward E., 1983. Let’s take the con out of econometrics. Am. Econ. Rev. 73(1 March), 31e43.

Link, E., 1993. Privatization, joint production and the comparative efficiencies ofprivate andpublic ownership: theUKwater industry case. Fiscal Stud.14, 98e116.

Marques, R., 2008. Comparing private and public performance of Portuguese waterservices. Water Policy 10, 25e42.

Martins, R., Fortunato, A., Coelho, F., 2006. Cost Structure of the Portuguese WaterIndustry: A Cubic Cost Function Application. Working Paper No. 9.. University ofCoimbra, Coimbra.

Martins, R., Coelho, F., Fortunato, A., 2008. Water Losses and Hydrographical CostStructure Industry. Working Paper No. 6. University of Coimbra, Coimbra.

Mizutani, F., Urakami, T., 2001. Identifying network density and scale economies forJapanese water supply organizations. Pap. Reg. Sci. 80, 211e230.

Mugisha, Silver, 2011. Utility benchmarking and regulation in developing countries:practical application of performance monitoring and incentives. Int. WaterAssoc. xi, 195.

Nauges, C., Berg, C., 2008. Economies of density, scale and scope in the water supplyand sewerage sector: a study of four developing and transition economies.J. Regul. Econ. 34, 144e163.

Panzar, J., Willig, R., 1977. Economies of scale in multi-output production. Quart. J.Econ. 91, 481e493.

Parker, D., Saal, D., 2001. The impact of the privatization and regulation on the waterand sewerage industry in England and Wales: a translog cost functionapproach. Manage. Decis. Econ. 21, 253e268.

Piacenza, M., Vanonni, D., 2004. Choosing among alternative cost function speci-fication: an application to Italian multi-utilities. Econ. Lett. 82, 415e422.

Prieto, Á, Zofio, J., Álvarez, I., 2009. Economías de escala, densidad y alcance en laprovisión pública de infraestructura básica municipal. Hacienda Pública Espa-ñola 190, 59e94.

Rassier, D., Earnhart, D., 2011. Short-run and long-run implications of environmentalregulation on financial performance. Contemporary Econ. Pol. 29, 357e373.

Renzetti, S., 1999. Municipal water supply and sewage treatment: costs, prices anddistortions. Can. J. Econ. 32, 688e704.

Renzetti, S., Dupont, D., 2009. Measuring the technical efficiency of municipal watersuppliers: the role of environmental factors. Land Econ. 8, 627e636.

Saal, D., Parker, D., 2000. The impact of privatization and regulation on the waterand sewerage industry in England and Wales: a translog cost function model.Manage. Decis. Econ. 21, 253e268.

Saal, D., Parker, D., 2004. The comparative impact of privatization and regulation onproductivity growth in the English and Welsh water and sewerage industry,1985e99. Int. J. Regul. Govern. 4, 139e170.

Saal, D., Parker, D., 2006. Assessing the performance of water operations in theEnglish andWelsh water industry: a panel input distance function approach. In:Coelli, T., Lawrence, D. (Eds.), Performance Measurement and Regulation ofNetwork Utilities. Edward Elgar Publishers, Northampton.

Saal, D., Parker, D., Weyman-Jones, T., 2007. Determining the contribution oftechnical change, efficiency change and scale change to productivity growth inthe privatized English and Welsh water and sewerage industry: 1985e2000.J. Prod. Anal. 28, 127e139.

Sauer, J., 2003. The Efficiency of Rural Infrastructure e Water Supply in Rural Areasof Transition. European Regional Science Association (ERSA), Vienna.

Sauer, J., 2005. Economies of scale and firm size optimum in rural water supply.Water Resour. Res. 41, 1e13.

Sharp, S., 1998. sbe23: meta-analysis regression. Stata Tech. Bull. 42, 16e22.Shih, J., Harrington, W., Pizer, W., Gillington, K., 2006. Economies of scale in

community water systems. J. Am. Water Works Assoc. 98, 100e108.Stanley, T.D., Jarrell, S.B., 1989. Meta-regression analysis: a quantitative method of

literature surveys. J. Econ. Surv. 3, 161e170.Stone & Webster Consultants, 2004. Investigation into the Evidence for Economies

of Scale in the Water and Sewerage Industry in England andWales. Final Report.OFWAT, Birmingham.

Torres, M., Morrison, C., 2006. Driving forces for consolidation or fragmentation ofthe US water utility industry: a cost function approach with endogenousoutput. J. Urban Econ. 59, 104e120.

Tsegai, D., Linz, T., Kloos, J., 2009. Economic Analysis of Water Supply Cost Structurein the Middle Olifants Sub-basin of South Africa. Working Paper. Social ScienceResearch Network (SSRN).

Urakami, T., 2007. Economies of vertical integration in the Japanese water supplyindustry. Jahrbuch für Regionalwissenschaft 27, 129e141.

Urakami, T., Parker, D., 2008. The Effects of Consolidation amongst Japanese WaterUtilities:AHedonicCostFunctionAnalysis.WorkingPaper. CranfieldUniversity,UK.

Zoric, J., 2006. Cost efficiency and regulation of Slovenian water distribution utili-ties: an application of stochastic frontier methods. PhD Thesis, LuganoUniversity, Lugano.

Zschille, M., Walter, M., 2010. Cost Efficiency and Economies of Scale and Density inGerman Water Distribution. Working Paper. German Institute for EconomicResearch (DIW), Berlin.