statistical analysis of philippine water district characteristics and how these affect water tariffs

11
This article was downloaded by: [Universitaetsbibliothek Potsdam] On: 09 January 2014, At: 20:35 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Water International Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rwin20 Statistical analysis of Philippine water district characteristics and how these affect water tariffs Carlos Primo C. David a , Peter Julian A. Cayton b , Theresa E. Lorenzo a & Eduardo C. Santos c a Environment Monitoring Laboratory, National Institute of Geological Sciences, University of the Philippines Diliman, Quezon City, Philippines b School of Statistics, University of the Philippines Diliman, Quezon City, Philippines c Local Water Utilities Administration, Quezon City, Philippines Published online: 13 Nov 2013. To cite this article: Carlos Primo C. David, Peter Julian A. Cayton, Theresa E. Lorenzo & Eduardo C. Santos (2014) Statistical analysis of Philippine water district characteristics and how these affect water tariffs, Water International, 39:1, 1-9, DOI: 10.1080/02508060.2013.847687 To link to this article: http://dx.doi.org/10.1080/02508060.2013.847687 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Upload: up-diliman

Post on 04-Feb-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

This article was downloaded by: [Universitaetsbibliothek Potsdam]On: 09 January 2014, At: 20:35Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Water InternationalPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rwin20

Statistical analysis of Philippine waterdistrict characteristics and how theseaffect water tariffsCarlos Primo C. Davida, Peter Julian A. Caytonb, Theresa E.Lorenzoa & Eduardo C. Santosc

a Environment Monitoring Laboratory, National Institute ofGeological Sciences, University of the Philippines Diliman, QuezonCity, Philippinesb School of Statistics, University of the Philippines Diliman,Quezon City, Philippinesc Local Water Utilities Administration, Quezon City, PhilippinesPublished online: 13 Nov 2013.

To cite this article: Carlos Primo C. David, Peter Julian A. Cayton, Theresa E. Lorenzo & EduardoC. Santos (2014) Statistical analysis of Philippine water district characteristics and how these affectwater tariffs, Water International, 39:1, 1-9, DOI: 10.1080/02508060.2013.847687

To link to this article: http://dx.doi.org/10.1080/02508060.2013.847687

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

Statistical analysis of Philippine water district characteristics and howthese affect water tariffs

Carlos Primo C. Davida*, Peter Julian A. Caytonb, Theresa E. Lorenzoa andEduardo C. Santosc

aEnvironment Monitoring Laboratory, National Institute of Geological Sciences, University of thePhilippines Diliman, Quezon City, Philippines; bSchool of Statistics, University of the PhilippinesDiliman, Quezon City, Philippines; cLocal Water Utilities Administration, Quezon City, Philippines

(Received 11 January 2013; accepted 15 September 2013)

Philippine water districts (WDs) provide water to over 17 million Filipinos. Each WDis an independent entity, and water tariffs vary widely across the 493 WDs due toperceived area-specific conditions. A statistical model was applied to available data todetermine how these conditions affect tariffs. Results confirm the direct influence ontariffs of factors such as location, water source and efficiency of service provision. It islikewise found that an optimal value exists for connection density and for capitaloutlay. This suggests an optimum size for a WD to be able to provide the lowestpossible tariff.

Keywords: water tariff; price inefficiency; stochastic frontier model; environmentalvariables

Introduction

A constant challenge for developing countries like the Philippines is the provision of cleanand affordable domestic water. To accomplish this, the Philippine water district (WD)system was established through the Provincial Water Utilities Act of 1973 (PresidentialDecree No. 198). Its aim was to reverse two prevailing conditions of public utilities in thecountryside: existing municipal systems are deteriorating faster than they are beingmaintained or replaced; and they are not being expanded at a rate sufficient to matchpopulation growth. A new system was therefore needed to replace the local government-controlled water utilities.

WDs are non-profit corporate entities, classified as government-owned corporations,under the guidance and monitoring of the Local Water Utilities Administration (LWUA).Under the decree, WDs are mandated to provide individual household water connectionswithin a particular political jurisdiction. Currently, there are 493 WDs in operation outside ofMetropolitanManila, serving more than 17million people (Figure 1).Many of theWDs are insmall municipalities, often only serving the town centre and with less than 2000 connections.

WDs are non-profit corporations; however, one of the guiding principles of the WDsystem in the country is the full cost recovery of the service they provide. What is passedon to concessionaires to be the WD’s water tariff includes, among other things: opera-tional and maintenance costs; capital costs; environmental-damage costs; and long-runmarginal costs. Except for national- and local-government subsidy for capital costs of

*Corresponding author. Email: [email protected]; [email protected]

Water International, 2014Vol. 39, No. 1, 1–9, http://dx.doi.org/10.1080/02508060.2013.847687

© 2013 International Water Resources Association

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

projects for small or newly established WDs, each WD operates independently and coversall other costs through its water-tariff collection. Such pricing of a basic good has beenthoroughly studied and is deemed ideal, as it provides for a sustainable scheme in mostcountries (Anderson & Gaines, 2009; Rogers, Bhatia, & Huber, 1998; Whittington, 2003;Zetland & Gasson, 2013). In the Philippines, it is the role of the LWUA to computeappropriate water rates based on a review of the individual WDs’ annual financial recordsand a list of guiding rules. This computation is then compared to any requested tariffadjustment from the WDs and approved accordingly.

Awater provider’s financial state is controlled by internal and external conditions, which inturn will affect its water rates. Renzetti and Dupont (2009) looked into how environmental

Figure 1. Map of the Philippines showing the coverage of the water-district system.

2 C.P.C. David et al.

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

factors such as weekly temperature, elevation, precipitation and population density affect thecost of providingwater services. DeWitte andMarques (2008) and Zetland andGasson (2013)compared country-specific environmental and institutional factors as these affect water rates.Baranzini et al. (2010) recognized the impact of technology and production-structure char-acteristics of water providers on their water rates, but also the significance of environmentalvariables that need to be considered when modelling water rates. When all these factors areconsidered, theoretical water rates can be calculated.

A common method utilized in previous resource valuation studies is to statisticallycalculate the cost inefficiency, which is roughly equivalent to how far an actual water rateis from the ideal water rate deduced from the characteristics of a water provider. De Witteand Marques (2008) demonstrated the use of data envelopment analysis in computing theefficiency of water utilities in different countries while taking the effect of environmentalvariables into account. Botasso and Conti (2007), in an analysis of the cost inefficiency ofEnglish and Welsh water utilities over a period of six years, investigated the role oftechnology under which water utilities operate. Kirkpatrick et al. (2006) also used astochastic cost function (specified as a Cobb-Douglas function) to compare the perfor-mance of private and public African water utilities using data from 13 countries, takenfrom the year 2000. Similarly, the stochastic cost approach, this time approximated by theTranslog approximation, was also utilized by Aubert and Reynaud (2005) to determine thecost efficiency of Wisconsin water utilities that used different regulatory approaches. Theanalysis included the volume sold by water utilities to their final customers, as well as thenumber of people served. Variable costs of the utilities, as well as technical factors thatinfluence the water utilities’ production, were also included in the function. It was alsoobserved that the cost efficiency of water utilities varied depending on the specificregulatory regime adopted by each water utility.

In this study, a stochastic cost frontier model is used to study all possible interactionsof the identified factors, and their overall effect on water rates for WDs will be described.In this method, it is assumed that deviation from the best-practices frontier is caused byinefficiency and random factors (De Fraja, 1993). Similar studies have been done usingthis method, including a spatio-temporal study of the technical efficiency of a specifictype of Philippine electric utility (electric cooperatives) using a time-invariant model, atime-decaying model, and a spatio-temporal stochastic frontier model (Lavado & Barrios,2008). The same study found that the stochastic frontier model yielded more stableefficiency estimates, as compared to the other two models, as well as robustness withrespect to the efficiency-affecting factors examined.

Methodology

The stochastic frontier model methodology involves modelling a response variable ofinterest with respect to an efficient system that uses information from other variables thatexplain the behaviour of the response, with a term in the model defining the behaviour ofthe inefficiencies (Aigner, Lovell, & Schmidt, 1977). In this paper, we use the model toinvestigate location and company-specific characteristics that affect the price of water inwater districts.

In equation form, the stochastic cost frontier model is defined in this manner (Lita &Stamule, 2011):

yi ¼ f xi; β� �

exp uif g exp vif g; i ¼ 1; 2; :::; n;

Water International 3

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

where yi is the price of water per cubic meter per ith water district; f ðxi; βÞ is an efficientpricing system, which depends on the vector xi ¼ x1i; x2i; :::; xkið Þ of k inputs and otheraccounted factors and the vector β of constants (parameters of the model); expfuig is thefirm-specific inefficiency, where a high value of the term would signify a multiplicativedistance of the observed price yi from the efficient price with stochastic partf ðxi; βÞ exp vif g; and expfvig is the stochastic component of the model, which measuresthe difference between the actual and predicted price due to factors that were notconsidered in the study.

To perform statistical analysis, assumptions on the form of the model are made. Thestochastic frontier model is adapted into a linear regression form by using logarithms:

ln yi ¼ ln f ðxi; βÞh i

þ ui þ vi; i ¼ 1; 2; :::; n:

The log-inefficiency term ui is assumed to have a one-sided distribution, which for thispaper was the half-normal distribution with parameters μ ¼ 0; σ2u > 0, where μ is a prob-ability constant that is set to 0 while σ is a parameter that indicates data dispersion in themodel. This means that log-inefficiency scores can only take positive values. The log-stochastic term vi is assumed to have a two-sided distribution, which allows the model toscale down or up the efficient price due to other factors or unpredictability, which for thepaper has normal distribution with μ ¼ 0; σ2v > 0. In the statistical model, there are k + 2unknown parameter values to be estimated: the k constants of β in the efficient system,plus σ2u and σ2v . These are estimated using maximum-likelihood estimation techniques instatistical inference. Analysis of the results is similar to regression analysis, with addi-tional analysis on the parameter σ2u, specifically testing for the hypothesis that this termmay be equal to zero (Aigner et al., 1977; Lita & Stamule, 2011).

All the variables considered in this paper can be categorized into two groups (Table 1):environmental variables, which deal with the natural environment and in-place conditions of aWD; and institutional variables, which are related toWDmanagement. The majority of the dataused came from the database of the LWUA and are up to date as of May 2012. The databaseincluded information for only 391 WDs; for the other WDs, either the information was notcurrent or therewas no information as yet because they had only been establishedwithin the year.

The source of water (groundwater and surface water, or a mix of both) that a WD useswas considered important for analysis. For those tapping groundwater, the number ofproduction wells was also taken into account. WD location was classified as coastal orinland, as coastal municipalities will usually have low elevation variance. Municipalityincome data classifications were also included, as a gauge of the inhabitants’ ability topay, with rates possibly being lower in the third-, fourth- and fifth-class municipalitiesthan in the first- and second-class municipalities as well as in cities or urbanized towns.

Table 1. Water-district characteristics considered in the study.

Environmental/in-place variables Institutional variables

Source of water (surface, spring or groundwater) Non-revenue waterNumber of groundwater wells maintained Number of connections/connection densityLocation (coastal vs. inland) Number of WD employeesTopography/elevation variance Total water volume producedHost municipality income (indirect measure ofability to pay)

Capital outlay/loans

4 C.P.C. David et al.

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

Non-revenue water (NRW) reduces the amount of revenue available to waterutilities given a fixed volume of water produced, and so is likely to cause waterutilities to raise water rates in order to cover costs. Connection density was alsoincluded in the study, as both a measure of revenue-generation capacity and an indirectindicator of capital outlay required per connection. The number of employees, espe-cially when related to the number of connections, was used as an operational efficiencymeasure. Lastly, the amount of existing loans availed by WDs from LWUA is likewiseincluded in the model. Note that the figures used in this study exclude external loansand subsidies from local government units. These subsidies may allow WDs to keeptheir water rates lower than would otherwise be normal given their environmental andinstitutional conditions.

Results and discussion

The results of the study confirmed many of the generalizations (i.e. economies of scale) inthe delivery of basic utilities but also resulted in new observations specific to the watersector. The average water rate for the 391 WDs sampled is PHP 240.21 (USD 5.78) forthe first block (up to 10 m3 per month consumption). On average, WDs serve around 7000connections; however, 171 WDs serve less than 2000 households each. The small size ofmost Philippine WDs is also reflected in the median number of employees per WD, whichis 19 (range, 3 to 881).

The final equation generated by the stochastic frontier model is:

ln water rateð Þ ¼ 5:242709þ 0:0108234 ln capital outlayð Þ½ �2þ 0:029709 ln capital outlayð Þ½ � ln connections per areað Þ½ �� 0:0288475 ln capital outlayð Þ½ � ln total volumeð Þ½ �þ 0:0158001 #groundwater sources½ � � 0:0967505 coastal area½ �þ 0:0202934 ln connections per areað Þ½ �2 � 0:0000713 NRW%½ �2

þ 0:0210036 ln total volumeð Þ½ �2 þ 0:0021264 NRW%½ � ln #employeesð Þ½ �þ 0:0955487 #cities½ � #surface‑water sources½ �� 0:002706 #groundwater sources½ � ln #employeesð Þ½ �� 0:0143363 #surface‑water sources½ � ln #employeesð Þ½ �� 0:0489662 ln total volumeð Þ½ � ln connections per areað Þ½ �

From this equation one can deduce which parameters positively and predominantly affectwater rates (e.g. capital outlay and connections per area). Each individual variable wasanalyzed for its effect on water rates as well as for possible linkages with fellow variables.All linkages and relation for each variable discussed below are observed when all othervariables are held equal.

It was found that on average, coastal WDs have rates that are lower by 9.22%compared to those which are inland. This may be attributed to coastal WDs being morelikely to be located in areas with low-to-medium elevation variance and medium-to-highpopulation density, which would make it easier to supply water to a large population.Several coastal WDs also have the advantage of having their production wells inland, at asignificantly higher elevation, so that their distribution system is almost entirely run bygravity.

Water International 5

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

In terms of source, WDs which utilized surface-water sources expectedly were foundto have higher water rates than those that tap groundwater, as their operational costs due totreatment requirements are higher. Still, while generally less expensive, water ratesincrease quickly as the number of wells a WD utilizes increases. Pumping and main-tenance costs for numerous wells, as well as the infrastructure needed to convey the waterfrom different sources to the user base, could contribute to the observed trend. In bothcases, the effect of source (surface water vs. groundwater) is reduced in WDs distributinga large volume of water. This suggests that sourcing water from surface sources, deemedmore sustainable, will be cost-competitive for larger WDs.

Comparing urban and rural WDs, the statistical results generally did not yield anysignificant trend. The expected higher willingness to pay, translating to higher waterprices in urban areas, may have been masked by other factors present in rural WDs,particularly their lower connection density. Average water prices are comparable over thedifferent WD sizes.

More than half of the WDs have non-revenue water (NRW) greater than 23%, andtherefore its impact on water rates would be significant. An overwhelming trend is analgebraic increase in water rates as NRW increases. While the actual increase perpercentage point varies from one WD to another, a medium-size WD will typicallyhave a PHP 11 (USD 0.26) increase in its monthly minimum charge per NRW percentagepoint increase. Moreover, an increase in NRW translates to a relatively higher increase inwater rates for smaller WDs and lower-volume production, compared to bigger WDs,which will have a higher capacity to absorb NRW.

Two relations were observed in the examination of connection density. An increase inconnection density results in lower water prices. However, an inverse relation – higherrates with more connections – emerges when the water produced by a WD exceeds acertain total volume. This suggests that there exists an optimal point for connectiondensity at which water rates are lowest. As can be seen in the equation below, an increasein connections will also result in an increase in capital outlay, which directly increaseswater rates:

connections per area

¼ exp�0:029709 ln capital outlayð Þ½ � þ 0:0489662 ln total volumeð Þ½ �

2 0:0202934ð Þ� �

A convex relation is also observed in the effect of the total volume produced by WDson water rates: as total production volume increases, water rates decrease until an optimalpoint is reached, after which water rates start to increase with volume. This may be causedby an increase in production being equivalent to an increase in revenue at low levels ofproduction. However, higher capital outlay is required as production increases (e.g. needfor extra groundwater wells, upgrade of main pipe capacity) beyond an optimal produc-tion point, thus resulting in higher water prices. The optimal annual production pointdepends on both connection density and capital outlay:

total volume

¼ exp0:0288475 ln capital outlayð Þ½ � þ 0:0489662 ln connections per areað Þ½ �

2 0:0210036ð Þ� �

6 C.P.C. David et al.

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

Several WDs were observed to have price inefficiency (actual rates higher than thecomputed ideal rate). However, the magnitudes were very small, from slightly less than1.00208 to around 1.00214 (Figure 2). Thus, the WD with the highest price inefficiency isonly one-fifth of a percentage point away from its ideal price.

As shown in Figure 3, there is variability observed when the ideal water rates ascomputed by the model are compared with the actual water rates of all WDs. Thissuggests that there is still a need to consider other natural and institutional factors inorder to create a more accurate model. One parameter that needs to be included in the

Figure 2. Water rates and price inefficiency for all water districts. The crossed lines denote 2standard deviations of the x and y parameters.

Figure 3. Comparison of actual and predicted water rates (in PHP) across water districts. The middleline represents 100 per cent correlation while the two outer lines represent 2 standard deviations.

Water International 7

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

calculation but may not be easily quantified is the ease with which water can be drawnfrom the groundwater in different areas (e.g. depth, production per well, etc.).

The interplay between variables may also be more complicated than perceived,resulting in variable relationships that cannot be completely resolved by the model.These are confirmed by the σu ¼ 0 test of the parameters of the stochastic frontiermodel, which resulted in a test statistic value of zero, or a p-value of 100%. This verylarge p-value indicates that based on the data at hand regarding natural and institutionalfactors affecting price in the study’s model, there was no significant evidence thatinefficiency is prevalent in the assignment of prices by the districts.

Conclusions

The investigation of natural and institutional factors using the stochastic frontier modelconfirmed the impact of various factors on water rates. The magnitude of their impact onwater rates was not only quantified but was determined as to when their influence ishighest (e.g. the impact of the source of water is negated in larger WDs). Based on theresults of the analysis of institutional variables such as total volume produced, connectiondensity and capital outlay, it can be concluded that there seems to be an optimum WDsize, one which would provide the lowest possible water rate. Therefore, one policyrecommendation would be to consider merging small adjacent WDs, which would resultin lower costs and therefore more affordable water rates.

The range of the technical inefficiency values of all the WDs was shown to be verynarrow and close to the ideal technical efficiency, meaning that the computed rate for eachWD is close to its current rate. However, the high variance of the predicted ideal waterrate versus actual water rates suggests that other variables not covered in the study couldalso come into play. This is one of the limitations of the current model. Succeeding modelruns would include other natural variables, financial information (e.g. subsidies, other WDassets), and costs associated with water treatment and delivery (i.e. bulk water supplierprice). These data are currently not integrated with the WD database. Furthermore, theincorporation of other service metrics (Majuru, Jagals, & Hunter, 2012) such as reliabilityof service, water-quality improvement and accessibility must be done, as these also haveimpacts on the water tariff.

This statistical study may be used as guide for the LWUA in setting up new WDswith a rough assumption of ideal water rates given a district’s environmental variables.It can also be used in monitoring and managing operating WDs, particularly inpinpointing those with potentially inefficient pricing. In fact, a general water rates–based policy is now being implemented by the LWUA in the country. This is imple-mented by determining the average water rates per region and per WD size category.This benchmark price is used as a basis for water rate increase approval, whereinproposed new water rates below the average are automatically approved. Proposalsfrom WDs that are above the average are likewise approved; however, the approvalcomes with corresponding performance indicators (e.g. increase in service hours, servicecoverage, lower operational expense, etc.) that need to be satisfied during the course ofimplementation of the new rates.

AcknowledgmentsThe authors wish to thank the LWUA, specifically its Research Division and the office of the SeniorDeputy Administrator, for providing most of the data used for this study.

8 C.P.C. David et al.

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014

ReferencesAigner, D., Lovell, C., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier

production function models. Journal of Econometrics, 6, 21–37.Anderson, K., & Gaines, L. (2009). International water pricing: An overview and historic and

modern case studies. Case Studies of Transboundary Dispute Resolution. Cambridge:Cambridge University Press.

Aubert, C., & Reynaud, A. (2005). The Impact of regulation on cost efficiency: An empiricalanalysis of Wisconsin water utilities. Journal of Productivity Analysis, 23(3), 383–409.

Baranzini, A., Faust, A., & Maradan, D. (2010). Water supply: Costs and performance of waterutilities: Evidence from Switzerland. Haute école de gestion de Genève CRAG- Centre deRecherche Appliquée en Gestion.

Botasso, A., & Conti, M. (2007). Efficiency incentives for a regulated monopoly: Some lessonsfrom the English and welsh water industry. Rivista di Politica Economica, 97(4), 115–144.

De Fraja, G. (1993). Productive efficiency in public and private firms. Journal of Public Economics,50, 15–30.

De Witte, K., & Marques, R. (2008). Capturing the environment, a metafrontier approach to thedrinking water sector. International Transactions in Operational Research, 16(2), 257–271.

Kirkpatrick, C., Parker, D., & Zhang, F. (2006). State versus private sector provision of waterservices in Africa: An empirical analysis. The World Bank Economic Review, 20(1), 143–187.

Lavado, R., & Barrios, E. (2008). Spatial-temporal dimensions of efficiency among electric coop-eratives in the Philippines. Philippine Institute of Development Studies, Discussion Paper SeriesNo. 2008–29.

Lita, J., & Stamule, T. (2011). Stochastic frontier analysis of production function and cost functionestimation methods. Study efficiency in the industry level. Management and MarketingChallenges for the Knowledge Society, 6(1), 163–176.

Majuru, B., Jagals, P., & Hunter, P. R. (2012). Assessing rural small community water supply inLimpopo, South Africa: Water service benchmarks and reliability. Science of the TotalEnvironment, 435, 479–486.

Renzetti, S., & Dupont, D. (2009). Measuring the technical efficiency of municipal water suppliers:The role of environmental factors. Land Economics, 85, 627–636.

Rogers, P., Bhatia, R., & Huber, N. (1998). Water as a social and economic good: How to put theprinciple into practice. Global Water Partnership Technical Advisory Committee. Global WaterPartnership/Swedish International Development Cooperation Agency, Stockholm, Sweden.

Whittington, D. (2003). Municipal water pricing and tariff design: A reform agenda for South Asia.Water Policy, 5, 61–76.

Zetland, D., & Gasson, C. (2013). A global survey of urban water tariffs – Are they sustainable,efficient and fair? International Journal of Water Resources Development, 29, 327–342.

Water International 9

Dow

nloa

ded

by [

Uni

vers

itaet

sbib

lioth

ek P

otsd

am]

at 2

0:35

09

Janu

ary

2014