IMPROVEMENTS IN DOMESTIC WATER SERVICES IN KELANTAN: ARE PEOPLE WILLING TO PAY?

Journal of Sustainability Siene and Management Volume 8 Number 2, Deember 2013: 61-70 ISSN: 1823-8556 Penerbit UMT IMPROVEMENTS IN DOMESTIC WATER SERVICES IN KELANTAN: ARE PEOPLE WILLING TO PAY? MAHIRAH KAMALUDIN 1*, KHALID ABDUL RAHIM 2, ALIAS RADAM 3 AND MOHD RUSLI YACOB 4 1 Agriultural and Food Poliy Studies Institute, 2 Agriultural and Food Poliy Studies Institute, 3 Faulty of Eonomis and Management, 4 Faulty of Environmental Studies, University Putra Malaysia, 43400 UPM Serdang, Selangor. *Corresponding author: mahirahkamaludin@gmail.om Abstrat: This study evaluated the willingness to pay by the people in Kelantan for improved domesti water servies. Unsatisfatory servies by the Kelantan water ompany are a long-standing issue in the state. Exessive water onsumption, frequent interruptions, rapid soial and eonomi growths are putting higher demands for quality of servies and the water ompany is unable to ope with the growing number of populations. This study employed Choie Experiment (CE) method to examine other possible preferenes in domesti water servies. This method highlighted four attributes suh as water quality (QUAL), water interruption (INT), non revenue water (NRW) and water pries (WP) in its servies. Respondents were required to trade off their inome for positive hanges in the servies. Basi and interations in Conditional Logit (CL) models were estimated to aount for heterogeneity in the seleted hoies of the onsumers from options provided in water servie attributes. A survey has been arried out with 552 respondents based in every distrit (stratum) in the state. It is interesting to note that of the four attributes onsidered in this study, reduing the frequeny of water interruption beomes a priority in the servies for the population. KEYWORDS: Choie experiment, willingness to pay, water servies, onditional logit, marginal value. Introdution The importane of water to life annot be denied as water is the most preious substane among all of our natural resoures. High growth rates in population, inreasing water onsumption, and water pollution are among the fators whih are putting pressure on water resoures. Hene, any new development of water resoures and improvement in water servies should be found immediately to meet the inreasing water demand. Interruptions in servies and other failures in water industry are a ontroversial issue in Kelantan. The 9 th Malaysia Plan states that Kelantan is ranked 13 th in Malaysia based on development omposite index by state and a large share of the population lives in rural areas. A few states whih are less developed in Malaysia struggle to improve water overage mostly in rural areas inluding Kelantan (Lee, 2011). There are many problems in Kelantan s water industry whih show low water supply overage, the highest Non Revenue Water (NRW) rate in the ountry, low prodution apaity and a low quantity of water supply (National Eonomi Planning Unit, 2006). Kelantan may not have adequate finanial resoures to enhane water supply overage. The sole water provider in Kelantan is Air Kelantan Sdn. Bhd. (AKSB). The water setor was privatized in Otober 1995 and before this it was under the Water Supply Division, Publi Works Department. A researh by the Assoiation of Water and Energy Researh Malaysia (2011) highlighted some ases in Kelantan with low overage performane, dirty and smelly water supply and frequent unsheduled interruption. Presently, water tariffs are low and unable to generate enough revenue to over the full ost of apital investment, operation and maintenane. Water tariff in 2009 only overed 78% of operating expenses (National Eonomi Planning Unit, 2010). Kelantan ranked 3 rd in

Mahirah Kamaludin et al. 62 2011 whih demonstrates lowest water pries for domesti water servies (RM0.55 applied for first 35m 3 ). If the prie is too low, the water ompany will not be able to maintain and sustain its operations and if the prie of water is too heap it leads to wastage. Inexpensive water and its ample supply makes onsumers take water for granted. Low priing of water an ause serious wastage of water. Thus, poor ondition of water supplied, low water pries, lak of finanial resoures and failities are worsening the industry in the state. An important question remains to be answered; are onsumers willing to pay for higher servie standards with the intention to improve the domesti water servies? The opinion and preferenes from onsumers will help the industry with upgrading projets and improvement in failities, so that they gain benefit from higher standard of servies in the future. The main objetive of this study is to valuate water servie attributes in Kelantan. People have been presented with a set of preferene then they selet the best preferene for improvement. This study estimates the monetary trade off between all attribute levels and prie attribute. Marginal willingness to pay is alulated by the differenes in oeffiients between the two attribute levels. People preferenes and tastes will be able to determine willingness to pay and demonstrate the ability to pay. Consumers are willing to pay a higher prie if they obtain a high soial value from the water servie. Many previous researhers suh as Vasquez et al., (2009), Farolfi, Mabugu and Ntshingila (2007) and Larson, Lew and Onozaka (2001) have used willingness to pay (WTP) approah for improving water supply servies in their studies. WTP reveals that if something is worth having, then it is worth paying for. Adamowiz (1994) was the pioneer in CE method in non market valuation of environmental resoures. Hensher et al., (2005) onduts a study to know how muh onsumers are willing to pay for speifi servie levels and finds that people are willing to pay in order to redue the frequeny of water interruptions and wastewater overflows. The strength of CE method is that it will seek people s preferenes based on harateristis/attributes of the goods and servies and it is able to tell the soial preferenes based on the different management senarios of environmental assets (Perni et al., 2012).Besides, it provides a riher data set that inludes benefit transfer potential, ontext flexibility, strategi bias redution and framing effet ontrol (Do and Bennet, 2007). However, the CE involves greater demands on onentration and patiene of respondents in order to hoose the best preferene in a hoie set. The respondents are faed with the ognitive burden due to omplex tasks (Bennet and Blamey, 2001). This appliation takes time as it needs the respondents to study and understand the attributes and their levels. Methodology This study used Choie Experiment (CE) to determine onsumer s willingness to pay by eliiting onsumer preferenes sine an improvement in water servies an be explained by the attributes and onsumer s valuation relies on attribute levels. The respondents were provided with a hoie set whih inludes three hoies inluding status quo or urrent water servies. The hoies are explained in terms of a ommon set of attributes and the hoies are differentiated from one another. The attributes and attribute levels indiate the values on the program for improved domesti water servies. The seleted attributes are water quality, water interruption, non-revenue water, and water pries. Too many attributes may put high burden on respondents, however if the attributes are too few, it redues the auray of the results sine some information may be absent (Mohd et al., 2008). Conditional Logit (CL) model is used to estimate the CE method. The utility funtion inludes U i as dependent random variable, attributes of the investigated program and their attribute levels is V i and random unobservable omponent ε i. The model shows that the utility whih a onsumer orrelates with alternative i is stated as: U i = V i + ε i (Equation 1) Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

IMPROVEMENTS IN DOMESTIC WATER SERVICES IN KELANTAN 63 Equation 1 offers onsumers, with = 1, 2, 3,... and a set of alternative hoies whih is i = 1, 2, 3,... Probability of onsumer hoosing alternative i an be denoted in parametri funtion of general form represented as follows: Prob i = f(x i, x j ; i j, β ) (Equation 2) Where; Prob i = Probability of onsumer in hoosing alternative i x i = Parameter of observable harateristis of alternative i for onsumer x j = Parameter of observable harateristis of alternative j for onsumer. MFadden (1974) as ited by Hanley et al., (1998) states that the analysis of CL model assumes that the random error terms are independently and identially distributed (IID) based on Gumbel distribution (Type 1). The probability that onsumer hooses alternative i over alternative j with the presene of the random term an be written as below: Prob i = Prob {(U i > U j ) } ; i j (Equation 3) = Prob{(V i + ε i ) > ( V j + ε j )} ; i j (Equation 4) = Prob{(V i -V j > ( ε i - ε j )} ; i j (Equation 5) Probability of onsumer hoosing alternative i over alternatives j in the hoie set C an be shown as below: Prob i = (Equation 6) Assuming that the demonstrates a linear parametergeneralized speifiation of utility funtion an be written as below: V i = β 1 x i + β 2 x 1i + β 3 x 2i +... + β s x si (Equation 7) Where β is vetor of parameters to be estimated and x signifies for all explanatory variables in the model. Prob i = (Equation 8) Aording to Equation 7, the β oeffiients an be used to estimate the marginal rate of substitution (MRS) or at whih onsumers are willing to trade-off between the attributes. The substitution rate an be estimated by dividing the β oeffiient with another β oeffiient (monetary attribute) and multiply it by -1. So, the equation is as follows: (Equation 9) If onsumers indued hanges in water industry, this study presents a set of attributes and the hoies an be made by respondents so that they an reveal their preferenes. The attributes onsist of qualitative (e.g., water quality) and quantitative (prie) items and are onstruted based on disussions with supervisors, expertise, offiers in AKSB and information from previous studies. Table 1 shows the hosen set of attributes and attribute levels to provide options in making deisions for a hange in their domesti water servies. The first attribute is water quality (QUAL). There are three levels hosen for this attribute suh as satisfatory, good and very good. The attribute levels illustrate whether the indiators meet WHO standard or not. This study follows the WHO guideline whih is speified through the ontents suh as Aidity(pH), level of Nitrate(NO 3 ),Nitrite (NO 2 ) and Ammonia (NH 3 ), Turbidity (NTU) and Hardness or Calium (Ca2+)in the water supplied. The level should not exeed level for onsumption beause it may pose risks to human health. The seond attribute is water interruption (INT). The levels assigned for this attribute are frequently, sometimes and never. The attribute levels desribe the frequeny of water supply interruption happening in their homes. Frequent water supply interruptions an affet various daily ativities and ompliate onsumers in many ways. The third attribute is Non Revenue Water (NRW). High rates of NRW an be viewed as poor performane of delivery servies to onsumers. Three levels are seleted suh as high, moderate and low. The targeted and expeted perentages of NRW are seleted as the attribute levels and they an be onsidered as a benhmark of their performane. Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

Mahirah Kamaludin et al. 64 Table 1: List of attributes and their Levels Seleted for the Study. Table 2: Example of CE Question in Questionnaire. Table 3: Total of Respondents Aording to the Distrits. Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

IMPROVEMENTS IN DOMESTIC WATER SERVICES IN KELANTAN 65 Monetary attribute in this study is water prie (WP). Inlusion of prie attribute is important in CE studies as it permits the estimation of marginal willingness to pay for other attribute. The adjustment in the pries brings a major impat on onsumer s budget deisions. For the attribute levels, this study hooses the national average water prie whih is RM 0.65 on the first 35 ubi meters and maximum water prie in the ountry whih is domesti average water prie in Johor (RM0.98 on the first 35 ubi meters). The seletion of attribute levels is based on urrent water prie of the state, national average prie and maximum water prie in Malaysia. Therefore, onsumers deided to selet the best prie in order to have higher level of servies. During the survey, the respondents have been told that hanges in water prie will affet their budget alloation for other expenditures. Respondents an hoose status quo or urrent servie option if they do not intend to have any hanges in the servies and they are very omfortable with the urrent situation. They need to hoose the best option for hanges in water servies based on their best preferene. Table 2 portrays an example of CE question in the questionnaire. Suppose Servie Option 1 and 2 below are the only possible alternatives to the Current Servie Option for domesti onsumers, whih option do you prefer? (Please hoose ONE and tik in the box). The sample onsists of 552 respondents in every distrit in Kelantan who are registered domesti users of Air Kelantan Sdn. Bhd. The sample size is valid as it represents the population of Kelantan whih is 1.6 million in 2011 (Department of Statisti, 2012). They have been told that the study will help the water industry and the water ompany to understand onsumers expetations regarding water servie improvement. The data olletion was onduted in four months starting in September 2011 until Deember 2011. Table 3 shows the total number of respondents who were seleted based on the distrits in the final survey. This study used random sampling in all distrits sine the survey sample is from a broad population and it is problemati to identify every member in the population. In household surveys, the distribution of households is bigger in urban areas (Kota Bharu) than in rural areas (Jeli). Besides, the distribution of the respondents was not equal in eah distrit beause of the distrits do not have the same total number of population. The respondents were randomly seleted in distrits but we were restrited to take 3 perent from the number of sample in eah distrit in order to ome out to 552 respondents from the all distrits in the survey. Results Table 4 presents the average age of respondents is 38 years old and most of them are female (50.2%). Average household size is 5 people with average household inome at RM 4077.90 per month. Estimation proedures for CM have been ompleted by using eonometri software whih is LIMDEP, NLogit Version 9. The findings are the results of the basi and interation onditional logit (CL) models. Table 5 shows the results of the estimated CL model for eah attribute levels in the Choie Experiment. In general, attribute signs are as expeted exept for a few attributes suh as never in water interruption attribute, moderate and low in non revenue water attribute. The variables that are insignifiant may be aused by the lose link to the base level whih makes people prefer the base level. The sign of water quality attributes are both positive at level two (good) and three (very good) signifiant at 1% level. The situation shows that people hoose to have an improvement in water quality sine they realize it is an important resoure in daily life. Poor water quality may threaten their family health onditions. The attribute level of water interruption (sometimes) shows a positive in its relation, implying that people prefer disruption in water servies about one a year ompared Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

Mahirah Kamaludin et al. 66 Table 4: Soio Demographi Charateristis of Respondents. Table 5: Results of Conditional Logit Model (Model 1). to frequent water supply interruption (urrent ondition). However, when the respondents were offered with the new level 3 of water servies (never been interrupted), they have rejeted it beause they believed it is impossible to be implemented. The NRW attribute demonstrates negative signs in the model. The people believed that the water ompany should be aountable in reduing non revenue water and should bear the ost of improving infrastrutures. The basi CL model an be more preise by improving the model fit itself. This study hooses to inlude soioeonomi attributes in the model in order to improve the model fit. The insertions of soio-eonomi variables also provide heterogeneity in hoies (Nam, 2004). The insertion of new attributes slightly improved the model fit ompared to the base model. The soio-eonomi Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

IMPROVEMENTS IN DOMESTIC WATER SERVICES IN KELANTAN 67 Table 6: CL Model with Interation. variables interat with the main attributes that help to evade the diffiulties in Hessian Singularities. Hene, the new interations offer a rih data set based on preise onsumer preferenes at every levels. The result of CL interations model is shown in Table 6 below: By adding the soio-eonomi variables with the new interations, the model fit progress has improved. The log likelihood gets better ompared to Model 1 (Table 5), as shown by inreases in value from -2823.53 to -2757.027. The Pseudo R 2 also inreases ompared to Model 1 (0.0688 to 0.0907), and adjusted R 2 also inreases onsiderably from 0.0676 to 0.0883. Thus, it indiates that the model has improved until it has a high explanatory power. It implies that a preise model speifiation has been aomplished. The sign of QUAL is both positive at level two and three and signifiant at 1% level. The situation shows that people hoose to have an improvementin water quality sine they realize the effets of bad water quality espeially on their health. The main attribute of INT2 (level 2) demonstrates positive relationship, implying that people prefer sometimes rather than frequent water supply servies interruption (urrent ondition). However, when the respondents were proposed with the new level 3 water servies (INT3), they rejeted it. It proves that respondents still hoose the urrent situation of water servies. The oeffiient of NRW2 is Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

Mahirah Kamaludin et al. 68 positive in its relation implying that onsumers prefer an improvement in redution of NRW problems in the state, as they deide to move away from urrent level of NRW (status quo). The interations with soio-eonomi fators produe a negative sign to the variable of NRW3_Ageand signifiant at 1% level, whih indiates that young people tend to support redution of NRW in water servies ompared to older people in the state. It implies that the younger generation is aware that NRW an reate losses to the ountry and it may have a bad effet on the industry. Eduation is signifiant and illustrates a positive relationship in these variables suh as INT3_Eduation and NRW2_ Eduation. Respondents with higher level of eduation are onerned with water interruption and non-revenue water problems as shown by signifiant level both at 1%. This group of respondents ertainly believed that interruption in water servies and NRW problems may affet their onsumption and they plae these as priority elements in water onsumption. Respondents with higher household inome really take water quality and disturbane in water servies very seriously as indiated by the positive signs for the variables QUAL3_Inome and INT2_Inome, with eah variable signifiant being at 1% level. Thus, it is onsistent with the eonomi theory whih says that WTP inreases with level of inome. Conversely, the variable NRW3_Inome shows a negative sign at 1% signifiant level. It implies that some people with higher inome are not onerned withnrw issue as they believe that the water ompany as water provider should solve the problems. Gender variables, QUAL2_Gender and QUAL3_Gender are positive in their relationships and signifiant at 1%. The relations showed that male has a tendeny to be more onerned about water quality ompared to female sine probably male are aountable as head of household. Results on Marginal Values of Willingness to Pay Estimated oeffiients for all variables are used to measure the outome of hanges in attributes based on the prie that onsumers are willing Table 7: Marginal Values for Differene in Attribute Levels. to pay for hanges in water servies. Marginal WTP are alulated based on Equation 9. Table 7 demonstrates the results of CL basi and interation models, for estimated marginal values of the differenes in attribute levels. The result of MRS at level satisfatory to good is at RM0.29 in Model 1. An improvement of servies from level of good to very good is at RM0.28. These values illustrate how muh respondents are willing to pay for any improvement in water servies for eah level. Model 2 shows that the value of MRS is higher at satisfatory to good whih is RM0.21 ompared to good to very good at RM0.01. There is a large differene in both values in this model. The marginal values for differenes in water supply interruption attribute for improvement in level frequently to sometimes is at RM0.44 in Model 1 and RM0.26 in Model 2. Meanwhile, the differenes in MRS values at sometimes to never is at negative values RM0.48 in Model 1 and RM0.1 in Model 2. The values of differenes in MRS at Non revenue water attribute demonstrate relatively low values ompared to other attributes and most levels inurred negative values. Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

IMPROVEMENTS IN DOMESTIC WATER SERVICES IN KELANTAN 69 Conlusion This study applies CE method to measure onsumer preferenes for improvement in domesti water servies. The CE offers benefit transfer potential through its water servie attributes and their levels. The respondents deided their preferred option whether to move on from urrent ondition to improved ondition. This study revealed that onsumers are willing to pay new proposed water pries as long as they are guaranteed of a high standard in water servies. The CE offers options for trade-off suh as water quality (QUAL), water supply interruption (INT), Non revenue water (NRW) and water prie (WP). The study alulated marginal values and it showed the monetary value that people are willing to plae for eah hange in attributes. The findings demonstrated that people are willing to ontribute more on water supply interruption attribute sine it produes highest marginal value from base level to moderate level. People prefered that the disruptions are redued to at least one a year. Presently, the water interruption is too frequent and it is worsening the situation in their homes and disturbing many ativities. The findings of the study an onvine the government to alloate more resoures and finanial aids, suh as soft loans to the water ompany. This ation will benefit those involved in the industry suh as poliy makers, water ompanies and onsumers. The government should regulate the market by ontrolling water pries so as to be in balane that protet both sides; produers and onsumers. When the government sets the eiling pries and the water ompany is unable to over the high ost of prodution, then the government should play its role to help the industry by giving subsidies. Sine water is onsidered as a publi good, it should be provided to the onsumers in the best ondition. From water provider s view, the higher tariff an be used for improving infrastruture to provide higher level of servies. It an be used for further maintenane works suh as the replaement of old and rusty pipes, upgrading water treatment plants and reduing Non Revenue Water in the state. Revising pries an help the water ompanies to redue ost somewhat but they still have to operate until the optimum prodution is ahieved. By inreasing the pries the publi will be aware of the importane to onserve water and avoid water wastage. It will lead to eonomi effetiveness of water utilization among onsumers. Consumers need to be eduated about the speifi value and benefits that they would get if the servies are improved. What they would gain and what they would lose if the program of imposing higher water pries is implemented will affet their utilisation of water. Referenes Adamowiz, W., Louviere, J., & Williams, M. (1994). Combining Revealed and Stated Preferene Methods. Journal of Environmental Eonomis and Management, 26: 271-292. Assoiation of Water and Energy Researh Malaysia. (2011). National Water Servies Industry Restruturing Kelantan State Case Study. Retrieved from: http://www. awer.org.my/publiation/report/national_ water_servies_industry_restruturingkelantan_ase_study_eng.pdf Bennett, J., & Blamey, R. (2001). The Choie Modelling Approah to Environmental Valuation. Massahusetts: Edward Elgar Publishing Limited. Department of Statisti. (2012). Kelantan at a Glane: Population in 2011. Retrieved from: http://www.statistis.gov.my/portal/ index.php?option=om_ontent&id=526&l ang=en&negeri=kelantan Do, T. N., & Bennet, J. (2007). Estimating Wetland Biodiversity Values: A Choie Modeling Appliation in Vietnam s Mekong River Delta (Working Paper EEN0704). Program at Australian National University Eonomis and Environment Network. Eonomi Planning Unit. (2009). Inidene of Poverty, by Ethniity, Strata and State, Malaysia 1970-2009. Household Inome Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70

Mahirah Kamaludin et al. 70 and Poverty. Retrieved from: http:// www.epu.gov.my//doument_library/ get_file?uuid=5bf3a7a-37a7-4ebb-966-634e17141d&groupid=34492 Farolfi, S., Mabugu, R. E., & Ntshingila, S. N. (2007). Domesti Water Use and Values in Swaziland: A Contingent Valuation Analysis. Agrekon, 46 (2): 157-169. Hanley, N., Wright, R. E., & Adamowiz, W. (1998). Using Choie Experiments to Value the Environment: Design Issues, Current Experiene and Future Prospets. Environmental and Resoure Eonomis, 11(3-4): 413-428. Hensher, D. A., Shore, N., & Train, K. (2005). Households willingness to Pay for Water Servie Attributes. Environmental & Resoure Eonomis, 32: 509-531. Larson, D. M., Lew, D. K., & Onozaka, Y. (2001). The Publi s Willingness to Pay for Improving California s Water Quality, Annual Meeting of Western Regional projet W-133. Miami: Florida. Lee, C. H. K. (2011). Privatization, Water Aess and Affordability: Evidene from Malaysian Household Expenditure Data. Eonomi Modelling, 28: 2121-2128. Malaysian Water Assoiation. (2011). Malaysia Water Industry Guide. Kuala Lumpur: MWA. MFadden, D. (1974). Conditional Logit Analysis of Qualitative Choie Behaviour. In: Zarembka, P. Frontiers in Eonometris. New York: Aademi Press. Mohd, R. Y., Alias, R., & Khairil, W. (2008). Eonomi Valuation of Marine Parks Eotourism Malaysia: The Case of Redang Island Marine Park. Serdang, Malaysia: Universiti Putra Malaysia Press. Mohd, R. Y., Alias, R., & Zaiton, S. (2011). Willingness to Pay for Domesti Water Servie Improvement in Selangor, Malaysia: A Choie Modelling Approah. International Business and Management, 2(2): 30-39. Nam, P., & Son, T. (2004). Household Demand for Improved Water Servies in Ho Chi Minh City: A Comparison of Contingent Valuation and Choie Modelling Estimates. In Researh Report 2005-RR, Eonomy and Environment Program of South East Asia. Retrieved from: http://www.eepsea.org National Eonomi Planning Unit. (2006). 9 th Malaysia Plan. Kuala Lumpur: NEPU, Malaysia. National Eonomi Planning Unit. (2006). 10 th Malaysia Plan. Kuala Lumpur: NEPU, Malaysia. Perni, A., Jose, M. P., & Federio, M. C. (2012). Soial Preferenes and Eonomi Valuation for Water Quality and River Restoration: the Segura River, Spain. Water and Environment Journal, 26: 274-284. Vasquez, W. F., Mozumder, P., Hernandez-Are, J., & Berrens, R. P. (2009). Willingness to Pay for Safe Drinking Water: Evidene from Parral, Mexio. Journal of Environment Management, 90: 3391-3400. Journal of Sustainability Siene and Management Volume 8 (2) 2013: 61-70