Willingness to pay and municipal water pricing in transition: a case study

Abstract

Water utilities in Calcutta, India are historically subsidized. This has kept residents quite unfamiliar with the concept of volumetric pricing of water. But pricing strategies have recently been receiving attention and the charges regime appears to be in transition. However, pricing with the objective of full cost recovery may not be feasible or even desirable in the transitional phase. The purpose of the article is to formulate a water pricing principle by using consumers' willingness to pay (WTP). A contingent valuation study revealed that people were willing to pay a charge of Rs. 3.18 per kiloliter for a monthly average household water use of 12.59 kiloliters. Such a principle, which considers consumers' WTP, may neutralize the political and public resentment in the phase of transition and can be a base for full cost recovery in the future. Respondents also expressed an opinion that the poor should receive some support from the government and contribute a part for smooth functioning of the water supply system.

Keywords:

• contingent valuation
• volumetric charge
• water pricing
• willingness to pay

1. Introduction

Revenue sufficiency is one of the important objectives for pricing municipal water supply. But pricing with the objective of full cost recovery is expected to be legally and politically unacceptable and thus not feasible or even desirable in the phase of transition. A transitional phase implies a phase when a historically subsidized system is moving into a system where cost recovery is seriously targeted. A possible way out, in the transitional phase, is to obtain consumers' willingness to pay (WTP) on payment of water dues to the municipal corporation. This could be done by a contingent valuation (CV) study, based upon a survey of a suitable sample. A water pricing principle, which considers consumers' WTP, can be a useful tool and a starting point for full cost recovery in the future.

Water utilities in India are in the transitional phase. Researchers have tried to investigate whether consumers are at all eager to pay for water. Griffin et al.'s (1995) study in Kerala, Baidya's (1995) study in Baroda, Reddy's (1999) study in Jaipur and Jodhpur, Zerah's (2000) study in Delhi, the World Bank (2001) study in Calcutta, Raje et al.'s (2002) study in Mumbai, and Dasgupta P and Dasgupta R's (2004) study in Delhi have found that consumers are willing to pay for better and more reliable water services.

Calcutta, the capital of West Bengal, India, is a city of 12 million people. The jurisdiction defined by the Calcutta Municipal Corporation (CMC) divides Calcutta into 141 wards. The CMC is responsible for supplying drinking water to its citizens under section 234 of the CMC Act. The CMC relies heavily on the supply of surface water lifted from the river Hooghly by two treatment centres situated at Palta and Garden Reach. Additionally, the CMC installs small and large diameter tubewells utilizing groundwater. Households often install tubewells within their premises especially in areas where surface water supply is absent or insufficient. The water demand of the poor in Calcutta is met by municipal stand post water supply. It is to be noted that the present stand post water supply that caters for the needs of the poor is characterized by huge wastage of water. A sample survey undertaken indicated that about 95 million litre per day (MLD) of water flows through these stand posts of which 50 MLD is used by consumers and the balance of 45 MLD is wasted (World Bank 2001). Apart from unidentified water, the municipal water infrastructure in Calcutta suffers from intermittent supply, inadequate pressure, high energy consumption, dependence on groundwater etc. But unacceptable water quality is the most important concern for citizens of Calcutta. Unacceptable water quality compels households to make expenditure on home purification methods such as filters. Though home purification methods are employed, households still suffer from waterborne illnesses such as diarrhoea and have to spend more money on treatment of these diseases. Households also have to spend a reasonable time on using home purification methods and treatment of waterborne illness.

In the absence of water meters, water pricing in Calcutta is based on annual property tax and the size of the ferule 1 of the supply line. Commercial and industrial customers and a few domestic consumers having connections of ferule size 20 millimetres and above have been paying water charges since the 1980s. It is to be mentioned that domestic consumers are very rarely provided with connections with a ferule size above 15 mm and thus most domestic consumers do not pay anything for water. Recently, the CMC has taken steps to include a large section of domestic consumers within the sphere of water tariffs. Thus the water pricing system in Calcutta is in a phase of transition. More detailed information about water infrastructure and water pricing in Calcutta can be obtained from Majumdar and Gupta (2007).

In Calcutta, the direct subsidy to water systems has resulted in huge budgetary expenditure on the part of the municipal authorities. Lack of finance has obstructed proper development of the infrastructure in Calcutta. The World Bank (2001) report estimated that the finance required for improvement of the water infrastructure in Calcutta is ∼264 million US $ at the 2001 price level. But income generation from the sector is very low. A decadal analysis between 1992 and 1993 to 2001 and 2002 revealed that the expenditure on water supply and sewerage has increased five times whereas the earning has only doubled. Inability to extend the surface water supply has resulted in indiscriminate withdrawal of groundwater that has far reaching hydrological effects. The Calcutta Metropolitan Development Authority (CMDA) has already declared areas like Park Street and Moulali in Calcutta as dry areas. Thus a rational pricing policy, with the objective to mop up resources as far as possible, is necessary for proper management of the system. There is ongoing debate within the municipal corporation and among the public regarding introduction of a suitable water charge. The major opposition political party in West Bengal stubbornly opposed the proposal to introduce a water charge. There was news of difference of opinion within the ruling party over the issue.

The purpose of the present article is to propose a water charge based on the WTP estimate. It also investigates how the poor can be safeguarded during the transitional phase. The poor in Calcutta are people living below the poverty line having a per capita monthly expenditure of Rs. 352 or less at the time of the survey. The investigation is limited to asking the non-poor households about the method of water provision for the poor. The reason for asking the non-poor households rests on the assumption that any subsidy to the poor must be borne by non-poor households and unless the non-poor support the idea of a subsidy to the poor, this may not become feasible. This is crucial that the equity concern during the phase of transition is given proper attention.

The rest of the article is organized as follows. Section 2 describes the methodology used in this study. Section 3 is on the survey design and procedure. Section 4 is on the CV design issues. The model for estimation of WTP is in section 5. The results obtained from the CV survey are described in section 6. Determining factors that influence consumers' WTP were identified and reported in this section. Section 7 is on policy implications. The implication of a pricing principle suitable for the transitional phase and support schemes for the poor are discussed. Section 8 gives a concluding summary.

2 Methodology

CV is a method that uses survey questions to elicit peoples' WTP for non-market goods and services by creating a hypothetical market and a means of payment. CV studies have become more and more acceptable as a useful tool to estimate WTP for desirable quality water and are often used (Jordan and Elnagheeb 1993, Kwak and Russell 1994, Griffin et al. 1995, Whittington et al. 2002, Dasgupta and Dasgupta 2004) as an instrument where markets are absent, imperfect or incomplete. Studies by Kwak and Russel (1994), Whittington et al. (2002), Dasgupta and Dasgupta (2004) have proved that a CV study, though unfamiliar in developing countries, is applicable for assessing WTP for water.

One important issue in a CV study is the questioning format. Traditionally, most CV surveys used an open-ended format for the WTP question. In an open-ended format, respondents are asked to report her maximum WTP. There is evidence that respondents find it difficult to answer an open-ended question. Kriström (1990) has shown evidence of lower response rates and higher item non-response rates with an open-ended question. Strategic behaviour also poses serious problems in the case of an open-ended valuation question. Bidding games have been used to elicit values in some studies (Randall et al. 1974). Desvousges et al. (1987) have shown that the bidding game format may lead to starting point bias. The dichotomous choice type question is another questioning format, which has been widely used in recent CV studies. The dichotomous choice technique asks respondents to state whether they will be willing to pay specific amounts and the amount is varied across respondents and ‘yes’ and ‘no’ answers are elicited. Hanneman (1984, 1985) argued that a dichotomous choice question yields a more meaningful response than an open-ended format. Blamey et al. (1999) argued that respondents might not respond to a dichotomous choice valuation question carefully and there is a chance of ‘yea-say’, that is respondents will say ‘yes’ without paying attention to the stated amount. Thus, there is a need to make a consistency check. A dichotomous choice question, if followed by an open-ended question, offers the opportunity to check whether the respondent answers consistently for both the set of questions. Deshazo (1999) employed a double bounded referendum with an open-ended follow-up question to estimate WTP for 24 h per day water service with adequate pressure and agreeable quality. A similar follow-up questioning strategy was used by Ready et al. (2002) for a CV study in Latvia.

A measure of WTP generated from a valid CV survey, if used for implementing a water charge, may not recover the full cost of operation. Saleth and Dinar's (2001) study in Hyderabad, India found net WTP to be inadequate to justify supply augmentation options. This is probable as WTP depends on how much the respondents can afford. But it is obvious that such system of pricing is able to mop up a reasonable quantity of resources that can be used for proper maintenance of the infrastructure.

3 Sample selection and survey design

In this survey, for collection of the data, the wards of the CMC were divided into five more or less equal parts. Wards receiving the municipal surface water supply completely and wards receiving it partially were separated in each part. One ward that receives the municipal surface water supply completely and one ward where the surface water supply reaches partially were selected randomly from each part. In this way, 10 wards were selected among which five receive the municipal surface water completely and five receive it partially. Fifty sample households from each ward were taken. This led to selection of some households that already have a CMC surface water connection and some households that do not have such a connection. This was done to capture the difference in the WTP of households having and not having a municipal surface water supply connection in their residence. There is a chance that respondents already having a surface water connection may express a different WTP than households having no surface water connection.

For selecting samples in each ward, following a similar practice as the one for geography, a transect was drawn from north to south and some equidistant neighbourhoods were selected. A suitable number of households were chosen in each neighbourhood along a track. The number chosen from each neighbourhood was so adjusted that it mimicked the expenditure distribution of the city as found in a CMDA (Chatterjee et al. 1999) study of 20,000 households. The intention behind choosing purposive sampling as a method was to obtain representative data. As it was felt that a separate plan for water provision for the poor is desired, the survey was not conducted among households lying below the poverty line. The distribution of the total of 500 samples according to total monthly household expenditure class is listed in Table 1. Other details of the sample characteristics are available from Majumdar and Gupta (2006).

Table 1. Number of samples according total monthly household expenditure (percentages in parenthesis).

Expenditure class	Frequency
2001–3000	208 (41.6)
3001–5000	192 (38.4)
5001–7500	72 (14.4)
7501–above	28 (5.6)

Quantitative and qualitative data have been collected through a questionnaire that was drafted in Bengali to make it more understandable to the respondents. Trained surveyors collected information from the head of the household as far as possible by the direct interview method. The total number of households contacted for the survey was 568 from which the completed response of 500 households could be taken for the final analysis. Thus the response rate for the survey came out to be 88 percent. The average time taken to interview a particular respondent was found to be ∼16 min. This is in conformity with average time taken in direct CV surveys. Larson and Gnedenko (1999) in their study reported interaction of 12 min on average with the respondent. About 95.8 percent of respondents themselves acknowledged that the survey method was more or less understandable to them.

4 Contingent valuation survey design issues

The final preparation of the questionnaire for the study was preceded by verbal protocol and repeated pre-testing with draft versions of the questionnaire. The pre-test revealed that there is general awareness about undesirable water quality among the citizens of Calcutta. The pre-test was performed with great caution to improve the ordering of questions, the language of the questionnaire, the choice of the stated preference bids, the payment vehicle, etc.

The questionnaire described the structure of the hypothetical market before asking the WTP question. Respondents were informed about the impurities in the water they use and the diseases that these impurities may cause. They were informed about the water quality specification of World Health Organization (WHO) and the divergence of present water quality from the WHO standard. The respondents were told that the central government has recently directed municipal authorities to open the water sector for private operators and to consider imposition of rational water charges to recover the cost. Respondents were presented with a meaningful and understandable hypothetical market scenario where a private water operator will supply water of WHO specified quality. The private operator offers the initial connection free of cost but charges monthly for water use. A consumer can use the WHO quality water supplied by the private operator or use any alternative source supplying water of unreliable quality. A household can choose any form of water source but will have to bear some monthly cost whichever source of water she uses. The use of the WHO specified quality water is expected to reduce the monthly expense for home purification methods and waterborne diseases. But consumers have to face a more stringent economic constraint if they opt for the WHO quality water and might have to revise purchase decisions of other commodities. The hypothetical construct was meant to obtain the maximum amount of money the respondents would actually pay monthly for a connection supplying WHO specified quality water. The reference level is the present state of water supply with unreliable water quality and the target level is a private water supply system supplying water of a standard specified by the WHO.

A dichotomous choice valuation question was used primarily as an elicitation method. The bids used in this study were Rs. 40, 60, 80, 100, 120 and 140 per month per household. The values were randomly assigned to respondents so that each value is administered to an equal sub-sample. An open-ended WTP question followed the initial dichotomous choice WTP question. Alternative payment vehicles were tried during the pre-test and it was felt that a payment vehicle like ‘water tax’ is not suitable. It appeared that the word ‘tax’ associated with water was unacceptable to most respondents. The study used ‘higher expenditure as utility bill’ as the payment vehicle. A CV survey requires clear specification of the periodicity of the payment to be made. The WTP question informed respondents that they would have to pay a monthly bill for water of desirable quality.

The dichotomous choice question after describing the scenario asked ‘Are you willing to pay (one bid only for each respondent) each month for the amount of water your household presently uses to a private water operator supplying water of WHO specified quality as expenditure for water supply?’ The quantity aspect of the CV query was incorporated by asking respondents intricate questions about their present use of water and whether they were willing to pay a specific amount for that quantity. The average use was estimated as 93.5 litres per capita per day (LPCPD).The follow-up open-ended question asked the maximum amount the respondent would be willing to pay.

5 The model for willingness to pay estimation

For a respondent i, utility (U _i) is a function of the level of income (Y _i) and the water quality (W _i).The respondent is asked whether she is eager to pay a specific amount (A _i) for a water quality improvement programme from W ₀ to W ₁. The respondent's response (R _i) can be either ‘yes’ (R _i  = 1) or ‘no’ (R _i = 0). Here, the respondent faces a decision of choosing (Y _i,W ₀) or (Y_i−A _i,W ₁). The respondent will choose to respond yes (R _i = 1) if the utility obtained from a lower income, better water quality combination is greater than the utility level associated with a higher income, lower water quality combination. Here, the response made by the respondent is known but the exact information regarding the WTP is not available. The respondent has some unobserved true point valuation and if the offered bid is above that, she responds ‘no’, otherwise ‘yes’. Thus, the response of the respondent is a binary variable.

The responses obtained from a dichotomous choice WTP question can be used to estimate the probability of the occurrence of a yes response by applying the logit estimation technique. The probability of a yes response is given by

where Z, in the present context, is

These responses can be used to estimate average WTP by regressing R _i on A _i and average WTP = −β₀/β₁.

The follow-up maximum WTP question is an open-ended question from which a mean or median estimate of WTP can be obtained. As mean is sensitive to extreme values, median is preferred as an average estimate especially in policy formulation studies. Nevertheless, these estimates can be used to obtain some bounds of the WTP value.

The open-ended WTP question obtained the maximum amount respondents were willing to pay. It was assumed that the maximum WTP is a linear function of several demographic, socio-economic and other variables. Households making higher out-of-pocket expenditure and spending more time on home purification methods and treatment of waterborne illness and those expecting that the proposed program will make a substantial reduction in such expenditure are expected to pay more for water. Samples from areas where the municipal surface water supply reaches partially were taken to search whether households already within the scope of in-house municipal water connection were less willing to get connected to an alternative surface water connection supplying WHO quality water. The details of the explanatory variables used are listed in Table 2. The ordinary least square (OLS) method was used to explain the variation in maximum WTP and significant explanatory factors were searched.

Table 2. Descriptive statistics for selected variables.

Abbreviated form	Description of the variable	Maximum	Minimum	Average	Standard deviation
IHSURWD	= 1, If the household has an in-house CMC surface water connection	1	0	0.35	0.48
	= 0, Otherwise
MOUPAVIL	Total monthly out-of-pocket expenditure on home purification methods and treatment of waterborne illness in Rs.	485.73	0	28.46	50.43
DTLAVIL	Time spent daily on home purification methods and treatment of waterborne illness in minutes	90	0	8.71	12.98
XREDUCD	= 1, If expected reduction in expenditure on home purification methods and treatment of waterborne illness is substantial, when WHO standard water is provided	1	0	0.44	0.5
	= 0, Otherwise
MWTPBID	Maximum WTP per month expressed by respondents in Rs.	250	0	47.6	44.71
RESPSEXD	= 1, If the respondent is female	1	0	0.28	0.45
	= 0, Otherwise
RELGD	= 1, If the respondent is Hindu	1	0	0.89	0.31
	= 0, Otherwise
NFAMMEM	Number of family members in the household	13	1	4.49	1.75
PKIDD	= 1, If children below five years of age is present in the household	1	0	0.214	0.41
	= 0, Otherwise
MEANEXP	Average monthly family expenditure of the household in Rs.	25000	2500	4139.9	2516.42
YRSCHOL	Number of years devoted on education by the respondent	17	5	11.07	4.12
OCCUPD	= 1, If the respondent is employed	1	0	0.54	0.5
	= 0, Otherwise
Number of observations used 500.

6 Results

About 415 (83%) out of 497 respondents (three respondents could not answer the question) replied that they feel the importance of having a water supply connection supplying water of WHO specified quality in their residence. Only 115 (26%) respondents out of 448 (others did not reply) reported that such a supply of water would not reduce their present expenditure on home purification methods and treatment of waterborne diseases.

6.1 Willingness to pay estimate and the water charge

As expected, yes responses to lower bids were more than yes responses to higher bids. People are not expected to state their preference for a high monthly bid such as Rs. 120 or Rs. 140 for water, which is historically provided almost free to the residents of Calcutta. The mean WTP obtained from the dichotomous choice valuation question was found to be Rs. 59.68 per household per month. The mean and median WTP obtained from the follow-up open-ended question were Rs. 47.60 and Rs. 40 per household per month, respectively. Table 3 reports the mean and median WTP as found from the open-ended question by different expenditure groups and related statistics.

Table 3. Willingness to pay by different expenditure groups (willingness to pay as a percentage of monthly household expenditure in parenthesis).

	Willingness to pay in Rs.
Expenditure groups	Average	Median	Maximum	Minimum	Standard deviation	Confidence level (95%)
2001–3000	24.49 (0.98)	10 (0.4)	150	0	30.04	4.11
3001–5000	56.68 (1.42)	50 (1.25)	250	0	43.19	6.15
5001–7500	68.53 (1.10)	60 (0.96)	200	0	42.81	10.06
7501–above	103.21 (0.87)	100 (0.83)	250	10	54.19	21.02
Aggregate	47.60	40	250	0	44.71	3.93

It is common practice in CV studies to identify protest responses. There was a structured question to identify protestors. The study found 17 protests. As the present CV study is used for policy formulations, protestors were included and the effect of such inclusion will lower the WTP value and thus yield a conservative WTP estimate. A very low amount of ‘yes-saying’ behaviour was found. Those responses were made consistent by considering the open-ended follow-up response.

The present study estimated average household monthly use of water and the WTP for the said quantity if WHO standard water is provided. Respondents were asked intricate questions to estimate per head per day use of water. The average use stated by respondents was estimated to be 93.5 LPCPD. The estimates of WTP and water use were used to calculate the rate at which a water charge can be imposed. Thus, the proposed pricing principle incorporates consumers' WTP for the amount of water consumers are expected to use. As the transitional phase pricing structure should be conservative, the actual charge was calculated by dividing the median WTP, obtained from the follow-up open-ended question, by the monthly household water use. As average household size obtained in this sample was 4.49, the household use of water for a month came out to be 93.5 × 30 × 4.49 = 12.59 kiloliters. Thus, the calculated charge was 40/12.59 = Rs. 3.18 per kiloliter.

6.2 Explanatory variables of willingness to pay

The purpose of the section is to identify the explanatory factors of stated preference behaviour and to check factors that explain variation in WTP. The open-ended WTP question obtained the maximum amount respondents were willing to pay. The total number of observations used for investigating variation in WTP with the open-ended follow-up was 500 and 483 when protest responses were included and excluded, respectively. The OLS method was used to explain the variation in the dependent variable maximum WTP bid (MWTPBID).

Residents of Calcutta use drinking water either from a ground water source or from a surface water source. Many households, in this study, were found using a ground or a surface water source situated outside their residence, but some households have an in-house CMC surface water connection. It was investigated whether households already having an in-house CMC surface water connection were less willing to get connected to an alternative surface water connection that will supply the WHO quality water. Hence, a dummy IHSURWD was taken where in-house surface water connection is represented by unity, zero otherwise. The variable was found significant with a negative sign. The sign implies that households which already have an in-house surface water connection have lower WTP in comparison to households not having an in-house surface water connection. Table 4 supports such findings. The mean or median WTP for the households which already have a municipal surface water supply connection is less than those not having a surface water supply connection. Jordan and Elnagheeb (1993) in their study found that private well water users were willing to pay more than households using the public water supply system. They concluded that as people connected to the public water supply system are presently paying the water companies for water, they are less willing to pay. Moreover, it may be more economical for households having no in-house surface water connection to have an in-house surface water connection.

Table 4. Willingness to pay (per month in Rs.) for different consumer class.

	Willingness-to-pay
Consumer class	Average	Median	Maximum	Minimum	Standard deviation	Confidence level (95%)
Households already having an in-house CMC surface water connection	41.41	30	150	0	38.96	5.83
Households not having an in-house surface water connection	50.90	40	250	0	47.22	5.15

Expenditure incurred on home purification methods and treatment of waterborne illness is an expenditure already incurred by a household before the specific WTP query in the survey is asked. It is expected that households making substantial expenditure on purification methods or for medical treatment would be willing to pay more for the WHO standard water. This is expected that a desirable quality water supply will reduce expenditure on home purification methods and treatment of waterborne illness. Households incur out-of-pocket expenditure on home purification methods and treatment of waterborne illness. They have to spend time on such activities. Thus, expenditure on home purification methods or treatment of waterborne illness comprises of out-of-pocket expenditure and opportunity cost of lost time due to the activities. Monthly out-of-pocket expenditure on home purification methods and treatment of waterborne illness jointly (MOUPAVIL) was taken as an explanatory variable. The variable was found significant at the 5% level whether the model incorporates protestors or not. The positive sign implies that people making higher out-of-pocket expenditure on home purification methods and treatment of waterborne illness were more eager to pay for an alternative water supply of desirable quality. It was found that a 100% increase in MOUPAVIL results in about 7.6% increase in the WTP. Lost time daily on such activities (DTLAVIL) was found significant for both the cases. The negative sign implies that people losing more time in such activities have a lower chance to pay for the improved service. The puzzling sign might be due to the fact that people have enough idle time to spare and put a low value on time.

If a household gets the opportunity of using water of WHO standards, then expenditure on home purification methods and treatment of waterborne illness is liable to decrease. Respondents were asked, in the survey, whether such alternative supply would reduce their expenditure on home purification methods and medical treatment substantially or not. A substantial reduction is defined as that reduction when the respondent perceives that the original expenditure will be reduced by half or more than half. A respondent who perceives a substantial reduction is expected to pay more for the improved service. A dummy (XREDUCD) where substantial reduction is represented by unity, zero otherwise should be a significant explanatory variable with a positive sign. The variable was found significant for both cases at the 1% level. This supports the expected result that respondents expecting substantial reduction in expenditure on home purification methods and medical treatment of waterborne illness are likely to pay more.

Alberini et al. (1997) found educational attainment to be a statistically significant determinant of the WTP. Jordan and Elnagheeb (1993) have also found the education variable to be positive and significant indicating that respondents with a higher level of education were willing to pay more than respondents with a lower level of education. Awareness really matters in a CV study where a hypothetical scenario is presented to the respondents. As respondents in Calcutta are not familiar with such studies, awareness gains added importance. The level of education was listed as the number of years devoted to education by the respondent. It was found that the variable is significant at the 1% level. The effect of a year increase in education of the respondent will lead to an increase in the WTP by about Rs. 3. The sign implies that awareness affects WTP positively.

The number of family members was not found significant in explaining variation in the WTP. Concern about children is supposed to affect WTP positively. The dummy presence/absence of children in the household (PKIDD) was found significant at the 10% level when protestors were included. Richer households are expected to pay more for desirable quality water. Total monthly household expenditure (MEANEXP), taken as a proxy of household income, was found significant at the 1% level for both the cases. The effect of a Rs. 1000 increase in MEANEXP on the WTP was found to be about Rs. 6. Jordan and Elnagheeb (1993), Alberini and Krupnick (2000) have also found that WTP increases with income. Male respondents were found to be less willing to pay for water quality improvement than female respondents by Jordan and Elnagheeb (1993). Studies of Vanliere and Dunlap (1980), Poe and Bishop (1999) suggest that the sex of the respondent is not related to the WTP decisions. The result found in this study supports such findings. A dummy variable related to occupation was not found significant for both the cases. The regression results obtained are summarized in Table 5.

Table 5. Explanatory factors of willingness-to-pay.

Explanatory variables	Dependent variable = MWTPBID (Model including protest bids)	Dependent variable = MWTPBID (Model excluding protest bids)
Constant	−8.1015 (0.3767)	−7.0938 (0.4475)
IHSURWD	−8.1409 (0.0409)∗∗	−6.8573 (0.0911)∗∗∗
MOUPAVIL	0.0764 (0.0274)∗∗	0.0823 (0.0268)∗∗
DTLAVIL	−0.3942 (0.0074)∗	−0.3997 (0.0099)∗
XREDUCD	13.7316 (0.0000)∗	14.9795 (0.0000)∗
RESPSEXD	−1.1055 (0.7940)	0.1704 (0.9684)
RELGD	−2.6322 (0.6444)	−3.3793 (0.5639)
NFAMMEM	−0.1752 (0.8643)	−0.2351 (0.8196)
PKIDD	7.5435 (0.0834)∗∗∗	6.7404 (0.1255)
MEANEXP	0.0058 (0.0000)∗	0.0058 (0.0000)∗
YRSCHOL	2.8838 (0.0000)∗	2.8184 (0.0000)∗
OCCUPD	−0.6479 (0.8601)	0.0544 (0.9883)
Number of observations	500	483
Joint significance	F value = 23.79∗	F value = 22.92∗
R^2	Adjusted R^2 = 0.33	Adjusted R^2 = 0.33
∗ denotes significance at 1% level.
∗∗ denotes significance at 5% level.
∗∗∗ denotes significance at 10% level.
P values are reported in parentheses next to the coefficient values.

6.3 Water provision for the poor

There is a general consensus, in underdeveloped countries, that the government should take necessary steps to subsidize supplies of water for the poor. Respondents in this study were asked to state their view regarding water provision methods for the poor. 494 out of 500 respondents could give an answer. The opinions given are listed in Table 6. Only five respondents opined that the full expenditure for water provision should be borne by the poor. The result showed that people are in favor of some subsidy for the poor. There is also a growing opinion that the poor should pay a part of their water bill themselves.

Table 6. Opinions regarding water provision for the poor (percentages in parenthesis).

Opinion	Frequency
Poor should pay in full	05 (1.01)
Part government subsidy, part payment from poor and part cross-subsidy	118 (23.88)
Part government subsidy and part payment from poor	143 (28.96)
Full government subsidy to the poor	228 (46.15)

7 Policy implications

The purpose of the study is to formulate a water pricing principle that uses consumers' WTP. A pricing principle based on WTP may not be revenue sufficient but can be a useful tool of resource mobilization in the phase of transition. A transitional phase pricing structure should be simple, easily implementable and understandable. As seen above, the rate came out to be Rs. 3.18 per kiloliter. A water operator can employ such a pricing strategy in the transitional phase. Private utility operators can enter the sector even in a developing country city like Calcutta and have enough scope to mop up resources. As households already having the municipal surface water supply connection have expressed less WTP than households not having the connection, it may be more profitable to extend or open service in areas where the municipal surface water supply is yet to reach.

The study found the opinion of the non-poor respondents regarding water provision for the poor. The opinion can be incorporated in a water charge with some rebate offered to the poor. The rebate can be given for the amount of water needed for fulfilling basic requirements. It is estimated that, in an urban area, people require 40 LPCPD for such purposes. As the average household size in this sample was 4.49, the total water demand to meet up basic needs amounted to 180 litres per day per household. It is proposed that the households lying below the poverty line should face the same charge of Rs. 3.18 per kilolitre but will be entitled to get average basic water requirement of 180 litres per day per household free. The proposed pricing structure jointly incorporates proposals of subsidizing connections (Whittington 2003) and provision of a rebate to the poor (Hite and Ulbrich 1988). It is obvious that average WTP for water will be much lower for households lying below the poverty line in comparison to their counterparts. As people below the poverty line will face a rate that is above their WTP, they will be induced to prevent wastage. A poor household, according this pricing structure, can get a water supply connection free of cost and can use water to meet essential needs without making any monthly expenditure. But they would be required to pay for any usage in excess of that quantity. As the charge will be the same as faced by households lying above the poverty line, it will not invite administrative difficulty. The authorities have to offer a concession for households lying below the poverty line.

8 Conclusion

The present article used the CV data obtained from the residents of Calcutta to measure WTP for water. Information was purposively collected from a representative sample household lying above the poverty line. A dichotomous choice valuation question was asked and the estimated monthly WTP per household amounted to Rs. 59.68. The follow-up open-ended question yielded mean and median WTP of Rs. 47.60 and Rs. 40 per household per month, respectively. Respondents were asked about their per capita per day use of water. The average use stated by respondents was estimated to be 93.5 LPCPD. The proposed charge was calculated by dividing the median WTP, obtained from the follow-up open-ended question, by the monthly household water use. The charge was estimated Rs. 3.18 per kilolitre for monthly average household water use of 12.59 kilolitres. It is proposed that households lying below the poverty line should face the same charge but will get water for essential needs up to 180 litres per day per household free. Though a dichotomous choice question format is more in vogue to elicit WTP, the present study used the WTP obtained from the follow-up open-ended question format to propose a water charge. It is thought that a conservative estimate of WTP, obtained from the open-ended format, is more suitable to formulate a water charge in the transitional phase.

One limitation of the study is that it does not take into account the actual preference of the urban poor. Studies in Bangalore, India has shown that the poor in slums get community water supply connections and pay for it on a monthly basis. Poor people in Coimbatore, Tamilnadu are willing to pay for public taps if the reliability of the water supply is ensured. It is obvious that a charge should not be imposed on the poor in such a manner that they are prohibited from getting the basic water requirement. The proposed charge structure for the poor captures such sentiments.

The proposed charge of Rs. 3.18 per kilolitre will favour conservation of the resource and expected to check wastage of water. As there is no fixed part in the pricing structure, this will give enough scope of cross-subsidization for the low-income users. This charge, as it uses consumers' WTP, may not face political or public resentment at the time of implementation. The proposed charge, if implemented, has the potential to raise resources for modernization of the water infrastructure. Households, when they will be under a priced surface water supply network, will reduce water demand. Extension of surface water supply and consequent reduction in groundwater use would lead to positive hydrological consequences. The charge may act as a good starting point for full cost recovery in future. The proposed charge can be increased gradually so as to make it revenue sufficient.

Acknowledgements

The comments received from Eric Massey, the Managing Editor of Environmental Sciences and two anonymous reviewers were helpful in improving the content and organization of the article.

Notes

1. A ferule is a check valve, which allows water to flow in, but does not allow going back. The diameter of the ferule controls the volume of water.