In this study the contingent valuation method is applied in order to estimate the willingness to pay (WTP) of the inhabitants of Sucre (Bolivia) for an improvement in the urban water supply system. The study finds that about 55 per cent of households would be willing to pay an increase in their water bill for an improvement in the service. In order to deal with the problem of protest responses and the possible presence of a sample selection bias, a Heckman two-step model was estimated. More specifically, the econometric analysis undertaken reveals that there is no evidence of sample selection bias and that WTP positively relates to the respondents' household income, their level of education, the continuity of the water supply service, and the fact of being forced to carry water to cover their basic needs of drinking, cooking and hygiene.

Introduction

Considerable effort has been made in recent years to improve access to water worldwide as the situation is far from perfect, particularly in less-developed countries where about 884 million people do not have access to an improved water source (WHO/UNICEF, 2012). In order to create a water supply system and maintain and improve the service, there must be sufficient financial resources. Regardless of how each country decides to meet the cost of the investment necessary to accomplish this goal, the users of the service must contribute either partially or in full to the cost of the service through the water bill (Lee & Floris, 2003). The problem in developing countries is the low cost-recovery rates of these services as a consequence of the extra effort required, in relative terms, from households. Therefore, in this particular context, when water improvement projects are going to be implemented, they face serious difficulties in recovering investment costs since there is a huge gap between the finance required to improve the water supply system and the revenue generated by the existing water tariff system (Tarfasa & Brouwer, 2013).

Before making an investment, the local public and private authorities responsible for the urban water service will be interested in ascertaining users' willingness to pay (WTP) for the service. The literature on this topic has focused on various issues: improving access to water (Venkatachalam, 2006; Wang et al., 2010; Lee et al., 2013), service continuity (Hensher et al., 2006; Genius et al., 2008), water quality (Nallathiga, 2009; Bilgic, 2010; Polyzou et al., 2011) and wastewater treatment (Kontogianni et al., 2003; Genius et al., 2005).

In this paper, the contingent valuation method (CVM) (Mitchell & Carson, 1989) is applied in order to estimate how much the inhabitants of Sucre (Bolivia) would be willing to pay to benefit from an improvement in the water supply system. Better access to water and sanitation is necessary in Bolivia, and not only in Sucre, as lack of access is an important factor in explaining the high risk of early death (OMS, 2011). Bolivia, jointly with Guatemala, Honduras and Nicaragua, is one of the South American countries with the highest risks arising from inefficient access to water and sanitation (OMS, 2011). In Bolivia, the death ratio derived from diseases related to water is around 0.05 per 100,000 inhabitants (Prüss-Üstün et al., 2008). In the infant population the risk is even greater. Each year, around 30,000 children die in Bolivia because of diarrhoea caused mainly by diseases or parasites from unsafe water. Moreover, 46 per cent of Bolivian children younger than five years suffer diarrhoea caused by the lack of safe water access and a lack of hygienic habits such as washing hands with soap (UNICEF, 2009).

In a typical CVM survey, respondents are asked about their WTP for the hypothetical provision of a public good or their willingness to accept (WTA) for its hypothetical loss. The measures of value obtained represent the economic benefits (or costs) of the proposed change and therefore should be aggregated in a cost-benefit framework to obtain the social benefits (or costs) from public policies that usually improve (or worsen) social well-being (Hanley & Barbier, 2009). However, CVM is not a method free of controversy as critics argue that this technique is unable to generate reliable estimates of value, given its hypothetical nature and other sources of error (Hausman, 2012). Nevertheless, and despite these limitations, today a considerable body of evidence supports the view that contingent valuation done appropriately can provide a reliable basis for valuing well-defined public goods (Carson, 2012).

The case study presented is based on information gathered from 324 households. In order to deal with the problem of the large number of zero responses obtained, a sample-selection model was estimated following Calia & Strazzera (2001) and Strazzera et al. (2003). Another interesting feature of this research is that we analyse household WTP in a climate of confrontation between the community, the local government and the company that manages the water service. Thus, the study is interesting because it explores respondents' WTP in a city where the urban water supply system is deficient and people are dissatisfied with the company that manages the service. A priori, WTP would be expected to be lower than in other circumstances. The reasoning for this is that if the population is dissatisfied with the company that manages the service, they will react negatively to an increase in the water bill, thus the people of Sucre may decide that the company should improve the management of the service provided before they are asked to pay more for it.

The remainder of the paper is organised as follows. Section 2 analyses the causes behind the conflict in the management of the water service in Sucre that are influencing the decisions taken by the water utility company. Section 3 explains the survey process, the data and the methodology used in the study. Section 4 presents the results, and the concluding section presents the policy implications.

Causes of the water management conflict in Sucre

The Municipal Drinking Water and Sewerage Company of Sucre (Empresa Local de Agua Potable y Alcantarillado de Sucre (ELAPAS)) runs the water service in Sucre. ELAPAS is a publicly owned company that apparently operates independently from the city council, although the president of this company is the city mayor. A public concession from the Bolivian Ministry of the Environment and Water gave the right to exploit this utility to ELAPAS during a 40-year period that began in 1999.

The water service can clearly be improved since serious shortcomings remain (Guidi et al., 2013). However, the current climate of confrontation may hamper the implementation of the required improvements. Several factors contribute to the tense relationship between the company, the local government and the people of Sucre.

The first factor to highlight is the dispute over the area where the service is provided. According to a contract signed by both parties, ELAPAS must provide the service to five of the eight districts in the city, which account for close to 94 per cent of the population. The remaining 6 per cent of the population who do not reside in the districts covered by the service area, oblivious to those who signed the contract, are demanding that the local government and ELAPAS provide them with access to water. Meanwhile the problem is mitigated by distributing water from tankers.

The second cause for conflict is related to the supply cuts suffered by the neighbourhoods in the higher part of the city, which accounts for approximately 25 per cent of the population. The water supply is mainly based on a gravity system. As a result, service regularity to these neighbourhoods depends on the water level of the eight storage tanks located in the highest part of the city. When there is a shortage of water in the upper part of the city, the solution provided by the company is to distribute water from tankers.

Another source of frequent tension between the citizens and ELAPAS is that pipes burst frequently due to a lack of investment in improvements to an old and obsolete supply network, which, in turn, causes water supply cuts in the lower part of the city. The temporary solution, once again, is to distribute water from tankers until the burst pipe is repaired.

The lack of dialogue and understanding between ELAPAS and the community has already produced significant confrontations and social conflicts in the city, giving rise to demonstrations outside the ELAPAS main office (Unidad de Análisis y Conflictos, 2010).

To improve the poor service provided to the public, ELAPAS could consider, among other strategies, raising water prices in the city of Sucre. Thus, an increase in water bills could generate enough revenue to improve the service. The question we raise is: what would happen if the company decided to increase the price of water, bearing in mind the current climate of confrontation described above? Would the people of Sucre be willing to pay more for water if the company promised to improve the service in the current climate of conflict?

Survey process, sampling and methodology

Survey process and sampling

When implementing a CVM study, the design of the survey instrument is a crucial stage since the values obtained are dependent on the information provided to the respondents. Thus the pre-test of the questionnaire, along with focus groups and in-depth interviews, can be extremely valuable in determining the background information needed and the way to effectively communicate it to the respondents (Chilton & Hutchinson, 1999). In this study, a random sample of thirty-five inhabitants of the city of Sucre was used for pre-testing the questionnaire. This was very useful to make sure that everything worked as intended while allowing us to identify any potential problems. In particular, the wording of the valuation scenario was improved and the suitability of the payment vehicle was tested. In any case, the pre-test should be as large as your budget and time constraints allow (Whitehead, 2006).

After the pre-test, a survey of 324 households was conducted in the six urban districts of Sucre where ELAPAS has to supply water. According to the Bolivian Population Census, in 2012 the population of Sucre was 237,480 inhabitants while the number of households was 69,252, therefore the sample size represents approximately 0.5 per cent of the entire population. We did not take into account Districts 7 and 8, which are predominantly rural and where approximately 16,000 people live on farms and do handicrafts. Sampling was done by selecting different stratification variables (district, sex, age, income and education), then different strata were defined for each variable, and finally from each stratum respondents were chosen randomly. Stratified sampling, compared with simple random sampling, requires smaller sample sizes for the same margin of error (Daniel, 2012).

The survey was carried out between November 2010 and January 2011. The 52-question questionnaire used was divided into four sections as suggested by Bateman et al. (2002): (i) attitudes towards the environment, access to water at home, and level of satisfaction with the service provided; (ii) respondents' profile; (iii) contingent scenario with the WTP elicitation mechanism; and (iv) a final section with the respondents' main demographic variables, which help to interpret and validate WTP estimates.

The specific wording of the valuation scenario read to respondents was as follows:

‘As previously explained, the ELAPAS is intending to invest in improvements to the urban water service. As a result of these investments, all the people of Sucre would be guaranteed access to tap water in their homes. Furthermore, the water would not smell; it would be colourless and would not taste of anything and could be consumed directly from the tap without any danger to people's health. However, these improvement measures cost a great deal of money. Given limited public resources, in order to fund this policy all the citizens would be asked to pay a monthly increase on their current water bill. If the majority of households are in favour, this project will be carried out, while if a majority are against the proposal, then this service will remain as it is today.’

In a similar way to Polyzou et al. (2011), we used two valuation questions to ascertain respondents' WTP. The first question was:

‘Considering all the benefits that stem from this project, would you be willing to contribute financially to such a project?

1. Yes: ___ 2. No: ___ 3. Don't know: ___ ’

Respondents who answered affirmatively to this question were asked the following open-ended question in order to obtain their maximum WTP:

‘How much more would you be willing to pay in your monthly water bill in exchange for an overall improvement in the service? ___________’

Although in the wording of this question it was not mentioned that we were asking about the respondents' maximum WTP, respondents were verbally informed about this by the interviewers. The overall improvement in the service was defined as an improvement in each of the three following variables: the source of supply, continuity in the service without cuts, and quality of the water supplied.

Although the Report of the NOAA panel on contingent valuation (Arrow et al., 1993) recommends the use of dichotomous-choice questions for eliciting respondents' WTP, Genius et al. (2008) point out that it has long been recognised that the information conveyed by yes-no type questions is limited, while requiring the collection of data from a large number of interviews to be statistically efficient. Hence, as we faced severe funding constraints, the open-ended question appeared to be the most convenient despite its shortcomings (more difficult to answer, more prone to strategic behaviour, etc.). In the CVM literature on water resources, the use of open-ended questions is not an uncommon occurrence. For example, Birol et al. (2006) used this elicitation format to estimate the non-use values of a wetland in Greece, while Cooper et al. (2004) addressed the issue of motivation for contingent values in a study about water quality improvement in a lake.

Following Ramajo-Hernández & Saz-Salazar (2012), the payment vehicle used was an increase in the current water bill since it was considered the most appropriate with regard to the credibility of the hypothetical market, while being plausible and familiar to the population surveyed. In addition, this obligatory payment avoided the free-rider behaviour typical of voluntary payments (Carson, 1997).

Methodology

A major concern in the CVM literature is the treatment of zero responses since non-participation can have a substantial impact on WTP estimates (Lindsey, 1994; Haab, 1999; Dziegielewska & Mendelsohn, 2007). Hence this problem is frequently viewed as a threat to the validity of CVM in informing decision-making. In order to distinguish between ‘true zero’ values and ‘protest’ responses, CVM practitioners have long used debriefing questions to clarify the reasons behind the refusal to participate in the hypothetical market created. In our particular case, on the basis of previous research by Lindsey (1994) and Jorgensen et al. (1999), respondents who declined to pay were allowed to choose different reasons for their response, such as ‘I pay enough already’, ‘I cannot afford to pay’, ‘I consider it unfair to ask me to pay anything’, ‘It is my right to have a good water service’, etc.

The standard procedure for treating protest responses is to remove them from the sample. However, this may not be the correct procedure if the protest responses induce a selectivity bias, that is, when the group of protesters is significantly different from the rest of the respondents so these individuals are self-selecting when protesting.

To test for the presence of sample selection bias in our data set, we applied a two-step Heckman selection model that has often been used to check the existence of selection bias in CVM literature. This model consists of two steps: in the first step, the decision of the respondent to pay (‘yes’ response) or not to pay (‘no’ response) is modelled. In the second step, how much the respondents are willing to pay is modelled for all observations with a positive WTP. Thus the responses of the respondents can be modelled simultaneously using two equations: the first one is the ‘selection’ equation, and the second one is the ‘elicitation’ or ‘valuation’ equation.

Following Strazzera et al. (2003) and Messonnier et al. (2000), if Y1 denotes the amount an individual is willing to pay, Y2 is a dichotomous variable that takes the value 1 if the individual is willing to participate but 0 otherwise (does not participate), and x and z are vectors of explanatory variables for the valuation and participation equations respectively, then the valuation equation can be written: 
formula
1
where σ is a scale factor, Y1i is observed only when the individual participates in the market (Y2i = 1), and for the participation equation: 
formula
2
where ui and εi are independent and identically distributed (i.i.d.) normal errors. When estimating the model, in the first step the participation equation is estimated using a probit model, which in turn gives an estimate of the inverse Mills ratio (λ). The second step consists of an ordinary least squares (OLS) regression for the valuation equation (Y1i) where apart from x, λ is also included as an explanatory variable. If the coefficient of λ in this second regression equation is not significant, it indicates that selection bias may have not been present.

Results and discussion

The aim of this section is twofold. First, we try to identify the determinants of respondents' WTP through the construction of an equation that predicts WTP for the public good with reasonable explanatory power and coefficients with the expected signs, thus validating the results obtained from a theoretical point of view (Carson, 2000). Second, in order to deal with the non-response problem a Heckman two-stage model is estimated.

Of the respondents in the sample, 54.63 per cent stated that they would be willing to pay more to improve the water supply service (see Table 1). The main reasons behind a ‘no’ WTP response were that users believed they were already paying enough for water; that the company was inefficient; or that it was the responsibility of the public sector. Respondents stated that they would be willing to pay an average of 10.53 Bolivian pesos (1.47 US dollars) more per month to improve the water service. Regarding the spatial analysis of WTP, a lower value was obtained in District 1 (6.46 Bolivian pesos) with only 26.83 per cent of the respondents willing to pay, while District 3 exhibited the highest value (15.79 Bolivian pesos). This result will be reinforced when analysing WTP determinants in the following paragraphs.

Table 1.

Mean WTP and percentage of respondents with WTP > 0 by district.

District Mean WTP (Bolivian pesos) Std. Dev. % WTP > 0 
6.46 16.93 26.83 
10.17 17.07 56.39 
15.79 20.56 62.07 
8.65 11.63 56.82 
10.13 14.33 57.89 
11.3 8.18 80.00 
Full sample 10.53 16.75 54.63 
District Mean WTP (Bolivian pesos) Std. Dev. % WTP > 0 
6.46 16.93 26.83 
10.17 17.07 56.39 
15.79 20.56 62.07 
8.65 11.63 56.82 
10.13 14.33 57.89 
11.3 8.18 80.00 
Full sample 10.53 16.75 54.63 

Taking this full-sample mean WTP as a benchmark, and considering that the average water bill paid in Sucre amounts to 88 Bolivian pesos, it implies that the hypothetical increase in the water bill at the level of the individual household would be 12 per cent. Although it is difficult to compare different contingent valuation studies, since it is well known that the results of any contingent valuation study are sensitive to the assumed econometric specification (Bengochea-Morancho et al., 2005), we can say that WTP for an improvement in the water service in the city of Sucre is low in comparison with the results of similar research. For example, Casey et al. (2006) found that only 8 per cent of the population of Manaus (Brazil) said they would not be willing to pay for an improvement in the service; Genius et al. (2008) found that 29.4 per cent of a sample interviewed in the Municipality of Rethymo (Crete) would not be willing to pay more to improve the water service; and Arouna & Dabbert (2012) found that 94 per cent of respondents were willing to pay to improve their rural water supply in Benin. Nevertheless, there are also cases in which WTP is lower than in our study. For example, in the city of Mytilene (Greece), Polyzou et al. (2011) found that only 40 per cent of the respondents were willing to pay to improve water quality, while in a study conducted by Raje et al. (2002) nearly 50 per cent of respondents were ready to pay slightly more than their current bill amounts.

The explanation for this relatively low WTP could be the current climate of confrontation in the city of Sucre. If users are dissatisfied with the service provided by the company, they would be expected to provide greater opposition to an increase in the water bill. However, it should be said that the research provides no conclusive evidence of this relationship as the questionnaire did not include any questions aimed at capturing the ‘degree of conflict’ between the community and the water management company.

Table 2 shows summary statistics of the data relevant to the analysis, sorted by groups of respondents, considering some of the main variables used in the estimation as well as other variables related to the socio-economic profile of the respondents. It can be seen that the number of ‘true zero’ responses is 90, which amounts to 27.78 per cent of the entire sample, while the number of ‘protest’ responses is considerably lower (55) making up almost 17 per cent of the sample. This latter result is very similar to the one obtained by Strazzera et al. (2003). A noticeable result is that for all the variables apart from ‘bottled’ and ‘how much’ the differences between protesters and respondents with positive response are negligible. These two variables indicate, respectively, whether the respondents buy bottled water and how much they spend on bottled water. In this case, the respondents who protested showed higher values on these variables than the rest of the respondents. In the event that these two variables influenced the WTP to improve the quality of the water supply service in Sucre, we would expect the existence of sample selection bias.

Table 2.

Mean and standard deviations (in parenthesis) by groups of respondents.

 Type of respondent 
 All True zeros Protester Yes response 
Mean WTP (Bolivian pesos) 10.53 (16.75) 0.00 (0.00) – – 19.28 (18.57) 
Income 2.48 (1.02) 2.55 (1.05) 2.33 (1.00) 2.49 (1.00) 
Education 3.77 (1.11) 3.87 (1.06) 3.85 (1.11) 3.69 (1.15) 
Age 41.27 (15.58) 43.77 (14.98) 45.14 (16.74) 38.90 (15.20) 
Sex (male = 1) 0.43 (0.49) 0.41 (0.49) 0.42 (0.49) 0.43 (0.42) 
Family size 5.87 (2.53) 5.83 (2.32) 5.44 (2.42) 6.03 (2.66) 
Children 2.50 (1.70) 2.47 (1.64) 2.58 (1.90) 2.50 (1.66) 
Bottled 0.26 (0.44) 0.25 (0.44) 0.38 (0.49) 0.23 (0.42) 
How much (Bolivian pesos) 4.38 (10.30) 4.37 (9.48) 5.6 (10.23) 3.94 (10.75) 
N 324 90 55 179 
 Type of respondent 
 All True zeros Protester Yes response 
Mean WTP (Bolivian pesos) 10.53 (16.75) 0.00 (0.00) – – 19.28 (18.57) 
Income 2.48 (1.02) 2.55 (1.05) 2.33 (1.00) 2.49 (1.00) 
Education 3.77 (1.11) 3.87 (1.06) 3.85 (1.11) 3.69 (1.15) 
Age 41.27 (15.58) 43.77 (14.98) 45.14 (16.74) 38.90 (15.20) 
Sex (male = 1) 0.43 (0.49) 0.41 (0.49) 0.42 (0.49) 0.43 (0.42) 
Family size 5.87 (2.53) 5.83 (2.32) 5.44 (2.42) 6.03 (2.66) 
Children 2.50 (1.70) 2.47 (1.64) 2.58 (1.90) 2.50 (1.66) 
Bottled 0.26 (0.44) 0.25 (0.44) 0.38 (0.49) 0.23 (0.42) 
How much (Bolivian pesos) 4.38 (10.30) 4.37 (9.48) 5.6 (10.23) 3.94 (10.75) 
N 324 90 55 179 

The whole set of explanatory variables used in the estimation of the Heckman two-step model and their main descriptive statistics are listed in Table 3, while the estimated equations are shown in Table 4 (participation equation) and Table 5 (valuation equation), respectively. The two-step model was estimated using different covariate specifications and considering only those variables statistically significant at the 10 per cent level or higher. Model selection was done using a stepwise procedure. In the participation equation (Table 4), the dependent variable takes the value ‘1’ if the respondent decides to participate in the market, and the value ‘0’ otherwise. The explanatory variables considered in this first equation are: the family income; a dummy variable indicating what characteristics of the water supply service (supply source, continuity of the service, and water quality) the respondent would improve (water continuity = 1); a dummy variable indicating whether the interview was carried out in District 1 or elsewhere (dummy variables for all the districts in which the interviews were carried out were considered, however, only the statistically significant ones were kept in the estimated equations); a dummy variable indicating whether the respondent owns a plot of land; a dummy variable indicating whether the respondent buys bottled water to satisfy their consumption needs; and a dummy variable indicating whether or not the water pipes are installed outside the house.

Table 3.

Descriptive statistics of the explanatory variables.

Variable Description Mean Std. Dev. 
Income Respondent's household monthly income after taxes in five different intervals ranging from 0 to 8,000 Bolivian pesos 2.484 1.021 
Education Respondent's education in five levels (from illiterate to university degree) 3.771 1.199 
Age Respondent's age 41.27 15.58 
Sex Respondent's sex (male = 1) 0.425 0.495 
Family size Number of family members 5.873 2.527 
Children Number of children under 18 2.508 1.695 
Enough water Dummy variable about the respondent's perception related to the quantity of water supplied (enough water = 1) 0.589 0.492 
Tap Dummy variable about the source of water supply (tap connected to the supply net = 1) 0.808 0.393 
Pump Dummy variable about the source of water supply (tap connected to a water pump = 1) 0.033 0.181 
Public tap Dummy variable about the source of water supply (public tap = 1) 0.018 0.135 
Tanker Dummy variable about the source of water supply (water tanker = 1) 0.111 0.314 
Water well Dummy variable about the source of water supply (water well or waterwheel =1) 0.031 0.173 
River Dummy variable about the source of water supply (river = 1) 0.009 0.095 
Rain Dummy variable about the source of water supply (rain = 1) 0.008 0.096 
Transparency Dummy variable about the transparency of the water (transparent = 1) 0.759 0.428 
Taste Dummy variable about the taste of the water (good = 1) 0.691 0.462 
Smell Dummy variable about the smell of the water (smells = 1) 0.679 0.467 
Turbidity Dummy variable about the turbidity of the water (turbidity or solids in suspension = 1) 0.663 0.473 
Pipes inside Dummy variable about where the water pipes are installed (inside house = 1) 0.860 0.347 
External pipes Dummy variable about where the water pipes are installed (outside house = 1) 0.123 0.329 
Wquality Dummy variable about what characteristic of the water supply service – supply source, continuity of the service and water quality- the respondent would improve (water quality = 1) 0.636 0.482 
Wsource Dummy variable about what characteristic of the water supply service – supply source, continuity of the service and water quality- the respondent would improve (water source =1) 0.450 0.498 
Wcontinuity Dummy variable about what characteristic of the water supply service – supply source, continuity of the service and water quality- the respondent would improve (water continuity = 1) 0.547 0.498 
Carry water Dummy variable related to the fact that the respondent has to carry water home to cover the basic needs of drinking and hygiene (yes = 1) 0.089 0.286 
Trips Number of trips per day made to collect water outside the home 2.769 3.409 
Time Average time spent collecting water outside the home (minutes) 25.192 48.010 
Litres Litres of water collected outside the home per day 112.53 145.60 
Wexpense Water expenses (Bolivian pesos) 11.803 18.561 
Water cuts Dummy variable related to the quality of the water supply (if respondent suffers water cuts = 1) 0.518 0.500 
Satisfaction Level of satisfaction related to the water supply service (from 1 = very satisfied to 5= not satisfied at all) 3.250 0.948 
Bottled Dummy variable about the purchase of bottled water (yes = 1) 0.265 0.442 
Bottled expense Bottled water expenses (Bolivian pesos) 4.385 10.291 
Floor 4-category variable related to the quality of the house (floor= parquet flooring, tile flooring, concrete flooring, rough flooring) 2.713 0.701 
Roof 3-category variable related to the quality of the house (roof= tiled roof, concrete roof, straw roof) 1.437 0.509 
Wall 4-category variable related to the quality of the house (walls= concrete or stone, wood, adobe bricks, straw) 1.594 0.924 
Kitchen Dummy variable about where the kitchen is located (inside the house = 1) 0.891 0.310 
Bathroom 3-category variable about the type of bathroom (bathroom = complete bathroom, latrine, outdoor) 1.244 0.628 
Car owner Dummy variable about car ownership (yes = 1) 0.425 0.495 
Landowner Dummy variable about land ownership (yes = 1). The land is used for plant crops or for raising cattle. 0.175 0.381 
Abundance 5-category variable about the respondent's perception related to the family's abundance of resources (from 1 = very scarce to 5 = very abundant) 3.543 0.754 
Life satisfaction Level of satisfaction with life (from 1 = very satisfied to 5= not satisfied at all) 3.137 0.873 
District 6-category variable related to the district of the city in which the interview was conducted (from District 1 = 1 to District 6 = 1) 2.799 1.328 
Variable Description Mean Std. Dev. 
Income Respondent's household monthly income after taxes in five different intervals ranging from 0 to 8,000 Bolivian pesos 2.484 1.021 
Education Respondent's education in five levels (from illiterate to university degree) 3.771 1.199 
Age Respondent's age 41.27 15.58 
Sex Respondent's sex (male = 1) 0.425 0.495 
Family size Number of family members 5.873 2.527 
Children Number of children under 18 2.508 1.695 
Enough water Dummy variable about the respondent's perception related to the quantity of water supplied (enough water = 1) 0.589 0.492 
Tap Dummy variable about the source of water supply (tap connected to the supply net = 1) 0.808 0.393 
Pump Dummy variable about the source of water supply (tap connected to a water pump = 1) 0.033 0.181 
Public tap Dummy variable about the source of water supply (public tap = 1) 0.018 0.135 
Tanker Dummy variable about the source of water supply (water tanker = 1) 0.111 0.314 
Water well Dummy variable about the source of water supply (water well or waterwheel =1) 0.031 0.173 
River Dummy variable about the source of water supply (river = 1) 0.009 0.095 
Rain Dummy variable about the source of water supply (rain = 1) 0.008 0.096 
Transparency Dummy variable about the transparency of the water (transparent = 1) 0.759 0.428 
Taste Dummy variable about the taste of the water (good = 1) 0.691 0.462 
Smell Dummy variable about the smell of the water (smells = 1) 0.679 0.467 
Turbidity Dummy variable about the turbidity of the water (turbidity or solids in suspension = 1) 0.663 0.473 
Pipes inside Dummy variable about where the water pipes are installed (inside house = 1) 0.860 0.347 
External pipes Dummy variable about where the water pipes are installed (outside house = 1) 0.123 0.329 
Wquality Dummy variable about what characteristic of the water supply service – supply source, continuity of the service and water quality- the respondent would improve (water quality = 1) 0.636 0.482 
Wsource Dummy variable about what characteristic of the water supply service – supply source, continuity of the service and water quality- the respondent would improve (water source =1) 0.450 0.498 
Wcontinuity Dummy variable about what characteristic of the water supply service – supply source, continuity of the service and water quality- the respondent would improve (water continuity = 1) 0.547 0.498 
Carry water Dummy variable related to the fact that the respondent has to carry water home to cover the basic needs of drinking and hygiene (yes = 1) 0.089 0.286 
Trips Number of trips per day made to collect water outside the home 2.769 3.409 
Time Average time spent collecting water outside the home (minutes) 25.192 48.010 
Litres Litres of water collected outside the home per day 112.53 145.60 
Wexpense Water expenses (Bolivian pesos) 11.803 18.561 
Water cuts Dummy variable related to the quality of the water supply (if respondent suffers water cuts = 1) 0.518 0.500 
Satisfaction Level of satisfaction related to the water supply service (from 1 = very satisfied to 5= not satisfied at all) 3.250 0.948 
Bottled Dummy variable about the purchase of bottled water (yes = 1) 0.265 0.442 
Bottled expense Bottled water expenses (Bolivian pesos) 4.385 10.291 
Floor 4-category variable related to the quality of the house (floor= parquet flooring, tile flooring, concrete flooring, rough flooring) 2.713 0.701 
Roof 3-category variable related to the quality of the house (roof= tiled roof, concrete roof, straw roof) 1.437 0.509 
Wall 4-category variable related to the quality of the house (walls= concrete or stone, wood, adobe bricks, straw) 1.594 0.924 
Kitchen Dummy variable about where the kitchen is located (inside the house = 1) 0.891 0.310 
Bathroom 3-category variable about the type of bathroom (bathroom = complete bathroom, latrine, outdoor) 1.244 0.628 
Car owner Dummy variable about car ownership (yes = 1) 0.425 0.495 
Landowner Dummy variable about land ownership (yes = 1). The land is used for plant crops or for raising cattle. 0.175 0.381 
Abundance 5-category variable about the respondent's perception related to the family's abundance of resources (from 1 = very scarce to 5 = very abundant) 3.543 0.754 
Life satisfaction Level of satisfaction with life (from 1 = very satisfied to 5= not satisfied at all) 3.137 0.873 
District 6-category variable related to the district of the city in which the interview was conducted (from District 1 = 1 to District 6 = 1) 2.799 1.328 
Table 4.

Heckman's two-step model – participation equation.

 Probit model 
 Coefficients T-statistic 
Constant −0.1273 −0.42 
Income 0.3598*** 2.91 
Wcontinuity 1.2967*** 5.05 
District1 −0.6417** −2.04 
Landowner 0.6485* 1.74 
Bottled −0.4709* −1.87 
External pipes −0.7594** −2.35 
LR chi2(6) = 59.08 Prob > chi2 = 0.0000 Pseudo R2 = 0.2720 % Correctly predicted = 85.04 N = 324   
 Probit model 
 Coefficients T-statistic 
Constant −0.1273 −0.42 
Income 0.3598*** 2.91 
Wcontinuity 1.2967*** 5.05 
District1 −0.6417** −2.04 
Landowner 0.6485* 1.74 
Bottled −0.4709* −1.87 
External pipes −0.7594** −2.35 
LR chi2(6) = 59.08 Prob > chi2 = 0.0000 Pseudo R2 = 0.2720 % Correctly predicted = 85.04 N = 324   

***, **, *Indicate significance at 1%, 5%, and 10% levels respectively.

The probit coefficients show that, as expected, the probability of participating is positively related to the respondent's household income. In the same way, those individuals who, from the different characteristics of the current water supply service (source of supply, continuity of the service, and quality of water supplied), would give priority to improving the first one, are more willing to participate in the market. Owning a plot of land for vegetable crops or raising cattle also has a positive effect on the probability of participating in the market. This result suggests that these individuals consume a higher amount of water because, besides their own needs, they also have to use water for irrigating the plants and watering the cattle, hence they are more interested in improving the overall quality of the water supplied. On the other hand, individuals who buy bottled water to satisfy their needs are more prone to protest. This is an unexpected result since an improvement in the water supply service would imply not being forced to buy bottled water, with its corresponding saving of money. We think that maybe these individuals are showing some kind of reaction against having to pay for something that they believe they have the right to for free. Having the pipe system installed outside the house also has an effect on the probability of protesting. This latter result can be explained by the fact that usually the public company is responsible for extending the distribution network to reach the building, but the cost of taking the network to the household is borne by the families. Hence the extra cost that this improvement implies for the families is difficult to assume in the current circumstances. Finally, if the interview were conducted in District 1 of Sucre the probability of participating in the market would be lower than in the rest of the Districts since residents in this area have better access to public services whatever their nature. This result conforms to the previous findings shown in Table 1 where District 1 exhibited a mean WTP clearly lower than the full-sample value.

Table 5 reports the estimates pertaining to the valuation equation with the stated WTP as the dependent variable. As explained previously, an OLS model was estimated adding a new variable (λ) obtained from the probit model estimates at the first stage. If the coefficient of λ is significantly different from zero, there is evidence of sample selection bias. However, in this case there was no evidence of bias since the inverse Mills ratio (λ) was not significantly different from zero. The coefficients show that, as expected, stated WTP is positively related to the respondent's household income. The non-market valuation literature strongly suggests that income is positively related to environmental quality improvements (Hanley et al., 2009). Previous studies conducted by Virjee & Gaskin (2010) and Wang et al. (2010) are examples that ratify this relationship. Another variable that also shows a positive coefficient is the level of education. This is a common result in the CVM literature since usually the higher the education of the respondent, the higher their WTP (see, for example, Birol et al. (2006) and Jones et al. (2008)).

Table 5.

Heckman's two-step model – valuation equation.

 OLS OLS with selectivity 
 Coefficients T-statistic Coefficients T-statistic 
Constant − 7.7033 −1.87 −12.9037* −1.71 
Income 1.6758* 1.81 2.2194* 1.72 
Wcontinuity 10.6289*** 5.47 12.6749** 3.05 
Education 1.6406* 1.90 1.9880** 1.96 
Carrywater 11.4635*** 3.54 13.6484*** 3.62 
Adjusted R2 0.1950 – 
λ (Inverse Mills ratio)   8.3882 0.317 
σ   0.5310 
ρ   15.7944 
Wald chi2(4)   28.63 
Prob > chi2   0.0000 
N 324 324 
 OLS OLS with selectivity 
 Coefficients T-statistic Coefficients T-statistic 
Constant − 7.7033 −1.87 −12.9037* −1.71 
Income 1.6758* 1.81 2.2194* 1.72 
Wcontinuity 10.6289*** 5.47 12.6749** 3.05 
Education 1.6406* 1.90 1.9880** 1.96 
Carrywater 11.4635*** 3.54 13.6484*** 3.62 
Adjusted R2 0.1950 – 
λ (Inverse Mills ratio)   8.3882 0.317 
σ   0.5310 
ρ   15.7944 
Wald chi2(4)   28.63 
Prob > chi2   0.0000 
N 324 324 

***, **, *Indicate significance at 1%, 5%, and 10% levels respectively.

Wcontinuity is a dummy variable that takes the value ‘1’ if the respondent stated that, from the different characteristics of the current water supply service (source of supply, continuity of the service, and quality of water supplied), they would give priority to improving the second one. Water supply continuity is a crucial factor when valuing the quality of the service provided to the community (Um et al., 2002). Supply cuts in Sucre can be due to different causes and affect the entire city. The upper neighbourhoods suffer cuts when water storage tanks have insufficient capacity during periods of low rainfall, while lower neighbourhoods suffer strategic cuts in order to supply higher neighbourhoods at the times during the year when the city suffers the most shortages. Furthermore, the water supply is frequently interrupted due to breakages in the obsolete supply network. On the other hand, the positive coefficient, the variable Carrywater implies that, as expected, respondents who do not have direct access to water at home and are thus forced to carry water are more willing to pay to improve the service because they are aware of the trouble that it causes, as well as the implicit costs in terms of time.

Summing up, our results have been able to pass some minimal test of theoretical validity in explaining the determinants of the stated WTP, since the main variables were statistically significant and had the expected signs. In particular, WTP was shown to be positively related to respondents' household income, as expected. In the same way, respondents who had to carry water home were more willing to pay. It is also worth noting that respondents gave priority to the continuity of the service above an improvement in the perceived quality of the water and the way of accessing this resource. The low percentage of people willing to pay more, and the causes, allow important policy implications to be drawn, which are addressed in next section.

Conclusions

The results of our research suggest that in a climate of confrontation between citizens and water utility managers an increase in tariffs could be unsuccessful. A lack of governance and poor management are further contributing to worsen an already poor water service and create confrontation. Moreover, the recent demonstrations in Cochabamba and El Alto do not make raising the bill advisable. All this is reflected in the low percentage of households found in this research who would be willing to pay extra to improve the water service. This result contrasts with the current shortcomings of the service regarding key factors affecting its quality, which are the network coverage, continuity of the service, and quality of the water supplied. Nevertheless, the current shortcomings of the water service may well explain the low WTP; users may be dissatisfied with the water utility, which they hold responsible for the poor service provided and hence are not willing to pay extra in their water bill. Accordingly, if improvement measures were going to be implemented with the subsequent increase in water tariffs, the water utility should be very interested in first seeing the responses of the users. Therefore, a joint campaign between the water utility and the local administration, aimed at raising public awareness about the benefits of improving the water service, could be of great help in increasing the low WTP observed while paving the way for increasing water tariffs in the future. Nevertheless, it should also not be forgotten that our results show that the higher the respondents' income, the higher their WTP. In the case of Sucre, therefore, an increase in water tariffs would be understandable only if it were part of a broader reform aimed at helping people escape from poverty, thus enabling them to afford to pay higher prices for water without having to renounce other needs.

Acknowledgements

The authors gratefully acknowledge the highly constructive comments and suggestions from the associate editor and two anonymous reviewers. We are also grateful for the financial support from the Spanish Ministry of Economy and Competitiveness (Project ECO2012-32189) and the Government of Andalusia (Project P11-SEJ-7039).

References

References
Arrow
K.
Solow
R.
Portney
P. R.
Leamer
E. E.
Radner
R.
Schuman
H.
(
1993
).
Report of the NOAA Panel on Contingent Valuation, National Oceanic and Atmospheric Administration. Federal Register 58
, pp.
4601
4614
.
Bateman
I. J.
Carson
R. T.
Day
B.
Hanemann
M.
Hanley
N.
Hett
T.
Jones-Lee
M.
Loomes
G.
Mourato
S.
Ozdemiroglu
E.
Pearce
D. W.
Sugden
R.
Swanson
J.
(
2002
).
Economic Valuation with Stated Preference Techniques: A Manual
.
Edward Elgar
,
Cheltenham
.
Bengochea-Morancho
A.
Fuertes-Eugenio
A. M.
del Saz-Salazar
S.
(
2005
).
A comparison of empirical models used to infer the willingness to pay in contingent valuation
.
Empirical Economics
30
,
235
244
.
Calia
P.
Strazzera
E.
(
2001
).
A sample selection model for protest responses in contingent valuation analysis
.
Statistica
61
,
473
485
.
Carson
R. T.
(
1997
).
Contingent valuation: theoretical advances and empirical tests since the NOAA panel
.
American Journal of Agricultural Economics
79
,
1501
1507
.
Carson
R. T.
(
2000
).
Contingent valuation: a user's guide
.
Environmental Science & Technology
34
,
1413
1418
.
Casey
J. F.
Kahn
J. R.
Rivas
A.
(
2006
).
Willingness to pay for improved water service in Manaus, Amazonas, Brazil
.
Ecological Economics
58
,
365
372
.
Chilton
S. M.
Hutchinson
W. G.
(
1999
).
Do focus groups contribute anything to the contingent valuation process?
Journal of Economic Psychology
20
,
465
483
.
Daniel
J.
(
2012
).
Sampling Essentials: Practical Guidelines for Making Sampling Choices
.
SAGE Publications
,
Thousand Oaks, California
.
Dziegielewska
D. A.
Mendelsohn
R.
(
2007
).
Does no mean no? A protest methodology
.
Environmental and Resource Economics
38
,
71
87
.
Genius
M.
Hatzaki
E.
Kouromichelaki
E. M.
Kouvakis
G.
Nikiforaki
S.
Tsagarakis
K. P.
(
2008
).
Evaluating consumers' willingness to pay for improved potable water quality and quantity
.
Water Resources Management
22
,
1825
1834
.
Genius
M.
Manioudaki
M.
Mokas
E.
Pantagakis
E.
Tampakakis
D.
Tsagarakis
K. P.
(
2005
).
Estimation of willingness to pay for wastewater treatment
.
Water Science Technology: Water Supply
5
,
105
113
.
Guidi
E.
González-Gómez
F.
Guardiola
J.
(
2013
).
Water access in Sucre, Bolivia: a case of governance deficit
.
International Journal of Water Resources Development
29
,
636
649
.
Hanley
N.
Barbier
E. B.
(
2009
).
Pricing Nature: Cost-Benefit Analysis and Environmental Policy.
Edward Elgar
,
Cheltenham
.
Hanley
N.
Colombo
S.
Kriström
B.
Watson
F.
(
2009
).
Accounting for negative, zero and positive willingness to pay for landscape change in a national park
.
Journal of Agricultural Economics
60
,
1
16
.
Hausman
J.
(
2012
).
Contingent valuation: from dubious to hopeless
.
Journal of Economic Perspectives
26
,
43
56
.
Jones
N.
Sophoulis
C. M.
Malesios
C.
(
2008
).
Economic valuation of coastal water quality and protest responses: a case study in Mitilini, Greece
.
Journal of Socio-Economics
37
,
2478
2491
.
Jorgensen
B. S.
Syme
G. J.
Bishop
B. J.
Nancarrow
B. E.
(
1999
).
Protest responses in contingent valuation
.
Environmental and Resource Economics
14
,
131
150
.
Kontogianni
A.
Langford
I. H.
Papandreou
A.
Skourtos
M. S.
(
2003
).
Social preferences for improving water quality: an economic analysis of benefits from wastewater treatment
.
Water Resources Management
17
,
317
336
.
Lindsey
G.
(
1994
).
Market models, protest bids, and outliers in contingent valuation
.
Journal of Water Resources Planning and Management
120
,
121
129
.
Messonnier
M. L.
Bergstrom
J. C.
Cornwell
C. M.
Teasley
R. J.
Cordell
H. K.
(
2000
).
Survey response-related biases in contingent valuation: concepts, remedies and empirical application to valuing aquatic plant management
.
American Journal of Agricultural Economics
83
,
438
450
.
Mitchell
R. C.
Carson
R. T.
(
1989
).
Using Surveys to Value Public Goods: The Contingent Valuation Method
.
Resources for the Future
,
Washington, DC
.
OMS
(
2011
).
Agua y Saneamiento: Evidencias Para Políticas Públicas con Enfoque en Derechos Humanos y Resultados en Salud Pública
.
[Water and Sanitation: Evidence for public policies focused on human rights and public health results.] Organización Panamericana de la Salud
,
Washington, DC.
Polyzou
E.
Jones
N.
Evangelinos
K. I.
Halvadakis
C. P.
(
2011
).
Willingness to pay for drinking water quality improvement and the influence of social capital
.
Journal of Socio-Economics
40
,
74
80
.
Prüss-Üstün
A.
Bos
R.
Gore
F.
Bartram
J.
(
2008
).
Safer Water, Better Health
: Costs, Benefits and Sustainability of Interventions to Protect and Promote Health.
World Health Organization
,
Geneva
.
Raje
D. V.
Dhobe
P. S.
Deshpande
A. W.
(
2002
).
Consumer's willingness to pay more for municipal supplied water: a case study
.
Ecological Economics
42
,
391
400
.
Strazzera
E.
Genius
M.
Scarpa
R.
Hutchinson
G.
(
2003
).
The effect of protest votes on the estimates of willingness to pay for use values of recreational sites
.
Environmental and Resource Economics
25
,
461
476
.
UNICEF
(
2009
).
Día Mundial de Lavado de Manos. Bolivia, panorama nacional. Resumen informativo
.
Unidad de Análisis de Conflictos
(
2010
).
Informe de Seguimiento y Análisis de la Conflictividad en Bolivia, Septiembre de 2010
.
Fundación UNIR
,
Cochabamba, Bolivia
.
Whitehead
J. C.
(
2006
).
A practitioner's primer on contingent valuation
. In:
Contingent Valuation Handbook
.
Alberini
A.
Kahn
J.
(eds).
Edward Elgar
,
Cheltenham
.
WHO/UNICEF
(
2012
).
Progress on Drinking Water and Sanitation
.
Joint Monitoring Programme for Water Supply and Sanitation
,
UNICEF
,
New York and WHO, Geneva
.