Cape Verde is an African country where consumers pay one of the highest tariffs for water supply services in the world. However, the current levels of service coverage and quality are still far from adequate, as acknowledged by local authorities and international agencies. Thus, an assessment of how much Cape Verdean households are willing to pay (WTP) for service improvements is a fundamental ingredient informing sector policy formulation going forward. This is the focus of this study. Valuation functions were estimated using suitable estimation methods. The results show that, taking into consideration their current levels of expenditure with this service, households would in general not be WTP more for significant improvements in their current water supply services, and that household income and the age of the head of the household both have significant effects on their WTP.

INTRODUCTION

Water supply services are usually provided under territorially defined monopoly conditions, as a reflection of both the technological characteristics of the industry and the public service nature of their provision. When service position is universal and the levels of service provided to the population are deemed adequate, key public policy concerns focus on incentives for increased efficiency, infrastructural sustainability, avoiding monopoly rents via excessive tariffs and improved environmental performance (Kanakoudis & Tsitsifli 2014).

However, this is not the case in most of the developing world as well as in several countries within the Organisation for Economic Co-operation and Development (OECD). In these instances, reflecting the fact that these services are deemed fundamental ‘merit goods’ for human wellbeing, extending access, improving quality and ensuring their financial sustainability under affordability constraints come forward as dominant concerns, namely in setting suitable and affordable tariffs which are socially fair for their users (Gunatilake et al. 2006; Hadipuro et al. 2013).

These issues are especially acute in Cape Verde due to the critical situation of these essential services in this country, which faces major natural water resource shortages and, therefore, almost all water supplied for human consumption is being sourced via seawater desalination processes (Marques et al. 2013). Furthermore, energy costs (based on imported fuels) are very high, making this a burdensome service with one of the most expensive water tariffs across the African continent (see Figure 1) and even around the world (Garmendia & Benitez 2011).
Figure 1

Tariffs of water utilities in Africa (year 2011).

Figure 1

Tariffs of water utilities in Africa (year 2011).

One of the most frequently used methods in the literature for the purpose of assessing how much consumers would be willing to pay (WTP) for services or their improvement has been the contingent valuation (CV) method (Alberini & Cooper 2000). This method consists of questioning people directly through surveys about how much they would either be WTP, as a maximum, to be provided with a certain service, or, alternatively, to enjoy a definite improvement in the quality of that service (Littlefair 1998).

The present study follows the first approach and aims at estimating how much the households of three islands in Cape Verde (Santiago, Santo Antão and São Nicolau) would be WTP, in total, for an improved water supply service. This study also aims to evaluate the influence of certain household characteristics on consumer WTP, such as the socio-economic features of the household and several variables related to current water service usage.

The population of these three islands (out of nine inhabited islands in this archipelago) represents 68% of the total Cape Verdean population and includes the capital, Praia, which is located on the island of Santiago and where more than half of the country's population lives. The population of Cape Verde in 2013 was estimated at 530,000 inhabitants. That same year, the Gross Domestic Product (GDP) per capita was about 4,000 US$ and the last GINI (coefficient) index was 50.5 (in OECD countries, values tend to fall in the range of 25 to 35). The Gini index measures the extent to which the distribution of income among households or individuals within a society deviates from a completely equal distribution. The most important industry in the country is tourism and, as a country where the emigrant population is greater than the one resident in Cape Verde (mostly living in Portugal, the Netherlands and the US Boston area), their monetary remittances are an important resource for the country's external account balance. The water supply service is most expensive on Santiago (where water derives primarily from desalination), followed by Santo Antão (where water is abundant in two of the three municipalities, with the third one resorting to desalination) and finally by São Nicolau (where, despite being scarce, groundwater is still the main source).

The CV method comprises several ways to pose the question of how much people would be WTP for a good or service. In this study, the payment card format was used because of its advantages over other formats, in particular, over the dichotomous choice format, open-ended questions and bidding games (Bateman et al. 2002; Pearce & Ozdemiroglu 2002). The payment card format provides the respondents with a series of ordered values (or value intervals) and questions them which values, up to a maximum, they would be WTP for the improved service (Clark et al. 2002; Saz-Salazar et al. 2015).

In the literature, to estimate WTP based on information originating from CV questionnaires in which the payment card format is used, it is usual to calculate valuation functions by parametric approaches (Carson et al. 2003), called ‘interval data models’ (Cameron & Huppert 1989). These approaches are applied even in situations where not only value intervals of the WTP are obtained, but also point values. In other studies, it is frequent for the observations corresponding to these point values to be either removed from the sample or to be treated as censored answers (Hurlimann 2009). However, these procedures are not entirely suitable because, in the first case, statistical information is lost due to the small sample size and, in the second case, the available information is not fully exploited (Mahieu et al. 2012). For these reasons, the present research applied ‘the point and interval data model’. This approach enables us to include in the estimation of valuation functions both the ranges of values of the WTP stated by the respondents and the point values (for instance, null values corresponding to situations where respondents are not WTP for the service) and thus to include as much information as possible in the estimation of the valuation functions. In order to assess the robustness of WTP estimation results, a ‘multinomial logistic regression’ was also applied and the results obtained with both models were compared.

This paper makes several contributions to the literature due not only to the methods applied and the case study itself, but mainly due to the fact that water resources in Cape Verde are very scarce and drinking water is quite expensive and this fact is expected to lead to different reactions of the population. This research was carried out under the scope of a consultancy project developed for the Millennium Challenge Account of Cape Verde, the local agency of the Millennium Challenge Corporation in this country, which is currently supporting an extensive program to reform the water sector.

LITERATURE REVIEW

There are numerous studies in the literature that aim to assess the WTP for the provision of water services. However, most of them focus on the methods used and their potential improvement, being essentially concerned with issues such as reducing the bias, improving information collection procedures (from surveys) and statistical and econometric methods (Gunatilake & Tachiiri 2012).

This section will address mainly studies from the literature that have been conducted in developing countries. For example, the study by Whittington et al. (1991) in Onitsha (Nigeria) focused on the understanding of the water service market (which has some similarities with the Cape Verdean market, such as the structure and multiplicity of actors) and evaluated the WTP for this service. This study showed that households were paying informal water vendors twice more than the level of the operation and maintenance expenses of a piped network and equally highlighted some WTP for better service access and quality.

More recently, also in Nigeria, in the Osogbo metropolis, the WTP for an improvement in the water supply service was investigated by Adenike & Titus (2009). The study revealed that households are WTP for improved water supply. Age, education level and the income level of households showed a positive influence on WTP. Other studies conducted in Nigeria have reached similar conclusions, such as the study by Olanrewaju & Omonona (2012) in Kosofe, in Lagos State. The study by Kanayo et al. (2013) in Nsukka (in Enugu State, a region which suffers from water scarcity), also demonstrates that the price of alternative sources of access to water supply significantly influence WTP.

In Tanzania, Kaliba et al. (2003) also sought to estimate the WTP for improved water supply services (rural community-based) in 30 villages in the Dodoma and Singida regions, located in the central part of the country. This study revealed that households in these regions are also WTP more than they currently pay for a better service. However, the authors also documented that the age and the wealth level of the respondents exhibited a negative relationship with reported WTP. This apparently counter-intuitive finding is likely to reflect the fact that older people typically do not participate directly in water collection tasks and that the richest households may already have their own water sources or resort to others to provide them with water at lower costs. The study also confirmed that the size of the household and WTP have a positive relationship, indicating that the bigger the household size, the greater the needs are and therefore, the largest households are WTP more for an improvement in their water supply service.

In Botswana, research by Mazvimavi & Mmopelwa (2006) was conducted with the purpose of analyzing the level of access to water supply services, and discussing the problems with this service in villages located along the Boteti River in the North West district, where there was no access to piped water and other basic services. The study indicated that most households used untreated water from rivers and wells to meet their domestic needs and, moreover, they had to travel long distances to fetch water. This scenario led to low water consumption levels (less than 20 liters per inhabitant/day for most households), affecting the hygiene and endangering the health of the population. According to the study, for these households, the opportunity cost of time spent fetching water was higher than the price paid by households that live in rural areas where the population obtains water from piped systems. The study also concluded that most households were willing to contribute to improve access to water supply with a little bit over one third of the agricultural minimum wage.

Relevant studies have also been carried out in other African countries, such as Ghana (Nyarko et al. 2007), Ethiopia (Bogale & Urgessa 2012), Sierra Leone (Kargbo 2003), and Kenya (Whittington et al. 1989), among others. For example, the research of Nyarko et al. (2007) in Ghana aimed to assess the cost recovery levels of water supply services in order to estimate the affordability to domestic users of the real cost of such services and the available WTP to support their improvement. The study was conducted in five of the 30 communities in the Ashanti region and it showed that current tariffs are not sufficient to cover the operating expenses of providing water supply services and that, assuming 5% of household income as the limit of their expenditure on water supply services, only between 67% and 87% of users could pay the current costs for these services. The study also concluded that the majority of respondents are not WTP more than they were already paying. Also in Ethiopia, in the Haramaya district, another study was performed for the purpose of estimating the WTP of rural households for improved water supply services (Bogale & Urgessa 2012). The results showed that household income, education level, gender, time spent fetching water, water treatment type, water quality, water charges and the interviewee's age had significant effects on a household's WTP. Similar conclusions were drawn by Kargbo (2003), in Makeni, in Sierra Leone.

In the American continent, one can point out, for example, the study performed in El Salvador in 1998 with the objective of estimating the value assigned by four semi-rural communities to the improvement in the quality of water supply (Pineda & Armijo 2013). The results showed that access to piped water services was a priority for the population of these communities. The authors concluded that it could be profitable, from a private sector point of view, to implement water supply projects if at an appropriate scale. Thus, this study demonstrated that the design of financially viable projects with a positive impact on social conditions in poor communities was possible. In Mexico, a study conducted by Vásquez et al. (2009) also showed that the population was WTP more for water services if the supply conditions improved significantly.

The literature of the Asian continent also provides many studies on this theme. In India, WTP research was undertaken in order to determine the propensity of people to accept future increases in water supply service tariffs (Raje et al. 2002). The results showed a strong correlation between dwelling types and WTP. In poor neighborhoods, increases in the water bill were rejected, justified by the financial burden of generalized increases in other essential goods and services such as food, clothing and housing. In dwellings with better living conditions, the WTP for the water supply service was greater. Also in India, in Rajasthan state, Reddy (1999) sought to estimate both the willingness and the capability of households to pay for the water supply service in six localities with scarce water resources. The WTP analysis supported the finding that rural households were willing to spend five percent of their income on water supply services, provided they were delivered with better quality.

In Dhaka, Bangladesh, a study was performed by Nurul Islam et al. (1994) with the aim of analyzing the WTP of households for water supply services. The authors concluded that households with low-medium incomes faced a burden of about 10% of their income with water supply services, which exceeded by far the recommendations of international organizations, such as the OECD.

Another study, in China, performed by Wang et al. (2010) aimed to assess the levels of WTP and how they differ across the population, so that decision makers can adjust the tariffs of the water supply service to the specific socioeconomic realities. The result was a proposal for the subsidization of households with the lowest income, while making the whole system, which faced technical difficulties and required significant investment, financially viable.

From the analysis of the studies found in the literature regarding the WTP for water services, it may be inferred that, in most cases, households showed positive WTP and exhibited a perception of the social benefits of these services in terms of public health, community wellbeing and local development. This is particularly evident in those studies conducted in developing countries. However, the WTP for water supply services is closely linked to the particular characteristics of each region, such as socio-economic and environmental features, practical aspects of service provision and how the water industry is organized, among other factors. For this reason, different results were found in the literature. Nevertheless, in general, a significant influence of certain variables on the WTP has been identified, such as: income; household size; education level, age, gender and occupation of the head of the household; the volume of water consumed daily by the household; available alternative water sources (network, fountain, tank truck, etc.), their quality and cost of usage, including the distance to the main source of water and the time spent fetching water.

SURVEY

The present study focused on a subset of Cape Verdean islands with more than one municipality and where a process of implementing Multi-Municipal Utilities is ongoing, namely the Santiago, Santo Antão and São Nicolau islands. Interviews were thus performed in a representative sample of households in the city of Praia and in the Santa Cruz municipality (Santiago Island) and in all the municipalities of Santo Antão and São Nicolau islands.

The sampling procedure consisted of a stratified random sample. First, the population was divided into subgroups (strata) which were as homogeneous as possible according to the household comfort level. The household comfort level was obtained via the ‘Comfort Index’ (Instituto Nacional de Estatística, 2004), which had values from 0 to 100 and was divided into the following five categories: Very Low [0,20], Low ]20,40], Medium ]40,60], High ]60,80] and Very High ]80,100].) Subsequently, the estimated number of elements within each stratum was randomly selected. The distribution process of questionnaires to different strata was performed using the proportional allocation method (Särndal et al. 2003).

The minimum number of elements of the population (questionnaires) required for a representative sample was estimated at 460 elements (households), assuming a sampling error of 2% for the ‘Comfort Index’ for a 95% confidence level. However, a bigger sample of 710 households was surveyed in order to prevent possible problems with the questionnaires collected (e.g. absent data, major inconsistencies). For this group of three islands, this sample of 710 households ensures a maximum sampling error of 1.65% for the ‘Comfort Index’ indicator and a 95% confidence level.

The drafting process of the survey involved several pre-tests with stakeholder representatives (operators, MCA) and sample consumers in order to design, in as much detail as possible, a questionnaire appropriate to the Cape Verde situation. (The questionnaire is presented as supplementary material, available with the online version of this paper.) The level of awareness was also an issue taken into account, but water is a major concern of the Cape Verdean population due to its very high price. The content of the survey was structured into six groups of questions. The first group contained questions about the geographical location of the household, type of dwelling, ownership status and the type of water storage available at home. The second group of questions concerned the socio-economic characteristics of the household (size and composition of the household, expenditure and income levels, gender, age and education level of the head of the household). The third group focused on the available access modes to water supply (piped network to the home, community standpipes, wells, tank truck, etc.), the charges currently incurred by the household, the water consumption habits, the regularity and continuity of service, the perception of households about quality of service and on how much more they were WTP for an improved system of their main source of water supply, and why. The fourth group of questions sought information about the relationship between users and service providers, including the level and nature of complaints. In this set of questions, customers who did not have access to piped water at home were questioned about how they would spend their time if they did not have to fetch water. The fifth group of questions was dedicated to sanitation and access to wastewater collection services and was designed to extract information about the level of connection to sewerage networks, the existence of household sanitary facilities and also the WTP for this service. Finally, the last group of questions was intended to help understand the overall perceptions of the respondents about the access to water supply services (network or off-network).

The questionnaire was organized in order to allow the immediate detection of possible inconsistencies in the answers provided by respondents (by successive questions with the same aim) and, hence, to enable the inquirer to confront the respondents with these incongruences and solve them. It was mainly focused on the water supply service, although questions on the wastewater service were equally asked. However, the results obtained concerning the wastewater service will not be discussed in this paper.

Given the diversity in socio-economic contexts and in the relationship with water service operators, especially between the city of Praia and rural locations and the other islands, the questionnaire used different formats of questions (closed-ended questions, semi-open questions and open-ended questions). Specifically, as far as WTP was concerned, respondents were first asked if they were WTP for a better water supply service, and if the answer was ‘yes’, they were asked to choose one of the following five intervals in which the total amount they would be WTP would fit (1 euro = 110.256 CVE):

  1. Less than 1,000 CVE/month

  2. 1,001–1,500 CVE/month

  3. 1,501–2,000 CVE/month

  4. 2,001–3,000 CVE/month

  5. More than 3,000 CVE/month

The closed-ended questions were mostly ‘Yes/No’ choices. The semi-open question type was allowed for issues in which the pre-test revealed the prevalence of certain trends, and the open-ended question type was adopted for issues where the respondent should not feel embarrassed.

SURVEY RESULTS

Most users are aware and sensitive to the fact that the water supply service involves costs and, therefore, they generally recognize that the water they consume must be paid for. A high proportion of households highlight a WTP less than 1,000 CVE/month and, on average, for all the three islands, the estimated WTP is 890 CVE/month. For each island, individually, the WTP is 1,010 CVE/month on Santiago and 623 CVE/month and 650 CVE/month on Santo Antão and São Nicolau, respectively.

Globally, only 11% of households say that they are not WTP anything for the service. Higher WTP is also observed in places where there is no piped network service and lower WTP where water resources are more abundant.

The information collected by the survey also highlights the major positive and significant correlations (at a 5% level) between the WTP for an improved water supply service by Cape Verdean households and their income and total monthly expenditure (Figure 2).
Figure 2

Major correlation coefficients between the WTP and the remaining variables.

Figure 2

Major correlation coefficients between the WTP and the remaining variables.

It is also observed that the WTP has moderate positive and significant correlations (at a 5% level) with the type of dwellings, the existence of in-house tanks for water storage, the daily volume of water consumed by the household and the age of the head of the household.

Moreover, it is also observed that the WTP is negatively correlated with the variable ‘having a bathroom with toilet’. As expected this result means that households having a bathroom with toilet show higher WTP, as water availability is more important for them. The WTP is equally negatively correlated with the variable ‘going to the operator's premises (except for complaints)’. This means that households that have already been to the operator's premises (to address a specific subject but not to complain) tend to show higher WTP. This was also expected since they place a greater value on water issues.

Type of water storage available at home

From the first set of questions one observes that the most usual type of water storage at home on the islands of Santiago and São Nicolau is a barrel (typically a plastic container with a lid and a 200 liter capacity). On the island of Santo Antão, the tank is more common (a fixed cistern located at a higher point than the housing supply points). It is common practice for the households to use a barrel, not only as a way of storing water, but also as a means of measuring and controlling the quantity of water consumed. This practice prevails in situations where households have a tank and even when the households have a network connection and enjoy continuous supply. In this last instance, it is related to the fact that households do not want to take the risk of exceeding the upper limit of the tariff block they are willing to belong to.

Socio-economic characteristics of the household

From the second group of questions, it is noticed that on all three islands the typical profile of the head of the household is predominantly male, with an age between 35 and 45 years, educational grade four completed and working for an employer. Still, there are frequent cases where the head of the household is aged between 46 and 55 years, has no qualifications and is self-employed. On the island of Santo Antão, most heads of households do not have any academic qualifications.

As far as household size is concerned, on all three islands, households with five members predominate, closely followed by households with six, four and three members, and households with two members under the age of 18. On the islands of Santo Antão and São Nicolau, households tend to be smaller, with a predominance of households with only four members, but on Santo Antão the households tend to have more members under 18 years. Also noteworthy is the fact that about 6.5% of households have members that need special care.

Regarding the economic activities performed by heads of households, on Santiago, informal trade, jobs in the private sector, paid domestic work, blue collar work and agriculture are the major occupations. On Santo Antão, agriculture, small trade and the civil service are the main occupations and, on São Nicolau, it is common to find people with multiple jobs (typically agriculture, cattle breeding and fishing) and employed in industry.

On average, for the three islands, households display total monthly expenditure levels of 21.8 thousand CVE. The island which has the highest level is Santo Antão with 31.6 thousand CVE/month, followed by Santiago with an average of 20.4 thousand CVE/month and, finally, São Nicolau with an average of 18.1 thousand CVE/month. The largest portion of household expenditure is allocated to food, which, for all three islands, represents on average 62% of the total expenditure (see Figure 3). Expenditure on electricity is also an important part of the household budget representing an average of 11% of the total. Finally, expenditure on drinking water corresponds to an average of 7% of total expenditure on all three islands (1,537 CVE/month). Island by island, on Santiago the average is relatively higher (about 8%, i.e. 1,629 CVE/month) and visibly lower on the other two islands, with an average of 4% (1,211 CVE/month) and 5% (823 CVE/month) on Santo Antão and São Nicolau, respectively.
Figure 3

Average monthly expenses of households for food, water, electricity, rent, phone and other expenses.

Figure 3

Average monthly expenses of households for food, water, electricity, rent, phone and other expenses.

On household income, the majority of respondents stated monthly incomes between 11 and 30 thousand CVE on the three islands as a whole. Individually, on Santiago, most households report incomes between 11 and 20 thousand CVE/month, although incomes below 10 thousand CVE/month and between 21 and 50 thousand CVE/month are also quite common. On the other hand, on Santo Antão and São Nicolau most households report incomes between 21 and 30 thousand CVE/month. It should be noted that addressing the issue of household income levels was very delicate and problematic, since it was observed that the income declared by households often contrasted with exterior signs of wealth. This is because households tend not to reveal the aid they receive from emigrant relatives, which has a very significant contribution to disposable income. To minimize bias, this question was asked immediately after the one about the level of expenditure, which made it possible to confront respondents immediately in situations where they claimed to incur a level of expenditure materially higher than reported income. It should also be emphasized that many households, especially the ones where the income is derived from agricultural activities, reported problems of access to regular financial resources in order to cope with periodic expenses, such as drinking water charges. Despite having access to food products, these households do not have the same ease and frequency in accessing money as an employee with a regular salary.

Available modes of access to water supply and usage patterns

Through the third group of questions, it was observed that most households are served by a water distribution network (58%), followed by community standpipes and tank trucks. In rural areas, resorting to small ‘levadas’ (natural water-courses/canals), wells and boreholes is also very frequent.

Across the islands, on Santiago there is a smaller proportion of households with piped water (51%) while on the other two islands these shares are much higher (87% and 88% on Santo Antão and São Nicolau, respectively). There is a reasonable number of cases where households are connected to the water network but do not use the service. This is because either the supply has been cut off for nonpayment, or they believe that the water from the network is not safe and has a poorer quality than that of other sources, or because of the level of water supply tariffs.

Concerning water consumption, the usual daily level of household consumption ranges between 31 and 50 liters for all the three islands and levels between 51 and 100 liters/day are also quite common.

Relationship between users and service providers

The fourth set of questions revealed that most households had never made complaints. Those who had, only made them orally and, in these cases, only very rarely was the problem solved. Most respondents express concerns about the price of the water supply service, its continuity and the quality of the water supplied.

Sanitation, access to wastewater collection services and WTP for this service

Most households have a bathroom but most of them do not have a connection to a sewer. On Santiago, respondents report a positive WTP for a wastewater collection service justifying this with concerns about hygiene and health. On São Nicolau, although respondents also expressed WTP for the wastewater service, they declared lower WTP values, which may be because on this island there are public bathrooms and water-closets where wastewater discharges are made. In contrast, on the island of Santo Antão, households state that they are satisfied with the septic tanks they already have and, therefore, are not WTP for the access to a public wastewater system.

Overall perceptions on the access to water supply services

Finally, the sixth group of questions shows that the preferred means of access to water supply is the public network because it is the most convenient (it does not require physical effort or loss of time fetching water). The main concerns raised by households are the physical access to the service, the level of water supply tariffs, which they consider excessive, the continuity of the service and the quality of the water supplied. Respondents also support reducing the price of water, facilitating payment by installments, or even providing free delivery for people with greater economic difficulties.

ECONOMETRIC ANALYSIS

In this study the CV method was applied. Households were first asked if they were WTP for a better water supply service (more continuity, improved quality, …), and if ‘yes’, which of five value intervals represented the maximum total amount they would be WTP. This resulted in a WTP variable extracted from questionnaires with six categories: WTP = 0 category (for cases in which the households are not WTP) and the five value categories presented above.

In order to estimate the WTP for the water supply service as well as to evaluate the influence of various variables on the WTP, valuation functions were estimated. As discussed in the introduction, in this study, ‘the point and interval data model’ is applied, which considers in the estimation of valuation functions not only the various interval of values of WTP, but also point values and the observations corresponding to situations where respondents are not WTP (point value when WTP = 0).

To validate the results obtained by this method, a multinomial logistic regression was also applied and the results obtained by both models were compared. The first model (‘the point and interval data model’) is represented by the following expression: 
formula
1
where WTP is the willingness to pay for the water supply service declared by households, xi is the vector of explanatory variables, βi is the vector of parameters to be estimated and ɛ is the error assumed to be independent and normally distributed with a mean of zero and a standard deviation of σ.
The second model (multinomial logistic regression) is represented by the equation: 
formula
2
where is the probability of a household being WTP for the water supply service a value within the interval of class i, relative to the probability of being WTP an amount in the interval of class 0 (in this case it was considered the class corresponding to respondents which ‘are not willing to pay’, or, WTP = 0), and xi, βi and ɛ have the same meaning as in the previous model.

The variables selected for both models were the variables that might have an influence on the WTP, namely those usually considered in the literature, such as personal attributes of the head of the household (gender, age, education level and employment situation), size and income level of the household, habits of water consumption and also information about the type of dwelling, its location (island) and the main source of household water supply. Table A1 (in the supplementary material, available with the online version of this paper) lists the (explanatory) variables included in the models.

In the estimation of the first model it was assumed that the dependent variable (WTP) follows a logistic probability distribution, since it was the model that led to the highest log-likelihood. Both models were estimated using the R software, the first from the ‘survreg’ command and the second from the ‘multinom’ command.

In Tables A2 and A3 (in the supplementary material, available with the online version of this paper) the estimated coefficients and the standard errors associated with each one of the explanatory variables are presented for the first and second models, respectively.

Both models led to similar conclusions. The second model allowed further demonstration of the existence of two distinct realities in the sample: one of Santiago and the other of the remaining two islands (São Nicolau and Santo Antão). According to the results obtained by this model, on these last two islands there was a high probability (greater than 0.8 in most cases) of observing a WTP smaller than 1,000 CVE/month, while on Santiago it was also quite likely for the WTP to occur between 1,001 and 1,501 CVE/month (Figures 47). This was expected since water resources are scarcer in Santiago.
Figure 4

Probabilities associated with each WTP category for various accommodation types (left) and for different age groups (right).

Figure 4

Probabilities associated with each WTP category for various accommodation types (left) and for different age groups (right).

Figure 5

Probabilities associated with each WTP category for different education levels of the head of the household (left) and for various household sizes (right).

Figure 5

Probabilities associated with each WTP category for different education levels of the head of the household (left) and for various household sizes (right).

Figure 6

Probabilities associated with each WTP category for different monthly household income classes (left) and for various types of water supply sources (right).

Figure 6

Probabilities associated with each WTP category for different monthly household income classes (left) and for various types of water supply sources (right).

Figure 7

Probabilities associated with each WTP category for various classes of daily water consumption of a household.

Figure 7

Probabilities associated with each WTP category for various classes of daily water consumption of a household.

According to the results of the first model, the island of Santiago (base category of the variable ‘Island’) was the one that presented the highest WTP, while Santo Antão was the one that exhibited the lowest WTP. On average, Santo Antão revealed a lower WTP than the island of Santiago by about 415 CVE/month (São Nicolau showed a lower WTP by about 324 CVE/month) (Table A2).

Although not proven to be statistically significant in the first model, the second model showed that there was a higher probability of households who lived in informal settlements to express a higher WTP (Figure 4) and the opposite for households who lived in apartments.

As for the ‘gender of the head of the household’, although the results were not statistically significant, they seemed to show that men tended to show a lower WTP for the water supply service (Table A2). This result is understandable, since women are the ones who usually fetch water when the household does not have a connection to the water network.

The age of the head of the household also has a significant effect on the WTP. There is a tendency for the WTP to increase with the age of the head of the household (Table A2). In instances in which the head of the household is more than 67 years old, respondents revealed a WTP greater by 532 CVE/month when compared with households where the head of the household was under 25 years old (Table A2).

As far as the level of education of the head of the household is concerned, the results do not show a clear relationship with WTP (Table A2 and Figure 5). Still, although the results are not statistically significant, it is noted that in overall terms, households whose head has a higher degree of education are those who seem to display the highest WTP.

As for the size of the household, it was not proved that this variable had a positive effect on the WTP. In the first model this effect, although positive, was not statistically significant (Table A2) and in the second model there was only a slight tendency to increase the probability associated with a greater WTP (greater than 1,000 CVE/month) as the number of people in the household increased (Figure 5).

A variable that turned out to have a significant effect on the WTP in Cape Verde was the level of monthly household income. According to the results, there was a clear trend towards an increase in the WTP as the household income increased (Table A2 and Figure 6).

Regarding the main source of household water supply, it appears that households using tank trucks are those which report greater WTP (Table A2). Compared to households with a connection to the water network, the households using tank trucks (the most expensive source of water) showed a higher WTP of about 360 CVE/month. On the other hand, the households with a lower WTP are those which obtain water from other sources such as wells, boreholes or water courses (Figure 6). As for households which are supplied by standpipes, when compared with tank trucks, they also show a lower WTP but still slightly higher than those supplied by other sources.

Finally, in examining the influence of the daily water consumption of a household, the results show a certain tendency for an increase in the WTP as the daily consumption also increases (Table A2 and Figure 7). For example, households with a daily consumption level between 150 and 200 liters/day reveal a higher WTP than households with a consumption of less than 25 liters/day, estimated at 326 CVE/month (Table A2).

CONCLUSIONS

This study aimed to assess the WTP for an improved water supply service on three Cape Verdean islands, on which the majority of the country's population lives. For this purpose, valuation functions were estimated using methods deemed more appropriate than those commonly applied in the literature. One of the methods, ‘the point and interval data model’, is considered more appropriate because it allows the integration of intervals of values of WTP with point values of WTP in the estimation of valuation functions. In addition to this method, a multinomial logistic regression was also applied. The results of both models led to similar conclusions. However, we drew the main conclusions from the second model because it allowed us to distinguish between two realities in the sample: one on the island of Santiago and the other on the other two islands (São Nicolau and Santo Antão).

According to the results, the islands of São Nicolau and Santo Antão exhibited a lower WTP than the island of Santiago, since water resources in the latter case are scarcer and the availability of alternative water sources is lower. On average, all three islands showed a WTP of 890 CVE/month. As for each island, individually, the WTP on Santiago is 1,010 CVE/month and 623 and 650 CVE/month for Santo Antão and São Nicolau, respectively.

The results also showed that household income, the age and level of education of the head of the household, and the level of daily water consumption have a positive and significant effect on the WTP.

Furthermore, the results equally show that the households living in informal settlements, those whose head is female, those with a higher number of people, and those using tank trucks as their main source of water supply were the ones which showed a greater WTP. Conversely, households living in apartments, whose head was male and low-skilled, the ones with a smaller number of people and those which obtained water from other sources (such as wells, boreholes or water courses) and standpipes were the ones that demonstrated lower WTP.

The conclusions of this research might be very useful for the Cape Verdean policy makers and stakeholders, particularly in establishing the priority of investments, the design and provision of subsidies and the reform and restructuring of the water tariff schemes applied by the water utilities. Most importantly, the overall results showed that levels of willingness to pay (even for an improved water supply service) fall short of the reported current expenditure levels of water services. Although under-reporting is an acknowledged widespread phenomenon in these instances. Although under-reporting is an acknowledged widespread phenomenon in these instances, it seems fairly obvious that the willingness of households in Cape Verde to finance water service improvements, via tariff increases, is limited.

ACKNOWLEDGEMENTS

This research was carried out under the scope of a consultancy project developed for the Millennium Challenge Account (MCA) of Cape Verde, the local agency of the Millennium Challenge Corporation in this country, which is currently supporting an extensive program to reform the water sector. We thank MCA for taking this program and giving us the opportunity to develop this study.

REFERENCES

REFERENCES
Adenike
A.
Titus
O.
2009
Determinants of willingness to pay for improved water supply in Osogbo metropolis, Osun State, Nigeria
.
Research Journal of Social Sciences
4
,
1
6
.
Alberini
A.
Cooper
J.
2000
Applications of the Contingent Valuation Method in Developing Countries: a Survey
.
FAO Economic and Social Development Paper 146, Food and Agriculture Organisation
,
Rome
.
Bateman
I. J.
Carson
R. T.
Day
B.
Hanemann
M.
Hanley
N.
Hett
T.
Jones-Lee
M.
Loomes
G.
Mourato
S.
Ozedemiroglu
E.
Pearce
D.
Sugden
J.
Swanson
J.
2002
Economic Valuation with Stated Preference Techniques: A Manual
.
Edward Elgar
,
Cheltenham
,
UK
.
Bogale
A.
Urgessa
B.
2012
Households’ willingness to pay for improved rural water service provision: application of contingent valuation method in Eastern Ethiopia
.
Journal of Human Ecology
38
,
145
154
.
Cameron
T. A.
Huppert
D. D.
1989
OLS Versus ML estimation of non-market resource values with payment card interval data
.
Journal of Environmental Economics and Management
17
,
230
246
.
Carson
R. T.
Mitchell
R. C.
Hanemann
M.
Kopp
R. J.
Presser
S.
Ruud
P. A.
2003
Contingent valuation and lost passive use: damages from the Exxon Valdez oil spill
.
Environmental & Resource Economics
25
,
257
286
.
Clark
D. E.
Novotny
V.
Griffin
R.
Booth
D.
Bartošová
A.
Daun
M. C.
Hutchinson
M.
2002
Willingness to pay for flood and ecological risk reduction in an urban watershed
.
Water Science & Technology
45
,
235
242
.
Garmendia
C.
Benitez
D.
2011
Cape Verde's Infrastructure: A Continental Perspective
.
Africa Infrastructure Country Diagnostic, Country Report, World Bank Policy Research Working Paper No. 5687, IBRD, Washington DC, USA
.
Gunatilake
H.
Tachiiri
M.
2012
Willingness to Pay and Inclusive Tariff Designs for Improved Water Supply Services in Khulna, Bangladesh
.
Working Paper No. 9
,
Asian Development Bank
,
Manila, The Philippines
.
Gunatilake
H.
Yang
J. C.
Pattanayak
S. K.
Van den Berg
C.
2006
Willingness to Pay Studies for Designing Water Supply and Sanitation Projects: a Good Practice Case Study
.
Economics and Research Department ERD, Technical Note No. 17.
,
Asian Development Bank
,
Manila, The Philippines
.
Hadipuro
W.
Wiering
M.
van Naerssen
T.
2013
The sustainability of urban water supply in low income countries: a livelihoods model
.
Journal of Water, Sanitation and Hygiene for Development
3
,
156
164
.
Instituto Nacional de Estatística
,
2004
IDRF 2001/2002 - Condições de vida dos agregados familiares
.
Cape Verde
.
Kaliba
A. R. M.
Normanb
D. W.
Chang
Y.
2003
Willingness to pay to improve domestic water supply in rural areas of Central Tanzania: implications for policy
.
International Journal of Sustainable Development & World Ecology
10
,
119
132
.
Kanakoudis
V.
Tsitsifli
S.
2014
Doing the urban water supply job: from privatization to remunicipalisation and the third pillar of the performance based service contracts
.
Water Utility
6
,
31
46
.
Kargbo
I. B.
2003
Willingness to Pay for Improved Water Services in Sierra Leone: Evidence from Makeni
.
MSc Thesis
,
School of Graduate Studies of Addis Ababa University
,
Addis Ababa, Ethiopia
.
Littlefair
K.
1998
Willingness to pay for water at the household level: individual financial responsibility for water consumption
.
MEWEREW Occasional Paper No. 26
,
Water Issues Study Group, School of Oriental and African Studies SOAS), University of London, UK
.
Mazvimavi
D.
Mmopelwa
G.
2006
Access to water in gazetted and ungazetted rural settlements in Ngamiland, Botswana
.
Physics and Chemistry of the Earth
31
,
713
722
.
Nurul Islam
A. K. M.
Kitawaki
H.
Rahmanm
H.
1994
Willingness to pay/use for Water Supply and Sanitation
. In:
Affordable Water Supply and Sanitation
,
20th WEDC Conference, Colombo
,
Sri Lanka
.
Nyarko
K. B.
Oduro-Kwarteng
S.
Adama
I.
2007
Cost recovery of community-managed piped water systems in Ashanti region, Ghana
.
Water and Environment Journal
21
,
92
99
.
Olanrewaju
E. C.
Omonona
B. T.
2012
Determinants of households’ willingness to pay for improved water supply service in Kosofe local government area of Lagos State
.
Continental Journal of Sustainable Development
3
,
20
31
.
Pearce
D.
Ozdemiroglu
E.
2002
Economic Valuation with Stated Preference Techniques
.
Department for Transport, Local Government and the Regions
,
Rotterdam, The Netherlands
.
Raje
D. V.
Dhobe
P.
Deshpande
A.
2002
Consumer's willingness to pay more for municipal supplied water: a case study
.
Ecological Economics
42
,
391
400
.
Särndal
C. E.
Swensson
B.
Wretman
J.
2003
Model Assisted Survey Sampling
.
Springer
,
New York
.
Saz-Salazar
S.
González-Gómez
F.
Guardiola
J.
2015
Willingness to pay to improve urban water supply: the case of Sucre, Bolivia
.
Water Policy
17
,
112
125
.
Vásquez
W. P.
Mozumder
P.
Hernández-Arce
J.
Berrens
R. P.
2009
Willingness to pay for safe drinking water: evidence from Parral, Mexico
.
Journal of Environmental Management
90
,
3391
3400
.
Whittington
D.
Lauria
D. T.
Okun
D. A.
1989
Water vending activities in developing countries
.
International Journal of Water Resources Development
5
,
158
168
.
Whittington
D.
Lauria
D. T.
Mu
X.
1991
A study of water vending and willingness to pay for water in Onitsha, Nigeria
.
World Development
19
,
179
198
.

Supplementary data