Estimating household willingness to pay for water services in a rural economy: The case of Kanye in southern Botswana

Abstract

This study uses the contingent valuation approach to examine the relationship between selected socio-economic characteristics of households and their willingness to pay for private water connection. An iterative bidding method was used to obtain survey data from 135 households randomly selected from Kanye in southern Botswana and the ordinary least-square regression technique was used to estimate the effects of selected exogenous variables on willingness to pay. The results show that household income, level of education and employment status of the head of the household and level of consumers' awareness are the principal factors influencing willingness. This suggests that any government policy that increases these factors will increase households' willingness. In addition, because consumers' income is associated with their willingness to pay, it is suggested that policies on connection and user fees recognise income inequalities when determining the amount that households should pay.

1. Introduction

Botswana is a landlocked country with an estimated population of 1,68 million. Before independence, beef production was the mainstay of the economy in terms of output and export earnings. However, since independence in 1966 the country has witnessed a remarkable economic transformation. There has been increased access to education, roads, health, telecommunications and electricity, and of course water (Government of Botswana, 1999).

The first comprehensive and meaningful attempt by the Government of Botswana to provide clean and easily accessible water to the majority of the households was through the Village Water Supply Programme (VWSP) launched in 1972. This is because rural water projects in the past had been largely associated with the presence of piped water supplies to the villages and this explains the various amendments to the laws and acts governing rural water supply in Botswana, which date as far back as 1968. These amendments include the Borehole Act 1956 Cap 34:02, the Waterworks Act 1962 Cap 34:03, the Water Act 1968 Cap 34:01, the Water Utilities Corporation Act 1970 Cap 74:02, the Water Utilities Corporation (WUC) Amendment Act 1978 Cap3: 78, and the Waterworks Amendment Act 1983 Cap16: 83.

Today the responsibility for water supply in Botswana is shared among a number of ministries, departments and other bodies. This reflects the way in which the country has developed over the years since gaining independence, with the institutions and laws that governed the country being modified to adapt to the nation's changing needs and circumstances. The institutional framework for rural water supply in Botswana involves the Ministry of Mineral Resources and Water affairs, which is responsible for policy formulation, planning, development and management of water and mineral resources (Government of Botswana, 1994).

Unfortunately, however, water has become one of the largest and certainly the most universal of all problems facing mankind in the 21st century. The tasks of supplying enough water of the required quality to growing populations and the safe disposal of wastewater are straining many authorities to the limit (Winpenny, 1994). Although the problem comes in different forms to different nations, it is a big challenge to all governments of all countries at all stages of development. The essence of the problem is the failure to treat water as a scarce resource. This is because the need to conserve it and allocate it to socially more valuable uses has not always been evident (Smith, 1996; Reddy, 1999). In some societies, water has long been treated as a scarce and valuable resource but in the majority, however, it has been treated as if it were available in unlimited quantities and supplied at zero or low or highly subsidised cost to consumers. For most households, access to a high-quality, reliable source of water is one of the most important components of the bundle of goods which constitute an improved standard of living. However, progress in improving the quality and quantity of water used by people in rural areas of the developing world has been very unsatisfactory (World Bank, 1993). One of the reasons for this, most researchers argue, is that the pipe-water supplies which have been constructed are frequently not used appropriately or properly maintained. In addition, the extension of services to populations with no water supply has been very slow.

1.1 Problem statement and justification for the study

Botswana has very limited resources of surface water. The majority of the population is dependent on groundwater for their supply of drinking water. To give the rural population access to a clean and safe supply is therefore important for developing the rural areas. However, the distribution of water in Botswana is uneven. Most originates from rainfall, which is erratic and unreliable (Gould, 1995). The long-term mean annual rainfall varies from a minimum of 250 mm in the southwest to a maximum of 650 mm in the extreme north (Sharma, 1996). Except for the Okavango Delta, river flows occur only in the eastern part of the country and they are all seasonal. As if this were not enough Kalahari sands cover most of the country, with the exception of the areas of surface streams and the delta area. In rural areas, where the majority lives, water facilities are inadequate compared to urban areas. In terms of service coverage and distribution, Table 1 shows that although Botswana has achieved a relatively healthy water coverage in the urban areas, the majority of the rural areas still rely on rain water and public standpoints for their daily water needs. The major source of water supply in the rural areas is from groundwater (Snowy Mountains Engineering Corporation, 1991).

Table 1: Coverage of water service in urban and rural areas of Botswana (population in thousands)

Water supply	Population	Population with private connection	Population with public standpoint	Population served	Population not served
Urban water	815	494	111	815	0
Rural water	807	61	551	N/A	N/A
Total population served with water	1 622	555	662	N/A	N/A
Source: Stone and Webster Consultants, Appropriate Regulatory Reforms for Infrastructure and Utility Services in Botswana, report submitted to World Bank and Public Enterprises Evaluation and Privatization Agency (PEEPA), January 2005.

The Village Water Supply Programme (VWSP) was launched with the main aim of providing clean and easily accessible water to the rural communities of Botswana. In the approximately 30 years of the existence of the VWSP, more than 500 000 people (50 per cent of them living in small villages) have been supplied with drinking water within 400 m of their houses (Government of Botswana, 1994). However, a nearby guaranteed supply of this essential resource is needed daily throughout the year, because water is heavy to carry for any reasonable distance, as well as the common problem of low pressure in the public supply system. A standpipe means queuing for water: for villagers the wait can be extremely long and sometimes women spend many hours fetching water elsewhere. A more reliable and zero distance water supply is much more productive.

There are both social and private benefits associated with private residential water connection in Botswana. From the social analysis viewpoint, the operational performance of the WUC, the main body responsible for water administration in Botswana, shows that private water connection is profitable, as the marginal benefits from water connections are higher than the opportunity cost of funds used in providing the service. Table 2 shows the financial analysis of the operations of the WUC and indicates a capital recovery ratio of more than 100 per cent, an indication that tariffs are sufficient to cover the utility's operating and capital costs. This implies that the WUC is profitable and that all money invested in the corporation is being recovered (Stone & Webster Consultants, 2005) and thus the provision of water services in Botswana is both economically profitable and socially desirable. Table 3 shows the current tariff schedule for WUC domestic and business customers for the 2003/2004 financial year.

Table 2: Financial operation performance indicators of Botswana Water Utilities Corporation

Indicator	1999	2000	2001	2002	2003	2004
Average consumption (m^3/customer/month)		61	58	54	58	55
Average tariffs (pula/billed m^3)	4,74	5,31	6,22	7,62	9,34	10,75
Operating cost recovery ratio	225%	266%	242%	261%	307%	365%
Capital cost recovery ratio				113%	134%	136%
Total debt/equity ratio				0,74	0,63	0,53
Cash flow from operations (P'000)				80 457	157 586	239 054
Source: Stone and Webster Consultants, Appropriate Regulatory Reforms for Infrastructure and Utility Services in Botswana, report submitted to World Bank and PEEPA, January 2005.

Table 3: Tariff schedule for WUC – domestic and business customers 2003/2004

Tariff band	Consumption per month (m^3)	Gaborone/Lobaste (thebe/m^3)	Jwannag (thebe/m^3)*	Francistown (thebe/m^3)	Selebi/Phikwe (thebe/m^3)
1	0–10	210 (190)	165 (150)	240 (190)	165 (150)
2	11–15	640 (580)	330 (300)	575 (460)	265 (240)
3	16–25	815 (740)	430 (390)	840 (670)	330 (300)
4	Above 25	1 130 (1 025)	495 (450)	940 (750)	430 (390)
Rural water (untreated)		330 (300)	N/A	200 (160)	235 (215)
Source: Stone and Webster Consultants (2005) Appropriate Regulatory Reforms for Infrastructure and Utility Services in Botswana, report submitted to World Bank and Public Enterprises Evaluation and Privatization Agency (PEEPA), January 2005.
Notes: Figures in parenthesis are the tariffs for 2002/2003 financial year. *Note that 100 thebe equal one pula, the Botswana unit of currency.

However, despite enormous improvements in the provision of water in rural villages, periodic water shortages remain a problem in the rural areas of Botswana because of drought, occasional breakdowns and the rapidly growing demand, which is due partly to high population growth rates (Snowy Mountains Engineering Corporation, 1991). Because of the vagaries of weather that sometimes result in erratic rainfall and frequent drought in Botswana, the traditional rainwater catchment system cannot be used as a substitute for reliable and safe water supply that is zero distance from the homesteads (Gould, 1995). This has compounded the problem of access to water for the majority of rural households.

There is an established relationship between water-borne diseases and unsafe sources of domestic water supply. Water-borne diseases are those caused by water that has been contaminated by human, animal or chemical wastes or related in some way to contaminated water in the environment or to impurities within the water (Freeman, 1979). If water is not safe for drinking or is insufficient for personal hygiene or sewage disposal, water-borne diseases such as cholera, typhoid and dysentery, to mention but a few, will spread more easily (Bourne, 1984). Other intestinal infections, which not only cause much illness among adults but are often fatal to infants or undernourished children, can also be spread in this way (Falkenmark & Widstrandt, 1992). Table 4 shows some cases of water-related diseases over the years in Botswana.

Table 4: Water-related disease cases in Botswana: 1990–1997

	1990		1991		1992		1993		1994		1995		1996
	C	D	C	D	C	D	C	D	C	D	C	D	C	D
Malaria	2 027	1	1 942	2	502	2	15 462	54	5 164	8	2 338	16	25 497	124
Diarrhoea	53 420	48	58 330	36	45 619	52	61 591	64	66 950	51	93 178	35	104 664	43
Typhoid													1	1
Source: Ministry of Health (1990–2000).
Notes: C = cases; D = deaths. Only confirmed cases are tabulated.

1.2 Research questions and objectives

This study therefore presents an empirical study of people's willingness to pay (WTP) for private water connection in the rural areas of Botswana, with Kanye in the Southern District as a case study. The main objective of the study is to link some selected socio-economic characteristics of households with their willingness to pay for private water connection. The paper asks three important questions:

•	Is there any quantifiable relationship between some selected socio-economic characteristics of households and their willingness to pay for private water connection in Kanye Village?
•	Is it possible to isolate these socio-economic characteristics of households that influence their decisions to make financial contribution to potable water supply projects?
•	What recommendations, if any, can be deduced from the study, as a source of effective policy that can increase the demand for private water in Kanye District?

To address these research questions effectively, the contingent valuation approach is adopted in this study. This approach, which seeks to identify the true demand for affordable and socially acceptable services, allows individuals to take account of all factors (disposable income, education, etc.) that are important to them in obtaining or preventing a hypothetical specified change in their present situation (Amin & Khondoker, 2004). The tool helps to establish the basis for the implementation of a self-sustaining development based on the demand-driven approach.

According to Whittington et al. (1992), if rural water projects are to be both sustainable and replicable an improved planning methodology is required which includes a procedure for eliciting information about the value placed on different levels of service. In addition, the tariffs must be designed so that at least operation and maintenance costs (and preferably capital costs) can be recovered (Rodriguez, 2003). One concept in such an improved planning methodology is WTP. If people are willing to pay the full costs of a particular service, then it is a clear indication that the service is valued (and therefore will most probably be used and maintained) and that it will be possible to generate the funds required to sustain and even replicate the project. According to Carson & Mitchell (1993), however, most attempts to incorporate WTP considerations into project design have at best been ad hoc. This is because of the absence of validated field-tested methodologies for assessing WTP for water or any housing or service good in the context of rural communities in developing countries (Ready et al., 2001).

It is hoped that the findings from this study will provide some basic policy information that could guide the Government of Botswana and other developmental agencies in improving water supply to the majority of rural households in Kanye in particular and in other rural areas of Botswana.

2. Study area

Kanye is located in the Southern District of Botswana. With an estimated population of 39 891, it is the district headquarters of Southern District and serves as one of the main economic nerve centres of the country because of its contribution to the agricultural sector of the economy. In terms of water supply, Kanye is one of the many villages serviced by Botswana's Department of Water Affairs. It is one of the largest villages in Botswana and the second largest client of the Department of Water Affairs. Its size makes it very strategic to the success and expansion of the rural water development (Department of Water Affairs Annual Reports, 1996–2000).

Kanye Water Affairs is charged with the responsibility for supplying potable or clean water to the whole population of Kanye and some other towns. The town is supplied with two boreholes, Ramonnedi and Kgwakgwe, which are located about 40 km and 25 km, respectively, from the Kanye Water Affairs office. The total production from the two well-fields is 5 100 kl per day (Department of Water Affairs, Annual Reports, 1996–2000). The daily demand for water by Kanye residents is about 3 000 kl per day, although the overall demand is 4 300 kl per day, because the well-fields supply other villages.

The water quality from the Kanye Water Affairs is within World Health Organization (WHO) standards, although adjusted accordingly to meet Botswana's standards. The water is very hard, with an acceptable concentration of salts. Kanye, as expected, falls within what is officially termed ‘the gazetted water works area’. By government regulations, only consumers within the waterworks area have water supplied at government subsidised rates. The implication is that residents outside Kanye, for example, will have to pay the full charges for water connection. As of 2000, the rates were as follows: P650,00 for consumers whose trenches are dug and back-filled by Water Affairs; P450,00 for consumers who dig their own trenches and lay their own pipes; and P75,00 for consumers who would like to be supplied within a few metres of their compounds. [The pula is Botswana's unit of currency. As at June 2001, when the study was conducted, P1 = $0,1765.] For consumers not included in the Kanye Water Supply Project (consumers who are outside the gazetted waterworks area), the Minister of Mineral Resources and Water Affairs is the only authority who can extend water to these consumers. However, the consumers outside the waterworks area bear the full cost of the connection.

In 2000, there were 5 184 house and yard connections and 164 standpipes in the Kanye Water Works Area (Department of Water Affairs, Annual Report, 2000). However, these connections were extremely inadequate, considering the total population that was serviced by Kanye Water Project. This is because people who lived outside the Water Works Area had to come to Kanye for their water supply, although this meant travelling long distances. The result was that long queues formed at the public standpipes and people waited for hours before having an opportunity to fetch water.

3. Theoretical background

This study is based on the theory of contingent valuation. Economic theory provides the necessary elements to model the decision process of a household's choice of water supply site. The contingent valuation model is based on the consumer demand theory, which is an expressed preference method that minimises the expenditure of the consumer subject to his utility constraint (Boccalett & Moro, 2000).

Mathematically, the model is specified as:

where (X₁, X₂, X₃, … X_n) represents the basket of goods and services, P(P₁, P₂, P₃, …, P_n) is the vector of prices of the respective goods and services and E is the initial environmental condition of the individual consumer, in this case his status quo without private water connection. U defines the utility a representative household derives from the consumption of a particular good or service, in this case potable water that is connected to the house.

Solving equation (1) with respect to equation (2), the first order condition necessary for a minimum yields the compensated or Hicksian Demand Function:

Using the solution from the Hicksian Demand Function, the indirect expenditure function can be obtained as:

From the indirect expenditure function, both the compensating variation (CV) and the equivalent variation (EV) can be derived (Whittington et al., 1988; Hanemann, 1994):

where E¹ is the expected transformation, assuming one is willing to accept an amount that will make the individual as well off as he was before the change in environmental quality. The superscript 0 refers to the status quo situation and 1 refers to the changed condition. In other words, the CV is the minimum amount of money that could be given to an individual to make him as well off as he was before the change in his environmental status (Schlapfer et al., 2005).

In like manner, EV can be calculated as:

if the expenditure function is e(P¹, E⁰, U⁰), the amount the consumer is willing to pay for an improvement in the environmental quality in order to maintain the initial level of utility is measured as:

This is given by the area under the Hicksian demand curve and is measured by CV. From equation (5) CV = e(P¹,E⁰,U⁰) − e(P¹,E¹,U⁰), it then follows that:

In the CVM (contingent valuation model), the valuation task is to determine how much better or worse off an individual is or would be as a result of a change in environmental quality. While the compensating variation measures how much income the individual is willing to pay (WTP) to secure a welfare improvement, the equivalent variation on the other hand measures how much income an individual is willing to accept (WTA) for deterioration in the environmental quality (Zeiler & Plott, 2004). In other words, theoretically, WTP and WTA are supposed to provide similar results (Isik, 2004), but empirically they give different estimates such that the estimates based on WTA tend to be greater than those based on WTP. Randall et al. (1983) have suggested that compensation measure (WTP) is preferable, as it relates valuation of gains or losses to the status given and thus fits more easily with potential Pareto improvement criterion (Mpenda, 2000).

For the above reasons, the study employed WTP to discover how much Kanye residents would be willing to pay to secure welfare improvement in terms of better water quality and supply.

4. Methodology

A simple random sampling technique was used to collect data. First, the households were numbered using existing information at the District headquarters. Thereafter, 135 households were randomly selected for the purposes of data collection. For the survey, a contingent valuation questionnaire was designed and administered to the selected households. The questionnaire was structured into three specific categories: household information, information on health and water use, and information on households' WTP for private connection.

The valuation task is to determine how much better or worse off individuals are or would be as a result of a change in environmental quality. An econometric model estimated using a multiple linear regression equation of the following form was postulated:

where: WTP = WTP bid by household; INC = household income (P/month); HHS = household size (number of all persons of all ages); ED = education level of respondents (number of years of formal schooling); AGE = age of respondent (years); DIS = distance of household from existing water sources (km); EMP = employment status of household (dummy, 1 if employed, 0 otherwise); GEN = male or female (dummy, 1 for male, 0 otherwise); IWD = incidence of water-borne disease in the family (dummy, 1 if the household has been infected with any water related disease, 0 otherwise); and U_i = random error term assumed distributed as normal with mean and variance i.e. (0,σ²).

The ordinary least-square regression technique was used to estimate the coefficients of the inadequate variables.

The characteristics of the households included in the study are as follows, with the expected signs in parenthesis, and Table 5 shows the variables and their units of measurement and expected a priori signs.

Table 5: Variables, units of measurement and expected signs

Variable	Unit of measurement	Expected sign
Income	Money (pula)	>0
Household size	Number	<0
Education	Years	0
Age	Years	??
Relative distance from existing sources	Kilometres	>0
Employment status	Dummy 1 if the respondent is employed and 0 otherwise	>0
Gender	Dummy 1 = male 0 = female	<0
Awareness of water borne diseases	Dummy 1 = if the respondent is aware and 0 otherwise	>0
Note: ?? Implies that the sign cannot be determined a priori.

4.1 Income (+)

This variable included the total monthly disposable income available to the household. Economic theory on environmental issues assumes that the higher the income the more the demand for improved environmental quality.

4.2 Household size (−)

The household includes all people who are under the direct responsibility of the household head and make common decisions about what to produce and consume. At a given income level, households with many members are less likely to pay for private connections since the cost of water increases with usage.

4.3 Education (+)

Household heads with some formal education are more likely to be aware of the health implications of alternative sources of water.

4.4 Age (?)

[? implies that the sign could not be determined a priori]

Older household heads who are used to the traditional free sources of supply may be less willing to switch to a new source, especially when the switch entails user fees. However, because water is heavy to carry, and where the sources are far away from the residence, older people who may not be able to transport water a reasonable distance would seek private water connection.

4.5 Distance from existing sources (+)

This variable was measured in kilometres. All other things remaining constant, it is assumed that households located furthest away from existing water sources may be more willing to pay for a conveniently located source of supply.

4.6 Employment status (+)

The household head's ability to earn income periodically makes him eligible to connect water into the house.

4.7 Gender (+)

Women heads of households are expected to be more willing to pay because traditionally it is the role of women to ensure that water is available in the house. In addition, it is assumed that women generally have a higher demand for household water than do men.

4.8 Incidence of water-borne diseases (+)

It is hypothesised that households that have experienced water-borne diseases will be more willing to pay for private pipe-borne water supply. The variable is specified as a dummy: 1 for households that have experienced water related diseases and illness and 0 otherwise.

5. Empirical results and discussion

The data were entered in SPSS and Microsoft Excel for ease of data cleaning and for descriptive analysis. Frequencies and cross-table analyses were carried out as a prelude to the regression analysis.

5.1 Socio-economic characteristics of the respondent

First, the socio-economic characteristics of the respondents were investigated. The results suggest that the data are a good representation of the characteristics of the population under study. In particular, the mean of the variables such as level of education, income and age of the respondents compare favourably with those that have been reported elsewhere (see Government of Botswana, 1994). Table 6 shows the mean values of the variables included in the regression analysis. The mean WTP of the 135 respondents in the survey was obtained as P171,50. This result implies that the respondents interviewed are willing, on average, to pay P171 to have a private water connection for the household.

Table 6: Descriptive statistics of the respondents

Variable	Mean	SD
Willingness to pay bid	171,50	124,32
Income	1 333,90	217,19
Household size	6,66	2,61
Education	4,89	3,60
Age	44,89	13,70
Distance	0,79	0,29
Employment*	0,74	0,44
Gender*	0,53	0,50
Incidence of water-borne* diseases	0,85	0,43
Note: * Variables specified as dummy.

The data also revealed that the mean monthly household income was about P1 300. Households in the income group of between P1 000 and P1 500 constituted the model income group in the study area, with about 22 per cent of the respondents belonging to this class. The average family size was six, with about 43 per cent of the households having more than six members in the family. The average numbers of years of formal education completed was five years, with only about 27 per cent of the respondents having completed seven or more years of formal education. About 63 per cent of the respondents did not have more than primary school education. The average age of the respondents was 45 years and 47 per cent of the households were female-headed households. The average distance from respondents' home to a water stand was 0,8 km, with the recorded furthest distance being 1,2 km. The individual household WTP bids for private water connection ranged from P10 to P600 for all the respondents interviewed. The study shows that in general about 99 per cent of the households interviewed were willing to pay for private water connection.

5.2 Willingness to pay for private water connection: results from multivariate analysis

In this section, the results of the relationship between household's willingness to pay for private water connection and selected socio-economic variables are presented. The correlation matrix obtained from the Shazam program used in analysing the data shows there is no problem of multicollinearity in the variables used. In addition, and as expected, the data were also free from the problem of auto or serial correlation, as shown by the value of the Durbin Waston statistic, which was obtained to be 2,0125. According to Studenmund (2000), d ≈ 2 indicates that there is no serial correlation. Table 7 summarises the effect of selected factors on a household's WTP for private connections.

Table 7: Regression results for WTP for private water connections

Variable	Parameter estimates	SE	t-value
Constant	30,953	46,935	0,659
Income	0,437***	0,057	7,623
Household size	−6,555*	3,654	−1,794
Education	1,456**	0,606	2,403
Age	0,844	0,685	1,231
Distance	11,756***	4,074	2,886
Employment	1,038***	0,288	3,064
Gender	8,101	18,037	0,449
Awareness	0,967***	0,299	3,234
Notes:
N = 135, d.f. = 126; F-statistic = 12,88; R^2 = 0,647.
*Significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

The overall fit (R² = 0.647) is reasonably good, considering that data for the study were obtained from a cross-sectional survey of selected individuals from the study area. Next, the influence of the independent variables on the willingness to pay for water connection was examined. The result shows that all the variables included in the model came out with the expected signs. In particular, the variables described by household income, employment, distance from the existing water sources and awareness of water-borne diseases were statistically significant at 1 per cent level of significance. On the other hand, the variable represented by education level of the respondent was statistically significant at 5 per cent, while the size of household variable was significant at the 10 per cent level of significance.

The F-ratio, which had a value of 12,88, was significant at the 1 per cent level indicating, as expected, that all the variables included in the model jointly influenced the respondents' willingness to pay for private water connection. Thus the null hypothesis that the included explanatory variables have no effect on the WTP for private water connection is rejected. The large number of the included explanatory variables that were statistically different from zero and the overall power of the model indicate that the relationship between willingness to pay and the included explanatory variables is not due to chance. Rather, the results show the systematic relationship between households' willingness to pay for improved environmental quality (in this case, private water connection) and households' characteristics. In addition, the results are consistent with established economic theory on contingent valuations.

Furthermore, the following inferences can be drawn from the results of the regression analysis. An increase in household income increases the amount that households are willing to pay for private water connection. In effect, anything that increases households' income would increase the amount they are willing to pay for private water connection. This finding is line with the economic theory of the demand for a good or service, which hypothesises that an increase in income increases the demand for that good, all other things being equal. The finding is also in line with Boadu (1992), who established that income is an important explanatory variable affecting the willingness to pay for private water connection in rural Ghana. However, because the amount respondents are willing to pay is positively related to total household income, it is therefore plausible that not all residents of Kanye would be able to afford private water connection as their marginal utility per dollar for water services would be different. It may therefore be necessary to support low-income families if the objective of providing adequate and clean water for all is to be achieved.

These findings complement the results of a few other studies that have tried to model rural household willingness to pay for improved social services. In effect, although the economies of most rural communities in Africa, Botswana included, may not be adequately monetised, it is possible to analyse the behaviour of rural households using established economic theories such as the demand and consumer behaviour theory, as in the present study.

There is a positive relationship between willingness to pay and employment status of the head of a household. This positive coefficient of the employment variable, which was specified as a dummy, underscores the importance of employment in relation to willingness to pay for private water connection. On the other hand, and as expected, the variable represented by household size exhibited a negative influence on the willingness to pay for private water connection. As defined in the study, the household variable measures the total number of people in the household and does not account for renters and others who may be sharing utility bills with the home-owner. If other occupants of a dwelling share water user fees then per-unit costs could be significantly lower, leading to the non-significance of the estimate of this variable.

The results show that the age of the household head is not a significant factor in explaining the willingness to pay for private connection, although on a priori consideration it possesses the right sign. The plausible explanation for this finding is that age in itself is not an important sociological variable affecting WTP for private water connection and that old and young heads of households value private water connection equally.

6. Policy recommendations

The findings of the study show that the socio-economic conditions of rural consumers (as with urban consumers) do influence their willingness to pay for improved environmental conditions (in this case water services) as hypothesised. In particular, consumers' income tends to influence willingness to pay for private water connection in the study area and is indicative of the fact that not all rural households in the area may be able to pay for it. It is therefore logical that a system of connection and user fees that is tied to individual households' income levels may be a more effective way of recouping investment funds that may be spent on the Kanye Water project. That is, the current policy or practice where the Water Utility Cooperation charges flat connection and user fees for all the consumers in a community, irrespective of their income status, may be problematic for effective delivery of services as it leads to high rates of default and disconnection among the less affluent members of the society. In other words, a Water Project such as Kanye's that seeks to extend private water connection to all the households may be more successful if price, in this case user and connection fees, is differentiated among different income levels for a more effective delivery of water services in the community. Similarly, the fact that the head of household's employment status positively influences private water connection suggests that any government policy that provides jobs for all will also increase consumers' willingness to pay for private water connection. In effect, unemployment militates against the willingness to pay for private water connection.

The significance of the education and level of awareness variables suggests that an increase in the level of education would raise the respondents' level of awareness about safe and reliable sources of domestic water supply. It is suggested that the government should strengthen its effort to increase the level of literacy among the residents of Kanye Village in particular and the population at large. In effect the government and other stakeholders in economic development should consider establishing adult education or any other type of informal education that would highlight the benefits of private water connection. Because the average amount that consumers are willing to pay is higher than that currently charged in most other rural areas of Botswana, it is suggested that the extra revenue that the project would generate be channelled to other developmental efforts such as intensifying the publicity campaign to educate people about the benefits associated with the need to access and pay for improved services. This would go a long way towards helping the government achieve its objective of providing a reliable and safe water supply to all citizens.

7. Conclusion

The study centred on the use of the iterative bidding method of contingent valuation to identify the determinants of willingness to pay for private water connection in Botswana, with the Kanye water project as a case study. This method is especially relevant in the case of Kanye, because this is the way business is conducted in the local markets in most parts of Botswana, including Kanye Village. Furthermore, the study has shown that using established economic theories such as the demand and willingness to pay theory is as effective in rural communities as it is in urban areas in estimating households' willingness to pay for improved services. From the results, the author concludes that some sociological and economic variables such as income, status of employment, level of education and distance from existing sources are the main determinants of households' WTP for private water connection.

Finally, water schemes must be demand-driven rather than supply-driven, as is the case in some other parts of the country. Future proposals on water supply projects should take particular account of the ecological and socio-economic characteristics of the rural population. This will enable the government to determine in a systematic and objective manner the connection and user fees that prospective buyers of services, such as water supply, would be willing to pay. This recommendation is in line with the fact that the average amount that the households are willing to pay for private water connection may vary from one community to another depending on their socio-economic characteristics.

Acknowledgments

The contribution of the research assistants, particularly O Otsetswe, in collecting the data used in this study is gratefully acknowledged.

Additional information

Notes on contributors

James N Mbata

formerly of Department of Economics, University of Botswana.