AN ECONOMIC ANALYSIS OF WATER AND WASTEWATER INVESTMENTS IN CAIRO, EGYPT

Source: Evaluation Review, Dec2000, Vol. 24 Issue 6, p579, 30p

By John P. Hoehn

Nonmarket valuation methods have proved useful in planning and evaluating investments in water and wastewater infrastructure in developing countries. This study used contingent referendum methods to estimate household willingness to pay for each of four types of service improvements stemming from water and wastewater investments in Cairo, Egypt. An analysis of the net economic benefits of the investments concluded that benefits exceeded costs for all projects. Cost recovery was not assured with a fixed tariff. Willingness to pay for some households was less than the per household cost necessary for cost recovery. Cost recovery was also sensitive to whether tariffs were set for individual services or charged for a combined package of services.

Research during the past decade demonstrated that nonmarket valuation methods contribute to effective planning of water and wastewater investments. Nonmarket methods, especially contingent valuation, were readily adapted to estimating water and wastewater benefits. Research showed that benefits, measured by willingness to pay, are significant under a wide range of cultural and demographic conditions (Altaf 1994a, 1994b; Altaf and Hughes 1994; Briscoe et al. 1990; Whittington, Briscoe, Mu, and Baron 1990; Whittington, Lauria, and Mu 1989). But households' willingness to pay was not uniform. Variation in willingness to pay made it difficult to generalize about the benefits and costs of water supply. In any particular situation, willingness to pay might be greater than or less than the costs of water supply improvement (World Bank Water Demand Research Team 1993). Nonmarket valuation was an effective and flexible tool for assessing economic demands, project efficiency, and financial viability (Altaf et al. 1993).

Several factors were linked to variations in the size of willingness to pay. One factor was the quality of water service. Poor quality and unreliable service diminished willingness to pay. A water agency could not expect to simply lay pipes and collect revenue. The quality of service had to be good enough to justify the conversion of willingness to pay into actual payments (Altaf 1994b; Briscoe 1992; Singh et al. 1993). A second factor was the availability of alternative supplies. Private suppliers were active in areas where public systems were inadequate or insufficient. Willingness to pay for a new service varied with the quality and quantity of existing alternatives (Whittington, Lauria, and Mu 1991; Whittington, Briscoe, et al. 1990; Whittington, Okorafor, et al. 1990). A third factor was household characteristics such as income, number of family members, age, and gender. Differences among household characteristics led to differences in willingness to pay (Briscoe et al. 1990; McPhail 1993; Whittington et al. 1992).

The variation in willingness to pay has an important implication for financial viability: Investment revenues collected through voluntary charges might not be sufficient to generate revenues in excess of costs, even if benefits exceed costs. This is especially true for the lump sum tariffs or fees that agencies often charge. Households with willingness to pay less than the tariff will not voluntarily sign up for the service, and no revenue is collected from this portion of the population (McPhail 1994). Households with willingness to pay greater than the tariff sign up for service, but the tariff they pay is less than their willingness to pay. The net result is that revenues are certain to be less than total willingness to pay and may be less than total costs.

This article examines the valuation of water and wastewater investments in one of the world's largest urban areas: Cairo, Egypt. Previous research examined water or wastewater improvements in small villages and secondary cities. The present research examines an investment in an urban environment where substitute sources of water are limited and water use generates immediate wastewater externalities. Unlike previous work that examined one or two services, this study examined four: water connections, improved reliability of existing water service, wastewater connections, and network maintenance to eliminate sewer overflows.[sup1] The results provide insight into the relative value of extending water service to new users versus improving the reliability of service to existing users. It was also clear that wastewater disposal was an almost essential complement to urban water service. Without adequate wastewater disposal, the externalities of urban water use were all too apparent in the city's streets, alleys, and canals.

A benefit-cost analysis examined the net economic values of alternative water and wastewater investments. Overall, the investments showed a positive economic value, even at a discount rate of 10%. Net economic values varied across specific services. The net value of allocating water to new service connections was slightly larger than that of improved reliability--at least within the range of reliability within Cairo. The net value of wastewater maintenance was several times larger than extension of the wastewater network to add new connections. The net values of wastewater connections were reduced by the high cost of wastewater treatment. The high cost of wastewater treatment raised the issue of how to allocate treatment costs. Finally, a tariff analysis examined the potential for collecting revenues in excess of costs.

This article is organized in the following manner: The first section describes the investments and the service conditions before and after the infrastructure improvements. The second section discusses the research design. Subsequent sections present the data collection procedures, the willingness-to-pay results, and the benefit-cost and tariff analysis. The article concludes with a brief discussion.

BACKGROUND ON WATER AND WASTEWATER INVESTMENTS

Assessments of water (ES-Parsons 1979) and wastewater (John Taylor and Sons and Binnie and Partners [Taylor-Binnie] 1977) facilities in Cairo in the late 1970s identified substantial constraints to the provision of adequate services. In particular, much of the existing infrastructure was in poor repair and lacked the capacity necessary to serve the population connected to the networks. Furthermore, a substantial portion of households was not connected to the water and wastewater networks. The network reviews provided a basis for negotiations between the Government of Egypt (GOE) and international assistance agencies to improve water and wastewater services. Between 1979 and 1996, a series of capital investments substantially improved and extended the water and wastewater infrastructure. Economic evaluation of the investments began with a technical review of the capital improvements with the objective of identifying likely impacts of the projects on services experienced by households. This section reviews preproject water and wastewater service conditions, outlines the projects, and identifies four key service improvements associated with the projects.

WATER SERVICE CONDITIONS, PROJECTS, AND IMPACTS

The 1979 review of Cairo's potable water network concluded that water pressure in much of Cairo was too low and too variable to provide adequate water service to city residents (ES-Parsons 1979). As a result, many households experienced daily interruptions of water service--particularly during hours of peak demand (Taylor-Binnie 1977). Households compensated for low pressure by scheduling water-intensive chores for times when adequate water was available--often late at night, installing electric pumps on water lines to augment pressure and storing water in the home and in rooftop tanks for use when regular water service was interrupted (AMBRIC 1987; Tecke, Oldham, and Shorter 1994).

In addition, the water network did not extend to all households in the city. The 1976 census reported that 36% of buildings in greater Cairo were not connected to the water network (CAPMAS 1986a, 1986b, 1986c). Thus, as many as 2.7 million people may have been without piped water in their homes. Households not connected to the network obtained water from households and businesses connected to the water network, public standpipes, mosques, canals and drains, shallow wells, and water deliverers. Women, who bear the primary responsibility for household chores in Egypt, spent substantial time and effort carrying water to their homes. In addition, they were often embarrassed to ask neighbors for water and experienced frequent arguments and fights over access to public standpipes. When households had to pay money for water, it could also be expensive (Nadim et al. 1980).

The water network review identified three primary causes of service deficiencies. First, the network contained inadequate water treatment capacity to serve the population or to permit regular maintenance and needed repairs of water treatment plants. Existing treatment plants were thus deteriorating and in need of rehabilitation. Second, the system lacked adequate storage capacity. Inadequate storage capacity meant that water could not be stored during slack times to augment water available during hours of peak demand. Third, deteriorating transmission and distribution networks could not Withstand the pressures needed to provide reliable water service without breaking.

Based on recommendations from the review, the GOE, with the aid of various international assistance agencies, began a series of rehabilitation and construction projects designed to relieve constraints in the water network. Projects focused on the central 63 meter zone--so called because it required 63 meters of water pressure to provide reliable water service to residents. The 63 meter zone encompassed the primary business and tourist districts in downtown Cairo. In 1986, it contained a population of 3.1 million people, about 32% of the population of greater Cairo (CAPMAS 1986a, 1986b, 1986c). Capital investments began in 1982 and included rehabilitation and expansion of water treatment plants, expansion of storage capacity, and rehabilitation and expansion of the transmission and distribution networks.

The water projects were intended to provide reliable, all-day water service to all residents of the 63 meter zone. In the short run, however, the projects appear to have increased treatment capacity over and above what was needed to improve pressure in the 63 meter zone. It is thus likely that the projects made water available to serve new network connections for households not previously connected. The benefits assessment thus estimated benefits for two service improvements: (a) reliable, all-day water service for households connected to the water network and (b) a new network connection for households without in-home water service.

WASTEWATER SERVICE CONDITIONS, PROJECTS, AND IMPACTS

The review of Cairo's wastewater infrastructure identified serious inadequacies in wastewater conveyance and treatment (Taylor-Binnie 1977). Sewers and pumping stations in many Cairo neighborhoods were undersized and in poor repair. As a result, sewage routinely spilled into the streets as sewer mains overflowed and pumping stations broke down (AMBRIC 1981). Residents of areas with chronic sewer overflows experienced foul odors, flies and mosquitoes, and associated health risks. Sewage in the streets also impeded traffic and made it difficult to keep children, homes, and clothing clean (EQI 1988). In some areas, sewage flowed into homes and businesses, damaging buildings and merchandise. Sewer overflows were so common in some neighborhoods that residents constructed raised paths of dirt, planks, or stones to walk on and built dikes of dirt to keep sewage out of buildings.

The wastewater infrastructure also served only a portion of Cairo residents. The 1976 census reported that 55% of buildings in Cairo were not connected to the sewers at all (CAPMAS 1986a, 1986b, 1986c). As many as 4 million people may have been without sewer services. Most of the unsewered areas were on the rapidly urbanizing west bank of the Nile River. Most households that were not connected to the sewer system constructed masonry vaults under the streets in front of their homes into which household wastewater flowed (Nadim et al. 1980; Oldham, El Hadidi, and Tamaa 1987). A small number of households were able to construct private sewers that emptied directly into nearby canals and drains.

Although vaults appeared to work well in rural areas, they tended to raise groundwater levels in congested urban areas and require frequent cleaning. Vaults were typically emptied by pumping the contents into a truck, which then hauled the sewage to a nearby canal or drain or into the desert. Emptying vaults was expensive, dirty, and inconvenient (Hoehn and Krieger 1996). Vault overflows caused local environmental conditions similar to those associated with flooding sewers in sewered areas. Even when vaults were cleaned before they overflowed, much of the sewage was dumped into local drains and canals where it continued to impact local environmental conditions and health (AMBRIC 1981).

In response to the wastewater conditions, the GOE and international donor agencies began a series of capital investments to improve wastewater services. Capital investments began in 1979 and included projects to improve conveyance and treatment capacity and expand the sewer network. The intended impact of the conveyance investments was to eliminate sewer overflows in areas with sewers and expand the sewer network into some areas without sewers. The assessment thus estimated benefits for two service improvements: (a) elimination of sewer flooding and (b) installation of sewer connections. Investments in treatment capacity were necessary to treat existing sewage before discharge as well as to treat the additional sewage collected from the intended network expansion.

DESIGN OF THE BENEFIT-COST ANALYSIS

A lack of data on the value of water and wastewater services to Cairo households shaped the design of the benefits assessment. Market prices typically provide an important source of value information, but adequate price series did not exist in this case. Most important, as an obstacle to valuation, charges for water and wastewater services were determined by tariffs that bore no relationship to the quantity of services used. Households were thus unable to adjust the quantities of water and wastewater services consumed in relationship to the values they placed on the services. The lump sum nature of the tariff meant that household payments did not reveal the values households placed on water and wastewater services. In addition, the size of the tariffs was unrelated to the costs of providing the services. Water and wastewater services in Cairo were administered as independent bureaus of the national government. Tariffs paid to these agencies were regarded as revenues of the national government rather than revenues of the agencies themselves. The size of the tariffs were fixed in a manner similar to other taxes and traditionally had no relationship to the cost or value of supplying water and wastewater services.

Private water and wastewater haulers set prices for substitute services in many neighborhoods that were not served by the public systems. These haulers, however, suffered from declining markets due to the expansion of the public system and often operated at the margins of legality. Laws prevented the sale of water in some situations, and wastewater haulers often disposed of wastewater surreptitiously in canals and empty lots at the outskirts of the city. As a result, data on prices and quantities of these privately supplied services were irregular at best and completely unavailable at worst.

In the absence of prices, the study used nonmarket methods to estimate the benefits of water and wastewater service improvements. The study relied primarily on one type of stated-preference method, a contingent referendum. A contingent referendum is one variant of a contingent valuation choice experiment (Hoehn and Randall 1987; Mitchell and Carson 1989). A contingent referendum presents a respondent with the choice of accepting or rejecting a program that would create a nonmarket good, such as household connections to a neighborhood water or wastewater system. The referendum describes the program and offers the household the choice of voting for or against the program at a cost to the household specified as part of the program.

The referendum approach presents respondents with a relatively easy and familiar valuation decision. A person answering for a household does not have to know exactly how much the household is willing to pay for the program. The person only has to decide whether the household's willingness to pay is greater than or less than the stated program cost. If the household's willingness to pay is greater than the stated per household cost, the person will make the household better off by voting for the program. By voting for the program, the household gains the prospect of getting the program at a cost less than its maximum willingness to pay. If the household's willingness to pay is less than the specified per household cost, the household is better off to vote against the program and reduce the chance of paying a cost that exceeds the household's willingness to pay. The net effect is that the referendum encourages an accurate statement of preferences.

To estimate willingness to pay for a program, the contingent referendum approach is administered to a sample of households. A set of cost thresholds is selected and randomly assigned to households. Thus, within the sample, different households receive a different stated per household cost. Each household is asked whether it would vote for or against the program at the stated cost to the household. Economic behavior suggests that households receiving a high cost threshold will be less likely to support the program than households receiving a low cost threshold. Hence, there should be a correlation between the pattern of voting and the pattern of stated costs. Because households make their voting choice by comparing what the program is worth with what it costs, the pattern of votes implicitly reveals the distribution of willingness to pay across households.

The implicit pattern is made explicit by statistically modeling the probability of a vote in favor of the program. The probability that the ith household will vote yes on program j is the same probability that the ith household's willingness to pay for the jth program, w[subij], is greater than the program cost presented to the household, c[subij],

Multiple line equation(s) can not be represented in ASCII text.

where w[subj] is mean willingness to pay for the jth project and epsilon[subij] is the deviation between the mean and w[subij].

For benefit-cost analysis, mean willingness to pay is of particular interest. Total benefits are simply mean willingness to pay multiplied by the affected population. The first step in estimating mean willingness to pay is to select a specific cumulative distribution to represent the probability distribution, Prob(ש) in Equation 1. For instance, Cameron and James (CJ) (1987) assume that epsilon[subij] is distributed N(0, sigma) and replace Prob(ש) with a normal distribution. With PHI(ש) representing the cumulative normal distribution, the CJ approach thus models the probability of a yes response as

Multiple line equation(s) can not be represented in ASCII text.

where ??[subj] and ??[subj] are parameters representing the mean and standard deviation, respectively, of the normal distribution, PHI(ש). Equation 2 is similar to a probit estimator (Maddala 1977), but the CJ construction identifies both ??[subj] and ??[subj] and sets the coefficient of the variable c[subij] to one.

Willingness to pay is either positive or zero. It cannot be negative for a program that provides a real good. Because the left tail of the normal distribution is defined over negative values, the mean of the normal distribution is not the mean of willingness to pay (Habb and McConnell 1998; Hanemann and Kanninen forthcoming). To estimate mean willingness to pay using the normal distribution, the domain of willingness to pay was restricted to the nonnegative values. Mean willingness to pay is a conditional expected value that incorporated the restriction to nonnegative values

Multiple line equation(s) can not be represented in ASCII text.

where z is the variable of integration and psi(ש) is the normal probability density function (Hanemann and Kanennin forthcoming). In Equation 3, the cumulative normal probability of a negative value is redefined as the probability that willingness to pay is zero.

The third line of Equation 3 can be integrated and rewritten in terms of the parameters of the normal distribution. Mean willingness to pay for the jth program is

Multiple line equation(s) can not be represented in ASCII text.

where psi(--??[subj]/??[subj]) is the normal probability density function at zero (Pudney 1989). Because the standard deviation, ??[subj], and the density function, ??(ש), are positive, the estimator of mean willingness to pay, w[subi], is larger than the estimated mean of the normal distribution, ??[subj].

Mean willingness to pay as estimated by Equation 4 rests upon a specific distributional assumption, that of a normal distribution. Other specific distributions may yield different estimates of mean willingness to pay. To provide a check on the validity of the normality assumption, distribution-free estimates were also computed. The distribution-free estimator provides both upper and lower bounds on mean willingness to pay (Cosslett 1983; Haab and McConnell 1998; Kristrom 1990). If the normal distribution is valid, Equation 4 should yield an estimate that lies between the distribution-free lower and upper bounds. The distribution-free estimator is sensitive to the values and number of the cost thresholds, the c[subij]. This sensitivity declines as sample size increases (Hanemann and Kanninen forthcoming). This study used a relatively large number of thresholds, evenly spaced across the value distribution.

Mean willingness to pay provides a key component of a benefit-cost evaluation. The benefit-cost evaluation compares the present value of investment benefits with the present value of investment costs. The present value of benefits, b[subi], for the jth program is the jth mean willingness to pay times the number of households affected by the program in year y, n[suby], discounted at the appropriate discount rate, r, and then summed over the life of the program,

Multiple line equation(s) can not be represented in ASCII text.

where L[subj] is the life of the jth program, and a denotes the base year for the bene-fit-cost evaluation. The summation in the second line of Equation 5 is the present value of the affected population, P[subi]. The present value of benefits can thus be written as mean willingness to pay times the present value of the affected population,

b[subj] = w[subj]P[subj]. (6)

Dividing both sides of Equation 6 by P[subj] results in mean willingness to pay as the per household measure of program benefits,

w[subj] = b[subj] / P[subj]. (7)

Similarly, the per household measure of program costs, c[subj], may be written as the present value of costs divided by P[subi], or

Multiple line equation(s) can not be represented in ASCII text.

where TC[subjy] represents investment costs associated with program j in year y.

Equations 7 and 8 permit a comparison of benefits and costs at the scale of the household. In addition, by taking a population weighted average of Equations 7 and 8, the benefits and costs of different sets of programs may be compared. For instance, the population weighted benefit of the set of four water and wastewater programs is

Multiple line equation(s) can not be represented in ASCII text.

and the comparable measure of aggregate costs is

Multiple line equation(s) can not be represented in ASCII text.

where

Multiple line equation(s) can not be represented in ASCII text.

is the population-based weight for the jth project and the set A includes the water connection program, the water reliability program, the wastewater maintenance program, and the wastewater extension program.

Equations 7 through 10 are the basic templates for the benefit assessment and benefit-cost analysis. The analysis requires a measure of household willingness to pay for each of the four programs as specified by Equation 7. These may then be compared to program costs per household as given by Equation 8. Weighted average aggregate benefits per household may be computed by Equation 9 and compared to average aggregate investment costs per household computed by Equation 10.

QUESTIONNAIRES AND DATA COLLECTION

The data collection effort consisted of designing contingent referendum questionnaires for each of the four programs, selecting an appropriate sample for the survey, and administering the survey effort. The questionnaire development and survey administration effort employed both qualitative and quantitative research methods. Qualitative research provided an understanding of Cairo residents' experience with water and wastewater services and contributed to development of a questionnaire that communicated unambiguously with respondents. Quantitative research consisted of a large-scale survey of Cairo households. This section reviews the questionnaire development process and details of the household survey.

QUESTIONNAIRE DEVELOPMENT

Questionnaire development began with qualitative research to determine whether Cairo residents were sufficiently aware of water and wastewater services to be able to respond deliberatively to the questionnaires. Valid contingent referendum results required that residents viewed water and wastewater services as economic goods. It seemed possible that residents might view adequate water and wastewater services as basic entitlements to be provided free by government rather than as services that required payment. If the entitlement view was common, respondents might reject a contingent referendum that required households to pay for improved service.

The qualitative research consisted of a series of 15 focus groups. Participants were carefully selected to represent a cross-section of socioeconomic backgrounds and a range of water and wastewater conditions.[sup2] A professional market research firm[sup3] conducted the focus groups in Arabic and provided simultaneous oral translation to English. The discussion guides for the focus groups drew heavily on ethnographic methods and avoided the vagaries of general attitudinal questions (Spradley 1979). The discussions focused on participants' experiences with water and wastewater services, the perceived quality of the services, the characteristics of services that were important to participants, and the language they used to describe their experiences.

The qualitative work demonstrated that Cairo residents were aware of water and wastewater services, that the services were important, and that the services were economic goods. It provided the insights into daily life and language necessary for the development of credible, contingent referendum questionnaires. Four questionnaire prototypes were constructed to value each of the services provided by the four investment programs. These services were (a) reliable, all-day water service; (b) the opportunity to connect to the water network; (c) wastewater network maintenance to eliminate sewer overflows; and (d) an in-home connection to the wastewater network. Each questionnaire was designed to be administered as a door-to-door personal interview conducted by a professional interviewer.

Two waves of pretesting were used to edit and refine the questionnaires. A local survey research firm conducted the first wave of 135 interviews at its offices. Respondents were drawn from neighborhoods where the final survey was expected to be conducted. Pretesting sought to identify potential ambiguities in the questionnaires and develop credible program and referenda descriptions. The descriptions needed to contain enough detail about a proposed water or wastewater program to be credible. As pretesting proceeded, the texts of the questionnaires grew in length to address the wide variety of questions asked by respondents about the program proposals. As the text grew in length, respondents' attention seemed to waiver.

To address the limits on respondents' attention, a major revision midway through pretesting gave the program descriptions and referenda a more conversational form (Sudman, Bradburn, and Schwarz 1996). The description of the programs and the referenda were broken up into interactive segments of the text. For instance, the first segment introduced the program and then asked the respondent for an open-ended evaluation of how the project would affect his or her daily life. The second segment described details, such as the program's location and duration, and then asked the respondent if he or she had questions about the project. The next two segments described the cost of the program and the referendum. A final segment asked respondents to describe their reasons for voting as they did.

The revised questionnaires were tested and refined in a second wave of 170 in-office pretests. The pretests showed a significant increase in the amount of interaction between the interviewer and respondent regarding the program. With the revisions, the interviewer read a brief segment, and the respondent reacted to the segment in response to an open-ended question. The length of uninterrupted narration dropped to 185 words in the revised draft from 450 words in the previous drafts. Respondents reacted favorably to the opportunity to think out loud about how the programs might affect their daily lives. Given the opportunity to voice questions about the project, many respondents raised a question about costs--a question answered in the next segment. Hence, the sequence in which the questionnaire presented information seemed to match the cognitive pattern of many respondents. Finally, in the debriefing questions following their votes, most respondents had little difficulty in voicing specific and relevant reasons for voting the way they did. Key reasons were the programs' per household cost relative to the household budget, the impacts on respondents' lives, and the quality of services offered. Thus, the final questionnaires appeared to encourage carefully considered choices that were consistent with economic parameters such as price and income.

The final stratified survey sample was drawn from areas of Cairo that had water and wastewater services similar to preinvestment conditions. Sampling from these areas would result in willingness-to-pay values consistent with a change from without-program conditions to with-program conditions. A separate stratum was selected for each of the four questionnaires. Thus, the water connection questionnaire was given to households that did not have in-home connections to the water network. The water reliability questionnaire targeted households that had water service similar to that in water reliability program areas before the completion of those projects. The wastewater connection questionnaire was given to households in neighborhoods outside the wastewater network.

With the sewer-flooding questionnaire, it was not possible to identify areas of Cairo that were certain to have preprogram conditions. The stratum for this questionnaire was defined as the sewered areas of Cairo. The questionnaire contained a response-dependent bifurcation. If a sampled household lived in an area subject to occasional overflows, the household received a referendum on a maintenance program to eliminate flooding. If a sampled household lived in a area that was free of occasional overflows, the household received a referendum regarding a maintenance program to prevent the occurrence of future flooding.

Forty sampling points were drawn randomly from a geographical information system (GIS) for each questionnaire. The GIS cross-referenced population, census area boundaries, settlement patterns from recent satellite data, and current and preinvestment water and wastewater service conditions. Because street addresses were unknown in many lower income areas of Cairo, sample points were identified by longitude and latitude and located using a handheld geopositioning device. Geosampling resulted in a probability sample based on the most recent census of population.

A team of specially trained interviewers visited each sampling point. The interviewers were instructed to complete approximately 25 questionnaire interviews for each sampling point. Interviewers asked to speak to either the male or female head of the household. As interviewing progressed, the interviewers targeted a specific gender to obtain approximately equal numbers of male and female respondents.

SURVEY SAMPLE RESULTS

The survey obtained completed questionnaires from 3,918 Cairo households. Approximately 1,000 households responded to each of the water reliability, water connection, and wastewater maintenance questionnaires. A total of 903 households responded to the wastewater connection questionnaire. The mean age of respondents was 43 years. Of respondents, 56% had completed a primary education and 15% had completed a university degree. The mean household size was 5.1 people, with a mean of 3.6 persons younger than age 14.

Respondents within the water reliability stratum were selected from neighborhoods in Cairo that had water reliability conditions similar to those existing prior to the infrastructure improvements. In this stratum, 90% of responding households experienced low water pressure and cutoffs; 67% experienced cutoffs more than once per week, and the mean duration of the last cutoff was 8 hours. Of respondents, 17% were able to maintain water pressure and avoid water service interruptions by using pumps and storage tanks. The mean cost of operating the pumps and tanks was $1.86 per household per month.

Of respondents, 98% paid a cash amount for water service, and 2% paid for water as part of a rental payment. The mean cash payment was $1.54 per household per month. The mean addition to rent to pay for water service was $1.39 per household per month. Although commonly perceived as water service payments, water service fees included surcharges for sewer service as well. However, water service charges differed by only 39?? per month between households connected and not connected to the wastewater network. Of the households within this stratum, 81% had both water and wastewater service. These households paid a mean of $1.63 per household per month. The remaining 19% with network water service but no wastewater service paid a mean of $1.24 per household per month.

Respondents in the water connection stratum were selected from neighborhoods outside the water network. Households in this stratum spent a mean of 28 hours and $1.18 per household per month to obtain water. Of respondents in this stratum, 71% purchased no water but spent a mean of 34 hours per household per month obtaining water; 20% purchased some water. These households spent a mean of 26 hours and a mean of $6.80 per household per month to obtain water.

The wastewater connection stratum was drawn from neighborhoods without access to the wastewater network. Sewerage vaults in these neighborhoods were associated with problems similar to those mentioned in the focus groups. Of respondents, 75% said that the vaults were frequently a source of mosquitoes and flies, 60% mentioned frequent problems with vault odors, and 45% said that they were a frequent cause of arguments between neighbors. About a third of respondents reported frequent problems with vaults leading to wet streets, soiling of clothing, and property damage due to leakage.

The costs of maintaining and evacuating the sewage vaults varied across respondents. Of respondents, 36% experienced no vault-related costs in the 12 months prior to the interview, 11% paid more than $17.60 for vault cleaning in the 4 weeks prior to the interview, and 39% paid a mean of $5.03 for vault cleaning during the same period. The sample mean was $4.14 for vault cleaning during the 4 weeks prior to the interview.

Respondents to the wastewater maintenance questionnaire were selected from among all neighborhoods in Cairo with wastewater services. Of respondents, 38% reported no wastewater overflows during the previous 12 months, and 50% reported that wastewater in their neighborhood had overflowed frequently or all the time in the year prior to the interview. Those who had experienced wastewater overflows described impacts similar to those associated with overflows of sewage vaults.

The results indicated that households not connected to the water and wastewater networks paid more in direct charges for water and wastewater services than connected households. Households in the water reliability stratum with water and wastewater paid tariffs of $1.63 per household per month. Households outside the networks had direct water and wastewater costs that averaged $5.32 and 28 hours of labor per household per month. Outside the network, households with the fewest resources paid higher direct costs for services that were poorer in quality.

THE REFERENDA AND WILLINGNESS TO PAY

Each contingent referendum began with a brief text segment that described the relevant infrastructure program. The interviewer first read the text and then asked the respondent to describe how the program would affect the daily life of the household. Table 1 lists the types and frequency of responses to the impacts question. Most notable is the high percentage of respondents who described at least one impact on their lives. All respondents in the water and wastewater connections strata described at least one impact following the brief program description. The lower response rates for the water reliability and sewer flooding programs indicated that the effects of these programs were less obvious or less immediate for about 20% of respondents.

On average, respondents from the water connection stratum mentioned 5.1 separate impacts. Respondents who were offered the program appeared to have little difficulty relating the program to impacts on their daily lives. They mentioned primarily impacts on household chores and time savings. Specific comments included items such as cleaning would be easier and done more frequently, clothes could be washed more often and would be cleaner, and tea would taste better than tea made from salty water drawn from shallow wells. Specific comments also indicated that women would have more time for other activities, including caring for children and working outside the home. Respondents also thought a water connection would make their neighborhood and local environment cleaner.

The average respondent in the wastewater connection stratum mentioned 4.2 separate impacts. The two most commonly mentioned categories of responses were those regarding household chores and activities and effects on the neighborhood. Effects on household chores and activities stemmed primarily from removing the capacity constraint of the sewage vaults. Households would be able to take showers and wash clothes without worrying about filling their vaults. Children's clothes would be cleaner and need less washing when vault overflows were eliminated. Almost 90% of respondents knew that the neighborhood and local environmental would be cleaner and free of odor when the vaults were replaced by the wastewater network.

Impacts due to reliable water were similar but less numerous than those due to a water connection. The main impacts of a reliable water supply were on household chores and the neighborhood. Households would be able to plan and carry out chores such as washing, cooking, and cleaning in a set routine rather than having to work around water outages. As with a water connection, respondents thought that a reliable water supply would result in a cleaner and more appealing neighborhood. Unlike those evaluating water connections, most respondents did not mention time savings in the context of the water reliability program. Unreliable water apparently did not mean that household members had to spend time bringing water in from other neighborhoods. Instead, households rearranged their chores and activities to fit the availability of water.

Impacts due to maintenance to eliminate or prevent wastewater flooding were similar but less numerous than those of the wastewater connection program. The mean number of impacts cited was 2.8 versus 4.2 for the wastewater connection program. As with the connection program, the most frequently cited impacts were impacts on household chores and activities and the impacts on the neighborhood environment.

After eliciting perceived impacts, the referendum portion of the question detailed the timing, location, and resulting service to be obtained from the program. Respondents were then asked for any questions they might have about the project. As in the pretests, most questions were about the cost of the programs. The portion of respondents with questions about cost ranged from a high of 38% in the water connection questionnaire to a low of 28% in the wastewater maintenance questionnaire. Less than 10% of respondents had questions about some other aspect of the infrastructure programs that were unanswered by the standard questionnaire narrative.

The next segment of the referendum described the necessity of paying for the programs, stated a specific household cost, and elicited the respondent's vote on whether to go ahead with the program. To remove the chance of biasing the votes, the questionnaire narrative stated that "some people we had talked to previously" had voted for the program and others had voted against it. The questionnaire then gave a brief list of reasons to remind respondents of the trade-offs and budget constraints involved in a voting decision.

Nine cost thresholds were selected for each infrastructure program. The pretests provided information for selecting appropriate cost thresholds. Each respondent was randomly assigned one of the nine cost thresholds as the per household cost for a given program. The lowest threshold for each program was selected so that approximately 90% of respondents would accept the program if they were offered the program at that cost threshold. The highest threshold for each program was selected so that approximately 10% would accept the program if they were offered it at that cost threshold (Alberini 1995). This meant that thresholds for more valuable services were higher than the thresholds for less valuable services. Cost thresholds for the water connection program ranged from 30?? to $2.66 per week. Cost thresholds for the wastewater connection program were higher by about 18?? per week than the water connection thresholds. The water reliability thresholds ranged from about 6?? to $1.78 per week. The wastewater maintenance program thresholds ranged from 7?? to $1.18 per week.[sup4]

Voting patterns within strata were consistent with valid economic choices. The frequency of yes votes declined as the stated per household cost increased. For example, within the water connection strata, 96% of those receiving the lowest cost threshold voted for the program. In contrast, only 21% of those receiving the water program's highest threshold voted for the program.[sup5] Across all programs, the number of respondents who voted for the program tended to be higher than suggested from the pretest data. Favorable responses at the highest cost threshold ranged from 11% to 21% instead of the expected 10%. Excluding the water program, favorable responses at the lowest thresholds ranged from 64% to 88% instead of the expected 90%. Thus, even with a large pretest sample, predicting the tails of the distribution in the full sample was difficult.

Table 2 presents the normal distribution parameters, willingness-to-pay estimates, and distribution-free bounds for each program. Normal parameters were estimated by applying maximum likelihood methods to Equation 2 and the referenda response data. Parameter estimates are significantly different from zero at all conventional levels of significance. The largest mean was for the water connection program, and the smallest was for the wastewater maintenance program. The standard deviation as a percentage of the normal mean increases as the program means decline. The standard deviation for the water connection program is about 40% smaller than the program mean, whereas the standard deviation for the wastewater maintenance program is about 70% larger than the program mean. A likelihood ratio test was conducted to determine if these differences were statistically significant. The null hypothesis was that the means and standard deviations were the same across programs. A chi[sup2] statistic rejected the null at all conventional significance levels.

The fourth column of Table 2 lists willingness-to-pay estimates for each program. These were estimated from the estimated normal parameters and Equation 4. The size relationship between willingness-to-pay estimates was similar to the relationship among the normal means. The largest willingness to pay estimate was $7.77 per month for the water connection program, and the smallest was $2.22 per month for the wastewater maintenance program. The difference between the normal mean and the willingness-to-pay estimate for a particular program was attributable, through Equation 4, to the normal standard deviation. The percentage effect of the standard deviation was smallest for the water connection willingness-to-pay estimate and largest for the wastewater maintenance program.

Differences between willingness-to-pay estimates across programs seemed consistent with pretest data and open-ended statements of program impacts. The largest willingness-to-pay estimates were obtained for the programs to install either a water network or a sewer network. These programs appeared to have the greatest impacts on households' daily routines. In addition, the scope of these impacts seemed similar. It is therefore reassuring that the willingness-to-pay estimates for these programs are similar.

The relationship between willingness to pay for the water connection and reliability programs also seems consistent with intuition. A household that has unreliable water at least has water for some fraction of a day. As shown in the impact responses, such a household rearranges its use of water, but unreliable water does not have the impact of no connection at all. Because improved reliability obtains only a portion of the services of a water connection, willingness to pay for reliability should reflect only a portion of the value of a water connection. Table 2 shows this expectation to be borne out in the data; willingness to pay for a water connection is larger than willingness to pay for reliable water.

A similar expectation holds for the relationship between willingness to pay for a wastewater connection and willingness to pay for wastewater maintenance. Respondents reported similar impacts from sewer flooding and vault overflows. Thus, wastewater network maintenance aimed at eliminating sewer flooding and sewer connection projects will have similar impacts with respect to the neighborhood environment. A wastewater connection, however, has the added benefit of disposing of household wastewater. Because the services of a wastewater connection program dominate the services of a wastewater maintenance program, willingness to pay for a connection should be larger than willingness to pay for maintenance. Again, willingness to pay estimates in Table 2 are consistent with this hypothesis.

It is also notable that the mean willingness-to-pay values lie within the distribution-free upper and lower bounds. The same is not true of the estimated normal means. The estimated normal means lie below the lower bounds for both the reliable water and wastewater maintenance programs. This result demonstrates the error of using the normal parameters directly without accounting for the economic structure of willingness to pay.

Almost all of the water and wastewater connection respondents and more than 80% of those responding to the two other programs had clear reasons for their choices. Reasons for respondents' referenda choices were similar to their descriptions of program impacts. The exception was that financial reasons were stated more often than in the description of impacts on daily life. The financial aspects of the program were revealed after respondents had described the expected impacts and just before respondents voted on the program. Thus, at the point of voting, respondents were well aware of the financial aspects. In addition, if respondents were making a decision that was truly based on their income constraints, they would have to trade off the beneficial impacts versus the negative on household budgets. Thus, it is consistent with economic decision making that more than 90% of respondents gave reasons that included reference to program cost, program payments, or the household budget. Other common reasons for voting the way they did included impacts on household chores and, for the wastewater connection and maintenance programs, impacts on the neighborhood environment.

BENEFIT-COST RESULTS

Equations 7 through 10 provided the theoretical structure of the benefit-cost analysis. These equations describe benefits and costs in per capita terms. Essential inputs into the analysis were (a) the willingness-to-pay estimates reported in the previous section, (b) investment costs as reported in agency documents (Hanrahan et al. 1993; Hoehn and Krieger 1997), and (c) projections of the number of households affected by each project.

Investment costs included capital expenditures by both the GOE and donor agencies (Hanrahan et al. 1993). The cost of improvements in water services included investments in treatment, storage, and improved conveyance. The costs of the wastewater improvements included conveyance and treatment. Treatment costs were allocated to new wastewater connections based on the share of households receiving new connections relative to the total number of households connected by the conveyance network to the treatment facilities. The present value of total water investment costs was $526 million. The present value of total wastewater costs was $837 million.[sup6]

Two key considerations figured into the calculation of the number of households affected by the investments. First, population projections for each of the areas affected by the projects were calculated using the extensive socioeconomic data compiled for planning the water and wastewater investments (AMBRIC 1991). Second, the portion of households served by an investment declined as the investment depreciated over time. The depreciation schedule was based on the straight-line method and a 40-year useful life.

The affected population estimated for a given year in the life of the investment was the population of all areas served by a project in that year multiplied by the depreciated fraction of the investment remaining in that year. Thus, in the first year of a completed project, 100% of the population was serviced by an investment, whereas in year 40, only one fortieth of the population was served. Projections showed that, in 1995, the water investments affected approximately 1 million households, whereas the wastewater investments affected about 1.2 million households.[sup7]

Table 3 reports benefits per household, costs per household, and benefit-cost ratios. Benefits per household were calculated using Equations 7 and 9, depending on whether the benefit measure pertained to the jth program or an aggregation of programs. The listed cost measures were calculated using Equations 8 and 10. Benefit-cost ratios were calculated for three different discount rates. The different rates provide some insight into the sensitivity of the ratios to a discount rate assumption. Ten percent is a rate commonly used by international agencies, whereas 7% is commonly applied by the U.S. government to domestic policies; 3% reflects a representative real return on low-risk financial investments.

The benefit-cost ratios show that the aggregate water and wastewater investments have a positive benefit-cost ratio at each of the three discount rates. The net economic value, benefit minus cost, of the water and wastewater investments was $1.2 per household. Water investments contributed relatively more to the aggregate economic value than did the wastewater investments. On average, the wastewater investments had smaller per household benefits and greater per household costs than did the water investments. Nevertheless, both water and wastewater contributed positive economic value to the overall investments. The average net economic value of water investments was $2.2 per household, whereas the net economic value for wastewater was $0.6.

Differences between net economic values are more pronounced at the level of individual programs. The net economic value of improved reliability was $0.5 per household, whereas the net value of water connections was $5.4 per household. In the present evaluation scenario, 70% of the water produced by the investments was allocated to meeting system objectives for reliable water service. In this scenario, 30% remained for water connections. Water connections had larger benefits and larger net economic values than reliability. Because the water produced by the investments could be used for either water connections or reliability, the benefit-cost results underscore the trade-off faced by water managers. Project goals made reliability the primary target, but the net economic values suggest that allocating more water to water connections would increase the net economic value of the overall water investments. Extending service to additional households, however, implies some sacrifice in water service reliability for households with existing connections.

The differences in benefit-cost results for wastewater programs are more pronounced than those for water programs. The net economic value of maintenance to update the existing system and eliminate sewer overflows was $2.0 per household. The net value of new connections depends on the allocation of treatment costs. The net value of new connections is -$4.5 per households if new connections are charged a full share of treatment costs. However, wastewater treatment provides no direct benefit to newly connected households. Treatment reduces the externality of wastewater disposal on downstream users. Reduction of the externality creates benefits that are not captured in the benefits of a household connection. Thus, it may be more economically meaningful to exclude treatment costs from an assessment of the benefits and costs of a sewer connection. If so, the net economic value of a connection considers only the cost of conveyance and excludes the cost of treatment. This net value is $1.3 per household.

The water and wastewater investments described in this report were supported by the general funds of the Egyptian and U.S. governments. However, the willingness-to-pay results and cost estimates can be used to determine the potential financial viability of such projects. The estimated normal parameters describe the distribution of willingness to pay among households. For a given tariff, the cumulative normal describes the portion of households with a willingness to pay higher than the tariff. This is the estimated portion of households that would voluntarily sign up and pay for the service at the given cost. The revenue forecast at a given cost is therefore the total number of households affected times the portion willing to sign up times the given tariff. Forecast revenue can be compared to costs to determine financial viability. A program is independently viable if the ratio of forecast revenues to costs is equal to or greater than 1.

Figure 1 presents the financial viability potential for the water program with costs discounted at a 10% discount rate. The horizontal axis describes the size of alternative tariffs in dollars per month per household.[sup8] The narrow beaded line is the cumulative normal plotted for different tariffs. It plots the voluntary sign-up and connection rate at each tariff. For instance, a $2.0 per month tariff results in a connection rate of 90%, whereas a $12 per month tariff results in a connection rate of a little more than 10%. The bold line plots the forecast total revenue to total cost ratio. With tariffs less than about $3 per month, the connection rate is high, but the tariff is too low to collect sufficient revenue to cover costs. With tariffs greater than about $11 per month, the connection rate is too low for revenue to cover costs. Tariffs between $3 and $11 per month result in connection rates associated with a revenues-to-cost ratio greater than 1. Notably, a fixed tariff that covers costs always excludes a portion of households from the service.

Similar financial viability analyses were carried out for reliability, wastewater maintenance, and wastewater connections. The wastewater maintenance analysis resulted in break-even tariffs ranging from about 30?? to $5.5 per month, with sign-up rates ranging from 70% to 10%, respectively. Sign-up rates for both the water reliability and wastewater connections were insufficient for revenues to cover costs at any level of the tariff.

Figure 2 illustrates the financial viability of a combined water and sewer connection tariff. At a 10% discount rate, the sign-up rates for a combined program of water and sewer connections are sufficient to cover cost for tariffs between $12 and $15.5 per month. Connection rates at this charge range from about 60% to 40%, respectively. Figure 2 also shows that the financial viability analysis is sensitive to the selected discount rate. A lower discount rate reduces the present value of costs. At a 7% rate, the combined water and wastewater program is viable for a tariff as low as $4 per month per household. The sign-up rate is approximately 80%.

DISCUSSION

The Cairo water and wastewater investments generated economic benefits in excess of costs even at relatively high discount rates. The infrastructure projects involved billions of dollars in investment and millions of beneficiaries. The results underscore some of the choices and trade-offs involved with investments of this magnitude and complexity.

Urban water and wastewater investments involved trade-offs across services. Managers have a choice as to the allocation of water between different services. The selected allocation has a direct impact on households and net economic values. In Cairo, the net economic values for new water connections were greater than the economic values associated with improving reliability for existing connections. It would be possible to increase the economic benefits of water production by allocating more water to connections and reducing the reliability of water supply to existing users. Of course, the economics of such a reallocation depend on the size of the reallocatio, its impact on reliability, and other initial conditions. In another setting, the relative economic values may be reversed. The exact relationship depends on the economic and supply characteristics of a given city.

A second issue that arose was correct allocation of conveyance and treatment costs for the wastewater connection program. Without the wastewater program, externalities occurred in both the conveyance and discharge of wastewater. Local neighborhoods experienced the conveyance externalities in the form of vault overflows. Downstream parties suffered discharge externalities once the wastewater was discharged on land or in a canal With the wastewater program in place, local households obtained conveyance benefits, whereas downstream water users gained treatment benefits. The correct benefit-cost questions are whether conveyance costs are offset by connection benefits and whether treatment costs are offset by downstream benefits.

In the Cairo situation, we did find that wastewater conveyance benefits offset conveyance costs, but it was not possible to evaluate downstream benefits due to the absence of adequate technical and monitoring data. We are left with partial knowledge. Local benefits were large enough to offset investments in local conveyance, and the overall net value of wastewater investments was positive. Future research is needed to consider the economic contribution of wastewater treatment alternatives.

A third issue was the question of cost recovery through fixed tariffs. Willingness to pay for both the water connection and wastewater maintenance programs was sufficient to generate revenues in excess of costs. A combined water and wastewater connection program would also generate positive net revenues. The issue with a fixed tariff is that willingness to pay for some fraction of households is less than costs. Depending on the program, 60% of potentially affected households might be unwilling to sign up for a program.

Cost recovery methods other than fixed tariffs need to be examined. For example, simple quantity-based pricing would spread program costs across households in a manner more in line with willingness to pay. With quantity-based pricing, households with low willingness to pay could purchase smaller amounts of a service. Households with higher willingness to pay would purchase larger quantities of the service. The drawback of pricing is that it requires metering, and metering is costly. Nevertheless, the gains in efficiency and fairness due to pricing could more than justify the costs.

NOTES

[sup1]. The 2 l/2-year economic evaluation project was funded by the U.S. Agency for International Development for a total cost of $550,000.

[sup2]. Average household income in Cairo in 1995 was about U.S.$260 per month. The average income of households without water or without wastewater connections may have been as low as U.S.$106 per month. Egyptian law prohibited noncensus researchers from asking income and expenditure questions in either qualitative or quantitative survey research.

[sup3]. RADA Research, Inc., of Heliopolis, Egypt.

[sup4]. Cost thresholds were stated in Egyptian pounds in the original questionnaires. The 1995 exchange rate was 3.38 pounds to the U.S. dollar. The pound amounts were given in numbers that were multiples of 5, such as 0.20 LE and 1.75 LE. Weekly amounts were used because weeks tended to be an interval of time that most households understood for budgeting purposes.

[sup5]. The economic structure of the data was assessed by testing whether the inverse relationship between project cost and acceptance was statistically significant. The null hypothesis was that the negative correlation was not statistically significant. If the null hypothesis was true, the percentages of accept responses should be statistically equal across all thresholds for a scenario. A separate test was conducted for the data from each scenario. The null hypotheses of no relationship between acceptance rate and program cost were rejected in each case at at least the 99% level.

[sup6]. The stream of investment costs began in 1977 and ended in 1997. The present value procedure used a discount rate of 10%. The present value was computed for the base year of 1995 in dollars at the 1995 price level.

[sup7]. The year 1995 is representative of the impact in the early years of most of the projects. It is not the first year of any of the projects.

[sup8]. A tariff is a fixed charge per month that does not vary with the quantity of service used. Tariffs are common in the developing world and were used by the Cairo water and wastewater agencies. However, the Cairo tariffs were set well below the break-even point (Hanrahan et al. 1993).

TABLE 1: Program Impacts Described by Respondents

Legend for chart:
 
A0=Water Connection
A1=Wastewater Connection
A2=Reliable Water
A3=Sewer Flooding
 Program Type
Impact on Daily Life A0 A1 A2 A3
One or more impacts mentioned (%) 100 100 79 78
Mean number of impacts 5.1 4.2 2.5 2.8
Financial (%) 18 3 21 16
Household chores and activities (%) 97 58 66 51
Time savings (%) 70 4 12 2
Neighborhood environment (%) 39 89 25 68

TABLE 2: Parameter Estimates and Willingness to Pay (WTP) per Month per Household by Program (1995 dollars)

Legend for chart:
 
A0=Project
A1=Water connection
A2=Wastewater connection
A3=Reliable water
A4=Wastewater maintenance
A5=Mean
A6=7.25 (0.22)
A7=5.92 (0.30)
A8=2.22 (0.14)
A9=1.43 (0.12)
B0=Standard Deviation
B1=Mean WTP[supa]
B2=Normal Parameters[supa]
 B2
 Bounds on WTP[supa]
A0 A5 B0 B1 Lower Upper
A1 A6 4.44 (0.28) 7.77 (0.14) 6.15 (0.15) 8.70 (0.21)
A2 A7 6.36 (0.55) 7.57 (0.20) 5.33 (0.24) 8.70 (0.36)
A3 A8 3.17 (0.22) 3.20 (0.10) 2.28 (0.12) 3.28 (0.15)
A4 A9 2.40 (0.19) 2.22 (0.07) 1.51 (0.12) 2.54 (0.12)
 
[supa]. Standard errors are given in
parentheses

TABLE 3: Economic Evaluation of Water and Wastewater Investments

Legend for chart:
 
A0=Project and Components
A1=Water and wastewater
A2=Water: Baseline
A3=Reliability
A4=Connections
A5=Wastewater
A6=Eliminate flooding
A7=Connection with treatment
A8=Connection only
A9=WTP per Household ($/month)
B0=Cost per Household ($/month)
B1=Net Economic Value ($/month)
B2=Benefit-Cost Ratios by Discount Rate
 B2
A0 A9 B0 B1  10% 7% 3%
 
A1 3.9[supa] 2.7 1.2 1.4 1.9 2.9
A2 4.7[supa] 2.5 2.2 1.9 2.3 3.5
 A3 3.2 2.7 0.5 1.2 1.5 2.2
 A4 7.7 2.3 5.4 3.3 4.2 6.4
A5 3.4[supa] 2.8 0.6 1.2  1.5 2.6
 A6 2.2 0.2 2.0 11.0 14.5 37.9
 A7 7.6 12.1 -4.5 0.6 0.9 1.5
 A8 7.6 6.3 1.3 1.2 1.7 3.2
 
NOTE: WTP = willingness to pay.
[supa]. Computed as a weighted average of the WTP per specific
service. Weights were the fractions of the total beneficiary
population that received the specific services (e.g., water
connection, reliable service, reduced sewer flooding, and sewer
connections).

GRAPH: Figure 1: Water Connection Program Financial Viability, 10% Discount Rate

GRAPH: Figure 2: Combined Water and Wastewater Connection Program

REFERENCES

Alberini, Anna. 1995. Optimal designs for discrete choice contingent valuation surveys: Single-bounded, double-bounded, and bivariate models. Journal of Environmental Economics and Management 28:287-306.

Altaf, Mir Anjum. 1994a. The economics of household response to inadequate water supplies. Third Worm Planning Review 16:14-53.

-----. 1994b. Household demand for improved water and sanitation in a large secondary city: Findings from a study in Gujranwala, Pakistan. Habitat International 18:45-55.

Altaf, Mir Anjum, and Jeffrey A. Hughes. 1994. Measuring the demand for improved urban sanitation services: Results of a contingent valuation study in Ouagadougou, Birkina Faso. Urban Studies 31:1763-76.

Altaf, Mir Anjum, Dale Whittington, Haroon Jamal, and V. Kerry Smith. 1993. Rethinking rural water supply policy in the Punjab, Pakistan. Water Resources Research 29:1943-54.

American British Consultants (AMBRIC). 1981. Investigations of sewage flooding incidents: Final report. Cairo, Egypt: Ministry of Housing and Reconstruction, Arab Republic of Egypt.

-----. 1987. Unsewered areas demonstration project: Final report. Cairo, Egypt: Ministry of Housing and Reconstruction, Arab Republic of Egypt.

-----. 1991. System load review, Volume I--Report. Cairo, Egypt: Ministry of Housing and Reconstruction, Arab Republic of Egypt

Briscoe, John. 1992. Poverty and water supply: How to move forward. Finance and Development 30: 16-19.

Briscoe, John, Paulo Furtado de Castro, Charles Griffin, James North, and Orjan Olsen. 1990. Toward equitable and sustainable rural water supplies: A contingent valuation study in Brazil. The Worm Bank Economic Review 4:115-34.

Cameron, Trudy A., and Michele D. James. 1987. Efficient estimation methods for "closed-ended" contingent valuation surveys. Review of Economics and Statistics 69:269-76.

Central Agency for Public Mobilization and Statistics (CAPMAS). 1986a. Cairo--Census of building and housing characteristic, 1986. Cairo, Egypt: Central Agency for Public Mobilization and Statistics, Arab Republic of Egypt.

-----. 1986b. Giza--Census of building and housing characteristic, 1986. Cairo, Egypt Central Agency for Public Mobilization and Statistics, Arab Republic of Egypt.

-----. 1986c. Kalyoubia--Census of building and housing characteristic, 1986. Cairo, Egypt: Central Agency for Public Mobilization and Statistics, Arab Republic of Egypt.

Cosslett, Stephen R. 1983. Distribution-free maximum likelihood estimator of the binary choice model. Econometrica 41:765-82.

Environmental Quality International (EQI). 1988. Evaluation of Cairo sewerage I rehabilitation. Cairo, Egypt: Environmental Quality International.

ES-Parsons. 1979. Final report: Waterworks master plan executive summary. Cairo, Egypt: Greater Cairo Waterworks Development Program, Arab Republic of Egypt.

Haab, Timothy C., and Kenneth E. McConnell. 1998. Referendum models and economic values: Theoretical, intuitive, and practical bounds on willingness to pay. Land Economics 74:216-29.

Hanemann, Michael, and Barbara Kanninen. Forthcoming. The statistical analysis of discrete-response CV data. In Valuing environmental preferences: Theory and practice of the contingent valuation method in the U.S., E.C., and developing countries, edited by Ian Bate-man and Ken Willis. Oxford, UK: Oxford University Press.

Hanrahan, Michael, Samir Makary, James Walker, Suzanne Goode, Rasha Abdul Hakim, Ingi Lofti, and Wasseem Mina. 1993. Promoting development and U.S. commercial interests: The A.I.D. capital projects program in Egypt. Appendix VIII economic analysis. Cairo, Egypt: U.S. Agency for International Development.

Hoehn, John P., and Douglas J. Krieger. 1996. Cairo water and wastewater economic benefit assessment: Volume I. Cairo, Egypt: United States Agency for International Development.

-----. 1997. Benefits.wk1: Aggregate benefits of water and wastewater investments in 1995 U.S. dollars. East Lansing: Michigan State University.

Hoehn, John P., and Alan Randall. 1987. A satisfactory benefit cost indicator from contingent valuation. Journal of Environmental Economics and Management 14:226-47.

John Taylor and Sons and Binnie and Partners (Taylor-Binnie). 1977. Cairo wastewater project Special Report No. 1: Improvement of plant operation, wastewater system repair, and rehabilitation. Cairo, Egypt: Ministry of Housing and Reconstruction, Arab Republic of Egypt.

Kristrom, Bengt. 1990. A non-parametric approach to the estimation of welfare measures in discrete response valuation studies. Land Economics 66:135-39.

Maddala, G. S. 1977. Econometrics. New York: McGraw-Hill.

McPhail, Alexander A. 1993. The 5-percent rule for improved water service-can households afford more. World Development 21:963-73.

-----. 1994. Why don't households connect to the piped water system? Observations from Tunis, Tunisia. Land Economics 70:189-96.

Mitchell, Robert Cameron, and Richarc T. Carson. 1989. Using surveys to value public goods: The contingent valuation method. Washington, DC: Resources for the Future.

Nadim, A., N. Nadim, S. Mahanna, and J. Nixon. 1980. Living without water. Cairo Papers in Social Science. Cairo, Egypt: The American University in Cairo.

Oldham, L., H. El Hadidi, and H. Tamaa. 1987. Informal communities in Cairo: The basis of a typology. Cairo Papers in Social Science. Cairo, Egypt: The American University of Cairo.

Pudney, Stephen. 1989. Modelling individual choice the econometrics of corners, kinks, and holes. Oxford, UK: Blackwell.

Singh, Bhanwar, Radhika Ramasubban, Ramesh Bhatia, John Briscoe, Charles Griffin, and C. Kim. 1993. Rural water supply in Kerala, India--How to emerge from a low-level equilibrium trap. Water Resources Research 29:1931-42.

Spradley, James P. 1979. The ethnographic interview. New York: Holt, Rinehart and Winston.

Sudman, Seymour, Norman M. Bradburn, and Norbert Schwarz. 1996. Thinking about answers: The application of cognitive processes to survey methodology. San Francisco: Jossey-Bass.

Tecke, B., L. Oldham, and Frank Shorter. 1994. A place to live: Families and child health in a Cairo neighborhood. Cairo, Egypt: The American University of Cairo Press.

Whittington, Dale, John Briscoe, Xinming Mu, and William Baron. 1990. Estimating the willingness to pay for water services in developing countries: A case study of the use of contingent valuation surveys in Southern Haiti. Economic Development and Cultural Change 38:293-311.

Whittington, Dale, Donald T. Lauria, and Xinming Mu. 1989. Paying for urban services: A study of water vending and willingness to pay for water in Onitsha, Nigeria. Washington, DC: The World Bank.

-----. 1991. A study of water vending and willingness to pay for water in Onitsha, Nigeria. Worm Development 19:179-98.

Whittington, Dale, Apia Okorafor, Augustine Okore, and Alexander McPhail. 1990. Strategy for cost recovering in the rural water sector: A case study of Nsukka District, Anambra State, Nigeria. Water Resources Research 26:1899-913.

Whittington, Dale, V. Kerry Smith, Apia Okorafor, Augustine Okore, Jin Long Liu, and Alexander McPhail. 1992. Giving respondents time to think in contingent valuation studies: A developing country application. Journal of Environmental Economics and Management 22:205-25.

World Bank Water Demand Research Team. 1993. The demand for water in rural areas: Deter-minants and policy implications. World Bank Research Observer 8:47-70.