The environment, low income and crime: Valuation of a lake in Cape Town, South Africa

Transcription

1 1 WORK-IN-PROGRESS NOT FOR DISTRIBUTION The environment, low income and crime: Valuation of a lake in Cape Town, South Africa Abstract Environmental values are often public goods, and depleted by pollution externalities. Sometimes they are provided in a setting of low income, crime and urban congestion. In developing countries such factors make it more difficult to observe the welfare effects of the environment, and even more difficult to compete with other pressing needs inside strapped public budgets. This study aims to show how public environmental goods may be beneficial to the local society. The focus is on Zeekoevlei Lake, situated on the Cape Flats in Cape Town, South Africa. The lake is an important regional recreational venue in the middle of the metropolitan area, but it is highly polluted and surrounded by areas with a high crime rate and mainly low-income households. A Contingent Valuation Method (CVM) was used to determine the environmental value of Zeekoevlei. The study focuses on its value for the local, low-income urban households, both for non-use values and for recreation. The impact of security problems on these values, due to crime, is also addressed. Communities close to the lake do feel that they benefit from the lake. Calculation of the mean willingness to pay for improved water quality was established to be R26.74 per month per household. The mean willingness to pay for improved recreational facilities is an entrance fee of R7.91 per person per visit. The willingness to pay varies with socio-economic variables such as household size, income and education. Different attitudes to the crime problems yield conflicting effects on the environmental valuation (often not significant ones). It was established that the highincome group is actually less willing to pay for the environment. JEL Classification: Q25, Q26, C24, C25 Keywords: Water pollution, urban setting, low income, crime, contingent valuation method 1. Introduction Water resources are a vital part of human activity: water is used for consumption and for production activities. Water bodies are also used for recreation and the enjoyment of their aesthetic beauty. Water management therefore does not only involve looking after the volume of the resource, but also its quality. In the developing world, urban settlements have experienced an influx of people, especially from rural areas in search of employment. This puts pressure on local governments to provide more and better public services. Less emphasis is then usually placed on improving the water quality of lakes and rivers utilised for recreational purposes. On the other hand, areas set aside for environmental purposes are also under pressure from urban development. Land is scarce and all open areas have alternative uses. Hence their existence must be justified based on the welfare added for the people living in the area. Lakes and rivers face negative externalities from economic activities in the area, and often pollution from sewage and from human use of surrounding areas. This study focuses on the valuation of water quality for environmental use in a low-income urban setting. The empirical focus is Zeekoevlei, the largest freshwater lake (256 hectares) on the Cape Flats, Cape Town, South Africa. It is surrounded by urban settlements and the area primarily consists of low- and middle-income households. There is however an enclave on a peninsula of the lake that is dominated by more well-off households. Some residents live on the edge of the lake while many are within walking distance. The lake is used for recreational purposes,

2 and sailing and fishing are some of the more frequent activities. Zeekoevlei is heavily polluted with high levels of phosphorous (World Lakes Database, n.d.). The pollution has not only impacted on the recreational use of the lake but it also has implications for the preservation of natural wetlands on the Cape Flats. This part of the Cape Flats is well known for a high crime rate, and the use of the lake for recreational purposes is most probably influenced by this crime and pollution. The areas close to the lake are more remote and people visiting the lake are more exposed to crime. Use of the lake as a recreational area is therefore less attractive, and the lake area is less known among the local households. This contributes to making valuation more difficult. The residents attitude to the environmental value of the lake will most likely be influenced by crime that makes it difficult to go there. The main aim of this study is to establish the value of the lake to the surrounding communities. The methodology used is the contingent valuation method (CVM). The CVM is a common method used in environmental valuation, and entails carrying out a survey. The primary purpose of the questionnaire was to obtain responses on the willingness to pay for improvements in the environmental and recreational quality of the lake. We investigated the factors that influence the willingness to pay of households, the focus being on both the recreational and the non-use values of the lake. Determining the economic value of the lake is especially relevant since cleaning it up will be costly, and generally the primary focus of local governments is the provision of basic goods and services. Special attention is given to the influence of crime on the valuation of the lake. It is possible that households see crime and pollution as a bundle of goods that jointly harms their use of the lake. If so, better policing and cleaning up the area are factors that can be combined to increase use values it is not easy to split these two factors. Since there is no ready market where these values can be established, the direct contingent valuation method is applied. The paper commences with a theoretical exposition of the case for government involvement in the provision of environmental goods. It explains why environmental goods are classified as public goods and the difficulties associated with obtaining values for these goods. The next section briefly elaborates on the budgetary trade-offs faced by the local government of Cape Town, given the pressing needs within its jurisdiction, and the implications thereof for environmental goods. Urban development patterns in the Western Cape are one of the underlying reasons for the increased demand for local government services. This is followed by a background section on Zeekoevlei and the communities in the surrounding suburbs. Section 4 discusses the design of the survey, the data-collection process and methodology. Results are presented and discussed in Section 5. The final section presents our conclusions and highlights certain policy implications. 2. Theoretical exposition on the provision of environmental goods 2.1. Environmental goods as public goods Environmental goods are usually classified as a subset of public goods. The latter exists due to two properties: the first one is non-excludability in consumption (which implies that no individual can be excluded from the benefit of the environmental good), and the second stems from the non-rivalrousness of consumption (which implies that when one person consumes the good there is not less left for the next person to consume). Environmental goods often have both these properties, and provision of the goods does not usually take place in markets. In many cases no markets exist for environmental goods and therefore no prices and costs are available. Environmental goods can have both direct and indirect benefits. The preservation of a wetland, for example, can have direct benefits, which include use values such as bird watching and non-use or existence values. Indirect benefits can emanate from the pollution costs that will be saved if the wetland is preserved. Most environmental benefits from the use of wetlands are non-consumptive, i.e. they do not exhaust the water body. 2

3 Environmental valuation and some empirical applications Since no markets exist for many environmental services, the standard method of consumer surplus cannot be used to obtain market information via the price system. Direct valuation involves obtaining the values from individuals (stated preference methods), while indirect valuation attempts to establish these values by observing the behaviour of individuals in related markets. In the case of stated preference methods Grafton, Adamowicz, Dupont, Nelson, Hill and Renzetti (2004, p. 229) argue that this entails conversations with individuals through structured surveys, i.e. asking the respondents what they would be willing to pay for changes in environmental quality. The Contingent Valuation Method (CVM) is the most common stated preference method. It involves creating a hypothetical market in which the respondents are offered a good and asked to indicate whether they are willing to purchase the good. O Doherty (1993, p. 1) states that where a policy change will cause an environmental effect, those who benefit (or suffer a disbenefit) are encouraged to reveal their willingness to pay for the policy change (or to avoid it). There exists a vast literature on the CVM (which will not be surveyed here): see Mitchell and Carson (1989), Arrow, Solow, Pornoy, Learner, Radner and Schuman (1993) and Hanemann (1994) for classical articles on the CVM. The CVM has been utilised in a number of studies in developing countries. Navrud and Mungatana (1994) used the method to study wildlife viewing. Kohlin (2001) applied the method to establish the value of a social forestry project in India. Another study carried out in developing countries is that of Carlsson, Kohlin and Mekonnen (2004). This study focused on community plantations in Ethiopia and attempted to draw both methodological and policy conclusions using the CVM Budgetary trade-offs for local governments implications for environmental goods Migration to urban areas is a common trend in the developing world, and South Africa is no exception. People migrate to Cape Town primarily from the rural parts of the Western Cape, Eastern Cape and Northern Cape provinces. This places a burden on the local government as it increases the demand for basic services such as water and sanitation. Over the past few decades the local government of the City of Cape Town (CCT) has experienced an influx of people from rural areas and other parts of the country. For the period 1996 to 2006 the CCT experienced an annual average growth rate of 3% in population (CCT, 2007b). This has lead to higher population densities, especially in the lower-income areas, which has in turn lead to overcrowding and therefore health problems. According to the CCT (2007b), from 1993 to 2006 the number of households in informal settlements grew from to Urban population growth places increased pressure on local governments to provide basic social and economic services, increasing their share of the budget. Less than 1% of the total budget of CCT is spent on environmental resource management as compared to almost 50% on utility services (see Table 1). 3. An overview of Zeekoevlei Lake and the surrounding suburbs The Zeekoevlei Lake is situated on the Cape Flats amidst an urban residential area. It is surrounded primarily by the (traditionally called) coloured suburbs of Grassy Park, Lotus River, Pelican Park and Zeekoevlei. A section of Zeekoevlei is also referred to as the peninsula as it is isolated (in the sense that residential plots are located on a piece of land surrounded by the lake). The residents living on the peninsula are relatively more affluent (and racially mixed) than in the other suburbs.

4 The suburbs surrounding Zeekoevlei Lake There are four suburbs within the immediate vicinity of the lake, namely Grassy Park, Lotus River, Pelican Park and Zeekoevlei. Table 2 provides a brief overview of the socio-economic information of these suburbs. The data reflect that coloureds are the main racial group in the suburbs. Pelican Park and Zeekoevlei, however, seem to be more racially mixed. If one compares the level of education across suburbs, Zeekoevlei is clearly different to the others, with 57% of the suburb s population having completed secondary and higher education. On the other hand the level of education in Lotus River is especially low (compared to the other suburbs), with only 28% having obtained secondary and higher education. Zeekoevlei is the more affluent among the suburbs. This is confirmed by this study s survey data. Table 3 includes an indication of the income groups of the suburbs (the table itself shows the distribution of questionnaires completed in the different suburbs, by income group). The majority of the lower income groups (1 to 3, amounting to more than 60% of all households) live in Grassy Park, Lotus River and Pelican Park. In Zeekoevlei more than 50% of the respondents earn above R per month. 1 For households living on the peninsula, about 40% earn above R per month. This illustrates the economic differences between the suburbs Factors impacting on the value of Zeekoevlei Lake Zeekoevlei Lake is an area that is open to the public and used for recreational purposes. It is particularly an area for swimming, although water pollution may not always allow such direct contact. It therefore seems more reasonable to study the values of boating, fishing and bird watching, as well as picnicking or simply relaxing along the shores of the lake. The lake also has value in terms of the nature of the area. A user assessment survey conducted by Quick and Johansson (1992) supports this by indicating these activities as being the most popular uses of the lake. Zeekoevlei Lake is fed by two rivers, the Big Lotus River and the Little Lotus River. It has become increasingly polluted as the areas surrounding the lake becomes more urbanised and industrialised. Southern Waters (2000) has identified the causes of the high nutrient loading (phosphorous) as emanating primarily from the following sources: agricultural runoff into the Big Lotus River and the Little Lotus River (from farming and horticultural activities), seepages from the nearby wastewater treatment plant (adjacent to the lake), and stock-based sediments in the lake itself. This stock pollution has accumulated over the years and at present approximately 25% of the pollution is generated by internal loading from the volume of organic-rich sediments. These sources constitute a negative externality to the lake and impair its functioning. The value of the lake for human use is also influenced by crime. The surrounding residential areas experience high levels of criminal activity. 2 Remote areas (such as the lake) are especially susceptible to these criminal activities. The lake area is subsequently often avoided due to the higher risk of exposure to crime. In an environment and tourism survey completed by the CCT it was found that crime and fear for personal safety are the most common reasons for stopping or limiting visits to protected natural areas, especially amongst the people from the Cape Flats (CCT, 2001). Pollution and crime therefore work in tandem to constitute the negative external effects on the use of Zeekoevlei Lake. In principle, crime will only affect the use value of the lake. The non-use values should not be influenced as long as the crime only affects the potential use of the lake and does not damage it. However, the estimated non-use values of the lake may be affected by the use value because the 1 This amounts to just a little over $1 300 per month. (One South African Rand (R) equaled 0.13 USD at the time of writing.) 2 In its 2006 Sustainability Report, the CCT (2007b) states that drug-related criminal activity has increased from 2004/05 to 2005/06 and the number of murders has increased from 54.6 to 57.3 (per population).

5 respondents may find it difficult to accurately state their interest in the lake without including the aspects of crime and access. The total effect of pollution, crime and the lack of recreational facilities together make the lake less attractive for the people in the area. The bad smell is linked partly to the pollution of the lake, and partly to a negative externality of the sewage treatment plant close to the lake. These factors are complements. It is rather difficult to ensure that the respondents reveal these different aspects separately, in a way that can be included in a regression analysis. This study therefore does not endeavour to separate these effects. Yet it is desirable to, in some way, gain as much information as possible on the importance of crime for environmental valuation. The negative externalities work more like a bundle of effects that influences the utilisation of the lake in a negative way. A polluted lake gives a low recreational value even if it is well policed while a clean lake gives a low recreational value if the area is ridden by crime. Hence it is not possible to completely separate the two effects; they are complementary negative externalities. Most people in South Africa are concerned about crime, and the residents of the Cape Flats are no exception. Certain questions in the questionnaire for this study focused specifically on crime and adequate policing. The aim was to ascertain the impact thereof on respondents perception of the relevance of crime, as well as the implications thereof on the environmental values. Results that emerged from the responses to these questions, pertaining to factors impacting on the value of Zeekoevlei Lake, are described in Table 4. An investigation into the correlation between the respondents visiting the lake and crime in the area revealed that crime is indeed a very important factor. It influences the respondents decision on whether to visit the lake or not. On the question of whether they visit or use the lake, 53% of the 384 respondents (203 respondents) indicated that they do not visit or use the lake. Twenty five per cent (of the 203 respondents, i.e. 50) also indicated that they do not value the lake as an environmental good, particularly because of the pollution and crime at the lake. Table 4 shows how both pollution and crime make the lake less attractive as an environmental asset. Seventy per cent of the 50 respondents who indicated that crime at the lake is an extremely important factor influencing why they do not value the lake, also indicated that pollution was another extremely important reason for not valuing the lake. Most households who participated in the survey indicated that effective policing at the lake is important the majority find it even extremely important. Table 5 shows that more than 90% of the households indicated that effective policing will be crucial for them to change their perceptions about visiting the lake Determining the value of Zeekoevlei Lake The complementary factors influencing the value of the lake (see Section 3.1) make it difficult to estimate the environmental value of the lake. An attempt was nonetheless made to establish the impact of crime on people s valuation of the lake. Access to an area with human activity and traffic close by are all outside factors that can overshadow the pollution factor. Further, in a study of the Zandvlei Lake (another natural wetlands on the Cape Flats, close to Zeekoevlei Lake), it was determined that the value of the wetlands was influenced by its potential effect, namely acting as a buffer to surrounding crime-infected suburbs (Van Zyl and Leiman, 2002). Even without environmental externalities such as pollution the value of the lake is influenced by the safety of using it. The opposite argument also applies: in determining the value of the lake in a situation with security enforced, it must be influenced by the presence of environmental externalities. A polluted lake gives a low recreational value even if it is well policed and a clean lake gives a low recreational value if the area is crime ridden.

6 The recreational area has different qualities, all of which are part of the utility function of the households in the area. It seems reasonable that the demand will differ with income, and that culture will also have an influence. (For example, the typical nature conservationist s attitude may not be shared by all people in the area.) The households distance from the lake is probably an important factor influencing the lake s value to residents. In the suburb of Zeekoevlei (for example, on the peninsula) some households have direct access to the lake. People living on the peninsula constitute a local community that is very different from the rest of the surrounding area. It makes sense to assume that their cultural background and income will lead to a different attitude towards the environmental value of the lake. Pelican Park is within walking distance of the lake, but it is an area where security is a more pressing issue. Lotus River and Grassy Park are further away, but some households do not live far from the lake. From observations and discussions in the focus groups it was established that many households use the nearby Zandvlei Lake for recreation. This can be seen as a substitute for using Zeekoevlei as a recreational value. From these discussions it was apparent that Zandvlei Lake provides better recreational facilities and people generally feel safer to take their families there for an outing. For access to sections of the Zandvlei Lake visitors must pay an entrance fee Methodology, survey design and data collection 4.1. Methodology This study used the CVM to establish the value of Zeekoevlei Lake to its surrounding communities. The CVM is applied especially when no market exists for goods and their value cannot easily be determined. Jennings and Curtis (2002, p. 1) argue that the CVM has been extensively used to assess the use and non-use value of environmental amenities, and continues to be developed as a tool for this purpose. The CVM is a hypothetical method that directly asks the respondents valuation of a particular environmental good. Jacobsson and Dragun (1996, p. 93) state that the development of the CVM has enabled the measurement of non-market values. This is especially the case for option or existence values. The focus is on estimating the willingness to pay for an environmental good. Stenger and Willinger (1998) indicate the usage of the CVM as an appropriate method in estimating the willingness to pay for the preservation of water quality. The method therefore aims to reveal the valuation of the lake in a way that makes it possible to use standard welfare analysis, such as consumer surplus theory. This provides an estimate of the welfare changes experienced by the respondent, since, if provided truthfully, it will provide a measure of compensating surplus. Compensating surplus is a monetary measure of the utility change associated with a change in environmental quantity or quality (Perman, Yue, McGilvray and Common, 1999, p. 385). According to FAO (2000), the objective of the CVM is to ascertain the compensating variation for a particular good, where the latter is used as a measure when the person needs to buy an improvement in environmental quality. The respondent can therefore be asked what his/her willingness to pay (WTP) would be to obtain this good Survey design and data collection Survey design The design of the survey is based on existing literature on empirical studies completed using the CVM, particularly according to Carlsson et al. (2004) and Loomis, Kent, Strange, Fausch and Covich (2000). Two focus group meetings and a small pilot survey were conducted before the main survey. One meeting took place with residents from Grassy Park and another with residents living in the suburb of Zeekoevlei. The meetings were generally informative

7 with respect to how prospective respondents would react to the survey questions. The participants were excluded from the main survey. Freeman (1999, p. 170) identified three important aspects to focus on in the survey instrument: (i) there must be a description of the choice setting within which the respondent will find him/herself; (ii) the choice questions from which environmental values will be inferred (to establish an economic value expressed in monetary terms); and (iii) questions posed to the respondents in order to obtain information on their demographic and socioeconomic background that may influence their utility from the goods. The questionnaire included questions on background information, such as on the social, economic and cultural characteristics of the households, as well as their attitudes towards the use of Zeekoevlei Lake, including their existing use of the lake. There were also questions pertaining to information about the residents of the households, such as household size, income, education, the age of the respondent, as well as questions on the existence value of the lake. As mentioned earlier, some questions pertained to whether crime had any significant impact in the household s decision to visit or use the lake. Questions on the willingness to pay were divided into two sections: one section focused specifically on the use value of the lake (hypothetical questions on the recreational facilities at the lake) and the other section focused on the environmental value (use and non-use) assuming hypothetical improvements in the lake s water quality. The reason for this two-pronged approach was that some households live next to the lake but do not use the recreational facilities at the lake. Their responses on the value of the lake encompassed both use and non-use values since they had already made an explicit choice to buy property next to the lake. Then, for households in suburbs surrounding the lake (but not adjacent to the lake), there would be a possible risk that some value would not be captured, especially if they did not visit the lake. The question format is primarily referendum-type (dichotomous choice questions). Referendum questions basically ask the respondent whether he or she would be willing to pay a specified amount for a particular (environmental) change to occur. A positive response to the questions indicates that the willingness to pay is equal or higher than the amount specified. A refusal to the bid indicates that the specified amount can be taken as the upper limit on the true willingness to pay, for the respondent concerned. An advantage of the referendum type question format is that it allows a more simple revelation of the willingness to pay, and is likened to a market transaction where the consumer can decide to say yes or no to a price. However, researchers still face other methodological problems such as strategic answers. For the environmental quality closed-ended question, individuals were asked whether they would be willing to pay a monthly cleanup fee, which would be added to their monthly municipality bill. The questionnaire used four different monthly fees as possible prices for the change in environmental quality: R10, R20, R30 and R50. 3 These amounts were chosen based on the information gathered from the focus groups and the pilot study. Each respondent was offered one of these amounts. The distribution of the bid amounts was determined beforehand and was not influenced by the research assistant. This payment vehicle was chosen to reduce possible biases arising from people s responses to other types of payment options, such as taxes or donations. Stenger and Willinger (1998, p. 185) used the water bill as their payment vehicle. The closed-ended question was followed up by asking the respondents what their maximum willingness to pay to clean up the lake would be. Using both an open-ended and a closed-ended question allows a comparison of the results between the two methods. Brown et al. (1996, in Stenger and Willinger, 1998) indicate that usually the closed-ended type question overestimates WTP compared to the open-ended question. In the case of improved recreational facilities at the lake, the questions used were of the same format as described above. The questions commenced with a closed-ended question using an entrance fee as payment vehicle (i.e. the respondents were asked whether they 7 3 One South African Rand (R) equaled 0.13 USD at the time of writing.

8 would be willing to pay an entrance fee. The entrance fee used four different amounts as possible entrance fees: R2, R3, R5 and R10. These were indicated as being paid per person per visit. The fees were based on amounts that emanated from discussions in focus groups, as well as comparative fees used in existing national parks and at other lakes. This was followed up with an open-ended question asking what maximum entrance fee they would be willing to pay Data collection Data were collected by students from the University of the Western Cape. Five students were appointed to participate in the data-collection process. All students selected were Economics majors with some of them engaged in completing postgraduate qualifications. Most had completed a course in Public Economics and were familiar with environmental externalities. All students were familiar with the study areas. The research assistants were given training in the method used. They attended a half-day session where the basics of environmental externalities and valuation techniques were discussed, focusing on CVM. Training was also provided on the questionnaire itself and the fieldwork process. Students were advised about the relevance of each of the questions in the questionnaire, especially how to approach the hypothetical scenario for the WTP questions. Students received detailed explanations about the importance of not influencing the responses of respondents, when to use the appropriate photographs that accompanied the questionnaire, and what information to provide to the respondents about the study. The survey area consisted of four suburbs: Grassy Park (GP), Lotus River (LR), Pelican Park (PP) and Zeekoevlei (Z) and those living on the peninsula (P) in Zeekoevlei. Table 6 indicates the proportion of yes responses to the closed-ended question on improvements in water quality. Column 3 in the table specifically indicates a decrease in the proportion of yes responses as the bid amount increases. This decrease is similar to what Carlsson et al. (2004) found in a study carried out to examine the determinants of the willingness to pay for community woodlots. In our case, however, the share of yes responses is relatively close to each other, given the bid amounts, showing that perhaps higher bid amounts should have been included in the survey. 5. Regression analysis and results The total number of questionnaires completed amounted to 384. These consisted of 26 questionnaires to households adjacent to the lake and 358 questionnaires to households within the vicinity of the lake Econometric models The econometric analysis for referendum questions usually entails estimating the probability that a respondent would pay (Loomis et al, 2000). The present study, however, uses an interval regression as the main econometric model. We have also used the logit model and an open-ended question, where the households give a specified amount they are willing to pay. We refer shortly the results from these estimations to confirm our main estimation. For some policy issues the WTP questions can generate zero responses. If this is the case then it is possible that, when applying the logit model, the mean WTP can be negative (Del Saz-Salazar and Garcia-Menendez, p. 2121). It is also possible that a fat tail problem can arise, which gives rise to an unrealistically high WTP (Ready and Hu, 1995). A way of dealing with data with a fat tail distribution is to apply censoring. In this study there is a long right-hand tail distribution in the willingness to pay data and it is censored to fall within a realistic budget share of the household. The regressions discussed in this section focus entirely on assessing the determinants influencing the willingness to pay for an improvement in the water quality of Zeekoevlei, i.e. 8

9 a reduction in the pollution levels. The estimates are thus for both use and non-use values. The variables in the regression analysis are given in Table 7. The regression equation is weighted according to the actual number of households in the different suburbs. 4 Interval regression An interval regression was applied to estimate the WTP for improved recreational facilities. Each household accepts or refuses a specified bid for their willingness to pay for cleaning up the lake. If the bid is accepted we use this as the lower limit and set an upper limit of 3% of the income of the respondent (assuming that this is the maximum realistic amount the respondents would be willing to pay). If the bid is not accepted then the bid is the upper limit while the lower limit is zero. This interval is regressed against the independent regressors. Table 8 tabulates the results for the interval regression on WTP for improved water quality. The mean WTP was calculated to fall between an interval of 0 and the upper limit income share. This gives a mean monthly WTP of R26.74 per household, which is substantially lower than the findings of the logit model (R37.66). The open-ended question yielded R Household size seems to be important. The positive sign indicates that the larger the household the higher the willingness to pay. This can possibly be explained by the hypothesis that households with larger families (especially where children are present) are more likely to spend time at the lake and would therefore be more willing to pay for improvements in environmental quality. Other wetlands within the vicinity of Zeekoevlei (such as Rondevlei, Princess Lake and Zandvlei) are considered as alternative recreational venues for households and used as substitute products in the analysis. The other-lake variable has a positive sign. The coefficient is however not significant. This gives a weak indication that those who visit other lakes do value these recreational spots and may be willing to pay for the cleanup of Zeekoevlei. The policing variable has a negative coefficient but it is not statistically significant. Table 5, however, clearly reflects that policing has an important impact on the value that people attach to the lake. The coefficients of the education variables both have positive signs (as compared to the base group). This is a reasonable outcome as it is often argued that more educated people will care more for the environment. Only edu2, though, has a significant coefficient. The suburb dummies provide some interesting insights into the differences between communities living around the lake. The dummy for Grassy Park is not significantly different from the base suburb (Lotus River). Pelican Park is significantly different from the base suburb and the negative coefficient indicates less willingness to pay compared to Lotus River. The reason for this may be that Pelican Park is within walking distance of the lake. The peninsula (part of the suburb of Zeekoevlei) is positive, and significantly different to Lotus River. The income dummies are mostly negative (although not significant for all). Hence there is no evidence that the willingness to pay for the environment increases with income. For the highest income group the results show a negative and significant coefficient. This is a strange result (although significant), since one normally expects that higher-income earners will be more willing to contribute to a cleaner environment. Investigations into the characteristics of the respondents in this category revealed the following information: - 44 of the 384 respondents fall into the highest income category (earning more than R per month). - Most of these respondents have a university degree. - Very few have time over weekends to spend on activities involving the lake. - When asked whether they actively use the lake, 28 of the 44 respondents indicated positively. 4 Since stratified sampling was used to ensure that all suburbs are represented, the mean willingness to pay has to be weighted. This was done by giving each suburb a weight according to the actual number of households in the respective suburbs. The statistics on the actual number of households in each suburb were obtained from the 2001 RSA National Census Survey. 9

10 - Activities such as water sports, angling and bird watching are not regarded as important on their activity list of these respondents indicated they value the lake as an environmental good. - Households living at the peninsula are over-represented in the high-income group (y4). In the total sample the peninsula represents 10% of the households, while for this income group (y4) 38% live in the peninsula. When we remove respondents in this group who reside in Zeekoevlei, only 28 respondents remain (residing in Lotus River, Grassy Park or Pelican Park). However, the characteristics of these 28 respondents appear to be no different. A possible (but untested) argument may be that the highest income group in the sample has other options for recreation available, making their interest in Zeekoevlei less dominant. The high-income earners have the ability to move out of the area more easily and can afford other recreational activities, which may possibly make the lake an inferior good for them Estimating the willingness to pay for improved recreational facilities at Zeekoevlei Here we focus on the use value of the lake, specifically on the recreational activities. As explained in Section 4.2, we now exclude households with direct access to the lake. (Only 14 households at the peninsula do not have direct access to the lake.) The payment vehicle used is an entrance fee (per person, per visit) to improve the recreational facilities. Interval regression Data collected on the willingness to pay for improved recreational facilities (using an entrance fee as vehicle payment) show that 79% of all the respondents accepted the fee presented to them. For the maximum entrance fee (i.e. R10 per person per visit), 66% of the respondents facing this fee accepted it. Once again, this is indicative of a fat tail distribution to the right. Therefore, an interval regression is applied to determine the WTP for improved recreational facilities. If the bid is accepted we use this as the lower limit and set an upper limit of 0.5% of the income of the respondent (assuming, as before, that this is the maximum amount the respondents would be willing to pay). If the bid is not accepted, the bid is the upper limit while the lower limit is zero. Table 9 shows the results of the interval regression for improved recreational facilities. Only Pelican Park is significantly less willing to pay an entrance fee. This perhaps reflects that these residents live close to the lake and feel they should have free access. Higher-income groups are mostly significantly more willing to pay an entrance fee (except the highest group). The negative effect of household size probably reflects that the larger households face a higher total fee to visit the lake than the smaller households do. This is contrary to the situation for the environmental qualities, where the fee is part of the monthly municipality bill and does not increase with the household size. We find that households who indicate that better policing is important for visiting the lake are more willing to pay, although this result is not significant. This is contrary to information obtained from the survey questionnaire, where most of the respondents indicated that adequate policing would matter significantly in their decision to visit the lake. A possible reason for this outcome may be that the respondents feel that an entrance fee will constitute some way of policing the area. We find the mean WTP of R7.91 per person per visit (allowing only non-negative values and less than 0.5% of monthly income). Using a logit regression yields higher results (R9.21) and the open-ended question even higher (R11.40). All estimates demonstrate a rather high WTP for using the area. 10

11 11 6. Conclusions and Policy Implications The analysis demonstrates that there is a willingness among the residents of the low-income suburbs to pay for the environmental qualities of Zeekoeivlei Lake. Aggregating over all households in the suburbs gives a monthly contribution of R , or R4.9 million annually for the clean-up of Zeekoevlei Lake. It is important to stress that this result is based on local households only. Visitors to the yacht club or potential tourists to the area (like to the nearby nature reserve at Rondevlei) are not included. It is suggested that this willingness to pay should be included in the policy and budget considerations of the City of Cape Town. It will be costly to clean up the area, and these costs must therefore be considered together with the benefits of doing so. The environment is usually assumed to be a normal good, for which poor people are less willing to pay due to their strict budget constraint. In a low-income urban area this could create a serious long-term problem if the environment is damaged due to the pressure arising from the increase in low-income households. This study found the opposite, however: there seems to be a willingness to pay by the low-income groups living close to the lake but without direct access to it. Furthermore, corrected for other household differences, these lowincome families are more willing to pay for cleanup than their richer neighbours. Local environmental quality may be important for less mobile low-income groups. The willingness to improve the lake is linked both to the lake as such and also to the potential recreational value of the area. The quality of the lake cannot be substituted by better conditions elsewhere there are no indications that households already actively using other lakes do not want to be involved in the cleanup of Zeekoevlei Lake. However, households with high income seem less willing to clean up. This may indicate that safeguarding the environment is not a policy for the rich only. The richer enclave of households on the peninsula of the lake is also willing to pay for cleaning up the area, notwithstanding the fact that they have already decided to settle in the middle of nature. However, they are not more willing to push for paid access to improved recreational facilities. It is not an easy task to constitute the value of the lake. We argue that the pollution and the crime of the area are close complements in destroying the value of the lake. We do not supply a way to distinguish between these effects, but results of our survey clearly indicate that crime influences the attitude of the households to the environmental cleanup. We find some indications that the recreational areas will increase in value if an entrance control regulates crime in a better way. The results are quite diffuse for many of the background variables, indicating that it is difficult to conclude that an improved quality of the area will support specific interests. It is mainly a policy that has a general backing among the households. Only the households of Pelican Park (already within walking distance to the lake) seem to have less interest in the process. Overall, the outcome of the study is quite optimistic in terms of support for the local environment. The Cape Flats is an area with quite a large population consisting primarily of low-income to very poor suburbs. When one finds significant support for cleaning up the Zeekoevlei Lake among the local households it supports evidence for the requirement for stronger activity in this respect from the CCT and the Province. Cost implications also need to be considered. Local governments in the developing world usually have pressing needs for service delivery and few resources are set aside for the provision of improved environmental services. This study reveals that improving the environmental quality and recreational facilities of Zeekoevlei will be of benefit to local low-income communities surrounding the lake. The most significant result is the establishment of a negative slope for the bid functions both for improved water quality and for improved recreational facilities. This shows a downward-sloping demand for both goods. Most of the other explanatory variables were, however, not significant in explaining what affected willingness to pay. More research seems needed on how crime influences the willingness to care for nature.

12 12 7. Acknowledgements The authors are grateful for considerable feedback and comments from Servaas van der Berg (Stellenbosch University) and S. Navrud and O. Bergland (Norwegian University of Life Sciences). 8. References Armirnejad, H., Khalilian, S., Assareh, H. and Ahmadian, M Estimating the existence value of north forests of Iran by using a contingent valuation method. Ecological Economics 58: Arrow, K., Solow, R., Pornoy, P., Learner, E., Radner, R. and Schuman, H Report of the NOAA panel on contingent valuation. Federal Register Carlsson, F., Kohlin, G. and Mekonnen, A Contingent valuation of community plantations in Ethiopia: a look into value elicitation formats and intra-household preference variations. Working Papers in Economics no Department of Economics, Gothenburg University. City of Cape Town, CCT-CPNP household survey of environment and tourism. Summary findings. [Online]. Available: City of Cape Town, 2007a. City of Cape Town Budget 2007/2008 to 2009/2010. City of Cape Town, 2007b. City of Cape Town, Sustainability Report [Online]. Available: 465_pdf Del Saz-Salazar, S. and Garcia-Menendez, L The nonmarket benefits of redeveloping dockland areas for recreational purposes: the case of Castellón, Spain. Environment and Planning 35: Freeman, A.M. III The measurement of environmental and resource values: theory and methods. Resources for the Future. Washington D.C. FAO Economic and Social Development Papers Applications of the contingent valuation method in developing countries: a survey. [Online]. Available: Grafton, R.Q., Adamowicz, W., Dupont, D., Nelson, H., Hill, R.J. and Renzetti, S Economics of the environment and natural resources. Oxford: Blackwell Publishing. Hanemann, W.M Welfare evaluations in contingent valuation experiments with discrete response: reply. American Journal of Agricultural Economics 71: Hanemann, W.M Valuing the environment through contingent valuation. Journal of Economic Perspectives 8: Huszar, E., Bergstrom, J. Holmes, T. and Kask, S. N.d. A methodology for valuing watershed ecosystem restoration. [Online]. Available: Jacobsson, K. and Dragun, A Contingent valuation and endangered species: methodological issues and applications. United Kingdom: Edgar Elgar. Jennings, M.I. and Curtis, J.A Assessing the public service element of Irish public broadcasting the validity of the contingent valuation method. Paper for CVM Conference, University of Chicago, 1 2 February Kohlin, G Contingent valuation in project planning and evaluation: the case of social forestry in Orissa, India. Environment and Development Economics 6: Loomis, J., Kent, P., Strange, L., Fausch, K. and Covich, A Measuring total economic value of restoring ecosystem services in an impaired river basin: results from a contingent valuation survey. Ecological Economics 3: Mitchell, R.C. and Carson, R.T Using surveys to value public goods: the contingent valuation method. Resources for the Future. Washington DC.

13 Navrud, S. and Mungatana, E.D Environmental valuation in developing countries: the recreational value of wildlife viewing. Ecological Economics 11: O Doherty, R The contingent valuation method. Centre for Social and Economic Research on the Global Environment. CSERGE Working Paper PA Quick, A.J.R. and Johansson, A.R User assessment survey of a shallow freshwater lake, Zeekoevlei, Cape Town, with particular emphasis on water quality. Water SA 18: Perman, R., Yue, M., McGilvray, J. and Common, M Natural resource and environmental economics. 2 nd edition. United Kingdom: Pearson Education Limited. Ready, R.C. and Hu, D Statistical approaches to the fat tail problem for dichotomous choice contingent valuation. Land Economics 71: Southern Waters Ecological Research and Consulting cc Zeekoevlei/Rondevlei rehabilitation study final report. Prepared for South Peninsula Municipality. Stenger, A. and Willinger, M Preservation value for groundwater quality in a large aquifer: a contingent-valuation study of the Alsatian aquifer. Journal of Environmental Management 53: World Lakes Database. N.d. Zeekoevlei. [Online] Available: Van Zyl, H. and Leiman, A The residential property price impacts of proximity to selected open spaces in Cape Town. Paper to FEE Conference, Cape Town February

14 14 Tables Table 1. Major spending categories for the City of Cape Town: 2007 Budget Category Capital expenditure R 000 Operating expenditure R 000 Total expenditure (TE) R 000 TE as percentage of total budget (%) Utility services Transport Environmental resource management Source: City of Cape Town. 2007a. The City of Cape Town Budget for 2007/2008 to 2009/2010