A community-based impact assessment of the Wild Coast spatial development initiative, 1997–2004

Abstract

Much of the debate around the spatial development initiatives (SDIs) in South Africa is based on economic theory. To add the community perspective, an evaluation of local economic development on the Wild Coast followed a baseline in 1997 with comparable surveys in 2000 and 2004. Apart from an increase in access to piped water and a reduction in unofficial payments for health services, there was little community evidence of development over this period. Residents reported decreasing knowledge of the SDI and there was no increase in numbers considering small business ownership. In the investment-intensive ‘anchor’ areas, as in the SDI as a whole, there was no significant increase in employment and more households received remitted incomes from migrant workers. In an SDI for small businesses in tourism and agriculture, there was a dramatic fall-off in food production. No more households had loans in 2004 than in 1997, but more were taking loans from loan sharks. Other spatial planning initiatives might learn from the Wild Coast, not least through the perspectives gained from community-based impact assessments.

1. Introduction

Improvements in South Africa since the new dispensation in 1994 include new schools, health clinics, housing projects, improved water facilities and an emphasis on training programmes (Harsch, 2001). The gross domestic product (GDP) increased at an average rate of 2.7 per cent between 1994 and 2000 and continued to improve through to 2004 (World Bank, 2006) although it was still far below the initial target of 6 per cent growth per year. Eradication of poverty remains the official priority of the ruling party, yet unemployment has doubled since 1994 and levels of poverty and inequality have risen apace. Statistics South Africa (2004) data show that unemployment rose from 19 per cent in 1996 to 29 per cent in 2001. Many South Africans are becoming disillusioned at the lack of progress and losing faith in the promises made to them by the new government, particularly in terms of employment and standards of living (Mahadea, 2003).

The reasons for the continuing problems are manifold. Some authors refer to lack of experience among elected officials, lack of accountability and the failure of the national government to achieve working partnerships with traditional leaders at local levels, all of which are fuelling an already increasing divide between officials at all levels of government (Cousins & Kepe, 2004). This has not been helped by high rates of violence and crime and the AIDS epidemic. With nearly one-fifth of the population infected, HIV/AIDS stagnates economic growth not only through the loss of human capital but also by redirecting scarce public funds for medical care (Bhorat & Cassim, 2004).

The 2005 Development Report published by the Development Bank of Southern Africa, Human Sciences Research Council and United Nations Development Programme (Aliber, 2005) asked a painful question: why, if the origins of economic dualism are rooted in the cheap forced migrant labour introduced by the mining industry and reinforced during apartheid, does dualism persist under democracy – when all the laws and many of the practices of the past have been abolished?

The explanation of a ‘second economy’ was offered as a framework for understanding or explaining how increased poverty and marginalisation could coexist with South Africa's economic growth (Aliber et al., 2006). For some authors, however, the postulation of a dual economy merely covers over the real reasons for continuing poverty and unemployment. Frye (no date), for example, dismissed the use of ‘two economies’ as masking a bias in favour of business, a bias that promotes separate and unequal development of the rich and the poor almost as deliberately as did the apartheid era Bantustans. Frye (no date) laid the blame for the skewed allocation of income and resources directly on the conservative macroeconomic policies adopted by the South African government in 1996.

Faced with extremely uneven distributions of wealth, high levels of poverty and the fact that many basic needs across the country are not being met, the government created the Reconstruction and Development Programme (RDP) in 1994. The RDP aimed to reduce poverty and distribute income more evenly, but was seen as a quick if short-sighted programme of basic needs fulfilment (Luiz, 2003). Within two years the RDP gave way to a new longer-term strategy called Growth, Employment & Redistribution (GEAR). Rooted in neoliberal paradigms, GEAR was to reduce the role of the state and increase corporate and private investment in the macroeconomy (Soderbaum & Taylor, 2001; Rogerson, 2002). A dramatic departure from the RDP, it is likely that GEAR represented the South African government's acceptance of the needs of globalisation (Mashalaba & Langa, 2005).

GEAR was publicly proposed as a way to provide a fast growing economy, create jobs, redistribute income and hasten universal access to basic needs (Republic of South Africa, 1996). However, it failed to meet its goals, including increased local and foreign investment and employment creation, although it coincided with the reduction of the budget deficit and reduction in inflation. One of the policies adopted along with this strategy was the creation of spatial development initiatives (SDIs), which were intended to promote investment and development in areas that were considered to have the greatest potential for growth.

2. Background

The SDIs, which were first conceived in 1995 and implemented in 1996 with the Maputo Development Corridor, were supposed to target areas of untapped potential for growth that had been under-resourced and neglected with past spatial development strategies. There were three types of SDIs: regional industrial, agro-tourism and mixed industrial and agro-tourism (Rogerson, 2002). Each SDI was to take on its own personality, with different interventions designed around its available economic and entrepreneurial resources.

Some SDIs were tied into local economic development theories, two key components of which are the encouragement of local entrepreneurship and the establishment of strong locally controlled economic sectors (Coffey & Polese, 1984). Including the Wild Coast SDI, these focused on short-term interventions designed to attract private sector investment, stimulate the growth of small, medium and micro-enterprises (SMMEs) and empower local communities (Rogerson, 2001; Cousins & Kepe, 2004). It was thought they would promote new businesses rather than redistribute or concentrate existing ones, as had happened with incentive programmes in earlier strategies. Furthermore, perceived obstacles to investment, such as inadequate infrastructure (roads, electricity and communications), were to be identified and overcome (Jourdan, 1998).

The 13 SDIs included the Fish River (industrial), the Wild Coast (agro-tourism) and the Maputo Development Corridor (mixed). Rogerson (2002), quoting official data from an audit conducted in 2000, reported 688 active SDI projects receiving roughly R165 million in investments and responsible for 100 000 jobs.

Since their inception, however, the SDIs have been strongly criticised. Some analysts have maintained that the neoliberal globalisation-friendly approach of the national government has overridden all regional processes, making local economic development and local empowerment impossible (Taylor, 2003). The government itself has been reduced to an investment promotion agency and many of the SDIs were crowded with private investments without any long-term strategic plan (Soderbaum & Taylor, 2001). Common problems with SDIs include a lack of communication with the public, particularly at the local level, capacity deficiencies at various levels of government, unresolved land claims and other local tensions and poor understanding of local practices (Kepe et al., 2000; Kepe, 2001; Luiz, 2003; Cousins & Kepe, 2004).

3. The Wild Coast SDI

The Wild Coast SDI is located in the Eastern Cape, the second poorest province in the country. In the areas formerly known as the Transkei, the Wild Coast is named for its rough shoreline along the Indian Ocean and is often described as being an area of unspoilt natural beauty and untapped economic potential. After years of labour migration, by 1994 the Wild Coast population was predominantly female and unemployed. Like much of the Eastern Cape Province, the Wild Coast had little access to clean water or public service infrastructure. Unemployment was higher than the national average and nearly three-quarters of the population lived in poverty (Luiz, 2003). There were major income divisions between urban areas (such as East London and Port Elizabeth) and the surrounding rural areas.

In 1997, the South African Department of Trade and Industry (DTI) started the Wild Coast SDI in an attempt to kick-start economic growth through public and private investment in SMMEs. The SDI was supposed to empower local communities to find solutions to their economic problems. Its specific objectives largely followed the general SDI objectives: increase employment, promote entrepreneurship and enhance community empowerment. The focus of the strategies was on the emergence and clustering of SMMEs, particularly those in the eco-tourism sector. The Wild Coast SDI identified four coastal anchor areas as the focus for both public and private investment: Mkambati, Port St Johns, Coffee Bay and Dwesa/Cwebe (see Figure 1).

Figure 1. Map of the Wild Coast SDI showing anchor locations Source: CIET, 2004.

The goal in the anchor areas was to promote eco-tourism through improvements to infrastructure and other investments and to encourage complementary and additional ventures (Cousins & Kepe, 2004). Its physical beauty and coastal proximity made it a defensible argument to promote the Wild Coast as a tourist location. However, Koch et al. (1998) highlighted some problems with tourism-led economies. First, there was too much reliance on external and largely uncontrollable forces (for example, it can take years for a region to live down the bad publicity caused by a report of an isolated violent activity and regain tourists' confidence). Second, the tourism industry was seasonal, with no alternative sources of income for large parts of the year. Yet tourism-based businesses do offer employment opportunities to marginal groups, including women, the young and the unskilled – all of whom make up a large proportion of the population of the Wild Coast.

A primary component of the eco-tourism strategy (and other industries such as forestry and agriculture) was the promotion of SMMEs in order to reduce unemployment and retrenchment, giving people a local opportunity to generate income through self-employment. However, there are many documented challenges to the success and sustainability of SMMEs (see Mead & Liedholm, 1998; McCormick, 1999). Among these, Kirsten & Rogerson (2002) pointed out that the view of ‘opportunity’ was overstated and that many SMMEs are started out of dire need, in more of a survivalist vein and not as well thought-out ventures. Wild Coast SDI planners, however, carried on regardless, believing that SMMEs would provide income for the poor and opportunities for women and play an important role in local economic development initiatives (Arkwright et al., 1998; Lefakane, 1998).

Cousins & Kepe (2004) outlined problems from a failed eco-tourism project in the Mkambati area that may be relevant to the entire Wild Coast SDI. The SDI ‘barely recognised the subtle realities of the use of natural resources in local livelihood strategies, or the local institutional arrangements through which access to resources is gained’ (Cousins & Kepe, 2004:42). The attraction of external investors to an area that shows promise for eco-tourism was complicated by uncertainties about who owns the land on which these projects are situated. Kepe et al. (2002) reported 65 pending land claims in the area between Port St Johns and Port Edward (the northern third of the Wild Coast SDI coastline). Furthermore, there was a lack of local capacity for small business development and many parts of the area are extremely remote. With levels of unemployment in the area ranging from 70 to 85 per cent (Kepe et al., 2001; Mitchell et al., 2001) the fact that so few people had considered starting their own businesses was informative.

Much of the criticism of SDIs was based on economic theory and qualitative studies (Kleynhans et al., 2003; Cousins & Kepe, 2004). At least on the Wild Coast, the authors were impressed by the need for empirical studies of local development from the community perspective. What is the ‘second economy’ at ground level? This article presents findings from follow-up surveys in 2000 and 2004, following a 1997 baseline at the inception of the Wild Coast SDI. It offers a community perspective supported by empirical evidence felt to be lacking in the existing literature. The paper describes methods for collecting and analysing data from the communities, in part to demonstrate the relevance of epidemiological approaches in assessing development impact. The discussion focuses on the extent to which the SDI has met its stated objectives and provides some possible explanations for its failures.

4. Methods

Leading up to a 1997 baseline survey, a consultative process with provincial planners, non-governmental organisations and community groups defined development indicators. These covered the community knowledge about the SDI, markers of local buy-in and empowerment. The study used coverage of essential public services to describe the extent to which the SDI met basic needs. Employment and attitudes to small business development provided the main indicators of economic impact.

The authors conducted the 1997 baseline in partnership with the Eastern Cape Socio-economic Consultative Council. Findings from the baseline showed that people of the Wild Coast SDI region were unaware of what they could do to improve their socio-economic conditions. In addition, there were high levels of unemployment, a low proportion of households getting their water from protected sources such as taps, a lack of food security, a significant level of corruption in the public services and little knowledge of the SDI project itself (Andersson & Galt, 1998).

The multi-method follow-up surveys and analysis methods relied on modern epidemiology and participatory research techniques used in over 50 countries worldwide (Andersson, 1985; Andersson et al., 1989). The last stage random sample of communities was stratified by anchor/non-anchor status, geographic location (such as coastal/non-coastal), proximity to infrastructure and road accessibility. The sample was weighted to make the sample proportions match proportions of the population of the Wild Coast. The 2000 and 2004 follow-up surveys returned to the same communities visited in the 1997 baseline to interview the same number of households (but not necessarily the same households) at each site.

The data collection instruments included household questionnaires, small business reviews, community profiles and gender-stratified focus groups. All of the instruments were translated into the local language (isiXhosa) and then translated back into English by non-members of the research team to ensure that the questions remained true to their intended meaning. The instruments were then piloted extensively before being implemented in the field. The pilot exercises helped to refine the instruments, test for clarity and ensure accurate translation.

The fieldworkers recorded and stored household data without any identifying fields, thus ensuring the confidentiality of the respondents. Confidentiality of the sample community identities was maintained as far as possible, especially with regard to the non-anchor areas. The exact sample sites were not included in any reporting.

Data entry and analysis relied on the public domain software EpiInfo and the open source analysis and geomatics software CIETmap. The indicators were weighted in order to account for the effect of sampling and the weighted results were reported. The study derived the main test of significance differences for comparisons between years from the Mantel–Haenszel χ², expressed as single tailed p-values (Mantel & Haenszel, 1959; Mantel, 1963). The reported p-values compare the years 1997 and 2004 unless otherwise noted. Contrasts and associations are described using the odds ratio (OR) accompanied by a 95 per cent confidence interval (CI). Averages are accompanied by a measurement of the standard error of the mean (SE).

The authors returned preliminary findings to the sample communities for discussion in gender-stratified focus groups. One researcher facilitated the group while another recorded the main points and quotes from the discussion. The discussions focused on the interpretation of results and how these should inform planning strategies. Key findings from the baseline survey and first follow-up cycle were shared with provincial planners and other stakeholders in a series of workshops and through the mass media (Eastern Cape News, 1999; Mail & Guardian, 2001).

The authors also shared findings through the use of population-weighted raster maps to give the data a visual interpretation (Andersson & Mitchell, 2006). Raster maps are derived from a set of sentinel sites, the data of which are interpolated to create a surface across the study area. The interpolation module accounts for population weighting and transforms the geographic space into population space. This transformation allows planners to interpret the maps as percentages of people affected and not as percentages of space affected – an important consideration when measuring impact at the community level (see Figure 2).

Figure 2. Percentage of household respondents who had heard of the Wild Coast SDI. Time-series map showing decreasing knowledge of the Wild Coast SDI (1997-2004) Source: CIET, 2004.

5. Results

5.1 Evidence base

The authors collected data from 2457 households in 1997, 2363 households in 2000 and 2383 households in 2004. Approximately one-third of the sample households were located in the anchor areas. In 2000 and 2004 three out of every four respondents were female, slightly higher than in 1997. Approximately one-half of the dwellings were made of mud with grass roofs in each sample, with a slightly lower proportion reporting this in 2004.

5.2 Knowledge and empowerment

Overall knowledge of the SDI project decreased from already low levels (roughly down from 20 per cent in 1997 to 9 per cent in 2004), as did the proportion who felt they had a say in it. The proportion of respondents who felt they had a say in the SDI, which is an indicator of empowerment, dropped from 64 per cent in 1997 to 21 per cent in 2004 (see Table 1).

Table 1: Evidence base and knowledge of the SDI

	1997	2000	2004	p -value
Households	2457	2363	2383	–
Female respondents	70% (1736/2457)	78% (1817/2363)	76% (1781/2378)	p < 0.05
Households' structure: mud and grass thatch	53% (1269/2455)	58% (1285/2342)	50% (1088/2374)	p < 0.05
Heard of the SDI	22% (577/2453)	20% (530/2360)	9% (250/2362)	p < 0.05
Feel they have a say in the SDI (among those who had heard of it)	64% (414/567)	31% (217/528)	21% (65/244)	p < 0.05
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

5.3 Basic needs (food and water)

In 2000 the household respondents and focus groups reported that lack of money and increasing unemployment were their biggest challenges. By 2004, when these questions were asked again in the same communities, more people reported basic needs, such as food and water, as their main problem (Table 2). By 2004 more households in the anchor areas reported getting their water from a protected source, consistent with the provincial trend (Table 3). The respondents in 2004 were nearly six times more likely to report getting water from a protected source than those in 2000 (OR 5.97 and 95 per cent CI 4.61–7.74). Food security overall had worsened. There had been a dramatic fall-off in food production (68 per cent in 1997, 55 per cent in 2000 and only 21 per cent in 2004). The average monthly expenditure on food items had increased and there had been a steady decline in household food production since 1997 (Table 4).

Table 2: Main problems reported by household

	2000	2004	p -value
Money	44% (1033/2324)	20% (443/2227)	p < 0.05
Unemployment	23% (518/2324)	23% (504/2227)	p = 0.78
Food	3% (75/2324)	8% (186/2227)	p < 0.05
Water	2% (56/2324)	14% (294/2227)	p < 0.05
Utilities	2% (58/2324)	10% (237/2227)	p < 0.05
Other problems	13% (305/2324)	15% (332/2227)	p = 0.88
None	12% (279/2324)	9% (231/2227)	p = 0.08
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

Table 3: Access to protected sources of water

	1997	2000	2004	p -value
Overall	18% (554/2475)	15% (477/2359)	36% (1032/2301)	p < 0.05
Anchor areas	19% (286/858)	14% (236/810)	56% (540/777)	p < 0.05
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

Table 4: Food security

	1997	2000	2004	p -value
Average monthly expenditure on food	R301 (SE 4.2, n = 2409)	R316 (SE 3.9, n = 2307)	R390 (SE 5.0, n = 2269)	–
Produced food	68% (1295/1901)	55% (846/1532)	21% (411/1942)	p < 0.05
Sold food (among those who produced)	12% (178/1291)	7% (61/838)	22% (63/401)	p = 0.34
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

5.4 Infrastructure

More people reported access to sewage, electricity and telephones as their main problem. Community leaders supported this, claiming that fewer sites now had access to both public telephones and local transport.

5.5 Use of government administration and health services

Satisfaction with government health services increased from 2000 to 2004 (Table 5). There was also a significant reduction in unofficial payments for services in the clinics, although it was hard to link this to the SDI (Table 6). Satisfaction with government administration services had remained consistent since the baseline survey, even though waiting times for service had increased. The use of government administration offices showed a significant increase in households seeking welfare grants and no increase in those seeking employment (Table 7).

Table 5: Levels of satisfaction (government services)

	1997	2000	2004	p -value
Education system	–	75% (1745/2358)	79% (1795/2306)	p < 0.05^a
Health services	86% (1965/2335)	65% (1517/2360)	70% (1616/2326)	p < 0.05
Government administration	68% (1200/1737)	68% (685/988)	70% (799/1159)	p = 0.93
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.
^a2000–4.

Table 6: Use of government health services in the previous year (individuals)

	2000	2004	p -value
Did not use any health service in the previous year	28% (2917/10433)	32% (2812/8383)	p < 0.05
Used government clinic (among those who used it)	73% (5604/7516)	77% (4343/5571)	p < 0.05
Users of government clinics: made payment	24% (1159/5571)	7% (201/4313)	p < 0.05
Users of government clinics: waited more than one hour	27% (1321/4920)	32% (1238/3822)	p < 0.05
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

Table 7: Use of government administration in the previous year

	1997	2000	2004	p -value
Reason for contact: collect welfare grants	8% (136/1727)	11% (109/978)	34% (365/1198)	p < 0.05
Reason for contact: seek employment	2% (26/1720)	2% (21/978)	3% (49/1198)	p = 0.01
Made payment	26% (462/1734)	28% (262/967)	13% (150/1189)	p < 0.05
Waiting times: more than one hour	–	42% (372/880)	52% (520/1009)	p < 0.05^a
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.
^a2000–4.

5.6 Employment

The study grouped working age adults (18–65 years) into five economic categories: those who were employed, those who were unemployed but looking for work, those who were unemployed and not looking for work, those who owned their own business and those who were considering owning one (Table 8). Male employment remained constant between 2000 and 2004, when 22 per cent of men reported employment. Fewer women than men had formal employment, but there was a small increase from 12 per cent in 2000 to 14 per cent in 2004. Employment levels were nearly identical in both anchor and non-anchor areas, showing no measurable improvement between 2000 and 2004 (Table 9). There was no significant decrease in the proportion of households receiving income from migrant workers: in the anchor areas the proportion receiving remitted income increased from 28 per cent in 2000 to 34 per cent in 2004 (Table 10).

Table 8: Employment and entrepreneurship among males and females (individuals aged 18–65 years)

	Males			Females
	2000	2004	p-value	2000	2004	p -value
Employed	20% (390/1821)	22% (367/1558)	p = 0.07	12% (354/3016)	14% (383/2723)	p < 0.05
Unemployed and looking for work	46% (834/1821)	39% (601/1558)	p < 0.05	46% (1386/3016)	41% (1102/2723)	p < 0.05
Unemployed and not looking for work	29% (522/1821)	33% (494/1558)	p = 0.24	34% (1032/3016)	37% (985/2723)	p = 0.12
Currently owns business	4% (75/1821)	6% (96/1558)	p < 0.05	7% (244/3016)	8% (253/2723)	p = 0.11
Considered owning business	28% (513/1782)	26% (418/1563)	p = 0.18	32% (978/3011)	30% (852/2747)	p = 0.23
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

Table 9: Employment and entrepreneurship among anchor and non-anchor sites (individuals aged 18–65years)

	Anchor			Non-anchor
	2000	2004	p-value	2000	2004	p -value
Employed	14% (271/1664)	17% (272/1347)	p = 0.45	15% (473/3173)	17% (483/2956)	p = 0.32
Owns business	6% (108/1664)	8% (140/1347)	p < 0.05	6% (211/3173)	7% (211/2956)	p < 0.05
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

Table 10: Households receiving income from migrant workers

	2000	2004	p -value
Overall	32% (724/2341)	34% (744/2211)	p = 0.16
Anchor areas	28% (214/802)	34% (239/719)	p < 0.05
Non-anchor areas	34% (510/1539)	34% (505/1492)	p = 0.68
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.

5.7 Entrepreneurship

There was an increase in the small proportion of male respondents who owned small businesses, from 4 per cent in 2000 to 6 per cent in 2004 (Table 8). There was no increase among the female respondents. There was no difference in SMME ownership between the anchor and non-anchor areas (Table 9). There was no increase in the proportion (male or female) of those who had considered starting their own business. Production-related small businesses showed a decrease in average monthly earnings (R283 in 2000 to R241 in 2004), while service-related businesses showed an increase (R492 in 2000 to R512 in 2004). In 2000 small business owners complained of a lack of resources, the lack of training and poor capacity of employees and customers buying on credit. In 2004 businesses complained about the same factors, although many of those interviewed in 2000 were no longer in existence in 2004.

5.8 Loans and credit

Between 2000 and 2004 there was a significant decrease in households that reported loans, returning to 1997 levels (16 per cent). The source of the loans changed, however, as far more households in 2004 reported getting their loans from a loan shark than in either 1997 or 2000 (Table 11). Fewer loans were for groceries and there were increases in loans for school fees and to pay doctors. Only 2 per cent of loans were for setting up a business in 2004, no change from 1997.

Table 11: Household loans and credit

	1997	2000	2004	p -value
Households with loans	17% (416/2437)	42% (951/2302)	16% (390/2255)	p < 0.05^a
Source of the loan: loan shark	1% (6/414)	2% (23/917)	32% (149/405)	p < 0.05
Reason for loan: to set up business	2% (10/407)	1% (6/866)	2% (9/387)	p = 0.83
Reason for loan: groceries	63% (255/407)	78% (666/866)	53% (204/387)	p < 0.05
Reason for loan: school fees	3% (14/407)	2% (19/866)	10% (36/387)	p < 0.05
Reason for loan: pay doctor	1% (2/407)	1% (5/866)	13% (51/387)	p < 0.05
All indicators are weighted and p-values represent 1997–2004 unless indicated otherwise.
^a2000–4.

5.9 Education

Enrolment was high (up to 90 per cent in some age categories) and there was evidence that satisfaction with the education services was increasing (Table 5).

6. Conclusions

A planners' forum after the authors' 2000 impact assessment recommended an urgent shift to change the SDI from an external investment-driven initiative to focus more on local skills building and incentives for development within the community (Mitchell et al., 2001). The authors' 2004 evaluation showed a continued low level of ownership of small businesses (6 per cent of male and 8 per cent of female respondents in 2004), an increase in the proportion living on remitted income even in the anchor areas, an increase in borrowing from loan sharks, no improvement in employment for men and only a little improvement in employment for women. Three-quarters of the working age population in 2004 were without a wage-earning opportunity. There had been a dramatic fall-off in food production (68 per cent in 1997, 55 per cent in 2000 and only 21 per cent in 2004).

One area of improvement in recent years was access to protected water. The study's findings confirm those of Statistics South Africa, which reported access rising from 54.2 per cent in 1996 to 62.4 per cent in 2001. This still leaves a large proportion of the population having to collect their water from unprotected sources such as rivers, sometimes several kilometres away, using up large amounts of time solely on this activity.

The poor performance of the Wild Coast SDI in the authors' 2000 evaluation might be explained by an unfavourable environment after years of apartheid rule, widespread poverty, the weak infrastructure for SMMEs, high levels of crime and the corruption of public services. By 2004, however, the SDI was no closer to taking off.

Evaluated from the community perspective, there are several reasons for the failure of the Wild Coast SDI. Other than an information campaign in the planning stages, there was little communication between SDI planners and communities or local leaders. The SDI facilitators, who were initially expected to provide this link, disappeared completely at the time of the 1997 baseline study. Many respondents reported that the only time the SDI had been mentioned to them was by fieldworkers conducting the evaluation surveys. Of the few respondents who did know about the SDI, most claimed this was a series of promises that never came to fruition.

The Wild Coast SDI was not alone in failing to meet its objectives. Also in the Eastern Cape, the Fish River SDI was primarily an industrial SDI and was supposed to expand to include tourism and agriculture. This SDI extends from East London to Port Elizabeth, each pole having an industrial development zone (IDZ). According to Luiz (2003), progress has been slow and neither IDZ has come to full development. Unclear boundaries and communication between various government agencies, inability to access finances and problems associated with two IDZs in close proximity are possible explanations. In similar fashion to the Wild Coast SDI's inability to move decisively beyond a single economic identity, i.e. tourism, the Fish River SDI remained focused on the industrial sector, ignoring tourism and agriculture.

Since the authors' 2004 evaluation, the Wild Coast SDI has been terminated. The responsibility for developing the Wild Coast has moved from the DTI to the Department of Environmental Affairs and Tourism. Current initiatives include the EU community-based tourism initiative, the highly controversial N2 toll road, the establishment of the Pondoland National Park and a new Wild Coast Development Project (Project Management Unit, Department of Environmental Affairs and Tourism, January 2006, personal communication).

If these initiatives were interested in learning from the SDI's failure to take off, the first step to local empowerment would be to increase knowledge and buy-in from the communities. This would necessitate two-way communication. Engaging with and listening to local leaders might increase understanding of the processes that fuel local tensions. Changes to the school curriculum – or even the addition of adult education options – might provide an emerging workforce with more tools that could support small enterprise development and eventually reduce unemployment. The Wild Coast SDI failed to take note, much less take advantage, of community-based evidence. The current spatially focused development projects in the Wild Coast would benefit from community-based evidence on the true impact of the economic strategies, while taking into account the impact of other initiatives. Probably the biggest lesson, notwithstanding the official discourse of dualism, is that spatially focused development initiatives will not be understandable and will not prosper so long as they are regarded as disconnected from what is, in effect, the single national economy.

The Wild Coast SDI evaluation was designed in collaboration with the Eastern Cape Socio-economic Consultative Council and other SDI stakeholders. The authors thank the households and community leaders of the Wild Coast who provided information about their daily lives and participated in focus groups and small business interviews. Special recognition is due to the local supervisors and fieldworkers who performed the data collection and data entry. The baseline in 1997 and follow-up in 2000 were partially funded by the Canadian International Development Research Centre (IDRC) as a project entitled The Wild Coast SDI: A Community-based Planning Support Platform for SMEs (IDRC File no: 97-8906-01).