Working it out: Labour geographies of the poor in Soweto, South Africa

Abstract

Local economic development (LED) research and policy grapple with the informal economy and township transformation. While most current thinking centres on firms, this paper argues that non-firm worlds of work and their spatiality are not adequately understood. Representations of the places where poor people work remain abstract and incomplete. The paper reports on a survey of 320 low-income Sowetan residents and in-depth interviews with 20 workers about their work roles in the urban space economy. The findings, which show poor workers engaging with diverse sectors and locations in complex ways, challenge the dominant spatial narratives about isolated poor residential areas. Poor workers deliberately create their own social capital in work realms. This being the case, a more finely tuned conceptualisation of these workers and their roles in urban space is essential to sharpen LED discussions so that policies can be based more on real rather than imagined spatiality.

Keywords:

• informal settlements
• space economy
• social capital
• labour geographies

1. Introduction

These days everyone is smitten with local economic development (LED). It offers local government, the private sector, the not-for-profit sector and the local community an opportunity to work together to improve the local economy from the bottom up. LED has become both the subject of social policy and the object of serious economic development policy. LED initiatives form a core part of social policy agendas for reducing poverty and unemployment and of economic policy agendas for targeting local economic growth to reduce marginalisation. Despite this warranted attention, considerable confusion exists about the roles of LED, particularly in the informal economy, and about how they are to be implemented in the most marginalised economic contexts (Rogerson, 2006, 2008).

The single greatest challenge for LED is understanding what is actually happening on the ground in informal economic spaces in Africa and elsewhere (Grant, 2009). South Africa has developed the concept of ‘second economy’, according to which the first (formal) and second (informal) economies are distinct entities. The official position of the South African government is that the second economy is structurally disconnected from the first and is incapable of self-generating growth and development. According to the government, state intervention is necessary through such programmes as skills development, entrepreneurship, micro credit, and home-based food production; when combined with massive public works programmes and township economic development programmes, local demand can be stimulated and local markets can be made thicker (Webster, 2005; Trade and Industrial Policy Strategies, 2009).

There has been much criticism of the all-encompassing conceptualisation of the second economy, although scholars argue that it is reviving interest in devising policy supports for assisting those working in the informal economy (Huchzermeyer, 2004; Rogerson, 2006, 2008). There is a general understanding of homogeneous characteristics of informal work but little knowledge about the details (Webster, 2005). For instance, informal work subsumes a number of different employment statuses (such as self-employed, part-time, subcontracted) as well as various combinations. The lack of precision in discussing work is surprising considering the many noteworthy studies on the spatial life of informals and subgroups in various developing world settings (e.g. Breman, 1996; Dierwechter, 2004; Biles, 2008). While many fields (such as entrepreneurship, development and geography) acknowledge a rich diversity of experiences of informality, representations of the spaces in which poor people work remain abstract and a-spatial (Marx, 2007; Nijman, 2010). One must move beyond concepts of under-spatialised informal workers as ‘urban hunter-gatherers’, ‘nomads’, ‘hawkers’ and so forth (Sitas, 2001).

Despite efforts since 1992 to refocus LED on the development of the pro-poor, this arena still lacks conceptual specificity. The sharpening of pro-poor LED now entails a township focus (City of Johannesburg, 2008c). Spurred by the proliferation of informal settlements1 and the continued structural underdevelopment of areas disadvantaged by apartheid, the government is now tackling township economic transformation. Soweto presents a vital test case. Township LED emphasises housing, the building up of general infrastructure and the development of firms, but not the workers themselves. Soweto's first international conference on entrepreneurship and small business development in January 2010 illustrated this policy tilt. Firms are singled out for job creation, employment and economic growth potential. They are to be strengthened by increased business support, being given a more effective voice in local development (via the Greater Soweto Business Forum) and by incubation in clustered industrial estates.

Focusing on under-theorised and under-studied workers, this paper turns the assumptions about second economy, firms and marginalised areas on their head by assessing the worlds of work from below. Building on Webster's (2005) conceptualisation, I argue that work includes various categories and that each of these categories encompasses spatial dimensions that need examining closely for better clarity and policy relevance. This research also connects to the literature on working space and the creation of social capital (Lyons & Snoxell, 2005), with the aim of putting the working space of the poor back into LED research, policy and planning. The paper presents evidence of work activities through a case study of low-income settlements in Soweto, South Africa's largest township. Researchers determined it was home to 1.1 million residents (Crankshaw et al., 2000:846). It is a well-defined setting with a well-documented historical past as a dormitory area and as a major site of political resistance during the apartheid era.

The paper is divided into a further seven sections. Section 2 reviews recent literature on the worlds of work and adds a spatial element, Section 3 presents the most salient spatial narratives about marginalised areas and evidence of linkages to wider economic space, Section 4 discusses the study sites and surveys of households in Soweto, Sections 5 and 6 present the study's findings, drawing on household questionnaires, Section 7 details the 20 follow-up interviews with heads of households, and Section 8 concludes.

2. Worlds of work at the bottom: A neglected spatiality

To date, the major emphasis in the social sciences and in LED policy has been on the role of firms in the informal economy. This focus is reinforced by official statistical data; for example, South African National Labour Statistics used in most research on the informal economy has used an enterprise-based definition (Devey et al., 2006). This firm-based orientation has limitations for understanding geographies of work. First, informal work is rarely done in formally zoned business locations; rather, it is most often done in homes, in backyards, in converted buildings, on pavements, and so on, or moves from place to place. Second, concentrating on larger informal firms (rather than on workers) biases the examination towards firms more likely to be plugged into wider supply networks and reinforces the misdirected incorporation of informal work into the formal sphere (Du Toit & Neves, 2007). Third, there are sites such as informal settlements and rehabilitated residential areas (that often accommodate former informal settlers) that we know virtually nothing about in terms of the extent of their participation in the wider space economy.

Data on the informal economy in South Africa are limited. For example, no manufacturing census has been undertaken in recent years, and there is no governmental economic data on informal settlements. It is therefore puzzling how widely this concept of a second economy has been accepted, given that most types of work and their spatial manifestations have been neglected.

Obviously, workers and their households exhibit different dynamics from informal firms. For the most part, much of the policy and academic literature ignores the role of workers in making their own economic geographies: instead, labour is conceived in a passive manner. Analyses of the dynamics of space economies typically consider workers primarily from the point of view of how capital (in the guise of powerful firms) and, to a lesser degree, the state makes investment decisions on the basis of the differences between workers located in particular places (Herod, 1997). Webster reminds us why workers are key active agents in the economy:

Opportunities for employment and the conditions under which people are employed still have more impact on most individuals' chances than many other fashionable concerns. Work and employment structure our lives and shape inequalities of condition and opportunity to a greater extent than most, if not all, other areas of social life. (2005:56, citing Brown 1991:1)

While refinements of notions of work and employment have been addressed (Webster, 2005; Webster & Von Holdt, 2005), working space remains an under-explored frontier (Herod et al., 2007). To extend our thinking about working space, I draw on literature about building trust in economic space (Mackinnon et al., 2002; Murphy, 2003, 2006), a literature that focuses on firms but also pays close attention to workers. I respond to the strong calls for more empirical research (as opposed to theorising) to assess what brings workers together in a particular space at a particular time and what opportunities draw them there (Hudson, 2004; Murphy, 2006; Kanai, 2010). Putting workers at centre stage, the paper also draws on a discussion of social capital (Lyons & Snoxell, 2005), which stresses that the urban poor actively create new relations centred on their main economic livelihood (a finding that challenges the overemphasis on family and kin relations common to social capital thinking in African contexts) (Meagher, 2005). Integrating the two areas of the literature (working space and creating social capital), I tease out the way trust is embodied in encounters in real workplaces which evolve from specific, unbounded movements in broader socio-spatial relationships (Holt, 2008). This enables a spatial rethinking about social capital – geographical accounts of which have been considered ‘almost dead’ (Holt, 2008:227).

3. Spatial narratives on possible links of low income settlements to the wider space economy

We do not know the extent to which low-income Soweto workers are linked or delinked to a wider space economy. We must, however, acknowledge that there are competing spatial narratives that challenge the idea of a second economy, emphasising movements of workers and raising different possibilities for potential linkages. Table 1 presents scholarly narratives that inform government thinking and identifies very different work possibilities.

Table 1: Spatial narratives on potential links of low-income settlements to the wider space economy

Narrative	Work	Potential linkage	Geographic reach	Space
Exclusion	Workless, survivalist self-employment	Virtually non-existent, some potential homeworkers	Remote location, isolation	Constraining, non-economic
Entrepreneurship	Firms and self-employment, entrepreneurship	Informal links, subcontracting networks, backward links to formal economy	Local and wider networks	Enabling
Home-based enterprises (HBEs)	Firms and self-employment	Backward links to formal economy	Local	Enabling but constrained
Apartheid legacy	Domestic service workers	Informal networks, hidden links inside formal economy	Far from residential location	Constraining
Self-help/mutual aid community	Households/informal links, state support	Local, rural and government	Local and elsewhere	Enabling and constraining

The first, the exclusion narrative, dominates official South African second economy thinking (Beall et al., 2002; Davis, 2006). Government snapshot data highlight increasing inequality and a polarised labour market, as Webster emphasises:

Official data show the resources of 6.8 million formal permanent workers at the core of the economy are increasing while at the same time the resources for those at the margins are reduced (an undistinguished 2.1 million temporary, part-time and outsourced non-core workers and a peripheral workforce of 2.3 million informal and 4.3 million unemployed). (Webster, 2005:16, citing 2003 labour data from The Presidency)

In some instances formal firms can reach inside informal spaces to employ homeworkers (e.g. Kenny, 2001, found evidence in manufacturing, retailing, transport and construction), but for the most part the economic realms have no ties with each other. It is assumed, therefore, that core and peripheral workers are separate and structurally disconnected from each other. Core formal workers and ‘jobless’ workers operate in different spaces: sealed off from each other, living in separate and often distant locations. The most excluded areas are informal settlements characterised by spatial confinement (with large numbers of migrants) on denuded urban fringes, where survivalist strategies dominate. Here large swathes of the population are condemned to permanent worklessness. This binary framing of the economy as being two separate realms, one subordinate to the other, obviously creates a negative image of the subordinate one, particularly when it comes to spaces such as informal settlements being conceptualised as non-economic misfits.

The second narrative emphasises economic vibrancy and potential. A feature of this thinking, rooted in the De Soto (2000) thesis, portrays informal settlements as ‘housing’ petty entrepreneurs and pre-entrepreneurs in the waiting (the latter's exclusion is because they lack access to formal credit). This strand of entrepreneurialism puts more emphasis on agency, whereby exiting the formal economy can be a strategy, or the only option, for immigrants to reduce operating costs and to avoid registration, labour regulations, large rental payments, and so forth. Some preliminary pieces of evidence of these entrepreneurial efforts include informal firms operating in quasi-industrial areas around informal settlements and entrepreneurs hawking, for example, pirated DVDs by the side of the road and selling imbada (local shoes made out of recycled tyres) to dance troupes and tourist craft shops (Zeilig & Ceruti, 2007). This type of work can also be conducted within subcontracting networks that reach and extend deeply into informal settlements, incorporating workers into outside businesses (Business in Africa, 2006:3). Studies show that this happens when entrepreneurs from the vicinity operate as labour brokers and when independent contractors function as part of a larger network (Zeilig & Ceruti, 2007). In general, the entrepreneurial thesis concentrates on discovering informal interrelationships and interdependencies that come from backward linkages to formal entities.

The third narrative, a variant of the entrepreneurial thesis, emphasises self-employment through using the home as a base (typically referred to as home-based enterprises [HBEs]) (Gough et al., 2003). The intermingling of domestic and work space is common throughout low-income areas; around 13 per cent of the Soweto population may be involved (Gordon & Nell, 2006:viii). Advantages of operating an HBE are low start-up costs, use of space in the home to maximise economic benefits, removal of the time and costs associated with work travel, and effective use of human capital (friends, relatives and neighbours) in exchange for remuneration or benefits in kind (Rogerson, 2007). HBEs typically involve small-scale production and service enterprises possible in limited space. A number of different activities, mostly with backward linkages, are highlighted (Gordon & Nell, 2006; Rogerson, 2007). There are retailers, typically selling food items; there are producers, making clothes, preparing food, constructing gates and fences, building shacks; there are people offering services, running hair salons and shebeens, doing auto repairs; there are traditional healers, selling medicine and treating patients at home. In general, space constraints make it difficult to expand the business beyond the immediate vicinity, so HBEs are characterised by very local linkages.

The fourth narrative centres on the historical legacy of apartheid policies and their spatial manifestation in townships, which had a specific purpose during the apartheid regime as remote dormitory areas for people working in the city. Townships were non-generative economic spaces, although under-the-radar businesses always operated (Webster, 1984). This narrative emphasises the continuation of adverse incorporation in the post-apartheid era (Du Toit, 2004), in which the geography of apartheid-era work patterns remains intact. Moreover, high population growth in townships adds further pressures and increases competition for poorly paid service jobs, making unemployment and poorly paid work more interdependent than ever before. However, at the same time a subset of the population composed of recent international migrants – some of them more skilled than the local people and willing to work for lower wages – creates another world of work under the radar. In general, this historical legacy thesis emphasises the inertia of geographical work patterns (even if the workers have changed), a labour geography at ‘the beck and call of individuals who demand chores and favours’, necessitating extensive travel for work (Sitas, 2001:2).

Finally, the fifth narrative focuses on the household, the family and the local community, emphasising self-help and mutual aid and the household as a site of refuge (Webster, 2005). Welfare economies are significant to low-income households that qualify for support. Qualifying households share and sometimes pool resources (shelter, wages, income from state grants such as old-age, child and disability pensions, etc.). Investments are made in friends and in some instances these relationships are considered ‘work’ (Lyons & Snoxell, 2005:1093). Work is often paid in kind. There can be considerable family and community pressures to involve friends and family in business activities based on community rather than on business logic. In some instances, links can extend to rural or international households in support chains, and mutual arrangements can even allow for poorer rural relatives with AIDS to hide away in urban homes (Webster, 2005:64). The existing literature on Soweto uncovers ‘few prima facie links … indicative of social capital (i.e. deriving from time spent in local formal groups such as churches, burial societies and the like)’ (Piazza-Georgi, 2002:638). However, it may be that while trust in strangers is low, trust in neighbours is high and networking beyond the home community is significant (Jooste, 2005).

4. Study sites and methods

Soweto, an amalgamation of numerous different townships, has become an integral planning unit for the City of Johannesburg. It was incorporated into the Greater Johannesburg Metropolitan Area in 2001. Officially, its economy has been shown to have few connections with the wider space economy of Johannesburg (City of Johannesburg, 2008a). According to official estimates (largely based on formal economic activities), Soweto contributes only 4 per cent of the city's total economic activity even though Sowetans make up 43 per cent of its population (City of Johannesburg, 2008a:1). Furthermore, as a result of the constraints enforced by apartheid spatial economic planning, there is little connectivity between areas inside the township and no internal urban hierarchy. City of Johannesburg (2005b) descriptions of economic nodal activity in Soweto identify specialised businesses in various industrial estates and scattered ribbons of small-scale economic development. Officially, 173 business areas are identified (City of Johannesburg, 2005b:Annex 3).

Soweto is a diverse urban residential area (Crankshaw et al., 2000; Ceruti, 2007), ranging from upper middle income to very low income areas. Shacks are ubiquitous in the latter – 200 000 people live in backyard accommodation in Soweto (Crankshaw et al., 2000:845), and the informal settlement areas consist almost exclusively of shacks, although some areas, such as Kliptown, have older brick houses, now encircled by shacks. Data on informal settlements will always be disputable: 55 066 families have been counted officially in Sowetan informal settlements (City of Johannesburg, 2005a:2). Settlement profiles also record low-income residents in council houses (these dwellings account for 57 per cent of all units) (Crankshaw et al., 2000:845). In the council housing category, newer Reconstruction and Development Programme (RDP) housing areas have been created since 1994 to alleviate some aspects of poverty by providing improved shelter to many informal settlement residents. For instance, at the time of writing the City of Johannesburg (2004:47) had plans for improved housing for 48 700 informal settlement families in Soweto. However, RDP-subsidised housing does not target informal settlers exclusively, and many informal settlers who qualify for a council house have yet to secure a unit in a rehabilitated area. Figure 1 shows the areas of Soweto covered by the present study.

Figure 1: Location of study areas in Soweto

In 2003 there was a new policy emphasis on economic development and internal employment opportunities in townships (Endres & Schenker, 2007) and especially on retailing in the early stages (Ligthelm, 2007). A range of projects, such as tarring roads, planting 60 000 trees and constructing cycle lanes, parks and new recreation spaces, improved the aesthetics, health and safety of the general urban environment, and this improvement was further enhanced by a City of Johannesburg township economic programme for Soweto in 2008. For example, the 2008/09 budget earmarked R200 million for projects to stimulate the township economy (City of Johannesburg, 2008b:39). City policy-makers aimed to promote ‘the economic competitiveness of the township as a viable urban location’ and to build Soweto ‘into a productive and sustainable local economy’ (City of Johannesburg, 2008a:4). The LED framework is being used to transform Soweto by ‘harnessing the local resources of the township’ as a ‘basis for creating locally driven urban economic development’ (City of Johannesburg, 2008c:3).

It is difficult to sample informal settlement areas accurately because comprehensive data are collected only for higher priority areas for redevelopment (e.g. Kliptown) and because settlements are always in flux. The City of Johannesburg (2005a) delineated 39 Soweto informal settlements using aerial photography, but debates about classifications and delineations are ongoing. Together with the survey team leader, I undertook site reconnaissance of all Soweto informal settlements. These varied in size and locational attributes (some were in more hazardous areas than others) and in terms of the presence or absence of informal businesses. The study sample was chosen from those that showed evidence of informal businesses and local work possibilities, and to cover a spread of Soweto that avoided the bias of the chosen settlements being either all close to or all distant from transport nodes. More established settlements (unlike the newer areas, which are more likely to be populated by refugees) were selected to ensure that residents had had ample time to create networks.

Where there was more than one designated informal settlement in an area, the largest was included in the study. For example, in Kliptown's 10 different settlements the number of shacks ranged from 200 in Winnie Camp to 2000 in Chris Hani (Chicken Farm), so the latter was chosen. Sampling from the largest settlements prevented an over-representation of small settlements and an under-representation of large ones, but correcting weights were not applied to each settlement because settlement size and boundaries could not be determined in all cases. To capture a geographically extensive mix of areas, six prominent Soweto informal settlement areas in various City of Johannesburg-defined business areas were selected: Kliptown, Klipspruit, Emdeni, Dobsonville (Snake Park), Ellis Motsoaledi and Midway. Besides these, two rehabilitated housing areas (Dobsonville and Braamfisherville) were chosen to compare their businesses with those of the informal settlements. Again, the selection was based on knowledge of considerable informal economic business activity. In both rehabilitation areas, residents had been relocated from informal settlements throughout metropolitan Johannesburg and given new housing in RDP-subsidised housing estates. The RDP areas are well-serviced in terms of infrastructure, whereas the informal settlements are not. Here, water is mainly obtained from a household yard tap, residents must use communal mobile chemical toilets, and not all have electricity, or a reliable supply.

A pilot test of a household survey took place in September 2008. With the assistance of a research coordinator and a team of 10 trained multilingual fieldworkers, a structured questionnaire comprising 66 questions was administered in November 2008 until we had obtained 40 complete household questionnaires in each of the housing areas. In all of the areas, we undertook more than 40 surveys to ensure data quality. We did not encounter any refusals, perhaps because we paid a modest sum to obtain a completed questionnaire. Paying respondents for their time has obvious drawbacks and meant that we had to vet the completed questionnaires returned by the field assistants. Nevertheless, we had ample time to examine questionnaires in detail and in specific cases we eliminated some from the final survey pool. For instance, seven surveys were too incomplete (20 of 66 questions were unanswered), six contained obvious inaccuracies (e.g. the respondent claimed to be from Johannesburg but all evidence suggested this was an international migrant) and one survey was duplicated. This paper therefore reports on 320 acceptable completed questionnaires.

The housing units were sampled randomly by rows of numbered houses in the RDP areas and by numbered stands or rows in the informal settlements. For example, we randomly sampled three houses in each row until we had worked through all rows of the settlement. The starting row was randomly selected to ensure that we did not bias the selection toward the perimeter or interior of the settlement. I conducted follow-up interviews with 20 respondents in February 2009 (10 men and 10 women). At least two interviewees were selected in each area, and four interviews were conducted opportunistically when interviewees wanted to tell their story. These 20 interviews focused on assessing economic trust in work and on identifying examples of activities that qualified as ‘work’ for the purpose of the survey in the home, the residential area and the workplace.

5. Examining the meaning of ‘work’ in the informal settlement context

Most assessments of work examine formal employment and unemployment and miss other types of work. Various categories of work in the Soweto study areas are presented in Figure 2, which shows that only 5 per cent are full-time workers at formal firms or organisations (such as the City of Johannesburg). Unemployment is widespread: 37 per cent of household heads do not work (one-quarter are well outside the workforce, having not worked for 5 years or more). Among the remaining 58 per cent (and missing from most analyses), one-third have other employers (sometimes these are firms but often non-firms, such as households), while the rest are self-employed casual workers (such as part-time manufacturers, or shelf-stockers) and those who could be working but are currently inactive (retirees and full-time students).

Figure 2: Geographies of worlds of workers

Categories of work performed at each of the spatial scales – in settlements, in the neighbourhood, in the city, outside the city – indicate considerable circulation and ties: 10 per cent work exclusively in the home, 11 per cent in the neighbourhood, 23 per cent elsewhere in Soweto and the largest group (56 per cent) in Johannesburg. An important finding is the diverse and multiple arenas in which individuals work. There is strong evidence to support the apartheid legacy thesis: this accounts for significant non-local work. More than half travel to Johannesburg for work, mostly to the Central Business District and the wealthier northern suburbs. Rail and the frequency of informal taxis out of Soweto to all the surrounding areas make commuting possible, and the average commute time is 1 hour. Almost one-quarter work at home or in their neighbourhood, showing evidence of efforts at home-based and locally oriented entrepreneurial endeavours. The recent economic boom in Soweto provides others with work opportunities, particularly in the construction and security industries. Only a very small number move from place to place to work, which suggests that the poorest workers from Soweto do not dominate street trading and also that hawking has been overtraded for some time (Rogerson, 1999).

Those who work do so at a range of firms and non-firm entities (see Figure 3). Not surprisingly, informal firms employ the most people; more importantly, 10 per cent of these workers did not know the status of their employer, which they presumably would do if they worked for a formal employer (unless they were doing casual labour). Twenty per cent were self-employed, on a continuum from working alone as a sole proprietor to having a few employees. Self-employment was not linear: many switched between self-employment and other part-time work. It is not uncommon for the self-employed to change professions periodically, depending on opportunities (sometimes in the same industry, e.g. from plumber to bricklayer, but often in different industries, e.g. from auto mechanic to security guard). Some self-employed men mentioned ‘inventing’ jobs, such as urban gardening (growing vegetables in urban spaces such as backyards and vacant stands), as a new sideline entrepreneurial effort that had yet to materialise fully. A man from Elias Motsoaledi justified this by saying ‘it is easier to obtain work when you have something to do’, and ‘going to work eases tensions and lifts spirits at the house’. By far the largest numbers are employed in household service jobs. Overall, the large number of people working in non-firm circumstances reaffirms the need for more attention to non-firm contexts in policy debates.

Figure 3: Work in firm and non-firm entities

6. Examining place and sectoral work niches

Figure 4 shows the areas where Soweto residents work – both close to home and in the wider space economy. Work is subdivided into employment niches and the places where the work is done – home, block or neighbourhood, elsewhere in Soweto, elsewhere in Johannesburg or a variety of locations used by mobile workers. Diverse work geographies emerge.

Figure 4: Place and sectoral niches for work

The finding that workers engage the urban economy at all spatial scales once again supports the need to combine various working space information and knowledge rather than to deploy a single spatial narrative. Several sectoral and spatial niches are evident. Work involving retail and food and beverages is heavily home oriented. Individuals in domestic services, business services, tourism and transport work mainly outside their neighbourhood, particularly in other parts of Soweto or elsewhere in Johannesburg, or both. Domestic workers and gardeners work mainly in the northern suburbs of Johannesburg. Construction and transport workers are the most mobile, and they are often hired by various employers; taxi drivers, for example, report different work arrangements in different parts of the city on different days of the week. Some of the more skilled construction workers, such as electricians and plumbers, appear to be moved from site to site by contractors more than general construction labourers who have regular arrangements. Other construction workers say their work is unstable, necessitating daily or weekly work searches on the track of new building investments.

Manufacturing and retailing workers engage all arenas in the wider space economy. Manufacturing activities (furniture, arts and crafts, clothing, refurbishment of engines, silk screening, shack construction, window frame and brickmaking) are significantly concentrated in the vicinity of residential areas. Labourers find work in local informal enterprises in industrial estates, such as Emdeni and parts of Orlando East. These estates, originally manufacturing incubator sites (Rogerson & Da Silva, 1988), have only partly succeeded in transforming informal firms into formal ones and over the last decade and one-half have moved back into informal manufacturing. The Emdeni estate, in particular, draws workers from an adjacent informal settlement, who walk to panel beating and gate making jobs through a gap in the back fence. Other informal manufacturing workers find work in nearby workshops, backyard shacks, reconverted buildings and other business clusters close to shopping centres.

Of the workers in this study, 20 per cent said they engaged in multi-sited work even though they mentioned only main employer. Multi-sited work is distinct from mobile work in that workers have a fixed base or bases. They split tasks between two sites (e.g. making crafts at home and at a fixed market site), or engage in multiple jobbing, or do part-time work that shifts according to employer need (in the case, e.g., of security guards who reported working in different sites in Soweto for one or more employers). In most cases and in most sectors (except tourism and transport, where work locations are most irregular and unpredictable), the different kinds of work that are combined are done within the vicinity of each other. For instance, housekeepers work full-time for one employer and work extra hours for others nearby.

In the study sample there were manufacturing workers who worked 3 or 4 days at one firm and did extra time at another firm close by. There were labour pooling arrangements, in which the more skilled and experienced workers were involved in large orders where specialised parts of the work were shared among firms (e.g. in the auto repair industry, a mechanic who repaired cylinders 3 days a week was loaned to a different and nearby firm that fixed crank shafts on other days). There were cases of workers in printing being shared among firms depending on the ebb and flow of work orders. Some printing firms used personal networks that had developed from bonds formed during apprenticeships. Such relationships between firm owners make it possible to trade workers. Workers also reported cases where different work arrangements are made at their own request. These often involved working for a family member or friend as well as working for an informal firm or non-firm employer. Responses to questions about the locations of various jobs revealed that the largest group did multiple jobs (especially moonlighting) in areas near their main place of work, and most of the others work at different sites only in their home area. Using the home as a workplace eased some of the time pressures when distance was involved in different working places, but some said they had great difficulty moving between multiple jobs.

7. Examining working space from home outwards and creating social capital in work

For many, home is a place that links workers and non-workers. Almost half of the households interviewed include various mixes of the unemployed, employed, self-employed and casually employed. And in low-income households there is no separation of working and non-working space (Webster, 2005); instead the poor and the poorer live in propinquity and remain connected in a number of spheres (such as finding work, budgeting and managing economic crises).

To discover the more complex interrelationships within households and how social capital was created, I relied on personal interviews. No doubt social capital is created in many places, but interviewees claimed they could tell which relationships were useful for work and which were not. Murphy (2003) found the opposite for informal entrepreneurs in Tanzania. In Soweto, interviews revealed that relationships are cemented (for better and sometimes for worse) through households devising several strategies to amplify income: renting rooms and space (when available), ensuring that household members qualify for state grants, and pooling resources to engage in micro entrepreneurship. Many of the interviewees said it was their combining of different work efforts, and often in various locations (at home, or in one workplace or several), that stretched working space out from the home. A gender difference became evident in that women expressed a much stronger preference for combining work and home in close proximity, because of domestic and child-related responsibilities. As highlighted in the literature on multiple livelihoods (e.g. Owusu, 2007), multiple jobbing often involves working the same sector in different locations (e.g. a health professional working at a health facility during the day and making house calls at night). Extended work efforts are responses to immediate family financial responsibilities, some of which also involve raising funds to cover the costs of relatives' work searches.

Workers maintain local residential ties as much as possible while they create and build new relationships. Socialising with neighbours, attending church services and participating in burial societies are typical social activities with high levels of participation and commitment, but no worker attributed a job to these ties alone. The concept of ubuntu (which means that a person is a person through other people, through being part of a collective) was often mentioned as a principle to live by but increasingly perceived as a major challenge because of the harsh, competitive and individualistic urban economy. Despite high levels of local networking activities, workers said they were uneasy about separating their social and professional selves into informal and formal local group participation. They had strong doubts as to whether they received any help in finding work from local groups, organisations, clubs, and so forth. Men were unanimous on this point. As one Emdeni man put it, ‘I am in the do-it-yourself economy’. By contrast, four women received help from other women in the same neighbourhood. For the most part, men and most women interviewees emphasised a decoupling of interpersonal relations within and outside residential areas. One Braamfisherville woman noted that ‘when it comes to money and work, few can be trusted outside your inner circle’. This kind of sentiment was reflected in interviewees' strong preferences for developing more personalised networks.

On the surface, interviewees were dismissive of any notion that locally expressed ethnic solidarity was pertinent for securing work (in fact, many claimed that international immigrants were obtaining much of the local work). An Emdeni man claimed that only African National Congress political connections mattered, noting that ‘if your granddaddy was Mandela's shoeshine boy then you get a job – if he wasn't, then you have to create new contacts from scratch’. Two people reported that locals might have even sabotaged a work opportunity for them by bad-mouthing them and undercutting them by encouraging someone to work for less (allegedly contacts received a payment-in-kind for brokering this deal). Still, there were many (more women than men) who were convinced that locals had helped them get work in the past, particularly among the self-employed. Employed workers said locals had helped them get their first job and helped them learn on the job (Lyons & Snoxell, 2005, found similar evidence among Nairobi traders). When it came to subsequent jobs, workers mentioned a different scale of operation, emphasising newer contacts they had made outside a residential area, around their place of work, with others from Soweto and Johannesburg, particularly those more on a professional than social level. Male workers remarked that over time the area beyond their homes, where they had forged new personal ties, became the main circuit in their world of work, whereas women's ties remained neighbourhood centred.

Most interviewees deliberately extended their circle of friends, and indeed made this a priority. Besides calling on and familiarising local employers with themselves, workers emphasised workplaces as important venues for making friends that could be helpful in keeping a job or hearing about new and better positions. Opportunities to enlarge social circles presented themselves as workers travelled to and from work and spent time together waiting for taxis and trains. Public meeting places along work routes often became important locales for meeting and talking about work. At the same time, some did not trust all co-workers, viewing them as competitors. Most claimed to be able to know instinctively (through interactions, reflections about stories shared, common experiences, mannerisms, etc.) when a work contact could be trusted; when pressed as to how to validate this, interviewees responded that reciprocity was the proof. Yet one must note that workers were not always able to engineer other work possibilities and that working space interactions were not always productive. Some mentioned ways that work came their way by chance. It was not unusual for individuals to be phoned and told about work.

There is evidence that some informal subcontracting networks reach into low-income areas; however, no individual reported doing work at home for a formal firm, as has been found in other South African settings (Kenny, 2001). Instead, there is a network through which workers are recruited by known labour brokers from Soweto. A few did not know who exactly had hired them, but more women than men stated that they would not trust someone they did not know. This process typically involved a home visit followed by phone exchanges about possible positions and remuneration. Most brokers appeared not to charge any fees (presumably the fees were paid by the employer). However, a few workers mentioned paying job-finding fees in advance, and others reported that payment-in-kind was expected after pay day. Some said work was slow after initial recruitment but tended to pick up once trust was established among the parties involved. Trust seemed to develop best after two or three face-to-face meetings and after two or more part-time jobs had been completed. Most interviewees said it was better ‘not to ask any questions’, ‘to mind one's own business’ at the place of work and to trust that someone was looking after you. Casual workers reported seeing ‘eye-to-eye’ with labour brokers on the basis of some shared understanding on how to find work when one had neither the right address nor government connections. Sowetan labour brokers were described by one male interviewee from Midway as ‘people with similar backgrounds and experiences who think the same, who are big and well-connected’ as well as ‘knowledgeable about where and when to find work’.

8. Conclusions

The study findings raise questions about the wisdom of relying on any single spatial narrative about work done by low-income residents in Soweto. They also encourage us to consider a wider and more robust role of poor workers in the space economy. Narrow conceptualisations of workers become magnified when we take into account the spatial selectivity embodied in dominant spatial narratives. The more accurate representation is that residents work it out in complex ways: weaving a strong web of different work activities, both formal and formal. Importantly, the workers' space of flows is more complex than that described in firm-centred analogies.

An assessment of the places where the poor work revealed intermingled spaces (within firms, between individuals, within subcontracting networks and so forth). Du Toit & Neves (2007:10) assert that workers use networks, routes and ‘wormholes’ to connect distant spaces, while employers, in turn, draw from the abundant labour pool and deploy labour within their spaces (in houses, firms, workshops, etc.). This paper maintains that poor residents' working spaces are more extensive than those identified in any single spatial narrative. A finer tuned understanding reveals the importance of examining all aspects of work and location specificities in tandem as well as the parts of various work networks. Understanding work geographies requires us to hold more than one locale, more than one economic unit and more than one impulse in our geographical imagination at the same time. This vortex spatiality of workers (sucked into central Johannesburg and the northern suburbs) is not captured by current firm theory and policy: the overemphasis on firms is misleading for understanding workers.

The findings confirm that every place (home based enterprises, informal settlements, rehabilitated areas, informal business clusters) has an economy and that even marginalised poor residents are connected to wider space economies (albeit incompletely incorporated) (Du Toit, 2004). The working poor use particular locales and informal circuits in and around their own locales to enhance work possibilities. This entails a localisation of informal networking in meeting places that operate as frontiers for working space: places that are distinct from the typical (official) employment and entrepreneurial advice centres and from the standard social group locales (within family, ethnic, kin and local organisations settings) emphasised in the African social capital literature. The way the working poor use space often entails a local capital of resources being accessed and mobilised with local development potential. We can consider this capital to embody an intrinsic value through its collectivity. Like all forms of capital, it is enhanced by use and accumulated slowly, but can erode quickly given its informal underpinnings. It is inherently unequal, and sometimes it is its unevenness that allows this form of capital to take shape: interviews confirmed that it takes shape within a habitus 2 and that personal work networks are important.

Townships originated out of political circumstances centred on a particular incorporation of workers (not firms) into urban economies, and this incorporation continues to be largely non-firm based. Without proper acknowledgement of this spatial reality, township economic development planning will be misguided. Policy interventions are fumbling because they are heavily firm-focused: current policies can be anticipated to have little impact on the many working in non-firm circumstances. It is unreasonable to expect firm-based policies to work, given the historical absence of firms from townships. Moreover, economic theory about firms, entrepreneurship and growth has been developed on the basis of non-apartheid conditions and cannot be transposed to townships without taking into consideration existing patterns of work that involve heavy worker leakage. It is only by developing policy interventions that are based on the disaggregation of township workers that policies can be effective in transforming townships. It is evident that township economies are not being conceptualised adequately: the everyday experiences of workers remain under-examined in existing theory and policy.

Acknowledgements

This research was funded by a US National Science Foundation grant (number 0721025).

Notes

¹In Gauteng, for example, the number of informal settlements increased steadily from 47 in 1990 to 200 in 2006 (Murray, 2008:94) to 500 in 2010.

²A system of dispositions, expressed in everyday living, seeing, acting and thinking, common to people who share similar positions in social space (Bourdieu, 1977).