Domestic Work and the Gig Economy in South Africa: Old wine in new bottles?

Based on innovative, mixed-methods research, this article examines the entry of on-demand platform models into the domestic work sector in South Africa. This sector has long been characterised by high levels of informality, precarity, and exploitation, though recent regulatory advances have provided labour and social protections to some domestic workers. We locate the rise of the on-demand economy within the longer-term trajectory of domestic work in South Africa, identifying the ‘traditional’ sector as a key site of undervalued labour. On-demand domestic work platforms create much-needed economic opportunities in a context of pervasive un(der)-employment, opportunities that come with some incremental improvements over traditional working arrangements. Yet we contend that platform models maintain the patterns of everyday abuse found elsewhere in the domestic work sector. These models are premised on an ability to navigate regulatory contexts to provide clients with readily available, flexible labour without longer-term commitment, therefore sidestepping employer obligations to provide labour rights and protections. As a result, on-demand companies reinforce the undervalued and largely unprotected labour of marginalised women domestic workers.


Introduction
The gig economy, in which Uber-like digital platforms unite workers and purchasers of their services, is expanding globally. The model requires workers to perform task-based 'gigs', mediated through digital platforms, without the security or benefits usually associated with formal employment. 1 Though exponential growth is forecast in traditionally female-dominated sectorsnotably on-demand household services including cooking, cleaning and care work 2 -relatively little research to date has focused on gendered experiences of gig work outside of North America and Europe. 3 This article discusses ondemand domestic work in South Africa. It explores platform models' effects on working conditions, their impact on the three key constituents of the gig economy (workers, platform companies, and clients), and the implications of their rise for the valuation of domestic work.
Domestic work is persistently undervalued in South Africa (as elsewhere), where it is overwhelmingly the preserve of poor black African women. However, the domestic work sector is relatively large, occupying 6 per cent of the country's workforce, 4 and advocacy by unions and allies has led to incremental improvements to the regulatory framework governing the sector. Though these regulations are neither comprehensive nor generous-the relatively low entitlements they stipulate reinforce the marginal status of domestic workersthey have given advocates a foundation from which to argue that working conditions could be further improved through additional formalisation. The 1 See V De Stefano, 'The Rise of the "Just-In-Time Workforce": On-demand work, crowd work and labour protection in the "gig-economy"', Comparative Labor Law and Policy Journal, vol. 37, issue 3, 2016, pp. 461-471; A Hunt and E Samman, Gender and the Gig Economy: Critical steps for evidence-based policy, Working Paper 546, Overseas Development Institute, London, January 2019, https://www.odi.org/sites/odi.org. uk/files/resource-documents/12586.pdf.

2
Projections by PricewaterhouseCoopers, for example, forecast that 'on-demand household services will be the fastest growing sector' of the gig economy in the European Union (EU), 'with revenues estimated to expand at roughly 50% yearly through 2025'. (See J Hawksworth and R Vaughan, 'The Sharing Economy-Sizing the Revenue Opportunity', PricewaterhouseCoopers, 2014, http://www.pwc.co.uk/issues/ megatrends/collisions/sharingeconomy/the-sharing-economysizing-therevenueopportunity.html; R Vaughan and R Daverio, Assessing the Size and Presence of the Collaborative Economy in Europe, Publications Office of the European Union, 2016, cited in Hunt and Samman, p. 10.) 3 On-demand services are provided locally, with the purchaser and provider in geographic proximity (in contrast to crowdwork, which takes place online).

4
'Quarterly Labour Force Survey: Quarter 3', Statistics South Africa, Pretoria, October 2018, retrieved 13 July 2020, http://www.statssa.gov.za/publications/P0211/ P02113rdQuarter2018.pdf. ostensibly different operating model underpinning the gig economy has the potential to undermine this effort, and thus it is important to understand the impact of its entry into the domestic work sector.
We begin by examining the characteristics of traditional and 'on demand' domestic workers, and then explore the undervaluation of domestic work within South Africa. We argue that while on-demand platforms offer some improvements to workers over traditional employment arrangements, their goal of facilitating flexible labour leads to the continued normalisation of the labour exploitation of domestic workers. We conclude that both models undervalue domestic labour and perpetuate breaches in workers' labour rights, leaving workers in a highly precarious position.

Characteristics of Domestic Workers
Traditional and on-demand domestic workers share many common characteristics. This is unsurprising, given that many on-demand workers have previously worked under traditional domestic work arrangements or have continued with traditional work alongside platform-mediated gigs. The domestic workforce is overwhelmingly comprised of poor black African women. 5 Indeed, 98 per cent of our survey respondents were female and 97 per cent were black African. 6 Migrant workers from South Africa's rural areas or from adjoining countries furthermore form a significant share of the paid domestic workforce, especially in its less formalised segments. 7 Finally, we should note that domestic workers are relatively young, although we found that platform-based workers are on average slightly younger. In our sample, on-demand workers had a median age of 35, while traditional domestic workers had a median age of 41. 8 5 L Orr and T van Meelis, 'Women and Gender Relations in the South African Labour Market: A 20-year review ', Labour Research Service, Cape Town, 2014, pp. 2-27. 6 See footnote 29 below for details of the survey methodology, including the sample size.

The Undervaluation of Domestic Work
The historical undervaluation of domestic work is evident in both the traditional and on-demand models. At its core, this undervaluation stems from the gendered way in which the dominant economic concepts of 'productive' and 'unproductive' work are differentiated. Unpaid domestic work-mostly carried out by women-is categorised as an element of the 'household and care economy', and as a result domestic work is not seen as having intrinsic economic value. This undervaluation persists when domestic work is commodified: 'The gender stereotyping of unpaid care work, and the association of care with women's "natural" inclinations and "innate" abilities, rather than with skills acquired through formal education or training, lies behind the high level of feminization of care employment'. 9 Consequentially, 'the fact that women's unpaid domestic work has been undervalued has had a negative impact on the salary and working conditions of remunerated domestic workers'. 10 In other words, paid domestic work, which is disproportionately carried out by women, is perceived as an extension of the unpaid work within the household. This has contributed to the frequent exclusion of domestic work from formal labour relations frameworks, and therefore to its perception as 'undeserving' of good working conditions, including decent remuneration.
The undervaluation of domestic work relates to the characteristics of domestic workers, who are typically marginalised and subject to intersecting inequalities alongside continued systematic discrimination. These women experience sites of power disparity that go beyond gender to include race, migratory status, and social class. 11 In South Africa, 'the low social status and undervalued nature of domestic work has roots in the historical use of specific racial and cultural groups as servants and slaves', 12 exacerbated by the racialised nature of relations 9 L Addati, U Cattaneo, V Esquivel, and I Valarino, 'Care Work and Care Jobs for the Future of Decent Work', International Labour Organization (ILO), Geneva, 2018, p. 8, http://www.ilo.org/wcmsp5/groups/public/---dgreports/---dcomm/---publ/ documents/publication/wcms_633135.pdf. 10 G Labadie-Jackson, 'Reflections on Domestic Work and the Feminization of Migration ', Campbell Law Review, vol. 31, issue 1, 2008, pp. 67-90, p. 82. 11 On intersectionality in the context of domestic work in South Africa, see D Gaitskell, J Kimble, M Maconachie, and E Unterhalter, 'Class, Race and Gender: Domestic workers in South Africa ', Review of African Political Economy, vol. 10, issue 27/28, 1983, pp. 86-108. On migratory status, see L Griffin, 'Unravelling Rights: "Illegal" migrant domestic workers in South Africa', South African Review of Sociology, vol. 42, issue 2, 2011Sociology, vol. 42, issue 2, , pp. 83-101, https://doi.org/10.1080Sociology, vol. 42, issue 2, /21528586.2011 D du Toit and E Huysamen, 'Implementing Domestic Workers' Labour Rights in a Framework of Transformative Constitutionalism', in D du Toit (ed.), Exploited, Undervalued -and Essential: Domestic workers and the realisation of their rights, Pretoria University Law Press, Pretoria, 2013, p. 79. between black domestic workers and their white 'madams', 13 a pattern that has lingered despite Apartheid ending. 14 With growing black affluence, class has featured more prominently in domestic worker-employer relations. Nonetheless, '[t]he result of the complex interplay between gender, race and class is, in many cases, a perception amongst employers that the domestic worker is a lesser creature'. 15 One outcome is the persistence of paternalist relationships between domestic workers and their employers. 16 Another is that the mobilisation of women, which is 'generally a necessary condition for changes in care-related policies', becomes less likely. 17 The location of domestic work also contributes to its undervaluation. Domestic work is largely conducted by isolated workers in the private sphere, which makes worker organising for better conditions with employers or stronger government regulation more difficult. 18 It should be clear from this analysis that there are many obstacles to raising the value of domestic work in South Africa. They go well beyond the gig economy. However, we argue that existing intersecting inequalities, discrimination, and power differentials tend to be reinforced in the on-demand economy, deepening the existing analysis of domestic work and care platforms in the However, see also S A Ally, From Servants to Workers, South African domestic workers and the democratic state, University of KwaZulu-Natal Press, Scottsville, 2010. Ally argues that the post-Apartheid state's regulation of domestic work 'depersonalised' employeremployee relations, thereby threatening domestic workers' use of personal relations to negotiate their working conditions, which is itself a unique characteristic of the nature of domestic work.
United States 19 as well as India, Kenya,Mexico,and South Africa. 20 In the remainder of this article, we demonstrate that the current operating model of platforms in South Africa is likely to perpetuate the labour exploitation of domestic workers. The next section explores working conditions for traditional domestic workers. This is followed by our analysis of the emergence of the ondemand economy. This section includes our methodology, outlines our empirical findings on labour conditions within the on-demand sector, and analyses the 'winners and losers' under this new model. We conclude with reflections on the policy implications of this research.

Labour Conditions in the 'Traditional' Sector
Working conditions for domestic workers in South Africa have been historically poor, characterised by informality and exploitation. There have, however, been recent attempts to improve the situation. Unions such as the South African Domestic Service and Allied Workers Union (SADSAWU) have been leading sustained campaigns for decent wages and adequate workers' protection. Government attempts to establish a regulatory framework include the introduction of 'Sectoral Determination 7' in 2002, which mandated a minimum wage and basic working conditions such as formal employment contracts and the compulsory registration of workers with the Department of Labour-a change that enables them to benefit from the Unemployment Insurance Fund (UIF). In 2013, South Africa ratified ILO Convention 189 on Domestic Work, setting a new benchmark for improved conditions in the sector based on the key pillars of 'decent work'. These include recognition of domestic work as 'real work', formalisation through contracts, adequate wages, social protection, health and safety in the workplace, and rights to organising and social dialogue. 21 In 2018, the Department of Labour proposed extending workers' compensation to domestic workers, and in May 2019, the Pretoria High Court ruled that their exclusion from the Compensation for Occupational Injuries and 19 N van Doorn, 'Platform Labor: On the gendered and racialized exploitation of lowincome service work in the "on-demand" economy', Information, Communication & Society, vol. 20, issue 6, 2017, pp. 898-914, https Diseases Act of 1993 (COIDA) was unconstitutional. As of early March 2020, the Constitutional Court began proceedings over whether to instruct Parliament to amend COIDA to include domestic workers. 22 Nonetheless, significant challenges remain. While the National Minimum Wage Act of 2018 specified a minimum hourly wage of ZAR 20 (approx. USD 1.40 at the time), as of January 2019, the minimum wage for domestic workers was set at only 75 per cent of the national minimum. 23 SADSAWU and other labour rights organisations continue to highlight the insufficiency of this wage to meet the cost of living, as well as its symbolism for the undervaluation of domestic work vis-à-vis other forms of work to which a higher minimum wage applies.
Implementation of regulation also remains patchy. An unknown (but presumably sizeable) number of domestic workers continue to work informally, and several categories of domestic workers remain excluded from social protection provisions. 24 For example, one recent estimate suggests that approximately one-third of the domestic workers who work the requisite 24 hours or more per month remain unregistered with UIF. 25 These include foreign individuals working on contracts, as well as individuals employed for less than 24 hours a month by a single employer-a key barrier given that many domestic workers work part-time for multiple employers. 26 Employer non-compliance and domestic workers' limited awareness of their rights further impede 22 'Why the Concourt Case for Domestic Workers is So Important -for Employers Too', Eyewitness News, 10 March 2020, https://ewn.co.za/2020/03/10/why-theconcourt-case-for-domestic-workers-is-so-important.

23
The Sectoral Determination of Minimum Wages for Domestic Workers (December 2018) adds detail based on location and weekly hours worked. As of 20 March 2020, the minimum wage was raised to ZAR 20.76 for most workers and ZAR 15.57 for domestic workers. See 'This is South Africa's New Minimum Wage', Business Tech, 18 February 2020, https://businesstech.co.za/news/finance/374890/this-is-southafricas-new-minimum-wage. 24 du Toit.

The On-Demand Economy: Labour conditions, winners and losers
The entry of digital platforms into the domestic work sector in South Africa builds upon an established model while also adding new features. Compared to other options open to domestic workers (notably but not exclusively traditional domestic work arrangements), digital platforms offer some positive features that workers value and which improve their working conditions. However, workers also identified several ways in which the on-demand model perpetuates their precarious working conditions.
The data that informs this article was collected as part of a broader two-year research project exploring gender and the gig economy in Kenya and South Africa. 29 In South Africa, novel methods of data collection included a nineround, automated voice response (AVR) survey with workers active on a domestic work platform, and the analysis of data from this same company. 30 It should be underlined that while the platform provided access to its data and contact information for registered workers, the study was fully independent: the survey was conducted through an independent company who secured consent from workers to survey them and anonymised the data that was collected. The  Nearly 650 workers (around one-third of the total) who were on the platform as of August 2018 responded to an invitation to complete the first round of a survey covering their background and motivations for engaging with the platform. We could not investigate self-selection into the survey nor non-response comprehensively because, for privacy reasons, the platform deliberately collected minimal personal details regarding registered workers. Subsequent response rates varied between 25% and 42% for the first five rounds of the survey, after which we had to reduce the sample size due to budgetary restrictions and a rise in the price of mobile phone airtime. See Hunt et al. for more details. study also involved qualitative interviews with workers who were active or had previously been active on gig platforms (16 direct interviews and one focus group discussion comprised of 10 participants). We also conducted three key informant interviews, with a team of academics, one domestic worker union representative, and one platform representative.
Several domestic work-focused platforms exist in South Africa. Although platforms evolve and change regularly, they typically offer a smartphone-operated app that allows clients to access the profiles of workers whose availability and profile match their preferences for domestic service provision. These same apps also offer ways for workers to sign up, manage gigs, and receive payment. At the time of data collection, the platform studied for this research enabled clients to make bookings of three hours or more and gave them a way to tip workers. On the worker side, it offered an hourly rate based on their tenure with the platform and a premium for taking on gigs cancelled by others. The platform's method of recruitment included an application and selection process, migration status and criminal record checks, and orientation sessions on using the platform. It covered the cost of cleaning supplies, while workers paid for transport and the cost of their airtime (the platform has since developed a datafree app, eliminating airtime costs for workers). 31

Labour Conditions in On-Demand Domestic Work
Our exploration of the conditions of gig work focused on: earnings and income stability; flexibility in the location and timing of work; safety and security; social protection; opportunities for learning and the professionalisation of service provision; and possibilities for collective organisation and bargaining. We briefly outline our findings on each in turn.

Earnings and Income Stability
Our analysis demonstrates that, as of December 2018, workers on the platform with five days of availability were earning ZAR 900 (USD 65) on average per week. This was around 45-50 per cent higher than the minimum wage for domestic workers (working at least 27 hours per week) of ZAR 616 (USD 45), but it still falls short of the amount needed for a family of four to exceed the poverty line (estimated at between ZAR 1,031 and ZAR 1,319 per week per 31 Detailed information is provided in Hunt et al. member). 32 Moreover, the overhead financial costs of gig work-e.g. airtime and transport costs between gigs-depress platform-based earnings.
Most survey respondents (84 per cent) reported being their household's primary earner, while nearly all (95 per cent) had financial dependents. Many also signalled that their household incomes were insufficient to meet their basic needs and financial responsibilities. Utilisation rates for 'full-time' workers (those available for work five or more days weekly) averaged around 60 per cent over a one-year period. 33 In addition, the irregularity in receiving bookings meant that some gig workers experienced significant changes in their incomes from week to week, as demonstrated in average variation from mean earnings of close to 50 per cent weekly. Some workers fared better than others on the platform. The top 10 per cent of full-time workers were taking on around one quarter of the available hours of work carried out by full-time workers, with this 'success' linked to ratings, length of tenure on the platform, and being relatively more available to take up gigs.
Nevertheless, over half of survey respondents (56 per cent) reported being satisfied or very satisfied with their pay. A significant share also reported that their hourly earnings were higher than they would have been in other types of work: 37 per cent reported that working through the platform was more lucrative than other jobs on an hourly basis, and 40 per cent indicated this was 'sometimes' the case.
Once registered, gig workers tended to engage in other forms of paid work alongside platform work. Around half (52 per cent) of survey respondents reported having an additional job or business, or that they also worked for another platform. However, the platform typically provided the bulk of workers' income: 73 per cent identified the platform as the main source of their earnings in the previous month. Participants in face-to-face interviews mentioned having recently undertaken other types of casual or informal work, with several reporting street vending, working in shops, and commercial cleaning work. However, paid domestic work was the most frequently cited work engaged in  33 We computed utilisation rates, discounting voluntary 'days off ' between November 2017 and December 2018. before joining the platform, whether live-in or day labour within private homes, sometimes obtained via an agency, and many continued to provide domestic labour through traditional means alongside gig work. Taken together, these findings suggest that despite gig earnings being inadequate, they are still better than other options-notably domestic work in traditional households.

Flexibility
Flexibility in line with workers' preferences is a core offering platform companies advertise to their workforce. It is often portrayed as particularly advantageous to women due to the potential it offers to balance paid work with unpaid care and domestic work. 34 We found some evidence of workers being able to work on days that they preferred-92 per cent of survey respondents reported having worked on convenient days during the previous week and 88 per cent reported having worked at convenient times.
Yet our interviews suggested mixed experiences among workers. Several interviewees agreed that platform work was more flexible than other types of work, including traditional domestic work. Alongside low pay in previous roles and/or persistent unemployment, this flexibility was cited as a reason to join the platform. However, this ostensible flexibility must be interpreted alongside other features of the platform model. First, the model allows platforms and clients to contract workers only when they need them. This means that the platform can respond to fluctuating demand at minimal cost, and that client demand for bookings de facto take precedence over workers' timing preferences. Second, the ability of clients both to book and cancel cleaners on an ad hoc basis-a key aspect attracting clients to the platform model-introduces considerable uncertainty for workers. Third, fixing gig booking lengths in advance increases the likelihood that clients will insist on more work than can reasonably be done in the agreed time, putting pressure on workers to acquiesce or risk being rated negatively and/or lose the client entirely.
The location of gigs was also a challenge. The persistent legacy of racial and economic segregation in South Africa means that many workers live in townships or other low-income areas. These are geographically far away from the more affluent neighbourhoods of their clients, and travelling between the two on public transport is rarely easy. While workers can specify on the platform where they wish to work, this aspect of flexibility was often shaped by logistical and/or financial concerns.

Safety and Security
As in the traditional sector, violence against on-demand domestic workers is a concern. On-demand domestic work also comes with safety risks particular to providing services to a range of different and unknown clients in their homes. 35 Some workers reported instances of rude, aggressive, or abusive treatment while working behind closed doors. The physical urban environment in South Africa, characterised by long distances, poor transport links, and extremely high levels of crime and insecurity, presents further risks. Early gig start times were raised multiple times as a safety issue, and workers reported several instances of armed and aggressive robbery while travelling to and from gigs.

Labour and Social Protection
A chief critique of gig platforms is that the classification of workers as 'independent contractors' (which denies them the status of employees) restricts their access to labour and contributory social protections, while removing the need for platforms to make contributions on their behalf. 36 The model does not guarantee entitlements mandated within South Africa's domestic worker employee regulation. Moreover, platform company representatives frequently express aversion to becoming recognised employers (as opposed to technology companies, as discussed further below), although some have extended basic protections to workers through limited private schemes. The relatively progressive platform involved in this research, for example, had instituted various measures aimed at improving working conditions, including making accidental death and disability coverage available to workers via a private insurance company.
Workers' status on the platform meant that routine life events risk further exacerbating economic precariousness. Workers often had limited or no income during maternity periods in which they could not work, which was especially pertinent since a majority were single mothers. Furthermore, workers' coverage by public social protection was low: the platform's polling of its workforce in 2019 suggested that just 5 per cent of on-demand domestic workers reported being registered for UIF (which would give them access to public maternity benefits), while 32 per cent did not know whether or not they were registered. 37 35 Hunt and Machingura.

Learning and the Professionalisation of Service Provision
Workers expressed satisfaction with the professional development the platform afforded; 91 per cent of survey respondents believed that the work they did through the platform gave them opportunities for 'learning on the job'. The model also appeared to enable a significant share of workers-26 per centto pursue studies alongside work. The platform representative we interviewed spoke of plans to provide training in soft skills such as scheduling and customer interaction. Despite this, attempts to professionalise on-demand domestic work, including through increasing and certifying worker skills, have not yet translated into widespread increased valuation of workers or the labour they provide.
Some evidence suggests that investing in skills development, certification, and other forms of domestic work 'professionalisation' is important for increasing its societal and economic valuation. 38 Indeed, the platform we collaborated with had sought to challenge client perceptions of domestic work as a low-value commodity by presenting it as professional service meriting 'above market' rates. 39 But although the company had started out with higher prices for clients, they did not make bookings until prices were lowered. 'Razor thin margins and no willingness to pay' among clients made raising earnings an extremely challenging proposition for the start-up company. 40 Moreover, several interviewees spoke of a continued lack of respect and poor treatment from clients, suggesting ondemand models have not caused clients to value domestic workers more.

Collective Organisation and Bargaining
Formal gig-worker organisation is nascent in South Africa, with few signs of successful collective action in the on-demand domestic work sector. 41 Indeed, the platform model excludes workers from fundamental labour rights such as freedom of association, collective bargaining, or protection against discrimination or unfair dismissal. None of our survey respondents reported membership in any formal group that would advocate for their rights: 32 per cent said they did not know how to join such a group and 26 per cent felt that such organisations were for workers in the 'formal economy'. While SADSAWU reported receiving some complaints from platform workers, it had not yet had the capacity to focus on them. It also noted that workers would  Ibid.

41
With notable exceptions, such as the legal case against Uber in South Africa discussed above, which the company later successfully appealed. need to be members of the union to receive structured assistance. 42 That said, many workers reported being in informal communication with one another: 74 per cent reported interacting with others on a regular basis, most commonly through WhatsApp. While certainly a source of support to these workers in lieu of other options, the informal nature of these private groups-and lack of any formal organising or bargaining mechanism-prevents them from transforming into meaningful collective action.

Winners and Losers of On-Demand Domestic Work
Building on the analysis of working conditions, this section explores beneficiaries and losers from the rise of on-demand domestic work. We consider three core constituencies: workers, companies, and clients. The core challenge for workers in South Africa is that of employment 'quality vs. quantity'. Although platforms play a growing role in generating paid work, some clearly provide better conditions than others-as evident in the results of the University of Oxford's 'Fairwork' index, which ranks platform companies according to principles covering fairness in pay, health and safety provisions, contracting, management, and representation. 43 Nonetheless, even where platform representatives report a wish to provide quality economic opportunities, a context of high unemployment, informality, and a weak regulatory environment make it possible for decent work standards to remain unmet. Clients are likely to benefit from securing flexible on-demand domestic work with few employer obligations.

Workers
The chief motivation for domestic workers to engage in gig work in South Africa, despite the multiple challenges it presents, appears to be economic necessity. It is important to recall the broader structural constraints that limit the availability and quality of work available to marginalised and disadvantaged women in South Africa, including an economy characterised by widespread un(der)employment and informality, persistent discrimination, and a challenging physical urban environment. Many interviewees highlighted a lack of other options and reported that platform work offered them some tangible benefits over both unemployment and the other forms of work realistically available to them. These included higher hourly earnings, some choice over work hours, and having an intermediary between them and clients. Indeed, 91 per cent of survey 42 Key informant interview, SADSAWU representative. respondents reported that gig work gave them greater freedom and control in their work. From this perspective, any constraints to platform operations through further regulation are likely to restrict workers' economic opportunities, under a model that many perceive as having relative advantages.
This backdrop is hardly promising for improvements in working conditions, even where platforms seek to charge higher rates to clients than in the traditional sector and pass on (some of) this surplus to workers. The traditional domestic sector has been a key source of work for many marginalised women in South Africa, and platform companies are fully cognisant that they are operating within a context of poor labour conditions. Indeed, platforms are reliant on having a large pool of workers willing to provide cheap and readily available labour. This means that their offering can come with only minimal security, rights, and protections, and it will in some ways be better than what is found in the traditional sector. In other words, it is a relatively better option. But by neither fully meeting workers' needs nor by advancing a quality work framework, it can also be argued that they are helping to maintain the traditionally inadequate working conditions that have long characterised domestic work. Indeed, they can do this because weak regulatory institutions (and enforcement), widespread unemployment (currently averaging 30 per cent for women), 44 and deeply entrenched structural challenges give workers little choice but to take whatever paid work comes their way.
Lack of protections typify South Africa's informal economy (and other low-and middle-income contexts). But what distinguishes the platform economy is that these are built in by design. Domestic work, as we have seen, provides an income to the most insecure workers who often lack other forms of social protection, such as living in a household with a member who has social insurance or who receives a government grant. Only 27 per cent of our survey respondents lived in households receiving a South African Social Security Agency (SASSA) grant (while 12 per cent were unsure), compared with 70 per cent of households nationally. It follows that these workers are most in need of the rights and protections that employee status would confer.
The platform has provided limited protections to workers, including privatelyprovided microinsurance for accident and disability coverage, instead of contributions to public social protection which would normally be provided by employers. However, public schemes are more likely to confer protection upon workers, while 'more individualized forms of protection, such as private insurance or individual accounts, do not comply with most social security 44 'Unemployment Drops in the Fourth Quarter of 2018', Statistics South Africa, 13 February 2019, retrieved 13 July 2020, http://www.statssa.gov.za/?p=11897. principles, and therefore are outside the core of social protection systems'. 45 Indeed, the privatisation wave in the 1980s and 1990s demonstrated the underperformance of such schemes and raised serious doubts about an increased role for private provision. 46 Accordingly, public social protection systems financed through an appropriate blend of taxes and contributions are more likely to guarantee adequate social protection, ensure fiscal and economic sustainability, and give due regard to social justice and equity. Such an approach has the potential to promote a stronger social contract by allowing for risk pooling and redistribution among different groups within the population. 47 Behrendt et al. conclude that 'proposals that weaken social insurance in favour of private insurance and individual savings arrangements, with their limited potential for risk pooling and redistribution, will likely increase poverty, especially for vulnerable low-income earners and those with non-linear work careers, and exacerbate inequality, including gender gaps, and thus can only be voluntary mechanisms to complement stable, equitable and mandatory social insurance benefits '. 48 The delinking of platform companies from social insurance is not inevitable.
GoJek, the largest gig platform in Indonesia, is notable for having developed the pioneering SWADAYA programme in 2018 in partnership with the country's public social security system, which adds a social insurance contribution to the price of its services. 49 However, because it voluntarily opts to provide this scheme rather than being mandated to do so by law, it has the option of changing or revoking the programme at any time.
In short, even if workers perceive short term benefits from engaging in platform work, the concern is that its operating model could undermine legislative gains achieved within the traditional sector in the longer term. In turn, this could worsen working conditions and make workers dependent on company goodwill rather than concrete entitlements to labour rights and government social 45 C Behrendt, Q A Nguyen, and U Rani, 'Social Protection Systems and the Future of Work: Ensuring social security for digital platform workers ', International Social Security Review, vol. 72, issue 3, 2019, pp. 17-41, https://doi.org/10.1111 Ortiz, cited in Behrendt et al. protection. Domestic work is inherently insecure work in which marginalised women are overrepresented, yet their lack of power and socio-economic marginalisation means they are too often excluded from such protections, especially since they often do not qualify as or are not recognised as employees. Indeed, this motivated a long-fought effort by domestic worker unions and other allies in South Africa, leading to one of the strongest regulatory and social protection frameworks for traditional domestic work globally, which the ondemand economy risks undermining.

Companies
The legal framework underpinning gig work is a recurring challenge. Should gig workers be classified as employees and platform companies as their employers? This issue's importance is reflected in litigation seeking the application of regulation and/or confirmation of employee status (with its associated protections and benefits), which is being pursued by workers and labour advocates in many countries. Some analysts argue that on-demand models herald a new form of working which renders current regulatory approaches ambiguous or even obsolete, and that a new classification is needed. 50 Others argue instead that such a reappraisal would merely undermine existing standards by evading the application of current sectoral regulation. 51 Legislative debates over gig workers' employment status in South Africa have been confined to the ridesharing sector, most notably a case for unfair dismissal brought against Uber by deactivated drivers in 2017. Despite Uber's defence that drivers were not employees, and therefore could not be dismissed, the Commission for Conciliation, Mediation and Arbitration (CCMA) ruled in the drivers' favour-although the decision was later overturned on appeal. So far, this burgeoning advocacy has not led to the wholesale recategorisation of gig workers as employees. This means that platform companies 'win' from the growing gig economy chiefly by positioning themselves as brokers between clients and workers, rather than as employers. They capture value from workers' labour by charging commissions on gigs, while at the same time circumventing the responsibility to uphold labour rights and contribute to social insurance on workers' behalf. Per Aloisi and De Stefano, 'the lack of compliance with labour- Platform companies, in turn, argue that innovation is needed to provide employment (particularly in high unemployment contexts like South Africa); that they provide their own support to workers where viable (e.g. private insurance); and that their operating model bolsters work quality in settings where poor quality work is endemic. They contend that any attempt to reclassify users of their platform as employees would severely hamper their profit-making ability, due to the attendant obligations in terms of worker taxation and employee contributions, and consequentially jeopardise their very existence and therefore the economic opportunities they facilitate.
Furthermore, by arguing that they need a favourable operating environment to 'create' jobs, platform companies may reduce the South African government's political will to carry out oversight. The government may well gamble that it is more politically expedient to support the creation of 'digital jobs' amid high unemployment, as Kenya's government has done, 53 than to increase the regulation of labour conditions and taxation. Indeed, a 'social partners' framework agreement for addressing South Africa's unemployment crisis through 'broad-based improvement in the business environment and conditions for entrepreneurial development' and strong encouragement of 'adopters of new technology to use innovation as a means to save and grow jobs' was agreed during the national Presidential Jobs Summit held in October 2018, with scant reference to job quality. 54 In short, there is a strong case that the profit-making model of on-demand companies in South Africa currently depends on the historic inability of domestic workers to establish a de facto employment relationship (and the better conditions that accompany it), as well as poor enforcement of existing regulations governing traditional domestic work. If these challenges were tackled in a meaningful way, companies would likely be obliged to emulate traditional employers in paying employee taxes and UIF contributions, which could in turn lead them to reduce the opportunities available (e.g. to ensure no worker gets over 24 hours per month work, which would render them exempt from having to pay UIF contributions).

Clients
At the time of writing, the platform charged its clients a variable rate that depended on the number of hours booked-ranging from ZAR 48 (USD 3.48) per hour for a four-hour booking to ZAR 30.5 (USD 2.21) per hour for a ten-hour booking, which represents the maximum length. This is clearly far higher than the government-mandated minimum wage for domestic workers (ZAR 16.03 [USD 1.16] hourly for a worker employed for fewer than 27 hours weekly, per the 2018 Sectoral Determination). However, the higher hourly cost to clients of hiring a platform worker is offset by lower transaction costs, e.g. those associated with selecting, screening, and supervising a worker found independently. This is a key attribute of the on-demand model that was highlighted by a platform representative we interviewed. They explained that, by allowing the platform to carry out these processes, clients were ensured a 'professionalised' service in return for paying higher prices. In addition, clients avoided the economic commitment of guaranteeing employment for a set number of hours' work and bureaucratic processes associated with being an employer as stipulated by South African labour law. Therefore, it could be argued that, from a client perspective, an important advantage of the platform model is that it de facto provides a service that evades compliance with labour or social security regulations. Such a trend significantly threatens the hard-won gains of the domestic worker movement and risks eroding the better-quality formal jobs where they exist, should the platform economy secure a sizable market share. This is likely to impact negatively upon the cohort of domestic workers who remain relatively marginalised but have managed to secure access to higher standards and securities in the traditional sector.

Conclusion
At present, neither the traditional nor the on-demand models can be said to offer decent domestic work. In both spheres higher standards and their enforcement are needed to redress historical power inequalities and ongoing breaches of the labour rights of South Africa's domestic workforce. The trajectory of the gig economy to date suggests that platform companies, with an inherent profit motive, are unlikely to lead the charge towards a wide-scale revaluation of domestic work; nor are household purchasers of workers' labour. Broader societal reforms are therefore needed to shift the social norms underpinning the discrimination and structural inequality characteristic of the domestic work sector. Government action is also needed, so that traditional domestic workers benefit from the same labour protections as workers in other more highly valued sectors, and to ensure that existing regulation is enforced in the platform economy. Indeed, without compliance with labour rights and protections, on-demand workers are unlikely to benefit fully from 'collective bargaining, protection from unfair dismissal and all the legal protection that goes with formal employment that goes in inverted commas if and when they become employees', per a Social Law Project representative. 55 As it stands, incremental improvements notwithstanding, we find that ondemand models can be seen as largely 'more of the same'. They capitalise on the undervalued labour of marginalised women workers and uphold the power held by the purchasers of their labour that characterise the traditional sector. Therefore, the platform economy represents a continuation of the normalisation of the labour exploitation of domestic workers. It is critical to extend labour and social protections to all domestic workers in a sustainable and comprehensive way, for which an increased societal valuation of domestic work-and workers-is a prerequisite. Policy-makers and platform companies have a central role to play in ensuring these rights, which notably include regular, fair, and adequate earnings, facilitating access to public social protection, ensuring their health and safety, and supporting collective action to ensure that policy and practice reflect workers' own priorities.
Abigail Hunt is a Research Fellow at the Overseas Development Institute, where she leads research focused on gender and the world of work. She is particularly interested in marginalised women workers' experiences of new and emerging labour market trends, unpaid care and domestic work, and social protection. Email: a.hunt@odi.org.uk Dr Emma Samman is a Research Associate with the Overseas Development Institute and an independent consultant. Her research centres on the analysis of poverty and inequality, particularly gender inequality, the human development approach, the future of work, and the use of subjective measures of wellbeing to inform research and policy. Email: e.samman@odi.org.uk