Household Social Mobility for Paid Domestic Workers and Other Low-Skilled Women Employed in South Africa

ABSTRACT

This paper explores the theme of patronage by examining how the social mobility prospects of paid domestic workers differ from other vulnerable low-skilled black and colored women in post-apartheid South Africa. The literature provides contradictory predictions about the effects of a relationship with an affluent employer on a vulnerable employed woman and her household. Using data from the 2002–8 General Household Survey, this study uses propensity score matching (PSM) to compare paid domestic workers versus employed women with similar labor market characteristics. It finds that the household members of paid domestic workers tend to have a lower likelihood of unemployment, lower unemployment duration, higher likelihood of owning assets, and lower prevalence of hunger. It is, however, important to see evidence of such benefits in the context of a complicated employment relationship and to highlight that such benefits can reflect both altruistic and self-serving employer motivations.

KEYWORDS:

• Domestic workers
• economic development
• low-wage employment
• South Africa
• gender discrimination
• inequality

JEL Codes::

• Z13
• D63
• J16

SUPPLEMENTAL DATA

Supplemental data for this article can be accessed at https://doi.org/10.1080/13545701.2017.1414951.

Notes

1 Although “colored” is a contested term, it is a self-identified label for a multiracial South African ethnic group.

2 We use the term “patronage” reluctantly here because we could not find a suitable alternative that is widely understood. Despite legislative reforms in South Africa, the relationships between employers and domestic workers are typically still marked by unequal power and exploitation, but also generosity and trust.

3 We are confined to this period because Statistics South Africa introduced a number of changes to the GHS in 2009.

4 Increase in within-group inequality is driving the increase in overall inequality.

5 Tewodaj Mogues and Michael R. Carter (2005) use the term “socially embedded” to describe the type of income inequality that results in a setting where socioeconomic polarization is high and reinforced along other social dimensions such as racial, ethnic, or linguistic divides. This makes it more costly and burdensome for the poor to build relationships with the more affluent members of society.

6 Mogues and Carter (2005) show how the social capital patterns of individuals can exacerbate income inequality in a situation where inequality is socially embedded and markets are incomplete.

7 The traditional “master–servant” relationship between a paid domestic worker and her employer typically meant that the employer had much greater power and superiority, while the paid domestic worker was placed in a position of dependence and servitude. This gave scope for either a paternalistic relationship, which extended “kindly patronage” to the paid domestic worker, or an extremely exploitative and disrespectful relationship between a paid domestic worker and employer (Berhardien, Lehulere, and Shaw 1984).

8 In 2005 Rands.

9 Interviews with paid domestic workers in the late 1970s showed that they had long paid working hours and consequently no family or social life. Paid domestic work was also characterized by low wages and disrespectful treatment by the employer (Cock 1980; Berhardien, Lehulere, and Shaw 1984). Gaitskell et al. refer to the “isolation, dependence, invisibility, and low level of union organization” as being characteristic of domestic service in South Africa (1983: 87).

10 In 1995, the Labour Relations Act was amended to cover paid domestic workers, thus providing paid domestic workers with the right to belong to a union and with protection against unfair treatment and unfair dismissal via dispute resolution mechanisms. Levels of union membership, however, have continued to be low in the sector.

11 In 2002, the Basic Conditions of Employment Act 75 of 1997 was expanded to make allowances for minimum wages for employed domestic workers through Government Notice No. R. 1068, Basic Conditions of Employment Act (75/1997): Sectoral Determination 7: Domestic Worker Sector, South Africa.

12 Paid domestic workers became eligible for unemployment benefits with the enactment of the Unemployment Insurance Contributions Act, No. 4 of 2002.

13 Research suggests that wages for employed domestic workers have indeed increased following the introduction of the regulation (Budlender 2005; Hertz 2005; Dinkelman and Ranchhod 2012). There is some evidence that weekly paid working hours have decreased (Hertz 2005), but this was disputed by Taryn Dinkelman and Vimal Ranchhod (2012).

14 During apartheid, the state legislated separate neighborhoods, and black individuals were not allowed to live in neighborhoods that were allocated to white individuals. Paid domestic workers were an exception and were allowed to live at their employment site in accommodations provided by their white employers (Goodlad 1996).

15 See Parreñas (2001), Helma Lutz (2011), and Dirk Hoerder, Elise van Nederveen Meerkerk, and Silke Neunsinger (2015) for examples of the international feminist literature on paid domestic workers.

16 Self-assessed estimates provided by donors are likely to overstate the market value of the transfers, especially given that the literature suggests such transfers to paid domestic workers are often old, unwanted, and discarded food or household items (Ally 2009).

17 These include street vendors and other street service workers, cleaners and launderers, building caretakers, garbage collectors, messengers and porters, entry-level sales occupations, agricultural and fishery laborers, mining and construction laborers, and transport and freight handlers.

18 The study is not longitudinal, and, accordingly, each year's sample is drawn independently. All the results in this paper are from the sample obtained after pooling the households across all these survey years.

19 In the GHS for the years 2002–8, only 4.7 percent of paid domestic workers were men.

20 This analysis is based on the PALMS 3.1 dataset with paid domestic workers and other elementary workers. Using the log of monthly earnings as our variable of interest, we find evidence of both cohort effects (with older cohorts earning less) and age effects (with a drop after age 50). We use the intrinsic estimator of Yang Yang, Wenjiang J. Fu, and Kenneth C. Land (2004) to decompose unskilled earnings into a life-cycle, generational, and period effect.

21 In other words, because we observe that paid domestic workers are more vulnerable and desperate participants in the labor market than those women included in the control group, we would expect that paid domestic workers and their households would, on average, be less likely to access quality healthcare, education, and employment opportunities.

22 This assumption is, of course, not testable, which emphasizes the importance of the choice of covariates.

23 Given the obvious differences in wealth and living standards between urban and rural areas, we also expect there to be differences in the number and types of individuals who are employed as paid domestic workers between urban and rural areas. However, the GHS data do not include information on whether households reside in urban or rural areas for the years 2005–8, and we therefore have not been able to include an urban–rural variable in our regressions. Instead, we have included provincial dummies to control for any geographic variation.

24 We cannot run the second model specification on the household income outcomes because that would involve using the same variable as both control and outcome. This is the last outcome variable in our list of employment-related outcome variables.

25 Since each of the outcome variables is associated with a different sample, the results from the first-stage estimation of the propensity score differ slightly. Results for the other outcome variables are not reported here due to space constraints but are available from the authors on request.

26 Again, as indicated above, what is reported here is only the balancing outcome for the estimation of the duration of unemployment outcome variable. The other results are available on request.

27 Mark Granovetter (1983) shows that employees hired via referrals are more likely to be promoted.

28 Children are defined to include any individual in the household between ages 10 and 24 who is either the child or grandchild of the household head. The fact that we find no positive impact for the education outcomes may be explained in part by the fact that paid domestic workers are unable to support their children's school careers in the ways that matter, due to their long paid working hours. Paid domestic work often requires long work hours and weekend work (South African–German Development Co-Operation 2001). The Department of Labour's investigation of employment conditions concluded that paid domestic workers often have very limited free time and that some paid live-in domestic workers are on call day and night (Department of Labour of the Republic of South Africa 2001).

29 It is also important to note that, according to the definition of “household” used in the GHS, paid live-in domestic workers would not be counted as part of the household of their employers, as the survey requires household members to “live together and share resources as a unit … [to be] ‘eating from the same pot’ and [to] ‘cook and eat together’” (Statistics South Africa 2008).

Additional information

Notes on contributors

Ronelle Burger

Ronelle Burger is Professor in the Economics Department at Stellenbosch University. She is interested in pro-poor development policy in African countries.

Marisa Von Fintel

Marisa von Fintel is Researcher at Stellenbosch University. Her work considers poverty measurement and social integration.

Carina Van der Watt

Carina Van der Watt is Research Student in the Economics Department of Stellenbosch University.