What is at issue in the minimum wage debate?

Reads 24,603
The debate about implementing a national minimum wage obscures the key point, which is the level at which a national minimum wage should be set. A national minimum wage at a much higher level than the sectoral minimum wages currently set by the Employment Conditions Commission or agreed upon by unions under the Labour Relations Act is likely to result in job destruction, especially in the tradable sectors, with the result that poverty might be increased rather than reduced.


Most contributions to the debate on a national minimum wage in South Africa generate more heat than light by misrepresenting the difficult choices facing any policy-maker concerned with poverty and inequality.

It appears that participants in the debate such as Cosatu’s Coleman (2013; 2014a; 2014b) or Isaacs & Fine (2014), who favour a high national minimum wage, try to simplify the choice so that it supports their case (which is in line with Cosatu’s demands in this regard).

Unfortunately, the choices are not as simple as these commentators would have us believe. This article attempts to provide a broader perspective on the various positions and issues at stake with regard to the question of a national minimum wage.

Background: how minimum wages work currently

The state sets minimum wages by means of two mechanisms. In one of these, used since 2002–3, sectoral minimum wages are recommended by the Employment Conditions Commission (ECC) and gazetted by the Minister of Labour in terms of the Basic Conditions of Employment Act. (Prior to the 2000s, minimum wages were set by the Wages Board.) At present, the ECC and Minister of Labour set wages in eleven sectors, including the Farmworker, Wholesale and Retail, Domestic Workers, Forestry, Taxi, Private Security, Civil Engineering, Contract Cleaning, and Hospitality sectors. This mechanism serves to set minimum wages in sectors where workers are not organized into strong trade unions.

The second mechanism for setting minimum wages entails the Minister of Labour’s ‘extending’ collective agreements that were negotiated between employers and unions in bargaining councils established in terms of the Labour Relations Act. Prominent examples of these are agreements in the clothing and metalworking industries.

A legislated and properly enforced national minimum wage would create a wage floor below which no wage in any sector would legally be allowed to fall.

The main positions on a national minimum wage

There are three main approaches to the issue of a national minimum wage. The first is the free market or libertarian view, advocated by the likes of the Free Market Foundation. It holds that any regulation of wages should be opposed, in part because wage regulation necessarily inhibits economic growth and job creation. The labour market should be left free to determine wages through the interaction of supply and demand. Seen from this point of view, both the existing ways of setting minimum wages are bad and the prospect of a national minimum wage is even worse.

The opposite argument, put forward by Coleman (and Cosatu), is that a national minimum wage should be imposed – and that it should be substantially higher than the current sectoral minimum wages, i.e. possibly between about R4 800 and R6 000 per month in 2011 prices (Coleman 2013: 34). He arrives at this target range on the basis that internationally a minimum wage is often set at between 40% and 50% of the median wage. Coleman also suggests a phasing-in period, with an initial level of R2 700–R3 000 in 2011 prices (2013: 43; 2014a).

Coleman’s main argument is that a higher minimum wage would not only reduce the exploitation of lower-paid workers, but would also generate job creation through consumption-fuelled growth: the higher wages would enable higher consumption by workers, thereby boosting employment and growth, at least if the requisite other policies were in place (Coleman 2013: 40; 2014a).

There is, however, a middle view to which we subscribe. From this point of view, wages should indeed be regulated in a context such as that in South Africa, but the level at which minimum wages are set should take into account the likely negative effects on employment. Such wage regulations might include a national minimum wage as well as sectoral minima. Minimum wages should not be set at a level that results in substantial job destruction because this would result in higher wages for a shrinking group of workers while unemployment would grow. This would be likely to increase, rather than reduce, both poverty and inequality.

This middle position on the minimum wage issue differs from the first (free-market) position in that it endorses wage regulation in the form of wage minima. It differs from the second (Coleman) position in that it takes seriously the concern that setting a national minimum wage too high might result in significant job losses.

The real debate

The real debate is between the middle position and the Coleman position. The tough decision does not concern the principle of a national minimum wage, but rather the level at which such a national minimum wage should be set as well as the procedures for setting that minimum (including possible phasing in).

The difference between the middle position and Coleman’s approach corresponds with the difference between the logic of setting a minimum wage under the Basic Conditions of Employment Act and the logic of the extension mechanism under the Labour Relations Act (LRA).

The Basic Conditions of Employment Act requires the ECC to take employment effects into account when recommending a sectoral determination regarding minimum wages. In late 2011, the ECC proposed that the minimum wage in agriculture be set at just over R1 500 per month from March 2012. In 2013, in the wake of dramatic protests by farmworkers, the ECC proposed a higher minimum wage of almost R2 300 per month (or R105 per day). The ECC recommended this increase on the basis of new research that suggested that the increase would not lead to extensive job destruction. The ECC decided against raising the minimum wage to the level of at least R3 600 per month (i.e. R150 per day in a six-day week) that was demanded by striking farmworkers, because it assessed that raising wages to that level would result in massive job destruction.

Similarly anxious about the possibility of job destruction, the Department of Labour took the unprecedented step of allowing farmers to apply for exemption from these minimum wages on the grounds of unaffordability. It is not clear how many exemptions were submitted or approved. But press and anecdotal reports suggest that the eventual negative effects on employment might be higher than the ECC expected, and that it affects women in particular (e.g. Cape Times, 3 December 2014). When the ECC considers a new sectoral determination for agriculture, it will no doubt take into account the actual effects that its 2013 sectoral determination has had on employment.

The Labour Relations Act, in contrast to the Basic Conditions of Employment Act, provides for minimum wages to be extended without any consideration of the possible ensuing job destruction. First, unions and employers’ associations – often dominated by the larger, more capital- and skill-intensive employers – negotiate collective agreements in bargaining councils. A collective agreement is binding only on the employers who are party to the agreement. But the Minister of Labour can – and under some circumstances must – ‘extend’ this to all employers in the particular industry. Trade unions and large employers, pursuing their own interests, typically do not raise wages to levels that would destroy their own businesses and members’ jobs (although trade unions might miscalculate). However, they can and sometimes do set wages which, when extended by the Minister of Labour, result in the closure of other firms in the industry and the destruction of other workers’ jobs. This outcome is possible, even likely, because the Labour Relations Act and the bargaining council process are not expressly concerned with the risks of job losses, but rather with the conditions of work of people with jobs.

Critical analysis

Coleman proposes a new system whereby a national minimum wage will be set without worrying about negative employment effects. In his various writings he downplays – or dismisses as a myth and a discredited idea – the concern that a national minimum wage will result in job losses (arguing that there is no ‘mechanical relationship’ between wage levels and employment levels; Coleman 2014b).

We do not disagree with him on this, depending on the level at which a minimum wage is set. A national minimum wage that has been set at the level of the ECC’s current minimum wages in sectors such as domestic work or agriculture would probably not result in significant job destruction. A much higher national minimum wage would in all likelihood result in significant job losses. 

Neither Coleman nor Isaacs & Fine address possible job-destruction effects of a R4 800–R6 000 minimum wage as such; they just reiterate that high minimum wages would be good for employment because of the consumption effect on growth (without explaining why wages should not be set much higher than the R4 800 to R6 000 band if that effect is so potent).

So, who is right? Coleman and Isaacs & Fine, who appear to think that minimum wages can be raised significantly without job destruction, or the ECC and even some unions (such as the SA Clothing and Textile Workers Union, SACTWU) whose recommended minimum wages in various sectors were lower than R4 800 per month because they were worried about job destruction?

International and local evidence: conclusive or not?

Coleman and Isaacs & Fine cite two sets of evidence in support of high national minimum wages. Neither is nearly as compelling as they would like us to believe.

First, they cite research in the USA and the experience of Brazil which suggest that, in those cases, increases in minimum wages did not result in significant job destruction. The difficulty with this research is that the USA and Brazil are low-unemployment economies, while South Africa has a very high unemployment rate. In the USA and Brazil, most poor people work. Raising the wages of the working poor in growing economies with ‘tight’ labour markets can indeed reduce poverty without much job destruction.

These countries’ experiences are not appropriate to South Africa. The case that is more relevant is Mauritius. In the 1970s Mauritius suffered from high unemployment. Acting on the recommendations of the (progressive) economist James Meade, Mauritius encouraged low-wage job creation. The Mauritian government waited until unemployment had dropped – i.e. the labour market had tightened – before raising statutory minimum wages because it wanted to ensure that there would be no job destruction (see Dabee & Greenaway 2001).

Secondly, Coleman and Isaacs & Fine point to research by Bhorat and his colleagues on the effects of sectoral minimum wages set by the ECC in the early 2000s (Bhorat et al. 2013; DPRU 2010). The DPRU researchers found no significant negative employment effects in non-agricultural sectors, except in terms of working hours in some cases (see Bhorat & Mayet, Econ3x3, 2013). Presumably, as dictated by its mandate, the ECC properly took employment effects into account in setting those (sectoral) minima. Thus, it is not clear that these experiences can be used to argue for setting high general minimum wages that do not take such effects into account and do not differentiate on a sectoral basis.

Taking sectoral characteristics and conditions into account

Coleman and Isaacs & Fine do not do full justice to the research by Bhorat et al., which is rather more nuanced than Coleman and Isaacs & Fine acknowledge. First, many of the sectors that they analysed were ones where, in the period under scrutiny, the employers or customers were enjoying rising real earnings and the demand for labour was buoyant. In most of these sectors higher minimum wages reduced employment in terms of reduced working hours, but did not destroy jobs. In agriculture, there was stronger evidence of job destruction.

More importantly, most of the ECC sectoral determinations cover non-tradable sectors, i.e. sectors (such as private security, domestic work, retail and restaurants) which do not face competition from imports. This reduces the likelihood that modest wage increases would result in major employment losses. In sectors such as these, the arguments for raised minimum wages need to be taken seriously by anyone in the middle ground of the minimum wage debate.

With regard to the tradable sectors, notably agriculture, the evidence points to a rather different conclusion. South African farmers – whether they produce grapes for wine or lambs for slaughter or sheep for wool – compete with farmers elsewhere in the world for foreign and local markets. Case studies in different parts of the country (Conradie 2007; Murray & van Walbeek 2007), as well the Bhorat et al. analyses of national sectoral data, suggest that the original (2002) sectoral determination in agriculture raised wages modestly and improved compliance with non-wage regulations – but it also resulted in a significant reduction in total employment and/or hours worked. In the (tradable) forestry sector, Bhorat et al. find that the sectoral determination did not lead to any observed improvement in total earnings, because higher wages were offset by a reduction in working hours.

From this research it seems that the ECC’s sectoral determinations in the early 2000s in non-tradable sectors probably resulted in higher real wages that more than offset the reduced hours of employment; as a result, the working poor were less poor. In contrast, in tradable sectors the negative employment effects were much larger.

The clothing sector is a tradable sector and is subject to constraints similar to those in agriculture. In the clothing sector the minimum wage was set through the extension mechanism provided by the LRA. But SACTWU itself recognized that higher wages could lead to job destruction and has moderated its wage demands in order to protect its members’ jobs. Nevertheless, it is apparent that the level at which this minimum wage was set has led to significant job destruction in parts of the clothing industry, notably in non-urban areas (Nattrass & Seekings 2014).

In related work Pauw & Leibbrandt (2012), using a microsimulation framework, found that minimum wages are likely to lead to significant job losses among unskilled workers. Minimum wages may have reduced wage inequality in terms of hourly wages, but perhaps not in terms of total wages and probably at the expense of diminished demand for unskilled labour.

Whilst no evidence can definitively prove or disprove that a much higher minimum wage in these sectors – e.g. at the level suggested by Coleman – would necessarily result in significant job destruction, it is very likely that this would be the case and that poverty would increase rather than be reduced.

The macro-economic argument?

The macro-economic argument that paying workers more money will fuel consumption and thus stimulate economic growth and job creation is a central element in the arguments of Coleman and Isaacs & Fine. There are many reasons for doubting the strength of any such effect in South Africa. First, if higher wages result in job destruction, the net effect on spending power will be reduced and may perhaps even turn negative. Secondly, while recent economic growth in South Africa has been driven primarily by increased consumption, this growth has generated only modest employment growth. Indeed, the ratio between employment growth and economic growth in South Africa is small (approximately 0.5). Consumption-led growth typically has not led to significant employment growth in South Africa.


The available empirical evidence suggests that the ECC has been right to worry about ‘negative employment effects’ (i.e. job destruction), especially in tradable sectors. The fundamental point is that the effects of higher minimum wages vary between sectors according to factors such as their exposure to international competition, the possibilities for mechanisation, and the incomes of employers or customers. In other words, there is more scope for higher minimum wages in some sectors than in others.

This suggests that minimum wages should vary between sectors, i.e. that the ECC model is the one to follow (amended as necessary, but not discarded). Such sectoral minimum wages can be combined with a national minimum wage. But – and this is our key dispute with Coleman and Cosatu – the eventual national minimum wage should be at a relatively low level, e.g. at the level of the lowest sectoral determination. Higher minimum wages might then be set in other sectors if the expected negative employment effects (i.e. job destruction) in these sectors are assessed to be modest.

Arguing for a high national minimum wage floor, even if phased in, seems to us to entail what might be called ‘unemployment denialism’, i.e. the denial that poverty in South Africa is primarily the result of massive unemployment and the assertion instead that the poverty is due primarily to low wages (as is the case in Brazil and the USA). To calculate a proposed minimum wage from a ‘median wage’ based on the wages of the low percentage of people that do have jobs is to ignore blithely the zero wages of the millions who are unemployed.

The mere possibility of unprecedented consumption-led employment growth hardly seems to warrant setting a national minimum at a level that has a high probability of destroying jobs in sectors such as agriculture, destroying entire sectors such as clothing and reducing the possibility that labour-intensive sectors could grow in future.

Minimum wages present one of the most difficult choices to any policymaker concerned with poverty, inequality and unemployment. It demands careful analysis and nuanced policy interventions – not a blanket approach to a very complex matter with high risks of unintended and perverse consequences.

* An earlier version of this article was published on www.groundup.org.za.


Bhorat H, Kanbur R & Mayet N. 2013. The impact of sectoral minimum wage laws on employment, wages, and hours of work in South Africa. IZA Journal of Labour and Development 2:1.

Bhorat H, Kanbur R & Stanwix B. 2012. Estimating the impact of minimum wages on employment, wages and non-wage benefits: the case of agriculture in South Africa. DPRU Working Paper 149.

Coleman N. 2013. Towards new collective bargaining, wage and social protection strategies in South Africa – Learning from the Brazilian experience. Working Paper 17. Global Labour University.

------------ 2014a. Many benefits to decent pay  and no, it does not kill jobs. Business Day. 2 August 2014.

------------ 2014b. Myths and so-called ‘facts’ about the National Minimum Wage. Daily Maverick. 10 November 2014.

Conradie B. 2007. Revisiting labour casualisation on fruit farms in the Western Cape. CSSR Working Paper 177. University of Cape Town.

Dabee R & Greenaway D. (eds). 2001. The Mauritian Economy: A Reader.  Basingstoke: Palgrave.

DPRU 2010. Addressing the plight of vulnerable workers: the role of sectoral determinations. Report for the Department of Labour. University of Cape Town.

Isaacs G & Fine B. 2014. The minimum wage debate – logic, evidence and need. Daily Maverick. 25 November 2014.

Murray J & Van Walbeek C. 2007. Impact of the sectoral determination for farm workers on the South African sugar industry: Case study of the KwaZulu-Natal North and South Coasts, CSSR Working Paper 181. University of Cape Town).

Nattrass N & Seekings J. 2014.  Job destruction in Newcastle: Minimum wage-setting and low-wage employment in the South African clothing industry. Transformation 84: 1-30.

Pauw K & Leibbrandt M. 2012. Minimum wages and household poverty: general equilibrium macro-micro simulations for South Africa. World Development 40(4):771-83.


Tags: Minimum wage

Download article

PDF icon Seekings & Nattrass 2015 Minimum wage debate FINAL.pdf
Post a commentary

This comment facility is intended for considered commentaries to stimulate substantive debate. Comments may be screened by an editor before they appear online. To comment one must be registered and logged in.

This comment facility is intended for considered commentaries to stimulate substantive debate. Comments may be screened by an editor before they appear online. Please view "Submitting a commentary" for more information.