Skip to content

South Africa's Manufacturing Wage Gaps with the BRICs: Measurement and Accounting

Literature › Works

Mengistae, T. (2011). South Africa's Manufacturing Wage Gaps with the BRICs: Measurement and Accounting. World Bank (apparent working paper; no publisher printed on the text).
Type working paper
Scope Post-1994 South African economy
Method econometric
Period studied 1991–2007
Regions South Africa, Brazil, China, India, Russia, Colombia, Chile, Malaysia, Mexico, Thailand, Argentina, Mauritius
Themes labour markets wages, unions strikes, manufacturing industrial policy, inequality poverty unemployment, comparative theory, methods data sources
Citation key mengistae2011

Argument

South African manufacturing unit labour costs fell over the decade to 2007 but stayed above those of 11 middle-income comparators, including the BRICs. Cost-of-living differences explain only part of the wage gap, and only with China and India. Higher South African productivity pushes wages up, but this is offset by wages being far less sensitive to productivity. The main driver is wage rigidity: wages respond less to unemployment and persist more over time, on top of historically high initial wages. The author infers that bargaining councils, through ergo omnes extension of agreements, probably contribute to this rigidity. The argument rests on UNIDO three-digit panel data, System GMM estimation of wage-bargaining models, and Oaxaca-Blinder decompositions.

Evidence and method

UNIDO industrial statistics at three-digit ISIC level (about 8,500-10,000 industry-year observations, 1991-2007) for South Africa and 11 comparators. Deflators, exchange rates and PPP converters from World Bank WDI. Union and bargaining-council coverage figures drawn from Bhorat et al. (2007) and Butcher and Rouse (2001).

Findings

  • South African unit labour costs fell but remained above all comparators, mainly because wages were higher and rose faster. (p. 1)

    South Africa's manufacturing unit labor costs were consistently higher than those in comparators and did not decline as fast essentially because South African manufacturing wages were also initially higher and rose faster.

  • Cost-of-living differences explain only a minority of the wage gap with China and India, and none of the gap with Brazil or Malaysia. (p. 3)

    differences in the cost of living as measured in this way account for only 17 percent and 38 percent, respectively, of South Africa's wage gap with China and India at market exchange rates.

  • Wage gaps are mirrored in unit labour cost gaps; in 2007 South Africa's unit labour costs exceeded those of Russia, Brazil and China by 5, 9 and 19 percentage points respectively. (p. 7)

    South Africa's average manufacturing unit labor costs in 2007 were, for example, 5 percentage points higher than Russia's, 9 percentage points higher than Brazil's and 19 percentage points higher than China's.

  • Union membership and bargaining-council coverage in the formal sector around 2005 were limited, with wage premiums from both. (p. 9)

    about 38 percent of formal sector workers belonged to unions in 2005, which was up from a union membership rate of 34 percent back in 1995.

  • The sliding-scale wage equation fits the data best of the models tested (Nickell-Wadhwani, Abowd-Lemieux, Oswald et al., sliding scale). (p. 16)

    Overall, the equation estimated in table 10 provides a better fit to the data than any of the other equations we have considered thus far, as reported in tables 7 through to 9.

  • Higher South African productivity was largely offset by lower wage sensitivity to productivity, as in the Brazil comparison. (p. 19)

    Brazilian wages would have been less than a third of what they actually were if their sensitivity to productivity growth were on a par with that of South African wages.

  • South African wages respond less to unemployment than comparator wages, muting the downward effect of very high unemployment. (p. 20)

    it is also clear from table 10 that South African wages were significantly less responsive to rises in unemployment than wages in any of comparators.

  • Wage persistence (stickiness) accounts for a large share of the gap with Brazil: 22 points from the initial-wage difference and 54 from higher persistence. (p. 22)

    South African wages are higher than Brazilian wages by 76 percentage points on account of the fact that South African wages were less sticky than wages in Brazil

  • The core conclusion: wage rigidity, combined with historically high wages, produced the international wage gaps. (p. 24)

    Combined with South Africa's historically high wages, South Africa's greater wage rigidity largely produced the international wage gaps seen in our data.

  • The paper infers a role for bargaining councils and ergo omnes extension in wage rigidity, but only by inference. (p. 24)

    it is difficult to see how unions in South Africa would have influenced wage flexibility independently of the functioning of bargaining councils.