Changing Labour Market Conditions in South Africa: A Sectoral Analysis of the Period 1970-1997¶
JW Fedderke; Martine Mariotti (2002). Changing Labour Market Conditions in South Africa: A Sectoral Analysis of the Period 1970-1997. The South African Journal of Economics. Economic Society of South Africa.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | mixed |
| Period studied | 1970–1997 |
| Regions | South Africa |
| Themes | labour markets wages, inequality poverty unemployment, manufacturing industrial policy, mining migrant labour, agriculture land, education human capital, methods data sources, transition post apartheid |
| Citation key | fedderke2002a |
Argument¶
The paper reviews a sectoral panel of 48 SIC sectors (28 in manufacturing) to show that South Africa's inability to create employment is a structural problem dating from at least the 1970s, not a product of 1990s policy change. Descriptive correlations and a pooled mean group labour demand estimate for manufacturing are used to argue that the real wage, and its link to labour productivity, matters for employment. Sectors where wages track productivity show stronger employment and wage growth. The authors conclude that supply-side labour market reform, labour market flexibility and human capital investment are needed, while also noting a possible structural break in 1990.
Evidence and method¶
TIPS panel of 48 SIC version 5 sectors for 1970-97, using benchmark years 1970, 1980, 1990 and 1997 for employment, real remuneration, value added and the skills composition of manufacturing. Population growth is based on Sadie (1993) and World Bank indicators. Panel econometrics cover 28 manufacturing sectors annually.
Place in the debates¶
- Real wages and employment (wage rigidity vs structural/technological explanations of unemployment) (see Unemployment: causes, measurement and labour market institutions). Argues that real wages relative to productivity are an important determinant of employment, and that labour market policy raising wage and non-wage costs reduced incentives to hire. It also accepts that technological change was capital-augmenting and that structural problems predate the 1990s.
- Whether post-1990s policy (trade liberalisation, macro policy) caused poor employment creation (see Unemployment: causes, measurement and labour market institutions). Rejects this. Poor job creation is a long-standing structural feature since the 1970s, and trade liberalisation demand factors do not explain it, citing Fedderke, Shin and Vaze (2000).
- Whether South African manufacturing growth is 'jobless' (see Unemployment: causes, measurement and labour market institutions). Finds an output elasticity of labour demand of about unity, so manufacturing growth is not jobless.
Findings¶
-
Employment creation has been weak since the 1970s and below population growth, so it is not a new development. (p. 8)
It is notable that the rate of employment creation across sectors has been below average annual population growth
-
Average sectoral employment growth differs sharply between 1970-90 and 1990-97, suggesting a structural break in 1990. (p. 8)
the average growth rate in employment (again across all sectors, unadjusted for absolute levels of employment) from 1970-90, and 1990-97 was 2.16 per cent and -1.26 per cent per annum respectively.
-
Most sectors shed labour in the 1990s, compared with very few in 1970-90. (p. 10)
Of the total of 48 sectors recorded, 31 had negative growth rates in employment over the course of the 1990's. By contrast, over 1970-90 only three sectors had negative employment growth rates.
-
Large employment sectors tended to grow slowly, indicating long-term restructuring away from traditional employers. (p. 9)
The correlation between the level of employment in 1970 and the average rate of change of employment by sector, is -0.27
-
The skills intensity of manufacturing is rising; by 1997 skilled and unskilled labour were approaching parity. (p. 13)
Even on average, by 1997 the ratio had reached 0.78, suggesting that skilled and unskilled labour may soon constitute an equal proportion of the labour force in manufacturing sectors.
-
Every sector with falling employment shows a negative employment-real wage correlation, consistent with a negative wage-employment link. (p. 19)
Every sector in South Africa that has showed negative growth rates in employment, also shows a negative correlation between employment and the real wage.
-
Differentials in real pay per worker between the highest and lowest paying sectors narrowed over 1970-97. (p. 15)
Thus from 1970 to 1980, 1990 and 1997 the differential has changed from 30:1, to 35:1, to 22:1 and 26:1.
-
Where wages are tightly linked to productivity, employment and sustained wage growth are stronger. (p. 24)
Where the real wage is less closely linked to real labour productivity, the growth in employment also tends to be lower.
-
The pooled mean group estimate gives a manufacturing wage elasticity of labour demand of about -0.5. (p. 31)
The implication is that the manufacturing sector is characterised by a negative wage elasticity in the region of -0.5 to -0.55.
-
Output elasticity near one means manufacturing growth is not jobless. (p. 32)
It cannot, therefore, be claimed that South African manufacturing growth is jobless. Instead, demand effects on labour demand prove to be relatively strong.
-
Technological change has been capital-augmenting, reducing industry's capacity to expand employment. (p. 14)
What does emerge is that technological change has effectively been capital- rather than labour-augmenting over time, thus decreasing the capacity of SA industry to expand employment.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| TIPS sectoral panel data set (48 three-digit SIC version 5 sectors), employment, remuneration, value added, skills composition, 1970-1997 | official statistical series | 1970-1997 | Core dataset for all descriptive tables (employment, real remuneration, productivity, skills ratio) and the manufacturing panel regressions. |
| TIPS full data set with documentation (availability statement) | official statistical series | 1970-1997 | Establishes that the underlying dataset is available from TIPS with documentation. |
| Manufacturing panel of 28 three-digit SIC version 5 sectors, annual 1970-97 (including real wage, user cost of capital, output) | official statistical series | 1970-1997 | Panel used to estimate the dynamic heterogeneous labour demand function (PMG, MG, DFE). |
| Skills composition of manufacturing labour force (highly skilled, skilled, unskilled employees) for 24 manufacturing sectors | official statistical series | 1970-1997 | Used to compute the skills ratio and show rising skills intensity in manufacturing. |
| Sadie (1993) demographic data, A projection of the South African Population 1991-2011, Research Report 196 | secondary literature compilation | 1970-1997 average | Provides the 2.38 per cent population growth estimate against which employment growth is compared. |
| World Bank Development Indicators (economically active population growth) | international trade or finance data | 1970-1997 | Corroborates that employment growth lagged labour force growth (2.56 per cent p.a.). |
| Fedderke, Shin and Vaze (2000), Trade & Labour Usage, ERSA Working Paper No 15 (econometric evidence) | secondary literature compilation | 1970-1997 | Cited for the claims that trade liberalisation does not explain employment decline and that technological change was capital-augmenting. |
| Fedderke et al (2001), capital market sectoral description 1970-1997 | secondary literature compilation | 1970-1997 | Cited for a parallel structural break in capital markets in the 1990s and for data construction. |
Inference¶
Quantification: Econometric (causal design).
Descriptive exploratory analysis of sectoral panel data at benchmark years (rank orderings, growth rates, cross-sector correlations of employment, real wages and productivity), followed by dynamic heterogeneous panel estimation (Pooled Mean Group) of a log-linear labour demand function for 28 manufacturing sectors, 1970-97.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Weak employment creation is a long-standing structural problem dating to the 1970s, not a product of recent policy regimes, and 1990 marks a structural break with employment contraction across most sectors. (p. 10) | TIPS sectoral employment levels and average growth rates (Tables 1-2), sector counts with negative growth, rank and growth correlations. | Partial data |
| Employment growth across sectors has been below population growth; average sectoral employment growth was 2.16 per cent in 1970-90 and -1.26 per cent in 1990-97. (p. 8) | Unweighted average of sector growth rates from TIPS data, compared with Sadie (1993) and World Bank population/labour force growth. | Partial data |
| Manufacturing skills intensity rose steadily over 1970-97 (average skills ratio 0.78 by 1997), with all sectors but Footwear improving. (p. 11) | TIPS manufacturing skill-category employment for 24 sectors; skills ratio computed in Table 3. | Systematic data |
| A negative association between employment and the real wage is present in the South African labour market; all sectors with negative employment growth show a negative employment-real wage correlation. (p. 19) | Sector-level correlations of employment and real remuneration per labourer (Table 5) from TIPS data. | Partial data |
| Sectors where real wages track labour productivity closely show stronger employment growth and sustained real wage gains. (p. 24) | Table 7 correlations of productivity and real wage grouped by correlation strength and compared with growth in Figure 1. | Partial data |
| Manufacturing labour demand has a long-run wage elasticity of about -0.5 to -0.55, with output elasticity near unity and negative user cost effects. (p. 31) | PMG estimation on the 28-sector manufacturing panel, 1970-97, with 5% and 10% depreciation assumptions (Table 9). | Systematic data |
| South African manufacturing growth is not jobless; demand effects on employment are strong and roughly proportional. (p. 32) | Estimated long-run output coefficient of about 0.86 to 1.00 in PMG regressions. | Systematic data |
The author's caveats¶
The authors note that benchmark years (1970, 1980, 1990, 1997) are used so yearly changes are not shown. Manufacturing is disaggregated far more than services and mining, which disadvantages cross-sector comparison, and no more disaggregated public data exist. Non-manufacturing data are less reliable and consistent, so the panel regressions are limited to manufacturing. Skills data exist only for manufacturing, and the productivity-wage correlation is only a proxy for the marginal product/marginal cost link. The paper is described as preliminary and exploratory.
Gaps in the evidence¶
The paper does not document the original sources behind the TIPS series (for example the censuses or surveys on employment, remuneration, capital stock), so data quality, coverage of informal, homeland and agricultural employment, and deflators cannot be assessed. Causal claims about real wages rest on bivariate correlations and a manufacturing-only panel with no explicit tests for structural breaks, with exogeneity assumed rather than tested. Wage elasticities by skill category and sector, and micro-level firm or worker data (for example the October Household Surveys, firm surveys or wage-setting records on bargaining council extensions) would test the argument better; some of these existed in the 1990s.
Replicability: Data described but not public.