Black Trade Unions and the Wage Gap in South Africa¶
W. S. Siebert (1987). Black Trade Unions and the Wage Gap in South Africa. Managerial and Decision Economics, Vol. 8, No. 1. John Wiley & Sons.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | econometric |
| Period studied | 1947–1984 |
| Regions | South Africa, gold mining, manufacturing |
| Themes | labour markets wages, unions strikes, mining migrant labour, bantustans removals influx, inequality poverty unemployment, methods data sources |
| Citation key | siebert1987a |
Argument¶
Siebert asks why the black-to-white pay ratio in gold mining and manufacturing rose so sharply after 1971. He builds a simultaneous demand-and-supply model of black employment and real wages for the two sectors and estimates it on 1947-84 data. He then decomposes wage changes for 1975-80 and 1980-84 into contributions from gold prices, union density, foreign labour supply, pass-law enforcement, capital and import prices. Rising gold prices and shrinking foreign labour supplies drive mining wages in the 1970s. Black unionization explains most of the real manufacturing wage rise after 1980. Influx-control enforcement depressed black wages in both sectors. The paper predicts small employment losses from unions and larger wage gains if the pass laws are abolished.
Evidence and method¶
Annual 1947-84 series from the Bulletin of Statistics and Statistical Yearbooks: earnings, employment, capital stock, output and prices for gold mining and manufacturing. Also census-based non-farm workforce figures, SAIRR Survey estimates of pass-law prosecutions, union membership (including unregistered unions) from SAIRR, Clack, the Wiehahn Commission and the Department of Manpower, education expenditure, and foreign mine labour (Lipton, SAIRR). Government commission reports (Riekert, Wiehahn) supply institutional detail.
Place in the debates¶
- Causes of the narrowing black-white wage gap after 1971 (gold price vs unions vs labour supply vs influx control) (see Racial wage gaps, wage convergence and wage determination). Multi-causal. Gold prices and falling foreign labour mattered for mining in the 1970s, but unions explain most of the real manufacturing wage rise after 1980. Relaxed influx control also helped. Gold prices alone are not the full answer.
- Are black mine wages demand-determined or set by an excess labour supply (Lucas)? (see Racial wage gaps, wage convergence and wage determination). Treats this as an empirical question. Finds an elastic mining labour supply as Lucas does, but with a supply elasticity above one (1.32) where Lucas found it below one, and keeps demand in the wage system.
- Interpreting strikes as evidence of union power (see Black trade unions: growth, wage effects, Wiehahn and strategy). Sceptical. Strikes reflect unexpected changes in the bargaining environment and the many new bargaining pairs after 1979, so they are not a reliable index of militancy. Union density is used instead.
- Employment costs of black unionization and effects on non-union sectors (see Black trade unions: growth, wage effects, Wiehahn and strategy). Predicts small job losses in manufacturing (about 20,000), but possibly around 100,000 jobs shed across mining and manufacturing. This would put downward pressure on informal-sector and rural wages.
Findings¶
-
The black-white earnings ratio in mining rose sharply after 1971, after falling since the war. (p. 2)
The increase in the case of mining has been from 5% to 19%.
-
The foreign share of the mine workforce collapsed after 1974, reducing labour supply and pushing up mine wages. (p. 3)
there has been a marked reduction in the proportion of the mining labour force made up by foreign blacks-from 70% in 1974 to 30% now.
-
Black unions cannot explain the 1970s wage gains because membership was low in that decade. (p. 2)
According to these estimates the decade of the 1970s was one of quite low union membership.
-
Union density has a significant wage effect in manufacturing after 1974 but none in mining, where unions came late. (p. 9)
Black trade union density is insignificant in mining, as expected, but significant for manufacturing after 1974.
-
Stricter pass-law enforcement lowered black manufacturing wages, as it lowered mining wages. (p. 9)
Pass law prosecutions also appear to reduce manufacturing wages.
-
Union density rise in 1980-84 accounts for the largest part of the predicted manufacturing wage increase. (p. 10)
This gives a log wage increase of 0.155.
-
The rising gold price was the main driver of mining wages in 1975-80 after allowing for consumer prices. (p. 10)
In the period 1975-80 the most important cause of the rise in mining wages, after allowing for the effect of rising consumer prices, is the rise in the gold price.
-
The model fits mining less well than manufacturing, so the mining decomposition is only indicative. (p. 10)
The model is less successful for this sector than for manufacturing.
-
Black real wages rose strongly after 1975 while white real wages were static, narrowing the gap. (p. 11)
Black real wages, on the other hand, have increased by leaps and bounds-by 35% in the case of mining and 17% in the case of manufacturing.
-
Union-induced employment losses in manufacturing are small. (p. 11)
These are rather small employment effects-20000 jobs out of a black manufacturing workforce of 700 000.
-
Abolishing pass laws is predicted to raise black real wages, especially in mining. (p. 11)
The abolition of controls might raise black real wages in manufacturing by 7% and in mining by a larger amount.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Bulletin of Statistics (earnings, employment, pass law prosecutions 1947-50) | official statistical series | 1945-1984 | Source of average annual earnings by colour group (wages bill divided by average monthly employment) and employment series used to construct the black/white wage ratios and the regression variables. |
| South African Statistical Yearbooks | official statistical series | 1947-1984 | Used with the Bulletin of Statistics to build the spliced manufacturing employment series. |
| Population censuses 1946, 1951, 1960, 1970, 1980 | population census | 1946-1980 | Used to construct the non-farm workforce denominator (union density), with geometric interpolation between census years; also occupational shares of black workers (1951 vs 1980). |
| Capital stock series (end value of fixed assets / machinery plus buildings; gross capital formation) | official statistical series | 1947-1984 | Real capital stock series built from benchmark years with perpetual-inventory depreciation, entering the labour demand equations. |
| Output, value of output and implicit product price series (kilos of gold, physical volume index, value of gold sales, value of net output) | official statistical series | 1947-1984 | Used to derive the implicit sectoral product-price deflator (1975=100) used to deflate wages in demand equations. |
| M. Lipton, Men of two worlds (Optima, 1982) and SAIRR Surveys, for foreign blacks in mining | secondary literature compilation | 1947-1984 | Source for the series on foreign black mine workers (including ex-High Commission territories), used as a supply-side variable. |
| SAIRR Annual Survey (Surveys 1975-84; 1979 Survey for Wage Board determinations; 1983 for union estimates) | secondary literature compilation | 1961-1984 | Source for pass law prosecutions, black union membership figures, Wage Board determination dates, and unregistered union estimates. |
| SAIRR (1963) pass law prosecution figures 1950-62 | secondary literature compilation | 1950-1962 | Supplies pass law prosecution figures for 1950-62, spliced with other sources into a consistent series. |
| SAIRR Yearbooks (education expenditure on blacks in white areas) | secondary literature compilation | 1945-1983 | Series of per pupil education expenditure used for the white/black education ratio, later dropped from the regressions as unstable. |
| G. Clack, South African Industrial Relations (PhD thesis, LSE) | secondary literature compilation | 1945-1958 | Source of early black trade union membership figures for 1945, 1950 and 1958. |
| 1984 Department of Manpower Report | commission or official report | 1984 | Source of the 1984 union membership figure and basis for estimating unregistered union membership. |
| Riekert Commission (1979), Report of the Commission of Inquiry into Legislation Affecting the Utilization of Manpower, RP 32 | commission or official report | 1979 | Used for labour bureau and pass law descriptions, farm workers, evasion, contract lengths, definition of pass law offences, and the proposed gold price explanation. |
| Wiehahn Commission (1979), Report of the Commission of Inquiry into Labour Legislation, Part I, RP47 | commission or official report | 1979 | Source for the registration of black unions, the number of black workers in unregistered unions, and the 1978 estimate of unregistered membership. |
| Blacks (Urban Areas) Consolidation Act of 1945, Section 10 | legislation gazette | 1945 | Defines Section 10 rights governing which black workers may live and work in towns, used to describe labour market segmentation. |
| W. S. Siebert (1975), The Relative Pay of White and Non-White Workers in South Africa, unpublished PhD thesis, LSE | author survey or fieldwork | post-war to 1975 | Cited for details on mobility restrictions and Wage Board behaviour in 1961/2. |
| Lucas (1985), Mines and migrants in South Africa, demand specifications and results | secondary literature compilation | 1985 | Demand specifications adopted and supply elasticity results compared against those of Lucas. |
New data¶
The author assembled annual 1947-84 time series for gold mining and manufacturing (black/white earnings, employment, real capital stock, product prices), plus constructed series on pass law prosecutions, black union density including estimated unregistered union membership (interpolated), and white/black education expenditure ratios.
Inference¶
Quantification: Econometric (causal design).
Structural supply and demand equations for black labour in gold mining and manufacturing are estimated on annual 1947-84 data (instrumental variables, AR error terms). The system is then solved to decompose changes in log wages over 1975-80 and 1980-84 into contributions of exogenous variables, including union density, gold price, foreign labour supply and pass law prosecutions.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Black trade union density was the main factor behind the manufacturing real wage rise in 1980-84. (p. 63) | Manufacturing supply equation (Table 2) with post-1974 union density coefficient, multiplied by the rise in log union density (Table 3); union series built from Clack, SAIRR and Wiehahn estimates. | Partial data |
| Rising gold prices were the most important cause of the rise in mining wages in 1975-80, but not the full explanation, since wages kept rising in the 1980s when gold prices were static. (p. 63) | Mining supply and demand estimates and the Table 4 decomposition using the gold price series. | Partial data |
| Relaxation of pass law enforcement raised black wages in both sectors, and abolition would raise manufacturing wages by about 7% and mining wages by more. (p. 63) | Pass law prosecution index (SAIRR, Bulletin of Statistics) with negative coefficients in both supply equations; linear extrapolation. | Partial data |
| Reduced foreign labour supply after 1974 raised mining wages, with a smaller effect in the 1980s. (p. 62) | Foreign mining labour series (Lipton, SAIRR) with a negative coefficient in the mining supply equation (Table 2) and the Table 4 decomposition. | Partial data |
| Union-induced wage increases cost few manufacturing jobs (about 3.4% or roughly 20,000 jobs), though future union effects in mining could shed about 100,000 jobs across both sectors. (p. 64) | Estimated manufacturing demand elasticity (-0.22) applied to the union wage effect; mining projections extrapolate the manufacturing union coefficient. | Illustrative or anecdotal |
| Black trade unions had no impact on mining wages in the period studied, since unionisation there began only in 1982. (p. 62) | Insignificant union coefficient in the mining supply equation (Table 2); historical note on the Black Mine Workers Union. | Partial data |
| The black-to-white pay ratio rose from 1971, from 16% to 26% in manufacturing and from 5% to 19% in mining by 1984. (p. 55) | Ratios of annual earnings from the Bulletin of Statistics, charted in Figures 1 and 2. | Systematic data |
The author's caveats¶
Author says the results should be interpreted cautiously given data and specification problems. Union series quality is too poor for yearly rates of change, though secular trends are reliable. The education expenditure variable was unstable and was dropped. The model is less successful for mining, over- and under-predicting actual wage change. Wage data for agriculture and informal sectors are sparse.
Gaps in the evidence¶
Union density is an interpolated series with estimated unregistered membership, and the post-1974 union effect rests on few observations of a rapidly changing variable, so firm-level or industry-level wage and union panel data would test the claim better. Pass law prosecutions are only a crude proxy for enforcement of influx control, and exogeneity of unionisation is assumed. Mine-level wage and recruitment records from the Chamber of Mines, and the microdata behind the censuses, could test the mining results. These sources existed at the time but were not used.
Replicability: Data described but not public.