Compulsory Centralization of Collective Bargaining in South Africa¶
Peter Moll (1996). Compulsory Centralization of Collective Bargaining in South Africa. The American Economic Review, Vol. 86, No. 2 (Papers and Proceedings of the 108th Annual Meeting of the American Economic Association). American Economic Association.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | mixed |
| Period studied | 1920s–1994 |
| Regions | South Africa, Witwatersrand |
| Themes | labour markets wages, unions strikes, inequality poverty unemployment, business and state, transition post apartheid |
| Citation key | moll1996 |
Argument¶
Moll argues that the 'ergo omnes' rule, under which the Minister of Labor extends industrial council (IC) wage agreements to all employers in an industry, compulsorily centralizes bargaining. Through a simple model he claims this sets a minimum wage that benefits large capital-intensive firms and unions' insiders, and pushes small and labor-intensive firms out of business or into non-compliance. The result is a large wage gap between covered and uncovered sectors, a barrier to small-firm 'graduation', weaker dynamic efficiency and growth, and higher measured unemployment as displaced workers queue for covered jobs. He tests the model with minimum-wage comparisons, wage trends and industry-level Mincerian regressions, and concludes that abolishing ergo omnes or exempting small firms would raise employment.
Evidence and method¶
Detailed minimum-remuneration data from two ICs (Steel and Engineering; Motor Transport (Goods)). Informal-sector and Witwatersrand domestic-service wages. Wage Determination minima. Four real unskilled wage series (1970s-1990s). October Household Survey 1994 (Central Statistical Service) for Mincerian regressions. An earlier estimate of IC coverage (Moll 1993). A du Toit et al. (1995) sample of 25 small firms. Commentary from industry and union officials. Bendix and Levy.
Place in the debates¶
- Does compulsory centralization of bargaining promote industrial peace, or cause inefficiency and unemployment (insider-outsider effects)? (see Unemployment: causes, measurement and labour market institutions). Moll sides with the second view: ergo omnes creates a 'wage cliff', protects insiders and large firms, depresses small-firm entry and growth, and raises measured unemployment. He grants that the evidence is suggestive, not conclusive.
- Role of unions and wage-setting institutions in South African labor-market dualism and unemployment (see Unemployment: causes, measurement and labour market institutions). He treats institutional wage floors, not only apartheid-era racial barriers, as a source of the covered/uncovered divide that persists after apartheid, shifting the divide from white vs African to insider vs outsider.
Findings¶
-
The IC system covered most manufacturing workers by the mid-1980s. (p. 2)
I have estimated that some 64 percent of workers in the manufacturing sector were covered by IC agreements in 1985 (Moll, 1993).
-
African-dominated unions captured the IC system in the 1980s. (p. 2)
During the 1980's the fast-expanding union movement, dominated by African workers, succeeded in wresting the IC's from white control.
-
The racial divide in labor institutions became an insider/outsider divide. (p. 2)
With an ironic twist, the original wedge between white and African workers transmuted, as we shall see, into a wedge between "insider" or employed workers and "outsider" or unprotected workers.
-
Large covered-sector wage premia over uncovered sectors for unskilled workers. (p. 4)
The covered differentials were considerable: 77 percent for Motor Transport (Goods) and 131 percent for Steel and Engineering.
-
Covered wages rose while uncovered wages fell during a period of declining per capita GDP. (p. 4)
Both series rose between the 1970's and the 1990's, even though GDP per capita fell steadily between 1980 and 1993.
-
Union wage effects are smaller in industries with strong ICs, which suggests that ICs standardize wages. (p. 4)
The mean of the union coefficients in the ergo omnes industries is 0.11 (SE = 0.032), and that in the uncovered industries is 0.23 (SE = 0.027).
-
Small firms report that IC agreements hurt them. (p. 4)
In a sample of 25 small firms studied by Darcy du Toit et al. (1995), 22 stated that IC agreements negatively affected their operations (p. 59).
-
Township microenterprises evade ergo omnes, which limits its employment impact; enforcement is costly. (p. 5)
Microenterprises located in the townships do not comply with ergo omnes because the costs of its enforcement would be too great.
-
Reform of ergo omnes would face strong political opposition. (p. 5)
Abolition of ergo omnes would probably encounter fierce opposition from unions and large firms.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Moll (1993) estimate of Industrial Council coverage in manufacturing, 1985 (Industrial and Labor Relations Review) | secondary literature compilation | 1985 | Supplies the estimate that about 64 percent of manufacturing workers were covered by IC agreements in 1985, establishing the scale of coverage. |
| Minimum-remuneration data from the Steel and Engineering and Motor Transport (Goods) Industrial Councils | employer association records | early/mid 1990s (undated in text) | Gives the covered-sector minimum wages for unskilled workers, which are compared with uncovered-sector pay to measure the covered-uncovered differential. |
| Remuneration data for informal-sector employees and domestic servants (Witwatersrand) | household or labour survey microdata | early/mid 1990s | Provides the uncovered-sector wage benchmark (R643, R815, R656 per month) against which IC minima are compared. |
| Wage Determinations (Wage Board minimum wages) for unskilled workers, minimum-wage data from SALDRU | commission or official report | 1970s-1990s | Series (iv), the median across sectors of Wage Determination minima, is used as an uncovered-sector wage series falling over time. |
| Real wage series for unskilled workers in Moll (1995b): Engineering actual wages and median of IC minima across sectors | official statistical series | 1970s-1990s | Used to show covered-sector real wages rising between the 1970s and 1990s while uncovered-sector wages fell. |
| GDP per capita, 1980-1993 (national accounts) | national accounts macro | 1980-1993 | Establishes that per capita output fell while covered-sector wages rose, supporting the claim of a widening differential. |
| October Household Survey 1994 (microdata), Central Statistical Service | household or labour survey microdata | 1994 | Used for industry-by-industry Mincerian wage regressions with a union dummy to compare union wage effects in IC and non-IC industries. |
| 1980 and 1991 census microdata and 1993 October Household Survey, Central Statistical Service; PSLSD (SALDRU) | population census | 1980, 1991, 1993 | Acknowledged as data supplied to the research program; the text does not say how they are used in the summarized results. |
| du Toit et al. (1995) survey of 25 small firms, 'Industrial Councils and Small Enterprises' | secondary literature compilation | 1995 | Supplies survey evidence that small firms oppose IC agreements and see minimum wages as a problem. |
| Bendix, Industrial Relations in South Africa (Juta) | secondary literature compilation | 1989-1995 editions | Quoted for employer opinion that ICs would be of little use without extension of agreements to non-parties. |
| Levy (1995), World Bank manuscript on South African industrial SMEs | secondary literature compilation | 1995 | Used for the finding that small and medium establishments are underrepresented relative to international norms. |
| Interviews/commentary from industry and union officials | interviews oral history | c. 1995 | Anecdotal support that large firms and unions favor ergo omnes extension and want to eliminate small non-complying firms. |
| Industrial Conciliation Act (1979 amendments) and ergo omnes extension provisions | legislation gazette | 1920s-1979 and after | Background on the legal framework allowing the Minister of Labor to extend IC agreements and Africans to join registered unions. |
Inference¶
Quantification: Descriptive tables.
A simple theoretical model of voluntary versus compulsory (ergo omnes) centralization is tested with indirect evidence: comparison of IC minimum wages with uncovered-sector wages, trends in real wage series during a period of falling GDP per capita, cross-industry Mincerian union-coefficient comparisons, and small-firm survey and interview evidence. The author calls the tests rudimentary and merely suggestive.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| About 64 percent of manufacturing workers were covered by IC agreements in 1985. (p. 326) | Author's earlier estimate (Moll 1993); no method is described here. | Partial data |
| IC minimum wages for unskilled workers are far above uncovered-sector pay (77 percent above in Motor Transport (Goods) and 131 percent above in Steel and Engineering), implying a 'wage cliff' that encourages queuing for covered jobs. (p. 328) | Minimum-remuneration data from two ICs compared with the average of informal-sector and domestic-service wages. | Partial data |
| The covered-uncovered wage differential widened during a decade of negative per capita growth, suggestively linked to ergo omnes. (p. 328) | Four real wage series for unskilled workers in Moll (1995b) set against GDP per capita. | Partial data |
| Union wage effects are smaller in industries with strong ICs (mean coefficient 0.11 against 0.23), implying that ICs standardize wages. (p. 328) | Industry-by-industry Mincerian regressions on the 1994 October Household Survey. | Systematic data |
| Small firms oppose IC activity, while large firms and unions support ergo omnes extension. (p. 328) | du Toit et al. survey of 25 small firms, Bendix, and the author's discussions with commentators and union officials. | Illustrative or anecdotal |
| Abolishing ergo omnes would raise employment at the small-firm end and strengthen growth, with a redistribution from large to small and medium firms. (p. 329) | Theoretical model and the suggestive evidence above; no direct test. | Asserted |
| Compulsory centralization raises measured unemployment and slows growth by blocking small-firm 'graduation', especially for black township firms. (p. 327) | Theoretical model, with partial support from Levy's finding on the underrepresentation of small firms. | Asserted |
The author's caveats¶
The author says the ideal dataset (worker characteristics with union, IC coverage, firm size and capital intensity) does not exist, so the tests are 'simple and admittedly incomplete' and 'do not amount to incontrovertible evidence'. He notes other influences on wages such as rising union density, so ergo omnes cannot be held solely responsible. Compliance by intermediate firms remains an open empirical question.
Gaps in the evidence¶
The argument needs linked worker-firm data covering IC coverage, union status, firm size and capital intensity, to test the model's predictions directly. It also needs measurement of employment and firm-entry effects (for example, firm-size distributions before and after extensions, or variation in extension timing across industries) and of actual compliance by small firms. Some of this could have been built from IC gazette notices and manufacturing census establishment data. The wage comparisons rely on a few ICs and on unweighted averages of heterogeneous uncovered sectors.
Replicability: Data poorly documented.