Industrialisation, protection and industrial structure¶
What drove South African industrialisation: tariffs and ISI, the minerals-energy complex, state corporations or business groups, and with what effects on efficiency?
Works engaging the debate: 30
Positions¶
- ISI and protection were central but ultimately inefficient (Moll, Feinstein)
- Minerals-energy complex as the core of accumulation (Fine and Rustomjee)
- MEC overstated; manufacturers' own capabilities and gradual pre-1925 growth (Bell and Farrell, Clark)
- State corporations integrated with private capital rather than challenging it; developmental-state readings
Evidence base¶
The debate rests mainly on official aggregate statistics, reworked by individual authors into long sectoral series. For the interwar tariff question the core sources are the Census of Industrial Establishments 1915/16-1941/42, the Annual Statements of Trade and Shipping 1910-1939, the Official Year Books and Union Statistics for Fifty Years. Lumby (1974) deflates net manufacturing output and builds industry case series for footwear, printing and cement, using import shares, firm-size classes and race-specific employment. Schirmer (2008) uses the Year Book series on 20 'principal industries' 1918-29, about £176m net output and 60,000 workers, and sets it against tariff-level changes. Marais (1981) builds his own constant-price (1956/57) output, import and export series by industrial group for 1916/17-1975, drawing heavily on T.A. du Plessis's unpublished UNISA constant-price thesis for 1916/17-1956/57. The same thesis also underpins (Bell (1995)), (Bell and Madula (2001)) and the pre-1948 GDP figures in Moll (1991b). For the post-war period the workhorses are Central Statistical Services manufacturing censuses and national accounts. McCarthy (1988) uses them for light/heavy shares 1925-82 and employment elasticities 1970-82, and National Productivity Institute capital-labour and multifactor productivity series for 1970-86. Other key sources are SARB Quarterly Bulletin series on investment, the balance of payments and capacity utilisation (Bell (1995)) (Bell and Madula (2001)), the WEFA 3-digit database (Bell and Madula (2001)), and the IDC 76-sector capital-stock and investment database (Kaplinsky (1995)). Fedderke and Szalontai (2003) uses Census of Manufacturing cumulative-output (Lorenz) tables for 1972-96, including unpublished tables for 1988-96, to compute Gini and Rosenbluth indices for 24 sectors. International benchmarking draws on World Bank World Tables, Summers-Heston and UNCTAD manufactured-export shares 1955-85, together with Maddison's TFP figures (Moll (1991b)) (Moll (1991d)), UNIDO data for 44 countries (Kaplinsky (1995)), and Page and Nishimizu-Robinson TFP estimates (Bell (1995)). The MEC thesis rests above all on reclassified 1984 and 1988 Input-Output Tables, plus Minerals Bureau, Customs and Census on Mining data (Fine and Rustomjee (1996)).
Policy and institutional evidence comes from commission and Board of Trade and Industries reports: Cullinan 1912, Holloway 1934-35, BTI Report 282 of 1945, Viljoen, Reynders RP 69/1972 and Kleu (Lumby (1974)) (Marais (1981)) (McCarthy (1988)) (Fine and Rustomjee (1996)). Houghton and Dagut (1973) anthologises these sources and Hansard. Genuinely archival work is rare. Clark (1987) works through SABA files of Mines and Industries, the Electricity Control Board, Railways, Labour and Commerce, along with Iscor board minutes, Escom and Iscor annual reports and Standard Bank inspection reports. Bozzoli (1975) and Schirmer (2008) mine manufacturers' trade journals: SA Commerce and Manufacturers' Record, the SA Journal of Industries and Industrial South Africa. Work on business groups rests on McGregor's Who Owns Whom and Financial Mail Top 100 rankings for 1979-99 (Nishiura (2006)) (Nishiura (n.d.)); Fine and Rustomjee (1996) uses the same type of ownership compilations. Kaplan (1983) draws on the annual reports of about 250 JSE companies plus 18 management interviews, and Kaplan (1989) on about 30 confidential interviews and Post Office correspondence. Synthetic essays such as Freund (2009), (Freund (2012)), (Iliffe (1999)), (Schneider (2000)) and Nattrass and Seekings (2010b) largely recycle these sources. Note that moll1991a/b/c and nishiura2006/nishiurand are duplicate records of single works.
Contested evidence¶
The tariff question is argued from overlapping census data with different methods and periods. Lumby (1974) finds gold and the post-1932 devaluation the larger stimulus, citing real net output for 1915/16-1938/39 and the growing dependence on imported raw materials paid for by gold exports. He allows only a presumptive tariff effect in footwear and cement, where import shares fell after the duty rises of 1930-32. Marais (1981) attributes the 1926-36 and 1960-69 spurts to tariffs and 1949-60 to import control, but only by matching growth-rate timing to policy dates. Schirmer (2008) uses Year Book data to argue that growth was steady through the 1920s, with tariff/value-added correlations below 0.5. That test is a bivariate correlation on roughly ten annual observations with no deflators, lags or effective protection rates. Iliffe (1999) repeats the low-tariff verdict only at second hand. On industrial sequence, (Marais (1981)) and McCarthy (1988) read import/output and value-added shares as a classic ISI shift from consumer to intermediate and capital goods. Fine and Rustomjee (1996) reads similar import/value-added ratios as showing that heavy MEC-linked sectors industrialised early, which in their view makes ISI a misleading frame.
On efficiency the sides partly rest on incompatible productivity measures. Moll (1991b) reports near-zero TFP for 1950-73, the lowest among comparators, from his own thesis growth accounting set against Maddison. He also adjusts pre-1948 GDP using Chamber of Mines ore-milled data to argue that post-1948 growth was no faster. Sources of Growth in the... (1990), from Nattrass's series and Kuznets comparisons, instead calls growth of about 5 per cent a year relatively rapid. McCarthy (1988) uses NPI series showing falling multifactor productivity. Bell (1995) counters with Wright's capacity-utilisation-adjusted TFP and cross-country estimates, treating measured TFP decline as a by-product of macro and foreign-exchange shocks. Concentration is similarly disputed. Fedderke and Szalontai (2003) links rising census-based concentration to lower productivity and employment, whereas Bell (1995) notes that the ISP presented no evidence on concentration. Kaplinsky (1995) and Fine and Rustomjee (1996) agree that synfuels and basic metals dominate the capital stock but draw opposite lessons. Kaplinsky sees underinvestment in labour-intensive sectors, while Fine and Rustomjee see MEC-centred accumulation.
For state corporations, (Clark (1987)) uses archival evidence of Escom's dependence on the VFPC (almost 80 per cent of 1932 sales went to redistributors) and of Iscor's cartel pricing. Richards's series shows structural sections rising from £6.11s.8d in 1934 to over £9 after the 1936 European cartel deal. Schneider (2000) instead calls Iscor a success on the strength of Norval's official assessment, and Freund (2012) reads parastatals developmentally through IDC house histories. The assertion in (ed.) et al. (1987) of a two-tier Iscor price system rests on an interested trade association bulletin. Figures for group concentration converge: over 80 per cent of the JSE in 1985-95 (Nishiura (2006)), 82.5 per cent in 1988 (Fine and Rustomjee (1996)) and 83 per cent in 1987 (Iliffe (1999)). The disagreement is over meaning. Nishiura's claim that the groups contributed little to industrialisation, because they grew by acquisition, is not backed by output or investment data.
Underused and missing sources¶
Industry-level effective protection rates are the most conspicuous missing evidence. They could be computed by combining tariff schedules with census input-output data, as the evidence ledgers for Lumby (1974) and McCarthy (1988) note. Six works cite Board of Trade and Industries reports, yet the unpublished BTI application files and Department of Census tabulations that Lumby (1974) touched have not been systematically exploited. Only (Clark (1987)) uses firm-level archival accounts for state corporations. Iscor, Escom and IDC financial records on costs, wage bills by race and investment over time would settle the dispute over integration versus developmental leadership, and the efficiency question for parastatals. Establishment-level manufacturing census microdata are not used anywhere. Such data would allow Herfindahl indices (Fedderke and Szalontai (2003)), tests of the capital-intensity mechanism (Kaplinsky (1995)) and plant-level productivity estimates. Input-output tables for years other than 1984 and 1988 would let the MEC claim of rising dependence be tested over time (Fine and Rustomjee (1996)). Reserve Bank exchange-control approval records would test the drivers of conglomerate diversification and outward investment (Kaplan (1983)) (Nishiura (2006)), as would firm balance sheets and Registrar of Companies filings.
Acquiring the Year Books and Union Statistics would allow Schirmer's tariff test to be re-run with deflators and import data.
Strength of key claims¶
| Basis | Share of key claims |
|---|---|
| Systematic data | 16% |
| Partial data | 52% |
| Illustrative or anecdotal | 20% |
| Asserted | 12% |
Comparative figures for all debates are on the Literature overview.
Best-evidenced works¶
Lumby (1974), Clark (1987), Fine and Rustomjee (1996), Fedderke and Szalontai (2003), Moll (1991b), Kaplinsky (1995)
Most-used primary sources¶
Holdings are matched by machine against the archive's file manifest; a dash means no match was found, not that the project certainly lacks the source.
Works and their positions¶
| Work | Position |
|---|---|
| (ed.) et al. (1987) | Mostly pro-privatisation and pro-deregulation, but Thomas is sceptical. He argues that privatisation is no panacea and that efficiency gains can come without it; Macaskill and Nel are ideological free-marketeers. |
| (ed.) et al. (1987) | Thomas argues that fear of white unemployment and right-wing opposition in the civil service explains the toned-down 1987 White Paper. |
| (ed.) et al. (1987) | Bosomworth and Cairncross argue that protection and two-tier pricing hurt secondary industry; deregulation should come first. |
| Bell (1995) | Sceptical of liberalisation. Its macro costs may outweigh efficiency gains, export growth has limits, and effective import substitution may still be needed. |
| Bell (1995) | Argues the ISP's evidence is unpublished or opinion-based. Cites Helleiner's finding that concentration was positively associated with productivity growth in some countries. |
| Bell and Madula (2001) | Questions the view that Sasol II and III distorted allocation and slowed MVA growth; they raised the overall investment/GDP ratio rather than reallocating a fixed pool. |
| Bell and Madula (2001) | Argues manufacturing remains vital because services growth in South Africa came from job losses elsewhere, and growth cannot revive without durable-goods recovery. |
| Bell and Madula (2001) | Suggests natural-resource dependence may now be 'more of a blessing and less of a curse', following Wood and Mayer, while still needing diversification. |
| Bozzoli (1975) | Rejects the conventional view. Manufacturers existed before 1925 and lobbied actively for protection, and Pact protection was less in conflict with imperial interests than its rhetoric suggested. |
| Clark (1987) | Rejects the conflict thesis (Bozzoli, Davies et al., Adam and Giliomee). Argues the corporations were integrated with foreign and mining capital and partnered with it. |
| Clark (1987) | The state corporations were neither challengers nor mere 'tools' of private enterprise, but a growing link between state and private sector, and they did broaden the industrial base. |
| Fedderke and Szalontai (2003) | Finds concentration does not raise efficiency. It is linked to lower output growth, lower labour productivity and higher unit labour costs, which the authors attribute to managerial objectives pursued at the expense of efficiency and profitability. |
| Fedderke and Szalontai (2003) | Argues the strongest returns come from keeping the number of firms in a market up (even to 5-10), rather than from regulating market share. |
| Fine and Rustomjee (1996) | Rejects state autonomy and the state/market dichotomy. State and market are forms through which class interests operate, and privatisation is a form of state intervention. |
| Fine and Rustomjee (1996) | Argues that ISI from consumer to capital goods is a false description. Industrialisation ran forward from MEC core activities, and heavy sectors were developed early. |
| Fine and Rustomjee (1996) | Contends that the division is statistical and arbitrary (transfer pricing, SIC classification). The MEC's GDP share and export dependence persist. |
| Fine and Rustomjee (1996) | Concentration is high and competition law never challenged large capital. They doubt that unbundling would by itself improve performance. |
| Freund (2009) | Sides with the MEC's linkage and institutional approach (Fine and Rustomjee, building on Christie and Clark) over the primary/secondary division. He regards this division as inadequate. |
| Freund (2012) | Yes for the 1940s and in part for the 1950s-60s, using the Woo-Cumings and Leftwich frameworks. It then decayed into a security-driven, inward-looking state after 1970. |
| Freund (2012) | Follows Grundy and O'Meara: military procurement benefited heavy industry, but there was no real joint strategic planning, and civilian spillover was weak. |
| Hilsenrath and Pogue (2012) | Argues that South Africa is a mixed case. Gold-cyanide and Sasol generated knowledge spillovers and industrial capacity, but the gold industry rested on exploitative labour and post-1994 reinvestment of rents has been poor. |
| Hilsenrath and Pogue (2012) | Concedes oil-from-coal is normally uncompetitive and was subsidised, but argues that opportunity cost is low in a capital-abundant, high-unemployment economy. The case is judged 'not clear' rather than clearly inefficient. |
| Hilsenrath and Pogue (2012) | Follows Kubicek in seeing cooperation in operations alongside competition for finance. Stresses Marshallian 'knowledge in the air', the fixed gold price, and COMSA coordination. |
| Hilsenrath and Pogue (2012) | Suggests such sympathies may have helped but notes the patents went to Anglovaal, a Jewish-owned, non-Afrikaner firm. |
| Houghton and Dagut (1973) | Anthology presents both sides: the Mills-Clay-Martin free-trade case and Hofmeyr's agricultural-first view against Major Richards and the ISCOR/protectionist case; the Van Eck report later criticises indiscriminate protection. |
| Iliffe (1999) | Notes that the liberal-radical consensus on rapid success is now disputed. He sees industrialization as gradual, complex, incomplete and traumatic in cost, and rejects take-off points and single-factor explanations. |
| Iliffe (1999) | Reports that research has qualified the radical view: tariffs were generally low and the parastatals collaborated with private capital rather than challenging it. |
| Kaplan (1983) | Argues South African DFI departs from the Argentine/Indian pattern (small-scale, regional, minority joint ventures, adapted technology), so generalizations from those studies need re-appraisal. |
| Kaplan (1983) | Treats monopoly concentration, rooted in the mining houses, as the key characteristic of South African capitalism and the main driver of overseas expansion. |
| Kaplan (1983) | Contends that dependency is not yet properly specified and that many South African firms develop and internationally capitalize 'know-how', though R&D is often located abroad. |
| Kaplan (1983) | Finds loose, decentralized integration with little intra-firm trade and head-office functions; partly accepts but also treats as ideological the firms' claims of autonomy. |
| Kaplan (1989) | Argues the military's role in establishing local production, through the Defence Resources Board and the Electronics Committee, is omitted in the literature, and criticises Frankel for ignoring it. |
| Kaplan (1989) | Held effective under slow electro-mechanical technology but ineffective after digitalisation, which raised capital intensity, cut employment, made the local market too small and weakened design capability. |
| Kaplan (1989) | Supports Hobday's claim that multinationals will transfer core technology to keep markets; sees the Siemens case in South Africa as further evidence. |
| Kaplinsky (1995) | Argues the crowding-out hypothesis is difficult to sustain. The evidence points to underinvestment in labour-intensive sectors, not overinvestment in capital-intensive ones. |
| Kaplinsky (1995) | Acknowledges the MEC argument and agrees the MEC's heavy industries explain much of the high aggregate capital intensity, while using the conventional manufacturing categories. |
| Kaplinsky (1995) | Argues against the assumption that state involvement crowds out investment. In the 1980s investment held up only where the state played a central role, so the state may need to crowd in investment. |
| Lumby (1974) | Distinguishes incidental from deliberate and ad hoc from systematic protection. Concludes there was no definite policy before 1924-25 apart from the Transvaal concessions policy (1881-1900). Closer to De Kock, Botha and Pearsall than to Bruwer and Knowles, who date protection to the 1880s. |
| Lumby (1974) | Tariff important after 1925 but secondary. The mineral discoveries gave the original stimulus (indirectly, via market growth). The greatest stimulus was the gold boom after the 1932 devaluation. The First World War also mattered but is exaggerated (inflation). Rejects the official claim that industrial revolution began in 1925. |
| Lumby (1974) | Accepts Schumann's warning against post hoc reasoning and criticises Rupert, van Biljon and the Board of Trade for asserting causation without proof. Concludes there is only a presumption in favour of a causal link, based on industry case studies. |
| Lumby (1974) | Multi-causal: employment of 'civilised labour' and poor whites, diversification and reduced import dependence, replacement of the 'wasting' gold mines, and Hertzog's economic nationalism. Judged by these aims, the policy was only partially successful. |
| Lumby (1974) | Argues dumping legislation was arbitrary, poorly framed and sometimes harmful to consumers, invoking Haberler that only spasmodic dumping injures the importing country. Notes the hypocrisy of tolerating internal cement dumping. |
| Malherbe and Segal (2001a) | Argues that autarky and protection fostered market concentration, management complacency, falling competitiveness and stagnant labour productivity. Concedes that protectionism built a diversified industrial base unlike any other in sub-Saharan Africa. |
| Marais (1981) | Credits import replacement with building the existing industrial structure. Expects its future scope to be limited, so growth must shift to manufactured exports, especially capital-intensive protected industries that become 'forced exporters'. |
| Marais (1981) | Favours capital-intensive mineral processors on comparative-advantage grounds, arguing capital deepening may raise productivity with a lag. Concedes manufacturing alone cannot solve black unemployment, which requires training and removal of labour-market discrimination. |
| Marais (1981) | Adopts the Schumann/Du Plessis stage framework, dating the shift to a manufacturing stage to after 1925 (or around 1939). Treats gold mining as having suppressed manufacturing's share early on. |
| McCarthy (1988) | Holds that import substitution has run its course and has not lowered the import coefficient. Exports are the orthodox alternative, but a massive export re-orientation is doubtful under sanctions, so he favours a dual approach of selective protection plus export incentives. |
| McCarthy (1988) | Sceptical. Employment elasticities are mostly below one, heavy industries such as chemicals and metals created as many jobs as light industry, and manufacturing growth alone is no 'magic wand' for unemployment. |
| McCarthy (1988) | Rejects the 'maturity' reading. Since the mid-1970s the expected S-curve relationship between manufacturing share and income has broken down, and South Africa is a developing economy at risk of slipping down the income ladder. |
| McCarthy (1988) | Accepts the case for deliberate industrial policy (including selective, 'made-to-measure' export assistance) to shift resources to higher-productivity uses. |
| Moll (1991b) | Faults inefficient import-substituting protection and the neglect of labour-intensive manufactured exports. He also argues that East Asian-style directed state intervention, not free markets, is the lesson. |
| Moll (1991c) | Argues that an inefficient import-substitution model, with no export push, constrained growth. He also stresses a recurring foreign exchange constraint. |
| Moll (1991c) | Rejects a simple free-market prescription. He argues that the quality of state intervention is what matters, citing the East Asian 'late industrialisation' paradigm. |
| Moll (1991d) | The critique of apartheid does not imply that free markets are preferable. The East Asian experience suggests directed state intervention is essential, so the quality and intentions of intervention are what matter. |
| Moll (1991d) | Sceptical, citing the fast growth of most developing countries after 1945 in an opening world economy, while stressing that state capability is crucial. |
| Nattrass and Seekings (2010b) | Notes industrialisation mixed ISI with forward linkages from mining, and sides with Bell and Farrell that the MEC is overstated; manufacturing was protected and inefficient. |
| Nishiura (2006) | Mentioned but not engaged. The paper focuses on business groups, and stresses that apartheid-linked exchange control and sanctions shaped diversification. |
| Nishiura (2006) | Accepts that mining and energy drove industrialization (Fine and Rustomjee) but argues groups' own contribution to manufacturing was limited and mostly through acquisition. |
| Nishiura (2006) | Partly applies it: the 'imperfection' was the restriction on investing abroad caused by exchange control. Group formation within a sector is not explained by it. |
| Nishiura (n.d.) | Cited as background. The paper does not take a side but locates business groups as an under-studied actor in industrialization and finds their direct contribution limited. |
| Nishiura (n.d.) | Accepts the mining and energy origin of industrial capital and shows mining-house groups (AAC, Anglovaal) diversifying into manufacturing from the 1940s and 1960s. |
| Nishiura (n.d.) | Partly adapts it. Group formation within a sector shows no consistent pattern, but diversification reflects exchange control and sanctions, which acted as a form of market imperfection. |
| Petterson (1951) | Treats gold as the foundation of the industrial nucleus (rail, steel, heat and power industries), while noting that since the First World War gold no longer dominates the industrial economy entirely. |
| Schirmer (2008) | Rejects dependence on tariffs. Argues the tariff-value added correlation is weak and that supply-side entrepreneurial capabilities mattered at least as much as protection. |
| Schirmer (2008) | Disputes this. Output growth was 8.76% p.a. in 1916-24 and 6.56% in 1924-29, which suggests continuity rather than a break. |
| Schirmer (2008) | Sceptical of the capture thesis (Yudelman) and of dependence on mining. Stresses Cape roots and internal dynamics while acknowledging Witwatersrand mining demand and the WWI boost. |
| Schneider (2000) | Argues that protection behind tariffs and efficient public investment (ISCOR, IDC) produced robust industrialization, against World Bank-style conventional wisdom. |
| Sender (1994) | Sender holds that Korea and Japan show institutions can be built through struggle and learning while intervening. He rejects Nattrass's claim that bureaucratic capability must come first, and rejects the Krueger-style rent-seeking view. |
| Sources of Growth in the... (1990) | Notes mixed consequences. Tariff-protected import substitution limited scale economies and raised mining costs, and the arms embargo prompted costly self-sufficiency. |