Industrial Decentralisation in South Africa¶
Trevor Bell (1973). Industrial Decentralisation in South Africa. Oxford University Press, Cape Town.
| Type | book |
| Scope | Economics and economic history of apartheid |
| Method | mixed |
| Period studied | 1910–1970 |
| Regions | South Africa, Bantu areas/Reserves (Transkei, Ciskei, Zululand), Border areas (East London/King William's Town, Natal, Rosslyn, Hammarsdale), Southern Transvaal, Durban-Pinetown, Port Elizabeth-Uitenhage, Western Cape |
| Themes | manufacturing industrial policy, bantustans removals influx, business and state, labour markets wages, demography urbanisation, inequality poverty unemployment, macro fiscal finance debt, methods data sources, politics ideology elections |
| Citation key | bell1973 |
Argument¶
Bell, an economist, argues that the border industries policy is chiefly an instrument of racial separation, not a defensible economic decentralisation policy. He shows that the concentration of manufacturing in four metropolitan areas was driven by market demand (above all gold-mining on the Witwatersrand) and materials, and probably helped growth. Welfare-economics arguments (externalities, congestion, infant areas) do not establish a general economic case for diverting industry to the border areas. Treating the government as a 'buyer' of African employment in the border areas, he builds a 'divertibility index' of 129 sectors showing that marginal excess costs rise steeply as more employment is diverted. Using the Permanent Committee's own figures, he finds that incentives and the 1967 Physical Planning Act produced little induced employment (about 2,200 extra jobs a year) at rising cost, so racial separation is bought at the price of growth.
Evidence and method¶
Official commission reports (Tomlinson, Viljoen, Schumann, Native Economic Commission, Barlow), unpublished Department/Bureau of Statistics industrial census data for 1934-5 to 1962-3, Permanent Committee progress reports, Wage Board reports, McCrystal's survey of 76 firms, Lombard's Transkei income estimates, Barker's clothing-industry study, a 1961 East London questionnaire to firms, and Ackley and Dini's method for Italian incentives applied to South African tax rules. The theory draws on Weber, Lewis, Johnson and Hutchison.
Place in the debates¶
- Economic versus political rationale for border industries/decentralisation (see Homelands, separate development, decentralisation and removals). Rejects the view that decentralisation is justified on economic grounds; the 'overriding' motive is racial separation, and policy should be judged against that aim and its growth costs.
- Border areas versus industrialising inside the Bantu areas (Tomlinson majority versus government) (see Homelands, separate development, decentralisation and removals). Notes the government rejected in-Reserve development for political reasons (avoiding 'white spots') and that border areas are cheaper on infrastructure; presents the border policy as a political compromise.
- Market versus state direction of location (positive inducements versus physical controls) (see Homelands, separate development, decentralisation and removals). Prefers market-based inducements, but argues controls are more effective at limiting African urban growth and cheaper in budget terms. Controls shift costs onto firms and workers and hurt confidence.
- Did regional wage differentials/cheap labour draw industry to the border areas? (see Homelands, separate development, decentralisation and removals). Labour cost is secondary to markets and materials; differentials are real and large (border areas 45-66% of southern Transvaal African earnings) and defensible, and cheap labour is decisive only in a few industries such as clothing and textiles.
- Agglomeration and infant-area arguments for subsidising decentralisation (see Homelands, separate development, decentralisation and removals). Sceptical: agglomeration helped early growth, the Schumann Committee contradicts itself, and there is no evidence that border areas can reach self-sustaining growth.
Findings¶
-
The border industries policy arose as a compromise between in-Reserve industrialisation and continued urban influx. (p. 8)
Thus the border industries policy may be seen as a compromise between internal industrial development of the Reserves and the continuation of the rapid rural-urban movement of Africans
-
Reserve per capita output was tiny relative to the modern sector and the gap widened between 1951 and 1965. (p. 14)
Thus the per capita output of non-Reserve Africans was 4.4 times as great as that of the inhabitants of the Reserves in 1965.
-
Regional concentration of manufacturing was stable for decades and driven by gold-mining demand, probably favouring growth. (p. 31)
The present degree of regional concentration, which goes back to the early thirties, and which was basically due to a highly concentrated demand, has therefore almost certainly been favourable to the country's economic growth.
-
Border and Bantu areas held only about a tenth of manufacturing output in 1956-60 and were dominated by food processing. (p. 45)
Since the corresponding figures for the border and Bantu areas, given in table 9 below, were 10.0 and 10.6 per cent respectively
-
Large wage differentials existed between metropolitan and border areas for African male workers. (p. 133)
the average earnings of male Africans in border and Bantu areas in the Cape (excluding East London) and the Transvaal were 59.8 and 44.8 per cent respectively of the southern Transvaal level in 1956-7
-
Economic arguments for decentralisation are unproven; its aim is racial separation. (p. 123)
The objective overriding all others is racial separation.
-
Border areas offered no escape from the industrial colour bar for African workers. (p. 150)
It appears that even in the border areas the African cannot break through the skill ceiling, and there are limitations on the responsibilities he may shoulder.
-
Divertibility costs rise steeply as the share of diverted employment grows, because mobile light industries such as clothing are quickly exhausted. (p. 188)
Thus the divertibility index (which is directly related to the marginal excess private costs of diversion) rises quite rapidly as the diverted percentage of employment is increased.
-
The heavy industries that are growing fastest are the hardest to divert. (p. 192)
It is worth noting that generally the most rapidly growing sectors are the least easily divertible, that is, have the highest divertibility indices.
-
In 1960-6 only a small fraction of national manufacturing job growth was induced by government intervention. (p. 212)
Thus not more than 5.2 per cent of the national increase in this period was diverted to border areas.
-
Even after licensing and enlarged incentives, induced employment was minimal (about 2,200 jobs a year). (p. 235)
Thus it appears that the maximum increase attributable to government intervention jn this 10| year period was 23,000 or 2,200 additional jobs per annum.
-
Under the Physical Planning Act nearly 30% of requested African labour was refused in 1968-9. (p. 229)
industrialists who had applied for permission to employ 91,468 African workers were allowed 26,736 fewer than they had requested, a refusal rate of 29.2 per cent.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Department/Bureau of Statistics unpublished records (magisterial-district manufacturing statistics, 1956-7 and 1959-60, regional employment, output, wages) | official statistical series | 1934-5 to 1959-60 | Basis for regional concentration tables, border/Bantu area contribution, earnings differentials and stage-of-production analysis. |
| Industrial Census 1956-7, Special Report No. 240 | industrial mining census | 1956-7 | Source of employee numbers and average earnings of African males by border-area district (Table 17) and employment growth. |
| Industrial Census 1959-60, Special Report No. 250 (129 manufacturing sectors) | industrial mining census | 1959-60 | Source of labour cost ratio, non-white share and horse-power per worker used to construct divertibility indices. |
| Industrial Census 1960-1, Special Report No. 257 | industrial mining census | 1960-1 | Employment by sector for ubiquitous industries and cumulative divertible employment. |
| Population Census 1960, Sample Tabulation nos. 1, 2 and 5 | population census | 1960 | Regional distribution of economically active population for localisation coefficients. |
| Statistical Year Book 1964, 1966, 1968; Bulletin of Statistics | official statistical series | 1951-1968 | Manufacturing employment totals, population, rural white and urban Bantu numbers, industrial production indices. |
| Lombard, Stadler and van der Merwe, The Concept of Economic Co-operation in Southern Africa (1968); Lombard on Transkei income | secondary literature compilation | 1960-7 | Land area, population, GDP and per capita GDP of Bantu areas and Transkei, migrant earnings and remittances. |
| Tomlinson Commission Report (Summary), U.G. 61/1955, and White Paper W.P. F/1956 | commission or official report | 1951-6 | Baseline 1951 population, income, migration, carrying capacity and the border versus Bantu areas industrialisation dispute. |
| Viljoen Commission (Protection of Industries), U.G. 36/1958 | commission or official report | 1958 | Cited for costs of concentration, social costs, and recommended positive measures. |
| Schumann Committee Report on Railway Rating Policy and Industrial Location in South Africa (1964) | commission or official report | 1964 | Authority for market and gold-mining demand as location determinants and prediction of continued centralisation. |
| Native Economic Commission (1930-32), U.G. 22/1932; Witwatersrand Mine Natives' Wages Commission U.G. 21/1944; Social and Economic Planning Council Report No. 9, U.G. 32/1946 | commission or official report | 1932-46 | Establish earlier official thinking on Reserve poverty, migrancy and absence of industrialisation proposals. |
| Board of Trade and Industries Report No. 282 (1945), Investigation into Manufacturing Industries | commission or official report | 1945 | Cited on decongestion, future decline of Witwatersrand gold demand and social aspects of concentration. |
| Permanent Committee for the Location of Industry and Development of Border Areas, annual progress reports 1960-70 | commission or official report | 1960-1970 | Source of employment, investment, incentive costs and tax-concession values in border areas. |
| Department of Planning, unpublished information on Physical Planning Act applications; Department's report year ended June 1968 | state archives | 1968-1969 | Applications, African labour requested, allowed and refused under Physical Planning Act. |
| Prime Minister's Statement on establishment of industries in the Border areas, 2 June 1960 | commission or official report | 1960 | Defines border areas (30 miles from Bantu areas) and 1960 incentive package and wage policy. |
| Physical Planning and Utilization of Resources Act No. 88 of 1967; Natives Land Act 1913; Native Trust and Land Act 1936; Wage Act 1925; Bantu Transport Services Act; Income Tax Acts 1962, 1964 | legislation gazette | 1913-1969 | Provide legal framework for land, controls on location, incentives and labour levies. |
| Wage Board reports (clothing industry 1961, stockings 1963, other industries) | commission or official report | 1961-63 | Illustrates wage-setting criteria and regional differentials. |
| McCrystal survey of 76 firms, City, Town or Country (1969) | secondary literature compilation | c.1960s | Bell reanalyses the survey to test claims of locational ignorance and wrongly sited firms. |
| Author's questionnaire survey of East London firms, 1961 | author survey or fieldwork | 1961 | Impression that substantial firms would relocate if possible. |
| Ackley and Dini study of Italian incentives (1959) and Bell's South African adaptation | secondary literature compilation | 1959; 1964-69 | Hypothetical-firm method to compute permissible cost differences from incentives. |
| Press reports (Financial Mail, Natal Mercury, Daily News) and Riekert 1970 address | newspapers periodicals | 1963-1970 | Evidence on wages in Hammarsdale, skill ceilings in border areas, ministerial statements and the 24 per cent employment target. |
New data¶
Bell compiled from unpublished Department of Statistics records a border-and-Bantu-areas manufacturing series (about 100 magisterial districts, 1956-7 and 1959-60, later 1967-8), regional African earnings tables, localisation coefficients and divertibility indices for 129 manufacturing sectors (1959-60).
Inference¶
Quantification: Descriptive tables.
Descriptive statistical comparison of regional concentration, border-area growth and wage differentials, combined with neoclassical welfare-economic argument. A constructed ordinal divertibility index (labour cost ratio, non-white share, horse-power per worker, localisation) yields an illustrative marginal-cost schedule, supplemented by hypothetical-firm calculations of incentives and residual estimates of induced employment.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Regional concentration of manufacturing in four main industrial areas (about 80% of output and employment) stayed essentially constant from 1934-5 to 1959-60, driven by market demand from gold mining and ports. (p. 29) | Department of Statistics regional manufacturing data, Table 5 and Schumann Committee. | Systematic data |
| Border and Bantu areas produced about 10% of gross manufacturing output in 1956-60, dominated by food and processing, with Natal accounting for the majority and the Cape lagging. (p. 47) | Bureau of Statistics magisterial-district data compiled into Tables 9-14, only 1956-7 and 1959-60 available. | Partial data |
| African male earnings in border and Bantu areas were far below the southern Transvaal (roughly 45-66%), but earnings in the main industrial centres were similar. (p. 133) | Industrial Census 1956-7 Special Report 240 and unpublished Department of Statistics data (Tables 15-18). | Systematic data |
| Per capita output of the Bantu areas was far below that of non-Reserve Africans and probably stagnant or falling, with rising dependence on migrant earnings. (p. 14) | Lombard et al. 1965 GDP figures, Tomlinson 1951 data, Leistner, Lombard Transkei accounts. | Partial data |
| Marginal excess private costs of diverting manufacturing employment to border areas rise steeply with the diverted share, since easily divertible industries (clothing, textiles) are few and slow-growing. (p. 188) | Constructed divertibility index from 1959-60 census data, ranking and cumulative 1960-1 employment (Tables 20-30); admittedly simplified, with no actual cost data. | Partial data |
| Only a small fraction (at most about 5% of the national increase in 1960-6, about 23,000 jobs by 1970) of border-area employment growth can be attributed to government intervention. (p. 235) | Permanent Committee employment figures against assumption that 12.2% of national growth would have occurred spontaneously; cross-checked with Department of Statistics 1967-8 data. | Partial data |
| Available incentives offered permissible cost differences of only a few per cent, well below Italian incentives, explaining limited success; physical controls under the Physical Planning Act refused 29% of African labour requested. (p. 229) | Hypothetical-firm calculations following Ackley and Dini; Department of Planning application data (Table 36). | Partial data |
| The economic case for decentralisation (market failure, external effects, infant area) is not established empirically; border industries policy is driven mainly by racial separation. (p. 97) | Theoretical analysis, critique of McCrystal's survey and official statements. | Asserted |
The author's caveats¶
Bell notes that magisterial districts are the smallest unit so border and Bantu areas cannot be separated; border-area data exist only for 1956-7 and 1959-60 (later 1967-8); regional earnings data stop in 1956-7; Department of Statistics disclosure rules hide some sectors; Permanent Committee employment figures are probably overstated; the divertibility index rests on simplifying assumptions (racial composition as proxy for skill, a fixed 40 per cent wage differential, horse-power as proxy for weight, constant marginal costs per sector); external effects are not measured.
Gaps in the evidence¶
The argument lacks firm-level cost data (plant-level comparative cost studies) and actual location decisions to test the divertibility ranking, which depends on proxies. No counterfactual for spontaneous border-area growth exists beyond extrapolating a pre-1960 share, so attribution of induced jobs is crude. Better evidence would be Permanent Committee and IDC project files, firm-level incentive and investment records, and labour productivity comparisons, some of which probably existed in government archives then and are partly accessible now.
Replicability: Data described but not public.