Industrial Decentralization in South Africa by Trevor Bell (review)¶
Leo Katzen (1974). Industrial Decentralization in South Africa by Trevor Bell (review). Journal of Southern African Studies, Vol. 1, No. 1. Taylor & Francis, Ltd.
| Type | other |
| Scope | Economics and economic history of apartheid |
| Method | synthesis review |
| Period studied | 1932–1970 |
| Regions | South Africa, Border areas, African reserves/Bantu areas, Southern Transvaal, Natal, Italy (comparison) |
| Themes | manufacturing industrial policy, bantustans removals influx, labour markets wages, business and state, comparative theory |
| Citation key | katzen1974 |
Argument¶
Katzen reviews Trevor Bell's 1973 book, which assesses the Border Industries (decentralization) policy of separate development using cost-benefit and welfare economics, history and cross-country comparison. Bell finds no valid purely economic case for decentralization: industrial location was stable, agglomeration economies outweighed congestion costs, and border areas offered only cheap unskilled African labour. Since the government's real objective was racial separation, the policy is judged by its effectiveness and cost in achieving this. Inducements (1960-66) and controls (the 1967 Physical Planning Act) diverted few jobs at high cost, and the controls mainly pushed employers toward capital-intensive methods. Katzen praises the book as careful, objective and a useful case study in regional and welfare economics.
Evidence and method¶
This is a review, so it reports Bell's evidence: location data to 1960, Bell's own sample survey of businessmen's evaluation of sites, an index of the divertibility of 129 industrial sectors, employment diversion statistics for 1960-66 and 1969-70, cost estimates for 1960-69, and the Italian comparison.
Place in the debates¶
- Economic rationale for industrial decentralization / border industries (see Homelands, separate development, decentralisation and removals). Reporting Bell, the review holds that neither market imperfections nor external diseconomies justify decentralization in South Africa; agglomeration economies are more likely, and any case must rest on socio-political grounds.
- Effectiveness of the apartheid decentralization policy and its link to capital intensity (see Homelands, separate development, decentralisation and removals). The review reports that the policy was costly and ineffective at diverting jobs, and that controls over African employment encouraged more capital-intensive production.
Findings¶
-
Bell traces the Border Industries policy to the Native Economic Commission and the Tomlinson Report; the government chose border rather than reserve industry, institutionalised in the Permanent Committee in 1960. (p. 2)
This policy was embodied in the establishment of the Permanent Committee for the Location of Industry and the Development of Border Areas in mid 1960.
-
Industrial location was highly concentrated and stable before 1960. (p. 3)
the pattern of location remained virtually unaltered in the 25 years preceding 1960. 80 per cent of all industry was concentrated in the four main industrial areas
-
Bell's survey found more locational mistakes in rural and border sites, and he doubts the external diseconomies of the main industrial areas. (p. 3)
His own sample survey of business men's evaluation of the sites chosen shows a larger proportion thought that mistakes had been made in siting in rural and border areas than elsewhere.
-
The government's objective function is racial separation, so decentralization is judged by its effectiveness in achieving it. (p. 3)
In the South African case the 'objective function' of the government (to use the jargon of welfare economics) is racial separation
-
The border areas' main locational advantage is cheap unskilled African labour, which makes the labour share of costs the crucial variable. (p. 3)
the only locational advantage of border areas, apart from certain raw material sources, is the relatively cheap price of unskilled African labour.
-
Policy based on inducements in 1960-66 diverted only a small share of national employment growth, and less than a fifth of the average annual entry of Africans into secondary industry. (p. 4)
5.2 per cent of the national increase in employment or 18,55 3 new jobs were diverted to border areas over this period
-
The 1967 Physical Planning Act gave the Minister power to control African employment in factories in 'controlled areas'. (p. 4)
gave the Minister of Planning the power to control the employment of Africans in factories in 'controlled areas'.
-
The Act did not boost border-area employment; its main effect was to push employers toward capital-intensive methods. (p. 4)
The main effect of the Act seems to have been to encourage employers to reduce African labour by more capital intensive methods of production.
-
Decentralization was costly per job created, and ignored costs of lost output. (p. 4)
From 1960 to the end of 1969 the costs of decentralization were R34.5 million working out at R 1500 per worker
-
Comparable cost and ineffectiveness are seen elsewhere, e.g. Italy's policy for the south. (p. 4)
South Africa is not alone in the high cost and comparative ineffectiveness of measures to divert industry. In Italy
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Native Economic Commission report, 1932 | commission or official report | 1932 | Cited as the earliest origin of the Border Industries idea, through its proposals to improve and develop reserve agriculture. |
| Tomlinson Commission Report, 1956 | commission or official report | 1956 | Establishes the policy background: it favoured industrial development in the reserves to raise carrying capacity and stem African flow to white areas, which the government rejected in favour of border industries. |
| Permanent Committee for the Location of Industry and the Development of Border Areas (establishment 1960, and its assessments of results) | commission or official report | 1960-1966 | Institutional policy framework; the Committee's judgement that results were unsatisfactory led to demands for wider controls. |
| Physical Planning and Utilization of Resources Act No. 88 of 1967 | legislation gazette | 1967 | The Act gave the Minister of Planning power to control African employment in factories in controlled areas, and its effects are assessed in Chapter 10. |
| Industrial location statistics for South Africa and border areas to 1960 (industry distribution across the four main industrial areas and border/Bantu areas) | official statistical series | c.1935-1960 | Establishes that the pattern of industrial location remained virtually unchanged for 25 years before 1960, with 80 per cent in four main areas. |
| Bell's sample survey of businessmen's evaluation of chosen industrial sites | author survey or fieldwork | c.1960s-1970s (not stated) | Tests whether entrepreneurs make poor location decisions; finds more perceived siting mistakes in rural and border areas. |
| Industry-sector data for 129 industrial sectors (potential materials used, horse-power per worker, African labour proportion and share in costs), used to construct a divertibility index | industrial mining census | not stated | Used to construct an index of the divertibility of industries to border areas, ranking industries by marginal excess private cost of diversion. |
| Employment and job-diversion data for border areas 1960-66 and 1969-70, including cost of decentralization concessions to 1969 | official statistical series | 1960-1970 | Used to estimate the number of jobs diverted, the border areas' share of national employment increase, and the cost per job. |
| Cross-country comparison data (Italy's regional policy for the south) | secondary literature compilation | post-war (not stated) | Comparative evidence that costly and largely ineffective decentralization measures are not unique to South Africa. |
New data¶
Per the review, Bell conducted his own sample survey of businessmen's evaluations of siting decisions and built an index of divertibility across 129 industrial sectors, and he estimated the jobs diverted and costs of decentralization for 1960-69. Sample size and coverage are not given.
Inference¶
Quantification: Statistical analysis.
The review describes a mix of cost-benefit and welfare-economic analysis, historical policy tracing, descriptive location statistics, a constructed sector divertibility index, counterfactual estimation of jobs diverted (assuming at least the 1959/60 rate would have occurred anyway), a business survey, and cross-country comparison with Italy.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| The pattern of industrial location in South Africa remained virtually unchanged in the 25 years before 1960, dominated by four main industrial areas, so left alone little change should be expected. (p. 131) | Descriptive location statistics for industry by area to 1960; the source is not specified in the review. | Partial data |
| No economic case for decentralization is valid in South Africa; businessmen's survey responses suggest more siting mistakes in rural and border areas, and evidence points to agglomeration economies. (p. 131) | Bell's own sample survey of businessmen plus theoretical argument on market imperfections and external diseconomies. | Partial data |
| The divertibility of industry to border areas varies widely, with low diversion costs in industries like clothing and high costs in heavy industries like iron and steel. (p. 132) | Bell's constructed divertibility index across 129 sectors, based on materials use and horse-power per worker. | Systematic data |
| Positive inducements in 1960-66 diverted only about 5.2 per cent of the national employment increase (18,553 jobs) to border areas, under one-fifth of average annual African entry into secondary industry. (p. 132) | Employment statistics for border areas against a counterfactual assuming at least the 1959/60 rate of job location. | Partial data |
| The 1967 Physical Planning Act did not boost border-area employment (14.2 per cent of the national increase in 1969-70 against 17.4 per cent in 1960-66) and mainly encouraged more capital-intensive methods. (p. 132) | Comparison of border-area shares of national employment increase across periods, plus data on permissions granted to employers. | Partial data |
| Decentralization cost R34.5 million from 1960 to 1969, about R1500 per induced job on 23,000 extra jobs, excluding lost output and investment costs. (p. 132) | Cost totals of concessions and inducements divided by estimated policy-induced jobs. | Partial data |
The author's caveats¶
The review reports no data caveats from Bell beyond noting the assumption behind the jobs-diverted estimate and that the cost figures ignore lost output and investment in controlled areas.
Gaps in the evidence¶
This is a book review, so the underlying sources, survey design and sample size are not specified. The counterfactual of jobs that would have gone to border areas anyway rests on a single 1959/60 baseline, and firm-level employment and investment records or the industrial censuses would test it better. Such data existed then in the Census of Manufacturing and the Permanent Committee's files, and could now be used for firm-level panel analysis.
Replicability: Data poorly documented.