Trends in Industrial Decentralization: An Examination of Bell's Hypothesis¶
R. Tomlinson; M. Addleson (1986). Trends in Industrial Decentralization: An Examination of Bell's Hypothesis. South African Journal of Economics, 54(4).
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | mixed |
| Period studied | 1965–1985 |
| Regions | Pretoria-Witwatersrand-Vereeniging (PWV), Ciskei, Transkei, Homelands, East London, Durban and Cape Town metropolitan regions |
| Themes | manufacturing industrial policy, bantustans removals influx, labour markets wages, business and state, demography urbanisation |
| Citation key | tomlinson1986 |
Argument¶
The authors test R. T. Bell's claim that industrial decentralization since the mid-1960s was a spontaneous response to foreign competition, with firms moving to the border areas for cheaper labour, and not a result of state policy. They accept that Bell has found a real trend in the shift of manufacturing employment away from the main industrial areas. They argue that his explanation is 'not proven'. Their grounds are ambiguities in the data (coarse spatial categories, the textile case, possible under-reporting of PWV employment) and their own survey of decentralized industrialists, which ranks the state incentives above cheap labour. They conclude that the truth lies between Bell's account and their own. They also propose a three-way classification of core, deconcentration and decentralization areas.
Evidence and method¶
Bell's (1983, 1984, 1985) manufacturing employment series, reproduced in Tables 1-3, and Pretorius et al. (1985) data on enterprise location. Decentralization Board annual reports (Table 4 and the sector breakdown for 1984/85), the IDC 1982 Annual Report, and the authors' survey of industrialists in the homelands (Addleson et al., 1985). Also their interviews with foreign textile managers in the Ciskei and Transkei, Ciskei case material (Tomlinson, 1986), Krieg's survey of PWV industrialists, and secondary literature on export processing zones and the Environment Planning Act.
Place in the debates¶
- Spontaneous market-driven decentralization versus policy-induced decentralization (Bell's hypothesis) (see Homelands, separate development, decentralisation and removals). Partly sceptical of Bell. They accept that a trend exists and that much of the employment shift cannot be counted as policy-induced jobs. They hold that his case is not proven and that incentives, the Environment Planning Act and labour costs together shaped the outcome. For firms locating in the homelands they would reverse his ranking, putting policy first and labour second.
- Whether decentralization policy was ineffective and a needless burden on the fiscus (see Homelands, separate development, decentralisation and removals). Bell's implication is that the policy was ineffective and the concessions a needless cost. The authors question this, because the post-1982 incentives now direct location decisions, though they concede that pre-1982 decentralization largely occurred without them.
- Purpose of decentralization: discipline of labour and industry, or a response to the high price of labour relative to productivity (see Homelands, separate development, decentralisation and removals). They note Bell's labour-discipline argument and question its narrowness. They argue the state is responding to high labour costs by relaxing influx control, promoting 'orderly urbanization' just inside homeland borders and favouring deconcentration points.
Findings¶
-
Official decentralization produced few jobs before the 1980s, far below Tomlinson Commission targets. (p. 1)
only 34 900 decentralized jobs mere claimed by the Decentralized Board (Maasdorp, 1982)
-
Non-metropolitan employment growth exceeded the jobs attributed to the ID programme, so Bell identified a real trend. (p. 1)
employment in the 'rest of South Africa's*(4) increased by 124 862 between 1965/66 and 1976
-
The main industrial areas' share of manufacturing employment fell between 1965/66 and 1979. (p. 1)
The relative balance has shifted from 81, 9 and 18,1 per cent in 1965/66 to 76, 9 and 23,1 per cent in 1979.
-
Decentralized job creation surged after the 1982 incentive package, though the figures are approved applications and overstate realized jobs. (p. 2)
The number of jobs which were decentralized each year increased from 11 792 in 1981 to about 32 000 in 1982/83.
-
Branch plants of larger firms, which depend less on agglomeration economies, are a large part of decentralized investment. (p. 3)
about 50 per cent of the decentralized investment is in the form of branch plants of larger companies.
-
Textiles, clothing and leather dominated decentralized employment creation in 1984/85, supporting the EPZ comparison. (p. 4)
textiles, clothing and leather accounted for 37 per cent of the employment opportunities reflected in approved applications in 1984/85
-
The authors' central verdict on Bell is that his hypothesis is unproven rather than wrong. (p. 4)
Our view is not so much that Bell is incorrect as that his case is not proven.
-
Their survey finds most homeland industrialists would not have located there without the incentives. (p. 5)
Sixty-four companies (82 per cent of the responses) answered 'no'.
-
Policy incentives rather than labour conditions are the primary location motive for homeland firms, reversing Bell's ranking. (p. 5)
Government policy via the incentives is the main motivating factor; with the labour situation in the homelands playing a supporting role.
-
Textile investment in the Ciskei fell when the wage concession was dropped under the tax-haven policy, despite very low wages, which indicates that the concessions direct location. (p. 5)
With the implementation of this tax-haven policy, which involved discarding the wage concession, textile investment in the Ciskei has declined.
-
Employment data for the PWV may be understated because firms evaded Section 3 restrictions, which would exaggerate the apparent shift. (p. 6)
Krieg reports that 15 of the 37 surveyed industrialists were deliberately contravening Section 3.
-
Net output stayed concentrated in the main industrial areas, so the shift involved labour-intensive, less productive operations moving to the periphery. (p. 6)
the share of the main industrial areas rose from 80,3 per cent in 1965 - 66 to 81,6 in 1979.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Bell (1983) manufacturing employment series by Main Industrial Areas and Rest of South Africa, 1965/66-1976 (Table 1) | secondary literature compilation | 1965/66-1976 | Supplies the manufacturing employment trend data whose reliability and interpretation the paper examines and partly accepts. |
| Bell (1984) growth rates and employment shares of manufacturing employment (Tables 2 and 3) | secondary literature compilation | 1961/62-1979 | Provides growth-rate and share series showing the relative shift of manufacturing employment away from the main industrial areas. |
| Bell (1985) manufacturing net output location data | secondary literature compilation | 1965/66-1979 | Shows that the main industrial areas' share of net output rose even as their employment share fell. |
| Decentralization Board Annual Reports (Table 4, and Annual Report 1984/85 Annexure D) | commission or official report | 1975-1985 | Documents projects, private investment and approved employment, showing a sharp rise after the 1982 concessions, plus the sectoral breakdown of approved applications. |
| Pretorius et al. (1985/1986) data on numbers and location of manufacturing enterprises (Table 5) | secondary literature compilation | 1965/66-1976 | Shows the share of manufacturing enterprises located in the main industrial areas rose, contrary to a decentralization trend. |
| Addleson, Pretorius and Tomlinson (1985) survey of industrialists who decentralized to the homelands | author survey or fieldwork | c.1984-1985 | Supplies survey evidence that government incentives were the main location motive, challenging Bell's ranking of forces. |
| Authors' interviews with managers of foreign textile companies in the Ciskei and Transkei, and with decentralized industrialists | interviews oral history | c.1984-1985 | Substantiates the claim that current Far-Eastern textile investment is motivated by wage concessions and export-quota circumvention. |
| Tomlinson (1986) Ciskei data (MBA research report) | author survey or fieldwork | 1982-1985 | Shows the number of Ciskei industries growing from 33 to about 100 between June 1982 and December 1985. |
| Krieg (1985) survey of PWV-based industrialists | author survey or fieldwork | 1985 | Suggests firms deliberately breached Section 3 restrictions, raising doubt about reported employment data. |
| Tomlinson Commission Report (UG 55/1955) | commission or official report | 1955 | Benchmark target of 50 000 jobs per annum against which decentralization achievement is judged. |
| Industrial Development Corporation Annual Report 1982 | company records or reports | 1982 | Indicates IDC funding of decentralized industries (R152m, 30 per cent of applications of funds), a policy effect not captured in Board statistics. |
| Environment Planning Act (formerly Physical Planning Act 1967), Sections 2 and 3 | legislation gazette | 1967 onwards | Policy instrument restricting black labour-intensive industry in the PWV, offered as a possible cause of the decentralization trend. |
| 1982 decentralization concession package documents (Dept of Foreign Affairs and Information 1982a, 1982b) | commission or official report | 1982 | Describes the content of the 1982 incentive package (cash grants, long-term and short-term incentives). |
| Survey of Race Relations in South Africa 1979 (Gordon ed.) | secondary literature compilation | 1979 (1976 events) | Cited for evidence that most PWV textile firms hired unregistered workers to evade Section 3. |
| Heeger (1985) SA Builder: MBA Focus (wage differential quote) | newspapers periodicals | October 1985 | Quoted to illustrate wage differentials between South Africa and the independent homelands. |
| Maasdorp (1982) decentralized jobs claimed by Decentralization Board 1970-1978 | secondary literature compilation | 1970-1978 | Establishes that only 34 900 decentralized jobs were claimed, indicating limited ID policy success. |
Inference¶
Quantification: Descriptive tables.
Critical re-reading of Bell's secondary regional employment series against Decentralization Board statistics, enterprise-location and net-output counts, survey responses from decentralized firms, and qualitative sectoral (textile/EPZ) reasoning; no formal statistical testing.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Manufacturing employment share shifted from the main industrial areas to the rest of South Africa between 1965/66 and 1979, by more than ID-programme-claimed jobs. (p. 2) | Bell's employment series (Tables 1-3) compared with Maasdorp's figure on Decentralization Board-claimed jobs. | Partial data |
| Private-sector participation in decentralization rose sharply after the 1982 concessions. (p. 3) | Decentralization Board Annual Reports (Table 4) and Ciskei industry counts. | Systematic data |
| For firms locating in the homelands, government incentives rather than cheap labour are the main motivation, reversing Bell's ranking. (p. 5) | Addleson et al. (1985) survey of decentralized industrialists. | Partial data |
| Bell's regional categories are too aggregated and obscure decentralization trends because deconcentration points are lumped into the 'rest of South Africa'. (p. 6) | Recent Decentralization Board statistics and argument from locational motives; no recomputation of Bell's series. | Asserted |
| Reported PWV employment decline may be overstated because firms evaded Section 3 by employing unregistered workers. (p. 7) | Gordon (1980) and Krieg's 1985 survey (15 of 37 firms contravening). | Illustrative or anecdotal |
| The enterprise-location share and net output share of the main industrial areas did not fall, so the movement is of labour-intensive, less productive operations to the periphery. (p. 7) | Table 5 (Pretorius et al.) and Bell (1985) net output shares. | Partial data |
| Textiles and clothing lead decentralization and were specifically targeted by policy, complicating Bell's 'spontaneous' interpretation. (p. 4) | Decentralization Board Annual Report 1984/85 Annexure D; Emden and Rogerson and Kobben on Section 3; Bell (1983). | Partial data |
| Ciskei textile investment declined when the wage concession was dropped under the tax-haven policy, indicating that concessions direct peripheral location. (p. 6) | Authors' interviews and Ciskei experience (Tomlinson 1986; Tomlinson and Addleson 1985). | Illustrative or anecdotal |
The author's caveats¶
Authors note uncertainties and ambiguities in Bell's data, that Table 4 reflects approved applications (typically slightly less than half of anticipated investment materialised), covers only South Africa and non-independent homelands, that data for independent homelands are less accessible, and that the magnitude of Section 3 contraventions is unknown. They also note no South African productivity comparisons exist and no published data on the sex composition of the decentralized workforce.
Gaps in the evidence¶
The argument would be better tested with establishment-level or magisterial-district census-of-manufactures data reclassified into core, deconcentration and decentralization areas, which existed in unpublished form at the time. Firm-level panel data on relocations by sector, with wages and import competition, would test Bell's competition hypothesis directly. The survey covers only post-1982 homeland firms and cannot speak to the 1965-82 trend, and the evasion of Section 3 is untested.
Replicability: Data poorly documented.