Distribution, investment, and employment in South Africa¶
James Heintz (2000). Distribution, investment, and employment in South Africa. TIPS (Trade and Industrial Policy Strategies) 2000 Annual Forum, Glenburn Lodge, Muldersdrift, 18-20 September 2000. TIPS.
| Type | conference paper |
| Scope | Economics and economic history of apartheid |
| Method | econometric |
| Period studied | 1970–1993 |
| Regions | South Africa |
| Themes | macro fiscal finance debt, inequality poverty unemployment, resistance social movements, unions strikes, manufacturing industrial policy, labour markets wages, methods data sources |
| Citation key | heintz2000 |
Argument¶
Heintz argues that both distributive outcomes (after-tax profit rates) and unresolved distributive conflict (political unrest and strikes) shape the rate of productive fixed investment in South Africa. Low accumulation of labour-absorbing capital since the mid-1970s helps explain 'classical' unemployment and jobless growth, in Malinvaud's sense. He builds a principal-components index of unrest and estimates fixed-effects panel investment functions for seven sectors over 1970-93. Profit rates, sectoral growth and unrest are significant, while the user cost of capital is not. Policy implications follow: asset-based redistribution, expansionary macro policy, targeted industrial policy and shared productivity gains.
Evidence and method¶
Panel dataset for seven private non-agricultural sectors, 1970-93, from the Reserve Bank, Statistics South Africa and the Institute of Race Relations' Survey of Race Relations (prison population, detentions without trial, strike counts). The Fallon-Lucas user cost of capital series was supplied by Fallon. Employer interviews (Heintz 1998) gave the typical investment lag of one year. Sources also include Fedderke et al. (1998) as a comparator index.
Place in the debates¶
- Determinants of investment: neoclassical cost of capital vs profitability, accelerator and conflict (see Did the apartheid economy fail? Growth, productivity, investment and the post-1970s crisis). Finds that the user cost of capital is insignificant; after-tax profit rates, sectoral growth and social unrest drive investment, siding with the post-Keynesian/Kaleckian profitability tradition.
- Keynesian vs classical unemployment (jobless growth) in South Africa (see Unemployment: causes, measurement and labour market institutions). Treats South African unemployment as largely classical (insufficient labour-absorbing capital accumulation), with some cyclical Keynesian elements.
- Does inequality and political instability depress investment and growth? (see Did the apartheid economy fail? Growth, productivity, investment and the post-1970s crisis). Provides South African time-series and panel support that unrest lowers investment independently of profit outcomes, while noting broader claims need more empirical work.
- Growth-first vs redistribution strategies for job creation (see Post-apartheid economic policy, governance and corruption). Argues growth alone is insufficient and favours asset-based redistribution, mild expansion with capital controls, targeted industrial policy and shared productivity gains.
Findings¶
-
The rate of accumulation of non-residential fixed capital fell sharply in the mid-1970s, setting the stage for the unemployment crisis. (p. 5)
A dramatic drop-off in the rate of investment is evident in the mid-1970s.
-
After-tax profit rate has the largest standardized coefficient and unrest the second largest; the user cost of capital is insignificant. (p. 8)
it is clear that the largest standardized coefficients belong to the after-tax profit rate. The index of political rest, reflecting social and distributive conflict, has the second largest standardized coefficient.
-
Between the 1970-75 and 1985-92 cycles, rising social conflict had roughly twice the estimated effect on accumulation of any other variable. (p. 8)
the largest effect can be attributed to growing social conflict – estimated to be approximately twice the effect of any of the other variables, taken independently.
-
Capital costs had the smallest effect and were statistically insignificant. (p. 9)
the smallest effect comes from rising capital costs (even if the estimated coefficients are taken to be significant), a finding consistent with many other empirical studies of investment.
-
Social conflict affects investment independently of actual distributive outcomes such as the profit rate. (p. 12)
Interestingly, social conflict has an effect independent from the actual distributive outcomes (that is, changes in the profit rate).
-
The unrest index was built from prison population, security detentions and strikes, capturing the post-1973 labour conflict and township mobilisation. (p. 13)
The indicators selected include the average annual prison population of South Africa, the estimated number of persons held in detention without trial under the apartheid security laws12, and the number of recorded strikes.
-
The unrest index peaks around Soweto 1976, the UDF build-up and the 1985-88 emergency, and declines with de Klerk and the transition. (p. 14)
the Soweto uprising of 1976; the build-up to the formation of the United Democratic Front (UDF) in 1983; and the state of emergency declared from 1985-88 are all associated with local peaks in the index
-
Validation: the unrest index raises the long-short government yield spread, significant only at the 10 percent level, suggesting it affects investor expectations. (p. 14)
In both estimates the coefficient on the index of political unrest has the expected positive sign and is statistically significant at the 10 percent level.
-
Fixed-effects results show the unrest coefficient is negative and significant across specifications, and the accelerator and profit terms are positive and significant. (p. 19)
More importantly for the focus of this paper, the index of political unrest has a coefficient that is both significant and negative.
-
Mining, manufacturing, utilities and transport enjoyed higher accumulation rates, consistent with state priority for capital-intensive industry. (p. 20)
These results are consistent with the priorities of industrial policy during this period, which emphasized the development of capital-intensive industries.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Reserve Bank of South Africa time series (national accounts, interest rates, fixed capital stock, investment, value added) | official statistical series | 1970-1993 | Supplies the dependent variable (sectoral rate of net fixed capital accumulation), operating surplus, capital stock and value added used to build profit rates and the accelerator proxy. |
| Statistics South Africa (formerly Central Statistical Service) labour market, wage, employment, industrial dispute and price index data | official statistical series | 1970-1993 | Provides labour-cost data for computing sectoral profit rates and the strike counts used in the political unrest index. |
| Survey of Race Relations (Institute of Race Relations) | newspapers periodicals | 1970-1992 | Source for prison population and detention-without-trial series used to build the index of political unrest/social conflict. |
| Fallon-Lucas user cost of capital index (World Bank) | secondary literature compilation | c.1970-1993 | Used as the user cost of capital variable (interest rate, capital goods prices, tax adjustments), deflated by implied sectoral producer prices. |
| Implied sectoral producer prices derived from Reserve Bank real and nominal value added series | official statistical series | 1970-1993 | Used to deflate the Fallon-Lucas cost of capital index to sectoral level. |
| Interviews with large employers in South Africa (Heintz 1998, NALEDI research report) | interviews oral history | c.1998 | Justifies the one-year lag between investment decision and realized investment used in all regressions. |
| Prinsloo and Smith (1996), Reserve Bank Quarterly Bulletin on fixed capital stock 1960-1995 | secondary literature compilation | 1960-1995 | Cited as a fuller overview of fixed capital stock and investment series underlying Figure 1. |
| Fedderke, de Kadt and Luiz (1998) index of political instability | secondary literature compilation | 1935-97 | Benchmark alternative index; Heintz builds his own index because theirs omits strikes. |
| Apartheid security legislation (Suppression of Communism Act 1950, Riotous Assemblies Act 1956, Internal Security Act 1976, state of emergency 1985) | legislation gazette | 1950-1985 | Context for the detention-without-trial variable in the unrest index. |
| Reserve Bank data on yields of long-term government securities and three-month treasury bills | official statistical series | 1970-1992 | Dependent variable (yield spread) in Table A.1 regressions validating the political unrest index as correlated with investor uncertainty. |
| Survey-based manufacturing capacity utilization series | official statistical series | mid-1980s onward | Noted as existing but too short to use; lagged value-added growth is used as proxy instead. |
New data¶
Author-compiled panel of seven private non-agricultural sectors (mining, manufacturing, electricity/gas/water, construction, trade, transport/communications, finance) for 1970-1993 (1970-92 lagged), n=154 observations, plus a principal-components index of political unrest built from prison population, security detentions and strikes, 1970-92.
Inference¶
Quantification: Econometric (causal design).
Fixed-effects panel regressions of sectoral net investment rates on lagged value-added growth, after-tax profit rate, user cost of capital and a constructed political unrest index. Includes a lagged dependent variable estimated by instrumental variables with autocorrelation correction, ADF stationarity tests, standardized and long-run coefficients, and business-cycle decomposition. Grounded in a stylized firm investment model; policy discussion is theoretical.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| After-tax profit rate and political unrest/social conflict are the largest determinants of sectoral investment; the user cost of capital has no significant effect. (p. 8) | Panel fixed-effects regressions (Table C.3) using Reserve Bank, Stats SA and Survey of Race Relations data; standardized coefficients in Table 1. | Partial data |
| Rising social conflict between 1970-75 and 1985-92 had the largest estimated effect on reducing accumulation, about twice any other variable. (p. 8) | Changes in business-cycle averages of regressors multiplied by estimated coefficients (Table 1, eq. 4). | Partial data |
| The rate of accumulation of non-residential fixed capital fell dramatically in the mid-1970s, setting the stage for the unemployment crisis. (p. 5) | Figure 1: Reserve Bank capital stock series for seven sectors, 1965-95. | Illustrative or anecdotal |
| South African unemployment has classical attributes: insufficient accumulation of capital contributes to the lack of jobs. (p. 4) | Theoretical Malinvaud framework, Chadha (1995) citation and Figure 1; no direct employment estimates. | Asserted |
| The constructed unrest index is a valid indicator of investor expectations because it raises the long-short government yield spread. (p. 14) | Table A.1 regressions (n=23) with coefficients significant only at the 10 percent level and low R2. | Partial data |
| Social unrest has an effect on investment independent of actual distributive outcomes (profit rate changes). (p. 12) | Negative, significant coefficient on lagged POL alongside profit rate in all Table C.3 specifications. | Partial data |
| Four capital-intensive sectors (mining, manufacturing, electricity/gas/water, transport/communications) enjoyed higher accumulation than others, consistent with industrial policy priorities. (p. 20) | Fixed-effects coefficients in Table C.3 (equations 1-3). | Partial data |
The author's caveats¶
Admits that capacity utilization series are unavailable for a long period, so lagged value-added growth proxies for it. Notes that the unrest index captures turning points only imperfectly and that the political measure is irrelevant after apartheid. Missing data in the detentions variable were corrected for two years. The yield-spread tests are 'not meant to be conclusive', and the findings need more empirical work to generalize.
Gaps in the evidence¶
The unrest index uses only three aggregate indicators (prison population, detentions, strikes) that could proxy for state repression or the business cycle rather than investor uncertainty. The paper has no firm-level investment data, survey data on investor expectations, or direct measures of property-rights risk, and the capital stock series are not described in detail. The link from investment to employment is never tested with employment data, and the unemployment claims are theoretical. Sector-level employment, firm-level panel data and expectation surveys existed in some form and could be used.
Replicability: Data described but not public.