Infrastructural Investment in Long-run Economic Growth: South Africa 1875–2001¶
J. W. Fedderke; P. Perkins; J. M. Luiz (2006). Infrastructural Investment in Long-run Economic Growth: South Africa 1875–2001. World Development 34(6). Elsevier.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | econometric |
| Period studied | 1875–2002 |
| Regions | South Africa |
| Themes | macro fiscal finance debt, business and state, comparative theory, methods data sources, transition post apartheid |
| Citation key | fedderke2006 |
Argument¶
The paper tests whether public economic infrastructure (rail, roads, ports, electricity, telecoms) drives long-run growth in South Africa. It builds on Barro-style endogenous growth theory, in which infrastructure raises the marginal product of private capital. The authors use a long time series, PSS bounds tests and VECM cointegration systems. They find that infrastructure affects output directly (mainly through electricity generation) and indirectly by raising private fixed investment. Evidence of feedback from output to infrastructure is weak and fragile across specifications. The infrastructure effect survives the addition of property-rights and political-instability variables. The authors read the post-1975 collapse in public infrastructure investment as a likely drag on growth, and they favour renewed public investment.
Evidence and method¶
South African national accounts and Statistics South Africa series on rail, roads, ports, air travel, telephones and electricity, updated from individual organisations after official series lapsed in the late 1980s and 1990s. SARB data on public-sector infrastructural gross fixed capital formation and capital stock. Property-rights and political-instability indices from Fedderke, De Kadt and Luiz (2001). Perkins et al. (2005) supplies the historical description.
Place in the debates¶
- Productivity of public infrastructure capital (Aschauer vs critics of spurious OLS results) (see Did the apartheid economy fail? Growth, productivity, investment and the post-1970s crisis). Takes the cointegration/VECM route to avoid spurious OLS results. Finds a significant but moderate direct output elasticity (about 0.1–0.2, up to 0.5 with institutions) and a strong crowding-in effect on private investment.
- Direction of causality between infrastructure and growth (see Did the apartheid economy fail? Growth, productivity, investment and the post-1970s crisis). Infrastructure leads growth robustly. Growth leading infrastructure receives some support but is not robust to alternative infrastructure measures.
- Economic vs social infrastructure and spending trade-offs under and after apartheid (see Did the apartheid economy fail? Growth, productivity, investment and the post-1970s crisis). Deliberately sets social infrastructure aside as a separate modelling problem, since apartheid created differential access to social rather than physical infrastructure. Suggests the growth payoff of economic infrastructure could help finance social spending.
- Why growth and investment declined after the mid-1970s (see Did the apartheid economy fail? Growth, productivity, investment and the post-1970s crisis). Points to collapsing public infrastructure investment and political instability and weak property rights as depressing private investment. Notes that after 1980 government consumption rose at the expense of investment.
Findings¶
-
Headline result: infrastructure investment leads growth, both directly and indirectly. (p. 1)
investment in infrastructure does appear to lead economic growth in South Africa and does so both directly and indirectly
-
Public infrastructure investment per capita and as a share of GDP collapsed from the mid-1970s. (p. 5)
Investment per capita fell from R1 268 in 1976 to R356 in 2002 (1995 prices), a collapse of 72%. Investment fell from 8.1% of GDP to 2.4% of GDP
-
Infrastructure was rolled out in sequential phases, and the port expansion of the 1970s doubled cargo volumes. (p. 6)
substantial expansion in port capacity was constrained to the 1970s through the construction of two new ports, doubling the volume of cargo handled.
-
Infrastructure rolled out in sequential phases rather than continuously. (p. 6)
infrastructural development tends to take place in phases of development—with a series of sequential periods of infrastructural roll-out.
-
In the baseline system, public infrastructure raises output only indirectly, by raising physical capital investment. (p. 9)
Public-sector infrastructural investment does not enter into the determination of output directly, but does serve to raise investment in physical capital stock
-
Electricity generation has an output elasticity of about 0.20. (p. 11)
The elasticity of output with respect to electricity generation is 0.20
-
Private capital investment is highly elastic to public infrastructure investment. (p. 11)
The implied elasticity at the mean values of physical capital and infrastructural investment is 1.37.
-
The feedback equation (output to infrastructure investment) is unstable across alternative infrastructure measures. (p. 11)
The elasticity of infrastructural investment with respect to output ranges over the 2.73–24.77 range
-
The authors concede the fragility of the feedback result. (p. 14)
Notwithstanding this possibility, the fragility of the result is striking—even to alternative measures of the roads and related transport infrastructure.
-
The direct electricity elasticity is 0.10–0.20 in most specifications and rises to 0.50 once institutional controls are added. (p. 16)
Electricity generation consistently is shown to impact positively and directly on output in South Africa—generally with an elasticity of approximately 0.10–0.20, but rising to 0.50 where we control for the impact of institutions.
-
Property rights strongly raise private capital investment. (p. 16)
In the case of the property rights measure, the direct elasticity of 1.68 on investment is even more dramatic.
-
The growth-to-infrastructure feedback is only weakly supported. (p. 16)
the existence of a feedback from economic growth to infrastructural investment, while receiving some support, proves considerably less robust
-
Growth has been stuck below 3% since the mid-1970s alongside falling public investment. (p. 18)
In South Africa, annual economic growth has struggled to break the 3% growth ceiling since the mid-1970s.
-
From 1980 fiscal priorities shifted from investment to consumption. (p. 19)
After 1980, government consumption expenditure increased at the expense of government investment
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Statistics South Africa and predecessors statistical publications (South African Statistics, 2002 and other years) – infrastructure time series | official statistical series | 1875–2001 | Supplies the long-run series on rail, roads, ports, air travel, telephones and electricity used in the descriptive indices, the bounds tests and the VECM estimation. |
| Individual organizations' infrastructure series (rail, roads, ports, air travel, telephones) used to update discontinued Stats SA series | company records or reports | late 1980s/1990s–2001 | Extends infrastructure series after Statistics South Africa stopped publishing them. |
| South African national accounts data (GDP, per capita output) | national accounts macro | c.1960–2002 | Provides real GDP and per capita GDP, the output variable in the bounds tests and VECM. |
| South African Reserve Bank gross fixed capital formation and fixed capital stock of the public sector, economic infrastructure component | official statistical series | 1960–2002 | Provides public-sector infrastructural investment and capital stock series (per capita indices, Figure 1) and the aggregate infrastructure variables in the VECM. |
| SARB Annual Economic Reports, 1996–2000 (and various issues) | commission or official report | 1996–2000 | Cited to explain the 1990s recovery and subsequent slump in infrastructural investment (Telkom, Eskom, SAA expansion). |
| Perkins, Fedderke and Luiz (2005) descriptive infrastructure dataset | secondary literature compilation | 1875–2001 | Provides the comprehensive historical description of infrastructure roll-out on which Figure 2 and Table 1 draw. |
| Fedderke, De Kadt and Luiz (2001) indices of property rights (PROP) and political instability (INST) | secondary literature compilation | to c.2001 | Supplies institutional control variables used in the extended growth/investment specification. |
| Development Bank of Southern Africa (1998) Infrastructure: a foundation for development | commission or official report | 1967–96 | Prior South African econometric evidence on infrastructure and output, reviewed as context. |
| Personal communication with civil engineering firm on road construction cost | interviews oral history | c.1995 prices | Supplies the R5.4 million per km cost of road used to interpret the large investment-to-stock elasticity. |
| National Integrated Resource Plan 2 2003/2004, National Electricity Regulator | commission or official report | 2003/2004 | Cited, with Bogetić and Fedderke, for projections of investment needs. |
New data¶
The authors compiled an annual time-series dataset of South African economic infrastructure measures from 1875 to 2001 (railway lines, rolling stock, freight and passengers, roads, vehicles, ports cargo, airline passengers, telephone lines, electricity generated), updated from individual organizations after Statistics South Africa discontinued the series, and combined with SARB public infrastructure investment and capital stock. A printout is said to be available on request.
Inference¶
Quantification: Econometric (causal design).
Time-series econometrics: Pesaran–Shin–Smith bounds F-tests on GDP and individual infrastructure measures, ADF unit-root tests, and Johansen VECM cointegration with just-identified structural vectors. Robustness is checked by substituting alternative infrastructure measures and adding property-rights and instability variables. Descriptive indices and growth-rate tables supply the historical narrative.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Public infrastructure investment affects output both directly (electricity generation elasticity about 0.16–0.20, up to 0.5 with institutions) and indirectly by raising private fixed capital investment. (p. 1045) | VECM on per capita GDP, capital stock, SARB infrastructure investment and electricity series; Tables 5, 8 and 9. | Partial data |
| Public infrastructure investment per capita fell 72% (R1,268 in 1976 to R356 in 2002) and from 8.1% to 2.4% of GDP. (p. 1041) | SARB gross fixed capital formation series for the economic infrastructure component of public-sector investment, in 1995 prices. | Systematic data |
| Infrastructure development occurred in sequential phases (railways to 1930, roads to 1940 then paving, ports in the 1970s, then telephones and electricity). (p. 1042) | Index and growth-rate series from Statistics South Africa and other organizations (Figure 2, Table 1), drawing on Perkins et al. (2005). | Systematic data |
| Causality between infrastructure and output runs in both directions, but the evidence is mixed across infrastructure types. (p. 1042) | PSS bounds F-tests, Table 2, for GDP against each infrastructure measure. | Partial data |
| The finding of an infrastructure growth impact is robust to parsimonious and fuller specifications, while the feedback from output to infrastructure is not robust. (p. 1052) | Sensitivity of VECM across alternative infrastructure measures and with PROP/INST controls (Tables 6a, 6b, 8, 9). | Partial data |
| Political instability reduces, and property rights increase, private fixed capital investment in South Africa. (p. 1052) | Fedderke, De Kadt & Luiz (2001) indices included in the VECM (Tables 8 and 9). | Partial data |
| Falling infrastructure investment since the mid-1970s reflects an overall fall in fixed investment, possibly overcapacity, and a shift of government spending from investment to consumption after 1980. (p. 1041) | SARB investment series and secondary discussion of Merrifield (2000), with budget deficit figures cited. | Illustrative or anecdotal |
The author's caveats¶
The authors note that Statistics South Africa discontinued most infrastructure series in the late 1980s and 1990s, so they were updated from individual organizations. Electricity generated is only a proxy for generation capacity. Human capital measures end in 1992, which would have cut the sample, so they are excluded. The third cointegrating vector is unstable and the evidence for three vectors is weak. Aggregate and infrastructural investment are definitionally linked, which biases the elasticity upward.
Gaps in the evidence¶
The argument rests on aggregate national time series with few observations and no direct data on project-level costs, returns or regional incidence of infrastructure. This limits identification of causality, and the authors admit instability in the feedback equation. Archival records of Eskom, SATS and the Railways, and of provincial road programmes, would allow sectoral and spatial tests, and they exist. Social infrastructure and apartheid-era racial differences in access are explicitly excluded.
Replicability: Data described but not public.