Investment in South Africa – a comment on recent contributions

Abstract

This article questions the interpretation of investment made by Fedderke et al. in the October 2001 issue of Development Southern Africa, in particular the conclusion that the higher investment rates of manufacturing in the 1990s are associated with the reduction in state intervention. Rather than improved manufacturing investment, the striking feature of the 1990s is the low level of investment by the government and parastatals. After making a brief review of investment performance over the past decade, manufacturing investment at the sectoral level is assessed, highlighting the very large investments in heavy industries. It is state support in the form of development finance from the Industrial Development Corporation that underlies these patterns, and not the withdrawal of the state as argued by Fedderke et al.

1 INTRODUCTION

Cross‐country studies of growth have consistently found investment to be a significant determinant, especially investment in infrastructure, machinery and equipment (Easterly & Rebelo, 1993; De Long & Summers, 1993; Temple, 1998). The high levels of growth attained by newly industrialising countries have also been characterised as ‘investment‐led’ (see, for example, Rodrik, 1995). The South African government placed investment at the core of the Growth, Employment and Redistribution (GEAR) macroeconomic strategy (Department of Finance, 1996). Yet, few direct mechanisms were put in place to increase investment. It was largely assumed that macroeconomic stabilisation would in itself encourage investment.

The South African investment record has been very poor. According to the South African Reserve Bank, resources allocated to investment as a proportion of gross domestic product (GDP) declined from 19,1 per cent in 1990 to 14,7 per cent in 2001. In the 1970s it had averaged 26,9 per cent. By comparison, high‐growth countries such as Malaysia have had investment rates over 30 per cent and, even with the financial crisis of 1997, the rates did not fall below 20 per cent. An understanding of what drives investment is therefore of critical importance.

In addition, investment is doubly important in South Africa as the economy undergoes protracted and painful restructuring associated with liberalisation. Restructuring implies that some sectors contract and production shifts to sectors in which the economy is more competitive and/or has greater future potential. Without new investment, productive capacity in potential growth sectors does not expand and the economy gets the ‘pain’ with little of the supposed ‘gain’.

Based on an assessment of investment rates over the three periods 1970–9, 1980–9 and 1990–7, Fedderke et al. (2001) argue that South African investment rates can be explained largely by the real user cost of capital, capital productivity and uncertainty associated with political transition. This, they believe, is exactly as economic theory would predict. However, the causal relationships underlying investment are very complex and there is a great deal of disagreement among economists about them. This article comments on Fedderke et al., together with articles by Fedderke (2001) and Gelb (2001). The discussion highlights the difficulties in understanding investment, primarily through a review of data on South African manufacturing investment.

Section 2 gives a brief overview of the investment data. Section 3 makes a more detailed evaluation of the data on manufacturing investment since 1990 and the interpretation of Fedderke et al. Section 4 concludes, and suggests alternative factors that need to be taken into account in understanding investment.

2 THE RECENT INVESTMENT RECORD IN SOUTH AFRICA

At 14,7 per cent of GDP in 2001, the ratio of gross fixed capital formation (GFCF) to GDP was at its lowest level for more than 20 years (Figure 1). While there are many factors contributing to the fall, the real cost of borrowing (i.e. the interest rate corrected for inflation) rose sharply in the 1990s as the government tightened monetary policy in order to reduce inflation. This mechanism works by putting downward pressure on lending, thereby reducing demand. It also makes investment more expensive to finance.

Investment and real interest rates in South Africa

Notes: Real interest rate is calculated from the rate on government bonds (0–3 years) and consumer price inflation. The interest rate therefore does not fully reflect changes in short‐term interest rates, including the four 100‐basis point increases in interest rates instituted by the South African Reserve Bank in 2002.

The South African investment and growth record is also poor in comparison with other middle‐income countries (Table 1). Middle‐income countries grew on average by 3,5 per cent from 1990 to 1999, compared with just 1,9 per cent for South Africa. South Africa had one of the lowest investment rates in 1999, even compared with Asian countries affected by financial crisis and macroeconomic instability. South Africa also had the lowest rate of capital expenditure by the government, compared with countries for which data were available. Part of the rationale for constraining government expenditure was to lower the government's budget deficit, but South Africa's deficit was already relatively low in 1998 after the tightening of fiscal policy in the preceding three years.

International comparisons of growth and investment

	GNP per capital (US$)	Average real growth in GDP		Government deficit % of GDP	Capital expenditure by the government % of GDP	Investment % of GDP		Real growth in GDP
	1999	1980–90	1990–9	1998	1998	1990	1999	1990–9
South Africa	3 160	1,0	1,9	−2,9	1,2	12	16	3,0
Brazil	4 420	2,7	2,9	–	–	20	21	3,1
Chile	4 740	4,2	7,2	0,4	3,6	25	24	11,4
Malaysia	3 400	5,3	6,3	2,9	4,5	34	32	6,2
South Korea	8 490	9,4	5,7	−1,3	3,8	38	27	1,6
Colombia	2 250	3,6	3,3	−4,7	3,2	20	17	7,5
Egypt	1 400	5,4	4,4	−2,0	7,4	29	23	6,7
Costa Rica	2 740	3,0	4,1	−3,8	2,9	27	28	3,4
Venezuela	3 670	1,1	1,7	−2,8	3,8	10	15	2,9
Ghana	390	3,0	4,3	–	–	14	22	4,2
Uganda	320	2,9	7,2	–	–	13	17	9,9
All middle‐income countries	2 000	3,2	3,5			26	24	4,0
Notes: – indicates data not available. Italics indicate that data are for a year other than that specified. Source: World Bank (2001).

The private sector accounts for the majority of investment in South Africa, with a share of 74,9 per cent in 2001. While private investment rates have not been strong in the 1990s, weaker investment has been due largely to lower levels of investment by the government and public corporations. The share of the general government in total investment fell from 17,9 per cent in 1990 to 15,5 per cent in 2001, and the share of public corporations dropped by a far greater extent, from 16,5 to 10,6 per cent.

The value of economic infrastructure under the control of the government and public corporations, including roads, bridges, dams, water supply, telecommunications and electricity increased in real terms at an annual average rate of just 0,4 per cent from 1994 to 2001. This is far below the rate of growth of the economy, and far below the target growth rates. It is calculated from the South African Reserve Bank data on fixed capital stock at constant 1995 prices. The rate of growth is low despite the increased investment by Telkom in telecommunications, classified under public corporations. Indeed, the lower aggregate rates of investment in the 1990s are due almost entirely to the lower investment in infrastructure by the government and public corporations.

The recent study of the capital market and investment from 1970 to 1997 by Fedderke et al. (2001) finds the most striking feature to be the stronger investment performance of manufacturing relative to other sectors in the 1990s. While this is due at least partly to the decline in public sector and parastatal infrastructure investment, Fedderke et al. find that manufacturing investment in machinery and equipment has indeed risen in the 1990s. The rise is attributed to the more efficient functioning of financial markets due to reduced state involvement, and the declining influence of primary products (mining). They also argue that market forces have brought the returns to investment more in line with the costs of capital. The continued relatively poor investment performance into the 1990s is further attributed to increased political uncertainty. This article addresses these issues in turn, critically appraising the analysis made and assessing it in the light of more recent data and a closer assessment of manufacturing performance at the sector level.

3 INVESTMENT IN MANUFACTURING AND SECTOR DEVELOPMENTS

The fact that the highest rates of growth in capital stock in the 1990s were recorded by manufacturing sectors is argued by Fedderke et al. (2001) to be evidence of the removal of distortions in South African capital markets. At several points, it is also suggested that the government has crowded out private investment in the past, but that this has altered during the 1990s. This is referred to as a ‘structural change’. These benefits from removing government distortions are also at the centre of Fedderke's (2001) work on growth.

The analysis of Fedderke et al. proceeds through several descriptive stages by sector for the period 1970 to 1997. They examine the total capital stock of different sectors of the economy, the growth in capital stock and investment rates. The higher ranking of manufacturing sectors in the 1990s is noted as a key development and is explained by the declining ‘distortions’ due to primary commodities and government‐led investments. For example:

Once again, therefore, the evidence suggests the plausibility of a distortion in the South African capital markets owing to heavy reliance on the mining of primary commodities during earlier phases of development of the South African economy, and the presence of substantial government‐led investment in capital stock in a number of core sectors (electricity, gas and water; petroleum refining). (p. 449)

And so:

The gradual disappearance of a reliance on primary commodities in the South African economy, and reduced state involvement in ‘strategic’ investments, have at least plausibly triggered a restructuring of the South African capital market. (p. 500)

They go on to argue that sectors that have been ‘crowded out’ from funds for investment are those showing strong growth in capital stock during the 1990s.

They continue in Section 3 to explore possible explanations of investment changes in the user cost of capital and capital productivity, and find both to be important. Where productivity is strongly correlated with the user cost, they find higher growth in capital stock which they read to be evidence that well‐functioning capital markets generate higher investment rates. In another article (Fedderke, 2000, repeated in Fedderke, 2001), a third determinant of investment is found to be the uncertainty associated with political transition.

The present article examines the sectoral data on manufacturing in more detail before assessing to what extent this can, in fact, be characterised as a withdrawal of the state, and in particular, we look at the role of the Industrial Development Corporation (IDC). We then address some issues of uncertainty and investment in a separate section below.

At the outset, it is worth noting that the use of rankings by Fedderke et al. to assess fixed capital stock and investment rates across sectors means that higher ranks for manufacturing sectors can equally be viewed as falls in the ranks of other sectors. The ‘structural change’ is therefore due partly to the very low levels of investment by the government and parastatals already noted. Low levels of machinery investment in areas such as electricity, gas and water in the 1990s are mainly due to Eskom's large investments in electricity generation in the previous decades. New investments will have to be made in the future as demand catches up with supply capacity. This is in sharp contrast to government investment in electricity, gas and water being characterised as a distortion. Viewing government infrastructure investment as a distortion also flies in the face of studies that have found that investment in infrastructure is a key determinant of growth (see, for example, Easterly & Rebelo, 1993).

Contrary to the patterns noted by Fedderke et al., the performance of the manufacturing sector has been poor over the last decade, with average annual growth in output of just 1 per cent from 1990 to 2001. There have also been large employment losses. There were, however, significant fluctuations over the decade, which significantly influenced their results due to 1997 being the last year in their analysis. A year‐by‐year appraisal indicates that both output and investment grew strongly in the mid‐1990s, with a 6,5 per cent increase in output in 1995 and a huge 20,6 per cent increase in investment (Figure 2). It is notable that in the period 1993–5, capacity utilisation increased alongside growth in investment, suggesting that demand factors were largely driving the growth. In 1996 and 1997, the pattern reversed. While investment continued to increase, as did output, the growth in output did not keep up with the increased production capacity, as indicated by falling capacity utilisation.

Percentage change in manufacturing output, employment and investment

Furthermore, the increase in investment in 1994 and 1995 was due largely to non‐ferrous metals, and in 1996 and 1997 it was due to just two subsectors, namely basic chemicals and iron and steel. Excluding these subsectors, manufacturing investment was stagnant after 1995. In several sectors, such as machinery, metal products, textiles, footwear, wood products, glass products and furniture, higher rates of investment in the early to mid‐1990s faltered in 1996 and 1997. These are relatively labour‐intensive subsectors in which there needs to be sustained growth in production capacity if there is to be increased employment generation (see also Machaka & Roberts, 2003). The stagnation in the second half of the decade coincided with high real interest rates and weak aggregate demand.

Investments in basic iron and steel alone accounted for 14 per cent of total manufacturing investment in 1996 and 16 per cent in 1997 due mainly to large‐scale projects with IDC support, such as Columbus stainless steel and Saldanha (Table 2). Similarly, very large investments in non‐ferrous metals made in 1993, 1994 and 1995 included those in the Alusaf project. These investments represent a large increase in production capacity, but are lumpy in nature and quite different from sustained investments in increased capacity across manufacturing. Note that taking just investment in machinery and equipment, as do Fedderke et al. (2001), lessens the relative size of these investments, as an important part of the investments were construction of new factories. Greenfield investments such as these clearly represent fixed capital stock and should not be omitted.

Manufacturing gross domestic fixed investment (R million, constant 1995 prices)

	1990	1991	1992	1993	1994	1995	1996	1997	1998	1999	2000	2001
Food	1 150	1 212	1 698	1 818	2 220	2 964	2 856	2 473	2 315	2 731	1 946	1 604
Beverages	1 374	1 041	1 464	1 218	1 117	1 013	1 185	891	1 422	1 969	2 557	1 698
Tobacco	43	71	56	42	23	27	33	15	55	45	48	55
Textiles	425	185	349	138	261	470	516	433	366	366	606	440
Wearing apparel	129	98	100	72	104	234	157	95	273	144	108	230
Leather and leather products	26	38	23	38	65	25	35	26	27	22	118	75
Footwear	40	38	48	34	44	93	34	28	17	37	34	32
Wood and wood products	217	148	157	118	121	242	280	251	232	211	201	309
Paper and paper products	907	449	554	703	1 042	1 783	1 320	1 371	1 317	2 493	1 948	1 960
Printing and publishing	134	130	228	214	436	622	645	593	702	408	492	293
Coke and refined petroleum products	7 144	7 251	4 423	2 472	1 524	2 171	1 764	2 465	2 231	2 910	3 430	2 905
Basic chemicals	558	517	390	441	537	973	1 521	2 141	2 035	1 208	2 170	3 367
Other chemicals and man‐made fibres	918	705	915	696	633	949	659	462	846	980	746	841
Rubber products	179	204	210	156	230	281	283	209	293	233	187	154
Plastic products	381	86	84	143	162	202	189	130	247	610	724	1 047
Glass and glass products	92	42	93	87	151	340	499	226	243	211	197	208
Non‐metallic minerals	419	480	451	323	392	616	789	828	737	552	602	606
Basic iron and steel	2 420	1 732	1 479	2 600	2 805	2 306	5 114	7 704	4 916	3 681	2 080	2 188
Basic non‐ferrous metals	567	603	288	2 262	3 587	2 582	1 762	1 032	1 781	941	757	943
Metal products (excl. machinery)	463	574	874	868	709	1 148	1 081	675	738	654	859	767
Machinery and equipment	328	313	339	212	253	437	363	256	375	244	366	831
Electrical machinery and apparatus	319	153	166	161	375	310	257	307	221	329	366	556
Radio and communication equipment, etc.	98	66	124	92	121	241	161	107	159	143	199	379
Motor vehicles and parts	1 162	941	1 203	1 379	854	1 318	1 288	746	1 566	1 801	2 826	2 345
Other transport equipment	43	44	77	152	185	285	233	379	205	135	223	547
Furniture	80	77	44	84	130	145	179	149	163	142	85	212
Other manufacturing	94	188	102	46	20	44	46	83	123	105	101	120
Total	19 709	17 384	15 938	16 568	18 099	21 820	23 246	24 073	23 600	23 307	23 976	24 712

The study's comparison of 1997 with 1990, 1980 and 1970 in order to identify changes in investment patterns means that it captures the five years of positive growth in manufacturing investment which ended in 1997, while by the end of the decade investment levels had fallen again.

The main increases in manufacturing investment in the mid‐1990s were, therefore, very closely linked to primary resources – both directly in the form of smelting of ores, and indirectly in being very energy intensive (and hence linked to coal). In addition, the decline in the significance of primary products of which Fedderke et al. make much is also overstated, especially when more recent data are examined. If one examines only primary and secondary sectors (in order to abstract from the increasing importance of services), then one finds that the share of mining has in fact increased in recent years. Mining accounted for 23 per cent of the output of primary and secondary sectors in 2002, up from a low of 18 per cent in 1997 (calculated from current price GDP data).

3.1 Withdrawal of the state from financial markets?

It is difficult to argue that the major increases in manufacturing investment in the mid‐1990s in iron and steel and non‐ferrous metals represent improved allocation of finance due to the withdrawal of the state. These investments were strongly supported by the state‐owned IDC and involved a very small number of large companies. Economies of scale and capacity utilisation are important factors for the large and lumpy investments in these sectors (see, for example, Makhaya et al., 2002). Together with the very large investment by telecommunications parastatal Telkom in 1998, the major capital projects in the iron and steel and non‐ferrous metals sub‐sectors account for almost all of the overall rise in the GFCF to GDP ratio in the mid‐1990s up to 1998. Rather than being surprising, therefore, the fall of the investment ratio from 1999 in fact represented a return to the lower levels at which it has remained under the prevailing macroeconomic environment.

Active state policy can be seen in other manufacturing sectors that have increased investment, such as motor vehicles and leather products. These sectors have been heavily influenced by the Motor Industry Development Programme (MIDP) (see, for example, Black, 2001). It is notable that the performance of leather products contrasts with footwear, which performed poorly. Growth in leather products is mainly accounted for by leather seat‐covers for cars. The motor vehicle sector itself also had above‐average rates of growth in investment.

It is notable that the incentives of the MIDP rely on the ongoing protection of the local market, as this provides the benefit of earning duty‐free import licences. The scheme is similar in some ways to the programme used to develop the car industry in South Korea, where firms were forced to export if they were to be allowed to import some foreign manufactured models (Amsden, 1989). This achieved the dual goal of forcing firms to work for competitiveness in international markets, while maintaining a level of protection in the domestic market to enable capabilities to be developed. The MIDP is also notable as a sector‐specific programme based on a clear understanding of the structural changes required in the sector, such as the rationalisation of models and the achievement of scale economies. It had clearly specified objectives, time‐frames and mechanisms. Its weakness is perhaps that it favours the export of very high‐value items, such as catalytic converters and seat leather, rather than more labour‐intensive components.

Fedderke et al.'s (2001) analysis of the underlying causes rests on an assertion that more efficient markets have brought returns to investment in line with the costs of capital across sectors. However, the study neither effectively measures the cost of capital across sectors nor the productivity of capital, and so cannot reach any conclusions about the relationship between them. Instead, in the study, the cost of capital is measured by the economy‐wide long‐term interest rate and not the interest rate actually faced by companies in different subsectors (including cheap finance made available by the IDC). Similarly, no sectoral data are available on the effective taxes paid. The only variable in the calculation that differs by sector is the depreciation on capital stock. Instead of the marginal productivity of capital (i.e. the increased output that would result from an increased unit of capital), the study calculates only the average output produced per unit of machinery. Unsurprisingly, this is highest for sectors intensive in labour, including skilled labour. After the ‘other manufacturing’ category, the highest output to machinery ratios are found in wearing apparel, instruments, machinery manufacture, and footwear.

In addition, recent manufacturing sector studies, such as of the auto and textile sectors, point to the importance of scale effects and the need to modernise capital stock in response to increased import competition (Barnes, 2000; Black, 2001; Roberts & Thoburn, 2003). For example, much of the textile sector in the 1990s operated with looms that were more than 20 years old, producing for the domestic market. Trade liberalisation and international competition meant that investment in new machinery was required in order to improve production capabilities and survive.

3.2 The IDC and the financing of investment

The IDC provides both loan and equity finance, mainly for new projects, at low interest rates. (This is true for the 1990s being focused on here, although the IDC has increased interest rates to commercial levels in recent years.) Its advances directly accounted for around 12 per cent of gross domestic fixed investment in manufacturing from 1998 to 2000 (IDC, 2001a, 2001b). The IDC's impact is much greater than this, as it invests alongside private sector agents, meaning that projects with significant IDC participation account for a very large proportion of manufacturing investment. The projects in the non‐ferrous metals and basic iron and steel sectors with major IDC participation alone accounted for approximately 25 per cent of total manufacturing investment from 1992 to 1997 (IDC, 1997). The IDC provided R14,1 billion of the R25,4 billion of investment in these projects which, given the submarket interest rates, implies a very significant level of support.

This means that the IDC's decision making is certainly one of the largest determinants of manufacturing investment. It has historically been oriented to the development of extremely large‐scale minerals beneficiation projects and has close links with previously state‐owned industrial enterprises such as Iscor and Sasol, as well as the major conglomerates. In 1998, the IDC owned 11,3 per cent of Gencor, 14,4 per cent of Iscor, 9,1 per cent of Billiton and 8,2 per cent of Sasol. This has changed to some extent in recent years with the sale of some of its equity holdings, but the IDC is still a major player in heavy industry.

The significance of the IDC in manufacturing investment contradicts the assertion by Fedderke et al. (2001) that the improvements in manufacturing investment up to 1997 represent the better functioning of the financial system with the reduction of state involvement. It appears to be precisely the success of public institutions that is responsible for the higher investment, as the IDC has been very successful in its own right in being entirely self‐sustaining due to the good performance of most of its investments. Fedderke et al. essentially start from an assumption that markets are efficient. Most economists would agree that financing of investment is an area where this assumption is least warranted (Stiglitz, 1993, 1998; Rodrik, 1999). The nature of finance, and in particular the importance of information, mean that market ‘failures’ are intrinsic to the financial sector rather than being rare and temporary aberrations. The far‐reaching interventions in the financial system by East Asian governments such as in South Korea and Taiwan were a very important part of their ability to support large‐scale investments in developing new industrial capabilities (Harris, 1988; Amsden, 1989; Wade, 1990; Rodrik, 1999).

The IDC's orientation did perpetuate the bias to capital‐intensive manufacturing linked to its heavy industry orientation. From June 1994 to June 1999, over half the IDC's investments by value were classified as being in basic metals (IDC, 1999). Based on data for 1993 to 1998, the IDC invested an average of R8,2 million in creating an additional job in the basic iron and steel sector and R5,5 million in non‐ferrous metals. A greater focus on small and medium enterprises (SMEs) and on tourism and agriculture projects in the last three years has shifted the project mix to some extent in favour of employment creation. Lending to SMEs accounted for around 75 per cent of the number of projects and 15 per cent of the amount of finance authorised in 2001. (SMEs are defined by the IDC as those with total assets of less than R30 million.)

3.3 Uncertainty and investment

In addition to the cost of capital and capital productivity, Fedderke (2000) also finds both systemic and sectoral uncertainty to be important determinants of investment. The implications of uncertainty for investment and related policy recommendations are also emphasised in Gelb (2001).

In particular, Fedderke employs an index of ‘systemic uncertainty’, which includes measures of political instability and property rights, and finds it to be an important explanation of investment and growth. This leads to discussions of low growth in the 1990s and appropriate economic policies in Fedderke (2001), which are essentially focused on an orthodox menu of enforcing property rights, lowering taxes, trade liberalisation and macroeconomic stability.

The index of political instability is based on a subjective weighting of selected legal measures provided by an anonymous group of ‘leading authorities’ in political science, sociology, law, economic history and history. By using measures such as prosecutions under the 1951 Suppression of Communism Act, the measure captures major periods of uprising against apartheid. It is, however, a long step from the effects of upheaval and protest before 1994 to the claims made by Fedderke for the implications of his regression results for government policy post‐1994, and in particular the constraints he claims should be placed on redistribution.

It is also notable that the improved investment performance in 1994 and 1995 does not fit with the political instability factor. The best performance straddled the transition period when presumably uncertainty was relatively high, and then investment rates declined after the commitments made in the government's 1996 macroeconomic framework and cuts in tax rates. Rather, the reductions in manufacturing investment rates in the late 1990s are consistent with the higher levels of interest rates and constraints on aggregate demand from a tighter fiscal policy.

In addition, while much is made by Fedderke of systemic uncertainty, the implications of the estimates are that both capacity utilisation and the user cost of capital also have significant effects (Fedderke, 2000: 29). Despite the implications that these have for aggregate demand and monetary policy, the effects of macroeconomic policy are not explored. This is also despite wide‐ranging research, by Fagerberg (1988) and others, that finds reinforcing effects running from strong demand to high levels of capacity utilisation, and further to investment and improved competitiveness and productivity resulting from firms' upgrading of capital stock. Instead, it is taken for granted that macroeconomic stability is the narrow concern with lower inflation and a reduced fiscal deficit rather than stability in real variables. Interestingly, the regression results in Fedderke (2000) also find no link between openness in international trade and investment, although this does not deter Fedderke (2001) from making strong arguments in favour of further trade liberalisation.

In addition to political stability, systemic uncertainty draws on a proxy of property rights given by the number of patents registered in South Africa. The rationale for this is that filing a patent is evidence of faith in the protection of intellectual property rights. Other possible explanations for patents, such as research and development effort, often linked to government support for research, are not considered.

Uncertainty over economic policy, taxes and the security of property rights (among other areas) was also found to be important in an extensive firm‐level study conducted by Gelb (2001). While it is plausible that uncertainty inhibits investment, Fedderke et al.'s and Gelb's views of it are open to question, especially as explanations of the changes in investment performance in the 1990s. Political uncertainty and instability have undoubtedly decreased rather than increased since the 1980s and early 1990s. Tax rates have also come down. However, uncertainty over demand growth, infrastructure spending by the government and parastatals, and social stability are all potentially important factors for investment. Failure to deliver the levels of public investment that were set out in the macroeconomic framework could, in this interpretation, be seen as contributing to uncertainty.

The reference to uncertainty factors by the firms surveyed by Gelb may also be an after‐the‐event rationalisation of decisions. There is some similarity here to allusions that are often made in the press to ‘business confidence’. Business confidence measures, such as those used by the South African Chamber of Business, are generally based on backward‐looking indicators (such as new car sales and movements in the currency) rather than measures of expectations linked to firm decision making and strategies. (This has led to the confidence index improving in late 2003 due to the strong rand, while companies are simultaneously issuing profit warnings due largely also to currency strength.) If these are portrayed as causal factors in investment, then the argument risks becoming circular, with low growth and investment being attributed to weak business confidence. The latter, in turn, is measured in terms of historical patterns of growth, investment and demand.

Gelb (2001) also highlights the ‘hold‐up’ of investment by firms based on their concerns about government policies, crime, and protection of property rights under widening inequality. The evidence for this is, however, speculative. Other sources of uncertainty, such as volatile exchange rates given capital account liberalisation, may also be important and suggest that current market signals are unreliable for planning future exchanges. Recent work has found that firms view exchange rate instability to be the single most important deterrent to investment (Kaplan, 2003).

Finally, the discussion of uncertainty highlights major gaps in the methodology of neoclassical economics in addressing issues of firm strategy and decision making. This is even more so given the relatively small number of firms that dominate much of investment and economic activity in the South African economy. To date, research on investment has done little to explore these firms' strategic decision making.

4 SOME CONCLUSIONS

There is little disagreement that the investment record has been very poor and that it is closely related to the poor growth of the South African economy. In the single largest sector, namely manufacturing, investment improved in the mid‐1990s but the growth was not sustained. Closer examination reveals that it was due to only two industries rather than being part of an expansion of the broad productive base of the economy. The investments in iron and steel and non‐ferrous metals were supported by the government in the form of the IDC and were made by a small number of large companies.

Fedderke et al.'s (2001) finding that better manufacturing investment in the 1990s (to 1997) is due to lessening of distortions and a ‘withdrawal of the state’ is simply not consistent with a more careful examination of the data. In addition to ignoring the concentration of investment in several resource‐based sectors of manufacturing and the role of the IDC, it is a partial characterisation that ignores many of Fedderke's own results, such as the positive effects of government investment and capacity utilisation on private investment (Fedderke, 2000). As noted, a major feature of the 1990s has been the low investment by the government and parastatals through to the end of the decade.

Instead of simplistic answers about the withdrawal of the state and an exercise that appears largely one of fitting data into predetermined conclusions, the complexity of factors affecting investment needs to be acknowledged. Investments in increased productive capacity are affected by expectations about a range of factors such as available inputs, growth of markets and the ability to access technology and skills. Government economic policies, including spending on infrastructure and support for research and development, affect these expectations. Uncertainty about demand and exchange rates is important, along with factors that can be characterised as ‘political’. Investment needs to be understood as part of the corporate strategies of companies, especially in large‐scale industries with lumpy investments.

Exploring reasons for investment needs to acknowledge contributions from different theoretical perspectives. In orthodox neoclassical economics, investment is attracted to activities with a higher productivity of capital and hence a higher rate of return. While this incentive is undoubtedly important there is a whole missing dimension, which deals with where the productivity and improved capabilities come from. Similarly, rather than the simplistic dichotomy of markets versus the government employed by Fedderke et al., it needs to be acknowledged that the ways in which financial markets work mean that they are intrinsically biased against financing investment in relatively undeveloped sectors and by newer firms (Stiglitz, 1993). This has major implications in a restructuring economy and suggests that detailed work on the actual flow of funds through the financial system is required.

In many ways South Africa has been relatively successful in developing scale‐intensive industries such as iron and steel, as well as some knowledge‐intensive industries, such as polymer chemicals and defence‐related electronics. These capabilities, along with world‐class capabilities in areas such as deep‐mining, have been developed by big businesses with close support from the state, in the form of direct ownership and finance, amongst others. Investment, therefore, cannot be separated from the different production systems and business models that govern the organisation of production.

It is difficult to generalise about what drives the adoption of different business models and production systems. It depends on the different interests that shape the orientation of business in a given country, and the broader political economy. The government has an important role to play through industrial policy, and support for research (Amsden, 1997a, 1997b; Best, 2001). The processes are path‐dependent in that the options facing firms and the economy at any given point in time depend on previous decisions and choices that cannot simply be reversed. Many of the important factors are also country‐specific. The difficulty in making generalisations provides a note of caution, but not reason for inaction. Rather we require an appropriate methodology, detailed research and appropriate strategies based on country‐specific conditions.

Additional information

Notes on contributors

Simon Roberts 1

Associate Professor, School of Economic and Business Sciences, University of the Witwaters‐rand, Johannesburg, South Africa. Research assistance was provided by Johannes Machaka. The article also benefited from the comments of two anonymous referees.

Notes

Associate Professor, School of Economic and Business Sciences, University of the Witwaters‐rand, Johannesburg, South Africa. Research assistance was provided by Johannes Machaka. The article also benefited from the comments of two anonymous referees.