Manufacturing in South Africa over the last decade: a review of industrial performance and policy

Abstract

This article outlines South Africa's comparative industrial performance over approximately the last two decades. It then examines the policies effected by the Department of Trade and Industry since 1994 to promote the development of industry. The focus is on the sector‐specific supports available to the auto and auto components and the clothing and textiles sectors. The final section locates these sector‐specific support measures within a broader discussion of industrial policy.

1 SOUTH AFRICA'S INDUSTRIAL PERFORMANCE – A COMPARATIVE PERSPECTIVE

1.1 Manufacturing value added

South Africa's manufacturing growth in the 1990s was only marginally higher than in the 1980s and significantly lower than that of the industrialised countries, the developing countries and the world (Table 1).

Growth of MVA, 1980–2002, South Africa and major regions

Total MVA annual average growth rate	South Africa	Industrialised countries	Developing countries	Developing countries (excl. China)	World
1980–90	1,1	2,8	5,1	4,3	3,1
1990–2000	1,2	2,3	6,4	4,5	2,8
2000–2	2,5*	1,1	4,9	3,9	2,0
Total MVA index
1990	100	100	100	100	100
1997	106	108	157	137	–
2000	106	118	181	152	139
2001	108	120	190	157	141
2002	113*	121	199	162	144
Note: *Estimate, 2000–2 data calculated by applying the UNIDO forecast to the 2000 data. Sources: UNIDO (2003), Table 1.3 for 1980–2000; Table J for world and 2000–2; South African data for 2000 and 2001 supplied by Olga Memedovic, UNIDO.

The manufacturing growth rate for South Africa accelerated in 2001, and even more so in 2002. During 2003, however, manufacturing value added (MVA) has declined quite significantly. Consequently, over the last two decades, South Africa's share of developed market and world MVA has declined persistently.

Given the more rapid rate of population growth in South Africa, the relative decline in South African MVA per capita has been even more pronounced (Table 2).

South Africa's share in MVA and MVA per capita, 1980–2001

	Year	Developed market economies	World total
MVA (share in %)	1980	0,7	0,6
	1990	0,7	0,5
	1999	0,6	0,4
		Developed market economies	China as % of South Africa
MVA per capita (% chare)	1980	20	12
	1990	15	25
	2000	12	70
	2001	12	72
Source: UNIDO (2003) database, Tables A and K. 2001 data are projections.

Sub‐Saharan Africa has experienced the lowest regional growth rates for manufacturing. However, the growth rate for sub‐Saharan Africa is considerably more rapid than for South Africa. There has accordingly been a consistent decline in South Africa's share of sub‐Saharan Africa's MVA, particularly in the second half of the 1990s (Table 3).

South Africa's share of sub‐Saharan MVA, 1980–2000

1980	1991	1995	1997	1999	2000
52,8	50	51	49	48	47
Source: World Bank (2003), Table 2–3.

1.2 Exports

Lower growth rates are also manifest in a poor export performance. Over the last two decades, there has been a pronounced decline in South Africa's share of world and developed country manufactured exports (Table 4).

South Africa's share of developed market economies and world manufactured exports, 1980–99

	Year	Developed market economies	World total
Manufactured exports (share in %)	1980	0,5	0,5
	1999	0,4	0,3
Source: UNIDO (2003), country tables.

The weak performance of South African manufactured exports is also evident at the product level. In 1998, the 20 most market‐dynamic products had an average annual growth rate of 12,9 per cent and constituted 22,6 per cent of world exports. In 1998, these products constituted 28,7 per cent of developing country exports, but only 3 per cent of South Africa's exports. By 2001, the share of these products in South African exports had barely increased, to 3,06 per cent.

1.3 Industrial structure

South Africa's poor relative aggregate output performance is evident also at the sector or branch level. Almost all of the main branches of industry exhibited lower growth rates in the 1990s, compared with the developed and the developing countries and the world.

The labour‐intensive branches – food and beverages and textiles, clothing and footwear – have grown more slowly than the average. As a result, they have declined, from 23 per cent of MVA in 1980 to 20 per cent in 2000.

1.4 Employment

Low rates of growth in MVA, and particularly low rates of growth in the labour‐intensive sectors, have combined with rising capital intensity overall to result in consistent declines in manufacturing employment.

Prior to 1990, manufacturing employment exhibited a slow but persistent increase, with cyclical fluctuations. Since approximately 1995, manufacturing employment has been on a persistent downward trend that has not reversed even in the upturn. With the very rapid rate of manufacturing growth in 2001–2, manufacturing employment rose marginally. However, employment fell significantly again in 2003 (Figure 1).

Manufacturing employment, 1990–2002 Source: DTI database.

1.5 Productivity and remuneration

A slow rise in MVA and a significant decline in employment have resulted in a steady rise in labour productivity. This, in turn, has resulted in a steady increase in labour remuneration. Average manufacturing wages in June 2002 were 21 per cent higher than in June 1990 – with two‐thirds of the increase post‐June 1995 (Figure 2).

Figure 2: Manufacturing remuneration, 1990–2002 Source: DTI database.

Moreover, the indications are that earnings of unskilled and lower‐paid employees in manufacturing increased more rapidly than those of the skilled and higher‐paid employees. For those in formal manufacturing employment, and more particularly for lower‐paid employees, there have been persistent and solid gains in remuneration.

1.6 Geographical spread

One of the explicit objectives of South Africa's manufacturing policy for the new government was a ‘more equitable geographic spread of economic activities’ (DTI, 2003: 10). A number of incentives to encourage manufacturing investments in less industrialised areas were accordingly developed. However, the aggregate share of MVA of the three dominating industrial provinces hardly changed between 1995 and 2001. While Mpumalanga and Free State, and more recently Limpopo, have had some marginal increases in their relative share, none of the less industrialised provinces has exhibited a significant or consistent change in its position relative to the major industrialised provinces (Statistics South Africa, 2002: 32).

There is, therefore, little evidence to suggest that the investment incentives have had a significant impact in altering the provincial location of industry.

2 POLICIES TO ENHANCE INDUSTRIAL PERFORMANCE

2.1 General support measures

2.1.1 Supply‐side measures

After 1994, the General Export Incentive Scheme was phased out and replaced with a large number of supply‐side measures. No clear objectives were developed for the supply‐side measures; monitoring in many cases has been deficient; and no systematic evaluative data exist as to the effectiveness of these measures (DTI, 2001: 43).

In this context, it is difficult to pass definitive assessments on the effectiveness of any particular supply‐side measure or of the collectivity of measures. However, anecdotal evidence together with surveys of a representative sample of manufacturing firms commissioned by the Department of Trade and Industry (DTI), suggests that the number of firms making use of the supply‐side measures is not high and that the effectiveness of the measures is limited. Moreover, there is little evidence to suggest that the situation is improving significantly.

2.1.2 Overall sector strategies

After 1994, the DTI spent considerable time and resources developing overall strategies for different sectors. How does business perceive the DTI's strategies for the development of their sectors? Once again, anecdotal evidence and a survey commissioned by the DTI show that few firms regard the Department as having industrial and trade policies for their particular sector, and only a limited number of those regard these policies as being effective.

However, in two sectors – clothing and textiles and, more especially, transport equipment – a significant percentage of firms regard the DTI as having trade and industrial policies for their sector and further regard these policies as exerting a considerable impact on the development of their sectors. These two sectors have been the only recipients of sectorally specific support measures. For both sectors, support has taken the form of import–export complementation – an exporter earns either the right to import on a duty‐free basis (the Import Credit Certificate Scheme in autos and components), or an offset against the duty paid on imports (the Duty Credit Certificate Scheme in clothing and textiles).

2.2 Sector‐specific policy support

2.2.1 Autos and auto components

The Motor Industry Development Programme (MIDP), initiated in 1995, entailed a phase‐down of tariffs; a removal of local content requirements; duty‐free imports of components up to 27 per cent of the wholesale value of the vehicle; and duty rebate credits to be earned on exports.

Exporters of autos and auto components earn an Import Rebate Credit Certificate (IRCC) equal to the local value of the export. The extent of the IRCC benefit depends on whether autos or auto components are exported, and on whether the IRCCs are used to import autos or components. Flatters (2002) summarises the additional benefits conferred through the IRCCs (Table 5).

Effective protection given to exports by IRCC facility

	Item exported
Use of IRCCs	Vehicles	Components
Vehicle imports	40%	26%
Components imports	30%	30%
Source: Flatters (2002: 9).

This is a very significant support to exporters of vehicles and components. Flatters (2002) argues that it is therefore unsurprising that exports and resultant investments have risen since the IRCCs were introduced. The IRCCs entail support to exporters with the incidence falling on local consumers in the form of prices that are higher than they would be in the absence of tariff protection.

Barnes et al. (2003: 10, 22) provide an alternative position. They argue that the export success of the South African auto industry results not from any support, but from the industry's competitiveness and efficiency. They argue further that South African consumers do not face higher prices. On the contrary, the MIDP has resulted in lower prices for domestic consumers by comparison with prices prevailing elsewhere. These authors also provide a comparison of the efficiency of a number of South African, Western European and emerging economies' auto component firms. Between 1998 and 2001, the performance of these South African auto component firms improved significantly. Nevertheless, in 11 of the 13 measures, they still lag behind both Western Europe and emerging economies.

There are difficulties in interpreting the data provided for these auto component firms and particularly in drawing conclusions that are applicable to the auto and auto components sector as a whole. Productivity data that encompass the sector en toto do not suggest a rapid rise in productivity after the introduction of the MIDP in 1995. For the entire period, growth in labour productivity has been marginally above the average for manufacturing – substantially below in the period to 1998, and substantially above since then. Capital productivity has persistently declined throughout the period and the decline has been more pronounced than the average for manufacturing. By contrast, there has been significant growth in capital productivity for the manufacturing sector since 1999. While the auto industry has expanded output rapidly, there has also been a considerable expansion in inputs (particularly imported inputs), such that the increase in value added is far less impressive and the growth in capital stock has exceeded growth in value added.

Multi‐factor productivity provides the best measure of efficiency growth. Over the period 1995–2001, motor vehicle parts and accessories experienced lower growth than the average for South African manufacturing. For the period 1995–2001, with the marginal exception of labour, motor vehicle parts and accessories experienced lower growth than the average for the manufacturing industry. The South African manufacturing industry has itself enjoyed only modest productivity growth (Table 6).

Motor vehicle parts and accessories and manufacturing productivity, 1995–2001 (1995 = 100)

	1995	1996	1997	1998	1999	2000	2001
Labour production
Manufacturing	100	100	106	108	110	118	124
Motor v.p. and a.	100	96	97	96	105	117	126
Fixed capital production
Manufacturing	100	98	96	92	90	94	95
Motor v.p. and a.	100	94	94	87	87	83	84
Multifactor production
Manufacturing	100	99	101	100	100	105	109
Motor v.p. and a.	100	95	95	92	96	100	104
Notes: Motor vehicle part and accessories (381–383): manufacturing (3). Source: TIPS (2003).

While there have been some productivity gains in autos and auto components in the last few years, these gains have not been exceptional by comparison with many other local manufacturing sectors. Nor does it seem likely that productivity gains have been rapid enough as to underpin an industry‐wide transition from what is universally agreed to have been a highly inefficient industry prior to 1995, to a situation where, in a short space of time, the industry is now internationally competitive.

Despite some productivity gains, the South African auto and auto components industry has been and remains internationally uncompetitive. This would suggest that the rapid growth of auto and auto component exports has been underpinned by other factors.

The IRCC is an additional ‘benefit’ earned by auto and auto component exporters. IRCCs enable most auto and auto component imports to enter without payment of tariffs. The collection efficiency rate – the actual duties collected, compared with the potential duties collected – for the broad category of vehicles, aircraft and ships is only 18,5 per cent. The effective rate of protection for motor vehicles is 81 per cent and for motor vehicles parts 64,8 per cent. Based on collection rates, however, the effective rate of protection for motor vehicles is 10,6 per cent and 0,5 per cent for motor vehicle parts (Cassim et al., 2002: 52, 56, 59). The government clearly collects less tariff revenue. The question is whether local consumers, despite the fact that imports enter without payment of duty, still confront prices that are higher than the duty‐free (i.e. world) price.

Barnes et al. (2003) compare retail car prices in South Africa in 2001 and 2002 to retail car prices in the United Kingdom and in the European Union, differentiating between countries with and without a significant domestic auto industry. They argue that, at equivalent levels of quality and specifications, South African consumers face lower product prices.

Flatters (2002) raises four objections to this price comparison:

	There are very significant difficulties involved in international price comparisons of complex consumer durables and in controlling for all the variables and product characteristics.
	Several models are only available in South Africa. This makes cross‐country price comparisons over the entire product range, which are necessary to pick up possible cross‐subsidisation between different models, impossible.
	Retail selling prices include distribution, selling and financing costs in addition to the ex‐factory cost. The customer list price is not therefore equivalent to the ex‐factory price.
	Most critically, the question of whether consumer prices are higher in South Africa than they would be in the absence of the MIDP, is not resolved by the comparison presented. The resolution of this question requires a comparison of ex‐factory prices of South African assembled autos with the cost, insurance and freight import price for the same vehicle.

A further objection is that changes in the nominal value of the currency make comparisons very difficult, particularly so for an industry that has large fixed capital costs amortised over a lengthy period. With the rand constantly depreciating, auto producers did increase their local prices, but could not increase their prices rapidly enough to compensate for the declines in the currency. This was especially the case following the accelerated depreciation of the currency from the third quarter of 2001. As a result, for a number of years – but particularly after the third quarter of 2001 – local consumer car prices, when converted to foreign currencies, were substantially lower than the auto producers would have liked. This, in turn, reduced the profitability of the local manufacturers. In the last few years, therefore, South African auto prices have been well below their long‐term equilibrium level. As Toyota SA Chief Executive Johan van Zyl explained, ‘Although we put up our prices, we could not catch up with the deteriorating currency –not just last year, but in previous years. However, the recovery of the rand towards the end of last year helped and if things continue as they are, we will return to profitability this year’ (Business Day, 2003a: 11).

South African auto prices for 2001 and 2002 were artificially low. Local auto and auto component prices are likely to increase despite the appreciation of the rand. While this is hardly definitive evidence, utilising the manufacturer's published list prices and the monthly exchange rates published by the South African Reserve Bank, in the six‐month period September 2002 to March 2003, the South African list price of the BMW 318i, for example, converted to euros, increased by 34 per cent (8 per cent due to the increase in the rand list price, the rest due to the appreciation of the rand); and the South African list price of the Mercedes C‐Class 180K Classic increased by 28 per cent (3,5 per cent due to the increase in the rand list price, the rest due to the appreciation of the rand). At the prevailing exchange rates, the current list prices for South African autos are higher than those in Europe.

The debate as to whether or not the MIDP provides a rent to auto and auto component exporters, the incidence of which falls on South African consumers in the form of higher prices for locally produced and imported autos and auto components, has not been resolved, and cannot be resolved – either way – on these price comparisons, or through any such price comparisons.

A consideration of the process by which an exporter can benefit from utilising an IRCC provides an alternative way of ‘demonstrating’ that the IRCC constitutes an export support, whose incidence falls on local consumers. Any exporter can choose to realise the IRCC earned in one of three ‘modes’:

	The exporting firm itself imports vehicles and/or components to the value of the IRCC duty free.
	The exporting firm transfers the IRCC directly to a firm that utilises the IRCC to import vehicles or components.
	The exporting firm sells the IRCC to a local firm who then utilises the IRCC to import vehicles or components.

Where the exporter utilises the IRCC ‘in‐house’, or where the exporter transfers the IRCC to another firm in return for some service rendered (often this is in exchange for marketing and distribution services rendered in respect of the export by the firm acquiring the IRCC), no ‘price’ is fixed and no money changes hands. There is no evident means of determining the benefit that the exporter derives from using the IRCC in either of these two modes.

By contrast, the third mode, by which an exporter can realise the benefit of the IRCC, namely via direct sale, reveals the situation in a transparent manner – the value of the IRCC to the exporter is immediately and directly evident.

An exporter who earns an IRCC and who then sells it, quite evidently receives a return in addition to what is earned on the export market. This is the export subsidy or support. Where the IRCC is sold at arms‐length in the market at a recorded price, the revenue earned by the exporter from the IRCC can be clearly established. There is, in fact, a well‐established spot‐market for arms‐length sales of IRCCs known as the Export Credit Exchange (ECE). The price obtained through the sale of the IRCCs in the ECE provides a clear indication of the value of the IRCC to the exporter, namely the export subsidy.

Who bears the incidence of this export support? The firm that purchased the IRCC will utilise it for importation. The IRCC will accordingly allow the firm to import at lower cost through paying less duty. This reduction in duty paid will not necessarily be passed on to the consumer. The firm will recoup the costs of purchasing the IRCC by adding this cost to the selling price to be paid by the consumer. The consumer therefore pays the import cost less the tariff, but must also bear the cost of the importer acquiring the IRCC. The price of the IRCCs prevailing on the ECE therefore gives a fair indication of the difference between the price paid by the consumer and the tariff‐free price.

In 2002, R550 million of IRCCs issued were sold through the ECE out of a total of some R21 billion. Thus, only 2,6 per cent of total IRCCs are sold on the ECE. A further 10–15 per cent of IRCCs are sold, but outside the ECE. It might therefore be objected that the value to exporters of the IRCCs they earn cannot be ‘read’ from the prices prevailing in a market in which only a small percentage of IRCCs are traded.

However, there are many instances where only a tiny share of the product is traded and the market nevertheless provides a fair indication of value. Suburban housing and grain markets in largely subsistence economies are just two examples. The producers of grain or suburban houses who choose to use their home or their grain (the vast majority), forgo the revenue given by the return they would have obtained in the market had they sold their grain or suburban home. Abstracting from transaction costs, consumers will use their home or their grain only as long as the ‘benefit’ obtained from so doing is equal to or greater than the benefit or revenue foregone through not realising the sale of their house or their grain on the market. Similarly, those who choose not to sell their IRCCs on the open market will only do so if the ‘benefit’ they so derive is equal to, or greater than that which they could obtain on the open market. The benefit derived through alternative uses of IRCCs can therefore be imputed from the price prevailing for IRCCs in the open market.

In order to maximise profits, firms will allocate IRCCs so that the returns from the various usages to which IRCCs can be put are equal at the margin. Returns at the margin obtained through any usage of the IRCC will be equal to the return at the margin obtained through the sale of the IRCC. The equivalent of what could be obtained through the sale of an IRCC will accordingly be added to the price charged to the local consumer for the import.

The IRCC therefore does constitute a support to exporters borne by local consumers in the form of prices in excess of the tariff‐free or world price. This support is substantial. The existence of a market price for the IRCCs allows for the precise extent of the support to be calculated, but this is not attempted here.

There are a number of concerns in regard to the MIDP that suggest that it may not have been as effective as is generally perceived:

	The first concern is local content. Local content for completely built‐up vehicles has actually declined, particularly in exports of vehicles and first‐tier components. Many local component exporters established under the MIDP do not supply domestic assemblers, but rely rather on export markets and the sale of IRCCs to subsidise their exports. Given that the IRCC benefit is realised through duty‐free importation, this result is unsurprising.
	The second concern is the range of components exported. This has not increased significantly – much of the export growth has been in catalytic converters and leather seats. These two items make up 62 per cent of component exports, with catalytic converters accounting for 48,4 per cent of component exports (Financial Mail, 2003a: 14). The value added in catalytic components is very low. Leather export handicaps other potential users of leather where value addition and employment are much higher. Where component production is solely for export, the IRCC leads to exports below average costs, resulting in a clear welfare loss (Black & Mitchell, 2002: 1291).
	The third concern is the distribution of the IRCC benefits within the industry. Exporters are most likely to sell a high proportion of their IRCCs on the market where exports have a low import content, such that they have limited ‘in‐house’ use for the IRCCs earned, and where they have their own independent export channels, such that they do not require the services of other auto firms to engage in exporting. This is especially characteristic of independent component producers. Indeed, they are the preponderant sellers of IRCCs on the EDE. The revenues from the sale of IRCCs is critical to many component suppliers who would otherwise not be competitive on export markets.

‘Some component makers rely heavily on the proceeds of the sale of certificates for profitability. Some of the component manufacturers would have used the proceeds from the sale of import rebate credit certifications to make up what they lose on the export markets – where they are constantly under pressure from multinational motor companies to keep prices low’ (Sunday Times, 2003: 3). In respect of these exporters, the IRCCs are quite evidently an export subsidy. The main buyers are large producers with high import content and, more particularly, groups such as tyre manufacturers and the distributors of imported vehicles who do not export and are ‘pure’ importers. With the expansion of exports and the consequent increase in the supply of IRCCs, the price of IRCCs has fallen. Previously IRCCs were being sold on the EDE at between 82,5 and 86 per cent of their value. However, in March 2003 this had fallen to 67 per cent. The result is a redirecting of a significant and increasing share of the benefit from exporters selling, to firms purchasing IRCCs to import. This considerably dilutes the support obtained by the independent, smaller, locally owned, component exporters from the IRCC – precisely those firms that, it could be contended, should receive the highest level of support.

The final concern is that domestic demand has been effectively stagnant since the MIDP was introduced in 1995. There clearly are a number of factors at play. The high cost of financing auto purchases for many, particularly lower‐income, consumers, for example, is a significant constraint. However, the lack of low‐cost cars at the bottom end of the market and the higher prices borne by consumers generally are also important factors. The increases in local vehicle prices have consistently exceeded the increase in consumer prices. Thus, according to WesBank's CEO, ‘the rise in vehicle prices outstripped consumer price inflation. In 1985 it took 40 weeks’ net income to pay off an entry‐level executive vehicle, but today it takes 72 weeks' net income' (Business Day, 2003b: 4). Furthermore, the official data may understate the extent of the price rises. In 2002, Statistics South Africa reported a 9,4 per cent rise in vehicle prices, while the List Price Index provided by the National Association of Automobile Manufacturers (Naamsa) reported an increase of 18,4 per cent (Financial Mail, 2003b: 16).

It is clear that, despite evident success in terms of exports and investment, an assessment of the desirability and the impact of government support for the auto and auto components sector must be cast far more widely.

2.2.2 Clothing and textiles

In contrast with the auto and auto component sectors, clothing and textiles have experienced stagnant or declining production post‐1995.

Of particular concern is the slow rate of growth of clothing exports (Table 7). The growth of South African clothing exports to the principal market, the United States, following the implementation of African Growth and Opportunity Act (AGOA) has been much slower than expected. Indeed, South Africa's clothing exports to the United States were increasing more rapidly in the period immediately prior to the AGOA than in the post‐AGOA period. This weak export performance for clothing is particularly evident when contrasted with the export performance of other countries in the region. While the shares of other African countries eligible under the AGOA are rising rapidly, South Africa's share of the American market has been declining.

Clothing exports from Africa to the US and the EU, leading suppliers, 1990–2002 (US$ million)

	Kenya	Kenya	Lesotho	Lesotho	Madagascar	Madagascar	Mauritius	Mauritius	SA	SA
	>US	>EU	>US	>EU	>US	>EU	>US	>EU	>US	>EU
1990	2,5	2,5	24,5	5,6	0,4	10,8	121,2	522,7	0,0	32,3
1991	4,5	6,3	27,0	18,2	0,1	15,1	97,7	536,5	0,7	72,7
1992	7,8	17,4	50,8	18,3	0,2	18,5	113,1	533,9	2,4	73,2
1993	22,1	10,3	55,1	14,7	1,5	46,3	161,2	501,0	12,7	75,5
1994	35,2	7,1	62,4	13,5	2,8	92,6	186,2	518,8	34,7	73,4
1995	34,0	6,3	61,7	12,6	6,7	122,0	190,3	573,3	16,6	66,9
1996	27,1	3,3	64,9	12,7	11,0	147,7	164,7	616,0	60,4	67,1
1997	31,3	2,6	86,5	4,5	15,3	177,1	184,4	658,0	70,9	62,3
1998	33,5	2,3	100,2	0,8	22,0	218,0	233,3	693,2	78,7	69,4
1999	39,3	2,5	110,7	0,2	45,7	213,9	231,6	625,2	96,9	68,3
2000	43,8	1,7	140,1	1,6	109,5	244,7	244,7	638,5	140,9	78,6
2001	64,4	1,7	216,7	3,2	178,2	238,3	238,3	591,2	173,3	69,0
2002	125,5	1,1	321,1	2,1	89,3	145,6	254,5	642,3	181,0	68,7
Note: Euro/US$ based on rates for 31 December for the relevant year. Source: Gibbon (2003: 47).

Concomitantly, despite the AGOA, there have been very few new investments in South Africa, particularly on the part of large integrated clothing firms focused on the American market. There has also been little new investment on the part of existing export‐oriented clothing firms. By contrast, other African countries have attracted significant new investments in clothing and, more recently, in some countries (notably Lesotho) in textiles as well.

South African exports to the United States are located overwhelmingly in those categories that are most regulated by quotas – especially cotton trousers and cotton knit shirts and blouses, where 75 per cent of South African exports to the United States are located. These are the two top quota‐constrained categories and account for 33 per cent of American imports overall (Minor et al., 2002: 4). Thus, South African exports of clothing will be very vulnerable when quotas are removed in 2005. Moreover, with the removal of quotas, all the indications are that American buyers will significantly reduce the number of source countries.

As in autos and auto components, policy in the clothing and textiles sector has entailed a tariff phase‐down combined with an export incentive. Following accession to the World Trade Organisation, the Swart Panel recommended a ten‐year phase‐down period, together with investment and input subsidies and export and training incentives. However, this was rejected and the government introduced a seven‐year tariff phase‐down for clothing, with the elimination of specific tariffs and the reduction of ad valorem tariffs to a maximum of 40 per cent by 2002. For textiles, tariffs were also significantly reduced, with average tariffs for yarn falling from 50 per cent in 1993 to 22 per cent in 2003, and for yarns from 35 to 15 per cent. In addition, firms importing yarn, fibre or fabrics in order to produce clothing for export were able to do so on a duty free basis. This is the regulation known as 470.03.

An explicit export incentive, known as the Duty Credit Certificate Scheme (DCCS), was introduced to coincide with the phase‐down and this was subsequently extended to the end of 2005. The DCCS allows firms to claim a remission of duty (hence duty credit) for proven exports. The level of support depends on the product exported – with highest support for clothing, followed by fabric and then yarn.

Thus, in addition to their export revenues, clothing and textile exporters earn via the DCCS a remission of duty paid on imports equivalent to a significant percentage of their export revenues. This would appear to be a powerful export incentive. Why has it not had a more significant effect on exports of clothing and textiles – more particularly for clothing where the DCCS incentive is far higher? Four factors reduce the impact of the DCCS on exports:

	The DCCS is an alternative to regulation 470.03. Firms availing themselves of 470.03 are prohibited from utilising the DCCS. It is not known exactly how many firms or what share of total exports are subject to 470.03. In 1997, 67 exporters were exporting under 470.03 (Gibbon, 2002: 18). The best estimate is that they currently account for between 55 and 60 per cent of total clothing exports. While there are some variations, these firms typically focus their entire, or close to their entire production almost exclusively on the United States. It is important to note, therefore, that more than half of South African clothing exports enjoy access to imported inputs duty free, but they enjoy no export incentive. By contrast with the auto and auto components industry where all exporters receive imported inputs required for production for export duty free and the IRCCs, the bulk of South African clothing exports only receive imported inputs required for production for the export market duty free. One major reason for the limited impact of the DCCS on exports is simply that more than half of South Africa's clothing exports do not receive the DCCS, opting rather for duty‐free importation.
	To earn a DCCS, exporters must have paid all duties in respect of their imported inputs. The incentive to exporting provided by the DCCS is considerably diluted, in that the DCCS can only be accessed by firms that face high tariffs on their imported inputs. These duties are considerable. In important part, the DCCS serves to compensate for the fact that South African exporters, availing themselves of this incentive, have to pay substantial duties on their imported inputs.
	There are significant restrictions on the usage of the DCCS. The DCCS can only be used for inputs in respect of the domestic market and then only for a same stage product or one stage back. Thus, a clothing exporter, for example, may only use the DCCS in respect of importation for sale or production for the domestic market and then only in respect of clothing or fabric. One consequence of this restriction is that exporting firms earning the DCCS utilise only a minority of the DCCS in‐house for own account imports. The majority of the DCCS is accordingly sold. The major purchasers of the DCCS are the large retailers, who then utilise the DCCS in respect of their imports – particularly for clothing. The large retailers have considerable market influence and the discount incurred on the sale of the DCCS is typically 30–40 per cent. As with the IRCCs, the discount represents that value ‘captured’ by the importer and the discount is a measure of the extent to which the actual value of the incentive to the exporter is below that of its nominal value. Largely as a consequence of the restrictions placed on in‐house usage of the DCCS, a large share of the DCCS is sold at a high discount.
	There are two other, albeit less important, factors relating to the DCCS. First, in order to qualify for the DCCS, a clothing or textile exporter has to meet certain requirements. Firms are required to spend a certain share of their wage budget on training and to participate in achieving competitiveness targets set by a Productivity Performance Monitoring Scheme. Amenities and facilities are inspected and management practices are reviewed. Compliance costs are therefore not insignificant. Second, only exports on which duty has already been paid qualify for the DCCS. Thus, the expenditure in respect of duty payment will be incurred generally several months before the DCCS is earned. Time delay imposes further costs.

The vast majority of clothing exporters utilising the DCCS are adamant that exporting would not be profitable without it (Gibbon, 2002: 42; Velia, 2002: 41). Without the DCCS, fewer firms would engage in export. Nevertheless, the costs and restrictions entailed in the DCCS and the applicability of the DCCS to less than half of the export total, have significantly limited the impact of the DCCS on exports.

How might exports, particularly of clothing, be more effectively encouraged? There are currently two distinct categories of clothing exporters. The first – ‘the 470.03ers’ – are almost entirely export oriented, invariably foreign‐owned (mostly Taiwanese) and located in decentralised areas. These firms have a long‐established track record of successful exporting and, as noted above, have done so without any export incentives. As South Africa already has a well‐established and internationally cost‐efficient clothing sector exporting to the United States, one would have expected that easing of entry into the American market under the AGOA would have resulted in substantial expansion of this part of the clothing industry. However, the existing firms are not significantly expanding their production or investment for the export market and South Africa is not currently attracting new export‐oriented firms. By contrast, other countries in the region are seeing the rapid expansion of existing firms and are attracting new entrants.

One part of the reason is labour costs. However, these firms are invariably located in the former decentralised areas where wages are very low and comparable to those elsewhere in the region. There is also a concern about labour regulations, particularly inflexibility with regard to hiring and firing. However, these also do not appear to be onerous. Moreover, any disadvantages of locating in South Africa as a consequence of higher labour costs or labour regulations are offset by generally higher productivity and better infrastructure.

For these firms, the major constraint in expanding their exports to the United States is access to fabric. In order to qualify under the AGOA, South African clothing exporters are subject to a three‐stage rule of origin – spinning, weaving or knitting and assembly and finishing must take place in South Africa, the United States or another beneficiary country. Less developed beneficiary country (LDBC) status is conferred on other countries that qualify under the AGOA, except for Mauritius. LDBC status entails only a one‐stage rule of origin, i.e. only assembly and finishing need to be undertaken in the exporting country.

The large internationalised clothing producers source cloth globally in large quantities. This is an important part of their competitive advantage. They then allocate production amongst their different plants located in a variety of low‐cost locations. Rules‐of‐origin restrictions in respect of plants located in South Africa interfere with this process and render this country a less advantageous location by comparison with other sites that are free of such restrictions.

In addition to the chorus of cries emanating from clothing exporters (Export Council for the Clothing Industry in South Africa, 2002a, 2002b), evidence suggests that fabric shortage is a major constraint on clothing exports to the United States.

First, where South African clothing producers do have access to the requisite fabric – manmade fibres and wool, in particular – there has been significant export growth and South African producers have been expanding their share of the American market. By contrast, in the case of cotton apparel, exporters have had major difficulties in sourcing the requisite fabric locally and the consequence has been a significant decline in cotton‐based apparel exports (Minor et al., 2002). Prior to the AGOA, cotton apparel constituted more than 90 per cent of South Africa's apparel exports to the United States (Minor, 2002: 10), and it is cotton apparel that has accounted for the comparatively poor performance of South African clothing exporters to the United States. This decline is unique to South Africa.

Second, recent producer price indices for textile inputs are very revealing. Textile and particularly spinning and weaving manufacturers were able to increase their prices very significantly early in 2002 (Figure 3). While total manufacturing and clothing prices also increased with the sharp depreciation of the currency, the increases were far larger in respect of textile products. Moreover, the rate of increase of manufacturing and clothing prices began to decline thereafter. But, as the AGOA came into effect, textile product prices maintained their rate of price increases. This is strongly suggestive of considerable fabric shortage.

Price comparison – clothing, textiles and total manufacturing Source: Export Council for the Clothing Industry in South Africa, private communication, May 2003.

Third, in the 12‐month period ending in May 2003, 48 per cent of South Africa's clothing exports to the United States were AGOA eligible. The figure for Mauritius is 43 per cent. For all the other countries that have less developed country status, the percentage is far lower – above 90 per cent being AGOA eligible.

The fabric shortage will have to be addressed if South Africa is to attract further investment for export to the United States. This can only be satisfactorily achieved if rules‐of‐origin regulations for South African clothing exports to the United States are altered significantly. Preferably, they should be the same as those enjoyed by other AGOA‐eligible countries, i.e. that South African clothing exports wholly manufactured from fabric derived from third countries be permitted to the United States under the AGOA.

Any change in rules‐of‐origin regulations in respect of South African clothing would almost certainly require reciprocally that American textile exports into the South African Customs Union (SACU) face substantially reduced tariffs or be allowed tariff‐free entry. The clothing industry is currently advancing this as a central objective for the proposed US‐SACU Free Trade Agreement (Clotrade & the Export Council for the Clothing Industry in South Africa, 2003).

Shortage of fabric also impacts negatively on the other clothing exporters who receive DCCS support. These exporters are also subject to additional constraints. Many of these firms are located in established metropoles – Cape Town and Durban especially‐and are faced accordingly with much higher labour costs and a strongly unionised workforce. Higher labour costs and high tariff levels for inputs result in a high cost structure, which severely constrains the ability of these firms to compete in export markets.

Largely as a consequence of these constraints, the firms adopt a different production model compared with the model employed by firms that strongly focus on exporting to the United States (Gibbon, 2003). By contrast with the latter, that specialise in the assembly and finishing of long runs of basic garments, the former focus on shorter runs with much more investment in other activities such as design and sampling. Indeed, this different model severely constrains the capacity and willingness of such firms to engage in exporting to the United States. Many firms export only a small share of their total production, often only what is surplus to their domestic production, and more often to the European Union. Unlike the ‘470.03ers’, these firms cannot export competitively and they rely heavily on the DCCS to subsidise their exports.

A high cost structure resulting in a particular production model that further raises costs, significantly inhibits export on the part of these firms. The DCCS is designed to overcome these obstacles and to compensate for the higher costs that exporters incur in respect of labour and imported inputs.

As outlined earlier, while the DCCS does provide a significant incentive to exporting, its efficacy is constrained by a number of factors. Most critical is the limited usage to which the DCCS can be applied. To reiterate, the DCCS can only be used for imported inputs in respect of the domestic market and then only for a same‐stage product or one stage back. This limited usage for the DCCS results in exporters selling a large part of their DCCS to importers at a considerable discount. The DCCS would be a more effective export incentive if its usage could be widened. It could be used as against imports not only for production for the domestic market, but also for the export market. Furthermore, the DCCS could be used not only for the import of the same product plus one stage back, but also for all imported inputs.

Widening the use to which the DCCS could be put would not result in an exporter earning more of the DCCS, but it would allow the exporter more possibility of using the DCCS ‘in‐house’ rather than selling it at a discount. Less of the DCCS would be supplied to the market and the price of the DCCS would rise, reducing the discount and the value captured by the importer as opposed to the exporter.

The DCCS is necessary to compensate clothing exporters who face higher imported input costs. In the longer term, policy should aim at allowing clothing exporters access to inputs at international prices. If tariffs on textiles are reduced, pari passu this will allow the level of the DCCS benefit to be accordingly reduced, without a reduction in the incentive to export. Thus, the level of benefit enjoyed by clothing exporters through the DCCS can be adjusted and synchronised in line with tariff reductions on textile inputs.

Addressing the fabric shortage and widening the usage of the DCCS will give a significant fillip to export expansion on the part of the South African clothing industry. However, there would clearly be some negative impacts on the local textile industry, more especially if, as suggested, this is combined with tariff reductions for textiles in general and, as suggested above, tariffs are reduced for American textiles. The DTI's industrial strategy is committed to developing the textile‐clothing value chain (DTI, 2002: 30). Further consideration will need to be given to deciding what form of policy support would, in this context, be most appropriate for the textile industry.

However, it is important to recognise that an expanding export‐oriented local clothing industry will be of immediate benefit to the local textile industry, at least that part of the textile industry that produces inputs for clothing production (currently approximately 30 per cent of textile output is for clothing). International experience is that expansion in clothing leads expansion in textiles. There is little likelihood of growth in textiles without a significant expansion in the local production of clothing.

However, without the expansion of the local textile industry the local clothing industry will be significantly constrained. In the immediate term, this constraint arises from the shortage of suitable AGOA‐compliant fabric. In the short to medium term, a very close integration with the supplier of fabric will be essential for the success of any clothing manufacturer intending to export to the United States. Buyers are demanding ‘full‐package producers’. This is particularly important in the fashion‐basic segment of the market where South African producers have the best chance of competing (Minor, 2002: 21; Gereffi & Memedovic, 2003: 31).

Clothing exporters need the local textile sector to expand, both in the immediate short term to alleviate the fabric shortage and in the longer term to secure markets. Local textile firms are hanging back because they are not convinced that clothing exports will take off. Textile firms are accordingly focusing on reducing fabric variety and producing longer runs of higher‐quality fabrics. This further limits the access of potential clothing exporters to the requisite fabrics. The scepticism of the local textile producers as to the expansion of a local export‐oriented clothing industry is thus rapidly becoming a self‐fulfilling prophecy.

What role is there for industrial policy in this conundrum?

2.3 Sectoral support measures and industrial policy

This section discusses the support extended to the auto and auto components and clothing and textiles sectors in South Africa in the light of some broad perspectives on government support for specific industrial activities. Drawing on these perspectives, some directions for future policy design are proposed – very generally in respect of autos and auto components and, somewhat more specifically, in respect of clothing and textiles.

2.3.1 Autos and auto components

In a developing country context, the promotion of non‐traditional activities may require government‐supported inducements (Rodrik, 2003). Potential investors in non‐traditional products in a developing country context operate in a situation of a high level of information uncertainty. Once the investment is successfully made, however, the information is disseminated – other producers can readily see what the costs and gains might be of further investment. Industrial policy thus acts as a coordination device (Rodrik, 1995: 21).

Industrial and broader economic development in South Africa, as elsewhere, requires that new areas of investment and of specialisation be developed – more specifically, new non‐traditional products that have a higher value added and have high growth rates in international markets. Autos and auto components represent one such category of products, and have indeed been the major addition to South Africa's export product basket that has otherwise changed very little.

Producing highly sophisticated German designed autos in South Africa for the most demanding export markets of Europe and Japan could certainly qualify as a non‐traditional activity with a highly uncertain outcome for the auto producers. Ex ante the decision to produce high‐quality German‐designed cars in Africa for discriminating Japanese and European consumers was a considerable gamble. The question was whether the South African firms would be able to produce at the demanding standards of the markets in industrialised countries, and whether the consumers would recognise and accept these standards in products produced in Africa. The auto producers were prepared to take that gamble as long as support was forthcoming from the government. Demonstrated success on the part of the initial investments has removed the uncertainty. Combined with continuing government support, the removal of uncertainty has subsequently crowded in new investors.

Very broadly, this is typically the ‘arena’ where an active industrial policy on the part of the government could potentially play a positive role. There have indeed been many positive spin‐offs resulting from the expansion of the auto and auto component sector. This is more especially so in that production was aimed at highly discriminating export markets that would discipline inefficiencies. Exports could be coupled to policies that, while they would continue to ensure high levels of protection in the local market, would simultaneously aim at enhancing efficiencies in the local market, principally through boosting the scale of production by reducing the proliferation of locally produced models. Moreover, there are considerable externalities, particularly as the auto exporters encourage and support their local suppliers to improve the quality of their products and as they facilitate the opening of new export markets for auto component producers. More widely, the demonstrated capability of making sophisticated products for the most discriminating markets provides a positive lesson and reduces the risks for producers in other areas contemplating South Africa as a production site for other sophisticated products.

Now that the considerable uncertainty surrounding the ability of South African firms to produce autos and auto components for the most discriminating export markets has been removed and consumer acceptance secured, there is a clear case for significantly reducing the levels of government support. Reductions in government support have indeed been forthcoming in subsequent revisions to the MIDP (Black & Mitchell, 2002: 1295). However, support extended under the MIDP remains very significant. Moreover, additional support for the sector has recently been implemented. Despite its name, the MIDP is a trade measure confined to export facilitation. The two hallmarks of an active industrial policy, namely support for investment and technological advance in this sector, have however been added very recently. Support for investment is provided through the Productive Asset Allowance, whereby new capital investment for export earns import duty credits of 20 per cent of the value of the investment spread over five years. Support for technological upgrading is provided by the national government and provincial government, for example, the Advanced Manufacturing Strategy of the Department of Science and Technology, and the Gauteng government's Blue IQ project.

Sector‐specific trade and industrial policies should be predicated on an assessment of dynamic competitive advantage; that is, the sector's prospect, in a defined period, of competing internationally without government support. The South African domestic market is small and the regional market very minor. Most of the inputs must be imported over a considerable distance. This is a scale‐intensive industry and volumes are critical. Accordingly, even if in their own operations, domestic firms become internationally efficient, location imposes considerable transport, logistics and marketing costs. Critically, there is no evidence that this is becoming less constraining – domestic demand, in particular, has been essentially static for several years.

This would suggest that this sector will continue to rely on government support. A careful assessment is needed of the impact on local consumers, particularly business consumers, and more especially small and undercapitalised businesses, and on employment, as downstream of production the employment numbers are considerably larger and production far less capital intensive. The results of an economy‐wide assessment should be factored into future support for the industry.

2.3.2 Clothing and textiles

Currently, both textile and clothing firms are hanging back on investment and blaming each other for the lack of any movement. Firms in the two sectors are taking different paths – textile firms are looking to longer runs for the export market, whereas the clothing exporters are increasingly looking elsewhere for fabric. As this translates into different strategies on the part of the firms, the chasm between the textile and clothing sectors grows ever wider. There is accordingly very little common purpose between the associations representing the two sectors.

The further development of the South African clothing and textiles sector requires simultaneous investments by both clothing exporters who then provide an impetus to textile investors to supply the necessary inputs, and by the textile manufacturers so that the clothing exporters, assured of the fabric that they need to be successful in export markets, will now be induced to invest. Investments on the part of both clothing exporters and textile input suppliers entail significant extra‐sector spillovers in the form of intermediate inputs or market demand spillovers for final goods whose production entails significant scale economies. Private returns on the investments of both clothing exporters and textile input suppliers are accordingly significantly below the social returns.

This is a classic case where the state needs to take the lead in order to secure coordinated and simultaneous investments. The sector associations are not likely to produce a unified vision for the development of the entire value chain. Left to the market, there will be no forward movement; only the state can play this role. This role of the state in coordinating simultaneous investments across linked industries has a long tradition in economics, going back to Rosenstein‐Rodan whose planned industrialisation for Eastern Europe comprised ‘a simultaneous planning of several complementary industries’ (Rosenstein‐Rodan, 1943: 204; Hirschman, 1958).

There is no attempt here to provide a policy blueprint for clothing and textiles. There would, however, appear to be a strong case for reducing investment risk through some form of subsidy or support for investment. This will be particularly important in respect of investments in textiles, where the capital investment component is large and much more significant than in clothing. Any subsidy to investment should also critically include support to clothing firms who seek to integrate backwards into textiles in order to advance their clothing exports.

There is also a strong case for critically assessing other factors that prevent the investment in textiles required to enhance the development of clothing exports. Of central importance here is the requirement that textile producers have to purchase the local cotton crop (Coughlin et al., 2001). This constrains not only the textile producers directly, but also clothing exports and, by so doing, indirectly but significantly further constrains textile investments.

3 CONCLUSION

Over the last decade, South Africa's manufacturing growth and export performance, when viewed in a comparative perspective, has been weak. In terms of equity, apart from an increase in remuneration, there has been little progress. The generalised supply‐side support measures available to all sectors, which have been at the centre of industrial policy, appear to have had limited reach. Where firms have utilised these measures, the indications are that they have not been very effective. Coherent and integrated sector strategies are similarly not evident.

The explanatory factors underlying the limited impact of the supply‐side measures are not immediately obvious and much more investigation is required. It may be that the problem lies with poor implementation rather than the measures themselves. However, the fact that many firms that make use of the support measures report that their impact is limited, suggests that something more than poor implementation is at issue. Moreover, the efficacy of supply‐side measures depends in important part on ensuring that they are integrally linked to an overall strategy for the sector. Despite the DTI's integrated manufacturing strategy, cohesive and integrated strategies are not yet in evidence for any of the manufacturing sectors.

With respect to two critical sectors, autos and auto components and clothing and textiles, substantive specific measures for export facilitation have been implemented. These measures have certainly provided a significant impetus to exports of autos and auto components. However, they have also entailed significant economy‐wide distortion and cost. In clothing and textiles, the impact on exports has been much more muted, despite the existence of new export opportunities under the AGOA.

The limitations of the DTI's specific support measures for the auto and auto components and clothing and textiles industries should not, however, be seen as an argument against support measures to specific industry sectors per se. There is a considerable role for industrial policy and, at least in respect of the clothing and textiles industry, it is argued, there is a need for the government to play a more active role in coordinating and supporting private investment.

However, sector strategies and associated interventions stretch governmental capacities. Governments will accordingly need to be limited in the number and complexity of the sectors that they select for specific attention and support. The DTI's industrial strategy (DTI, 2002) identifies eight priority sectors for specific encouragement and support. In addition to automotive and transport and clothing and textiles, these sectors are: agro‐processing; metals and minerals; tourism; crafts; chemicals and biotechnology; and knowledge‐intensive services. Other priority sectors have been added since, including aerospace, call centres and back‐office operations.

In the light of the limited success achieved with respect to its two selected sectors thus far, the DTI's industrial strategy would appear to be over‐ambitious. The DTI will need to be more focused and more limited in the number and scope of the sectors that are targeted for customised policy support.

There is a need to re‐examine South Africa's industrial policies in order to chart a more effective way forward. A review should entail not merely an examination of existing policies with a view to producing better policies. It should also critically entail a consideration of the institutional and organisational capacities underpinning industrial policy. An assessment of organisational capacities will be critical in order to determine the optimal scope of industrial policy and to identify which policies can be effectively implemented (Kaplan, 2003: 38–45).

Additional information

Notes on contributors

David Kaplan 1

Professor, Department of Economics, as well as Allan Gray Professor of Business Government Relations, Graduate School of Business, University of Cape Town, Rondebosch, South Africa. This is an adapted and shortened version of a plenary paper given at the Annual Forum of the Trade and Industrial Policy Strategies and Development Policy Research Unit, The challenge of growth and poverty: the South African economy since democracy. Johannesburg, 8–10 September 2003.

Notes

Professor, Department of Economics, as well as Allan Gray Professor of Business Government Relations, Graduate School of Business, University of Cape Town, Rondebosch, South Africa. This is an adapted and shortened version of a plenary paper given at the Annual Forum of the Trade and Industrial Policy Strategies and Development Policy Research Unit, The challenge of growth and poverty: the South African economy since democracy. Johannesburg, 8–10 September 2003.