Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
3. Microelectronics and the third world
3.1
Introduction
3.2 Patterns of adoption and diffusion in the
third world
3.3
Impacts of microelectronics
3.4 Policy implications and future research
directions
Acknowledgements
Notes
References
For
further reading
Jeffrey James
This survey will be guided by recognition of two points which I hope will facilitate understanding of the impacts that innovations in microelectronics have on developing countries. The first is that the social impacts of technical change in general, and thus of microelectronic innovations in particular, are importantly influenced by the way in which innovations are generated and diffused. The second is that a wider range of economic and social analysis than is usually considered is in fact relevant to understanding the innovation processes and impacts associated with microelectronics. With reference to these two points, this chapter will attempt an integrative focus.
The purpose is not, however, to meet all the formidable integrative requirements imposed by this general framework. The purpose, rather, is to undertake part of the integration and to urge a far greater effort along similar synthetic lines than is typical of the existing literature. Perhaps mainly because it tends to emphasize the singular features of microelectronics-based innovations, this literature is inclined to disassociate itself methodologically from the economics of technical change on the one hand and from development economics on the other.
Yet, in our view, these bodies of thought can contribute greatly to understanding the economic impact of microelectronics on the Third World. Thus, the first part of this chapter (Section 3.2) will attempt to show how aspects of the literature on technology transfer choice of technology, multinational enterprises (MNEs), and the diffusion of innovations can be used to help explain prevailing patterns of adoption and diffusion of microelectronics in developing countries. On the basis of these findings and with reference to an equally broad range of literature, the discussion is led, in Section 3.3, to identify the major development issues to which these patterns give rise, to suggest ways in which they may be tackled, and to assemble the evidence that bears on each of them. Emerging from the arguments of both these sections are issues for policy and for research directions. These are taken up in Section 3.4.
3.2 Patterns of adoption and diffusion in the third world
3.2.1
Supply-side factors
3.2.2
Demand-side factors
Understanding generation of new technology is concerned with the forces that determine the rate at which new technologies are created and the form that they take. Diffusion, in contrast, concerns the spread of new technologies so created. As such, it deals, on the one hand, with the conditions in which the owner of a new technology chooses to make his innovation (which may be a new process or a new product) available to potential users (who are firms in the case of new processes and households in the case of new products). We shall refer to these conditions as constituting the supply side of the diffusion of new technologies. In addition, there is a demand-side component, which relates to the factors that underlie demands for the new technologies by relevant groups of potential users.
In relation to the agricultural sector, the literature tends to focus almost entirely on the question of diffusion and, in particular, on the demand side of the problem (for example, the characteristics that predispose individual agents to accept or reject innovations). This preoccupation is mainly a reflection of the public goods character of many of the innovations introduced into this sector. For example,
With the exception of a few countries, seed varieties are not effectively protected by licenses. In the case of many varieties, once a farmer has bought a single lot he can reproduce the seed himself and sell it to neighbours without having to pay more to the plant-breeding firm.... It was necessary to create a public agricultural research system to do research in the biological area in order to overcome the lack of incentive brought about by this situation.
(Binswanger and Ruttan, 1978, p. 117)
Once created, the innovations of such public research become freely accessible: that is, nonproprietary. Diffusion of such innovations, therefore, depends only on the side of demand, namely, whether and to what extent the innovations are taken up by the relevant groups of adopters.
Literature on the industrial sector focuses a great deal more attention on the generation of new technologies, because in this sector (unlike agriculture) innovations tend to be proprietary in character. With respect to many microelectronics-based innovations, for example, 'whether final goods or components, vital technical information regarding design, engineering specifications, process know-how, testing procedures etc., is proprietary in nature, covered by patents or copyrights, or closely held within various electronics firms from developed countries' (O'Connor, 1985b, p. 12). There is relatively little discussion, however, of the circumstances that govern the way owners of proprietary knowledge in microelectronics, or indeed new technologies in general, choose to exploit these assets (that is, on the supply-side aspects of diffusion). Yet, in the case of developing countries-for most of whom the generation of microelectronics innovations takes place entirely exogenously-it is precisely this question that one needs to answer to determine the availability of new technologies. Accordingly, we devote the first part of this section to a discussion of supply-side factors in the diffusion of microelectronics technologies in the Third World. Thereafter, we will examine the major determinants of demand for these innovations.¹
It is convenient to discuss the supply-side determinants of diffusion of microelectronics to developing countries under the two broad headings of foreign direct investment and arm's-length markets. We shall begin with the issues raised under the former and then proceed to define the circumstances in which certain new technologies become available to the Third World without the accompaniment of foreign investment (e.g., through licensing).
3.2.1.1 Foreign direct investment
Literature on the activities of multinational corporations in developing countries tends to distinguish sharply between exporters on the one hand and firms producing for domestic markets on the other (Caves, 1982). This same distinction is necessary also as a starting point for explaining the presence of multinational subsidiaries in the electronics-based industries of the Third World.
Export-oriented foreign investment There is a sizeable literature describing the evolution of competitive forces within markets in developed countries that compel multinational electronics corporations to locate production for export in the Third World (Plesch, 1979; ESCAP, 1979; Ernst, 1983; O'Connor, 1985a,b; Henderson, 1989). Many of the most important of these forces-ease of entry, the threat of internal and external competition and the rapidity of technical change-are illustrated with particular clarity in the history of the computer industry.
What is interesting, in particular, is the contrast between the mainframe and the later, smaller, computers. Thus,
Mainframe computer manufacture has historically been concentrated almost exclusively in the advanced industrialised countries. Since the demand for mainframes has not generally been price elastic, lowering production costs through offshore assembly has not been considered a requirement of competition in that product range. Demand for smaller computer systems, however, tends to be more price sensitive, with the result that computer firms have had to economise on production costs in whatever way possible so as to expand their market shares. In addition to this fact, in the low end of the computer market ... firms have employed aggressive market entry pricing strategies which have resulted in periodic rounds of steep price reductions. Combined with the rapid pace of product innovation in this segment of the industry, the intense price competition has forced a number of computer manufacturers to shift production to low wage locations as a short term survival strategy.
(O'Connor, 1985b, p. 6)
Somewhat less numerous than the explanations of why MNEs choose to undertake export-oriented investments in the Third World (such as that quoted in the previous paragraph) are attempts to account for the particular countries selected. One approach employs a cross-country framework to identify the common characteristics of the countries selected (for example, Nayyar, 1978).2 Wheeler and Mody (1988) used this approach to investigate location decisions by U.S. MNEs in manufacturing. What they found, in the case of export-oriented electronics firms, is that these decisions are not a function of labour costs alone: the quality of infrastructure and agglomeration economies appear to be equally important. These results, in turn, help to explain the popularity of East Asian locations among U.S. electronics firms, as reflected, for example, in the fact that in 1983 the vast majority of semiconductor imports under sections 806.30/807.00 of the U.S. Tariff Schedule came from assembly plants in four East Asian countries.3 On the one hand, many countries in the region have a well-developed infrastructure: 'In fact, some Far East locations today have an excellent infrastructure which, more so than in OECD [Organization for Economic Cooperation and Development] countries, is geared to the requirements of global sourcing in electronics' (Ernst, 1985, p. 31). On the other hand, the newly industrialized countries (NICs) appear to have benefited from what Wheeler and Mody refer to as an 'historical first-mover advantage', that is the agglomeration economies that accrued to these countries from the substantial offshore assembly investment undertaken by U.S. firms during the 1970s in the face of strong Japanese competition.
While one can progress reasonably far on the basis of this sort of cross-country analysis, it is apparent that in many cases an adequate explanation needs to invoke a much more specific set of factors. Consider, for example, the case of computer disk drives, in which Singapore has become the principal location for production by multinational subsidiaries (O'Connor, 1985b; UNCTC, 1987). This has occurred, according to O'Connor (1985b), for reasons associated with the presence in that country, because of its industrial history, of a well-developed machining industry. Television assembly, in contrast to semiconductor assembly, is space intensive, which could explain in part why Hong Kong and Singapore do not figure as centrally in this segment of the electronics industry as their more 'land-rich neighbours, Korea and Taiwan.... In addition, Hong Kong and Singapore lack domestic manufacturing capacity for cathode ray tubes (CRTs), which put them at a competitive disadvantage not only in TV production but in the production of computer terminals and monitors, since the CRTs used in such products are bulky and costly to transport' (ibid., p. 6 of notes and references).
What these examples suggest is the need for an approach to the study of the location of foreign electronics investment which can, on the one hand, specify in detail the requirements of each of the new technologies (for components, infrastructure, managerial skills, and so on) and, on the other, discern the extent to which different developing countries can satisfy these diverse requirements. Though clearly highly demanding of data and research input, such an approach is often likely to be highly rewarding. It enables one to bring out, for example, the relevance of the pattern of past industrial (and other) development in a developing country to the determination of its present comparative advantage (and, by extension, points to the way in which current patterns might impinge on its future endowment).4
Henderson (1989) made a systematic effort along these disaggregated lines to explain the changed (and changing) pattern of foreign investment in semiconductor production in East Asia. In particular, he suggests that a 'specifically East Asian division of labour has developed' (ibid., p. 55). This new pattern of foreign investment is essentially one in which 'more and more of the investment in assembly plants for large-batch standardised outputs tended to go to an increasing degree to Thailand, Malaysia, the Philippines, and so on, while Hong Kong, Singapore and now, seemingly Malaysia, have tended to be upgraded as to the quality and complexity of their production technologies and labour processes' (ibid., p. 59). Henderson also offered a very detailed explanation of these important changes, to which we will refer in later sections.
Local-market oriented foreign investment The model of the multinational firm which (as we shall see) underlies much of the literature on the impact of microelectronics on the Third World is the export-oriented subsidiary. This model applies best to the semiconductor industry, in which there is a preponderance of direct foreign investment that is heavily oriented towards production for export. Yet in many other sectors it is manufacturing subsidiaries with an orientation to the local, rather than the export market, that tend to dominate.5 In fact, 'Like investments by computer and consumer electronics TNCs [transnational corporations] in developing countries, those by telecommunications TNCs tend to be oriented largely toward local or regional markets' (O'Connor, 1985b, p. 37). Fragmentary data for Japanese MNEs operating in consumer electronics suggest that local-market-oriented investment may have comprised roughly 50% of total global investment in the mid-1970s (ESCAP, 1979). A recent study of the electronics industry in Brazil concludes that foreign computer firms were 'much more interested in Brazil's market than in the price of its labour' (Hewitt, 1988, p. 170).
At a general level, it is important to recognize that the location of local-market-oriented investment needs to be viewed, as Vernon (1973) has pointed out, from a different realm of economic theory than that which governs the export-oriented variety. Whereas 'the classical model provides a fairly good basis for the description of locational behaviour' for export-oriented investment, one may often need to look to oligopoly theory to explain patterns of domestic-market-oriented foreign investment. To illustrate this important point, consider the locational decisions prompted by the local substitution of imported consumer electronics goods in most of developing Asia after 1960. The sequence of events has been described as follows:
The erection of protective trade barriers compelled the foreign exporters to commence local production if their market was not to be lost. The Japanese in particular stood to lose as they had been very active in the export field. Moreover, there appears to have developed a 'bandwagon effect' whereby Japanese firms upon witnessing the relocation of their competition, perceived a threat to their market share and subsequently followed the move.
(ESCAP, 1979, p. 29, emphasis added)
It is plainly not standard comparative cost theory that lies behind this concentration of foreign investment for the local markets, which appears to be explicable instead in terms of a 'follow-the-leader' type of reaction function in oligopolistic markets (Vernon, 1973). That is, 'the decision of one firm in an international oligopoly to establish a new producing facility outside its home country often seems to trigger the establishment of similar facilities in that country by others'6 (ibid., p. 21).
Although very little is known about the inter-country distribution of local-market-oriented investment (which, as in the case of export-oriented production, will vary according to the specific fiscal and other attractions offered by particular developing countries), a few general observations can be made. The first is that insofar as economies of scale are a significant aspect of the new technologies, large [in the sense of having a high gross domestic product (GDP)] countries will be favoured. Since, for example, 'there are substantial economies of scale in the production of many pieces of telecommunications equipment, only those developing countries with the largest potential markets have attracted substantial investments' (O'Connor, 1985b, p. 37). A second observation is that adaptation of new technologies to serve particular user needs will tend to be profitable only if the size of the market is sufficiently large.7 It seems plausible, for example, to view the specialized development efforts that have been undertaken by computer MNEs to overcome the language processing requirements in Korea, China, and Taiwan (ibid.) in terms of the potential afforded by the combined market size of these countries. The last observation is that the market size for different products in a developing country is likely to depend on the pattern of income distribution, as well as on per capita income. When a given average income is evenly distributed, large markets are created for relatively simple goods (e.g., radios). But if income is highly unequally distributed, a small market is created for more sophisticated, 'high-income' products (such as VCRs).
Taken together, these general observations suggest a broad pattern of local-market-oriented investment that will tend to comprise individual countries with a relatively high GDP, a high per capita GDP, or groups of countries which possess one or both of these characteristics. Combined with the types of countries that appear to be favoured by export-oriented investments in microelectronics described above, several tentative generalizations about the overall distribution of foreign investment in the Third World may be offered. The first is that many countries (such as the Asian NICs, Mexico, parts of the Caribbean, etc.) will be favoured by both forms of foreign investment. Insofar as (is broadly the case) these are among the more affluent developing countries, an inegalitarian bias in the distribution of foreign investment within the Third World is created. And this bias is not generally likely to be offset by the relatively low incomes of countries that fall in only one or other of the two investment categories (though of course, there are such cases, such as India and the Philippines). On the contrary, the income levels of countries forming part of these categories may well be such as to contribute to, rather than ameliorate, the tendency to inequality.8 Though there are no data that can throw any conclusive light on these hypotheses, some very crude supportive evidence for the computer industry is given in Table 3.1. Though the classification of countries according to the stance that they adopt vis-à-vis MNEs ('assertive' versus 'accommodating') is somewhat questionable, the table is useful in showing that the total involvement of American computer MNEs in the Third World was concentrated among a narrow group of countries during the late 1970s.
Table 3.1 U.S. computer enterprises in upper-tier developing countries, late 1970s
Category of developing countrya |
Number of unitsb |
Percentage of units in all LDCs |
| Assertive | 12 | 40.0 |
| Accommodating | 11 | 36.6 |
aAssertive LDCs are defined to include India, Brazil, Mexico, Colombia, Indonesia, Nigeria, and Venezuela; accommodating LDCs are defined to include Hong Kong, Malaysia, the Philippines, Singapore, Taiwan, South Korea, and Thailand.bNo manufacturing subsidiary was operating at this time in Indonesia, Nigeria, or Venezuela.
Source: Grieco (1984, p. 65).
In particular, 76% of the total was invested in India, Brazil, Mexico, Colombia, Hong Kong, Malaysia, the Philippines, Singapore, Taiwan, South Korea, and Thailand.9 The predominance of relatively affluent segments of the Third World (namely, the groups of large Latin American countries and East Asian NICs) in this list of countries also bears on the arguments presented immediately above.
Another piece of evidence is provided by the global distribution of overseas subsidiaries of Japanese electronics firms. Of the total number of these subsidiaries, 46% are found in four Asian countries (Singapore, Korea, Taiwan, and Malaysia), while another 11% are located in Latin America (Humbert, 1988). When these tendencies towards concentration are combined with the finding that agglomeration economies appear to be very important to the location decisions of multinational firms, the likelihood of a future pattern of cumulative causation emerges (Wheeler and Mody, 1988). That is, 'newcomer' countries will find it increasingly difficult to attract foreign investment in microelectronics-related industries, while countries already in the location 'tournament' will become increasingly attractive.
Distinctions between export- and import-oriented investment An interesting and important question is whether any systematic differences can be found in the behaviour of subsidiaries engaged in production for export as opposed to those that are oriented to the local market. Unfortunately, however, only one study-of the consumer electronics industry in a group of developing Asian countries-has addressed this question systematically. On the basis of a sample of 63 firms drawn from Malaysia, Korea, Singapore, and Thailand, this study 'suggests the existence of a tendency for outward-looking foreign investments vis-à-vis domestic market-oriented businesses to have larger production volumes, a higher degree of product specialization, greater foreign equity participation, a lower percentage of domestic value-added and a higher proportion of expatriate management' (ESCAP, 1979, p. 50). In several respects, such as differences in ownership and value added, these findings match those of other studies that have compared the behaviour of inward- and outward-looking manufacturing subsidiaries (see, e.g., Reuber et al., 1973).
They are also confirmed in part by a more recent study of MNEs in the electronics industries of some of the same economies (UNCTC, 1987). But because they were not subjected to tests of statistical significance, the results of the Asian survey need to be treated with the corresponding degree of caution.
3.2.1.2 Transfer mechanisms not involving foreign investment
So far, our review of the factors that determine the availability of microelectronics technologies to developing countries has relied in effect upon two assumptions: (l) that some degree of monopoly control over these technologies is held by agents in the advanced countries and (2) that these agents choose to exploit their proprietary rights in the form of foreign direct investment in the Third World. The fact that these assumptions are not always applicable, however, gives rise to a set of alternative mechanisms through which the new technologies become available to certain developing countries. To begin, we will consider the circumstances in which, though proprietary rights in these technologies are held, they are exploited in forms that do not involve foreign direct investment.
Arm's-length licensing From the general literature on technology transfer, it is possible to identify the variables that seem to govern the firm's choice between establishing a foreign subsidiary and licensing its proprietary assets at arm's length. This literature suggests, first, that 'companies do contemplate foreign investment and licensing as direct alternatives, preferring foreign investment for its greater rent-extracting potential, turning to licensing only if that potential cannot be realized' (Caves, 1982, p. 204). In the semiconductor industry, for example, the preference of American firms for wholly owned subsidiaries in East Asia continues to be reflected in the predominance of this form of production in the region (Henderson, 1989, p. 55).
It is accordingly mostly to the circumstances that are relatively unconducive to foreign investment that one must look in predicting when licensing will occur. In particular, 'Arm's-length licensing is encouraged by risks to foreign investors and barriers to entry of subsidiaries, by short economic life of the knowledge asset, by simplicity of the technology, by high capital costs for the potential foreign investor, and by certain types of product market competition that favor reciprocal licensing' (Caves, 1982, p. 224).
Tigre's (1983) study of the computer industry in Brazil throws some degree of light on these propositions in the context of transfers of microelectronics technologies to the Third World. Among the foreign computer firms operating in that country in 1980, Tigre found a tendency for foreign investment to be associated mainly with very large firms. 'Smaller firms either licensed their technology or entered into joint ventures in order to compete in the Brazilian market' (Tigre, 1983, p. 143). His explanation of this finding is that while foreign ownership was the preferred mode of transfer for most firms 'only very large firms could actually afford to pursue this strategy' (ibid.). Underlying this distinction between actual and desired behaviour is the finding that several of the circumstances mentioned in the previous paragraph as being unfavourable for foreign investment (or favourable to licensing) bore in this case differently on firms of different size. In particular, these factors appeared to be so arranged as to make licensing the preferred option only for firms that were relatively small.
The absence of direct investment by all except large firms arose because 'medium-size firms usually lacked the managerial and financial resources to compete directly in overseas markets. A licensing agreement did not require any additional investment and helped the firms recover previous R&D expenditure' (ibid., p. 144). A second reason was that the smaller firms appeared to be less able than their very large counterparts to resist demands by the Brazilian authorities to share technology. (The same point is made by Evans and Tigre, 1989, in relation to 'super-mini' computers in Brazil.) Finally, the competitive forces to which the smaller firms were subjected, to an apparently greater degree, meant that Brazil obtained access to some of the new technologies in the computer industry relatively early in the product cycle and on terms other than through foreign investment. For example, in 1979 the Brazilian firm Flexidisk tried to obtain a licence from Shugart to manufacture floppy-disk drives in Brazil. Shugart refused on the grounds that its corporate policy was not to sell newly developed technology. Later, however, when Flexidisk was on the brink of signing a technology transfer agreement with another American manufacturer (Pertec), Shugart realized that its policy could cause it to lose the fast-growing Brazilian market. Thus it changed its policy and agreed to license floppy-disk drives to Flexidisk (Tigre, 1983, p. 165).10
Though there is very little hard evidence on the subject, one suspects that this last example constitutes an exception to the more general rule that licensed technology is available to developing countries only at fairly advanced stages of the product cycle.11
Clarke and Cable, for example, observed that
Japanese firms (even in the case of joint ventures) have displayed a reluctance to sell licenses for more advanced technology.... Japanese firms prefer to transfer know-how only for simple products such as black and white sets and simple colour portables, using older technology in tubes and semiconductors. ... Japanese producers have also been unwilling to provide Korean firms with technical know-how for new products such as VCRs and video discs.
(Clarke and Cable, 1982, p. 31 )
Much is likely to depend, however, on the prevalence of small firms which are innovative but unable to gain access to the resources required for direct investments. These firms will tend to license their innovations early, because there is no alternative way to exploit the market (Evans and Tigre, 1989).
Subcontracting Just as licensing arrangements may, under some circumstances, replace foreign investment as a mechanism of horizontal transfer of microelectronics technology (i.e., between firms producing the same line of goods), so too may subcontracting sometimes serve as a substitute for the multinational subsidiary in production that is vertically integrated (e.g., semiconductors). This latter phenomenon has been noted by several authors (Plesch 1979; ESCAP, 1979), and, not surprisingly, it seems to be facilitated by several of the same factors that were described above as tending to favour licensing over foreign investment. Plesch, for example, cites the following specific determining factors:12
• If a company's production ... is too small to justify offshore production and take advantage of possible economies of scale, subcontracting an independent LDC manufacturer would seem a more economical solution, especially so when it requires lower investment;13
• If a company's production volume is subject to fluctuations, subcontracting may be a short-term solution during peak production periods;
• Subcontracting may involve less risks than offshore investment;14 products or processes that are subcontracted are generally the most standardised and unsophisticated.
(Plesch, 1979, pp. 87-88)
The first two of these factors appear also to explain the emergence of a network of subcontracting relationships in the overwhelmingly foreign-owned semiconductor industry in East Asia. Such arrangements have emerged particularly in the Philippines and to a lesser extent in Korea (Henderson, 1989).
Loss of effective proprietary rights In the discussion above, the availability to developing countries of the new microelectronics-based innovations was shown to be circumscribed by the various mechanisms that owners of proprietary rights in the innovations choose to exploit them. But insofar as these rights may themselves become ineffective, further possibilities of access are opened up to developing countries. The replication in the Third World of microcomputers developed by MNEs is undoubtedly one of the most important manifestations of these possibilities.15
In part, this phenomenon appears to have resulted from the reliance of MNEs on outside sourcing which, by disclosing technical information to supplier firms, 'made the replication of standard computer architectures somewhat simpler' (O'Connor, 1985a, p. 320). But the loss of proprietary control over this technology also has to be viewed against the particular industrial structure in the developed countries from which it emerged. This structure was one in which the innovation of the microprocessor (the central element of the microcomputer) occurred in firms that happened to have no incentive to exploit the innovation as proprietary knowledge. Specifically, 'Because microprocessors were produced by semi-conductor companies, like Zilog, Intel and Motorola, which did not themselves produce computers, they were available on a merchandise basis rather than being incorporated in final demand products as proprietary technology'16 (Evans, 1985, p. 21, emphasis added). Countries with the relevant technological capabilities (such as Brazil, Taiwan, and Korea) were thereby enabled to use the microprocessor (purchased on the open market) as a basis upon which to imitate existing models, such as Apple.17 Indeed, 'It was the micro computer that really generated a dynamic set of locally-owned hardware producers in Brazil' (ibid., p. 21). More recently, Brazil and Korea have also used the availability of microprocessors on international markets as a basis for some of their attempts to promote an indigenous 'super-mini' computer industry (Evans and Tigre, 1989).
It bears emphasizing, however, that access by developing countries to the new technologies may often continue to depend on the behaviour of suppliers even when the technologies themselves are internationally available. In their study of the diffusion of flexible automation techniques, Edquist and Jacobsson (1988), for example, refer to this uncertainty as an important constraint on the adoption of these techniques in the newly industrializing countries. As they see it, the general problem is that 'in the fast-growing industries which are intensive in the use of service and/or application engineering, the suppliers may not be interested in sending scarce service or application engineers to marginal markets' (ibid., pp. 155-156).