Index of content

The essential difference between internal and external technology transfer is that in the case of internal technology transfer, there still will exist a considerable or total foreign control on the technology involved. In the case of external technology transfer, the foreign control has diminished or even disappeared and the technology can be spread freely in the host economy.

The concept of technology transfer is being used in different ways. Most scholars consider all possible technological impacts as technology transfer. Some restrict technology transfer to technology transmission and diffusion, others only refer to technology diffusion, and a few use the concept of technology transfer only in case of indirect or informal diffusion of technology transfer.

2.3.4. Summary

The presence and actions of foreign affiliates can have technological impacts on host economies in three different ways: by means of technology production, transmission and dissemination, together called technology transfer.

Now that we have defined the several channels of technology transfer through a foreign affiliate we will describe theoretical backgrounds and operationalise theses channels in the following sections. In the next section we will first compare the role of foreign affiliates with other sources of technology available to economies.

2.4. Some more aspects relating to technology transfer

2.4.1. Channels of international technology transfer

Technology transfer among nations and regions can and does take place through a number of different channels and mechanisms. To get an idea of the different channels, table 2.2 illustrates some commercial examples by nature of transferring and recipient entities. For reasons of simplicity, we will not take into consideration non-commercial channels of technology transfer, such as imitation of imported (capital) goods (or reversed engineering), international journals, etc.

Table 2.2: Examples of commercial channels for international transfer of technology

Source: Aggarwal, 1991, p64

As an example of foreign governments transferring technology to LDCs we want to mention the Japanese International Cupertino Agency (JICA). Japan is sending yearly more than 50 experts to Mexico in order to stimulate technology transfer and is training 150 Mexican technicians in Japan. Until 1994 1,256 experts have visited Mexico and 2,975 Mexican technicians have been trained in Japan. The costs of this exchange programme almost equals 5,300 millions of yens, which is roughly 50 millions of dollars. As an example of commercial institutes specialised in technology transfer we can refer to 'Spectrum International', whose mission is to provide products and services maximising on the technical and economic strengths of Mexico and the USA.

Of the different channels of technology transfer mentioned in table 2.2, MNEs are considered to be the dominant form of resource and technology transfer from developed to developing countries (e.g. Quinn, 1969; Dunning, 1993; Ostry & Gestrin, 1993, p18).

The benefits from MNEs as a source of technology transfer arise from the fact that MNCs are the main producers, and in many cases the only ones, of technology in manufacturing industries (Lall & Streeten, 1977, p69).

At the international level, the current state of technological knowledge and capacities is highly concentrated within the richer industrial economies, notably in Japan, Germany, United Kingdom, United States and Switzerland. Taken the world total as 100 percent, developed countries accounted in the 1980s for about 30 percent of the population, 90 per cent of the R&D personnel and 96 percent of the R&D expenditure. As a consequence, most countries must import the bulk of their technology (Gomulka, 1990, p192; Dunning, 1993, p288).

Furthermore, it is also well documented that technological creative resources are highly concentrated at the sectoral level among industries and at the industrial level among MNEs (Madeuf, 1984, p132). Since the production of technology is an expensive business, it tends to be concentrated among and controlled by MNEs located in a few developed economies, which can attract global markets to finance the necessary R&D. Therefore MNEs account for the great bulk of expenditure on knowledge-creating and skill-enhancing activities and dominate the cross-border trade in technology or technology-intensive products (Dunning, 1994, p288; Agmon, 1991, p91).

MNEs have proven to be the most efficient channels for the transfer of technology because they are able to internalise, within one organisational and decision-making framework, all the various means for moving information, people, and equipment across national boundaries (Rosegger, 1996, p289). Furthermore, through their centralised co-ordination, a TNC is better able to deal with the various externalities usually associated with the transfer of technology, especially learning externalities (Dunning, 1993; Tilton, 1971).

But why are MNEs involved in technology transfer?

MNEs possess certain ownership-specific advantages, which are property rights and/or intangible assets providing them competitive advantages over domestic enterprises in the host economy. For Such advantages can result from size, product diversity and learning experiences of firms (e.g. economies of scope and specialisation), to name but a few. The ability of MNEs to produce, acquire, master the understanding of and organise the use of technological assets across national boundaries is among the most important ownership-specific assets. "Partly this reflects other ownership-advantages of these companies, in particular size and experience, as well as their ability to tap the global market for scientific managerial and professional personnel, and to capture the cross-border economies of the joint governance of related technologies, or of the same technologies which may be differently used" (Dunning, 1993, p290).

The larger these ownership-specific advantages, the greater MNEs find it in their interest to exploit them from a foreign location, and the more they are likely to engage in outbound production. "The appropriation of the economic rents derived from those advantages is a principal motive for FDI" (UN, 1987, p177). In other words, MNEs want to make profits (or fulfil other corporate targets).

In literature reference is often made to the examples of "the transfers of the British companies at the turn of the century and, even more so, the Americans after the Second World War. These multinational firms were not simply exporting capital when they were developing their activities abroad: they also exported management capabilities and technology. That permitted the host countries, mainly Western Europe and Japan, to develop their technology and catch up little by little, in particular with the USA (Ghertman & Allen, 1982, p83; Ostry & Gestrin, 1993, p7).

2.4.2. Subsidiaries and other technology transfer mechanisms of MNEs

MNEs can be involved in technology transfers in different ways. A listing of these include the following: affiliates, joint ventures, licensing, franchising, management contracts, marketing contracts, technological service contracts, turnkey contracts and international sub-contracting. Of these, affiliates and joint ventures involve equity participation by MNEs, others generally represent arm's-length transactions between MNEs and independent enterprises in host countries. Each mechanism has its own costs and benefits (see Vernon & Wells, 1991, p22-27, Robinson, 1988, p39).

One important factor taken in consideration by MNEs is maintenance of control over technology, in many cases their raison-d'ètre. Traditionally, MNEs have had a strong preference for fully owned and to lesser extent majority-owned affiliates in developing countries to enable the internalisation of benefits from specialised technological know-how and capability, and to retain technology within the corporation. Since the 1980s other mechanisms, especially joint ventures, are gaining some importance (UN, 1987, p177), mainly because of improved international protection of properties and trademarks. But "even at present, it is generally considered that MNCs prefer 100 percent ownership of foreign affiliates. An important item of evidence for this statement is the fact that the majority of FDI is still undertaken through wholly-owned subsidiaries" (Martin, 1986, p69).

The decision process of the MNEs in which they choose between the mentioned mechanisms has been described by many scholars, but is out of scope of this study. In this study we depart from the point at which a foreign affiliate is already functioning and we will focus on the technological impacts of this affiliate.

"TNCs, through their foreign affiliates, serve as the primary channel of innovation from the developed to developing countries" (Ostry & Gestrin, 1993, p18). Given the complex nature of technology and technological knowledge, in particular its level of tacitness, foreign affiliates are well placed to take advantage of technological innovations developed elsewhere within a MNE. In particular, foreign affiliates can benefit from a MNE's technicians and researchers, the presence of complementary equipment, software, and internal communication networks, and from the availability of numerous secondary innovations that are complementary to the main innovation being adopted by the affiliates (Antonelli, 1991; TNCs, December 1995, p135).

In short, MNEs still prefer fully owned or majority owned subsidiaries in general. At the same time, potential technological impacts of affiliates are most probably larger than those of other mechanisms.

2.4.3. Some earlier thoughts on technology transfer

Technology transfer through foreign affiliates has resulted in much controversy. The large body of literature on this subject can be categorised into three distinct schools of thought: the orthodox, the radical and the pragmatic.

The orthodox school of thought (Mason, 1973; Helleiner, 1975; Dunning, 1972) proposes that because of their technological capabilities, combined with their managerial, marketing, and financial skills, the MNEs are best equipped to transfer technology through their affiliates and subsidiaries. Mason (1973) argued that well-managed MNEs could make substantial and unique contributions to the recipient country's growth. The orthodox school believes that restrictions and controls applied by developing countries to regulate the transfer of technology are generally harmful to their economies. Controls such as restrictions on foreign exchange and requirements for local training and purchasing will result in inefficiencies, loss of competitiveness in export markets, and higher costs of transferred technology. The orthodox school argues that appropriateness of technology transferred is limited by the size of domestic LDC markets and by lack of local management and technical skills.

Most of the preceding arguments are rejected by the radical school of thought (Stewart, 1974, Vaitsos, 1974), which contends that MNEs are mainly profit-motivated and will not devote any attention to the needs (social, economic, or other) of the recipient developing countries. Thus, MNEs tend to transfer capital-intensive or 'inappropriate' technologies and hardly carry out R&D in the host developing countries. They do not make any attempts to adapt their technology to the conditions in the host developing countries. On the other hand, the companies extract as much profit as they can engage in various unfair practices, such as restrictive clauses (Helleiner, 1975) and transfer pricing (Stewart, 1974; Vaitsos, 1974).

Pragmatists share some of the radical school's criticisms concerning the present system of technology transfer. However, they depart from the radical position when solutions and policies are concerned (Baranson, 1978, Balasubramaniam, 1973; Silvere, 1976). This school of thought proposes that developing countries need an appropriate mix of technologies; each should be transferred through channels that involve the least cost. MNEs may contribute to such better relations by appreciating the social, economic, and political problems of individual recipient countries (Abdullah, 1988, p3+4; UN/ST/CTC/11, 1979, p2-9).

Nowadays "the international business environment has changed fundamentally with a world-wide liberalisation of national FDI policies. A shift in policy emphasis has occurred from one focusing on firms' obligations and governments' rights to one emphasising firms' rights and governments' obligations" (Lan & Young, 1996, p60). This means a shift towards orthodox thinking, in which the MNE may play a dual role.

On the one hand, MNEs may serve to increase technological divergences between countries (in terms of the general capability of research centres as against other locations, and in terms of the specialisation of centres); but, on the other, it may decrease differences in technological capabilities of countries. Whatever the results of technology transfer, technological linkages between countries are increasing. "Whether countries are brought closer together or pushed farther apart, the interdependence between them is increased, and the development of technological systems or galaxies within the MNEs reinforces such interdependence between the innovation in different countries" (Cantwell & Dunning, 1991, p55+56).

It is against this background that the present study was undertaken.

Let us now take a look at the hypothesised extent, content characteristics and restrictions of intra-company flows of technology.

2.5. Foreign affiliates and technology production

2.5.1. Introduction

In this part we will focus on pre-R&D activities of foreign affiliates (adaptations) and on their pure R&D activities and their consequences for local technological capabilities.

2.5.2. Quasi R&D activities: product and process adaption

Adaptation may either take the form of modifications of the types of products (product adaption) or modification in the methods by which products are produced (process adaptations).

To what extent do MNEs help to provide host countries with the technological assets they need to upgrade their indigenous resources and capabilities? Put slightly differently: how far, in practice, are MNEs prepared to adapt their technologies to meet the particular requirements of host countries?

In theory the maximum contribution affiliates can make to indigenous resources and capabilities is to modify their technologies in such a way that they can be fully understood by host economies.

Main facts restricting a foreign affiliate to make maximum adaptations are (Lall & Streeten, 1977, p72+73):

1) Uncertainty

The technologies of MNEs have proven to be efficient (else the MNE would not have become a multinational), while the results of adaption are uncertain.

2) Higher costs (at least on short term)

Process, and to lesser extent, product adaptations require high costs.

3) Globally uniform product image and quality

"The appeal of their products is precisely that they are standardised goods with a reputation uniform quality (...) Other multinationals are concerned with maintaining product uniformity because the interchangeability of parts and equipment is critical for a servicing network (Frank, 1980, p74+75).

4) Higher risk of imitation

If adaption results in more simple technologies, the danger of having these technologies, often an (most) important competitive advantage, 'stolen' is increasing.

5) Labour issues

To minimise threats of labour problems affiliates prefer to rely on their relatively capital-intensive technologies.

6) Insufficient market size

Affiliates producing for the local market will adapt their products only if the market size permits recovery costs.

If affiliates are involved in product adaption, main reason will be to meet local market requirements. Where local market conditions, i.e. tastes or climate, require adaptations, such adaption are normally being made.

If affiliates are involved in process adaptations, this will be to make efficient use of the lower labour costs in LDCs.

Results of other studies show that MNEs make rather infrequent adaptations of technologies that they take to LDCs, the process technology being unchanged in 73 percent of all cases (Caves, 1982, p267+268) and that the amount of adaption increases with the subsidiary's age and experience. (Forsyth, 1972, p124-7).

Our expectations are that the majority of affiliates will be involved in any type of adaptations, necessary to meet the local requirements of the fairly large Mexican market. We will test whether market orientation is a significant influence or not, assuming that product adaptations are more often mentioned by affiliates the more they are producing for the Mexican market. Another firm-specific factor of importance is number of production years as a proxy for experience.

Because of the high costs involved and many uncertainties and risks we expect that less affiliates will report process adaptations than product adaptations. Furthermore, most product adaptations will be confined to simple modifications. The more complex adaptations require more contacts and/or support with the parent firm, who will be willing to assist if they have a higher interest in affiliate's efficiency and/or if the risks of technology leakages are small. Both conditions are present the higher the technological and financial dependence of the affiliate is. Majority-owned affiliates with low technological and/or financial freedom should thus be more involved in complex adaptations.

Assuming that machinery and equipment (M&E) obtained from parent companies requires more adaptations than M&E obtained otherwise, we expect that affiliates importing higher shares of their M&E from their parents will be more involved in process adaption.

A final important question is whether affiliates do adapt more or less than local firms.

In the situation of limited technology modifications affiliates can still attribute to the indigenous technological capacities if they are relatively more involved in adaptations than indigenous firms (Abdullah, 1988, p44). Lacking data restrain us to analyse this matter.

2.5.3. Pure R&D activities: experimental and basic research

Affiliates' contributions to the host economy are maximum in case they dedicate a large number of research personnel, all having the indigenous nationality, and capital to high quality research (i.e. basic research), resulting in high numbers of granted patents. To which extent do affiliates upgrade the innovatory capacities of host economies?

In literature the affiliate's involvement in R&D activities has received a lot of attention. As we have already seen, R&D activities are heavily concentrated within MNEs in developed economies (Dunning, 1993; Malecki, 1991, p200; Caves, 1982, p198-200).

In theory two forces, influencing the geographical location of R&D activities within MNEs are working at the same time: centralising and decentralising forces. Centralising forces favour concentration of R&D activities in a few developed economies, both within firms and on a global scale. Dunning (1993, p310) puts forward the following reasons for concentration of R&D:

1) the need for critical mass to gain economies of scale;

A minimal critical mass at R&D centres is required to allow for economies of scale, for the benefits of complementarities and externalities of inventive capacity within a centre and to facilitate the badly-needed communication and co-ordination among the centres (Safarian, p107, 1988). In LDCs the inadequate size of the market is one of the main reasons for affiliates not to execute R&D.

2) the presence of supporting industries and agglomerative economies;

3) availability of resources and capabilities (e.g. R&D facilities, skilled manpower)

"Manufacturing firms that do little or no R&D in Third World countries generally explained their position in terms of the lack of skilled local personnel" (Frank, 1980, p84).

4) the accumulated experience of R&D know-how and organisation of innovative, generating activities, and

5) the avoidance of cross-border communications and co-ordination problems and of leakages of core technologies.

Decentralising forces, or reasons that are in favour of the geographical spread of R&D activities within MNEs, are (Chen, p51+52, 1983):

1) the need to cater for local market needs

MNCs undertake R&D overseas principally because of the needs to respond to special design needs of overseas markets (Chen, 1983, p51; Marton, 1986, p281). It is advantageous to have R&D personnel to be in close touch with the local market requirements so that the type of R&D carried out can be more appropriate and effective. Related to local market needs is the expectation of the future market growth. In the beginnings of the 1980s Frank (1980, p83) concluded that manufacturing firms established applied R&D and product development facilities in several developing countries where there are large or rapidly growing markets. "Most of these institutions are located in Brazil and India; particular companies also mentioned Nigeria, Mexico, Saudi Arabia, and Argentina".

2) 'on the spot R&D desirable' (e.g. into tropical diseases, pesticides and new varieties of seeds and crops);

3) differences in local materials and need to test products locally (Frank, 1980, p83);

4) to satisfy host government pressures or regulatory instruments; or as part of a regional or global strategy,

5) to raise the quality of the output of at least some subsidiaries, and

6) to defend a competitive position in R&D-intensive sectors;

7) the need to be where there are clusters of cutting-edge R&D activity;

8) the need to acquire new technological assets or specialised skills and talents;

9) to scan and monitor R&D activities of foreign firms, and

10) to take advantage of differences in cross-border location-bound resources and capabilities, and markets.

The last four mentioned reasons, playing an important role in regions such as of Silicon Valley, apply more to developed countries than to LDCs.

In practice the centralising forces seem to have been dominating the decentralising forces, which is especially true for LDCs. "Most studies lend support to the hypothesis that affiliates of TNCs do little R&D work in host developing countries" (UN, 1986, p39). Marton (1986, p64), for example, found in her analysis of industrial transfers that "despite the fact that the costs of research are much lower in countries such as Brazil, Mexico, or India compared to the United States and Western Europe, MNCs have not made significant investments in R&D in developing countries so far".

Other findings, such as Lan and Young's (1996, p67) in China indicate that 11% of the affiliates execute R&D.

The first question to be answered is whether an affiliate is involved in R&D or not. Affiliates fulfilling the following conditions will be classified as executing R&D activities:

1) The affiliate has to employ research personnel, defined as employees "engaged in the conception or creation of new knowledge, products, processes, methods and systems" (OECD, 1981).

2) The affiliate has to dedicate a certain part of its annual sales to R&D activities.

3) The affiliate has to perform its R&D activities in a separate department.

4) The affiliate has to be constantly involved in applied or basic research or experimental development.

We expect that only a small minority of affiliates will be involved in R&D activities.

Many firms not involved in R&D are restricted to retain themselves from R&D by their parent companies. In a study of Young, Hood and Hamill (1985; cited in Dunning, 1993, p227) 49% of the interviewed subsidiaries felt that the headquarter influence on decisions concerning R&D are strong or decisive. In the second place, we expect that lack of qualitative personnel will be of main importance.

If affiliates do have a R&D unit in their establishment, run by own R&D personnel, a second question is in which R&D activities affiliates are involved.

Two types of R&D activities are experimental research and basic/fundamental research.

The primary objective of experimental research is to improve or develop products and processes tailored to meet local needs. "R&D of this type both depends on, and affects, the local technological capacity (...) Moreover, except in a few sectors, notably food processing, textiles and clothing and pharmaceuticals (...) and in a few countries, such as Brazil, India, Korea, Mexico, Singapore and Taiwan, it is the only R&D carried out in developing countries" (Dunning, 1993, p307). This type of R&D is more of a know-why than a know-how variety. It is increasing the innovatory capacities of host countries, especially if local research personnel perform improvements and developments.

Basic research is most likely to be located in the MNE's home country. "Where this research is undertaken abroad, it is likely to be of two main kinds. The first is where the inputs required for the research are themselves immobile (e.g. tea plantations, bauxite mining, a particular climate of ecological condition, quality improvement techniques for agribusiness). The second is that the need for continual testing and interaction with customer needs requires a local R&D facility. (...) This latter type of research is likely to make more demands on the local innovatory infrastructure. It is most likely to flourish where it is part of a cluster of similar R&D activities and/or has access to university or co-operative research facilities. At the same time, MNE activity might itself influence, for good or bad, the development of innovatory clusters" (Dunning, 1993, p308).

On the one hand, this type of research is considered to have positive influence on the host economy since it is increasing the technological capabilities of the latter. On the other hand, it might also have negative effects, such as substituting for or even restricting local research projects. Also much is depending on the way of entry and the level of competitiveness of local firms operating in the same industry.

However, we will not be able to conclude whether the net effects will be positive or negative. Therefore a dynamic analysis, also including local firms as an explicit study object, is requested with a stronger accent on subsectoral developments.

Being aware of its limitations, we will consider as indications of positive R&D influences the way of establishment (being other than by take-over) and ownership (being other than fully-owned).

It is perfectly clear from a variety of studies (e.g. Casson, 1991) that the great majority of R&D undertaken by the foreign affiliates of MNEs is directed to the improvement of products and processes rather than to basic or fundamental research. For example, in a survey of the R&D activities by 108 foreign subsidiaries in Greece in 1981, Dokopoulou (1985) found that only nine (8%) undertook basic research (Dunning, 1993, p306).

As a consequence of performing R&D activities mainly to cater for local market needs, we expect that MNCs' overseas R&D activities tend to focus on development rather than research, and on product and process improvements rather than on new products and processes.

Finally we will analyse the intensity of R&D activities.

Three indicators will be used: number of R&D personnel, R&D costs, and number of patents.

Using the first two indicators is based on the assumption that more input factors of R&D (operationalised as relative number of researchers and R&D costs) means more output of R&D (or the level of R&D).

In other studies low percentages of R&D were found. For example, in South Korea, foreign affiliates spent 1.8% of their sales on R&D in 1980 (Stoneman, 1989, cited in Dunning, 1993, p304). US data reveals that, in 1994, US manufacturing MNEs spent 4.0% of their sales on R&D, whereas their foreign affiliates spent only 2.4%. (US Department of Commerce, December 1996, p26-32).

However, the data also shows that European-based MNEs generally spend more on foreign-based R&D activities than do their US counterparts (Dunning & Pearce, 1985). "When compared to American MNEs, the inventive activity of European MNEs is more decentralised, while Japanese inventive capacity is very centralised (...) The greater decentralisation among European MNEs is explained by the proximity of large foreign markets, the attraction of the large American market and the relatively small size of their home market. On the contrary, the greater - if not complete - centralisation of R&D activity of Japanese MNEs is linked to their recent internationalisation, their low foreign production involvement and low domestic R&D costs" (Safarian, 1988, p107).

Another factor influencing the extent of R&D is time. Many studies such as Cantwell (1991), Pearce (1990) and Warrant (1991) have argued that R&D tends to follow behind the establishment of manufacturing abroad but with a time-lag (Hagendoorn, 1995, p50).

A final indicator, measuring the level of R&D in terms of its output, is the number of patents granted to foreign affiliates.

The third issue to be addressed is the level of affiliate's R&D costs in comparison with those of local firms. Affiliates' R&D activities might be low when expressed as percentages of annual sales, but if these percentages are still higher than those of local firms operating in the same sector, one can say that affiliates create more technology than local firms. In that way foreign affiliates can still have a considerable impact on the host country's innovatory capacities.

2.6. Foreign affiliates and technology transmission (internal technology transfer)

2.6.1. Introduction

Internal technology transfer involves intra-firm flows of technology from the parent to the affiliate. In theory technology transmission is optimal if all affiliates introduce new technologies in the host economy by importing state of the art technology from their foreign partners.

We will find out to which extent this is realised in practice and which channels are most complete.

We will end this section with an overall view of the several types and characteristics of internal technology transfer.

2.6.2. Internal transfer of hardware technology

We treat the issue of internally transferred hardware technology under three headings: machinery and equipment (M&E), patents and licensing.

We expect the majority of affiliates' M&E to be imported from the mother company, as M&E are essential parts of the (more advanced) production process of the parent company. Reasons that can explain lower percentages of M&E inputs from the parent company are differences in industry, in products, in scale of production and in labour costs. Closely related with this issue is to which extent adaptations are made to local circumstances. More attention to this issue is paid in section 2.5.2.

According to Kokko & Blomström (1995, p460) affiliates in LDCs prefer to imports of capital equipment before imports of disembodied or software technologies. An explanation for this preference may be "that technologies embodied in M&E require less human capital and technological skills, i.e. assets that are scarce in developing countries".

Other types of hardware technology are patents and licenses. "Most licenses are granted to subsidiaries and affiliates and to other multinational firms. The smallest share of licences goes to local firms in developing countries" (Malecki, 1991, p100).

The extent of hardware technology obtained from the parent company is a function of autonomy of the foreign affiliate, local integration and other affiliate's characteristics.

We expect a negative relationship between the extent of autonomy and hardware transfers. The lower the autonomy of the affiliate, the more the MNE will be prepared to share its competitive advantages, for two reasons. First of all, the majority-owned affiliate will have less room to choose its own technology in comparison with minority-owned affiliates and, secondly, the willingness of MNEs to transfer their latest technologies to their affiliates will be larger in the case of fully-owed (or majority-owned) affiliates because the risk of spread of these technologies will be lower. Autonomy will be operationalised by percentage of foreign ownership and freedom of control in the fields of finance and technology.

We expect a negative relationship between the extent of local integration and hardware transfers. The more an affiliate is externally orientated, the more the affiliate will be inclined to import hardware technologies from its parent company in order to maintain a constant product quality. Local integration will be operationalised by percentage of inputs obtained from Nuevo Leon, from Mexico and from North America (NAFTA countries), by percentage of exports and area of sales.

2.6.3. Internal transfer of software technologies

One of the key ownership-specific advantages that MNEs enjoy, and can utilise in the countries in which they produce, is their ability to train and upgrade human resources. They derive this advantage, in part at least, from their cross-border experiences in manpower management. The extent and pattern of human resource management by MNEs in Mexico is likely to depend upon its philosophy and general strategy towards its foreign operations, the length of its involvement in Mexico, the nature of activities in which it is engaged, the availability and quality of local training institutions, its market share and the role played by government in promoting in-house or other kinds of training programmes.

This internal technology transfer becomes an external one in case the trained employees quit working in the foreign affiliate and start to work for themselves or for indigenous firms.

Other ways of internal software technology transfer are maintenance and repair of M&R, management and marketing know-how.

We expect the same relations between extent of software transfers and affiliate's characteristics as we expected with hardware transfers.

Lan & Young (1996) found that most foreign affiliates in China (70%) were involved in transfer of hardware technology. More than half of the affiliates was also involved in training (60%) and other channels of software technologies (50%).

Mataloni (1993, p40) concluded that "foreign affiliates remained far less technology-intensive than U.S. parents".

2.7. Foreign affiliates and technology dissemination (external technology transfer)

2.7.1. Introduction

In theory the impact of foreign affiliates is at its maximum if all affiliates diffuse their technologies to local firms. We will examine to which extent this is found in practice and pay attention to the main restricting factors. Although in literature great emphasis is laid upon informal or indirect transfers, we believe that formal ones can have a positive contribution on the host country's technological capabilities as well.

2.7.2. Direct ways of external technology transfer

Mechanisms of formal or direct external technology transfer are sale and lease of M&E to indigenous firms, maintenance and repair of indigenous firm's M&E, training of their personnel, assistance of management and/or marketing know-how and the sale of patents. As demonstrated in figure 2.3 (section 2.8) local firms must have the technological capabilities to adapt the transferred technology and affiliates must be involved in technology transfer. If the affiliate is not involved in any of the above mentioned activities, there will be no external technology transfer.

We will begin with giving an impression of the technologies transferred from affiliates to local firms. We expect that M&E will be the most transferred technology, since M&E transfers require less human capital and technological skills than disembodied technology transfers.

Most frequently mentioned restrictions will probably be fear of loss over control over technology and lacking indigenous skills to use the M&E.

2.7.3. Informal/indirect ways of external TT

Diffusion, or external technology transfer, can "result in indirect productivity gains for the host developing countries, through the realisation of external economies. Generally these benefits are referred as spillovers, which indicates the importance of the way in which the influence is transmitted. There are several ways in which these spillovers may occur. Presumably the most important channel is via competition. Existing inefficient local firms may be forced by the competition of foreigners to make themselves more productive by investing in physical or human capital or importing new technology" (Blomström & Wolff, 1994, p264). "Even if their is no direct transfer of knowledge, innovations or introduction of new techniques and products by foreign firms may force local firms to adopt more efficient techniques or to ensure more effective quality control procedures. Finally, perhaps by demonstration effects, the success and profitability of manufacturing activities by foreign firms and joint ventures in Malaysia have induced local entrepreneurs to invest in manufacturing. In this sense, foreign investors have indirectly motivated the local entrepreneurs" (Abdullah, 1988, p42).

Another source of gain is the training of labour and management provided by the multinationals, which may then become available to the economy in general. Since such resources are in a short supply in developing countries, this type of spillover efficiency is expected to be more important here. Especially in Singapore, but also in other Asian economies "several local engineers and technicians working for MNCs have acquired sufficient skills and expertise to set up their own enterprises in supporting industries to supply parts and components to MNCs. Such transfers of expertise have played an important part in building up a pool of indigenous supporting industries in Singapore" (Natarajan & Tan Juay Miang, 1992, p51). Lan & Young (1996, p73) found in China improvements in management skills and a stimulus to entrepreneurship. Some illustrations existed of personnel moving back to state enterprises and helping to promote reform, and of local personnel establishing new enterprises based on market and other knowledge obtained in foreign affiliates.

"A third potential resource of spillover efficiency benefits is through the impact made by the foreign subsidiaries in the host economies on their local suppliers, by insisting that they meet standards of quality control, delivery dates, prices, and the like" (Blomström & Wolff, 1994, p264+265).

Other mechanisms are the imitation of products and/or production processes and 'abuse' of patents, licences and technology contracts by indigenous firms.

Most external technology transfer is likely to occur between firms along the same value-added chain (e.g. to suppliers or industrial consumers) or across value-added chains, e.g. to competitors or non-competing firms in the same locality (Dunning, 1993, p289).

On the basis of a cross-section study of Mexican manufacturing, Kokko demonstrated that positive spillovers are less likely in industries where high foreign market shares and large productivity gaps between foreign and local firms coincide (Kokko et al, 1996, p603+604). In his regression analysis of Uruguayan manufacturing plants in 1988 Kokko et al found no signs of spillovers. "However, looking separately at two sub-samples of plants characterised by the size of the technology gap between the locally-owned plants and it foreign-owned competitors, we found significant differences in the impact of foreign presence. Spillovers appeared to be positive and statistically significant in the subsample of plants with moderate technology gaps vis-à-vis foreign firms, but not in the group of local plants facing large technology gaps" (Kokko et al, 1996, p609).

Finally we will look at the extent of local contacts for affiliates executing R&D activities. While foreign affiliates serve as the primary channel of innovation from the developed to developing countries, the welfare gains to the latter of adopting new technologies depend upon the extent to which these innovations are diffused locally (Ostry & Gestrin, 1993, p18). To the extent that subsidiaries are engaged in local R&D or product development, they maintain very limited linkages with local technical and scientific institutions, mostly for confidentiality reasons (Nadal, 1997; Ostry & Gestrin, 1993, p22). This leads to the expectation that affiliates are performing R&D on their own instead of in co-operation with local research institutes or local firms.

2.8. Local technological capabilities and technology gaps

2.8.1 Introduction

Until now we have been discussing modes of technology transfer and possible influencing firm characteristics. However, local technological capabilities are at least as important as the already treated aspects of technology transfer. Therefore in this section we will touch briefly upon the complicated relationships between the affiliate's technological impact and local technological capabilities, or host country aspects.

In the next section we will look at the role of foreign affiliates in producing and transmitting new technology to LDCs and at the role of local technological capabilities.

Then we will elaborate briefly on affiliates as disseminators of technology and local technological capabilities.

2.8.2. Local technological capabilities and new technologies of foreign affiliates

Between the technological impact of foreign affiliates and indigenous technological capabilities exist complicated inter-dependent relationships. Indigenous technological capabilities are influencing to a large extent the R&D activities and quality of internal technology transfers.

The optimising affiliate will import the technology only to the point where marginal revenues equal marginal costs. Figure 2.2 illustrates the affiliate's optimal technology import decision. The curve TC indicates that the transfer costs will be increasing with the complexity of technology. The curve TR shows that the total revenues are an increasing function of the technology level of the foreign affiliate, such as the slope of the curve shows that the marginal revenues of new technologies are decreasing but always positive.

In figure 2.2 the optimal import quantity is indicated by Q1. Here the slopes of the curves TC and TR are equal and the distance between them is maximised. In the case of competitiveness, where local firms start to invest, the technology gap narrows and the affiliates' profits are decreasing because indigenous firms have captured a share of the market. In this case the curve of total revenues moves to the right in the position of TR' . The affiliate will increase the technology imports until the new equilibrium Q2, where the two curves TC and TR' are equal.

The implication is that the larger the technological capabilities of the host economy, the more an affiliate will be inclined to perform high-quality R&D and introduce state of the art technology.

To conclude, the technological impact of foreign affiliates will be larger the more they are forced by local firms to use latest technologies available.

Figure 2.2 The influence of indigenous technological capabilities on behaviour of affiliates

Source: Blomström, Kokko & Zejan, 1992

2.8.3. Local technological capabilities and diffusion of the affiliates' technologies

For a technology transfer or diffusion to take place, three conditions need to be fulfilled:

1) There must be differences in technological capabilities between those of the foreign affiliates and those of local firms.

2) The recipient (indigenous firm) must have sufficient technological capabilities in order to be able to absorb the transferred technology.

3) The technological supplier (foreign affiliate) must be transferring technology to the recipient.

A technology gap can be defined as the differences in technological capabilities between those of the foreign affiliates and those of local firms. If the MNE and local firms were operating at the same technological level, no technology would be transferred.

"For a technology transfer to occur, there must be a technology gap among firms or the countries. Theoretically, free trade in technology should close the gap and equally endow all countries. The scientific and technical resources needed to develop new theories are heavily concentrated, however, as are the engineering resources required to develop new products or processes and to carry out essential modifications in the new product after its embryonic commercial introduction" (in Agmon, p91, 1991).

As a result of the current emphasis on firms'' rights and governments' obligations in literature the importance of developing local technological capabilities is stressed. Why?

If technology is defined as human knowledge applied in production it follows that "the mere transmission of information about techniques is a necessary but not a sufficient condition for technology transfer (...) But knowing about technology is not the same as using it" (Rosegger, 1996, p286).

Why are technological capabilities of the technology receivers so important?

Where the technological capacity of a host country is weak in the sector concerned, the investment of MNEs may reduce local technological capability (Cantwell & Dunning, 1983, p52).

Furthermore, the lack of technological capabilities in less developed countries not only adds to the difficulties of assimilating the imported technology, but will also lead to perpetual dependence on technology of developed countries.

"If technology is firm-specific information and is cumulative in nature, technical change is a continuum in the sense that present-day technical change is based largely on previous changes and is a link in a long chain of continuous technical change (...) These characteristics of technical change present great difficulties for technology importing countries. As mentioned before, to transfer an advanced technology is actually to introduce a discontinuity into the production technology of the importing country (...) However, the real problem emerges after the transfer is completed. The problem is caused by the fact that the technology exporter still keeps on carrying out R&D activities, accumulating information and experience, and strengthening expertise continuously. On a macro scale, the world technology frontier is moving ahead constantly, possibly at an accelerating rate. Unless technology importers move rapidly to generate a corresponding rate of technical change, the end result is similar to the technology importer jumping on a fast-moving train to immediately jump off" (Liu, 1995, p110). So, what the firm can hope to do technologically in the future is heavily constrained by what it was capable of doing in the past (OECD, TEP, 1991, p63+64).

The relationship between local capability building and technology imports is partly complementary and partly competitive. Inputs of knowledge from technologically more advanced countries are clearly necessary to launch local efforts. But too passive a dependence on imports of know-how can stifle the development of more advanced local technological capabilities, to improve technologies, design new products and develop new processes.

This can have three deleterious effects (Lall, 1993, p102):

1) First the cost of absorbing and deploying a technology is higher if local capabilities are weak.

2) Second, the mastery, adaption and further development of a technology, and its beneficial externalities, are reduced if locals are unable to undertake a number of technological tasks.

3) Third, the 'quality' of FDI cannot be raised if the local economy is not able to offer higher levels of complementary skills and technological effort.

This can be summarised as is done in table 2.3. In the first quadrant, MNEs transfer technology but LDCs do not have the absorptive capacity. Under these circumstances MNEs will enjoy their technological monopoly power and LDCs will suffer from the syndromes of technological dependency. An example of technology being transferred without being received is a turnkey project where the donor builds an entire factory but no one in the recipient country knows how to replicate it.

Neither MNE nor LDC gain in the second quadrant, in which the MNE is not willing to transfer its technology to the LDC and, at the same time, the LDC is lacking the capability to benefit from existing technology used by the MNE.

In the other two quadrants the LDC does possess absorptive capacity. The MNE in the third quadrant eventually will take advantage of opportunities to expand their market through FDI. An example of this situation, characterised by receiving technology without being transferred, is the case in which the engineers in the recipient country create a new product through reverse engineering.

Then we arrive in the fourth quadrant. Here LDCs will not end up being technologically dependent on MNEs. Here, internationally technology transfers benefit both the MNE and LDC.

Table 2.3: Interaction between suppliers and recipients in technology transfer

Source: Kim, in Agmon & Von Glinow, 1991

"Even though Japan has been a large-scale importer of technology, it is important to note that in her case the process has involved not only imitation, but also creative adaption which is, in effect, innovation" (T. Ozawa in Wei, 1995, p114). "A policy conclusion of these findings is that host country governments may choose to support local investment, competition, and education rather than to rely on controls and direct supervision of FDI to secure inflows of technology to the affiliates of foreign MNCs" (Kokko & Blomström, 1995, p466).

"The convergence hypothesis asserts that when the productivity level of one (or several) country (ies) is substantially superior to that of a number of other economies, largely as a result of differences in their production techniques, those laggard countries that are not too far behind the leaders will be in a position to embark on a catch-up process. This catch-up process will continue as long as the economies that are approaching the leader's performance continue to be able to learn from the leader. But as the distance between the two groups narrows, the stock of knowledge unabsorbed by the laggards will grow smaller and approach exhaustion. The catch-up process then usually terminates unless some supplementary and unrelated influence fortuitously comes into play. Meanwhile, those countries that are so far behind the leaders that it is impractical for them to participate in the convergence process at all, and so many such economies will find themselves falling even further behind. The most important influence underlying this hypothesis is the transfer of technology that constantly takes place among economies" (Blomström & Wolff, 1994, p274).

Summarising, in this section we have seen that three conditions need to be fulfilled for a successful technology transfer to occur. First of all, there have to be technological differences between foreign affiliates and local firms. Secondly, the host economy needs to have certain technological capabilities. If these capabilities do not exist or are too small, there will be no transfer of technology. LDCs can not catch-up by simply importing technology. If they want to develop their technological capabilities by importing technology they also need to adapt and improve this technology in an active way.

Thirdly, the MNE has to (be willing to) transfer a technology.

2.8.3. Operationalisation of indigenous technological capabilities

We will operationalise the concept of technological capabilities in three ways: on an international level by comparing several measures for input and output of scientific and technological activities in Mexico with those of other OECD countries, on state level by comparing roughly the same measures for Nuevo Leon with those of other states, and on basis of our questionnaire.

On the international level we will use OECD data on several measures of national scientific and technological activities. By using these data, collected according to internationally agreed definitions as registered in the so called Frascati-Manual, we avoid to a high extent the problems one usually confronts when making international comparison with national defined data. One severe limitation is that OECD data for Mexico are only available for 1993 making a dynamic analysis impossible.

Input indicators, such as R&D expenditure and R&D personnel give an idea of the resources devoted to develop national technological capabilities.

"As yet, there are no direct measures of the output of scientific and technological activities, only proxy indicators based on data originally collected for other purposes. Three such types of data are presented in this research: patents, the technology balance of payments and trade in highly R&D intensive industries. Unfortunately, for Mexico only data for patents are available. These patent data can be considered as partial proxy measures of the output of R&D in the form of inventions (OECD 1995-2, p4+5).

The use of these indicators gives an overall view of Mexico's level of technological capabilities (Bhalla & Fluitman, 1985, p182+183; Dunning, 1993, p300+301). We expect large differences between the technological capabilities of Mexico and those of other OECD countries , indicating a technology gap.

However, whether a technology gap exists or not, we have all reasons to believe there are large national differences in technological capabilities of Mexico. Those of Nuevo Leon are expected to be much higher than those of Mexico as a whole. Because of lacking data on state level, we will include data collected by surveying affiliates.

We will classify the affiliates on the basis of the following two criteria:

affiliate's technological level of 1991 and of 1996 in comparison with that of local firms, and,

explicit reference to the existence of a technology gap between foreign and local firms and the affiliate's technology capacity must possess a too complicated nature to transfer.

2.9. Costs of TT

"Technology transfer almost always involves resource costs. These are all the costs that must be incurred to obtain a foreign technology and to ensure its effective utilisation. Typically, they comprise royalty fees and the cost of transmitting and absorbing all the relevant know-how, which includes the peripheral information relating to plant design and construction, equipment installation, organisation and operation of production, quality control, modifying the technology, and training labour" (Gomulka, 1990, p197).

Teece (1977) gives a detailed discussion of these costs among others. In his case study of 26 investment projects involving international technology transfer, the total transfer costs were found to average 19 percent of total project costs, with the percentage ranging from two to 59 percent. Teece identified four variables as particularly important in influencing the size of the transfer costs:

1) the extent to which the technology is understood by the transferee;

2) the age of the technology;

3) the number of firms utilising the same or similar technology, and

4) the cultural and systematic characteristics of the importing country.

His econometric estimates imply that "transfer costs decline as the number of firms with identical or similar and competitive technology increases, and as the experience of the transferee increases" (Teece, 1977, p253; cited in Gomulka, 1990, p197+198).

Chapter 8: Technology in practise

8.1. Introduction

In this chapter we will analyse the economic impact of foreign affiliates on the technological capabilities of Nuevo Leon. First we briefly describe the institutional context (8.2). In the following sections we will look at the technology production of affiliates (8.3), technology transmission through foreign affiliates (8.4), and technology diffusion (8.5)

8.2. Technology policies of Mexico

8.2.1. Science federal institutional framework

"After the war and for four decades, Mexican industrial development was defined by an import-substitution model. This resulted in an inward-looking strategy that favoured protection and regulation over open markets and foreign investment inflows unaccompanied by technology transfer and industrial modernisation. These foreign investment inflows were dependent on high barriers to external modernisation" (OECD, Stpol, 1994, p119). Economic growth was based upon low labour costs and government protection rather than upon efficient production and technological attention. This explains why Mexican industry has very few research centres. It had no need to exploit proprietary technologies to compete: the manufacturing of undifferentiated goods only called for process innovation, which could be easily acquired in the form of embodied technology in plant and equipment" (OECD, 1994a, p165).

By the mid-1980s, the Mexican government policy changed dramatically, towards an open and deregulated economy. Domestic industry, therefore, is now exposed to competition both at home and abroad. "In the short-run, success will probably depend on the capacity of the industrial sector to become efficient and to adapt and diffuse modern technology. In the long run, however, the ability to succeed, for the companies and the country as a whole, will depend upon the capacity to innovate. The capacity to innovate, in turn, is a function of R&D facilities, highly trained personnel, investment and productive sector linkages to basic and applied science (...)

The move towards technological modernisation has just begun. It will take years for technology and innovation to become an integral part of company activities. However, the increasing number of large companies with their own R&D centres - e.g. Vitro, Alfa, Condumex and Cydsa - offer some indication of change" (OECD, 1994a, p120).

8.2.2. The current Mexican technology policy

The most important laws on technology have been the 'Ley para Coordinar y Promover el Desarrollo Científico y Tecnológico' (Science and Technology Development Act), promulgated in 1984, and the 'Ley que Crea el Consejo Nacional de Ciencia y Tecnología' (the law creating the National Council for Science and Technology, Conacyt), promulgated in 1970. Conacyt is the primary agency responsible for defining and implementing science and technology policy. It distributes the allocated budgetary resources among its different programmes, provides incentives to increase R&D activities and establishes networks with international institutions to identify new opportunities for international co-operation in R&D and technology transfer.

Conacyt has entered a period of profound change. It has reduced its personnel by one-third and moved to significantly smaller premises, which are better equipped with a modern computer network system. In this way, Conacyt has freed resources to strengthen R&D financing capabilities. Thus, in 1991, 20 percent of the budget went to administrative expenses; in 1992, the proportion decreased to 8 percent" (OECD, 1994a, p37+46). Still Conacyt is being criticised because of its complex and bureaucratic procedures and its strong emphasis on science (especially on scholarships) at the expense of assistance to Mexican enterprises. As a result, only five percent of its budget is spent on the latter (interviews with Conacyt and Caintra).

"In May 1995, the Mexican government introduced the National Development Plan (1995-2000) which maintains Mexico's commitment to modernise its industrial sector and achieve economic growth in part by reducing regulatory burdens on the private sector" (OECD, 1996, p160).

Its main purpose is to achieve a significant progress in the national growth of science and technology in order to arrive at the level, both in quality and results, of developed countries (SEP/Conacyt, 1995, p6). Current limitations to achieve this hoal are "the lack of sufficient infrastructure, quality defects and the cost of basic inputs, technological shortcomings and the existence of burdensome regulations and inequitable competition" (Poder Ejecutivo Federal, 1995, p43).

"Technological policy should be conductive to making more efficient and effective use of the country's limited resources in the endeavour to attain greater growth and raise the general level of well-being in a manner compatible with environment protection. It should contribute to phasing out dual technologies and technological dispersion among the productive sectors and to closing the technological gap between domestic companies and their foreign competitors" (Poder Ejecutivo Federal, 1995, p43).

8.2.3. Tax policy

The current Income Tax Law provides for deductions of up to 1 percent of total sales for investments in R&D. This law also permits accelerated depreciation of equipment, at a rate of up to 35 percent, when it is linked to domestic product and process R&D" (OECD, 1994a, p129). Existing schemes in favour of industrial R&D appear to be insufficient given the challenges faced by Mexico. Concerning tax concessions for R&D and technological modernisation, there seems to be a real need to overhaul the system and adapt it to the needs of technological modernisation (OECD, 1994a, p167).

"Tax legislation was passed in 1995 that allows for deduction of costs of employee training programs equal to up to 1 percent of revenues for the corresponding period" (OECD, 1996, p161+162).

8.2.4. The evolution of Mexico's transfer of technology policies

Mexico's industrialisation has made extensive use of imported technology to establish new production capabilities. Licenses and other forms of technology transfer have fulfilled a demand for technology that could not have been met immediately by local technological capabilities. However, during the late 1960s, the government became aware of several negative effects of technology imports. In particular, attention was drawn to the impact if royalties and fees on the balance of payments and of the imposition of restrictive clauses by technology suppliers. The need for government regulation of technology was also justified on general considerations, such as the need to develop local technological capabilities. Hence, in 1973, Mexico initiated the regulation of technology transfer by passing the Technology Transfer Law (TTL) and establishing the National Register for the Transfer of Technology. The objective was to help recipients of foreign technology to strengthen their bargaining position vis-à-vis technology suppliers.

In 1982, the regulation of technology transfer and the technology transfer law were modified, putting more emphasis on the need to develop local technological capabilities, to be assured by imposing specific programmes for each enterprise receiving technology by way of a technology transfer contract. The number of programmes agreed upon with recipients of foreign technology (both nationals and affiliates of MNCs) amounted to 1,166 from 1983 to 1988. Technical agreements accounted for 72 percent of the total number of agreements, and the most frequent concern was the assimilation of foreign technology (33%). Only thirteen percent of the agreements were dedicated to R&D. Peres Nuñez (1986, p67) found out that none of the interviewed foreign companies felt restricted by the commitments resulting from the TTL. Nor did this law bring about major product development or research that would not have been done anyway.

Technological change has accelerated in recent years and the pace of technological progress is now understood to be of crucial importance for international trade and competitiveness. The government decided to change its role from a regulator of technology transfer to a promoter of technology transfer. Accordingly, in 1989, the government announced that laws would be issued to give technical and administrative facilities for enterprises to effectively acquire technology in the international market, which supposedly would involve a reorientation of the functions of the National Register for the Transfer of Technology. In January 1990, the administration issued a regulation concerning the TTL, which established that the conditions of technology transfer contracts are now the sole responsibility of the parties involved, i.e., the technology supplier and the recipient firm.

Issued in June 1991, the 'Ley de Protección a la Propiedad Industrial' (LPPI, Law to Protect Industrial Property Rights) has a secondary relationship to the field of science and technology. It regulates and co-ordinates patents; defines industrial secrets and protection and creates the 'Instituto Mexicano de la Propiedad Industrial' (IMPI, Mexican Institute for Industrial Property, operative since March 1994) (OECD, 1994a, p43). This law to protect industrial property rights is comparable to those of industrialised countries, protecting patents for a period of 20 years from the date the application for registration therefore is filed. Under the LPPI, any patent license or assignment of patent rights must be registered with Secofi.

The LPPI not only introduced important reforms in the areas of patent and trademark protection, but it also repealed the TTL and its regulations. Under the TTL all agreements which involved the transfer of technology (e.g. patent and trademark license agreements, know-how and technical assistance agreements) had to be approved by and registered with the National Registry of Transfers of Technology (NRTT). The NRTT would not register agreements which contained 'objectionable clauses', such as grant-backs, tie-ins, choice of foreign law and foreign forum clauses, and 'excessive' royalties. With the abrogation of TTL, parties may now freely negotiate technology agreements in Mexico (interview with Baker & McKenzie; UN 1992, in: UN ST/CTC/SER.B/10, 1992, p 45+46).

In short, the previous policy goals of reducing payments for foreign technology and strengthening the bargaining position of the local recipient have given way to procuring modern and competitive technology. The liberalisation of the technology transfer regulations must be seen in the context of the general liberalisation of the economy. It was assumed that, in an open and more competitive environment, firms have a self interest in acquiring foreign technology on competitive terms and in developing local technological capabilities (TNCs, December 1995, p147).

8.3. Quasi-R&D and pure R&D activities of foreign affiliates

8.3.1. Introduction

In this paragraph we will analyse the technological impacts of foreign affiliates related with their adaption and R&D activities.

8.3.2. Product and process adaptations

Do affiliates adapt their products and processes, and if they do, to which extent and for which reasons?

Table 8.1 summarises the research findings. Of the 24 affiliates who fully replied the related questions the overwhelming majority (87%) reports adaptations. Relatively more affiliates are involved in process adaption than in product adaption, although differences are small in an absolute sense. As expected simple product adaptations are mentioned more frequently than complex product adaptations.

To which extent affiliates are modifying their technologies?

Nearly half of the subsidiaries (11) is frequently adapting its products or M&E, or both. However, of these eleven affiliates only one is frequently involved in simple and complex product adaption and in process adaption at the same time.

Differences between affiliates involved in either product or process adaptations are small. Involvement in process (product) adaptations is reported by 18 (17) affiliates of which six (7) adapt their processes (products) frequently.

Product adaptations can be divided in occasional and frequent involvement on the one hand and in simple and complex adaptations on the other. Most affiliates (12) are occasionally involved in simple product adaptations and the least common is complex product adaption on a frequent basis, found in only four cases.

Table 8.1: Involvement in product and process adaptations

Source: Research 1996

Another way of determining the affiliate's involvement in adapting products and processes is by asking them the priority of these activities. In five cases adaptations are given the highest priority (see table 8.2). These affiliates consider their adaption activities are more important than their R&D activities, with a strong preference for product adaption. However, adding all priorities, differences between product and process adaptations appear to be marginal. In total 13 affiliates (62%) report that adapting products and or processes have a high priority in their company.

Product adaption is mainly due differences between Mexican and foreign parts and components and to satisfy local demand and to reduce costs. Adaption of M&E is executed to increase the level of efficiency and in case of M&E imported from the parent company.

Most important reason not to adapt is the 'too high costs involved'.

Table 8.2: Frequency of adaptation and R&D priorities

Source: Research 1996

*Since the same affiliate can be involved in both simple and complex adaptations at the same time, the number of affiliates involved (13) is not corresponding to the number of times mentioned (15).

Do any relations between firm characteristics and extent of adaption exist?

In order to answer this question we compare the averaged exports, sales in Mexico and years of production of affiliates who are not involved with those of affiliates who are frequently involved.

Results, depicted in table 8.3, do not confirm the expected relation between frequency of product adaption and market orientation. Seven affiliates reporting no involvement in simple product adaption, export on average 16 percent of sales in 1995, while this average is higher for affiliates who are involved in simple product adaption. The same holds for complex product adaption.

However, the low numbers of involved affiliates (depicted in bracelets) might heavily influence the results. The average number of years an affiliate has been producing in Nuevo Leon, calculated with slightly more cases, is for all type of adaption higher among affiliates adapting frequently. However, this relationship is not statistically significant.

There appears to be a significant relationship between frequency of simple and complex adaptations, which is fairly obvious. Affiliates involved in complex adaptations will be more inclined to be involved in simple adaptations as well.

Table 8.3: Averages of some firm-specific variables by extent of adaption, 1995

Source: Research 1996

Let us now take a closer look at the relation between imports of M&E and process adaption. Five affiliates operate with M&E obtained from their parent companies (table 8.4). Four of these affiliates report adaptations of their imported M&E. Only one affiliate is producing without making any modifications of its imported M&E. Process adaptations seem to be less common among affiliates operating without imported M&E. Due to the low numbers of affiliates we have to be careful in drawing conclusions. This being said, two observations for our sample of affiliates can be made.

Firstly, affiliates producing with foreign M&E are significantly more involved in process adaption than affiliates operating without M&E obtained from their parents. Secondly, M&E obtained from the parent company is nearly always adapted to local production circumstances.

Table 8.4: Estimated number of affiliates adapting M&E imported from their parent company

Source: Research 1996

Importing M&E: number of affiliates importing more than 50% (yes) and less (no) of M&E from parent company

Adapting M&E: number of affiliates giving priority to process adaptations (yes) and no priority (no)

(Phi=0.58, significance according to Fisher's exact test significance=0.05)

The last question, whether foreign affiliates are more involved in modifications than local firms, cannot be answered because of lacking data.

8.3.3. Summary

Between 62 and 87% of the sample affiliates do adapt their products or M&E, depending on the method of analysis. There appears to be no clear tendency that affiliates are more involved in product adaption than in process adaption. Simple product adaptations are more frequently mentioned than complex product adaptations.

Further conclusions are all based on low numbers of affiliates, and might be less reliable.

The expected relation between local market orientation and degree of adaption could not be confirmed. There is some evidence that local production experience is influencing the degree of adaptations.

Most important reason to adapt products is to satisfy local demand, while process adaption is important if affiliates have imported M&E from their parent company.

Furthermore we found evidence that affiliates producing with foreign M&E are more involved in process adaption than affiliates operating without M&E obtained from their parents.

8.3.4. R&D activities of affiliates

According to our definition of R&D eleven out of 29 affiliates are involved in R&D activities. Eleven other affiliates report they do not employ any personnel dedicated to R&D activities, the majority of them being relatively small-sized firms.

Seven other affiliates report certain expenditure on R&D, but are excluded from further analysis because these firms do not execute R&D activities in their own establishment (5 cases) and/or they undertake some very limited development work only (modifying products for local markets).

Finally, six affiliates refuse to report any information concerning their R&D activities, mainly because of strategic reasons.

The number of affiliates involved in R&D activities, eleven or 38%, is much higher in comparison with other research findings and much higher than we expected. However, Robinson (1976, p98) found the exact percentage among more than 200 foreign affiliates involved in R&D activities in Mexico in the 1970s.

Main reasons for not executing any R&D at all (mentioned by affiliates not involved in R&D costs) are 'too high costs involved' (6 times), policy of foreign partner (6 times) and low priority of R&D activities (2 times).

In which kind of R&D activities affiliates are involved?

A first indication can be found in table 8.5 The majority of R&D efforts is nearly completely in experimental development. Within this category most efforts are dedicated to improvement of products and of M&E, while developing new products and M&E is receiving less attention. A second characteristic of experimental development activities is the dominance of product technology. Only one affiliate reports involvement in (improving) process technology.

Finally, only one affiliate reports involvement in applied or basic research.

From this table we can conclude that although a large number of affiliates report possessing R&D capabilities, these seem to be mainly confined to product improvement (55%). R&D seems to be executed mainly to cater for local market needs, as hypothesised.

Table 8.5: R&D activities by main component (What do you consider your main R&D activity?)

Source: Research 1996

The one affiliate reporting basic research as its main R&D activity very strongly feels restricted in executing more basic research because of lacking qualified personnel from Mexico. It is the only affiliate hiring foreign research personnel (one person out of its four researchers). There are indications that the impact of its involvement in basic research is probably a positive one for the technological capacity of Nuevo Leon. The affiliate was founded as a joint venture in an industry and until now the affiliate never faced competitors in Nuevo Leon. The only competition the firms is facing, comes from foreign firms outside Mexico. This means that the establishment of this affiliate did not push Mexican firms out of the market, still leaving open the possibility that the presence of this affiliate might have prevented Mexican firms from entering its market. However, the affiliate is contributing directly to the stock of Nuevoleonese know-why, directly by executing basic research and indirectly by hiring Mexican researchers, through its relations with local research institutions, and through its local suppliers by giving them often technological support.

In table 8.6 some main characteristics are given for all of the eleven affiliates performing R&D. Absolute R&D costs, in total more than twelve million dollars, are being estimated by multiplying the reported percentage of R&D with reported sales in 1995. On average the absolute R&D costs are more than 1,300 US dollars per affiliate. These data become mutual comparable by expressing these absolute data in units of researchers. Two affiliates (cases 9 and 26) report both absolute and relative R&D costs being high above average. Three other affiliates (cases 2, 8 and 23) report absolute and relative R&D costs both below average.

Another qualitative measure of R&D is the share of research personnel in total employment.

For the period 1993-1995 five affiliates report an increase of research personnel. The other affiliates report no changes.

Table 8.6 shows that one extreme score heavily influences the averages for both absolute and relative numbers of researchers. Between parentheses the averages without these outliers are being displayed. The corrected average percentage researchers, 2.2 percent, still is high in comparison with other research projects. Most affiliates with a high percentage of researchers generally have low R&D costs (and vice versa).

In the last column the nationality of research personnel is displayed. In all but one cases all research personnel is local.

Table 8.6: Estimated R&D expenditure figures, 1995 (US dollars)

Source: Research 1996*

*Complemented with Mercametrica figures on sales (case2) and on sales and number of total employees (cases 6 and 34), some being of previous years. In order to obtain absolute figures on R&D, R&D data, collected on an ordinal scale, have been transformed as follows: 0.1-1% of sales becomes 0.5%, 1.1-2% becomes 1.5%, 2.1-3% becomes 2.5%, etc.

Comparing our findings with those of USA affiliates in Mexico (table 8.7) it becomes clear that the sample affiliates report a higher R&D expenditure. This is true for USA affiliates and for the whole sample. This can mean that R&D expenditure of affiliates in Mexico has been rising during the last five years or that affiliates in Monterrey spend more on R&D than other affiliates in Mexico. Probably it will be a combination of these mentioned explanations.

Table 8.7: Total sales and R&D expenditure of majority-owned USA affiliates in Mexico, 1990-1994 (millions of dollars)

Source: Based on Survey of Current Business (various issues)

Affiliates involved in R&D have the following firm-specific characteristics. The majority of the eleven affiliates is classified as a large firm (more than 250 employees) and operating in subsector 38 (machinery and equipment) for more than 35 years. Three affiliates are relatively new investments, made four to six years ago. As expected, the average number of years operating in Nuevo Leon is higher for affiliates executing R&D (38 years) than for other affiliates (operating on average 28 years). Six researching affiliates have USA or Canadian parent companies, while only one European affiliate is involved in R&D. One other affiliate is a Japanese investment and for other R&D affiliates no national data are available. The relatively high numbers of researching USA affiliates, which is contrary to our expectations, can be attributed to their dominance of the total sample (consisting of 19 USA affiliates, 5 European and one Japanese).

As expected most of the affiliates are orientated to the Mexican market, exporting not at all or little. One exception on this observation is an affiliate exporting 73 percent of its sales.

Unfortunately autonomy data are lacking for more than 50 percent of the researching affiliates.

Affiliates involved in R&D feel restricted in executing more research because of insufficient economies of scale in Mexico, thereby increasing the necessary costs (cases 2, 8 and 10) and because of lacking qualified personnel (cases 17 and 26).

As a final measure of R&D activities of foreign affiliates we take a look at patents granted to foreign affiliates in Nuevo Leon. Unfortunately we have no Mexican data of patents granted to subsidiaries. However, we do have some data of Mexican patents granted in the USA.

In the period November 1995 - June 1996 25 patents were granted in the USA to Mexican inventors, of which twelve to foreign firms and thirteen to Mexican firms or individuals. Two of the four patents originating from Nuevo Leon are from foreign affiliates (USPTO, 1997a). This implies that although few foreign affiliates are involved in R&D activities they play a very important role in the technology production within Mexico.

Table 8.9 gives an impression of the earlier number of patents granted in the USA to Mexican inventors. We are especially interested in the category 'enterprises or institutions', but we have to keep in mind that we are dealing with only half of all granted Mexican inventions.

During the period 1970-1984 the ratio of dependency has been declining at a fast rate, indicating that Mexican firms have been increasing their innovatory capacities. However, this trend did not continue after the mid 1980s. The dependency ratio has been rising again which can be an indication of a declining innovatory capacity of Mexico.

The most recent figures (as mentioned earlier) point at a repeated foreign dominance of innovatory capacities.

In short, patent data show two important things:

1) After a period of reduced dependency on foreign patents, the absolute and relative number of foreign patents is rising again with regard to Mexican patents.

2) The number of patents granted to affiliates in Mexico is nearly equal to that of Mexican firms, meaning that foreign affiliates play a crucial role in Mexican technology development and in increasing the innovatory capacity of Mexico.

Table 8.9: Patents granted in the USA to Mexican inventions, 1970-1995

Source: Based on US Patent & Trademark Office, 1996

*Patents granted to foreign firms divided by patent granted to Mexican firms.

Having concluded that on average eleven affiliates roughly spend 1.6% of their annual sales on R&D activities, the last question is if they are spending more on R&D than local firms. If they do, they are increasing the innovatory capabilities of Mexico. In order to find out whether affiliates are spending a higher percentage of their sales on R&D than local firms we will analyse data of the 'National survey of employment, salaries, technology and training in the manufacturing sector' (Enestyc, STPS & Inegi,1995). Some 140,000 firms were interviewed, among others about their R&D expenditure. There appears to be no significant differences between ownership and R&D expenditures, nor calculated on basis of all 51 subsectors (called ramas), nor within the 14 ramas constituting subsector 38 (machinery and equipment).

8.3.5. Summary

Eleven out of the 29 surveyed affiliates are involved in R&D activities. Most of these affiliates can be described relatively larger and older majority-owned affiliates, oriented towards the Mexican market. Most affiliates have USA or Canadian.

In all but one cases these activities can be classified as experimental research laboratories. This means that the Mexican stock of know-why is being increased by the existence of foreign affiliates. One affiliate is, by performing basic research, influencing the technological capabilities of Nuevo Leon. Indicators point at a probably positive contribution.

Looking at the number of patents granted in the USA, it seems that foreign affiliates play a crucial role in Mexican technology development and in increasing the innovatory capacity of Mexico.

Most frequently mentioned R&D restriction is diseconomies of scale in Mexico, while affiliates not involved refer to high costs involved and restrictive parent policies.

On a national level we did not find significant differences between ownership and input indicator R&D. A the same level the evidence is indicating that foreign affiliates are at least as important as local firms in terms of output indicator patents granted in the USA.

8.4. Internal technology transfer

8.4.1. Introduction

In the beginning of the 1990s obstacles to the importation of specific technologies and services have been removed in Mexico. "For example, in 1991, the government rescinded the decrees which made imports of computers and related products subject to an import permit system. Although this system applied only to hardware, it had a chilling effect on the market for technological services. Another example is the strengthening of intellectual property laws, which have improved prospects for exporting computer software and biotechnology products, among others" (Ruben Moreno, 1996). As these examples illustrate, the process of trade liberalisation, deregulation and privatisation have facilitated the entry of foreign technology into Mexico.

To which extent and in which way are the surveyed affiliates involved in internal technology transfer?

8.4.2. Experiences of surveyed affiliates

Do all affiliates introduce foreign technologies by obtaining their technologies from their foreign partner?

Technology flows to foreign affiliates are depicted in the left half of table 8.10. Nineteen of the responding affiliates, or 79 percent, report one or more technology flows from its foreign partner. Only a minority of foreign affiliates is not obtaining any form of technology from its parents at all. For the early 1990s the UN found a roughly equal percentage (86%) among affiliates throughout Mexico (ST/CTC/SER.A/18, 1992, p38).

The same number of firms as software transfers mentions channels of hardware transfers.

Main channels of internal technology transfer appear to be training and the imports of M&E. Less than half of the affiliates reports managerial services and maintenance and/or repair of their M&E, while licenses and patents from the foreign partner are most uncommon among the sample affiliates.

Table 8.10: Components of internal TT

Source: Research 1996

Do MNEs transfer their technologies completely to their affiliates?

Fifteen affiliates, the large majority of affiliates involved in internal TT, report that they obtain less than 50 percent of their technologies from their parent companies. On average, nearly one quarter of the affiliate's technology (23%) has its origin in its foreign parent. Furthermore, the average of hardware transfer obtained from the partner (29%) is evidently higher than that of software technologies (17%).

Taking a closer look at hardware technologies, it seems that (M&E) is the largest technology flow from parent to affiliate. More than one half of the affiliates (58%) report inflows of M&E. On average 41 percent of the M&E used by these affiliate is obtained from their parent company.

We asked the affiliates obtaining more than 50 percent of their M&E to estimate the level of this type of hardware. Three affiliates consider their M&E as state of the art technology and claim their M&E is more modern than normally used by their parents. Two of them are recently opened greenfield production sites (1 and 4 years ago).They also report the most complete transfers, respectively 92 and 65 percent. The remaining six responding affiliates are operating with M&E of the same technological level as their parent companies. None of the affiliates importing more than half of their M&E from their parents reports the use of obsolete or relatively older process technologies. As an example of financial investments in transfers of M&E we want to mention that in 1991 the expenditures for new plant and equipment in USA affiliates amounted 173 millions of dollars (Survey of Current Business, May 1993), or 5.3 percent of sales.

Affiliates not importing any M&E at all mainly refer to different products they are making, differences in plant size and in sector. However, affiliates who do receive M&E from their parent companies feel restricted to import more M&E, mainly because of differences in labour costs, and to a far lesser extent differences in plant size and in products.

Besides M&E other channels of hardware transfers are patents and licensing. Although few affiliates are involved in transfers of patents and/or licenses, it appears that five out of six affiliates have no other sources but their foreign partner. Obviously these figures can be explained by the restrictive nature of licenses and patents.

As an indication of absolute size of these flows, we can use USA data. In 1992 USA parent companies recorded net inflows of royalties and license fees of 407 millions of dollars from Mexico. USA affiliates operating in Mexico paid 350 millions of dollars, while payments of Mexican firms amounted to 57 millions of dollars (Survey of Current Business, November 1996).

Software channels are less completely than hardware channels. On average 23% of all training in an affiliate comes from the parent. Managerial and maintenance services are even less completely. On average less than 17 percent of these services comes from the parent company.

In the chapter concerning labour impacts we concluded that a main motive to train personnel is the specific nature of job-related tasks. At this place we can add that inflows of training are correlated with flows of M&E and maintenance.

The implication of the above is that training received from the foreign partner is confined to operation and maintenance skills, transferred through on-the-job training.

Elsewhere we explained our expectations of certain firm characteristics to influence the extent of internal TT. We expect the extent of internal TT, both as a whole and its separate components, to rise with local production experience, plant size, and integration in the global production system, and to decrease with autonomy.

Looking at the survey results, depicted in table 8.11, we see that production experience is not significantly influencing the extent of internal TT.

Size appears to be positively influencing the extent of software transfers.

The percentage of foreign ownership, although not significantly correlating with software transfers, is positively influencing the transfer of patents and (as a consequence) with hardware flows and total inflows of foreign technology.

The more technological and financial freedom an affiliate has, the lower the reported technology flows are. These data are confirming theories that multinationals will transfer more to their affiliated firms the more these affiliates can be controlled.

With the affiliate's integration in the global economy we mean the directions of the affiliate's inputs and outputs. Looking at the affiliates' input orientation (or purchases) we can conclude that the less an affiliate is obtaining its inputs locally, the higher the technology flows it is receiving from its parent company. Since the large majority of affiliates has USA capital, there is a high correlation between inputs obtained from Canada and the USA.

Furthermore we see that the larger is the export orientation, the larger are the technological inflows. In the case of exports to other Nafta countries there is a strong relationship with inflows of M&E and training. In order to be competitive on the Nafta markets, the affiliates in Nuevo Leon need foreign M&E.

Somewhat surprising are the findings that more exports to Asia correlate with more inflows of software technology, notably maintenance and management services. While the same relationship exists for exports to Europe, we did not find these relations for Nafta exports.

Table 8.11: Pearson's correlations between firm characteristics and channels of internal TT

Source: Own research

a) Significant at the 0.01 level (two-tailed test)

b) Significant at the 0.05 level (two-tailed test)

c) Significant at the 0.10 level (two-tailed test)

* Spearman's rank correlations

8.4.3 Summary

The large majority of the affiliates is obtaining technology from its parent. The number of affiliates receiving software flows equals that of hardware flows. Although more affiliates mention training transfers than M&E transfers, the latter seem to be the most complete transfers. On average an affiliate obtains 41 percent of its M&E from its parent company. Averages for other transfers are less than 25 percent. Most imported M&E seems to be state of the art technology.

Problems related with internal technology transfer relate especially the differences in labour costs and differences in plant size. Furthermore training appears to be confined to operation and maintenance skills. Furthermore we see that the larger is the export-orientation, the larger are the technological inflows.

8.5. External technology transfer

8.5.1 Introduction

We define external technology transfer as all technology flows from foreign subsidiaries to local firms. These flows diffuse technological know-how and increase the technological capacities of the host country.

In section 8.5.2 we will pay attention to formal channels of external technology transfer, i.e. external TT which are compensated by opposite financial flows (running from local countries to affiliates). In section 8.5.3 we will pay attention to external TT which are not compensated by opposite financial flows, called informal channels of external TT.

8.5.2. Formal TT

In this section we will analyse the involvement of affiliates in external TT. In this analysis we will make a distinction between rarely and frequently transfers. Since frequently external transfers obviously have a larger positive impact on the host economy we will pay most attention to channels of frequently transferred transfers.

According to a technology expert at ITESM, "MNEs with production sites in Nuevo Leon made this choice in order not to loose their core technologies. Their plants are totally closed, resulting in isolated islands of state of the art technology. It is a myth that local firms can obtain the latest technology from foreign affiliates and adapt them for their own purposes. Instead local firms obtain older technologies and do not adapt them" (interview of 17/06/96).

Of the 33 sample affiliates five report they never transfer technology to local firms at all. This means that 85 percent of the affiliates is involved in any type of external TT, half of them (14 or 42%) on a frequent basis.

Through which channels technology is being transferred to local firms?

Most of the affiliates (79%) are involved in any form of software technology transfer. Less affiliates are involved in hardware transfers (61%). However, most flows of external TT take place on an irregular basis. Only 42 percent of the affiliates report frequent involvement in external TT.

Looking at frequent used channels of transfer only, it seems that sale of M&E is the most frequently mentioned channel of external TT (24%). Other important channels of external TT are training (21%) and maintenance of M&E (21%). Leasing M&E and management services are of relatively few importance, while selling patents is not mentioned at all.

Table 8.12: Channels of external TT and intensity*

Source: Research 1996

*Because affiliates can be involved in several types of hardware and software transfers at the same time, figures can not always be added.

Asking the affiliates directly which is their main channel through which they transfer technology to local firms delivers results confirming the above sketched picture, although less affiliates did respond. Eleven out of 20 subsidiaries (61%) indicate that sale of M&E is the most important channel of external TT. Six affiliates (30%) mention training and three affiliates mention other transfer channels.

Summarising, we have seen different figures of affiliates' involvement in transferring technology to local firms, ranging between 61 and 85 percent. The emerging picture, however, is a consequent one. Main channel of external TT is sale of M&E followed by training.

Why affiliates are not involved in (more) external TT?

Table 8.13 identifies the main restricting reasons. In total six affiliates feel restricted in transferring technology because technological differences are too large. None of these affiliates is reporting external TT on a frequent basis. Too complicated nature of the affiliate's technology is mentioned twice, once by an affiliate frequently involved and once by an affiliate not involved in external TT. Most often mentioned reason is preventing loss of control over affiliate's technological skills. Spreading of these skills would mean loss of the affiliate's competitive advantage. Eight of ten affiliates reporting this reason are not involved in external TT, which is confirming our expectations.

Table 8.13: Most important restriction of (further) external TT

Source: Research 1996

*The number of affiliates printed in this table are slightly lower than those depicted in table [channels of external TT] because not all affiliates report their most important restriction of external TT.

Looking at the firm characteristics and the extent of technology diffusion (table 8.14) we can draw the following conclusions. Local production experience is to some extent influencing with extent of lease of MNE and training of local firm's personnel.

The smaller the size of the affiliate, the more likely it is that this affiliates is selling its M&E and/or maintenance services to Mexican firms. However, the opposite is true for managerial services, which are being diffused by larger affiliates.

Autonomy of the affiliates seem not to play a significant role.

Our expectations of a strong relation between local integration and extent of technology diffusion are confirmed only in case of leasing M&E.

The expectations about relationships between channels of internal TT and technology diffusion can not be confirmed for sale of M&E and managerial services to Mexican firms.

Table 8.14: Pearson's correlations between firm characteristics and channels of direct technology diffusion

Source: Research 1996

Summary

Between 61 and 85 percent of the interviewed affiliates report transfers of technology to local firms, mainly by sale of M&E and training. Most important reason to be not involved in external TT are avoidance of loss of control over technologies, being essential competitive advantages, and secondly the too complicated nature of affiliates' technology.

Affiliates importing more software technologies from their parents appear to have larger software technology appear to diffuse more technology in Mexico. Furthermore there is evidence that larger affiliates diffuse more management skills, while smaller affiliates are selling more M&E and M&E services to Mexican firms.

8.5.3. Informal technology transfer

Blomström and Wolff (1994, p265) proved with Mexican data for the period 1970-1975 that technology spillovers from foreign direct investment do exist. In this part we can take a closer look at the channels through which technology is informally being transferred to Mexican firms.

The large majority of affiliates, 68 percent (or 21of 31valid questionnaires), feels that local firms are obtaining technological capabilities from affiliates without payments to these affiliates. Eight affiliates estimate the extent of informal TT is low, three affiliates estimate a high extent, and experience an informal TT to a moderate extent.

Table 8.15: Perceived extent of informal technology transfer

Source: Research 1996

The three most important channels of informal TT appear to be knowledge of former employees (6), imitation of products (6) and contacts with local suppliers (3). When we operationalise importance as number of times mentioned, the same channels are being mentioned, though in different sequence. Most important channel appears to be product imitation, followed by contacts with local suppliers, former employees ranking thirdly.

Table 8.16: Channels of informal technology transfer

Source: Research 1996

*MF=Mexican firms

Eight affiliates (36%) feels the demonstration effect (product and process imitation) is the main source of technological leakage.

Affiliates reporting former employees as an important source do not have a larger labour turnover than other affiliates.

To which extent and in which ways affiliates experience uncontrolled spread of technology through supports of their local suppliers?

The most strongest relationships between perceived of uncontrolled spread of technology and supplier support appear to be between procurement, price, and management support (see table 8.17). Between management and financial support significant, but lower, relations are found as well. Surprisingly no statistical relations were found for technological and information assistance.

To summarise, affiliates rendering notably procurement, price and management assistance transfer more technology than affiliates not involved in these kinds of supplier supports.

Table 8.17: Relations between affiliates reporting local suppliers as a source of informal TT and several types of supplier support

Source: Research 1996

A final channel of informal technology transfer is occurring between affiliates executing R&D activities and participating local firms and research institutions.

Unfortunately of the eleven affiliates involved in R&D activities only six respond questions concerning local co-operation. In this research three of the eleven affiliates involved in R&D activities are executing their R&D projects in co-operation with local firms and two are reporting R&D relations with Mexican research institutions.

According to an interview with Proexport, the former state agency attracting and analysing foreign MNEs, R&D co-operation between foreign firms and local organisations are rare. Earlier research in Mexico (Nadal, 1977) concluded the same.

Although we can not generalise because of the low numbers involved, it seems that R&D activities of foreign affiliates are isolated. Most probably the impact of R&D activities as a channel of external TT is very small.

One big exception on this conclusion is the R&D project of Mercedes Benz SA de CV. In 1994 this subsidiary announced that it would start a project in collaboration with the ITESM called 'El autobus del año 2000'. The objective is to develop a construction of a chassis for buses which responds to the NAFTA norms. This is the first time that Mercedes Benz is developing and building an autobus outside its own factories (Division de graduados e investigacion del ITESM, 1994, p6).

Table 8.18: Contacts with local firms and research institutions of affiliates involved in R&D

Source: Research 1996

Which firm-specific characteristics determine the extent of external technology transfer?

There appear to be no differences in channels between affiliates reporting low levels of informal TT on the one hand and with moderate or high levels of informal TT on the other. Nor can differences be found for firm-specific characteristics such as country of origin, size, way of establishment, years of operations and percentage of foreign ownership. Nor can the relationship between the extent of labour turnover and extent of informal TT statistically be found.

However, there is some (weak) evidence that the more an affiliate is integrated in the local economy, the more an affiliate will face informal TT. This is true for both backward (operationalised as support to suppliers) and forward (operationalised as sales within Mexico) linkages. Furthermore table 8.19 confirms the expected relations between informal and formal TT.

Firm characteristics seem to play a fairly indifferent role in influencing the perceived extent of informal technology transfer. There is some modest evidence that affiliates receiving software transfers of their parents experience larger spillover effects. Affiliates involved in diffusion of technology (external TT) through lease of M&E, training and management skills experience more severe informal TT as well. Local production experience, size and autonomy seem not be significant factors at all.

As expected the more the affiliate is integrated in Mexico, the more it is experiencing spillover effects. Procurement and price support to local suppliers, and to a lesser extent managerial, location and price assistance, are more often mentioned by affiliates reporting larger spillover effects.

Table 8.19: Spearman rank correlations between firm characteristics and extent of informal TT

Source: Research 1996

a) Significant at the 0.01 level (two-tailed test)

b) Significant at the 0.05 level (two-tailed test)

c) Significant at the 0.10 level (two-tailed test)

Summary

To conclude, 68 percent of the affiliates perceive a certain extent of informal TT, most of them reporting a moderate extent. The three most important channels of informal TT appear to be knowledge of former employees, imitation of products and contacts with local suppliers. In the latter case, especially procurement, price and management assistance appear to be important in this context. As predicted, the extent of informal TT depends on the extent of formal TT and on the level of local integration.

8.6. Technology capabilities of Mexico and Nuevo Leon

8.6.1. Mexican technological capabilities compared vis-à-vis other OECD countries

Table 8.20 gives an overview of several input measures of Mexico's science and technology activities and other OECD countries.

In 1993 Mexican gross domestic expenditure on R&D were 1,562.9 million current PPP$. Its share in total OECD expenditure on R&D was 0.41 percent, while its share of population was 9.45 percent.

Other indicators of technological capabilities clearly show that Mexico is at the bottom of all ORCD countries. Mexico has the lowest GERD per capita population, the lowest GERD as a percentage of GDP, the lowest number of R&D personnel and researchers and the lowest percentage of GERD financed by industry.

Mexican R&D expenditures have increased sharply in real terms since the onset of economic reforms in the 1980s, but, at a 0.3 percent of GDP, remain low in comparison to other OECD countries. The Mexican government is aiming to double its national expenditure on R&D to 0.7 percent of its GDP in 2000, of which indigenous firms have to account for 45 percent (SEP/Conacyt, 1995, p32+33).

Looking at the distribution of R&D actors we see that the majority of R&D expenditures is concentrated in government and higher education.

Most R&D financed by industry can be attributed to the large Mexican companies, although interest among the many small firms in intensified R&D efforts is growing because of the domestic and international increasing competition. However, small and micro sized firms have not been able to modernise because their main goal has been surviving the crisis (Ernst & O'Conner, 1989, p81)."The low level of business R&D reflects in part Mexico's disadvantage in knowledge-intensive industries and its technological dependence on multinational enterprises. Other factors include the lack of an adequate venture capital market and the hesitancy of commercial banks to fund technological development projects" (OECD, 1996a, p162-164).

When compared with other OECD countries, it becomes clear from table that Mexico's spending on its resources devoted to R&D, both measured in terms of expenditure and of labour, are very limited.

In table 8.21 we focus on resources devoted to R&D in the business enterprise sector. In terms of expenditures, Mexico had the lowest spending of all OECD countries. BERD taken as a percentage of domestic product of industry (DPI) and of gross domestic product (GDP) is only 0.03 percent in the Mexican case. Also in terms of labour devoted to R&D activities Mexico's experiences are extremely low in comparison with other OECD countries. Only 7 percent of all Mexican R&D personnel was working in the business enterprise sector, and 6 percent of all Mexican researchers can be found in the business sector. Although the economy has become open and deregulated, the attitude of Mexican firms did not change yet. Nowadays, in manufacturing industries firms spend on average 2.5 percent of their incomes on purchasing technology and spend only 0.6 percent on R&D activities (SEP & Conacyt, 1995, p16).

The majority of BERD (87%) is performed in manufacturing industries, notably in the drug industry (9.6%), computer industry (3.6%) and electronics (0.9%).

However, in a more encompassing research, covering more countries than only the OECD member states, the OECD classified all countries in four different categories (technology leaders, other high-tech countries, middle tech countries and low high-tech countries). Based on the development of R&D intensities and GDP over the 1980s the OECD placed Mexico in the third category (middle tech countries) which also consists of India, Italy, Ireland, Denmark, Canada and Australia (OECD, TEP, 1991, p59+60).

A third input measure is the expenditure on R&D in the higher education sector (HERD). Only 0.13 percent of Mexican GDP was directed to higher education, which is the lowest percentage of all OECD countries and thrice as less as the OECD average. "Mexico science faces two kinds of problems. On the one hand, there is the historical lack of a scientific tradition, and, on the other, there are the problems produced by the financial crisis of the 1980s" (OECD, 1994a, p102; interviews with technology director of ITESM and with government of economic development).

Table 8.20: International comparison of input measures of technological activities: Gross domestic expenditure on R&D (GERD), labour and distribution, 1993

Source: OECD 1996-2

*1992

Table 8.21: International comparison of input measures of technological activities: expenditure on R&D in the business enterprise sector (BERD) and expenditure on R&D in the higher education sector (HERD), 1993

Source: OECD 1996-2

*1992

Looking at the output indicators in table we get a similar picture.

With 25,845 national applied patents Mexico's share in the total number of applied patents in the OECD is only 0.6 percent. The number of national patent applications is the sum of resident and non-resident patent applications. "It shows the size of the technological 'market' that the country represents and also, in a certain manner, the potential market for industrial goods and services" (OECD, 1996-2, p63).

The number of resident patent applications gives an idea of the production of technology, while the number of non-resident patent applications indicates the technological penetration. A fourth category of technological output indicators is the number of external patent applications (i.e. applications made abroad), indicating the technological 'diffusion' of each country.

Looking at the ratio of dependency (non-resident/resident patent applications), we see non-resident patents were almost 19 times more numerous in Mexico than resident patents. Although this figure is much higher than the OECD average, countries such as Greece, Portugal and Spain had even higher dependency ratios.

The ratio of auto-sufficiency (resident/national patent applications) is only 0.05, meaning that only 5 percent of all applied patents were of domestic origin in Mexico. Again this figure is rather low when compared with the OECD average, but is even lower in the above mentioned countries. During the last 15 years, main patents have been granted with a USA origins (23,206), followed by Germany (2,847) (Conacyt & SEP, 1996, p66).

When we express the resident patent applications in terms of population, we obtain the inventiveness coefficient. Together with Turkey Mexico's inventiveness approaches zero Furthermore the low rate of diffusion (external/resident patent applications) makes it clear that Mexico is also lacking in applying for patents in foreign countries.

Summarising, the used input measures of technological activities all showed Mexico is performing on a relatively, and sometimes absolutely, low level in comparison with other OECD member states.

The output measures showed to a certain extent the same picture. Mexico is performing far below the OECD average, but, is performing slightly better than in terms of input indicators.

The question of whether Mexico's technological capabilities are sufficient or not too far behind to benefit from MNEs can not be answered in this basis. Therefore we need more disaggregated data by industry and by region.

8.6.2. Nuevoleonese technological capabilities compared vis-à-vis Mexican ones

The analysis of technological capabilities of Nuevo Leon is more qualitative because of lacking data. We feel there are large differences between the technological capabilities of Nuevo Leon and Mexico as a whole. This stems from the presence of the grupos, competitive MNEs themselves, and of good universities, such as TEC and UANL, boosting the technological capabilities of Nuevo Leon. "The growth of domestically-owned industrial groups and conglomerates in developing countries [like Mexico] has been important in providing a strong domestic entrepreneurial and technological base, one that can usually compete effectively with MNC subsidiaries. The growth of such large private sector companies has led to increased domestic capital accumulation, along with technological and managerial know-how" (Marton, 1986, p34). One of these firms, HYLSA S.A. de C.V., was Mexico's second most important organisation receiving US patents (USPTO, 1997b).

Although table 8.23 only gives input measures of science and technological activities, it becomes very clear that Nuevo Leon (NL) has in most cases the highest scores on several (rough) BERD and HERD indicators.

Other research from ITESM (1994, p393) shows Nuevo Leon is representing one of the Mexican states with the highest incentives of technological growth for private firms.

The higher Nuevoleonese percentage of large firms having a R&D-unit might be explained by the existence of the grupos, of which most of them are performing important R&D-activities in their founding city (ITESM, 1994, p398).

Another source is an unpublished ITESM research (Molino & Villanuava, 1992) on growth and technology transfer in Nuevo Leon. Of the interviewed local firms, 8.5 percent used obsolete or regular technology, 42 percent use good technologies, 31 percent use very good technologies and 17 percent of the firms use state-of-the-art-technology.

Finally, in all interviews with local government, consulates and employer's and employee's organisations the high quantitative and qualitative level of technicians in Nuevo Leon is being mentioned.

We can conclude the above analyses by citing the following OECD conclusion (1994, p142): "Scientific and technological policy must become a central agent of Mexico's structural change: the country needs many more professional engineers, scientists and technicians; it needs a much larger technological infrastructure to support the modernisation of industry, services, and public administration; it needs a much wider base of enterprises with high technical standards able to compete on the international markets on the basic of product quality rather than low labour costs; it needs strong scientific input in all economic and production activities in order to protect the environment".

Table 8.22 International comparison of output measures of technological activities: patents, 1992-1994

Source: Based on OECD 1996-2

*Comprising 1993 and 1994 only.

Table 8.23: Technological capabilities of Nuevo Leon compared to those of other Mexican states, 1996

Source: ITESM, Centro de estudios estratégicos, 1995 and 1997

Note: NL=Nuevo Leon, DF=Districto Federal

8.6.3. Perceived technology gap of surveyed affiliates

The technology gap between affiliates and local firms is estimated at the basis of the following two indicators:

1) the perceived technological position of affiliates vis-à-vis local firms in 1991 and in 1996, and

2) explicit reference made to the existence of a technology gap.

Eleven out of 34 affiliates feel a large technology gap is present, while ten other affiliates feel the same to a lesser extent. Thirteen affiliates indicate there are no (large) technological differences between foreign affiliates and local firms and one affiliates could not be classified because of lacking data.

Combining these data with data on R&D laboratories, it can be seen that most R&D is being performed by affiliates reporting a large or moderate technology gap.

Technology inflows from foreign partners are highest among affiliates perceiving a large technology gap and decreasing with perceived technological differences.

Furthermore we hypothesised that the smaller the technology gap, the more commercial technology transfer would take place. The findings in table 8.24 do not confirm these expectations. For the interviewed affiliates it appears that commercial technology transfer is highest among affiliate who experience a large technology gap. Nor did we find the expected combination of large spillovers and small/no technology gap (or the opposite). Affiliates perceiving a large technology gap are at the same time reporting largest spillovers.

Table 8.24: Technology gap and estimated technology transfer through foreign affiliates

Source: Based on Lan & Young, 1996, p62 and questionnaire survey

8.7. Costs of technology transfer

Calculating financial aspects related to internal and external technology transfer is a difficult task because of lacking data and because of transfer pricing, causing inaccurate data. Keeping these limitations in mind, we will use the available data to get an impression of the involved financial flows.

"The Technology Balance of Payments (TBP) registers the commercial transactions related to international technology transfers. It consists of money paid or received for the use of patents, licences, trademarks, designs, know-how and closely related technical services (including technical assistance) and for industrial R&D carried out abroad, etc." (OECD 1996-2, p64).

In 1994 the Mexican technology balance of payments was, as the years before, negative: 417 millions of dollars, with the receipts being 106 and the payments being 522 millions of dollars, resulting in a coverage ratio of 0.2. These data correspond to international transactions for royalties and technical assistance (Banco de Mexico; in CONACYT/SCP, Mexico indicadores de actividades cientificas y tecnologicas 1995, 1996, p73+111).

The coverage-ratio is the coefficient obtained by dividing receipt by payments. It shows to what extent a country covers its own requirements of technological imports by its corresponding exports. Total transactions is the sum of receipts and payments. When compared with other countries it gives the weight of each country in international technology trade. In 1992, for example, Mexico's weight in the international technology trade, while that of Spain was more than seven times as high.

In 1991 more than three-quarters of firms in Mexico reported payments for technology transfer and/or technology buying to other firms within Mexico. Second source is the US, reported by 16 percent of all firms. On average local firms spend a higher percentage (3.2%) of their incomes on technology transfer and/or buying than foreign firms (2.9). A student t-test suggests that we should not reject the hypothesis that there are no differences in such spending between foreign and local firms. For both types of firms these payments are on average much higher than expenditure on R&D (calculated with data of Enestyc, STPS & Inegi,1995).

Table 8.25: Mexican technology balance of payments, 1990-1995*

Source: Conacyt/SEP (1996, p73)

*Until June

However, it is not possible to track if the technology buying and selling firms or foreign-owned or local firms. Most probably the technology selling firms are foreign affiliates. "In 1980, for example, 75 percent of the registered technology agreements were between MNC parent companies and their Mexican subsidiaries and affiliates" (Marton, 1986, p266).

Looking at the net payments of USA affiliates and indigenous firms in Mexico to MNEs in table 8.26, we see the following trends. Local firms seem to have net financial inflows for private services, implying that local firms received more for private services than they have paid to USA MNEs. However, this is only the case for two categories: telecommunications and other services.

For the categories education, management and maintenance and repair of M&E financial flows are, as expected, outward. Payments on management and maintenance have been increasing until 1994, and dropped in 1995 as a consequence of the Mexican crisis.

Table 8.26: Net payments of USA affiliates and Mexican firms for other private services, 1992-1995 (millions of dollars)

Source: Survey of Current Business (various issues)

Summary

Local firms dedicate on average higher shares of their incomes to obtain technology than foreign firms, although not significantly more. Both foreign and local firms spend more on buying technology than on R&D.

Main summary

The presence and actions of foreign affiliates can have technological impacts on host economies in three different ways: by means of technology production, transmission and dissemination, together called technology transfer.

I Affiliates and technology production

Adaptations

The large majority of foreign affiliates is involved in any kind of adaptations. The number of affiliates adapting its products is slightly higher than the number of affiliates making process adaptations, though differences are small. However, most reported product adaptations appear to be relatively simple modifications.

Affiliates with less autonomy are more active in process modifications, while affiliates involved in complicated product adaptations are less dependent on their parent companies.

Most important reason to adapt products is to satisfy local demand, while process adaption is important in affiliates obtaining machinery and equipment from their parents.

Adaptations are not a pure R&D activity according the OECD and do not contribute to the innovatory capabilities of host economies. They affect the know-how rather than the know-why capabilities of the host country.

R&D

Eleven out of 29 foreign affiliates, or 38%, report R&D-activities. Most of these affiliates can be described as relatively large affiliates with a Mexican market orientation. The extent of dependency on the foreign parent can not be analysed because of lacking data.

Direct contributions to the innovatory capabilities can be measured in capital, number of employees and in granted patents.

The R&D expenditure is worth more than twelve millions of dollars, or 1.6 percent of annual sales in 1995.

At least 90 researchers are employed in these affiliates, all but one having the Mexican nationality.

Most affiliates are involved in product improvement and to a less extent in product development. Only one affiliate reports process improvement and applied research is reported by one other affiliate.

Since the 1990s the US patent office has been granting as many patents to foreign affiliates operating in Mexico as to Mexican firms, indicating that the former are making important contributions to Mexico's innovatory capacities.

On national level there seems to be no significant differences between R&D-expenditure of foreign and Mexican firms.

Most frequently mentioned R&D restrictions are diseconomies of scale in Mexico. Affiliates involved in applied research and in process improvement also report lack of qualified personnel.

Most affiliates not performing R&D report that too high costs are involved and that policies of their parent companies are restricting R&D.

We can conclude that the contribution of affiliates in the innovatory capacities of Nuevo Leon is fairly large.

II Affiliates and technology transmission

The large majority of the affiliates is obtaining technology from its parent. The number of affiliates receiving software flows equals that of hardware flows. Although more affiliates mention training transfers than M&E transfers, the latter seem to be the most complete transfers. On average an affiliate obtains 41 percent of its M&E from its parent company. This imported M&E seems to be state of the art technology. Averages for other transfers are less than 25 percent. We found evidence that training is mainly confined to improvement of operation and maintenance skills.

The two main factors hampering affiliates in obtaining technologies from their parent are the differences in labour costs and differences in plant size.

The extent of internal technology transfer is mainly determined by the autonomy of the foreign affiliate. The higher its dependency on its foreign partner, the higher the extent of obtained technologies of the latter. This can be explained by the fact that MNEs prefer to transfer their technologies to fully-owned and/or majority-owned affiliates, because the risk of loosing its technologies is lower in these cases.

Looking at the affiliates' input orientation (or purchases) we can conclude that the less an affiliate is obtaining its inputs locally, the higher the technology flows it is receiving from its parent company. Since the large majority of affiliates has USA capital, there is a high correlation between inputs obtained from Canada and the USA.

Furthermore we see that the larger is the export-orientation, the larger are the technological inflows. In the case of exports to other Nafta countries there is a strong relationship with inflows of M&E and training. In order to be competitive on the Nafta markets, the affiliates in Nuevo Leon need foreign M&E.

We did not find a relationship between production experience and extent of internal technology transfer. Firm-size appears to be positively influencing the extent of software transfers.

III Affiliates and technology diffusion

Formal technology transfer

Between 61 percent and 85 percent of the interviewed affiliates reports transfers of technology to local firms, mainly by sale of M&E and training. Most important reason to be not involved in external TT are avoidance of loss of control over technologies, being an essential competitive advantages, and secondly the too complicated nature of affiliates' technology.

Affiliates importing more software technologies from their parents appear to diffuse more technology in Mexico. Furthermore there is evidence that larger affiliates diffuse more management skills, while smaller affiliates are selling more M&E and M&E services to Mexican firms.

Informal technology transfer

Almost 70 percent of the affiliates perceive a certain extent of informal technology transfer, most of them reporting a moderate extent. The three most important channels of informal technology transfer appear to be knowledge of former employees, imitation of products and contacts with local suppliers. As predicted, the extent of informal technology transfer depends on the extent of formal technology transfer and on the level of local integration.

Looking at the technological capabilities of Mexico and Nuevo Leon we find that Mexico's technological capabilities are very low in relation with other OECD countries. Technological policies are not transparent and lack financial support. Mexico is missing both capital and culture to improve its technological capabilities. Nuevo Leon is one of Mexican states with the highest technological capabilities. In Nuevo Leon affiliates perceiving a high technology gap report the highest contribution as producers, transmitters and diffusers of technology.

Local firms dedicate on average higher shares of their incomes to obtain technology than foreign firms, although not significantly more. Both foreign and local firms spend more on buying technology than on R&D.

Main findings of questionnaire on technological impact of foreign firms in the manufacturing sector of Nuevo Leon

1) The large majority of foreign affiliates is involved in any type of modifications; process adaptations being as important as product adaptations.

2) Main R&D activity is product improvement, basic research is executed by one affiliate.

3) R&D activities are being executed by 38 percent of all affiliates, contributing more than twelve millions of dollars, hiring more than 90 Mexican researchers and receiving half of all Mexican patents granted in the USA.

4) Lack of economies of scale seem to be the main reason restricting affiliates already performing R&D. Other affiliates, not involved in R&D, face too high costs and restricting policies of their parent companies.

5) Training, followed by M&E, is the most important technological flows from parent company to affiliate.

6) Technology transfers from parent to affiliate are most complete for M&E: more than 40 percent of all M&E is obtained from the foreign partner. This imported M&E is state of the art technology.

7) Affiliates importing technology from their parent are more dependent on their parent and less oriented towards the Mexican market.

8) Between 61 and 85 percent of the interviewed affiliates transfer technologies to local firms, mainly by sale of M&E and training.

9) Obstacles in transferring technologies to Mexican firms are fear of loosing control over essential technologies and the too complicated nature of affiliates' technology.

10) Almost 70 percent of the affiliates feel that Mexican firms increase their technological capacities at the expense of the affiliates. Main channels of this 'uncontrolled' spread of technology are the knowledge of former employees, imitation of products and contacts with local suppliers.

11) Compared to other OECD countries, Mexican technological capabilities are very low.

12) Within Mexico Nuevo Leon, is one of the states with the highest, probably highest, technological capabilities.

Affiliates perceiving the largest technological differences seem to be the largest contributors to local technological capabilities.

Appendix I: List of selected literature/bibliography