2

Industrial and Corporate Change Advance Access published March 24, 2007 Industrial and Corporate Change, pp. 1 of 24 doi:10.1093/icc/dtm003

Learning, product innovation, and firm heterogeneity in developing countries, evidence from Tanzania Micheline Goedhuys

Using firm data, this study investigates the various sources of firm learning, investment and linkages, and their importance for product innovation in Tanzania. The analysis reveals important differences in innovation strategies for foreign and local firms. Foreign innovative firms have stronger vertical linkages with other foreign firms and invest more in human and physical capital. Local firms offset these disadvantages through in-house R&D, connectivity, and collaboration with other local firms, proving to be more embedded in the local industrial structure.

1. Introduction In recent years, there has been an increased interest, by both policymakers and academics, to understand the processes of innovation that underlie corporate success and international competitiveness of firms and states. This is especially true for developing countries, as the knowledge intensity of production increased worldwide and competitiveness of firms active on world markets became increasingly determined by their ability to innovate (Mytelka and Farinelli, 2003). However, also in their home markets local firms’ positions in developing countries are more and more challenged by the competition from cheap imports. Following the liberalisation of the trade regime of many least developed countries including a number of African countries, basic goods are now entering least developed country (LDC) markets, putting local firms under fierce competitive pressure. At the same time, many LDC governments engaged in a race to attract foreign direct investment (FDI), to bring new technology and managerial and production skills to the country, and to increase the capital stock. It was also expected to create opportunities for local firms to tap from the world knowledge stock and to benefit from technology transfer from foreign firms. However, evidence of positive spillovers from foreign to domestic firms in least developed countries is scarce and mixed. Too wide a technology gap between local and foreign firms seems to hinder transfer of technology.

ß The Author 2007. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved.

2 of 24

M. Goedhuys

This has led to the recognition of the importance of learning and innovation within local firms. The literature on firm learning developed over the last 20 years has identified the learning process as a costly and risky one, as firms have to invest in R&D, training, and information search, the outcome of which is uncertain. However, also the broader socio-economic and institutional context in which firms operate, the innovation system, is found important in driving innovation. Firms undertake innovation activities in collaboration with other firms and non-firm institutions to source knowledge from sources external to the firm. For least developed countries, where financial markets are poorly developed and biased towards larger and foreign owned firms, the importance of such linkages among local and smaller firms is potentially very large as they act as non-market mechanisms through which firms share knowledge and risk. While linkages are fully recognised as important determinants of knowledge flows and firm learning, the empirical evidence on learning and collaboration for innovation is very thin for least developed countries. It is also mostly based on a small number of firms or on case studies. Additionally, the role that foreign firms play in stimulating local innovation efforts is scarcely documented. This article addresses this problem by analysing the learning and collaboration behaviour of a sample of both local and foreign firms and firms of different size, using a unique data set from Tanzania. Tanzania has liberalised its economy over the past two decades, and privatised its strong direct government participation mainly in favour of foreign investment, making the study of the foreign–domestic dimension highly interesting. Using firm level data, the study first provides empirical evidence on the various sources of learning, including internal sources such as R&D, skills and training of the workforce, investment in new equipment, and telecommunication, as well as external sources through linkages and collaboration with other firms. The article subsequently relates learning and collaboration to product innovation to see how these learning activities are related to successful product innovation. The data of the World Bank’s ‘‘Investment Climate Survey’’, conducted in Tanzania in 2003, are used in this study. Though the survey was not designed as an innovation survey, a large amount of useful information could be extracted, giving valuable insights on largely unexplored questions. First, by using information from both foreign and local firms, not only the linkages between foreign and local firms are documented. Also collaboration among domestic firms on the one hand and linkages among foreign firms on the other hand are taken into account in the analysis, revealing important differences in innovation strategies between foreign and local firms. Second, the learning mechanisms and linkage intensities are also analysed directly with respect to product innovation, shedding light on the very processes that may lead to productivity differences, measured in other studies. Third, there are, to my knowledge, no studies on LDC’s linking innovation activities of firms to the eventual introduction of new products. Especially for Africa

Learning, product innovation, and firm heterogeneity in developing countries

3 of 24

this is missing, possibly due to a scarcity of innovation survey data. The rising interest for conducting innovation surveys in developing countries, including Africa,1 shows the perceived importance of product and process innovation for development and the need to understand the driving forces behind it in a developing country context. Product innovation is indeed recognised as an important form of technical change in developing countries (Fransman, 1985; Semboja and Kweka, 2001, for Tanzania), as firms need to put deliberate efforts in the search for new products suited to their markets. The structure of the article is as follows: Section 2 discusses a selection of the literature on innovation and learning for developing countries that is relevant to the empirical section and specifies the research questions to be analysed. Section 3 highlights some of the main characteristics of the Tanzanian economy relative to learning and innovation. Section 4 discusses the methodology and data and the construction of the variables. Section 5 presents the results. Section 6 concludes.

2. Innovation and learning in a developing country context: findings from the literature An extensive literature emerged since the foundations of evolutionary theory were articulated, criticising the prevailing ‘‘neo-classical’’2 theory and its assumptions for its inability to study industry dynamics (Nelson and Winter, 1982: 4; Rosenberg, 1982; Dosi et al., 1988: 120–121). Also, the empirical studies that were based on conventional models and focussed on total factor productivity growth with the ‘‘unexplained’’ residual capturing technical change were seriously questioned. The very process of innovation and technical change underlying the unexplained productivity growth remained largely unexplored and the models did not provide any significant guidance for policy making. Therefore, shifting the focus of attention and starting from the earlier documents on technical change in developing countries [including, among others, Fransman (1985), Katz (1987), Lall (1992)3], research of 1

The first NEPAD Ministerial Conference on Science and Technology requested the NEPAD Secretariat to undertake activities that would generate an ‘‘African Innovation Outlook’’ (AIO), that is, a comprehensive data set that allows profiling the innovation landscape. As a basis for an AIO two distinct but complementary surveys, one on science and technology and one on innovation, were designed (UNU/INTECH, 2004). 2

Dosi et al. (1988) outlines what is meant with ‘‘neo-classical theory’’ and criticised especially the assumptions of rational agents, equilibrium thinking, and the weak incorporation of uncertainty in the models.

3

Lall developed the concepts of ‘‘technological capabilities’’ to study the dynamics of technological learning in firms that are operating far from the frontier technology. Other researchers built further on this concept, resulting in a large body of literature, an overview of which is given in UNCTAD (1996).

4 of 24

M. Goedhuys

the last 20 years has explored the technological learning processes that drive innovation in firms. This research revealed that importing new technologies from abroad is not sufficient to generate productivity growth as technologies developed elsewhere have to be adapted to or mastered by LDC firms. This requires technological learning which is costly and risky. Often it involves a tacit component that needs to be acquired through experience and use, and through the development of new technological capabilities. It requires a financial commitment on behalf of the firms to master technology or build up competences and skills, through R&D, training, engineering activities, and information search, the outcome of which is uncertain and bears considerable risk (Fransman, 1985: 575–576; Katz, 1987; Malerba, 1992). Another central finding is that innovation activities within firms also depend heavily on sources external to the firm. Learning and innovation are interactive and systemic processes rather than a linear process driven by R&D. Firms do not innovate in isolation but in collaboration and interaction with clients, other firms and non-firm agents (Edquist, 2004). Lundvall (1988) showed that user–producer interactions can be important for successful product innovation. The widely used concept of ‘‘innovation systems’’ is based on the idea that the innovative performance of firms is not only determined by their own efforts, but is strongly influenced by the wider institutional and socio-economic framework in which their activities are embedded (Cassiolato et al., 2003; Edquist, 2004; Fagerberg, 2004). The interactive nature of learning is also a key element in the more recently developed literature on clustering, which studies knowledge flows between firms in geographical or sector agglomerations. Schmitz and Nadvi (1999) introduced the concept of ‘‘collective efficiency’’ to refer to the competitive advantages of geographical or sector agglomeration, derived from external economies and joint action. External economies are the benefits firms enjoy of being merely located geographically close, including the availability of skilled and specialised labour, specialised suppliers, superior provision of utilities and infrastructure, and improved access to markets and information (Mccormick, 1999). Joint action on the other hand refers to intentional collaboration, linkages, and networking of firms in the cluster. They can reduce the costs related to information and communication, the time needed and risks associated with the introduction of new products. Bell and Albu (1999) developed a conceptual framework to investigate clusters capacity to generate and diffuse knowledge based on their openness to external knowledge sources and the intensity of collaboration between firms. More recent studies go on modelling and analysing the relationship between clustering and intra-firm learning, and they look at the different role individual cluster firms play in absorbing extra-cluster knowledge and diffusing

Learning, product innovation, and firm heterogeneity in developing countries

5 of 24

knowledge within the cluster (e.g. Caniels and Romijn, 2003; Giuliani and Bell, 2005).4 Inter-firm knowledge flows are also studied in the empirical literature that focuses on knowledge spillovers from foreign owned to domestic firms in host countries. This is mostly done by analysing productivity differences of firms over time, at different levels of exposure to foreign investment in the same sector or in supplier or user sectors. While for developing countries there seems to be no robust empirical evidence of horizontal spillovers—knowledge spillovers from foreign firms to competing local firms in the same industry—(Aitkin and Harrison, 1999; Djankov and Hoekman, 2000; Konings, 2001), evidence of spillovers through backward linkages—from foreign firms to their local suppliers—is more likely to be observed and is indeed found for some countries (Blalock, 2001; Schoors and van der Tol, 2002; Javorcik, 2004). The innovation process that leads to this productivity shift and the importance of local knowledge flows between domestic firms is in these studies not taken into account. The costly and interactive nature of technological learning has important implications for firms in LDCs, where financial markets are typically poorly developed. Some authors have shown that in LDCs, financial markets are biased against SMEs and favouring larger and foreign owned firms that enjoy a more legitimate status in the industry, facilitating their growth and investment opportunities (e.g. Nugent and Nabli, 1992; Harrison and McMillan, 2002; Sleuwaegen and Goedhuys, 2002, 2003; Beck et al., 2005). This institutional aspect of the business environment affects firms’ opportunities for learning and their choice of innovation strategy. The more financially constrained micro-enterprises and small and medium sized firms may find themselves in a disadvantages position to engage in learning efforts that require a strong financial commitment. Linkages can then become an even more important non-market mechanisms to share information, knowledge, investment, and risk, and they may offer unique opportunities for local SMEs to engage in a process of continuous learning and innovation. This study integrates these findings of the literature into the empirical analysis. It analyses differences in learning mechanisms and collaboration intensity of firms in Tanzania, to see subsequently how these translate into successful product innovation. The focus of the analysis is threefold.

4

Interesting contributions include Caniels and Romijn (2003) who integrate the meso-perspective of ‘‘collective efficiency’’ studies and the micro-economics perspective of technological learning at the firm level. Also Giuliani and Bell (2005) illustrate the different roles played by firms in a Chilean wine cluster with respect to the acquisition of knowledge external to the cluster, and their contribution to the intra-cluster diffusion of knowledge, driving cluster performance. Cassiolato et al. (2003) present selected case studies from Brazil, discussing the conditions under which learning and the use of capabilities occur in Brazilian clusters.

6 of 24

M. Goedhuys

First, though it is not possible to trace from the data whether firms belong to a cluster and how they interact with other cluster firms, it is possible to analyse whether more intensive collaboration with other firms in the local market is associated with the introduction of new products. This sheds light on the benefits of interaction and the possible effects on firm learning in an African country for which the empirical evidence is really thin. Second, the impact of linkages with foreign firms and interaction through subcontracting relationships will be analysed empirically. It is indeed likely that domestic firms, becoming suppliers of foreign firms, are asked to produce specific goods. This may lead to activities in the firm that result in new product innovation. The analysis thus sheds light on the very processes that may lead to productivity differences, as measured in other studies. Third, when financial markets are poorly developed, SMEs may rely more heavily on linkages and less on learning efforts with a strong financial impact. The learning mechanisms are thus first analysed in relation to size and ownership.

3. Tanzania Tanzania is a typical least developed country, still heavily based on agriculture, which accounts for 45% of GDP in 2003 (World Bank, 2005), while industry accounts for 16.4% of GDP, including an important share of mining activities in the North. The country has been undergoing reforms since 1985 when it started to move away from a socialist centrally planned economy with large state participation towards a more liberal market-based economy. Over the last decade, a large scale privatisation programme was implemented, in which state participation in industry was reduced, mainly in favour of foreign participation. Foreign direct investment increased sharply since 1992,5 putting Tanzania among the major FDI recipient countries of Africa (UNCTAD, 2002). FDI was mainly concentrated in manufacturing, especially in the food and beverages industry. The government of Tanzania expected FDI to bring along an injection of capital and investment, and a transfer of technology, skills, and superior management techniques that would benefit the country at large. There is no doubt that the recent foreign investments have increased the stock of technology, especially the embodied technology such as machinery and equipment. Using a case study of two manufacturing firms, Portelli and Narula (2003) find evidence that technological upgrading has occurred and that its magnitude is determined by the capabilities within the industrial base in Tanzania. Narrower technology gaps between firms are more likely to result in backward linkages. However, as some studies indicate, the scale of technology diffusion from foreign 5

In 1992, FDI was US$ 12 million, but it rose to US$ 193 million in 2002.

Learning, product innovation, and firm heterogeneity in developing countries

7 of 24

firms to local firms is still limited. Transfer of know-how, design, and R&D capabilities is not observed (UNCTAD, 2002: 18). According to Szogs (2004) the linkages between foreign and local private firms remain weak. At the origin of this lies the low level of human resources and technological capabilities in Tanzanian firms. The educational and training systems have been insufficiently oriented towards science and engineering, generating managerial, and technical skills. In general, the technology policies pursued in Tanzania paid little attention to the technological needs and problems of local private firms (UNCTAD, 2002; Utz, 2006). Domestic research capability was built in public research centres, doing research in priority areas determined by the Tanzania Commission for Science and Technology. The choice of sectors and research areas was supply driven, rather than based on an analysis of technological needs and problems of productive private enterprises. Some state-owned technology support institutions were established, but today they lack awareness of private sector needs as well as resources and motivation to carry out their mandate successfully (Lall, 1999; Bongenaar and Szirmai, 2001; Utz, 2006). The linkages between industry and university and research institutions are weak, as described in Bangens (2004) and Mwamila and Katalambula (2004). Under these constraints it can be expected that product innovation in local firms is mainly taking place as a result of internal learning and inter-firm linkages among domestic firms, as linkages with research centres and support institutions are weak. This is also found by Murphy (2002) who points at the importance of networks and trust for innovation in Mwanza, Tanzania. He found that social networks of business people support innovation, as trust in these relationships improve the quality of information exchanges.

4. Data and methodology 4.1 Data The data used for the empirical analysis were gathered in Tanzania in 2003, in the framework of the World Bank programme ‘Investment Climate Survey’ (ICS). Survey data were collected through intensive interviews with owners and managers of firms. The objective of the ICS was to obtain firm level data that allow analysing the conditions for investment and enterprise growth in the country. Hence, the many aspects and constraints of the African business environment that influence the investment decisions and performance of the firms were tackled, in a number of sub-questionnaires. Survey data are interesting for analysing innovative behaviour of firms, especially in Africa where the traditional measures for innovation such as R&D intensity and patenting reveal little information. Since innovation surveys were

8 of 24

M. Goedhuys

developed and the OECD’s ‘Oslo Manual’ (OECD, 1992 and 1997) appeared as a common practice in this field, a new type of innovation analysis emerged, focussing on firm level innovation activity. The surveys are designed to gather information on both innovation inputs, including also non-R&D activities with a strong learning component, and innovation outputs: product and process innovation. They focus on the innovating firm, the ‘subject’ of innovation who is the respondent to the surveys. Inherent to this type of data collection and the resulting indicator construction is a certain degree of subjectivity in the data, which also holds for the ICS. Despite this criticism, a number of interesting studies based on innovation survey data appeared, measuring and mapping innovation, and linking it to firm performance (see e.g. Smith, 2004 for an overview of studies). Scholars increasingly use the data to increase their knowledge on the very process of innovation as it takes place within the firm. The Investment Climate Survey sampled a heterogeneous group of firms for the interview in Tanzania, including firms of different size, ownership structure and active in manufacturing and commercial agriculture.6 A total of 276 firms in manufacturing and 98 firms in commercial agriculture were surveyed. Due to missing responses to some of the variables used, the actual sample of firms used in the empirical analysis is reduced to 257, distributed over the different sectors of activity and size classes as shown in Table 1. The number of firms with foreign participation, 60 in total, is also presented per sector. They are mostly large (23 firms) and medium sized (19 firms). Twenty foreign owned firms are owned by a foreign company. The other foreign owned firms belong to foreign individuals or families, about half of which belong to the large Asian (Indian) minority active in Tanzanian trade and industry. Fifteen foreign firms, or about one quarter, were previously state-owned firms and later on privatised.

4.2 Methodology The rich data set created by the World Bank ICS includes a large amount of variables that capture learning, investment, linkages, and innovation. In the literature, there are a number of empirical studies that proceed to construct composite indexes measuring technological capabilities or technological intensity (see Wignaraja, 2002 and Rasiah, 2004 for an overview), subsequently linking the index to exports or other firm performance indicators. I have deliberately not chosen to proceed with the construction of an index, as I am specifically interested in the different learning 6 The Investment Climate Survey was also conducted for firms active in construction and tourism. However, the questionnaires used in these sectors did not include the key questions on innovation. Hence in our study, manufacturing and commercial agriculture are the sectors retained for the empirical analysis. The latter sector includes the produce of tea, coffee, cocoa, flowers, livestock, cotton, vegetables, grains, sugar, and other products.

Learning, product innovation, and firm heterogeneity in developing countries

9 of 24

Table 1 Distribution of the sample of firms, by employment size class, sector and foreign ownership Numbers of firms

Micro

Small

Medium

Large

1–9

10–29

30–99

100þ

Foreign

20

18

9

8

14

Total sample

Sector of activity Commercial farming

55

Manufacturing sub-sectors Agro-industries

10

17

18

12

15

57

Textiles, garments, leather

5

10

6

5

6

26

Wood working, furniture Chemicals and paints

13 3

17 6

5 9

4 4

5 7

39 22

Plastics

0

1

2

4

4

7

Paper, printing, publishing

1

11

7

1

2

20

Metal working

6

8

7

1

4

22

Construction materials

1

3

3

2

3

9

59

91

66

41

60

257

Total sample

mechanisms and their importance for firms operating under different financial and competitive conditions. The differences would be levelled out if an index would be constructed for the entire sample of firms. Therefore, the various sources of learning are first discussed separately. This is done by presenting a number of key variables capturing skills, training, research and development, connectivity, investment, linkages and collaboration along the firm size, sector of activity, and ownership dimension. Some statistical tests are done to investigate the existence of a relationship between the specific learning or linkage activity and the firm characteristics. Subsequently, product innovation is modelled following a probit model that relates the probability of being an innovative firm to the characteristics of the firm and the underlying learning activities: INNO
i ¼ a Xi þ bLi þ gIi þ dCi þ ei INNOi ¼ 1 if INNO
i  0 INNOi ¼ 0 if INNO
i 50 where Xi is a vector with relevant firm characteristics, including firm size and age, foreign ownership and sector of activity. Li is a set of learning or technological capability variables, including the skills level of the work force and training, access to the Internet and R&D. Ii is investment in new machinery and equipment and Ci represents collaboration or linkages with other firms for product development. The estimation is also done for local and foreign firms separately, to uncover potential differences in their innovation strategies.

10 of 24

M. Goedhuys

4.3 Variables Table 2 presents the construction of the variables for the empirical analysis and some summary statistics: the mean value and, for continuous variables, also the standard deviation. In line with the discussion above, the firm’s human capital stock is measured by the skills level of the labour force (SKILLS), and formal training (TRAINING). Connectivity is measured by access to the Internet in the firm (INTERNET), the use of an own website (WEBSITE) and the use of email for interacting with clients and suppliers (EMAIL). Internal efforts further encompass R&D (RD) and investment in new machinery and equipment (NEWEQ). With respect to linkages and collaboration, the firms were also asked to rate the intensity of collaboration with other local firms for product development, by indicating on a Likert scale the frequency of interaction, ranging from 1 (always) to 6 (never). From this, the variable was restructured so that higher values indicate higher intensity of collaboration and rescaled to vary between 0 and 1. Linkages through subcontracting (SUBCONTR) and backward linkages with foreign firms (BACKLINK) are also measured (Table 2). In the probit estimation, the dependent variable is a binary variable for firms that report to have introduced a major new product line in the period 2000–20027. In this estimation, the impact of R&D is measured by R&D intensity, to capture the cost of innovation. Likewise, the value of the investment over 2000–2002 is used instead of the dummy to come to grips with the costs related to innovation efforts. Firms are considered as foreign (FOREIGN) if there is foreign participation in the ownership structure. However, the group of foreign firms includes affiliates of multinational firms as well as entrepreneurial firms owned and managed by a foreign individual or family, mostly of Asian or Middle-Eastern origin. To look into differences of these groups of firms, the probit analysis includes the dummy for firms owned by a foreign company. Additionally, to control for sector variations, the probit estimation also includes a set of sector dummies, the reference sector being firms active in commercial farming.

4.4 Estimation Before proceeding to the discussion of the results, a number of comments need to be made regarding the probit estimation. 7 Product innovation is the introduction of products that are new to the firm. In line with the studies based on Community Innovation Survey data, the analysis does not focus on innovation through upgrading existing product lines, nor on process innovation or innovation through organisational change, due to the generally more stringent definitional concerns about what is meant by ‘new’ or ‘improved’.

Learning, product innovation, and firm heterogeneity in developing countries

11 of 24

Table 2 Construction of the variables and summary statistics Variable

Definition

Mean (Std)

SKILLS

(number of professional and skilled production workers)/

2.18 (3.33)

(number of unskilled and non-production workers) TRAINING

¼ 1 if the firm has offered formal training to its permanent

0.40

employees, in 2002. RD

¼ 1 if the firm has invested in design or R&D in 2002.

0.20

WEBSITE

¼ 1 if the firm has a website that it uses for interacting with

0.21

clients and suppliers. INTERNET EMAIL

¼ 1 if the firm has internet access. ¼ 1 if the firm regularly uses email for interacting with clients

NEWEQ

¼ 1 if the firm has invested in new machinery and equipment

0.47 0.56

and suppliers. 0.44

over the period 2000–2002 COLLAB

Intensity of collaboration with other local firms for product

0.19 (0.29)

development, [0,1] SUBCONTR

¼ 1 for firms that produce as subcontractor for other firms

0.12

BACKLINK

proportion of output that is sold to multinationals located in Tanzania

0.03 (0.14)

INNO

¼ 1 if the firm has introduced a major new product line in

0.32

2000–2002 FOREIGN

¼ 1 if there is foreign ownership participation

0.23

MNE

¼ 1 if the owner is a foreign company

0.08

LEMPL

Total number of employees, in 2002, in logarithmic terms

3.75 (1.49)

RDINTENSITY

R&D expenditures/sales, for 2002, expressed in%

0.87 (8.02)

LNEWEQ

Value of investment in new machinery and equipment over 2000–2002, in logarithmic terms

5.80 (4.43)

The equation was subject to simultaneity bias, especially with respect to R&D and training, both of which refer to 2002, and to the investment in new machinery and equipment, that is measured over 2000–2002, the same period over which product innovation is measured. Ideally, the use of lagged variables could solve this problem and facilitate an interpretation on the direction of causality. Unfortunately, these data are not available, and there is no information at hand to reconstruct them somehow. Therefore, to correct for the simultaneity bias and given the limited number of observations in the sample, the endogenous variables were instrumented and their predicted values were used in the probit model on product innovation (following Maddala, 1983). The incidence of offering training to the employees (TRAINING) was estimated in a probit using as instruments the presence of a labour

12 of 24

M. Goedhuys

union in the firm and the education of the general manager. For R&D intensity (RDPERSAL), the profitability of the firm is used (following Klette and Griliches, 2000: 376), measured by the value of sales minus the purchase of raw materials, energy cost and labour costs, and over sales. The value of investment in new machinery and equipment (LNEWEQ) was instrumented by a binary variable for firms that have an overdraft facility with their banks, thus measuring firms’ access to flexible forms of liquidity or credit. Firm size, ownership, sector and location variables were used as additional explanatory variables in all equations, as well as one binary variable equalling one if the firm reported not to be credit constraint. It also has to be mentioned that due to high correlation between the connectivity variables, only INTERNET was included to prevent multicollinearity problems from obscuring the real impact of the internet.8 Admittedly, the results still need to be interpreted with caution, especially with respect to the causal relationship between the variables. Nevertheless, it is still interesting to note the positive correlations between particular innovative activities and strategies and product innovation and this for different types of firms, as will be demonstrated in the next sections on the results.

5. Empirical results 5.1 Learning and linkages in Tanzania The different learning mechanisms and their importance for firms of different size, sector and ownership structure are shown in Table 3. For binary variables, a chisquare statistic and its significance are reported. As SKILLS is a continuous variable, a T-test on differences in mean values was done, for the respective subgroup of firms versus the rest of the sample. The significance of this test is indicated after the mean values of the respective subgroup. Looking first at the size dimension, it is clear that all of the learning variables are strongly and significantly related to size, with higher skill levels, training, and R&D activities and higher occurrence of connectivity and investment in larger firms than in smaller ones, and the relationship is gradual. In a similar way, foreign firms outperform their local counterparts in every way. The difference is significant for all the variables, except for the skills level of the workforce and the incidence of R&D or design activities. This is in line with the findings discussed by UNCTAD (2002), which indicated that the transfer of knowhow, design, and R&D capabilities is weak in Tanzania. Foreign firms invest more in new machinery and equipment, they are better connected to the Web, and train 8

Correlation coefficients being 0.80 for internet and email and 0.47 for Internet and website. The use of internet is indeed systematically in between the use of email, which is most common, and the ownership of an own website, which is least common, as shown in Table 3.

Learning, product innovation, and firm heterogeneity in developing countries

13 of 24

Table 3 Sources of internal learning and investment, by employment size class, ownership and sector of activity (1)

(2)

(4)

(5)

(6)

(7)

SKILLS

TRAINING RD

WEBSITE

INTERNET

EMAIL

NEWEQ

% of

% of

% of

% of

% of

% of

firms

firms

firms

firms

firms

firms

Mean

(3)

Size 1–9

1.68*** 17.0

10–29 30–99

1.90 2.30

28.6 60.6

100þ

3.34* n.a.

68.3 48.8 39.0 78.1 87.8 68.3 43.179*** 32.158*** 19.486*** 65.661*** 65.980*** 16.611***

Local

2.01

34.0

18.8

17.8

40.6

49.2

39.1

Foreign

2.76

61.7

23.3

33.3

68.3

80.0

60.0

X2

n.a.

14.602***

X2 Ownership

Sector Agro-industries 2.17

6.8

6.8

8.5

17.0

28.8

12.1 24.2

16.5 30.3

41.8 69.7

52.8 77.3

39.6 48.5

0.599

6.626**

14.189*** 17.700***

8.165***

49.1

21.1

24.6

61.4

68.4

47.4

2.80

58.6

34.5

37.9

58.6

72.4

58.6

1.77

31.0

6.9

24.1

62.1

62.1

41.4

Metal working 4.21*** 36.4

27.3

13.6

45.5

59.1

50.0

Wood working 1.55** Textiles 3.05

30.8 57.7

15.4 23.1

20.5 26.9

35.9 53.9

38.5 69.2

35.9 50.0

1.31*** 27.3

16.4

9.1

23.6

38.2

34.6

Chemicals & plastics Publishing & construction materials

Commerical farming X2

n.a.

15.663**

8.856

11.411*

23.450*** 21.088***

6.596

SKILLS (column 1) is a continuous variable. Instead of a X2 statistic, a T-test was done, for the difference in mean value of each size category, ownership and sector, versus the rest of the sample firms. The significance of this T-statistic is reported. Note: *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

the workforce more intensively. Additional information from the survey indicated that on average 9% of the workforce is trained in foreign firms, against only 2% in local firms. Some sector variations are also observed, with the highest incidence of formal training, R&D, connectivity and investment being found in the sector of chemicals

14 of 24

M. Goedhuys

and plastics. The metal working and textiles sectors are also doing relatively well, in contrast to the sector of wood working and commercial farming, which are performing generally poorly on the learning and investment indicators. Table 4 gives information on the firms’ linkages to other firms. The first column presents the reported intensity of collaboration with local firms in the field of product development. The proportion of firms that is doing subcontracting work is shown in column 2, and for those firms that are working in subcontracting relationships, the proportion of their production produced under this subcontracting relationship is also shown in column 3. When looking at linkages, the picture is less clear than for the internal learning variables. Local firms report to collaborate more intensely with other local firms than do foreign firms, but the difference is not significant. Also the size dimension does not reveal a strong link with collaboration. The same holds for subcontracting relationships, which prove to be established independently of size, sector or ownership. Only a small number of large and foreign owned firms are active as subcontractor, yet this activity accounts for a large share of the firms’ production. Interestingly, backward linkages established by supplying foreign firms seem stronger among foreign owned firms, than between foreign and local firms, a finding that will be explored more when analysing its impact on product innovation. The relation between size and the proportion of firms that innovate (column 5) follows an inverted U-shape, and foreign firms also have a slightly higher propensity to innovate. It is highest in the chemicals and plastics industry, followed by metal working. The findings of Tables 3 and 4 jointly indicate that internal learning activities that require certain financial commitments, involving risk, are indeed found to be related to size and foreign ownership. In a developing country like Tanzania, smaller firms that face more severe financial constraints in imperfect markets instead rely on networking and collaboration with other firms, to acquire information on markets to improve or upgrade products and production processes. In the survey, firms were also asked to indicate the three major ways of acquiring new technology, out of a list of 14 ways, so as to get the managers’ perception on the relative importance of the various sources for new knowledge. Though this question addresses the subjective view of the respondents and the question was not directly focussing on acquisition of knowledge for product innovation, the analysis of the responses does show interesting additional and corroborating evidence on how small versus large, domestic versus foreign firms learn. For all the 14 ways to acquire technology, an average score could be calculated,9 which is presented in Table 5. As could be expected, the investment in newer vintage 9

For each firm, the most important way to acquire technology was subsequently given a weight of three, the second most important way was given a weight of two, the third a weight of one, the other options keeping a zero rating.

Learning, product innovation, and firm heterogeneity in developing countries

15 of 24

Table 4 Collaboration, linkages and product innovation, by employment size class, ownership and sector of activity (1)

(2)

(3)

(4)

(5)

BACKLINKa

PRODUCT

% of COLLABa

SUBCONTR

subcontracted productiona

Size 1–9

INNO

Mean

% of firms

Mean

Mean

% of firms

0.16

15.3

22.1

0.25***

13.6

10–29

0.20

9.9

11.8**

6.90***

30.8

30–99

0.17

18.2

39.8*

2.18

48.5

100þ

0.22 n.a.

51.5 n.a.

2.43 n.a.

36.6 17.826***

X2 Ownership

4.9 5.118

Local

0.20

13.2

25.2

2.55

29.4

Foreign X2

0.15 n.a.

10.0 0.432

38.4 n.a.

6.40* n.a.

41.7 3.144*

Agro-industries

0.18

12.3

50.4

2.11

24.6

Chemicals & Plastics

0.04***

10.3

38.3

1.72

55.2

Publishing &

0.19

10.3

9.5

3.97

27.6

Metal working

0.24

27.3

15.3

8.45

40.9

Wood working Textiles

0.24 0.09***

12.8 15.4

24.2 11.0

1.36** 7.04

33.3 38.5

Commercial farming

0.26** n.a.

33.0 n.a.

3.27 n.a.

23.6 11.898*

Sector

construction materials

X2

7.27 6.234

a

a T-test was done, for the difference in mean value of each size category, ownership and sector, versus the rest of the sample firms. The significance of this T-statistic is reported. Note: *significant at the 10% level; **significant at the 5% level; ***significant at the 1% level.

machinery and equipment is by far the most important way to acquire new technology. This is followed by technology developed in-house and through the hiring of key personnel, as can be seen from the magnitude of the score presented in the last column. However, some ways of acquiring technology are positively related with size, such as, investment in machinery and equipment, the hiring of key personnel, and technology transferred from the parent company. Inversely related to firm size are technology developed in-house, trade fairs, technology adapted from the

16 of 24

M. Goedhuys

Table 5 Relative importance of different ways to acquire new technology, by firm size and ownership Average score

Micro

Small

Medium

Large

Local

Foreign

All firms

New machinery

0.80

1.01

0.97

1.43

0.92

1.33

1.02

Developed in-house

1.22

0.90

0.94

0.64

1.03

0.67

0.94

Hiring key personnel

0.34

0.45

0.90

1.05

0.46

1.21

0.63

Trade fairs

0.44

0.35

0.34

0.33

0.41

0.23

0.37

Consultant

0.34

0.33

0.24

0.52

0.34

0.34

0.34

Adapted from competitors

0.47

0.46

0.13

0.10

0.36

0.20

0.32

Cooperation with suppliers Study tours

0.10 0.14

0.33 0.29

0.45 0.13

0.07 0.12

0.29 0.21

0.20 0.11

0.27 0.19

Cooperation with clients

0.02

0.27

0.07

0.17

0.15

0.15

0.15

Parent company

0.00

0.09

0.24

0.31

0.10

0.28

0.14

University

0.05

0.17

0.10

0.00

0.11

0.07

0.10

Business association

0.10

0.07

0.04

0.00

0.05

0.10

0.06

Licence from intl. Sources

0.03

0.02

0.06

0.17

0.06

0.05

0.06

Licence from domestic sources

0.03

0.03

0.00

0.07

0.04

0.00

0.03

competitors and through business associations. In other words, smaller firms acquire technologies in ways that are less resource intensive and more relying on the other agents in the system. Likewise, it can be seen that domestic firms are more actively working internally through in-house activities, study tours, and trade fairs, in cooperation with suppliers, copying competitors, and sourcing from universities than foreign firms, which implies they are more strongly rooted in the local productive structure. Foreign firms on the other hand rate investment, the hiring of key personnel and sourcing from the parent company as major ways of acquiring technology, revealing a better financial endowment and a smaller involvement in the local industrial structure.

5.2 Learning, linkages, and product innovation The impact of these different learning and collaboration mechanisms on product innovation is shown in Table 6. Not the estimated coefficients of the variables are shown, but the marginal effects, so that the magnitude of the effects can be seen from the table. For dummy variables, the reported figure reflects the impact of the discrete change of the dummy variable from 0 to 1. For continuous variables, the change in the probability of being a product innovator for an infinitesimal change in the continuous variable is presented (dF/dX). The significances refer to the significance of the test of the underlying coefficient being 0.

Learning, product innovation, and firm heterogeneity in developing countries

17 of 24

Table 6 Results of the probit model on the probability of being a product innovating firm All firms dF/dX

All firms SE

dF/dX

SE

0.053

(0.050)

Local firms

Foreign firms

dF/dX

SE

dF/dX

SE

0.032

(0.052)

0.241

(0.132)*

LFIRMAGE

0.001 (0.035)

LEMPL

0.365

LEMPL sq.

0.032 (0.012)**

–

FOREIGN

0.003 (0.081)

0.019

(0.094)

–

–

MNE

0.014

0.206 (0.102)

–

–

SKILLS TRAINING

– –

0.021 0.158

(0.010)** 0.013 (0.013) (0.354) 0.097 (0.380)

INTERNET

–

0.187

(0.073)** 0.237

(0.087)*** 0.147 (0.254)

RDPERSAL

–

0.041

(0.018)** 0.046

(0.020)**

0.057 (0.042)

LNEWEQ

–

0.033

(0.015)** 0.029

(0.017)*

0.111

COLLAB

–

0.213

(0.101)** 0.282

(0.101)*** 0.471 (0.319)

BACKLINK

–

0.305

(0.191)

0.004

(0.270)

1.528

(0.509)***

SUBCONTR

–

0.195

(0.095)** 0.150

(0.106)

0.353

(0.224)

Metal Paper

0.208 0.019

(0.133) (0.115)

0.020 (0.149) 0.023 (0.119)

0.186 0.037

(0.184) (0.140)

0.482 (0.105)* 0.356 (0.192)

Textiles

0.118

(0.121)

0.059

(0.163)

0.231

(0.205)

0.394 (0.208)

Wood

0.186

(0.114)*

0.153

(0.127)

0.202

(0.156)

0.288

Chemicals

0.239

(0.125)**

0.025

(0.163)

0.071

(0.208)

0.251 (0.275)

(0.107)*** 0.026 (0.043)

(0.128)

0.000 (0.050)

0.220 (0.125)*

–

–

0.122 0.809

(0.048)*** (1.080)

(0.045)**

(0.328)

Agro-ind.

0.052 (0.092)

0.113 (0.109)

0.010 (0.146)

0.529 (0.230)

Obs.

257 0.110

257 0.191

197 0.210

60 0.356

MC Fadden’s R2

Note: Marginal effects are reported, standard errors in parentheses; Reference sector is commercial farming; Paper includes paper, publishing, printing, and construction materials, Textiles includes textiles, garments, leather products, Wood includes wood working and furniture, Chemicals includes chemicals, paints and plastics. *** significant at the 1% level; **significant at the 5% level; *significant at the 10% level.

The most interesting result that can be derived from the basic estimation without learning variables (column 1) is with respect to firm size. Firm size has a non-linear effect on the propensity to innovate. This is also found when the estimation was repeated including size dummy variables. Then, medium sized firms (30–99 employees) are found the more innovative ones, increasing the probability of being a product innovator by 39%, followed by large (100þ employees) and small firms (10–29 employees) with a 26 and 22% increase in the probability of being a product innovator, respectively. Bearing in mind from the previous section that investment and learning activities were more pertinent in the largest size class,

18 of 24

M. Goedhuys

medium and small firms seem to have found ways to offset this disadvantage. Microenterprises, scoring low on both learning, and investment and linkage indicators, are clearly the least innovative firms. From the probit analysis, foreign firms do not seem to be more innovative than local firms, a finding that contrasts with the bi-variate results from the previous section. Also the distinction between foreign firms owned by foreign individuals and families on the one hand and multinationals on the other hand does not show any significant result when firm size and age and sector of activity are controlled for. The sector of chemicals and plastics shows superior performance in product innovation (þ24% increase in the probability as compared with commercial farms), as expected from the previous section, followed by metal working (þ21%) and wood working (þ19%). The inclusion of the learning and linkages indicators (column 2) provides additional interesting findings. The size and sector effects turn insignificant, as learning and linkage effects take over in explaining differences in innovative propensity. The learning variables all have the expected impact, the underlying coefficients have the right sign. Apart from backward linkages and training, they are all significant showing that successful product innovation is related to both internal efforts and investments as well as knowledge sourced from external sources. Strikingly, when splitting the sample and doing the estimation for local (column 3) and foreign firms (column 4) separately, the mechanisms that come to play a role for product innovation appear to be very different for both groups of firms. While local product innovators are mainly characterised by them having access to the Internet, by their superior internal research and development efforts and more intense collaboration with other local firms, foreign firms’ innovative propensity is not driven by any of these factors. Rather, foreign firms that introduced new products are mainly characterised by superior skill levels of their workforce in terms of more professional and skilled workers, by a larger proportion of their output sold to other foreign firms active in Tanzania and more investments in machinery and equipment. Hence, the idea that being a supplier of foreign companies active in Tanzania can conduce firms to develop new products in that context seems to hold only for firms that are also foreign owned. This finding points at stronger linkages among foreign owned firms, than between foreign and domestic firms. Evidence that the gap between foreign and local firms in Tanzania is large has already been described in previous studies, but it is striking that it shows up again in a much larger sample. It would have been interesting to repeat this estimation for MNEs only instead of foreign firms only, to see the impact of minority groups and their linkage behaviour, but the sample size gets too small to get meaningful results. Domestic firms, on the other hand, seem to be more embedded in the local structure, with more intensive collaboration being related to product innovation. This finding illustrates the importance of the interactive nature of the innovation

Learning, product innovation, and firm heterogeneity in developing countries

19 of 24

process. Indeed, by collaborating, and by performing in-house R&D, they succeed in performing equally well in terms of introducing new products to the market than their foreign counterparts. Also for local firms, larger levels of investment in new machinery and equipment are positively related to innovation, revealing the innovation process as a costly and risky undertaking. For foreign firms, the impact of the level of investment is even larger in explaining the probability that the firm is a product innovator, pointing at the better financial resource endowment that generally characterises foreign firms as compared to local firms in developing countries.

6. Summary and conclusion Recent findings in the literature stress the importance of technological learning for innovation, through in-house efforts and through collaboration with other actors in the system. In developing countries, where markets for finance, technology, and information are highly imperfect and more favourable to larger firms, inter-firm collaboration is likely be an even more important mechanism for SMEs to overcome the constraints related to the inputs into the innovation process. The role of FDI herein is also studied and the existence of spillovers from foreign to local firms investigated, which is a relevant research question, given the eagerness of LDC governments to attract FDI. However, mixed results of spillovers and the persistence of a technology gap between local and foreign firms are found in the literature. Using a unique data set, this article has analysed the learning processes and linkage behaviour of small and large, local and foreign firms active in Tanzania, and has investigated the effect of different learning mechanisms on product innovation. The evidence supports the idea that collaboration can indeed enable local SMEs in developing countries to be product innovators, even when they show to invest less in new machinery and to engage less in training and research, development and design activities than their larger or foreign counterparts, probably due to more severe financial constraints. Local firms prove to be more embedded in the domestic industrial structure as they indeed innovate in collaboration with other local firms and through internal R&D, and by sourcing information from the Internet. Supplying foreign firms active in Tanzania does not seem a strong enough linkage to lead to the introduction of new products, pointing at the limited role of foreign firms in stimulating product innovation in domestic firms. Foreign firms are equally product innovators, yet the mechanisms leading to product innovation are different: they hire better skilled personnel, train them more intensively, develop products for other foreign firms, and invest heavily in machinery and equipment. As the linkages of foreign firms with the local industrial base remain weak, impact of foreign investment on the level technical competences in Tanzania mainly runs

20 of 24

M. Goedhuys

through its contribution to human capital formation. This finding that is in line with the evidence presented by Blomstro¨m and Kokko (2003). Foreign firms train the labour force intensively to enable them to operate new vintage machinery and equipment. Foreign firms may also provide the so-needed job opportunities for highly skilled Tanzanians. In this way, they can help smoothening the brain drain problem that marginalises the LDCs that invest in formal education without reaping the benefits of the investment (Archibugi and Pietrobelli, 2003; Utz, 2006; Wamae, 2006). However, there is potentially more to gain from foreign investment than the mere human capital development within foreign production facilities. Public policy should be oriented at stimulating technological learning in the local industrial base by fostering inter-firm collaboration and partnerships with foreign firms. This may require investigating the supply side constraints of domestic firms first and identifying their needs for technological and managerial skills development and education. To strengthen networks with support and research institutions, it may be essential to evaluate the effectiveness of the public research agenda and to put research priorities in line with private sector needs. Putting in place information mechanisms to foreign investors on local production technology and designing incentives for technological partnerships between foreign firms and local organisations can be expected to considerably benefit the country and reduce the observed innovation system weaknesses. Finally, the analysis raises an alternative question that could not be fully explored with the used data set, namely: who benefits from product innovation in a poor country? Often foreign firms develop new or improved products for export markets or a small upper segment of the local market and the evidence presented here seems to confirm this. Local firms are more likely to be familiar with the needs of the poor and this puts into perspective the value of product innovation by foreign versus local firms, giving more weight to the latter for their focus on local needs. However, a recent and controversial strand in business literature, inspired by Prahalad (2005), describes the enormous opportunities and successful experiences of MNEs that have started to serve the mass of poor in the lowest income classes, termed the ‘‘bottom of the pyramid.’’ Using case studies it is shown that producing and distributing innovative products and services tailored to the needs and problems of the poor can be potentially very rewarding, due to the magnitude of the market that can be served. To be successful MNEs not only have to develop highly innovative low cost products to satisfy specific needs of the poor, they need to form new alliances with governments, local firms, cooperatives, and NGOs for their superior insights into local culture and their knowledge of the market. In that respect, policy makers could consider articulating local research needs and priorities in certain productive sectors, establish supportive relations with foreign firms and organise the local community so that they can participate and benefit from such a process. The value of product innovation in developing countries and the role of foreign and local firms is still

Learning, product innovation, and firm heterogeneity in developing countries

21 of 24

insufficiently investigated and understood though, yet pro-poor product innovation is a challenging and important field and more research is expected in this area.

Acknowledgements I wish to thank Pierre Mohnen, Norbert Janz, Harry Bowen, Leo Sleuwaegen, Lynn Mytelka, and the participants of the Globelics Conference 2005, Pretoria, South Africa, for their comments and suggestions. I am particularly grateful to the two anonymous referees for their constructive comments, which helped me to improve the article substantially. I also thank Josaphat Kweka, ESRF, Tanzania, and Vijaya Ramachandran, Manju Kedia Shah and John Nasir at the World Bank for helping me accessing the data.

Address for correspondence Micheline Goedhuys, UNU-MERIT, Keizer Karelplein 19, 6211 TC Maastricht, the Netherlands and Institute of Development Policy and Management, University of Antwerp, Venusstraat 35, Belgium. E-mail: [email protected]

References Aitkin, B. J. and A. E. Harrison (1999), ‘Do domestic firms benefit from direct foreign investment? Evidence from Venezuela,’ The American Economic Review, 89, 605–618. Archibugi, D. and C. Pietrobelli (2003), ‘The globalisation of technology and its implications for developing countries: windows of opportunity or further Burden,’ Technological Forecasting & Social Change, 70, 861–883. Bangens, L. (2004), ‘University–industry linkages: forming successful alliances between university and industry in East Africa,’ in B. Mwamila, L. Trojer, B. Diyamett and A. Temu (eds), Innovation Systems and Innovative Clusters in Africa. Proceedings of the Regional Conference. 18–20 February 2004, Bagamoyo, Tanzania. Beck, T., A. Demirgu¨c¸-Kunt and V. Maksimovic (2005), ‘Financial and legal constraints to growth: does firm size matter?,’ The Journal of Finance, 60, 137–177. Bell, M. and M. Albu (1999), ‘Knowledge systems and technological dynamism in industrial clusters in developing countries,’ World Development, 27, 1715–1734. Blalock, G. (2001), ‘Technology from foreign direct investment: strategic transfer through supply chains,’ Paper presented at the ‘Empirical Investigations in International Trade’ Conference, Purdue University, 9–11 November 2001. Blomstro¨m, M. and A. Kokko (2003), ‘Human Capital and Inward FDI,’ CEPR Working Paper 167.

22 of 24

M. Goedhuys

Bongenaar, B. and A. Szirmai (2001), ‘Development and Diffusion of Technology: the case of TIRDO,’ in A. Szirmai and P. Lapperre‘ (eds), The Industrial Experience of Tanzania. Palgrave: New York. Caniels, M. C. J. and H. A. Romijn (2003), ‘Firm-level Accumulation and Regional Dynamics,’ Industrial and Corporate Change, 12, 1253–1278. Cassiolato, J. E., H. M. Lastres and M. L. Maciel (2003), Systems of Innovation and Development, Evidence from Brazil. Edward Elgar: Cheltenham, MA, UK, Northhampton, MA, USA. Djankov, S. and B. Hoekman (2000), ‘Foreign Investment and Productivity Growth in Czech Enterprises,’ World Bank Economic Review, 14, 49–64. Dosi, G., C. Freeman, R. Nelson, G. Silverberg and L. Soete (1988), Technical Change and Economic Theory. Pinter Publishers: London, New York. Edquist, C. (2004), ‘Systems of Innovation: Perspectives and Challenges,’ in J. Fagerberg, D. Mowery and R. Nelson‘ (eds), The Oxford Handbook of Innovation. Oxford University Press: Oxford. Fagerberg, J. (2004), ‘Innovation: A Guide to the Literature,’ in J. Fagerberg, D. Mowery and R. Nelson‘ (eds), The Oxford Handbook of Innovation. Oxford University Press: Oxford. Fransman, M. (1985), ‘Conceptualising Technical Change in the Third World in the 1980s: An Interpretive Survey,’ Journal of Development Studies, 21, 572–652. Giuliani, E. and M. Bell (2005), ‘The micro-determinants of meso-level learning and innovation: evidence from a Chilean Wine Cluster,’ Research Policy, 34, 47–68. Harrison, A. and M. S. McMillan (2002), Does Direct Foreign Investment Affect Domestic Firms Credit Constraints?. mimeo, University of California: Berkeley. Javorcik, B. S. (2004), ‘Does foreign direct investment increase the productivity of domestic firms? In search of spillovers through backward linkages,’ The American Economic Review, 94, 605–627. Katz, J. (1987), Technology Generation in Latin American Manufacturing Industries: Theory and Case-studies Concerning its Nature, Magnitude and Consequences. Macmillan: Basingstoke. Klette, T. J. and Z. Griliches (2000), ‘Empirical Patterns of Firm Growth and R&D Investment: a Quality Ladder Model Interpretation,’ The Economic Journal, 110, 363–387. Konings, J. (2001), ‘The Effects of Foreign Direct Investment on Domestic Firms,’ Economics of Transition, 9, 619–633. Lall, S. (1992), ‘Technological Capabilities and Industrialisation,’ World Development, 20, 165–186. Lall, S. (1999), The Technological Response to Import Liberalisation in Sub-Saharan Africa. McMillan: London. Lundvall, B.-A. (1988), ‘Innovation as an interactive process: from user-producer interaction to the national system of innovation,’ in G. Dosi, C. Freeman, R. Nelson,

Learning, product innovation, and firm heterogeneity in developing countries

23 of 24

G. Silverberg and L. Soete‘ (eds), Technical Change and Economic Theory. Pinter Publishers: London. Maddala, G. (1983), Limited-Dependent and Qualitative Variables in Econometrics. Cambridge University Press: Cambridge. Malerba, F. (1992), ‘Learning by Firms and Incremental Technical Change,’ The Economic Journal, 102, 845–859. Mccormick, D. (1999), ‘African enterprise clusters and industrialization: theory and reality,’ World Development, 27, 1531–1551. Murphy, J. T. (2002), ‘Networks, trust and innovation in Tanzania’s manufacturing sector,’ World Development, 30, 591–619. Mwamila, B. and H. Katalambula, (2004), ‘University-Industry linkage: the case of the prospective college of Engineering and Technology at the University of Dar es Salaam,’ in B. Mwamila, L. Trojer, B. Diyamett and A. Temu, (eds), Innovation Systems and Innovative Clusters in Africa. Proceedings of the Regional Conference, 18–20 February 2004 Bagamoyo, Tanzania. Mytelka, L. K. and F. Farinelli (2003), ‘From Local Clusters to Innovation Systems,’ in J. E. Cassiolato, H. M. Lastres and M. L. Maciel‘ (eds), Systems of Innovation and Development, Evidence from Brazil. Edward Elgar: Cheltenham, UK, Northhampton, MA, USA. Nelson, R. and S. G. Winter (1982), An Evolutionary Theory of Economic Change. Harvard University Press: Cambridge, Northhampton, MA, USA. Nugent, J. B. and M. K. Nabli (1992), ‘Development of financial markets and the size distribution of manufacturing establishments: international comparisons,’ World Development, 20, 1489–1499. OECD (1992), 1997, Oslo Manual, Proposed Guidelines for Collecting and Interpreting Technological Innovation Data. OECD: Paris. Portelli, B. and R. Narula (2003), ‘Foreign Direct Investment through Acquisitions and Implications for Technological Upgrading: case evidence from Tanzania,’ paper presented at the University of Oslo, May, 2003. Prahalad, C. K. (2005), The Fortune at the Bottom of the Pyramid: Eradicating Poverty through Profits. Wharton School Publishing: New Jersey. Rasiah, R. (2004), ‘Technological Intensities in East and Southeast Asian electronics firms: does network strength matter?,’ Oxford Development Studies, 32, 433–455. Rosenberg, N. (1982), Inside the Black Box: Technology and Economics. Cambridge University Press: Cambridge. Schmitz, H. and K. Nadvi (1999), ‘Clustering and Industrialization: Introduction,’ World Development, 27, 1503–1514. Schoors, K. and B. van der Tol (2002), ‘Foreign Direct Investment within and between Sectors: Evidence from Hungarian data,’ Working Paper, University of Ghent, p. 35. Semboja, H. H. and J. P. Kweka (2001), ‘The form and role of industrial innovativeness in enhancing firms’ productivity: the case of selected manufacturing firms in Tanzania’,’

24 of 24

M. Goedhuys

in A. Szirmai and P. Lapperre‘ (eds), The Industrial Experience of Tanzania. Palgrave: New York. Sleuwaegen, L. and M. Goedhuys (2002), ‘Growth of firms in developing countries, evidence from Coˆte d’Ivoire,’ Journal of Development Economics, 68, 117–135. Sleuwaegen, L. and M. Goedhuys (2003), ‘Technical efficiency, market share and profitability of manufacturing firms in Coˆte d’Ivoire: a Technology Trap?,’ Cambridge Journal of Economics, 27, 851–866. Smith, K. (2004), ‘Measuring Innovation,’ in J. Fagerberg, D. C. Mowery and R. Nelson‘ (eds), The Oxford Handbook of Innovation. Oxford University Press: Oxford. Szogs, A. (2004), ‘The Making of Innovation Systems in Least Developed Countries. Evidence from Tanzania,’ DRUID Summer Conference on Industrial Dynamics. Innovation and Development, Denmark, 14–16 June. UNCTAD (1996), Fostering Technological Dynamism: Evolution of Thought on Technological Development Processes and Competitiveness. A Review of the Literature. United Nations: Geneva, New York. UNCTAD (2002), Investment Policy Review. The United Republic of Tanzania. UNCTAD: Geneva. UNU/INTECH (2004), Designing a Policy relevant Innovation Survey for NEPAD. Mimeo UNU/MERIT, Maastricht: the Netherlands. Utz, A. (2006), ‘Tanzania in the Knowledge Economy: Fostering Innovation, Productivity and Technological Change,’ WBI Working Paper, the World Bank, Washington, DC. Wamae, W. (2006) ‘Why Technological Spillovers elude Developing Countries: a Dynamic Non-linear Model,’ DRUID Working Paper 06-2. Wignaraja, G. (2002), ‘Firm Size, Technological Capabilities and Market-oriented Policies in Mauritius,’ Oxford Development Studies, 30, 89–104. World Bank (2005), World Development Indicators 2005, IBRD/The World Bank: Washington D.C., USA.