Centro de Gestão e Estudos Estratégicos Ciência, Tecnologia e Inovação

Projeto “Estudo Comparativo dos Sistemas de Inovação no Brasil, Rússia, Índia, China e África do Sul” - BRICS

Brazilian Software Industry: a general view of its structure, specialization and competence building processes

Nota Técnica Final

José Eduardo Cassiolato Jorge Britto Vicente Guimarães Fabio Stallivieri

Rio de Janeiro Julho, 2007

BBRRIICCSS

Projeto “Estudo Comparativo dos Sistemas de Inovação no Brasil, Rússia, Índia, China e África do Sul” - BRICS

Brazilian software industry: a general view of its structure, specialization and competence building processes

José Eduardo Cassiolato Jorge Britto Vicente Guimarães Fabio Stallivieri

Julho 2007

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 2 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Brazilian software industry: a general view of its structure, specialization and competence building processes

José Eduardo Cassiolato - IE/UFRJ and RedeSist Jorge Britto – Fluminense Federal University (UFF) and RedeSist Vicente Guimarães – IE/UFRJ and RedeSist Fabio Stallivieri – UFF and RedeSist

Summary

1 – Introduction...... 4 2 - Technological Regime in Software Industry: a general view...... 5 3 - Brazilian software Industry: an historical perspective focused on policies issues...... 9 4 - Brazilian software Industry: market evolution, industrial structure and firms´ strategies...... 15 4.1 General figures of the industry...... 15 4.2 – Software Labour market...... 20 4.3 - Industrial structure...... 24 4.4 - Regional Distribution of Software Activities...... 32 5. Export orientation and outsourcing...... 40 6 - Brazilian Software Specialization and new opportunities: the major trends...... 48 7 - Sectoral system of innovation in Brazilian software industry: general characteristics...... 53 8 - Local Productive Arrangements in Brazilian Software Industry: an exploratory analysis..... 68 9 - Concluding Remarks...... 77 Bibliography...... 80

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 3 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

1 - Introduction

The analysis seeks to explore analytical and methodological issues associated to the study of learning, innovation and competence building, departing from a sectoral system of innovation (SSI) approach (Malerba, 2004) to analyze recent developments of the Brazilian software industry. This industry is fairly complex, sophisticated, diversified and heterogeneous. The analysis of the sectoral system of innovation should deal with these elements, enlarging its analytical limits to capture the amplitude of its impacts in the whole economy. That industry has faced an impressive growth in the last twenty years, based on the construction of productive and innovative capabilities. Experiencing two digit growth rates during the last decade, the Brazilian software was, in 2001, one of the worlds´ largest markets, with sales amounting around US$ 7,7 billion. Domestic market is responsible for approximately 98% of total sales; with 10 thousands firms employing 165.000 workers. Between 1991 and 2001, the share of the software industry in GNP more than tripled, going from 0.27% to 0.71%. The share of software in the IT sector as a whole is 2/3 now. As part of this process, Brazil has developed an entrepreneurship and technological capacity that allows her to be an active participant of the softare market in the world. This growth can be inserted in a more general tendency, related to the emergence of some developing countries that become important players into the software industry, particularly some BRICs countries as India, China and Brazil (see Botelho, Tschang and Amsden, 2003; Arora and Gambardella, 2004; Commander, 2003; Carmel 2003).

In the context of a “knowledge economy” the software industry seems quite relevant not only due to the embodiment of knowledge into products, systems and services but also because it provides the tools to adapt the characteristics of systems, channels, nets and organizations to the requirements of new IT technologies. Despite the advances in modern communication technologies that make the long-distance interaction fairly quick and easy nowadays, there are a lot of evidences that the contacts, interactions and exchange of information and knowledge developed trough a local basis constitutes an important root to the improvement of innovative competences. According to this interpretation, the strengthening of industrial competitiveness might be articulated with local conditions that improve the potential of learning, innovation and competence building processes. In the software industry, the complex and tacit character of the knowledge required to generate innovations reinforces the importance of the construction of proper channels of contact and communication between providers and users, in order to allow a systematic interchange of information that permits the adaptation of customized solutions according to their needs. The capacity to develop these solutions in a local basis permits the accumulation of capabilities that makes possible the expansion to more sophisticated markets, including international ones.

The analysis seeks to explore analytical and methodological issues associated to the study of learning, innovation and competence building in Brazilian software industry. Contrasting with a strictly sectoral focus, our hypothesis is that the territorial proximity and the local institutional basis play an important role to the improvement of learning and to the strengthening of firms’ competitiveness and innovativeness. In fact, empirical evidence collected from case studies on local productive systems in Brazil illustrates the remarkable differences on innovative and learning capabilities of different systems belonging to the same industry. In this sense, a solely sectoral approach tends to neglect specificities emerging from institutional and historical Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 4 http://brics.redesist.ie.ufrj.br

BBRRIICCSS contexts, which are specific to local productive systems. So, the analysis tries to articulate a more enlarged concept of industry to a discussion about the characteristics of local production systems in the Brazilian software industry.

The paper is organized as follow. The second section is an attempt to clarify the main characteristics of technological regime in software industry, stressing some contrasts between the different segments of the industry. In the third section an historical perspective is adopted to discuss the changes experienced by Brazilian software industry in the last thirty years. The fourth section presents some figures about the Brazilian software market, its industrial structure and territorial distribution. The fifth section discuss the export performance of Brazilian software, stressing the challenges it faced in order to exploit the opportunities opened by the consolidation of global outsourcing markets. The sixth section presents an analysis of the competence building processes in some segments of the industry. Subsequently, some general characteristics of the sectoral system of innovation in Brazilian software industry are explored in the seventh section. The eighth section develops an exploratory analysis about the territorial heterogeneity of the industry, connecting it to a discussion about learning, innovation and competence building processes in local software productive systems. Finally, the last concluding section presents some normative implications of the analysis.

2 - Technological Regime in Software Industry: a general view

The software industry is important not only because of its potential of direct economic impact – associated with a huge growth of industry and services productivity - but also due the critical leverage it performs for innovation across virtually every area of activity, playing a decisive role to improve intra and inter organizational learning. Moreover, the software industry presents some characteristics that might be considered in the analysis of learning, innovation and competence building processes. Traditional concepts of industrial sector - strictly defined according to characteristics of technologies or the marked base – seem not very useful to understand those characteristics. In this activity the critical productive factor is the knowledge embodied in qualified people with a creative and intellectual work capacity. The industry is also characterized by the dynamism at which innovations are converted into new products, services and technical solutions.

Software industry is consensually viewed as one of the driving forces of the structural changes generated by the ICT paradigm. According to Fatima Gaio (1990:145): “Software is an intrinsic and pervasive part of the ICT paradigm which has peculiar and complex characteristics. First, software is a core element for implementing the fundamental features of microelectronic-based systems, which are the capacity to perform a wide range of functions with structural plasticity. Second, software is a disembodied technology and its product is intangible, but its value is determined by how effectively computational operations are undertaken and by how reliable the presentation of the application domains' being modeled is. In synthesis, software represents the embodiment of knowledge into products and production systems.”

The production of software is a design-intensive activity of intangible products, which are independent from the carrier media, and information is its basic material (Gaio 1992). The production flexibility brought about by IT systems is allowed by its software component. This flexibility defined 'ex ante' – at the project design stage of IT - is often well suited to user Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 5 http://brics.redesist.ie.ufrj.br

BBRRIICCSS intervention (Aksoy 1990). User-producer relations are not only crucial for successful software development but gain importance in the present paradigm precisely because of possibilities brought by software.

The pervasive impacts of innovations generated from “software industry” assume some particular characteristics. In fact, the “products” generated by this industry are totally different from traditional industrial products, being basically made up of knowledge, an intangible factor associated with the creativity and intellectual capacity of specialized workers .The dynamism of the sector has an impact in terms of the high birth and mortality rate of firms as well as in terms of the volatility of company’s profile and market shares. The structural heterogeneity of the sector can be associated not only to the variety of products and services provided but also to diversity of competitive conditions associated with industries’ software segments.

In this sense, a clarification about some characteristics of the software industry is usual in the literature. According to a distinction that is almost consensually accepted (Frick and Nunes, 1996; OECD, 1998; Schware, 1992; Mowery, 1999) two analytical cuts can be used to capture the diversity of software activities. The first regards the form of software arrival to the market, establishing a distinction between “package”, “services” or “embedded” software. The second regards the market destination of software, signalizing a distinction between “horizontal markets” and “vertical markets”.

The package software is usually standard, targeting an ample number of clients and not requiring a direct interaction between user and developer during the making of the product. In this segment the software must be ready to meet a generalized demand, with the marketing strategy being quite similar to those used by hardware equipment. Competition is carried out through technical development and mass distribution, with the costs of creating and launching new software being very high. The possibility of obtaining scale economies is very important in this segment, requiring the adoption of aggressive sales strategies, based on the construction of sales network and on the consolidation of a broader known brand. Software product companies normally show insignificant marginal costs and growing incomes to the scale in products. They are directly and necessarily exposed to and focused on customer’s requirements, since the definition of the product outlines and functionality is crucial to its success. They have to offer a solution that meets the requirements of a more or less large group of customers, to whom it sells a license to use, eventually associated with some customization work per service for specific needs. Package software requires very high investments – in market research, technological innovation and marketing – before any sale can be made, with the risk of failure being equally high. On the other hand, success brings high rewards, as a growing base of users is captured without increasing sales costs.

Service software involves custom-made programs, developed through an intense interaction with specialized users, from which specific features are specified. This software can be seemed more as a service than as a product. The active involvement with the clients becomes a decisive competitive factor, reducing the market risks but increasing the costs and lead-time of development activities. In this segment, a solid client relationship, as well as a capacity to provide sophisticated solutions, is very important to strengthen firms´ competitiveness. In this segment, a distinction between low value and high value services must also be stressed. Low value services normally involve aspects like software maintenance or code generation, being Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 6 http://brics.redesist.ie.ufrj.br

BBRRIICCSS developed according to the tasks specified by the customer. To low value services competition is normally based on price, with the process quality requirements being given by the company history, or by the certification of the processes. In this segment, local companies, particularly for services that involve strong interaction with customers, normally dominate the market. However, in a situation in which there is less need for interaction, this service may be contracted at a distance. On the contrary, the high value services involve uncertainty relative to the result, becoming the evaluation of supplier’s capability a difficult task. Thus, aspects like producers´ reputation become especially important, which tends to benefit multinational consulting companies.

Contrasting with the two others alternatives, the embedded software is usually set in equipment, permitting to operate different functions it in an efficient and automate way. This software works in conjunction with a machine like, for example, the software of a cellular telephone, called shipped software.

Software can also be grouped according to the market it targets. In the “Horizontal Segment” the attributes of the software are designed according to standards of the informatics area, with little specific context of another area of information. Despite the fact that it tend to be sold by means of packages, this software can also be flexible, in order to solve information problems in diverse areas. Examples of horizontal segment products can be associated with the development of operational systems, spreadsheets, databases, and word processors, among others. Often, they become software development tools in the vertical segment (for example, used as databases for administrative management of a medical office). In the “Vertical Segment”, the software is custom-made developed for a specific economical activity (health, education, transports, research, etc.), incorporating other types of specific knowledge from distinct areas. In this segment, is also necessary to accompany the development of informatics, through the creation of new versions associated with more efficient and friendly operational environments (for example, network communication systems). According to the segment targeted by company strategy, a different pattern of innovative efforts can be privileged.”

There is a tendency to generate an industry marked stratification according to the characteristics of each segment, implying a co-existence between fragmentation and concentration forces. In fact, the industrial structure tends to be divided between horizontal segments, dominated by large corporations with standard products and a worldwide production, such as Microsoft, IBM, Oracle among others, and specific niches in the vertical segment, where a multiplication of small and mid-size companies took place. Although the barriers to entry are not prohibitive in the vertical segment, the growth barriers can be significant, since the large corporations occupy the most strategic and lucrative segments.

Despite these tendencies, we also observe that companies’ strategies usually target different segments at the same time. In fact, software firms often have various lines of business and simultaneously develop service and product activities. The intermediate strategies usually involve the production of standardized software components, shipped software and customized software. So, there is a dynamics by which the structure of the industry has been constantly reconfigured by the appearance of new products and new market segments and niches. Technological innovations are the driving force of such dynamics, defining the amplitude of life cycles of existing products and opening new opportunities to producers and new needs to customers. For Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 7 http://brics.redesist.ie.ufrj.br

BBRRIICCSS local firms in developing countries such as Brazil, the building up of innovative competences becomes a key factor to overcome the limits imposed by the control of more attractive segments of industry by large multinational companies. In the segment of high value services, for example, local companies in developing countries can position themselves equally, replacing some of their reputation deficit by experience, strong process qualification (using the best software engineering practices, such as ISO 9001, certification CMM or SPICE) and drastically lowering costs.

This possibility stresses another important characteristic of software industry, related to a process of market internationalization through which firms may overcome the limits of local markets, converting external trade as an alternative to their survival and growth. This tendency is relevant not only for the package segment, in which scale economies play a decisive role to firms´ competitiveness, but also in the custom-made segment, in which globalized companies are constantly searching for international suppliers. In order to internationalize their sales, it is important to create an image of trustworthiness in the targeted segment, trying to work directly with selected clients. Quality and productivity are also of prime importance to maintain competitiveness. The adequate balance between those elements depends on a complex articulation among various factors, requiring a permanent effort to adapt product characteristics, to qualify the working force and to consolidate innovative strategies and marketing competences. Finally it is worth mentioning that software development is not necessarily R&D or technology intensive. It involves totally different things that have different needs in terms of knowledge and capabilities and consequently have different impacts in terms of job quality and income generation.

Several authors have tried to classify software development as a series stages. Some of the best known are the waterfall model (Royce, 1970) and the spiral model (Boehm (1981). With recent improvements we can propose splitting software developments in nine stages: i) technical proposal: this stage defines the scope of the system; here the functionalities and the technological architecture are defined; ii) specification of requirements: in this stage the system’s analyst and project manager specify the conditions which the final product should satisfy; iii) system’s analysis: stage in which the analyst identify the business classes of the application, main attributes and methods; iv) prototype making: phase in which software models are created in order to allow the user to navigate among screens and simulate the functionality of the system; v) system’s project: started after the prototype is approved; should be organized in a framework that is essential for obtaining scale gains in the production. vi) implementation: phase in which the programmer codifies the system; it is the simplest activity of the whole process; vii) tests: stages where system’s specification and the environment of the final user are validated; viii) Installation: it is when the system is introduced; it involves operational change in user’s routines; and ix) maintenance: brings improvements and corrections in the system. This distinction is important for the discussion in this paper since there has been a tendency to treat software development activities and exports are technology intensive activities in developing countries. In fact most software development activities in developing countries refer to stages vi) to ix) that are in fact very low in the value-chain. Hence the need to be careful to treat export levels as a proxy for competitiveness.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 8 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

3 - Brazilian software Industry: an historical perspective focused on policies issues

The emergence of software industry in Brazil can be associated with the industrial policy based on high import tariffs and government controls protectionism implemented by the military government in the 1970s. The focus of this policy was on hardware, but it had induced an improvement of the labor force in IT sector, giving a decisive stimulus for the nascent software industry. So, the origins of Brazilian software industry lie back in a period of high protection and the “market reserve” policies of the 1970/80s. As a corollary of a long history of computer use that characterized the market reserve policy, particularly in the financial and government sector, computer users established in-house teams and organization competencies to develop software systems. Despite the huge increase in local hardware production (from US$ 200 million in 1979 to about US$ 4 billion in 1990), the domestic industry was growing in insulation from the dynamic technological change that was taking place in the international IT industry during the 80s and gradually became uncompetitive. Additionally, financial and liberalization pressures reduced government’s direct financial subsidies and inducements to develop new technology, and firms systematically started failing to innovate.

As a consequence of the focus on import substitution, the IT sector as a whole remained very much domestic oriented. The market reserve policy appears to have created a demand for IT related training and to an important increase on the supply of skilled personnel for the sector. Further, it led to a more general upgrade in IT use and networked activity, a process that was stronger in particularly sectors, such as banking (Cassiolato 1992). Despite the focus on import substitution, Brazilian governments since the 1970s have also pursued a policy of export promotion, using a mix of instruments, such as tax rebates and credit subsidies.

One of the key arguments of the liberalizing IT policy regime in the 1990s was that emphasis on local production was hampering the diffusion of IT products in Brazil. However, according to figures from the US Department of Commerce, Brazil was in fact, the second fastest growing microcomputer market (after China) during the second half of the 1980s, precisely the period of the “market reserve” policy. In fact, from 1984 to 1987 the Brazilian market of microcomputers grew by an average of 74 per cent per year (from US$ 189m to US$ 992m) and became the sixth largest in size (Cassiolato 1992). An important shortcoming was the inadequate and insufficient technological learning by local IT firms.

The strategy of evolving from technology licensing to in-house development was anchored in learning processes sustained by continuous R&D expenditures. These, however, have been limited by the size of the market and, most importantly, by the size of local firms. The adoption of more aggressive technological strategies was restricted by the absence of conglomerates in the Brazilian economy, which could provide the financial means for a long run strategy. With few exceptions - usually associated with firms that were created by financial institutions such as Itautec - Brazilian IT firms were relatively small and independent. Financial means to foster long- run investment was also almost always missing. The government expenditure on general science and technology also faltered during the period. Concerning the demand side, we can stress the impact of the built of National Research Network (RNP in Portuguese), probably the most successful government-sponsored technology program in IT since it has been a major driving force behind Internet diffusion in Brazil. Managed by a committee of academics and representatives of the IT-users, the program allocated US$ 20m in 1997 to local IT service Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 9 http://brics.redesist.ie.ufrj.br

BBRRIICCSS providers, schools, and infrastructure, such as high-speed backbones, and linking universities and business centers (Cassiolato 1992).

The nineties were characterized by the implementation of liberalization policies, aiming a reduction in the role of state in the economy and the progressive integration of Brazil in the global economy. Based on the fear of being left behind in diffusion of the IT paradigm and subject to huge pressures by the United States to open the local market, the policy based on market reserve was replaced by a more market-oriented policy in 1992. Once that protection was finally lifted in the 1990s, the market share of domestic producers declined substantially. The first significant impact of the new policy regime was that most Brazilian hardware makers who had been producing minicomputers and peripherals disappeared after liberalization, moved into other markets such as services and distribution or were bought out by foreign firms. In the components market, Brazil has lost much of its production capacity as companies such as Itautec have abandoned the components business and scaled down R&D in face of international competition. Hence, a dramatic increase in the participation of foreign capital in the Brazilian IT industry was observed. The liberalizing changes in the policy regime were meant to bring an increase in the efficiency of the IT industry and its overall performance. However, only the software and computer services industries experienced some growth. For these industries net sales increased more than three-fold from 1990 to 2000, as shown in Figure 1. The hardware industry which received massive fiscal incentives never recovered the dynamism of the 1980s. Before liberalization the net sales increased from US$ 1b in 1981 to more than US$ 7b in 1989. After liberalization sales collapsed to less than US$ 5b a year during the 1991-1995 and a slight recovery ever since was not sufficient to bring sales to levels above those achieved in the late 1980s.

Figure 1 - Brazilian IT Industry – Net Sales (1981-2000)

17.8 17.3 Hardware 15.0 15.1 Software + ServiçosServices 13.7

TIIT 11.1 10.3 9.8 10.3 9.0 8.4 9.4 9.0 7.9 7.1 7.1 6.9 7.5 ' 8.4 5.3 5.9 7.5 4.6 6.8 7.2 4.5 4.9 7.0 3.5 4.1 4.5 6.1 5.3 6.3 5.2 4.9 2.3 4.0 1.5 1.8 3.4 3.9 1.0 2.6 3.1 3.0 3.4 2.7 1.8 1.5 1.5 1.3 1.8 1.0 0.9 1.2 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 Source: Own elaboration from MCT-Sepin (1988), SCT-Sepin (1991), and www.mct.gov.br/sepin

Although recession has increased under liberalization, several movements of entry of foreign firms have been observed in the IT sector since the early 1990s. Different mechanisms have been used: acquisition of local firms, setting up of sales offices and the establishment of different types

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 10 http://brics.redesist.ie.ufrj.br

BBRRIICCSS of partnerships with local capital. These movements, besides changing the structure of the industry in Brazil, showed the importance of Brazil in the global expansion strategies of leading international firms. Partnerships allowed entry without the need for a substantial amount of capital. They made possible for foreign firms to overcome important barriers to entry such as the existence of a network of commercialization channels, maintenance and technical assistance in all Brazilian territory. An important part of the liberalization strategy was to attract MNCs that would use Brazil as an export platform. However, these firms were only interested in the internal market and hardware exports did not materialize. IT goods manufactured in Brazil are, with rare exception, designed abroad, imported into Brazil as complete kits prepared to be assembled. This reduces extremely the supply chain while, at the same time makes not viable the setting up of a components sector. Most subsidiaries of MNCs even subcontract the assemblage and still receive the fiscal benefits directed to stimulate local production.

The inversion of priorities in policies led also to a deep re-examination in technological strategies by local firms. Total R&D expenditures by local firms of the IT sector fell from US$ 183.8m (or 4.2 per cent of sales) in 1989 to US$ 60.7m (1.4 per cent of sales) in 1992. However it was not only a significant decrease in total R&D expenditures by firms what was perceptible. In parallel, there has been a significant change in the nature of innovation activities performed within. They downgraded those activities with higher innovative intensity such as R&D for new products and increased activities of lower technology intensity, such as training.

Considering these changes, we can conclude that the institutional basis of the sectoral innovation system has faced major transformations in its characteristics during the last two decades. In this context, an interesting question is, whether innovation system for the software industry follows the general pattern of IT or has specifics characteristics that affect the dynamics of the learning, innovation and competence building processes.

The organization of software Brazilian industry is deeply articulated with the orientation of industrial policy to the IT sector as a whole. In this sense, the evolution of software sector in Brazil is usually divided in two main stages (Evans, 1995; Botelho, 1991; Tigre et al, 2001). The first period, before 1990, is characterized by import substitution, and the second, after 1990, by a progressive liberalization that induces a process of global competition. Until 1990, Brazilian policy for the software industry was mostly indirect, being subsidiary to a hardware policy oriented to the search of technological autonomy. During the “market reserve” period, the software industry got a pattern characterized by the development of copycats of the most popular programs. Despite the learning gains associated with this pattern there were also negative consequences related to the increase of hardware costs to final users, delaying and reducing the scope of IT diffusion and, more importantly, blocking their early development as sophisticated and mature software users. The impact on the emerging software firms was higher barriers to entry and a bias towards those segments that could afford the extra hardware cost, such as banking and finance. Although these problems, the domestic software market reached US$ 1.1 billion in 1991, which corresponded to about 1/3 of total IT sales (Kraemer et al, 2001).

After the liberalization of the 90s, a relevant change to software industry was caused by Law 8.249/91, which intended to give fiscal benefits to R&D activities. The essence of the legislation was to balance the liberalization with some supportive mechanisms for firms. It specifically aimed at providing assistance for local equipment manufacturing and R&D activities in the IT Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 11 http://brics.redesist.ie.ufrj.br

BBRRIICCSS sector. Instead of emphasizing local technological development, the new legislation offered fiscal benefits if companies would commit themselves to a minimum level of local production. The policy consisted of fiscal benefits comprising a waiver on the industrialized goods tax (which was responsible for 15 per cent in the final cost of production), a reduction of 50 per cent on income tax for R&D expenditures and a cut of 1 per cent of the income tax payable by companies investing in IT firms (available until 1997). Government procurement policy should also, in theory, favors the acquisition of IT goods developed and produced in Brazil, as long as they had similar prices to imported equipment. By 1997, 248 firms had benefited from these measures. In order to get these benefits firms should invest at least 5 per cent of their sales from IT products (excluding software and professional services) on R&D activities, of which 2 per cent should be through joint projects with universities, research institutes, or in government-sanctioned IT programs. The definition of “R&D activities” has been too broad (including training for example) and the government never implemented a monitoring program to assess the effectiveness of these measures. From 1993 (when it came into effect) to November 2001, it benefited 428 firms and directed R$ 2.9 billion in resources for R&D, of which 63% was spent in corporate R&D and 33% in sponsored research in universities and research institutes. In the same period, about ¼ of the resources were directed to software development (and another 24% to the development of software systems and hardware). By 2000, R&D expenditures in the IT industry reached US$ 530 million, 56% provided by the fiscal incentives (SEPIN, 2002). Despite the significant amount of resources, the Law ultimately benefited mostly large hardware firms, which could make use of corporate tax rebates, particularly foreign firms. Software firms benefits derived from the law were much smaller and indirect.

The focus on promoting an export drive was maintained when the general focus of industrial policy change from import substitution to trade liberalization in the 90´s. To software industry, the major policy initiative to getting this aim was the creation, in 1992 of the SOFTEX-2000. The SOFTEX 2000 – an initiative of the Science and Technology Ministry (MCT/CNPq) - was one of the three Priority Programs of Informatics in 1994, relating to incentive applications of the Informatics Law of 1991 (Law 8248/91). The program wanted to strength the software national industry and to promote the commercializing of its products and services abroad. SOFTEX 2000 had the target of exporting US$ 2 billion worth of software in the year 2000 and estimated that 50,000 new skilled jobs could be created. The strategy was to have agents responsible for the development of a foreign region through the establishment of local offices (Beijing, Tokyo, Silicon Valley, Boston, etc.). Offices assisted interested firms in market studies, setting up sales and marketing partnerships and participating in major trade shows (e.g. Comdex, Cebit, etc.). Since its creation, about 400 firms have participated in international software trade shows.

At the end of 1996, the SOFTEX Society, a private non-profit organization, was established to manage the SOFTEX program. The emphasis of the renewed initiative became the promotion of entrepreneurship, training, funding and market orientation, as well as exports, although less so than before. Among other aspects, the idea of having offices around the world was dropped. The new perspective tried to replicate the pattern of Silicon Valley and Route 128 regions, which ascribed a critical role to incubators for the growth of high-tech firms, which did not exist at the time in Brazil. In 1997, SOFTEX launched Prosoft, a pilot credit program for software firms in partnership with the National Banks for Social and Economic Development (BNDES), with a budget of USD 35 million to run until December 2003. Its funding analysis takes into account the risky nature of the software business and thus it is managed in portfolio logic. It provides loans Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 12 http://brics.redesist.ie.ufrj.br

BBRRIICCSS up to USD 2 million to software firms with gross revenues under USD 30 million. By the end of 2002, it had invested half of the budget in 29 firms. At the end of the 1990s, the SOFTEX program underwent a major restructuring, reflecting its adaptability to the different context of the Brazilian software industry. In the software industry even though ambitious target to export US$ 2b by the year 2000 was not fulfilled, exports grew considerably from US$ 1m in 1990 to US$ 72m in 2000 and to US$ 300m in 2004. However, such growth was not sufficient to surpass a bigger surge in imports (basically of package software) that helped to increase the trade deficit in software to about US$ 1b in 2000 and 2001 (Guimarães, 2005). The impact on the balance of payments was even bigger if one should add the remittance abroad of payments for patents, licenses and royalties of IT goods and software developed in the home countries of MNCs.

The limits of the export strategies coordinated by SOFTEX might stressed the difficult that firms form emerging markets face with they try to shift their orientation to export markets given the lack of experience and reputation. Evidences also suggest that a shift from price to quality as the source of competitiveness requires a significant changes in industrial organization. At its inception, in 1993, SOFTEX estimates deemed possible that by the close of the century the software exports of SOFTEX-supported companies would have reached USD$2 billion. However, by 2000 all software exports, supported by SOFTEX or not, may have reached, USD$100 million, twenty times less than promised. In face of these results, many analysts of the industry stress the importance of some points that explained this trend.

First, even without the benefit of hindsight, one would have to readily admit that to expect to bring Brazilian exports of software products from naught to $2 billion in seven years would have been an exaggeration. Second, there was a problem of the general focus of the program. In this sense, the main real reason for the failure of Softex export strategy was the absence of a client-led drive. Rather than trying to find first who are the clients that might be attended and then sorting out how and what products could be sold to them, the drive was put behind pushing what was available, without a focus in which segments and which markets could be properly exploited. Yet Softex authorities went about assuming that the foreign market would readily take what Brazil had to offer for the domestic market, without a clear strategy of specialization in some segments where the local capabilities were already consolidated - low-cost competence in software coding for spatial close markets, for instance. It would be also almost impossible to reach the $2 billion exports target with start-ups, converting SOFTEX in an agency to take care of promoting incubators, converting the increase of exports in a subsidiary aim. Thus, Softex strategy had a focus oriented to new companies which needed to “learn” to export software in spite of being oriented to companies which already proven to have a potential to exploit new markets.

Despite these problems, SOFTEX has secured a niche for itself as the clearing house for information regarding software development and publishing in Brazil. Under the aegis of SOFTEX several initiatives have prospered, including a Brazilian Software online catalogue in four languages, the consolidation of regional software nuclei and the promotion of quality aspects of Brazilian software, permitting to reach almost 900 members. SOFTEX has also contributed to foster software entrepreneurship, to create a sense of community in the industry and to expose national firms to the global market. However, the initiative to promote a decentralize growth of software industry was also partially succeeded: most of the industry continues to be located in the South East of the country near the major sources of demand, while the incubators supported by SOFTEX have continued to suffer from weak market orientation and ability to launch viable Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 13 http://brics.redesist.ie.ufrj.br

BBRRIICCSS projects. However, by focusing on training and supporting mechanisms to start ups firms (such as incubators programs) the program contributed to improve the quality of software development directed towards local needs (by firms and other organizations).

There are also another signals that the main policy focus – understood in a broader sense, including no only directly public agencies – began to change to the improvement of quality and productivity in the software sector. These changes underlie a process of “institutional learning” that was experienced by Brazilian institutions that support software industry. As an example of this process we can mention the built of the “New Brazilian Software and Services Industry Support Program” carried by SOFTEX and APEX, a medium term national program of certification, lined up to the international parameters, to reduce barriers to the entrance of national software in the international market. The general purpose of the program is to increase the competitiveness and the productive capacity of the sector, with the strengthening and magnifying of national companies penetration in the internal and external markets. The diagnostics behind the program stressed the lack of an articulated strategy that explores the demand of the great national (also the government) and international technology purchasers, trying to fortify national companies and promote their access in the foreign market. The program wants to increase the Brazilian companies’ software and services exports, utilizing the high degree of quality that was reached in some segments, such as: financial and banking, security, telecommunication, electronic business, enterprise management and public services automation. Specifically, the program intends to magnify the shared infrastructures for exportation companies, anchored in the implantation of national companies and multinationals development centers, which can be act as develop exports platforms to offer outsourcing/offshore services. The idea is to identify windows of opportunities in contracts for outsourcing of services, progressively building an international image of Brazil as supplying of software and services based on IT. It also try to foster the development of emerging segments with high growth potential, as wireless communication; digital TV; agro business; visualization and entertainment; applications for clusters and computer grids. One of the main ideas is to privilege solutions developed in the country, for digital inclusion, in the scope of other governmental programs such as Digital TV, Program of Communitarian Centers, City Halls Financing, among others. Another focus can be target to the use of government purchasing power to strength competitive solutions developed by Brazilian companies.

To get these aims, the industry has to overcome the delay in terms of international certification, what is basic to compete for world contracts. In fact, Brazil software process improvement in the last ten years have favored the ISO 9000 certification, generating a situation in which the number of local firms that have reached high levels of CMM is extremely limited. The MPS-BR Project focus on this problem, involving a set of public and private agencies - SOFTEX Society (Project Leader), COPPE/UFRJ, CenPRA – Campinas, CESAR & SOFTEX - Recife, ABNT & CELEPAR (Standards and Guides) – in order to create and improve a Process Reference Model (PRM) and a Process Assessment Model (PAM), compatible with CMMI, based on the reality of the Brazilian firms. The idea is to implement and assess the MPS Model in the Brazilian software companies, mainly oriented to the small and medium-size enterprises (SMEs), in all regions of the country, with feasible costs. The program is also based on the built of an “Accreditation Forum” responsible for certification of the competence of the organizations conducting Process Implementation Services (II-MPS.BR) and Conformity Assessment Services (IA-MPS.BR)

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 14 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

4 - Brazilian software Industry: market evolution, industrial structure and firms´ strategies

4.1 General figures of the industry

As we have already mentioned, software has a huge impact in the productive base of all the economy, becoming a permanent source of innovation and differentiation of products. In fact, the introduction of technological and organizational innovations strongly supported by software products and services is the basis for the increasing of productivity in virtually all the setors of economic activities. In the last two decades Brazil develops a dynamic and large software industry, becoming an important player in a sector that has faced an impressive growth. However, a rigorous identification of the dimension of Brazilian software market is not an easy task. Data form different sources are often contradictory in the evaluation of the size this market. Two elements make particularly difficult this kind of evaluation. First, software activities are usually extremely integrated with other IT activities at the business level, in a way that ones of the most important firms of the market do not have software as its core activity. Second, the business model of the relevant players is also quite heterogeneous, sometimes being based on product, sometimes with the services assuring the larger slice of their sales.

Despite the complexity of its characteristics, is usual an estimation of a market of around US$ 7,7 billion to Brazilian software industry. The sector has experienced two digit growth rates during the last decade, becoming the world 10th larger market, with the domestic industry being responsible for approximately 98% of the market sales, employing around 160.000 workers in 10 thousands firms. Between 1991 and 2001, the software segment’s share as a percentage of the GDP more than tripled, reaching 1,5% (see Table 1 below). However, the imports sustain a share of around 44% of Brazilian software market1.

Table 1 – Main Indicators for Software Industry in Selected Countries – 2001 Counties Sales Exports US$ Employees sales/GDP US$ Billions millions ) USA 295,700 n.a. 1.042.000 2,0% Japan * 66,040 73 534.000 2,0% Germany 41,720 n.a. 300.000 2,2% United 41,160 n.a. n.a. 1,0% Kingdom France 35,800 n.a. 158.000** n.d Canada 18,340 n.a. 123.000** n.d Italy 15,730 n.a. n.a. n.d Holland 10,630 n.a. 62.000** n.d India 8,200 6.220 350.000 1,7% Brazil 7,700 100 158.000 1,5%

1 These figures show that the exports are responsible for a significant share of software sales for some countries - India, Israel and Ireland, for instance, get exports coefficients of respectively 75%, 70% e 84% of software sales. On the other hand, in Japan and Korea software sales are strongly oriented to the internal market. The software represents of around 1-2% do GDP to the countries selected which this index being higher in most developed countries. Israel and Ireland present a disproportionably high share of software in the economy as a whole, respectively, 3,4% and 7,,4% of GDP Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 15 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Korea 7,694 35 n.a 1,8% Ireland 7,650 6.500 25.000 7,4% China 7,400 400 186.000 0,6% Spain* 4,330 n.a. 20.000 0,7% Taiwan* 3,801 349 n.a. 1,2% Israel 4,200 3.066 15.000 3,4% Finland 1,910 185 20.000 1,6% Singapore 1,660 476 n.a. 1,9% Argentina* 1,340 35 15.000 0,5% Mexico <1.000 N.A. N.A. <0.2% Sources: OCDE, Veloso et.al (2003, p.5), WITSA apud Softex (2003), Campos (2000), Arora (2004).* data from 2000, ** data from 1995, na = not available.

A precise evaluation of the market of the Brazilian software industry is a complicated task. However, we can consider some data from IBGE, the Brazilian official statistics agency, collected through the PAS (“Pesquisa Anual de Serviços” - Annual Services Survey) for the year 2003. PAS is a research the captures the entire universe of firms with more than 30 employees, being also based on a sample to firms with less than 30 employees. According to data showed in Table 2 the set of informatics activities – including software, hardware and services – reached sales of US$ 5,6 billions in 2003, from which software has the greater slice, with sales of around US$ 1,4 billion. The difference between this data and the traditional estimative of the size of Brazilian software market can be attributed either to methodological issues of PAS as well as to intrinsic difficulties to identify firms that has software as a secondary or complementary activity. Data from PAS also shows that firms involved with the development of software had a sales amount of around US$ 2,7 million in 2003, which is smaller that the sales obtained in other segments of informatics activities. We can also observe that, compared with firms involved with the development of software, firms oriented to outsourcing activities and to representation and licensing of software not produced internally present an amount of sales per firm particularly higher.

Table 3 – Sales Indicators for Informatics enterprises surveyed by PAS - IBGE- 2003 Net Number of operational Total Segment informants sales (US$ (%) Sales per 1000) firm 5 019 Informatics 1 129 5 666 865 100,0 366 3 934 Hardware Consulting 176 692 406 12,2 127 2 752 Software Consulting 507 1 395 720 24,6 900 Development and edition of package 2 336 software 287 670 624 11,8 668 Development and edition of custom made 2 324 software 312 725 096 12,8 027 Internet services (except Internet providers) 92 83 166 1,5 903 984 Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 16 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Database specific services (except for the 1 671 Internet) 45 75 229 1,3 747 1 942 Data processing services 394 765 539 13,5 992 1 828 Informatics security services and support 210 383 907 6,8 126 5 685 Outsourcing activities 116 659 475 11,6 127 Licensing and representing of software not 5 732 produced internally 52 298 099 5,3 678 2 203 Maintenance and repairing services 203 447 400 7,9 941 3 573 Other activities related to informatics 210 750 398 13,2 324 1 674 Other services 69 115 526 2,0 286 Source: IBGE, Annual Research about Services (Pesquisa Anual de Serviços -PAS) 2003. Observations: (1) The research covers the entire universe composed by firms with more than 30 employees. To firms with less than 30 employees, the research is based on a sample; (2) the sum of total of informants from the different categories can be greater than the informants of the informatics services as a whole because they can be present in more than one

Beyond the official statistics, we can also consider information from consultancy companies specialized in the analysis of the informatics sector. According to IDC Consultancy Group (2006) the Brazilian Software and Services market occupies the 12th position in the global market (considering the amount of domestics sales) in 2005, with sales of approximately 7,45 billion dollars, equivalent to 0,95% of the Brazilian GDP that year and of 1,1% of the global market (see Figure 2). Out of this total, 2.72 billion referred to software licensing, which represented close to 1.2% of the global market and 41% of the Latin American market. The remaining 4,69 billion referred to related services. According to IDC, an amount of 7.760 companies would exploited the sector, from which 6,040 would be involved directly with the commercialization of software (being 1,850 dedicated mainly to the development and production of software and 4,190 to resale and distribution activities) and 1,720 mainly oriented to the supply of software services. Among the firms oriented to the development and production of software, 667 could be classified as “micro” firms (with less than 10 employees), 1,075 as small firms (between 10 - 99 employees), 94 as medium firms (between 100 - 500 employees) and only 14 as large firms (with more than 500 employees). Studies also forecast an annual average growth rate over 11% by 2009. Despite the fact that the Brazilian software market is still largely supplied with foreign software, whose participation is about 71%, forecasts indicate that this participation should drop to 66% in the next five years, with increase of domestic products participation in the market.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 17 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Figure 2 – Main Brazilian Software markets Indicators - 2005

Source: IDC(2006)

Data divulged by SEPIN-MCT for the year 2002 also show that, for an amount of 1406 firms directly involved with the development of software, more than 62% have five or less employees. On the other hand, 1,7% of those firms have more than 100 employees and only 0,2% more than 500 employees (see Table 4). Information extracted from micro data of PAS2 for the year 2002 by Roselino (2006) also indicate that approximately 87% of the firms in the informatics sector as a whole had less than 5 employees. This share increases to 88% in the case of firms directly involved with software development. Contrasting with these figures, only 0,4% of the firms had more than 100 employees in the informatics sector and only 0,8% in the segment of software development (see Table 5). Finally, data about RAIS-MT for the year 2005 - which are based on information of the formal labor market disaggregated trough different CNAE (equivalent to the international ISIC) industrial classification – indicate that more than 60% of the firms included in the segments of software development have five or less employees (see Table 7). All these information stress the relevance and preponderance of micro and small firms in the industrial structure of the software sector.

2 These data are based on a different methodology of systematizing the information, involving an expansion of the original basis to include, through estimation, all the firms with less than one employee (even the self employment firms). So, these data are not directly compared with other data extracted from PAS presented in Table 3. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 18 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 4 – Distribution of software firms according to company’s size Company’s size 1000 (no of 5 or 10 to 20 to 50 to 100 to 250 to 500 to or employees) less 6 to 9 19 49 99 249 499 999 more Total Absolute values (number of firms) Informatics 15.64 activities 9.384 1.883 1.402 713 213 112 58 24 17 7 Software activities 699 216 167 98 27 17 4 2 1 1406 package software 247 64 63 38 9 4 0 0 0 489 custom made software 452 152 104 60 18 13 4 2 1 917 Distribution (%) Informatics 100,0 activities 71,7% 12,0% 9,0% 4,6% 1,4% 0,7% 0,4% 0,2% 0,1% % Software 100,0 activities 62,2% 15,4% 11,9% 7,0% 1,9% 1,2% 0,3% 0,1% 0,1% % 100,0 package software 63,6% 13,1% 12,9% 7,8% 1,8% 0,8% 0,0% 0,0% 0,0% % custom made 100,0 software 61,4% 16,6% 11,3% 6,5% 2,0% 1,4% 0,4% 0,2% 0,1% % Source: MCT – SEPIN (2002)

Table 5- Distribution of software firms according to company’s size (2002) CNAE Classes (0-4) (5-9) (10- (20- (50- (>100) Total Total 19) 49) 99) 7210 –Hardware Consultancy 92,7% 3,6% 2,5% 0,7% 0,3% 0,3% 100,0% 8.071 7220 –Software Development 88,0% 5,7% 1,4% 3,1% 0,9% 0,8% 100,0% 10.064 (package and custom made) 7230 – Data processing 89,1% 5,4% 3,4% 1,4% 0,3% 0,4% 100,0% 12.330 services 7240 – Database and on-line 67,7% 0,0% 18,3% 8,6% 2,2% 3,2% 100,0% 93 electronic services 7250 – Informatics 78,2% 18,7% 1,8% 0,9% 0,2% 0,2% 100,0% 10.700 maintenance and repairing Total 86,7% 8,5% 2,4% 1,6% 0,4% 0,4% 100,0% 41.258 Source: data from PAS-IBGE compiled by Roselino (2006)

Table 5- Distribution of software firms according to company’s size (2005) Betwe More en 5 - 20 to 100 to than No of 19 99 499 500 establi Less emplo emplo emplo emplo shmen CNAE Classes then 5 yees yee yee yees Total ts Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 19 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

100,0 (i) Hardware Consultancy 64,1% 25,0% 8,0% 2,3% 0,5% % 1.904 (ii) Development of package 100,0 software 59,6% 27,3% 11,8% 1,2% 0,1% % 755 (iii) Development of custom made 100,0 software 63,3% 24,5% 9,7% 2,0% 0,5% % 1.409 100,0 (iv) Data processing services 57,5% 36,2% 5,5% 0,7% 0,1% % 7.600 100,0 (v) Database activities 67,2% 23,1% 7,1% 1,3% 1,3% % 238 (vi) Informatics maintenance and 100,0 repairing 72,0% 21,9% 5,3% 0,7% 0,1% % 3.018 100,0 (vii) Other informatics activities 67,1% 25,3% 6,1% 1,3% 0,2% % 4.813 100,0 TOTAL 63,3% 28,9% 6,4% 1,1% 0,2% % 19.737 Source: data extracted from RAIS-MT (2005)

4.2 – Software Labour market

The inherent difficulties to map the characteristics of Brazilian software market remain strong when we consider characteristics of the labour market. Specifically, we can identify a movement of workers across different segments of informatics industry according to the competitive pressures they faced. In the case of the software segment, we have to go back until the period of liberalization of IT industry to understand the dynamics of jobs creation. When the government removed the protection of the IT hardware sector, in 1991, many specialized workers, who had been trained on hardware issues, or who were working in that field, turned to software development. By this period, we could already found a competent software development core working on operational system issues, but not much of a software industry, except that related to data processing bureaus. So, a lot of new companies oriented to software development have started up during the nineties. By 1995, software development companies were employing almost five thousands staff, half of them with university degrees and no more to 15 with PhD degrees. With the turn of the century the software industry has maintained its capacity of generating new jobs. Information extracted from PAS presented in Table 6 show that software segments employed around 60 thousand workers in 2003, which corresponds to 41% of total workers in the informatics sector. According to information collected by PAS firms in the package segment have a mean of 60 employees per firm, contrasting with a mean of 150 employees per firm in the custom made segment.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 20 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 6 – Jobs generated by Informatics enterprises surveyed by PAS - IBGE- 2003 % of the Number number Size Number of Segment of (employe of firms employee employee es per s s firm) Informatics 1 038 144 954 100,0% 140 Hardware Consulting 91 29 261 20,2% 322 Development and edition of package software 169 10 155 7,0% 60 Development and edition of custom made software 331 49 739 34,3% 150 Data processing , scanning and web hosting 338 46 809 32,3% 138 Data base activities, on-line editions and web design 23 1 918 1,3% 83 Other informatics activities 86 7 072 4,9% 82 Source: IBGE, Annual Research about Services (Pesquisa Anual de Serviços -PAS) 2003. Observations: (1) The research covers the entire universe composed by firms with more than 30 employees. To firms with less than 30 employees, the research is based on a sample

Data extracted from RAIS-MT3 permit to analyze the importance of informatics and software activities in the entire labour market. The informatics industry accounts for only 0,7% of all formally employed Brazilian labour, employing 219 thousands workers in 2005. Concerning specifically the development of software (packaged or custom made) data from RAIS indicate that approximately 35 thousands workers were employed in 2005, corresponding to 0,15% of the more then 33 million Brazilians formally employed in 2005. According to RAIS the average size of a firm in the informatics industry reached 11 employees per firm, being higher in the segment of custom-made software compared with the segment of package software. Data from RAIS presented in Table 7 also show that almost 34% of the jobs created in formal firms of the informatics sector were concentrated in firms with less then 20 employees. Comparing the two segments directly involved with software development, we can see that the share of firms with less then 20 employees is higher in package software; meanwhile, in custom made software segment almost 39% of the jobs are concentrated in firms with more than 500 employees.

3 The Brazilian Ministry of Labour and Employment, through the RAIS database, collects annual data on formally employed workers by all companies in the whole spectrum of economic activity. In any year, software companies may have hired formally employees who are not longer active by the reporting date, when such employees are said to be inactive. In this research piece, aimed at assessing the overall size of the Brazilian software sector, both active and inactive employees are considered. This database has also some informational constraints. First, it covers only the formal labor market and recently the relative share of the informal labor market has been increasing in Brazil. Second, its data is less reliable for regions and activities where the labor market is less organized. Third, as software activities are usually only a part of firms´ IT business, data about this sector informed by RAIS include only the firms that selected software as its core business. As a consequence, data from these source tend to under valuate the figures of software industry, particularly to the package segment where the largest firms have a diversify presence in IT industry Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 21 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 7 - Distribution of employees according to company’s size (2005) Betwe More Size en 5 - 20 to 100 to than of Less 19 99 499 500 No of firms then emplo emplo emplo emplo empl No of CNAE Classes 5 yees yee yee yees oyee frims 32.77 (i) Hardware Consultancy 5,7% 13,5% 18,3% 30,7% 31,8% 0 1.904 17,2 (ii) Development of package software 8,6% 25,2% 39,3% 20,3% 6,6% 8.193 755 10,9 (iii) Development of custom 26.51 made software 5,0% 12,1% 21,2% 22,9% 38,8% 1 1.409 18,8 73.67 (iv) Data processing services 11,7% 32,3% 20,7% 16,3% 19,1% 2 7.600 9,7 (v) Database activities 4,3% 8,4% 9,2% 7,5% 70,6% 6.099 238 25,6 (vi) Informatics maintenance and 22.09 repairing 15,1% 26,4% 27,0% 22,2% 9,4% 7 3.018 7,3 49.97 (vii) Other informatics activities 10,9% 21,4% 22,9% 26,9% 17,8% 9 4.813 10,4 219.3 19.73 TOTAL 9,8% 23,0% 21,9% 22,2% 23,1% 21 7 11,1 Source: data extracted from RAIS-MT (2005)

Because of the high level of qualification, we could expect to find well-paid jobs in the software industry, particularly when compared to general figures of Brazilian labour market. According to information collected by PAS-IBGE, the firms in the software segment spent around US$ 15.000 with each one of its workers in 2003, from which approximately US$ 9.600 was directly spent with salaries (see Table 8). The salaries paid in segments oriented to the development of software are usually higher than the salaries paid in informatics sector as a whole, being particularly higher in the package segment, compared with the custom made segment.

Table 8 – Jobs and salaries paid by Informatics enterprises surveyed by PAS - IBGE- 2003 Annual Total Total Numbe Annual expenses expens expens r of expenses with Segment es whit es whit employ per salaries employ salarie ees employe per ees s es employees 144 1 794 1 103 Informatics 954 405 831 12 379 7 615 29 247 148 Hardware Consulting 261 735 756 8 466 5 084 Development and edition of package 10 193 117 software 155 854 917 19 089 11 612 Development and edition of custom 49 712 457 14 318 9 205 Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 22 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

made software 739 158 858 Data processing , scanning and web 46 507 296 hosting 809 619 927 10 844 6 343 Data base activities, on-line editions and 29 17 web design 1 918 634 520 15 451 9 135 103 64 Other informatics activities 7 072 405 853 14 622 9 170 Source: IBGE, Annual Research about Services (Pesquisa Anual de Serviços -PAS) 2003. Observations: (1) The research covers the entire universe composed by firms with more than 30 employees. To firms with less than 30 employees, the research is based on a sample

Data about formal labour market extracted from RAIS show a similar portrait about the salaries paid in the software industry. The data presented in Table 9 indicate that the annual salary of an employee in software industry was around US$ 10.000 in 2005, which is higher than the salaries paid in informatics sector as a whole (US$ 8.700) and much higher than salaries paid for all the spectrum of all formal economic activities (around US$ 5.400). Comparing the two segments of software development, data from RAIS show that, contrary to data from PAS, salaries in the custom made segment tended to be higher. This difference can be explained by the fact that data from RAIS are based only on the formal firms that are active and specialized in the custom made segment, probably paying higher salaries than those surveyed by PAS.

Table 8 – Salaries paid by Informatics activities - 2005 Monthly Annual Amount of Salary per Salary per Annual Salary Monthly employee employee per employee Employees Salaries (R$) (R$) (R$) (US$) Informatics Activities 219.321 379.201.420 1.728,98 20.747,7 8.703,5 (i) Hardware Consultancy 32.770 75.188.327 2.294,43 27.533,1 11.549,9 (ii) Development of package 1.702,73 software 8.193 13.950.450 20.432,7 8.571,4 (iii) Development of custom 2.159,96 made software 26.511 57.262.792 25.919,6 10.873,1 (iv) Data processing services 73.672 96.361.537 1.307,98 15.695,8 6.584,3 (v) Database activities 6.099 16.267.483 2.667,24 32.006,9 13.426,6 (vi) Informatics maintenance and 1.565,52 repairing 22.097 34.593.240 18.786,2 7.880,7 (vii) Other informatics activities 49.979 85.577.592 1.712,27 20.547,3 8.619,4 All economic activities - 1.078,92 BRAZIL 33.238.617 35.861.972.723 12.947,1 5.431,2 Source: data extracted from RAIS-MT (2005)

These data suggest that, compared with other economic activities in Brazil, the software industry can be seen as an elite sector in terms of the qualification of its manpower. In fact, the level of schooling in software companies is substantially higher than the national average: a full 93% of the labour force of software companies has at least a complete secondary education (eleven full years of schooling) in 2005 while only 53% of Brazilian employed labour shows a comparable Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 23 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

level of schooling (see Table 9). The share of employees with less than four years of education is almost zero (0,4%) in the software industry, compared with a share of almost 6% for the entire spectrum of economic activities. Further, the share of labour with at least a university degree is almost the triple of the comparable average share for Brazil. Compared with other segments of informatics sector, the shares of employees with complete and incomplete university education is higher in the segments oriented the development of software, especially in the custom made segment. On the other hand, a research coordinated by MCT-Sepin4 in 2001, with a sample of 446 firms, indicates that there was 2,5 post-graduated workers per firm.

Table 9 - Distribution of employees according to the level of schooling (2005) Full 8 Full 4 years Incom years and plete Unde and under univer BA r 4 under 11 Full 11 sity and years 8 years years years studies above Total Employees Informatics Activities 0,4% 2,6% 10,3% 41,3% 14,7% 30,6% 100,0% 219.321 (i) Hardware Consultancy 0,3% 2,2% 7,0% 33,8% 15,8% 41,0% 100,0% 32.770 (ii) Development of package software 0,7% 1,8% 7,9% 30,9% 19,1% 39,6% 100,0% 8.193 (iii) Development of custom made software 0,2% 1,1% 4,7% 28,1% 20,4% 45,4% 100,0% 26.511 (iv) Data processing services 0,6% 3,5% 13,8% 45,3% 14,3% 22,5% 100,0% 73.672 (v) Database activities 0,3% 3,1% 8,3% 38,1% 7,2% 43,1% 100,0% 6.099 (vi) Informatics maintenance and repairing 0,5% 2,9% 11,8% 52,8% 11,6% 20,4% 100,0% 22.097 (vii) Other informatics activities 0,3% 2,2% 10,4% 44,3% 13,3% 29,5% 100,0% 49.979 All economic activities - 33.238.617 BRAZIL 5,7% 17,4% 24,3% 33,4% 4,1% 15,1% 100,0% Source: data extracted from RAIS-MT (2005)

4.3 - Industrial structure

The origins of Brazilian software companies are quite diverse. We can observe an intense process of firm creation during the previous decade, but there are some evidences that this process lost part of the impetus during the last years, being replaced by a tendency of firms´ consolidation and specialization. According to a survey coordinated by Botelho et alli (2002) the majority of software companies were established from existing firms, part as a true “spin-off” of the mother company and the majority as independent companies created by employees of the mother company that spotted a new opportunity in the course of their work. Three others alternatives

4 MCT – Ministério da Ciência e Tecnologia, Sepin – Secretaria de Política de Informática. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 24 http://brics.redesist.ie.ufrj.br

BBRRIICCSS seem to be relevant as patterns of firm creation in Brazilian software industry. First, there are a lot of individual start-ups created by the talented entrepreneurs to explore new opportunities emerged form the industry growth. Second, there are a limited number of university spin-offs, most of them oriented to segments where the previously market experience seems not to be indispensable. Finally, we can mention the prevalence of strong government software firms, usually specialized in the manipulation of a high volume of data (SERPRO, as example).

The main focus of Brazilian software industry is the internal market, contrasting with the international market orientation of software industry in countries like India, Ireland and Israel. The emphasis on internal market must be explained by the huge internal demand and by the sophistication of productive structure of Brazilian economy, despite the macroeconomics problems it have faced during the last two decades. The extensive diffusion of informatics since the 70´s and the explosive growth of IT in the last part of the 80`s and 90`s generated a string demand for virtually all kind of software products and services. This tendency must be associate with an impressive growth of expenses with IT as a share of GDP, which reached an amount of 8,3% in 2001, against 7,9% in USA, 5,7% in China and 3,9% in India [Veloso et al. (2003, p.7)]. Considering that software segment was responsible for around 42% of IT markets in 2001, we can show evidences of the hardness of software internal market and of its potential to generate an impressive demand for product and related services.

According to information of the International Data Group -IDC the Brazilian software market was estimated in around USS 7,4 billion for the year 2005, being integrated in a wider market of Information Technologies (ITs) estimates in around 11,9 billion (see Table 10). Data from IDC can also be used to stress some distribution of software sales among different segments of products and services in 2005. A first analytical cut may be based on the kind of software that was produced. In this sense a distinction between package, parameterizable and custom-made software5. Table 11 shows that almost two third of the software sales were associated with "parameterizable software" with mixed characteristics of package and strictly custom-made software. Yet, when the origin of software is considered (see Table 11), we can observe that 71% of software commercialized was developed abroad. The local production of custom made software was responsible for 20,5% of the software sales, corresponding to an amount of around USS 558 millions in 2005. In the segment of software services, almost the totality (97%) of the sales was oriented to the local market, being particularly restricted the amount of exports.

5 According to the definition used in this survey, the followed class are considered :(i) Standard Software: products that can be installed by the user itself, without need of additional services. In this market we have operating systems, productivity suites, entertainment and collaboration consumer products, among others; (ii) Parameterizable Software: products that require the hiring of additional services for their installation and parameterization; (iii) Custom Developed Software: systems developed according to the specifications of a single user and that will meet the needs of that user alone; (iv) Services: are technical services added to the software, such as consultancy, training, support, data entry, data processing, development and maintenance of internet page content, etc Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 25 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 10- Segmentation of Software Market by Class (2005) Class Volume (Us$ Share (%) Growth millions) 2005/2004 (%) Standard Software 397,4 14,6% + 13,5 % Parameterzable Software 1.764,1 64,8% +11,9 % Custom Developed Software 558,5 20,5% +28,5 % Software Subtotal 2.720 100% +15,2 % Services Subtotal 4.690 100% + 29,5 % Software and Services Total 7.410 --- + 24 % Source: ABES-IDC (2006)

Table 10- Segmentation of Software Market by Origin (2005) Origin Volume (Us$ Share (%) Growth millions) 2005/2004 (%) Foreign Production 1.924,1 71,0 11,6 Domestic Custom Development 558,5 20,5 28,5 Domestic Standard Production 202,1 7,3 14,2 Domestic Production for Export 35,3 1,2 38,9 Software Subtotal 2.720 100 15,2 Domestic Market Services 4.548 97 29,2 Services Export 142 3 40,5 Services Subtotal 4.690 100% 29,5 Software and Services Total 7.410 --- 23,9 Source: ABES-IDC (2006)

The sophistication of productive structure of software industry in Brazil can be associated with the distribution of firms and sales among its different market segments. These segments can be grouped and classified according to vertical or horizontal criteria. Concerning a horizontal criteria, data collected by Gutierrez and Alexandre (2004) show that approximately 36% of sales generated by software products in 2002 were obtained in the infra-structure segment, followed by the generation of software tools (see Table 11). Data collected from the same source indicate that approximately 31,1% of sales generated by software services in 2002 were obtained in by outsourcing services, followed by integration services (26,4% ) and support services (13,4%). More recently data from IDC Consulting Group (2006) indicate that approximately 47% of the sales of software products in 2005 are from applications for specific users, followed by sales of infrastructure software (including management software for systems and networks, security software, storage and backup software and operating system software) with 33% of the sales of software products (see Table 12). Concerning software services, the sales are almost equally distributed between system integration (development of IT solutions to meet technical specifications defined by the customer to meet its business needs), outsourcing (in which a service provider external to the organization assumes responsibility for the management and operation of all or part of the customer's IT infrastructure) and support (software installation, customization and configuration, as well as technical support services to users).

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 26 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 11 – Sales of software products and services segments (2002) Segment % Software products Infrastructure 36 Software tools 29 General applicative 21 Corporate applicative 15 Software services Outsourcing 31,1 Integration 26,4 Support 13,4 Consultancy 5,4 Training 5,2 development 4,8 Others 13,6 Source: Série Estudos Tecnologias da Informação – software 92003) adapted by Gutierrez and Alexandre (2004) Table12 - Sales of software products and services according to horizontal segments (2005) Volume (US$ % Growth 2004- millions) 2005 Applications 1.285,5 47,2% + 19,6 % Development Environments 539,4 19,8% +10,3 % Infrastructure 895,1 32,9% +12,3 % Software Subtotal 2.720 100% + 15,2 % Consultancy 562,8 12,0% + 19,6 System integration 1.360,1 29,0% +83,7 % Outsourcing 1.360,1 29,0% + 21,7 % Support 1.266,3 27,0% + 10, 4 % Training 140.7 3,0% - 3,1 % Services Subtotal 4.690 100,0% + 29,5 Software and Services Total 7.410 --- + 24 % Source: ABES-IDC (2006)

We can also consider data collected in a survey coordinated by Mayer & Bunge Informatics Consulting in 2004 with 461 firms, equivalent to approximately 16,5% of the universe of Brazilian software firms. Data presented in Table 13 show how the firms distribute themselves among the diverse horizontal markets of software industry. Corporate ERP, customers’ management and systems´ management seem to be the most important horizontal segments, stressing the connection between software firms and the productive basis of Brazilian economy. When we compare data from Brazil with Latin America as a whole, we can note that the number of firms per each horizontal segment is expressively minor in Brazil, indicating that the software firms tend to specialize themselves to attend properly the demands of sophisticated users. In fact, the presence of sophisticated clients contributes to stimulate a process of market specialization of software firms, from which technological and organizational competences might be accumulated. The majority of the leading companies have their business model based on a core product, but related services assure the larger slice of their revenues. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 27 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 13 – Distribution of firms surveyed according to vertical segments of software industry (2004) Latin Segments Brazil % America % BR/LA Business Intelligence and Data Warehousing 55 8,2% 134 8,8% 41,0% Corporate ERP 125 18,7% 215 14,1% 58,1% Management of relations with customers 117 17,5% 208 13,6% 56,3% Electronic commerce 45 6,7% 122 8,0% 36,9% Information publishing 13 1,9% 79 5,2% 16,5% Infrastructure 23 3,4% 83 5,4% 27,7% Knowledge management 43 6,4% 91 6,0% 47,3% Medium size ERP and accountancy 74 11,1% 164 10,7% 45,1% Data security 18 2,7% 84 5,5% 21,4% Software tools 32 4,8% 116 7,6% 27,6% Supply chain 12 1,8% 57 3,7% 21,1% Systems management 104 15,6% 163 10,7% 63,8% Others 6 0,9% 11 0,7% 54,5% Total 667 100,0% 1528 100,0% 43,7% Number of firms surveyed 461 682 Number of segments per firm 1,4 2,2 Source: Panorama da Industria latino-americana de software ; MBI Mayer & Bunge Informatica Ltda (2005)

The software markets may also be classified according to vertical market criteria. These criteria considerer the productive segment attended by software products and services. Table 14 indicates that financial services, general services and industry are the segments that concentrated the larger number of firms in the sample. Although they offer services for diversified areas, the majority of the activity is concentrated in a few sectors, with which they have been working for some time and for which they develop sophisticated solutions. When we consider the share of each segment in the total sales of the sample, the importance of the financial sector and of the public sector tends to increase. In fact, since the 70´s the financial segment and the public sector have been responsible for a huge demand of software products and services. According to data from SOFTEX (2002), in 2001 the demand from the financial segment corresponds to approximately R$ 3 billion and from public sector of around R$ 2 billion. Recently information collected by IDC Consulting Group (2006) confirms this trend, indicating that approximately an half of software sales are divided between industry and finances services, followed by services, trade, government, agricultural industry and others (see Table 15). IDC also points out that, from the point of view of software users, IT investments is supposed to increase by 16% in 2006.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 28 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 14 – Distribution of firms surveyed and sales according to horizontal segments of software industry (2004) No of firms in Segments the segment % of total firms % of total sales Education 32 4,3% 2,6% Financial services 122 16,2% 32,1% Healthy 50 6,7% 4,3% Industry 152 20,2% 16,9% Media and entertainment 11 1,5% 3,2% Non-profit organizations 3 0,4% 0,5% Public sector 80 10,7% 14,0% Retail and hospitality 98 13,0% 5,2% Services 161 21,4% 18,0% Telecommunications 32 4,3% 3,2% Others 10 1,3% nd Total 751 100,0% 100,0% Source: Panorama da Industria latino-americana de software ; MBI Mayer & Bunge Informatica Ltda (2005)

Table15- Sales of software products and services according to vertical (2005) Volume (US$ % Growth 2004-2005 millions) Industry 742,4 27,6% + 15,8 % Commerce 204,4 7,6% +13,6 % Agricultural industry 28,7 1,0% +15,7 % Government 172,4 6,4% + 16,7 % Finances 607,9 22,7% + 15,2 % Services 377,2 14,0% + 13,4 % Oil and Gas 70,6 2,6% + 14,0 % Others 481,4 18,0% + 14,8 % Total 2.685 100% + 15,0 % Source: ABES-IDC (2006)

The structure of Brazilian software followed a pattern in which the multinational enterprises tend to reinforce their positions in the horizontal market, leaving for the Brazilian companies the vertical segment, as is the case of national banking software. In fact, software firms often have various lines of business and simultaneously develop service and product activities, with a greater or lesser preponderance of each of them, a pattern that can be found in countries in which the majority of companies develop services, like Brazil, India or China. Since national companies that develop software are small-sized and that competition in this sector is closely related to concentration and fragmentation, the market strategies may follow the following alternatives: i) “niche strategy”, when a company tries to specialize in meeting the needs of a particular group of clients, accumulates knowledge/information that will allow for the development of products that incorporate specific functions; and ii) “interstice strategy”, where companies must take advantage Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 29 http://brics.redesist.ie.ufrj.br

BBRRIICCSS of the multidimensional character of software for the implementation of product differentiation in order to occupy markets of large proportions, left behind by leading companies, keeping in mind that the line of products of the later will never have total inclusion, offering all possible varieties.

Although the possibility of generating continued gains from those strategies, we can identify a general process of industry concentration in Brazilian software industry. The increase of international competition and the sophistication of consumer demands - elements that tend do reinforce the importance of availability of financial resources and distribution channels - contribute to this process, making harder the maintenance of market shares To local firms the impacts of these tendencies might be disruptive: large users tend to adopt a risk averse purchasing attitude that favors the selection of established foreign suppliers, whereas small and medium sized enterprises (SMEs) are not willing to pay a premium for the domestic firms’ unique knowledge. These tendencies might also be observed from data extracted from annual surveys about the Brazilian software industry. Table 16 shows a list of software firms that was classified between the 200 largest firms of the Informatics sectors for the years between 2002-2004, according to an annual inquired organized by the “Exame Informática” magazine. The data reflect the process of industrial concentration and internalization. So, between 2002 and 2004 the number of software firms between the largest of the Informatics sector decreased from 33 to only 20 firms. On the other hand, the total sales of the firms included in the list increase from US$ 128 billion in 2002 to US$ 1,68 billion in 2004. As a consequence of the market concentration, the sales per firm included in the list also increased from US$ 38,9 million in 2002 to US$84 million in 2004. There was also evidence that the internalization process has maintained its impetus. In fact, in 2004 only six firms between the twenty listed could be classified as domestic software firms and one of them (Veritas – SP) was acquired by a multinational company (Symantec) in 2005.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 30 http://brics.redesist.ie.ufrj.br

Table 16 – Largest Software Firms in Brazil (2002-2004) – sales in US$ 1000

2004 Firm Origin Sales 2003 Firm Sales 2002 Firm Sales 1 MICROSOFT SÃO PAULO (SP) USA 365.400 1 MICROSOFT - São Paulo (SP) 301.134 1 MICROSOFT 299.819 2 COMPUTER ASSOC. S.PAULO (SP) USA 333.000 2 COMPUTER ASSOC. - São Paulo (SP) 277.700 2 COMPUTER 245.725 ASSOCIATES 3 ORACLE SÃO PAULO (SP) USA 208.207 3 ORACLE - São Paulo (SP) 144.559 3 ORACLE 158.259 4 SAP SÃO PAULO (SP) GER 169.173 4 SAP - São Paulo (SP) 123.532 4 SAP BRASIL 112.624 5 CONSIST SÃO PAULO (SP) USA 107.369 5 CONSIST - São Paulo (SP) 84.671 5 CONSIST 76.789 6 CPQD CAMPINAS (SP) BR 78.421 6 CPQD - Campinas (SP) 59.650 6 CPQD 56.290 7 PEOPLESOFT SÃO PAULO (SP) USA 75.000 7 PEOPLESOFT - São Paulo (SP) 59.340 7 MICROSIGA 36.417 8 MICROSIGA SÃO PAULO (SP) BR 65.303 8 DATASUL - Joinville (SC) 41.768 8 DATASUL 35.220 9 DATASUL JOINVILLE (SC) BR 56.316 9 MICROSIGA - São Paulo (SP) 40.950 9 PEOPLESOFT 29.000 10 RM SISTEMAS B.HORIZONTEE (MG) USA 41.556 10 SYMANTEC - São Paulo (SP) 34.000 10 RM SISTEMAS 23.734 11 SYMANTEC SÃO PAULO (SP) USA 37.500 11 RM SISTEMAS - Belo Horizonte (MG) 30.548 11 J.D. EDWARDS 22.600 12 ADOBE SYSTEMS SÃO PAULO (SP) USA 24.000 12 COREL - São Paulo (SP) 30.000 12 NOVELL 20.000 13 DÍGITRO FLORIANÓPOLIS (SC) BR 22.090 13 VERITAS - São Paulo (SP) 19.500 13 VERITAS 15.000 14 VERITAS SÃO PAULO (SP) BR* 20.000 14 DIGITRO - Florianópolis (SC) 14.433 14 SYMANTEC DO 14.000 BRASIL 15 SPRING WIRELESS SÃO PAULO (SP) USA 19.694 15 SSA GT - São Paulo (SP) 14.390 15 LOGOCENTER 13.259 16 SAS INSTITUTE BR- S.PAULO (SP) USA 15.300 16 LOGOCENTER - Joinville (SC) 13.284 16 SSA GT 12.995 17 BEA SYSTEMS SÃO PAULO (SP) USA 12.000 17 SENIOR SISTEMAS - Blumenau (SC) 8.716 17 EVERSYSTEMS 11.695 18 MCAFEE SÃO PAULO (SP) USA 10.700 18 INTERQUADRAM - Rio de Janeiro 8.467 18 CIASC 11.582 (RJ) 19 BENNER SÃO PAULO (SP) BR 10.253 19 PROGRESS - São Paulo (SP) 6.813 19 ADOBE SYSTEMS 11.000 20 PROGRESS SÃO PAULO (SP) USA 8.977 20 CETIL SISTEMAS - Blumenau (SC) 5.936 20 DÍGITRO 10.812 TOTAL 1.680.259 21 BENNER - Blumenau (SC) 5.839 21 INTERQUADRAM 7.508 22 MICROSTRATEGY - São Paulo (SP) 5.403 22 SENIOR SISTEMAS 6.469 23 STERLING COMMERCE - São Paulo 5.368 23 BORLAND 6.100 (SP) 24 PROCENGE - Recife (PE) 5.251 24 MICROSTRATEGY 5.820 25 MEGA SIST. CORPorativos - Itú (SP) 5.227 25 PROCENGE 5.562 26 STERLING 5.225 TOTAL 1.346.479 COMMERCE 27 CETIL SISTEMAS 5.216 28 PROCESSOR 4.700 29 BENNER 4.683 30 SISPRO 4.416 31 NEXXY CAPITAL 4.198 32 MEGA SISTEMAS 3.969 33 MASTERMAQ 3.468 TOTAL 1.284.154 Source: data extracted from “Exame Informática” magazine compiled by the Authors

BBRRIICCSS

These tendencies point out a challenge for domestic software firms. Due to the increasing competitive pressure of foreign firms on their vertical niches they may try to stay ahead of the foreign competitors by constantly entering new growth market segments. These markets should be large enough to generate margins to finance the development of component-like products, which can assist suppliers in facing up foreign competitors when these niche markets mature into full blown markets. Local firms should also be able to provide a superior service and semi-customized products at low cost to an expanding number of SMEs, especially in areas where these are also growing to become larger and more competitive, perhaps with international ambitions. They can also try to ship software/software components and services of high added value, followed by customizable products. A corollary of this process is the attempt to maintain and improve relationship with more profitable customers. In this sense, the survey coordinated by Mayer & Bunge Informatics in 2004 shows that 54,4% of the firms interviewed maintained systematic market relationships with no more than two corporate customers (see Table 17).

Table 17 – Distribution of firms surveyed according to the number of customers(2004) No of corporate customers % of firms of the sample 1 or 2 54,4 3 or 4 14,0 5 to 8 10,8 9 to 15 10,0 16 to 31 9,8 more than 32 1,0 Total 100,0 No of medium enterprises as clients % of firms of the sample less than 15 77,3 16 to 60 17,5 61 to 249 3,0 more than 250 2,2 Total 100,0 Source: Panorama da Industria latino-americana de software ; MBI Mayer & Bunge Informatica Ltda (2005)

4.4 - Regional Distribution of Software Activities

The software industry has been characterized by significant spatial concentration in developing countries. Concerning this trend, the economic geography literature has argued that it is the combination of economies of scale and transaction costs that determine this process of concentration. The former are necessary for concentration to happen at all and the latter will limit concentration by raising the benefits to locating production close to demand. Reproducing a pattern that can also be observed in counties like India and China, there is also a spatial concentration of software industry in Brazil. In fact, according to micro data from PAS-IBGE presented by Roselino (2006), more than 60% of the employments in informatics activities (including software) are concentrated in the southeast

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 32 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

of the country (see Table 18). Moreover, the same source demonstrate that almost 80% of the sales revenues in informatics activities come from southeast, while more than 70% of informatics firms are located in this region. Beyond that, those firms have a high level of labour productivity calculated as a proxis by the amount of sales revenues divided by the number of employees, The middle west region follow that region in terms of sales revenues and workers. This trend can be explained by the presence of large firms oriented to heavy bureaucratic dataprocessing services, due to the presence of the central government in Brasília. Because of this, middle west companies have sales revenues and number of workers expressively higher than the general means of the country.

Table 18 – Distribution of Software Activities by Geographical Regions (2002) Region Total sales % Number % Employees % a/b a/c c/b (a) of firms (c) (b) Southeast 15.955.217.994 79,34 29.402 71,26 157.344 61,79 542.639 101.403 5,35 South 1.428.668.055 7,10 6.653 16,12 32.901 12,92 214.740 43.423 4,95 Middle 2.085.955.576 10,37 1.923 4,66 45.002 17,67 1.084.740 46.353 23,40 West North 69.911.918 0,35 305 0,74 2.076 0,82 229.219 33.676 6,81 Northeast 569.603.496 2,83 2.975 7,21 17.327 6,80 191.463 32.874 5,82 Total 20.109.357.039 100 41.258 100 254.650 100 487.393 78.969 6,17 Source: Roselino (2006) based on data from PAS-IBGE (2002)

Data about the distribution of informatics activities among the different states of Brazilian federation reinforce these figures. The regional distribution of employment is strongly concentrated in most heavily industrialized core of the country, including the states of São Paulo, Rio de Janeiro and Minas Gerais (see Table 19). Aggregated with the Brasília Federal District, those states are responsible for more than 87% sales revenues and for 70% of the jobs generated in the informatics activities. São Paulo alone is responsible for almost an half of firms and sales generated by software activities. The south of the country accounts for 17% of formal employment, mostly concentrated in the states Rio Grande do Sul and Santa Catarina. Concerning an index of labour productivity calculated by the amount of sales revenues divided by the number of employees, high levels can be found in the states of Rio de Janeiro and São Paulo. The number of employees per firm is usually low, with the exception of Brasília due to the presence of state-owned companies in the capital, as well as large private firms oriented to the attendance of government demands.

Table 19 – Distribution of Software Activities by Federative States(200) Federat Number % Sales Revenues % Employ % Size Sales per ive of Firms (R$) ees (employ Emplyoyee States ee per firm) SP 19.845 48,1 9.279.701.977 46,1 95.059 37,5 4,79 97.620

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 33 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

RJ 5.400 13,1 5.954.684.491 29,6 43.131 17,0 7,99 138.060 DF 890 2,2 1.510.053.083 7,5 23.543 9,3 26,45 64.140 MG 3.604 8,7 646.269.812 3,2 14.540 5,7 4,03 44.448 RS 2.745 6,7 636.922.603 3,2 13.579 5,4 4,95 46.905 SC 2.036 4,9 466.033.380 2,3 9.092 3,6 4,47 51.258 GO 546 1,3 410.920.283 2,0 8.431 3,3 15,44 48.739 PR 1.871 4,5 325.712.072 1,6 11.167 4,4 5,97 29.167 BA 928 2,2 229.522.154 1,1 6.146 2,4 6,62 37.345 PE 834 2,0 182.970.399 0,9 4.385 1,7 5,26 41.726 Others 2.559 6,2 466.566.785 2,3 24.092 9,5 9,41 19.366 TOTAL 41.258 100,0 20.109.357.039 100,0 253.165 100,0 6,14 79.432 Source: Roselino (2006) based on data from PAS-IBGE (2002)

Considering those aspects, we can develop an attempt to map the spatial distribution of agglomerations in Brazilian economic space. The methodology used to map those spatial agglomerations is based on data extracted from the Relação Anual de Informações Sociais (RAIS - Social Information Annual List), the annual survey produced by the Ministry of Labor and Employment (MTb) in 2004. Data from the RAIS-2004 refers to around 34 million registered formal workers on December 31, 2004. File registers contain a lot of useful information, such as: (1) data about jobs and salaries paid in each industrial activity; (2) the location of the industrial activity (municipality, micro-region, state); (3) a detailed classification of economic activities (based on the CNAE industrial classification, which is compatible with the ISIC international classification), from which two segments of software industry were selected (package and custom made software); (4) the type and size of establishment. The methodology uses the micro-region (a geographical cut based an aggregation of municipalities that gravitated around larges cities) in which the industrial activities are located as a spatial reference. The aim of such methodology is to measure, at first, four elements: (1) the agglomeration of activities in the space; (2) labor and income distribution of the spatially agglomerated companies through the two segments of software industry considered (package or custom made software); (3) the amount of salaries paid per year to software employees; (4) the relative specialization of the micro-region in software activities which might be related to an index that compares the share of software activities in the total employment of the region with the share of them in the whole country. . Thirty agglomerations with more than 0,3% of the total employment generated by software sector in 2004 was selected. Table 20 presents the general characteristics of those agglomerations. They were responsible for 94% of the total employment in the software industry, generating 24.761 employments, 6.330 (25,6%) in the package segment and 18.431 (74,4%) in the custom–made segment. Those agglomerations comprise 1.463 firms paying of around USD 8.625 in salaries to employee per year, amount that was bigger in the custom-made segment. Compared with the whole local economies, the software activities are particularly well paid, with salaries being approximately 58% higher. The average size of firms in the agglomerations reaches 16,9 employees per firm To the set of agglomerations selected the relative specialization index in software activities reaches around 1,73. The limited value of this index might be explained by the presence of some agglomerations in micro-regions where that index tends not to be so high because of two

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 34 http://brics.redesist.ie.ufrj.br

BBRRIICCSS alternative situations: (i) a very diversify productive base; (ii) a huge predominance of traditional activities to the jobs generated locally.

From an inter-territorial comparison of the data, some aspects can be pointed out. First, it is observed that, as expected, most of those agglomerations were located in the most developed areas of the country. In fact, eleven agglomerations were located in São Paulo state, comprising approximately 35,3% (25,2% only in the capital of the state) of the jobs generated by the industry. São Paulo was followed by Distrito Federal with 16,5% of the jobs generated in software industry (due to huge presence of the public sector), Belo Horizonte (12,9%) and a set of agglomerations located in the south region of the country (comprising the states of Parana, Santa Catarina and Rio Grande do Sul) which are responsible for 13,3% of the jobs created.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 35 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 20 – Territorial agglomerations of software activities in Brazilian economy - 2004 Package Segment Custom-made segment Software industry % of sala ries No of % of No of No of in Size empl empl firms employ salar cust No of No of (emp % in comparis No of oyees salar oyees - ees - y in om firms emplo loyee total Relat on firms - - y in in Cust Custo custo mad - yees - s per empl ive between Packa Pack pack pack om m m e Total Total salar firm oyees Speci salary in ge age age age made made made soft of of y in ) in of alizat software Softw softw softw softw Softw softwa softw war Softw softwa softw soft softw ion and all are are are are are re are e are re are ware are Index activities Total 667 7.069 7.145 26,8 1.190 19.302 8.754 77,0 1857 26.371 8.322 14,2 100,0 1,00 1,86 23.016 – Fortaleza 10 111 3.309 77,6 13 32 3.475 23,2 23 143 3.346 6,2 0,5 0,31 0,92 26.017 – Recife 11 61 3.969 18,7 26 266 6.488 87,7 37 327 6.018 8,8 1,2 0,63 1,55 29.021 – Salvador 13 17 3.159 2,6 16 647 4.466 98,2 29 664 4.433 22,9 2,5 1,03 1,00 31.030 - Belo Horizonte 54 882 6.604 25,9 77 2.528 3.833 62,5 131 3.410 4.549 26,0 12,9 3,06 0,96 32.009 – Vitória 11 69 6.046 36,3 20 121 5.901 63,1 31 190 5.954 6,1 0,7 0,66 1,32 33.015 – Serrana 2 13 1.322 15,7 6 70 4.659 95,0 8 83 4.137 10,4 0,3 1,19 1,26 33.018 - Rio de Janeiro 46 646 7.759 38,5 115 1.034 7.812 61,7 161 1.680 7.792 10,4 6,4 0,85 1,41 35.004 - Sao Jose do Rio Preto 3 27 3.757 34,6 5 51 5.496 73,4 8 78 4.894 9,8 0,3 0,68 1,30 35.014 - Ribeirao Preto 10 40 3.549 21,9 11 143 6.590 86,9 21 183 5.925 8,7 0,7 1,02 1,34 35.017 – Araçatuba 4 120 3.417 94,5 2 7 3.398 5,5 6 127 3.416 21,2 0,5 3,36 0,91 35.027 – Limeira 3 56 3.993 69,1 3 25 4.263 32,3 6 81 4.076 13,5 0,3 0,80 0,98 35.032 – Campinas 13 178 5.670 26,6 41 492 5.444 72,6 54 670 5.504 12,4 2,5 1,36 0,95 35.046 – Sorocaba 6 70 7.503 77,8 6 20 3.659 12,2 12 90 6.648 7,5 0,3 0,45 1,40 35.050 - Sao Jose dos Campos 9 72 6.434 78,3 8 20 6.163 21,0 17 92 6.375 5,4 0,3 0,41 0,97 35.057 – Osasco 23 407 9.429 36,7 53 702 14.39 72,5 76 1.109 12.57 14,6 4,2 3,10 2,40

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 36 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

8 5 35.060 - Itapecerica da Serra 3 12 4.838 10,3 13 105 7.178 92,8 16 117 6.938 7,3 0,4 0,80 1,49 12.75 12.68 12.69 35.061 - Sao Paulo 112 1.269 7 19,1 246 5.375 3 80,8 358 6.644 7 18,6 25,2 1,93 1,93 35.062 - Moji das Cruzes 5 60 5.559 54,1 18 51 3.709 36,2 23 111 4.709 4,8 0,4 0,69 1,09 41.009 – Maringá 7 61 5.272 64,9 8 33 3.914 28,7 15 94 4.795 6,3 0,4 0,96 1,49 41.011 – Londrina 3 27 2.445 29,7 8 64 4.924 82,7 11 91 4.188 8,3 0,3 0,70 1,17 41.037 – Curitiba 33 197 5.951 49,1 39 204 7.237 55,7 72 401 6.606 5,6 1,5 0,58 1,27 42.008 – Joinville 11 90 9.736 16,9 20 443 7.269 78,6 31 533 7.685 17,2 2,0 3,05 1,71 42.012 – Blumenau 27 491 6.567 75,4 21 160 6.512 24,4 48 651 6.554 13,6 2,5 4,03 1,80 42.016 – Florianópolis 25 146 7.377 33,7 35 287 5.820 60,8 60 433 6.345 7,2 1,6 1,88 1,13 43.016 - Caxias do Sul 10 75 4.093 26,7 20 206 6.152 80,5 30 281 5.603 9,4 1,1 1,58 1,26 10.03 43.023 – Montenegro 2 13 2.937 15,1 2 73 1 95,0 4 86 8.959 21,5 0,3 2,14 2,72 43.026 - Porto Alegre 26 212 5.764 22,7 65 722 6.277 78,8 91 934 6.161 10,3 3,5 1,12 1,12 50.004 - Campo Grande 5 527 2.445 96,5 5 19 5.119 7,0 10 546 2.538 54,6 2,1 3,41 0,56 52.010 – Goiânia 5 185 9.169 32,7 16 380 6.114 57,8 21 565 7.114 26,9 2,1 1,39 1,76 10.02 53.001 – Brasília 11 196 8.423 4,5 42 4.151 4 96,2 53 4.347 9.952 82,0 16,5 6,06 1,18 Sub-total 503 6330 7.609 25,6 960 18.431 8.974 77,4 1463 24.761 8.625 16,9 93,9 1,72 1,58 Source: data extracted from RAIS – Mte (2004) compiled by the authors

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 37 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Second, we can observe strong differences between the patterns of software specialization to the different agglomerations selected. Comprising the specialization index, we can point out seven micro-regions where that index is higher than three indicating a strong specialization in software activities: Brasilia, Blumenau, Campo Grande, Aracatuba, Osasco, Belo Horizonte and Joinville. Patterns of specialization can also be articulated to a distinction between package and custom made software. Concerning this point, two groups of strong specialization might be contrasted. The first is composed by eight agglomerations with more than 80% of employment concentrated in the package segment: Salvador; Brasília; Itapecerica da Serra; Montenegro; Serrana; Joinville; Recife; Sao Paulo The other group is composed by six agglomerations with more than 75% of employment concentrated in the custom-made segment: Blumenau; Fortaleza; Sorocaba; Sao Jose dos Campos; Araçatuba; Campo Grande. Those territorial contrasts might be considered to format policy measures well adapted to the requirements of competences building processes in each region. . Third, there are also differences in the average size of firms in the agglomerations selected. Despite the general mean of 16,9 employees per firm, we can contrast the differences between two groups. The first involves four agglomerations with more than 25 employees per firm, located in the micro-regions of Brasília; Campo Grande; Goiânia and Belo Horizonte. The second group is composed by agglomerations with less than 7 employees per firm, being characterized by a huge presence of micro firms. These agglomerations are located in the micro-regions of Maringá; Fortaleza; Vitória; Curitiba; Sao Jose dos Campos and Moji das Cruzes. Those contrasts also stress the fact that the exigencies of firms’ integration and consolidation can vary a lot from one region to another.

Finally, the analysis of the salaries paid by the firms in the agglomerations selected also points out strong differences between them. In fact, only four agglomerations paid more than USD 8.000 per employee in 2004: Sao Paulo; Osasco; Brasília; Montenegro. On the other hand, eleven agglomerations paid salaries that were inferior to USD 5.000 per employee in 2004. These salaries were particularly low in agglomerations located in some micro-regions such as Serrana, Limeira, Araçatuba, Fortaleza and Campo Grande. According to those data, the general argument that the salaries paid in Brazilian software are relatively high – when compared with the level of firms´ competitiveness – might be qualified.

The concern about the effectiveness of policies well adapted to the demands of software sector has resulted in the incorporation of spatial focus in the actions implemented by the central government and specialized agencies. Softex, for instance, had change its focus in the middle of the 90´s in order to support autonomous and geographically dispersed nuclei, fostering ties with local authorities and clients, in order to permit software development and the built of exporting capabilities. These independent nuclei would promote the sponsoring of local software labs, incubators and software training, including training on business issues and responding to local needs. Originally set up to raise software exports SOFTEX moved into sponsoring incubators and regional software nuclei. In order to promote a decentralized growth the program created 20 regional nuclei throughout Brazil (see Figure 3). These nuclei were responsible for coordinating and assisting company activities that associate to this program, trying to take advantage of the potential of each region. Today, it Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 38 http://brics.redesist.ie.ufrj.br

BBRRIICCSS has a network of regional agents established in 22 cities in 12 States, with 37% of the software development firms in Brazil as its associates.

Figure 3 – Localization of Softex Regional Nuclei

Fortaleza

Campina Grande

Recife

Salvador

Brasília

Uberlândia São Carlos Belo Vitória Horizonte Juiz de Fora Campinas Rio de Janeiro Londrina São José dos Campos São Paulo Curitiba

Blumenau Joinville Florianópolis Porto Alegre

The creation of stimulus to the consolidation of technological poles oriented to the development of software activities were widely adopted as a tool to accelerate the process of local competences building. The use of incubators to strengthen these local competences was also a strategy that had been disseminated. According from data collected by Anprotec (Brazilian Association of Technological Incubators) approximately 25% of the 283 incubators identified in Brazilian economy are involved with software and informatics activities in 2004. An emphasis on this kind of support can be seemed not only in Softex strategy but also in a lot of actions formatted by government agencies, such as Finep and CNPq. As a consequence, spatially dispersed nuclei of software poles were created during this period (see Table 21). Some of these nuclei seem to be particularly dynamics in terms of the process of competences building, such as Parqtec (São Carlos), Softex Recife, Blusoft (Blumenau-SC), Softex Campinas and Petrópolis – Tecnópolis. Some of these poles seem particularly dynamic. The Recife Digital Port, for instance, has 86 firms in its infrastructure which generate annual sales of approximately R$ 250 millions, most of them already involved in an export effort. Some of these poles can also be seem as an intermediary step in order to achieve a progressive construction of Local Productive Arrangements (LPAs) based on the strengthening of local ties among producers and local institutions, a topic that will be discussed later.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 39 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 21 – Brazilian Software Poles • 1) Adetec – Associação do Desenvolvimento Tecnológico de Londrina • 2) Blusoft – Blumenau pólo tecnológico de informática • 3) PaqTcPB – Fundação Parque Tecnológico da Paraíba • 4) CITS – Centro Internacional de Tecnologia de Software • 5) Fumsoft – Sociedade mineira de software • 6) Infomar – Incubadora Tecnológica do Paraná • 7) Insoft – Instituto do Software do Ceará • 8) ITS – Instituto de Tecnologia do Software • 9) Softex Campinas • 10) Trisoft – núcleo Softex do Triângulo • 11)Núcleo Softex Salvador • 12)Parasoft – núcleo de tecnologia de software do Pará • 13) Parqtec - Pólo de Alta Tecnologia de São Carlos • 14) Petrópolis – Tecnópolis • 15) Riosoft • 16) Softex Recife • 17) Softpólis – Santa Catarina • 18) Softsul • 19) Softville • 20) Tecsoft - Centro de Tecnologia de Software de Brasília • 21) Tecvitória Source: MDIC (2005)

5. Export orientation and outsourcing

An argument that has been systematically mentioned as a deficiency of Brazilian software industry comprises the limits of its export orientation. If Brazil has an internal market that is at least half a decade ahead of India, in terms of its export sector, it is now where India was probably about fifteen years ago. In fact, current Brazilian export level of US$ 300 million is comparable to that of India in 1991-1992, when it started its export growth take- off. Table 22 shows that, in 2001, five major Indian companies export four times more than the ten major Brazilian companies, which sells only for the internal market. Among the weaknesses identified to justify this trend, is usual to stress the presence of a policy structure that is adverse an export drive, particularly due to the so-called “Brazil Cost” and to the absence of incentives for exportation. The limited experience of companies in the open market, which only began in 1990, is also mentioned. The hypothesis is that the fragmented structure of the market also generates disincentives to cooperation between small and medium firms, preventing a process of the inter-organizational consolidation that is necessary to improve the capacity to compete internationally. The absence of an image that can be associated with Brazilian Software abroad is also mentioned as an obstacle. Finally, it is also argued that Brazilian companies do not have a salary level difference for competing in the low value segment, whereas the small size of companies and their insufficient qualification limit the possibilities of establishing relations with potential customers in high value segments. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 40 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 22 – Comparison between leading Brazilian software firms and leading exporters from India - 2001 Brazil Sales(US$ India Exports MM) US$ MM CPM 175 Tata Consultancy Services 963 Politec 112 Infosys Technologies Ltd 751 CTIS 87 Wilpro Technologies Ltd 591 Consist 77 Satyama computer Services Ltd 424 TBA 55 HCL Technologies Ltd 324 DBA 51 Stefanini 47 Microsiga 36 Datasul 35 G&P 34 Total 709 Total 3.053 Source: NASSCOM (Índia) e Info Exame (Brasil)

The characteristics of Brazilian industrial structure in the software sector affect decisively its export orientation. Approximately 79% of Brazilian exporters are multinational companies that use Brazilian subsidiaries to attend international customers. As a consequence of multinational intra-corporate transactions and of other exports that are hardly accounted by the Payments Balance, we can estimate an amount of approximately 85% of Brazilian exports as taking part of a “grey” market. The amount of tax duties is usually mentioned as the main argument to justify the phenomenon, but we must also consider that multinational could by pass these duties through the intra-corporate channels mobilized to export. In this context, the local firms, responsible for 21% of the amount of exports, tend to privilege the segments of industrial automation and integrated management in their exports. They also tend to use small offices abroad and customers´ connections as the main channels to access international markets. Sales by the Internet have been an effective alternative of distribution only to software products. Information from the Central Bank indicates that, in 2004, the twenty largest software exporters were responsible for an amount of export close to US$ 230 millions. From this amount, approximately US$ 110 millions would come from software products and US$ 120 millions from software services. Data collected from IDC consultancy group (2006) era more modest, point out an amount of US$ 177 millions of software exports in 2005, from which US$ 35,3 millions were exports of software products and US$ 142 millions of software services. Although the controversies about the proper way to evaluate the export performance of Brazilian software industry, all the figures seem to be disproportionably low when compared with the economic dimensions of that industry, especially when they are contrasted with the amount of exports generated by developing countries with a pattern relatively similar.

The Brazilian government would aims to grow software exports five-fold between 2004 and 2007 to reach 5.4billion reais (US$ 1.5 bi) at the end of the period, thereby approaching the level of revenue generated by countries like China. Despite the fact that this target probably will not be attained, the software market was selected as one of four business areas the government has prioritized for growth by the industrial policy, the others being

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 41 http://brics.redesist.ie.ufrj.br

BBRRIICCSS biotechnology, semi-conductors and pharmaceuticals. According from data collected by BRASSCOM (Brazilian Export Association of Software & Services Companies) 300 software firms have demonstrated an interest in become exporters, searching for the different kind of support provided by Brazilian export agencies. Moreover, data from BRASSCON indicate that 71 companies exported software products or services in 2004, generating an amount of approximately US$ 235 million of exports, from which US$ 110 million came directly from software exports and US$ 125 million from services and outsourcing. The total sales of export firms reached approximately US$ 1,5 billion in the same year, with software exports corresponding to 15% of the amount. The twenty largest exporters were responsible for approximately 98% of the total value of software exportations. Figure 3 show that the firms survey by BRASSCOM have an expectation to increase the amount of exports from US$ 230 million in 2004 to US$ 470 million in 2009. There are still only a few companies present in the international market, and those that have already taken this route prefer the North American market, accessing this market through the internal channels of multi-national companies or by their own subsidiaries abroad. Countries such as Germany also buy software systems from Brazil. Germany's Siemens plans to spend US$ 79 million on Brazilian telephone software during 2004-2008.Table 23 indicates that the largest markets for Brazilian exports in 2004 were Mexico (with US$ 78,3 million of exports, corresponding to 33,3% of the total) and USA (with US$ 48,1 million of exports, corresponding to 18,3% of the total. The table also shows that the amount of exports per firm is greater to the firms that exports to Mexico, comparing with other international markets. Probably, part of this flow can be attributed to a regional specialization of multinational companies in Latina America area.

Figure 2– Exports Sales of Brazilian Software Companies (2004) and expected Sales (2005-2009)

Source: BRASSCOM (Brazilian Export Association of Software & Services Companies)

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 42 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 23 – Main markets of Brazilian software exports, number of exporters and exportation per firm

Source: BRASSCOM (Brazilian Export Association of Software & Services Companies)

Although the problems mentioned, we can identify some elements that might act as stimulus to strengthen the export drive of Brazilian software companies. First, the huge increase of global outsourcing in software industry opens new opportunities for countries like Brazil, where we can find a complex industrial base and a lot of qualified working teams specialized in software activities. Second, the competences accumulated by the industry permits to develop applicative and software solutions to broader range of activities, particularly to financial markets, insurance, social welfare and governments activities. Third, there are a lot of new dynamic niches where some small Brazilian software firms began to position themselves, such as the areas of digital games and specific applicative to telecommunications.

Some local initiatives can be particularly important to exploit those opportunities. BRASSCOM, for instance, was created as copy of the Indian NASSCOM, being oriented to consultancy services of quality management and marketing. BRASSCOM was founded in 2004 as a new effort to position Brazil in the growing market of software and IT Global Sourcing, mobilizing some of Brazil’s leading software companies such as CPM, Datasul, DBA, Itautec, Microsiga, Politec, Promon Tecnologia and Stefanini that together project US$ 2 billion in revenues for 2005 . BRASSCOM wishes to play an important role on the creation of Brazil’s Software brand. The first action taken in this direction was the contracting of A.T.Kearney, one of the leading International Consulting Firm in the IT Global Sourcing Market, to develop a study about the new directions and ways to strengthen Brazilian offer in the IT Offshore outsourcing market. Public entities such as FINEP, MDIC, MCT/SEPIN, BNDES and ABDI are taking a strong role in this effort, Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 43 http://brics.redesist.ie.ufrj.br

BBRRIICCSS contributing directly in discussions. Other Public Organs involved are APEX, Banco do Brasil, Ministério do Planejamento and Softex. Another initiative that can be mentioned is the creation of export consortiums such as the Núcleo Brasileiro de Exportação de Tecnologia (NEXT), which includes 15 Brazilian software companies, with more than 11 thousands workers and annual sales revenues of R$ 1,19 billions.

These initiatives are particularly concerned about the new opportunities opened to Brazilian firms by the rapid growth of offshore outsourcing processes in the software sector. Today, with highly qualified personnel available at lower cost in emerging economies, many companies in developed countries rely on foreign outsourcing for software development, resulting in a more cost-effective concentration on their core business. In recent years, the number and size of companies specializing in outsourcing services has significantly increased outside North America and Western Europe. When most businesses think of outsourcing to offshore companies, developing countries such as India, and more recently China and Russia, come first to mind. In 2004, the global outsourcing market of IT services reaches an amount of US$ 607 billions, with the perspective of getting a 6% annual growth until 2008. The offshore outsourcing was responsible for a share of US$ 18 billion of this market, facing a perspective of annual growth of 40% until 2008.

Trying to follow the experience of India, many countries are trying to qualify themselves as effective alternatives for off-shoring practices. The Brazilian competitors in this market may be grouped in two categories: (i) traditional exporters of IT services (such as India, Canada, Ireland and Israel), to which competences and reputation are already consolidated; (ii) emergent exporters countries (such as China, Malaysia, Philippines, Pakistan, Russia, Eastern Europe countries, Mexico and Costa Rica). The last countries are trying to position themselves as low-cost exporters of IT services or as attractive alternatives generated by the availability of some local resources that can be used to improve an export drive. The possibility of being selected as a regional hub of offshore outsourcing might also be exploited by those countries. As a general trend, there are some evidences that competitive advantages strictly based on low salaries differential of software programmers and developers (illustrated by Table 24) are loosing importance compared to the importance of the quality of the services provided and to the reputation of the agents involved.

Table 24 – Annual Salaries Paid for Software Programmers (US$) Country Salary Range USA* 60.000 - 80.000 Canada** 57.000 United Kingdom** 45.000 – 99.000 Japan** 44.000 Ireland* 23.000 - 34.000 Israel* 15.000 - 38.000 Brazil** 20.000 India* 5.880 - 11.000 Ukraine** 5.000 Poland ** 4.800 - 8.000 China** 3.000 - 4.700 Source: Roselino (2006) base on (*) Bardhan & Kroll (2003)and (**) Carmel & Tjia (2005). Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 44 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Though mentioned less frequently, Brazil has become an appealing outsourcing alternative in terms of technological standards, competitive pricing, and enforceability of intellectual property rights. There are a number of indications that Brazil possesses outsourcing conditions superior to those found in other developing countries. In addition to a highly trained work force, Brazil has comparatively low labor costs, a very large internal market, and intellectual property protection legislation of long date substantially similar to that of Western Europe and the United States. Though Brazil still has unresolved piracy issues, these tend to be related more to consumer products. Brazilian enforcement of existing intellectual property protections is increasing and recognized as essential to Brazil's economic development. Internal economic pressures to expand exports, the increased use of Brazil as the South American base for foreign companies with operations in the sub- continent and the strategic need of Brazil to come to an international agreement with respect to the enforcement of intellectual property rights to participate in the Free Trade Area of the Americas, all evidence a substantial evolution towards a highly competitive, technologically competent and secure business environment.

Outsourcing is already a trend in Brazil. For example, the annual business volume in the services sector for the wireline and wireless service providers is around R$ 3.8 billion – approximately US$ 1.8 billion. Information and communication technology firms present the largest amount of alliances between industrialized countries and usually they are large multinational firms in the areas of telecommunications, computers, electronic components and consumer electronics. Brazilian wireline service providers outsource their networks to companies like Siemens and Alcatel – German and French multinationals – traditional manufacturers of the gear used by these service providers. Thus, more and more these service providers are focusing in market-related activities and less in technology itself. On partnerships involving firms from developing countries, studies show that a great deal of them regards the creation and exchange of technological knowledge, with telecommunications and computers sectors outnumbering alliances in any other field.

Outsourcing of software services requires also the subcontracting of specific IT activities, opening new opportunities to qualified agents. In this sense, geographic specialization permits that outsource services might be distributed according to resources costs and local technological competences. Currently, México is the greater exporter of IT services in Latin American with an amount of US$ 500 million of exports per year, with Argentina, Brazil, Chile, Costa Rica being viewed as exporters potentially competitive in those outsourcing services. According to the evaluation periodically developed by A.T.Kearney, one of the leading International Consulting Firm in the IT Global Sourcing Market, Brazil occupied the 10th position in the “Offshore Location Attractiveness Index” of 2005, falling down from the 7th place occupied in 2004 (see Table 25). In the three years of publication of this index, Brazil has remained itself as one of the 10 more attractive destinations for offshoring. While India and China are recurrently mentioned because of its low cost and qualification of human resources, Brazil is inserted in the group of countries that compete to be established as an effective alternative to offshore activities. In the 2005 index, Brazil faced a light fall for the tenth position, because other countries - Thailand, Chile and Canada - had improved its results, supported by pro-active actions of its governments and local associations which had improved its attractiveness. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 45 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 25 - A.T.Kearney Offshore Location Index 2005 Environment Financial al People Attractivenes Attractivenes Attractivenes General Country s s s Index India 3,5 2,1 1,3 6,9 China 3,2 1,8 1,2 6,2 Malaysia 3,0 1,1 2,0 6,1 Philippines 3,6 1,2 1,1 5,9 Thailand 3,3 0,9 1,5 5,7 Singapore 1,6 1,4 2,7 5,7 Egypt 3,6 1,0 1,0 5,6 Chile 2,7 1,0 1,9 5,6 Czech Republic 2,6 1,1 1,9 5,6 Brazil 2,9 1,4 1,2 5,5 Indonesia 3,5 1,1 0,9 5,5 Canada 1,1 2,0 2,4 5,5 US (Tier II) 0,5 2,7 2,2 5,4 Slovakia 2,7 1,0 1,6 5,3 Bulgaria 3,3 0,9 1,1 5,3 Mexico 2,9 1,2 1,2 5,3 Jordan 3,0 0,9 1,4 5,3 UAE 2,7 0,6 1,9 5,2 Poland 2,7 1,1 1,4 5,2 Ghana 3,6 0,6 0,9 5,1 Vietnam 3,6 0,7 0,8 5,1 Romania 3,1 0,9 1,1 5,1 Costa Rica 3,0 0,8 1,3 5,1 Hungary 2,6 0,9 1,6 5,1 Source: A.T.Kearney (2005)

In this sense, we can also consider the evaluation of CIO Whiteboard, which intends to provide a thorough country by country analysis of the current outsourcing market, and to point out where to find the best opportunities for a variety of services. Through this analysis, the Global Outsourcing Index 2005 was constructed, rating each country on the basis of a combination of relative cost, risk, and market opportunity. Because of the world wide fast economic changes, the whiteboard also looks at where offshoring opportunities can be found in the future, based on an evaluation of experts. The Future Outsourcing Rank compares each country by its competitive position in the overall outsourcing market as it will look in ten years, based on such factors as population growth, GDP growth, labor supply and IT expertise. According to this evaluation, Brazil’s huge labor pool with good IT skills, decent infrastructure and low costs make it a relevant player in the Americas, but it must focus on controlling the economy and improving IP protection. In the current ranking of best opportunities for outsourcing, in which India takes the lead followed by China, Brazil occupies the 15th position. On the other hand, if we consider the prospective Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 46 http://brics.redesist.ie.ufrj.br

BBRRIICCSS ranking, Brazil could occupy the 4th positioned (in a rank leaded by China, India and US), particularly because of the IT expertise, which would permit to become a major outsourcing center for U.S. industry.

These indexes suggest that the attractiveness of Brazil as a player in the offshore market has been affected by structural conditions, such as the general level of education, social costs that affects the contraction of workers, the relative inflexibility of the labour market and the appreciation of the local currency. However, Brazilian supply of IT services also presents some relative advantages when compared with its main competitors such as India, China, Canada and, more recently, Russia. Among these advantages, we can mention the size and sophistication of the internal demand, the qualification already demonstrated to develop world-class solutions for specific segments (such as financial services, retail, government and telecommunications), the business culture similar to the main market targets (North America and Europe), the good infrastructure of telecommunications (compared with countries like India or China) and the proximity with the main market targets, also when compared with India or China. Due to the intensification of the concurrence, the main challenges that might to be faced by Brazilian companies refer, basically, to the construction of an image of technological reputation, as well as to the improvement of the quality of services provided. The improvement of local human resources seem also to be very important, as well as investments in innovation to domain new technological platforms. Finally, an adaptation of the local the tax burden with international patterns should also be pursued.

Concerning the improvement of export drive it is important to consider the specificities of different segments of offshore outsourcing markets. For low aggregate-value services such as coding (of computer programs) it is only necessary that IT infrastructure works well – the determinant factor is the cost per tested line of code, and communication is simple and determinative. The client-supplier relationship does not need to be long-term, and contracts may be short-term. Going up the value scale, communication becomes more complex, more interactive: the dialogue between client and supplier needs then a common language, and English is the sector’s universal language. Knowledge of other languages may become a strong differential in certain areas (Chinese, Japanese, Spanish), but English is always indispensable. On higher aggregate-value deals, with contracts of longer duration, it is fundamental the perspective of continuity, because the client must assure that the fulfilled requirements will continue to exist for a long time after contracts are signed.

On service development, it is difficult to compete in the lowest market segments, given that these are based above all on the use of very cheap labor where Brazil is unable to compete with China and India, or in historical specificities, like the English language and the Diaspora of professionals. Competition in this segment should become even fiercer with the entry of second line competitors (e.g. the Philippines and Russia). However, the increase in sales, and especially of exports via outsourcing, is of vital importance to the Industry’s sustainable development, because it increases its credibility and establishes confidence with the market agents. Moreover, it may also have an important impact on the trade balance, generating a considerable growth in the number of jobs and, indirectly, increasing the industry’s talent pool.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 47 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Despite the problems to gain competitiveness in the low added segment, it is possible to explore the competences accumulated in higher added value software services in vertical areas where current Brazilian customers are of world class. Telecommunications, e- government, data and network security, financial or retail services constitute niches in which the qualification and maturity of Brazilian software companies may be able to standout. Another field that seems to be profitable concerns the complex integration outsourcing, trough the development of large scale customized systems for governments and large companies with many users spread out over a large geographic extension. In these big projects, Brazilian integration firms making use of software factories in regions with lower labor costs than those of big domestic demand centers, and their expertise in defining requisites and breaking up development into parts will show a unique competitive advantage, which would be difficult for potential competitors to replicate. But to explore these opportunities, the requirements related to cost and quality must be attended, what can be done either through reputation or investment in process certification schemes, such as CMM and CMNI.

6 - Brazilian Software Specialization and new opportunities: the major trends

Brazilian software sector has largely developed for the domestic market and remains quite heterogeneous in terms of skill content and firm organization. Thus, most firms are small and involved in customized software development or packages, although there appears to be comparatively little specialization. However, there are a number of major software ventures which have developed highly complex software products. There has been little attempt to market these products beyond Brazil’s borders and the industry as a whole remains almost exclusively devoted to the domestic market. Part of this can be attributed to language and other barriers, but also to apparent deficiencies in business strategies, including a weak client focus.

Despite the limitations stressed, there are also a set of market forces that might contribute to maintain the dynamism of Brazilian software industry. In fact, the dynamism of Brazilian domestic market as a driving force to the software industry seems still quite incontestable. So, we should broad the focus to evaluate the possibilities of Brazilian software industry sustain a trajectory based on a sustainable market growth. Due to the extension of internal market, Brazil can explore a lot of comparative advantages is areas such as banking and commercial automation, telecommunications, enterprise management and automation of public services and free software. Some of these segments might be discussed in more details.

A world-class financial sector is individually the largest and most sophisticated Brazilian software market. Currently, the Brazilian financial sector is one of the most competitive and efficient in the world, according to research conducted by the consulting company Accenture in conjunction with the country’s main financial institutions. Financial institutions have recorded IT investment indices much higher than the economy in general. Accenture study also showed that 15.8% of investments are oriented to IT technologies, with an emphasis towards Internet Banking, where Brazil is one of the world leaders. A survey carried out by the Getulio Vargas Foundation (FGV) also indicates that 8.5% of financial institutions’ assets are invested annually in IT – double the average of other Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 48 http://brics.redesist.ie.ufrj.br

BBRRIICCSS national industries. Currently, there are 180 financial institutions in Brazil, which operate highly sophisticated automated systems. Among the solutions that stand out in the banking system are: intelligent card (smart card); electronic cash, modality in which the money is stored in the hard disk of the computer; electronic check, to accomplish commercial transactions over the internet with digital signature; and, finally, virtual wallet, a system used by major companies to store their customers' data securely, so that for each new purchase only a password identification, for instance, is required.

From 1999 to 2001, spending on software acquired from third parties by banks grew by 70% and internally developed software by 46%. The emphasis in banking automation has moved from the hardware side to the software side, although developments in the two are still complementary. The bank has to be able to talk much the same language as its software or hardware suppliers if the systems that it acquires are to be useable in the ways it intends. An example is the built and modernization of Brazilian Payment System (SPB – Sistema Brasileiro de Pagamentos) which is extremely demanding of software services, due to exigencies of readiness, security and reliability: In 2003, approximately 80% of 20 billion bank transactions are made automatically (through ATM, phone, fixed and mobile, web). Another example can be found in the built of SERASA, the biggest credit information company in Latin-America, whose shareholders are 76 financial institutions. The customers of the system came from Brazilian financial institutions and from over 300.000 retail, manufacturing and service companies, generating around 3 million inquiries to Serasa’s database per day. The huge growth of Brazilian financial system also generates an impressive demand for software related to the building of Credit Risk Models. Finally, the conjunction of the growth of financial system with the expansion of internet basis results in a huge growth of internet banking operations, converting the Brazilian market in the 3rd world’s biggest internet banking operation basis. In this sense, a surprising fact is that four Brazilian banks are among the world's ten most accessed sites from homes. Eighteen percent of banking clients in Brazil handle transactions online.

Recently, financial institutions invested primarily in telecommunications infrastructure and data security. This was necessary in order to bring them up to date with SPB’s technology guidelines. The financial sector is comprised of a strong niche market for IP voice technology (VoIP), useful in carrying out companies’ financial transactions in the market. Also there is room for growth in IT supply and in the area of information integration for banks’ myriad activities, as well as in distance customer service. Beyond home banking/Internet, data warehouse and security stand out as solutions demanded by the financial institutions. Market source also indicates that about 80% of the Brazilian banks are expected to use data warehousing in the near future, integrating data banks with consumer profiles. However, the biggest concern for the financial institutions is to be able to offer secure transactions through the internet. Thus, access security software and hardware and biometrics identification become excellent business opportunities The losses caused by Internet fraud against Brazilian banks and credit card administrators are estimated in US $132 million in 2005. According to Febraban, banks in Brazil spent ten percent of their budgets in IT to protect their systems from hackers. During this same period, they expended US$ 1 billion for guards and electronic security. IT investments in financial institutions are not limited to electronic transactions: CRM, billing, ERP, B2B and B2C are also important solutions in the financial sector investment list, following the trend Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 49 http://brics.redesist.ie.ufrj.br

BBRRIICCSS observed in other sectors. There are also some evidences that, after modernizing their infrastructure, Brazilian banks will begin a second phase of investments, focusing mainly on safety for more complex transactions through the internet, including integration of wholesale networks and retail operations for electronic commerce. Thus, safety is expected to be the Brazilian banks’ top investment focus. In this context, access security software and hardware, as well as biometrics identification, are viewed as excellent business opportunities

The Federal Government is also a large and sophisticated user of software. Brazil’s lead in e-government is broadly recognized, with flag projects such as electronic voting and 98% of all personal income tax delivered electronically. The share of internet users that access public services is greater than in several developed countries (see Figure 3). Concerning the e-Government base, there are approximately 1.700 services available on the Internet, with 23.000 links installed, 35 million pages accessed a month and already 6.000 auctions made, corresponding to amount of R$ 1.07 billion per year. Another success case can be found in the general elections, carried out in electronic ballot boxes specially designed, which are based on hardware and software developed by Brazilian firms. As a consequence of the competences accumulated in this area, the period of time to count the ballots of approximately 115 million voters had been reduced to 10 hours and the solution has been exported to Mexico, Argentina and Uruguay. We can also mention the receiving of income tax filing by the web. The number of on-line declarants reached 15 million in 2003, corresponding to 90% of the income tax base (95% of the natural persons and 100% of the corporations). Payment of the tax on line, transferring the value regarding the payable tax from the tax payers banking account to the Government banking account has also experienced a huge growth in the last years.

Figure 3 - Share of internet users that access public services

Source: IBOPE/NetRatings apud Correio Popular, 29/02/2004, pp. B-2

Beyond the experiences of financial and government segments, there are others opportunities that can be exploited by Brazilian software companies. Concerning software products, there are some evidences that competition among software packages is very difficult. In this sense, the segments where Brazilian companies may aim for a relevant position include embedded software, software components and customizable software. Among these, the more relevant opportunities seem to arise for specific components and solutions for more complex products aimed at vertical segments, where there are already Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 50 http://brics.redesist.ie.ufrj.br

BBRRIICCSS several companies with successful products in Brazil. There are also opportunities for companies with experience in customization and services that are qualified to make products out of modular solutions. The e-business is another area in which Brazil is leader in Latin America and which may leverage the international competitiveness of Brazilian companies. With regard to international competition in services, it is important to consider software with greater added value, potentially in vertical areas where the current Brazilian customers are of world class or complex integration outsourcing, with the development of large scale customized systems for governments and large companies with a high number of users dispersed over a vast geographical area. Specific components and solutions should also be considered for the more complex products, as well as the exploitation of horizontal software technologies/ components with few customers in global markets.

Concerning the vertical market for Brazilian software companies, there are also a tendency to increase the impact of the products and services generated to traditional industries that have been deeply impacted by IT, such as the construction industry and the agribusiness market. In these sectors, the huge extension of informatics technology use has made possible the processing of a large amount of data and information, the constitution and management of integrated databases and the implementation of communication networks that constitute what is called “integrated systems of business management” also in the construction industry. Thus, software becomes a crystallized element of structural changes of this sector, often being tailored according to the needs of the segment. The insertion of software in a traditional sectors shows that the new ITs are becoming widespread, not only serving as administrative support, but as strategic tools to increase firms´ competitiveness. The example of the use of IT by Brazilian framers – particularly in high productivity areas – seem quite interesting, opening new opportunities to the development of software products and services well adapted to their specific needs.

We can also mention a set of new dynamic segments where the possibilities of rapidly accumulating technological competences are articulated with attractive market opportunities. Because of the broadly impact of IT, the development of digital games to PC, mobile phones and Internet is a software segment that had experienced a huge growth during the last years. This segment is very young not only in terms of firms that have recently entered into the market but also due to the age of technicians and entrepreneurs that are responsible by the development of those games. There are evidences that around 80% of these enterprises have less than 20 employees, being geographically dispersed in sates of the South (PR, SC e RS), Southeast (SP e RJ) and Northeast (PE). Despite their small size, approximately 50% of the firms have already a presence in international markets. Some research centers in the area have a pattern of excellence, being articulated with universities such as PUC-RJ and UFPE. The local market for digital games is estimated in somewhere between R$ 70 million and R$ 300 million. The firms from the segment have already developed 35 games to PC in the last two years. Currently, we have also 25 games being developed by the sector. The international market to the segment is estimate in around US$ 22 billion. The segments that seem particularly attractive to Brazilian firms comprehends serious games oriented to professional applications (such as for training), simulations related to environment, healthy and education issues. The development of digital games through outsourcing practices demanded by international clients seems also to be a window of opportunity. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 51 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Another segment that seems to be attractive is related to the development of visual applicative to petroleum industries in areas such as geosciences, engineering, environment management and operational security. Petrobras, the Brazilian state petroleum company, is the main client of those technologies, particularly through 10 centers of virtual reality installed in its units. Despite the fact that the majority of equipment came from abroad, the adaptation of software to the needs of Petrobras open opportunities for the increase of local learning, inducing a process of technological capacitating for local firms. The technological competences accumulated by local firms can be mobilized to diversify the applicative generated to another fields - such as digital games, broadcasting and defense - enlarging the scale of their markets. In the broadcasting area there are also opportunities for software firms that have been opened by the diffusion of digital signals associated with the advent of high definition television market. The transition from analogical to digital TV seems to be quite intensive in the use of software applicative because of some processes that have to be carried out. These processes involve not only the necessary compression of data to the digital pattern but also the development of digital subject, portability, interactive applicative, electronic commerce. The competences that might be constructed during this process can also allow a diversification and internalization of software firms.

The dynamics of innovative efforts in software industry can also be articulated with the growth of free software / open source software (FS/OS) that can be produced by Brazilian firms. In this sense, the rise of a self-organized, complex and virtual network of developers and users with varying motivations and of new forms of licensing software indicates that there are new variables in the sector. Free software has become a strategic option for technological development that creates opportunities for social inclusion as discovered by a number of successful experiments in Brazil. In this segment, Brazilian developers tend to become highly professionalized, with a predominance of qualified professionals: systems managers, network technicians, entrepreneurs, researchers, and college students. Among development companies, however, small companies prevail, but larger companies have already adopted this model as well. The market for FS/OS-based operating systems in Brazil is estimated to have a minimum volume of R$ 77 million, considering only the sale of Linux distributions and related services, with potential to grow 2.5 to 3 times by 2008. Brazil spends about US$ 34m per year for licenses to administer government ICT infrastructure. Due to this high cost, the government, like in many other developing countries, has been encouraging its ministries to use Linux and other free software systems (Alerigi, 2003).

FS/OS is becoming professional in Brazil, and is beginning to move from the industry’s periphery towards its center. Three types of firms working in FS/OS might be detected: (a) small and medium-size companies founded in the 1980´s and 1990´s, which are dedicated particularly to proprietary software, but which entered FS/OS, some because of market requirements; (b) small and medium-size companies founded more recently, which have a large part of their activities in FS/OS; and (c) large companies, some of them multinationals, which have also entered the FS/OS market to provide pre and post-sale support for Linux and to make their proprietary products available to run on this platform. The amplitude of potential impacts of FS/OS might be articulate not only to challenges imposed to the appropriability conditions of the technological regime of software industry Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 52 http://brics.redesist.ie.ufrj.br

BBRRIICCSS but also to political, institutional, and cultural elements that affect the dynamic of innovation in the sector. Among these elements, particularly importance can be attributed to cultural elements related to how the people and firms position themselves in face of the general principles of open source software.

7 - Sectoral system of innovation in Brazilian software industry: general characteristics

This section uses a sectoral system of innovation (SSI) approach to analyze the recent development of Brazilian software industry. Initially, we might consider that software has a huge impact in the productive base of all the economy, becoming a permanent source of innovation and differentiation of products. This characteristic makes the delimitation of the sectoral system of innovation a complicated task. In fact, the introduction of technological and organizational innovations strongly supported by software products and services is the base of profits increasing of productivity in virtually all the sectors of economic activities.

The size, structure and sophistication of the demand for software in the Brazilian market comprise vectors that accelerate learning for Brazilian companies, increasing competence building processes. In this sense, the Brazilian government policy has been explicitly aimed at raising the networked level of the economy and encouraging wider IT diffusion. IT adoption has became widely diffused, as exemplified by both the levels of ecommerce and the integration in industrial production, as well as the level of e-government, that have been attained. Certainly, this has been associated with increased local demand for software. However, income barriers remain significant at the level of households while organizational impediments, combined with deficiencies in the business environment, principally piracy, have come together to limit, in particular, the market for software products.

In this context, there are some trends that justify the optimism about the future of Brazilian software industry. First, we can note that the dynamism of software sector is intrinsically articulated with the dynamism of the IT as a whole. Brazil has the largest IT sector in Latin America as a share of the whole economy, and it has experienced a significant growth during the 1990s and 2000s. According to IDC data (2006), The Brazilian IT market corresponds to an amount of Us$ 11,9 billion in 2005, representing 1,1% of world market and 39% of Latin America market. In this market, we can fin 5,5 million PC's sold each year, generating a 17 million PC'sinstaled base, from which there are 46 million users of internet. The telecom sector has thrived as an international powerhouse, both through local initiative – often coming from CPqD, the former government research labs– as well as the presence of important multinationals. Like in many other countries, IT application and diffusion first started mainly in the business world. Particularly, it started with the financial sector and then it spread to other sectors. Between 1998 and 2002, the use of IT applications in different sectors of the Brazilian industry has grown significantly. Table 26 demonstrates that IT market had an explosive growth from 1999 to 2005, not only in terms of Internet users and transactions but also in terms of other IT products (such as mobile phones, PCs and broadband access) that are extremely demanding of software services.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 53 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 26 – Indicators of IT Diffusion in Brazilian Economy 1999 2000 2001 2002 2003 2004 2005 Internet users (mi) 4,8 9 11 14,3 17,4 20,9 25,1 Active domestic Internet users (mi) 2,5 4,7 5,9 7,91 0,9 13,4 16,1 Phone (mi) 27,8 38,3 38,6 38,9 40 42 44,1 Mobile phone (mi) 15 23,2 28,7 34,9 45,5 53,2 60,7 PCs (mi)I 7,8 9,5 11,3 12,7 14,2 16,3 18,7 Broadband access (mi) 0 0,2 0,3 0,7 0,9 1,3 1,7 E-Commerce (US$bi) 0,3 0,6 2,1 5,1 8 10,5 13,5 B2B Marketplaces (US$bi) 0,2 0,3 1,6 3,7 6,3 8,5 11,1 B2B Companies(US$bi) 32,8 44,3 57,6 74,8 B2B total (US$bi) 36,5 50,6 66,1 85,8 B2G (US$bi) 1,2 2,6 5,3 8,4 B2C (US$bi) 0,1 0,3 0,5 1,4 1,7 2 2,4 M-Commerce (US$bi) 0,01 0,03 0,05 Expenses with advertising online (US$mi) 32,3 52,6 47,5 60,3 63,4 67,1 % expenses / total advertising 0,65% 1,15% 1,42% 1,40% 1,56% 1,79% IT expenses (US$ bi) 16,3 17,2 18 14,9 15,8 16 16,1 IT expenses (% invoicing of the companies) 3,90% 4,20% 4,50% 4,70% 4,90% 5,10% 5,20% Source: Data from E-consultingcorp

The dynamism of IT market in Brazil is extremely connected to the expansion and sophistication of Internet basis. Concerning the Internet, we can observed that Brazil is the 11th country in connected hosts, presenting around 9 million individual users and a 36 million inertial audience and a US$ 10 billion inertial market for e-business. The e-business area is one in which Brazil is leader in Latin America (40% of the total and 60% of the B2B) and which may leverage the international competitiveness of Brazilian companies (see Table 27). The turnover of e-commerce in Brazil reached R$ 1 billion in 2002 and the forecast annual growth rates are placed in the order of two to three digits. Evidences also suggest that São Paulo has now emerged as a major hub for international internet traffic exchange in the Latin America region, trailing only Miami.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 54 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 27 -Use of E-Commerce Technologies in Brazil -- 2002 (in %) Type of ICT Application Establishmen Sector b Total t Size a SME Larg Mfg. Distri. Finan Brazil Global e ce c d E-mail 100.0 99.9 100.0 100.0 99.9 100.0 98.5 Website 70.4 80.8 77.7 67.1 84.5 70.7 74.1 Intranet 36.8 71.7 45.2 34.0 50.1 37.7 63.6 Extranet: 32.9 44.6 41.4 29.2 46.4 33.2 32.7 Accessible by Suppliers/ Business 10.1 33.7 22.3 4.8 35.6 10.7 20.9 Partners e 15.5 28.6 21.8 12.5 35.1 15.9 17.8 Accessible by Customers e EDI: 35.7 71.9 38.9 35.8 35.1 36.7 44.3 Over Private Networks Only e 7.0 25.8 4.2 9.2 2.7 7.5 19.4 Internet Based Only e 6.9 10.4 11.6 4.6 15.6 7.0 8.4 Both e 21.8 35.7 23.1 22.0 16.8 22.2 15.9 EFT 52.0 66.5 45.5 54.8 51.8 52.4 43.4 Call Centre 45.6 62.5 46.8 46.3 35.1 46.1 32.3 Source: GRITO, Global E-Commerce Survey, 2002 (see: "E-Commerce Readiness and Diffusion: The Case of Brazil," Federal University of Rio de Janeiro. http://www.crito.uci.edu/publications/pdf/gec/brazil.pdf).Notes: a SME (small and medium sized enterprises) are those with 25-250 employees, large are those with more than 250 employees; b Manufacturing include all establishments, distribution includes wholesale and retail, and finance includes banking and insurance; c Responses were weighted based on total number of establishments by employee size within the sector. Survey sample size by sector are 68 establishments in manufacturing, 68 in wholesale and retail distribution, and 64 in banking & insurance; by size are 98 establishments classified as SME and 102 as large; d Consists of weighted survey responses in 10 countries: US, Mexico, Brazil, Germany, France, Denmark, Singapore, China and Japan. ‘Global’ sample size by sector are 743 in manufacturing, 701 in wholesale/retail distribution, and 695 in banking & insurance; by size are 1,088 establishments classified as SME and 1,053 as large; and e Percentage based on total sample. Mfg. – Manufacturing; and Distri. - Distribution

However, in relative terms, the general situation is not so good since only 8 per cent of the population had access to the internet, with Brazil being in the 46th position in the world table. The real problem of the Brazilian economy and society was that diffusion of ICT was affected and hampered mostly by the very uneven income distribution of the country. In this sense, Brazil has also put a high priority on improving access to advanced info- communication technologies, promoting digital literacy and improved access to government public services. This is considered pivotal to improving social and economic development for society at large. The term ‘digital divide’ refers to a common problem among many developing and emerging nations where large segments of the population do not have access to IT such as Internet, telephones, and computers. Brazil is no exception to this problem, with large sections of the country not having access to IT.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 55 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

A series of government programs have been envisaged to cope with this situation. In 1999, the government launched the Program for the Information Society which put forward an ambitious set of actions to be implemented in partnership with private capital aiming at creating socio-economic, cultural, political and technological conditions to a wider diffusion of ICT in Brazil. The program aimed to expand Internet infrastructure, facilitate the interconnection of all public libraries, and create thousands of community access centers throughout the country. The idea was that the availability of ICT, in conjunction with relevant ICT training programs and the availability of new low-cost computers will inevitably promote more ICT users in the country. It also aimed to promote computer literacy with free learning material, courses and community centers (to be developed from an initial 1000 to 5500). The program was a major attempt to spread ICT in the society as a whole. Although there were some very interesting ideas in the project (such as giving fiscal incentives for the production of microcomputers costing up to US$ 1,000, and the development of popular micros with prices around US$ 200) this program never really took off and was eventually abandoned by the new government.

In 2003, the Federal Government introduced a new project called ‘PC Connected’ that aims at universalizing the use of micros via selling low cost micros (Popular PCs) to low income people. The ‘Popular PC’ is a basic personal computer that comes with all of the necessary hardware minus storage capacity (hard drives and floppy drive). The idea of ‘PC Connected’ is to foster the production of a micro with free operational systems and basic software and it is predicted that final price should be around US$ 560. The production is to be subsidized by fiscal incentives. People could buy their own computers using a payment plan of about $15/month for 24 months. The ministry of communications is to negotiate with telephone operators agreements to make less expensive the Internet Access (up to US$ 3.00 per month for 15 hours of access). This is a program that is still on the “pipeline” and, although it presents the obvious merit of targeting the poorest sections of the Brazilian society, there is skepticism about its success. First, as in the past, there is no guarantee that sufficient funding will be available. Second, planned final price of the PC is still high, given the income patterns in Brazil – 30 per cent of the population receiving a monthly income below US$ 60

At the supply side, there are also some factors that condition the dynamism of sectoral system of innovation in the Brazilian software industry. Specifically, the structure of this industry has some characteristics that affect the possibility of developing innovative capabilities internally. Despite the fact the country represents the 6th largest market of the industry; it is also true that this market is controlled mainly by multinational companies, which tend just to sell solutions, investing and amount relatively low of resources in internal R&D. In fact, R&D location decisions of multinational companies are likely to be fraught with important bandwagon effect, i.e., if several firms opt to go for a country or region, others will follow. This process tends to be particularly strong to areas that are recognized as attractive opportunities for these companies. Despite the concurrence with countries that have been beneficed by the process of software development migration – such as India for a long time and China and Russia more recently - Brazil still constitutes an attractive market, not only because of its large dimension but particularly because of the

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 56 http://brics.redesist.ie.ufrj.br

BBRRIICCSS competences that has already constructed in areas such as telecommunications, banking and e-commerce, where it already possesses a strong base.

Brazilian national software companies perform an important role to the deepening of the competence building processes in the industry. Despite the fact that they control only 20% of the sector, we can found an important presence of Brazilian companies - most micro and small firms - in almost all areas of the software market. Brazilian companies comprise a dynamic industry, capable of innovation, but very much fragmented, which tend to restrict the possibilities of investing in formal R&D activities. An important shortcoming was the inadequate and insufficient technological learning by local firms. The strategy of evolving from technology licensing to in-house development was anchored in learning processes to be sustained by continuous R&D expenditures. These, however, have been limited by the size of the market and, most importantly, by the size of local firms. The adoption of more aggressive technological strategies has been restricted by the absence of conglomerates in the Brazilian economy which could provide the financial means for a long run strategy. With a few exceptions Brazilian software firms were relatively small and independent.

The study coordinated by MCT-Sepin in 2001 with a sample of 446 firms tried to evaluate the innovative performance of Brazilian software firms through data from patents. Data collected by MCT-Sepin indicated that 16,5% of firms used to register copyright rights in a systematic way and 6,5% occasionally. In this sense, Brazil’s legal system is moving in step with international intellectual property protection standards. As a general matter, a foreign contracting company can intervene, protect, and enforce its intellectual property rights related to software developed in Brazil with reasonable commercial certainty. In addition, foreign arbitration awards are now enforceable in Brazil, and have become a plausible alternative to enforce intellectual property rights. Brazil is also a signatory of the Agreement on Trade-Related Aspects of Intellectual Property Rights (“TRIPS”). In May of 1996, approximately one year after signing TRIPS, the new Industrial Property Law No. 9279/96 (the “Patent Law”) came into force. This law brought Brazil’s patent and trademark regime up to par with the international standards specified in TRIPS. Intellectual Property protection includes Copyright Law No. 9610/98 (the “Copyright Law”), Software Law No. 9609/98 (the “Software Law”), and several other secondary laws. Brazil was the first country in Latin America to enact legislation specifically designed to protect software by copyright. Concerning the question of property rights of software, we can note that the piracy rate of 64% is still high, but not as high as China or India, and Brazil is making efforts to reduce it (see Figure 4).

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 57 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Figure 7 - Software piracy rates 2005

100 87 90 86 83 77 80 72 71 66 70 65 64 60 46 50 43 36 37 40 32 30 20 10 0 China India Indonesia Phillipines South Taiwan Russia Argentina M exico Brazil Costa Israel South Ireland Korea Rica Africa

Source: BSA – Business Software alliance , December 2005

The study coordinated by MCT-Sepin in 2001 also shows that software firms have invested around 14% of their revenues, being 7,5% of them in R&D. As regard the technical qualification, data collected demonstrate that only 18% of software firms got quality certification standards such as CMM or SPICE. In this context, quality and innovation are intrinsically articulated, if quality is understood as the capacity to meet the needs of the client. In order to develop quality software, not only technical and standard norms are needed, but it is necessary to have the capacity to incorporate innovative technological characteristics that not only satisfy user needs, but being capable of foreseeing them. A harmonious match between quality and innovation may guarantee that a product remain in the competitive market for a profitable period. With regard to process qualification, it was observed that only a small number of Brazilian software firms have high maturity certification in the software development process (CMM level 3 or higher)6. In fact, software process improvement concern has emerged over the past twelve years, but local firms have favored the ISO 9000. Around 2003, India had 32 firms that have reached level 5, CMM highest level, China only had one thus far and Brazil had none. Among the companies with (CMM/ISSO) certification, the majority were associated with products, while the service companies employed their own methods, but without certification, an important gap that limits their expansion to international markets. To illustrate this process, we can also consider data collected from a survey carried out by Mayer & Bunge Informatics Consulting in 2004 with 461 firms Brazilian software firms. Table 28 show that only 5% of the firms surveyed had got CMM quality certification in 2004 and only 0,8% a level equivalent to CMM level 4 or higher. Since 2004, these figures have evolved in a positive way. The Integrated System Diagnostics– ISD-Brasil, a subsidiary of the American ISD Inc is authorized to make official evaluations about CMN conditions. According to the last evaluation – relative to August 2006 – there are 49 firms that are in process of getting CMN certification and 21 firms getting CMNI certification in Brazil (see Tables 29 and 30). Among them, there are one firm that have already gotten a level 5 CMNI (EDS, an

6 The CMM - Capability Maturity Model is a model developed by the Software Engineering Institute – SEI of Carnegie Mellon University to evaluate the maturity of software process adopted by a company. This model defines five level of maturity: initial; repetitive; defined; management; optimized. The CMMI - Capability Maturity Model Integration was created as an evolution of CMN, which were merged with two other models to evaluate software performance – the SECM - System Engineering Capability Model and the IPD-CMM - Integrated Product Development CMM. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 58 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

US-based MNC) and 6 firms that have already gotten a level 5 CMNI certification (BRQ Soluções em Informática; IBM, Stefanini and Tata Consultancy Services do Brasil). Brazil also became the 14th country in terms of CMN certification and the 11th in terms of CMNI certification in September 2005, according to the periodic evaluation of ISD Inc.

Table 28 – Certification procedures of Brazilian software firms Certification pattern % of firms No of firms Microsoft certification 10,8 50 Oracle certification 5,2 24 IBM certification 2,5 12 SUN certification 1,0 5 Novell certification 1,0 5 ISSO 9000 7,2 33 PMI 1,4 6 CMN level 2 3,0 14 CMN level 3 1,2 6 CMN level 4 or 5 0,8 4 Total of certificication 34,1 157 Total 100,0 461 Note: Information about software certification was extracted from a basis of 461 enterprises syrveyed in 2004 Source: Panorama da Industria latino-americana de software ; MBI Mayer & Bunge Informatica Ltda (2005)

Table 29 – Brazilian firms with CMN Certification – situation in August 2006

Source: MCT/SEPIN/DIA

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 59 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 30 – Brazilian firms with CMNI Certification – situation in August 2006

Source: MCT/SEPIN/DIA

The patterns of segments´ specialization of Brazilian software companies affect directly the intensity and distribution of innovative efforts. Concerning this point, there are evidences that companies that are active in more structured market segments, in the areas of telecommunications, integrated management software and industrial automation, as well as in banking and financial software, form part of a group that is outstanding from the point of view of investments in technological qualification and Research & Development (R&D). Software for e-business and document and content management occupies an intermediate position, appearing after a group in which systems integration and the governmental area are included, which shows that is it associated with activities having fewer R&D requirements.

Re-investment of their own capital constituted the main source for financing the growth of companies, but recently almost all the companies obtained external financing, through private risk capital and governmental programs. The amount of investment necessary to finance innovative efforts is relatively less in comparison to other products that demand industrial processes for its production. Despite that, the financial mechanisms, especially for small and mid-sized innovative companies, seem to be inadequate to give the proper infrastructure to finance R&D efforts in the software sector. In this sense, we can observe that one of the main financial instruments for software company growth is risk capital, with the consolidation of venture capitals markets by the initiative of institutions like FINEP (the Brazilian government agency dedicated to the financial support of innovative activities) can give an important contribution to solve this deficiency.

In face the structural and institutional conditions of Brazilian software market, company technology is usually developed by the company itself, and only a small part of them made use of technology originating from the university. In relation to infrastructure basis that supports sectoral system of innovation, the strengthening of the educational system seems to be a fundamental aspect. Brazil has a strong education system that over the years produced a significant pool of skilled workers, scientists, engineers, and technologists. Due to the high level of education of its IT professionals, Brazil has been considered one of the better destinations for US companies to outsource software development (Horowitz, 2003). However, there are still a number of weaknesses in the education system which the government has been trying to address. One of them is the low level of IT literacy. Due to low levels of general literacy and the prohibitive cost of computers, IT literacy and skills are below target levels. Formal education, as well as training and qualified hand labor, besides available means of cheap and modern communications must be faced as a basis of an innovative and competitive system. The lack of a dominium of English language seems

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 60 http://brics.redesist.ie.ufrj.br

BBRRIICCSS to be harmless for the diffusion of Brazilian software abroad, especially when compared with countries like India, Israel and Ireland, but this picture might change with the increase of international exposure of Brazilian economy. On the other hand, compared with India and other countries that came to the international software market with the “second wave” of new entrants, such as Russia, and China, Brazil might take some advantage in terms of the dynamism of national innovation system (see Table 31), particularly in terms of IT infrastructure, scientific competences and some experience in the export of high tech products.

Table 31 - Key Statistics of National Innovation Systems in Selected Countries Brazil India China Rússia Annual GDP Growth (%), 2001-2005 2.18 6.84 9.48 6.12 GDP per Capita (in/nal current $ PPP), 2005 8,729.70 3,485.90 6,571.60 10,897.10 GDP (current US$ bill), 2005 794.10 785.50 2,228.90 763.70 Gross Capital Formation as % of GDP, 1995-2004 20.96 24.81 36.09 20.31 Trade as % of GDP, 2004 31.40 41.60 65.30 57.30 Tariff & Non-tariff Barriers (0-5), 2006 3.50 5.00 3.00 3.50 Intellectual Property Protection (1-7), 2006 3.50 4.50 3.30 2.40 Exports of Goods and Services as % of GDP, 2004 18.00 19.00 34.00 34.40 Cost to Register a Business as % of GNI Per Capita, 2006 9.90 73.70 9.30 2.70 Days to Start a Business, 2006 152.00 35.00 35.00 28.00 FDI Outflows as % of GDP, 2000-04 0.42 0.22 0.20 1.39 FDI Inflows as % of GDP, 2000-04 3.72 0.68 3.89 1.36 Science and Engineering Enrolment Ratio (%), 2004 15.89 22.14 n/a n/a Science Enrolment Ratio (%), 2004 8.35 15.62 n/a n/a Researchers in R&D, 2004 59,838.00 117,528.00 926,252.00 477,647.00 Researchers in R&D / Mil. People, 2004 344.00 119.00 708.00 3,319.00 Total Expenditure for R&D as % of GDP, 2004 0.98 0.85 1.44 1.17 Manuf. Trade as % of GDP, 2004 16.38 15.29 50.35 17.85 University-Company Research Collaboration (1-7), 2006 3.50 3.60 3.90 3.20 Scientific and Technical Journal Articles, 2003 8,684.00 12,774.00 29,186.00 15,782.00

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 61 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Scientific and Technical Journal Articles / Mil. People, 2003 47.87 12.00 22.65 109.14 Availability of Venture Capital (1-7), 2006 2.60 4.60 2.70 3.20 Patents Granted by USPTO, avg 2001- 05 135.20 316.40 448.20 194.40 Patents Granted by USPTO / Mil. People, avg 2001-05 0.75 0.30 0.35 1.34 High-Tech Exports as % of Manuf. Exports, 2004 11.60 4.90 29.80 9.10 Private Sector Spending on R&D (1-7), 2006 3.80 4.20 3.60 3.40 Adult Literacy Rate (% age 15 and above), 2004 88.60 61.00 90.90 99.40 Average Years of Schooling, 2000 4.88 5.06 6.35 10.03 Gross Secondary Enrollment, 2004 102.00 53.50 72.50 92.90 Gross Tertiary Enrollment, 2004 22.30 11.80 19.10 68.20 Internet Access in Schools (1-7), 2006 3.60 3.80 4.00 3.80 Public Spending on Education as % of GDP, 2003 4.10 3.30 2.10 3.70 Quality of Science and Math Education (1-7), 2006 2.90 5.70 4.10 4.50 Females in Labor Force (% of total labor force), 2005 42.90 28.40 44.50 49.00 School Enrollment, Secondary, Female (% gross), 2004 107.30 47.30 72.50 92.50 School Enrollment, Tertiary, Female (% gross), 2004 25.40 9.30 17.40 78.80 Total Telephones per 1,000 People, 2004 587.10 84.50 499.40 773.10 Main Telephone Lines per 1000 People, 2004 230.40 40.70 241.10 255.80 Mobile Phones per 1,000 People, 2004 356.70 43.80 258.30 517.30 Computers per 1,000 People, 2004 105.20 12.10 40.90 132.20 Households with Television (%), 2004 90.00 37.00 91.00 98.00 Daily Newspapers per 1,000 People, 2000 46.00 60.00 59.00 n/a International Internet Bandwidth (bits per person), 2004 149.30 11.40 57.40 99.90 Internet Users per 1,000 People, 2004 119.60 32.40 72.50 111.20 Price Basket for Internet (US$ per 28.00 8.70 10.10 10.00

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 62 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

month), 2003 Availability of e-Government Services (1-7), 2006 4.51 3.55 3.96 2.57 Extent of Business Internet Use (1-7), 2006 5.00 4.60 3.50 4.10 ICT Expenditure as % of GDP, 2005" 7.82 5.91 5.28 3.58 Source: World Bank – KAM Database - info.worldbank.org/etools/kam2

Human capital formation has a reasonable history in Brazil, and the country has in general a widespread infrastructure capable of educating and training professionals to the various IT technology areas. Since the period of protectionism a significant amount of resources were ploughed into higher education and training. The number of universities and course offerings increased and this has been maintained in the recent period. The university system annually provide approximately 30.000 IT bachelors, most of them qualified people with proper technical skills, creativity and ability for identifying problems and conceiving solutions for sophisticated software and services solutions (see Table 32). As an example, we can mention the case of Marcelo Tosatti, 20 years old, from Curitiba, Paraná that was selected to be the leader supporter for the current Linux kernel. The basis for the generation of qualified people came from around 875 Graduation Courses in the areas of computer sciences, systems´ management and correlated disciplines, comprising 159,984 enrolled students in 2001. At the same time, IT technical education has been increasingly provided at secondary or non-university levels and this has raised the effective supply of labor to the software sector, albeit at the lower end of the skill spectrum.

Table 32 - Number of students finishing traditional University courses in IT for 2005 Total Private Nature of course Public Institutions Institutions Total Non- Federal State Municipal Private profit Total 29.834 1.927 2.147 992 14.959 9.809 Computer science 15.604 1.623 1.167 557 7.076 5.181 Information Processing 14.050 304 980 435 7.837 4.494 Computer usage 180 - - - 46 134 Source: Min.Education/Inep/Deaes.

The picture at postgraduate education is more problematic. In fact, we can also note an impressive growth of post-graduated basis in the last period. Concerning software activities we can found 29 Mastering Courses in 2002, with 2467 enrolled students and 847 dissertations defended. We can also found 11 Doctoral Courses, with 2467 enrolled students and 80 PhD thesis approved. However, we can also note a relatively low exposure to education abroad and a weaker science orientation, compared with countries like India or China, which has limited the availability of high end skills for the sector. As an evidence of this process we can note that in the area of computer sciences the share of scientific articles of Brazilian residents published in indexed international scientific periodic seems to be quite inferior to the share of other scientific areas such as agronomy, natural sciences and Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 63 http://brics.redesist.ie.ufrj.br

BBRRIICCSS physics (see Table 33). We can also observe that in the last evaluation of the Post-Graduate Academic System developed by CAPES (the Brazilian government agency responsible for the monitoring of that system) only two universities (PUC-RJ and UFRJ) got an grade six or seven (the highest in the scale adopted to evaluate the post-graduate institutions) indicating that they were hardly integrated in the international academic system (see Table 34).

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 64 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 33 -Share of scientific articles of Brazilian residents published in indexed international scientific periodics according to academic fields 2002-2004 % of articles of Brazilian Residents Área 2002 2003 2004 Agronomy 3,00 3,36 3,34 Natural Sciences / Plants 2,10 2,29 2,58 Physics 2,31 1,99 2,48 Pharmacology 1,75 2,15 2,41 Microbiology 2,18 2,48 2,33 Space and Aeronautics 2,01 1,75 2,11 Biology and Biochemistry 1,75 1,89 1,99 Immunology 1,53 1,71 1,98 Ecology and Environment 1,74 1,67 1,89 Mathematics 1,73 1,85 1,81 Chemistry 1,68 1,59 1,66 Multidisciplinary 1,57 1,50 1,58 Materials and Metals 1,49 1,44 1,57 Neurosciences 1,38 1,43 1,42 Molecular Biology and Genetics 1,28 1,68 1,40 Medicine 1,16 1,24 1,40 Geosciences 1,46 1,48 1,38 Engineering 1,14 1,29 1,36 Social Sciences in General 0,80 0,79 1,02 Computer Science 0,86 0,88 0,83 Psychology and Psychiatrics 0,42 0,41 0,46 Economics and Business 0,41 0,35 0,44 Education 0,30 0,28 0,40 Law 0,05 0,10 0,11 Source: Institute for Scientific Information (ISI). National Science Indicators (NSI). Elaboration: ASCAV/SEXEC – Ministério da Ciência e Tecnologia.

Table 34 – Ranking of Brazilian Universities in Computers Sciences - 2004- Area of the University UF Program Level Degree University UF Area of the Program Level Degree PUC-RIO RJ Informatics MD 7 USP SP Bioinformatics D 4 System engineering UFRJ RJ and computers MD 6 UFAM AM Informatics M 3 Computers UFMG MG science MD 5 UNIFACS BA Computers networks F 3 Computers UFPE PE science MDF 5 UNIFOR CE Applied Informatics M 3 UFRGS RS Computers MDF 5 UNB DF Informatics M 3

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 65 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

science Computers UNICAMP SP science MDF 5 UFES ES Informatics M 3 Computers USP SP science MD 5 UFU MG Computers science M 3 Computers science and USP/SC SP math MD 5 FUFMS MS Computers science M 3 Computers UFC CE science M 4 UEM PR Computers science M 3 UFCG PB Informatics M 4 UFPR PR Informatics M 3 Applied System engineering PUC/PR PR Informatics M 4 IME RJ and computers M 3 Computers System engineering UFF RJ science MD 4 UFRN RN and computers M 3 Applied computers UFRJ RJ Informatics M 4 UNISINOS RS science M 3 Computers PUC/RS RS science M 4 UFSC SC Computers science M 3 Computers INPE SP applied MD 4 UFSCAR SP Computers science M 3 Level: D = Doctorate (PHd); M = Master (MSc);F = Professional Master Source: CAPES

In addition, while Brazilian universities – rather than firms - have tended to produce the major stream of R&D, much of this work has had limited market application while serious legal difficulties in assigning patent rules for work done in universities has also had adverse consequences for incentives. Table 35 can be used to illustrate the weakness of systematic links between university and software companies presenting an evaluation of cooperative links extracted from the Directory of Research Groups of CNPq the most important federal agency that supports academic research in Brazilian economic. The data comprises the period between 2002 and 2004, indicating that university-enterprise links related to the development of software was responsible for only 4,9% of the total links of the base7. They also show that approximately 59% of these links are associated with engineering and computer sciences, privileging a flow that came from the university to attend industry demands. Secondarily, the university-industry links to development of software in the area of agronomy and natural sciences seems also to be relevant. Concerning the size of enterprises, there are also evidences that the links tend to be more intense in two extremes case. The first involves very small firms that probably emerge from the academic field, as a direct star-up or with the support of academic incubators. To lustrate this trend, we can mention data collected by Anprotec indicating that approximately 25% of the 283

7 Data extracted from the directory of the academic research groups of CNPq contain a detailed description about the links established between those groups and the productive sector, including information about the subject and the means of knowledge and technology transfer. The information extracted from that directory reveal qualitative aspects of the established relationships of those groups with firms that are not necessarily innovative. Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 66 http://brics.redesist.ie.ufrj.br

BBRRIICCSS incubators identified in Brazilian economy are involved with software and informatics activities in 2004; however, this share had decreased from 50% to 25% between 2003 and 2004. The second alternative involves large companies that can use university competences as a complementary mean of its own innovative efforts. In this context, the range of firms that seems to be misarticulated with the academic field are medium size firms, probably the segment for which this links could have a broader impacts in terms of strengthening industrial competitiveness.

Table 35 - Links between Research Groups and Enterprises by Science & Engineering Fields and by Enterprise Size % of the field in the flow % of the of Develop Develop Develop ment of No of % of the ment of ment of Software links in field in Software Software s in total the field total flow Links s flow Science and Engineering Fields Agronomy 1.640 19,5 46 11,3 2,8 Biological Sciences 819 9,7 32 7,8 3,9 Healthy Sciences 489 5, 8 13 3,3 2,8 Engineering and Computer Sciences 3.688 43,8 212 52,0 5,8 Natural sciences 864 10,3 56 13,8 6,5 Human Sciences 454 5,4 24 6,0 5,4 Social Sciences 402 4,8 19 4,5 4,6 Linguistics, literature and arts 61 0,7 4 1,0 6,9 Total 8.417 100,0 408 100,0 4,9 Size of enterprises % of the field in the flow % of the of Develop Develop Develop ment of ment of ment of Software No of % in total Software Software s in total links flow Links s flow very small enterprises 2.261 26,9 124 31,3 5,5 small enterprises 1.462 17,4 52 13,2 3,6 medium enterprises 1.842 21,9 66 16,8 3,6 large enterprises 2.853 33,9 142 36,0 5,0 Total 8.417 100,0 396 100,0 4,7 Source: data extracted from Directory of Research Groups of CNPq compiled by Rapini (2004)

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 67 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

8 - Local Productive Arrangements in Brazilian Software Industry: an exploratory analysis

The general heterogeneity of Brazilian software industry might be considered not only through a strictly sectoral approach but also through a focus on the territorial dynamics of industry organization. Specifically, we can admit that territorial proximity plays an important role to the improvement of learning and to the strengthening of firms’ competitiveness and innovativeness. In this sense, a solely sectoral approach tends to neglect specificities emerging from institutional and historical contexts, which are specific to local productive systems. Spatial proximity seems important not only because it facilitates inter-firm learning which offers a material base for the development of innovation processes but also because it can provide a basis for the presence of institutions and firms with complementary competencies. Interactions between those agents generate collective learning processes, which contribute to sustain innovative dynamism and industrial competitiveness. The argument is that innovative capabilities can be enhanced trough cooperation between firms and other actors - such as universities, suppliers, customers or even competitors - developed on a local basis. The need to cope with increasingly complex and/or tacit knowledge and the benefits of sharing the costs of innovative activities are among the reasons for firms joining local cooperative networks. However, it is also undeniable that there is a huge diversity in the patterns of cooperative practices among firms, even in the same sector, across countries and regions.

The central question addressed in how identify the presence of Local Productive Arrangements (LPAs) in Brazilian software industry and how to measure and differentiate knowledge and information flows that affect the process of competence building in those arrangements. An analysis based on LPAs concept points out the positive impacts of spatial agglomeration of firms and of inter-industrial articulations to the generation of learning effects and to the increase of productive efficiency. Some general principles should guide the identification and analysis of such arrangements. First, it is assumed that the spatial agglomeration of industrial activities with some degree of similarity is a basic requisite to setting up competences building processes. Second, the emphasis on the interdependence of activities leads the discussion towards the pattern of overlapping of the complementary industrial activities in the same territory or space. Third, the aspects above indicate that the existence of a minimum density of companies and activities is a pre-requisite to set up effective characteristics of such arrangements. Besides, it is supposed that the presence of a dense set of relationships is fundamental in order to deepen interactive learning mechanisms, from which more favorable conditions are created for the consolidation of an innovation system (Lundvall, 1992) structured at either local or regional scale. Fourth, the structuring of such arrangements tend also to stimulate multiple interactions between companies and other institutions thus making possible a dynamic invigoration of innovative processes at both the local and regional level. Flexibility and the existence of strong networks of agents, which result in specialization, are understood to be the major characteristics of tem, creating a local system that boost the technological dynamism and competitiveness of local firms

The main idea behind this notion is that in the contemporary stage of capitalism the economic performance of firms, regions and nations is increasingly dependent on their Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 68 http://brics.redesist.ie.ufrj.br

BBRRIICCSS ability to learn. Furthermore, although information and codified knowledge can be easily transferable across space some crucial elements of knowledge remain deeply rooted in specific locations and institutional designs (Lundvall and Borras, 1997). Yet, it is widely recognised that one of the most important limits to globalisation remains in the spatial mobility of some tacit forms of knowledge. Despite the huge advances in modern communication technologies that make the long-distance interaction fairly quick and easy nowadays, there are still types of information and tacit-forms of knowledge exchange that continue to require regular and face-to-face contacts. Following this arguments, the importance of territorial proximity for localized learning and innovative processes is reinforced by the perception that the tacit forms knowledge have become a key element for competitiveness and innovativeness (Maskell and Malmberg, 1999). Moreover, although the sectorial determinants of innovation systems are important for analysing aspects regarding the knowledge base and technological regimes, the empirical evidences emerging from case studies on local productive systems in Brazil illustrate the remarkable differences on innovative and learning capabilities within local productive systems belonging to the same industry. In this sense, a solely sectorial approach tends to neglect specificities emerging from institutional and historical contexts which are specific to territories. Such diversity seems also to be very important when dealing with the adoption of policy measures.

Based on those hypotheses, we can develop an exploratory analysis about the factors that affect the patterns of learning, cooperation and innovative performance in a set of Brazilian software LPAs. Specifically, the analysis explores the empirical evidences emerging from a survey of four case studies of local productive systems in the industry, in order to analyze and measure the learning, innovation and competence building processes in those systems, discussing the impacts of those processes to the strengthening of innovativeness and industrial competitiveness

This analysis was based on empirical data collected from four case studies of productive agglomerations located in different regions of the country being based on a set of indicators constructed from information collected in the empirical evaluation of those agglomerations. Through this procedure, we seek to explore analytical and methodological issues associated with the study of learning, innovation and competence building processes within software local productive systems, emphasizing specific characteristics of these processes in diverse geographical and institutional context. In doing so, we intend to contribute to the theoretical debate about LPAs in a knowledge based activity of a developing country. In this context, two main analytical issues can be addressed. The first is related to the relationship between territorial proximity, learning strategies, cooperation forms and innovative and competitive performance of firms in agglomerations. The second regards the impact of individual firms’ learning mechanisms on the organization of knowledge systems within agglomerations.

The material on which the followed discussion is based was extracted from a database of LPAs constructed by the team of RedeSist a research network about the subject coordinated by the Rio de Janeiro Federal University (UFRJ)8. The survey covered a sample of 131 firms in four software LPAs located in the provinces of Petropolis (RJ), Curitiba (PR),

8 For more details about this academic network, access www.ufrj.com.br/RedeSist Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 69 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Recife (PE) and Brasilia (DF). All the case studies are characterized by the predominance of SMEs (see Table 36).

Table 36– Characteristics of the firms inserted in LPAs surveyed Number of firms % of the total Firms size Micro 94 71,8% Small 27 20,6% Medium 8 6,1% Large 2 1,5% Location of LPAs Petropolis - RJ 18 13,7% Curitiba - PR 25 19,1% Recife - PE 36 27,5% Brasilia - DF 52 39,7% Total 131 100,0% Source: RedeSist database

The empirical evaluation of learning, innovation and competence building process in LPAs is a complex task, implying the adoption of some methodological procedures to deal with the structural characteristics and performance of these systems. It should be emphasized that, among the different dimensions that might be considered, particular emphasis can be attributed to the systematization of information about the mechanisms of learning, the forms of cooperation and the impacts in terms of innovative performance of firms inserted in them. The analysis is based on a set of indicators that was extracted from data collected in the empirical evaluation of LPAs in software industry. These indicators have been calculated individually for each agent, stressing the heterogeneity of local innovation systems and the specific characteristics of each governance structure. They were manipulated by statistical tools in order to identify different clusters of firms with some similarity in terms of the characteristics of the learning, innovation and competence building processes. Through this evaluation, we can get some tracks about the relative importance of aspects that are extremely useful for the understanding of learning processes within local productive systems.

Four groups of indicators, extracted and normalized from the data base, were articulated to discuss those aspects. A first group of indicators tries to capture the internal or external origins of the main information flows to innovation and learning process. A second group tries to evaluate innovative efforts of firms inserted in local productive systems, conceived as a way to improve the potential of learning gains. The third block of indicators comprises an evaluation of the intensity, nature and impact of cooperative practices implemented by the firms. The forth set of indicators comprises some aspects related with the innovative performance of firms inserted in the local productive systems surveyed. The set of indicators of each group is presented in Table 37. These indicators were calculated from a general questionnaire applied to LPAs located in different industrial sectors – including the four LPAs of software activities mentioned. Through this questionnaire, the firms were

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 70 http://brics.redesist.ie.ufrj.br

BBRRIICCSS asked about the importance of different aspects that affect their competences, competitiveness and innovative strategies. Then, the answers were normalized in a score from zero to one to construct those indicators.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 71 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 37 - Normalized Indicators constructed from data collected in empirical investigations 1 - Learning activities In-hose learning (APRINT) Learning from productive links(APRAGPR) Learning from S&T links (APRC&T) Learning from other agents(APRDMAG) 2 - Innovative efforts Internal training efforts (ESFTRE) Efforts to contract qualified people externally (ESFABS) Constancy of innovative activities(COATIN) Constancy of R&D efforts(CONP&D) Constancy of the acquisition of new technologies (CONOUTC) Constancy of marketing efforts(COFORCOM) 3 - Cooperative Practices Vertical cooperation(COPVER) Horizontal cooperation (COPHOR) Cooperation with services suppliers (COPSRESP) Cooperation with other agents (COPDMAG) 4 - Innovative performance Radical Innovations in products (INPD1) Radical Innovations in processes (INPC1) Incremental Innovations in products(INPD2) Incremental Innovations in processes (INPC2) Organizational innovations (INORG) Note: indicators were normalized from zero to 1 Table 38 presents the values of each group of indicators to the LPAs analyzed, constructed as a mean of the individual normalized indicators of the group. These data permit to contrast some characteristics of the competence building process in the arrangements surveyed. First, it should be noted that any conclusion extracted from the data must be qualified because of the expressive value of the standard deviation compared with the means that were calculated. This value points out the huge heterogeneity among the firms surveyed stressing the strong difference among their strategies and competences. This aspect also reinforces the idea of having different patterns of learning, cooperation and innovation in the software sector. Despite this heterogeneity, the data presented stress also the relevance of some learning activities and the low level of cooperative practices among firms in the LPAs surveyed. Learning activities are particularly high in Petropolis and Curitiba compared with the other LPAs. Innovative efforts seem to be higher in Curitiba and cooperative practices are extremely limited in almost all the LPAs, perhaps with the exception of Petropolis, where the indicator associated with those practices assumes a higher value compared with other LPAs. The general asymmetry between learning index and cooperative index permit to conclude that cooperation is not viewed as a relevant element of technological strategies in software sector.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 72 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Table 38 – Mean ad standard deviation of Indicators constructed for LPAS All the Sample Petropolis Curitiba Recife Brasilia Mean Std.D Mean Std.D Mean Std.D Mean Std.D Mean Std.D ev. ev. ev. ev. ev. Learning activities 0,434 0,255 0,547 0,217 0,518 0,223 0,370 0,249 0,399 0,256 Innovative efforts 0,402 0,313 0,383 0,306 0,487 0,289 0,350 0,323 0,404 0,308 Cooperative 0,164 0,241 0,360 0,278 0,147 0,223 0,063 0,191 0,174 0,226 Practices Innovative 0,428 0,408 0,447 0,388 0,539 0,413 0,359 0,394 0,416 0,404 performance General Mean 0,365 0,311 0,430 0,303 0,436 0,294 0,296 0,298 0,357 0,305 Source: data extracted and manipulated from RedeSist database

The followed step of the analysis was based on a manipulation of the indicators constructed by statistical tools in order to identify different clusters of firms with some similarity in terms of the characteristics of the learning, innovation and competence building processes. Through multivariate techniques, based on not-hierarchical clustering procedures, we have tested the alternative presence of 3, 4, 5, 6 or 7 clusters from the data. Comparing the results generated, the alternative of 5 clusters was selected due to value reached by F statistic. The mean and standard deviation of each cluster to the indictors used is presented in Table 39. We can see that all the indicators seem statistically significant at 5%, indicating that they contribute to the formation of clusters. According to the F statistic, the indicators that seems more relevant for the formation of clusters were respectively incremental Innovations in processes (INPC2); radical innovations in products (INPD1); constancy of innovative activities (COATIN); constancy of innovative activities (COATIN); constancy of R&D efforts (CONP&D); constancy of marketing efforts (COFORCOM) and efforts to contract qualified people externally (ESFABS).

Table 39 – Vale of indicators for each cluster and basic statistics of the clustering procedure Clust Clust Clust Clust Clust Médi er 1 er 2 er 3 er 4 er 5 a (131 Stand Signif Indicators/ Clusters (22 (25 (34 (20 (30 F Firms Devo ic. P Firms Firms Firms Firms Firms ) ) ) ) ) ) 0,714 0,615 0,554 0,373 0,748 0,609 0,000 0,307 6,293 In-hose learning (APRINT) 8 5 6 8 0 8 1 Learning from productive 0,452 0,424 0,410 0,282 0,544 0,431 0,001 0,221 4,939 links(APRAGPR) 8 3 0 1 8 3 0 0,316 0,198 0,141 0,128 0,373 0,233 0,000 0,256 5,566 Learning from S&T links (APRC&T) 1 8 5 6 8 0 4 Learning from other 0,583 0,489 0,382 0,296 0,555 0,463 0,000 0,238 7,161 agents(APRDMAG) 4 0 4 5 7 1 0 0,377 0,239 0,185 0,074 0,387 0,257 0,001 0,316 4,667 Vertical cooperation(COPVER) 5 2 6 5 3 3 5

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 73 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

0,195 0,145 0,087 0,099 0,309 0,169 0,011 0,272 3,411 Horizontal cooperation (COPHOR) 9 8 9 8 5 7 0 Cooperation with services suppliers 0,198 0,071 0,054 0,033 0,241 0,121 0,000 0,199 6,924 (COPSRESP) 7 4 5 1 1 4 0 Cooperation with other agents 0,180 0,102 0,058 0,033 0,157 0,106 0,011 0,175 3,389 (COPDMAG) 9 6 5 1 5 3 4 0,367 0,379 0,306 0,282 0,502 0,372 0,000 0,210 5,241 Internal training efforts (ESFTRE) 9 8 8 4 5 1 6 Efforts to contract qualified people 0,286 0,264 0,170 0,070 0,467 0,260 13,02 0,000 0,246 externally (ESFABS) 1 9 7 1 2 6 2 0 Constancy of innovative 0,511 0,505 0,165 0,253 0,622 0,406 42,18 0,000 0,241 activities(COATIN) 4 0 4 1 9 5 6 0 Constancy of R&D 0,590 0,550 0,117 0,237 0,675 0,425 23,29 0,000 0,350 efforts(CONP&D) 9 0 6 5 0 6 7 0 Constancy of acquisition of new 0,545 0,800 0,220 0,425 0,716 0,530 13,30 0,000 0,406 technologies (CONOUTC) 5 0 6 0 7 5 6 0 Constancy of marketing 0,681 0,600 0,147 0,050 0,616 0,416 18,58 0,000 0,426 efforts(COFORCOM) 8 0 1 0 7 0 1 0 Radical Innovations in products 0,386 0,540 0,073 0,175 0,516 0,332 11,62 0,000 0,375 (INPD1) 4 0 5 0 7 1 4 0 Radical Innovations in processes 0,000 0,480 0,029 0,450 1,000 0,396 49,73 0,000 0,491 (INPC1) 0 0 4 0 0 9 0 0 Incremental Innovations in 0,621 0,413 0,225 0,350 0,500 0,409 0,000 0,322 7,025 products(INPD2) 2 3 5 0 0 7 0 Incremental Innovations in processes 0,954 0,000 0,000 1,000 1,000 0,542 1041, 0,000 0,500 (INPC2) 5 0 0 0 0 0 630 0 0,518 0,552 0,264 0,340 0,640 0,459 0,000 0,351 6,819 Organizational innovations (INORG) 2 0 7 0 0 5 1 Source: data extracted and manipulated from RedeSist database

To the sample as a whole the analysis of indicators shows that in-hose learning (APRINT = 0,60) is particularly relevant. Concerning external information sources, learning from productive links (APRAGPR = 0,43) and learning from other agents (APRDMAG = 0,46) are also important, but learning from S&T links (APRC&T = 0,23) is expressively inferior. Despite the general limitation of cooperative practices, cooperation tends to occur mainly through vertical links with suppliers and customers (COPVER = 0,25). Indicators related to internal training efforts (ESFTRE = 0,37) are higher than those related to the efforts to contract qualified people externally (ESFABS = 0,26). The constancy of innovative activities (COATIN = 0,40) is high, but the main innovative activity is related to the acquisition of new technologies (CONOUTC = 0,53), followed by R&D efforts (CONP&D = 0,42) and by the constancy of marketing efforts (COFORCOM = 0,41). Concerning the innovative performance, data show that incremental innovations tend to be more relevant, as demonstrated by the higher value of incremental innovations indexes (INPC2 = 0,54 and

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 74 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

INPD2=0,40), as well as organizational innovations (INORG = 0,45), specially when compared to radical innovations indexes (INPC1 = 0,39 and INPD1 = 0,33)

Figure 4 illustrate the procedures of plotting the means of the indicators calculate for each cluster. Table 40 summarizes the results that were obtained for each cluster permitting to identify some important differences between them. The Cluster 5 seems to be more dynamic when we consider the indicators as a whole, followed with some distance by Cluster 1, particularly in terms of innovative efforts and performance. Cluster 5 is also the only where cooperative practices seem to have some relevance to competence building processes. Cluster 2 and Cluster 3 are positioned in an intermediary position, with Cluster 2 being better positioned in terms of learning and cooperative practices and Cluster 3 presenting a superior innovative performance. Finally, Cluster 4 seems to be in an inferior position compared with the other clusters, despite the fact that it also presented a reasonable innovative performance, based mainly on the imitation of processes. Figure 4

Plot of Means for Each Cluster 1,4

1,2

1,0

0,8

0,6

0,4

0,2

0,0

-0,2 Cluster 1 Cluster 2 -0,4 Cluster 3 INPC1 ESFTRE Const Outras Tecno APRC&T CPOSRESP Cluster 4 INPC2 COATIN APRINT CPOVER Cluster 5 Variables Table 40 – Main Characteristics of Software Clusters Characteristi Cluster 1 (22 Cluster 2 (25 Cluster 3 (34 Cluster 4 (20 Cluster 5 (30 cs Firms) Firms.) Firms.) Firms.) Firms.) Intensity of internal High Medium/ High Medium/ High Low High sources for learning Medium/ High Intensity of Medium and Medium/ High Medium mainly with external mainly with mainly with mainly with Low other sources sources for productive other sources other sources and productive learning links links s

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 75 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Medium/ high Medium / low Medium and Low and and Intensity of and concentrated concentrated Low / absence concentrated cooperative concentrated in vertical in vertical of cooperation in vertical and practices in vertical actions actions horizontal actions actions Training Medium / low Medium / low effort s and mainly mainly contraction of Medium Medium High internal internal qualified training training people Low and High and Medium/ High Medium/ High Medium/ Low equally equally mainly R&D mainly mainly Innovative distributed distributed and e acquisition of acquisition of Activities among among marketing new new different different efforts technologies technologies activities activities Medium and High and Medium based based on new Low and Medium and based on new on imitation of Innovative products and restricted to oriented to the products and processes Performance processes and the imitation imitation of processes and (mainly )and imitation of of products processes imitation of productions products processes

The analytical exercise is obviously exploratory, but it gives some important tracks about the heterogeneity of the competence building processes in software industry, particularly in a country like Brazil where the amplitude and complexity of internal market gives to different kind of software firms the opportunity of surviving and become profitable. Specifically, we can argue that the capabilities of firms inserted in local productive systems might be upgraded based on intra-agglomeration knowledge systems. In this context, competence building processes might be related to the reduction of innovation lags and to an acceleration of incremental innovations. We can also suppose that, the more tacit and people embodied is the knowledge required to generate technological innovations, the more important will be the construction of the proper channels of contact and communication, in order to allow a systematic interchange of information between the different agents integrated in the local productive systems. Another aspect that must be stressed refers to the impacts of the interchange of information to the definition of standards, certification procedures and quality control techniques that guide the behavior of the agents integrated in local productive systems. These mechanisms play a very important role for the strengthening of industrial competitiveness of the companies integrated in those arrangements. Finally, the analysis reinforces the necessity of policy measures being well informed and well adapted to the characteristics of the LPAs they focused.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 76 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

9 - Concluding Remarks

In the context of a “knowledge economy” the software industry seems particularly relevant not only due to the embodiment of knowledge into products, systems and services but also because it provides the tools to adapt the characteristics of systems, channels, nets and organizations, accelerating the diffusion of new information and telecommunication technologies. Starting from this hypothesis, the analysis developed focus on the complexity, heterogeneity and sophistication of Brazilian software industry. In fact, despite the huge problems and challenges imposed by the trajectory it faced in the last twenty years, software remains a very dynamic industry in Brazil with potential to amplify its economic impact and to stimulate the adoption of innovations, contributing decisively to the growth of productivity and to the deepening of learning and competence building processes in virtually all the economic activities. Throughout this trajectory, the sector has invested significantly in the accumulation of technological capabilities and important backward and forward linkages were developed. A policy regime which emphasized human capital formation, infrastructure building and support to new ventures appear to have also helped significantly the development of the industry.

We can also point out some characteristics of Brazilian software industry that reveal an important set of strengths, weaknesses and opportunities. Besides the amplitude of the internal market, the basis of the Industry’s strength can be associated with the flexibility and creativity of companies and technical personnel, with the sophistication and attractiveness of some of its target markets and to the capacity of generating satisfactory technical solutions for a wide range of economic activities. Among the weaknesses faced by the industry, we can highlight the fragmentation of industrial structure and the presence of policy issues that create some troubles to improve a strategy of effective internalization based on the construction of a well known image abroad. The small amount of exports by Brazilian companies emerges from these characteristics. Despite these problems, we can also identify a lot of opportunities that can be exploited to maintain and enlarges the dynamism of Brazilian software industry. These opportunities include not only segments where Brazilian firms had also consolidated a strong position – such as in the financial sector, ERPs, e-business, telecommunications applicative, e-governament, agribusiness and retail - but also new attractive segments – such as those related to complex integration outsourcing, to free software and to the development of digital games, virtual reality software and broadcasting solutions.

In this context, some challenges concerning policy measures might be stressed. Confronted with the heterogeneity of software industrial structure and to the complexity of its technological dynamics, the policy measures might be flexible and properly adapted to the focus they target. The formulation of a software policy should consider the intrinsic characteristics of the different software activities to elaborate their programs and to construct institutional arrangements. However, the policy also has to consider some important integrated instruments and some general conditions that might be attended to consolidate a sustained growth trajectory. In a country like Brazil, where the socio-

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 77 http://brics.redesist.ie.ufrj.br

BBRRIICCSS economic contrasts are still very dramatic, these policies might also pursue a trajectory based on the reduction of social and spatial disparities.

Concerning the software industry, this perspective implies a focus on the creation of incentives that widen the possibilities of learning, innovating and competence building for the whole industry. Specifically, there are a lot of evidences that the contacts, interactions and exchange of information and knowledge developed trough a local base constitutes an important root to the improvement of innovative competences in the software industry. Moreover, the strengthening of industrial competitiveness might also be articulated with local conditions that improve the potential of learning, innovation and competence building processes. In the software industry, the complex and tacit character of the knowledge required to generate innovations reinforces the importance of the construction of proper channels of contact and communication between providers and users, in order to allow a systematic interchange of information that permits the adaptation of customized solutions according to their needs. The capacity to develop these solutions in a local base permits the accumulation of capabilities that make possible the expansion to more sophisticated markets, including international ones. Considering the territorial heterogeneity of software industry in Brazilian economy, policy measures should try to be well informed and well adapted to the characteristics of the local productive systems in which software activities are embedded.

Concerning the strengthening of export orientation, some requirements might also be stressed. As mentioned, the main challenges that might to be faced by Brazilian companies refer, basically, to the construction of an image of technological reputation, as well as to the improvement of the quality of services provided. Among the aspects that should be improved at the firm level, we can mention the search for quality in all service delivery phases, the concern about competitive cost/benefit solutions and about on-time delivery, and the consolidation of a perspective of supplier continuity in market relations. The continuity of investments in innovation to domain new technological platforms is also extremely important. At the infrastructural level, the availability of an agile, trustable and cost-competitive telecom infrastructure seems to be very important. The availability of human resources is also key element to provide the industry with competitive conditions to follow the market evolution, keeping the original quality and evolving as new technologies and new requirements arrive. The increase of inter-firm and firm-government cooperation is also a prerequisite to reinforces the export drive of Brazilian software industry, as demonstrate some recently initiatives, such as those implemented by BRASSCOM and NEXT. In this sense, the possibility of mobilizing those agents in order to elaborate and implement a strategic plan with a diagnostics and recommended actions to exploit the possibilities opened by the growth of offshore outsourcing market for software and related services seems also to be quite important.

A final point should be raised in the sense that one cannot disregard the intense pressures both internal and external and geo-political concerns in any serious analysis of this kind. In this sense, the Brazilian experience is typical of the difficulties faced by a developing country in a period of intense change like the one we are facing today. It is a period where capabilities in software activities are important components of national competitiveness, being vital for any country to bypass both the ‘digital divide’ and the ‘learning divide’. So, Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 78 http://brics.redesist.ie.ufrj.br

BBRRIICCSS policies initiatives might also be compromised with the achieving of a wider socio- economic development, making the necessary improvements in infrastructure building (and particularly in the educational and vocational system) to effectively include the society in the trajectory of software development.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 79 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Bibliography Aksoy, A. (1990) 'Innovation and diffusion dynamics of the information technology paradigm', Working Paper, No. 11, Centre for Information and Communications Technologies, Brighton: Science Policy Research Unit, University of Sussex. Alerigi A. (2003): “Brazil's Government Leans Towards Linux” in Computer World, 24 November (see: www.computerworld.com/softwaretopics/os/linux/story/0,10801,87507,00.html?nas =AM-87507). Arbache, J. (2002) “Análise do Setor de Software Brasileiro”, Ministério do Desenvolvimento, Indústria e Comércio Exterior Arevolo, W. (2002) “Innovative Program to Develop and Export Brazil’s Software”, Gartner Group, CS-15-5090, March Arora, A. and Gambardella, A. (2004) “The globalization of the software industry: perspectives and opportunities for developed and developing countries”, NBER WORKING PAPER SERIES, Working Paper 10538, NATIONAL BUREAU OF ECONOMIC RESEARCH, Cambridge, MA, June 2004 Arora, A. Gambardella, A. and Torrisi, S. (2001), “In the Footsteps of Silicon Valley Indian and Irish Software in the International Division of Labor,” paper for the “Silicon Valley and its Imitators” project, Standford Institute for Economic Policy Research. Aspray, W.; Mayadas, F. and Vardi, M.Y. (Editors) (2006) Globalization and Offshoring of Software - A Report of the ACM Job Migration Task Force, Association for Computing Machinery ASSESPRO, MBI e ITS (2005) “Exportações Brasileiras de TI & Software”, Pesquisa e relatório desenvolvido pela MBI, disponível em http://www.assespro-sp.org.br/ ATKEARNEY (2005) “Making Offshore Decisions: ATKearney´s 2004 Offshore location Attractiveness Index”. ATKEARNEY e BRASSCOM (2005) “Desenvolvimento de uma Agenda Estratégica para o Setor de IT Off-shore Outsourcing”, Relatório final, Brasília, novembro de 2005 Bastos M. I. (1992): “State policies and private interest; the struggle over information technology policy in Brazil” in Schmitz H. and Cassiolato J. (ed.), Hi-Tech for Industrial Development (London, Routledge). Botelho, A.; Stefanuto, G. and Veloso, F. (2003) “The Brazilian Software Industry”, mimeo, September Botelho, A.J., Stefanuto, G. and Veloso, F., Strengthening the Knowledge Economy: The Software Industry in Brazil. Sociedade SOFTEX., 2003. Breschi, S. and Malerba, F. (1997). “Sectorial innovation systems: technological regimes, schumpeterian dynamics, and spatial boundaries”. IN: EDQUIST, C. (ed.) “Systems of innovation technologies, institutions and organizations”. London: Pinter, UK, p.130-156. Britto, J.; Cassiolato, J.E.; Stallivieri, F. e Guimarães, V. (2006) “Sectoral system of innovation and local productive systems in the Brazilian software industry: a focus on

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 80 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

competence building processes”, Catch up Milano Meeting – Milão, -5-8 Setembro, Milão – Itália Carmel, E. (2003) , “The New Software Exporting Nations: Success Factors” The Electronic Journal on Information Systems in Developing Countries, 13, 4, 1-12 Cassiolato J. (1992): “The User-Producer Connection in Hi-Tech: a Case-Study of Banking Automation in Brazil,” in Schmitz H. and Cassiolato J. (ed.), Hi-Tech for Industrial Development (London: Routledge). Cassiolato J. (1997): “Information and communications technologies in developing countries: the main features and problems surrounding their scientific and technological capabilities”, in M. R. Bhagavan (ed.), Development Aid Approaches to New Generic Technologies in Developing Countries (London, MacMillan). Cassiolato J. and Szapiro M. (2003): “Privatising telecommunications in Brazil: impacts on the innovation system” in Technology Policy Briefs, 2, (2): 3-5, INTECH, United Nations University, Maastricht. Cassiolato J. Lastres H. Maciel M. L. (ed.) (2003): Systems of Innovation and Development: Evidence from Brazil (Cheltenham, Edward Elgar). Cassiolato, J. E., Guimarães, V. and Lastres, H. M. M. (2006) “Innovation systems for information and communication technologies: the case of Brazil”. In BASKARAN, A.; MUCHIE, M. (eds) Bridging the digital divide: innovation systems for ICT in Brazil, China, Thailand and Southern Africa, London: Adonis & Abbey, Commander, S. (2003) “What explains the growth of a Software Industry in some Emerging Markets”, DRC Working Papers no 22, Centre for New and Emerging Markets, London Business School, April CompTIA, (2004) “The Critical Role of the Software Industry in Latin America: A Summary of Studies Conducted in Eight Latin American Markets”, Mimeo, CompTIA, Duarte, C.H.C. (2002 )“Brazil: Cooperative Development of a Software Industry” IEEE SOFTWARE May/June Freire, E. and Brisolla, S.N. (2003) “Innovation and Competitiveness: the challenge to be faced by Brazilian software industry”, paper presented at the Conference “The Future of Innovation Studies, Eindhoven University of Technology, The Netherlands, 20-23 September Frick, S. and Nunes, R. (1996) ‘Produtos, Estruturas de Mercado e Estratégias Competitivas no Setor de Software’, in Economia & Empresa, vol.3, n.1, pp 34-44 Gaio, F. (1990) 'The development of computer software technological capabilities in developing countries - a case study of Brazil', D. Phil. Thesis, Brighton: Science Policy research Unit, University of Sussex. Gaio, F. (1992) 'Software strategies for developing countries', in H. Schmitz and J. Cassiolato (eds) Hi-Tech for Industrial Development – Lessons from the Brazilian experience in eletronics and automation, London: Routledge. Garcia R. and Roselino J (2004): “Uma avaliação da lei de informática e de seus resultados como instrumento indutor de desenvolvimento tecnológico e industrial” in Gestão & Produção, 11(2): 77-185, mai.-ago. (see: www.scielo.br/pdf/gp/v11n2/a04v11n2.pdf) Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 81 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Guimarães V. (2005): “A indústria de software de Petrópolis: Um estudo de caso de Arranjo Produtivo e Inovativo Local” (MSc. Dissertation, UFRJ, Brazil). Gutierrez R. and Alexandre P. (2003): “Complexo Eletrônico Brasileiro e Competitividade” (BNDES Setorial, Rio de Janeiro, Brazil). Horowitz Alan. (2003): “Brazil: It’s a sleeping giant with a tradition of high-quality software” in Computer World, 15 September IDC – ABES(2006) – Associação Brasileira de Empresas de Software “Mercado Brasileiro de Software Panorama e Tendências 2006” (disponível em wwww.abes.org.br) Lastres H. and Cassiolato J. (2005): “Innovation Systems and Local Productive Arrangements: new strategies to promote the generation, acquisition and diffusion of knowledge” in Innovation: Management, Policy and Practice, 7(2-3): 172-187. Lundvall, B.; Bórras, S (1997)“The Globalising Learning Economy: Implications for Innovation Policy”. OECD, DG XII – Science, Research & Development. TSER programme., Lundvall, B.A.; Johnson, B. (1994 )“The learning economy”. Journal of Industry Studies, v.1, n.2, dec, p.23-42., Malerba, F. (Ed.) (2004): Sectoral systems of innovation, Cambridge UK, Cambridge University Press. Malerba, F. e Orsenigo, L. (1996) “The dynamics and evolution of industries”, Industrial Corporate Change, volume 5, no1 Malerba, F. e Orsenigo, L.(1996) “Schumpeterian patterns of innovation are technology- specific”, Research Policy, 25, 451-478, Malerba, F. e Orsenigo, L.. (1993) “Technological Regimes and Firm Behaviour”, Industrial Corporate Change, Volume 2, Number 1, 41-71 Malerba, F. e Torrisi, S. (1992) “Internal capabilities and external networks in innovative activities: evidence from the software industry”, Economic Innovation and New Tecnnolgy, Harwood Academic Publishers, Vol 2, pp 49-71 Maskell, P.; Malmberg, A. (1999) “Localised Learning and Industrial Competitiveness”. Cambridge Journal of Economics, v.23, n.2, p.167-185,. MBI (2005) Panorama da Industria latino-americana de software ; MBI Mayer & Bunge Informatica Ltda MCT-SEPIN (Various Years): (Brazilian Ministry of S&T/IT Policy Secretariat, see: www.mct.gov.br/sepin). MDIC (2004) Ministério do Desenvolvimento, Indústria e Comércio Exterior (org) “O futuro da indústria de software: a perspectiva do Brasil: coletânea de artigos”, Instituto Euvaldo Lodi / Núcleo Central. Brasília : MDIC/STI : IEL/NC,. MIT & SOFTEX (2002). “Indústria de software no Brasil – 2002: Fortalecendo a economia do conhecimento”. Massachussets Institute of Technology – MIT, Brasil, coordenação geral Sociedade SOFTEX. Campinas. 80p.

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 82 http://brics.redesist.ie.ufrj.br

BBRRIICCSS

Prochnik, V. (1996) “Cooperation between Universities, Companies and Government in the National Export Software Program - SOFTEX 2000”, CEPAL/UNO Roselino, J.E. (2006) “Análise da Indústria Brasileira de Software com Base em uma taxonomia das Empresas: subsídios para apolítica industrial”, Revista Brasileira de Inovação, Volume 5, no 1, janeiro-junho 2006, pp.157-201 Roselino, J.E. (2006)”A Indústria de Software: o “modelo brasileiro” em perspectiva comparada”, Tese de Doutorado, IE-UNICAMP Royce, W.W. (1970). Managing the development of large software systems: concepts and techniques, 1970 WESCON technical papers, Vol. 14, p. 723. Schmitz H. and Cassiolato J. (ed.) (1992): Hi-Tech for Industrial Development (London, Routledge). SOFTEX (2004). Mercado de Software – Notícias, disponível em www.softex.org.br Spence J. (2001): “Access in Brazil,” in Trends in Latin American Networking, May Stefanuto, G.N (2004)., O Programa Softex e a Indústria de Software no Brasil, Tese de Doutoramento, Instituto de Geociências, DPCT, UNICAMP, Campinas, 2004 Stefanuto, G.N e Carvalho, R.Q.(2005), Perfil das Empresas Brasileiras Exportadoras de Software, DPCT/IG Unicamp, e Observatório Digital Softex, Torrisi, S. (1998): Industrial Organization and innovation. An international study of the software industry, Cheltenham, Elgar. Veloso F. Botelho A. Tschang T. and Amsdem A. (2003): “Slicing the knowledge-based economy in Brazil, China and Índia: a tale of 3 software industries” (Softex, Campinas, Brazil). Website A: www.e-commerce.org.br Website B: www.ipeadata.gov.br Website C: www.softex.com.br

Av. Pasteur 250 Urca Rio de Janeiro RJ CEP 22290-240 Tel 55-21-3873.5279 Fax 55-21-2541 8148 83 http://brics.redesist.ie.ufrj.br