Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

10. Women in software programming

Software as a technology and its production process
The software and computer services sector
An analysis of patterns of women's employment in software activities

The experience of Brazil

Fatima Janine Gaio


The extraordinary technical progress in microelectronics, coupled with the convergence of a diverse set of technologies, has favoured the pervasive diffusion of information technologies (ITs). As the neo-Schumpeterians have pointed out, such a process involves extensive structural changes, threatens previous economic, social and organizational paradigms with 'creative destruction', and heralds new ones. But this revolutionary process seems, so far, to have left women behind. The broad 'conventional' pattern of feminine employment has not been significantly altered. Women tend to be found clustered in low-skilled and low-paid jobs, and to be systematically excluded from the techno-scientific occupations. This picture may change in the future, at least in some areas of IT such as the production of software.

Software production, a core component of IT, is a labour-intensive, design- and science-based activity. It is an area in which both industrialized and developing economies could reap considerable economic benefits. The question addressed in this chapter is whether its growing commercial importance could create a window of opportunity for women, and so change the conventional pattern of women's employment in developing countries. The outcome will depend on the direction of movements in three dimensions: the participation of women in software activities; whether the employment offered in these activities, and particularly to women, suffers deskilling; and the level and nature of software activities on both the internal and external markets. The international competitiveness of the software and computer services industry plays an increasingly crucial role in developing countries, and thus has a significant influence on the total volume of employment which the industry can offer. The optimum out- come for women would be their increasing participation, at higher levels, in a skill-intensive and growing industry.

Software as a technology and its production process

This section will seek to relate recent changes in software production to changes in work organization, and consider the implications for job opportunities for women. The background here is the ongoing labour process debate around the question of whether the long-term trend in productive processes is in the direction of deskilling and job fragmentation. But there are a number of differences between software development and more conventional production systems, so that theories based on the latter may not provide an adequate framework for understanding the evolution of software production, and software-related occupations.

Software programming as a technology and occupation

Software production is a design-intensive process producing logical and non-material products whose core function is to incorporate coded knowledge and structural flexibility into products and processes. The idiosyncrasies which arise from the non-material nature of software are magnified at present because the industry, and the technology, have not yet matured.

Software can best be understood in the context of its evolutionary interdependence with user requirements and hardware.1 The dynamic interplay of these three components has influenced the pace and direction of the generation and diffusion of IT-based systems, as well as generating a series of imbalances and bottlenecks. There have been exceptional improvements in the cost and performance of electronic components and hardware, due to a continuous flow of technological innovations. Software production, in contrast, has remained an essentially labour-intensive activity, based on intellectual skills. This evolutionary 'desynchronization' has been exacerbated by steep increases in the variety and complexity of software which is required.

In the 1970s there was growing apprehension that a 'software crisis' might hinder the diffusion of ITs. The 'crisis' was substantiated by a range of forecasts such as the reversal of the comparative cost relationship between hardware and software, with software being responsible for an increasing share of system costs;2 a growing gap between the demand and supply of software products and services; and a scarcity of skilled labour.

The 1970s could be called a turning point in the development trajectory of software. Its recognition as a major bottleneck in the IT complex was paralleled by, and influenced, significant technological, economic and social changes. The technological evolution in hardware accelerated, with the advent of semiconductors and microprocessors, coupled with increasing integration and miniaturization of electronic components.3 This made it possible, in the late 1970s, to launch microcomputers, creating a potential mass market. An independent economic sector developed, specializing in the development and selling of software, with packages (software products) as the major growth area. There was a growing managerial emphasis on controlling the software development process, together with a tendency to apply industrial scientific management principles to job fragmentation and deskilling.

The evolution of software as a computing occupation

From the 1980s on, as information technologies have become more central to business strategies, three broad trends have emerged. First, productivity and reliability in software production have continued to be constraints, and there has been increasing emphasis on the concept of software engineering. Second, there has been an increasing emphasis on the adequacy of emerging systems to meet the needs of users. Software development, and its evolution, have come to be seen as an interactive process in which both users and producers have significant roles. Third, there has been a growing shift to 'downsizing', i.e. to decentralize computer resources to user departments, and transfer applications from mainframes to microcomputers linked by telecommunications networks. The two latter trends, in particular, are bound to create major disruptions in organizations and to change the composition of the computing staff they require.

Despite these trends, most labour process analyses of software development work relate to mainframe computers. Since the emergence of centralized data processing installations, mainframes have been the focus of attention. These computers have the aura of complex high-tech, and the staff directly linked to them have had a high status. Complementary tasks, such as the operation of input and output devices, have been regarded as subsidiary activities which involved much lower skills and, therefore, are much more easily subject to scientific management practices. From the outset. data-entry has been seen as a clerical job with low educational requirements and skill levels, and thus as 'appropriate' for women (Braverman, 1974). Friedman ( 1986) has observed that the Braverman/ Kraft hypothesis of task fragmentation and deskilling does apply to data entry work. Data entry has been deskilled to the point of alienation, producing both physical and mental health hazards.

Although more 'flexible' management strategies have been adopted in computing' a hierarchic structure seems to remain (e.g. systems and applications analysts, analyst/programmers, programmers, computer operators, data preparation and data entry staff).4 Women's participation is greatest at the bottom of this pyramid. Case studies by Soares (1989) in local firms in Brazil showed that women accounted for the major share of employment only in jobs related to data entry. In the largest firm studied, around 60 per cent of the data entry and data preparation staff were women. In another firm, women were responsible for 40 per cent of total computing jobs, but while 87 per cent of data entry employees were females, only 37 per cent of programmers and analysts were women. These findings appear to support the broad scenario outlined by Cockburn (1985), that the generation and diffusion of information technologies has not substantially altered the traditional association of women's employment with low skill and education requirements, low status and low paid work.

But there is also scattered evidence that some national software markets and areas of work may provide more conducive environments for women's employment, and should be interesting for future investigations.5 For example, recently, software was evaluated as a very promising area, in India and in the USA, for females as professionals and managers.6 In Brazil, from the late 1980s to the early 1990s, women have accounted for about 50 per cent of graduate degrees in computer science and mathematics, conventionally seen as predominantly male areas.7

The limitation of mainstream labour process theory, as applied to software development work, is that it has focused on some kinds of 'hard' technical knowledge, and so has found that automation and scientific management will lead to deskilling.8 It has tended to overlook the 'soft' technical knowledge involved in areas such as the analysis and basic design of systems, the complex work of modelling a problem to be computerized, communications and interactions tasks, and keeping a software product in tune with the evolving environment it is supposed to reflect.

Evolving and expanding software requirements are bringing major changes for software development occupations. The emergence of analyst-programmers, already noted by Friedman and Greenbaum (1984) in the 1980s, not only entails a reversal of the Braverman/Kraft fragmentation hypothesis, it also points to reskilling of the labour force. The new occupational structure of development work entails different sets of skills, changing the content of programmers' and analysts' work and combining parts of them in new job descriptions. The analyst-programmers may be an indicator of new trends in the division of labour and job contents.

Analyst-programmers would be more closely linked to the technical 'know-how' of computing, being responsible for the detailed design of systems. A new category of professionals appears likely to emerge - the business or information analysts. Probably at the top of the hierarchy of development work, these will be senior staff who combine a deep knowledge of business with the technical know-how of computing. These professionals would be responsible for identifying strategic new applications of information technologies and specifying the systems requirements in close interaction with users. Their work could be called the phase of basic design, or the 'know-why'.

With the increase in more advanced applications, another new kind of professional is emerging in development work: 'specialists' with specific knowledge in the application fields being modelled.

The software and computer services sector

This section provides a brief picture of international markets for software and computer services, and contrasts the Brazilian experience in the sector with the theoretical strategies for developing countries suggested by the current literature.9 The question here is whether the total volume of employment which this industry offers to both men and women in a country such as Brazil can be maintained at significant levels. If it cannot, women's entrance to this high-status work will not have a significant effect on the overall profile of women's employment.

The international context

As observed by the OECD (1989), many of the existing interpretations of the software sector are based upon quantitative and simplified structural analyses which do not reflect the industry's intricate differentiation or dynamic nature. Quantitative analyses10 have generally been applied to the relatively more mature 'computer services' segments, such as commercial electronic data processing (EDP) and general purpose computer systems.

Because there has not been any thorough understanding of the supply sector, there has been a tendency to associate the industry structure with the USA model, as an international benchmark. The USA is both the leading world market for computer services and the major software exporter. The USA is the most conducive market for producing standardized software products, due to its large internal market, closeness to sources of technical change in complementary technologies (e.g. computers, microelectronics and telecommunications), and tight labour markets. The local software and hardware producers' led by a small number of large firms have played significant roles in the industry's success. The independent industry is active in various market segments, but their market share has been higher in packages for microcomputers.11

The OECD (1989: Table 2, pp. 24-25) has presented a broad picture of the world market for computer services and software. In 1987, it was estimated that the total revenues of the industry would amount to US$ 78199.6 million (m), of which about 61 per cent would come from software (US$ 47667.3m), with the USA, Japan, France, Germany and the UK accounting for around 77 per cent of these sums. The USA alone earned about 50 per cent of world revenues. The participation of newly industrializing economies was very small. Brazil, India, South Korea, Mexico, Singapore and Taiwan together represented only about 6.5 per cent of the world market (US$ 5068.2m) and 5.4 per cent of the total software market (US$ 2575.9m). Brazil was the most important country within this group, with a market of US$ 4217.6m (83 per cent of this group) and software turnover of US$ 2186.2m (about 85 per cent).

The literature regarding the computer services markets has often stressed software and software packages as a source of significant economic opportunities. However, it is debatable whether the USA phenomenon has been replicated even by other industrialized countries. As Correa (1989) noted, the national markets of both industrially advanced and newly industrializing economies are generally highly dependent on software imports, particularly in the segments of systems and utilities and applications tools. In cross-industry packages, trade is increasingly internationalized, and the few standard products traded in the internal markets come from foreign sources.

In the literature it is generally argued that software could be a point of entry to information technology for developing countries, due to its labor-intensive nature, relatively low capital requirements, and slow rhythm of technical change (see e.g. Ernst, 1981; Hoffman, 1982; O'Connor, 1985). Two main strategies have been recommended: to concentrate on the production of applications solutions, exploiting the 'natural' market reserve created by diversified market conditions, and to generate export opportunities by reaping the benefits of low-waged labour as a comparative advantage (UNIDO Secretariat, 1983; Kopetz, 1984; Narasimhan, 1984). The export-led strategy of India, since the 1980s, shows that there is room for international outsourcing. India's software exports have grown remarkably, and are expected to reach US$ 350 m by 1993, but this is still a very small share of the world market revenues. Indian export activities have been based primarily on subcontracting programming and writing components for large systems designed abroad.12 There are already indicators of the Indian experience being followed by countries such as Singapore, Taiwan and Russia.

The Brazilian experience13

Brazilian industrial development in general has been associated with Government intervention, under a broad 'import-substitution' policy. From the mid- 1970s to 1990, Brazil pursued a policy of promoting an indigenous IT complex within its economy in an integrated manner. The strategy aimed at 'self-reliance' through fostering the accumulation of internal technological capabilities in combination with local entrepreneurship. This strategy proved to be too ambitious.

The policy measures were gradual, focusing on the supply side and favouring the hardware industry, while software activities were not assigned a high priority. Yet there was some early recognition of the importance of devising a specific policy for software (SCSS, 1981), and the promotion of an indigenous software industry was a goal. Particular emphasis was given to the production of packages for locally-manufactured mini and microcomputers.

Until the mid-1980s, the main policy instruments included a registry of computer programmes combined with Government procurement of local products; import control coupled with a refusal to accept any legal protection for intellectual property rights (IPRs); and making approval of manufacturing projects for microcomputers conditional on the use of indigenous operating systems. In addition, a three year diploma course in computer science was established at the university level, with a practical emphasis. The objective was to accelerate the education of skilled personnel for software development. It is not equivalent to a full degree course, which takes 5 years.

By 1987, the revenues of the computer industry as a whole amounted to US$ 2,578m, and total employment to 30,947. Local hardware firms, manufacturing in the 'market reserve' segments of mini and microcomputers, were already responsible for around 53.4 per cent of this amount and for 76.2 per cent of the total labour employed in the industry. The indigenous microcomputer segment was characterized as the most dynamic in the industry (SKI, 1987; DEPIN, 1991). But local demand for software was still largely met by imports, particularly in the mainframe and microcomputer segments (SKI, 1986).14 Some 90 per cent of software turnover was estimated to come from imports, channelled mainly through subsidiaries of multinational corporations (MNCs). Local private firms had 19.6 per cent (US$ 450.3m) of the computer services market in 1987, and employed 26,388 people. The internal market for computer services does not appear to have been conducive to fostering local firms' activities, yet local services firms, with revenues amounting to less than one third of the revenues of the local computer industry, generated more employment than the computer industry.

Another important feature of the Brazilian computer services market is the strong presence of public firms, which mainly act as captive (i.e. 'in-house') data processing bureaux for Government institutions. The revenues of these firms rose from US$ 484.2m (26.9 per cent of the total market) in 1985 to US$ 789.9m (a 34.3 per cent share) in 1987. This was about twice the revenues of local private firms.

The policies which provided supply inducements for local computers in the domestic market also promoted the externalization of demand for software products and services, without creating any comparable conducive market for local software products. Imported packets continued to dominate the market. Effective control of software imports would in fact have jeopardized the expansion of the local hardware industry, particularly in the microcomputer segment, which was the actual centre of the Brazilian informatics policy. No clear rules for the internal market were adopted, and there was increasing internal and external political pressure for liberalization. The 1988 Software Law merely provided a legal framework for the de facto market liberalization which was already in place. Thus, in practical terms, software activities have never been the subject of effective 'infant industry' protection. Local firms operating in this sector have been exposed to international competition from the early stages.

The Brazilian case provides some empirical evidence as to the validity of the main software strategies being suggested for developing economies. First, Brazil does not provide any significant evidence to support the conventional theoretical view that economies endowed with low-waged skilled labour enjoy comparative advantages in software. On the contrary, in the early stages of the diffusion of information technologies, when cross-industry packages tend to predominate, the demand tends to be met through imports.

Second, in 1993 around 15 Brazilian software houses were exporting software, mainly to the USA and the European Union, either through joint ventures with foreign partners or by establishing subsidiaries abroad. The majority produced cross-industry packages for microcomputers, with total revenues estimated at US$ 20 to 30m.15 Although this could be seen as a very modest result, these firms have proven to be internationally competitive, carving out niches in highly internationalized market segments. Following what Schware (1992) called a 'two-legged' model, they have struggled to accumulate technological and marketing capabilities in the internal market and have then moved to exports. Whether such a strategy will remain successful and can be expanded on is an interesting topic for future research.

Table 10.1 The local private sector for software and computer services in Brazil, 1985-1987: distribution of revenue, employment and number of firms by market segment

Main activity

Revenues US$ million

Average annual growth rate of employees

No. of firms



















% total





Data processing






% total











% total











% total





Trade & maintenance of computers, peripherals and software






% total





Other services (including systems integration)






% total






Source: Elaborated from ASSESPRO (The Brazilian Association of Processing Services Firms) ( 1988).

Note: Data covers ASSESPRO members only. In this survey. firms declared their revenue by main type of activity.

Finally, the evidence of Brazil confirms that applications solutions have been a major area of opportunity for local firms, but are not a sufficient base for promoting the dynamic growth of an independent software sector. The software industry has, in practice, remained stalled at an incipient stage. To a large extent, local opportunities in applications software have remained limited to the 'in-house' activities of IT users. The predominance of imported packages in the internal market has limited the scope for local production at a number of levels. Not only the existing distribution channels, but also users, have shown a marked preference for well-finished and well-known international brands, and a reluctance to interact with local producers of still amorphous products. Such interaction is necessary to the development of new applications. Thus imports seriously undermine the future prospects of local information technology initiatives, by reducing both firm-level learning and the positive employment externalities for computing personnel.

Contents - Previous - Next