MacroScan - Concentration in the Competitive Software Business

Statistics on the global production and trade in software are notoriously inconsistent, because of difference in coverage and methodology. Any single source can at most provide an indication of the structure of and trends in the industry, rather than an exact measure of its size.

According to leading market research firm Datamonitor, the global market for software totalled $143.7 billion in 2004 (Chart 1). The interesting feature is, of course, the structure of this market. In terms of market segments the North American market clearly dominates, accounting for slightly more than 50 per cent of the total. Europe is a distant second (28.8 per cent) followed by the Asia-Pacific (16 per cent) and the rest of the world (4.9 per cent) (Chart 2).
Chart 1 >>

A large home market, among other things, has served the US industry well. It dominates the global trade in software as well. As Table 1 indicates the US has accounted for a quarter to a half of global trade in one area (packaged software) for which figures are available in the UN's commodity trade database.
Chart 2   >>

The industry is also highly concentrated in terms of market shares of leading firms, with US firms dominating the market. The top four firms (Microsoft, IBM, Oracle and Computer Associates), accounting for 41.5 per cent of the global market, are from the US (Chart 3). This structure belies the conventional notion that the software sector is characterised by extremely low barriers to entry.
Table 1   >>

The software sector is seen by many as characterised by low costs of entry and an easily accessed and almost universally available knowledge-base for innovation. What is of special significance is that the sources of this knowledge, such as journals, conferences, seminars and publicly or privately financed training programmes, are easily accessed. This makes it easy for wholly new entrants to acquire the knowledge base required for cutting edge technological contributions to the industry, as was and is true of at least some of the myriad start-ups in Silicon Valley. Thus, the software sector is seen as one where knowledge is easily acquired and innovations easily replicated, requiring skilled labour but little by way of capital investment.

This perception of the industry fails to take account of the heterogenous nature of the industry which has added on different segments in the course of its evolution, driven by technological changes in hardware that have created new demands and opportunities. Writing in 1995 Martin Campbell-Kelly classified software firms into three distinct sectors, based on their historical evolution: software contractors, and the personal computer software industry. The role of software contractors was to develop one-of-a-kind programs for computer users and manufacturers, buying and selling expensive computer systems with limited capabilities by today's standards. Packaged software producers emerged in the 1980s attempting to provide standard software for applications needed by specific sets of clients. These two sets of software firms either served or competed with hardware firms, who sought to develop software applications for purchasers of their equipment. They were part of the computer industry establishment.
Chart 3   >>

It was for this reason that Campbell-Kelly chose to treat producers of software for personal computer users as a distinct category, even though they were involved in the same kind of business as the packaged software producers. In his view producers of personal computer software were in the business of generating products that would have a large number of users, if successful, and consisted of many firms that were outside the traditional software establishment.

Since then software contractors (including relatively small firms) and hybrid firms like IBM have developed into service companies providing enterprise software, systems integration and consultancy services to large corporations, adding a major software services component to the industry. However, the evidence seems to suggest that the structure of the US software industry has not changed very much over time.

Further, the evidence suggests that the market for large scale public and private projects were characterised by significant barriers to entry, with established firms and a few successful start-ups that grew rapidly in size dominating the market. However, entrepreneurial firms always had a place in the industry, providing custom programs and software maintenance services of modest scale to medium-sized firms. According to one estimate there were 2800 software contractors in the US in 1967, many of whom were small firms catering to smaller clients. In this market, the only barriers to entry were programming knowledge, technical knowledge of the applications domain and the availability of a client.

With the growth of the market for packaged software in the wake of IBM's decision to unbundle software and hardware in 1968, it would appear that smaller firms would have a new market. But in actual fact the need to develop a product fully, by investing a substantial number of programming man-hours, before testing it and entering the market increased sunk costs substantially. This was in itself a barrier to entry. And, though the arrival of the personal computer increased the scope for packaged software substantially, the problem of high sunk costs remained. As has been repeatedly noted, producing the first unit of a software product requires large investments in its generation, whereas producing an additional unit is almost costless. The larger the sales, therefore, the lower the average cost and the higher the return. But that is not all. When large sales imply a large share of the market as well, scale becomes a means of ensuring consumer loyalty and strengthening oligopolistic positions. This is the result of ''network externalities'' stemming from three sources. First, consumers get accustomed to the user interface of the product concerned and are loath to shift to an alternative product which involves some ''learning'' before the features of the product can be exploited in full. Second, the larger the number of users of a particular product, the greater is the compatibility of each user's files with the software available to others, and greater the degree to which files can be shared. The importance of this in an increasingly networked environment is obvious. Finally, all successful products have a large number of third-party software generators developing supporting software tools or ''plug-ins'', since the applications program interface of the original software in question also becomes a kind of industry standard, increasing the versatility of the product in question without much additional cost to the supplier. These ''network externalities'' help suppliers of a successful software package to ''lock-in'' consumers as well as third party developers and vendors, leading to substantial barriers to entry.

Partly because of these characteristics successful start-ups like Microsoft, which entered the market at the right time, came to dominate the industry. As a result, even though the history of Silicon Valley is full of anecdotes of tech-savvy entrepreneurs discovering new possibilities and new products, concentration is the dominant feature, with most start-ups with innovative products now being acquired rather early in their history.

The reasons for this need to be spelt out. Take the case of software products for mass use. Creating such a product starts with identifying a felt need (say, for a browser once the internet was opened up to the less computer savvy or for a web-publishing programme once the internet went commercial). The persons/firms identifying such a need must work out a strategy of generating the product, by hiring software engineers, at the lowest cost in the shortest possible time. Once out, the effort must be to make the product a proprietary, industry standard. This involves winning a large share of the target consumers, so that the product becomes the industry standard in its area. Once done, the product becomes a revenue generating profit centre.

The investment required is the sums involved in setting up the company, in investing in software generation during the gestation period, and in marketing the product once it is out so as to quickly win it a large share of the market. Needless to say, while entry by individuals or small players are not restricted by technology, they could be limited by the lack of seed capital. This is where the venture capitalists enter, betting sums on start-ups which if successful could give them revenues and capital gains that imply enormous returns.

There are, however, three problems here. The first is one of maintaining a monopoly on the idea during the stage when the idea is being translated into a product. The second is that of ensuring that once the product is in the public domain competitors who can win a share of the market before the originator of the idea consolidates her position do not replicate it. It is here that a feature of 'entrepreneurial technologies' – the easy acquisition and widespread prevalence of the knowledge base needed to generate new products - considered an advantage for small new entrants actually proves a disadvantage. Thirdly, no software product is complete, but has to evolve continuously over time to offer more features, to exploit the benefits of increasing computing power and to keep pace with developments in operating systems and related products. Thus large and financially strong competitors, even if they lag in terms of introducing a product 'replica', can in time lead in terms of product development, and erode the pioneer's competitive advantage.

There are two aspects of technology that are crucial in this regard. First, their source. Second, the appropriability of the benefits of a technology. As mentioned earlier, in the case of software the sources were in the public domain. This was where the advantage lay for the small operator. But once a technology is generated based on some expenditure in the form of sunk costs, there must be some way in which the innovator can recoup these costs and earn a profit as incentive to undertake the innovation. In the Schumpeterian world this occurred because of the 'pioneer profits' that the innovator obtained. The lead-time required to replicate a technology itself provides the original innovator with a monopoly for a period of time that generates the surplus which warrants innovation.

Most often this alone is not enough to warrant innovation and in the software sector lead times can be extremely low, especially if the competitor invests huge sums in software generation, reducing the lead-time substantially. It is for this reason that researchers have defended and invoked the benefits of patents, copyright and barriers to entry in production, which allow innovators to stave off competition during the period when sunk costs are being recouped. Unfortunately, neither is the status of patents and copyrights in the software area clear (as illustrated by the failure of Apple to win proprietary rights over icons in user interfaces), nor are there barriers to entry into software production.

This has had two implications. First, the importance of secrecy in the software business. The 'idea' behind the product must be kept secret right through the development stage, if not competitors can begin rival product developments even before the original product is in the market. A feeble attempt to institutionally guarantee such secrecy is the now infamous 'non-disclosure agreements' which prospective employees, financiers and suppliers are called upon to sign by the innovator who is forced to partially or fully reveal his idea. Secondly, even after the product is out, since the threat of replication remains, it is necessary to strive to sustain the monopoly that being a pioneer generates. This is where the possibility of locking in users with the help of an appropriate user interface which they become accustomed to and are reticent to migrate away from, and locking in producers of supportive software with an appropriate 'applications programming interface' becomes relevant. It should be obvious that sustaining monopoly to recoup sunk costs can indeed be difficult.

Such strategies did help the early start-ups, resulting in the jeans-to-riches stories (Microsoft, Netscape, etc.) with which Silicon Valley abounds. But more recently it has become clear that start-ups undertake innovative activities only to create winning products that the big fish acquire. This is because of the possibility of easy replication and development of an original product, which can be done by dominant firms with deep pockets that allow them to stay in place and spend massively to win dominant market shares. In the event, the likelihood that a small start-up would be able to recoup sunk costs, clear debts and make a reasonable profit is indeed low. Selling out ensures that such sums can indeed be garnered. And selling out is often a better option than investing further sums in developing the product, now faced with a competitive threat, in keeping with industry and market needs.

Given this feature of the software products market, it is not surprising that small players (such as Netscape with it Navigator and Vermeer Technologies that delivered Frontpage) are mere transient presences in key areas even in the developed countries. To expect developing country producers to fare better is to expect far too much. The latter can merely be software suppliers or outsourcers for the dominant players.

In sum, other than in the supply of services to medium-sized firms or serving as contractors for relatively small projects by industry standards or serving as sub-contractors to leading software contractors, the basic nature of the software sector seems to be such that concentration is the key.

This is a factor that firms from countries like India have to confront when attempting to exploit the benefits of their pool of skilled cheap labour. India has been successful in breaking into this sector. But that success is also predicated on having an extremely concentrated industry here. Thus a recent study of 65 small and medium enterprises in the IT sector (B.G. Shirsat, Business Strandard, July 14 2006), with revenues ranging from Rs.10 crore to Rs.200 crore, found that their revenues in 2005-06 amounted to Rs.3,400 crore, which was just 8.9 per cent of the Rs.38,169 crore revenue garnered by the top four IT firms (TCS, Wipro, Infosys Technologies and Satyam Computer). Their profits aggregated Rs 575 crore or 6.9 per cent of the Rs.8,386 crore earned by the top four. This concentrated structure is sustained either through the acquisition of smaller firms or by the exit from the industry because of unviability. This has implications for policies aimed at ensuring the proliferation of software firms as part of India's ongoing IT thrust.