Research Policy 30 (2001) 1267–1287

The Indian software services industry Ashish Arora a,∗ , V.S. Arunachalam b , Jai Asundi c , Ronald Fernandes a a

H. John Heinz III School of Public Policy and Management, Carnegie Mellon University, Pittsburgh, PA 15213-3890, USA b Engineering and Public Policy, Robotics and Materials Science and Engineering, Pittsburgh, PA 15213-3890, USA c Engineering and Public Policy, Pittsburgh, PA 15213-3890, USA Received 13 February 2000; received in revised form 14 June 2000; accepted 7 November 2000

Abstract The Indian software exports have grown in spectacular fashion. Its success has, for the most part, been a combination of resource endowments, a mixture of benign neglect and active encouragement from a normally intrusive government, and good timing. The bulk of the Indian software exports have consisted of fairly mundane services such as low level programming and maintenance. The marked reliance on access to low cost human capital has prompted considerable scepticism about the ability of the Indian software industry to sustain its performance, given the rapid growth in the demand for engineers and the relatively inelastic supply of engineers. This paper reports on the results of research on the Indian software industry. We use a variety of sources, including a questionnaire survey of Indian software firms, and field visits and interviews with industry participants, observers, and US based clients. Although, maintaining the current rate of growth will pose a number of challenges, these challenges are not insurmountable. Not only can the available pool of human capital be expanded by tapping and training the very large pool of English-speaking college graduates, the leading Indian firms are making strong efforts to move up the value chain by acquiring better software project management capability and deeper knowledge of business domains, and reducing costs and improving quality by developing superior methodologies and tools. Moreover, the greatest impact of the software industry on the Indian economy may well be indirect, in its role as an exemplar of the new business organisational form and as an inspiration to other entrepreneurs. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Software; Exports; Indian software industry; Economic growth; Human capital

1. Introduction Technological revolutions sometimes bring unexpected opportunities for countries. India, a relative laggard among developing countries in terms of economic growth, seems to have found such an opportunity in the information technology revolution as an increasingly favoured location for customised software ∗ Corresponding author. Tel.: +1-412-268-2191; fax: +1-412-268-5161. E-mail addresses: [email protected] (A. Arora), [email protected] (V.S. Arunachalam), [email protected] (J. Asundi), [email protected] (R. Fernandes).

development. India’s success at software has led to speculation about whether other developing countries can emulate its example, as well as whether this constitutes a competitive challenge to software industries in the developed world. In this essay, we focus on the Indian software export sector. After briefly describing the main features of the industry, we analyse the major challenges it faces and assess its prospects for the future. We also briefly discuss the implications of the Indian experience for other developing regions and for software industries in the developed world. Our analysis is based on field visits to over 40 Indian firms in Bangalore, Bombay, Hyderabad and

0048-7333/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 7 3 3 3 ( 0 0 ) 0 0 1 4 8 - 7

1268

A. Arora et al. / Research Policy 30 (2001) 1267–1287

Delhi, where we interviewed nearly 75 senior managers and software professionals. These interviews were loosely structured around a questionnaire that we developed in consultation with industry experts and were followed by interviews with 15 US based firms that had outsourced software development to these firms in India. 1 We complemented the field research with publicly available data on firms (NASSCOM, 1994–99), as well as information from a questionnaire survey administered to over a hundred Indian software exporters. 2 Finally, we had brief structured interviews with 60 software development professionals in Indian firms to understand better where and how they are trained and the nature of the work they do. Section 2 places the development of the Indian software industry in an international and historical context. Section 3 discusses the main features of the industry, and Section 4 describes how software service exports by Indian firms are organised. As late as 1998 or 1999, the picture that emerges is of an industry that is based primarily on providing relatively simple software programming and coding services, exploiting its superior access to a supply of cheap software programmers, and with only limited degree of technical and managerial contribution. Though this situation has improved in a number of ways since 1998, there is no doubt that low cost human capital remains an important source of advantage. 3 In Section 5 we analyse the supply of human capital, along with other factor inputs. In thinking about the future prospects for the Indian software sector one must recognise that this is different from the fate of individual Indian software firms. In principle, the Indian economy could do quite well merely for quite some time simply as a source of low cost programming talent, and by tapping the very large pool of English-speaking graduates to supply the growing demand for Internet enabled services, such 1

The questionnaire, along with other background information about the project and some of the outreach activities is available on the project website, at http://www.heinz.cmu.edu/project/india. 2 The National Association for Software Service Companies is an industry organisation that represents the interests of software firms to the various government departments. 3 Since then, a number of the leading firms have demonstrated that they may have accumulated sufficient market and technological knowledge to compete in other ways.

as medical transcription and call centres. However, clearly the Indian economy could do even better if indigenous firms could leverage this talent to produce higher value services and products. More precisely, in Section 6 we identify at least three distinct strategies that Indian software firms have adopted, albeit with varying degrees of success. The first involves the acquisition of greater user-sector specialisation and the accumulation of domain knowledge, enabling the firm to offer “solutions”. This is complemented by a second strategy that involves good software development methodology and high quality standards, and the development of tools to provide these solutions in a cost effective and timely fashion. The third strategy, which as yet very few firms have adopted, involves developing products. Section 7 summarises our findings and concludes.

2. Background 2.1. Packages, services and custom developed software Software development can be broadly categorised into custom developed software and packages or generic software products. Customised software development involves close interaction between the development team and the end-user. Software products may be targeted to a vertical segment or may cut across segments, but rarely to a specific user. In some cases, business software products, such as ERP packages that manage the flow of inputs, work in process and shipments in a company, are very large and complex. These require a great deal of customisation before they can be used. Often, this customisation is done by outside software consultants. Information technology consultants, such as Anderson Consulting, provide “solutions”, which may involve some combination of custom developed software and commercial off-the-shelf software and hardware products. Software development involves a number of stages: conceptualisation, requirement analysis, high-level design, low-level design, coding, testing and support. These stages roughly correspond to stages described in the waterfall model of software development, shown in Fig. 1 (Royce, 1970. See Boehm (1981) for an alternate model of software development, the

A. Arora et al. / Research Policy 30 (2001) 1267–1287

1269

Table 1 India: growth in domestic, export and total revenuesa Year

Exports in US$ M

Total US$ M

Employment

Revenue/ employee (US$)

1993–1994 1994–1995 1995–1996 1996–1997 1997–1998 1998–1999 1999–2000 (Est.) 2000–2001 (Est.)

330 485 734 1085 1800 2650 3900 6300

557.9 825.8 1249.4 1765.8 2700 3900 5600 8600

90000 118000 140000 160000 180000 250000 n.a. n.a.

6198.5 6998 8924.5 11036 15000 15600 n.a. n.a.

a

Source: Nasscom and CMU Software Dataset.

Fig. 1. The waterfall model of software development.

Spiral Model). Indian software exporters largely provide services rather than products. Further, Indian software exports consist largely of low-level design, coding, and maintenance services, which are considered to be low in the value-chain of software services. 2.2. The Indian software industry in an international context Even though its US$ 5.7 billion software revenues in 1999–2000 was a tiny fraction of the estimated world software market of over US$ 300–500 billion, the Indian software industry has attracted a disproportionate amount of interest as a source of software. 4 There are some compelling reasons, nonetheless, for the attention on India. The Indian software industry has captured a significant portion of the world trade in software services. One estimate suggests that India has 16% of the global market in customised software, and that more than 100 of the Fortune 500 had outsourced 4 One must note that this does not include the software developed by users themselves, nor does it include embedded software.

to India (Dataquest, 31 July 1996; pp. 43–44). 5 Table 1 describes the growth in revenues and employment in the software industry, and shows that the industry has grown at over 50% per year over the last 5 or 6 years. 6 Assuming this growth rate can be sustained, the industry association, NASSCOM, has projected 23% share of world customised software market and 5% of products and packages market 2003. NASSCOM also projects that software exports will constitute about 25 per cent of India’s total exports by 2003, up from its current level of 5%. Even if these projections are only partially fulfilled, the Indian software industry shall still have achieved a substantial role in the world software industry, especially in customised software and software services. The Indian industry is comparable to that of the Irish and Israeli software industries in terms of revenues and exports. Table 2 reflects the emergence of Israel as a source of entrepreneurial firms developing software products in areas such as security and anti-virus technology. In 1997, there were about 300 software firms in Israel, employing nearly 10,000 people, with 5

In 1995, the global market for computer services was estimated by IDC as being over US$ 220 billion. Of this, a substantial fraction involved outsourcing of some part of the software development and maintenance process: custom software development was estimated to be nearly US$ 16 billion, systems integration at US$ 32 billion, IT consulting at US$ 11 billion and business service outsourcing at US$ 9 billion. 6 Indian industry grew at an average rate of 7.6% while the service sector grew at an average rate of 8.2% over the same period. (Indian Budget 1999–2000 Table 1.2a., www.ieo.org/budget99/table 1.2a.html).

1270

A. Arora et al. / Research Policy 30 (2001) 1267–1287

Table 2 Israel: growth in export and total revenuesa Year

Exports (US$ M)

Total (US$ M)

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 (Est.)

65 89 110 135 175 220 300 400 540 700

380 450 540 600 700 800 950 1100 1300 1500

a

Source: Israeli Association of Software Houses, Israel.

Table 3 Ireland: growth in export and total revenuesa

and employment in the Irish software industry. The Irish industry employed over 18,000 people, with 1997 revenues of over US$ 5.5 billion as well. The software sector comprises more than 600 companies, of which about a fifth are overseas firms. Note that India has the largest number of people working in the industry as well as apparently the highest rate of growth of revenue, but also the lowest revenues per employee. Revenue per employee is about US$ 150,000 in Israel and Irish firms appear to earn US$ 60–800,000 per employee. 8 By comparison, Indian exporters earn only about US$ 20,000, and even the established firms earn only a little over US$ 30,000. Since the Irish industry is closer in its activities to India, the comparison is especially relevant. The Indian industry is much larger, is growing somewhat faster, but is markedly less productive.

Year

No. of firms

Total employment

Exports (US$ M)

Total (US$ M)

2.3. A brief historical background

1991 1993 1995 1997

365 417 483 679

7793 8943 11784 18300

2266.3 2594.5 3959.3 5942.5

2436.7 2805.7 4221.3 6283.3

These features of the Indian software industry are rooted in the factors that underpin its rapid rise. The Indian success story has, for the most part, been a combination of resource endowments, a mixture of benign neglect and active encouragement from a normally intrusive government, and good timing. By the late 1980s, India was graduating, approximately, 150,000 English-speaking engineers, with only a limited demand for their services from the rest of the economy. By the late 1980s as well, India’s economic liberalisation was also well under way. Around this time, the information technology revolution in the developed world had begun to take root and shortages of skilled programmers and IT professionals were beginning to develop. 9

a

Source: National Software Directorate, Ireland.

total revenue of over US$ 1.5 billion. A large fraction of the firms are engaged in developing software packages, often technically highly sophisticated, for export markets. As well, many of the world’s largest computer companies including Microsoft, IBM, Digital, Hewlett Packard, National Semiconductor, Motorola and others have set up software development centres in Israel. 7 The Irish software industry develops software products, as well as provides a variety of software development and support services. Table 3 provide data on the growth in exports, total revenues, firms 7 These features reflect the large number of technically skilled people with high entrepreneurial drive available in Israel. Reportedly, about 20% of Israel’s population consists of those who arrived in the last 5 years, doubling the number of technicians, engineers and scientists. Israel now has 135 engineers and technicians for every 10,000 people, compared with only 18 for the US and 40 for India. (Economist, 20 February 1999, p. 27). And venture capital has flowed. There are said to be over 50 venture capital funds operating in Israel and more than US$ 4 billion was invested in high technology start-ups in Israel in 1998.

8 Overseas firms in Ireland have revenues per employee in excess of US$ 400,000. However, these firms have strong tax incentives to book revenues in Ireland for products and services sold all over Europe, and therefore, the true productivity is likely to be substantially lower. 9 A frequently cited study concluded that there were an estimated 346,000 IT positions currently vacant in the US, in three core IT occupational clusters (programmers, systems analysts, computer scientists and engineers) (ITAA, 1998). Barr and Tessler (1996 and 1998) claim that the shortage is part of a secular trend that had, in the past, been disguised by cuts in defence spending in the US between 1988 and 1993, that resulted in 75,000 programmers being laid off in Orange county alone. The shortage was further masked by downsizing in the IS departments of many major corporations.

A. Arora et al. / Research Policy 30 (2001) 1267–1287

By this time a number of Indians were working in US firms. Some of them played an important, although as yet undocumented role, in bridging the gap and matching the buyers in the US with the suppliers in India. Responding quickly to the growing demand, a number of Indian firms arose in quick time. Contrary to its normal practice, the State encouraged this growth by considerably simplifying the process for obtaining the numerous clearances and permits that any firm in the organised sector in India typically needs. Finally, the low levels of initial investment required to start a software services firm meant that finance was not a serious constraint upon entry. The Indian software industry consists of a large and growing number of firms: using NASSCOM membership as a measure, the number of Indian software firms has grown from around 38 (accounting for 65% of industry revenue) in 1988 to over 545 (accounting for over 95% of industry revenue) in 1999. Table 4 shows that many of these firms entered the industry during or just before the economic liberalisation in 1991, and few have exited. A few big companies dominate the industry, with the top 25 companies accounting for 58.67% share of software exports revenue in 1997–98, whereas nearly 125 firms, or over a quarter of all firms, had revenues below US$ 250,000. The market leaders in the Indian software firms are, for the most part, relatively new themselves. Further, with a few exceptions, notably Wipro and Satyam, these firms specialise in software alone. This is in marked contrast to early entrants into the industry, who had close links with computer hardware development (Heeks, 1996, p. 69). With a growing need for maintenance services many firms entered the industry by Table 4 Age distribution of Nasscom member firms in 1997 (n = 426)a Year of establishment

Number

Age (years)

≤1980 1981–1985 1986–1990 1991 1992 1993 1994 1995 1996 1997

31 33 111 27 43 31 49 38 48 15

≥19 14–18 9–14 8 7 6 5 4 3 2

a

Source: CMU Software Dataset.

1271

providing these services, often by sending software programmers to the client on a temporary basis. These entrants were of two types. The first type consists of existing firms diversifying into software, including computer hardware firms, such as HCL and Wipro, as well as firms with large in-house data processing and system integration capabilities such as Larsen & Toubro (LTITL). Others such as BFL, Sonata, Satyam and Birla Horizons began as divisions of industrial groups. 10 The other type of entrants was new start-ups, such as PCS, Datamatics, Infosys and Silverline. Indeed, managers at a large number of software firms have worked in these companies earlier in their career. One of the best-known software exporters, Infosys, was founded by a group of seven PCS managers who broke away from PCS. Infosys’s first contract was a support and maintenance contract with a client in the apparel industry for whom PCS had finished a large project. Many MNCs have set up liaison offices and subsidiaries as well. Initially, they intended to sell hardware and software in the Indian market, often in partnership with a domestic firm. Increasingly, however, the objective is to use India as a place for software development. Many MNCs have established software development centres in India, and are exporting packages or components of systems to other countries from India. 11 As well, there are a number of US firms with large Indian operations that are very similar to Indian software firms. Firms such as Mastech (now IGate), Information Management Resources (IMR), Syntel, Cognizant (a subsidiary of Dunn and Bradstreet) and CBSL use their India operations much in the way that Indians software export firms do, to tap a large pool 10 In addition to these firms that focused on software exports, there were others that served domestic users, most notable Computer Maintenance Corporation (CMC). Responsible for maintaining computer systems after IBM left India, CMC has grown to over 2000 employees and developed the ability to develop and implement large and complex projects, especially for infrastructure systems. CMC has also proved to be a good training ground for managers that would later be employed by other, private sector firms. 11 For instance, the operating system for the “Network Computer” introduced by Oracle is said to have been designed entirely in India. Similarly, the Texas Instrument R&D centre in India is capable of fairly sophisticated work, including analogue chip design. Other prominent MNCs operating in India include Motorola, Siemens, Hughes Network, Computer Associates, Microsoft and Cadence.

1272

A. Arora et al. / Research Policy 30 (2001) 1267–1287

Table 5 Revenue distribution of NASSCOM member firms by geographic region, 1997a

Table 7 Composition of Indian software development and services (domestic and exports)a

Region

Revenue (US$ M)

Number

Software activity

Domestic (%)

Export (%)

Bombay & Pune Bangalore Delhi Chennai Hyderabad Calcutta Gujarat Kerala Other

597.5 323.6 285.8 130.9 62 44.8 3.1 1.4 8.3

107 84 95 34 21 26 13 7 18

Turnkey projects Professional services Products and packages Training Support and maintenance IT enabled services

28.6 4.1 52.0 6.1 3.2 6.0

31.5 48.4 8.8 1.5 3.0 6.8

a

Source: CMU Software Dataset, N = 405.

of relatively cheap but skilled workforce for providing software services to US based clients. Virtually, all are headed by entrepreneurs of Indian origin, and started their existence, as did many of the leading Indian firms, by supplying software professionals such as programmers and analysts to clients in the US. As Indian software exporters establish overseas subsidiaries, the distinction between the two will tend to diminish. Contrary to popular belief, as Table 5 shows, the industry is not concentrated exclusively in Bangalore, although Bangalore is certainly a very prominent location, along with Bombay, Pune, Madras and Hyderabad are important as well. However, with the exception of the region around Delhi, there are no noticeable clusters in the northern or the eastern regions of India. The distribution of engineering colleges, concentrated in the western and southern regions, closely mirrors the distribution of the software industry. As Table 6 shows, engineering colleges are heavily concentrated in these two regions, which also account for the greater part of employment in the Indian software industry.

a

Source: NASSCOM.

3. Characteristics of the Indian software industry The Indian software sector displays many unusual features compared to most Indian industries. The most obvious one is its export orientation: exports account for 65% of the total software revenue. Given India’s size and history of inward development, most industries tend to be driven by the domestic market. 3.1. Domestic market There are important qualitative differences between the export market and the domestic markets. The first relates to different types of software developed. Table 7 gives the composition of the domestic and export software development and services market. The domestic market has a higher proportion of revenues from the sale of software packages and products. Whereas products accounted for nearly 40% of the domestic market, they account for a little under 10% of exports. Over 80% of exports are software services, which includes custom software development, consultancy and professional services. Even though the bulk of the product revenues in the domestic

Table 6 Number and capacity of Indian engineering colleges, approved by 1998–1999, by regiona Region

Number of colleges

Sanctioned capacity (no. of students)

% of sanctioned capacity at self-financed colleges

Central East North (incl. north–west) West South (incl. south–west) Total

50 25 140 140 308 663

9470 4812 25449 34165 82597 156493

0.52 0.26 0.42 0.74 0.79 0.69

a

Source: Ramarao, 1998.

A. Arora et al. / Research Policy 30 (2001) 1267–1287

market are probably accounted for by imported software products, a number of Indian software firms have also developed software packages for the domestic market and other developing countries. However, with some exceptions, these packages, which included accounting software, Indian language word processing packages, and banking and accounting products, have not been very successful. 12 Although it is tempting to blame weak intellectual property rights for the failure of Indian firms to develop successful packages, our interviews suggest that, at least as important, if not more, has been the lack of experience in design and marketing. In many cases, firms simply overestimated the willingness to pay and underestimated the difficulties of developing and supporting products. The reluctance of Indian users to pay large sums for software products has undoubtedly been very important, as has the slow rate of computerisation of the Indian economy. The result was that since the early 1990s, firms recognised the higher profitability of service exports compared to other types of software development and even firms that are product focused have added software services and consulting to fund product development (see also Udell, 1993). The second difference between the domestic and export sectors relates to the stages of software development as described earlier. For domestic clients the industry provides a wider range of services that usually spans the entire lifecycle of software development. Domestic projects are larger and more challenging than export projects, with the screen based trading system for the Bombay Stock Exchange and the Reservation System for Railways, being two recent examples. Until recently, the bulk of Indian service exports have consisted of low-level design, coding and testing. Further, Table 8, showing results from our survey, shows that most Indian software firms provide services to a diverse set of industries including manufacturing, banking and insurance, retail and distribution and telecommunications. They rarely 12 An ERP package by Ramco has enjoyed some success. A subsidiary of Citicorp, which has successfully exported its banking products to a large number of developing countries. Other firms have also targeted other developing countries as outlets for specialised products for the financial, banking and hospitality industry. Firms like TCS and Infosys have their own banking packages, which they have exported to other Commonwealth countries with reasonable success.

1273

Table 8 Industries served by Indian software exportersa Domain area

Number of firms

Percentage of firms

Banking Medical Retail, warehousing Multimedia and entertainment Education Travel and tourism Manufacturing Government related Transport

45 23 47 27 23 16 46 22 27

50 25.6 52.2 30 25.6 17.8 51.1 24.4 30

a Source: CMU Software Dataset – CMU Survey of Indian Software Industry, N = 90.

specialise in a particular sector. This observation suggests that the firms are working in that area of software development where business knowledge is least required. However, there is anecdotal evidence that the extent to which Indian firms have provided higher value services like problem-conceptualisation and high-level design has increased. Interestingly enough, firms do not appear to derive any advantage from domestic consulting experience in the export market. The CEO of a software subsidiary of a very large Indian engineering firm explicitly noted that the considerable experience his firm had in executing large in-house software development projects was of limited use in exports. “As far as functional skills are concerned, these differ from country to country and client to client and there is a learning curve there. Domestic expertise may be useful in gaining technical expertise such as in coding and project management. However, domestic and export projects are two different ball games.” 3.2. Exports Although some of the leading firms are beginning to differentiate themselves from the rest, the leading Indian software export firms are remarkably similar in terms of their activities. These activities include maintenance tasks for applications on legacy systems such as IBM mainframe computers, development of small applications and enhancements for existing systems, migration to client–server systems, and increasingly, e-commerce related services and “solutions”. The

1274

A. Arora et al. / Research Policy 30 (2001) 1267–1287 Table 10 Destination of Indian software exports, 1997–1998a

Table 9 Nature of export projects undertaken by Indian firmsa Service

Number of firms

Percentage

Destination region

% of export revenues

Network designing and engineering Conversion project System integration Application solution software System/utility software Application tools Operation and network mgmt Help desk operation Datacentre management

19 25 22 71 28 22 7 12 12

20.4 26.9 23.7 76.3 30.1 23.7 7.5 12.9 12.9

USA Europe SE Asia Japan Australia & New Zealand West Asia Rest of the World

58 21 6 4 2 2 7

a Source: CMU Software Dataset – CMU Survey of Indian Software Industry, N = 93.

Year 2000 (Y2K) problem had also opened a large market for firms that were traditionally doing mainframe based maintenance projects. Table 9, displaying results from our survey, shows that application solutions are the most common type of export, followed by re-engineering (which is the term used to describe porting applications from one platform to another) and conversion projects, such as Y2K projects. Although Y2K projects were an important source of revenue, most of the leading Indian software firms have limited their dependence on such projects, and this is reflected in the sustained growth of the Indian software exports in 1999 and the first half of 2000. Managers at most of the US firms we interviewed in 1998 agreed that the type of work outsourced was neither technologically very sophisticated nor critical to their business. 13 Requirement analysis and high-level design is typically done either in-house or by US based consultants. However, smaller firms may rely more heavily upon their Indian suppliers, as was the case of a small firm developing medical software. 13 The managers at a leading electronics and telecom firm said they outsource work related to sophisticated but mature digital signal processing software to their Indian subsidiary. The telecom firms we interviewed outsourced domain related software maintenance or tool development for the maintenance or enhancement of existing applications. The manager at a value added telecom services firm said that they were outsourcing testing of their existing software and to some extent maintenance of their old UNIX based software. However, we did find one exception to the idea that the outsourced projects are not mission-critical: a leading computer manufacturer out sources critical device-driver software that is shipped directly from the Indian vendor for distribution.

a

Source: Nasscom.

Not only is the work outsourced technologically undemanding, the projects are typically small. The mean number of man-months involved in the most important export project in 1998 for firms that participated in our questionnaire survey is 510 man-months, whereas the median is only 150 man-months. Since the question related to the most important project, it implies that the typical export project is quite small. Table 10 shows that the US accounts for over half of all export revenues (58% in 1997-98), compared with 21% for Europe and 4% for Japan. Many of the larger US firms we interviewed are knowledgeable about outsourcing software development and the strengths and weaknesses of Indian software services firms. Some of these firms have also outsourced software development to firms based in other countries like the Ireland, Philippines, Russia and South Africa. The US is not only a major market; US based service providers are important competitors. Although competition from other countries such as Philippines and China is typically cited in the press, as Tables 11 and 12 show, most software exporters indicate that their Table 11 Location of primary competitors of Indian software firmsa Location of competitors

Number of firms

Percentage of firms

India Israel Ireland USA Singapore Philippines Eastern Europe/Russia

75 12 12 58 19 6 10

82 13 13 63 21 7 11

a Source: CMU Software Dataset – CMU Survey of Indian Software Industry, N = 92.

A. Arora et al. / Research Policy 30 (2001) 1267–1287 Table 12 Perceived location of primary competitorsa One of top three locations

Small

Medium

Large

India USA Other countries Total in category

39 27 24 48

21 17 14 28

15 14 9 16

a

Source: CMU Software Dataset – CMU Survey of Indian Software Industry, N = 92.

main competitors are located either in the US or in India itself. However, with few exceptions, most of the US based competitors are themselves firms that extensively recruit Indian software professionals.

4. The organisation of software outsourcing to India 4.1. A typology of software exports Software exports can be divided into three categories based on where software is developed and how the development is managed and organised. The first category is onsite consultancy or onsite projects, where the Indian company provides the US client with software professionals with the particular technical skills asked for by the client. In essence, the entire project is executed at the client’s site. The client manages the project, controlling the deliverables and deadlines. The software is developed according to the client’s processes and a more accurate description would be to label this supply of staff augmentation services to overseas clients. 14 The second category of exports has a mix of work done offshore (i.e. in India) as well as onsite. In this model, the Indian company sends a few software professionals to the client’s site for requirement analysis or training in a particular system. These professionals then bring back to India the specifications for the software and a larger team develops the software offshore. If the project is large, a couple of Indian professionals remain at the customers site acting as liaisons between the project leaders offshore and 14

A common term for some of these “onsite projects” is “bodyshopping” wherein all that the client gets is programmers. There is no value-add in terms of accumulated knowledge/expertise.

1275

the clients. Sometimes these onsite professionals are needed for emergency operations and for reassuring the client that the project is proceeding according to schedule. To execute such projects, a firm needs not only skilled professionals, but also a software development process and methodology, and an ability to manage software development. Unlike in onsite projects, the Indian firm provides technical and managerial expertise for offshore projects. The third method of software export, similar in some respects to offshore development, is in the form of an Offshore Development Centre. An Offshore Development Centre is a popular organisation form, especially for firms based in the US and Europe and who wish to take advantage of the skilled talent pool and lower wages in India. An offshore development centre involves an umbrella contract with a long-term agreement on prices for time and materials (usually standardised on a man-hour basis). Periodically, the client sends projects to the centre. For each project, the negotiations are largely restricted to the resources and time that will be required. In some cases, the place where the work is done is physically separate (from the rest of the Indian company) and secured. Many of the established Indian software firms will have more than one development centre, and we interviewed firms that had five or six offshore development centres or more. Firms that have been outsourcing software to Indian firms for a long period prefer this form of organisation since they are confident of the Indian firm’s capabilities and rely on their processes for delivering software. Our survey (sample of 65 software export firms) indicates that on average 42.7% of the total work was done offshore for the most important project, consistent with NASSCOM estimates of 41%. The distinction between offshore and onsite work is important because offshore development is more cost effective. Our interviews suggest that one man-year of onsite work is billed at about US$ 90,000–100,000 while comparable offshore work is billed at US$ 25,000–35,000. The bulk of the difference is accounted for by the higher cost of living in the US, as well as greater overheads and communication costs. Offshore work is widely believed to be more profitable for the vendor. However, there are some important limits to the extent of offshore work. An important reason is the need for face-to-face communication. Further, since many projects are cost plus (“time and

1276

A. Arora et al. / Research Policy 30 (2001) 1267–1287

Table 13 Fixed fee versus time and material export contractsa

Time and materials contract Fixed fee contract Firms that have had both Other

All firms (%)

Small (56 firms) (%)

Medium (32 firms) (%)

Large (14 firms) (%)

53 42 3 2

52 45 2 2

53 41 3 3

57 36 7 0

a

Respondents were asked about the nature of their most important export contract. Source: CMU Software Dataset – CMU Survey of Indian Software Industry, N = 102.

4.2. US experience with software outsourcing to India

software programmers to work onsite. The shift to offshore work requires substantial investment in physical infrastructure (including secure physical and computing infrastructure that some clients demand to protect their intellectual property). It also requires that the Indian firm be able to demonstrate project management capabilities. 15 The most frequently cited reasons for outsourcing have to do with the shortage of skilled professionals in the US. Firms claim that they simply cannot find enough software professionals fast enough. In addition, firms outsource because they do not want to invest in in-house capability in areas outside their core-competence (such as developing applications for old computing platforms) and to free their in-house IT staff from mundane maintenance tasks for more creative projects. US firms engaged in developing software products emphasised the need for accelerating product development in the face of ever-shorter product lifecycles. In some cases, US firms outsource to Indian firms to get access to more specialised engineering talent, particularly in the area of telecommunications. Other reasons include the option of round the clock operations and the ability of Indian vendors to assemble “functional” teams of engineers at a very short notice. Most of the US managers we interviewed commented on the excellent programming and coding skills available in India. They also noted that their Indian vendors were good and willing learners, receptive to new ideas, and flexible in terms of the software and hard-

Typically, the US firm begins by outsourcing a fairly small project to the Indian vendor, with the objective of gauging the latter’s capabilities. Not only are the initial outsourced projects small, much of the work boils down to the Indian firm supplying

15 Managers at a large software product firm said that given the rapid pace and nature of product development, they mostly relied upon onsite work. In effect, they use their Indian vendors as a staff augmentation agency, while for the Indian firm, this is an opportunity to train its staff in the tools and methodology used at one of the largest software producers in the world.

materials” is the term used in the industry), clients have to trust the supplier to not overcharge them (Gopal, 1996; Bannerjee and Duflo, 1998). Fixed fee contracts involve greater risk taking by the vendor in contrast to cost plus or time and material contracts. With greater risk also comes greater control over the organisation and management of work. The available evidence suggests a steady increase in the fixed price component of work, indicating the growing maturity of the Indian software firms, an impression confirmed by several of the managers we interviewed. Bannerjee and Duflo (1998) present data from a sample of 236 contracts (not including offshore development centre contracts) from 125 software companies. They find that 58% are fixed price contracts, while another 27% are “mixed”. Only 15% of the contracts are pure time and material contracts. This tends to exaggerate the differences because the dynamic nature of software development results in changes in requirements and specifications, resulting in frequent time and cost over-runs, which are often shared between the Indian vendor and the overseas client. From Table 13 we see that 53% of the firms we surveyed indicated that their most important export project in the last 12 months was a cost plus contract. The 42% of the firms had a fixed price contract for their most important export project. Interestingly, 32% of all the firms also claimed that their contract contained a penalty or incentive clause related to quality or schedule.

A. Arora et al. / Research Policy 30 (2001) 1267–1287

ware platforms for which they provide services. One reason that we expected to hear but frequently did not, was outsourcing to reduce costs. Many of the respondents downplayed this issue, insisting that cost was a relatively minor consideration. This appears to be a response to the prevailing concern in the US about the possible harm to US engineers from software outsourcing and inflow of foreign software programmers. The apparent lack of concern with cost is inconsistent with the extensive price competition that prevails in the Indian software industry, and with the actual behaviour of some of the firms that outsource to India. 16 Our interviews with US firms also revealed a number of areas of dissatisfaction. Many of the interviewees thought that the Indian firms had no domain knowledge and poor management skills. Even a highly rated Indian subsidiary of a leading electronics and communication firm was considered 4–5 years behind the latest communication technologies by managers in the US. Most of the managers believed that Indian firms could not work on high level specifications or project definition stages of a project, although for the most part, this belief had not been tested. Indian productivity levels are lower as well. For instance, managers at the electronics and telecom firm noted that they needed to assign more engineers to a task in their Indian subsidiary than would be assigned in the US. Many were critical of the Indian system of promoting software programmers to managers based on seniority rather than on proven managerial ability. Interviewees felt that this weakened project management. Indian firms, on the other hand, cited this practice as a way of providing a career path to their professionals and a major part of their attempts to hold down employee attrition. However, software firms are now actively recruiting MBA students from Indian colleges like the IIMs as project managers. Though typical business undergraduate or MBA courses do

1277

not cover issues such as design and software project management, firms themselves try to develop these skills through on the job training or other types of in-house programs. A very large fraction of the managers we interviewed considered employee attrition a big problem and wanted their Indian suppliers to tackle it quickly. Some of them recounted experiences where virtually the entire project team left after the first 6 months, causing substantial delays. A manager at one of the telecom firms that had had considerable experience with Indian software vendors spoke at some length about how frustrated he was with the apparent inability or unwillingness of Indian firms to move up the value chain. He claimed that his firm would like to be able to outsource more software design and development tasks to its Indian vendors. In particular, this firm would like its suppliers to display more initiative in identifying business problems that it faces and propose solutions. Even though this manager claimed to be willing to help the suppliers to acquire such capabilities, in his view, the suppliers were not responsive enough to this opportunity. In addition, there were a number of cultural and political issues that US managers perceive as irritants or barriers. One such issue is the apparent unwillingness of Indian software professionals to point out potential problems up-front, and in general, an unwillingness to say no for fear of offending the clients. Another related weakness is the lack of familiarity of many Indian firms and professionals with the work culture and work norms in the West, and especially in the United States. 17 Other difficulties included resistance within the US to foreign programmers, poor telecommunication infrastructure, and the delays in obtaining the required visas for Indian programmers.

5. Factor supplies 16 For instance, one of the telecom firm we interviewed had, in addition to a couple of prominent Indian firms, a firm largely in the business of “body-shopping” among its list of preferred vendors. Our interview suggested that this was a way of promoting competition and keeping prices down. This firm’s policy was to ensure that its business accounted for a substantial (but less than 50%) share of the vendor’s revenues and that no single vendor had more than 25% of its business. These policies ensured that it had considerable bargaining clout in the pricing negotiations with its Indian vendors.

5.1. Human resources It is widely believed that the key to the success of the Indian software exports is the supply of trained, low cost software professionals. Table 14 shows that 17 This includes the sideways movement of the head to indicate agreement, as opposed to the Western nod.

1278

A. Arora et al. / Research Policy 30 (2001) 1267–1287

Table 14 Software professionals: comparative salaries, 1997a Designation

United States (US$ per annum)

Indiab (US$ per annum)

Programmer System analyst Programmer analyst Network administrator Database administrator Help-desk support technician Software developer

32500–39000 46000–57500 39000–50000 36000–55000 54000–67500 25000–35500 49000–67500

2200–2900 8200–10700 5400–7000 15700–19200 15700–19200 5400–7000 15700–19200

a Figures are starting salaries for large establishments employing more than 50 software professional’s. Salaries for a particular designation would vary due to factors such as educational and experience profile of the professional, platform of operation, nature of assignment (contract/full-time) location of the employer and additional technical/professional certification. Source: INFAC (1998), Mumbai. b Converted at exchange rate of Rs. 41.50/US$.

estimated wage costs in India were about 1/3rd to 1/5th of the corresponding US levels for comparable work. Other estimates suggest that once all costs are factored in, the cost of software development is only half that in the US. Despite the large stock of engineering graduates in India, NASSCOM claims that by year 2000, demand will outstrip supply. If this were to happen, the present rates of growth could not be sustained. We believe that although investments in engineering education are necessary, a bigger part of the solution lies in a more efficient use of existing human capital resources. The tight labour market conditions are reflected in the 20% increase in wages and in attrition rates that are said to be nearly 20–25% for the industry. 18 We believe that demand already exceeds supply for experienced project managers. Many of our interviewees mentioned their difficulty in recruiting professionals with 4–6 years experience, even though nearly 40% of the workforce is reported to have 4–6 years experience. Experienced software professionals tend to move to the US through the H-1B visa route, adding

18

India graduates about 155,000 engineers of various sorts, and another 200,000 diploma holders from a variety of private training institutes per year. About 60,000 of these enter the IT sector. In 1999, the total number of software professionals in India was estimated to be about 250,000 compared with 200,000 in the previous year, with a median age of 26.2 years (NASSCOM, 1999, p. 16). Virtually all of these professionals have either engineering or computer science degree or a Masters in Computer Applications (MCA).

to the problem and the entry of new firms exacerbates the situation (Nidumolu and Goodman, 1993). 19 To a considerable extent, this reflects the relatively inelastic supply of engineers (as compared to students trained by private training institutes) and the marked preference for engineers (of all types, not just software engineers or computer scientists) by Indian software firms. Responses to our questionnaire survey indicate that 80% of the software professionals employed had engineering degrees, while 12% only had diplomas from private training institutes. Few of the firms we interviewed admitted to hiring graduates from private training institutes, although since the bulk of the engineering graduates are not trained in software engineers or computer science, a substantial fraction also have diplomas from private training institutes. Further some of the smaller and younger firms do hire graduates from these institutions, as do some domestic market focused firms. This preference from engineers is even more puzzling since the bulk of the work is relatively non-technical and requires mostly logical and methodical work and a familiarity with software development tools and languages. As one CEO we interviewed put it “Take somebody from a good college (any of the top 20 colleges in India), give him 3 months of orientation and they are ready to take up a programming assignment. I don’t need all these engineers. . . . But I 19

The US Consulate India estimates that around 30,000 H1-B visas were granted in 1999, mostly for software professionals (The Economic Times, 1999).

A. Arora et al. / Research Policy 30 (2001) 1267–1287

don’t want to be branded by my customers as a guy who hires NIIT graduates.” (Emphases added) Given the large number of science and arts graduates, and the widespread availability of private training, the pool of potential software professionals is much larger than merely the engineering graduates. However, software firms are reluctant to tap this pool because of the potential negative signals to their customers. 20 This is a clear instance of a “race to the top” rather than a race to the bottom. With only limited market power, Indian software exporters try to distinguish themselves from the competition by pointing to the quality of their processes and people, and when possible, their experience. Firms also have quality concerns. Managers we interviewed believe that an engineering education imparts a set of problem solving skills, methods of thinking logically and learning tools that help quick adaptation to changes in technology, domains and tasks. Since Indian firms provide services across a range of platforms and domains, this is an important asset. The CEO of a small firm developing innovative products stated Because things are changing so fast in this industry, knowledge of a particular operating system, a particular language, a particular technology is not as important as the ability to learn and adapt to change. Despite paying substantially above Indian standards, virtually all firms find it difficult to retain talented professionals. All the firms we interviewed and a very large fraction of the over one hundred firms we surveyed mentions employee turnover and difficulty in attracting suitable employees as a major problem (Table 15). High rates of employee turnover constitute one of the most important challenges to the ability of Indian firms to progress beyond providing low-end software coding, development and maintenance services. The loss of employees is especially severe among consultants (programmers and analysts) sent overseas to work onsite in the US, and a very substantial fraction quit within 2 years. 21 The situation is 20

US visa restrictions are another reason why firms prefer engineering graduates. 21 However, high turnover is also common in US software developers and in other service industries such as accountancy firms, and as the CEO of a rapidly growing software services firm put it-“paying a premium price for talented people. This is a world-wide phenomenon.. . . I am not worried”.

1279

Table 15 Major problems for Indian software firmsa Problem

Export

Domestic

Manpower shortage/skills Employee attrition Physical infrastructure Commercial infrastructure Quality certification Visas Finance/capital Marketing access Lack of domestic computerization Lack of government support Tarrifs and other barriers

57 44 12 24 11 33 20 42 6 10 11

32 27 12 17 6 NA 14 17 21 11 8

a The firms were asked to indicate their top three problems. Source: CMU Software Dataset – CMU Survey of Indian Software Industry, N = 104.

particularly acute for experienced project managers, typically software professionals with 4–6 years of experience, and for talented software designers. Virtually all the firms we interviewed noted the difficulty they have in retaining experienced software professionals, many of whom are lost to firms overseas. Firms are responding to the problem of employee attrition in a number of ways. One popular way is by providing opportunities to work in the US. Many managers also stressed quality of life issues. Some firms stressed their ability to provide a career path for their employees, wherein they could move to being managers and would not have to remain programmers, apparently something valued by Indian professionals. Interestingly enough, few claimed to pay more than their competitors although a number of firms were actively considering stock options, something that few firms in India had done by 1997, with Infosys being an important exception. Some firms expressed the need to make the organisation individual independent by addressing the loss of knowledge due to employee turnover. Thus, one CEO told us that although, Our attrition is . . . better than average. . . . it is difficult to retain the segment (of the workforce with) 6–8 years experience. We are tackling this by making (our) organisation individual independent, and stressing training & process methodologies. Our sense, however, was that efforts at insulating the organisation from employee turnover are very much

1280

A. Arora et al. / Research Policy 30 (2001) 1267–1287

Table 16 Revenue per employee, by quality certification: CMM level 3 in year 1999a Year

1995 1996 1997 1999 a

Non-CMM firms

CMM certified firms

Sample size

Mean revenue per employee, in US$ 1000

Sample size

Mean revenue per employee, in US$ 1000

38 46 46 46

21.31 22.80 20.13 23.71

17 18 18 18

27.51 32.11 29.14 35.82

t-statistic for difference of means

1.24 1.89 1.79 2.74

Firm sample does not include subsidiary firms like Motorola, Texas Instruments etc. Source: CMU Software Dataset.

Table 17 Revenue per employee, by size of firm in year 1997a Year

1995 1996 1997 1999 a

Small firms (250 empl.)

Sample size

Mean revenue per employee, in US$ 1000

Sample size

Mean revenue per employee, in US$ 1000

31 39 39 39

19.01 22.19 19.85 22.89

24 25 25 25

28.67 30.46 27.05 33.71

T statistic for difference of means

1.96 1.84 1.61 2.28

CMU Software Dataset.

in their infancy, and firms are still very vulnerable to the problem of retaining experienced and talented workers. In addition to improving retention rates, an important part of the solution is a more efficient use of the available human capital. As noted earlier, there are a large number of non-engineering graduates that are trained by private training firms such as NIIT and Aptech. Indeed, as Tables 16–18 shows, the training sector industry has grown along with the software industry, with total revenues in 1997–98 estimated at Rs. 8.56 billion (about US$ 225 million), up from Rs. 6.6 billion the year before. Even if one accepts the dubi-

ous proposition that such graduates are not well suited for software development, they are well suited for providing support and maintenance, for back office operations, and for a variety of IT enabled services such as medical transcription and claims processing for the insurance industry. Moreover, the current wage trends will provide incentives for firms to use engineering talent more efficiently and to substitute software professionals with other backgrounds where possible. Although the supply of engineering graduates is relatively inelastic in the short-run, there are a number of public sector and industry initiatives to increase the supply of software professionals. The government has

Table 18 Revenue per employee, by age of firms in 1999a Year

1995 1996 1997 1999 a

Young firms (9 years)

Sample size

Mean revenue per employee, in US$ 1000

Sample size

Mean revenue per employee, in US$ 1000

25 30 30 30

24.81 26.75 22.18 28.35

30 34 34 34

21.91 24.25 23.09 26.03

CMU Software Dataset.

T statistic for difference of means

−0.52 −0.55 0.21 −0.51

A. Arora et al. / Research Policy 30 (2001) 1267–1287

recently announced the establishment of Indian Institutes of Information Technology, along the lines of the well-known Indian Institutes of Technology. A number of engineering colleges have increased their emphasis on information technology and, in some cases, have started IT management programmes. A potentially serious constraint on the ability to rapidly increase the output of trained software and computer engineers is the shortage of engineering doctorates being awarded in India. Recent data show that the PhDs awarded in engineering disciplines have fallen from their high of 675 in 1987 to 375 in 1995. Concurrently, the number of engineers with postgraduate training has also risen only slowly, from a little over 12,000 in 1987–89 to a little over 17,000 in 1990–92. Surveys of the IITs shows that a very large fraction of postgraduates enter the IT sector, in some cases as many as 90%! Although there are a substantial number of engineering doctorate holders of Indian origin working outside India, a long run solution will require an increase in the postgraduate research and training infrastructure. 5.2. Communication and physical infrastructure Good communication infrastructure is considered vital for the continued growth of the industry. This is most obviously the case in software service exports, and especially for offshore software development. Overall, the data communication infrastructure in India is expensive and in limited supply. Some US based clients do find the slowness of the networks and the general problem with the communication infrastructure a major irritant. 22 However, as with finance, most firms consider the problem of communication infrastructure to be less important than the problem of finding and retaining qualified software professionals, and the problem of physical infrastructure such as roads and power. Indeed, most Indian firms seemed quite content with what was available. Two inferences, quite different in implication, are possible. The first is that communication infrastructure is adequate, or that firms have found ways around the high price and low availability constraint. The other is that the offshore component of the tasks being 22

The managers at one of the largest US software firms said that the connectivity to India was poor and the networks were very slow, making onsite presence almost a requirement.

1281

carried out in most Indian firms requires only limited communication so that the existing bandwidth is adequate. Consistent with the latter interpretation is that export projects are simple and small and reasonably easy to specify in detail in advance. 23 Clearly, the poor communication infrastructure has affected the diffusion of the Internet domestically, and through that, has discouraged the growth of new firms that could provide software services for and through the Internet. 5.3. Finance Software services, especially for export, are a very profitable business with good cash flows and limited requirements for up front investment. Therefore, finance is not a major problem for software service firms, unless a firm wishes to expand rapidly or wishes to expand overseas. More than half of the firms we surveyed indicated that they relied upon personal funds for start-up finance. Many of the firms we interviewed appeared to rely on equity financing as the primary source of capital, although they had other diversified sources of finance such as loans and lease finance. Others relied upon financing from their parent firm or from business groups with which they were affiliated. Obtaining finance is, however, a major concern for firms developing software products. In contrast to services, a substantial investment is required to develop the product, and even more to market the product. Firms that are trying to develop software products have faced problems in getting finance, in part because the inexperience and conservatism of Indian venture capital funds. 24 A common strategy has been to provide 23 In late 1998, the government of India announced that a high bandwidth (2.5 GB) fibre-optic backbone would be set up. This may open up many new possibilities, including a significant increase in the offshore component of software exports. 24 A manager for a venture fund run by a public sector investment bank noted that, although they had invested in as many as 32 firms in the past, only five were product focused, the rest being software service firms focusing on exports. However, this venture capital operation no longer invests in start-ups. Instead, they fund expansion or new product development by existing firms with good track records, or by proven star managers in the industry. Funding investments are in the range of Rs. 15–20 million (US$ 400,000–500,000) but the firm can come back for more as soon as the financing is exhausted.

1282

A. Arora et al. / Research Policy 30 (2001) 1267–1287

software services and use the cash generated to fund product development. An interesting example is a firm started by an entrepreneur who had worked for many years in a large US software firm who was financing the venture himself, but was also using supplying software services to his earlier employer to generate revenue. A firm affiliated with a large banking group also used its earnings from its services business to fund product development. A venture capitalist affiliated with a well-known Silicon Valley venture capital firm agreed with the view that venture financing in India for start-ups focusing on packaged products is limited. He indicated that the existing tax laws made it more profitable to make equity investments in a US firm rather than an Indian firm but to then have a wholly owned Indian subsidiary to make its financial reports in India and get tax exemptions. He also noted that debt rather than equity fund most start-ups. Therefore, having a service component to a product strategy made sense since services provide a relatively easy way of entering the market and gaining experience. Services also provide cash, which start-ups are unable to get from elsewhere. The problem, it appears, is as much on the demand side as on the supply of venture capital. Venture capital funds associated with well-known Silicon Valley venture capitalists, such as Draper International, have only been able to use 60% of the allocated funds. In addition to the usual problems involved in setting up businesses in India, venture capitalists are interested primarily in products developed for large markets, and therefore, for products that can succeed in the US. Developing products for the US market from India is widely thought to be very difficult. Table 15 shows that it does not appear that lack of venture capital is the major constraint for developing software products. Moreover, as discussed later, in recent years a number of firms focused on product development, or on internet and e-commerce have emerged, and alongside, there has been a concomitant increase in the supply of venture capital and venture capital firms. In 1998, about US$ 1.3 billion had been committed in venture capital and following policy changes in 1998, a number of firms, included overseas funds, have entered the market. The number of venture capital funds registered with SEBI increased from 8 in 1996–1998

to 14 in 1999–2000. 25 A number of prominent expatriate Indians are also increasingly important as angel investors, and as links between Indian firms and US based venture capital funds. The actual amount of venture capital and angel capital invested in the IT sector in India has increased from US$ 20 million in 1996 to over US$ 300 million in 2000. Though very small in magnitude, it does point to the growing demand for this type of risk capital in India.

6. Moving beyond low wage costs Industry leaders, academics and government officials alike have stressed that software firms need to provide more value-added services — to move up the value chain to ward off the growing wage pressures at home and the potential competition from China, eastern Europe and others. Many observers have also argued that the prospect of short term gain offered by the simple service exports described in Sections 3 and 4 are seducing Indian firms away from making the substantial investments required to develop products, or even to move up the value chain. Further, they argue that the absence of a sophisticated domestic market further exacerbates the problem (e.g. Heeks, 1996; D’Costa, 1998). Finally, they argue that this, coupled with the inevitable shortage of engineers, constitutes a serious impediment to the sustained growth of the Indian software industry. Although there is an important element of truth to this observation, our sense is that this is overly pessimistic. First, as noted earlier, a considerable fraction of the current services supplied by the Indian software industry does not require an engineering background and, consequently, suitably trained non-engineers, whose supply is far more elastic, can be substituted. Second, overseas firms can supply some of the required technology and managerial expertise. Indeed, a number of them are setting up research and product development centres in India, typically headed by managers of Indian origin. Finally, there is evidence, albeit anecdotal, that Indian firms are responding to the rising wage pressures by improving software development practice and developing tools to reduce 25 http://www.nasscom.org/template/Venture%20Capital1.htm downloaded on 10 August 2000.

A. Arora et al. / Research Policy 30 (2001) 1267–1287

1283

costs, and providing higher value services beyond simple coding. As well, some firms are investing in research and in developing new products.

success story is that of Pentamedia, a spin-off of from a traditional software service exporter, in animation software and related technologies.

6.1. Technology

6.2. Maturity of software process and improved quality

A brief look at the evolution of software services in India shows that firms were very quick to migrate to newer platforms and adapt to new technologies pertaining to software development. From the traditional mainframe based systems, firms quickly moved to open system platforms like UNIX and Windows. The skill level of firms in the open systems is considered to by customers to be on par with that of any US based firm. Relatively few firms have invested in R&D. As long as firms primarily provided coding services to foreign customers, the need for R&D was minimal as there was no new technology being developed and the services were mostly application solutions. In the past few years, the R&D is largely the province of firms focused on semiconductor intellectual property (IP) or the larger firms like TCS. More recently, some indigenous firms have begun to pay attention to technology development as well. Wipro has been developing semiconductor IP and had recently started a subsidiary, EnThink, to market the semiconductor designs developed in Bangalore. Another firm, Silicon Automation Systems has been developing a range of speech compression products and design tools for semiconductor design. Newgen Tech., a New Delhi based company, licensed their document management technology to Canon for 5 years for US$ 2 million (Computers Today, March 1998). Infosys built a CNC machine controller (Gamana) which they, consequently, sold. Infosys also invested US$ 3 million in CiDRA Corp., which manufactures photonic devices for high precision wavelength management and control of next-generation optical networks. TCS has a research division in Pune, where research in different areas of computer science and multimedia is carried out. Graycell technologies (now Unimobile) is probably the best example of a technology focused firm. Graycell develops software for mobile devices and its technology was attracted well known angel and venture capital investors from the US, who, subsequently, moved the company to California, much like many technology based Israeli software firms. Another

Most firms have responded to wage pressures and employee turnover by focusing on improving their software development process and developing tools. Since the early 90’s Indian software firms started to get their software processes certified according the ISO 9001 standard or got assessed according to the Capability Maturity Model (CMM). Indeed, over half of the CMM level 5 firms in the world are in India. Although quality certification is often to signal to potential clients rather than to improve the quality of software produced, there is growing evidence that firms are in fact developing software development methods and project management skills to be able to undertake larger and more complex projects. Firms certified at CMM level 4 and 5 are better able to predict time and costs involved in such projects, and thus better able to bid for fix priced contracts (as opposed to the more common time and material contracts). This ability also complements other skills important for moving up the value chain, such as requirements analysis and high level design. TCS has executed a number of large export projects and moved into newer areas like data warehousing which involves a closer link to the customer’s business objectives. Long term relationships between Indian firm and customers seem to be typical of most export projects. Of the firms we surveyed, over 93% said that their most important export contract involved work for a company they knew earlier or was part of an ongoing relationship with the client. The fact that 32% of the firms surveyed also said that they had a penalty clause in their contract points out that firms are getting increasingly mature and are able to make projects more predictable. Some of the leading firms have adopted a strategy of developing tools that embody some of the business and technical knowledge accumulated by the firm in the course of serving overseas clients. Not only do they lower costs and improve delivery time, they also help partially insulate the firm from employee

1284

A. Arora et al. / Research Policy 30 (2001) 1267–1287

turnover. Nortel Networks, a big long-term customer for the large software firms like TCS, Infosys, Wipro and Silicon Automation Systems, recently announced that they would share royalties with these companies for the tools and products developed. 6.3. Moving up the value chain: solutions & products Developing tools is one way of reducing costs. A complementary strategy is to try to “move up the value chain”, by providing services beyond simple programming services, intensive in industry specific business knowledge (domain expertise) and technical capability. Together, these enable the software supplier to provide “solutions” to business problems, rather than simply programming services to implement solutions that the customer or firms such as Anderson Consulting and Oracle provide. 26 Documenting the acquisition of such capabilities is not easy, however. There is ample anecdotal evidence that firms such as Wipro, Infosys, Mastek, NIIT, Satyam and TCS are already offering solutions in areas such as e-commerce and telecommunications. 27 26 However, some believe that such a strategy, which ultimately involves getting involved in business process re-engineering for overseas clients, is not feasible for Indian firms. The CEO of one such firm, a wholly owed subsidiary of a very large Indian engineering firm, believes that it is better remain a subcontractor to the established systems integration and IT consulting firms in the West.

“It is much better to piggy back on the Big Six who have large organisations, credibility and have done this (business process re-engineering) for years. Better for us to do level two work on logical and physical systems . . . maybe the implementation part but onsite. . . (Our parent firm) has ES 9000s and IBM mainframes. It was the first firm to use IBM mainframes in India for a very long time. . . . We have the most qualified experts on IBM mainframes. So as far as legacy maintenance on IBM mainframes is considered, we know the technology inside out.. . . (but) technology is not such a critical factor as compared to understanding business practices.” 27 One type of service that is considered to be moderately value-added but quite specialised is the writing of software for specific hardware components, such as printers, cell phones, and modems. Device drivers, as they are called, involve a varied level of sophistication as far as programming talent is concerned. Wipro, BFL software, Silicon Automation Systems and Srujana Technologies are among the firms that provide device driver related services to clients.

Sometimes, these solutions are based on products. For instance, TCS and Infosys have launched banking products in the Commonwealth countries and have had some success. CITIL (now I-Flex) has a banking product Flexcube, which they have just started to market throughout the western world after testing it in the eastern markets. The product strategy behind the larger companies is to build products out of services that they have provided to the past customers. In many cases, packages or tools built for the customers have to be modified slightly to be more applicable to market needs. This seems a viable strategy in the years to come, considering that Indian firms would continue to provide services for the export market. Despite the lack of success for most firms that focused on software products in the early 1990s, a few product focused firms have soldiered on. Ramco has offered an enterprise resource planning (ERP) product for some time, but are only now making inroads in the export market. In the past couple of years, there have been new entrants in the ERP market like Vishesh Infosystems and Nexgen Technologies. Nonetheless, most products being sold or developed today are extensions of the packages that the firms developed for their earlier customers. In the last couple of years, the entrance of a new set of entrepreneurs, either home-grown or US returned, has led to the resurgence of product oriented firms, seeking to exploit lower development costs. Improved communications and links with overseas markets reduces the disadvantage of being far away from customers, helping developers in India keep in touch with the latest trends in the markets overseas and keep tabs on their competitors and their product releases. 28 In the mid 1990s, when concerns about the future of Indian software exports were first raised, many exhorted software service firms to develop products, and some product developers used service exports to fund product development. But managing services and product development in the same firm has proved to be challenging, and many firms have separated their product development divisions from their software 28

Examples of products being developed include Smartifacts with WebVU and Ruksun Technologies with LiveJam, which allows music artists to “jam” across the Internet, and EasyDiary, which is an online calendar program similar to those offered by Internet portals like Yahoo.

A. Arora et al. / Research Policy 30 (2001) 1267–1287

operations, often by spinning them off as stand alone companies. This also suggests that one should look to later entrants, not the established software services firms, for successful product development. 29 6.4. Assessing success: revenue per employee One measure, albeit an imperfect one, of measuring the effectiveness of these strategies, is revenue per employee. This commonly used measure of labour productivity is particularly apt for comparing firms that are labour intensive and where there is no reason to suspect significant economies of scale from large fixed assets. In the Indian context, this measure has a significant problem in that firms that rely heavily on onsite work and body-shopping will tend to have higher revenue per employee, even though they are lower in the value chain and less productive. Tables 16–18 compare revenue per employee for a group of firms registered with NASSCOM for which have consistent data over several years. To facilitate comparisons we removed divisions and wholly owned subsidiaries of overseas firms (such as Texas Instruments and Motorola) and also excluded product focussed firms and domestic oriented firms. Table 16 shows that firms that are quality certified at CMM level 3 or higher have higher revenue per employee than non-certified firms and the statistical significance of the difference tends to be high. Table 17 shows that large firms (with more than 250 employees in 1997) similarly tend to have higher revenue per employee. Further, the absolute difference in revenues per employee between CMM and non-CMM certified firms, as well as larger and smaller firms has tended to grow over time. However, older firms (that entered the industry before 1990) do not have significantly higher revenue per employee. 29

One of the earliest instances was Yantra, till recently a US subsidiary of Infosys Technologies (which still owns a considerable stake in the company). Yantra, which develops supply-chain management software, got US venture capital funding and appears to be quite successful. Since then, Infosys has also spun off Onscan Inc., a wireless Internet solutions company that it had incubated. Another relatively well known case is of Talisma, a front office product from Aditi Technologies. This product was developed as an extension of the in-house tool that they used to provide e-mail support to customers. Talisma was recently hived off from Aditi as a separate entity.

1285

It is extremely unlikely that CMM level 3 and higher firms and larger firms have a higher fraction of onsite and body-shopping projects. It is more likely that the higher revenue per employee reflects higher value added. In a related study (Arora and Asundi, 1999) we explored the economic returns to ISO 9001 software quality certification. The empirical results show that firms that did get quality certified do not get better rates for their services. Quality certification, however, allowed the firms to take on larger projects and grow. Further, the results indicate that increased size enabled firms to increase revenue per employee. Our interviews also suggest that the ability to take on larger, more complex projects and manage projects better is an important type of value addition strategy for software firms. As well, many of the larger firm have developed industry specific solutions that yield higher revenues per employee. Finally, the absence of a substantial difference between older and younger firms indicates that this ability is not simply a matter of learning by doing but rather requires deliberate investment by the firm. Further, it points to the relative immaturity of the industry as a whole in that early entrants do not appear to have a significant advantage over later entrants. On the whole, these results do not support an overly pessimistic view of the future. Earnings per employee have increased, particularly for the more established firms and those that have invested in developing higher quality software development practices. Further, this growth has taken place even as these firms have decreased the extent of onsite work, implying a greater increase in value addition than reflected in the figures in revenues per employee.

7. Summary and conclusions The picture one gets of the Indian software services industry is a mixed one. On the one hand, there is a great deal of excitement about its rapid growth and its export success. On the other hand, the kind of work being performed is typically fairly mundane. The domestic market provides far more challenging projects but is financially less rewarding. Moreover, the links between the domestic market and exports is, at present, very tenuous. There is little evidence for the idea that experience with complex domestic projects has had a high payoff in the export market or that the “learning

1286

A. Arora et al. / Research Policy 30 (2001) 1267–1287

to walk on two legs entry strategy”(Schware, 1992) was being practised by Indian firms. 30 These observations are consistent with the other research in this area (e.g. Heeks, 1996; D’Costa, 1998). Both of these authors have argued that the export orientation and routine tasks that exports involve have limited learning potential for Indian firms. Heeks has noted that the reported export revenues are misleading insofar as a substantial fraction of these are accounted for by onsite work, that in turn requires that a substantial fraction of the revenues be spent outside India. D’Costa has highlighted the limited potential for leapfrogging of most export-oriented activities in which Indian software firms are currently engaged. In out interviews, US clients convey the sense that their Indian suppliers are competent (by and large) at providing a limited range of services but only a few have moved to where they offer solutions to client problems. In the last couple of years, there is evidence that the established Indian firms are maturing and growing in their ability to execute larger and more complex projects, as well as execute higher value added parts of such projects (such as requirement specification and high level design). These firms can act as the nucleus around which the industry can develop and mature. We also see a entry of new entrepreneurs that are experimenting new product ideas. The strategies that firms are adopting to climb the value chain differs but can be broadly divided into: (a) developing mature software development processes to take on larger, more complex, and higher paying projects and effectively do more value added work; (b) understanding business needs of the customer and proactively developing business and technical solutions to existing problems; (c) developing products from the services provided earlier (“productized” services) and (d) entering emerging technologies and businesses such as multimedia and e-commerce. Executing these strategies is not easy. Firms have to invest a great deal in hiring, training and retaining their employees, in expanding overseas and establishing (or acquiring) subsidiaries in the US and Western Europe, as well as in acquiring the technological and business 30

Many firms that began with a domestic market focus seem to have moved away towards less challenging but more lucrative export tasks.

expertise needed. Furthermore, such investments will have to be made for long periods, possibly well in advance of any return on the investments. As noted above, thus far Indian firms have been able to enjoy high profits without having to expose themselves to such risk. Thus, although most of the established firms have employed these strategies, the effectiveness with which they have deployed these strategies has varied. However, our optimism about the beneficial impact of the Indian software industry on the Indian economy in the long run is not based entirely on the quantitative importance of the relatively smaller number of successes among software service exporters. We think that in the shadow of the much more prominent software services firms, we are finding firms developing a variety of new software products, components and technologies. Further, the software service firms are exemplars of organisational forms and practices that are relatively new to India. A large number of software firms are de novo start-ups, indicating that the supply of entrepreneurial talent appears to be forthcoming when the opportunity arises, even in new and technology intensive sectors. These software firms are relatively flat organisations, with young management teams, informal but professional management styles, and with an emphasis on efficiency, punctuality and other virtues that an export orientation brings. Top managers of the leading software firms have been profiled in the popular press in India and are viewed favourably by many Indians, particularly in comparison to traditional Indian business leaders. Further, this industry has pioneered equity stakes and stock options for employees in India, and many of these companies are star performers on the Indian stock market. Thus, unlike in the past, the fruits of the success of the industry have been shared far more broadly. The implications of the success of this industry, at a time of slow but far ranging changes in the Indian economy, can be immense and far-reaching.

Acknowledgements We thank the Alfred P. Sloan Foundation for supporting this research. We are grateful to the many people in the software industry, both in India and the United States, who have given generously of their time and expertise. We have benefited from the help

A. Arora et al. / Research Policy 30 (2001) 1267–1287

and advice from many colleagues, especially Ashok Desai, Naushad Forbes, Raj Reddy and Bankim Shah, and seminar participants at the NBER, Boston and SPRU, Sussex. Seema Chawla provided helpful suggestions for the research. Zhaoli Rong, Hairong He and Jason Wang provided competent research assistance. To preserve confidentiality, firms and respondents have not been identified. We welcome comments and suggestions. References Arora, A., Asundi, J., 1999. Quality certification and the economics of contract software development: a study of the Indian software service companies, Heinz School of Engineering and Public Policy, Carnegie Mellon University, Pittsburgh, unpublished manuscript. Bannerjee, A., Duflo, E., 1998. Reputation effects and the limits of contracting: a study of the Indian software industry, MIT, Cambridge, MA, unpublished paper. D’Costa, 1998. Technology leapfrogging: software industry in India. In: The 2nd International Conference on Technology

1287

Policy and Innovation, Calouste Gulbenkian Foundation, Lisbon, 3–5 August 1998. Gopal, A., 1996. An empirical analysis of offshore software development: a first look at some explanatory factors, Communications of the ACM, submitted for publication. Heeks, R., 1996. India’s Software Industry: State Policy, Liberalisation and Industrial Development. Sage Publications, Beverly Hills, CA. ITAA, 1998. Help wanted 1998: a call for collaborative action for the new millennium, Report published by Information Technology Association of America and Virginia Polytechnic and State University, March, Washington, DC. Nidumolu, S.R., Goodman, S.E., 1993. Computing in India: an Asian elephant learning to dance, in Communications of the ACM, 36(4). Royce, W.W., 1970. Managing the development of large software systems: concepts and techniques, 1970 WESCON technical papers, Vol. 14, p. 723. Schware, R., 1992. Software industry entry strategies for developing countries: a “walking on two legs” proposition, World Development, Vol. 20, No. 2. Pergamon Press, Oxford, pp. 143–164. Wipro Launches University Degree Verification in The Economic Times, Hyderabad Edition, 30 June 1999. Udell, J., 1993. India’s Software Edge, Byte, pp. 55–60.