Pakistan Inc. -- The IT Industry Edition

Monday, July 04, 2005

IT's Lost Decade: Does Pakistan of Today Mirrors India of 1995? -- II

IT's Lost Decade: Does Pakistan of Today Mirrors India of 1995? -- II
By Athar Osama
An abridged version of this article series was published in Dawn, EBR, July 4, 2004

In this first of this series of two articles, I outlined the primary motivation behind undertaking this analysis. I also laid some ground rules and identified sources for comparing the early performance of India's software/IT industry with that of Pakistan today. I then went on to trace the beginnings of the Indian software "miracle". Finally, I attempted to introduce some statistics that are available through various sources, to develop a more concrete sense of the state and dynamics of the Indian software industry up until mid-1990s. As I did that analysis, I also briefly outlined and discussed major similarities and differences between the Indian software industry and that of Pakistan. In this article, I would continue my examination of the Indian software industry as it stood in 1994/95 with some additional insights on its qualitative features. Then I would turn to some analysis and prescription for other countries aspiring to do the same.

A Brief Recap

In the way of a brief recap of the first article, I discussed the early days of the Indian software industry as it rose from the shadow of a much larger and mature hardware industry. The article also noted a curious contradiction in early Indian policy towards software as it linked permissions to import computer hardware with future commitments to export software. Not only did this result in half-hearted attempts to export software--the earliest of Indian software companies were not really in it for the exports, per se, but rather for the ease with which they could import computer hardware in the name of writing exportable software--but also much of the imported hardware actually ended up being leased out to the domestic market.

As the Indian software industry moved ahead, however, the domestic market for software created its own demand. IBM's departure from India in 1978 provided 1,200 "surplus" employees and was partially responsible for jumpstarting an entrepreneurial software Industry in India. The growth of the Indian software industry--increasingly outward-looking in its outlook--in the early days, however, was not without stops and jerks. For individual companies it was often a roller coaster ride with huge swings in annual revenues. While the gross annual export revenues for the entire industry grew from $4million in 1980 to $480 million ($168 million net of onsite expenses) in 1994/95, it had repeatedly missed the government target of $1 billion mark. Thus in 1994/95--the reference year of our analysis--the Indian software industry stood at a much stronger position from its humble beginnings in 1974, albeit (perhaps) with little real sense of the dramatic transformation that was to happen in the following decade.

Product-Service Offering and Client Mix

Up until the mid-1990s, the Indian software industry mainly relied on export of software services--custom software development (conversion and application building) work (~90%)--rather than software products (<>Firm and Industry Structure

In the early days of the Indian software industry, we also observe over-reliance on a single (or a small set of) client(s). Between 1989/90 and 1994/95, eighteen of the top-twenty five software firms in India did majority of their work for a single foreign client. This over-reliance on a small set of clients accentuated the volatility of the companies' performance. For example, barring the top-3 companies in India, all of the companies that occupied the rest of the slots in the top-10 list had slipped down and off the list or gone out of business.

We find many of the above organizational characteristics prevailing in Pakistan's software Industry today. The over-reliance on a small set of clients has been a major bottleneck in the growth and diversification of the industry. Barring some exceptions, companies have resisted in investing in systematic marketing and brand-development with the result that they have failed to diversify their client base. Even the largest companies do not service more than a handful of clients and are dependent upon them for their continued viability. According to the PSEB study on Pakistan's Software Industry, companies, on average, derived as much as 50% of their revenues from a single client and between 70-90% from 5-largest clients. Although the trend was less extreme for product companies (generally operating in the domestic market) but it points in that direction, nonetheless.

The Indian software industry was also a very concentrated one. For a very long time, Tata Consultancy Services and Tata Uniysis Limited dominated the market share for Indian software exports. The two companies had the combined market share of upto 73% in early 1980s that only gradually decreased to 25% in 1994/95. According to DoE and Dataquest surveys, as late as 1989, 80% of the Indian software exporters had export revenues less than $50,000 per annum.

While Pakistani software Industry is not as concentrated as that of India's of the early days--primarily because there is no one leader that has made it big on the scale that Tatas did--but it too does have a very long tale. According to the above referenced PSEB survey, only 4 of the 60 largest software operations in Pakistan reported revenues greater than $5 million, although 50% of the companies surveyed had revenues in excess of a half-a-million dollars. While the dollar-figures are not comparable due to temporal and inflationary differences, the degree of concentration in the Pakistani industry of today is considerably less than the Indian industry of 1990s--a fact further amplified by the sheer number of companies in India (>1200 in 1994/95) vs. Pakistan (~3-400 in 2004, per PSEB/PASHA estimates).

HR and Manpower Situation

Numbers of software manpower in India are a little sketchy and of less credibility. Firms are reported to exaggerate manpower numbers to present a rosier picture of their capability than was actually true. Heeks, therefore, asks his readers to treat these numbers with caution. He puts together a set of "rough" figures gathered from a whole variety of government and private-sector sources as well as his interviews with industry executives prior to 1994/95. These data put the size of Indian software labor force at 15,000 in 1990 (around 6,500 of whom were engaged in export work) and around 50,000 in 1995 (27,500 of whom were engaged in export work). Other estimates suggest that, in addition to the above pure software work, as many as 150,000 workers were involved in data entry and processing work of various kinds around 1995. These latter figures, however, seem less credible than the pure software development figures as Heeks himself earlier suggested that the outsourcing/BPO type work at that time was nowhere as prevalent as these manpower figures suggest.

Compared to these figures, Pakistan's software industry appears to be relatively small in terms of manpower and employment. The PSEB Best Practices study puts the total employment in 60 of the most prominent (and larger) firms that it surveyed at around 5000 people. Another internal PSEB study whose results were shared with this author put the entire software industry size at around 8000 people. Even if one adds a tentative number of another 10,000 or so people who are involved in various kinds of ITES/BPO operations ( e.g. call-centers) in the mix, the total industry size is certainly not more than 20-25,000 people. This makes the Pakistani software industry today around half to a third of the size of India's software industry in 1995. While it would be a mistake to read too much into these figures--primarily because of their suspect credibility and lack of rigor--the broader trend is unmistakable.

The Indian software industry of that time, not very unlike Pakistan's software industry of today, suffered from high attrition and turnover rates. Software companies routinely lost 15-20% of their staff each year with some reporting an attrition rate as high as 50% in a single year. According to Heeks' estimates 10-15% of those who changed jobs went to work or study abroad. This was in addition to the regular rotation of employees that was common due to the practice of bodyshopping. According to one estimate, of the 20,000 people engaged in software exports in 1995, around 10,500 or so had worked abroad at some point during the year. Many of the people who worked on clients' sites, at one point or the other, broke off and went to work for the client.

Despite this healthy churn of employees, we still find complaints about difficulties in obtaining visas for Indian workers to work abroad. One interesting, yet confusing, anomaly that we find is the very little mention of India as the modern day manpower behemoth that it is widely acknowledged today. Heeks, for example, does not substantively talk about the role of IITs (Indian Institute of Technology) in his entire book. Infact, the word IIT is not even mentioned in the index of subjects. While there is no doubt today, about the role of IITs in the making of the Indian software miracle and much of that apparatus was already in place during the timeframe of interest for this study, this lack of interest and mention of IITs suggests that perhaps their role received national and international prominence and recognition after 1995.

Quality and Productivity

Worker quality and productivity has often been an issue of much anguish in Pakistan. The shortage of middle management, people with project management experience, and low-quality of manpower churned out by mom-and-pop IT schools (and even some chartered universities) have often been cited as reasons for lack of quality and growth in the Pakistani industry. Heeks study provides some, albeit tentative, insight into productivity and quality of Indian software workforce and industry. Some of these figures, however, are quite unreliable. For example, Tata Consultancy Services (TCS) had a productivity of around $18,000 per employee all through the 1980-94/95 period, yet it is suddenly reported to have shot up 25% (to $24,000/person) in just one year thereafter (p. 98). Despite these discrepancies, however, there are some discernable patterns. For example, productivity ($/person) of onsite work is twice as that of offshore work, of export work is two-to-three times as large as domestic work, and for multinational subsidiaries is higher than that of locally owned operations.

For Indian software developers working onsite, their personal productivity (i.e. lines/programmer) is as high (or even higher) than their western counterparts. The programmers working offshore, however, seem to lag in terms of productivity than either onsite or foreign programmers. Many of these productivity differences seem to be caused by lack of resources (capital per programmer) and managerial skills (p. 100). Heeks also talk about the overall quality culture in Indian companies and the practice of seeking quality certifications.

Very much like Pakistan which actually seems to be following and Indian model in this respect, an observer talks about "India being struck by...'ISO 9000 fever'...from 1993 onwards" ( p.111). By 1996, 30 or so companies are reported to have been ISO 9000 certified. Heeks also documents a healthy degree of skepticism among industry players and executives of the value of seeking certification as an end in itself rather than a means to and end. In 1995, the practice of CMM certification had not become as common as it is reported today. Infact, Heeks suggests that majority of the companies only operate at ISO 9000 equivalent of CMM level 1 and 2 "thus putting [CMM] certification beyond their capabilities for some time to come." As with several instances earlier, we find the above situation to be fairly accurate picture of Pakistani software Industry as it stands today.

The Bottomline: If It Seems Familiar, It Probably Is

In a nutshell, in many more ways than we all anticipate, Pakistan's software industry of today (2004/05) resembles that of India from a decade ago (1994/95). Many of the readers--software industry executives and policymakers--would probably look at the state and dynamics of the Indian software industry as described in this and the preceding article and would notice the striking similarities but also some differences. Both industries developed primarily in the shadow of their hardware counterparts in the respective countries, albeit with a lag of a few years. Although the Indian governments' early policies towards computer industry (and hardware in particular) was much more restrictive than that in Pakistan, the two industries share a strikingly similar evolutionary trajectories. Striking similarities are also evident in the target setting process, the product-services mix, and the early quality dilemmas etc.

In addition to the above similarities, there are several significant differences too. First, the degree of concentration in the Pakistani software industry is much less than that in India's. There are no Tatas or Wipros in Pakistan. The ratio of revenues between the most successful and the median firm in Pakistan is around 1:100 against 1:2000 in India. In Pakistan, as against India, we find very little involvement of the country's large business houses in the software industry. Not that Pakistani business houses have not ventured into software but they have failed to create sustainable software businesses. The performance of Cressoft, Atlas Software, Kalsoft stands in sharp contrast with Tata Consultancy Services. While the reasons for this fact maybe debatable, the trend itself is undeniable.

Secondly, with an exception of NCR, Pakistan has never been seriously considered by Multinational software firms as a software development hub for their activities. This is in contrast to the early recognition of India's potential by the likes of Citicorp, TI, and GE that gave a certain boost to the Indian software industry. Third, Pakistan has seriously suffered (and continues to do so) from a weak HR pipeline, both in terms of quality and quantity of its product. This, to our mind, is the single most important hindrance in the development of Pakistan's software industry and the key differentiator between where it stands today vs. where India's software industry stood a decade from now.

Nations' histories are often deeply and irreversibly affected by idiosyncratic factors and events. Economists use the term "path dependence" to convey the notion that where nations (or industries) stand at a given instance in time is basically a reflection of a set of choices or events at various points in the past. While it is often difficult to attribute the exact benefit or harm caused by a particular event or factor, or even imagine what the counterfactual would have been but the practice of doing so remains in vogue. The Indian software industry, throughout its history, also faced quite a few of these factors and events that stand out as having significant influence on its outcomes. From as far back as the creation of IITs in 1950s, to the departure of IBM from India in 1978, to the practice of "bodyshopping" in the mid-to-late 1990s, these factors played crucial roles in the evolution and strengthening of the Indian software industry at these particular instances in time.

They, infact, can be credited for putting India onto an altogether different "path" or trajectory of development. The first provided a strong foundation for the software miracle, the second provided the entrepreneurial impetus, and the third, the final touches by creating a professionally competent and well-connected workforce. While it is hard to imagine how the Indian software industry might have evolved had there been no IITs or had they not practiced and perfected bodyshopping, the fact remains that these events did happen, either consciously or unconsciously, and changed the face of the Indian software industry. They also probably set it apart in from the Pakistani software industry and essentially gave India the head-start--and a decisive first-mover advantage--that it enjoys till this day, not only over Pakistan but also the rest of the world.

Did the decade of the 1990s, the one that Joseph Stiglitz, a Nobel Economist describes as the "roaring nineties", constitute a lost decade for Pakistan's software industry? In many ways, the above analysis provides an inkling into a possible answer to that question. Having started at pretty much the same time (mid 1970s) and progressed in similar fashion, Pakistan's software industry of today stands, like shadow of a much larger and well-recognized Indian behemoth, but it is very similar to what India was in 1995.

Many Pakistani software executives and CEOs would probably differ with my characterization of the 1990s as a lost decade for Pakistan. They would highlight the tremendous hard work and toil that they have put into the software sector during all of the nineties. In that particular respect I would understand, and probably agree with their judgment. However, the fact on the ground remains that Pakistan's software industry seems to have lost out on the tremendous opportunity that India's software industry capitalized upon in the 1990s. We would probably be justified to look for reasons and determinants of this lackluster performance in the decades prior to nineties (1980s or even earlier).

As Pakistan's software professionals, executives, and policymakers woke up to the opportunity in early 1990s, India was already well-place and on its way to capitalizing on the windfall. As they say, "if you can see the bandwagon, you've already probably lost it". The nineties marked the loss of that opportunity for Pakistan.

What Does This Mean for Pakistan's Software Industry?

What does it all mean for Pakistan's software industry, and that of other countries at similar level of development and facing the near perfect dominance of India? Would Pakistan's software Industry, having lost the opportunity by a decade, continue to grow in the manner similar to India's? Does the above analysis point towards a future where much of the disparity between India and Pakistan might be bridged someday? These are not easy questions for any industry analyst to answer.

There are several trends and counter-trends that point to different outcomes and development trajectories for the players in question. What is quite clear to even the most cursory of the observers, however, is that the IT/Software industry worldwide is now entering a phase of maturity after the crazy nineties. The investors who lost big time in the dotcom revolution--if there was ever one--are now more intelligent about investing in IT as are corporate and individual users ( e.g. CIOs) and entrepreneurs. Many of the trends that accentuated India's emergence on the world software scene (e.g. the Internet bubble, the millennium bug, and bodyshopping etc.) are perhaps a thing of the past--never to be repeated again, not atleast in the manner in which their confluence led to the dotcom bubble and its burst.

There are some counter-forces and mitigating factors as well. Firstly, the India of today is not India of the early 1990s. With somewhere between half-a-million to a million people already engaged in IT/ITES area, India is suffering from a huge strain on its HR and infrastructure capacity. The costs of setting up businesses and hiring talented programmers and managers are increasing by the day. The trends are so strong that government and private sector have been forced to look beyond the traditional Indian IT/ITES destinations like Bangalore etc. for software talent.

Secondly, the Indian government recently revoked the tax holiday on offshore operations of foreign software companies operating in India. The 35% tax is likely to hugely affected the cost-benefit calculations of foreign companies who have long viewed India as a cheap development center. Many industry observers, among them Anthony Mitchell of who has long helped US companies outsource to India, view this as the single most significant action that has suddenly put other countries, like Pakistan, in the run for the big game.

Thirdly, while India has successfully milked its early-mover-advantage for years, there might be some late-mover-advantages up for grabs too. Pakistan, and other countries, might benefit from tremendous know-how and experience that has been generated as a result of a decade of experimentation that has gone into the Indian software industry and strategically place themselves in a well-defined and less competitive niche. This would, however, require business acumen and innovation to not follow the pack but rather innovate from a market-offering and organizational standpoint.

Looking at things as they stand today, one can probably make a fairly credible guess of the fact that Pakistan's software industry would not have the "luxury" of learning-by-doing at the clients' premises as India had in the form of bodyshopping in mid-to-late 1990s. But could there be something else? One of the very important lessons that one can take away from India's success story is the virtue of being prepared to capitalize on opportunities rather than seeking the "next big thing"--a lesson often lost on the Pakistani policymakers and business leaders alike.

When India's political leaders laid the foundation of a well-functioning tertiary education system in 1950s, they certainly did not have the current IT revolution in mind. IT had not even been invented in India at that time. Yet, by creating quality human resources--scientists, engineers, managers, and policymakers--they positioned India as a country that could capitalize on an opportunity as and when it arose. That's precisely what happened in early 1990s. The public policy lesson, for Pakistan and its likes, is to pay emphasis on long-term foundational investments in infrastructure, international goodwill, and human resources so as to quickly position themselves for the next big thing, as and when it becomes apparent what that is going to be.

Till that happens, Pakistani software professionals, business leaders, and policymakers must approach the strategic and policy-environmental challenges identified in the PSEB Best Practices study to forge ahead against formidable reputational and capability-based entry barriers. As many who toil for the cherished goal have discovered, it will certainly not be a walk-over that many naively expected in early 1990s, but rather a long and hard slog. But, in the end, it might be one well worth all the effort.

Athar Osama is a Doctoral Fellow in Policy Studies at the RAND Graduate School for Policy Analysis in Santa Monica, California. He specializes in technology and innovation strategy and policy areas. He maybe contacted at


Post a Comment

<< Home