Embed Size (px)
Citation preview
Stanford University John A. and Cynthia Fry Gunn Building
366 Galvez Street | Stanford, CA | 94305-6015
Working Paper No. 526
Let's make software patentable. . . or maybe let's not: Evidence from the Indian software
industry
By
Markus Eberhardt Christian Helmers
Marcel Fafchamps Manasa Patnam
April 2015
Let’s make software patentable. . . or maybe let’s not:
Evidence from the Indian software industry∗
Markus Eberhardta Marcel Fafchampsb
Christian Helmersc Manasa Patnamd
a University of Nottingham
b Stanford University
c Santa Clara University
d CREST (ENSAE)
October 15, 2014
ABSTRACT
This paper analyzes the impact of changes in the availability of patent protection for software
in India over the past decade. We find that a proposed broadening of patent eligible subject
matter to include software had a large positive effect on average on companies’ stock market
returns in India, but little effect on their patenting behavior. An unanticipated reversal
of this proposed policy change resulted in substantial negative stock market returns. Our
analysis of the link between the availability of patent protection on software and companies’
focus on (customized) software services versus software products does not yield any strong ev-
idence that the availability of patent protection affects the type of software produced by companies.
Keywords: Patents, software, innovation, India
JEL Classification: D22, O14, O34, O38
∗We thank Luis Aguiar, Brian Love, and participants at the 12th ZEW Economics of ICT Conference
for helpful comments. We are grateful to Rubina Anjum for excellent research assistance with the data
collection. Corresponding author: [email protected]
“This [software] is a world-class industry built from nothing, [. . . ] while overcoming
India’s [. . . ] inept, and sometimes venal, state.”The Economist (2011)
“This [Manual of Patent Practice and Procedure by the Indian Patent Office] implies that
software with a hardware implementation can be patented. [. . . ] [T]his will be devastating
for the Indian software industry [. . . ]”Indian Institute of Technology (2008)
1 Introduction
There is an important, long-standing debate on the effect of intellectual property (IP)
rights, in particular in form of patents, on innovation and economic development (Nord-
haus, 1969; Helpman, 1993; Grossman and Lai, 2004). One strand of this literature
focuses on the impact of patents on innovation in high-tech industries, mostly in devel-
oped economies (Scherer and Weisburst, 1995; Sakakibara and Branstetter, 2001; Jaffe
and Lerner, 2004), while another analyzes the role that patents can play for economic de-
velopment in emerging and developing economies (Penrose, 1973; Chen and Puttitanum,
2005; Hu and Jefferson, 2009).
We provide empirical evidence on both questions by relying on a number of policy
shocks which had opposing effects on the availability of patent protection for software in
India over the past decade. These shocks allow us to identify the effect of changes in the
patentability of software on the performance of software companies listed on the India
stock exchanges. We further analyze the effect of patent protection on the direction
of innovation in the Indian software industry. Innovation does occur in the absence of
patents, but if patent protection is not available, innovation is directed towards output
that can be more easily protected by alternative mechanisms such as secrecy (Moser,
2005, 2012). In fact, the larger part of the Indian software industry produces customized
software and provides software-related services, which raises the question whether limited
patent protection has directed the software industry away from generic software products
towards (customized) software services and similar activities that benefit less from patent
protection.
For our analysis we use various changes in the Indian patent law and rules applied
by the Indian Patent Office (IPO) which affected patent eligibility of software.1 In In-
dia, software per se is explicitly excluded from patentable subject matter. In December
1Patent eligibility defines whether a given invention represents patentable subject matter. A patenteligible invention is patentable if it satisfies the statutory requirements for patentability, most notable,novelty, inventive step, and industrial applicability.
2
2004, however, the Indian government announced an ordinance that would have substan-
tially loosened the restrictions on patent eligibility of software. The proposed ordinance
was unexpectedly rejected by parliament in April 2005 and the IPO subsequently is-
sued patent examination guidelines that made it clear that the restrictions would not
be lifted. Our analysis relies on this unique historical episode and the resulting policy
shocks to identify the effect of the presence or absence of patent protection on stock mar-
ket valuations of companies listed on the Bombay Stock Exchange (BSE) and India’s
National Stock Exchange (NSE) by means of an event study. Further, we trace the im-
pact of these policy changes on the patenting activity of the Indian software companies
by investigating patents filed with the IPO as well as the U.S. Patent and Trademark
Office (USPTO) before and after the policy changes. Since software was patent-eligible
in the U.S. throughout our period of analysis our data allow us to compare filings by
Indian software companies across jurisdictions. To investigate whether the restrictions
on software patents have influenced the type of software produced by companies, we use
information on companies’ sales broken down into products and services. We analyze
whether companies focused more on software services as it became clear that the restric-
tions on patent eligibility of software would not be loosened. To carry out our analysis,
we construct a firm-level dataset for the period 2000-2011 that combines patent filings
at the Indian and U.S. patent offices for all publicly traded software companies in India.
Our findings show that the announcement of a loosening of the restrictions on
patentability on average resulted in significant positive stock market returns for the com-
panies in our sample, although we find substantial heterogeneity in this effect. When
parliament rejected the proposed change four months later and the Patent Office issued
revised examination guidelines that clarified its unaltered stance and reinforced the re-
strictions on the patentability of software, companies on average experienced substantial
stock market losses. This suggests that the stock market expected at least some software
companies to capitalize on an improved ability to protect software inventions through
patents. Due to the short time window applied in our event study, the positive stock
market returns reflect expected future benefits rather than a reward for realized changes
in companies’ activities. Following the IPO’s release of the revised examination guide-
lines it became clear that these future benefits would not materialize such that stock
markets reacted accordingly and cut their valuations.
Our analysis of companies’ patent filings indicates at best a very minor change in
companies’ filing behavior following the 2004 ordinance. The release of the 2005 guide-
lines had a more important effect on patenting, most likely because it eliminated some
of the policy uncertainty surrounding patent eligibility of software and laid out the rules
3
more clearly. Finally, establishing a link between the policy shocks and companies’ prod-
uct mix is more difficult due to data limitations. Our results show no robust evidence
that companies increased their focus on software services at the expense of software
products. Hence, these results do not suggest that the availability of patent protection
generally affected the type of software created by companies. That said, we find some
evidence for a positive association between the share of services in total sales and the
policy shocks in 2005 for the subset of companies that had applied for software patents
before the undoing of the ordinance and the revision of the patent office’s guidelines in
2005. This further underscores the heterogeneity in firm response to the availability of
patent protection.
Our analysis contributes to the existing literature in at least two ways. First, we
offer empirical evidence on the impact of changes in the availability and strength of
patent protection on companies in a developing economy. Most of the existing literature
on IP in developing markets focuses either on pharmaceutical patents (Arora et al.,
2009; Sampat, 2010; Arora et al., 2011; Kyle and Qian, 2013) or relies on cross-country
analysis which often has difficulty arguing for exogenous variation in the measure of IP
strength (Lerner, 2002; Kanwar and Evenson, 2003; Qian, 2007; Hu and Png, 2013). Our
study provides evidence on the effect of patents on a highly-innovative industry that has
contributed enormously to India’s recent economic growth.2 Second, our results also
speak to the debate on the patentability of software more generally. Despite the fierce
debate on software patents in the U.S. (FTC, 2011; Graham and Vishnubhakat, 2013),
there is precious little empirical evidence on their effect on companies and innovation
(Lerner and Zhu, 2007; Lemley, 2012; Hall et al., 2012). Our analysis offers evidence on
the question whether software patents affect not only the rate of innovation but also the
type of products produced by companies. This analysis of the effect of software patents
on the direction of innovation also sets our analysis apart from the existing literature,
which is chiefly concerned with the effect of software patents on rates of innovation.
The remainder of this paper is organized as follows. Section 2 provides an overview
of the chronology of the legal treatment of software patents in India. Section 3 describes
our empirical approach. Section 4 contains a description of the data used in our analysis.
Section 5 presents our findings and Section 6 offers some concluding remarks.
2Software accounts for around 12-16% (depending on industry definition) of total Indian exports(2010/11) and employs around 2.5 million mostly skilled workers (UNCTAD, 2012). See online AppendixD for a detailed discussion of the Indian software industry.
4
2 Software Patents in India
India’s 1970 Patents Act predates the IT era and hence did not refer explicitly to soft-
ware, as this was not a relevant concern at the time.3 This means software was not
explicitly excluded from patentability. However, the Patents Act established that only
methods for the manufacture of products were patent eligible, which effectively excluded
software per se from patentability.4 Yet, due to the lack of clauses specifically regulating
whether software was patent eligible, the four patent offices had some discretion in their
decision to grant software patents.5 The 2002 amendment to the patent law added a
clause to Section 3 which defines subject matter excluded from patent eligibility. Accord-
ing to this amendment, inventions in the form of “a mathematical or business method or
a computer programme per se or algorithms” (Section 3(k) amended Patents Act) are
not patent eligible. This meant that software on its own cannot be patented, although
software in combination with hardware may be patentable. The 2002 amendment clar-
ified the position of the Indian patent office with regard to software patents, although
some discretion remained in determining what “per se” meant in practice. In fact,
according to the office’s first edition draft Manual of Patent Practice and Procedure,
software was only patentable in combination with hardware if the hardware required
some ‘non-standard’ modification specific to the software.
In December 2004, in preparation for another major amendment to the Patents Act,
the Indian government issued an ordinance which stated that the clause governing soft-
ware patentability added in 2002 should be modified to exclude “a computer programme
per se other than its technical application to industry or a combination with hardware”
[emphasis added].6 Hence, although the amendment did not propose the elimination of
the “software per se” restriction, it explicitly opened the possibility for software to be
patent eligible provided it had technical application. Also, according to the 2004 amend-
ment, the combination of software and hardware was patent eligible without requiring
the specific adaptation of the hardware. Although the proposed amendments left ample
room for interpretation, for example it was not clear how “technical application” was
defined, they still would have had a major impact on the granting practice of software
patents. However, parliament surprisingly rejected this modification and the amendment
3Software in India only started developing in the late 1980s through companies such as Datamatics,Patni, and Tata Consultancy Services.
4Chapter II, 3(d) of the 1970s Patents Act.5The national patent office in India is divided into four offices: Kolkata, Chennai, New Delhi, and
Mumbai.6The clause still maintained that “a mathematical method or a business method or algorithms” are
excluded from patentability.
5
adopted in April 2005 did not include the software-related amendment put forward in
the ordinance. This creates a window of just over 4 months during which it was expected
that software would become patent eligible. Parliament’s unexpected rejection of the
amendment made it clear that the law makers wanted the Indian Patent Office to adopt
the narrow application of the patent eligibility rules set forth in Section 3 of the Patents
Act. Nevertheless, ambiguity persisted in the office’s granting practice since the wording
remained open to interpretation and therefore discretion in its application.
But this state of affairs did not last long. Shortly thereafter, in June 2005, the office
issued the second edition of its draft Manual of Patent Practice and Procedure with
detailed guidelines to clarify the stance of the office with regard to software patents and
its practice in examining incoming patent applications. The guidelines affirmed that
software per se was not patentable, but software that made a ‘technical contribution’ in
combination with hardware was patentable (Manual of Patent Practice and Procedure
2005, 2.8). Hence, patentability of software was restricted by the requirement of solving
a technical problem as well as the combination with hardware. While this resembles
the ordinance, the guidelines also established that “claims relating to software program
product [sic] are nothing but computer program per se simply expressed on a computer
readable storage medium and as such are not allowable” (Manual of Patent Practice and
Procedure 2005, Annexure II 7.3).
In February 2008, the office issued another draft edition of the guidelines affirming
its stance expressed in the 2005 guidelines and clarifying the patentability of computer
implemented mathematical methods and algorithms. The revised guidelines suggest that
mathematical methods and algorithms may be patentable if they represent a technical
application, which would have softened a restriction thus far maintained by the 2005
amendment and the previous editions of the guidelines.7 In late 2010, the office however
published a revision of the 2008 draft guidelines, which now adopt a more restrictive
approach to the patentability of mathematical methods and algorithms than the 2008
guidelines. The 2010 guidelines also adopt clear language regarding the patentability
of software products, stating that “computer programme products are computer pro-
gramme per se [...] and as such are not allowable” and further a “computer programme
which may work on any general purpose known computer does not meet the requirement
of patentability” (Manual of Patent Practice and Procedure 2010, 08.03.06.10 d. and
g.). However, the guidelines also clarify that this does not mean that software across the
7The guidelines offer the following example of a patentable mathematical method: “a method ofimage processing which used the mathematical method to operate on numbers representing an image”(Manual of Patent Practice and Procedure 2008, 4.11.10).
6
board is not patentable. In fact, the guidelines state that “[i]f a claim in a patent appli-
cation is not directed at a computer programme per se it could be patentable” (Manual
of Patent Practice and Procedure 2010, 08.03.06.10 e.) and affirms that software in com-
bination with hardware may be patentable. The 2010 manual offers the most precise
delineation of the conditions that make software patent-eligible. The chronology of all
the events described is summarized in Figure 1.
Figure 1: Chronology of patentability of software de jure and de facto
Amendmentrestricting
patentability
25/06/2002
Ordinancebroadening
patentability
27/12/2004
Amendmentundoing
ordinance
05/04/2005
Guidelines
20/06/2005
Revisedguidelines
25/02/2008
Revisedguidelines
04/11/2010
Notes: Red colors indicate softening of restrictions on software patentability. Black color indicates a tighteningof restrictions on software patentability.
In summary, the Indian patent system transitioned from applying broad clauses
contained in the original 1970 Patents Act that were not designed with software in mind
to fairly specific guidelines contained in the 2010 manual on practice and procedure.
The transition was characterized by some back and forth, tightening and loosening the
restrictions on the patentability of software. Especially the 2004 ordinance, its rejection
by parliament four months later in April 2005, and the issuance of the revised guidelines
in June 2005 stand out in this process. The following sections explain how we use
these policy changes to study the effect of the patentability of software on companies’
performance and behavior.
3 Empirical Strategy
Apart from looking directly at software companies’ patenting behavior, we assess the
impact of the policy on software companies in two ways. First, we examine the imme-
diate effect of the policy on investor confidence by investigating changes in the market
value of firms around the date of the policy announcements. The underlying rationale is
that stock market prices fully reflect the present discounted value of a firm’s future prof-
itability, conditional on the effect of the policy change. Rational investors will therefore
7
immediately take this into account such that the post-event relative to the pre-event
price of the security indicates the difference in market valuation due to the effect of the
policy. Second, we analyze the direct effect of the different policy shocks on the share of
software services (as opposed to software products) in companies’ total sales.
3.1 Event Study Methodology
To analyze the effect of the policy shocks on the Indian software industry, we use an
event study approach looking at changes in companies’ market valuation, which acts as
a proxy for firm performance. The objective is to distinguish market reactions due to
normal factors from reactions induced by the policy shocks. To do this, we specify a
market model (‘normal returns’) and estimate deviations (‘abnormal returns’) from such
a model over the event window. To illustrate, consider the event timeline in Figure 2
where τ is the event date (policy shock):
Figure 2: Event timeline
- T
T0 T1 T3T2τ
Estimation Event Post-Event
The market model specifies the relationship between the period t return on security
i (rit) and the period t return on the market portfolio (rmt ):8
rit = α+ βrmt + εit (1)
We estimate this market model in the estimation window (T0−T1) prior to the event
τ to obtain estimates for α, β. Using these estimated parameters we predict expected
return r̂pit in the event window (T1 − T2), from which in turn we can then compute the
deviation from the observed return, thus providing a measure of ‘abnormal returns’ (ε̂it):
ε̂it = rit − r̂pit r̂pit = α̂+ β̂rmt (2)
The cumulative abnormal return (CAR) over the event window is then simply the
accumulated ε̂it from T1 to T2:
8Our discussion of the methodology follows MacKinlay (1997), Campbell et al. (1997) and Hall andMacGarvie (2010).
8
CARit =
T2∑τ=T1
ε̂it (3)
If the event had no effect on our subset of firms then ε̂it should have mean 0. If
the policy change had a positive effect then we observe abnormal returns, ε̂it > 0 which
produce an upward sloping CAR. The underlying assumption is that without the event
or singular information, the relation between a firm’s asset return and the market return
is stable. We present our results as average CAR estimates across all software firms,
reporting the outcome at t + 5 days as well as (in graphical form) for days t − 5 to
t + 5: while the latter will indicate any response to the specific event, the former will
concentrate on the likely permanence of the impact.
A number of choices have to be made before we can implement our event study:
we specify a pre-event window from t − 80 to t − 10 days, giving us substantial time
series observations to estimate the expected returns for firm i while at the same time
preventing any overlap between pre-event window for event τ and the event window for
event τ − 1. The event window encompasses the 11 days from t − 5 to t + 5. Both
of these choices are fairly standard in the literature and we convinced ourselves that
reasonable changes to these window sizes do not affect our results in any significant
manner. Finally, we investigate a number of alternatives for the market model outlined
above: firstly, we analyze a simpler ‘constant mean return’ model, where the abnormal
returns are computed for firm i without accounting for the return on the market portfolio,
i.e. rmt is dropped from equation (1). Secondly, in the standard ‘market return model’ we
investigate a number of alternative indices for market return, namely the BSE Sensex,
the CMIE Index for IT, the BSE Tech Index and the CNX IT Sector Index of the
National Stock Exchange of India.
3.2 Software Type
India’s software industry has grown to be world famous for its software services rather
than the production of software packages and products (see online Appendix D). There is
convincing empirical evidence that innovation occurs in the absence of patents, although
the unavailability of patents directs research towards innovative output that can be more
easily protected by alternative mechanisms such as secrecy (Moser, 2005, 2012).9 In the
case of the Indian software industry, this might imply that the restrictions imposed on
the patentability of software have directed the industry increasingly towards (customiz-
9For a detailed review of this literature see Hall et al. (2014).
9
able) software services. Patents are of little value to software services as they are often
specifically tailored to customer needs and made available only to specific customers.
This also means that often the (customized) software solutions are not visible to the
public and hence the risk of piracy is low.
Therefore, we also analyze whether the availability of patent protection has had
an effect on the type of innovation companies pursue. Specifically, we look for any
association between the different policy shocks and the share of software services (as
opposed to software products) in a company’s total sales. One of the limitations in this
analysis is the fact that product-level data is only available on an annual basis. This
makes it difficult to look at policy shocks that occurred in short sequence, such as the
Ordinance in December 2004 and its undoing in 2005 shortly thereafter.
We examine whether there is such an effect by estimating Equation (4). The depen-
dent variable is the share of software services in firm i’s total sales in year t. We include
a dummy variable I(t≥τ) equal to one for the time period following a policy shock. The
specification also includes company-level characteristics Xit: firm age, R&D intensity
and total sales. We also accommodate different time trends in services shares through
year dummies.
sit = α+ βI(t≥τ) + δXit + γt + εit (4)
The dependent variable, share of software services si in total sales, is a fractional
response variable, i.e. it is bounded between 0 and 1 and includes a significant amount of
observations taking on values at 1. In principle we could use a log-odds transformation of
the dependent variable to address the fact that it is bounded between 0 and 1. However,
this approach is unsatisfactory in our context because such a transformation does not
accommodate the possibility that si takes on values of 0 or 1 with positive probability. To
tackle this concern we estimate Equation (4) using a quasi-maximum likelihood estimator
(QMLE) for a fractional response model as proposed by Papke and Wooldridge (1996).
The QMLE is obtained by maximizing the Bernoulli log-likelihood function, given by:
li(β̂, δ̂, γ̂t) = sit log[G(E(sit|I(t≥τ), Xit, γt)
)] + (1− sit) log[1−G
(E(sit|I(t≥τ), Xit, γt)
)]
where G(·) is a logistic function, G(z) ≡ Λ(z) ≡ exp(z)1+exp(z) , satisfying 0 < G(z) < 1 for all
z ∈ R.
In addition to Equation (4), we also use a control sample of non-software firms.
Because data at the product-level is only available for a small subset of firms within
industries, we combine data from three industries to create the control set of firms:
10
information technology enabled services (ITES), telecommunication, and electronics in-
dustries.10 Although we cannot rule out that these companies are also engaged in the
production of software, their main line of business is elsewhere. Using the control sample,
we can estimate the following specification using the fractional response model described
above:
sit = α+ θsoftwarei + β[softwarei × I(t≥τ)
]+ δXit + γt + εit (5)
Specification (5) allows us to interpret the coefficient β on the interaction term as
evidence for any potential differential effect of the different policy shocks on software
companies relative to the companies in the control sample.
4 Data
To analyze the effects of changes in the availability of patent protection on software in
India, we construct a firm-level dataset for the period 2000-2011 that combines patent
filings of all publicly traded software companies registered in India with the national
patent office as well as filings with the USPTO (see online Appendix A). The patent
data are matched to firm-level data on all software companies listed in India – both on the
Bombay Stock Exchange (BSE) and the National Stock Exchange (NSE); a total of 294
software companies. The firm-level data come from the PROWESS database provided by
the Center for Monitoring of the Indian Economy (CMIE). For the product-level analysis,
we add data from PROWESS on all companies in the ITES, telecommunication, and
electronics industries that report information on sales at the product-level. Table A-1 in
the online Appendix contains descriptive statistics for a few core characteristics of the
companies used in the regression analysis.
5 Results
5.1 Patent Filings
Figure 3 shows total patent filings of publicly traded software companies in India by
quarter over the entire 2000-2011 period.11 The graph shows that the absolute number
10Our results are robust to dropping any of the three industries from the set of controls.11Although there is no official definition of software patents, we check all patent filings to ensure they
cover software inventions. In the literature, different definitions have been used to identify softwarerelated patents. Graham and Mowery (2003), for example, define software patents as patents in thefollowing patent classes: Electric Digital Data Processing (G06F), Recognition of Data; Presentation
11
of filings by Indian software companies at both the Indian and the U.S. patent offices is
modest.12 Having said that, a total of 1,240 patent filings with the Indian patent office
and an overall share of 15% of patenting companies (45 companies – see Figure A-1 in
the online Appendix) among all listed software companies is still substantial,13 keeping
in mind that software remained in principle unpatentable throughout the period studied.
Figure 3: Total patent filings – Indian Patent Office vs USPTO, 2000-2011
Figure 3 suggests that patent filings with the Indian Patent Office by publicly traded
software companies in India increased enormously over this 12-year period. The largest
increase in this period is however entirely due to two firms, Infosys and Tata Consultancy
Services. Online Appendix Table A-2 shows that their combined number of filings almost
tripled in a single year from 119 to 348 filings. We can also see that there was no similar
increase in filings at the USPTO, for which filings by our sample fell from 2008 onward.
of Data; Record Carriers; Handling Record Carriers (G06K), and “Electric Communication Technique”(H04L). Bessen and Hunt (2007) opt for an alternative approach, searching USPTO patents for keywords.Due to the relatively small number of patent filings, we were able to check each patent manually byreading its abstract.
12In comparison, Bessen and Hunt (2007) found 21,000 patent filings at the USPTO in 2000 alone,the majority of which originate from U.S. companies.
13In our analysis we do not distinguish between the different locations of the Indian Patent Office sincenearly all software patent applications (97%) are made at the Chennai (52%) and Mumbai (45%) officesand companies overwhelmingly file at the same office throughout the period studied in our analysis.
12
While Figure 3 indicates that despite the restrictions placed on patentability software
companies have been filing a sizeable number of software-related patent applications,
Figure 4 indicates that very few of these filings with the Indian Patent Office have been
granted. The grant share fluctuates until the end of 2004 and then plummets in early
2005 shortly after the announcement of the ordinance, remaining close to zero from the
fourth quarter of 2006 onward. Figure A-2 in the online Appendix visualizes the poor
grant rate even more clearly as it displays absolute counts of the Indian filings by legal
status. It shows that only 3% of all patent filings with the Indian Patent Office have been
granted by February 2014.14 The largest share of filings is still pending (62%), meaning
the examination has not been concluded. Since there are a large number of pending
patents that have been filed a decade or more ago, it is likely that these filings have
been abandoned by the applicants. This also means that the share of filings officially
withdrawn or deemed abandoned is most likely substantially larger than 14%. Perhaps
even more revealing, the only patent applications officially rejected by the Indian Patent
Office are applications filed immediately after the announcement of the 2004 ordinance.
For comparison, Figure 4 shows the share of filings granted by the USPTO. Until the
end of 2004, all filings at the USPTO were granted. However, shortly after the undoing
of the ordinance, the grant rate began a downward trend. Figure A-3 in the online
Appendix shows that the drop is not explained by a grant lag as only few patent filings
are still pending. Interestingly, following the announcement of the ordinance, we see for
the first time that filings with the USPTO that are eventually withdrawn. In any case,
the substantially larger grant rate at the USPTO underscores the restrictive approach
adopted by the Indian Patent Office and shows that despite a relatively large number
of software patent filings at the Indian Patent Office, hardly any software patents were
granted.
Turning our attention to the different policy shocks, we find that there was no no-
ticeable reaction to the first amendment in 2002. Although the amendment introduced
changes that could have provided incentives for increased patenting, such as the exten-
sion of the patent term to 20 years, it restricted the patentability of software. Hence it is
not surprising that there was no change in software patenting following the amendment.
The entire period 2000-2004 was characterized by very little patenting activity (on av-
erage 9 filings at the Indian and the U.S. patent offices combined per year). In contrast
14Since we check the legal status of filings in February 2014, it is possible that especially filings in 2010are affected by a truncation problem (i.e. not enough time has lapsed for these patents to have beenexamined). Figure A-2 suggests this is indeed the case as for a large number of patent filings in 2010 theapplicant has not yet requested examination. That said, the truncation does not affect the statistics forearlier years of the period studied.
13
Figure 4: Share of granted patents in total patent filings at the Indian Patent Officeand the USPTO – 2000-2011
RestrictingAmendment
Ordinance
UndoingOrdinance
Revised
Revised
Guidelines
Guidelines0
.2.4
.6.8
1S
hare
gra
nted
pat
ents
in to
tal f
iling
s
2000q1 2001q3 2003q1 2004q3 2006q1 2007q3 2009q1 2010q3
Indian Patent Office USPTO
Notes: The legal status was obtained in February 2014 for Indian patent filings from the iPairs website of theIndian Patent Office and from the PAIR website of the USPTO.
patent filings spike during the first quarter of 2005 immediately after the announcement
of the ordinance and drop immediately after the ordinance was rejected by parliament
in the second quarter of 2005. Strikingly, patent filings jump at the Indian as well as
the U.S. patent office. This is interesting because there was no concomitant change in
the law regulating the patentability of software in the U.S. As it turns out, the sudden
increase in filings is largely driven by three companies that had not patented previously:
KLG Systel, Megasoft, and Ramco Systems. Ramco filed eleven patents with the In-
dian patent office and 12 patents with the USPTO during the first quarter following
the announcement of the ordinance. Ramco had not filed any patents with the Indian
office before that (since January 2000). Moreover, its filings with the Indian office stand
out because they all claim priority in the U.S., that is, the company had previously filed
patents in the U.S. on the same software inventions.15 This suggests that Ramco decided
to file patents with the Indian patent office that it had already filed at the USPTO at
an earlier date. These USPTO filings do not show up in 2004 in Figure 3, however,
15The priority right allows applicants to file a patent at different offices around the world while freezingthe first filing date for 12 months (and hence the point in time that is relevant for assessing novelty andinventive step), which is referred to as the priority date.
14
because they were provisional filings, based on which the company subsequently filed
the non-provisional applications in the first quarter of 2005.16 It is obviously difficult
to know what motivated Ramco to transform its provisional U.S. filings into regular fil-
ings at the USPTO and to file the same patents with the Indian Patent Office after the
government had announced to loosen the restrictions on the patentability of software.
However, it bears noting that none of Ramco’s applications at the Indian Patent Office
were granted. Two applications have been rejected explicitly based on the fact that soft-
ware is not patent eligible;17 in contrast, their equivalent filings in the U.S. were granted
by the USPTO. Two applications at the Indian office and the USPTO were abandoned
by Ramco, and the rest is still pending in India but has been granted at the USPTO
(except for one abandoned application). Also, Ramco never filed a patent application
with the Indian office again in any of the following years, but kept filing applications with
the USPTO. Taken together, Ramco’s filing behavior suggests that its filings were at
least partly motivated by the prospect of a broadening of patent eligible subject matter
as foreseen by the ordinance. The other two companies filing domestically are NIIT and
KLG Systel. While NIIT had filed patent applications before the ordinance, KLG Systel
started filing only about a month after the ordinance was announced. At the USPTO, a
total of seven companies filed patents in the quarter following the ordinance. However,
apart from Ramco, only Megasoft started filing for the first time during that period.
Since both the amendment that ignored the ordinance and the subsequent guidelines
fall into the same quarter, Figure 3 does not distinguish between them. A closer look at
filings at weekly intervals, however, indicates that the drop in filings occurs immediately
after the undoing of the ordinance and that no change in filings occurred between the
undoing and the release of the revised guidelines. Shortly after the release of the guide-
lines, however, patenting picks up again, although filings fluctuate somewhat. Filings
start increasing more substantially in 2007, going from 21 to 36 filings between the first
and fourth quarter, well before the revised examination guidelines are released in early
2008. The increase continued after the release of the guidelines, as would be expected
since the guidelines made it slightly easier to patent software in India. Then there is that
jump in patents directly after the revision of the guidelines in 2010, which is entirely
driven by two companies as explained above.
The analysis of patent filings produces ambiguous findings. Although it appears that
16A provisional application is a patent application that requires less information, for example no formalclaims, but which does not lead to a granted patent. A provisional application can be transformed intoa non-provisional application within 12 months of filing.
17The decisions explicitly refer to Section 3(k) of the 1970 Indian Patents Act, which excludes softwarefrom patent eligible subject matter.
15
the ordinance has had a positive effect on patent filings, this effect appears to be driven
by a single company, which decided to file all of its U.S. patents domestically as well.
Since the company’s filings with the Indian Patent Office were strikingly unsuccessful
compared to their counterparts at the USPTO, it seems that the company filed the former
applications expecting the restrictions on patentability to loosen. Nevertheless, there is
no evidence for a broad-based change in patenting behavior immediately following the
ordinance. This might be explained by the short time frame of only four months before
parliament rejected the proposed changes or simply reflect the fact that the ordinance
never turned into binding law. Instead it appears that the release of revised guidelines
in 2005, 2008, and 2010 had a positive effect on filings at the Indian Patent Office. A
possible explanation is that companies preferred clarity over the rules that would lead to
the rejection of patent applications to avoid filing an application that would eventually
get rejected based on Section 3(k) or to be able to draft software patents in a way to
bypass the restrictions although the grant rate shows no evidence for that.
5.2 Abnormal Returns
Next, we analyze stock market reactions to the (announced) policy changes: if including
software in patentable subject matter is privately beneficial to firms, we would expect
the softening (tightening) of restrictions on software patentability to have a positive
(negative) effect on firm valuation. As our results below will show, it is crucial in this
analysis to go beyond averages across all firms and to distinguish different groups of firms
across events. Figure 5 provides a graphical representation of the Cumulative Abnormal
Returns (CAR) over the 11-day event window computed adopting the Constant Mean
Return Model – results for alternative Market Models, presented in Figure 6, do not
appear to deviate from the patterns revealed in the simpler empirical model. In all
graphs we indicate statistically significant (insignificant) average CAR across all listed
software firms in India with a filled (hollow) marker, adopting a 5% significance level.
Counter-intuitively, the ordinance broadening patentability in late December 2004
as well as its undoing in April 2005 are both shown to have had a positive impact, while
the second event in 2005, the publication of the draft Manual of Patent Practice and
Procedure with detailed guidelines to clarify the stance of the IPO represents a substan-
tial negative shock. The estimates presented in Table 1 suggest that the broadening of
the ordinance results in an average CAR of around 14%, while its undoing led to a 10%
average excess return. The drop in returns following the publication of the draft guide-
lines in June 2005 was of similar magnitude. All other events indicate either statistically
16
insignificant or very small average CAR across all firms and our graphs indicate that
these cannot be interpreted as convincing evidence for significant market reaction.
However, these averages are somewhat misleading and additional investigation re-
veals a more intricate yet consistent pattern across events: although the ordinance in
2004 and its undoing in April 2005 both had an average positive effect, this average in
each case was driven by different firms. In contrast, the positive reaction to events in
2004 and and the drop in average CAR in June 2005, were driven by the same firms.
In order to illustrate this matter empirically we plot firm-specific CAR at time t + 5
following each event in Figure 7:18 in the upper panel we observe that many firms with
high CAR in 2004 saw this measure drop in April 2005 (blue downward arrows) and
that the average positive CAR – indicated by the dashed horizontal line – is driven by
other firms with low CAR for the December 2004 event (red upward arrows). From
this we conclude that the identity of the firms which drove the average positive CAR in
either time period differed. Turning to the lower panel of the same Figure, we observe
a multitude of lengthy blue arrows: firms with high CAR following the broadening of
patentability in 2004 saw their CAR drop to zero or negative values when the June
2005 guidelines firmly shut the door on software patentability. This significant drop is
driving the average negative CAR in the sample – indicated by the dashed horizontal
line – whereas firms with comparatively low or negative CAR in 2004 (red arrows) now
represent only a limited positive counterweight. We conclude for the lower panel that
the average CAR in both periods was primarily driven by the same firms, namely those
which were deemed by the market to benefit from a broadening of patentability.
Next, given that we have detected differential groups within our sample, we ask
whether specific observable company characteristics can explain the substantial hetero-
geneity across firms associated with the policy changes illustrated in Figure 7.19 As
dependent variable we construct a dummy variable Idrop equal to unity (a) if the firm-
specific CAR at t+ 5 following the April 2005 undoing of the ordinance is smaller than
that t+ 5 following the December 2004 event, and (b) if the firm-specific CAR at t+ 5
for the June 2005 revised guidelines is smaller than its equivalent for the December 2004
ordinance broadening patentability — in terms of the plots in Figure 7 we are trying
to predict the blue arrows. An alternative dependent variable ‘∆CAR’ measures the
18In both panels ‘+’ and ‘x’ indicate firms which only had one CAR estimate, in the former case forthe respective 2005 event, in the latter for 2004.
19We also computed cross-tabulations for negative CAR in December 2004 versus negative CAR inApril 2005 and June 2005. Conducting a Pearson’s chi-squared test cannot reject the null of independenceacross columns and rows for the tabulation with the April 2005 event (χ2(1) = 0.67, p = .415) but wereject independence for the tabulation with the June 2005 event (χ2(1) = 4.44, p = .035).
17
Figure 5: CAR for Constant Mean Models
Notes: The event analysis in the upper panel adopts a 50-day estimation window (minimum number ofobservations: 30) while the lower panel adopts a 70-day one (minimum observations: 50). Within each plot theseries markers indicate statistically insignificant (hollow marker) and significant (filled marker) CARτ where weadopt a 5% level of significance. The CAR in Table 1 below report the estimates at t+ 5.
18
Figure 6: CAR for Market Return Models (various)
Notes: Each series corresponds to the Cumulative Abnormal Returns for the 11-day event window of each ofthe events detailed in Figure 1.
Table 1: Cumulative Abnormal Return Estimates (Mean, Market Models)
Date Event Mean Return Model Market Return Model
CAR[-5,+5] abs SE CAR[-5,+5] abs SE
June 25 2002 Amendment restrictingpatentability
-0.0227 [0.0194] -0.0226 [0.0193]
December 27 2004 Ordinance broadeningpatentability
0.1371 [0.0272]*** 0.1429 [0.0279]***
April 5 2005 Amendment undoingordinance
0.1048 [0.0201]*** 0.1067 [0.0201]***
June 20 2005 Revised guidelines -0.1029 [0.0212]*** -0.1320 [0.0219]***
February 25 2008 Revised guidelines -0.0123 [0.0126] -0.0175 [0.0126]
November 4 2010 Revised guidelines 0.0249 [0.0100]** 0.0233 [0.0100]**
Notes: The market return model estimation adopts the BSE Index; results for alternative indices arequalitatively identical. * p < 0.10, ** p < 0.05, *** p < 0.01.
19
Figure 7: CARt+5 by firm, comparing 2004 and 2005 events
Notes: The two graphs compare the firm-specific CARs for the broadening of patentability in December 2004and its subsequent undoing in April 2005 (event 2005a) – upper plot – as well as the CARs for the 2004 eventand the revised guidelines in June 2005 (event 2005b). Firms are ordered by id along the x-axis – this merelyacts to provide a visual ranking of the firms. The red and blue arrows indicate the difference in firm-specificCAR 5 days after the respective event: for red (blue) arrows firm-specific CAR in 2005 (2004) exceeds that in2004 (2005). Two symbols indicate firms where one or the other CAR is missing: ‘+’ are values for the respective2005 event (i.e. 2004 is missing), ‘x’ are values for 2004 (i.e. 2005 event observation is missing). The horizontallines indicate average sample CAR for event 2005a and 2005b in the upper and lower panel, respectively.
20
Tab
le2:
Wh
ich
firm
sw
her
em
ost
affec
ted
by
the
pol
icy
chan
ges?
Ap
ril
2005
&Ju
ne
2005
vs
Dec
emb
er20
04
Ap
ril
2005
vs
Dec
emb
er20
04Ju
ne
2005
vs
Dec
emb
er20
04
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
I dro
pI d
rop
∆CAR
∆CAR
I dro
pI d
rop
∆CAR
∆CAR
lnS
ale
s-0
.057
-0.0
72
0.0
450.
056
-0.0
38-0
.035
0.058
0.06
0[0
.027
]**
[0.0
23]*
**
[0.0
16]*
**[0
.015
]***
[0.0
24]
[0.0
24]
[0.0
24]*
*[0
.025]
**
Fir
mA
ge-0
.004
0.000
0.1
640.
154
0.00
20.
001
0.00
30.
002
[0.0
11]
[0.0
12]
[0.0
92]
*[0
.089
]*[0
.011
][0
.012
][0
.006
][0
.007
]
Ser
vic
eS
har
e-0
.054
-0.0
43
0.00
10.0
000.1
510.
142
-0.0
13-0
.014
[0.1
47]
[0.1
38]
[0.0
07]
[0.0
07]
[0.1
42]
[0.1
42]
[0.1
20]
[0.1
20]
I R&
D-0
.254
-0.1
65
0.21
40.
202
-0.2
13-0
.169
0.03
80.
038
[0.3
51]
[0.3
47]
[0.3
50]
[0.3
73]
[0.2
78]
[0.2
76]
[0.1
84]
[0.1
84]
I Pate
nts
0.288
-0.0
060.
133
-0.0
76[0
.209
][0
.146
][0
.159
][0
.104
]∑ U
Sp
ate
nts
0.0
03
0.00
3-0
.017
0.00
1[0
.022]
[0.0
19]
[0.0
35]
[0.0
13]
∑ Ind
ian
pate
nts
0.051
-0.0
180.
019
-0.0
13[0
.014]
***
[0.0
08]*
*[0
.023
][0
.008
]*
Con
stant
0.74
00.7
43
-0.3
38-0
.360
0.62
0.63
9-0
.473
-0.4
84[0
.257
]***
[0.2
55]*
**
[0.1
47]*
*[0
.155]
**[0
.219
]***
[0.2
19]*
**[0
.195
]**
[0.1
98]
**
Nu
mb
erof
firm
s91
91
9191
104
104
104
104
R-s
qu
ared
0.20
0.2
70.
210.2
30.0
90.
090.
210.
21
Note
s:T
he
sam
ple
of
firm
sanaly
sed
(n=
91)
inco
lum
ns
[1]-
[4]
incl
udes
all
those
wit
hC
AR
for
both
even
tsand
isid
enti
cal
toth
at
for
the
blu
eand
red
arr
ows
inth
eupp
erplo
tof
Fig
ure
7.
The
sam
ple
of
firm
sanaly
sed
(n=
104)
inco
lum
ns
[5-8
]in
cludes
all
those
wit
hC
AR
for
both
even
tsand
isid
enti
cal
toth
at
for
the
blu
eand
red
arr
ows
inth
elo
wer
plo
tof
Fig
ure
7.
The
dep
enden
tva
riable
inth
eL
PM
regre
ssio
ns
in[1
],[2
],[5
]and
[6]
isa
dum
my
for
those
firm
sfo
rw
hic
hth
eir
CA
Ratt
+5
follow
ing
the
2005
even
t(A
pri
lin
[1]
and
[2],
June
in[5
]and
[6])
wassm
aller
than
thatt
+5
follow
ing
the
Dec
emb
er2004
even
t.In
the
OL
Sre
gre
ssio
ns
in[3
],[4
],[7
]and
[8]
the
dep
enden
tva
riable
isth
em
agnit
ude
of
diff
eren
ceb
etw
een
thes
etw
oC
AR
s(b
oth
tim
es2005
less
2004):
larg
ep
osi
tive
(neg
ati
ve)
valu
esin
dic
ate
asi
gnifi
cant
incr
ease
(dec
rease
)in
CA
R.
Som
eof
the
vari
able
sare
indic
ato
rsfo
rver
ysm
all
gro
ups:
firm
sw
hic
hca
rry
out
som
eR
&D
(n=
6&n
=7
firm
s),
whic
hhav
eso
me
pate
nt(
s)(n
=8
&n
=10
for
[5]-
[8]
and
[1]-
[4]
resp
ecti
vel
y),
whic
hhav
eU
.S.
pate
nts
(n=
3&n
=4),
and
whic
hhav
eIn
dia
npate
nts
(n=
7&n
=9).
Each
model
incl
udes
als
odum
mie
sfo
rm
issi
ng
obse
rvati
ons
rela
ted
tosa
les
(n=
3&n
=4
mis
sing),
firm
age
(n=
5)
and
serv
ices
share
(n=
7).
We
indic
ate
stati
stic
al
signifi
cance
at
the
10%
,5%
and
1%
level
usi
ng∗,∗∗
and∗∗∗
resp
ecti
vel
y.
21
magnitude of change in CAR (in either case 2005 less 2004 event), whereby a large pos-
itive (negative) value suggests a significant rise (drop) in relative CAR between the two
events.
As firm-level characteristics, we use data on sales, firm age, share of services in total
sales and innovation-related information (R&D expenditure dummy, patent dummy,
separate patent counts for USPTO and the IPO). Since the various events occur within
less than six months and our dependent variables are akin to time differences we only use
the 2004 information on firm characteristics in these regressions.20 Table 2 presents our
results: columns [1], [2], [5], and [6] adopt a linear probability model (LPM) to predict
the Idrop dummy.21 Columns [3], [4], [7], and [8] estimate a least squares model for the
‘∆CAR’ variable (these models equate to predicting the length and signs of all arrows
in Figure 7). It should be borne in mind that the sample size (columns [1]-[4]: n = 91,
columns [5]-[8]: n = 104) is moderate, and that the innovation indicators identify only
a small subset of these firms.
Acknowledging this caveat, our regressions indicate that larger firms (in terms of
sales volume) ceteris paribus were less significantly affected by the two events tightening
patentability – it should however be noted that our sample only consists fairly large
publicly traded companies. The LPM results in columns [1] and [2] indicate that larger
firms were less likely to witness a significant drop in CAR between the 2004 and April
2005 events. The coefficients on sales are also negative in columns [5] and [6] (the drop in
CAR between the December 2004 and June 2005 events) but not statistically significantly
different from zero. The least squares results adopt the continuous difference between
CARs as their dependent variable, which varies from large negative to large positive
values: thus a positive coefficient on the log sales variable in models [3], [4], [7] and [8]
suggests that size is associated with positive or less negative values. Firm age or services
share22 are not consistently correlated with a drop in CAR or the change in CAR. The
coefficients on patent filings at the IPO suggest companies that filed for patents before
the 2004-2005 sequence of policy shocks were more likely to see a drop in market returns
when the ordinance was undone and when the revised guidelines were issued.
Our analysis of market returns shows some clear patterns highlighting the positive
market response to a softening of patenting restrictions in December 2004 and a negative
response once these restrictions were reinforced in June 2005. We further showed that
20Our results are qualitatively identical if we adopt the 2005 information instead.21A nonlinear model (logit) yields qualitatively identical results.22This measure is fairly concentrated: 15% of firms for which we have this information have zero share
of services in total sales, while around 7% have between 2% and 90%, leaving 78% with virtually noproduct-related sales.
22
the market indicated a heterogeneous response to the former event, which explains the
counter-intuitive positive average market response to the undoing of the ordinance in
April 2004. That is, companies that did not experience positive returns when the ordi-
nance was announced, experienced positive returns when it was undone. In contrast, the
negative effect of the tightening of patenting restrictions through the revised guidelines
on those software companies that had experienced positive returns when the ordinance
was announced drowned out the more modest positive reaction by the other companies.
Investigation of the characteristics of firms reveals that comparatively smaller software
companies were most adversely affected by the tightening of patentability.
5.3 Software Type
Our previous analysis provided insights into the patenting behavior and market valuation
of software firms in response to the series of policy shocks. We now turn to our product-
level analysis where we investigate whether companies change the share in software
services in total sales as a reaction to the changes in the patentability of software.
To this end we use sales data for our sample of companies that distinguishes between
sales from software products and software services.23 We also assembled information
on the detailed software products and services offered by the companies in our sample
from companies’ websites; however, these data do not include sales at such disaggregate
levels.24 Figure 8 plots companies’ average share of services in total annual sales over
the entire 2000-2011 period.
The figure shows that services account for a very high share of between 81% and
85% of total sales value, underscoring the emphasis of the Indian software industry on
software services as opposed to selling (generic) software packages. There is an increase
in the share of services between 2001 and 2004 and a drop between 2004 and 2005.
From 2006 onward, however, we see again an increase in the share of services in total
sales until 2009. These patterns suggests that the ordinance coincided with an increased
focus on software products whereas its undoing, in particular the issuance of the revised
guidelines in June 2005 which clarified the restrictions on the patentability of software,
coincides with an increase in software services among total sales.
It is tempting to conclude from Figure 8 that the Indian Patent Office’s affirmation
of the restrictions on software patentability has led companies to focus even more on
23We complemented the Prowess data with data collected manually directly from companies’ annualand financial reports.
24Moreover, at such disaggregate level, in most cases we only have information for a snapshot in timeand therefore cannot track changes in the range of products/services offered by companies over time.
23
Figure 8: Share of services in total sales, 2000-2011
RestrictingAmendment
Ordinance
UndoingOrdinance
Revised
Revised
Guidelines
Guidelines
.81
.82
.83
.84
.85
Ave
rage
sha
re s
ervi
ces
in to
tal s
ales
2000 2002 2004 2008 20102005
Notes: Figure shows average share of services in total sales for 212 software companies that reportproduct-level data before and after the 2004 Ordinance and the 2005 undoing of the Ordinance and the issuanceof the revised guidelines.
24
the provision of software services. To investigate this further, we use the product-level
data to estimate specification (4) and (5) using the sample of control companies. Our
main interest is in the interaction term softwarei×I(t≥τ) which captures any differential
effect of the policy events on the share of software services in total sales. To account
for time-varying firm-level characteristics, we include age, R&D intensity and total sales
among the regressors.
Table 3: Share of software services in total sales
Dep. Var.: Share ofservices in total sales [1] [2] [3] [4] [5] [6]
I(t≥2002) -0.011
[0.039]I(t≥2005) -0.011
[0.041]I(t≥2008) -0.011
[0.042]Software × I(t≥2002) -0.037
[0.025]Software × I(t≥2005) -0.013
[0.021]Software × I(t≥2008) 0.002
[0.020]Software 0.121** 0.094* 0.085*
[0.058] [0.053] [0.050]
Log Age 0.025 -0.090*** 0.025 -0.090*** 0.025 -0.090***[0.044] [0.035] [0.044] [0.035] [0.044] [0.035]
R&D Intensity 0.024 0.110 0.024 0.108 0.024 0.102[0.454] [0.451] [0.454] [0.452] [0.454] [0.453]
Log Sales -0.004 -0.001 -0.004 -0.001 -0.004 -0.001[0.008] [0.006] [0.008] [0.006] [0.008] [0.006]
Year dummies YES YES YES YES YES YESObs. 1,942 2,608 1,942 2,608 1,942 2,608Number of firms 208 286 208 286 208 286
Notes: Software is equal to one if firm i is in the software industry. Regressions include dummyvariables for missing data on age, R&D intensity and sales (results not reported). Regressions cover2000-2010 period. Marginal effects shown. Robust standard errors clustered at the firm-level inparentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
Table 3 shows the corresponding estimates for the 2002, 2005 and 2008 policy shocks
25
(Table A-4 in the online Appendix shows comparisons of means). We are unable to look
at the 2004 Ordinance because it occurred in December 2004 and hence coincides with
the 2005 events in the data that is available only on a yearly basis. Similarly we only
have a single year of data following the 2010 event and therefore limit our attention
instead to the period 2000-2010. Columns [1], [3], and [5] show results from estimating
specification (4) on the sample restricted to software companies. Columns [2], [4], and
[6] show the results when we add the sample of control firms and estimate specification
(5).
The results for specification (4) produce negative coefficients for all three policy
shocks – but the estimates are clearly not close to conventional levels of statistical sig-
nificance. When we look at the estimates obtained from using the control sample of
firms, the coefficients on the treatment variables (the interaction between the different
policy shocks and the software dummy) vary across the different policy shocks, but again
none of the coefficients is statistically significant. In contrast the dummy variable for
software companies on its own is positive and statistically significant, reflecting the fact
that these companies have on average a higher share of services in total sales. These re-
sults, therefore, do not offer any evidence in support of a switch from software products
to services as a response to a tightening of the restrictions imposed on software patents.
To account for the possibility that firms which rely on patents reacted differently to
the various policy shocks, Table 4 distinguishes between companies that have applied
for at least one patent (domestic or U.S.) before a policy shock and all non-patenting
firms.25 As in Table 3, columns [1], [3], and [5] show results from estimating specification
(4) whereas columns [2], [4], and [6] show the estimates from specification (5). Table 4
shows that our conclusions remain unaltered for the 2002 and 2008 policy shocks – there
is no significant change in the share of software services following these policy shocks
even if we restrict our attention to patenting firms. However, looking at the estimates
in column [4], we find a positive coefficient on the treatment variable for companies that
patented before 2005. This indicates a positive association between the restrictions on
the patentability imposed on software in the course of 2005 and the share of services
in total sales specifically for companies that had patented beforehand. It is also worth
mentioning that in columns [1] and [3] we observe a positive and statistically significant
coefficient on the dummy that captures any effects for software firms that have patented
before 2002 and 2005, respectively. In combination, these results offer some, albeit
tentative, evidence that in particular software companies that have been hoping to rely
25Distinguishing between Indian and U.S. patents makes no qualitative difference for the results shownin Table 4.
26
on patents to protect their software increased the share of software services in total sales
once it became clear that software would remain unpatentable. This result is also inline
with the evidence found for the number of patent filings with the Indian Patent Office
in Table 2 above.
Table 4: Share of software services in total sales
Dep. Var.: Share ofservices in total sales [1] [2] [3] [4] [5] [6]
I(t≥2002) × Patent (t < 2002) -0.201
[0.144]I(t≥2005) × Patent (t < 2005) -0.096
[0.116]I(t≥2008) × Patent (t < 2008) -0.009
[0.021]Software ×I(t≥2002) × Patent (t < 2002) 0.092
[0.072]Software ×I(t≥2005) × Patent (t < 2005) 0.130***
[0.040]Software ×I(t≥2008) × Patent (t < 2008) 0.035
[0.032]Software 0.120** 0.086 0.090*
[0.058] [0.053] [0.052]Patent (t < 2002) 0.108*** -0.022
(0.039) [0.135]Patent (t < 2005) 0.144*** 0.030
[0.027] [0.068]Patent (t < 2008) -0.011 -0.025
[0.071] [0.069]
Log Age 0.022 -0.092** 0.012 -0.093*** 0.026 -0.088**[0.046] [0.036] [0.045] [0.035] [0.045] [0.035]
R&D Intensity 0.023 0.085 -0.340 -0.316 0.047 0.105[0.449] [0.452] [0.472] [0.488] [0.458] [0.457]
Log Sales -0.004 -0.002 -0.010 -0.005 -0.003 -0.001[0.008] [0.007] [0.009] [0.007] [0.009] [0.007]
Year dummies YES YES YES YES YES YESObs. 1,942 2,608 1,942 2,608 1,942 2,608Number of firms 208 286 208 286 208 286
Notes: Software is equal to one if firm i is in the software industry. Patent is equal to one if firm ifiled for at least 1 U.S. and/or Indian patent before the corresponding policy shock. Regressionsinclude dummy variables for policy events (I(t≥τ) and Software × I(t≥τ) ), missing data on age, R&Dintensity and sales (results not reported). The coefficients on policy events are statisticallyinsignificant. Regressions cover 2000-2010 period. Marginal effects shown. Robust standard errorsclustered at the firm-level in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
27
6 Conclusion
The question of the effect of patent rights on innovation is highly controversial, especially
so in a developing country context. Even within this contentious debate, the controversy
surrounding the patentability of software stands out. Tensions currently run high in the
U.S. where some argue that software patents are at the heart of current problems con-
fronting the patent system as evidenced by the so-called ‘smartphone patent wars’, the
steep increase in litigation by ‘patent trolls’ and the emergence of patent mass aggrega-
tors.26 Software patents further have been shown to be the main driver behind ‘patent
portfolio races’ (Hall and Ziedonis, 2001) and ‘patent thickets’ (Hall et al., 2012). Our
study directly contributes to this debate and offers insights on the impact of restrictions
imposed on the definition of patent eligible subject matter in an emerging economy
context.
We study the effect of (announced) changes in the patent law and regulations that
affected the patentability of software on all publicly listed software companies in India.
We find that software companies file a relatively large number of patent applications
with the Indian patent office despite of the restriction on patent eligibility of software.
That said, a closer look at the legal status of these patent filings reveals that practically
no filings have been granted. In contrast, the majority of patent filings by the same
companies in the U.S. are granted, which highlights the restrictive approach in the ex-
amination of software filings by the Indian patent office. In our event study, we find that
software companies experienced large positive stock market returns (‘abnormal returns’)
in response to the government’s announcement to substantially loosen the restrictions
imposed on the patentability of software. This suggests that markets expected software
companies to benefit from an improved ability to protect their software through patents.
These positive returns might have reflected the expectation that Indian software com-
panies would move away from (customized) software services toward the production of
more generic software packages, for which patent protection is more important. However,
analysis at the product-level which allows us to distinguish between software products
and services provides no robust evidence that Indian software companies increased their
focus on software services at the expense of software products as a reaction to the con-
tinued restrictions imposed on software patentability. That said, there is some evidence
that companies that applied for patents before the policy reversals in 2005 increased
their share of services in total sales in response to the tightening of the restrictions on
the patentability. In sum, our analysis suggests that the restrictions on the patentability
26See for example The Economist (Dec 13th 2013) “Obituary for software patents.”
28
of software have been effective in keeping publicly traded Indian software companies
from obtaining patent rights on software and there is little evidence that the absence of
patents has impacted the type of software companies are producing.
29
References
Arora, A. (2008): “The Indian software industry and its prospects,” in Sustaining
India’s Growth Miracle, ed. by J. Bhagwati and C. Calomiris, Columbia Business
School Press.
Arora, A., V. S. Arunachalam, J. Asundi, and R. Fernandes (2001): “The
Indian Software Services Industry: Structure and Prospects,” Research Policy, 30,
1267–1288.
Arora, A., L. Branstetter, and C. Chatterjee (2011): “Strong Medicine: Patent
Reform & the emergence of a Research-Driven Pharmaceutical Industry in India,” in
The Location of Biopharmaceutical Activity, ed. by C. I. M. and M. J. Slaughter,
NBER.
Arora, A., L. Branstetter, C. Chatterjee, and K. Saggi (2009): “Strong
Medicine? Patent Reform and the Transformation of the Indian Pharmaceutical In-
dustry,” mimeo.
Athreye, S. S. (2005): “The Indian software industry and its evolving service capabil-
ity,” Industrial and Corporate Change, 14, 393–418.
Bessen, J. and R. Hunt (2007): “An Empirical Look at Software Patents,” Journal
of Economics & Management Strategy, 16, 157–189.
Campbell, J. Y., A. A. W.-C. Lo, and C. MacKinley (1997): The econometrics
of financial markets, princeton University press.
Chen, Y. and T. Puttitanum (2005): “Intellectual Property Rights and Innovation
in Developing Countries,” Journal of Development Economics, 78, 474–493.
Economist, T. (2011): “Indian technology firms: Seeking to avoid a mid-life crisis,”
Tech. rep., nov 5th 2011, print edition.
ESC (2012): Statistical Yearbook 2010/11, Electronics and Computer Software Export
Promotion Council.
FTC, F. T. C. (2011): “The Evolving IP Marketplace: Aligning Patent Notice and
Remedies with Competition,” Federal Trade Commission Policy Report.
30
Graham, S. and D. Mowery (2003): Patents in the Knowledge-Based Economy,
Washington, D.C.: National Academies Press, chap. Intellectual Property Protection
in the U.S. Software Industry.
Graham, S. and S. Vishnubhakat (2013): “Of Smart Phone Wars and Software
Patents,” Journal of Economic Perspectives, 27, 67–86.
Grossman, G. M. and E. L.-C. Lai (2004): “International Protection of Intellectual
Property,” American Economic Review, 95, 1635–1653.
Hall, B., C. Helmers, V. Sena, and M. Rogers (2014): “The choice between
formal and informal intellectual property: a review,” Journal of Economic Literature,
52, 375–423.
Hall, B., C. Helmers, G. von Graevenitz, and C. Rosazza-Bondibene (2012):
“Technology Entry in the Presence of Patent Thickets,” mimeo.
Hall, B. and M. MacGarvie (2010): “The private value of software patents,” Re-
search Policy, 39, 994–1009.
Hall, B. H. and R. Ziedonis (2001): “The patent paradox revisited: an empirical
study of patenting in the U.S. semiconductor industry, 1979-95,” RAND Journal of
Economics, 32, 101–128.
Helpman, E. (1993): “Innovation, Imitation, and Intellectual Property Rights,” Econo-
metrica, 61, 1247–1280.
Hu, A. and I. Png (2013): “Patent rights and economic growth: Evidence from cross-
country panels of manufacturing industries,” Oxford Economic Papers, 65, 675–698.
Hu, G. A. and G. Jefferson (2009): “A great wall of patents: What is behind
China’s recent patent explosion?” Journal of Development Economics, 90, 57–68.
Jaffe, A. and J. Lerner (2004): Innovation and Its Discontents: How Our Broken
Patent System is Endangering Innovation and Progress, and What to Do About It.
Kanwar, S. and R. Evenson (2003): “Does intellectual property protection spur
technical change?” Oxford Economic Papers, 55, 235–264.
Kyle, M. and Y. Qian (2013): “Intellectual Property Rights and Access to Innovation:
Evidence from TRIPS,” mimeo.
31
Lemley, M. (2012): “Software Patents and the Return of Functional Claiming,” Stan-
ford Public Law Working Paper No. 2117302.
Lerner, J. (2002): “150 Years of Patent Protection,” American Economic Review, 92,
221–225.
Lerner, J. and F. Zhu (2007): “What is the impact of software patent shifts? Ev-
idence from Lotus v. Borland,” International Journal of Industrial Organization, 25,
511–529.
MacKinlay, A. C. (1997): “Event studies in economics and finance,” Journal of eco-
nomic literature, 35, 13–39.
Moser, P. (2005): “How Do Patent Laws Influence Innovation? Evidence from
Nineteenth-Century World’s Fairs,” American Economic Review, 95, 1214–1236.
——— (2012): “Innovation without Patents - Evidence from the World Fairs,” Journal
of Law and Economics, forthcoming.
Niosi, J. and F. T. Tschang (2009): “The strategies of Chinese and Indian software
multinationals: implications for internationalization theory,” Industrial and Corporate
Change, 18, 269–294.
Nordhaus, W. D. (1969): Invention, Growth, and Welfare: A Theoretical Treatment
of Technological Change, MIT Press.
Papke, L. E. and J. M. Wooldridge (1996): “Econometric Methods for Fractional
Response Variables with an Application to 401 (K) Plan Participation Rates,” Journal
of Applied Econometrics, 11, 619–32.
Penrose, E. (1973): “International Patenting and the Less-Developed Countries,” The
Economic Journal, 83, 768–786.
Qian, Y. (2007): “Do additional national patent laws stimulate domestic innovation in
a global patenting environment: A cross-country analysis of pharmaceutical patent
protection, 19782002,” Review of Economics and Statistics, 89, 436–453.
Sakakibara, M. and L. Branstetter (2001): “Do stronger patents induce more
innovation? Evidence from the 1988 Japanese patent law reforms,” RAND Journal of
Economics, 32, 77–100.
32
Sampat, B. (2010): “Institutional Innovation or Institutional Imitation? The Impacts
of TRIPs on India’s Patent Law and Practice,” mimeo.
Scherer, F. and S. Weisburst (1995): “Economic Effects of Strengthening Pharma-
ceutical Patent Protection in Italy,” International Review of Industrial Property and
Copyright Law, 26, 1009–1024.
UNCTAD (2006): Information Economy Report 2006: The development perspective,
United Nations Conference for Trade and Development.
——— (2012): Information Economy Report 2012: The Software Industry and Devel-
oping Countries, United Nations Conference for Trade and Development.
33
A Online Appendix: Patent data
While the patent filings by Indian companies at the USPTO are readily available in
EPO’s PATSTAT database, patent filings with the Indian Patent Office have to be
assembled using three databases. First, the Indian Patent Offices’ electronic patent
search facility iPairs provides information on published patents from 2005 onwards. For
patent filings before 2005, we rely on the EKASWA database assembled by the Patent
Facilitating Centre (PFC). EKASWA contains all domestic patents published between
January 1995 and early 2005. We complement the dataset on domestic filings by using
the online portal BigPatents India. The patent data are matched to the Prowess firm-
level data. Due to the absence of a unique identifier shared by the firm-level and patent
data, we matched patent and firm-level data based on a combination of an automated
name-based matching algorithm and extensive manual matching and checking.
B Online Appendix: Figures
Figure A-1: Patenting companies – Indian Patent Office vs USPTO, 2000-2011
i
Figure A-2: Legal status of patent filings at the Indian Patent Office – 2000-2011
RestrictingAmendment
Ordinance
UndoingOrdinance
Revised
Revised
Guidelines
Guidelines
020
4060
80N
umbe
r of
pat
ent f
iling
s
2000q1 2001q3 2003q1 2004q3 2006q1 2007q3 2009q1 2010q3
Examination not yet requested GrantedPending RejectedWithdrawn
Notes: The legal status was obtained from the iPairs website of the Indian Patent Office in February 2014.
ii
Figure A-3: Legal status of patent filings at the USPTO – 2000-2011
RestrictingAmendment
Ordinance
UndoingOrdinance
Revised
Revised
Guidelines
Guidelines
05
1015
Num
ber
of p
aten
t fili
ngs
2001q1 2002q3 2004q1 2005q3 2007q1 2008q3 2010q1 2011q3
Withdrawn GrantedPending Rejected
Notes: The legal status was obtained from the PAIR website of the USPTO in February 2014.
iii
Figure A-4: CAR for patenting and non-patenting firms
Notes: Each plot compares the CAR for patenting versus non-patenting firms for the five ‘events’ outlinedabove (event window of 11 days). We label a firm as patenting if it has filed for a patent with the Indian PatentOffice or USPTO prior to each respective event date.
iv
C Online Appendix: Tables
Table A-1: Descriptive statistics – product-level regression sample, 2000-2010
Variable Mean Median St. dev. Min Max Obs
Software companies# patents Indian Patent Office 0.29 0 3.22 0 71 1,942# patents USPTO 0.11 0 0.97 0 28 1,942Age 15.30 14.00 7.45 1 66 1,942R&D intensity 0.002 0 0.02 0 0.54 1,942Total assets (100mio Rs) 44.40 3.60 214.29 0 3,443.50 1,942Sales (100mio Rs) 36.21 1.70 199.98 0 2,927.70 1,892
Control sample# patents Indian Patent Office 0.02 0 0.18 0 3 666# patents USPTO 0 0 0 0 0 666Age 16.10 12.00 12.94 2 79 666R&D intensity 0.001 0 0.005 0 0.45 666Total assets (100mio Rs) 345.09 15.28 1348.00 0 15,603.62 666Sales (100mio Rs) 133.68 12.43 424.68 0.2 38,017.70 615
v
Table A-2: Companies patenting in India (in alphabetical order), 2000-2011
Company 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
3I Infotech 1Aftek 1Bartronics India 1Bodhtree Consulting 1Cranes Software Intl. 2Flextronics Software Systems 1 2Geodesic 1 3 1 1 1HCL Technologies 1 3 3 3 8 30Hinduja Global Solutions 1ICSA (India) 1Infosys 1 14 31 24 69 21 48 163Infotech Enterprises 2 1Intellvisions Software 2KLG Systel 2 4 8KPIT Cummins Infosystems 2 3 6 11 10Mindtree 3 1 7 1Mold-Tek Technologies 1Mphasis 1NIIT 3 2 2 1 2 2 1Odyssey Technologies 1Om Energy 1Onmobile Global 4 9 2Persistent Systems 2 1 1 1 2Polaris Financial Technology 1 1Prithvi Information Solutions 2Ramco Systems 11Rolta India 1Sasken Communication Technologies 1 1 1 2 2 1 1 1 3Satyam Computer Services 3Tanla Solutions 2 2Tata Consultancy Services 5 3 2 6 3 8 10 19 30 63 71 185Tata Elxsi 9 9 1 1Tata Infotech 2 2Wipro 1 1 1 3 1 3 2Zensar Technologies 1 1
vi
Table A-3: Companies patenting in the US (in alphabetical order), 2000-2011
Company 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Encore Software 1Flextronics Software Systems 1Frontline Business Solutions 1Geodesic 2Geometric 1 1Infosys 1 3 17 28 7KPIT Cummins Infosystems 2 1Mascon Global 1 1Megasoft 2 1Mindtree 3Moschip Semiconductor Technology 2Onmobile Global 1 4 1 1 1Oracle Financial Services Software 2Patni Computer Systems 1 1Ramco Systems 1 6 2 2 1Sankhya Infotech 2 1 1Sasken Communication Technologies 1 6 6 4 2 2 2 1 2 4 1Satyam Computer Services 1 1 4 1 4 3 3 8 7Tanla Solutions 1 2 1Tata Consultancy Services 5 2 1 2 4 3 5 1 3 8Tech Mahindra 1Wipro 1 1 9 8 4 1
Table A-4: Before and after comparison of means – 2002, 2005, 2008 policy shocks,2000-2010
Variable Software companies Control sample
≤ 2002 > 2002 t-test† ≤ 2002 > 2002 t-test†
Share of services in sales 0.840 0.850 0.571 0.718 0.772 0.127
≤ 2005 > 2005 t-test† ≤ 2005 > 2005 t-test†
Share of services in sales 0.845 0.849 0.804 0.752 0.768 0.600
≤ 2008 > 2008 t-test† ≤ 2008 > 2008 t-test†
Share of services in sales 0.846 0.855 0.626 0.762 0.753 0.804
† P-value.
vii
D Online Appendix: India’s software industry
The history of the Indian software industry is a remarkable success story of a shift
from an original comparative advantage in factor endowment (cheap educated labour)
in the 1990s to a ‘created’ comparative advantage in services outsourcing enabled by
organizational capability less than a decade later (Athreye, 2005). Like many other
industries the software and IT services sector benefited from full financial liberalization
and increased openness in the early 1990s, which together with the Software Technology
Parks government infrastructure scheme (Athreye, 2005, refers to the role of the state
up to this point as ‘benign neglect’) enabled domestic firms such as Wipro and TCS to
answer the phenomenal growth in global demand for software services by establishing
dedicated Offshore Development Centres for overseas clients. During this phase the
dominant business model however had not yet entirely moved on from the ‘body-
shopping’ of the early export phase in the 1980s, when Indian firms simply provided
programmers on-site for whatever jobs overseas clients needed done (Arora, 2008). For
the time being the work carried out was limited to maintaining the software and systems
designed by others (Economist, 2011) and the dollar value of service contracts was still
modest, on average only around US$100,000 (Arora et al., 2001; Arora, 2008), although
then and up until today the lion’s share of the industry was produced for export.
Throughout this phase, however, following a period of experimentation, major Indian
players in the sector – see Table A-5 for the Top 5 firms – identified specific industrial
niches for themselves (firm-specific capability in servicing the finance and insurance
or the telecommunications industries) and the sector as a whole pushed for increased
quality certification Athreye (2005) — both factors which enabled these companies
to weather the period of slack demand following the burst of the dot-com bubble in
the early 2000s and emerging from this experience as key actors in a consolidated
industry.27 Before this episode, the late 1990s brought severe challenges for the domestic
industry, with a scarcity of experienced software engineers, rising wages and high levels
of attrition (up to 40% employee turnover per annum) forcing Indian software firms to
choose the ‘high road’ to export competitiveness rather than to continue its reliance
on a low cost model: domestic firms developed organisational capabilities to operate
within this challenging environment, building on in-house training and the adoption of
formalised procedures to ensure worker attrition did not translate into loss of knowledge
for the company Athreye (2005). The unique ‘created’ comparative advantage of Indian
27Other important factors included improved access to venture capital and a lifting of restrictions onprivate investment in Indian telecommunications, which dramatically reduced the cost of connecting tothe internet for these firms (Athreye, 2005).
viii
software and IT services firms relates to their capacity to ‘ramp up’ and ‘ramp down’
specific projects quickly in response to client demand, enabled by their expertise in
how to split tasks between high- and low-capability workers as well as their ability in
managing global operations (Arora, 2008). By the time the leadup to the Y2K bug
forced many Western companies to update their software systems (Niosi and Tschang,
2009) the primary business model in the Indian software and IT services sector had
moved away from body-shopping (on-site work) and toward an increased focus on
off-shore operation (i.e. India-based software services): between 1999/2000 and 2002/3
the share of revenue from off-shoring operations increased from 39% to 57% (Athreye,
2005). Individual, multi-year service contracts secured by Indian firms in the 2000s now
ran into the tens of millions or even hundreds of millions (Athreye, 2005; Arora, 2008),
a far cry from the mere scraps of the late 1990s described in Arora et al. (2001).
Table A-5: Top-5 Indian Software/IT Services providers (by export value)
Rank Firm Exports (2010/11) Global Workforce (2012)
1 Tata Consultancy Services (TCS) $5.1bn 254,0002 Infosys $4.5bn 155,6293 Wipro $3.5bn 140,0004 Cognizant∗ $3.3bn 156,7005 HCL $2.1bn 85,194
Notes: Ranking and exports taken from ESC (2012), global workforce in 2012 from company websites.* indicates foreign firms/MNCs.
Today (FY2010/11) the industry delivers the outsourced software requirements for
over half of all U.S. Fortune 500 companies, contributes around 4.3% to domestic GDP
and employs around 2.5 million mostly skilled ‘knowledge’ workers, with the top three
firms (TCS, Infosys and Wipro) accounting for half a million employees alone. Over
the past decade a steady 80% of total production in the Indian computer software and
services industry has been exported, amounting to around US$58bn of exports in 2010,
almost five times the 2003 figure of US$12.6bn — see Table A-6 for sectoral evolution of
production and exports. In terms of global players in the US$ 960bn world software/IT
services market, India ranks fourth behind the U.S. (market share: 39%), Japan (12%)
and the EU (8%), but ahead of its fellow BRIC China (6%) (all figures for 2010/11,
taken from ESC, 2012). The local industry association NASSCOM further estimates
that the potential market for software and IT services outsourcing will triple by 2020
(Economist, 2011).
ix
Table A-6: Evolution of the Indian Computer Software and Services IndustryProduction and Exports (in million US$)
Year 2003 2004 2005 2006 2007 2008 2009 2010
Total Production 16,141 22,058 30,404 42,312 55,144 61,984 64,956 74,890Annual growth rate 37% 38% 39% 30% 12% 5% 15%
(a) Exports 12,609 17,216 23,718 33,757 43,467 49,540 51,001 57,616Annual growth rate 42% 37% 38% 42% 29% 14% 3% 13%ITES/BPO 9,480 12,295 13,700 14,418 16,572w/out ITES/BPO 24,277 31,172 35,840 36,583 41,044World exports 73,153 108,200 130,800 160,900 201,400 190,300 215,200
(b) Domestic market 3,532 4,842 6,686 8,555 11,677 12,444 13,955 17,274Share of production 22% 22% 22% 20% 21% 20% 21% 23%
Notes: All figures in million US$ (nominal) for the financial year with the exception of world exports(for the calendar year). Definitions: Total Production: Computer Software/Services; Exports:Computer Software/Services (including ITES, IT-enabled Services, and BPO, Business ProcessOutsourcing); Exports of ITES/BPO: IT-enabled Services and Business Process Outsourcing; Worldexports: Computer and Information Services (excluding ITES/BPO) – note that these figures are notdirectly comparable due to likely discrepancy in the definition of computer software and services.Sources: Indian data from ESC (2012) Statistical Yearbook 2010/11, the ESC website and UNCTAD(2006) for 2000-2003; world exports from UNCTAD (2006, 2012).
This historical overview and present competitive state of the sector is however marred
by continued worries regarding the sustainability of the industry’s growth and corporate
strategy (Arora, 2008). High-end software ‘solutions,’ packages of software and IT ser-
vices which range from design of application to advising clients on how to restructure
their operations are estimated to account for a mere 10% of revenues today (Economist,
2011). As discussed in Arora (2008) there are three sets of activities to add value in
software production: (a) software design and development, with household names such
Microsoft or Adobe typical examples; (b) development and design of bespoke client soft-
ware, typified by German software giant SAP; and (c) services to ‘user industries,’ e.g.
banking and finance institutions, healthcare providers, or public institutions. It is in the
latter domain that most of India’s software providers operate in, successfully substituting
the provision of these services by in-house divisions within firms in these industries. The
importance of this third type of operations cannot be underestimated, with around two
thirds of software professionals in the U.S. suggested to work for IT-using rather than IT
firms. This corporate strategy may be seen as the root of the problems in securing the
future sustainability of the Indian software industry: while other recent export success
stories like Israel and Ireland have focused on fostering start-up software development
x
companies (the biggest of which are listed on U.S. stock markets but maintaining their
primary R&D in Israel) and hosting MNC subsidiaries, respectively, Indian firms stuck
to customised software services, often relatively low-value work and with revenue per
worker substantially lower than in other activities (Athreye, 2005). Since Indian firms
still primarily operate as contractor services to user-industries they face a natural bar-
rier in moving into the higher value end of the software value-chain as their clients are
unwilling to hand over this sensitive part of their business to outsiders or because Indian
firms do not have the necessary human resources to carry out these complex tasks (Niosi
and Tschang, 2009). In recent times, only one Indian firm, i-Flex, a provider of high-end
banking software until its acquisition by Oracle in 2006, has been able to successfully de-
velop own-brand software products, although a number of firms (in particular domestic
industry-leader TCS) have engaged in co-development of products with clients while a
considerable number has engaged in strategic acquisition to complement organic growth
(see in particular Table 3 in Niosi and Tschang, 2009).
xi
