The Software Dilemma

The Legal Classification of Software in India

Interestingly, it is often the most ubiquitous commercial commodity that poses the greatest difficult to the law. In many cases, this is because it is difficult to “classify” the commodity in question under existing legal pigeonholes, and yet necessary to do so in the interests of applying a coherent regulatory framework to it. Thus it is with software. Without attempting to be exhaustive, this post discusses developments primarily in Indian law over the past few years on various aspects of classifying software transactions.

Such an account must begin with the decision of a Constitution Bench of the Supreme Court in Tata Consultancy Services v. State of Andhra Pradesh. The decision is an extremely influential one, and arose out of a sales tax dispute concerning the taxability of software sold to customers of TCS in connection with services that the company rendered. Frequently, these products were “off-the-shelf” or “branded” software, manufactured on a large scale without a significant degree of customization. Such companies also “sell” software that is heavily customised to the needs of a particular consumer – referred to commonly as “bespoke” software. In both cases, consumers are usually asked to enter into an “End User Licence Agreement” [“EULA”] as a precondition for using the software. However, the Supreme Court in TCS was concerned with the former but not with the latter – a point that has significance for the scope of its observations in the case.

TCS challenged the levy of sales tax on the theory that the transfer of off-the-shelf software does not constitute a “sale”. There are two arguable points to support this contention in any case: first, that software “inherently” does not constitute “goods” for it predominantly represents intellectual property; and secondly, that even if it does, its transfer does not constitute a “sale”, because the rights accorded to the end-user are so severely limited as to militate against such a characterization. In TCS, the judgment reveals that the petitioner argued the first point, but not the second.

Consequently, the Constitution Bench’s analysis is mainly directed towards ascertaining whether software is inherently goods. The dominant view in several jurisdictions is that it does, for any entity represents underlying intellectual input manifested in a suitable physical form. This test, of course, is not very helpful when software is transferred without a physical medium. However, the Constitution Bench found that Indian law on the meaning of “goods” does not attach importance to the medium of transfer, adopting as it does the functional test of whether the entity in question is capable of “abstraction, consumption, transmission, transfer etc.” These software is undoubtedly capable of – and thus the question was answered against the petitioner. Paragraph 29 of the majority opinion is of vital significance, for the court notes that although there is “no distinction between branded and unbranded software”, it does not follow necessarily that the transfer of unbranded software is a sale because “other questions like situs of contract of sale and/or whether the contract is a service contract may arise”. As the Court was not required to classify unbranded software, it left this question open, and paragraph 29 continues to be a source of some ambiguity. Since the petitioner did not argue the second point we discussed above, no opinion was expressed on that at all.

I have argued elsewhere that the Supreme Court, with respect, may need to reconsider this issue because its first conclusion that software constitutes goods, although correct, does not conclude the matter. The reason, quite simply, is that the typical EULA transfers to the end-user very limited rights of use – many concede that it is not symptomatic of a typical sale contract. However, there is opposition to this view as well, and an analysis of the decisions of courts in England and America shows that the question is a close one. Intriguingly, courts and tribunals in India appear to have recognised in other contexts that the typical software transfer agreement may only constitute a limited licence.

To better understand the importance of this point, it is useful to refer back to the famous decision in State of Madras v. Gannon Dunkerley, and to the opinion of the legendary Justice Venkatarama Iyer. At issue was whether the State Government could charge the “sale” of cement to sales tax, when it occurred in the course of an architect building a house for a client. Clearly, neither the architect nor the client viewed the contract as “the sale of cement to be employed for the purpose of constructing a building” but as “a service contract for the construction of a house”. The Supreme Court held therefore that it is not open to tax authorities to “vivisect” or “dissect” a “composite” contract unless it is shown that the parties themselves intended to enter into two separate contracts. It further laid down three ingredients that any transaction must fulfill to qualify as a sale – the existence of “goods”, an intention to transfer “title” to those goods, and the consequent or actual transfer of title. A line of decisions applying Gannon Dunkerley made the principle a firm part of Indian jurisprudence, and parties began to use it as a means of tax avoidance (not evasion). For example, a petitioner who entered into a “99 year lease” successfully argued that the transaction was not a sale, relying on Gannon Dunkerley.

The decision was criticised by the Law Commission, and the 46^th Constitutional Amendment was passed to partially modify it. It inserted Art. 366(29A) into the Constitution, which provides that “tax on goods” include six specific transactions listed in sub-clauses (a)-(f). Each clause nullified a particular decision of the Supreme Court applying Gannon Dunkerley – for example, it is now possible to levy sales tax on the “sale” of food served at a restaurant. While the constitutional validity of the amendment was upheld in Builders Association, the Supreme Court (per Ruma Pal J.) noted in paragraph 29 of BSNL v. Union of India that Gannon Dunkerley is good law in every situation save the ones specifically enumerated in Art. 366(29A) [emphasis mine].  In the case of software, we must turn to Art. 366(29A)(d), which provides for sales tax on “the transfer of rights to use goods”. This appears to meet any objection based on the limited nature of a EULA. Or does it?

In a further twist, Parliament added s. 65(105)(zzzze) to the Finance Act in 2008, proposing to levy service tax on “any service provided or to be provided … in relation to information technology software…” The “development” of such software was specifically listed in the Act. Writ petitions were filed in several High Courts challenging the constitutional validity of this provision. Ironically, these petitions relied on TCS to argue that since the typical software transaction has been held to be a “sale”, it is within the exclusive legislative competence of State Legislatures, and therefore outside the purview of service tax. On 24 August, 2010, the Madras High Court rejected this challenge (hat tip: Shantanu Naravane), and I have posted a detailed account of the decision here.

The Union of India argued before the Madras High Court that a typical software transaction need not constitute a sale because the terms of the EULA that accompanies the software is characteristic of a licence – in short, the argument the petitioner in TCS could have advanced to take software beyond the purview of sales tax. The petitioner relied on Art. 366(29A)(d) in response. The High Court rejected the petitioner’s contention with a somewhat ambiguous conclusion in paragraph 31 that Art. 366(29A)(d) may not apply because the owner of software does not intend to transfer copyright, but ultimately found that it could not strike down a statutory provision that would turn on the circumstances of an individual transaction.

What are we to make of this? In my submission, the Supreme Court, with respect, reached the wrong conclusion in Tata Consultancy Services in holding that the transfer of off-the-shelf software constitutes a “sale”, although this undoubtedly had a lot to do with the fact that the petitioner in that case did not press the point. However, that is the law as it stands today, and while the Madras High Court’s opinion is not inconsistent with it, a clear conceptual articulation of the legal classification of software is absent. Perhaps the way forward is for the Court to revisit its opinion in TCS and attach greater importance to the circumstances of the particular transaction.