Search interesting materials

Thursday, April 27, 2017

Building blocks of Jio's predatory pricing analysis

by Smriti Parsheera.

In a recent post on predatory pricing and the telecom sector Ajay Shah questions whether the subsidised user base of Reliance Jio can set off a network effect. The post makes two claims. The explicit claim is that the combined effect of interconnection regulation; mobile number portability and open standards of TCP/IP ensures that there are no real network effects in the telecom sector. The underlying implicit claim is that the existence of network effects is central to a predatory pricing analysis in this context. This piece takes a closer look at both these claims and the other factors that should inform the Competition Commission of India (CCI)'s analysis in the complaint filed by Airtel against Jio's pricing practices.

Network effects in telecom

Modern day tariff plans, including that of Jio, comprise of three main components - voice, data and access to content - all bundled into one product. A competition law analysis of Jio's pricing strategy must focus on each of these segments individually, and then their collective effect.

Voice services: Telecommunication services are known to generate strong network effects - the value of having a phone number is linked to the number of people who can be called using it. This creates a classic case for concentration of market power in the hands of the incumbent. Telecom regulators have overcome this issue by mandating operators to link their networks with the networks of other operators, allowing users to communicate across networks. Research on telecom networks, however, finds that despite interoperability, users tend to display a preference for being on a larger network, particularly when operators offer lower tariffs for calls made within their networks (on-net/off-net price differentiation). Others suggest that the network effects in telecom are more 'local' in nature - the preference to be on the same network as one's family and friends leads to the formation of calling clubs. This is not necessarily dependent on the overall size of the network.

In summary, even with mandated interconnection norms, traditional telecom services display a certain level of network effects. Arguably, the relevance of being "on the same network" would have gone down with the convergence of voice and data services and availability of various over-the-top calling apps. This requires a deeper study of consumer behaviour and preferences in the post-data world.

Internet services: Network effects on the Internet are not about the provision of Internet access services (i.e. the data services offered by ISPs) but rather about the direct and indirect network effects that define the business models of many Internet-based platforms and businesses.

Telecom service providers are increasingly stepping into the role of Internet platforms by bundling access to online music, TV, movies and news along with their communication services. In Jio's case, every new SIM comes bundled with a bouquet of Jio-branded apps, making it one of the fastest growing content aggregators in the country. Its free offer period from September to March has helped Jio build a massive user base, which in turn helps in attracting other complementary users to its platform. For instance, Uber's recent decision to partner with Jio Money reflects the value that it sees in being able to access Jio's users. The same holds true for other merchants and suppliers, like providers of music, video and news content, who are attracted to platforms with a large number of users.

Integration of data services and content

The vertical integration of data services and content offers Jio many advantages. One, convenient access to free content along with free/discounted data services has helped Jio in promoting higher consumption patterns. The aggressive data usage on Jio's network, particularly of video content, will gradually translate into higher revenues. The company claims that its users "consume nearly as much mobile data as the entire United States of America...and nearly 50% more mobile data than all of China." It would be interesting to see what percentage of this data is being consumed within the Jio ecosystem and the change in consumption patterns after Jio starting charging for its data services.

Two, it promotes faster adoption of in-house services. To take an example, the AT&T/FaceTime case study in the United States found that less than 10 percent of iPhone users downloaded Skype while all of them had automatic access to Apple's FaceTime. Adoption of Jio Money versus rival payment apps (among Jio subscribers) is likely to show similar results. Reports about the launch of Jio's 4G feature phone with built-in Jio apps suggest the possibility of further entrenchment of new users in the Jio universe.

Can Jio's pricing strategy in telecom enable it to indulge in monopolistic behaviour in related markets like mobile payments? Unlike telecom services, the payments sector continues to suffer from the lack of interoperability among providers, leading to significant network effects. Safaricom's M-Pesa service in Kenya offers an example of how the company was able to leverage massive network effects in the mobile-money space to establish its dominance in calls and text messages. The situation in India is certainly different - we have higher levels of competition, both in telecom as well as online payments. Yet, the Kenyan example is a helpful reminder of the extent to which cross-linkages between bundled products can influence their adoption and usage, to the exclusion of other competitors.

Jio's dual role as a telecom provider and platform offering access to online content makes it difficult to outright dismiss the role of any network effects. Moreover, any subsequent recoupment of the losses suffered by Jio in its early days need not necessarily be through a significant markup in data tariffs. Increase in volume of data consumption, future monetisation of Jio apps and opportunities for utilisation of data collected from users, are all factors that must be considered.

The tests of predatory pricing

The law and jurisprudence on predatory pricing defines it as below cost pricing by a dominant firm, with a view to exclude competitors. Sustained discounting practices in a market with strong network effects certainly raises a red flag due to the tendency of a single network to dominate the market. In such a scenario, there is a strong likelihood of recoupment after other competitors have left the market and structural barriers deter the entry of new players. The determination of predatory pricing, however, does not hinge on the existence of these network effects.

When Jio first launched its services in September, 2016 it was a fresh entrant in a market with several established players. Its price point of zero was certainly below cost but there was no question of it being a "dominant player". Any regulatory intervention to stop the pricing plans at that stage, whether by the sectoral regulator TRAI or the CCI, would have been premature.

This position has come to change over the last few months. Jio has managed to acquire a sizable presence in the market for high-speed data services - it holds about one-third of the country's broadband subscriber base and about 85 percent of the market in terms of mobile data traffic. Its share in the overall market for telecom services (voice plus data) still remains small since telecom subscribers continue to outnumber Internet users by a wide margin. The manner in which CCI delienates the "relevant market" will therefore form the crux of its analysis in this case.

Accordingly, the first step for CCI would be to determine whether there is a market for data services that is distinct from the broader cellular services market? This will hinge on a factual analysis of whether users regard voice, data and high-speed data services as being interchangeable in terms of their end-use and characteristics, based on a number of factors. One, voice calls can be made using the Internet but the reverse is not true - this indicates a one-way substitutability between the services. Two, there are some differences in the utility of 2G and 4G networks based on the applications that they are able to support. Three, CCI will need to collect data on Jio's usage patterns, that of its competitors and the switching behaviour of consumers. Four, supply-side constraints (like spectrum holdings) that can make it difficult for providers to switch from one type of service to another will also need to be considered.

In the second stage, CCI will need to examine whether Jio can be regarded as being dominant player in the identified market. Besides looking at its market share, in terms of subscriber base and usage volumes, this analysis must also consider the various other factors that have been given under the Competition Act, 2002. These include:

  1. Size and resources of Jio and its competitors - Telecom being a capital-intensive industry has many big players. Each of them has access to significant capital resources, although there may be differences in the extent to which these firms have been leveraged.
  2. Vertical integration of the enterprise - As discussed above, the bundling of voice, data and content offers Jio certain clear advantages. Other players are also offering similar bundles but not necessarily at the same scale. In many cases, the prices and bundles offered by other players have come about as a response to Jio's entry strategy.
  3. Entry barriers - Telecom is a heavily regulated sector and there are entry barriers, both in terms of licensing requirements and the availability and price of spectrum.
  4. Relative advantage through contribution to economic development - The arrival of Jio's 4G LTE network, with its aggressive pricing strategy, could also have some pro-competitive effects. Arguably, it has nudged the telecom market towards greater price competition, resulting in lower tariffs. Over time, this could also push other operators towards faster upgradation of technology.

Assuming that CCI's analysis leads it to delineate a separate broadband market (and Jio is found to be dominant in it), the third challenge would be to assess whether its current prices are in fact "below cost". This again will require data on the costs incurred by Jio for delivering its voice and data services and the free apps that are on offer. Finally, CCI will have to determine whether Jio's current pricing continues to be in the nature of a genuine "promotional strategy" by a new entrant or is it a deliberate attempt to reduce competition in the market.

Many have linked the consolidation that we are seeing in the market today with Jio's entry strategy. One one hand, consolidation reduces the number of players, hence reducing competition. On the other, it might be a sign of the sector's movement towards a more mature market with fewer players who are able to focus better on infrastructure expansion and quality of services. CCI will need to weigh in all these factors while examining the impact of Jio's prices on consumer interests, competition in the market and overall economic development.

These are all complex questions, with no obvious answers. The solution lies in a multi-stage, data-driven analysis of predation that should be rooted in an understanding of competition policy and telecom economics. Co-operation and knowledge-sharing between CCI and TRAI is key to finding these solutions.


Smriti Parsheera is a researcher at the National Institute of Public Finance & Policy. The author would like to thank Amba Kak, Kaushik Krishnan and Faiza Rahman for useful discussions.

Tuesday, April 25, 2017

Interesting readings

The Modi government has a new model for managing Kashmir by Sushil Aaron in Hindustan Times, April 19, 2017.

How can ARCs help solve the banking crisis? by Ajay Shah in Business Standard, April 17, 2017.

Budgeting for the police by Renuka Sane and Neha Sinha in Mint, April 11, 2017.

Political reforms are direly needed to renew our tryst with democracy by Varun Gandhi in The Economic Times, April 10, 2017.

It's a tug-of-war out there by Somasekhar Sundaresan on Wordpress, April 9, 2017.

The cow as cause - Vigilantism and the BJP by Mukul Kesavan in The Telegraph, April 9, 2017.

A Mega-merger and a Bureaucrat's Transfer by Paranjoy Guha Thakurta in Economic and Political Weekly, April 8, 2017.

The Money Bill conundrum: Constitution bench should decide by T K Arun in The Economic Times, April 6, 2017.

NCLT needs 69 benches to handle current case load: report by Shreeja Sen in Mint, April 6, 2017.

Banks aware of Bankruptcy code, some initiated corporate insolvency resolution: Madhusudan Sahoo by Joel Rebello & Saikat Das in The Economics Times, April 5, 2017.

Shri Jairam Ramesh's Speech-Central GST, Integrated GST, UT-GST & GST Compensation bills in Rajya Sabha TV, April 5, 2017.

A How to Book for Wielding Civic Power by David Bornstein in The New York Times, April 5, 2017.

LOKNITI-CSDS-KAS survey: Mind of the youth in Indian Express, April 3, 2017.

On Tyranny: Twenty Lessons from the Twentieth Century by Timothy Snyder by Tim Duggan Books, February 28, 2017.

The Fighter by C. J. Chivers in The New York Times, December 28, 2016.

Thursday, April 13, 2017

Retreat from private infrastructure projects

by Ajay Shah.

The case for private participation in infrastructure

Many years ago, most infrastructure in India was government owned. Policy thinkers strenuously argued for greater private participation, for the following reasons:

  • Private ownership would give better hardware, as a private person cares about what is being built. It would also give better safekeeping of the assets, as a private person cares about his things.
  • Private owners would strenuously push for adequate user charges, and act as a counterweight against the biases of the Indian political system in favour of low user charges and thus a burden upon the exchequer.
  • If a project is unviable, the private sector will more clearly say so and walk away, in contrast with the government processes which will build infrastructure in respond to political pressures.
  • Indian public finance would be better off when its balance sheet is freed from infrastructure assets, and these are instead held by listed utilities who issue debt and equity. It would become possible to bring the vast global capital to bear on these markets and deliver low cost financing.

For some years, private participation in infrastructure grew well, but things have changed sharply. Here are three key pictures. At each point in the time-series, we sum up the value of infrastructure projects in the CMIE Capex database that are classified as being `under implementation' by CMIE. The time series of the stock of `under implementation' projects changes from $t$ to $t+1$ because some old projects are commissioned, some are abandoned, and some new projects appear into the list.

Private infrastructure projects that are Under implementation in the CMIE Capex database

As the graph above shows, we got a huge increase from 2003 to 2011: a gain of roughly 10x in nominal rupees. By 2011, there was a stock of roughly Rs.25 trillion rupees of private infrastructure investment projects that were under implementation.

After  that, private infrastructure projects have receded substantially. We have a decline of Rs.5 trillion in nominal terms. If inflation were taken into account, that is a decline of another 25%.

How has government infrastructure investment activity fared?

Government infrastructure projects that are Under  implementation in the CMIE Capex database

This shows a picture of steady growth. In 2011, both private and government projects were at roughly Rs.25 trillion. From there, the private projects have dropped to Rs.20 trillion while the government has gone on to Rs.38 trillion. There is growth in the stock of government infrastructure investment projects under implementation, even after you take out the 25% increase in prices from 2011 till today.

It is quite a reversal for the long-held objective of having greater private participation in infrastructure.

What's the overall picture of infrastructure investment?

Total infrastructure projects that are Under implementation in the CMIE Capex database

Putting the two together, there is broad stability from 2013 onwards (but a decline in real terms once you take out inflation). What has not been widely observed is the compositional change within this overall number: the private sector is losing ground and government infrastructure projects are gaining ground.


In my view, the original logic in favour of greater private participation in infrastructure remains. The private sector will use capital more effectively, deliver a better incremental capital-output ratio, and take care of assets better. Conversely, public sector domination of infrastructure investment is going to deliver reduced bang for the buck. The compositional shift in favour of public infrastructure projects is a weakness.

Where did we go wrong?

In the first wave of pushing private sector participation, we did not adequately understand that private participation in infrastructure requires complex institutional machinery. The government's role in infrastructure is in three parts: Planning, Contracting and Regulating. Clear structures needed to be established for each of these three pillars. Mechanisms were required for resolving disputes and protecting cashflows from user charges. We needed to keep our eye on the prize: the projects that come out of all the complexities of the early stage and make it into the listed space, as boring utilities who just collect user charges and do O&M. With the benefit of hindsight, we went  about private sector investment in infrastructure in a slipshod manner.

In the recent period, instead of fixing these institutional complexities, there has been an excessive willingness to give up on private sector participation and make do with muscular State-led investment. It feels like an entire generation of institutional memory, about the problems of public sector infrastructure investment, has been lost. We are now running a Chinese-style risk of large investments going in with low returns in terms of incremental GDP per unit investment.

Wednesday, April 12, 2017

Emerging themes around privacy and data protection

by Vrinda Bhandari, Amba Kak, Smriti Parsheera and Renuka Sane.

Issues of data protection and privacy have become the subject of intense discussion and debate, in India as in the rest of the world. In this post, we identify certain key themes that arise in the context of these issues, that can augment our understanding of privacy and data protection and help towards forging safeguards in the form of a privacy law. Many of these were discussed recently at a round table organised at NIPFP on 24th March 2017. The key themes that emerged are summarised below.

What do we understand by privacy?

The term privacy has many connotations, takes different forms in different contexts and is viewed differently depending on the individuals own subjectivity. Defining it has been a challenge, with many scholars leaning towards more conceptual, and less rigid formulations. In philosophical debates, privacy can be characterised in terms of defining a sphere of private life that is separate from political activity and government interference. The sociological argument traces its roots in the fundamental haracteristics of social life - social context determines what is considered private in different circumstances. Others, like Solove 2006), however, move away from these conceptual discussions to identify specific privacy harms that have been recognised by society. His taxonomy of privacy encompasses four aspects - first, information collection (through surveillance and interrogation); second, information processing (through aggregation, identification etc.); third, information dissemination (through disclosure, exposure, breach of confidentiality etc.); and fourth, invasion (through intrusion and decisional interference).

Taking a slightly broader view, Calo (2011) speaks about privacy through the boundaries of subjective and objective harms. A subjective harm is internal to the person harmed, and is caused by unwanted observation. This encompasses, for instance, the knowledge or perception that some negative information about oneself is out there, which leads to distress and anxiety. Conversely, objective harm is external to the person harmed, when coerced or unanticipated information about oneself is used by other persons. Understanding of the potential harms is extremely important for the design of a policy response.

Another debate that emerges is whether privacy should be viewed as a right, an interest, or a property? Interestingly, the early parameters of what is now regarded as privacy evolved in the context of property rights. In 1890 Warren and Brandeis argued in a seminal paper that the right to privacy goes much beyond the concept of personal property rights, and must be recognised as such (to include for instance, the principle of an inviolate personality). By now most countries view privacy through a rights lens, because property, by its very nature, once bought, can be destroyed, transacted, and shared without the consent of the original owner. The economic dimensions of private data in the digital age have, however, once again triggered these rights versus property debates focused around the concept of "propertarian privacy".

Discussions on privacy also raise the question of privacy from whom. Traditionally, privacy was viewed in the context of the surveillance and law enforcement powers of the State. However, with the rise in big data and the explosion of social media, we now have to think of privacy from private actors as well, whether in the context of data mining, data retention, or data sharing arrangements. Surveillance, in this context, includes what Roger Clarke terms dataveillance - systematic monitoring of actions or communications using information technology.

Do people in India really value privacy?

While a lot has been written about the value of privacy (for example, Westin (1968)), it is often argued that people do not really know how to gauge the value of their own privacy. Many view the debates on privacy protection as the privilege of the elite who do not have to worry about accessing basic services, or as refuge for those who have "something to hide".

It is, however, important to remember that privacy is context specific. It is not always about what one may have to hide, but also what one may have to lose. These considerations vary across class, gender, caste, age and are often be different for different intersections of these categories. For each person, there are aspects of their life that are "personal", that they do not wish to be revealed to the public at large- and the control over which is integral to their sense of autonomy. In the digital context, the oft-heard lament is that privacy does not seem to be valued enough perhaps because people either don't know or feel ambivalent about how much data they are sharing (unwittingly), to which entities and the picture of themselves that their data is able to generate to these entities.

For awareness to be effective it must move from the risks to the harm. Sunil Abraham offers a useful analogy of tobacco use. Most smokers are well aware of the risks of smoking, but do not bother to stop, until they face a health crisis. Similarly, most people, while well aware of the privacy risks associated with their activities, for instance careless use of social media, do not take any remedial action until and unless they face a data breach. Therefore, just as health policy workers have tried to change the attitudes of smokers by scaring them through the inclusion of graphic images on the cigarette packs, it might be useful to alert people to the harms caused by the loss of privacy.

"Privacy by design" holds important lessons

The principles of Privacy by Design (PBD) developed by Ann Cavoukian are worth emphasising. The approach highlights that measures to protect privacy should be proactive and preventive, and not remedial. Privacy should be the default setting, embedded into design of technologies and services.

This overcomes many of the problems associated with choice/consent based regimes although adoption still depends on voluntary buy-in from businesses and users. So far, businesses in India are said to find an unwillingness among users to pay for privacy. For this reason, most privacy-enhancing technologies (PET) based solutions are B2B rather than B2C, and even these are far and few. We, in India, need to think of innovative ways to bring about a regime of data protection. A law on the subject and privacy-enhancing design elements are both part of the solution.

Issues of surveillance

Perhaps the most contentious of all issues is the one on where to draw the line between privacy and security, which often requires the use of various surveillance tools by the state. The PBD framework calls for "full functionality" in this context, i.e. it seeks to accommodate all legitimate interests in a positive-sum manner. Instead of a dated zero-sum approach with unnecessary trade offs of privacy vs. security, PBD says that it is possible, and far more desirable, to have both.

Yet, in reality there remains no consensus on a) the extent to which the state is engaging in surveillance, b) the extent to which Aadhaar and other big data techniques are being deployed, and c) the relationship between national security and privacy (is balance the appropriate metaphor? what is the trade-off, if any). The State claims that surveillance fears are misguided and overstated, while civil society argues that surveillance is broad based, and inadequate checks and balances leave citizens vulnerable. Given that both national security and privacy remain nebulous terms, there is no clarity on when one gives way to the other, and it is undeniably the rhetoric of national security that invariably overwhelms privacy. This issue requires unpacking and principles-based resolution as unchecked intrusions by the State can damage the very essence of what it means to be a liberal democracy.

Problems of Aadhaar

Given the pervasiveness of Aadhaar in our lives today, a debate on data protection cannot be complete without evaluating the legal framework surrounding it. The current legal framework of Aadhaar is weak. The Act delegates a number of core functions to be specified by the regulations, and these regulations further defer these functions as matters 'to be specified' by the UIDAI in some undefined future. This suggests that Aadhaar is currently functioning in some sort of a legal vacuum in terms of the nuts and bolts of important issues such as enrollment, storage, and sharing of data.

The regulations that have been issued by UIDAI did not go though a rigorous consultative process - both while preparing the draft, and in seeking comments from the public. The UIDAI should voluntarily opt for greater transparency on issues that have implications for privacy and data protection.

There is a case for a horizontal law

In India, the Supreme Court is yet to decide, what was until recently regarded a settled position - whether the right to privacy constitutes a fundamental right under Part III of our Constitution. While this is being debated, we have sector specific frameworks, like Section 43A of the IT Act, for protection of personal information and data security. More recently, the Ministry of Electronics and Information Technology (MeitY) has released the draft Information Technology (Security of Prepaid Payment Instruments) Rules 2017 for public comments. The draft rules aim to ensure the integrity, security and confidentiality of electronic payments through prepaid instruments, although amid concerns over the scope of the draft rules, MeitY's jurisdiction, and overlaps and conflicts with existing laws. Several other regulators such as the RBI, telecom authorities and health departments also have, or are in the process of developing, privacy/data protection norms pertaining to their jurisdictions.

These are all notable moves, but in the absence of a horizontal law, they will lead to the development of certain pockets of protection in certain sectors, while many other facets of private data will remain unprotected. Another concern is that the current legal framework does not hold meta data to the same standards as data in privacy and data protection debates.

There is a case for a comprehensive, principles-based, horizontal privacy law with basic minimum standards of privacy. These standards can then be tuned further to meet the requirements of different sectors. Thus, regardless of whether the Supreme Court of India considers privacy as a fundamental right, the State must define the circumstances in which it, as well as other private sector entities, may intervene with an individual's rights. Work on the draft privacy bill which began a few years back needs to be pursued with haste.

Vrinda Bhandari is a practicing advocate in Delhi. Amba Kak, Smriti Parsheera and Renuka Sane are researchers at the National Institute of Public Finance & Policy. We thank all participants at the round table on privacy and data protection organised by NIPFP on 24th March, 2017 for their contributions. Any omissions are our own.

Tuesday, April 11, 2017

Replace voice votes

by Devendra Damle, Shefali Malhotra and Shubho Roy.

Turning a bill (legislative proposal) into an Act of Parliament (law of the land) is a multi-step process. It involves placing the bill before the legislature; readings of the bill; publication in the official gazette; possible reference to relevant committees of the legislature; debates; and concludes with present members voting on the bill. Voting can be done in multiple ways, one of which is a voice vote. In the voice vote system, the members verbally communicate their assent by shouting 'Aye', or dissent by shouting 'No'. Based on which answer is most audible, the Speaker (person chairing the legislature) decides the outcome of the Bill. The system of voice votes, is obsolete. It slows down legislatures, grants excessive discretion to the Speaker, reduces the ability of citizens to hold their legislators accountable. It is still used in India.

For example, recently, the Punjab Assembly passed the Vote-on-Account of more than INR 251,990 million for the first quarter of the 2017-18 fiscal, by a voice vote. In the winter session of Parliament, the Taxation Laws (2nd Amendment) Act 2016 was passed through a voice vote, amidst protests and demonstrations. In 2014, opposition parties in Maharashtra questioned the legitimacy of the government after a confidence motion was decided in the government's favour through a voice vote. In the same year, the Lok Sabha Speaker was criticised for passing the Telangana Bill through a voice vote (Note: In this blog, we use the term 'Speaker' generically, for the Speaker of the Lok Sabha, the Chairman of the Rajya Sabha, the Speakers of state legislative assemblies, and Chairman of state legislative councils).

Voice votes

The Constitution leaves it to each house of legislature to set the rules of functioning of that house. Articles 118 and 208 of the Constitution empowers the Houses of Parliament and state legislatures to make rules governing procedure, respectively. Lok Sabha and Rajya Sabha have made Rules of Procedure and Conduct of Business (Procedure Rules) for their respective houses.Voting in the Lok Sabha is governed by rules 367, 367A, 367AA and 367B of the Lok Sabha Procedure Rules.

On the conclusion of a debate, the Speaker asks the members present whether a bill or a motion is passed. The member respond through a voice vote, the Speaker decides whether the motion is accepted or rejected. If a member challenges the Speaker's decision, the Speaker repeats the voice vote process for a second time. Any member can challenge the second voice vote by requesting for a division. The Speaker has the discretion to reject or grant the request for a division. If the Speaker rejects the demand for division, parliament employees take a head-count of members voting 'Aye' and members voting 'No'. Based on this head-count, the Speaker announces whether the motion is accepted or rejected. This decision cannot be challenged.

If the Speaker accepts the demand for division, he orders for voting by any one of the following three methods, at his discretion:

  1. Automatic Vote Recorders: Members press a button to vote 'Aye' or 'No' from their allocated seats. The result appears on an electronic display, and the Speaker announces whether a motion is accepted or rejected.
  2. Paper Slips: Members write 'Aye' or 'No' on paper voting-slips. Parliament officers collect the slips and count the votes. The Speaker announces whether a motion is accepted or rejected.
  3. Division Lobbies: The Speaker directs members voting 'Aye' to go to the right lobby, and those voting 'No' to go to the left lobby. Parliament officers count members in each lobby. The Speaker then announces whether a motion is accepted or rejected.

The Rajya Sabha and state legislatures follow a similar process with minor variations. In no case are individual voting records maintained. Even when a division is carried out, only the total number of votes for and against the motion are recorded.

The system of voice votes suffers from two weaknesses: It grants excessive discretion to the Speaker, and it reduces accountability of legislators to the citizens.

Speakers' excessive discretion

The discretionary power vested with the Speakers in a voice vote system is prone to abuse. In the Indian system, Speakers are inclined to side with the ruling party or alliance. Speakers of Lok Sabha and State Legislative Assemblies are elected from among the members of their respective legislatures, usually from the ruling party or alliance. Unlike in the UK, they continue to be members of their parties even after being elected as the Speaker.

The Maharashtra Assembly no-confidence vote in 2014 is an example of alleged abuse of discretionary power by the speaker. The Maharashtra Assembly speaker approved the confidence motion in favour of the present ruling party. The motion was approved through a voice vote, amongst allegations that the demand for division by the opposition was ignored. Effectively, it cast doubt on whether the government truly had a majority in the house.

According to the Parliament's Statistical Handbook 2014, five incidents of no-confidence motions and three incidents of confidence motions have been decided through a voice vote in Lok Sabha. While there has been no allegation of abuse in these cases, a voice-vote system can be easily manipulated, especially when it is used to determine crucial issues like the legitimacy of the ruling party.

Accountability to citizens

The Voice vote system lacks transparency. In a voice vote system, it is impossible to record individual votes of legislators. In the absence of individual voting records, a citizen has no way of judging the actions of his representatives. He is clueless about which way his elected representative voted, or whether that representative voted at all. This makes it difficult for the public to hold their representatives responsible.

An example of accountability using public voting records is Obama's criticism of Hillary Clinton, in the 2008 Democratic Party Primaries. In 2002, the then Senator Hillary Clinton voted in favour of the resolution to invade Iraq. By the 2008 primaries, public opinion (especially in the Democratic Party) had turned against the invasion. Obama repeatedly pointed out that Clinton voted in favour of the Iraq war, signalling to the Democratic Party that he is a better candidate. Obama could do this because each Senator's vote on each resolution is recorded against their name and published.

Solution: electronic voting

In the past, recording votes for each motion would have been time-consuming and costly. So, recorded votes would be reserved only for contentious issues. If the support or dissent for a motion was evident, it was left to a quick decision of the speaker. The cost of this efficiency was wide discretion to the speaker. Today, with electronic systems, we can gain this efficiency with no costs.

Efficiency gains

As the following examples show, electronic voting is a time-tested, efficient and, cheap technology:

  • Time-tested: Machine vote recording systems are not new; Thomas Edison patented a system in 1869. The World e-Parliament Report 2016 states that 67% of parliaments have adopted some form of electronic voting. Out of the remainder that still vote manually, 72% are considering to introduce electronic systems. The US House of Representatives has been using electronic voting systems since 1973. Recently, the Korean National Assembly adopted an electronic voting system.
  • Efficient: A 2016 Australian Parliamentary report found that adopting electronic voting systems reduces the time spent on counting votes, minimises human error, and expedites publication of results. A 2010 report to the UK House of Commons found that electronic voting can make the process less time-consuming. In turn, allowing MPs to devote more time to discussion and debate, the real function of legislatures. A 2003 Australian Parliamentary report finds that conducting a division vote in the Mexican Legislature used to take upto one hour; with the electronic voting system it now takes two minutes.
  • Cheap: The Mexican Legislature, with more than 500 members, has been using biometric authenticated electronic voting since 1998. The 2003 Australian Parliamentary report finds that the Mexican Legislature's electronic voting system has an operating cost equivalent to INR 54 million per year (at 2016 prices). For perspective, that is 0.86% of the total budget of the Lok Sabha for 2016-17.

Electronic voting systems are not alien to the Indian Parliament (the one in Rajya Sabha was installed in 1957). However, they are only used in case of a division. This means going through the process of two voice votes, calling for division, granting division, and then conducting a division. Even then, the decision to use electronic voting is subject to the Speaker's discretion. He can choose other inefficient methods like paper slips or the lobby method to reach a conclusion.

Ushering in transparency

The first step towards recording individual legislators' votes is by replacing voice votes with electronic voting systems. Carey, 2005, finds that when countries adopt electronic voting systems, demand for recording individual votes grows. Once the usage of recorded vote starts, pressure to make these records visible increases. An EU Parliament study of EU countries, where individual votes are recorded, finds electronic voting to be the most popular method.

The Indian electorate has been criticised for voting on caste/communal lines. However, in the absence of information regarding legislators' actions in the legislature, there is no other parameter for the average citizen to decide who to vote for. Bovitz and Carson, 2006, conducted a study examining the electoral consequences of individual voting records of legislators in the US House of Representatives. They found that legislators who vote against their constituents' preferences on controversial and politically prominent issues get lower vote shares in subsequent elections. Conversely, when legislators vote according to their constituents' preferences, especially against the party-line, they get higher vote shares. Legislators tend to vote strategically on prominent issues as they worry about taking the 'wrong' position in the eyes of their constituents.

Unlike the US, Indian legislators are subject to anti-defection laws. An Indian legislator cannot ignore a 'party whip/instruction' without risking losing his seat. It may be argued that, in the Indian scenario, individual voting records are useless. This argument has two weaknesses, namely that it:

  1. Contradicts the general principle of governance, that greater transparency in the working of government brings greater efficiency. It does so without providing any evidence for it.
  2. Ignores that once votes are made public the equilibrium shifts and individual voting records may act as a counter-balance to the negative aspects of anti-defection law.

We should always strive for greater transparency in governance. India still follows the Westminster system with people voting for individual legislators to represent them. On one hand, even if this information is useless, it does not harm anyone. On the other hand, when citizens get more information about their legislators, they can make more informed decisions. For example, merely because some candidates with criminal backgrounds are elected, does not mean we should stop requiring candidates to declare their criminal records. In addition, there are situations where anti-defection laws do not apply. In such cases this information can help voters. Anti-defection laws do not affect legislators who do-not vote or when there is no formal party whip. In such cases, individual voting records will provide valuable information about a legislator's behaviour. Did your legislator actually vote on a bill that was important to you or was he absent? It forces legislators to at least participate in issues important to their electorate and turn up to vote. Carey, 2009 examines the Corruption Perceptions Index, calculated by Transparency International, for most countries in the world. He finds that perceptions of corruption tend to be lower in countries where legislative votes are visible.

It may act as a counterbalance to anti-defection. Anti-defection laws have been criticised for reducing the voice of legislators. It puts party interests above the interests of the electorate. However, today the costs on the legislators following the 'party whip' and consequently harming his electorate is nil. Individual voting records may act as a counterbalance to this problem. Just like anti-defection pressurises legislators to vote in favour of the 'party whip', some studies show that individual voting records pressurise legislators to vote in accordance with the wishes of the electorate. Canes-Wrone et al., 2002 examined, through a study of the US elections between 1956 and 1996, the relationship between legislators' electoral performance and support for their party inside the legislature. Their study shows that in each election, an incumbent received a lower vote share when he supported his party. It also decreases the probability of retaining office. Crespin, 2010 finds that where votes are more likely to be noticed by the public, members of the US Congress adjust their votes in line with the demands of their constituency.

It is simplistic to argue that bringing transparency in individual voting records will not change the incentives/behaviour of legislators. Once individual voting records are available, the legislator will have two choices. First, vote against the decision of the party and get disqualified but, on the other hand, gain sympathy of the electorate. This may translate into more votes in the next election/by-election. The legislator can run as an independent candidate or on another party ticket and gain sympathy votes. Second, vote in accordance with the party line and hold on to his seat. However, now the entire electorate is likely to know he voted against their interest. The next election may not be in his favour.

We need to overhaul the functioning of the Parliament. Adopting compulsory electronic voting in our legislative bodies is a low-hanging fruit. It requires a small change in the Parliamentary procedure rules and trivial technological additions. This small change can go a long way in increasing efficiency, accountability and transparency in the functioning of the legislature.


Inter-Parliamentary Union, World e-Parliament Report 2016, 2016.

The Parliament of the Commonwealth of Australia, Division required? Electronic Voting in the House of Representatives, May 2, 2016.

European Parliament, Democratic scrutiny, transparency, and modalities of vote in the National Parliaments of the Member and in the European Parliament , 2012.

The UK House of Commons, The Case for Parliamentary Reform, 2010.

Michael H. Crespin, Serving Two Masters: Redistricting and Voting in the U.S. House of Representatives, Political Research Quarterly, 2010.

John M.Carey, Legislative Voting and Accountability, Cambridge University Press, 2009.

Gregory L. Bovitz and Jamie L.Carson, Position-Taking and Electoral Accountability in the U.S. House of Representatives, Political Research Quarterly, June, 2006.

John M.Carey, Visible Votes: Recorded Voting and Legislative Accountability in the Americas, Campbell Public Affairs Institute, September 9, 2005.

Judith Middlebrook, Voting methods in Parliament, Constitutional & Parliamentary Information, 2003.

Brandice Canes-Wrone et al., Out of Step, Out of Office: Electoral Accountability and House Members' Voting, American Political Science Review, March, 2002.

The authors are researchers at the National Institute of Public Finance and Policy, New Delhi. They thank Sanhita Sapatnekar, Anirudh Burman and Pratik Datta for useful discussions.

Thursday, April 06, 2017

Does the NCLT Have Enough Judges?

by Devendra Damle and Prasanth Regy.

The recently passed Finance Bill made headlines for combining tribunals, purportedly to rationalise their functioning. This is not the first such attempt. The Parliament set up the National Company Law Tribunal (NCLT) with a similar objective of streamlining all judicial proceedings under the Companies Act, 1956 and Companies Act, 2013. It has been operating since June 2016. The NCLT will hear all cases under these Acts which would earlier have gone to one of four existing courts and tribunals: the Company Law Board (CLB), Board for Industrial and Financial Reconstruction (BIFR), high courts (HCs) and Debt Recovery Tribunals (DRT).

However, commentators have pointed out that the NCLT is ill-equipped to cope with the pending cases it will inherit from the high courts and three tribunals. Others have pointed out that many of the legal and procedural issues which made the other tribunals ineffective will likely plague the NCLT too. One common concern is whether there are enough judges.

The popular discourse till date has largely focussed on pending cases which the NCLT will inherit. While the volume of these cases is substantial, we estimate that the volume of new cases which will instituted will be even larger. In this article we estimate how many judges the NCLT will need to handle this caseload. We find that the present strength of the NCLT is far lower than what is required. If this problem is not solved, NCLT is likely to end up a slow and inefficient tribunal.

Our Approach

We use the tribunals which would have originally heard the cases as a proxy for the different types of cases NCLT will hear. So we have HC-type cases, BIFR-type cases, DRT-type cases, and CLB-type cases. To calculate the number of judges NCLT needs, we need to know the:

  1. Annual rate of institution of cases of each type (I), and
  2. Annual rate of disposal of cases per judge for cases of each type (D).

To get I, we take the average number of cases instituted every year in HCs and each of the three tribunals. We only count the kind of cases which will be transferred to NCLT. For example, original jurisdiction company petitions were earlier being heard by HCs, but will now be heard by NCLT. So, from the total cases instituted in HCs every year, we only count original jurisdiction company petitions.

For D, we first calculate the average number of cases (of the relevant kind) disposed of by HCs and each of the three tribunals every year. Then we divide this average disposal rate for each of them by the number of judges. For example, the CLB has 5 benches. So D for CLB is equal to total cases disposed of by CLB in one year divided by 5. Thus, we get the average disposal rate per judge for each of the four types.

I/D gives us the number of judges required for disposing of each type of case instituted in one year. Adding all values of I/D gives the total judges NCLT will need to dispose of all cases of all types instituted every year.

What is the NCLT's caseload?

In an article by the consultancy Alvarez and Marsal, the authors estimate that a total of 24,900 existing cases — 4,000 from CLB, 700 from BIFR, 5,200 from HCs, and 15,000 from DRT — will be transferred to NCLT. Cases which were being heard by CLB will be be transferred to the NCLT automatically. Cases from BIFR will only be taken up by the NCLT if the parties file fresh applications. Cases from HCs which are eligible for transfer to NCLT, will be transferred in stages through notifications by the Ministry of Corporate Affairs. Assuming all cases do get transferred to NCLT, the tribunal will start with 24,900 cases.

What about the admission of new cases? From 2011-12 to 2014-15, on an average the CLB admitted about 10,170 cases per year (See CLB Annual Statistics).

In the same period, BIFR admitted an average 140 new cases per year (See BIFR Annual Statistics).

All HCs put together admitted approximately 14,000 original jurisdiction company matters in 2015-16 (See SC Annual Report 2015-16). In the Bombay, Delhi and Orissa HCs, approximately 90% of all original jurisdiction company matters are company petitions and applications made thereunder, cases likely to now go to NCLT. Assuming this proportion holds good for all HCs, about 12,700 cases which would earlier have gone to HCs will now be heard by NCLT.

DRTs admitted an average 21,470 Original Applications (OA cases) per year from 2012-13 to 2014-15 under the Recovery of Debts Due to Banks and Financial Institutions Act, 1993 (See here and here). In a sample of about 15,000 cases we collected from Delhi DRT-3 company and LLP related matters — matters which will now be heard by NCLT — constitute approximately 45% of the total cases. That brings the number to 9660 cases for all DRTs put together. These are the cases which will now go to NCLT instead of DRTs.

Table.1: Institution rate of fresh cases in NCLT
Type of Cases Institution Rate (I)
(cases per year)
HCs 12,700
BIFR 140
DRT 9,660
CLB 10,170
Total 32,670

This brings the total volume of fresh cases to 32,670. This will be the annual rate of institution of new cases in NCLT, assuming it stays constant over the years. The next question we have to tackle is: how many of these cases can be disposed of in a year?

What will be the disposal rate of cases?

The DRT had an average disposal rate of 360 cases per judge per year (from 2012-13 to 2014-15). The CLB had an average disposal rate of approximately 1,705 per judge per year (from 2012-13 to 2014-15). We assume the disposal rate for these types of cases will be the same even in NCLT.

For cases from DRT and CLB we can straightaway use the respective tribunals' disposal rates.

For HCs and BIFR, the disposal rate cannot be calculated in the same manner. HCs deal with a large variety of matters. Of these, company petitions — the kind of cases which will now be heard by NCLT — form a small fraction. The average per judge annual disposal rate for all HCs put together is 19. Bombay HC, which has the highest disposal rate for company petitions among all HCs, only disposes of 60 company petition cases per judge per year. It stands to reason that a specialised tribunal like NCLT would have a higher disposal rate than that. Therefore, we assume that in NCLT these cases will have the same disposal rate as DRTs' average disposal rate i.e. 360 per judge per year.

BIFR is notorious for cases pending for a very long time. In the three years from 2010 to 2013, BIFR disposed of just 169 cases (62 pending). Since the Companies Act, 2013 and the Insolvency and Bankruptcy Code, 2015 have more streamlined processes for winding-up of companies, it is expected that the disposal rate would be higher when NCLT takes over cases being heard by BIFR. Therefore, we assume that when NCLT hears these cases it will dispose of them at a rate equal to DRT's disposal rate i.e. 360 per judge per year.

For cases from HCs and BIFR, why do we take the disposal rate of DRTs and not CLB? Unlike DRT, 60–80% of the CLB's caseload consisted of compliance related matters or small matters. For example, in 2013-14 and 2014-15, 60% and 40% (respectively) of the cases instituted in CLBs were matters regarding taking deposits without advertising (Sec.58A(9) Companies Act, 1956). These are routine matters and have high disposal rates; the ratio of annual disposals to institutions is consistently close to 1. Substantive matters on the other hand constitute 20–40% of the instituted cases, and their disposal rate is much lower. For example, in cases regarding mismanagement of companies (Sec. 397, 398 Companies Act, 1956), the ratio of annual disposals to institutions is around 0.7. Between 2013-14 to 2014-15 cases under Sec. 397/398 constituted 4–6% of total cases instituted in CLB, but represented 40% of the total pending matters at the end of the year. Since most of the cases heard by HCs and BIFR are expected to be of a substantive nature, we cannot use CLB's average disposal rate. Therefore, we assume that NCLT's disposal rate for these cases will be similar to the DRTs' average disposal rate, rather than the CLB's.

How many judges does the NLCT need?

Armed with the institution and disposal rates of each type of case we can now calculate the number of judges required. One last consideration is the structural difference between the NCLT and the other tribunals. The NCLT has two types of benches, viz. division benches consisting of 2 judges and single benches consisting of 1 judge. Therefore, in the case of NCLT we calculate the number of benches rather than judges.

Table.2: NCLT Benches Required
Type of Cases Institution Rate (I) Disposal Rate (D) Benches Required (I/D)
(cases per year) (cases per bench per year)
HCs + BIFR 12,840 360 36
DRT 9,660 360 27
CLB 10,170 1,705 6
Total 32,670 473 69

Thus, we estimate that the NCLT will require 69 benches just to keep up with its caseload. We can also use these disposal rates to calculate the number of judges required for clearing the inherited backlog of 24,900 cases. If these cases are to be disposed of steadily over the next five years, NCLT would need about 80 benches.

The NCLT currently has 14 benches (in eleven locations) (See here, here and here). With a disposal rate of 473 per bench per year, and 14 benches, NCLT can dispose of 6,620 cases per year, i.e. only a fifth of the incoming fresh cases. At this rate, the NCLT will accumulate a backlog of 26,050 cases per year; a total backlog of 1,30,250 cases in five years. The Central Government is planning to establish one NCLT bench in every HC jurisdiction, i.e. a total of 24. Even with 24 benches, the NCLT would accumulate a backlog of 21,320 every year. That's a total backlog of 1,06,600 cases over the next five years.

It should be noted that in these calculations we haven't factored in any increase in the rate of filing fresh cases, nor have we considered the entirely new categories of cases (e.g.: class action suits) which the NCLT will hear. These will place an even greater burden on the tribunal.

Some corroborating evidence for our predictions

Some data on the disposal rates of NCLT are already available. In the six months from June 2016 to November 2016, NCLT disposed of 1,930 cases. From June to August only 240 cases were disposed of. This is probably because the NCLT just started functioning in June. The bulk of the cases — 1690 — were disposed of in the last three months, i.e. from September to November. If we take the disposal rate for just these three months and project it for the whole year, it translates to a disposal rate of approximately 6,750 cases per year. This is close to our estimated disposal rate of 6,620 cases per year.


There is evidence to suggest that badly designed procedures which allow unecessary adjournments, lost working days, and administrative inefficiency substantially contribute to judicial delays and pendency. In the case of NCLT, even if these issues are fixed and we manage to double its disposal rate, the current bench strength would still be far short of what is required to handle its caseload.

This analysis draws attention to the fact that we need to do a better job of estimating the judicial resources required to handle case loads. What we have presented in this article is an example of a Judicial Impact Assessment (JIA): estimating the resources required in the judiciary to handle the case load, using data on court productivity. A pre-requisite for JIA is the availability of high-quality empirical data on case loads and productivity. It is important that JIA should be institutionalised in the legislative process, so that courts are able to deliver timely justice.

The US has a specialised body called the National Centre for State Courts dedicated to conducting research on the functioning of state courts. It maintains databases of case-level data and has developed models for various kinds of judicial needs assessments for state courts. It has even developed software for court and case management, which are currently used in the US state courts. There is a similar system in place for federal courts. The estimates for judicial resource requirements in both cases are based on weighted caseloads which are derived from granular, case-level data.

One significant difference between India and the US is that in comparison to the US, it is easier for the Government of India to change the number of benches as needed. In the US, new judgeships can only be created by an act of the legislature. By contrast, in India, new benches for the NCLT can be created by the Central Government simply by notification. What is however lacking is a sound process for determining the number of judges/benches needed.

With its lack of adequate number of benches, the NCLT is likely to be plagued by delays just like its predecessors. A more efficient judicial procedure, and greater bench strength, are both required to effect a lasting solution.


Pratik Datta and Prasanth Regy. Judicial procedures will make or break the Insolvency and Bankruptcy Code. Ajay Shah's Blog, January 2017.

Prasanth Regy, Shubho Roy and Renuka Sane. Understanding Judicial Delays in India: Evidence from Debt Recovery Tribunals. Ajay Shah's Blog, May 18, 2016.

Pratik Datta and Ajay Shah. How to make courts work? Ajay Shah's Blog, February 22, 2015.

Reserve Bank of India. Report on Trend and Progress of Banking in India 2015-16

Nikhil Shah, Khushboo Vaish, Kavya Ramanathan. NCLT Readiness Report. Alvares and Marsal India, 2017

Company Law Board, annual statistics.

Lok Sabha Questions on cases pending in DRT dated 4th March 2016 and 4th December 2015.

The authors are researchers at the National Institute of Public Finance and Policy, New Delhi.

Monday, April 03, 2017

Predatory pricing and the telecom sector

by Ajay Shah.

  1. When there are network effects, we should be cautious about the business strategy of discounting. What looks like a gift to consumers today is often a plan to achieve market power and recoup those gains by extracting consumer surplus in the future.
  2. The burn rate at Reliance Jio is likely to be pretty large. However, the question that we should be asking: Can this subsidised user base set off a network effect?
  3. In telecom, interoperability regulation is in place. Even if Reliance Jio was able to establish a commanding market position through discounting, there is no way to close off its user base for rival firms. Interconnection regulation by TRAI imply that a phone call from a rival telecom company, to a Reliance Jio customer, will always go through. The open standards of TCP/IP mean that a data packet from a customer of any data communications company in the world will successfully reach a Reliance Jio customer. Even if all my friends and family are on Reliance Jio, it makes no difference to my decision to be on Reliance Jio. There is no network effect.
  4. Recoupment test: If in the future, Reliance Jio tries to increase prices, nothing prevents customers from switching to rival firms. There is no reason for a consumer to stay with Reliance Jio at future dates if it turns out that Reliance Jio is expensive.
  5. Market power in this industry has been checked by the three key building blocks -- interconnectivity regulation + mobile number portability + the open standards of TCP/IP.
  6. In fact, there is a negative network effect, as follows. Suppose a lot of customers switch from rival telephone companies to Reliance Jio. This will clog the airwaves of Reliance Jio's base stations, so the performance of Reliance Jio will go down while the performance of rival companies will go up. Through this channel, if Reliance Jio succeeds a lot in gaining customers, it will fail in delivering the best mobile data services.

Second order issues:

  1. Interconnectivity regulation imposes costs upon all regulated persons and these costs should be placed in a fair manner.
  2. There is an opportunity to obtain market power in JioMoney as payment regulation lacks all three components: interconnectivity regulation + number portability + open standards.

Interesting readings

Market power in the new economy by Ajay Shah in Business Standard, April 2, 2017. Paper landing page.

UW professor: The information war is real, and we're losing it by Danny Westneat in The Seattle Times, March 30, 2017.

Don't Worry, Be Happy - High Frequency Trading Is Over, Dead, It's Done by Tim Worstall in Forbes, March 25, 2017.

Seven Charts That Show the Gap Between Old and New IITs by Thomas Manuel in The Wire, March 24, 2017.

Aadhaar is a legal right, but the government can suspend a citizen's number without prior notice by Anumeha Yadav in Scroll, March 23, 2017.

What London Police Learned From the Last Big Attack by Henry Wilkins in The Atlantic, March 23, 2017.

Central Ministry, State Government Departments Publicly Expose Personal Data of Lakhs of Indians in The Wire, March 23, 2017.

Don't mistake good psephology for good policy by Somasekhar Sundaresan in Business Standard, March 22, 2017.

Is it too late to save Hong Kong from Beijing's authoritarian grasp? by Howard W. French in The Guardian, March 21, 2017.

There is nothing that can stop banks from systematically fleecing you. Here's why by Dhirendra Kumar in The Economic Times, March 14, 2017.

Lessons From The FPI Limit Breach In HDFC Bank by Bhargavi Zaveri and Radhika Pandey in Bloomberg, March 10, 2017.

Sebi chairman Ajay Tyagi has his task cut out by Mobis Philipose in Mint, March 9, 2017.

Inside Steve Bannon's Failed Breitbart India Scheme by Asawin Suebsaeng in The Daily Beast, March 2, 2017.


CUTS 5th Biennial Competition, Regulation & Development Conference 09-11 November, 2017 Jaipur, India


I. Introduction

CUTS and CIRC invite papers for the 5th Biennial Competition, Regulation & Development Conference to be held on 09-11 November, 2017 in Jaipur (India). Interested scholars and practitioners are invited to submit a 500 word abstract for a chance to participate in this Conference and present their paper.

The abstract should be based on any one of the four plenaries of the Conference (below) and should be submitted to the undersigned, along with a brief CV (not more than 2 pages) of the author. Authors are requested to mention the specific ‘Plenary’ their paper is based on while submitting the abstract.

Authors of selected abstracts would be invited to submit full conference papers (3,000 to 4,000 words) for a chance to participate in this conference. On successful selection, the organisers will provide support to the author (air travel, accommodation and meals) to participate in the conference, and present the paper.

II. Plenary Sessions

The abstract/paper should target any one of the following four plenaries of the conference.

  1. Plenary 1: Revisiting IPR and Competition
  2. Plenary 2: Disruptive Technologies and Economic Regulations
  3. Plenary 3: Building Organisational capacities for tackling policy and regulatory uncertainty
  4. Plenary 4: Challenges and Opportunities of Development Financing for Fostering an Innovation based Ecosystem

For more information, please visit:
Call for Papers
Background Note

III. Important Dates and Deadlines

The following dates/deadlines need to be noted:

  • Last date for submission of Abstract (500 words) is 15th April 2017
  • Selection of successful Abstract will be done by 30th April 2017
  • Last date of submission of Full paper (3,000 to 4,000 words) is 30th June 2017
  • Selection of successful papers will be done by 31st August 2017

IV. Further Information and Contacts

All submission and further queries should be directed to Mr. Ashutosh Soni ( and Mr. Parveer Ghuman (

Friday, March 31, 2017

Competition issues in India's online economy

Smriti Parsheera, Ajay Shah and Avirup Bose.

The world of high technology companies is seen as a dynamic area with a rapid pace of creative destruction. There is, however, a class of industries where there are strong network effects, where the market tends to collapse into a narrow set of players. After one burst of innovation where a new online business is born, there is the possibility of entrenched market power with the extraction of consumer surplus.

Many firms, global and Indian, have resorted to the strategy of making large losses by subsidising users, as a way to obtain those network effects. This has created a new class of concerns about predatory pricing, with unprecedented negative profit margins on a sustained basis, being supported by equity capital infusions. In the short run, discounts are popular, but recoupment is inevitable and market power will adversely affect consumers in the future.

In a recent Paper, we argue that the existing competition law regime in India needs to be fine tuned, for technology-enabled markets with significant network effects, to address the possibility of new kinds of abusive conduct. We offer a series of tangible proposals through which the Competition Commission of India can better handle these emerging situations. We also look into the role and responsibilities of the investors who back these online businesses and the impact of their conduct on competition in the underlying markets.

Entry barriers in the new economy

Innovation is the foundation of economic progress. While we normally revere technology companies for their disruptive innovations and the efficiencies that they create, we must recognise that some technology-driven businesses are susceptible to the acquisition and abuse of market power. The Indian competition regime is an evolving one, and has only recently started facing some of these concerns. Our paper brings new evidence and arguments to the table, on these questions.

Internet-based businesses, along with several other high-technology sectors, form part of the 'new economy', characterised by high rates of innovation; low marginal cost; increasing returns of scale; and, in many cases, network effects. Direct `network effects' arise where a user's benefit from a product or service increases with the number of other users on that network. The benefit of being on Facebook or WhatsApp, for instance, corresponds with the number of friends and family who use that service. Contrast this with the benefit of having an email address, where the benefits are not limited to closed proprietary networks. This became possible due to the early adoption of interoperability standards in email protocols.

Network effects are particularly important in two-sided markets where users on each side of the market derive a positive effect from the expansion of users on the other side. Commuters who use taxi aggregation platforms like Ola and Uber will logically be attracted to a service that has a large number of drivers on its network, which yields a lower waiting time. The same is true for the drivers working with these platforms. Similarly, in case of payments wallets, in the absence of interoperability regulation, merchants and customers will both prefer a service that has the most addressable users.

With the use of modern technology, the cost of running the marketplace itself has dropped to near zero levels. As an example, the online classifieds site Craigslist reports that it has about 40 employees who manage a network that sees over 80 million classified ads per month. The marginal cost of a transaction has gone to near-zero levels. This gives a unique class of problems where technological innovation that yields cost reductions cannot be a mechanism to take on an incumbent.

Brain versus brawn

How can market power be established, in this new world? One mechanism through which one player can obtain a competitive advantage is to attract users through technological innovation, and thus get a network effect started. This is an attractive strategy for firms which have deep human capital. Another mechanism is by using financial capital to pay subsidies that entice users. This is an attractive strategy for firms which have superior access to financial capital. Many online businesses have resorted to practices like deep discounting, cash-back offers and other schemes designed to attract new users and establish the network effect. Sometimes, heavy losses have been sustained for years on end.

As an example, the global taxi company 'Uber' made worldwide losses in the first half of 2016 of US\$1.27 billion (approximately Rs.86.5 billion). Uber's behaviour impacts upon the Indian economy as it has applied the strategy of using financial capital as a competitive lever in India also. On a similar note, the Indian taxi company 'Ola' reported a net loss of Rs.7.96 billion in March, 2015. The company's financial records for the periods after that are not yet available although it is reasonable to expect that the losses will be significantly higher due to the higher driver incentives. In the last two years, it is estimated that the two taxi companies, Uber and Ola, burned cash adding up to about Rs. 130 billion in India.

Such behaviour is found in other industries also. In the field of payments, where regulations have blocked interoperability and thus created the opportunity to kick off a network effect, the firm One97 Communications, which owns 'PayTM', reported a loss of Rs.15.49 billion in March, 2016.

The scale of these discounting practices, and the sustained periods for which they are continued, has created new barriers to competition. It is difficult to rationalise these sustained losses as being an introductory offer by a new player. Rather, these practices appear to be a systematic competitive strategy. Capital has become a competitive weapon. This gives rise to concerns that the market may eventually tip in favour of the player that may not necessarily have the most innovative product or service, but one that succeeds in obtaining more capital and enticing more users in the early days, using subsidies. While seeming beneficial for consumers in the short run, such practices raise concerns about competition on account of the creation of market power, and elevated prices for consumers in the following years when losses are recouped.

The FDI guidelines issued by the government in March, 2016 turned the spotlight on pricing practices of e-commerce firms. It clarified that the automatic route of foreign investment would be available only to those e-commerce marketplaces that avoided such subsidies.

These issues have also come to the attention of the CCI in a few recent cases. In April 2015, the CCI passed a prima facie order recommending a detailed investigation into the allegation that, armed with substantial funding received from various investors, Ola had indulged in abusive market practices to garner greater market power in the city of Bengaluru. More recently, the COMPAT directed the Director General of the CCI to initiate a similar investigation to assess Uber's dominance in the market for radio taxi services in the National Capital Region (NCR) of Delhi after the CCI had refused such an investigation. Uber has now challenged this decision before the Supreme Court, citing a 'jurisdictional flaw' in the Tribunal's ability to order such an investigation. Alongside these developments, CCI is also reported to have set up an in-house panel to understand the cash-back incentives being offered by various online companies from the perspective of predatory pricing provisions under the Act.

Our paper explores the recent developments in India in this area, in the light of foundations of economics and competition law. It argues that there are grounds for concern about the harm to competitive dynamics from these new business strategies. At the same time, it is important to avoid intrusive interventions that bring the State into excessive involvement in the world of business.

New economy requires new thinking

There is a need to take into account the distinct economic features of certain high-technology businesses when looking into allegations of anti-competitive conduct by them. Practices like deep discounting and cash back offers may be aimed at building sufficient scale in today's market to ensure that the business is able to fully capture tomorrow's market, to the exclusion of other competitors. A robust economic analysis of the impact of increasing returns to scale, and network effects, is required for understanding the present and future impact of these practices on competition and consumer interests. A novel dimension, which is addressed in the paper, concerns collaboration between the investors in the multiple firms that they invest in.

Transient gains to consumers

We examine the question about gains to consumers from discounting. We suggest that the gains in the short term need to be seen in a larger context. The recoupment test examines the extent to which market power can be achieved in the future, after which prices can be raised. If the CCI were to adopt this test in investigations relating to predatory pricing by online firms it would see that in certain areas, there are network effects, and once a small cartel of firms has acquired market power, it would be difficult for entrants to compete with them in the future. In that future scenario, it would be possible for incumbents to raise prices, and recoup earlier losses.

Interoperability as a tool for competition policy

In some situations, the CCI could rely on the essential facilities doctrine to mandate interoperability between a dominant player that is found to be indulging in the abuse of its position and other operators in the market. For instance, imposing interoperability requirements on a dominant payments network can help extend the network effects of digital payments to the economy as a whole, rather than being limited
to a closed network. The imposition of any such requirements will, however, need to be balanced against factors such as the payment of fair and reasonable access fees, the complexity of institutional arrangements required to monitor such arrangements and assessment of the impact on future innovation. More generally, open standards are an important element of interoperability, and various arms of the regulatory State need to push in favour of competitive markets through interoperable open standards.

Acting within Internet time

Given the fast-changing nature of online businesses, there are concerns about the elapsed time between a full-fledged investigation and the determination of a violation. We suggest a two-pronged approach to address this issue. On one hand, the CCI needs to work towards adopting stricter time frames for the disposal of cases, particularly those relating to new economy firms. On the other, we propose a voluntary settlement process that will allow a business that is under investigation to voluntarily alter its market behaviour, with the concurrence of the authority but without the need for a conclusive finding of violation by the CCI.


In India, technology companies are generally revered as the source of technological progress. However, the problems of competition policy are universal and cut across all industries. The basic principles do not change. The purpose of competition policy is to stave off situations where a narrow set of firms have market power, and new players are not able to enter. Society gains when firms obtain profits and valuation through innovation, not through the crafty use of financial capital to kick off network effects.

These issues were not faced in thinking about Indian competition policy as recently as five years ago. They are, however, likely to become increasingly important in the future. We argue that this calls for fresh think about the legal framework also. There is a case for competition authorities to look into the unilateral abusive conduct of a firm, which, although not dominant at the given point of time, is engaging in anti-competitive practices that create a strong and imminent possibility of its dominance. We highlight some pros and cons of this approach and leave this question open for further research.

Smriti Parsheera and Ajay Shah are researchers at NIPFP, and Avirup Bose is a researcher at Jindal Global Law School.