Gilbert & Tobin seminar: the data economy

Speakers: 
Mr Rod Sims, Chair
Conference: 
Gilbert & Tobin seminar
15 October 2018
Transcript: 

Check against delivery

Introduction

Thank you for inviting me along to speak with you today.

Gilbert and Tobin have asked me to provide some insight into a number of interesting questions from the perspective of the competition and consumer regulator.

These questions included:

  • How should market definition apply to data/information markets?
  • How should factual and counterfactual assessments be carried out where there are high levels of uncertainty?
  • And what is the relationship between the CDR and access regimes, and what is the ACCC’s approach to enforcement action in this area?

These are each excellent questions.

In doing so, I will be focusing on four main points:

  • First, the flexibility of the tools available to the ACCC in assessing the competitive effects of transactions 
  • Second, the challenges posed by transactions  in dynamic markets, and in particular transactions involving big tech companies and big data
  • Third, the CDR, what it is and what it isn’t
  • Fourth, the enforcement issues associated with big data.

1. The tools available to the ACCC in assessing mergers or conduct

G+T have asked me to consider whether there is a need to adapt the market definition and market share tools used in merger analysis when applying them to data/information markets.

In particular they have asked me to look at how the small but significant and non-transitory increase in price test, the SSNIP test, might apply to zero price markets.

To me, this is straightforward.

It is important to note that these tools are purposive in application: one of our main objectives when we define a market is to identify the field of rivalry that is likely to be affected by the transaction. This information helps us to determine the degree of overlap, and the extent of competitive constraint provided on the merged firm. 

Market definition rarely remains static, and will shift over time and from case-to-case in order to most effectively identify the boundaries of the competitive overlap. For example, where the competitive overlap is in data or personal information, we will be looking to identify markets which capture that overlap, as well as the activities which use that data. 

In the case of multi-sided platforms it can often be helpful to define separate markets for the services supplied to the different types of users of the platform, for example for newspapers the users are  readers and advertisers. In providing services to a user group, platforms are constrained by the prospect of losing customers to other suppliers of the service.  It is important to recognise the strength of this constraint is affected by the interdependencies between user groups.  Losing readers not only reduces a newspaper’s subscription revenues, but will also reduce its advertising revenue in turn.

The SSNIP test, which looks at the likely response of customers to a small but significant increase in price by a hypothetical monopolist is one way to define a market. While the SSNIP test refers to the impact such an increase, however, price is only one way of identifying the boundaries of competitive rivalry.

The flip-side to price is clearly a reduction in the quality of the service or product being supplied by the hypothetical monopolist. A degradation in the quality of service provided could be evident in a number of ways; a reduction in the features of the service or a change to the terms and conditions on which the service is offered are two examples. One potential example may be if the terms and conditions are altered in such a way as to allow an online business to collect more data on the users.

Where users value their privacy and their personal information, an erosion of privacy is equivalent to an increase in price in the case of zero price markets. 

We do acknowledge that determining the impact of a significant deterioration in quality may be more difficult than calculating the impact of a SSNIP, but it important to note that these are just tools to determine the boundaries of substitution.

Our tools for assessing market power are similarly flexible. While we often look at whether a company will be able to increase prices above competitive levels following a merger, our merger principles are flexible enough to enable us to look at the impact of a transaction on quality and other factors.  We have been doing this for many years.

The list of merger factors in section 50 paragraph 3 of the Competition and Consumer Act 2010 expressly allows us to take into account the dynamic characteristics of the market and it is also open-ended.

It is worth noting, however, that some jurisdictions have recently amended their law to address potential deficiencies in dealing with the challenges of big tech.

For example, in June last year the German Competition Act was amended to address the challenges of digitalisation.

Changes included:

  • clarification that where no monetary consideration is paid, the services offered can still constitute a market
  • identification of certain factors particularly relevant to digital markets and online competition can be taken into account in assessing market power, namely: access to competitively relevant data, innovation driven competitive pressure, direct and indirect effects and barriers to switching.

We don’t think we need the first change and have not yet reached a view on the second. While the ACCC may take into account such factors under the current law, there may be some benefit in making this clear in legislation, particularly given that the Tribunal and the courts also apply these tests.

Nevertheless, the German changes highlight the international focus on these issues.   

2. The challenges in dynamic markets

This brings me to the second point I wish to make. While the tools we have available are flexible, these are hard cases. The value of data, big data, is only beginning to become apparent. The factual and the counterfactual in all dynamic fast moving markets, however, are difficult for regulators to predict, particularly when the overlap from the consumers’ side of the market is not immediate or obvious.

As the UK’s Competition and Markets Authority pointed out recently in a submission to the OECD, one of the key challenges with cases in digital markets relates to assessing what may be a small possibility of a large reduction in competition and that this may require the competition authority to consider the possible future development paths of digital markets, for which there are unlikely to be precedent or historical data.

That is a difficult task indeed. Imagine if someone had asked you to predict in 2008 how Facebook would operate in the market in future, for example.

A number of acquisitions by the big-tech companies, and the extent to which these acquisitions contributed to the market power held by these companies, have been the subject of scrutiny: Google’s acquisition of DoubleClick in 2008, Facebook’s acquisition of Instagram in 2012 and Facebook’s acquisition of Whatsapp in 2014 are some of the key transactions, but there are others.  The ACCC looked at some of these cases, but the more in-depth analysis was carried out elsewhere.

Google/Double-Click

In that case, the key issue was the magnitude of data that Google was able to collect as a result of these two businesses.  While there was some appreciation of the significance of personal data to advertising businesses at the time, it is fair to say that the true significance of such data and the Google/DoubleClick combination has only become apparent recently.

But it is also worth noting that in that case, Google submitted to the regulators that it would be contractually prevented from using ‘that part of its enlarged database originating from DoubleClick to improve, for example, targeting of search ads on Google’s sites or contextual ads in the Adsense network.’   In 2014, Google combined the Google and DoubleClick databases and let users know by virtue of an update to its privacy policy. Their previous representation was clearly not accurate.  

Facebook-Instagram

The competitive impact of Facebook’s acquisition of Instagram in 2014, which at that point had no revenue and only 13 employees, was also difficult to establish.

The UK Office of Fair Trading looked at that transaction and considered whether Instagram would be a potential competitor to Facebook but concluded that Instagram would not be well placed to compete against Facebook in the short term and also that there would be other firms able to compete against Facebook for brand advertising.  

It would be extremely difficult to predict what was going to happen in social media in the last five years, including whether in the absence of the acquisition by FB, Instagram would have grown and become an independent challenges to FB.  

We do, however, know where we are now: Instagram is the second largest social media platform in Australia, next to Facebook.

Facebook/Whatsapp

This is a more recent decision and one which squarely looked at the potential for FB to use data collected from Whatsapp users to improve its targeted advertising possibilities.

While the European Commission, or EC, acknowledged that this was not a material factor relied upon in its decision, Facebook made representations in that case that it was not technically possible to merge the Whatsapp information with its Facebook network. This representation turned out to not be correct, when Facebook announced in 2016 that Whatsapp users could link their numbers to their FB identities. This lead to a fine on FB by the EC for 110 million Euros. 

I am not saying that the ACCC would have reached a different view but we, like our overseas equivalents, are now very much alive to the significance of data, the implications of the network effects associated with these platforms and the potentially far-reaching consequences of acquisitions of smaller rivals.

There are further lessons to be learnt from these cases.  The ACCC has long taken with a grain of salt statements regarding whether particular conduct is legally or technically possible.  The broader lesson, of course, is for regulators to retain a very healthy scepticism concerning statements of fact or intention made by merger parties and their advisors.

3. Consumer Data Right

Now, I want to turn to the third point, the Consumer Data Right, which we call the CDR, and what it is and what it isn’t. 

The CDR: what it is and what it will do:

The CDR will enable customers to safely share their transaction, usage and product data with trusted service providers, but only if they choose to do so. It is essentially a data portability right. That bears repeating, it is purely a consumer right.

It will reduce the costs consumers incur when switching between providers and will lower the barriers to entry for new providers.

Accordingly, it will encourage competition between service providers, leading not only to better prices for customers but also more innovation of products and services

While an important objective, it is not just about facilitating switching but also enabling other services to be delivered to consumers such as budgeting advice, buying tips, data visualisation and any number of services that might emerge.

What it isn’t

It isn’t an access regime. There is no general right of access for competitors or potential rivals to data held by an incumbent, in the banking sector or elsewhere. The rights created in the regime belong to the consumer not to a competitor.

As you know, markets usually work more efficiently when consumers are well informed about the price and quality of offers available to them, the costs consumers incur when switching between providers are small, and barriers to entry for new providers are low.

In mortgage markets this is not the case.  As we observed in our Residential Mortgages Prices Inquiry, it is often difficult and costly for borrowers to compare the offers of mortgage providers. Discounts off standard variable interest rates are deliberately opaque. Borrowers often have to lodge an application and provide substantial personal information in order to confirm the interest rate a mortgage provider is willing to offer.

This all amounts to high search costs for the market to work effectively. Consumer data rights are therefore intended to reduce the cost to borrowers of discovering and comparing offers.

While starting with banking, a logical starting place, It is envisaged that the consumer data right will be expanded, initially to energy and communications, and consumers will be able to benefit from access to their data across the entire economy sector by sector over time. Parts of banking are due to commence from July 2019 though we are consciously looking at an achievable version on day 1 with more advanced versions to come.

This is an exciting new role for the ACCC which is throwing up many new issues for the ACCC and no doubt many more to come. We are out there talking to banks, consumer groups, fintechs and broader stakeholders. Our first job has been to develop rules with our framework document published and generating a lot of feedback. Next steps are to publish draft rules and progress our build of the electronic address book of accredited data receivers. Your interest in our work here and contribution to the landing point of tricky issues is appreciated.

Separately, I have been asked to speak about the possibility of an access regime for data.  

One option is clearly for the CDR to be extended to social media or other digital platforms. There are clearly a number of issues associated with this proposal including whether consumers would have an incentive to port their data when the service being provided is ‘zero-priced’. It is also unclear which data would be ported: the data which is valuable to the consumer, or the data which is valuable to prospective advertisers and therefore to rival digital platforms?  And is that data one and the same?

Extending the CDR is clearly not the only option, but any data sharing obligation runs into the same issue: privacy concerns and in particular whether it is in consumers’ interests for their personal information, even if de-identified or aggregated, to be shared with other businesses.

The merits of any kind of access regime or data sharing obligation would depend on the extent to which access to data is necessary for effective competition. Data is an important part of the business models of digital platforms and we are looking closely at the role it plays. But we acknowledge that data does not display the same characteristics as essential facilities type infrastructure; data can easily be duplicated.

We also acknowledge that data is only one source of digital platforms’ market power.

Other factors are also significant to competition in digital markets such as network effects, and economies of scale and scope. 

4. Enforcement challenges associated with big data

My fourth point is our enforcement experience in data.

The market issues we are encountering in data are basically fresh incarnations of the same issues that we encounter in market definition and merger assessments; there is nothing new under the sun, after all.

An example will make the point. In August this year, the ACCC commenced proceedings in the Federal Court against Trivago, alleging it made misleading hotel pricing representations in its television advertising and website, in breach of the Australian Consumer Law.

We allege that Trivago ran TV advertisements presenting its website as an impartial and objective price comparison service that would help consumers identify the cheapest prices for hotel rooms when in fact, Trivago’s website prioritised advertisers who were willing to pay the highest cost per click fee to Trivago.

The case, which is still before the court, underscores our growing concerns in relation to comparison platforms, and on how algorithms present search results to consumers.

In some respects, this is consumer protection 101; there is nothing new in allegations that a company says one thing, expressly or by implication, but does another either overtly or surreptitiously. But the new way in which this conduct is emerging does raise some challenges in terms of identifying matters and investigating the black box.

So how do we get behind these? As a starting point, we are very conscious of these issues. Matters involving data and algorithms were signalled as a priority in our 2018 compliance and enforcement policy.

We have also invested in an internal team, our Strategic Data Analysis Unit, staffed with the right expertise and focus to contribute to investigations and assessments.

The Unit has turned out to be instrumental in allowing us to analyse data and algorithms across a range of investigations that we are currently undertaking in both the competition and consumer areas, including in our growing market study work. The unit played a pivotal role in our Trivago case.

Beyond algorithms, we are also on the record as having an interest in and looking at the gathering and use of data by platforms and others online. On the gathering front, we are looking at instances in which firms may have failed to disclose the extent of gathering, changed their position on gathering or use of data and not adequately informed or obtained consent.

We are also looking at terms and conditions which go to gathering and use and whether they may be unfair contract terms. While new in some respects, nothing is new with the classic unrestricted unilateral variation terms that we have seen in many consumer and small business contracts and which really need to disappear from the landscape. We are seeing these in the online environment and in relation to the collection and use of data, and we will be following through with our concerns in this area.

Finally, I should mention that we look closely at the competition effects of conduct by businesses that share commercially sensitive data either directly or indirectly through a third party.  Such data might include prices or information allowing market shares to be determined or even information that discloses bids for work.  Huge amounts of data can now be shared very efficiently on a real time basis.  Often such sharing can make markets operate more efficiently. 

The nature of the information shared, the people with whom it is shared, the conditions of sharing and the impact of the sharing on the operation of markets is something that we are now finding ourselves looking at closely in many markets that rely of exchange of data.  We analyse the competitive problems looking carefully at any obstacles to competition created by market participants and the interactions between parties which might amount to anti-competitive concerted practices or even collusion.  Examples of investigations that have touched on these issues include our ‘informed sources’ and our current investigation into online travel agents.

Conclusion

Data is an issue that will becoming increasingly important over the coming years.

The consumer data right, and the digital platforms inquiry are two issues that concern themselves with how businesses and consumers are, and will be, interacting with the data that is increasingly collected about consumers as they go about their lives.

The ACCC has an important role to play. The consumer data right, and the digital platforms inquiry, set us up very well indeed for the issues to come.

Thank you for your time today.

Tags

Audience