VoxEU Column Competition Policy

The ACCC’s ‘bargaining code’: A path towards ‘decentralised regulation’ of dominant digital platforms?

New law for a Mandatory Bargaining Code has been introduced in Australia to implement a decision that publishers should be compensated for use of news content by giant digital platforms. This reflects a policy view that the large disparity in bargaining power between platforms and individual publishers requires positive intervention to support quality journalism and news production.  This column argues that the Code as formulated by the ACCC has desirable properties in line with bargaining theory (including the use of ‘final offer arbitration’ as a backstop); it also leaves implementation of the regulation to the parties involved, not to an agency suffering from extreme asymmetric information. At a time when the design of regulation for ‘gatekeeper’ platforms is very much top of the agenda, this ‘decentralised regulation’ approach should be considered as part of a menu of possibilities in multiple settings. 

Authors’ note: We would like to thank Monika Schnitzer and Patrick Rey for their helpful conversations on this topic, and Patrick again for suggesting the term “decentralised regulation”. The views expressed in this column are the authors’ only. We have not received any compensation for this column, nor have we advised any parties in either the Australian investigation or the current consultation. Caffarra used to advise NewsCorp but no longer does, and has consulted recently for Amazon, Apple and multiple others. Crawford has consulted recently for Apple.  Neither have consulted on matters related to those discussed in this column.  

The Australian Competition and Consumer Commission (ACCC) has just proposed new legislation to establish a Mandatory Bargaining Code to cut through a long-standing stalemate between giant ad-funded platforms (Google and Facebook) and the publishing industry about payment for use of publishers’ news content. While discussions around this issue have long been mired in controversy around the globe (with publishers arguing they are providing valuable content but are powerless to monetise it, and the platforms arguing that they are providing free traffic), the ACCC has introduced a legislative proposal involving an obligation for the two sides to reach a bargained solution for a payment in favour of publishers, specifying also a particular type of ‘backstop’ (‘final offer arbitration’) should negotiations fail.  Notwithstanding Google’s threats of lowering the quality of its service to Australian consumers in retaliation (Lomas 2020), this approach has desirable properties – because it leaves it to the parties themselves to shape an agreement, and the backstop of ‘baseball-style arbitration’ leverages their private information while reducing the arbitrator’s discretion. But can this approach have implications beyond compensating news publishers? Might it be a useful complement to a more normative approach to regulation, where ex-ante principled rules are set, to be implemented by an agency? Can it have broader application as we struggle with how to deal with the enormous disparity in bargaining power between platforms and businesses that operate on them? In this column, we argue that this form of ‘decentralised regulation’ shows promise as a worthwhile option in the design of future regulatory options for addressing the dominance of digital platforms in many settings. 

A Mandatory Bargaining Code: The ACCC’s innovative solution to compensating publishers

An exploitation problem, and a clever solution 

We are not alone in worrying about the quality of journalism, a key input into the standards of political debate and a pillar of democracy. Quality journalism needs funding. Yet publishers’ sites have been long used as a source of free ‘content’ by Google and Facebook to populate their pages; and while this has added to the consumer experience by linking to information of interest, it has also weakened the economics of publishing.  News outlets are finding themselves in a ‘pact with the devil’: on the one hand, they seek visibility on digital platforms given their pervasiveness as a go-to source for information on the web; on the other, to the extent that news is consumed ‘for free’ on a platform and less traffic goes directly to publishers’ outlets, publishers’ ability to build relationships with readers and pay for news production is undermined. And while consumers may benefit in the short run from Google/Facebook’s free use of publishers’ content by getting a convenient service ‘for free’, the tradeoffs get more complicated in the long run – factoring in the reduction in publishers’ quality and even ability to produce news.  Having Google and Facebook as intermediaries will mean that publishers’ choices between ‘ad-funded’ and ‘subscription’ models for their titles do not reflect the underlying distribution of consumer preferences (as they should), but are driven by Google’s and Facebook’s own preferences for the ad-funded model – a model in which the quality of news and journalism are under threat.1 The platforms’ argument that they in fact provide free traffic to news outlets, so what’s not to love, is a convenient posture with costs to both publishers and society. 

The dispute has rumbled along for some time without being taken up by antitrust agencies. Isn’t this a copyright issue? Platforms don’t compete with publishers in the production of news, do they? So why would they want to harm them? But as the debate on the scope of antitrust enforcement has progressed, with growing calls for it to tackle more vigorously not only exclusion, but also exploitation and expropriation of value, we are seeing this issue finally being tackled.  Building on its influential 2019 Digital Platforms Inquiry (ACCC 2019), the ACCC was asked by the Australian government in April to “develop a mandatory code of conduct to address bargaining power imbalances between Australian news media businesses and digital platforms, specifically Google and Facebook”. It produced a concepts paper in May, and has now put forward a solution that is innovative and creative. Importantly, it may also have the potential for further application to the question of resolving disputes between gatekeepers and their “dependents”.2

On 31 July, the ACCC issued draft legislation for consultation on a News Media and Digital Platforms Mandatory Bargaining Code.3 The draft bill has made waves because it eschews a prescriptive solution in which an agency/regulator has to make lengthy inquiries to form a view on the appropriate level of compensation, digest endless submission from both sides and then exercise judgement to come to ‘the right number’. Instead, it says to the two sides: some compensation to publishers is due, but I won’t calculate what it should be for you. You can make confidential information and data requests, both (for publishers) to determine the value of news to platforms and (for platforms) to determine the cost of producing news.  With this information, go find a solution yourselves. However – and here is the clever twist – do not think you can dilly-dally indefinitely without reaching a conclusion: after a while, if you don’t come to a private settlement you will need to appoint an arbitrator panel and the rules of the arbitration will be ‘final offer’. Meaning: after negotiations are exhausted each of you needs to make a single best and final offer, and the arbitrator can only choose one of them. No further round of offers, no ‘cutting it in the middle’ by the arbitrator: it is one or the other. This concentrates the mind, doing away with posturing over multiple rounds and encouraging ‘reasonable’ valuations to emerge on both sides. 

This feature, known as (American) ‘baseball-style arbitration’, has been previously advanced as a way of addressing disputes over ‘FRAND value’ in litigation around Standard Essential Patents, because it obviates the need for expensive litigation, court disputes, and a judge having to decide what a poorly defined ‘FRAND value’ is.4 It has also been used a few times, including by the CRTC and the FCC (in the form of merger conditions in NewsCorp/Direct TV and in Comcast/CNBC) for resolving disputes in pay television markets over payments from distributors to channels.  We discuss its economic properties below, and other features of the scheme in the rest of this section. But most interestingly, there may be more to it: in a world where regulation of ‘gatekeeper platforms’ appears inevitable and efforts are underway to design ‘good’ regulatory solutions to get to a ‘fair’ deal between two sides (the owner of a platform and the businesses operating on it), is this an approach worth thinking about more broadly? We think there is potential and discuss this in the next section of this column.

Before doing so, we should mention that not everyone shares our positive assessment of the ACCC’s proposal. Ben Thompson, a widely followed tech commentator, has described it in a colourful blog post as ‘a shakedown’ of Google and Facebook for no other reason that they “make a lot of money, which the Australian news media business feels entitled to” (Stratechery 2020). He further posits that “if anyone should be paid in this relationship it is Google and Facebook”, because news content “shows up for free” on them and they “do not charge [publishers] for any traffic [that they send their way]”. He asks “what exactly are the terms to be bargained over? … It is impossible to … determine the appropriate share of revenue that simply doesn’t exist”. Ben Evans, another widely followed tech analyst, has said “News has little direct value to internet platforms:  that’s not where the ads are” (Benedict’s Newsletter 2020).

We disagree with these assessments because it assumes that platforms get no benefit from displaying news. The issue is not that Google and Facebook make little or no advertising revenues directly from displaying news content. Rather, the issue is that there are other sources of value created for them in the relationship (for example garnering information on consumer preferences for online ad sales and other uses and/or keeping consumers inside their ecosystem by pre-empting publishers from creating direct relationships with readers).  To which we would add that there are values like the quality of journalism that governments (and democratic societies more generally) may want to preserve and are undermined by the ‘race to the bottom’ fuelled by the ad-funded model.  

Thompson indeed concludes that “[i]f Australia wishes to tax Google and Facebook and give that money to news media companies, it is free to do so. I could even entertain an argument that taxing Aggregators generally and funnelling it to a public good like the media is good for society”, but “(d)on’t pretend like this regulation is some sort of honest attempt at collective bargaining when it is the extraction of money at the point of the proverbial gun”.  We think that this overlooks that there is an economic rationale (on top of the societal rationale) to funnelling more of the joint value created by the consumption on news on dominant digital platforms to publishers. And because of the disparity in bargaining power, this needs to be mandatory. In other words, this is regulation.  But given that it’s regulation, it is better to let the size of the payment be bargained between the two sides (with an arbitrator as backstop) than have a government (or its appointed regulator) set a potentially arbitrary (and inefficiently homogenous) ‘tax’.

Key elements of the Australian Draft Bill

The Australian Draft Bill mandates that digital platforms (for now, Google and Facebook) negotiate with Australian ‘news media businesses’ over compensation payable to the latter over the use of news content (to populate carousels and other parts of the ‘page’). Having determined that compensation is due, the draft bill requires the parties “to negotiate in good faith over all issues relevant to news on digital platform services”. First, news businesses can bargain either individually or collectively.  Secondly, it also covers non-price elements such as giving 28 days’ notice for algorithmic changes that are likely to have a significant effect on the ranking and presentation of news; giving original news content recognition through branding; and providing information about how and when they use users’ data collected through their interactions with news content. The latter has been a persistent issue for publishers as they typically have no idea about how much data digital platforms collect on their users, and what use they make of them. Importantly, the draft Bill also grants to both sides information gathering powers to allow them to gauge the benefits (direct and indirect) to digital platforms of using news content and the costs to publishers of producing news. 

The rules for compensation are especially interesting. The idea of getting the parties to bargain privately is not new (and indeed it is part of the obligation issued in France by the Adlc as well). What is different is the adoption of a compulsory ‘final offer arbitration’ on payment if negotiations within a fairly tight deadline (three months) do not produce an agreement.5 The arbitrator must decide on one or the other offer within a month, taking into consideration “the direct and indirect benefit that the content of the news business provides to the digital platform’s service”, “the cost to the news business of producing news content”; and “whether a particular payment amount would place an undue burden on the commercial interests of the digital platform”. The outcome is binding. The regulator (the ACCC) stays out of the way, other than it can make a submission to the arbitrator (which the latter can however ignore).  There are also provisions that digital platforms cannot ‘discriminate’ in any way against news businesses on the basis of their participation in the code (specifically including favoring international over Australian content), and penalties for non-compliance. 

Why this set up has desirable properties 

The proposal has attractive features from an economics perspective.  Economists have studied bargaining for decades, with the principles underlying this literature seeking to capture the reality of business negotiations (Brandenberger and Nalebuff 2011).  The dominant paradigm for ‘non-cooperative’ settings like negotiations between business partners is the asymmetric Nash bargaining framework.  It says that outcomes of a bilateral bargain depend on three factors:  (1) the total profit to both parties from reaching an agreement, (2) each party’s ‘threat point’ profit in the case of a disagreement, and (3) each party’s ‘(Nash) bargaining power,’ written as a percentage between 0 and 100 (with the sum of the bargaining powers equal to 100%).6 The total profit from (1) less the profit from each party’s threat point in (2) defines the possible ‘gains from trade’, a.k.a. the size of the ‘pie to be split’, and the bargaining power in (3) defines how much each party gets of this pie (Muthoo 1999).  

Several elements of the ACCC’s proposed Bargaining Code can be understood naturally within this framework.  The first is the granting of information-gathering powers to both platforms and publishers.  The bargaining theory just described assumes both sides know the value to digital platforms of Australian news provision and the costs to publishers of producing news [(1)] as well as each party’s threat points [(2)].  But this is unrealistic unless publishers can learn how platforms benefit from their provision of Australian news and platforms can learn how much it costs to produce that news.7 Each has to also learn the other’s (and their own!) threat point profit, a harder undertaking, but also fostered by the ability to gather relevant information (e.g. how much does Google or Facebook profit from online advertising when they don’t have information on consumers’ news consumption; how many Australian users would go directly to publishers’ websites if Google or Facebook no longer offered news content?).8

The second is granting the ability for news publishers (if they so choose) to bargain collectively in the new system, as well as the non-discrimination requirement that would prevent them from favouring non-Australian news content relative to Australian news content.  In line with the concern that digital platforms have greater bargaining leverage in negotiations with news publishers, allowing collective bargaining allows (especially smaller) publishers to get together and achieve economies of scale in negotiation to improve outcomes relative to their negotiating individually.9 Non-discrimination is important because Google would have a stronger threat position if it could substitute international news content for Australian news in the absence of an agreement with all Australian publishers.  It can also be beneficial in circumstances where individual negotiations would lead to ‘divide and conquer’ strategies.  With collective bargaining (if chosen) and a non-discrimination rule, Google would have to consider not being able to link to any news, Australian or otherwise.  If such links have value, this will lower Google’s threat point, improving outcomes for Australian news publishers as a whole.10

But the important feature, that makes a real difference, is the use of ‘final offer’ (‘baseball-style’) arbitration as a backstop to voluntary negotiations.  We’ve discussed the three Nash factors in the context of governing outcomes of such voluntary negotiations.  But they also play an important role by defining (expected) threat points in the case no such voluntary agreement can be reached. A known benefit is that when submitting final offers to an arbitrator, each party has an incentive to present an offer relatively close to that which they consider a likely outcome of the voluntary agreement (as evaluated by the arbitration panel), for fear that if they present something outlandish that favours themselves too much, the panel will simply choose the other party’s proposal.  This advantage has long been recognized in the economics literature, both by fostering final offers that are close to each other relative to what might arise in bargaining outcomes backstopped by other arbitration rules, as well as producing agreements that don’t require the use of arbitration at all.11 But there is more: this backstop can also mean that collective bargaining is not the only way to protect publishers from uneven power. Should they choose to bargain individually, the fact that there is an ‘arbitration shadow’ to the negotiations means they can make the case to the arbitrator that they should not be offered just the incremental value to Google of their content. Arbitration can mitigate the ‘divide and conquer’ problem.   

To which we would add the final advantage of the proposed method:  it ultimately gets the parties to do the hard work of quantifying the key inputs into the value of news, not a regulator.  This is critically important as it goes towards addressing a key asymmetry of information problem that plagues the economics of regulation generally:  digital platforms and news publishers are far more likely to know the value of news content to digital platforms and the costs of news production to news publishers than would a regulator, regardless of that regulator’s information gathering powers.  The parties in the proposed method have the same powers, but in addition have the industry expertise to know much better how to use the gathered data to inform valuation estimates (or at least getting close).  Indeed, it is this insight – that mandatory bargaining supported by a final-offer arbitration backstop can likely out-perform a dedicated digital platform regulator – that inspires the second half of this column.

A few more observations

Before turning to these more general insights, we close with three final observations.  First, we think it could help to provide more clarity about the goal of the arbitrators.  When the arbitrator panel is presented with different final offers (even with FOA, there will be a gap), how should it decide between them?12 In various regulatory settings, so-called FRAND mechanisms propose “fair, reasonable, and non-discriminatory” rates and prices.  Would a 50/50 split of whatever is the arbitrator’s best estimate of a hypothetical ‘pie’ be fair and reasonable?  Or ‘is news different’ given the essential role of informing voters in democratic societies such that publishers should get more? Or are such considerations beyond the scope of an arbitrator’s remit?13 This is a non-trivial issue, as bargaining theory suggests that offers will be centred around the expected value of the arbitrator’s preferred settlement, with the gap between them increasing in the uncertainty parties have over this value (e.g. Gibbons 1992).  And more broadly, even though the ‘direction of travel’ is clear (that is, the belief that the status quo is biased in favour of the platforms), magnitudes matter, so it would be useful to provide more specific guidance in the final bill.  

Second, everything discussed so far focuses on the bargaining over the payment for news content.  But we also argued that the data collected by digital platforms on Australian consumers’ news consumption may have significant value.  The draft bill requires a list and explanation of the data digital platforms collect about news publishers’ users from all their services, as well as how publishers “can gain access to [this] data,” but one could imagine that such access is critical to the ongoing operation of a news business.  As such data is “jointly produced’ (i.e. by virtue of a user on a digital platform being presented with publishers’ news content),  it would seem reasonable that it would be ‘jointly owned,’ and that the minimum standards be expanded to include a right for news publishers to have access to the data that currently only the digital platforms have, particularly data that links news consumption of a given publisher to a specific individual over time.  Failing this, negotiations between digital platforms and news publishers should perhaps be broadened in the final bill to include both price and non-price (esp data) considerations.14

Finally, again our predictions that the proposed code would address digital platforms’ exploitation of news publishers is easiest to map out under the assumption that publishers would indeed bargain collectively.  But they may prefer not to, as there may be important differences between their individual circumstances.  The ‘final offer’ backstop could in principle provide protection against a ‘divide and conquer’ strategy of making ‘sweetheart deals’ with one or a small number of publishers, if clarity is given to arbitrators on how to evaluate offers.  In principle other options could be on the table, e.g. most favoured nations (MFN) provisions across publishers (perhaps within groups of publishers of similar size)15 or a softer requirement that the terms of agreed deals be made available to arbitrators (and other publishers?) to serve as benchmarks (ensuring any ‘sweetheart deal’ could - with possible adjustments - serve as a basis for compensating other publishers in similar circumstances).  It is therefore essential to think through the implications of such a strategy and ensure the code can address it.  

Can this approach have wider application to digital platform concerns around exploitation and expropriation?

The explicit purpose of the ACCC code is “to address bargaining power imbalances between news media businesses and Google/Facebook”. But could the approach be of wider application?  As well as concerns around ‘foreclosure’ (which are a major issue, for a platform can leverage power in multiple ways to marginalise actual or potential competitors), we worry a lot about the enormity of the inequality between platforms which unilaterally dictate rules and the businesses they host, and the dependency inherent in these relationships. This creates numerous opportunities for platforms to exploit and expropriate value from businesses that contribute to create that value.  Yes, this is not inevitable and not uniformly the case – different platforms with different monetization strategies have different incentives for this and examination of their conduct needs to consider the peculiarities of their specific setting.  But this said, concern about exploitation and expropriation of value are pervasive, and were centrestage at the recent House Judiciary’s Antitrust Subcommittee Hearings with the CEOs of Amazon, Apple, Google and Facebook.16 

On the shift from competition to regulation for dealing with exploitation

We have argued for some time in favour of expanding antitrust rules to deal directly and fully with exploitative abuse (and not just foreclosure) (Caffarra 2018, 2019).17 But while theories of exclusionary abuse are well developed in economics and also in antitrust precedent (‘Microsoft/Netscape’ being the paradigm agencies are comfortable with), the rubric on ‘exploitative’ abuse is much more limited. Exploitation cases have typically been limited to excess pricing in pharma, and while exploitative abuse is just as much a part of the Article 102 on abuse as exclusion,  it has been less traversed. Aside from the German BKA and the French Adlc, which have embraced this concept and acted on it a few times now, there is still reluctance on the part of many antitrust agencies to expand the toolbox in this direction.  Added to which, there is the vexed issue that antitrust enforcement does not move at adequate speed (for all sorts of reason, including strategic delay by defendants, plus enormous longueurs in both administrative and court disputes). France has sought to move the dial here by embracing interim measures, but they remain a rare breed.  

Frustration with the pace of enforcement and “we already left it too late”, plus “we need a specialist body” have brought forward regulatory initiatives, where exploitation seems an easier fit.  The most advanced is the UK, where following the Furman Report the government adopted the suggestion of creating a Digital Markets Unit (within the Competition and Markets Authority, and with ‘input’ from telecom regulator Ofcom). The CMA set out the proposed “rules of the road” to deal with “platforms of strategic significance” in its recent and well-regarded market study on online platforms and digital advertising (Competition and Markets Authority 2020). As stated in the report, the aim is to create “a new regulatory regime with strong and clear ex ante rules which can be enforced rapidly by a dedicated regulatory body” (para 7.8). The proposed approach is two-pronged: an “enforceable code of conduct” with behavioural rules to “govern the behaviour of platforms” with power over an “important gateway” (with objectives such as “fair trading”, “open choices”, “trust and transparency”); and a separate list of “pro-competitive interventions” that the DMU can also pursue to “tackle the underlying causes of market power” (mandating interoperability, mandating third-party access to data, mandating data separation, mandating changes in default standard and even separation/break-up). 

In Brussels, DG Connect is planning to build on the recent Platform-to-Business Regulation (EU) 2019/1150 and is gearing up for a major push in the fall with regulatory design. In its preliminary scoping documents, it has outlined the need to “adopt a new and flexible ex ante regulatory framework for large online platforms acting as gatekeepers”, and mentioned also a two-pronged approach with “prohibition or restriction of certain unfair trading practices (‘blacklisted’ practices)” as well as “adoption of tailor-made remedies … on a case-by-case basis where necessary and justified,” “adopted and enforced by a competent regulatory body (in principle acting at the EU level)” (see European Commission 2020). In effect, they are proposing a digital EU super-regulator.

While the CMA’s Final Report emphasises “the code for Google and Facebook should take the form of high-level principles rather than detailed and prescriptive rules” (para 7.67) and the Connect document also talks of a “flexible” framework, there expectation is that the platforms are in for fairly prescriptive regulatory regimes that are going to be drawn up in Brussels and London in a hurry. And while some of it will need to be prescriptive, the question of how to best design regulation rules is very hard.

Do we have enough expertise? From where? Regulatory economics as a discipline is in decline since most of the “hard” questions of the past (in electricity, in telecoms, in transport) have been mooted by the deregulation of the (presumed) potentially competitive elements of the supply chain.  It is no longer an active field of study among academics. Some regulators have spoken about applying “the principles of utility regulation” – indeed the Connect scoping document explicitly states “While recognising the many differences, experience from the targeted and tailor-made ex ante regulation of telecommunications services can serve as an inspiration in this regard, given the similarities deriving from network control and network effects”. But applying the principles of telecoms regulation (whether cost-plus or incentive/price-cap regulation) would likely be a mistake in markets with features very distinct from those of old-fashioned platforms (e.g. two-sidedness, significant networks effects and/or economies of scale and scope, hundreds if not thousands of affected industries, and more). 

We therefore ask, “Could a leaf be borrowed from the ACCC’s approach, at least for some of these issues?”  Could some of the proposed “private negotiation with a backstop” solution have broader application? Could it at least be a complementary approach that cuts through the difficulties of designing the right rules and getting a regulator up to speed in a world of asymmetric information and potential regulatory capture? 

Could the ACCC’s 'bargaining code' be a model for 'decentralised regulation'? 

Much of the current debate on platforms is the question of ‘rent extraction’ by a ‘gatekeeper’ that unilaterally (and uniformly) sets the rule of engagement for businesses that are reliant on it for distribution and visibility. We do not enter in this paper into the merit of claims set forth by complainants in various jurisdictions against various platforms that their rate of ‘extraction’ is excessive.  The typical background is that some uniform rule was originally adopted, years back, by the platform – based on some underlying principle, and affecting a relatively small number of counterparties; but as time has gone by, and the number of counterparties and their circumstances have grown and changed, uniform rules are increasingly perceived as ‘unfair’, and a path is being beaten to competition agencies and the courts. 

The classic defence from the platform in these cases is that the underlying principles that motivated the original solution have not changed, that a uniform rule creates a ‘level-playing’ field, and as terms have remained broadly the same over time, ‘rent extraction’ has not worsened even if market power increased. But as the number and heterogeneity of counterparties exploded, and platforms have become a ‘gatekeeper’ to a large number of businesses and volume of trade, is it credible that a single rule applying uniformly across the piece still does the best job at the getting the right tradeoffs? Is the claim one is still pursuing a ‘level playing field’ by extracting the same share of revenues from all adequate?  Circumstances have changed over time, what was thought of as a ‘level playing field’ once may no longer be fair to all. 

While antitrust agencies are reluctant to intervene as direct price setters, incoming regulation will undoubtedly be willing to mandate price setting rules.  Various ‘solutions’ will be discussed in the coming months, perhaps involving mandated access pricing and tiered fee schedules without a clear basis.  The question is: why could there not be also a solution over a ‘fair’ payment based around private bargaining, underpinned by a baseball-arbitration backstop? As outlined, this would have the benefit of allowing the parties to leverage their private information and come to a private agreement on terms that suit them both.  We think of it as ‘decentralised regulation’.

There are of course practical considerations.  First, we take for granted here that the hurdle of demonstrating a platform needs to be subjected to regulation (i.e. it has ‘strategic significance’ or is a ‘gatekeeper’ in the UK and EU terminology respectively) has been met – though in practice this will need to be established as a threshold matter in each case.   The second relates to scope.  The ACCC’s bargaining code focuses on news publishers, where individual bargaining is conceivable.  How feasible would it be to mandate platforms bargain with potentially thousands of counterparties across multiple industries? The notion of a single uniform rule certainly had the attraction of being easily implementable and minimise transaction costs. Negotiating with counterparties is time consuming and requires some effort. On the other hand, ‘negotiating’ is what companies do. Why should it not have some attraction here, especially if limited to larger counterparties?  One could imagine some hybrid rules: non-discrimination rules and smaller counterparties negotiating collectively as a category, with large players engaging in individual bargains.  Or solving the problem for the vast majority of operators (e.g. the smallest players) with an access price structure, with a generous rate (perhaps even zero) for the smallest players, collective bargains for medium players, and bilateral bargains for the large, accompanied by minimum standards on non-price elements, including data collection and use, for all.   

Lest this sound daunting, note that these problems must be solved in any regulatory design.  And it further highlights how regulating digital platforms will necessarily be different from traditional utility regulation:  we already worry that a regulator cannot understand the economics of a single industry as well as the firms it is mandated to regulate; the problem is multiplied in the case of digital platforms.  While it may not be trivial to populate arbitration panels with the necessary skillset, that seems easier than expecting a regulator to build expertise in each itself.  And even once industry-specific issues are settled, there remain multiple “yes, buts” in terms of practical implementation. For instance, what is the ‘right’ (homogenous?) surplus division rule (i.e. the ‘Nash bargaining power’ parameter)? Still, our guess is that it may be substantially more efficient to move forward with a rule that allows for private negotiations backstopped by arbitration, rather than to accept a prescription that “the right tax is X”.  

What would platforms make of such an approach?  Well, if the threat of regulation is serious and close enough, some may anticipate that some voluntary change may be better than a mandated solution imposed by a relatively ill-informed regulator – and bargaining is a commercial negotiation, after all. Alternatively, a regulator could itself include a mandatory bargaining code, with a backstop, as one aspect of a menu of approaches. There seems to be no reason why bargaining backed up by FOA could not be mapped into these discussions.  

One thing is clear:  change is coming to the way in which large digital platforms are going to manage the relationship with third parties that are subject to their rules.  We feel that ‘decentralised regulation’ inspired by the ACCC’s Mandatory Bargaining Code can be one part of the set of possible solutions.


ACCC (2019), Digital Platforms Inquiry Final Report, 26 July.

Aghion, P, M Dewatripont and P Rey (1994), “Renegotiation design with unverifiable information,” Econometrica 62(2): 257-282.  

Benedict’s Newsletter (2020), “Two stories of dysfunctional regulators this week.  First Australia”, 23 August (

Brandenberger, A and B Nalebuff (2011), Co-opetition, Currency.  

Caffarra, C (2018), "A Radical Antitrust Manifesto for Digital Platforms?", Association of Competition Economists Annual Conference, Bologna, December.

Caffarra, C (2019), “Antitrust Enforcement in Digital Space”, Keynote, Competition and Consumer Law Conference, Law Council of Australia, Melbourne, 30 August.

Caprice, S and P Rey (2015), “Buyer Power from Joint Listing Decision,” Economic Journal 125: 1677-1704. 

Çelen, B and O Özgür (2018), “Final-offer arbitration with uncertainty-averse parties,” Games and Economic Behavior 109: 484-500.  

Competition and Markets Authority (2020), Online platforms and digital advertising market study, 1 July.

De Klippel, G, K Eliaz, and B Knight (2014), “On the selection of arbitrators,” American Economic Review 104(11): 3434-3458.  

European Commission (2020), Digital Services Act, Inception Impact Assessment.

Farber, H (1980), “An analysis of final-offer arbitration,” Journal of Conflict Resolution 24(4): 683-705

Gibbons, R (1992), A Primer in Game Theory, Pearson 

Henning, L et al. (2020), "Making Facebook, Google pay for news content requires regulatory collaboration, not reinvention", 14 May.  

Larsen, B J (2019), “The efficiency of real-world bargaining: Evidence from wholesale used-auto auctions,” NBER Working Paper 20431.

Lemley, M and C Shapiro (2013), “A Simple Approach to Setting Reasonable Royalties for Standard-Essential Patents”, 28 Berkeley Technology Law Journal 1135.

Lomas, N (2020), “Google Warns Users in Australia Free Services are at Risk if Forced to Share Ad Revenue with Big Media”, Techcrunch, 17 August.

Muthoo, A (1999), Bargaining theory with applications, Cambridge University Press, Section 2.8.

OECD (2017), “Collective “bargaining in a changing world of work,” Chapter 4 in OECD Employment Outlook 2017, OECD Publishing, Paris.

Rolnik, G et al. (2019), “Protecting Journalism in the Age of Digital Platforms,” Stigler Center Committee for the Study of Digital Platforms, Media Subcommittee, 1 July.

Stratechery (2020), “Australia’s News Media Bargaining Code, Breaking Down the Code, Australia’s Fake News”, 20 August.

Yaïche, A (2020), "Google may see extension of French antitrust order to negotiate with press publishers", 12 August. 


1 For an extensive discussion from leading academics studying the media, see Rolnik et. al. (2019).

2 The French Adlc has also been investigating a complaint filed by French publishers accusing Google of abusing its dominant position by breaching part of the EU's new copyright directive, and issued a decision for interim measures as well as an order in May for Google to reach agreement with the publishers. Talks are underway though publishers were considering filing a complaint for non-compliance (see Yaïche 2020). Similar complaints are being considered in the UK. More generally, countries from Japan to Malaysia to Brazil to South Korea to Canada are seeing complaints from publishers to demand platforms pay their “fair share” (see Henning et al).

3 See,, and

4 See for instance and, among others, Lemley and Shapiro (2013).

5 The bargaining mandate and final-offer arbitration backstop is not available to Australian public broadcasters, the Australian Broadcasting Corporation (ABC) and Special Broadcasting Service Corporation (SBS), as “advertising revenue is not the principle source of funding for public broadcasters.”

 6 ‘Nash bargaining power’ is specifically defined as the share of the ‘pie to be split’ and is distinct from both (1) improved bargaining outcomes due to changes in one or the other party’s threat point (which some scholars call changes in (Nash) ‘bargaining leverage’) and (2) bargaining power as used in colloquial language (which often speaks to general bargaining outcomes, thus encompassing both ‘Nash bargaining power’ and ‘bargaining leverage’).

7 The value to a platform from providing news content does not mean simply the advertising revenue it sells on pages with news links. Indeed Google does not sell ads on its Google News service and so generates no value from this channel.  But that doesn’t mean Google doesn’t benefit from providing news content.  First and foremost, there is plausibly considerable information about consumer preferences that can be inferred from knowing which newspapers a person follows (The Guardian or The Times?) as well as the specific articles they search for, link to, and ultimately read.  Such data is likely to be an important and valuable complement to both Google and Facebook’s primary business of selling online advertising.  Furthermore, Google and Facebook arguably seek to be a “one-stop shop” for all consumers’ information needs.  In a world where neither provided news content, the quality of their services would decline, but importantly this would provide publishers with greater direct access to consumers.  Because many consumers care about news, this implies increased data to foster their own online ad businesses, a greater ability to build a platform or “ecosystem” distinct from Google and Facebook, and a greater ability to develop subscriber relationships with consumers.  Each of these things could have considerable value, value that currently isn’t being shared with publishers, and these possibilities form the basis for our belief that rebalancing bargaining power as the proposed rules seek to do would yield an outcome in which both would pay publishers for news content (and reasonably so).  

8 Indeed, recent academic research has shown negotiations can fail even when there are gains from trading when each side has private information.  See, e.g., summarizing Larsen (2019).

9 Caprice and Rey (2015) make the same point in the context of buyer groups. See also the growing interest in collective bargaining to address perceived imbalances in worker bargaining power summarised at, with further details in OECD (2017).

10 Of course, if they bargain collectively, Australian publishers would then have to decide how to divide whatever they can negotiate from the platforms, itself a non-trivial problem.  But collective bargaining would likely enhance their overall bargaining outcome in negotiations with digital platforms, whatever is this second-round decision rule.

11 See Çelen and Özgür (2018) for a recent contribution that summarizes the relevant economics literature and De Klippel et al. (2014) for a suggested method for selecting arbitrators.  See also Aghion et al. (1994) for a similar mechanism in the context of contract renegotiations. Final offer arbitration also has the value of not requiring an arbitrator to defend any particular award different from that proposed by either party, as often arises in commercial arbitration proceedings, leading to decisions that are simpler, faster, and less likely to be appealed by either side.

12  From a theory perspective, this would be analogous to providing guidance on the third, “Nash bargaining power,” factor that governs bargaining outcomes.

13 Or, going the other way, might such public-interest considerations impact other aspects of the code beyond the compensation provided for news?

14 While including non-price elements may complicate somewhat the selection of which of the two final offers to choose should voluntary negotiations break down and the negotiations go to an arbitration panel, in principle there is nothing in the structure listed above which would forbid having negotiations be over bundles of access to data and financial remuneration.  

15 Our thanks to Patrick Rey for raising this issue.  

16 See  Coverage has been extensive, including, and many more blogs, tweets and op-eds. 

17 For related views on the regulation of digital platforms, see also Caffarra (2020).

2,520 Reads