Big Tech, Jobs and Cass Sunstein’s Nirvana


In a short Bloomberg column titled “A New View of Antitrust Law That Favors Workers”, Professor Cass Sunstein warns that “Many labor markets are not competitive“, and that antitrust’s narrow focus on consumer welfare (CW) is not up to the predicament.

Sunstein’s opinion is based on research by Eric Posner, Glen Weyl and  Suresh Naidu, a group dubbed the “New Chicago School“.

Their work suggests that because wages are artificially low, qualified workers refuse to take jobs. Instead they “exit the workforce and rely on government benefits“. On some markets this problem is caused by “barriers to competition” and excessive “market concentration“, in plain words a small number of hirers. So far, so good.

But the low wage argument breaks down a sentence after when of all possible examples of monopsony power in american labor markets, Sunstein singles out the tech industry.  In tech, writes Sunstein, “only a few employers can survive, perhaps because of “network effects,” which arise when the value of a good or a service increases if a lot of people use it” think of Google or Facebook). And we are yet to understand how “big companies (like Apple or Google) might use their market power to hurt employees“.

This is incorrect as a matter of economics, law and logic. Let us start with the economics. In the past decade, the large tech industry has added more jobs to the US economy than any other sector.  The graph above shows the year over year employee headcount at each of the FAANG (including Microsoft) v job creation of the US economy. This data is orders of magnitude inconsistent with a theory of monopsony power. Anyone with basic economics knowledge understands that this would require a correlative decrease in labor participation at the FAANG. Moreover, there is every reason to believe that the FAANG did not only create new jobs for wealthy computer scientists. Amazon (blue line), the largest employer of all FAANG, has been predominantly adding jobs in warehousing. And Apple has essentially grown its workforce in retailing.

Perhaps more suprising is that Cass Sunstein turns a blind eye on several legal developments which show that the state of nature in tech is one of intense competition for talents. How else can one explain cases like US. v. Adobe Systems, Inc., et al., where the DoJ prosecuted several large technology firms – including Google, Apple, Intel, Pixar, Intuit and Adobe –  who had unlawfully agreed to refrain from soliciting, cold calling, recruiting or otherwise competing for each others’ computer engineers and scientists? And what about the ongoing policy discussion on employment contracts in the tech sector, which are said to lock the mobility of knowledge behind barriers like non-compete clauses, contractual perks, stock option plans or simply high wages. If anything, the very existence of those contractual arrangements does not bespeak monopsony power in tech, just the very opposite.

Last, there is logic. Think of the points that we just made on the economics and the law. How can a tech firm be a monopsonist and meanwhile compete with other firms as an oligopsonist on the labor market? Admittedly, in their original study, the writers of the new Chicago School write that the “problem boils down to excessive merger activity, which has led to concentrated labor markets“. Again, big mergers are quite infrequent in tech. Most M&A transactions instead involve the “acqui-hiring” of successful startups, like FB/Instagram or Microsoft/Skype.  But those acquisitions represent low headcounts. And their occasionally astronomical valuations leans against the very idea of monopsony power.

The only way to do justice to Sunstein’s piece is to read it as follows: would tech firms be smaller, competition for jobs would be even stronger and wages even higher. This is an example of what economist Harold Demsetz once called the Nirvana fallacy, namely an idealized, unrealistic situation. However, one would then expect that Sunstein brings evidence that the grass could be greener in tech jobs. And perhaps even more importantly, this shrouded reading that we make of Sunstein’s piece squares poorly with the author’s proven ability to speak clear to the public opinion in the mainstream press.


FAANG: A Firm to Country Comparison

Now that I have read all of the FAANGs 10-Ks, I would like to offer a quick ‘firm to country’ tech comparison. Note that I have excluded Netflix, and covered Microsoft. Disclaimer: the spirit is slightly humorous, and decidedly un-academic.

Microsoft (MSFT): Europe


Like Europe, MSFT is the perennial number two of the gang. MSFT has spent a life (or more) following the leader. It has great skills, committed people and an old history. But it is a tad bit disorganized, so it changes strategy every 2 or 3 years. Its weakness? It has historically attempted to enter in every possible tech business/segment/product: software, hardware, search, content, cloud, ML, etc. Its strength? Like Europe, it has a great legacy product that many people still visit: the old stones of the Windows castle, and its many Office dependences. Put simply, MSFT is civilization to the Western world. Many of us could not (psychologically?) survive if it was ever erased from the map. Note that MSFT could also be Japan.

Amazon (AMZN): China

china flag

Forget the cult of the leader, and focus on the big picture. Like China, Amazon is an army. It’s well organized, hugely hierarchical, and it’s here for the long game. Like China’s catchup growth econ policy, AMZN follows a catchup profit strategy: little by little, piece by piece it buids scale. As in East Asia where labor is cheap, abundant and intensive, low wages and employees are in no short supply at AMZN’s warehouses. And both China and AMZN have been unambiguous about their ambition of a full AI overall at some point: China to control its people, AMZN to deliver its products. Last, but not least, both China and AMZN have beef with President Trump.

Apple (AAPL): Australia


This is a tricky one. Moving to AAPL is like travelling down under. Enter a new world. Find there all of what the big blocks have to offer, but just a better version. It’s more sunny (design), healthy (no virus), and tribal (ozzies are true patriots) than in other organisations. Few outsiders seem to know what’s taking place there: AAPL is secret, Australia is very far. And both have undergone crazy economic growth due to the discovery of scarce resources, iron, gold and diamonds for one, the iMac – circa 20y ago – and the iPhone – circa 10y ago – rents for the other. Last, but not least, connectivity with both AAPL and Australia is not easy.

Facebook (FB): United Arab Emirates


It’s small, compared to the other giants, but it’s loaded with cash. As with oil, we are all addicted to it though we know it’s not a good thing. Like the UAE, FB is a single product entity. It has its social graph. And like with the UAE, we don’t like some things that happen there, but we don’t say much about it. Both the UAE and FB have been desperately trying to find a “blue ocean”. Investments in the future, read R&D intensity, is over the rooftop. FB bets on VR and a flying Internet network. The UAE is all in on museums, conferencing, and knowledge centers.

Google (GOOG): USA


We’ve all been there, or we all dream to go there one day. It’s not old and it’s gigantic. Like the US with the world, GOOG is tech’s big brother. It can tell us where to go, and where not to go. It knows what we do, what we know and what we want. Both GOOG and the US share the same motto: it’s better to ask for forgiveness than permission. Sometimes, that leads them to flout the law. Antitrust law for the former, international law for the latter. There are things that the outside world doesn’t understand about GOOG. The secret sauce behind the search algorithm is like Coke’s secret recipe. Last, but not least, GOOG is like the US: it’s a federation with decentralized business and R&D units and individuals. Some thrive, others shutdown, many compete.

The Big Tech Funding Sideshow

The ‘industry buys academics’ narrative has received significant coverage last week at the Stigler Center conference on Digital platforms and Concentration.

But this time, in relation to a dead person.

In a short exchange, Gary Reback and Bill Kovacic find themselves discussing Phil Areeda’s 1975 Harvard Law Review article on  predatory pricing.

Reback manifestly finds the paper too defendent-friendly.

The ‘funding bias’ attack comes at 49.55′.

Reback throws a:”that research was paid for by IBM“.

Kovacic replies with: I have “looked at that“, but confidentiality restrictions from Harvard university on Areeda’s archives make it impossible to “test that“.

Reback, however, does not leave it here.

He adds: “Bill Baxter told me“.

And follows with “that paid for academic research might be the most influential paid for academic research in the history of antitrust jurisprudence“.

Kovacic’s reply?: “by far … an AVC test for a company with high fixed cost and lower marginal costs … terrific! …“.

Kovacic’s sense of irony speaks volumes.

Attacking a person’s credibility on grounds of disclosed or suspected funding is an effective yet logically wrong way to discredit an opponent’s idea. It is one variant of Schopenhauer’s 38 logical fallacies, namely argumentum ad hominem (and in the case of an attack on a dead person, it falls close to the quandary of proving a negative).

As I wrote elsewhere, the proper way to think about this is counterfactually: but for private sector funding, would we consider that the same research is sound and worthy of discussion. Or, to make the thought experiment even easier: imagine that the originator of the idea is a fictional academic who has not received funding. Would we talk to him?

What I mean to convey here is a suggestion to the tech academics, policy makers and practitioners. The “paid for academic research” narrative is a sideshow, a waste of time.

If f you are not convinced, think carefully about the following questions:

  • Can’t people who inherited a family business talk about employment and job market reform?
  • Can’t people who enjoy substantial wealth make a political career or hold public office in welfare programme institutions?
  • Shall we no longer use the Haber-Bosch process because it was essential to the German War effort?

My personal hunch is that the answer to all three questions is no.

An idea is good or bad, regardless of the person that invents it.

So let’s not get carried away too much by this sideshow.

PS for European readers: I recall that René Joliet’s seminal book on the Rule of Reason started with an acknowledgment that he had received a scholarship from GE. Yet, no one has ever attacked him for excessive proximity to the private sector.


The China Risk – An Amazon Gem

téléchargementAmazon has traditionally operated in China through a subsidiary named

In its 2009 10-K filing to the SEC, Amazon has a characteristically euphemistic way to describe regulatory fiat in China

Although we believe that Joyo Amazon’s structure complies with existing PRC laws, it involves unique risks. There are substantial uncertainties regarding the interpretation of PRC laws and regulations, and it is possible that the PRC government will ultimately take a view contrary to ours.

Please look attentively, because it cuts deeper than meets the eye.

The 10-K makes subtle reference to the “PRC government” – not courts – as the ultimate ruler on compliance and interpretation of Chinese law.

Was Facebook Negligent?

FB risk factor

Those days, much of my research consists in reading the tech giants 10-K filings to the SEC.

10-Ks have an informative section called”Risk Factors” (Item 1A). This section conveys the reporting firm’s perception of the risks it faces in its daily business. Firms often report on competitive, technological or regulatory threats.  Information found in 10-Ks is trustworthy because it is against the law to make misrepresentations to public authorities and investors.

Facebook’s filings provide non contemporaneous information on Cambridge Analytica-type issues. Already in 2012, we see that FaceBook has an acute understanding of the problem. Facebook point out to a risk that “Improper access to or disclosure of our users’ information, or violation of our terms of service or policies, could harm our reputation and adversely affect our business

The discussion of the problem is even more informative.

Our efforts to protect the information that our users have chosen to share using Facebook may be unsuccessful due to the actions of third parties, software bugs or other technical malfunctions, employee error or malfeasance, or other factors. In addition, third parties may attempt to fraudulently induce employees or users to disclose information in order to gain access to our data or our users’ data. If
any of these events occur, our users’ information could be accessed or disclosed improperly. Our Data Use Policy governs the use of information that users have chosen to share using Facebook and how that information may be used by us and third parties. Some Platform developers may store information provided by our users through apps on the Facebook Platform or websites integrated with Facebook. If these third parties or Platform developers fail to adopt or adhere to adequate data security practices or fail to comply with our terms and policies, or in the event of a breach of their networks, our users’ data may be improperly accessed or disclosed.

Any incidents involving unauthorized access to or improper use of the information of our users or incidents involving violation of our terms of service or policies, including our Data Use Policy, could damage our reputation and our brand and diminish our competitive position. In addition, the affected users or government authorities could initiate legal or regulatory action against us in connection with such incidents, which could cause us to incur significant expense and liability or result in orders or consent decrees forcing us to modify our business practices. Any of these events could have a material and adverse effect on our business, reputation, or financial results

Moreover, Facebook describes this risk factor (“RF”) as an increasing concern over the years. I have tracked the progression of this RF between 2012 and 2017 (see table above). The change is not huge, but this RF still moves from position 12 to 9 in Facebooks’ 10Ks.

Take away: the Cambridge Analytica scandal was a scenario that FB had anticipated as early as 2012.

So was Facebook negligent? This would be a precipitous conclusion to draw. In a large ecosystem, it is easier for rogue players to hide. True, Facebook knew that there was a risk. But the “whether risk” question and the “where risk”/”when risk” questions are different issues. And the larger the company, the more difficult the answer to the second one.

Uber buys JUMP


Uber just bought bike sharing startup JUMP

Several hypos:

  • Illustration of the Kronos effect => Kill JUMP before it diverts too many commuters away from Uber?
  • Tit for tat replication of rivals moves?
  • Platform migration towards a less regulated environment => no drivers, no medallions?

The US v EU Tech Divide: Fake News?


I am in DC today, talking on a panel organised by CCIA. I have been asked to adress the following questions: “Is competition policy approach to tech dividing the US/EU? Increasing influence of the EU in the US debate or vice versa? Possible solutions? Future predictions?”. Hereafter are my remarks.

The idea that EU and US antitrust policies on tech diverge — with the EU being stricter against tech companies — is essentially a misrepresentation of the facts, conveyed in the works of business reporters incentivised to make headlines with good stories. All readers are familiar with the press idea that “when it bleeds it leads”.  Since GE-Honeywell, the US v EU divide is an abundant well that never runs dry for the antitrust press.

But when one looks closely at the facts, the tech clash on antitrust policy is much less clear. First, one should distinguish DG COMP’s antitrust policy from the European Commission broader initiatives. Interventionist policies introduced at the EU Commission level like the digital tax or the DSM platform regulation do not necessarily capture the mindset of DG COMP when it comes to antitrust cases in the tech industry.

Second, one must again draw a distinction between what AAG Delrahim called on Tuesday “speech-related advocacy work” and enforcement. The antitrust leadership in the EU occasionally speaks in somewhat hyperbolic terms of the tech industry. Like President Trump, Commissioner Vestager is never in short supply of tech threats, though they admitedly come in a gentler hand.  Last week, it was reported that the Commissioner did not exclude the idea of a Google “breakup”. But in the parallel world of enforcement, decisional initiatives have steered clear of media catchwords.  Concrete decisional initiatives have been humbler, as can be seen in the Google search case. If one looks at its scope and its remedy, the Google search case is quite a narrow one. True, we have seen a big €2.42 billion fine. But most antitrust experts know that fines are a sideshow, and not the primary concern for the deep pocket likes of Google, Facebook or Apple.

Third, the claims of bias against US tech firms should also be looked at more generally.  The EU traditionally allocates resources and set priorities on sectors where industry concentration is high bc this is where collusion and monopolization risks are deemed most serious. In banking, most cases concern firms from Switzerland and UK. In the chemicals, most cases featured German companies.  And in utilities, most if not all Member States incumbents have been in the crosshairs of the DG COMP. In tech, US firms dominates the industry. They are thus logically subject to antitrust scrutiny.

Fourth, it is short sighted to talk of a EU Commission bias against US tech firms, when most if not all tech disputes in the EU arise from complaints from US companies.  Microsoft was the initial complainant in the Google search case.  Apple is the de facto complainant in the Qualcomm case. Salesforce was very proactive in the Microsoft/LinkedIn merger investigation (btw is an old feat: Sun Microsystems was the complainant in the Microsoft server case).

This brings me to my fifth point. There has been a historic division of labor between the US and the EU in relation to large tech firm conduct, but this may come to an end.  Until recently, the picture was as follows: technology wars take place in the Silicon Valley, regulatory ones in Europe. That division of labour was made possible by the existence of a lower standard of liability in EU antitrust law in general, and in unilateral conduct law in particular. But the Grand Chamber judgment of the EU Court in Intel v Commission from last year has made clear that the competition enforcement should (i) be evidenced-based; (ii) that harm to competition is an empirical question, not a rhetorical one; (iii) that the rule of reason is the proper mode of operation to adjudicate unilateral conduct cases; and (iv) that efficiency is the theology underpinning Article 102 TFEU.  With this background, tech cases that were started before the Intel judgment, but concluded post Intel, may be based on an obsolete – and potentially unlawful – interpretation of unilateral conduct law.  Having sunk large investigative resources in such cases, the Commission may have had no other choices but to decide them on the basis of a now outdated approach (I develop this point in a recent op ed in Competition Law and Policy Debate).

As to future predictions, I don’t want to bet the farm but I see three things coming. They are ordered there from conventional to heretical.

In doctrinal terms, the EU and the US antitrust systems are converging towards an economic approach in unilateral conduct cases, based on a consumer welfare (“CW”) standard or something close. The US case-law has traditionally been oriented this way, and the leadership at the US DoJ has just this week reiterated its commitment to an effects-based approach in unilateral conduct cases.  In the EU, the case-law has moved more slowly, but has eventually seemed to embrace an efficiency standard.  True, the leadership of DG COMP has not yet come full circle on this, though a 10 years old soft law instrument already follows a CW approach. And granted, the legal service of the EU Commission which represents the Commission before the EU courts may stick for a time to the old version of the competition law software.  But this is essentially transitional, and the byproduct of short term litigation and political incentives.

As to enforcement initiatives, several interesting questions may arise for antitrust in the aftermath of the Cambridge Analytica scandal.  One is what do we do if Facebook implements the Zuckerberg threat of introducing an ad-free but paying version of its service? Is this a deviation from consumer welfare? My take is no. We may see more price discrimination on Internet platforms as a result of the Cambridge Analytica scandal.  And this possibly represents an opportunity for disruptive entry by competitive platforms whose business model is less advertisement-centric.

At the political economy level, there’s an elephant in the room, and this is China. Let’s not fake to ignore that Chinese competition policy is about helping Chinese firms dominate non-Chinese markets (and about preventing non-Chinese firms dominate the Chinese market, think of Google or Uber). Tech giants and startups in the Silicon Valey are growing increasingly concerned of the competition from Chinese firms.  They’re unhappy of their lack of market access and of the risk of technological disruption carried by Chinese firms.  As consumers, if the next big thing is invented by China, are we sure that we will obtain full access to functionality?  And more importantly on what terms? I wish to recall here that China is building a massive surveillance state, and as much as I don’t like my data leaked to Cambridge Analytica, I like even less the thought that it may be openly accessible by the Chinese Communist Party.

Video on Tech Giants Breakup




The video of my February talk on the tech giants is available!

A teaser: at 7.50′, I do something heretical.

Tx to my UniSA colleague for a fantastic evening.

Hat tip (literally) to Professors Phil Marsden and Spencer Webber Waller for supplying the necessary equipment.

And my gratitude to Professor Pinar Akman who appeared on a UK TV show and inspired the topic of that talk.

Rule-Based Antitrust & the #DeleteConsumerWelfare Movement

FullSizeRenderThe Consumer Welfare (CW) standard is under attack.

It is criticized as under-inclusive (see this Lina Khan piece).

CW’s focus on short term price-effects (and output) allegedly leaves other economic harms unchallenged (and perhaps even political ones). And tech firms benefit from a regime of unjustified per se immunity because they supply free goods and expand output.

As a result, some voices have suggested to do away with CW in antitrust policy. Matt Stoller – a member of the Open Markets Institute – has even tweeted that CW was an instrument of class domination, invented by Robert Bork “to eliminate constraints on plutocrats“.

All of this betrays a fundamental misunderstanding of what CW is, and of what it is not.

CW is not as much an applied decision-making tool as it is a rule designed to lift (some) discretion from policymakers, and with it to insulate them from political interference, rent seeking and crony capitalism.

This struck me as crystal clear as I was reading First Principles by John Taylor. John is a longtime advocate of a rule-based monetary policy.

Here is what he writes in his book:

In any organisation, a clear, well-specified goal usually results in a consistent and effective strategy for achieving that goal. Too many goals blur responsibility and accountability, causing decision-makers to choose one goal some times, and another goal at other times, in an effort to chart a middle course.

In the case of monetary policy, multiple goals enable politicians to lean on the central bank to do their bidding and thereby deviate from a sound money strategy. There is no justification for an independent agency of government to undertake interventions in those areas. They are best left to Congress and the President to handle through the regular appropriations process.

Last, but not least:

Central bank intervention is a poor substitute for sound fiscal policy and it removes incentives for Congress and the President to do their own jobs well

Notice the parallels with antitrust. All those words apply almost equally well to antitrust policy.

The upshot?

If we throw away CW, we are putting antitrust officials in a tough spot. It is not comfortable to face the demands of politicians. Compromizing is the easy exit. And this just does not happen when clear and predictable rules are in place.

The bottom line?

CW may have some limitations in tech markets. But until the literature advances a substitute devoid of its defects, no case has yet been made to #DeleteConsumerWelfare.