The folly of Cicilline's 'Glass-Steagall for Tech'
© Getty

Rep. David CicillineDavid CicillineJim Jordan, Val Demings get in shouting match about police during hearing House committee approves Big Tech antitrust blueprint House lawmakers fired up for hearing with tech CEOs MORE (D-R.I.), chairman of the House Antitrust Subcommittee, has said repeatedly that he may propose what he calls a ‘Glass-Steagall’ for tech companies — a ban on digital companies that operate platforms, like Amazon’s Marketplace and Apple’s App Store, from selling their own products on those platforms. A report by his committee will be released this month, and this proposal may be in it. Most recently, in a podcast for the Brookings Institution, he says that:

If you think about the inherent conflict, that you would allow Amazon to control the marketplace, set all the rules for your presence on the marketplace, gather information from people who are in the marketplace, learn about what products are selling, where they're selling, what consumers are most interested in, and then using that to roll out their own private label product to compete directly with the people who are in their marketplace

Objections to Amazon selling its own private label products are nothing new, but the analogy with Glass-Steagall is Cicilinne’s own contribution. The analogy is shallow, as are many other contemporary attempts to compare tech with finance to suggest that finance-like regulation of tech is needed.


Glass-Steagall was enacted in 1933 (and abolished in 1999), in the aftermath of thousands of bank failures across the United States during the Great Depression. The logic then was to create a barrier between banks’ investment arms and retail banking arms, on the assumption that this would shield more prudential retail arms from losses made by the relatively riskier investment sides of those banks.

It’s obvious why, in principle, you might want to protect ordinary people’s deposits from losses made by a bank’s more risk-seeking operations, although the wisdom of this is debatable in practice. As UK financial regulator Oonagh McDonagh points out, for example, Glass-Steagall wouldn’t have done much to avoid the 2008 financial crisis, since the root problems were to do with mortgages on retail-side lending, and it was the collapse of independent investment banks without retail arms, like Bear Stearns and Lehman Bros, that precipitated the wider financial crisis. And forcing banks to be less diversified — across regions, businesses, or asset types — tends to make them less stable.

But regardless of the propriety of Glass-Steagall in the banking sector, there’s no real analogy with tech platforms here at all.

Cicilline is concerned that tech companies have an unfair advantage over others by being able to set the rules and gather data from platforms, and also able to sell their own products on those platforms. That’s a completely different issue than the one Glass-Steagall was trying to address. The similarities are barely even cosmetic — it is as if he described the split between male and female athletes as “a Glass-Steagall for sport.”

That spurious analogy aside, limits on whether platform companies can sell their own products on their platforms are a bad idea if your goal is cheaper, better products and more choice for consumers. Imagine if we proposed a rule like that for grocery stores. Walmart sells its own brand of milk, but also manages the stores in which other milk brands compete and sell to customers. Would consumers benefit if Walmart either had to discontinue its milk offering, or stop selling milk made by other companies? If not, what is it about (say) Amazon that makes Amazon’s store brand products bad for consumers in a way that Walmart’s aren’t?


It may in fact be that Amazon is using data it has to figure out where the best profit opportunities are. But that’s a good thing! If it took Amazon to figure out that Lightning charging cables for iPhones were overpriced, and that a cheap, reliable offering would be popular, we should welcome and encourage that. The fact that consumers can now buy those charging cables for less than the price that Apple charges is a good thing. Maybe Amazon’s competitors don’t like it, but that’s what competition looks like.

Or think about Apple’s own “private label” products on its App Store. Right now I can choose from third party notes apps for my iPhone, including great products like Bear or Onenote, or I can choose Apple’s Notes app. Is this a bad thing? Would I and other iPhone users be better off if Apple either had to kick off products like Bear or discontinue its own Notes application? The idea seems ridiculous — and yet, that is the logic of the ban on companies that run platforms being able to offer their own products on those platforms.

Cicilline is not alone in comparing financial markets with digital markets and concluding that the similarities make regulation necessary. Dina Srinivasan, a former digital ad executive, argues that real time digital advertising exchanges like Google’s ad business are comparable to financial markets, and that Google in particular succeeds because of practices that are banned in financial markets.

For example, Srinivasan observes that an ad exchange, OpenX, moved to Google Cloud because (according to Google) it worked with Google’s ad products with less latency than other cloud hosting. She compares this with co-location, the practice in finance of traders locating their computers physically close to the exchange computers so their high-frequency trades can go through more quickly. And, she says, “in financial markets, colocation practices are tightly regulated”.

This is actually an overstatement: colocation practices are decided by private stock exchanges themselves, not by the regulator (the SEC). But in any case, as with Cicilline and Glass-Steagall, the comparison is simply inapt. To the extent that co-location is considered to be unfair or wasteful, it is because high-frequency trading is not considered to be useful — so any benefit co-location gives is a zero-sum benefit, where the closer bidder profits at the expense of other bidders, and society is no better off because the transaction has been done a fraction of a second faster. Whether or not this is true — high-frequency trading actually does have many benefits, including making markets more liquid and less volatile — to think that co-location itself needs to be stopped requires a belief that faster trading speeds from co-location do not have any overall positive-sum benefit.


But this is not the case when it comes to real-time ad bidding. In this case, the speed of ad bidding is enormously important, because it determines the speed at which a website can be delivered to the user. Speed is extremely valuable to all parties involved, and whatever the publishers, advertising intermediaries and advertisers themselves can do to speed up the process is good for everyone involved, especially the website publishers and their users.

Srinivasan makes many other points — too many to address here. But her basic mistake is the same as Cicilline’s: the unsupported assumption that superficial similarities between two markets imply that regulation that we may want for one must also be desirable for the other. A lot of people want to regulate the platforms that big tech companies have built up. Their reliance on such spurious analogies shows how weak their arguments are on the merits.

Sam Bowman is the director of competition policy at the International Center for Law & Economics, a think tank that promotes the use of law & economics methodologies to inform public policy debates.