Why a “Glass-Steagall for the Internet” Makes No Sense

Why a “Glass-Steagall for the Internet” Makes No Sense

On Thursday, the House Judiciary Committee will release its long-awaited report on Big Tech and antitrust. While the full extent of the report’s recommendations remain to be seen, Chairman David Cicilline has been dropping not-so-subtle hints that one of the proposed changes will be implementing a “Glass-Steagall for the internet.”

In a recent interview with the Brookings Institution, Cicilline endorsed the idea of having “a Glass-Steagall for the internet, that is not allowing someone to be both a seller of goods and services, and a person who controls the marketplace, that you could be one or the other, you can't set all the rules, control the marketplace and also sell on it in the way that Amazon does, for example.”

And in an interview with the Financial Times last year, he described how the rule would apply to social media companies (emphasis added):

Mr. Cicilline added that more stringent regulation might be needed, including a Glass-Steagall-style separation. This could work, he said, by forcing social media companies to run their platforms entirely separately from the parts of their business that sell customer data.
The Glass-Steagall rules, enacted in 1933, were repealed in 1999, a move that critics said helped lead to banks making riskier investments, thereby triggering the financial crash.
“What we did in the financial services sector, that seemed to work pretty well for a long time,” said Mr Cicilline. “We then repealed it and had a big problem.”

As I’ll show in more detail below, Cicilline managed to bungle a number of different concepts in these two interviews:

  1. Social media companies do not sell personal data.
  2. Private label goods are not unique to tech platforms.
  3. Repealing Glass-Steagall did not cause the 2008 financial crisis.
  4. Analogizing Glass-Steagall to the tech industry makes no sense.

Social media companies do not sell personal data

First, social media companies don’t “sell consumer data”. Rather, the platforms allow advertisers to target users with advertisements. Personal data is neither shared with or sold to advertisers during any stage of the transaction.

Importantly, these targeted ads are why the price users pay to access the platforms is zero. Take away the ads and the whole business model falls apart. Separating the functions of “selling ads” and “running a platform” would leave the former business with no eyeballs and the latter business with no revenue. Is Cicilline advocating for social media companies to start charging users monthly fees instead? If so, that would be a direct and tangible harm to consumer welfare, which, ironically, is precisely what antitrust law is designed to prevent.

Private label goods are not unique to tech platforms

Private label goods, or store brands, are neither unique to tech nor bad for consumers. Retailers have been selling private label goods for more than a century. As the chart below from Benedict Evans shows, selling store brands is practically de rigueur for a company competing in this industry.

Banning this widespread practice would likely harm consumers, as private label goods are generally similar in quality to but priced lower than brand name goods. And for those who think having “more data” is a reason to treat e-commerce retailers differently than brick and mortar retailers, it is important to note that physical retailers have enormous amounts of store scanner data and invest in information technology at levels near the tech companies (Walmart spends more on IT than Microsoft or Facebook).

Glass-Steagall was designed to limit the moral hazard caused by deposit insurance

Cicilline’s analogy to the financial services sector makes no sense for tech. To explain why, we’ll need to review the origin of Glass-Steagall in 1933 and why it was repealed in 1999.

At a high level, all financial crises are bank runs. A series of bank panics in the fall of 1930 precipitated the start of the Great Depression. The financial crisis of 2008 was a run on the shadow banking system (“the collection of non-bank financial intermediaries that provide services similar to traditional commercial banks but outside normal banking regulations”). Examples of shadow banks include money market funds, markets for repurchase agreements, investment banks, and mortgage companies.

The business of banking is maturity transformation. Banks transform deposits (liabilities with short-term maturities) into home mortgages and commercial loans (assets with long-term maturities). If all the depositors ask for their money back at the same time, the bank won’t be able to pay them all out because it holds in reserve an amount equal to only a fraction of its deposit liabilities. In It’s a Wonderful Life, when the owner of the local savings and loan association, George Bailey, is faced with a run on his bank, he tries to explain to one of his customers why he can’t pay out all the deposits on demand:

You’re thinking of this place all wrong. As if I had the money back in a safe. The money’s not here. Your money’s in Joe’s house right next to yours. And in the Kennedy house, and Mrs. Macklin’s house and a hundred others. Why, you’re lending them the money to build, and then, they’re going to pay it back to you as best they can. Now what are you going to do? Foreclose on them?

After the banking panics of 1930-1931, Congress passed the Banking Act of 1933, which included the Glass-Steagall provisions and set up the Federal Deposit Insurance Corporation (FDIC). The FDIC insures bank deposits up to a capped amount (currently $250,000 per depositor), which gives depositors peace of mind during a crisis that even if their bank goes out of business, their deposits will be safe. However, the downside of this government guarantee is the risk of moral hazard. Banks might take risks they otherwise wouldn’t if they know the government will bail them out in a crisis.

That’s where Glass-Steagall comes in. Glass-Steagall separated commercial banking and investment banking. By implementing this line of business restriction, Congress was trying to insulate regular savings and loan activity from risky speculation in securities markets. While this made sense in theory as a way of limiting moral hazard, it was less effective in practice.

The financial sector is an interdependent web of assets and liabilities. Counterparty risk is “the probability that the other party in an investment, credit, or trading transaction may not fulfill its part of the deal and may default on the contractual obligations.” The linkages between financial firms are so tight that if the government  provides a backstop for one set of “safe” entities, that guarantee will implicitly extend to risky entities that do business with them. And by limiting the types of business that commercial and investment banks could engage in, that meant they would be less diversified in the event of a crisis in one segment of the market. Recognizing this problem, Congress (and the courts) slowly eroded Glass-Steagall until finally repealing it entirely in 1999.

Repealing Glass-Steagall did not cause the 2008 financial crisis

Cicilline is perpetuating the myth that the repeal of the Glass-Steagall Act in 1999 was the spark that lit a long fuse leading to financial blowup in 2008. If only we had stopped greedy investment bankers from gambling with commercial bank deposits, the narrative goes, we could have avoided the whole mess altogether. But that gets the causes of the crisis wrong. As noted earlier, the 2008 meltdown was a run on the shadow banking system, which by then had become larger than the regular banking system. As Treasury Secretary Tim Geithner said at a public event in 2012, the crisis had “nothing to do with Glass-Steagall.”

In fact, the repeal of Glass-Steagall actually made resolving the crisis easier, enabling the acquisition of Bear Stearns (an investment bank) by JP Morgan (a commercial bank) and of Merrill Lynch (an investment bank) by Bank of America (a commercial bank). Recent research on the Great Depression-era banking crises also bears out the importance of banking diversification:

  • A paper in the American Economic Review by Kroszner and Rajan found that securities issued by unified banks were of higher quality than those issued by investment banks.
  • A paper in the Journal of Economic History by Carlos Ramirez found that the separation of commercial and investment banking “increased the cost for corporations of raising external funds for investment spending.”
  • A paper in Explorations in Economic History by Eugene White found that “one of the most convincing pieces of evidence that the union of commercial and investment banking posed no threat to parent banks is the significantly higher survival rate of banks with securities operations during the massive bank failures of 1930-1933.”

Megan McArdle summed it up best in a piece for The Atlantic in 2008:

Most importantly, commercial banks are not the main problems. If Glass-Steagall's repeal had meaningfully contributed to this crisis, we should see the failures concentrated among megabanks where speculation put deposits at risk. Instead we see the exact opposite: the failures are among either commercial banks with no significant investment arm (Washington Mutual, Countrywide), or standalone investment banks. It is the diversified financial institutions that are riding to the rescue.

Analogizing Glass-Steagall to the tech industry makes no sense

That background history is necessary to understand just how inapt Cicilline’s “Glass-Steagall for the internet” idea truly is. The phrase Glass-Steagall has become a kind of shibboleth in the public consciousness to connote the dangers of excessive government deregulation. By invoking it, Cicilline is trying to hitch his anti-tech crusade to the still simmering (and justified) resentment around the bank bailouts and the slow recovery from the Great Recession.

But trying to analogize Glass-Steagall to the tech industry is flabbergasting. In tech, there is no equivalent of a bank run or maturity transformation or deposit insurance or fractional reserve banking or minimum capital requirements. If anything, contemporary large tech companies are inverse banks. Big Tech borrows long-term and lends short-term. They issue long-term bonds and, infamously, hold large cash reserves (and other short-term securities). To take one example, in August, Apple was able to borrow billions of dollars at a 2.6% interest rate with a 40 year term. At the time of the issuance, Apple had $194 billion in cash and marketable securities. In the next financial crisis, the last companies to need a bailout will be the tech companies.

Tech companies don’t have explicit government guarantees (FDIC) or implicit government guarantees (Fannie Mae and Freddie Mac). Tech companies don’t create money (banks create the majority of the money in our economy in the form of bank deposits). And while tech companies create a lot of value for consumers, they’re not systemically important in the same way financial institutions are. If one Big Tech company fails, it will not bring the economy crashing down with it. For example, if Facebook were to fail, Google might actually benefit by scooping up more business in the digital ad market.

The same cannot be said of the largest banks. Due to leverage and counterparty risk, the failure of one financial institution can lead to a cascade of bankruptcies in other financial firms — and ultimately throw the entire economy into a recession. It is obvious why such a foundational sector of the economy needs to be highly regulated.


There are real problems in tech related to privacy, misinformation, radicalization, counterfeit goods, child sexual abuse material, and foreign interference in our elections. Each of those concerns deserves a targeted solution, not handwaving about a nearly century-old banking regulation. Banning platforms from offering low-cost goods and services will hurt rather than help consumers.

Perhaps Cicilline can look to the regulation of the financial industry for inspiration. After the 2008 crisis, the U.S. government did not bring back Glass-Steagall for banking. Instead, regulators increased the minimum capital requirements at financial institutions, which limits excess leverage and provides a buffer that can absorb losses during a downturn. This regulation is consistent with universal banking: allow financial firms to offer whatever services they want, but do not provide any exemptions from minimum capital ratios. In other words, bring the shadow banking system in from the dark. That’s what a targeted solution to a real market failure looks like.

We should be working toward similar solutions for the problems in tech, instead of wasting our time with buzzwords. Because whenever I hear someone bring up “Glass-Steagall” in the context of a tech policy conversation, I think of what Inigo Montoya said to Vizzini in The Princess Bride: “You keep using that word. I do not think it means what you think it means.”