Facebook admitted that only a binding regulation on political ads could prevent private corporations from influencing the outcome of US presidential elections. Without such regulation, digital platforms can favor a candidate by altering (or maintaining) their policies on digital advertising. Trump’s campaign was much more effective than Clinton’s in using micro-targeting to shape voters’ preferences in 2016, a new study shows. Facebook decided to confirm the same policies for the 2020 election.  

In a surprising moment of honesty, Facebook admitted that only binding regulation on political ads on digital platforms can prevent private corporations from actually choosing the next president of the United States. Since there is no such regulation, Facebook founder Mark Zuckerberg will still have a say in the 2020 election’s outcome (and we know some of his preferences from leaked audio: He doesn’t like Elizabeth Warren and her plan to break up Big Tech). “We don’t think decisions about political ads should be made by private companies, which is why we are arguing for regulation that would apply across the industry,” Facebook stated on January 9 in its long-awaited communication regarding a policy change around political ads before the 2020 US presidential elections.

The policy change is, actually, no change at all. This is not a trivial choice. Different policies regarding political ads on social media can favor or penalize different candidates, but only Facebook knows which changes will help or harm which candidate because the company does not disclose that information nor the data needed by researchers to run independent analyses. In other words, without a regulation that levels the playing field, Facebook could decide to adapt its political advertising policy to lead voters to choose not the best president for the American people, but the best president for Facebook. Donald Trump’s campaign proved able to use the current rules to influence voters’ behavior in 2016 through the controversial micro-targeting advertising strategy.

Micro-targeting is not fake news or disinformation. It is simply a new tool to win the usual political competition: to persuade swing voters to choose one candidate over another, to mobilize supporters and increase their turnout, or to prevent the other candidates’ potential voters from showing up at the polls. A definition of micro-targeting is the following: “A marketing strategy that uses people’s data—about what they like, who they’re connected to, what their demographics are, what they’ve purchased, and more—to segment them into small groups for content targeting.” Micro-targeted ads can be a very powerful tool not only because they are so tailor-made that they can reach the most precious voters’ attention, but also because a convinced swing voter can share each ad (or its narrative about the candidates) with his friends on Facebook, who are very likely swing voters themselves.

The New York Times has recently leaked a memo that a prominent Facebook executive, Andre Bosworth, posted to a Facebook employee-only page on December 30. According to Bosworth, the micro-targeting regulation is as powerful as the invisibility ring in J.R.R. Tolkien’s books. As all Lord of the Rings junkies remember, the ring was designed “to find them” and to “rule them all” (and also “to bring them all and in the darkness bind them”).

According to Bosworth, who in 2016 used to run the Facebook ads organization, “Facebook was responsible [for getting] Trump elected” not because of Russian interference or Cambridge Analytica’s abusive operations, but “because he ran the single best digital ad campaign I’ve ever seen from any advertiser. Period.”

Brad Parscale, a digital consultant who is currently running Trump’s 2020 campaign, used Facebook’s potential better than his Democratic competitors did, Bosworth says: “The use of custom audiences, video, e-commerce, and fresh creative [methods] remains the high watermark of digital ad campaigns in my opinion.”

There is a semantic dispute regarding whether or not this strategy should be considered micro-targeting: Both Parscale and Bosworth deny that what the Trump campaign did was “micro-targeting” simply because that term seems to imply some kind of information manipulation to tell different and sometimes misleading stories to different audiences. However, Gary Goby, digital director for Trump’s 2020 campaign and for digital advertising in 2016, proudly states on Twitter that “Facebook saw the very clear difference in sophistication between our paid effort & Hillary’s. There are many metrics which make this obvious, but none better than the number of ad variants Trump ran vs Hillary. We (Trump) ran 5.9M unique ad variants, while Hillary only ran 66K.”

The only reason to produce so many different advertisements is to send slightly different messages to different groups of voters through micro-targeting. The effects of such an aggressive digital strategy proved relevant, according to a group of scholars that found a clever way to overcome Facebook’s reluctance to share data on political advertising.

At the 2020 annual meeting of the American Economic Association in San Diego, Federica Liberini (University of Bath), Michela Redoano (University of Warwick), Antonio Russo (Loughborough University), Ángel Cuevas Rumin (UC3M), and Ruben Cuevas Rumin (UC3M) presented their paper Politics in the Facebook Era: Evidence from the 2016 US Presidential Elections. Since they could not get access to data on the actual buying of political ads on Facebook, they measured the intensity of micro-targeting in 2016 by looking at the Cost per Mille impressions (CPM) and at Cost per Click (CPC). In the Facebook advertisement auction system, CPM and CPC are the prices to reach specific audiences, segmented by gender, age, race, education level, and location. A surge in certain prices during an election campaign, the authors argue, reveals a surge in demand for reaching specific targets. The reason for the increase in demand is that those groups of voters become particularly relevant for one or both parties.

To measure the exposure of voters to social media campaigns, the authors use self-reported media and Facebook usage for respondents to the 2016 American National Election Survey. The paper then estimates the effect of exposure to Facebook campaigns on changes in voter behavior with respect to revealed intentions, exploiting the randomly-assigned interview date that determines different levels of exposure to social media in different moments for people with similar characteristics.

The study’s results are unambiguous: In 2016, political micro-targeting had significant effects when based on geographical location, ideology, and gender or race, but these effects were not the same for every kind of voter. Micro-targeting in 2016 increased turnout and support for Trump among moderate voters targeted with intense campaigning, while Democratic voters exposed to intense political ads were less likely to vote for Hillary Clinton.  Moreover, social media campaigns made it less likely for individuals to change their voting intentions, increasing ideological polarization.

An increase of 10 percent in the relative CPM between the interview and election week reduces the likelihood of changing one’s vote, compared to stated intentions, by 3.9 percent. The effect is almost double that amount for men and for conservative voters. As Trump’s digital strategist Brad Parscale once tweeted, Donald Trump “was a perfect candidate for Facebook.” Or, at least, Facebook proved to be the perfect tool for the then-candidate Donald Trump in 2016.

We do not know whether Democratic Party candidates have closed the skill gap with Trump’s digital team or if the 2020 presidential campaign in the digital realm will be more balanced than in 2016but Facebook knows. Zuckerberg and his team have full visibility of what is happening on their social network, while voters, journalists, and scholars can simply navigate the Facebook Ad Library, which provides no clue about micro-targeting strategies and effects.

Twitter CEO Jack Dorsey banned all political ads on his social network, stating, “We believe political message reach should be earned, not bought.” What he did not mention is that political ads are also less lucrative for Twitter than they are for Facebook. On the other hand, the Washington Post, owned by Amazon CEO Jeff Bezos, praised Facebook’s decision to permit micro-targeting during the 2020 campaign. Henry Olsen, a Washington Post columnist, presents a very original argument to defend Facebook’s discretionary political power: “Minorities of all types recognize that their interests and beliefs are often underserved and ignored by a majority that does not share them. Micro-targeting makes it cost-effective for political actors to address those concerns.” As you can expect from a newspaper owned by a Big Tech mogul, the Washington Post never advocates for more regulation of digital business.

What is most surprising is that a call for regulating micro-targeting on Facebook now comes from those who have benefited the most from current policies: Trump’s digital strategists. Gary Coby tweeted on January 4: “Why would Facebook insist on hiding historic ad spend[ing] and targeting data by the news media? Very simple, share all data, be transparent with us. If the media is targeting swing voters in swing states, it should not be hidden.”

Whether you think that Facebook’s rules are unfair because they favor Trump’s campaign or you believe in a digital conspiracy by liberal media to prevent the president’s reelection, transparency regarding the data and an enforceable regulation could fix the problem and increase the public’s trust in the democratic process.

The ProMarket blog is dedicated to discussing how competition tends to be subverted by special interests. The posts represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty. For more information, please visit ProMarket Blog Policy.