The U.S. Supreme Court issued a divided 5-to-4 decision in January stating that government could not limit political spending by corporations because doing so violated their free speech. The Court’s decision treated corporations as human beings by empowering them with a very broad interpretation of the First Amendment’s free speech principles. Campaign finance experts have said the ruling could fundamentally alter how political campaigns are executed. Opponents of the ruling fear that corporate money will swamp the democratic process, while supporters reported feeling empowered because the new decision represents a paradigm shift and overturns two important precedents that restrained corporate political spending.

Regardless of your take on the ruling, the case of the Citizens United v. Federal Election Commission embodies the main element of a real-world explanatory model developed by Mehmet Ekmekci, an assistant professor of Managerial Economics and Decision Sciences at the Kellogg School of Management. Ekmekci’s model describes how endorsers of political candidates work strategically to coordinate a divided electorate.

Ekmekci describes in his research the unfortunate case of the Condorcet loser, where the candidate voters least prefer wins an election because the majority of the electorate is divided between two other candidates. This concept traces back to an historic essay penned 225 years ago by the Marquis de Condorcet, Marie Jean Antoine Nicolas de Caritat, a French mathematician, early political scientist, and pioneer in applying math systematically to the social sciences. In a perfect system the most preferred candidate, the Condorcet winner, would always win.

But some voting systems are imperfect. Let’s imagine that multiple candidates are running for office and 60 percent of the electorate prefers candidate A or B, and 40 percent prefers candidate C. Imagine, too, that candidates A and B have platforms more similar to each other’s than to candidate C’s. If the 60 percent preferring candidate A or B does not coordinate, then a situation could arise where the least preferred candidate, the Condorcet loser, wins.

Ekmekci modeled how endorsers behave in pre-election activities to organize divided voters and prevent the Condorcet loser from winning. In his model, this endorser has some knowledge about the distribution of the voter preferences that is not publicly known. Perhaps the endorser is a media group that has conducted extensive polling, and from this polling can deduce where voters are split between candidates A and B. Or perhaps the endorser uses proxy data, say, measuring how many people watch specific TV news channels such as Fox News, read specific newspapers such as The New York Times, or are members of special interest groups such as the National Rifle Association or the National Association of Home Builders. Endorsers could be non-governmental organizations, religious alliances, corporations, or labor unions. But for the purpose of the paper, Ekmekci used the example of media and how it influenced his large, simulated electorate of rational voters who carefully deliberate over their voting decisions.

The model explains that when endorsers stand to gain some sort of benefit, they are capable of coordinating the 60 percent of the electorate that is split between candidates A and B. “Endorsement is costly,” Ekmekci says. “But if the candidate who the media endorses wins the election, then we know this endorser will get some private benefit from his actions.”

A media group, for example, would benefit from wider access to government officials or advance news tips. Labor unions or NGOs might gain legislation favorable to their agendas. But the point is that endorsers expect an intangible benefit for endorsing a winning candidate. And so the endorser must choose strategically and get behind the candidate most likely to win. “If the endorser thinks that A is more likely to be supported than B, if the distribution is 35 to 25 percent, then he is going to support the one who has the 35 percent excluding the effect of his endorsement,” Ekmekci says. “And when he makes his endorsement, the voters will coordinate on the candidate he is endorsing.”

The model shows that in all cases, the endorser can effectively coordinate the voters to prevent the Condorcet loser from gaining office, if there is a benefit to the endorser. It would appear then that the endorser is doing a public good by preventing a coordination failure of the voters. But what if the endorser is biased toward one candidate or the other? Ekmekci builds this into the second part of his paper by examining what he calls manipulation.

“If the endorser, in addition to his intangible benefit, is also a little bit biased politically towards A or B, then can he leverage this coordinating power to make the candidate that he is biased towards win?” Ekmekci asks rhetorically. “This is what I call manipulation.”

The model reveals that if there is not a large discrepancy between voter preference for candidate A or B, then the endorser can manage to manipulate voters toward the candidate he favors. He could even successfully help candidate A to win, even if this candidate were slightly less favored than candidate B. “This will happen even if the electorate knows the endorser is biased, and even if the bias is large,” Ekmekci says, “because the voters who support A, but really prefer B to C, are afraid of a potential miscoordination. So even if they prefer B, but they see the endorser support A, they are afraid of splitting the votes and end up having C win the election.”

Ekmekci says his model also explains situations where competition can lead to a coordination failure. “The endorsers may be dividing their support across A and B, in which case they may be dividing the 60 percent of the electorate that needs to be coordinated,” he explains.

Limitations of the model include assuming that voters are rational and make voting decisions after careful deliberations, and that the electorate is large.

“In some countries, the media group is not allowed to express opinions or give this sort of information to the public,” Ekmekci points out. “So this says, if you govern such that you allow this information to be revealed through media, it will do something good in terms of coordinating voters. But on the other hand, you will expose yourself to some manipulation as well.”

Ekmekci considers the most important insight of the model to be that biased endorsers can “fool” the public. “The public may know the endorser is biased, but there is this whole psychology of ‘what if’ he is really telling the truth,” he says. “So there is a really small chance they may end up selecting a wrong candidate. And this fear gives the whole hand to the manipulator, to the endorser.”