The fact that the United States, a country known for its long history of uninterrupted elections under the same fundamental rules, does not elect its president through a popular vote is a constant source of public attention and intense debate. According to a survey conducted by the Pew Research Center in January 2021, 55 percent of Americans favor electing the president based on the popular vote, while only 43 percent support the current system of the Electoral College. Even prior to the controversial 2016 and 2020 elections, headlines such as “How to Get Rid of the Electoral College” have been pervasive. Since the adoption of the Constitution in 1787, there have been more than 700 congressional proposals aimed at reforming or repealing the Electoral College. Typically, criticism of the Electoral College centers on the issues of unequal or unfair representation, as it assigns higher proportional weights to smaller states. Notably, the issue of election fraud rarely features prominently in these discussions.

We argue that the Electoral College creates strong disincentives for national parties to engage in election fraud, which might explain why discussions about the merits of the Electoral College often neglect the issue. Under the Electoral College, presidential candidates compete for votes on a state-by-state basis. The winner of each state obtains all the state’s electoral votes, the number of which is equal to the size of the state’s congressional delegation, which is roughly aligned with the state’s population. This system often results in very close elections in a few states. In the 2020 presidential election, for instance, the incumbent president Donald Trump faced losses of 10,000 votes (0.3 percent) in Arizona, 12,000 votes (0.2 percent) in Georgia, and 20,000 votes (0.6 percent) in Wisconsin. Similarly, in 2016, Hillary Clinton experienced a combined margin of defeat of 76,000 votes across three swing states. A superficial analysis suggests that the relatively small degree of fraud required to overturn the election in either case could present a significant opportunity, if not a temptation, to engage in such activities.

Under the Electoral College, the states where fraudulent votes would be most valuable are also the states where it would be most difficult to obtain them. Consider, hypothetically, what then president Trump would have needed to do to change the outcome of the election he lost in 2020. In the states that he lost closely—Arizona, Georgia, and Wisconsin—his opponents had significant representation at all levels of government, including the state supreme courts, both chambers of the state legislatures, and the states’ congressional delegations. Even if some critical offices in these states were held by Republicans (as one office in Georgia had been in 2020), these Republicans would likely not be hyperpartisan. Thus, organizing fraud sufficient to swing the outcome in these states despite the surveillance of election supervisors, prosecutors, and judges unsupportive of Trump would have been extremely challenging. It would have been easier to obtain a substantial number of fraudulent votes in Republican-dominated states (such as Alabama, Oklahoma, Tennessee, or Texas), where Trump won by significant margins and enjoyed widespread support at all levels of government. However, while it may be easier to steal votes in states dominated by one party, it would be pointless under the Electoral College system. But under the popular vote system, a vote in Tennessee would hold the same value as a vote in Wisconsin, and a party inclined to steal an election would have the opportunity to conduct larger-scale fraud more easily.

We developed a theoretical model to determine the resilience of the Electoral College and popular vote system to fraud. Our model measures the resilience of each electoral system as the difference between the fair vote tallies of the two parties that would be required to deter vote fraud. In other words, how far ahead does one party have to be for the other party to perceive committing fraud as futile? Our model demonstrates that when the difference between the fair vote tallies of the two parties is identical in the Electoral College and popular vote system, the incentive to commit fraud is higher in the popular vote system. This means that the Electoral College can tolerate a smaller difference between the fair vote tallies of each party than the popular vote system and result in the same degree of incentive to deter fraud. Perhaps surprisingly, our model shows that an increase in polarization (within or between states) does not hurt the ability of the Electoral College to deter fraud, but it does increase the incentive to commit fraud in the popular vote system.

One critical assumption of our model is that instances of fraud are checked by local authorities, regulators, and courts under both electoral systems. This is natural given the way the electoral system currently works in the United States and the way it would work in the National Popular Vote Interstate Compact (NPVIC)—the most likely way the popular vote system would be implemented in the United States. The NPVIC is an agreement among some states to award their electoral votes to the winner of the popular vote if the number of electoral votes among participating states becomes large enough to enforce the outcome of an election. Under this system, local officials would retain vote counting and election law enforcement authority even while implementing the national popular vote outcome. The incentive to commit fraud in states where one party enjoys overwhelming support would increase substantially. Thus, the Electoral College better protects against election fraud than the NPVIC system. Our model does not consider the implications of transferring the authority to conduct elections to the federal government because that is not a part of the NPVIC proposal. However, our analysis demonstrates that without a national election authority— that is, with states’ authorities responsible for counting votes—the popular vote system would likely result in a massive increase in vote fraud. Those who advocate for the NPVIC should be transparent about the fact that they may be implicitly advocating for a national election authority, a reform of the US political system far more profound than merely abandoning the Electoral College.

Note
This research brief is based on Georgy Egorov and Konstantin Sonin, “Electoral College and Election Fraud,” National Bureau for Economic Research Working Paper no. 31474, July 2023.