---

Summary

---

Introduction

Over 700 proposals have been introduced in Congress in the last 200 years to reform or eliminate the Electoral College system.

The president of the United States is arguably the world’s most powerful political leader, and the incentives created by the electoral process are important. 

It is not surprising, therefore, that the Electoral College system of electing the president has been under constant debate. According to federal historians, over 700 proposals have been introduced in Congress in the last 200 years to reform or eliminate the Electoral College system. A recent Pew Research Center poll found that, “Nearly two-thirds of U.S. adults (65%) say the way the president is elected should be changed so that the winner of the popular vote nationwide wins the presidency.” The drive toward reform is understandable given the perceived impact on the democratic process and economic policy.

Two features are crucial for this discussion. First, the number of electoral votes per state is not proportional to the number of voters. The Constitution allocates electoral votes to each state equal to that state’s representation in Congress, with an additional three for the District of Columbia. Since each state has two senators, small states are over-represented in terms of electoral votes. Second, all states except Maine and Nebraska have adopted a winner-take-all system in which all electoral votes are awarded to the candidate who receives the most popular votes in that state. 

This primer will focus on these two features and will not discuss, for example, the problem of faithless electors. In assessing the theoretical and empirical research on the likely effects of the Electoral College, this primer discusses the following issues:

1. Incentives Created by the Electoral College

The Electoral College creates incentives for presidential candidates to focus their resources, whether campaign time or policy promises, on a few battleground states. Which features of the Electoral College system create these incentives? How has the inequality of candidate attention across states changed over time, and why? How would the inequality change after a reform to a National Popular Vote system?

2. Vulnerability to Fraud and External Interference

How vulnerable is the outcome of the Electoral College system to fraud, external interference, influence operations, and incorrect vote counts relative to a National Popular Vote system?

3. Discrepancies Between the Electoral and Popular Vote

How likely is it that the Electoral College winner does not win the popular vote? Which party is more likely to win the Electoral College without popular support, and how has this changed over time? 

Unequal Attention

Researchers have analyzed the implications of the Electoral College for public policy and argued that it increases the incentives to select policies that benefit specific states and groups rather than the population at large (1).

More recent work integrates theory and empirical data into models that predict which states the candidates should target. This primer will contain numbers computed using this type of model (2).

The Electoral College creates incentives for presidential candidates to prioritize a few battleground states. Three out of four states are regularly ignored by their campaigns. This may lead voters in neglected states to conclude that they are politically irrelevant in the presidential election, thereby suppressing their participation in the political process. The consequences for people’s wellbeing are likely more significant if the Electoral College biases economic policy in favor of swing states. However, this primer will start by investigating election campaigns. The reason is that election campaigns provide a much clearer signal of presidential candidates’ priorities across states, as driven by the Electoral College. In contrast, policies are determined by a multitude of actors and factors, among which presidents and their election concerns may be one. Thus, campaigns let us test the theory.

Figure 1 (p. 9) shows the share of candidate visits across states in the 1976 and 2020 elections. The 1976 election, when President Gerald Ford faced Democratic challenger Jimmy Carter, was the most competitive of the three decades leading up to 2000. In that election, California and New York were the most targeted states, receiving 12% and 10% of candidate visits, respectively. By 2020, however, candidate attention had shifted. California and New York were completely neglected, while Florida and Pennsylvania became the most targeted states. We will return to the implications of this shift below. 

Another noticeable feature in this figure is that candidate visits in 2020 were much more concentrated, leaving many states without any visits. In other words, the distribution of candidate attention across states has become more unequal. To quantify this political inequality, this primer will use the Gini coefficient for the number of campaign visits per voter.

By this measure, the inequality in campaign visits more than doubled between 1976 and 2020 (3). The extreme inequality in campaign visits in 2020 may have been an outlier due to the pandemic, which reduced travel and campaign events. However, the inequality was nearly as high in 2016, when campaign visits were not affected by the pandemic. In sum, the inequality in candidate attention across states has risen and reached an unprecedented and extremely high level.

How do we know that the increase in inequality reflects changing incentives and is not simply an artifact? Perhaps the collection of data on campaign visits was very different in 1976 compared to 2000 and 2020? We can investigate this by using a model of optimal campaign visits. Theory tells us that campaigns should allocate more time to states where it is more likely that (i) their electoral votes are decisive for the Electoral College outcome at the same time as (ii) the state vote is expected to be close. The probability of such an event can be calculated based on a forecasting model. For example, Pennsylvania has the highest such probability in the 2024 election.

It turns out that this model explains the distribution of actual campaign visits well; see Figure 1 (p. 9). The model also predicted visits in 2024. Inequality of these predicted candidate visits in the 2024 election was just as high as in 2020. 

Since this predicted allocation is not affected by the measurement of visits, we know that the increasingly unequal candidate attention in recent years is not an artifact; something structural has changed that has exacerbated the inequality of candidate attention. The model is useful in understanding which factors drive the increased inequality. We can change one factor in the prediction model at the time and observe how inequality is affected.

Part of the political inequality created by the Electoral College is an intentional design feature, particularly the over-representation of small states in electoral votes. However, our measure of campaign visit inequality remains virtually unchanged if each state’s electoral votes were proportional to its number of voters. Thus, a discussion of political inequality created by the Electoral College system can largely disregard the disproportional allocation of electoral votes.

Instead, the largest contributing factor is the first-past-the-post system, where the winner in a state receives all of its electoral votes. Presidential candidates can only hope to move voter support by a few percent. As a result, states with a large lead for one party can safely be ignored by candidates. 

The primary reason political inequality created by the Electoral College has increased since 1976 is that a greater number of states can be safely ignored, due to a combination of growing political polarization across states and improved polling. The growing geographic polarization in the U.S. is well documented (3).

Red states have become redder, while blue states have become bluer. As a result, there were far more swing states with election results close to 50-50 in 1976 than in 2020, and far more states are now either solidly Democratic or solidly Republican. This increased geographic polarization has magnified the incentives produced by the Electoral College to focus attention on a narrow set of states. Improved polling has further exacerbated this effect by making it easier to predict what states can safely be ignored early in the campaign. 

But even if it is true that the Electoral College focuses candidates’ attention on a very narrow set of states, we must ask whether alternatives to the Electoral College might introduce equally problematic effects on presidential campaigning, such as an overemphasis on large media markets. While we cannot directly observe how campaign visits would be distributed under a National Popular Vote system, we can predict this distribution using a theoretical model, just as we did for the Electoral College. 

This prediction shows that there will also be inequality of candidate attention under National Popular Vote. However, the inequality is just much smaller than the under the Electoral College system. This disparity occurs because under National Popular Vote, politicians are chasing swing voters rather than swing states. The shares of swing voters across states, media markets, or demographic groups are much more evenly distributed than the probability of being a decisive swing state. In addition, under the Electoral College system, politicians are also chasing swing voters, although only in decisive swing states. Hence visits under the Electoral College system will be more concentrated on a few states and media markets than visits under National Popular Vote (4).

What are the consequences of the campaigns concentrating on a few states? Voters in neglected states may feel politically irrelevant in the presidential election, leading to decreased interest and participation in the political process. Consistent with this, voter turnout is significantly lower in states that receive few campaign visits. Estimates suggest that the 2012 presidential campaigns increased turnout in highly targeted states by 7 to 8 percentage points (5).

A related concern is that economic policy will disproportionately benefit battleground states. The incentives created by the Electoral College are similar for both campaigning and policy, in that politicians with electoral concerns should grant policy favors to the same states they prioritize during their campaigns. 

However, the strong incentives created by the Electoral College to favor certain states are counterbalanced by institutional features—specifically, the checks and balances inherent in the U.S. political system—that limit the president’s ability to act on these incentives. The geographic distribution of federal funds is mainly decided by the U.S. Congress, one of the most powerful and decentralized national legislatures in the world.

Nevertheless, presidents have opportunities to influence the budget during both the proposal and implementation stages. The president is responsible for preparing and submitting an annual budget proposal to Congress. By appointing like-minded officials to top management positions in departments and agencies responsible for crafting budgetary requests, as well as within the Office of Management and Budget, which coordinates and oversees the process, presidents can empower officials who align with their goals (6).

The fact that the states considered battlegrounds have varied considerably over time has been leveraged by researchers studying the allocation of federal spending. As mentioned, California and New York were battleground states in 1976 (see Figure 1). Today, these states are largely neglected, while campaigns prioritize states like Pennsylvania and Florida. A key question is whether the U.S. federal budget has shifted focus accordingly. The evidence is mixed. Some studies find that more federal funds are allocated to battleground states (7), while others do not support this conclusion (8). One possible explanation is that studies finding an effect focus on more recent time periods, and that spending may have become increasingly politicized.

The evidence is clearer in policy areas where the president has more direct control, such as the allocation of federal disaster relief and trade policy. The Federal Emergency Management Agency (FEMA) was established by an executive order from President Carter in 1979. FEMA is responsible for allocating federal funds to areas adversely affected by natural disasters. The process requires a presidential disaster declaration, and the authority to make such declarations was significantly expanded after the passage of the 1988 Stafford Act. 

FEMA disaster declarations are well-suited for investigating presidential influence for several reasons. First, the incentives to consider the political effects of disaster relief decisions are strong due to the media attention they receive, which can result in either credit or blame. For example, support for President George W. Bush plummeted after the widely criticized federal response to Hurricane Katrina, the catastrophic storm that devastated Gulf Coast states in late August 2005. Second, the need for relief is driven by external factors, such as extreme weather events, which can be reasonably well measured. By comparing similar events, such as those with comparable storm damage and objective economic need, one can detect whether political factors influence the allocation of federal disaster aid.

Empirical studies of thousands of presidential disaster declarations find that battleground states are systematically favored, and this pattern emerges after 1988, when presidential authority to declare disasters was expanded (9).

U.S. presidents also wield substantial power over trade policy. For example, a long-standing claim is that presidents, from Reagan to Trump, have backtracked from free-trade policies to protect the steel industry in key swing states. However, it is hard to judge from just this one industry whether trade policy is systematically influenced by electoral concerns. There may be other reasons to protect the steel industry, and industries in non-swing states may also have been protected. A group of researchers addressed this by studying all import tariffs in 2018 and found that tariffs systematically favored sectors concentrated in politically competitive areas (10). Other researchers have studied all WTO trade disputes initiated by the United States from 1995 to 2014. They found that U.S. presidents were more likely to initiate trade disputes involving key industries in swing states, particularly when facing re-election (11).

To sum up, the winner-take-all feature of the Electoral College focuses candidate attention on a small set of states. This reduces voter turnout in neglected states and distorts government policy. The incentives for unequal treatment would be dramatically reduced under a National Popular Vote system.

---

It is desirable for an electoral system to be robust against fraud, external interference, influence campaigns, and even simple counting errors. 

The fact that presidential election outcomes are decided in only a few states makes foreign interference easier, as the interference only needs to target a handful of states to inflict a high electoral cost on a U.S. president. For example, in response to tariffs introduced by the Trump administration, the EU imposed retaliatory tariffs on orange juice, impacting Florida, while China targeted soybeans with large production in battleground states such as Iowa, Ohio, and Minnesota.

The Electoral College also increases the risk of election interference for a related reason: it more often results in razor-thin margins of victory. In 2020, the incumbent president lost by seven million votes nationwide. Instead of conceding defeat, he contested the outcome, alleged fraud, and pressured state and local officials to alter vote counts. The president appealed to Georgia Secretary of State Brad Raffensperger on January 2, 2021, stating, “I just want to find 11,780 votes” (12). Needless to say, finding seven million votes would have been significantly more difficult. Razor-thin margins are problematic for several reasons; for example, they increase vulnerability to fraud and illegitimate influence. In 2000, the vote margin in Florida was so slim that almost any form of influence, from biased news media to foreign interference, or even just a confusing ballot design (13), could have affected the outcome.

But how likely are close elections under the Electoral College, and how likely are they under alternative methods of electing the president? Historically, the 2000 election was decided by just 537 votes in Florida. Similarly, the 1876, 1884, and 1916 elections were decided by just 889, 1,149, and 3,773 votes, respectively. However, these are extremely rare outcomes, and historical frequencies are not very informative of event probabilities. 

Instead, researchers have simulated likely election outcomes using election forecasting models. Estimates suggest that it is about 40 times more likely that an election is decided by a margin of a thousand votes or fewer under the Electoral College than under the National Popular Vote system (14). The main reason is simply that the population of the entire U.S. is much larger than that of the average state being decisive in the Electoral College. As a result, the National Popular Vote system is more robust against election interference than the Electoral College.

However, there is a countervailing factor that makes the Electoral College more resilient to fraud by state-level officials. Under the Electoral College, the swing states where fraudulent votes are most valuable are also the states they are most difficult to obtain. In swing states, both parties typically have significant representation across all branches of government, including the state Supreme Court, both chambers of the state legislature, and the state’s Congressional delegation. Therefore, organizing fraud significant enough to alter the outcome in these states, while expecting election supervisors, prosecutors, and judges to turn a blind eye, would be extremely difficult (15).

Estimates suggest that it is about 40 times more likely that an election is decided by a margin of a thousand votes or fewer under the Electoral College than under the National Popular Vote system.

Counting errors differ from fraud in that they are not targeted but are more likely random. If counting errors occur with some probability, it is equally likely that a given fraction of votes will be miscounted in the national election and in decisive states in the Electoral College, such as Florida. It is, however, less likely that miscounted votes will decide the election outcome under a National Popular Vote system. The reason is due to the law of large numbers. If counting errors are random (and not intentional election interference), then the risk that they favor any one party with more than, say, 0.1 percent of the vote share is smaller the larger the electorate.

A final vulnerability of the Electoral College system is that of faithless electors. One of the original justifications for the electoral college was to have a group of highly qualified electors who would serve as a check against the impulses of the less qualified masses. This justification is now abandoned and most states have laws requiring that electors faithfully cast votes in line with the will of the state. Still, electors occasionally do not cast their votes for president as prescribed by the state. However, faithless electors typically act alone and have never changed the outcome of an election. 

---

Another commonly criticized feature of the Electoral College is that it occasionally produces winners without the support of a popular majority. 

Some may not see this as a problem. After all, these electoral rules have been in use for a long time, and the U.S. Constitution deliberately avoided awarding the presidency based on the popular vote. However, some believe that this reduces the fairness of elections and consequently the legitimacy of the president-elect. Research can contribute by providing facts, for example, by calculating the probability of this mismatch and whether it systematically favors either party.

It is important to note that a mismatch does not necessarily mean that the Electoral College and National Popular Vote systems would produce different winners. Since the Electoral College influences platforms, campaign strategy, turnout, and other factors, we cannot determine this simply by observing the popular vote under the current system. For example, voter turnout is lower in non-battleground states under the Electoral College. A shift to a National Popular Vote system would likely increase voter turnout in these states while potentially decreasing it in battleground states. Thus, the popular vote would differ from what it is under the current system.

Historically, a mismatch between the Electoral College winner and the popular vote winner has occurred in 5 out of 59 presidential elections, or approximately 8 percent of elections. However, we may be more interested in the likelihood of such an event occurring in recent years or a future election, rather than over the entirety of U.S. history. To compute the likelihood of such events, researchers have simulated possible election outcomes based on the current distribution of party support across states. By taking the predicted outcomes in all 50 states and Washington, D.C., and drawing possible outcomes from the historical error distribution, we can generate a range of possible election outcomes. From these simulated outcomes, we can compute the likelihood of a mismatch.

Using this method, the estimated probability of a mismatch in the 2024 presidential election was just below 20 percent. There was an 18 percent chance that Trump would win the Electoral College without a popular majority, and a 1 percent chance that this would happen to Harris.

To have a 50 percent chance of winning, Harris needed around 51.5 percent of the national two-party popular vote. This year, the Electoral College produced a bias favoring the Republican candidate; however, this is a recent phenomenon. Before 2016, there was no systematic bias favoring either party (16). From 1948 to 2012, the average chance that either Republican or Democratic candidates would win the Electoral College without a majority of the popular vote was the same, at three percent each.

The primary reason for the discrepancy between the electoral vote and the National Popular Vote is the winner-take-all system at the state level. As a result, a candidate can win some states by very narrow margins, lose others by large margins, and hence win the electoral vote while losing the popular vote. For example, in 2016, Hillary Clinton won New Hampshire by a margin of just 0.4 percent of the popular vote. Although 346,000 people voted for Donald Trump, all of New Hampshire’s electoral votes went to Clinton. The 346,000 Republican votes in New Hampshire were lost for Trump and became completely inconsequential. Conversely, Trump won Michigan by 0.2 percent of the popular vote. In that state, 2.3 million Democratic votes were lost for Clinton. Across the nation, Trump won several large states by small margins (Michigan, Pennsylvania, Wisconsin, and Florida), causing many of Clinton’s popular votes to be lost. In contrast, the large states won by Clinton were secured by large margins (California, New York, Illinois). As a result, roughly ten million more of Clinton’s popular votes were wasted compared to Trump’s. 

The other salient feature of the Electoral College system, the malapportionment of electoral votes across states, plays a relatively minor role. If each state’s number of electoral votes were proportional to its voter population, then the probability of a mismatch between 1948 and 2020 would remain virtually unchanged. For the 2024 election, the probability of this event was 18 percent, with a 14 percent chance of Trump winning without a popular majority and a 4 percent chance of this happening to Harris. If one is concerned about the mismatch between the electoral and popular vote outcomes, then the winner-takes-all feature at the state level is the primary issue.

If each state’s number of electoral votes were proportional to its voter population, then the probability of a mismatch between 1948 and 2020 would remain virtually unchanged.

Another related concern is the possibility of an Electoral College tie, where each candidate receives 269 electoral votes. In such a scenario, the Twelfth Amendment dictates that the House of Representatives selects the president, while the Senate chooses the vice president. Again, some may see no issue with this, while others may strongly disagree.

A tie can occur in several ways. In 2020, the most likely scenario involved a small shift of votes from Biden to Trump in Arizona, Georgia, and Wisconsin, which would have resulted in an Electoral College tie. In this scenario, the House, with a Republican majority, would have most likely elected Donald Trump. Simulations show that a tie was not entirely unlikely in 11 other combinations of state-level outcomes in 2020. For example, shifting outcomes in Arizona, Georgia, and Minnesota, or in the four states of Pennsylvania, North Carolina, Michigan, and Georgia, could have resulted in a tie. However, the estimated probability of a tied election in 2020 from any of these scenarios was only 0.2 percent. The probability was higher in the 2024 election, at 0.6 percent. While this is a highly unlikely event, it could eventually occur. In fact, an Electoral College tie is about as unlikely as an election as close as Florida’s in 2000 deciding the outcome. 

---

Conclusion

The Electoral College directs candidates’ efforts and attention to a small set of states, leaving others neglected. 

This attention inequality is substantial and has increased dramatically over time due to increased political polarization and improved polling. These aspects of the presidential system have reduced political participation in neglected states and distorted policies over which the president has more control.

The fact that the election is decided in a few states (often by relatively few votes) makes the system vulnerable to fraud and external interference. 

In the current election, the Republican candidate is more likely to win the Electoral College than a national popular majority. However, before 2016, there was no systematic advantage favoring either party.

The problematic effects of the Electoral College primarily stem from the winner-take-all elections at the state level. This feature is not mandated by the U.S. Constitution. Instead, the Constitution grants states exclusive control over how they award their electoral votes, thus empowering them to reform the system. 

One popular reform alternative is the system currently used by Maine and Nebraska to award electoral votes. In this system, one electoral vote is awarded to the candidate who carries each congressional district within a state, while the two additional electoral votes go to the candidate who wins a plurality of the statewide vote. However, in light of the results presented above, this reform proposal presents its own issues. Winner-take-all elections at the district level would shift candidate attention to battleground districts rather than states, leaving districts that are safely won by either party neglected. Since partisan polarization within states (for example, between rural and urban districts) is at least as strong as it is between states, the problems associated with winner-take-all elections are likely to persist. Additionally, new problems could arise related to partisan gerrymandering of district boundaries. 

A system that more effectively addresses this problem is the National Popular Vote. This system can be implemented by the states in several different ways. One approach is to allocate electoral votes in proportion to each candidate’s share of the vote, allowing fractions of electoral votes to be awarded, rather than only whole numbers. This system would eliminate the issues associated with winner-take-all elections. However, a reform would be implemented only if all states agree to commit to it. Such a unanimous decision seems unlikely. 

A more easily implemented procedure is the ‘National Popular Vote Law,’ which specifies that each state awards all its electoral votes to the winner of the National Popular Vote (17). This reform does not require unanimous agreement. Instead, the National Popular Vote Law will take effect once it is enacted by states representing a majority of electoral votes (270 out of 538). The presidential candidate receiving the most popular votes across all 50 states and Washington, D.C., will then receive all the electoral votes from the enacting states. In other words, the candidate who receives the most popular votes nationwide would be guaranteed enough electoral votes to secure victory. Currently, 17 states and the District of Columbia have enacted the National Popular Vote Law. These jurisdictions control 209 of the 270 electoral votes needed to activate the law. 

However, receiving the additional electoral votes may prove challenging. While transitioning to a National Popular Vote might be beneficial on average, it would also create winners and losers. In the 2024 election, a reform would have disadvantaged the Republican candidate and the current battleground states. Consequently, the states that have enacted this law are predominantly Democratic, with only a few Democratic-leaning battleground states being exceptions. Nevertheless, which party benefits from the system and what states are considered battlegrounds tend to change over time. To gain broader support, a reform could be designed to take effect far enough into the future that the winners and losers of the reform would be uncertain. Reflecting on past trends, this suggests a timeline of at least two electoral cycles, as there was no systematic bias favoring either party prior to 2016.

---

References

---