Using Market Design to Reform the US Electoral College

1 Introduction

On November 6, 2024, Donald Trump won the Presidency of the United States with 312 Electoral votes. A potential crisis in voter confidence was averted because Mr. Trump also won the popular vote by just over 2.2 million votes. This “dual” win has temporarily assuaged a growing distrust of the democratic process in America. Still, his popular vote win was a narrow 49.8 percent to 48.3 percent. While some view Mr. Trump winning the Electoral College and popular vote as a mandate, the narrow margin in the popular vote shows just how close the United States was to a result that could have escalated the crisis in democratic confidence.

For the first two centuries of the United States, only three instances occurred where the popular vote winner did not win the presidency – John Quincy Adams in 1824, Rutherford B. Hayes in 1876, and Grover Cleveland in 1888. However, this has happened twice in the last twenty-four years – George W. Bush in 2000 and Trump in 2016. Whatever your political views, it is undeniable that this phenomenon undermines the winner’s legitimacy. Our position is that had this happened again in 2024, the current crisis would have escalated. Such a result is possible in the future and creates ill will towards the US democratic process.

To avoid such a crisis in democracy, we propose that now is the time for election reform. We propose using market design to look beyond the Electoral College and popular vote as the only two alternatives. The status quo is one where successful political candidates optimize campaign strategies for the voting rules. This obvious fact is why candidates are ruthlessly focused on the battleground states. Their goal is to tip enough voters in their favor to garner the state’s precious Electoral College votes. Two strategies are pursued: (1) attract undecided voters and (2) encourage more supporters to vote or non-supporters not to vote. The 2024 presidential election was clearly defined by Mr. Trump succeeding in this capacity by winning all seven key battleground states, although by slim margins.

Realistically, most other states don’t matter. Harris won California, and Trump won Texas, as anticipated. It is inconceivable that any one person’s vote in these states impacted the overall outcome. Campaigning in these states is primarily about winning the majority in Congress for the candidate’s party. Gerrymandering makes this difficult but not impossible in the House. Ask the Democrats residing in Austin, Texas. The Austin districts are drawn to bring in enough rural Republicans that the Republicans win. Knowing your vote does not matter frustrates voters.

These gaming behaviors also distort who we are apt to see at the top of the ticket. Chess grandmasters are good at chess. Likewise, top presidential candidates are good at gaming the Electoral College. The key attribute is succeeding is improving the margin in the marginal states. This is an implication of the Electoral College. Unfortunately, there is no reason to suspect that candidates who are superb at gaming the Electoral College will necessarily be good at serving the public interest. A better voting rule may better align incentives to avoid the possibility of electing a candidate who is exceptionally good at Electoral College gaming yet has no interest in serving the public at large.

Improving presidential voting rules is challenging because it would require passing a Constitutional Amendment, which requires a two-thirds supermajority vote in the House and Senate and subsequent ratification by three-quarters of the states. Such consensus is made all the more difficult because whichever party is in power wants to stay there. In recent presidential elections, the status quo has favored the Republican party, so Republican representatives may hesitate to alter a system that has worked for them. Still, it is worth exploring the issue.

Market design provides a pragmatic and powerful lens through which to study election reform. The market design has brought successful innovations in many critical markets, such as electricity and communications, two crucial commodities, but also to matching markets that do not use price as the key instrument to manage scarcity, such as matching doctors to residencies, students to public schools, and organ donors to recipients. Like all markets, these involve allocating scarce resources. The market rules define an algorithm for eliciting preferences from market participants and then mapping these preferences into an outcome that best satisfies the market’s goals.

The benefit of a market-designed system is that it is flexible and allows for agile, incremental change as needed, thus ensuring its durability. Specific details are crucial, and deviations from those details can be catastrophic. Unfortunately, system implementation is sometimes plagued by compromises needed to adopt the system, and these compromises can lead to catastrophic results.

The Constitution of the United States provides an excellent example of a good idea gone awry. The ideals of the Constitution, such as protection from authoritarians and freedom of the people, are unquestionable. However, for ratification, substantial deviations from those ideals were embedded in the country’s founding document, including several that affected elections. Amongst the defects in the Constitution, it let individual states decide who was allowed to vote in Federal elections, most of which did not extend suffrage to women and Black people. This is despite the Constitution declaring the rights of “people” and “persons.” Compromise led to the most egregious of the defections from the ideal when the “Three-Fifths Compromise” emerged from the 1787 Constitutional Convention. Not only could Black people not vote, but they were also not even counted as full human beings.

Although the Constitution does provide a pathway for change, in protecting against the tyranny of the majority, it installed a rigidity in its amendment process that nearly led to the destruction of the country in the Civil War. The amendment process ultimately gave Black men the right to vote in 1865, but women did not receive that right until 1920. A more agilely designed system may have averted the split of the Confederacy and spared more than half a million American lives. Although the market design principles were not developed then, they are now. We argue that it is time to apply them in politics. Election reform is an immediate need.

Still, a change to the Constitution would be difficult. From this perspective, we stress that the market design process could also be applied to local and state elections, producing better political leaders at those levels who then “fill the pipeline” and improve the pool of future presidential candidates. Likewise, the market design process can be applied to presidential primaries, where the parties can reform their primaries so that the winner can perform better in the general election.

We can think of voting rules as a means to elicit voters’ preferences and translate them into a winner that best satisfies the electoral goals. The approach combines insights from economics, computer science, and political science. The innovations we describe here are not new. Most have been implemented with success in various settings. We can learn from theory and practice (Foley and Maskin 2022; Foley 2023; Boudreau, Mathews, and Schwartz 2024). We show how the market design approach can be applied to reform presidential voting rules.

2 Historical Background

In the 60 US presidential elections, only five resulted in the winner differing from the popular vote.

1. 1824: John Quincy Adams was elected president by the House of Representatives despite Andrew Jackson receiving more popular and electoral votes.

2. 1876: Rutherford B. Hayes won the Electoral College by one vote, despite Samuel J. Tilden receiving a majority of the popular vote.

3. 1888: Benjamin Harrison won the Electoral College by 65 votes, despite Grover Cleveland receiving more popular votes.

4. 2000: George W. Bush won the Electoral College by five votes, despite Al Gore receiving more popular votes.

5. 2016: Donald Trump won the Electoral College by 77 votes despite Hillary Clinton receiving more popular votes.

The two most similar outcomes of these elections are in 1888 and 2016 when a candidate secured a solid majority in the Electoral College despite losing the popular vote. In contrast, in 1876, Hayes narrowly won the Electoral College by the slimmest of margins even though Tilden’s portion of the popular vote received was three percentage points higher. Specifically, Tilden got 50.9 percent of the popular vote to Hayes’ 47.9 percent. In 2000, the Electoral College and popular vote were very close. Gore received 48.4 percent of the popular vote to Bush’s 47.9 percent.

Many changes in how we elect the president have occurred since the first election 1788. The most significant changes were about who could vote. However, changes related to the Electoral College have also been made. The first came in 1804 as the Twelfth Amendment, which required electors to cast separate votes for president and vice president. Before then, each elector in the Electoral College cast two votes for president with no distinction between the presidency and vice presidency. The candidate with the most electoral votes became president, the runner-up vice president. The election in 1800 revealed a flaw. Thomas Jefferson and Aaron Burr each received an equal number of electoral votes, leading to a tie that had to be broken in the House. Tie-breaking rules are an important element of market design.

Individual states adopted other changes. State legislatures have the power to decide how their electors are chosen. Initially, they had varying methods of selecting electors, including popular vote, appointment by state legislatures, or a combination. In the 1824 election mentioned above, there was no popular vote in 6 of the then 24 states in the union. Since 1868, however, 48 states have adopted the winner-takes-all rule, where the candidate with the most votes in a state gets all its electoral votes. The only exceptions are Maine and Nebraska, where electoral votes are allocated based on the popular vote counts within each congressional district. A further change came in 1961 with the Twenty-third Amendment, which granted the District of Columbia three electoral votes so its residents could participate in presidential elections. The district also uses the winner-takes-all rule.

A more recent issue has been non-compliant electors – those who do not vote according to their state’s popular vote. Three electors from Washington state refused to vote for Hillary Clinton in 2016 when she won the popular vote. The three were fined $1,000 each for their non-compliance. The Supreme Court in Chiafalo v. Washington, 591 US 578 (2020) addressed the matter in a unanimous decision, affirming the state court’s ruling that states can require electors to vote by the state’s popular vote and that they can penalize or replace non-compliant electors. Though not an actual change to the system, the verdict has added to the debate over whether the Electoral College should be replaced.

The closest the US has come to abolishing the Electoral College was the Bayh-Celler amendment, proposed in 1969, which explicitly aimed to replace the Electoral College with the popular vote. It passed in the House but failed in the Senate due to a filibuster. Those who objected were from smaller states who argued that getting rid of the Electoral College would reduce their political influence, reinforcing the idea that those in power prioritize hanging onto it.

Another recent effort to remove the Electoral College, though not an official change to the national system, has been the National Popular Vote Interstate Compact. The Compact is a formal interstate agreement to award all participating states’ electoral votes to whichever presidential ticket wins the national popular vote. If it is ever adopted by states comprising 270 or more electoral votes, it would de facto guarantee that the nation’s popular vote winner becomes president without technically doing away with the Electoral College. As of this writing, the Compact has been adopted by 17 states, comprising 209 electoral votes, further adding to the debate over the role of the Electoral College. This approach could also be used to leverage US election reform. If the Compact has 270 votes, then these states can argue, “Reform or we will impose the popular vote, which we can do unilaterally.”

The evolution of the electoral process points to a desire for change. Although most proposals have favored a shift to the popular vote, including a recent call from vice-presidential candidate Tim Walz (Tait 2024), we suggest a more thoughtful approach based on insights from market design. The benefit comes from thinking beyond the popular vote and Electoral College options. Voting rules evolve in most political systems, and America should not be complacent with the status quo. Continuous improvement is essential. A looming crisis may provide an opportunity.

3 Market Design

After inventing the mobile phone in 1973, the US mobile market fell behind Europe in the early 1990s because the US regulator failed to license radio spectrum essential for mobile communications. This motivated Congress to pass legislation granting the regulator, the Federal Communications Commission (FCC), spectrum auction authority. The FCC, recognizing its inexperience in auctions, began a thoughtful process to identify the best approach to auctions informed by debate among experts (McMillan 1994). The result was an initial auction design that performed well in this complex setting. In 2020, Paul Milgrom and Robert Wilson won the Nobel Prize in Economics partly for their pioneering, practical auction design that was adopted globally.

A successful market design identifies practical tradeoffs and carefully tests them against prioritized goals. The FCC has continued to improve the design of its auctions to address issues identified from experience. Early on, for example, bidders encoded messages in their bids to collude (Cramton and Schwartz 2000). Later, auctions adjusted the rules to prevent this behavior. This kind of iterative adjustment is a hallmark of effective market design. Similarly, lessons from ranked-choice voting implementations – such as those in Alaska, Maine, and various municipalities – have shown that incomplete ballots can compromise results yet are an often-overlooked issue in elections that allow voters to list multiple preferences (Kilgour, Grègoire, and Foley 2020; Boudreau, Mathews, and Schwartz 2024).

The market design approach is pragmatic and problem-specific, similar to the clinical approach to poverty reduction policies in development (Sachs 2005). Solutions are tailored to specific problems: one country might require mosquito nets to combat malaria, while another needs debt relief. As market design adapts tools to the issue, successful programs are tested locally and broadly adopted if proven effective.

There are many other examples of market design in action. The Clean Air Act’s sulfur dioxide market, Uber’s redesign of ride-sharing markets, and Milgrom’s auction for milk products highlight the diverse applications of these principles. Market design is not static. It evolves as new problems and technologies emerge, with lab and field trials essential in refining theory and practice.

Experience also teaches us to account for real-world constraints. Alvin Roth’s work on organ donor matching (2016) provides a powerful example: given that monetary exchanges are not allowed, the mechanism must accommodate specific challenges like finding compatible donors and managing geographic and temporal constraints. Similarly, Roth (2008) described modifications to the National Resident Matching Program to address the needs of married medical school graduates who require placements in nearby hospitals. This constraint became more significant as more women entered the medical profession, illustrating how market design evolves to keep pace with societal changes.

Market design starts with broad goals. In spectrum auctions, the FCC wants to allocate the spectrum to those most effective at providing valuable communications services for consumers. In the residency match, hospitals and new doctors want efficient doctor assignments to hospitals. In wholesale electricity, the regulator seeks reliable electricity at the lowest cost.

Once a goal is established, an iterative process harnesses the scientific method of continuous improvement through design and experiment. From theory, promising trial mechanisms are identified. These designs are then tested in the laboratory or a low-stakes field environment. The mechanism is then rolled out, recognizing the importance of measuring and publicly reporting performance. Weaknesses are found, and improvements are developed. Proven designs are adopted more widely, geographically, and across sectors. Spectrum auctions, kidney exchanges, and electricity markets gradually improved, and adoption broadened over decades of continuous refinement.

Despite the success of market design, there are instances where not adopting the approach has led to failure. Troubled by high costs, Congress passed legislation requiring the Department of Health and Human Services (HHS) to use competitive bidding to price Medicare’s durable medical equipment. Cramton, Katzman, and Ellermeyer (2010) showed that HHS’s proposed auction design had fatal flaws. These flaws were confirmed experimentally (Merlob, Plott, and Zhang 2012). Nonetheless, HHS pushed forward with its flawed approach, and unsurprisingly, the identified flaws materialized in the field. Only after a costly failure and a ten-year delay did HHS begin to implement the recommendations of the market designers.

In the most challenging settings, even good market designs can fail. Texas’s winter storm Uri in February 2021 is an example. Sustained extreme cold produced an impossible gap between demand and supply, causing multi-day outages for one-quarter of Texans. The dollar cost exceeded $100 billion, and 262 Texans died. Despite having a best-practice wholesale electricity market design, it was vulnerable to extreme cold. The event highlighted the need for electricity markets globally to be resilient to extreme weather, which is becoming increasingly important with climate change and the transition to renewable electricity (Cramton 2022). The crisis has triggered design improvements that are gradually being adopted in markets globally (Cramton et al. 2025).

Market design is an iterative process that tailors mechanisms to specific problems and refines them based on empirical evidence and real-world constraints. By learning from successes and failures, we can better inform efforts to redesign complex systems, such as the one for electing the president of the United States.

4 Election Design

Long before Arrow’s (1950) impossibility theorem, a substantial body of literature explored the challenges of aggregating individual votes into collective social preferences. Market design principles become relevant because of the complex tradeoffs in crafting an effective electoral system. A critical debate is whether to judge an election system by how closely it mirrors an ideal or to take a more practical approach that weighs feasible alternatives on their merits. As Sen (2009) explains, in the context of social justice, the goal should be “comparative assessment and not merely identifying a transcendental ideal.” Arrow’s theorem undercuts any notion of measuring an election system against an unattainable ideal, leaving only practical, comparative assessments.

Electoral goals are many. Here, we outline some while acknowledging that the list is not exhaustive.

5 Prioritizing Goals

In designing elections, some tradeoffs are unavoidable. For instance, Condorcet consistency, where the candidate preferred by a majority wins, and responsiveness to voter intensity, as seen in quadratic voting, are mutually exclusive (Nitzan and Nitzan 2024). In a scenario where 51 out of 100 voters prefer Candidate A but with lukewarm enthusiasm, and 49 voters strongly prefer Candidate B, a Condorcet consistent system would elect A. However, a system like quadratic voting, which allows voters to weigh their preferences more heavily at some cost, might result in B winning. Decisions about which goal to prioritize – majority rule or preference intensity – are central to designing a fair and effective electoral system.

Ultimately, the design of an electoral system must weigh all tradeoffs and allow for a continuous process of refinement and adaptation, learning from past implementations and evolving political, social, and technological contexts.

6 Alternative Voting Rules

We now discuss voting rules tested in smaller settings to determine which components merit consideration. These are fledgling examples of change that could greatly benefit from a market design approach.

7 More Drastic Changes

8 Modernizing the Voting System

With modern technology, it is possible to design a voting system that is both secure and sophisticated. Today, even people experiencing homelessness often have access to the internet and cell phones. If financial institutions can securely protect millions of individual accounts, we could develop a system that protects the integrity of a more modern and complex voting system. As suggested by some voting theorists, transparency and accuracy are the top priorities. In that case, we should have room to innovate beyond a simple one-person, one-vote system within local jurisdictions.

However, more radical innovations might be limited if transparency and audibility are deemed more important than other goals. Our current system reflects the importance of continuity with past elections and the ability to transparently double-check the accuracy of vote counts. Shifting priorities away from this may open the door to more advanced and responsive voting systems.

9 Conclusions

The 2024 Presidential election temporarily turned a bubbling-over crisis into a more slow-burning situation. Still, citizens understand that the status quo of presidential elections leaves much room for improvement. Simply going from the Electoral College to the Popular Vote is undesirable if smaller states would be made worse off and would not be on board with a constitutional amendment. However, the market design approach that we have described provides a pragmatic way forward. Although existing theoretical and empirical work has not identified a definitive choice, promising alternatives exist, and we summarize some of them here with the understanding that the best opportunity for fixing the presidential election is to first experiment with these election designs at the state and local levels.

Ranked choice voting is already being tested, and expanded testing would be easily implemented. It is a strong option because it allows voters to express their preferences over candidates more fully and may, in time, lead to less polarizing candidates who must compete to be favored by voters as – if not their top choice – a second or third choice. For example, Maine adopted a version of rank-choice voting that was first used in the November 2018 US House elections. Alaska also has been using ranked-choice-voting, beginning with the August 2022 special election for the US House after the death of Republican incumbent Don Young. Further experimentation is on the horizon, as evidenced by Colorado, Idaho, Nevada, Oregon, and Washington, DC, offering referenda to do so on their November 2024 ballots.

However, ranked-choice voting is not a single process; there are many ways in which it can be implemented. Should candidate elimination be based on the fewest first-place votes, as in Alaska and Maine, or by the fewest total votes, as recommended by Foley and Maskin (2022)? This is an important detail in implementing ranked-choice voting: it may lead to drastic differences in how voters choose to rank their opponents on their ballots, how candidates position themselves in the race, and ultimately, who wins the election. For example, elimination by the fewest first-place votes is usually called “instant runoff” or a Hare system. It can eliminate a Condorcet winner, while eliminating by the fewest total votes is known as “Borda’s Elimination Rule” and never eliminates a Condorcet winner. For more on these and other voting rules and how they sometimes perform poorly, see Felsenthal and Nurmi (2019).

Other important rules within ranked-choice voting that need to be vetted are whether to require strict rankings or allow for tied rankings and whether or how to treat partial rankings in the elimination procedures. Open versus closed primaries can also impact the election. State elections provide the perfect opportunity to experiment with election rules. At the same time, metrics for election performance, such as voter turnout and voter satisfaction, can be worked out to see which version works most effectively and to make appropriate adjustments to improve the process.

Alaska in 2024 provides a cautionary tale concerning market design’s iterative improvement process. Their ballot included a proposition to revert from ranked-choice voting to the previous popular vote rule. In the end, ranked-choice voting survived by the narrowest of margins. However, the instinct to discard rather than improve experimental changes is concerning. We should not expect the first iteration of a new voting rule to be perfect but as a step towards a better system.

However, Alaska’s experimentation with ranked-choice voting highlights our final point: experimentation at local and state levels could not only provide benefits to the stakeholders in those elections, but the benefits should filter up to state and national levels in several ways. This includes potential changes in Democrat and Republican primaries. Currently, both parties have convoluted primary structures likely resulting from party “elders” wanting to have control over nominees. Much like passing a Constitutional Amendment, convincing party elders to relinquish their power will likely be a daunting obstacle to reform.

Currently, the Democrats allocate convention delegates to states half based on each state’s contribution to electors in the Electoral College and half based on what percentage of Democrat voters live there. That means California gets more delegates than is in proportion to their Electoral College representation. Some view Californians as being more progressive on many issues and pushing the party to nominate a more progressive candidate. Republicans have a similar system, although it awards delegates based on Gubernatorial and legislative control by Republicans in the state rather than by the percentage of Republican voters in that state. Having elected a Republican Governor, US Senators, or a majority of US House seats. Thus, solid red states push more conservative candidates.

These primary processes result in the country having highly polarized candidates. The moderates in the middle feel ignored. Most presidential nominees appeal to a broad swath of party members in the primaries and take positions left/right of center. However, once they become the party nominee, they must tack back to the middle to appeal to the general electorate, and this turns off many voters who get tired of these political machinations and view them as flip-flopping or downright lying. A primary system that produces a more centrist candidate in the first place would alleviate this problem for both parties.

This is the opposite of what makes sense for the parties. Parties should seek to nominate candidates who thrive in more middle-of-the-road swing states. Instead, under the current primary systems, more extreme candidates are nominated, requiring massive amounts of money to be spent trying to trick swing-state voters into believing that the extreme candidates are centrists.

Better-designed elections, whether at the local, state, or primary levels, have two potential benefits. First, what is learned at lower levels can be applied and further refined at higher levels. Second, if election rules emerge that elect more broadly appealing candidates, then we should start to see more broadly appealing candidates elevated to the state and national levels, and the investment locally will pay dividends more broadly over time.

We have recommended several possible election formats as a starting point, but our list is not exhaustive. Efforts should be part of an ongoing process involving public comment, experience, and experimentation to determine what works and what does not. The debate should be much broader and more thoughtful than whether to replace the Electoral College with a popular vote. The nation needs an electoral system that can evolve to represent voters’ interests best and reflect our societal changes. We propose that such a system be developed using a market design approach.