Cost of Elections

Understanding Elections from an Administrative Burden Perspective: Making Participation Easy by Minimizing Unnecessary Barriers

Much of the discussion of elections and election laws centers on access to voting and election security, and the ostensible tradeoff between the two. This paper presents an alternative perspective for evaluating election administration policies, drawing from the administrative burdens framework.

September 13, 2023
Donald Moynihan
McCourt School of Public Policy, Georgetown University

Much of the discussion of elections and election laws centers on access to voting and election security, and the ostensible tradeoff between the two. This paper presents an alternative perspective for evaluating election administration policies, drawing from the administrative burdens framework. Administrative burdens represent the costs that people encounter and experience from implemented policies, including learning, compliance and psychological costs. While many new election laws might not have easily detectable aggregate effects on turnout or election security, they may still impose or reduce burdens on citizens. All else equal, policymakers should have a bias toward reducing the time tax on citizens in voting and other interactions with the government, both to minimize unnecessary hassles on the public, and to eliminate barriers to basic political rights. An administrative burden framework can help us to better evaluate election laws and administration from a neutral perspective, separate and apart from often unsubstantiated claims about turnout, partisan impact, and security. Recent changes to election laws in Texas, for example, impose new administrative requirements for absentee voting that are non-intuitive, and resulted in many voters spoiling their ballots. Washington DC has taken the opposite approach, adopting fully automatic enrollment in voter registration in a way that effectively eliminates the most consequential burden in voting.

What are administrative burdens?

Administrative burdens represent the costs that people encounter in their experience with government policies as they have actually been implemented. These costs fall primarily in three areas, identified in Table 1.

Table 1.  Three Components of Administrative Burden

Learning costs Time and effort expended to learn about the program or service, ascertain eligibility status, the nature of benefits, conditions that must be satisfied, and how to gain access.
Compliance costs Provision of information and documentation to demonstrate eligibility; financial costs to access services (such as fees, legal representation, travel costs, costs of acquiring relevant documentation); avoiding or responding to discretionary demands or follow-up made by administrators.
Psychological costs Stigma arising from applying for and participating in an unpopular program; loss of autonomy that comes from intrusive administrative supervision; frustration at dealing with learning and compliance costs, unjust or unnecessary procedures; stresses that arise from uncertainty about whether a citizen can negotiate processes and compliance costs; fear about the coercive face of state power.

Adapted from Herd and Moynihan (2018)

The literature on the topic suggests some basic lessons relevant to elections.

Small hassles can have big effects. An extra step that might seem reasonable (or even minimal) to an administrator or policymaker can predictably lead to lower public participation in a public service or right, even one that citizens value. Burdens are especially likely to reduce participation when they are new and unfamiliar, or complex. The flip side of this observation is that removing a “pain point” in an administrative process can have a potentially large positive effect on uptake and participation.

Burdens have distributive effects, i.e. they fall heavier on some groups more than others (Herd, et al 2023). As such, burdens may reinforce existing patterns of inequality. One reason that some people experience burdens more than others is that they interact with programs or administrative processes that are more burdensome, such as means-tested welfare programs that largely require extensive documentation, while others avoid such processes. In terms of elections, those who move more frequently (who are also likely to be poorer) must engage with voter registration processes more frequently. Another reason is variation in human capital and other resources. Those with less education will struggle more with complex forms and legalese.  Those with less financial resources have less ability to hire someone to help them and may experience a sense of scarcity that makes it difficult to deal with administrative processes. Those with poorer physical and mental health are less likely to persevere through onerous requirements. (Bell et al., 2023). Another reason for distributive effects is discrimination by administrative actors: they may ask more of groups they are suspicious of relative to individuals they trust (Olsen, Khyse-Andersen and Moynihan 2022).

Policy actors may sometimes strategically use burdens to engage in policy making by other means. If policymakers oppose a policy, but cannot gain support to reverse or limit it directly, they may instead make the service onerous to access. This is by no means the only reason for administrative burdens. They may also arise due to a lack of bureaucratic attention to the costs of frictions on the public, a form of benign neglect, or a lack of administrative capacity to reduce those frictions. But administrative burdens can also be especially attractive to policymakers in contexts where it is politically or legally risky for them to express their true intent (Herd and Moynihan 2018). For example, policymakers who bear animus toward a racial group, or believe that their party will lose out if there is greater participation, cannot explicitly use those justifications for changes to election laws. But they can, instead, suggest that new requirements that significantly increase burdens will minimize fraud and increase voter trust.

Across policy areas, street level bureaucrats and third parties play a crucial role in helping individuals to overcome burdens resulting from state action. These can include for-profit helpers, such as tax preparers who help clients claim the Earned Income Tax Credit, or advocacy issue groups, or simply friends and families. In voting, political parties and aligned organizations play such a role, as do non-partisan voting-advocacy groups, such as the League of Women Voters. In addition to third parties, street level bureaucrats, most obviously election officials, can play a direct role in increasing or buffering burdens. In the area of elections, such officials can use their discretion in ways that increase or reduce turnout (Burden et al 2013).

Policymakers should consider evidence-based costs and benefits of state actions that create burdens on the public. Mirroring the rise of this new research framework, policymakers are increasingly aware of the effects of burdens. In two executive orders, the Biden administration has drawn attention to reducing burdens, tying it to the goal of improving social equity.1 At the broadest level, this implies that policymakers should explicitly weigh the costs and benefits of state actions that might give rise to burdens. In the context of election, the proposed benefits of new election laws are often to improve election integrity, or confidence in elections. The costs may be that voters spend more time and effort dealing with increasingly frustrating processes, which itself saps trust in the capacity of government to enable access to voting rights. Beyond simply stating the tradeoff between values, it is better to document their scale. The evidence that new election laws protect against fraud has to deal with the reality that there is simply not strong evidence that election fraud exists on a meaningful scale, for the very good reason that fraud comes with the risk of a felony conviction.2 On the other hand there is tangible evidence that the same election laws can be so confusing, onerous and time-consuming as to exclude people who want to participate. Alternatively, election laws designed to reduce burdens, such as vote-by-mail, or automatic registration can significantly help voters.

Applying administrative burdens to elections

Efforts to apply administrative burden theory to elections can draw on an extensive literature that seeks to examine what factors make voting easier or harder from the perspective of the voter, reflecting the idea that Anthony Downs proposed in An Economic Theory of Democracy: “Heretofore we have assumed that voting is a costless act, but this assumption is self-contradictory because every act takes time. In fact, time is the principal cost of voting: time to register, to discover what parties are running, to deliberate, to go to the polls, and to mark the ballot. Since time is a scarce resource, voting is inherently costly” (1957, 265).

While I won’t review this literature here in depth, some points are salient to incorporating an administrative burden framework to elections.

It is not just that voting is costly, but the costs can vary by group and location. Within a jurisdiction, voting is costlier for someone who does not own a car that brings them to a polling place, or for voters whose polling places are oversubscribed and understaffed. Voting costs also vary across locations, depending on the nature of election laws, and the resources that citizens, election officials, and third-party actors can draw on.

Burdens affect not just voters, but other actors in the voting process, such as third parties and election officials. Laws that excessively punish errors in third party voter registration, or the provision of a bottle of water to those waiting in line to vote, seek to minimize the role of civic groups in reducing burdens in elections. For example, Florida has ratcheted up fines for errors that third parties make in registering voters, e.g., by registering voters for the wrong district, or turning in registrations late (within 10 days of being collected). The scale of the fines – which can be up to $250,000 – and the likelihood of some errors – for example, voters who incorrectly attest they live in the wrong county could trigger a fine – has a clear chilling effect on voter registration drives, and on the number of people registered. This, in turn, can reinforce patterns of inequality in access, since voters of color in Florida are much more likely to be registered via voter registration drives (Levine and Witherspoon 2023).

Election officials are more likely to oppose election laws if it makes their jobs more onerous (Burden et al. 2012). In some cases, new laws can create a zero-sum distribution for burdens. For example, early voting processes or same day registration that demand more election officials over an extended period may reduce burdens on citizens, but add them to officials. In some cases, new laws can be lose-lose, imposing new burdens both on election officials and voters. The increased hassles in Texas vote by mail described below are an example. And in some cases, new laws can reduce burdens on both voters and election officials, as is the case with the Washington DC automating voter registration (AVR) law detailed below. The design of state actions should consider both the burdens imposed on the public, as well as the state actors implementing the policies.

Voter registration, and not voting itself, is the most consequential site of burdens in election processes. Voter registration adds an additional step in the process of voting. If not registered, voters cannot participate. For example, potential voters who become interested in voting at the last minute might google about how to register. But only in states without strict voter registration deadlines can the citizen act on what they learned and participate. Using such google searches as a barometer of interest, a study of the 2012 election estimated that 3-4 million more citizens would have registered without voter registration deadlines (Street, Murray, Blitzer and Patel 2015.).

The burdens of voter registration mechanically fall more heavily on those who move more frequently, who tend to be low-income, younger, and persons of color. Policies that ease registration are especially likely to help voters with lower incomes, reducing the income gap between low- and high-income voters in voting (Rigby and Springer 2011). These policies can similarly reduce participation gaps among young people and communities of color.

Historically, those intent on reducing turnout focused most intently on registration. Arguments about election integrity provided a rationale for the initial adoption of registration processes, even as these policies were simultaneously a method to exclude poorer, less educated and immigrants, and later, Black citizens. The 19th century shift towards requiring affirmative voter registration prior to voting contributed to a substantial decline in participation nationwide. Voter registration turned voting from a one-step process, where eligible voters could simply participate by showing up at the polls, into a two-step process, where those seeking to vote first had to engage in the separate and time-limited process of affirmatively indicating a desire to vote, typically by providing a registration form and supporting materials at a particular time and place. The greatest administrative efforts in the Jim Crow south to discourage Blacks from participation in elections were at the registration stage. Clerks would work slowly, not provide assistance, limit hours and days for registration, require that Black voters be accompanied by White sponsors, impose more onerous tests (including literacy test) and use their discretion to decide what answers were acceptable (Herd and Moynihan 2018).

Table 2: Policies that Raise or Reduce Administrative Burdens in Registration

Burdens imposed Policies that reduce burdens Policies that increase burdens
Learning costs: knowing registration policies, knowing documentation required and deadlines

Compliance costs: going to location to register; acquiring appropriate documentation

Election day registration and same day registration allow “one-stop shopping”

Registration opportunities at other routine interactions with state (DMV, welfare provider, high-schools for pre-registering individuals below 18 years)

Online registration: does not require special trip to register, but usually requires DMV-issued ID

Automatic registration: using administrative data to register citizens, changing default from opt-in to opt-out

Requiring documentary proof of citizenship or proof of residence to register

Reducing election day and same day registration opportunities

Purging the rolls based on inadequate data, requiring citizens to re-register rather than automatically updating information

Shifting risk to third parties who run registration drives, or preventing them from conducting registration drives

Lengthening the residency requirement

Adapted from Herd and Moynihan (2018)

Defaults matter enormously. A major lesson from behavioral science is that defaults matter immensely to outcomes. People tend to stick with the default choice, even when it is in their interest and relatively costless to switch to an alternative. This might reflect a variety of behavioral tendencies – a preference for the status quo, a fear of regret from loss as a result of action, or a tendency to put off an action for a future date that may never come.

The canonical example of this point is retirement savings. Simply changing the default from requiring individuals to opt-in to retirement accounts to opting-out generates enormous changes in outcomes. Since opting-in helps the individual (by gaining matching contributions from employers, or more generally planning for the future in a way they would not otherwise do), the behavioral change is beneficial both for the individual and the community, helping to minimize poverty among older adults. A number of countries and states have pushed private employers to switch to making enrollment in a retirement saving program the default option, and requiring individuals to affirmatively opt-out (GAO 2022).

Another example: in health and welfare programs, annual renewal periods are a time when many individuals and families lose coverage. They may miss deadlines, or struggle with paperwork and documentation, and lose resources that they highly value. Some will eventually find their way back to the program, but the “churn” on and off the program is costly, both for the people who lose benefits, and for state administrators who have to manage new enrollments for the formerly enrolled, which are more time-consuming than renewal of status. To offer a sense of how big a problem churn is, the suspension of Medicaid renewal procedures during the pandemic Public Health Emergency was associated with a greater increase in coverage than the implementation of the Affordable Care Act (Obamacare) over a similar time period. In other words, changing the administrative default to require individuals to opt-out had a greater effect than the most important health policy expansion in decades (Dague and Ukert 2023).

To varying degrees, policymakers have sought to automate both initial enrollments and renewals. One strategy is to have single applications for multiple programs with similar eligibility, or to use data from one of those programs to enroll an individual in another program. Another is to use existing administrative data to confirm the individual remains eligible. Such a tactic is easier when administrative data aligns with eligibility criteria.

Automatic voter registration might be viewed as a subcategory of a broader strategy of automatic enrollment applied in other settings. It is based on the same principles:

  • Minor administrative hassles may prevent people from acting in a way consistent with their personal interests and better societal outcomes
  • Defaults matter enormously in altering behavior, and;
  • With sufficient administrative data, state actors can alter the defaults in ways that improve individual and societal outcomes.

Shift from turnout, to burdens, as way to evaluate election laws and administration

The centrality of election administration in policy discussions since the 2000 election reflects an assumption that election policies have partisan effects on turnout. But our standard for evaluating new voting laws and administrative practices should not only be “do they reduce turnout?” or “will they benefit one party?” but rather “do new laws impose pointless burdens on voting?” and “are those burdens targeted toward some groups more than others?” A good law should create benefits for the public and avoid unnecessary costs. Many election laws are restricting clear benefits to voters — convenience and access — while imposing pointless new costs upon them.

One reason that election laws don’t have marked partisan effects is because they often only affect a small group of voters, and that small group may be relatively bipartisan (Grimmer and Hersh 2023). However, while the turnout and partisan effects of election laws may be limited, research shows significant changes to learning and compliance on the voters affected, often minority voters. For example, White, Nathan and Faller (2015) show that non-white voters were less likely to receive helpful information about new voting laws, suggesting a bias that resulted in higher learning costs for such voters. Racial and ethnic minorities were also the most likely not to have IDs that would satisfy new election laws, meaning they bore greater compliance costs. Grimmer and Yoder (2022) point to evidence that new voter ID laws do reduce turnout for those voters who previously lacked an ID, suggesting that the effects of compliance laws are specific to those who have to do the most to comply with new laws (Grimmer and Yoder 2022). In the aftermath of voter ID laws (Cantoni, Enrico and Pons 2019), and counties affected by post-Shelby laws (Komisarchik and White, 2022), racial and ethnic minorities engaged in extra mobilization efforts to overcome new burdens.

In short, election laws can still create new burdens, and even burdens that fall on some groups more than others, even if they don’t affect aggregate turnout. The implication is that the effects of new voting restrictions may be best detected by looking at the burdens that are created, and how those burdens are distributed. Voter turnout is only one, and perhaps not the best, measure of the effects of laws that impose burdens on voters. Measuring administrative burdens provides a more direct measure of the effects of these laws, which may or may not result in changes in turnout.3

I next turn to some examples of what it means in practical terms to focus on burdens in elections. I first discuss automatic enrollment, using Washington DC as an example, before discussing the imposition of new burdens in Texas.

Minimizing burdens through automatic enrollment

There are a variety of ways to minimize administrative burdens (Herd et al., 2023). One is to provide informational nudges or weak defaults. Another is to eliminate administrative processes entirely by shifting responsibilities and burdens from citizens to state actors better equipped to handle them. The tool of automatic enrollment takes this latter approach, and offers an example of the type of policy that awareness of burdens in elections could encourage.

For the past thirty years, the federal government has encouraged easing registration processes. The National Voter Registration Act (NVRA), also known as the motor voter law, was built around the logic that the state should use a wide array of citizen-state interactions with citizens to make registration easier. The law encouraged the use of DMVs as a site of registration, but other parts of the NVRA were less widely applied, including agencies providing welfare benefits to provide voter registration opportunities. However, the minimum processes required by the NVRA still leave burdens in place for registration. While state agencies are required to offer customers the opportunity to register to vote, such registration is only required to occur on an opt-in basis, and individuals must assess their own eligibility and are typically asked to provide additional information or duplicate information already provided to the agency in order to complete the form. While the NVRA has improved registration opportunities at state agencies, it still leaves in place administrative burdens for those seeking to register.

Recognizing these limitations, 22 states and Washington, DC, have embraced some form of automatic voter registration at DMV offices. The use of automatic enrollment creates the impression that this change has massively expanded registration. But what AVR means can vary from one setting to another, and the effects on registration differ depending on the extent to which the AVR system considers and removes administrative burdens.

Partial or Front-end AVR

Roughly half of the states with AVR have what is sometimes described as partial or front-end AVR. Here, people at a DMV or other covered agency are told that their information will be used for voter registration (or updating of any existing registration), unless they opt-out. They are then presented with the option to decline registration. Partial AVR (PAVR) has increased registration (Grimmer and Rodden 2022) and turnout (Rakick 2019). But it still leaves a significant amount of friction in place, requiring more effort to complete the registration process at that particular moment, even if it means minimizing effort compared to the basic NVRA system where DMV customers must affirmatively select a voter registration option. Partial AVR systems crucially still require customers to answer questions during the transaction – about whether they want to register, about their eligibility to register, and various voting preferences, including party affiliation, mail ballot options, or language preference. DMV customers, believing that these voter registration questions will lengthen the transaction, may want to complete their primary business and move on, to the point that many will not take any extra time to register to vote, or verify their registration information. A large share of eligible customers can decline unnecessarily due to imperfect information, because they incorrectly believe they are already registered to vote or that their registration is current, or because they are unsure of their status, in the case of teenagers, and individuals with past felony convictions. Other individuals may be confused about why they are being asked about voter registration at the DMV, or trust themselves to register or update their information later when they have more time. In a study of the PAVR process used by Colorado from 2017 to 2020, Grimmer and Rodden (2022) find that more than 70% of unregistered and potentially eligible people were opting-out of voter registration, even when presented with a default voter registration opportunity as part of their DMV transaction.4 Similarly, they found that one third of existing registered voters who declined to update their registration had out-of-date registration information and actually needed to update.

PAVR systems, if poorly designed, can also raise risks for non-citizens. If the system does not affirmatively filter out individuals who provide the DMV with proof of non-citizenship (such as a green card), a PAVR system at the DMV can also capture non-citizens who erroneously indicate citizenship as part of the voter registration questions and fail to opt-out. This could occur for any number of reasons, including limited English proficiency, momentary inattention, confusion, or belief that answering yes to these questions is required to obtain a license. The false claim of citizenship and erroneous requests to register that result can create significant immigration challenges for a non-citizen, including denial of a naturalization application and deportation under certain circumstances, even if the non-citizen never votes. Obviously, a system designed to reduce administrative burdens should not indirectly create significant legal risks, and further burdens for vulnerable subsets of the affected population.

Secure or Back-end AVR

A back-end or secure AVR (SAVR) approach creates a more frictionless process for registration than a PAVR system, and requires that an eligible individual make more effort to opt-out of the registration process than to be registered. A SAVR system uses administrative data from state agency transactions to establish eligibility and automatically register clearly eligible individuals, while affirmatively filtering out ineligible non-citizens. If the state can establish a person’s eligibility (including definitive proof of U.S. citizenship) based on the information provided as part of a DMV or other agency transaction, the person is automatically registered. Similarly, any existing registered voter has their information updated based on address and name information provided to a state agency, to ensure that address and name records reflect the information most recently provided by the voter. Following an agency transaction that results in a new registration or a registration update, affected individuals are sent a mailer advising them of the change and providing a prepaid return mailer that allows them to opt-out or correct any information.

In the past eight years, Oregon, Colorado, and Alaska set the benchmark for this approach, which has now been adopted in six more states, and relies on structuring choice architecture to favor registration. In Colorado, which shifted to SAVR in 2020, fewer than 1% of people who are automatically registered to vote return the opt-out mailer, resulting in a system where registration is the strong default for anyone the state knows is clearly eligible to vote. Only individuals who truly do not want to be registered to vote take the additional step of returning the post-transaction mailer, whereas the remainder of the covered population remains registered without any additional effort required on their part. These individuals are free to vote, if they wish, with registration having been accomplished through an essentially frictionless process. Grimmer and Rodden (2022) illustrate that changing the default to a post-transaction opt-out relative to within transaction opt-out has a large effect at DMV voter registration in Colorado, significantly decreasing the number of voters who opt out, and effectively doubling the registration rate at the DMV, with disproportionate impact on younger voters. These are very large effects for relatively low costs, an estimated $86,000 to implement the new system.

According to the Colorado Secretary of State: “Registering to vote and voting itself should not be a burden. These are our constitutional rights. State governments should be seamlessly offering potential voters the option to register.” This reflects the basic assumption that underpins automatic enrollment: where it can, the government should eliminate frictions to valued rights and services. In a SAVR system, learning and compliance costs for registering to vote are effectively eliminated. Eligible voters are automatically registered to vote or have their registration updated to reflect the most current information available, and are then informed of the change and provided an opportunity to decline.

Some SAVR states have also sought to use Medicaid data to automatically register eligible voters who enroll in Medicaid. Like SAVR at the DMV, this approach builds on the requirements of the NVRA, which requires a voter registration opportunity as part of Medicaid enrollment. However, these states have faced implementation roadblocks due to a lack of clear guidance from the federal Department of Health and Human Services on how Medicaid data can be used in this fashion (Burness 2023).

Table 3.  Minimizing Administrative Burdens via Secure Automatic Voter Registration

Learning costs Citizens do not have to learn about registration processes. Compared to PAVR, SAVR eliminates need for individuals to keep track of registration status if asked to register at DMV.
Compliance costs The extra step of registering to vote is removed, minimizing additional visits to public offices and extra paperwork. Compared to PAVR, SAVR eliminates need to devote addition time completing paperwork when at DMV.
Psychological costs Minimizes frustration and other potential psychological costs associated with registration processes.

SAVR in Washington, DC 

In 2023, Washington DC adopted a theoretical extension of the Colorado model, passing the Automatic Voter Registration Expansion Amendment Act. Washington, DC already had a PAVR approach in place, but, as in other PAVR states, saw that large numbers of eligible DMV customers were declining voter registration or were not updating their registration when presented with the opportunity as part of the DMV transaction. The new law keeps this system largely in place, but supplements it with a second system that relies on the SAVR model. This additional system aims to significantly reduce friction around voter registration, and indeed, fully eliminate voter registration as a barrier, by creating a new category of “pre-approved” voters.

DC’s new law keeps the existing PAVR system in place for potentially eligible people who are not already registered to vote in the District. An unregistered person will still be provided the opportunity to decline voter registration during a DMV transaction, unless they provide proof of non-citizenship as part of their DMV transaction. This filtering out of non-citizens is a protective measure that avoids erroneous registrations and false claims of citizenship. As in other SAVR states, anyone who is already registered to vote will have their voter registration automatically updated based on new address or name information provided to the DMV and will be sent a notice alerting them of the change.

The new system adds a back-up approach to avoid the limitations of the existing PAVR system. Even if an unregistered person declines voter registration during the DMV transaction, the DMV system will still check if the information provided by the person establishes their eligibility to vote. If the agency can verify and confirm the person’s U.S. citizenship and other eligibility requirements for registration based on its records, the person’s information will still be shared with the Board of Elections. The Board of Elections will not register the person, but instead designate the clearly eligible person as “pre-approved” for registration and send them a notice in the mail advising them of their status and providing an opportunity to be removed from the “pre-approved” list.

People deemed “pre-approved” are not formally registered to vote, but their pathway to voting is effectively cleared. If they show up at the polls, they can cast a ballot, and the act of voting activates their registration for that election and all future elections. They do not need to take additional steps to affirmatively register or use the election day registration process before casting a ballot. Similarly, people on the “pre-approved” list will be mailed a ballot for their first four years on the list. If they return the ballot, they likewise activate their registration without the need for additional steps, and their ballot is counted. The same model will be used at Medicaid if the federal Department of Health and Human Services permits DC and other SAVR jurisdictions to implement.

This “pre-approval” system manages to balance multiple goals without putting additional administrative burdens on eligible voters. The DC model uses a largely similar approach to other SAVR states like Colorado. Where an individual is clearly eligible based on DMV or other agency data, their information is shared with election officials. However, recognizing that some share of people declining registration at the DC DMV or other covered agency are doing so because they want to remain registered to vote in another jurisdiction (e.g. members of Congress and their staff), the law instead designates eligible people who declined a registration opportunity at the DMV as “pre-approved.” People on the pre-approved list are not officially registered in the District (recognizing the particular need to maintain registration in another state among certain DC residents), but they are still able to vote. The affirmative act of voting confirms their desire to register in DC and officially registers them to vote for future elections. Therefore, to the extent that the state knows a person is eligible to vote in DC, there is no additional barrier to voting. No affirmative or separate act of registration is required. The burden of registration is eliminated both in theory and in practice, since an eligible person is able to vote, regardless of whether they have affirmatively registered.

The pre-approval system, like SAVR systems generally, also reduces burdens on election officials. Rather than relying on voters to update their address when they move, the system automatically updates registration records with new name and address data that voters provide during DMV and Medicaid transactions, and provides the voter notice of the change with an opportunity to correct the information if needed. This ensures registration records are as accurate as possible, avoiding misdirected mail ballots or requiring registration updates at the polls. Similarly, the “pre-approval” process avoids the time and resources necessary for election day registration for most eligible voters. Pre-approved voters can simply return a mail ballot and activate their registration, without the need to appear in person. Alternatively, they can appear at the polls and cast a ballot without the need to provide their information again for election day registration. By taking on the responsibility of identifying eligible voters and automatically updating information for people on the rolls, the state can reduce administrative burdens both on itself and the voters it serves.

Maximizing burdens by adding more confusing requirements

If using tools like automatic enrollment reflects the successful application of an administrative burden framework to minimize the time tax on individuals, it is not hard to find examples where governments do the opposite, adopting new and confusing processes that make it more difficult to participate, with little discernible value.

A recent example from Texas stands out. The state introduced a slate of new voting laws after the 2020 election, which were claimed to improve election integrity and faith in the process. At the time, many states had adjusted voting processes in the context of the pandemic, such as the use of drop-boxes, drive-thru voting, and extended early voting. There is no credible evidence that such efforts did anything other than improve voter convenience and enable massive turnout in the midst of a pandemic. Nevertheless, the backlash to those flexibilities affected not only these pandemic-related expansions, but also extended to policies that preceded the pandemic.

In Texas, absentee voting came under the microscope, reflecting a polarized assumption that any flexibilities in voting would benefit Democrats and endanger Republicans. President Trump heightened this polarization by claiming that mail-in voting would result in voter fraud. In Texas, absentee voting is already highly restricted, generally limited only to people who are out of town on election day, ill or disabled, or over the age of 65. These voters must affirmatively apply for an absentee ballot for each election through a paper form. There is no method of applying for an absentee ballot online.

Nevertheless, a 2021 Texas law added an additional requirement that voters must write down a unique personal ID number both on their absentee ballot application and on their ballot envelope. This could be the last four digits of their Social Security Number, or the number from their Texas driver’s license or non-driver ID card. In addition to matching the voter’s signature against registration records, election officials were also required to match this ID number against the ID number contained in the person’s voter registration record before accepting the application for processing or the ballot for tabulation.

The problem is that voters were unused to including such a number on their applications and  ballot envelopes, which for decades had simply required the voter’s signature. Not surprisingly, people did what they usually do, which is not provide such detail when returning the application or ballot envelope. Such a small change therefore tripped them up.

The mailers that absentee ballots are sent in (Figure 1), and the instructions they receive about voting (Figure 2), compound the problems. Basic principles of human centered design emphasize the need to keep form instructions simple and clear. Otherwise, people become overwhelmed, and ignore the instructions.

Figure 1: Texas absentee ballot mailer 
Texas absentee ballot mailer

Figure 2: Instructions for absentee ballots
Instructions for absentee ballots

The ballot instructions include lots of text which has to be included because of state law, much of it in 10-point font. These lengthy and wordy instructions increase learning costs, making it less likely that voters would see or understand the new requirements. And the volume of legalese could be off-putting to many potential voters, potentially discouraging them from even voting by mail at all due to psychological costs. Changes that are embedded in this sea of verbiage will be missed.

Even if voters are aware of the change, they may not know which ID number is on their voter registration. This is because when registering to vote, voters could have provided either the driver’s license or state card number, or the last four digits of their social security number. Even in attempting to comply, they might make an error. To be safe and ensure that their ballot would be counted, voters with a driver’s license or state ID card are now encouraged by the state to provide both that number and the last four digits of their social security number. (To somewhat alleviate this problem, the Texas Department of State has made an effort to match voter registration records with motor vehicle records and supplement the voter file with a driver’s license or state ID card number, where possible.)

As Travis County Clerk Dana DeBeauvoir noted: “This is all brand new…These are new obstacles for voters. They have never had to deal with this kind of problem in the past.” In some counties, 25-40% of vote-by-mail ballots were tossed in the first election where the new standards were employed. During the March 2022 primary, more than 12,000 ballot applications and 24,000 mail ballots were rejected, a 12% rejection rate. The rates were higher for Asian (19%), Black (16%) and Latino voters (16%), illustrating how burdens can fall heavier on some groups more than others. This number of rejections declined as voters became more familiar with the process, with the assistance of third-party groups like the League of Women Voters. Instructions accompanying the ballot envelopes were also slightly revised to better highlight the new requirements. However, the resulting rejection rate was still higher than had historically been the case (about 1%), a predictable outcome of the change.

Table 4.  Administrative Burdens Arising from Confusing Absentee Requirements in Texas

Learning costs Potential absentee voters have to learn about the new policy change, which is poorly communicated in official messaging, know which multiple identification numbers are appropriate for them to write on absentee mailers, and know their specific number.
Compliance costs Once additional learning costs are accounted for, compliance costs in completing applications and ballots are not significantly different. But higher potential for rejection triggers more paperwork to cure defective applications and ballots.
Psychological costs Fear of potential legal liability from committing error, stress from uncertainty about whether ballot complies with law, or from needing to cure a defective application or ballot, frustration at confusion that comes from process.

If automatic enrollment offers an example of a win-win when it comes to burdens – reducing burdens both on clients and on street-level bureaucrats – the new Texas absentee ballot requirements represent an example of a lose-lose, simultaneously imposing new burdens on both members of the public and election officials. Both voters and election officials have a new and more complex task. The voter must understand the new additional requirements and provide an identification number. Election officials have to check those numbers against state records. If there is an error, election officials must try to resolve it by engaging in outreach to individual voters, beyond whatever outreach they have already done to communicate the new policy. If an application or ballot is received in sufficient time, and the election official has the inclination and resources, they can reach out through multiple methods of contact and help the voter to remedy the error by curing the issue. Offices with fewer resources are obviously less likely to engage in extensive and repeated outreach. Election clerks in the state did not lobby for the change, but were left with dealing with the challenges of implementing it.


This paper has argued for applying an administrative burden perspective in understanding how to assess election laws and improve election administration, offering two contrasting examples. The value of this approach is that it centers the process around the experience of the public, rather than speculative and typically unsupported claims about election turnout and security. To the extent changes are made, they should reflect how such changes will make public experiences more or less onerous.

Another aspect of the administrative burden literature bears noting here, as it suggests that as citizens have more negative encounters with government, they are likely both to withdraw from those processes and lose confidence and trust in government (Michener 2018). This point is worth highlighting since many of the justifications around imposing more burdensome election processes rely on the claim that they will increase voter confidence in elections. However, voter confidence in elections is just as apt to decline under more burdensome processes. Making the process of political participation a baffling labyrinth of administrative requirements communicates that people are not welcome to engage in their democracy, and are viewed with suspicion, discouraging registration and voting.

Those arguing for more burdensome requirements to reduce fraud face the two-fold problem that a) there is little fraud to reduce, and b) new restrictions are often unrelated to fraud. It is, therefore, unsurprising that making elections more onerous does not increase trust in elections. States with ever more burdensome voting procedures do not have higher trust in their elections. For example, the addition of voter ID did not increase trust in elections (Ansolabehere and Persily, 2008; Stewart, Ansolabehere, and Persily, 2016). Indeed, we have much stronger evidence that claims of widespread fraud, even if unsubstantiated and even if corrected by fact checkers, are what drives declines in confidence in elections (Berlinski et al 2021).

One way to actually improve confidence in elections is to make the process functional and accessible for citizens. Voters tend to agree in very high numbers (above 95%) that voting was a positive experience with minimal problems. Such surveys exclude non-voters, and so we know little about how negative experiences discourage any sort of participation. But it is clear that confusing or negative experiences are correlated with declining trust. According to Stewart (2022):

Among the 94 percent of in-person voters who agreed their experience was mostly positive, 94 percent were confident their vote was counted as intended; for the 6 percent who had a negative experience, only 68 percent trusted that their vote was counted as intended.

An election system that took seriously the principles of administrative burden theory – that citizen-state interactions should be simple, accessible, and respectful – would do more to actively broaden who participates through tools like fully streamlined AVR, while doing away with pointless roadblocks that predictably trip-up the average voter.


Stephen Ansolabehere and Nathaniel Persily. 2008. Vote Fraud in the Eye of the Beholder: The Role of Public Opinion in the Challenge to Voter Identification Requirements. Harvard Law Review 121 (7) : 1737–1775.

Barreto, Matt, Stephen Nuño, Gabriel Sanchez and Hannah Walker. 2019. The racial implications of voter identification laws in America. American Politics Research 47(2):238– 249.

Bell, Elizabeth, Julian Christensen, Pamela Herd and Donald Moynihan. 2023. Health in Citizen-State Interactions: How Physical and Mental Health Problems Shape Experiences of Administrative Burden and Reduce Take-Up. Public Administration Review. 83(2): 385-400.

Burden, Barry C., David T. Canon, Kenneth R. Mayer and Donald P. Moynihan. 2012.  The Effect of Administrative Burden on Bureaucratic Perception of Policies: Evidence from Election Administration. Public Administration Review 72(5): 741-51.

Burden, Barry C., David T. Canon, Stephane Lavertu, Kenneth R. Mayer, and Donald P. Moynihan. 2013. Selection method, partisanship, and the administration of elections. American Politics Research 41(6): 903-936.

Berlinski, Nicolas, Margaret Doyle, Andrew M. Guess, Gabrielle Levy, Benjamin Lyons, Jacob M. Montgomery, Brendan Nyhan, and Jason Reifler. 2023. The effects of unsubstantiated claims of voter fraud on confidence in elections. Journal of Experimental Political Science 10(1): 34-49.

Burness, Alex. 2023. Oregon Wants to Register Medicaid Recipients to Vote. Will Biden Officials Allow It?

Cantoni, Enrico and Vincent Pons. 2019. Strict ID Laws Don’t Stop Voters: Evidence from a US Nationwide Panel, 2008–2016. Technical report National Bureau of Economic Research

Government Accountability Office. 2022. Retirement Security: Recent Efforts by Other Countries to Expand Plan Coverage and Facilitate Savings.

Grimmer, Justin and Ethan Hersh. 2023. How Election Rules Affects Who Wins.

Grimmer, Justin and Jonathan Rodden. 2022. Changing the Default: The Impact of Motor-Voter Reform in Colorado.

Grimmer, Justin and Jesse Yoder. The durable differential deterrent effects of strict photo identification laws. Political Science Research and Methods. 10(3): 453-469.

Dague, Laura and Benjamin Ukert. 2023. Pandemic-Era Changes to Medicaid Enrollment and Funding: Implications for Future Policy and Research. NBER Working Paper No. 31342

Herd, Pamela and Donald Moynihan. 2018. Administrative Burden: Policymaking by Other Means. Russell Sage Foundation.

Herd, Pamela, Hilary Hoynes, Jamila Michener and Donald Moynihan. 2023. Introduction: Administrative Burden as a Mechanism of Inequality in Policy Implementation. RSF: The Russell Sage Foundation Journal of the Social Sciences 9(4) 1-31

Levine, Sam and Andrew Witherspoon. 2023. Revealed: Florida Republicans target voter registration groups with thousands in fines. The Guardian.

Michener, Jamila. 2018. Medicaid, Federalism, and Unequal Politics. Cambridge University Press.

Olsen, Asmus Leth, Jonas Høgh Kyhse‐Andersen, and Donald Moynihan. 2022. The unequal distribution of opportunity: A national audit study of bureaucratic discrimination in primary school access. American Journal of Political Science 66(3): 587-603.

Rigby, Elizabeth. and Springer, Melanie. 2011. Does electoral reform increase (or decrease) political equality? Political Research Quarterly. 64(2): 420-434.

Stewart, Charles III. 2022. Trust in Elections. Daedalus. 151(4): 234–253.

Stewart, Charles III, Stephen Ansolabehere, and Nathaniel Persily. 2016. Revisiting Public Opinion on Voter Identification and Voter Fraud in an Era of Increasing Partisan Polarization. Stanford Law Review. 68(6): 1455–1490.

Street, Alex, Thomas A. Murray, John Blitzer, and Rajan S. Patel. 2015. Estimating voter registration deadline effects with web search data. Political Analysis 23(2): 225-241.

1. See Executive Order on Advancing Racial Equity and Support for Underserved Communities Through the Federal Government. and Executive Order on Transforming Federal Customer Experience and Service Delivery to Rebuild Trust in Government. The administration also changed guidance for the implementation of the Paperwork Reduction Act to encourage agencies to better identify learning and psychological costs, and to reduce administrative burdens, consistent with Office of Management and Budget guidance. See:
2. For evidence on the risk of fraud see
3. One example of efforts to capture these burdens is the Cost of Voting Index.
4. Grimmer and Rodden (2022. 7) note that some of those declining are not eligible to vote, for citizenship or other reasons, but conclude that “it appears that a majority of eligible unregistered Colorado citizens turned down the opportunity to register, in spite of the gentle prompt in favor of registration.”