How to Accommodate a Massive Surge in Absentee Voting

In this fall’s election, we are almost certainly going to witness a dramatic increase in absentee or mail-in voting1 throughout the country. In Wisconsin’s recent high-profile April 7 primary, absentee voting skyrocketed by more than 650 percent compared to normal times. Kentuckians cast 80 percent of their votes by absentee ballot in their June 23, 2020 primary.  In Georgia’s  primary on June 9, 2020, about half of all voters cast their votes by absentee ballot, compared to 5 to 7 percent normally.2 Election officials in Pennsylvania recently described absentee ballot requests there as “off the charts.” 

To be sure, by the fall, people might be more comfortable going to work, traveling, eating out, and engaging in other activities under conditions of social distancing and other safety measures. They might be more comfortable voting in person then than now. Many states will likely expand their early-voting hours and number of sites, which would take further pressure off absentee voting. Perhaps effective treatments for coronavirus will even emerge by then, which will make some people more willing to vote in person. But even so, election officials must anticipate and prepare now for unprecedented levels of absentee voting that will put an enormous strain on running the election. That is so whether expanded use of absentee voting comes about through congressional legislation (unlikely), state legislation or the administrative actions of state election officials, court-ordered expanded access to absentee voting, or simply through voters taking advantage of long-standing options to vote absentee.

From the perspective of the more than 10,000 county, local government, and state entities that run our elections, accommodating and coordinating all these changes requires a buildup of administrative capacity that must take place virtually overnight. Every element in this process will be under unprecedented pressure: the systems for requesting, delivering, and counting absentee ballots; the processes for verifying that absentee ballots are valid; the supply chains for producing the ballots; the process for enabling voters to cure any defects in their absentee ballots; the need to find new polling sites and new poll workers. In addition to all these changes—and to make this new election infrastructure work as well as possible—states and Congress also need to turn their attention to changing certain key dates in their voting calendars. Those calendars, mostly fixed by statute, were not designed with this anticipated new flood of absentee ballots in mind.

Changing the key dates I identify below is important for at least three reasons. First, changing key dates enables voters to have sufficient practical opportunity to vote under these exceptional circumstances. Second, these crushing new administrative burdens mean breakdowns in the process will almost inevitably arise, and changing key dates gives officials and voters time to respond to these problems and correct them.  If tens of thousands of ballots do not reach voters in time, or similar problems emerge, the legitimacy of a close election might well be thrown into turmoil. Third, with massive increases in ballots mailed in, we are also likely to see unprecedented numbers of ballots that arrive after Election Day. As I have said, these late-arriving ballots open up one of the greatest risks of what might, in our era of hyperpolarized political parties and existential politics, destabilize the election result. If the apparent winner the morning after the election ends up losing due to late-arriving ballots, charges of a rigged election could explode. To preempt this risk, we want to encourage absentee ballots to arrive earlier rather than later; we want states to be ready to process these ballots as close to election night as possible; and we want to build in enough time on the back end to deal properly with the likely volume of these ballots.

In focusing on this expected surge in absentee ballots, we can think of the election process in three interconnected stages. Think of the first and third stages as the sides of a vise, each putting pressure on the critical second stage.3 The first involves the initial stages of requesting, filling out, and returning the ballot. In the second stage, state officials process these ballots, verify them, give absentee voters time to cure procedural defects in their ballot envelopes, count the votes, and resolve any disputes. Finally, in the last stage, the electoral college formally votes and Congress counts those votes. This fall, the second stage is likely to come under exceptional pressure. Those pressures will be all the more intensified if we expand the first stage by permitting absentees to be postmarked and received later than usual (as courts are already beginning to order).4 Unless Congress relaxes the dates for the last stage, the crucial middle process—when the votes are actually counted properly—will be compressed even more.

To relieve this pressure, policymakers need to consider a variety of changes to the election calendar. Some of these changes in dates should not be controversial; others might be. Some dates should be moved up; some should be moved back. There is also some tension between these three policy objectives, which is why there might be disagreement about some of these suggested changes.

1. Move back the dates for completing the canvassing and certification of the vote.

The most important calendar change is for states to move back the dates on which they are required to complete the all-important stages of canvassing and certifying the result. As a rough benchmark, a date of December 1st for certifying the winner of the election would give states three weeks to complete counting the vote accurately and dealing with potential problems. Some states permit even more time than that, but states that permit less should move their deadlines back.

On election night, jurisdictions release a preliminary tally of the ballots that have been counted up to that point. Many voters believe these are “the results.”  But the results are not official and no one has won the election until two post-election-day stages of the process are completed. The first is the official canvass of the vote; the second is the certification of the winner.

The canvass stage is designed to ensure that every valid vote, and no other, is included in the official result. This means counting and confirming all the various forms of ballots, such as absentee, early voting, Election Day votes, provisional ballots, and ballots from overseas and uniformed citizens. Among other tasks, the canvass enables election officials to take any actions needed to ensure the vote count is complete and accurate and, in some states, for any necessary recounts to take place (in other states, recounts take place only post-certification). Once the canvass is complete and numbers from across the state are aggregated, election officials then certify the results. In the presidential contest, the electors in the electoral college then vote for the person certified as the winner in their state.5

The dates set in current state laws for the canvass and certification process were not designed to deal with the crush of absentee ballots that will arrive this year at the last minute, such as on Election Day or in days after (provisional ballots, cast by voters who show up at the polls but whose eligibility is uncertain, will likely also rise this fall and can only be processed after Election Day). Georgia and Florida, as two examples, require that election officials certify the vote no later than 14 days after the election. That might well not be enough time this fall. Election officials should not be in the position of struggling to process and resolve any conflicts over this surge of absentee ballots under the pressure of deadlines too tight to provide time to do this critical job properly.

In addition, we know that legitimate voters—particularly those voting by absentee ballot for the first time—make inadvertent mistakes in using these ballots. When voting in person, voters have a quick opportunity with many of today’s voting technologies to correct certain inadvertent mistakes. With optical-scan machines, for example, those who make a mistake (such as by voting twice in the same race) will have their ballots kicked back by the machine and the voter is then able to correct the mistake and cast a valid vote. But absentee ballots are more complex to cast validly in the first place, and more complicated to cure if they are procedurally defective. A recent study of Georgia’s 2018 elections found that nearly 3 percent of absentee ballots were rejected, either because the voter made a mistake on the return envelope—such as failing to sign the oath on the outside of the ballot (some states also require a notary or witness signature); or because their signature is deemed not to match their signature in their voter registration file; or because the ballot was cast or received too late.

To have a 3 percent rejection rate when only 5 percent of the total votes are cast by absentee ballot is one thing. It’s quite another to reject 3 percent of absentee ballots when 70 percent of the vote is cast absentee. Moreover, we know that first-time users of absentee ballots commit more of these mistakes than those who regularly vote absentee—and the vast majority of those voting absentee this fall will be doing so for the first time. Indeed, in New Jersey’s recent special elections, a remarkable 10 percent of absentee ballots were rejected, precisely because there were so many first-time absentee voters. In a close election, an absentee-ballot rejection rate of 3 percent, 10 percent, or even higher, when 70 percent of the votes are absentee votes, could unleash charges that the election is being stolen.

Particularly because the number of these first-time absentee voters is so high, voters should be given notice if they have made a mistake on their ballot envelope and an opportunity to cure that defect. This is not just a matter of fairness to the individual voter; as just noted, the structural integrity and legitimacy of the election will be challenged if a large enough number of absentee ballots from eligible voters that could have affected the outcome are rejected.

Some states already provide notice and an opportunity to cure absentee ballots. Florida, for example, provides notice and permits voters to cure defects until 5:00 P.M. on Election Day; Nevada provides notice and gives voters until seven days after the election to resolve the issue. But some states currently provide no notice before rejecting an absentee ballot as defective. Given the likely volume of first-time absentee voters this fall, voters should have a reasonable opportunity to be notified of, and to cure, procedural defects in their ballots. Time to deal with these foreseeable problems should be built into this fall’s calendar.

States that do not change their canvass deadlines now and then find themselves facing this problem will be in a bad bind, particularly in a close election. Trying to change these canvass deadlines after the election could raise constitutional problems. In addition, partisan forces will be unleashed, based on calculations about who is likely to benefit, which may make it impossible to make those changes or, if courts order those changes, have them widely accepted as legitimate. Yet if forced to comply with those current deadlines, states might have to ignore large numbers of absentee votes from eligible voters.

To get out in front of these problems before they arise, states with canvass and certification deadlines unduly short for the likely context this fall should change them now.

2. Move up the deadline for requesting absentee ballots.

Moving those canvass and certification dates back addresses on the back end of the process the exceptional time pressures likely to arise this fall. But we should try to reduce them on the front end as well.

One place to start is by moving a bit earlier, in some states, the date by which a voter must mail in their request for an absentee ballot. This suggestion might be resisted by some. But moving this date a few days earlier would have significant benefits for administering the election smoothly without imposing a significant burden on voters. This reform targets states that permit these requests to come in much later than most states.

States already vary greatly in when the window closes for requesting an absentee ballot. Around ten states, including Arizona, Nevada, Florida, Rhode Island, and Missouri, close the window for mailed-in absentee ballot requests a full ten days or more before Election Day. Many states, instead, close their window four to five days before the election. But other states, including ones with large populations, permit requests to be made later than this. Ohio, for example, permits requests to come in as late as noon on the Saturday before the election (Ohio election administrators have already requested that the legislature move this date up and legislation appears likely to be enacted). And around 10 states (two of which, Minnesota and New Hampshire, might be swing states this fall) permit mailed-in requests to come in as late as the day before the election. These dates were not designed to accommodate the quantity of absentee balloting that will take place this year.

For election administrators to be able to process these requests properly this year, it is reasonable to close this absentee-ballot request window a bit earlier in those states that currently close their windows just a few days before the election. In a system facing a crushing number of these requests, it would be wise to build in more time than normal for election officials and the U.S. Postal Service to handle these requests and the return of these ballots.  Wisconsin, for example, currently permits requests to come in as late as the Friday before the election; in the recent primary there, more than 60,000 requests came in that day. This volume, for just one primary in one state, overwhelmed the U.S. Postal Service, and several thousand requested absentee ballots were not delivered as a result. In the recent Georgia primary, the later in the day an absentee ballot was requested, the less likely it was to be returned successfully. 

In our new circumstances, those states that permit absentee requests up to the last minute should move that date earlier to accommodate this reality. As noted above, many states already require these requests to be made no later than a week to five days before the election.

Some will perhaps express concern that bringing the “late-closing states” into line with most states would deprive some voters of the opportunity to participate. Those concerns may arise particularly if this change takes on a partisan valence within legislatures.

But while the burden to voters of this modest change is small, the benefits to pulling off a smooth election would be significant. For one, we already have plenty of advance knowledge that in-person voting may be complicated this fall; voters will not be faced with making last-minute adjustments, as in Wisconsin’s April primary. Voters concerned about in-person voting will thus have long lead times to apply for an absentee ballot. For another, the political parties and other voter-mobilization groups do a good job of educating voters of these dates and organizing their participation. The parties and these groups just need a clear date that is fixed well in advance for when ballot requests must be made so they can build their campaigns around that date. And even if some voters who want to vote absentee do not get around to requesting a ballot on time, they will still be able to vote in person early or on Election Day. The small burden of moving up a bit the date by which absentees must be requested is a price worth paying to reduce the problem of ballots not being delivered or returned on time. If tens of thousands of these ballots are invalid and could have made the difference, the risk that the losing side will conclude the election illegitimate is high.

Moving the date up a bit for absentee ballot requests would reduce this risk on the front end.

3. Move back the date by which absentee ballots must be returned.

Even if states move up the date by which absentee ballots must be requested, many more ballots than usual will nonetheless arrive after Election Day. Even under “optimal conditions,” for example, the Wisconsin Election Commission recently estimated that the turnaround time is at least five to seven days from the day an absentee ballot is requested, received, and returned. But this is under ideal circumstances. Given the dramatic increase in volume of such ballots this fall, the U.S. Postal Service is likely to take longer than usual to get these ballots to voters and get them back to election officials. Most importantly, this massive increase in volume will not be experienced in only one state. All states will be experiencing this simultaneously. The volume the U.S. Postal Service will have to contend with could be extraordinary.

States currently also vary greatly in the deadlines they impose for receipt of a mailed-in ballot to be a valid vote. Some require receipt by the time polls close on election night. Yet other states permit them to be received up to ten days after the election. Wisconsin is one of the states that normally require receipt by 8:00 P.M. on election night. For the April 7 election, which included the presidential primary, the federal district court held that, under the conditions of the pandemic, the U.S. Constitution required Wisconsin to accept absentee ballots received on or before April 13.  That court-ordered extension was not challenged in the case that went to the Supreme Court.

States that require absentees to be received by election night or shortly after should move this date back. Even if this fall the same percentage of absentee ballots as in normal elections would be rejected for coming in too late, the same point noted above holds true: a 3 percent rejection rate risks undermining the perceived legitimacy of the election if 70 percent of the vote is cast by absentee ballot.  And this problem would be compounded, of course, if mailing back ballots five days before the election is normally sufficient to get them back in time, but not this year. The overall burden on the U.S. Postal Service makes that five-day figure less realistic this time around. Moreover, if a significant number of votes come in after a receipt deadline that has not been changed and that is much tighter than in other states, ex post litigation challenging that deadline is easy to imagine. This is exactly what we do not want to face for a risk that can be mitigated in advance.

In Wisconsin’s election, the federal court pushed the date back six days. But that was for a presidential primary. In the general election, participation rates will be much higher. In choosing an updated receipt deadline that anticipates a dramatic rise in mailed-in ballots, policymakers face a trade-off. The longer the permitted time, the more ballots will be valid. But the longer that time, the longer it will take for the final result to be known. If we thought voters would be patient, that would not pose any risk. But in our climate of existential politics—with partisans all too prepared to believe or charge that elections are being manipulated, and a social-media environment poised to heap fuel onto the fire—the longer after Election Day any significant changes in vote totals take place, the greater the risk that the losing side will cry that the election has been stolen.

Election administrators in different states must weigh in on whether, in their circumstances, a six-day deadline post-election is appropriate, as the federal district court held for Wisconsin. The National Vote at Home Institute, one of the leading advocacy organizations for absentee and mail-in voting, suggests the deadline should be three business days after the election, which seems unduly short under our new circumstances. But state legislatures and election officials need to start facing this issue soon.

4. Move up the date on which returned absentee ballots can be processed and prepared for counting.

Among the biggest risks of a breakdown in the perceived legitimacy of the election arises from the potentially large number of mailed-in ballots not counted until after election night. If every day after the election, the race starts moving away from the candidate ahead on election night (and in the morning news the day after), and eventually shifts the outcome to the other candidate, it is no great leap to expect that charges of fraud, rigged elections, and the like will emerge. One easy way to diminish this risk is for state law to permit election officials to begin processing absentee ballots before Election Day, including taking all steps to enable those ballots to  be machine-read (but not actually counted), so that they can be counted as soon as polls close. This is important for another reason: voters should be given an adequate opportunity to cure any inadvertent defects, such as failing to sign the ballot envelope. The earlier the ballots are processed, the more time there is for voters to do so.

Some states permit this already. California, for example, begins processing absentee ballots up to 29 days before Election Day. Georgia, just recently, changed its rules so that absentee ballots for its upcoming primary can start being processed 8 days before Election Day for, as one election official said, “this once-in-a-lifetime unprecedented emergency.” 

Yet several of the fall’s potential swing states have still not made this change—even though election administrators are pleading with their state legislatures to do so. When Pennsylvania’s legislature recently delayed its April 28 primary, it did change the time at which mail-in votes could start being counted, but only from 8:00 P.M. election night to 7:00 A.M. that morning. But election administrators, understandably say this 13-hour shift is not nearly enough to accommodate the huge surge in expected absentees. As one election official there said recently, “[i]n terms of November, if they don’t let us start canvassing sooner than the day of the election, there’s no way anyone can responsibly call Pennsylvania on election night.” In Florida, a letter from the Supervisor of Elections, representing the election supervisors in all 67 counties, requested the Governor to make this change (among others). In Michigan, the Secretary of State has pressed the legislature to make this change, to no avail so far. North Carolina’s State Board of Elections requested this change; the legislature is currently considering it.

As of now, we thus face the possibility that Michigan, Pennsylvania, and Florida (at the very least), will end up with millions of ballots that cannot be prepared for counting until after the polls close. This almost certainly will lead to long delays before we know who won those states. Empowering election officials to begin preparing ballots for counting before Election Day would diminish this delay. This change should not be controversial. Failing to make this change in these or other states, however, could plunge us into partisan turmoil by unnecessarily forcing large numbers of absentee ballots to be counted well after Election Day.

5. Congress and the Electoral Count Act.

The last stages in the presidential election process are the casting of votes by the electors and the counting of those votes in Congress. To relieve the vise-like pressure on the state vote-counting process and enable it to be completed properly under the unique circumstances this fall, Congress could move back two key dates in federal law.

The framework statute governing the meeting of the Electoral College and the counting of the electoral votes is the Electoral Count Act, passed in 1887. The Act provides that Congress must count the electoral votes on January 6, 2021. That date should not be changed; January 6 is the first date the newly-elected Congress meets and the President must, according to the Constitution, be inaugurated on January 20. But the two other key dates in the Act, which might have made sense in the 19th century, can easily be moved back today; there is no contemporary policy reason these dates need to be fixed where they currently are. Pushing them back would not only provide breathing room for states to complete the vote count properly under the exceptional burdens this fall, but also for potential legal challenges.

The first is the date the Electoral College formally votes. By law, that date is currently December 14. But there is then a gap of more than three weeks until Congress receives and counts those votes on January 6. That gap might have been necessary with 19th century modes of transportation and communication. But there is no need for it now. Congress could easily push this date back several weeks. The electors could vote as late as January 3, the same day the new Congress convenes (the Act currently requires the certificates of election to be transmitted by registered mail, but that could be changed to permit those votes to be transmitted electronically). If Congress did not want to move the date back that far, it could move it back two weeks. Again, there is no reason not to move this date back.

Moving back the date the electors formally vote then enables Congress to move back the other critical date in the Act, the so-called safe-harbor date, currently December 8 (this safe-harbor deadline played a critical role in the Supreme Court’s Bush v. Gore decision). Moving this date back is key to relieving the vise-like pressure states will potentially experience in properly processing and counting the anticipated flood of absentee ballots. Under the Act, if states certify the winner of the election by this safe-harbor date (technically, if they appoint a slate of electors) then Congress will be bound by that determination. This means Congress will not challenge the validity of those electors if they have been appointed by December 8.

As the country learned in Bush v. Gore, this date puts states under tremendous pressure to complete their processes by then. But this date, too, can easily be moved back without compromising any policy concerns. If Congress moved back the date the electors vote by two weeks or so, it could move this safe-harbor date back by the same amount. It is possible, of course, that states will be able to complete their processes properly even without moving back these dates, particularly if they make the other adjustments I noted above. But to deal with the foreseeable and unforeseeable problems that could arise from transforming our election process almost overnight for this fall, pushing this date back, even just for this year’s election, would be good policy.

Even though these minor date changes to the Electoral Count Act should not be controversial, Congress might be reluctant to open up the Act at all. After all, this is the Act that plays a critical role in forestalling or resolving a potentially disputed presidential election; the stakes are high. Even so, Congress would be doing the country a service if it held hearings and addressed the Act, at least for these two minor date issues.6 Given the sensitivity of anything involving the Electoral Count Act, and Congress’s general propensity not to act before absolutely necessary, the prospects for Congress changing these dates in the Act are perhaps not promising. But moving these dates back would give election officials more time to manage successfully, and with less controversy, the extraordinary burdens they will likely face this fall.

* * *

State governments and local election officials must make myriad policy, administrative, and logistical changes before the fall election to minimize the extent to which voting in a time of pandemic disrupts the perceived fairness and legitimacy of the election. And at the same time they prepare for a surge in absentee ballots, they must ensure robust options for in-person voting are available as well.  The suggested changes in the election calendar flagged here are not the only changes that should be considered, but they are among the most important. In the midst of all the other changes legislatures and election officials must bring online almost overnight, they should not lose sight of the need to adjust certain key dates on the election calendar to build in an appropriate amount of time for an election likely to see never-before-confronted levels of absentee voting.

• 1Throughout, I will use “absentee voting” and “mail-in voting” interchangeably.  But in more technically precise terms, absentee voting typically means that a voter must request an absentee ballot, which can either be returned by mail or dropped off in person (at designated drop-off sites or at the polls).  “Vote by mail” is often used to refer to voting procedures in certain states, in which all voting is done by mail. Vote-by-mail states typically mail out absentee ballot request applications to all registered voters and often lack in-person voting options.  Most voting by mail this fall will be in the form of no-excuse absentee voting.  Professor Paul Gronke has offered an effort to clear up terminological confusion on these and related issues.
• 2At the time of this Essay’s publication, the website for Georgia’s Secretary of State lists the current absentee and in-person vote totals office by office. Examining the most competitive races, such as the Democratic primary for Senate, indicates that about half the votes (and perhaps more) were cast absentee.
• 3I am indebted to Ned Foley for this metaphor.
• 4See Richard H. Pildes, The Constitutional Emergency Powers of Federal Courts (forthcoming ___ 2020).
• 5At least, this is what has happened virtually all the time from early in our history. In two cases pending before the Supreme Court, the Court is being asked whether the Constitution requires that the electors be able to vote for whomever they prefer, regardless of how their state votes. If the Court accepts this view, we might expect the rate of “faithless electors” to increase.
• 6The Act is also notoriously ambiguous on other major issues—such as what happens if two slates of electors are purportedly chosen by the safe-harbor date and the two houses of Congress disagree about which to accept. Clearing up these ambiguities before our next disputed election would be wise.