Cost of Voting in the American States: 2020

The ease of voting across the United States is constantly changing. This research updates work which established the relative ‘‘cost of voting’’ during presidential election cycles, in each of the 50 states, from 1996 to 2016. The 2020 iteration takes into account the recent adoption of automatic voter registration processes, expansion of early voting, new absentee voting laws, and the elimination of polling stations in some states. We learn that Oregon, which has one of the most progressive automatic voter registration processes and mail-in voting, maintains the first position as the easiest state in which to vote. Texas falls to 50th, in part because it does not keep pace with reforms like online voter registration and no excuse absentee voting, which have taken place in most other states. Voters in both Michigan and Virginia will find voting more hassle free in 2020 because of changes to both voter registration and bal-loting processes that have occurred since 2016.


INTRODUCTION
T he decentralized nature of election laws in the United States, along with the dynamic nature of these laws, creates considerable confusion for the American voter. If a citizen moves from one state to another it will certainly take time to get used to the new voting practices they will encounter (Burden and Videl 2016). Indeed, voting laws across the 50 states are so dissimilar that we have been motivated to develop a measure that can capture the unique electoral climate of each state in presidential election years. We believe these snapshots of electoral conditions can be used to help elaborate state voter turnout rates, and also a number of political outcomes and quality of life considerations.
Earlier work finds that states vary considerably in their level of restrictions or what is termed the ''cost of voting'' (Li, Pomante, and Schraufnagel 2018). This update measures the relative restrictiveness of each state's electoral environment in mid-June of 2020. 1 There has been considerable movement in the relative cost of voting since data was last collected in 2016, with many states going out of their way to make voting less costly or hassle free. Yet other states have maintained the status quo and failed to innovate when technological advancements afford the opportunity to make things easier for the voting public. Furthermore, a few states have increased the cost of voting, principally by reducing the number of polling stations but also by adopting new restrictions on absentee voting. In order to keep pace with the many election law changes over the past four years it is necessary to update the Cost of Voting Index (COVI) to reflect the new reality in each of the 50 states.
Both stages of the voting process in the United States-registering to vote and casting a ballotare combined into a single index value. We pay special attention to developing a state rank, which particularizes the relative costs associated with voting in each of the 50 states. Importantly, the state ranking and the change in state rank allow for easy cross-state comparisons of electoral climate. Movement, over time, in the qualified cost of voting makes it clear that not all state legislatures value citizen participation alike. Our results suggest there remains considerable disparity in the relative cost of voting in 2020.

CALCULATING THE 2020 COVI
The 2020 indicator is developed using the same principal component analysis (PCA) that was drawn on to create indices from 1996 through 2016. Because state laws have been significantly modified since 2016, it is necessary to expand the number of issue areas or components used in the PCA from seven to nine. Most specifically, we have included a new four-item Likert scale that captures the complexion of new state laws surrounding the issue of ''automatic'' voter registration. Also, the number of early voting days has become a new stand-alone ratio variable because this consideration has taken on greater significance in the past four years and there is much more variability between states to capture than was the case in 2016. All nine issue areas used for the construction of the 2020 index are displayed in Table 1. The new considerations are issue areas 5 and 9, respectively. All newly added considerations are presented in italic font.
In the table, the first five issue areas deal with the act of registering to vote. Issue areas six through nine deal with the costs associated with casting a ballot. Regarding the registration issue areas, we have added two additional considerations to the Voter Registration Restrictions issue area. In the past we considered any felon restriction the same, but now are counting this consideration in two unique ways. The first is whether felons can register while incarcerated. The second, more costly, consideration denotes states that do not allow felons to register even after their incarceration is complete. The other new registration restriction designates whether a state has an online voter registration deadline (in days before the general election) greater than the median number of days. 2 The third issue area, Registration Drive Restrictions, combines state certification and training requirements into a single variable because in 2020 there is no longer any variance between these two considerations across the 50 states. Maintaining a four-point additive scale, we insert a consideration that taps the two states (New Hampshire and Wyoming) that have written laws that effectively ban registration drives. These two states are scored ''4'' on the index under the assumption that a ban is more limiting than a combination of three different laws that make registration drives more restrictive.
The last change to the registration half of the costliness puzzle reflect new laws passed in 20 states that provide nuanced versions of automatic voter registration. Importantly, state laws in only two states (Alaska and Oregon) have actual automatic registration of the type that is seen in many democracies around the world. These two states go out of their way to find and keep citizens registered when they change addresses. Other ''automatic'' voter registration laws combine a visit to a state Department of Motor Vehicles (DMV) licensing facility, or the Secretary of State office, with automatic voter registration unless you opt out. Still other states register you to vote if you visit additional state-run agencies besides the DMV. The Likert scale ranges from ''0,'' for Alaska, Oregon, and North Dakota, 3 to ''3'' for the 30 states that 2 The nine states that still do not have online voter registration are scored ''1,'' suggesting they have greater than the median number of days. Notably, the improved measurement strategy regarding felon registration has been adopted for earlier versions of the COVI, and index values from 1996 to 2016 have been revised accordingly and are available from the authors.
have not yet tried to make voter registration more automatic (see Table 1 for more detail).
Regarding the second stage of voting in the United States, actually casting a ballot, we have added four new considerations to the Voting Inconvenience scale. Particularly, we append a new consideration to the additive scale that indicates states that have reduced the number of polling locations since 2012 by 50% or more in some part of the state. Arizona, Georgia, and Texas are scored ''1'' but we also score Oregon and Washington ''1'' on this consideration because, although they have vote by mail, neither state provides an opportunity for in-person voting in the way the other three states (Colorado, Hawaii, and Utah) that have institutionalized a vote-by-mail process do. 4 A second consideration captures new state laws that somehow restrict absentee voting, limiting it to people of a certain age or not allowing first time voters to No voting centers 7 -Voter ID laws (5-item Likert scale) 0 = no ID required to cast a ballot, only signature 1 = non-photo ID required not strictly enforced 2 = photo ID required not strictly enforced 3 = non-photo ID required strictly enforced 4 = photo ID required strictly enforced 8 -Poll hours (ratio-level) Minimum and maximum poll hours (averaged and reversed) 9-Early voting (ratio-level) Number of early voting days (reversed) 4 We do not provide any accommodation for states that will allow mail-in voting in 2020 because of the COVID-19 pandemic. Because our intention is to develop a measure of electoral climate that is a reflection of more institutionalized practices, we only count states that have made a longer-term commitment to mail-in voting. However, the adoption of temporary vote-by-mail policies, due to the COVID-19 pandemic, is included in the COVID Cost of Voting Index discussed in Supplementary Appendix A1.

COST OF VOTING IN THE AMERICAN STATES 505
vote absentee. The third consideration measures whether the state recognizes Election Day as a state holiday (nine states in 2020) and the fourth consideration taps states that have created voting centers (17 states in 2020), which allow voters to cast a ballot somewhere other than their local precinct. The new Voting Inconvenience scale now has 12 items (see Table 1) and the effective range is ''0'' for Hawaii to ''11'' for Mississippi and South Carolina. The last change is the inclusion of a new issue area (number 9), a ratio-level variable which measures the number of early voting days in each state. We take care to account for weekends and count the actual number of early voting days available. The highest value, 46 days, belongs to Minnesota. Because the COVI is constructed in a manner that higher numbers represent greater cost, the values of this variable are reversed. The number of days a state allows is subtracted from the maximum of 46 days. The nine states that do not allow for early voting each receive a score of 46. The five states that have institutionalized vote-by-mail processes are scored ''0'' under the
assumption that voting by mail precludes the need to vote early. Importantly, the value of all our variables is available to the public for replication purposes. A detailed codebook provides the sources of our data. For a refresher on the principal component analysis used to create index values we ask the reader to view the article by Li, Pomante, and Schraufnagel (2018, 235-37) and the online Supplementary Appendix to that article made available by the publisher. 5

FIG. 2.
Change in state rank from 2016 to 2020. Note: Negative values indicate a decrease in state rank or a relative increase in the cost of voting.

5
The 2020 index was constructed using the same weighted linear combination of four principal components derived from the principal component analysis (PCA) of the indicators. The total variance explained by the four components is 73 percent. For more details on the methods employed see Li, Pomante, and Schraufnagel (2018, Supplementary Appendix) THE 2020 COVI Figure 1 presents the new values. Note that Oregon repeats as the state where it is easiest to vote. Mississippi had been ranked 50th in 2016, and now moves up marginally to 47th place. Texas takes over the 50th spot, representing the state with the most restrictive electoral climate. In particular, Texas has an in-person voter registration deadline 30 days prior to Election Day, has reduced the number of polling stations in some parts of the state by more than 50 percent, and has the most restrictive pre-registration law in the country. States that make voting less costly, like Oregon and Washington, have mail-in voting but also automatic voter registration processes. Other considerations that states with lower COVI values have in common include same day voter registration at the polling station, a significant number of early voting days and laws that make voting more convenient such as paid time off from work to vote. Figure 2 displays the change in state rank from 2016 to 2020. This is done to see which states in the past four years have made voting easier relative to other states. Note the dramatic change in Virginia. The state, in 2016, had laws that made it the 49th most difficult state to exercise the franchise. Changes passed early in 2020 move Virginia 37 places and it is now the 12th easiest state to vote in. Specifically, Virginia passed an automatic voter registration law, got rid of the in-person registration deadline, and made Election Day a national holiday, among several other considerations. Presumably, partisan change in majority control of the state legislature from the Republican Party to the Democratic Party in the 2019 off-year elections facilitated the aggressive move to make voting less costly.
But it is important to note that the relative cost of voting is not necessarily a partisan matter. Michigan voters in 2018, using a ballot initiative, changed the state constitution eight different ways to make voting easier. Among the changes were an ''automatic'' voter registration process, same day voter registration, and a no-excuse absentee voting provision. These changes, garnered 67 percent voter support and had, obvious broad bipartisan appeal. The changes have caused the state to move from the 45th easiest state to vote in to the 13th easiest, a move of 32 positions. Note, also, that the cost of voting does not always associate with Democratic or Republican Party influence over state legislative processes. More Democratic leaning states such as New Mexico and Delaware do not make voting particularly straightforward and Republican leaning states like Utah, North Dakota, and Idaho have less costly voting on average.

CONCLUSION
What is abundantly clear from the examples of Virginia and Michigan is that if a state wishes to make voting more accessible it is entirely possible to do so. Particularly interesting is the fact that some of the reforms, such as online voter registration, are found to come with a reduced monetary cost for states (PEW 2011). Moreover, researchers have identified cost savings associated with voteby-mail processes (Hamilton 1988) and have found that automatic voter registration reduces election administration costs (Gerken 2013). So, it is not the case that state legislatures cannot afford these reforms. What else is clear is that if a state does nothing to change their laws the relative cost of voting, compared to other states, will increase. When the relative cost of voting increased in 2020 it was often the case that state legislatures simply failed to keep pace with changes in other states. 6 Reforms take advantage of new technologies to make processes more efficient and less costly.