Cataloguing the Promises of Ranked Choice Voting in New York City
All that we thought it would be?
In June 2021, New York City’s party-registered voters used ranked-choice voting (RCV) in primary elections for the first time. Two years ago, NYC voters adopted RCV by an overwhelming margin after it was touted as a “faster and cheaper” alternative to the runoff system, that would eliminate the “spoiler effect” and lead to winners supported by majorities. After the NYC primaries, there are reasons to question the promises made by RCV advocates, as voters in Maine have continued to after adopting the reform.
It’s worth recounting RCVs recent history in the city. In 2019, Ballot Question 1 asked voters if they wanted to “establish ranked-choice voting to be used for primary and special elections beginning in 2021” as well as change the timing for vacancies to be filled and a change to petitioning requirements to place candidates on the ballot. The New York Times Editorial Board wrote in support of this question and included a short history of the reform and how it’s been used elsewhere.
It is not surprising that voters supported the reform. Most Americans lack any experience with alternative voting procedures for elected office, many Americans dislike parts of politics and will jump at the idea of a reform. and When the choice is something new with coordinated, advocacy group support and virtually no one arguing to keep standard simple plurality elections, voting for a change is understandable.
The supportive arguments put forward by the Times Board were:
- RCV saves time, money and effort by eliminating a potential second, runoff election for the mayoral election
- In the few places RCV had been adopted, everyone liked the change
- RCV could lead to higher turnout
- RCV leads to more positive campaigns and that makes people feel better about the process
- RCV leads to winners with majority support, which should be better than simple plurality elections.
- RCV can solve the “spoiler effect”
So, did the results from the first 21st century attempt of RCV in NYC live up to these arguments? With the recent release of Professor John Lovett’s The Politics of Herding Cats, this is a cat-themed analysis of the results from this specific election, on these specific claims. For those interested in other interesting takes on this election and the primacy of parties check out what Jack Santucci has to say on the matter or get into more of his longer form research.
There is no question that RCV can replace runoffs — which NYC previously had for Mayoral and other executive elections though not for the numerous city council primaries. RCV eliminates the need for a runoff as it is essentially a real time runoff, just with more than two candidates.
Whether RCV is cheaper than a runoff is a more complicated question. The last time we had a high profile runoff was in 2013 for the Public Advocate seat, and that cost the city $13 million dollars. RCV elections are heralded for encouraging more candidates to run, and as NYC also adopted more generous public funding for primary elections, the city still spent $100 million + in candidate matching funds for the debut of RCV.
An exit poll conducted after voters left the polls commissioned by RCV advocates Common Cause and Rank the Vote NYC, reported that over three-quarters (77%) of voters understood the reform and wanted to keep it. Though elsewhere, other recent survey research indicates that only a minority of voters (22%), mostly younger and Democratic partisans, prefer RCV to standard, single selection ballots. As it stands, surveys have indicated that voters in NYC generally liked RCV for the primaries. But for the original argument it’s simply not true that everyone who tries RCV likes it. In 2009, voters in Burlington, Vermont had a disastrous first go round with RCV and decided to go back to traditional elections, though there has since been movement by supporters to reinstate the reform.
Turnout was higher in 2021 with 25% of registered Democrats participating compared to the 15% participating in 2017. Yet it is difficult to assess the impact RCV has on turnout in this specific election because there were many differences between the 2021 NYC Democratic primary and the most recent, 2017 Democratic primary.
First, in 2021 New York City had a 10-day early voting window in addition to election day that was not in place in 2017. Second, in 2021 there was an open mayoral seat, multiple Borough President seats, and 46 out of 51 city council races had competitive elections due to term limits in 2021. In 2017 there were simply fewer open-seat races which are known to field fewer candidates, which in turn leads to lower levels of campaign activity and thus lower turnout overall. The 2021 primaries also operated with a new, generous public matching program, where candidates who raised specific sums of money were giving a multiplier match in public funds to conduct their campaigns. This influx of cash to more candidates led to greater overall campaign activity in 2021 than what was possible in 2017. And finally, the Board of Elections spent $15 million in voter education campaigns to make people aware of the change to RCV elections in the primaries — this across the board awareness of the election probably activated some voters.
This is all to say, though there was greater turnout in 2021 than in 2017, it’s hard to know how much to attribute to RCV alone.
Survey research on RCV elections in other cities finds general feelings of increased civility — at least in for the first few RCV elections. Research on the positivity of the first RCV elections in NYC is still in progress, but just from my experience, the NYC elections felt more positive than before.
However, these first effects will likely fade. After a while, candidates and political operatives realize that RCV campaigns in single-member winning districts are essentially the same as simple plurality elections. RCV elections take longer to tabulate, and allow voters to feel expressive, but at the end of the day, being on top is all that matters. Winning outright is easier to pull off by pointing out negative things about opponents rather than staying positive and hoping to game subsequent rounds by earning #2 spots from voters.
In the most recent Senate election in Maine — the state with the most experience with RCV — saw the most negative campaigns of the cycle. Whether or not more positive elections are good for informing voters, or if more positive campaigns will remain the norm in future primary elections are different questions, but on the claim that NYC would experience a kinder advertising environment, RCV seemed to deliver.
Majority Winners and the Elimination of Spoilers
The final two arguments, that RCV leads to majority winners and solves the so called, “spoiler effect” is testable using the newly released NYC Board of Elections Data. The specific claims made by the Editorial Board on these topics were,
In multicandidate races like this, the winning candidate often has less than majority support. The mayoral race is required to hold a runoff if no candidate breaks 40 percent of the vote, but no similar cutoff exists for City Council races. This can create a “spoiler effect,” where an unpopular candidate can win with, say, 25 percent of the vote, solely because his or her opponents split the rest. Ranked-choice voting solves this problem. [emphasis added]
Did RCV solve these “problems” in NYC? Not really.
In the New York City Democratic Primaries, more often than not, winners won with less than majority support, like what happens elsewhere with RCV. RCV in NYC or anywhere cannot not guarantee electing a candidate with majority support, the system merely selects a candidate with the majority in the final round. Given that voters are limited to five ranks, and that RCV elections tend to attract more candidates (at least in the first few subsequent elections held under RCV) plenty of races will be won by candidates without the support of a majority of voters casting ballots in an election.
Below is a table with the total votes cast, number of candidates, number of RCV rounds, and the final number and share of votes for the eventual winner for city and borough wide positions. Elections with only 1 round have the same outcome as that from standard, plurality election. Every candidate who would have won in round 1 under the standard plurality system, ended up winning with the RCV system, and though their shares of support are higher in the final round due to a smaller denominator and the transference of votes, only 1 race that went beyond round 1– the Staten Island Borough President — was a majority supported winner.
In the City Council, for most contests things were similar. On average, winning candidates did not earn the support of a majority of voters in their districts. Only 13 of 46 city council elections were decided in the first round with clear majorities — ranging from 51–79% — favoring one candidate. For the others most winners did not earn majorities of the ballots cast in their races.
Astute readers will note that it is more likely to have a non-majority winner when there are more candidates, and that’s true, but that’s also one of the reasons to question what the adoption of RCV means. RCV proponents will say one of the benefits is that people won’t feel the need to “wait their turn” to run, thus more people will enter elections. In 2021 the average number of candidates in each race was between 6-7, whereas in 2017 it was 3-4. There are solid reasons for and against more people running — but a clear tension is that elections with more candidates give rise to more opportunities for winners who lack a clear majority.
To be clear, no single-stage, single-winner electoral system in use today in the US can guarantee a winner with majority support. In some elections, where a candidate exists that has clear, and outright majority support, he or she will win, but the procedural style for tallying an election does not make this so. But the story that RCV can accomplish this feat is pervasive, as are other notions about the benefits of RCV that deserve more academic and empirical scrutiny before other cities race to consider changing their own elections.
The final argument used in favor of RCV takes on the oft-cited — though rare in reality — problem of “spoiler” candidates, who change the outcome of an election away from what a majority wanted by inadvertently siphoning votes away from an outcome that a majority preferred.
Like the New York Times Editorial Board, RCV interest groups assert that RCV will result in spoiler-free majorities and that multiple candidates will feel freer to jump in the ring without the risk of splitting constituencies. But does RCV eliminate election outcomes changed through spoilers? No, but it does make the task of identifying an individual “spoiler” more difficult, and RCV can lead to more elections with “spoiled” outcomes in some circumstances.
The classic spoiler story involves a Green candidate taking votes from a liberal, resulting in a conservative victory. This outcome is “bad” because the so-called “spoiler” candidate gave a win to the conservative despite the fact that there are more liberal and Green voters combined than conservative voters and that liberal and Green voters are closer to each other than they are to conservatives. The same story could be made with the ideologies reversed — two more conservative candidates losing to a liberal candidate despite the electorate being generally more conservative.
Not only does the spoiler effect lead to these “bad” election results in a specific race, there is also a corresponding effect believed to deter future candidates from entering future races because they fear splitting similar voters. This secondary effect, encourages like-minded candidates to coalesce around one candidate either through not entering a race, dropping out of a race as the election comes closer, and/or encouraging voters to support the one, strongest candidate to beat an opponent with views further away from the like-minded candidates. But this future effect is discouraged in the advocacy stories for RCV and this sort of spoiled outcome is potentially more likely under RCV, but more difficult to pin on just one candidate.
In a standard, single vote election a “spoiler” is an individual candidate that receives a sufficient number of votes that, if those votes had all gone to the second-place candidate, could have changed the result of the election. In a RCV election a spoiler can be a candidate or a group of candidates that receives a sufficient number of votes that, if those votes had all gone to the second-place candidate, could have changed the result of the election.
In standard, single vote elections it is relatively easy to discern if a race had a “spoiler” by simply transferred his/her vote total to the second place candidate and seeing if that total tops that of the winner. In RCV elections when there are greater numbers of candidates to start with, and when voters can rank multiple candidates — but fewer than the whole list of candidates — and when candidates are told there is no need to drop out and coalesce, because they need not fear being spoilers there will be people who are unwitting spoilers. At the very least, there will still be the potential for vote splitting, and at the very worst, RCV can actually create greater incentives for like minded candidates to accidentally spoil races for each other.
Here’s a way to think about this using the City Council races.
Thirteen of the 46 races had outright majority winners in round one. The remaining races are all sequentially narrowed until there are just two candidates competing head-to-head. In that final, two candidate match up the difference in votes between the winner and the second-place (loser) is the same margin that can be compared to the difference between winner and loser totals in standard elections. If the number of ballots that were cast on round 1 but were later “exhausted” or eliminated from consideration in the final round exceeds the difference in the vote totals between the RCV winner and second place (loser) it is possible that a spoiler candidate or candidates exist. If these candidate(s) who if otherwise would not have been in a standard plurality race — either through deterrence or the through a pressure to drop out and coalesce — and their supporters actually shared more in common politically with the runner up than with the winner, they are acting as spoilers.
In 19 of the 32 multiple-round RCV city council races the number of ballots exhausted exceeds the difference between the first (winning) and second place (losing) candidate, sometimes by a lot. Assessing the political closeness or likeness of losing candidates to the winner and the second-place loser is difficult, but the potential for spoilers still exists even if it is harder to point out.
Some successes, misses, and downsides
RCV is seductive partially because the electoral systems in place in most US elections — single winner, single member districts decided by a plurality vote — is a somewhat frustrating way to elect people to governing bodies. Any change might seem better than what we currently have. But if the justifications used to support RCV fall short, voters may have swapped the relatively simple, though troubling standard style of voting for a somewhat more complex style of voting that can reproduce similar sorts of frustrating outcomes, as well as new troublesome downsides.
When considering how to change the processes to get more representative and consensus candidates, the choice is not only to allow for ranking or pick one approaches. A potential promising style to further reduce spoiling or non-majority outcomes is known as “approval voting” where voters mark their approval for any number of the candidates that they approve of to hold a specific office — the candidate with the most approval votes wins. Another option is STAR voting which is sort of a mash up and RCV and approval, asking voters to give a ranking to candidates and then working through an instant runoff style of tabulation. All systems have pros and cons, and as different municipalities and states consider changes, it’s important to continue to evaluate empirical outcomes against supportive promises.
NYC is the largest place to adopt RCV, and there is much to learn from this first election. For some of the arguments supporting the change, the reform worked, for others, it’s less clear, and there are still under-appreciated downsides from the new system. As an earnest crop of researchers pour over the recently released ballot data, we will learn more. As political consultants and strategists refine their suggestions, campaigns will adapt. After all, each system has strengths and weaknesses and the more people are concerned in understanding these, the better.