POLITICS
Nonpartisan elections don’t reduce polarization
State-legislative edition
One big story coming out of 2020 has been the fusion of two movements: one for multiparty government, another for less of a gap between the major parties (polarization). The former promotes ballot reforms to ease third-party voting. The latter aims to get rid of partisan nominations. Its signature reform is the nonpartisan two-round electoral system, or NPTRS. Recent adoptions include Alaska, where NPTRS combines with instant-runoff voting, and St. Louis, where NPTRS comes with approval balloting. NPTRS itself has longer history.
Does it reduce polarization, as proponents claim it will?
My answer is “no,” based on the latest data from political scientists Boris Shor and Nolan McCarty. Their data capture state-legislative polarization, 1993–2018, using a procedure similar to NOMINATE.
I won’t wade into the voluminous literature on “top-two,” primary systems, etc., much of which finds no effect on polarization. I do want to acknowledge that Christian Grose has results consistent with a ‘moderating’ effect in Congress. Those results are interesting.
But if NPTRS have no effect, at least two constituencies should care. First are those investing in reforms — with dollars and/or sweat — to loosen political gridlock. Second are those who care about the long-run consequences of party-weakening reforms. Such reforms (e.g., direct primaries) tend to outlast the purposes for which they were enacted (e.g., throwing various bums out). It is possible that, over the long haul, they just give “special interests” an even greater role in politics.
What is NPTRS, and how might it reduce polarization?
Such systems do not permit a party to advance a single candidate (or slate) to the decisive round of an electoral cycle. Examples include “top-two” elections in California and Washington, Nebraska’s fully nonpartisan system (no party labels on ballots), and Louisiana’s so-called “jungle primary.”
NPTRS are distinct from ‘open primaries,’ which let voters choose which party’s nomination process in which to participate. NPTRS are also distinct from ‘blanket primaries,’ which let voters do the same on an office-by-office basis.
In other words, NPTRS creates a general election in which voters may be choosing among multiple candidates from the same party.
I have noted elsewhere that NPTRS were common in Western Europe of the late 1800s, just before the emergence of disciplined parties. From this perspective, NPTRS attempts to roll back the clock. The theory is that, by weakening parties as agents of nomination, legislators will be free to cross the aisle more often. In turn, we should see less polarization in the roll-call record.
An alternative theory is that NPTRS do nothing — at least with respect to aggregated measures based on legislative voting— because legislatures tend to generate stable coalitions.
Have NPTRS reduced polarization?
The Shor-McCarty data include two measures of polarization. One is the familiar distance between the median Democrat and median Republican in each state-legislative chamber. Another is the average of all pairwise differences in a given chamber.
The first measure can be thought of as “two-party polarization.” I call the second “coalition polarization,” as the coalition in control of a legislative chamber may not follow party lines. An example is the erstwhile deal in New York between Republicans and the Independent Democratic Caucus. “Coalition polarization” is not a perfect measure, but it does try to capture such deals.
Following the lead of Seth Masket and Boris Shor (2015), I plot the polarization trend for all 50 states. Four states are highlighted: California, Louisiana, Nebraska, Washington. Two of these adopted NPTRS during the period covered (see dots).
Focusing first on state upper chambers, we see no long-run moderating effect from NPTRS. No trend is downward-sloping. This holds for both measures.
Turning to state lower chambers, we again see no long-run moderating effect. (Nebraska is omitted from these plots because Shor and McCarty group its “Unicam” with state senates, above.)
Overall, there is no long-run effect of NPTRS on legislative polarization.
Some may point to brief reductions (e.g., CA Senate, two-party measure; WA House, both measures), but the “depolarization” does not last. Recall the alternative hypothesis: legislatures tend to generate stable coalitions.
Also, some states adopt NPTRS during ‘local lows’ for polarization (e.g., CA House, coalition measure; WA Senate, both measures). This is consistent with the experience of direct primaries —widely adopted during the de-polarization of the Progressive Era and New Deal, then followed (as we all know) by decades of re-polarization.
None of this is surprising. Legislatures tend to generate stable coalitions.
Yet few would deny that ‘things aren’t right’ in the wider party system. People probably disagree about exactly what’s not right. Let’s have that conversation. Let’s stop pretending that our problems — whatever they are — will go away by fighting a tenacious summary statistic. And let’s stop wasting energy trying to get parties out of politics.
