The following is a guest post from  Eric McGhee, Jennifer Paluch, and Mindy Romero:

The pandemic forced a lot of states to consider expanding access to voting by mail (VBM) in a way they hadn’t before to lower the risk of viral transmission at in-person voting sites. But President Trump charged that the practice was rife with fraud and designed to hurt Republicans, and criticized it at every opportunity. The result was a profound polarization over VBM among elected officials and voters alike and a crisis of confidence in our elections.  In the wake of this election, state legislatures across the country have considered scaling back VBM or have already done it.

What are the effects of increasing VBM access? Before the 2020 pandemic election, research generally showed it increased turnout at least a little, and did not favor one party or the other. But the pandemic election might have been different for a whole host of reasons, not least because a wider range of states experimented with these reforms than ever before. Some approaches were even tried that had rarely been used. Were the results any different?

Jennifer Paluch, Mindy Romero, and I have a new working paper that explores just this question. We look at the effect on turnout and partisan outcomes from three VBM reforms—opening VBM to anyone who wants it (no-excuse VBM), mailing all voters a VBM application (VBM applications), and mailing every voter a ballot (universal VBM)—and extend the research to the pandemic election itself.  In the process, we account for the complex mix of factors at work in the election:  including COVID, competitiveness, and other election reforms.

As was true in the research before 2020, universal VBM increased turnout in the 2020 election about 4 percentage points, while the other reforms had smaller and more contingent effects (turnout was actually lower in states that adopted no-excuse VBM).  The reforms didn’t help Democrats, either—in fact, if anything we found a modest effect in favor of Republicans.

There are caveats, of course. States and counties adopting these reforms were shifting strongly Democratic before the reform, and we accounted for that in our analysis. So these “pro-Republican” effects were measured against a baseline of steadily increasing Democratic support. In fact, almost all the results were at least a little sensitive to the exact way we sliced and diced the data. The big exception is mailing every voter a ballot:  the turnout boost from this reform seemed consistent across a wide range of approaches.

On balance, then, we think the effect of mailing a ballot on turnout is clear, but otherwise these reforms do not produce substantial increases in turnout or Democratic support, and in 2020 there is some evidence of the opposite in each case. Policymakers should change the lens through which they view this issue:  the question is not who gets an advantage from reform, but whether one wants more people to participate or not.