That’s the title of a study that’s been accepted for publication in The Quarterly Journal of Economics. The approved manuscript is paywalled except for the brief abstract, but you can read an earlier “working paper” version here.
The neat thing about the study is that, instead of relying on overall voting rates or survey data, it has administrative data on individuals’ registration and turnout. This is put together by Catalist, a business that maintains enormous amounts of voter data to help progressive campaigns. The study includes 1.6 billion observations of voters’ behavior between 2008 and 2018, along with demographic information about the voters.
And the upshot is that if voter ID does anything, it’s not a big enough effect to measure reliably, even with such a huge, detailed sample. Turnout in general doesn’t decline when these laws are in effect. Minority turnout in particular doesn’t decline. Fraud doesn’t decline, or at least detected cases of fraud don’t. Perceptions of fraud don’t change either. One thing that does happen is that “the likelihood that nonwhite voters were contacted by a campaign increases by 4.7 percentage points, suggesting that parties’ mobilization might have offset modest effects of the laws on the participation of ethnic minorities.”
The whole issue is kind of a nothingburger. Conservatives may overhype fraud — and a losing presidential candidate I can think of may have gone off the deep end over it. But it’s not unreasonable to ask people to show IDs to vote, and this requirement does not stop people from actually voting.