Michael Gilbert has posted this draft on SSRN (forthcoming, Columbia Law Review).  Here is the abstract:

Voter ID laws have provoked a fierce controversy in politics and public law. Supporters claim that such laws deter fraudulent votes and protect the integrity of American elections. Opponents, on the other hand, argue that such laws — like poll taxes and literacy tests before them — intentionally depress turnout by lawful voters. A vast literature, including legal scholarship and opinions of the Supreme Court, accept these two narratives. But these narratives are wrong, or at least incomplete. This paper presents a comprehensive analysis of the effects of voter ID laws, and it shows that they can have surprising effects, including this: they can exacerbate fraud. To illustrate, suppose that without a voter ID law candidates A and B would receive 13 and 10 lawful votes, respectively, and B would receive two fraudulent votes. Candidate A wins non-fraudulently, 13 to 12. Now suppose that with a voter ID law, candidates A and B would get nine and nine lawful votes, respectively (less than before because of depressed turnout), and B would get one fraudulent vote (less than before because of fraud deterrence). Candidate B wins fraudulently, 10-9. The conditions necessary for ID laws to have this effect are simple and may be common. The paper captures this risk with a formula, the Election Integrity Ratio, which judges and scholars could use to determine when ID laws represent good policy — and when they cause the very problem they aim to solve. The paper has important implications for constitutional law and public policy. It also has broad reach. Any law that deters fraudulent votes while also depressing lawful votes — citizenship and residency requirements, for example, which are used throughout the United States and around the world — are subject to the analysis herein.

Provocative and recommended!