April 19, 2011, 8:29 p.m.
The study of election fraud is largely undeveloped in the academic community. Although there are some exceptions, such as Walter Mebane's use of Bendford's Law and random digit forensics, most research has been case-specific without broad implications.
One type of election fraud, which often goes undiscussed, is when electors casting ballots in constituencies they don't reside in. Kentaro Fukumoto and Yusaku Horiuchi address this issue, in a paper, which explores the impact of registration requirements on electoral fraud. The authors use Japan as a case study, where weak registration laws automatically change where a citizen is registered to vote after they submit a change of address form to the local municipal office. In other words, any Japanese citizen can easily change their voter registration address by submitting a form without evidence of their actual address. By examining the 2003 local elections, the authors argue that there is considerable evidence of voters changing their registration status before the election, only to switch back right after. Because only select municipalities vote in a particular year, the authors claim they can use local areas where elections are held as a control group, contrasting it with those that don't. They notice that changes in registration occur at a statistically significant level, more often in municipalities where elections are being held. They further argue that the events leading up to why certain regions held elections on specific years was far enough removed from present day to be considered exogenous.
We argue that whether or not a municipal election was held in April 2003 can be regarded as an “as-if” randomly assigned treatment. The difference-in-difference analysis of municipality-month panel data shows that an increase in the new population just prior to April 2003 is significantly larger in treatment municipalities (with an election) than in control ones (without an election). The estimated effects are, in some cases, decisive enough to change the electoral results, especially when the election is competitive. We argue that our approach – “election timing as treatment” – can be applied to investigate not only this type of electoral fraud but also other “electoral connection[s]” (Mayhew 1974) in other countries.
It is also unclear why the authors discuss Japan's weak registration laws as the reason for the false-residency fraud. While the narrative they discuss seems sounds, their actual methodology has nothing to do with establishing the causal effect of registration laws and voter fraud. Instead, their research focuses on providing evidence that fraud exists, and by doing so promoting a new method for post-election fraud forensics.
It is difficult to tell how useful the authors' findings are for practitioners in developing democracies; few nations would even have the capacity to develop an electronic registration system that makes Japan's fraud so easy to accomplish. Furthermore, detecting fraud through staggered municipal elections would, by the authors own admission, not work in every environment. The paper does, however, provide useful evidence that local elections with small district magnitude increase incentives to engage in false-residency fraud in the lead up to an election. It is also admirable that the authors have contributed to the small, but hopefully nascent, field of electoral fraud studies.