There are lots of explanations out there, see work by @Evaanduiza @CatherineDVries @TaylorBoas @MSWinters1 @mgoldenProf @ainagallego @gvisconti among others. But we’ve usually tested individual explanations in isolation.

In a new paper in @JEPS_ed with Marko Klašnja and @j_a_tucker , we test some factors together to see which is more important.

We put survey experiments on @Lapop_Barometro in Argentina, Chile, and Uruguay. We tested whether voters are more willing to support a corrupt incumbent when (1) corruption is already widespread and (2) when corruption brings side benefits (in our case, construction jobs).

In general, of course, people prefer clean candidates. But they’re willing to tolerate corruption when it’s not all bad – and especially when it benefits them.

We find that corruption being widespread does not make corruption more palatable to voters, but side benefits do. And those who stand to gain from these side benefits are less likely to sanction corruption.

@NoamLupu Great evidence! In Brazil we have this saying about corrupt politicians "rouba, mas faz" (steals, but does). I think it speaks to the main conclusion of your paper. If the politician is helping the population (creating jobs), the voters are less likely to sanction corruption.

@maricarvalb Absolutely! See also work on this in Brazil by Rebecca Weitz-Shapiro and @MSWinters1

@NoamLupu Congrats Noam! Looks super cool. Straight to the lit review section of the book ms we are writing on Lava Jato.