In Chile's two-member legislative districts we show there are two groups of swing voters, one group for the first seat won by the governing coalition, another for the second. We build a model that allows us to identify the relative prevalence of these voters across communities. Using data on the allocation of discretionary agricultural loans, we find that communities with relatively many voters pivotal for the first seat receive more loans than they otherwise would have, but we find no systematic advantage for districts that are pivotal for the second seat.