The electoral benefits of unemployment, clientelism and distributive politics.

• We test competing hypotheses on distributive politics (swing, core, opposition) with Argentine data from 2001 to 2003. • Exploiting the political and economic crises, we use the allocation of a poverty relief program, to show the political bias. • Opposition strongholds received a disproportionally larger share, controlling for the number of potential recipients. • The clientelist purpose of spending is demonstrated. First, counties with "better" monitoring obtained more resources. • Second, counties with greater spending experienced greater turnout and greater vote shares for the incumbent. We show that in the month prior to the 2003 Argentine presidential election, the expenditures of an Argentine poverty relief program exhibit a partisan bias. Taking into consideration the number of potential recipients (the unemployed with children 18 years old or less), the counties that were ideologically against the incumbent received a disproportionately large amount of the spending. Hence, instead of targeting swing or core districts, as established in the distributive politics literature, in Argentina the incumbent targeted opposition strongholds. To identify these effects and to classify districts into swing, core and opposition ones, we exploit the 2001 crises in Argentina. In particular, we use the 2001 political discontent that led to 15% of blank and null votes to estimate swing districts. Finally, we also show that this tactical spending had an electoral purpose. In counties where it was arguably cheaper to monitor the recipients' voting behavior, the spending was significantly larger. In return, the clientelist exchange took place in districts with greater spending and monitoring. Thus, the politicians' incentives to design an appropriate poverty relief program may be at odds with their electoral motivation. [ABSTRACT FROM AUTHOR]