Upon signing the 2009 American Recovery and Reinvestment Act (ARRA) into law, President Obama said, expect you, the American people, to hold us accountable for the results of the new legislation. (1) In political science, however, there is substantial debate about the extent to which voters reward or punish their elected officials, and in particular presidents, for their spending choices. Most notably, the voter responsiveness literature has not yet fully considered the effects of polarization. Polling data suggest that the response to the ARRA in general appears to have been filtered through partisanship. Pew Research Center polling data from 2010 and 2012 reveal that almost 80% of Republicans disapproved of stimulus spending, compared with roughly one-fifth of Democrats (Pew Research Center 2012). While it is unsurprising that attitudes toward the program writ large are a reflection of the partisan divide concerning President Obama, it may be that voters respond to actual stimulus spending in a similarly partisan way. Such a result would be consistent with research on the rising salience of partisanship in America's polarized politics (e.g., Abramowitz 2010; Fiorina, Abrams, and Pope 2004; McCarty et al. 2006). Among elected officials, in the media, and among the mass public, partisanship has become increasingly predictive of political behavior. To explore the effects of polarization on electoral rewards for federal spending, we take advantage of a sudden influx of government spending courtesy of the ARRA. In addition to its highly politicized nature, the ARRA represents an excellent case to explore this issue for several reasons. First, the particularly massive and widespread nature of the stimulus offers an unusual opportunity to explore political responsiveness to federal government spending in a wide variety of geographic locales, including heavily Democratic and Republican places. Moreover, the federal government's attempts to claim credit for specific stimulus programs via road signs, speeches, and other publicity render it a particularly good test of political responsiveness: if any federal government spending program were likely to generate a constituent response, it would be something as highly visible as the ARRA. Finally, there have been, to our knowledge, no political science studies that examine the democratic implications of the ARRA. That is, while much scholarship--particularly in economics--has been devoted to the effectiveness and economic effects of the program (Congressional Budget Office [CBO] 2012; Council of Economic Advisers, Executive Office of the President 2011; Feyrer and Sacerdote 2011), there has been no research on the political effects of the initiative. We simply do not have any rigorous empirical evidence on whether the American people did indeed hold the Obama administration or Congress accountable for the successes or failures of this signature piece of legislation. This article proceeds as follows. First, we examine the existing research about voter responsiveness to spending and show that this literature does not create a consistent picture of how we should expect voters to react to public spending. Then, using data on the allocation of the ARRA, we estimate the effects of spending on support for the president's party at the county level. We find that ARRA investments have a positive effect on Democratic vote share only in those counties that are already Democratic leaning; in strongly Republican counties, the effect of increased spending is negative. In total, then, the stimulus had a polarizing effect on the electorate, driving liberal counties to the left and conservative counties to the right. This finding suggests that highly politicized spending can actually be counterproductive for an incumbent Democrat. Rather than being unresponsive (Kriner and Reeves 2012), conservative counties punished Democrats for the spending they received from the stimulus. …