We investigate the political power of credit rating agencies by building a theoretical model that illustrates how heterogeneous voters change their political preferences after receiving credit signals which infer the quality of their governments. We empirically test this hypothesis using a rich dataset of daily sovereign ratings, outlook and watch signals assigned by S&P, Moody’s and Fitch to EU countries from 2000 to 2017, along with a unique dataset measuring public support for governments. We find that negative rating signals lead to a significant decrease in government support, therefore influencing the electoral prospects of political parties. Both sociotropic and egocentric voters’ preferences are affected by sovereign ratings. Our results are confirmed across a battery of robustness tests and various modelling approaches, including fixed effects and difference in differences models and propensity score matching. Our findings offer wide-ranging implications for policy makers, political parties, governments, and the rating industry.