Rational partisan theory's exclusive focus on electoral uncertainty ignores the importance of policy uncertainty for the economy. I develop a theory of policy risk to account for this uncertainty. Using an innovative measure of electoral probabilities based on Iowa Electronic Markets futures data for the United States from 1988 to 2000, I test both theories. As predicted by rational partisan theory, positive changes in the probability that the Left wins the Presidency or the Congress lead to increases in nominal interest rates, implying that expectations of inflation have increased. As predicted by the policy risk theory, positive changes in the electoral probability of incumbent governments and divided governments lead to significant declines in interest rates, implying that expectations of inflation risk have decreased. And as an extension to both theories, I find that electoral margins matter for the economy—partisan and policy risk effects depend not only on which party controls the government, but how large its margin of victory is.