In this paper a new instrument for monetary policy shocks is presented. Exogenous variation of the policy rate may come from frictions of collective decision-making. Dissenting votes indicate how far the final decision of the decision making body is from the mean of the members' individually preferred interest rates and thus correlate with the policy shocks caused by the decision-making frictions. Measures of dissent are used as external instrument in a structural VAR. Results for the U.S. show significant effect of the Fed's interest rate policy on real variables with the expected sign. On the other hand, the estimated effect on nominal variables is reminiscent of the price puzzle. Usual remedies, such as inclusion of commodity prices, inflation expectations or starting the sample in the middle of the eighties do not change the qualitative results casting doubt on the usual interpretation that the price puzzle is a statistical artifact.