Measuring monetary policy with residual sign restrictions at known shock dates

We propose a novel identification strategy to measure monetary policy in a structural VAR. It is based exclusively on known past policy shocks, which are uncovered from high-frequency data, and does not rely on any theoretical a-priori restrictions. Our empirical analysis for the euro area reveals that interest rate decisions of the ECB surprised financial markets at least fifteen times since 1999. This information is used to restrict the sign and magnitude of the structural residuals of the policy rule equation at these shock dates accordingly. In spite of its utmost agnostic nature, this approach achieves strong identification, suggesting that unexpected ECB decisions have an immediate impact on the short-term money market rate, the narrow money stock, commodity prices, consumer prices and the euro-dollar exchange rate, and that real output responds gradually. Our close to assumption-free approach obtains as an outcome what traditional sign restrictions on impulse responses impose as an assumption.