Proxy-SVAR as a Bridge for Identification with Higher Frequency Data

High frequency identification around key events has recently solved many puzzles in empirical macroeconomics. This paper proposes a novel methodology, the Bridge Proxy-SVAR, to identify structural shocks in Vector Autoregressions (VARs) by exploiting high frequency information in a more general framework. Our methodology comprises three steps: (I) identify the structural shocks of interest in high frequency systems; (II) aggregate the series of high frequency shocks at a lower frequency employing the correct filter; (III) use the aggregated series of shocks as a proxy for the corresponding structural shock in lower frequency VARs. Both analytically and through simulations, we show that our methodology significantly improves the identification of VARs, recovering the true impact effect. In a first empirical application on US data, we show that financial shocks identified at daily frequency produce unambiguously macroeconomic effects consistent with a demand shock. In a second application, we identify U.S. monetary policy shocks that are highly correlated with the series of monetary policy surprises but, contrary to the latter ones, are invertible and so valid external instruments for low-frequency VARs.