SVAR identification with nowcasted macroeconomic data

Starting from the theoretical observation that the identification problem of SVAR models arises from contemporaneous dependence among macroeconomic variables, we show, both theoretically and empirically, that such dependence tends to vanish as the observation frequency increases. By adopting nowcasted high-frequency data, we exploit this feature to identify structural shocks using standard short-run restrictions, thereby reducing or even eliminating the reliance on variable ordering. Our empirical analysis is divided into two parts: an illustrative application comparing identification strategies across different frequencies, and a structural section featuring (i) a Proxy(HF-)SVAR to recover exogenous monetary policy shocks, and (ii) an uncertainty shock analysis using high-frequency data to replicate the well-known dynamics found in the literature. The results align with recent findings and highlight the feasibility and usefulness of preserving high-frequency information in all variables.