A0201
Title: An order and scale-invariant identification scheme for structural dynamic models
Authors: Neville Francis - UNC Chapel Hill (United States) [presenting]
Abstract: The purpose is to introduce OASIS, an order- and scale-invariant identification scheme for structural dynamic models, which includes structural vector autoregressions (SVARs), local projections, and impulse-response function matching within DSGE frameworks. OASIS identifies the orthogonal rotation that maximizes the correlation between structural shocks and their corresponding reduced-form innovations. This objective is similar to that of a Cholesky decomposition, but the latter is constrained by the ordering of variables. Additional theoretical insights regarding Cholesky identification are also provided. To evaluate OASIS against Cholesky, a significant number of SVAR studies are revisited. The correlations between structural and reduced-form shocks are generally very large. Notably, it is found that OASIS effectively resolves the price puzzle observed in impulse responses to monetary policy shocks.
