A1382
Title: Counterfactual variance analysis with local projections
Authors: Jooyoung Cha - University of Notre Dame (United States) [presenting]
Abstract: Counterfactual changes in unconditional volatility across regimes are studied. The counterfactual logic combining the shock environment of one regime with the propagation mechanism of another follows earlier counterfactual work in autoregressive and VAR-based settings. The contribution here is to recast that exercise in a local projections framework. A counterfactual long-run variance is defined, where propagation is represented by the regime-specific sequence of impulse responses. Estimating this propagation sequence with local projections and combining it with regime-specific estimates of shock variances in a plug-in estimator of the counterfactual variance is then proposed. Under standard structural assumptions, point identification of the population counterfactual is shown and conditions under which the plug-in estimator should be consistent and asymptotically normal are outlined. A Shapley-style averaging provides a possible additive attribution of volatility changes to shocks and propagation. The goal is to provide a flexible alternative to parametric VAR-based counterfactual analysis that avoids imposing cross-horizon restrictions on the impulse-response path.
