Local Projection Inference Is Simpler and More Robust Than You Think

Applied macroeconomists often compute confidence intervals for impulse responses using local projections, that is, direct linear regressions of future outcomes on current covariates. This paper proves projection inference robustly handles two issues commonly arise in applications: highly persistent data and the estimation at long horizons. We consider projections control lags variables regression. show lag?augmented with normal critical values are asymptotically valid uniformly over (i) both stationary non?stationary data, also (ii) a wide range response Moreover, lag augmentation obviates need to correct standard errors serial correlation regression residuals. Hence, is arguably simpler than previously thought more robust autoregressive inference, whose validity known depend sensitively persistence length horizon.