Salient Shocks: Estimation and Inference with Concentrated Identifying Variation (joint with Geert Mesters and Victor Sancibrian)
Martin Almuzara, Federal Reserve Bank of New York
Many empirical designs use long time series but draw a substantial fraction of their identifying variation from a few dates (often a single one). To study the implications of this, we introduce a novel asymptotic embedding in which total regressor variation grows while a fixed number of dates retain nonvanishing leverage. We characterize the large-sample properties of vector autoregression (VAR) and local projection (LP) estimators and inference procedure in that setup, and show that conventional approaches generally fail. We then propose valid estimation and inference strategies. Finally, we document that salient shocks are pervasive in economic applications.
| Building: | North Quad |
|---|---|
| Website: | |
| Event Type: | Workshop / Seminar |
| Tags: | Econometrics, Economics, Macroeconomics, seminar |
| Source: | Happening @ Michigan from Department of Economics, Econometrics, Michael Beauregard Seminar in Macroeconomics, Department of Economics Seminars |
