Essays in the Estimation and Use of Impulse Response Functions in Macroeconometrics

This thesis proposes practical solutions to address three challenges of local projections (LPs). First, it introduces subspace methods to address the issue of difficult estimation of impulse response functions (IRFs) by LP given many controls. Second, it shows that pooling vector autoregression (VAR) and LP IRFs with an equal weight across horizons may address the bias-variance trade-off in IRF estimation in the choice between VAR and LP. Finally, it studies the role of consumer sentiment in the transmission of monetary policy to consumer expenditure using counterfactual analysis.