This paper by Óscar Jordà and Alan M. Taylor reviews the state-of-the-art in local projections (LPs) for estimating the effects of exogenous interventions or shocks on outcomes. LPs are a sequence of regressions where the outcome at different horizons is regressed on the intervention, conditional on a set of controls. The authors discuss the advantages of LPs over vector autoregressions (VARs), including their simplicity, versatility, and ability to handle nonlinearities and state dependence. They also explore the connection between LPs and VARs, showing that while they are asymptotically equivalent under certain conditions, LPs can have advantages in small samples and when the lag length is misspecified. The paper covers various topics such as instrumental variables, impulse response decompositions, and extensions to panel data and difference-in-differences settings. Additionally, it discusses the estimation of multipliers, which measure the average effect per intervention, and smoothing techniques to improve the visual appearance and precision of LP responses. The authors provide practical guidance and best practices for using LPs in empirical research.This paper by Óscar Jordà and Alan M. Taylor reviews the state-of-the-art in local projections (LPs) for estimating the effects of exogenous interventions or shocks on outcomes. LPs are a sequence of regressions where the outcome at different horizons is regressed on the intervention, conditional on a set of controls. The authors discuss the advantages of LPs over vector autoregressions (VARs), including their simplicity, versatility, and ability to handle nonlinearities and state dependence. They also explore the connection between LPs and VARs, showing that while they are asymptotically equivalent under certain conditions, LPs can have advantages in small samples and when the lag length is misspecified. The paper covers various topics such as instrumental variables, impulse response decompositions, and extensions to panel data and difference-in-differences settings. Additionally, it discusses the estimation of multipliers, which measure the average effect per intervention, and smoothing techniques to improve the visual appearance and precision of LP responses. The authors provide practical guidance and best practices for using LPs in empirical research.