Gimme a Break! Identification and Estimation of the Macroeconomic Effects of Monetary Policy Shocks in the U.S.

Abstract

We employ a novel identiÖcation scheme to quantify the macroeconomic e §ects of monetary policy shocks in the United States. The identiÖcation of the shocks is achieved by exploiting the instabilities in the contemporaneous coe¢cients of the structural VAR (SVAR) and in the covariance matrix of the reduced-form residuals. Di §erent volatility regimes can be associated with di §erent transmission mechanisms of the identiÖed structural shocks. We formally test and reject the stability of our impulse responses estimated with post-WWII U.S. data by working with a break in macroeconomic volatilities occurred in the mid-1980s. We show that the impulse responses obtained with our non-recursive identiÖca-tion scheme are quite similar to those conditional on a standard Cholesky-SVARs estimated with pre-1984 data. In contrast, recursive vs. non-recursive identiÖca-tion schemes return substantially di §erent macroeconomic reactions conditional on Great Moderation data, in particular as for ináation and a long-term interest rate. Using our non-recursive SVARs as auxiliary models to estimate a small-scale new-Keynesian model of the business cycle with an impulse response function matching approach, we show that the instabilities in the estimated VAR impulse responses are informative as for the calibration of some key-structural parameters.

Identification in Structural Vector Autoregressive Models with Structural Changes, with an Application to US Monetary Policy

“…Thus, under the null of a valid DSGE model, triangular SVARs offer a misspecified representation of monetary policy shocks and their propagation, and can produce price puzzles and muted responses of inflation and the output gap to monetary shocks, see Castelnuovo and Surico (), Castelnuovo () and Bacchiocchi et al . (). In other words, given , the C matrix must be full with highly restricted non‐zero coefficients for the SVAR to be consistent with the predictions of a DSGE model.…”

“…Second, the system might omit, as it stands, important transmission mechanisms, see Bacchiocchi et al . ().…”

“…However, in equation the instantaneous impact of the shock on the variable varies from in the first volatility regime to in the second volatility regime. The specification in equation is interesting because it can be to some extent related to the debate about the consistency between SVAR analysis and dynamic stochastic general equilibrium (DSGE) modelling, see Bacchiocchi, Castelnuovo and Fanelli (). As is known, small‐scale new‐Keynesian DSGE models of the type discussed, among many others, in for example Lubik and Schorfheide () and Carlstrom, Fuerst and Paustian (), typically admit an immediate reaction of output and inflation to monetary policy impulses, while ‘conventional’ triangular (Cholesky‐based) SVARs feature a lag in such reactions.…”

Structural Vector Autoregressions: Checking Identifying Long‐run Restrictions via Heteroskedasticity

“…In practice, the latter condition may not be satisfied and it is important to note that heteroskedasticity can also be helpful for identification if other parameters are also varying. That case is considered, for example, by Bacchiocchi and Fanelli (2012) and Bacchiocchi et al (2014).…”

Uncertainty across volatility regimes

“…This topic has been explicitly analyzed in Ludvigson, Ma, and Ng (2018a) and Carriero, Clark, and Marcellino (2018b), reporting mixed evidence. The identification strategy we apply extends the standard "identification-through-heteroskedasticity" approach, popularized in the empirical macroeconomic literature by Rigobon (2003), Rigobon and Sack (2003), and Lanne and Lütkepohl (2008), to the case where the structural parameters (on-impact coefficients), and hence the associated impulse response functions (IRFs), may vary across volatility regimes (see Bacchiocchi, Castelnuovo, & Fanelli, 2018;Bacchiocchi & Fanelli, 2015). This concern has been addressed in the recent literature, and evidence that uncertainty shocks have time-varying effects has been provided by, among others, Alessandri and Mumtaz (2018), Caggiano, Castelnuovo, and Groshenny (2014), and Caggiano, Castelnuovo, and Pellegrino (2017).…”