This thesis consists of three essays in Applied Macroeconomics. Each essay is contained in

one chapter and can be read as a standalone piece.

Chapter 1 proposes a new strategy for the identification of monetary policy shocks in Structural Vector Autoregressions (SVARs). It combines traditional sign restrictions with external constraints on high-frequency monetary surprises and Federal Reserve internal forecasts. I use it to assess the transmission of

US monetary policy over the period 1965-2007. First, I find that contractionary monetary

shocks unequivocally decrease output, sharpening the ambiguous implications of standard

sign-restricted SVARs. Second, I show that the identified monetary shocks and monetary

policy equations are consistent with a narrative reading of the times and Taylor-type rules.

Finally, I implement an algorithm for robust Bayesian inference in set-identified SVARs,

providing further evidence in support of the methodology I propose in this chapter.

Chapter 2 studies the relationship between monetary policy decisions taken by the Federal

Open Market Committee (FOMC) and higher moments of expected economic outcomes.

First, I employ quantile factor models to characterize the conditional distribution of central

bank economic projections and construct indicators of uncertainty and skewness. Second, I

find that the skewness of expected output growth and inflation rate is a crucial predictor of the

changes in the intended federal funds rate deliberated by the FOMC. This empirical evidence

is found to be reconcilable with central bank’s optimal behavior under non-linear weighting

of probability. My findings suggest that considering central moments only is not enough to

fully capture the systematic component of monetary policy and lead therefore to important

implications for the identification of monetary shocks. Specifically, I find that conditioning

on higher moments allows to identify monetary shocks exhibiting lower predictability and

that generate theoretically consistent effects on the economy.

In Chapter 3, we study SVAR models that impose internal and external restrictions to setidentify

the Forecast Error Variance Decomposition (FEVD). This object measures the

importance of shocks for macroeconomic fluctuations and is therefore of first-order interest

in business cycle analysis. First, we characterize the endpoints of the FEVD as the extreme

eigenvalues of a symmetric reduced-form matrix coming from quadratic programming. Second,

we use the perturbation theory to prove that the endpoints of the FEVD are differentiable

with respect to the reduced-form parameters. Third, we rely on inference for eigenvalues

to construct confidence intervals that are uniformly consistent in level and have asymptotic

Bayesian interpretation. We also describe the conditions to derive uniformly consistent confidence

intervals for impulse responses. AMonte-Carlo exercise demonstrates the properties of

our approach in finite samples, while an empirical application based on credit supply shocks

illustrates our toolkit. Finally, our machinery can be also used to quantify the sensitivity of

the standard Bayesian inference to the choice of an unrevisable prior.