New Evidence on the Puzzles. Results from Agnostic Identification on Monetary Policy and Exchange Rates.

The identifying restrictions we shall impose to identify an impulse vector
characterizing monetary policy shocks are that (r_a(k))j ≥ 0, j ∈ J₊ and
(r_a(k))j ≤ 0, j ∈ J_- for some subsets of variables J₊ and J_- and some
horizon k = 0, . . . , K.

We use a Bayesian prior for the reduced form VAR parameters (B, Σ) and
an independent uniform prior for α. The uniform prior for α assures that
the implied prior for a is independent of the specific decomposition AA' = Σ
of Σ and can even be random, as long as the choice of the decomposition is
independent of α.

A Bayesian VAR with 6 lags in levels of the logs of the series has been
fitted to the data except for using interest rates directly. No constant or
time trend are included. The choice of 6 lags follows the choices made in
the literature¹ . The prior and therefore the posterior belong to the Normal-
Wishart family, see Uhlig (1994) for a detailed discussion of the properties.
Results are obtained by taking draws from the posterior for the VAR coef-
ficients B and draws from the space of possible impulse vectors. Inference
statements are based on those joint draws that satisfy the sign restrictions
for the impulse responses. We use 500 draws satisfying the restrictions for
drawing posterior inferences. We typically show the median as well as the
16% and 84% quantiles of the distribution for the points on the impulse
response functions.

For further methodological details, see Uhlig (2005).

3.2 Identification of Monetary Policy Shocks in Open
Economies

The first choice to be made is the selection of variables. To assure compa-
rability and similar to Faust and Rogers (2003), we shall use the specifica-

¹In fact, much of the evidence becomes considerably weaker, when using 12 lags instead,
which we did in a previous version of this paper.

12

<<   12   13   14   15   16   17   18   >>

More intriguing information