2. Strengths
• Interes7ng
topic.
• Methodologically
sound.
• A
novel
approach.
Using
TVP-‐VAR,
along
with
Bayesian
methods.
3. Sugges7on/comments
• Posterior
can
be
quite
sensi7ve
to
priors.
o Karagedikli
et
al
(2010),
“RBCs
and
DSGEs:
the
computa7onal
approach
to
business
cycle
theory
and
evidence",
Journal
of
Economic
Surveys.
• Should
under
take
sensi7vity
analysis
to
see
if
posterior
is
prior
driven
or
likelihood
driven.
4. Sugges7on/comments
• Bayesian
techniques
adapted
to
deal
with
model
uncertainty.
• There
usually
is
considerable
model
uncertainty
with
these
models.
– Variables
to
choose
(unit
labour
costs
etc,
integrated
health
of
world
economy
and/or
financial
markets
maer);
– Cointegra7on
rank;
– Lag
length;
and
– Structural
breaks.
5. Sugges7on/comments
• Could
construct
a
reasonable
model
space
and
use
Bayesian
Method
Averaging
to
see
how
robust
your
results
are.
o Garra,
Koop,
Mise
and
Vahey
(2009)
“Real-‐
7me
predic7on
with
UK
monetary
aggregates
in
the
presence
of
model
uncertainty",
Journal
of
Business
and
Economic
Sta7s7cs,
27(4),
480-‐491
7. Sugges7on/comments
• Some
interes7ng
results.
Which
need
to
be
explained.
• For
example
:
• “Posi7ve
spikes
in
machinery
investment
deliver
short-‐
and
long-‐term
Tankan
falls.
This
indicates,
according
to
the
assump7ons
made
here
about
the
strong
link
between
Tankan
business-‐confidence
and
uncertainty,
that
increasing
the
fixed
produc7ve
capacity
of
firms
on
a
macro
scale
increases
macro-‐level
uncertainty.
This
inverse
rela7onship
between
investment
and
confidence
requires
further
study.”
8. In
conclusion
• A
very
novel
approach
to
looking
at
fixed
investment,
uncertainty
and
financial
markets.
•
Sound
methodology.
• May
be
the
sugges7ons
given
will
help.