Table 2: Basic model augmented with house prices
Forecast Error Variance Decomposition of HPI:
Period |
Y |
HPI |
_______P_______ |
M |
IS |
2 |
0.0 |
98.0 |
079 |
0.3 |
0.7 |
(1.9) |
(3.5) |
(2.6) |
(0.8) |
(1.2) | |
4 |
0.3 |
87.8 |
3.2 |
0.8 |
7.9 |
(3.0) |
(7.4) |
(4.5) |
(1.8) |
(5.0) | |
8 |
0.5 |
66.4 |
6.3 |
3.4 |
23.3 |
(4.6) |
(12.1) |
(6.3) |
(4.2) |
(10.1) | |
16 |
0.2 |
41.7 |
9.0 |
14.7 |
34.3 |
(6.5) |
(14.2) |
(7-5) |
(9.4) |
(12.8) | |
Forecast Error Variance Decomposition of P: | |||||
Period |
Y |
HPI |
P |
M |
IS |
2 |
18.5 |
1.6 |
7877 |
1.0 |
0.2 |
(8.4) |
(2.0) |
(8.6) |
(1.8) |
(1.2) | |
4 |
25.0 |
5.3 |
66.2 |
1.4 |
2.1 |
(10.6) |
(5.1) |
(11.0) |
(2.5) |
(3.0) | |
8 |
33.2 |
17.4 |
45.4 |
2.4 |
1.6 |
(12.8) |
(10.2) |
(12.6) |
(3.6) |
(3.6) | |
16 |
23.4 |
44.5 |
18.8 |
1.9 |
11.3 |
(13.4) |
(13.4) |
(8.9) |
(3.5) |
(7.6) |
Cholesky Ordering: Y P HPI IS M
Standard Errors in parentheses
from housing price inflation to consumer prices from an empirical angle. From a
theoretical point of view these findings underline the relevance of wealth effects and
the balance sheet channel, which probably contribute to these spill-overs.
The house price index in our model does not only solve the ”price puzzle”, it
is also involved in many significant impulse responses and is a major factor in the
forecast error variance decomposition of the price level. Therefore, the house price
variable is too crucial to be omitted in the following. Consequently, we will augment
our model with stock prices while still including the house price index.
We now add the log of the MSCI World index to our model to represent
global stock markets. The vector of variables under consideration is therefore (in a
Cholesky ordering):
xt = (y p hpi IS m msw)'t
Figure 5 shows a selection of the impulse responses representing the relationships
that are of primary interest. No evidence can be found that either interest rate
shocks or liquidity shocks fuel stock markets. Furthermore, no significant spill-overs
from share prices to inflation occur in our model. At least, there is a significant
response of money to a stock market impulse. This may be due to wealth effects
21