Table 1: Forecast error variance decomposition of P, basic model
Period |
Y |
P |
M |
IS |
2 |
23.2 |
748 |
0.1 |
1.9 |
(8.7) |
(8.6) |
(0.9) |
(2.0) | |
4 |
31.4 |
59.2 |
0.3 |
9.1 |
(10.9) |
(11.1) |
(1.6) |
(5.6) | |
8 |
37.5 |
41.6 |
2.3 |
18.5 |
(13.5) |
(12.6) |
(3.9) |
(10.5) | |
16 |
50.2 |
23.3 |
9.1 |
17.4 |
(17.2) |
(11∙0) |
(8.3) |
(13.8) |
Cholesky Ordering: Y P IS M; Standard Errors in parentheses
puzzle” did not disappear.15 There will be further discussion of the ”price puzzle”
in the context of the following models, where the house price index helps us to solve
the ”price puzzle”.
The short-term interest rate moves up due to an output shock, but does not show
a significant reaction to a price or a money shock. These results may occur, because
either the system captures only the monetary policy stance in the short run which
could be dominated by the business cycle or because the monetary policy instrument
might be difficult to model from a global perspective where different central banks
with different strategies exist. The responses of money show, according to standard
money demand considerations, a positive response of money to an output innovation
and a decline of liquidity with growing interest rates. The latter effect might be
caused by rising opportunity costs of money holdings and/or due to central bank
driven shifts in the money supply.
Table 1 shows the forecast error variance decomposition of the GDP deflator.
Liquidity matters again in the long run, while most of the variance of the price level
is a result of fluctuations of the output variable. Notwithstanding the close long-run
relationship between money and prices, in the short run, business cycle fluctuations
seem to play the major role for price level volatility.
Overall, the results of the benchmark model provide a good starting point for the
15The same finding appears in Ruffer and Stracca (2006) as well as Sousa and Zaghini (2006) as
these authors used commodity prices as well but did not solve the ”price puzzle”.
18