The Importance of Global Shocks for National Policymakers: Rising Challenges for Central Banks

-33-

5.4. Robustness checks

We apply three robustness checks in order to test whether our results hold in general
independent of variations in the specification. One important robustness check is the
“cross-sectional” stability of the global forces. In dynamic factor analysis, it is usually
assumed that the cross-sectional dimension of the dataset is large, even possibly larger
than the temporal dimension. Hence, the exclusion of one country or variable should not
influence the results markedly. However, this may not be true in our approach, since we
focus on a rather limited set of countries (G-7 countries plus euro zone) often with only
one variable per country. For example, for broad monetary liquidity, a total of five figures
are used (M2 in the US and Canada, M2CD in Japan, M3 in the EMU and M4 in the UK)
to estimate the common monetary liquidity force. The same is true for other variables,
like house and share prices or GDP. It is therefore conceivable that the exclusion of one
variable may significantly alter the results received from principal component analysis.

In order to assess how vulnerable our results are, we estimate each global variable
excluding one country. Afterwards, the co-movement between the global force including
all five countries and the global force based on four countries is estimated. If the
correlation coefficient does not change markedly, it is concluded that the global variable
is symmetric, i.e. it is not overly influenced by one country. Our results indicate that this
is the case, as the co-movement is comparatively high. Interestingly, this is even true
when the US is excluded from the dataset. This may point in the direction that the global
economy may be less asymmetrically influenced by the US than usually thought.
However, it is important to stress that a simple correlation analysis cannot distinguish
between shocks and propagation.

Table 7 - Co-movement between global force and global force ex respective country

Note 1: Correlation coefficients

Note 2: *** Indicates significance at 1% level, ** at 5% level, * at 10% level

<<   <   33   34   35   36   37   38   39   >   >>