-5-
concerning the origin of global shocks, evidence of both global supply-side and demand-
side disturbances is found.
However, our work differs from Bagliano and Morana (2009) in three major
aspects. First, we explicitly analyze the structural relationship between various global
variables in order to get a better understanding of the world economy. This part is
missing in Bagliano and Morana (2009). Instead, the authors concentrate on four global
factors which they label “inflation factor”, “output growth factor”, “stock return factor”
and “oil price factor”. For example, the inflation factor contains not only inflation rates
but also interest rates as well as monetary aggregates. In our approach, the relationships
between such common forces are explicitly disentangled via a structural model on a
global level. We use a total of seven global factors, including a “house price factor”. The
latter enables us to investigate to what extent other global variables contributed to the
strong rise in property prices until the start of the financial crisis.
Second, we examine spillover effects from global to national variables using
structural VARs for the G-7 countries plus the euro area. In contrast to Bagliano and
Morana (2009), we do not estimate one model with all national variables taken as
endogenous but implement separate VARs for each country or region. Hence, we
intentionally neglect feedback effects between countries. Bagliano and Morana identify
structural idiosyncratic shocks by imposing exclusion restrictions on their
contemporaneous impact on all national variables across countries, estimating a total of
43 parameters in each equation (including the four factors). Our main motivation to apply
an approach different from theirs is that modeling both global and all national variables in
one complete structural framework would have been too costly in terms of degrees of
freedom. Hence, the typical empirical trade-off between using a “sufficient” number of
structural factors on a global level and establishing structural links between national
variables for various countries emerges once again.3
3 We use seven factors with a lag length of 2, thereby leading to the estimate of fourteen coefficients on the
global level (plus a constant and a deterministic component). Our sample covers the period from Q1 1984
to Q4 2007. Bagliano and Morana base their specification on four factors with only one lag (sample Q1
1980 - Q2 2005).