-15-
5. Empirical results
5.1. Baseline FAVAR
The common factors are estimated by principal component analysis. As is common in
factor model applications, the variables are initially de-trended and standardized. Apart
from the short-term interest rates, all Xt are log first-differenced. For the short-term
interest rates only, first differences are calculated. Afterwards, all the de-trended
variables are standardized so that each of them has a mean of zero and a variance of one.
Otherwise, the results would have been systematically affected by cross-country
differences in variability.
For each global variable, the proportion of the total variance of the series
attributable to each principal component is calculated. For the first principal component
(PC1) to suitably qualify as a factor capturing international co-movement, one important
condition must be met. PC1 should explain a sufficiently large fraction of the total
variance of the relevant data set in comparison to the remaining principal components of
higher order. As can be seen in the table below, the requirements are met for all of the
common factors. For example, in the case of global money, a significant part of the total
variance (48.6%) can be attributed to the first principal component. In contrast, PC2
accounts for only 18%. Even for the inflation factor which includes four different
measures of inflation for each country or region (CPI, PPI, import prices and the GDP
deflator), the first principal component’s share is 31.8% compared to 16.9% for PC2.
With 55.9% and 74.3%, the PC1’s share for commodity and share prices is clearly the
highest.
Table 1 - Share of variance explained by first three principal components
PC1 |
PC2 |
PC3 | |
Real GDP___________ |
36.5 |
23.0 |
17.0 |
Inflation___________________ |
31.8 |
16.9 |
9.6 |
Commodity prices_____ |
55.9 |
27.8 |
7.9 |
House prices__________ |
33.2 |
26.9 |
15.5 |
Broad money________ |
48.6 |
18.0 |
15.0 |
3M interest rate_________ |
42.8 |
19.7 |
18.5 |
Share price____________ |
74.3 |
12.1 |
6.4 |
Note: Calculations based on first standardized differences
More intriguing information
1. Credit Market Competition and Capital Regulation2. SAEA EDITOR'S REPORT, FEBRUARY 1988
3. Naïve Bayes vs. Decision Trees vs. Neural Networks in the Classification of Training Web Pages
4. The name is absent
5. Economic Evaluation of Positron Emission Tomography (PET) in Non Small Cell Lung Cancer (NSCLC), CHERE Working Paper 2007/6
6. On s-additive robust representation of convex risk measures for unbounded financial positions in the presence of uncertainty about the market model
7. An Empirical Analysis of the Curvature Factor of the Term Structure of Interest Rates
8. THE INTERNATIONAL OUTLOOK FOR U.S. TOBACCO
9. Needing to be ‘in the know’: strategies of subordination used by 10-11 year old school boys
10. The name is absent