12.07.2015 Views

"Frontmatter". In: Analysis of Financial Time Series

"Frontmatter". In: Analysis of Financial Time Series

"Frontmatter". In: Analysis of Financial Time Series

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

346 VECTOR TIME SERIESTable 8.9. Factor <strong>Analysis</strong> <strong>of</strong> the Monthly Log Stock Returns <strong>of</strong> IBM,Hewlett-Packard, <strong>In</strong>tel, Merrill Lynch, and Morgan Stanley Dean Witter.The Returns <strong>In</strong>clude Dividends and Are from January 1990 to December1999. The <strong>Analysis</strong> Is Based on the Sample Cross-Correlation Matrix andAssumes Two Common Factors.(a) The principal component methodEstimates <strong>of</strong>Rotatedfactor loadings factor loadings CommunalitiesVariable f 1 f 2 f1 ∗ f2 ∗ 1 − iIBM 0.536 0.561 0.011 0.776 0.602HWP 0.744 0.335 0.317 0.752 0.665INTC 0.607 0.433 0.151 0.730 0.556MER 0.788 −0.515 0.928 0.158 0.887MWD 0.791 −0.514 0.930 0.161 0.891Variance 2.4563 1.1448 1.8502 1.7509 3.6011Proportion 0.491 0.229 0.370 0.350 0.720(b) The maximum likelihood methodEstimates <strong>of</strong>Rotatedfactor loadings factor loadings CommunalitiesVariable f 1 f 2 f1 ∗ f2 ∗ 1 − iIBM 0.191 0.496 0.087 0.524 0.282HWP 0.394 0.689 0.247 0.755 0.630INTC 0.250 0.511 0.141 0.551 0.323MER 0.800 0.072 0.769 0.232 0.645MWD 0.994 −0.015 0.976 0.186 0.988Variance 1.8813 0.9866 1.6324 1.2355 2.8679Proportion 0.376 0.197 0.326 0.247 0.574• The two factors identified by the principal component method explain morevariability than those identified by the maximum likelihood method.• Based on the rotated factor loadings, the two estimation methods identify essentiallythe same two common factors for the data. The financial stocks (MerrillLynch and Morgan Stanley Dean Witter) load heavily on the first factor,whereas the technology stocks (IBM, Hewlett-Packard, and <strong>In</strong>tel) load highlyon the second factor. These two rotated factors jointly differentiate the industrialsectors.• <strong>In</strong> this particular instance, the varimax rotation does not change much the tw<strong>of</strong>actors identified by the maximum likelihood method. Yet the first unrotatedfactor identified by the principal component method was destroyed by the rotation.This is not surprising in view <strong>of</strong> the idea behind the varimax criterion.• The specific variances <strong>of</strong> IBM and <strong>In</strong>tel stock returns are relatively large basedon the maximum likelihood method, indicating that these two stocks have theirown features that are worth further investigation.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!