5-Matrix Algebra Page 147 Wednesday, February 4, 2004 12:49 PM
Matrix Algebra 147
are all zero except the elements on the diagonal that coincide with those
of the matrix A:
a 11 a 12 ··· a 1 n a 11 0 ··· 0
a 21 a 22 ··· a 2 n 0 a 22 ··· 0
A = · · · · ⇒ dgA = · · · ·
· · · · · · · ·
· · · · · · · ·
an 1 an 2 ··· ann 0 0 ··· ann
The trace of a square matrix A is the sum of its diagonal elements:
n
trA = ∑ aii
i = 1
A square matrix is called symmetric if the elements above the diago-
nal are equal to the corresponding elements below the diagonal: aij = aji.
A matrix is called skew-symmetric if the diagonal elements are zero and
the elements above the diagonal are the opposite of the corresponding
elements below the diagonal: aij = –aji, i ≠ j, aii = 0.
The most commonly used symmetric matrix in finance and econo-
metrics is the covariance matrix, also referred to as the variance-covari-
ance matrix. (See Chapter 6 for a detailed explanation of variances and
covariances.) For example, suppose that there are N risky assets and
that the variance of the excess return for each risky asset and the covari-
ances between each pair of risky assets are estimated. As the number of
credit risky assets is N there are N^2 elements, consisting of N variances
(along the diagonal) and N^2 – N covariances. Symmetry restrictions
reduce the number of independent elements. In fact the covariance σij(t)
between risky asset i and risky asset j will be equal to the covariance
between risky asset j and risky asset i. We can therefore arrange the
variances and covariances in the following square matrix V:
σ 11 , · σ 1 , i · σ 1 , N
· ··· ·
V = σ 1 , i · σii, · σiN,
· ··· ·
σ 1 , N · σiN, · σNN,
Notice that V is a symmetric matrix.