## ABSTRACT

In this chapter we analyse a situation where some

or all of the explanatory variables in the random

ly varying coefficients model

are collinear. The term ’pe r f e c t 1 m u l t i collin

earity refers to a situation where perfect inter

relationships exist among the explanatory va r i

ables. In this case we will not get a u n i q u e

solution to the generalised least squares normal

equations since the data matrix X of order nxK is

of a rank less than K, and hence the matrix X ’ IT^X

is singular.1 ’High' multicollinearity refers to a

situation where some or all the explanatory vari

ables are ’highly’ but not ’perfectly’ collinear.