ABSTRACT

Chapter VII has been concerned with hypotheses, database, and experimental design. In chapter VIII analyses will be performed using the database and experimental design explained in chapter VII to test the two hypotheses of this thesis: (i) segmental geographical disclosure affects systematic risk; and (ii) segmental geographical disclosure results in a lower systematic risk than would have been the case without such segmental geographical disclosure. This chapter proceeds as follows:

First, individual company moving betas will be computed using the fortran programme earlier cited (section 7.334, and appendix VII(C)). These moving betas will be used to perform cross sectional as well as time series a naly ses.

Moving betas will be examined using non parametri c as well as parametric tests. First nonparametric tests will be performed to see if the treatment group and the control group belong to the same population, assuming no underlying distribution of the moving be tas.

Parametric tests will then be performed to examine (i) independence (ii) stability and (iii) the impact of intervention on the treatment and control group betas.

The time series will be redefined for intervention analysis purposes, and the two hypotheses of this thesis will be examined in the light of intervention analysis.

Having tested the hypotheses, a duration-onset analysis will be performed on the beta changes in response to intervention to see if the duration is permanent or temporary, and wnether the onset is abrupt or gradual.

Finally, impact of possible confounding variables will be tested in an attempt to improve the external validity of our results. First,the moving betas.