100%(1)1 out of 1 people found this document helpful
This preview shows page 20 - 22 out of 35 pages.
The measure of risk utilised in this study was beta, a measure of the sensitivity of a company’s rateof return to the rate of return of a market index (Risk Measurement Service, 1998). Beta values,obtained from the Risk Measurement Service (1989), were only available for 28 of the samplecompanies. Thus, tests of hypotheses relating to this variable use this reduced sample ofcompanies.The final proxy for political sensitivity used in this study was membership of a politically sensitiveindustry. Industry membership was initially determined based on Standard Industrial Classification19
(SIC) codes, obtained from the Key Business Directory of Australiafor 1988. For those companiesin the sample not listed, industry was determined by reference to Jobson’s Yearbook of PublicCompanies, and by direct reference to the company’s annual report. Companies in the miningindustry were regarded as members of a politically sensitive industry, due to the environmental andsocial concerns surrounding mining. Also, in 1989 income derived from gold mining in Australiawas still exempt from tax, while exploration expenditures were also afforded concessional treatmentfor tax purposes. This suggests that the mining industry is politically sensitive and that abnormalprofits could lead to intervention affecting the concessions received.105.Results5.1Data PropertiesDescriptive statistics were calculated to provide information about the distributional properties ofthe data to be used in testing the hypotheses developed in Section 3. Erring on the side ofconservatism, only those variables satisfying the Kurtosis, Skewness and Kolmogorov-Smirnov onesample tests of normality are regarded as being normally distributed. On this basis the results ofthese tests indicate that only the variable BETA can be regarded as being normally distributed. Thissuggests that nonparametric rather than parametric methods are most appropriate to the data utilisedin the study. As such, statistical methods were confined to nonparametric tests of association.115.2Hypothesis TestingThe degree of association between the dependent variables and all the independent variables,excluding membership of a politically sensitive industry, are examined using Spearman RankCorrelation Coefficients. The results of these tests are provided in Table 4. Panel A provides thecorrelations with the dependent variables concerned with the issue of capitalisation of identifiable10The tax free status of income from gold mining and sales was removed from 1 January 1991.11In the hope of allowing parametric methods to also be used, all other variables were transformed in an attempt to bring them closer to normality. Neither the natural logarithmic, square root nor fourth root transformations were successful in significantly improving normality of the data. Specifically, only one measure of the dependent variable could be brought to normality by transformation, and only four independent variables were able to be made normal in this way.