Erasmus university rotterdam

Global	Local	Three Moment CAPM
Four Moment CAPM
Egypt Max.	1.246	1.149	1.243	0.015	1.210	0.066	0.006
Egypt Min.	0.017	0.182	-0.047	-0.061	-0.103	-0.042	-0.002
Egypt Average (SD)	0.52 (0.251)	.62 (.256)	.44 (.243)	-.02 (.015)	.39 (.241)	.006 (.019)	.001 (.001)
Israel Max.	1.619	1.566	1.673	0.057	1.786	0.047	0.001
Israel Min.	0.177	0.249	0.135	-0.043	0.222	-0.171	-0.008
Israel Average (SD)	.79 (.288)	.71 (.221)	.85 (.319)	.01 (.018)	.92 (.337)	-.01 (.028)	.00 (.002)
Morocco Max.	0.360	1.157	0.373	0.009	0.366	0.019	0.001
Morocco Min.	0.051	0.285	0.001	-0.030	0.0071	-0.067	-0.003
Morocco Average (SD)	.21 (.082)	.71 (.226)	.16 (.086)	-.01 (.010)	.20 (.090)	-.03 (.017)	.00 (.000)
Turkey Max.	1.869	1.191	1.659	0.022	1.659	0.060	0.006
Turkey Min.	0.538	0.485	0.501	-0.078	0.392	-0.148	-0.007
Turkey Average (SD)	1.14 (.213)	.81 (.136)	1.05 (.209)	-.02 (.018)	1.03 (.218)	-.01 (.228)	.00 (.002)

Note: Number of cross-sections; Egypt (79), Israel (135), Morocco (38), Turkey (264).

Table 7 shows the result of the cross-sectional regression of the Global CAPM. is the expected excess return on a zero beta portfolio. is the market price of risk (the difference between the expected rate of return on the market and a zero beta portfolio). The linear relationship between the dependent variable (return) and the explanatory variable (beta) is examined. In the case of Egypt, the constant variable is significantly different from zero in all three periods, therefore based on the intercept criterion the CAPM hypothesis can be rejected. The beta coefficients are significant, which means that beta does have an explanatory power on return for all periods. It can be seen from the F-test that the model as a whole is statistically significant. However, the adjusted value shows that the goodness of fit of the explanatory variable is very modest. If the residuals are normally distributed, the hypothesis of normaltity of the Jarque-Bera statistic must not be rejected. Table 7 shows that the null hypothesis of a normal distribution can be rejected for the total period and the period during the crisis. A heteroscedasticity test was conducted, since cross sectional data could suffer from the variance of the error tem being large for large beta stocks and small for small beta stock (Omran , 2007). No signs of heteroscedasticity were found. The Durbin Watson (DW) statistic showed no signs of autocorrelation, so OLS is justified.

When the results of the Global CAPM of Israel are investigated, it shows that the constant variable is only significantly different from zero when the period before the crisis is considered. During the crisis and for the total period the CAPM the intercept criterion cannot be rejected. The beta coefficients are significant, except when the total period is considered. It can be seen from the F-test that the model as a whole is statistically significant, except when the total period is considered. However, the adjusted value shows that the goodness of fit of the explanatory variables are again very modest. As for Egypt and Israel, the results obtained for Morocco and Turkey are mixed. Some periods show evidence supporting the CAPM, for instance Turkey during the crisis, where the constant variable is not significantly different from zero, therefore based on the intercept criterion the CAPM hypothesis cannot be rejected. The beta coefficient is significant on a 5% level, which means that beta does have an explanatory power on return for all periods. Furthermore, it can be seen from the F-test that the model as a whole is statistically significant. However, the adjusted value shows that the goodness of fit of the explanatory variable is once again very modest. Evidence against the CAPM comes from the period before the crisis for both Morocco and Turkey. In both cases the is significantly different from zero and the beta coefficients are insignificant.


Table 7: Global CAPM

Global CAPM