**4.5 Test for normality of TY-VAR residuals**

In the test for normality, to examine whether the residuals are normally distributed. We employed the null hypothesis H0: residuals are normally distributed. From **Table 7** we rejected the null hypothesis of normality of residuals of each equation as well as all the equations combined at 5% level of significance since p-value of all the variables are zero. Hence, we concluded that residuals are not normally distributed [62].


*\*df and Prob stands for the degree of freedom and probability. Source: Estimation was compiled using E-views Version 9 software.*

Although, the credibility of Iarque-Bera test of normality with application to VAR has been questioned specifically for an I(1). Jarque-Bera normality of the series does not guarantee normality of distributions, it only signifies normality of the first four moments of a distributions [58]. According to Lutz and Ufuk [63] in their remarks, they posited that Jarque-Bera test based on asymptotic critical values can be very unreliable. In their submissions, they gave the asymptotic critical values of 1–100% in their Monte Carlo analysis of VAR. They presented that the size distortions of the asymptotic test persevere even for sample sizes as large as 5000 observations.
