The log-returns of country or regional share indices are decomposed into 
two component: regional component and the World Index. World Index is 
modelled by non-Normal Random Walk, whose in-novations are modelled by 
skewed distribution. Regional  components are modelled by GARCH- or Random 
Walk process.  The selected models are used for estimating Value-at-Risk ( VaR ) 
and Con-ditional Value-at-Risk ( CVaR ) in different holding periods by simulations.   
In  short holding period  there is a significant difference between extreme risk 
of the Normal Random Walk and the skewed model.