Eko andryanto Prakasa
2012-Sep-20 18:22 UTC
[R] ES with time varying (GARCH model) under nonnormal (using cornish fisher expansion)
Hiii I have tried to measure ES with cornish fisher expansion using PerformanceAnalytics package, but i still confuse because to measure volatility i use GARCH model and i don't know how to consolidate it with ES in PerformanceAnalytics package...... i have measured ES under normality using formula ES=sqrt(S)*dnorm(qnorm(p))*value Where S is volatility that I got from GARCH model; p is probability that is 5%. and to get ES under nonnormal, I modified the formula above become ES=sqrt(S)*dnorm(ZES)*value where ZES is alpha prime and obtained from ZES = Zscore-(1/6*(p square -1)*skewness)..... but the result is less than VaR (that its volatility used GARCH model) so its not reasonable, cause ES should larger than VaR Do you have any idea, suggesition or solution ? Thanks for the attention... Eko [[alternative HTML version deleted]]