First of all, you need to have a sequence of values, usually losses f_L. It can be the output of a simulation in which case you have just a vector of thousand of losses in dollar, for example. So you just find the loss that corresponds to the say, 95% quantile, l(95\%) then keep all loss values equal or higher than l(95\%), and just take the average of those remaining values. In mathemtical terms:
\text{TVaR} = \frac{1}{n} \int_{\text{VaR(95%)}}^\infty f_L dL
where n indicates the number of loss values that are inexcess of the quantile l(95\%).
No comments:
Post a Comment