Correlation, co-dependency and risk aggregation [33]

Go to: Summary | Previous | Next   
Bullet points include: With some copulas a similar approach can be used as for Gaussian copula. But for others, e.g. Clayton, a different approach is required. Suppose we have a pair of uniform random numbers u1 and u2 with a copula    C(u1, u2) then: Hence given a pair of independent uniforms, u1 and u2, we may generate a pair of uniforms with joint distribution equal to the copula, u1* and u2* as follows, where   C-1(u2|u1) is the inverse of C(u2|u1) with respect to its first argument, u2. Relatively easy if derivatives of copula are simple to calculate analytically

Contents | Prev | Next | ERM Lecture Series

Desktop view | Switch to Mobile