Do you know how to perform a bias variance covariance decomposition of the mse?

Question

I have a vector ´x=[x1, x2, ..., xM] of size M and vector y=[y1, y2, ..., yM] of the same size. y is an estimate of x, where x is the real value. I can compute the mse by averaging (xi-yi)*(xi-yi) for i from 1 to M and then dividing by M. How can I decompose the mse, in the three components (bias, variance and covariance)?

Answer 1

Nope, I sure don't.