A New Method for Generating Random Correlation Matrices

We propose a new method for generating random correlation matrices that makes it simple to control both location and dispersion. The method is based on a vector parameterization, γ = g(C), which maps any distribution on R n(n−1)/2 to a distribution on the space of non-singular n × n correlation matrices. Correlation matrices with certain properties, such as being well-conditioned, having block structures, and having strictly positive elements, are simple to generate. We compare the new method with existing methods.