Conditioning information and variance bounds on pricing kernels with higher-order moments : theory and practice