The intertemporal capital asset pricing model (ICAPM), under the usual assumptions, produces a multifactor pricing model like the one described in this question.
When and how would it be possible to test this result when the assets' mean returns and variance-covariance matrix of returns are not constant over time?