If so, we should receive the different components (with different colour) of the light of distant stars with small time delays with respect to one another. This may be tested and precise observations are under way. The problem is that this effect is difficult to compute within the theory, and therefore it is not completely clear what the theory really predicts.
