It's a slightly more confusing issue, but that is a valid way of looking at it. A given photon is emitted, and then observed later cosmologically redshifted, but these two events happen in different frames of reference, so there's not any reason why the energy in the two frames should be measured the same. Even if you want to try to look at the universe as a whole, under general relativity there are no global reference frames, and it's not even clear that the total energy of the universe can be unambiguously defined. You can find a medium-depth discussion of the matter here.