Can we differentiate between “Atheists ought logically to be moral nihilists” and “If you are an atheist, you are necessarily a moral nihilist” ? I take you to mean the second of these, which is indeed plainly false.
The first of these statements is not obviously false. It is (epistemically) possible that there are no good non-religious grounds for moral realism (which is not to say that there are good religious grounds for it either). That said, I do wonder if Robertson actually believes it. If he ceased to believe in God, would he really start behaving “immorally” whenever it turned out to be in his self-interest?
Most forms of signalling fall into one of two categories—“ability” signalling and “commitment” signalling. (Not all forms of signalling—e.g. wealth—fall easily into one of these, but I think the distinction is a useful one). The clearest example of ability signalling is Bryan Caplan’s model of education—you go through a rigmarole in order to demonstrate intelligence, which employers look at when choosing employees. Engagement rings are a form of commitment signalling.
The whole point of a commitment signal is that it is costly to the signaller—it is a sacrifice which would be utterly senseless if one were not committed to a particular course of action. Hence banks building expensive buildings that mean they’re attached to a particular location; people from minority groups taking the time to learn languages which will not allow them to communicate with anyone outside the group; etc. Ability signalling, on the other hand, need not involve any such sacrifice for the signaller. Rather, it should be difficult or expensive for people who do not possess the trait being signalled; for those who do possess it, the cheaper the signal is the better! Hence the value of a degree is not that you found it difficult to get a top grade, but that someone of lower intelligence could not have done so (except perhaps with extremely hard work).
(One consequence of this: taxing commitment signals doesn’t necessarily harm the signaller, but taxing ability signals does.)
So ability signalling should be cheap to the signaller, commitment signalling expensive to the signaller, and both can involve costs to society (through wasted resources and through the weakening of other people’s signals). Weakening of other people’s signals is perhaps (?) inevitable, and the whole point of a commitment signal is to be expensive to the signaller. This suggests that ways to improve the efficiency of signalling should include at least two categories: reducing wasted resources from commitment signalling, and making it cheaper to signal ability.
Philanthropy is a mixture of wealth signalling (which I lack a good model for—intuitively it feels more like an ability, but seems to function more like a commitment) and commitment signalling (giving money away only makes sense if one cares about other people, or at least about other people’s opinions of oneself). Most of your suggestions fall into the category of reducing wasted resources; the exception is your idea of increasing on-the-job assessments as a way of reducing the need for higher education, which it seems to me is more about making ability-signalling cheaper.