Good initiative. I have thought about the relatively low "answered-questions rate" on Open Data SE and came to the conclusion that the questions we pose are generally harder to answer than the ones on other sites.
Like most others, I use this site mainly for data-hunting and I will only come here when my prior research did not produce any satisfactory results. With this in mind, it is highly unlikely that anyone can just come by the question, do a bit of research and come up with the answer unless the person has some knowledge about the data already. I suspect that this is similar for a lot of users. Such rather specialized questions seem to have no answer to us but at some point they might be read by a specialist who then provides an answer.
Should we accept No answers?
With this in mind, I think we should not put such questions off with a simple No. One reason is that you cannot prove and, in fact, never know for certain that the data does not exist. The second and more important reason is that a simple No will almost never help.
Instead, I suggest that for every question we try our best in providing a useful and not necessarily complete answer. Two concrete examples come to my mind where this approach was, I believe, helpful (either for the author or for future readers): example 1 and example 2. In both cases, it is reasonable to claim that no such "prepared" data will exist anywhere. At the same time, there exists similar data or methods ways how one could combine several sources to construct the data. In both cases a simple No might have been technically correct. The upvotes and the general interest in both questions, especially in the first one, show however that there are other, possibly more helpful, ways to answer such questions.