Here's another question: "How can a human being, which consists of many neurons, each of which has limited agency over the aggregate actions of that entity, be remotely capable of demonstrating moral agency?"
Incidentally, I noticed that you substituted 'human-like moral agency' for 'moral agency' in the OP. I never said states have human-like moral agency (nor am I making a claim to the contrary per se).
You might also note that the answer to your question is that they don't. It is a problem applicable at scale - and we do terrible things to single neurons all the time in the name of science.
Also, I intended 'limited agency' to be read as 'zero or very little agency'.
I believe each group of people have their own social and moral codes that guide their behavior. I don't believe its something easy to understand, determine or represent.