All of it. The theoretical work that showed this was done in the 1960s and early 1970s by Feynman, Deser, and others. One of the key insights was figuring out how to reformulate the theory so that it did not require an infinite series of terms, whose sum nobody knew how to calculate, in order to derive the exact field equation correct to all orders. IIRC Deser was the one who figured that out, and he published a paper in the early 1970s that summarized the research.