I've consulted Goodwin, Smyth and Chase & Phillips, and as far as I can tell they all agree that when a proclitic (like οὐκ) is before an enclitic (like ἐστίν) the proclitic has an accent and the proclitic has none. However a search with the Perseus database reveals that
οὔκ ἐστι occurs 5 times
οὔκ ἐστιν occurs 6 times
οὐκ ἔστι occurs 394 times
οὐκ ἔστιν occurs 770 times
And so the rule appears to be followed very infrequently, at least for this pair of words. Does anyone know what is going on here?