What I’m getting at is that, while Perot might have had an equal impact from a popular vote perspective, he did not necessarily have an equal impact from an electoral perspective. He probably did flip states like Montana and Nevada to Clinton; his impact in larger Midwestern states is harder to gauge.
Of course, the popular vote was a five-point Clinton win, and if Perot really did draw evenly between the two it seems extremely unlikely that Bush was going to win the Electoral College under any circumstances, even if Perot likely did cost Bush a few electoral votes.
But this helps to raise a second, deeper argument: If Perot cost Bush the election, he likely did so well before Election Day. Consider the horse race polling from 1992. Bush led handily in polls until Perot declared his candidacy (not shown), then saw his standing ebb. But Bush voters did not go to Clinton. Rather, they went to Perot, at least at first. Clinton was mired at around 27% of the vote throughout the spring and summer, at least until Perot suspended his campaign on the eve of the Democratic convention. After an initial “pop” for both candidates, Clinton surged to a 22-point lead over Bush, who found himself roughly where he was in the polls before Perot dropped out.