TL;DR: The EPI graph isn’t measuring productivity vs. pay, even for “typical workers”; it’s measuring wage inequality, and images like these are the visual equivalent of out-of-context half-truth soundbites:

https://www.reddit.com/r/badeconomics/comments/6rtoh4/productivity_pay_gap_in_epi_we_trust/