"Value-performance paradox" isn't one fixed academic term, so it depends which version you mean. Usually it's the high performer who doesn't live the company's values, and whether you keep them. A few well-documented cases:
Jack Welch's GE. His 2000 shareholder letter sorts managers into four types. The hard one is the manager who hits the numbers but doesn't share the values, and GE's answer was to remove them anyway. This is where the framing comes from.
Wells Fargo, 2016. Aggressive cross-selling targets led employees to open millions of accounts customers never asked for. The bank's stated values said customers first; the incentive system paid for volume. The CFPB settlement and the board's own 2017 investigation report are the primary sources.
Uber, 2017. The Holder report (Eric Holder and Covington & Burling, commissioned after Susan Fowler's blog post) found that high performers were protected despite misconduct. Kalanick resigned a few months later.
Netflix's culture memo takes the opposite stance on paper: "no brilliant jerks." Useful as a counterexample, though it's a policy document, not a study of outcomes.
If you mean the measurement version (the more you measure performance, the less your measures tell you), look up Meyer and Gupta's "performance paradox" (1994) in public administration research.
Sources:
https://en.wikipedia.org/wiki/Performance_paradox
https://en.wikipedia.org/wiki/Wells_Fargo_cross-selling_scandal
https://en.wikipedia.org/wiki/Travis_Kalanick