One of the patterns most visible in our dataset is that the largest white-collar Korean sentences rarely run their full course. Lee Jae-yong (Samsung): pardoned. Shin Dong-bin (Lotte): pardoned. Chey Tae-won (SK): pardoned mid-sentence. Lee Jay-hyun (CJ): pardoned twice. Chung Mong-koo (Hyundai): suspended at appeal, then pardoned. Kim Seung-youn (Hanwha): suspended, then pardoned.
Pardon is a constitutional power; the President of Korea has used it under every administration since 1948, conservative and progressive alike. The political function it has served — at least in chaebol cases — has been to soften final sentences whose mitigators the courts had already used to the maximum at trial.
This matters for sentencing analysis because it means the effective punishment for high-net-worth white-collar defendants is determined by *two* tribunals, not one: the court at trial, and the executive at pardon. A 7-year nominal sentence that runs for 2.5 years before pardon is effectively a 2.5-year sentence. Nominal harshness papers over an empirical pattern of leniency.
For proportionality analysis, the implication is uncomfortable: if the gap between low-income defendants and elite defendants is already wide at sentencing, and the executive pardon adds another reduction layer that low-income defendants almost never receive, the empirical disparity is larger than the trial transcripts alone show.