Why is this? I think there are (at least) two reasons to doubt outcome statistics for relievers. First, one season is a small sample size for a typical relief pitcher -- 50-70 innings. One bad/unlucky outing (4 earned runs, say) can take a month of shutout pitching to dilute down to a decent ERA. Second, more of their outcome statistics are out of their control than any other player. They often leave runners on base who may or may not score and they may end up facing clutch pinch hitters selected specifically to face them. I'm a skeptic about the tendency of some to denigrate outcome-based statistics, but this seems like one area where structural performance numbers such as velocity, spin rate, and detailed location data can be most helpful and ERA is less useful, at least season by season.