We study the labor market impacts of retroactively reducing felonies to misdemeanors in San Joaquin County, CA, where criminal justice agencies implemented Proposition 47 reductions in a quasi-random order, without requiring input or action from affected individuals. Linking records of reductions to administrative tax data, we find employment benefits for individuals who (likely) requested their reduction, consistent with selection, but no benefits among the larger subset of individuals whose records were reduced proactively. A field experiment notifying a subset of individuals about their proactive reduction also shows null results, implying that lack of awareness is unlikely to explain our findings.