Australia's 'income management' policy requires benefit recipients to spend at least half of their government transfers on essentials (e.g. food, housing). We estimate income management's impact on birth outcomes by exploiting its staggered rollout. By changing parents' consumption patterns, the policy aims to improve child outcomes. We find no evidence of this. Instead, our estimates suggest it reduced average birthweight by 95 grams and increased the probability of low birthweight by 3 percentage points. We explore the mechanisms that may explain this finding. Our study demonstrates how policies that are not carefully implemented and tested can unintentionally escalate existing inequalities.