Policy makers are generally interested in both the anti-poverty impact and the efficiency of reforms. To connect these two dimensions, I measure the poverty gap change per unit of net revenue that tax-benefit reforms produce. To isolate the impact of reforms and account for labour supply responses, I apply a microsimulation decomposition framework to poverty gap and net revenue changes. Labour supply responses are accounted for using reduced-form models, partly exploiting variation over time that reforms produce. I measure this indicator in Belgium between 2005 and 2014, focusing on revenue changes at the bottom half of the income distribution. Without considering labour supply reactions, reforms reduced the poverty gap among the poor by €0.6 for each euro of net revenue decline. However, this drops to €0.4 when negative labour supply reactions are included, which were caused by unemployment benefits growing faster than in-work compensations. These results highlight the importance of looking simultaneously at reforms to in- and out-of-work benefits.
