This paper studies how employment subsidies for start-ups shape their performance. We exploit an unexpected policy reform in Belgium that permanently exempted start-ups hiring their first employee from payroll taxes for that employee. Using firm-level administrative data and a regression-discontinuity-in-time design, we find that subsidized post-reform startups employed fewer workers and generated lower output, value added, and profits compared to pre-reform start-ups. However, post-reform start-ups were more likely to survive as employers. These effects emerged within the first year after hiring and remained stable over a medium horizon of three years. Our findings indicate a compositional shift: the subsidy primarily induced low-productivity firms to enter the market. As most firms nowadays are nonemployers, our results meaningfully generalize the theoretical implications of standard neoclassical entrepreneurship models (employee–employer margin) and fill the important gap of the nonemployer–employer margin.