This paper studies the cost-effectiveness of targeted payroll taxes for stimulating labor demand. It uses rich administrative data to study the effects of an Italian reform that raised social security contributions for apprenticeship contracts but granted a substantial discount for firms with 9 employees or less. The discount does not increase demand for apprenticeship contracts. Instead, it subsidizes inframarginal hiring. This reform is not cost-effective. Point estimates imply that each million euros of foregone social security contributions supports the employment of 29 apprentices for one year and no permanent contracts (these estimates are not statistically different from zero).