This study evaluates the effectiveness of contracting out mandatory publicly provided counselling and training for long-term unemployed in Flanders (Belgium) to private for-profit and non-profit organisations (FPOs and NPOs). A multivariate transition model exploits timing-of-events and novel exclusion restrictions to account for selection on unobservables. Overall, the intervention was highly effective in reducing unemployment duration, but also spurred employment instability and withdrawals from the labour force. FPOs slightly, but significantly enhanced exits to employment without reinforcing recidivism relative to the public provider but not significantly relative to NPOs. FPOs also charged lower prices and hence were the best performing providers.