<p>Public works are usually considered a less important and less effective type of Active Labour Market Policies (ALMP) as has been emphasized by various empirical studies, mainly based on micro-level data. However, very few studies have looked at local-level labour market effects in the case of large-scale public works intervention. In this study, we perform a municipality-level counterfactual evaluation of a large-scale intervention targeting unemployed women without tertiary education in Croatia to work as assistants for elderly people. The municipality-level effects are examined in Croatia by applying difference-in-differences (DID) estimation and following the approach for staggered rollout by Callaway and Sant’Anna (Callaway and Sant’Anna 2021a). Empirical findings suggest that intervention has resulted in significant reductions in long-term unemployment over the longer term, albeit partly caused by the implementation dynamics, in particular the return of participants to the unemployment register after completion. More genuine and lasting impact on long-term unemployment was noted only two years after the municipality’s exposure to the intervention. Despite the significant financial resources invested in the intervention, we find no evidence of displacement effects. We found no effects on women’s employment in other jobs or on their activity rate. Besides the empirical findings, research also involves methodological insights regarding the application of aggregate-level counterfactual studies in the evaluation of labour market interventions, providing lessons on evaluation setup, evaluation timeliness, research questions and data availability.</p>