Despite numerous existing studies on the role of microcredit on poverty reduction and households well-being, the issue has remain controversial. This study aims to contribute to this debate by replicating and extending the previous work of . This study proposes two alternative estimation strategies, namely the First Difference (FD) and Propensity Score Matching (PSM), for an experimental data used in  to re-analyze the impact of expanding a well-established microfinance system to “marginal borrower” in Bosnia and Herzegovina. These alternative estimation strategies are chosen in order to control for unobservable characteristics and estimate the impacts based on the correct counterfactual. Our results show that despite the di erences in their magnitudes and significance, FD and PSM estimations, compared to , produce similar sign of coefficient of treatment effects for all outcome variables. The results do not clearly shows that there is a support that a loan has increased self-employment activities and business ownership, as well as shift away from wage work as in . Moreover, our findings do not completely support the  claim that stated the labor supply of teens in the age of 16-19 who work in household business increased. Lastly, while the results afirm that a loan is likely to decrease the households savings, its impacts on consumption are still inconclusive once we control for unobservable characteristics.