This thesis exploits a unique micro dataset that uses a natural field experiment to identify indirect effects of formal savings access on de facto ineligibles residing in the same community. Despite widespread interest in microfinance as a poverty-reduction tool, the indirect effects on the very poor of expanding formal financial services remain largely unexplored. This study examines evidence from a large field experiment which helps fill this gap. It also contributes to an important emerging literature on the indirect impacts of policy interventions in developing countries, often (incompletely) evaluated solely on the basis of how they impact participants and beneficiaries. In developing regions, households vulnerable to extreme poverty often benefit from long-standing local safety nets based on cash gifts and other transfers from relatives and friends, which help them smooth consumption across food-deficits and household shocks. To date, little is known about how these pre-existing practices are affected as community members begin adopting newly available formal financial services, and there remains much unexplored in the interaction of formal financial markets with informal safety nets. This paper addresses that gap by examining how formal savings expansion affects inter-household wealth transfers, with a particular emphasis on receipts by the most vulnerable. Using a rich panel dataset from Central Malawi that includes over 2,000 households, I find that experimentally boosting local savings uptake in rural areas leads to a strong positive effect on assistance receipts by non service-users during peak periods of hunger. The difference is strongest among the most vulnerable households. That is, the entrance of formal savings appears to complement local informal support systems for the highly vulnerable through an indirect mechanism, channeling greater wealth to such households during periods of food-deficits. The positive impacts of formal savings expansion on non service-users suggests that formal savings may have substantially greater benefits than would be suggested by focusing exclusively on the impacts experienced by the service-users themselves.