Our study takes place in eight municipalities in the province of Ancona (Marche region), which are very similar in terms of population size, economic activities, rate of migrants over the total population. We will randomly select a sample of 5,000 households among the list of all poor households legally residing in each city. These administrative lists are created yearly following a standard and homogeneous procedure in order to provide social benefits to low-income people in each city of the Marche Region. Selected households will invited and incentivized to take part in an interview at the local town hall. Subjects showing up (which are expected to represent about half of the invited pool based on the pilot experience) will be administered a questionnaire and three lotteries to assess risk and time preferences. At the end of the baseline survey, they will be randomly assigned to treatment and control groups. Aware of the existence of the two types of profile among our target population, we will condition treatment on the baseline account ownership in order to tackle both the need of basic financial inclusion and the need of savings incentives. People with no bank account (expected to be about 25% of the entire pool) will receive an invitation to open a current account at the local branch of our partner bank, where a trained loan officer will be waiting for them. This is a basic transaction account, offering zero interest rate and being freely available through a debit card allowing payments and withdrawals. One out of three respondents will be offered the account at normal market price (T1) and the other two free of charge (T2). Since the first product is already present on the market, it represents a control group for the cost experiment. Yet, a positive response to our offer could imply the existence of a lack of trust and/or information about formal bank accounts. By contrast, a higher take-up of the free account would point at the cost issue. Provided we find a higher take-up in T2 than in T1 (which we expect based on the pilot results), we will then be able measure the real impact of holding a transaction account by following the two groups over time. Turning to the households who own already a current account at baseline, one out of five will be offered a free current account (T2), which will form the control group of the savings experiment, and two out of five a free savings account (T3). The savings account offers a competitive interest rate and is accessible only at the bank’s branch. The remaining participants will be offered a variant treatment consisting in the same savings account complemented with an initial monthly savings goal elicitation and a monetary incentive to respect it during one year (T4). For banked people, T2 could represent an improvement (in which case we would observe a positive take-up) by providing (i) a cheaper transaction account that would be used to replace their previous one, or (ii) a second account that would encourage savings through a ‘mental accounting’ effect. With respect to T2, T3 offers (i) a small remuneration on deposits (2% interest rate), (ii) a lower liquidity given that savings cannot be accessed through the debit card but only by going at the bank, and (iii) a specific savings label. Finally, T4 adds a commitment option that could help savers facing self-control problems. By comparing take-up rates between treatment groups, we will be able to assess the relative relevance of each of those. Moreover, by observing the allocation of funds between the current and savings accounts, we will be able to check if the savings account has indeed a specific added value. We will measure the impact of treatments by following the standard procedure of submitting a common baseline and endline questionnaire to all treated and control individuals, in order to quantify the change in different economic and behavioral outcomes.