Improving Financial Access in Rural Areas: Evidence from a Randomized Trial in Spain

2024-11-01

2025-12-01

- Withdrawals: Cash withdrawals per month.
- Receipts: Cash receipts per month.
- Transfers: Transfers (cash movement) between CaixaBank accounts per month.
- Own Checks: Deposits in account or cash of own checks or promissory notes (CaixaBank) per month.
- Foreign checks: Deposits of foreign checks or promissory notes (other entities) per month.
- Transfers: Transfers (cash movement) to non-CaixaBank and international accounts per month.
- Other: Other cash transactions per month: cancellation of contracts, contributions to NGOs, collection of non-cash receipts per month.

The outcomes are reported both for transactions performed at the ofibus cashier terminal attended by the sales employee (table and chair inside) and that involve cash movements in the customer's account and for those that are carried out at the ofibus ATM (self-service by the customer) and that involve cash movements in the customer's account. We will also exploit this categorization as a secondary outcome to evaluate whether the treatments change the probability of performing transactions in these two modalities using dummies for each of these and no transactions as well as a categorical variable and variables capturing the total number of operations performed in each modality.

We will analyze the effect of the different treatments on the outcomes of interest shown below through regressions:
Y_i= α+β_1 〖T1〗_i+β_2 〖T2〗_i+ 〖X'_i δ + u_i,
where Y_i represents the different outcomes of interest that we will observe for the program (see below for more details) for individual (or, alternatively, municipality) i, 〖T1〗_i is a dummy that takes a value of 1 if the individual is in a municipality (or, alternatively, if the municipality is) treated with a double ofibus, 〖T2〗_i is a dummy that takes a value of 1 if the individual is in a municipality (or, alternatively, if the municipality is) treated with information, and 〖X'〗_i is a vector of controls included for precision (we will show results with and without controlling for 〖X'〗_i). As controls we will consider: distance to the nearest CaixaBank bank branch, distance to the nearest bank branch of any entity other than CaixaBank, people of legal age who reside at home, educational level, age, gender, and municipality (alternatively, province) of residence. The coefficients of interest are β_1 and β_2, which measure the causal effect of intent-to-treat (ITT) on the different measures of knowledge or use of the program.
We will cluster standard errors by municipality. We will run a randomization balance check using pre-treatment municipality characteristics: distance to the nearest CaixaBank bank branch, distance to the nearest bank branch of any entity other than CaixaBank, average income, population, share of population with income lower than 40% of the median, among others

Not available

Randomization done in office by a computer.

We have stratified by the assignment to the 8 geographical blocks and by the distance from the municipality to the closest municipality having an ATM. Specifically, each of the eight blocks is divided into two groups, depending on whether the distance to the closest municipality with an ATM is above or below the median.

Yes

184 municipalities

We plan to get survey data for an average of 50 individuals per municipality.

24 municipalities in treatment 1, 32 municipalities in treatment 2, 128 municipalities in control.