Do Agents Improve Financial Inclusion? Evidence from Brazil

Blog Series

Agent banking is an innovative delivery model that has brought trained financial services providers and access points within closer reach for millions of people around the world. However, little is known about the quantitative impact of branchless banking on financial inclusion. Agent networks have succeeded in taking person-to-person (P2P) payments and bill pay to scale, but how successful is this channel in delivering savings, credit, and other financial products, especially to the poor?

With more than 400,000 agents as of December 2013 according to the Central Bank, Brazil has one of the largest agent networks in the world. To examine the impact of agents on financial inclusion in Brazil, BFA and the Bill & Melinda Gates Foundation implemented a nationally representative survey of Brazilian households focusing on financial inclusion and agent use. BFA published a longer focus note and report analyzing the data.

Men load oranges onto a truck
Men load oranges onto a truck. Photo by Luiz Grillo.

We find that a majority of respondents (heads of household who are most knowledgeable about household finances) do visit agents on a regular basis: nearly 70% of respondents regularly pay at least one bill through this channel. However, only a very small proportion of respondents use agents for transacting through a bank account or accessing new financial products: just 12% of banked respondents usually make their withdrawals at an agent, and 9% usually deposit at an agent. Even fewer have opened a bank account (4% of banked individuals) or accessed credit (6%) through the agent channel.

Although the percentage of Brazilian respondents who use agents for financial services beyond bill pay is small, we find evidence that agents have reduced exclusion by serving traditionally underserved populations. Those who do use the channel are more likely to be poor, less educated, female, and to live in small towns, rural areas, and the Northeast, Brazil’s poorest region.

Therefore, agent impact on financial inclusion in Brazil is mixed: few Brazilians use the channel for transformative access to financial services besides paying bills, but agents have succeeded in reaching poorer and more remote populations. Reflective of the tepid impact of agents on transformative financial inclusion is the fact that 79% of households without banking accounts regularly visit an agent to pay bills. This regular interaction represents a missed an opportunity to bring new clients into the formal financial system through the agent channel.

As banks and mobile network operators continue to build their agent networks, we should remember that agents offering cash-in and cash-out services for bill pay or person-to-person payments do not necessarily make access to formal savings, credit, and insurance possible for the poor. Agent coverage does not always beget transformative access to new financial products, and it is possible for clients to make formal payments while still lacking a bank account, as is the case for an estimated 32 million Brazilian households.


09 April 2014 Submitted by Eduardo (not verified)

Please, where the information on the 400,000 agents came from? It says here that is "according to Central Bank". Is this data available from any public source?

21 April 2014 Submitted by Eduardo Diniz (not verified)

Hi Caitlin,
Thanks for you time. You might know that from these 400,000 correspondents, about 250,000 are what is called as "pastinha", single person correspondent, offering only credit, mostly payroll loans. These guys are not the kind of correspondent we are investigating (shame on us). We better keep in mind the difference between the two kinds of correspondents to understand the real potential for the model to promote financial inclusion and local development.

23 April 2014 Submitted by Caitlin Sanford (not verified)

That is a great point, Eduardo. There are a few different types of agents in Brazil, and the impact for financial inclusion is likely to vary based the incentives each kind of agent faces. Indeed this is an opportunity for more research. Thank you.

Add new comment