The use of banks can generate benefits such as greater savings, financial education and access to credit. But some groups are less likely to use banks than others. Compared to the national average, low-income households are four times more likely to be unbanked; Black households are two and a half times more likely to fall into this category.
“Why do we see these continuing disparities, despite all these benefits?” says Alexander Zentefis, assistant professor of finance at Yale SOM.
Policymakers and researchers have speculated on two possible explanations. One is access: these households may not have high-quality bank branches nearby. The other is demand: perhaps these groups are less inclined or less capable of using banks because, for example, they have less capacity to save or do not trust the institutions.
In a new study, Zentefis and Jung Sakong of the Federal Reserve Bank of Chicago tried to quantify those factors. Using mobile device data on user visits to bank branches, they found distinct patterns. For low-income households, the main reason for less frequent use of banks appeared to be lower demand. Many banks were located nearby, but these households did not use them as much.
But for black households, the barrier was access: Branches tended to be further from their homes. And in some areas, demand for banks by black households was even higher than among white households. This finding contradicts earlier speculation that lower bank use by black households could be due to lower demand.
The results point, if anything, “in the opposite direction,” says Sakong.
With the rise of mobile and online banking, one might assume that branch visits are not a significant indicator of bank usage. But when Zentefis and Sakong analyzed data from the FDIC’s Household Use of Banking and Financial Services Survey, responses from participants suggested that people still rely heavily on in-person interactions at bank branches.
This was particularly true for low-income and black households that did not seem to be offsetting lower in-person use with higher digital use. For example, black respondents were 10% less likely than white respondents of similar ages and income levels to have visited a branch in the past year. But they were also 7% less likely to say they primarily relied on online or mobile methods to access banking services.
“They trust the branches more,” says Zentefis. “And yet they are visiting the branches less.”
To investigate why, the researchers examined anonymous data from the company SafeGraph on the movements of mobile phone users, which covers approximately 10% of all smartphones in the US, from January 2018 to December 2019. geographic area that includes between 600 and 3,000 people—visited every bank branch in the country per month, with slight distortions and noise added to protect user privacy.
The team considered three factors. The first was the lawsuit, which encapsulated any characteristics of people in a Census block group that affected whether they visited a bank. These traits could include your need for savings, financial education, trust in banks, and flexibility in your schedule.
The second factor was the effect of distance on branch visits, that is, the extent to which having a branch further away discouraged people from traveling to it. And the third factor was the quality of the branch, which covered whatever characteristics made a bank more attractive. These included having more products or services, better interest rates, more ATMs, more conveniences like a coffee shop, or being located in a desirable neighborhood.
Access to the bank was determined by both the quality of the branch and the effect of distance. People in a census block group would be considered to have high access to banks if they had many branches nearby or if the branches around them were of high quality.
Because the data had been slightly distorted for privacy protection, the team had to calculate the values of these factors indirectly. They ran computer simulations of user visits to the branches and performed the same types of data adjustments that SafeGraph did. By testing different parameter values until their simulated data closely matched the observed data, they were able to determine which values were most likely to be accurate.
The researchers found substantial variation in access to banks between regions. For example, people in eastern New England had much better access than those in the Deep South. Urban centers had more access than neighboring rural areas. And within a metropolitan area, some neighborhoods did substantially better than others.
“There is a lot of local variation in access, beyond the city or county level,” says Sakong.
The team then focused on demographic factors that were associated with lower bank usage. To determine the role of income, they looked at how branch quality, distance, demand, and number of visitors changed as the median household income of a group of blocks increased. And to determine the role of race, they extrapolated how those factors changed if they compared a hypothetical 100% black block group to a 100% white block group.
Surprisingly, low-income households had greater access to banks than wealthier ones. They tended to be closer to the branches, perhaps in part because they often lived near commercial areas, and the quality of those branches was similar to the quality of branches in wealthier communities.
In contrast, low-income households seemed to have lower demand. They may not have the minimum balance to open an account or want to avoid bank fees, or may not need certain in-person banking services, such as safe deposit boxes. Their lower demand outweighed their higher access, ultimately leading to fewer bank visits.
Among black households, a different pattern emerged. Compared to white households, “access was much worse,” says Sakong.
A census block group made up of all African-American residents would have 5.6% fewer visitors to the branch per month than an all-white group of similar age and income, the team estimated. This discrepancy was entirely due to lack of access. There was no evidence of lower demand; in fact, in large cities, black households had a slightly higher demand for banks than white households. But this higher demand was more than offset by weaker banking access, leading black communities in big cities to use branches less.
How might policymakers close these branch usage gaps? One proposal is to add banking services to post offices. This plan would increase access for all, but could potentially widen racial disparity because post offices tend to be located closer to white homes, the researchers found. However, it is possible that adding banking services to post offices could increase demand among low-income households, the researchers argue. If one of the reasons for the lower demand in these households is their lack of trust in banks, and they trust post offices more, they might be more inclined to use those services. Another policy option would be to subsidize or grant tax breaks to banks that establish branches in black communities, to help reduce the racial gap in access to banks.
In the future, researchers could use mobile phone data and similar analysis methods to investigate other disparities. For example, they could unravel whether gaps in healthcare use are due to differences in distance, quality of services, demand, or some combination.
“This paper reveals a way to use that data and better understand, ‘What are the driving forces?’” says Zentefis.