As a PhD student studying finance at Columbia University, Naz Koont learned how the personal relationships bankers develop with their clients build trust and help sell products. At the same time, walking around her Morningside Heights neighborhood and throughout Manhattan, she noticed most of the bank branches she passed were empty. She wondered: Where were the customers and signs of these pivotal in-person relationships?
Meanwhile, Koont watched classmates, friends, and other people around her check their bank balances, pay bills, and make deposits using apps on their phones. “I started to see this big disconnect and to get interested in the changing nature of banking,” says Koont, who recently joined Stanford Graduate School of Business as an assistant professor of finance.
In her research, Koont set out to answer a straightforward but significant question: Does digitization pose risks to the financial system’s stability? “I’ve always been interested in understanding the world through an economic lens and curious to find tools to understand the world in a quantitative way,” she says.
As more banks adopted digital platforms and apps, and customers forwent brick-and-mortar branches to do their banking online, she was curious how this shift was impacting competition among banks and whether it had altered the makeup of bank deposits and loans. “I wanted to see, are these innocuous changes, is it lossless, or is this really something that we have to take into account?”
Too digital to fail?
Koont set out to build a model that could shed light on how digitization has impacted the banking sector. First, she had to circumvent the lack of data on financial institutions’ adoption of digital platforms by spending a year and a half hand-collecting data on when the United States’ 5,000 banks first released mobile apps. By 2019, according to her data, 60% of banks had a mobile banking application – up from a handful in 2010.
Next, she combined those figures with data on branch locations, bank balance sheets, uninsured deposits, loans, and other figures to develop a framework that compares the digitized world of 2019 to a counterfactual scenario where banks did not go digital.
It turns out that these in-person relationships are actually really important for low-income customers to get access to credit.”Naz KoontAssistant Professor of Finance
Koont found that in the digitized banking world, there are, on average, 8% more banks providing services in a given market, even as the average bank closes nearly 6% of its physical branches. Mid-sized banks with assets between $10 billion and $100 billion have gained the most market share amid the shift to digital. They provide 29% more services within the banking sector and serve nearly 7% more markets, as digitization enables them to compete more effectively with bigger firms.
“These mid-sized banks had sufficient scale to develop high-quality digital platforms, and they didn’t already have branches everywhere, so they had the scope to grow with digital technology,” she says. Banking customers, for their part, have benefited from this increased competition through more choice.
At the same time, however, mid-sized banks’ expansion offers a cautionary tale, Koont says. These banks have traditionally been subject to fewer government regulations. Yet, in recent years, the collapse of banks like Silicon Valley Bank and Signature Bank in 2023 has cast doubt on this lighter regulation. Meanwhile, digitization has hurt community banks with less than $10 billion in assets. These banks lack sufficient resources to invest in technology at scale and have lost market share, becoming less profitable.
This shift, Koont says, “is something for policymakers to be aware of.”
“There’s the notion of banks being too big to fail. This suggests an increase in the number of banks that have the potential to disrupt the financial system,” she says. Policymakers may want to consider whether mid-sized banks warrant stricter regulation now that they’re growing due to digitization.
When Koont dug deeper into the data, she found that uninsured deposits in the banking sector (those exceeding the $250,000 insured by the Federal Deposit Insurance Corporation) have increased with digitization. The banking sector’s uninsured deposits rose by 9%, with larger banks attracting the bulk of the funds.
This could be cause for concern, she says, since uninsured deposits are less stable. Signs of economic distress often prompt skittish investors to pull out their unprotected cash, which can spark a bank run. The rise in uninsured deposits could be explained by businesses that adopted digital platforms for payroll and treasury functions finding that digital deposits streamline operations. It’s an area that warrants additional research, she says.
Disconnected borrowers
Looking at loans, Koont found that as banks expanded after adopting digital platforms, they more than doubled the number of counties in which they provided mortgages to high-income borrowers. Yet access to credit did not increase significantly for low-income borrowers. After digitization, in these new counties banks received 26% fewer loan applications from low-income borrowers and rejected 76% more loan applications from these borrowers, ultimately reducing their ratio of local low-income lending by 27%.
What’s likely at play here, Koont says, is a greater reliance on metrics such as credit scores in the absence of intangible knowledge that’s gained through in-person interactions. Personal relationships and trust used to factor into banks’ in-person local lending decisions, she says. “If you allow people to bank with you digitally, then you interact with them less, and it’s harder to know them,” Koont says. “It turns out that these in-person relationships are actually really important for low-income customers to get access to credit. There’s something lost there.”
When banks are more “at arm’s length” from their clients, it becomes more expensive for them to figure out which low-income borrowers are creditworthy. It’s easier to simply screen out most low-income borrowers. A lingering question, Koont adds, is whether intangible information traditionally conveyed through close personal relationships with customers will be better codified as technology continues to evolve, making the digital banking of the future a closer substitute for in-person banking.
Koont’s research also suggests that digital depositors may behave differently than traditional ones and may be more likely to switch banks. In a recent working paper, she finds that digital banks have to react more swiftly to changes in interest rates since it’s easier for their customers to move their money in search of a better rate. “When interest rates change, how do digital and non-digital banks respond differently, and what does that mean for customers? Is digitization changing that, too?” she asks.
Better data would help both researchers and regulators better understand digital banking, Koont adds. For example, banks don’t have to distinguish between digital and physical deposits in their disclosures. They also have significant discretion in how they log digital deposits. If a customer opens an online account in Houston but then moves to New York, their bank may still assign their digital deposits to the Texas branch.
“There is some movement to get banks to report more,” Koont says, “but they have no incentive to disclose. The current regulatory disclosure requirements that banks are subject to has just not caught up to the digital world that we’re living in.”
For more information
This story was originally published by the Stanford Graduate School of Business.