Skip to:
  1. Main navigation
  2. Main content
  3. Footer
Economic Commentary

Unbanked in America: A Review of the Literature

We review the recent literature on the causes and consequences of financial exclusion—that is, the lack of bank account ownership—in the United States. We examine existing work in a range of fields, including economics, finance, public policy, and sociology.

The views authors express in Economic Commentary are theirs and not necessarily those of the Federal Reserve Bank of Cleveland or the Board of Governors of the Federal Reserve System. The series editor is Tasia Hane. This paper and its data are subject to revision; please visit for updates.


In 2019, 5.4 percent of US households were unbanked, meaning that no one in the household had a checking or savings account at a bank or credit union (FDIC 2020).1 The likelihood of being unbanked was even higher for some segments of the population, such as low-income and racial and ethnic minority households.2 As shown in Figure 1, there has been a general improvement over time. Nonetheless, the United States still has a lower rate of access to banking services than most other developed countries (Demirgüç-Kunt et al., 2018).

Our aim in this Commentary is to understand the reasons for financial exclusion in the United States and its possible consequences. We do so by reviewing recent literature on access to transaction accounts in the United States. This literature, which began with the seminal work of Caskey (1994, 1997, 2002), has grown through contributions from researchers in different disciplines. Given this multidisciplinary effort, we draw from a range of fields, including economics, finance, public policy, and sociology. Much of the existing work focuses on the reasons for being unbanked, bank access for racial and ethnic minority households, and the consequences of financial exclusion on payments. We structure our Commentary around these same topics and use data from surveys to illustrate related issues.

Reasons for Not Having a Bank Account

The 2019 Federal Deposit Insurance Corporation (FDIC) Survey of Household Use of Banking and Financial Services asks households about their reasons for not having a bank account. As shown in Table 1, the most common answers focus on the costs of maintaining a bank account, including meeting minimum balance requirements and paying fees for overdrafts and other services. Indeed, according to Pew Center on the States (2012), about one-third of overdraft users close their accounts because of high fees.3 Because most of the literature so far has examined the effects of overdraft fees, problems with previous bank accounts, and distance from bank locations, we focus on these topics. A lack of trust in banks and a desire for privacy are also commonly cited by FDIC survey respondents, but we have found less literature on these issues.

Costs of having a bank account

There is considerable literature assessing the consequences of overdraft fees for bank account ownership. However, the overall effect of overdraft charges on financial inclusion is unclear. On the one hand, overdraft fees can discourage individuals from opening and maintaining bank accounts, excluding them from the financial system. On the other hand, revenue streams from fees may make low-balance accounts more profitable and thus incentivize banks to open accounts for a wider range of customers.


 Table 1: Reasons among the Unbanked for Not Having a Bank Account, 2019 (by percentage)
  Cited reason Main reason
Cannot meet minimum balance requirements 48.9 29.0
Lack of trust in banks 36.3 16.1
Desire for more privacy 36.0 7.1
High bank account fees 34.2 7.3
 Unpredictable bank account fees 31.3 1.6
Identification, credit, or former bank account problems 20.5 8.0
Banks do not offer needed products and services 19.6 1.9
Bank locations are inconvenient 14.1 2.2
Bank hours are inconvenient 13.0 2.4
Other reasons 17.8 13.9
No cited reason 10.4 10.4
Source: FDIC (2020)
Note: The FDIC survey asked unbanked households about their reasons for not having a bank account. Households were asked to cite all reasons that apply and then choose the main reason.


Overdraft fees have not always been common in the United States. Checking account overdraft fees as a business model became prevalent only in the 1980s. Berre, Blickle, and Chakrabarti (2021) discuss how, in the late 1980s, increased computing power and growing sophistication in cost accounting made it possible for banks to categorize individual customers based on their profitability. Banks began to charge higher fees for account services and overdrafts so that even low-balance accounts would be profitable on a stand-alone basis. This approach, in turn, had implications for financial inclusion. Higher fees make it rational for some low- and moderate-income (LMI) customers to close their accounts. This effect of higher fees on account closures is likely to be stronger during economic downturns, when LMI customers may need to watch their spending more carefully.

Servon (2017) provides an in-depth report on the banking experience for low-income individuals and families in the United States. She concludes that many find it difficult to predict when banks will charge them a fee and how much the fee might be. This lack of clarity can be costly because it makes financial planning difficult and penalizes mistakes. It can also induce people to use alternative financial services such as check cashers.4 Fees tend to be more transparent at check cashers than at banks, but also substantially higher. Nevertheless, Servon reports that for many lower-income families, the immediate availability of funds provided by check cashers—as well as the greater transparency and more personal service due to recurrent customer–teller interactions—can offset their higher fees. Using unique survey data, Rhine, Greene, and Toussaint-Comeau (2006) find evidence that a consumer’s decision to be unbanked is not made independently from the decision to obtain financial services from check cashers.

Many of Servon’s conclusions are supported by Barr (2012). Working with the University of Michigan, Barr designed and conducted a unique survey of more than a thousand LMI families in the Detroit area between 2005 and 2006. About 30 percent of the adults surveyed were unbanked. The study finds that LMI households face high monetary costs from using alternative financial services, including high fees and a lack of saving opportunities. They are also confronted with substantial nonpecuniary costs; for example, according to Barr, they have to wait in line to pay bills in person and may suffer strained relationships with friends and family as they ask for help with borrowing needs. Survey interviewers asked unbanked respondents about potential changes that could make them more likely to open a bank account. The most common answers were lower and less confusing fees, more convenient bank hours and locations, lower minimum balance requirements, and the ability to get faster access to new deposits.

Even though overdraft fees can be expensive for consumers, there is some evidence that they can help improve access to bank accounts. Dlugosz, Melzer, and Morgan (2021) analyze whether overdraft fee restrictions affect bank account ownership. They study the consequences of a 2001 ruling by the Office of the Comptroller of the Currency (OCC) that relaxed state-imposed overdraft fee limits for national banks. The authors identify this as a quasi-natural experiment because the fee caps varied by state and over time. This setup allows them to isolate the impact of fee caps on financial inclusion. They find evidence that when fee caps were abolished, national banks increased overdraft fees and expanded the supply of overdraft credit. Yet other costs came down; in particular, national banks lowered minimum balance requirements by at least 25 percent relative to the minimums at state banks, which were not subject to the OCC ruling. In addition, in states where national banks are exempted from overdraft fee caps, the share of LMI households with a checking account rose by 10 percent relative to shares in other states after the fee cap exemption. This rise in inclusion persisted for several years. These results suggest that overdraft fee caps may hamper financial inclusion by making it less profitable for banks to serve LMI customers.

Customers’ overdraft fee costs can also depend on how banks order the processing of transactions. Because a bank can charge additional overdraft fees every time it processes a transaction on an already overdrawn account, it can maximize overdraft revenue by processing the largest transactions first. This is called “high-to-low reordering.”5 Pew Charitable Trusts (2016) examines the practices of 50 of the largest banks and finds that high-to-low reordering is widespread. The study recommends that banks should process transactions either chronologically or from low to high. However, the benefits of a mandated change in ordering are not clear because reducing banks’ revenues can make them more reluctant to provide low-balance accounts. Di Maggio, Ma, and Williams (2020) study the effect of high-to-low reordering on demand and supply for banking services. They examine a series of lawsuits that challenged the practice of high-to low reordering at multiple US banks and find that banning it causes a permanent decline in borrowing from nonbanks such as payday lenders. They also find that affected households significantly increase their consumption of certain durable goods and of essential nondurable goods. However, the paper also finds that banks are significantly more likely to close their branches after being required to stop the practice of high-to-low reordering. Furthermore, such closures appear to be concentrated in low-income zip codes and in areas where banks already have fewer branches. Collectively, these results suggest that high-to-low reordering bans make it less likely that households borrow from payday lenders in the short run, but they can also mean less access to traditional banking services in the longer run, especially for low-income individuals.

Overall, the existing literature provides evidence that overdraft fees can be costly for low-income individuals. Prescott and Tatar (1999) argue that the costs of an account might even exceed the benefits for many unbanked households. Speedier payment settlement would likely help, as customers would then get faster access to funds rather than potentially waiting days for checks or electronic transfers to clear (Brainard 2020). Real-time payments could also make bank account management easier, since people would know instantly how much money they have available. Ultimately, however, low-balance accounts might not be profitable for banks. Indeed, the studies we reviewed show that when fees are constrained—for example, by overdraft caps or bans on high-to-low reordering—banks compensate for their lost revenue by imposing unfavorable terms for low-balance account holders, a situation which reduces access in the longer term.

Former bank account problems

Overdraft fees may not only cause households to forgo a bank account, but they can also have implications for involuntary closure of checking accounts, that is, account closures initiated by the bank. As discussed by Campbell, Martínez-Jerez, and Tufano (2012), virtually every large US bank uses data from ChexSystems, a banking reporting agency that collects information about depositors’ problems with bank accounts, including unpaid overdraft fees, checks bounced at retailers, and suspected fraud. Banks use these data to assess the riskiness of a customer and, in turn, to determine when an existing account should be closed and whether an applicant can open a checking account or obtain a debit card. This means that problems reported on ChexSystems can have long-term consequences for one’s ability to open a bank account.6 Indeed, there is strong anecdotal evidence that a customer’s being reported on ChexSystems can lead banks to deny that customer a checking account or to offer that customer only high-cost or limited-service accounts (see, for example, Servon, 2017).

There is, however, little formal research on how banks use ChexSystems data to determine who should be able to open an account. Campbell, Martínez-Jerez, and Tufano (2012) study ChexSystems data to identify the characteristics of people whose bank accounts are closed involuntarily. The authors find that involuntary closures are more frequent in US counties with overall lower education levels, lower wealth levels, higher unemployment rates, and a larger fraction of single mothers. Closures are also higher in communities with high property-crime rates, low electoral participation, and a high availability of payday lending credit. Bank characteristics also matter: counties with more competitive banking markets and more multimarket banks experience higher closure rates compared to counties with less banking competition and a higher presence of local banks. This difference may arise because local banks have better customer knowledge and so are less likely to accept applications from risky customers—and more likely to show forbearance—compared to bigger banks that are less informed about the local market and have fewer vested interests in a particular community. Also, banks appear more willing to accept risky clients when they face increased local competition for customers, a situation which can, in turn, lead to higher account closure activity. The latter suggests that an increase in involuntary account closures does not necessarily imply less financial inclusion.

Location of bank branches

In recent years, concerns have been raised about the effects that bank branch closures may have on financial inclusion (see Board of Governors of the Federal Reserve System, 2019). There is, however, no consensus in the academic literature regarding whether proximity to bank branches promotes financial inclusion.

Goodstein and Rhine (2017) examine the influence of geographic proximity to a financial institution on a household’s joint decision about whether to have a bank account and whether to use nonbank financial transaction services. Using 2011 data from the FDIC and US Census Bureau, they find that a household with reasonable geographic access to bank branches is more likely to have a bank account and is less likely to use nonbank financial transaction products, though the magnitudes are fairly modest. For lower-income households, the positive effect of bank branch proximity on bank account ownership is somewhat larger. However, the importance of branch proximity is less than that associated with income, education, or race.

In contrast, Célerier and Matray (2019) find that an expansion of bank branches driven by a supply shock increases financial inclusion. They study the consequences of the passage of the Riegle-Neal Interstate Banking and Branching Efficiency Act of 1994, which made bank branching across US states legal but also gave states the right to erect barriers to the entry of interstate branches. States lifted these barriers in a staggered way in the following years. Using FDIC data to identify the location of bank branches and using the Survey of Income and Program Participation (SIPP) to identify both banked and unbanked low-income households for the period 1993 to 2005, they find that the density of bank branches increases by around 20 percent in poor counties after a state fully deregulates.7 This increased presence of bank branches in turn leads to a 4 percent increase in the likelihood that a low-income household is financially included.

Changes in the structure of banking in recent decades may mean that physical proximity to a bank branch is less important now than in the past and that access to the internet is correspondingly more important. For example, the rise of online banking, mobile phone ownership, and new entry from fintech (financial technology) firms has made geographic proximity less important. Table 2 shows how mobile banking has grown since 2015, while physical visits to banks have fallen. Furthermore, Figure 2 demonstrates a positive association between access to the internet and to banking, suggesting the two issues are likely connected. For example, in 2019, 87 percent of banked households had access to a smartphone, compared to only 64 percent of unbanked households.

Table 2: Primary Method Used to Access Bank Account (percentage of banked households)
Year Bank teller ATM/Kiosk Telephone Online Mobile Other
2015 28.2 21.0 3.0 36.9 9.5 0.9
2017 24.3 19.9 2.9 36.0 15.6 0.7
2019 21.0 19.5 2.4 22.8 34.0 0.3
Source: FDIC (2020)
Notes: Data refer to the most common method used to access bank accounts in the previous 12 months: visiting a bank teller, using an ATM or bank kiosk, calling the bank on a telephone, using a computer or tablet (that is, online banking), using a mobile phone app, or using some other method.


In today’s financial system, internet access may well be a bigger factor in financial inclusion than physical access to a bank branch. Friedline and Chen (2021) investigate associations between the racial makeup of communities and rates of fintech penetration, measured as rates of high-speed internet access, smartphone ownership, and online and mobile banking. Using 2015 data from high-poverty zip codes, they find that low-income communities of color have the lowest fintech adoption rates. Friedline, Naraharisetti, and Weaver (2020) find similar results for rural communities: low-income rural communities of color have the lowest fintech adoption rates.

Access to Bank Accounts for Racial and Ethnic Minority Households

According to the FDIC (2020), US households belonging to racial and ethnic minority groups are less likely to be banked, as shown in Figure 3.8 Part of the recent literature has focused on this issue. For example, Faber and Friedline (2020) conducted an independent survey among 1,344 banks across the United States to investigate differences in the costs and fees of checking accounts at commercial banks. They found the costs of opening and maintaining accounts to be higher in areas with larger Black and Hispanic populations.

One problem with studying the effects of race and ethnicity on account ownership is that they are correlated with other household characteristics such as income and education. Rhine and Greene (2013) confront this problem by studying the 2004 SIPP, which is a longitudinal survey. By looking at changes in the banking status of individual households over time, they are able to identify the characteristics that explain account ownership. They find that race and ethnicity are important determinants of banking status, even after accounting for other characteristics. In contrast, Yogo, Whitten, and Cox (2021) examine a dataset of all US tax filers during 1999–2018 and find that the effect of race and ethnicity on access to bank accounts disappears once income and zip codes of households are taken into account. Rhine and Greene (2013) do not have geographic data at the zip code level, so it is possible that race and ethnicity may cease to be important determinants of financial inclusion once geography is fully taken into account. However, there is a strong correlation between race and zip code, so it is not clear which matters more (Aliprantis, Carroll, and Young, 2019).9

Immigration status is another important determinant of bank account access. Rhine and Greene (2006) use data from the SIPP over 1996–2000 to study immigrants’ decisions to open bank accounts. The analysis finds that immigrants as a group are significantly less likely to have bank accounts than people born in the United States, with Mexican and other Latin American immigrants displaying the highest unbanked rates. Using a probit model, Rhine and Greene (2006) find many of the same factors affect the likelihood of opening a bank account for both US-born and immigrant respondents. For example, having a higher net worth, earning a higher income, and being married all reduce the likelihood of being unbanked. Having lower levels of education, poverty-level income, or a large family all make being unbanked more likely.

Amuedo-Dorantes and Bansak (2006) examine the effect of access to banking services on immigrants to the United States and find that access to a bank is associated with larger savings. The authors examine a dataset from the Mexican Migration Project containing information on Mexican immigrants’ banking and remittance behavior and their legal immigrant status at the time of migration. Since 2002, recognition by some US financial institutions of Mexico's Consular Identification Card—the matrícula consular—as an acceptable form of identification has increased Mexican immigrants’ access to US banking services. Nonetheless, the authors find banking among Mexican immigrants remains limited, with only 9 percent of the respondents reporting having a US bank account. The authors use a tobit model to estimate the effects of banking on Mexican immigrants’ money transfers to their families. They find that having a US bank account does not appear to significantly raise monthly remittances by Mexican immigrants, but it does help increase the lump-sum amount they bring back home at the end of their migration periods.

Consequences for payments

Cash use is declining in the United States. Cash was used for 40 percent of transactions in 2012 (Greene, Schuh, and Stavins, 2018), 26 percent in 2019 (Greene and Stavins, 2020) and just 21 percent in 2020 (Foster, Greene, and Stavins, 2021). However, cash transactions remain prevalent for the unbanked. As Table 3 shows, unbanked consumers used cash for 60 percent of their payments in 2020, compared to only 19 percent for banked consumers.

Table 3: Shares of Payment Instrument Use, by Banking Status (percentage) 
Payment instrument  Unbanked Banked
Cash 60.4 18.9
Prepaid card 19.5 2.2
Credit card 14.9 26.2
Money order 3.2 0.3
Income deduction 2.0 0.7
Debit card 0.0 5.0
Check 0.0 5.0
Other 0.0 12.3
Source: Authors' calculations from the 2020 Survey of Consumer Payment Choice (SCPC)
Notes: Income deduction denotes direct payments from income, for example, automatic deductions for an employee's portion of health insurance. In the absence of a bank account, credit card bills can be paid using money orders. Cash can also be used, either by visiting the card issuer's branch location or the issuer's automatic teller machine (ATM). Not all credit card issuers accept cash payments. 


How do the unbanked fare in an increasingly cashless society? According to Coyle, Kim, and O’Brien (2021), only 60 percent of consumers made at least one in-person payment in the previous 30 days in August 2020 (in any payment medium), compared to 96 percent in October 2019, a reduction that suggests a smaller potential role for cash as a medium of payment during the COVID-19 pandemic. If fewer merchants accept cash and more commerce is done online, then unbanked households’ dependence on cash could shut them out of some markets. One solution is to use prepaid cards to conduct electronic payments. The 2020 Survey of Consumer Payment Choice (SCPC) shows that unbanked individuals used prepaid cards for roughly 20 percent of their transactions, compared to only 2 percent for the banked ones, a difference suggesting that prepaid cards are, indeed, a partial substitute for electronic bank accounts.

Overall, however, the extent to which prepaid cards address the needs of the unbanked remains an open question. According to Pew Charitable Trusts (2015), unbanked prepaid cardholders use their cards much like a traditional checking account. Most users would like prepaid cards to offer savings options, but they are less interested in the ability to overdraw their accounts. Hayashi (2016) explains that general purpose reloadable (GPR) prepaid cards—which share the same networks, such as Visa or MasterCard, as standard cards—can be used much like regular debit cards. That is, cardholders can make purchases at any physical or online store that operates on their card’s network, can withdraw cash at ATMs or merchants, and can send and receive bank payments. And, importantly, in 2016 the Consumer Financial Protection Bureau introduced additional protections for prepaid cards, making them more similar to traditional bank accounts.10 Consumers can also avoid unexpected overdraft and nonsufficient funds (NSF) fees entirely by choosing a GPR card that does not offer an overdraft capability, just like with a bank account.

Fees incurred by prepaid cardholders can vary widely, as shown by Wilshusen et al. (2012) and Hayashi and Cuddy (2014). Various fees can make GPR cards costlier than traditional checking accounts, even without overdraft or NSF fees. For example, cardholders who do not receive periodic direct deposits may incur a per-purchase fee. Also, some retailers charge fees when cardholders reload their cards using cash. According to Shy (2020), respondents to the SCPC and the Diary of Consumer Payment Choice find prepaid cards to be slightly less secure and less convenient than debit cards and slightly costlier.

However, GPR cards may have some advantages over traditional bank accounts. For example, it is generally much easier to qualify for a GPR card than for a traditional bank account, since prepaid cards tend not to use ChexSystems as a reporting tool. Moreover, GPR cards typically do not have minimum balance requirements to open an account or to qualify for lower or zero monthly fees. In addition, overdraft fees on GPR cards tends to be much lower than for checking accounts; see Hayashi, Hanson, and Maniff (2015).


The unbanked rate has been decreasing in the United States, but it remains high compared to unbanked rates in other developed economies. Lower-income and racial and ethnic minority households are more likely to lack access to a bank account than white households with higher incomes. Fees and minimum balance requirements are important factors in the decision to be unbanked. However, several studies show that when fees are constrained, banks find low-balance accounts unprofitable, so access to bank accounts can actually decline.

It is also important to point out that many of the studies we cite, even the most recent, might not fully reflect the fast speed at which the payments landscape is evolving and the consequences of such changes for the unbanked. The move away from cash and the increasing tendency to shop remotely might increase the costs of exclusion for the unbanked. At the same time, the proliferation of fintech products could make access to a traditional bank account less essential for US households provided they have access to the internet.

The most effective way to reduce financial exclusion remains an open question. Several existing initiatives and proposals could be interesting topics of study. For example, the Cities for Financial Empowerment Fund’s Bank On project promotes the use of low-cost commercial bank accounts (Federal Reserve Bank of St. Louis, 2021). Baradaran (2015) proposes reintroducing postal banking. The Board of Governors of the Federal Reserve System (2022) paper on the potential costs and benefits of a US central bank digital currency (CBDC) says, “Further study would be helpful to assess the potential for CBDC to expand financial inclusion, including cases targeted to underserved and lower income households.” Finally, it is possible that private market initiatives, such as continued innovation with GPR cards, will lead to better and cheaper substitutes for bank accounts.

  1. The Federal Deposit Insurance Corporation (FDIC) Survey of Household Use of Banking and Financial Services has become a standard reference on access to bank accounts in the United States because of the survey’s large sample size and its focus on bank account ownership. Return to 1
  2. For example, the FDIC (2020) reports that the probability of being unbanked rises to 23.3 percent if annual household income is lower than $15,000 and 13.8 percent if the householder was Black. Return to 2
  3. A typical overdraft fee at the largest banks in the United States is $35, although some banks have recently reduced these fees. See Horowitz and Liang (2022). Return to 3
  4. The FDIC defines “alternative financial services” as those offered by providers that operate outside of federally insured banks and thrifts. Examples are check-cashing outlets, money transmitters, car title lenders, payday loan stores, pawnshops, and rent-to-own stores. Return to 4
  5. For example, suppose an individual has an account balance of $100. In one day, she makes a payment of $20 in the morning, $50 in the afternoon, and $120 in the evening. If the bank processes these in order, her bank account becomes overdrawn only when the third payment is made, and so she would pay one overdraft fee of, to use a common example, $35. But if the bank processes the largest payment first, her bank account is overdrawn on the first transaction and thus incurs three $35 fees upon processing the subsequent two payments, for a total of $105 in fees. Return to 5
  6. ChexSystems generally keeps a closed account on its database for a period of five years from the report date. Some banks choose not to use ChexSystems or decide to apply a shorter time horizon for consideration of involuntary account closures. Return to 6
  7. SIPP reports data on the socioeconomic, demographic, and financial characteristics of households in the United States. Return to 7
  8. The FDIC survey collects data at the household, not individual, level. The race or ethnicity of a household is determined by that of the owner or renter of the home. We drop the category “Native Hawaiian or Other Pacific Islander” because the sample size in 2019 is too small to produce a precise estimate. Return to 8
  9. Moreover, the dataset in Yogo, Whitten, and Cox (2021) is restricted to people who are 50–59 years old, so their results might not apply beyond that age cohort. Return to 9
  10. These rules covered reimbursement of funds on lost or stolen cards, error resolution and investigation, upfront fee disclosures, and access to account information. See 81 FR 83934 at Return to 10
Suggested Citation

Boel, Paola, and Peter Zimmerman. 2022. “Unbanked in America: A Review of the Literature.” Federal Reserve Bank of Cleveland, Economic Commentary 2022-07.

This work by Federal Reserve Bank of Cleveland is licensed under Creative Commons Attribution-NonCommercial 4.0 International