برگشت سیستم بانکداری: تجزیه و تحلیل تجربی از بسته شدن ناخواسته حساب بانکی

Using a new database, we document the determinants of involuntary consumer bank account closures. During 2001–2005, approximately 30 million debit accounts were involuntarily closed for excessive overdrafting. We focus on multiple factors to explain this phenomenon: household economics and financial decision-making ability, social capital, bank policies, and the alternative financial services sector. Involuntary closures are more frequent in US counties with a larger fraction of single mothers, lower education levels, lower wealth, and higher unemployment. Closures are higher in communities with high property crime rates and low electoral participation. Bank policies have an independent relation to closures, with counties having more competitive banking markets and more multi-market banks experiencing higher closure rates; bank structure also seems to affect the speed at which banks adjust their policies to changes in household income. Finally, using both national data and a state-level shift in regulation, we find evidence that access to payday lending leads to higher rates of involuntary account closure.

There is substantial policy interest in the extent to which Americans are banked. According to the FDIC’s 2009 National Survey of Unbanked and Underbanked Households, 25.6% of Americans are un- or underbanked (FDIC, 2009). A sizeable fraction of these have experienced the involuntary closure of their checking and debit accounts.1 For example, in 2005 about 6.4 million accounts were involuntarily closed, adding to more than 30 million such closures over the prior 5 years. Virtually all of these closures were due to repeated overdraft or non-sufficient funds (NSFs) activity. While banks routinely honor overdrafts, they typically charge customers “courtesy overdraft fees” that generate substantial revenues. After some number of offenses, however, banks’ risk management systems close the accounts of recidivist “bouncers.” Almost all banks report these involuntary closures to a clearinghouse, ChexSystemsSM, which maintains a national database on closure activity. Apart from Jacob et al. (2006), there is no published empirical work on the incidence of involuntary bank account closures. However, virtually every large American bank uses ChexSystems’ debit histories to determine eligibility for basic transaction accounts with checking or debit privileges. ChexSystems provides the data; bank practices determine which customers qualify for an account.2 Involuntary closure at one institution may lead other banks to deny customers checking or savings accounts,3 or to offer them only at high cost or with limited service (Lamb and Leonard, 2007, Michael, 2004, Manning, 2000, Beckett, 2000 and Caskey, 1994). The consequences of an involuntary closure can follow customers for years after the incident. Often, they may be left with very limited and costly banking alternatives, or with access only to the fee-for-service check cashing or money service businesses. In short, customers who bounce too many checks can find themselves bounced out of the formal banking system and into the fee-based alternative financial service (AFS) sector. As Desmond and Sprenger’s (2007) summary shows, the high costs of these services are compounded by indirect costs resulting from the lack of access to traditional bank accounts and related credit opportunities. This paper uses a new national database to understand the determinants of involuntary account closure. We match county-level closure statistics over the period 1999–2006 to corresponding demographic, economic, and industry data. While a number of studies examine advanced financial decision-making such as how to allocate funds in a 401(k), we examine a much simpler financial condition: the inability to balance cash flows to avoid recurrent overdrafts. We posit that involuntary closure is the joint product of behavior by individuals and banks in the context of a broader social network and financial economy. First, closures reflect the consumer behavior of repeated overdrafts. We consider a number of potential factors influencing this behavior. First, as an economic proposition, overdrafts occur when expenses exceed current income, so we look at the level and volatility of both. We also consider buffer savings and certain family traits as they relate to narrow income–expense margins. We find evidence that involuntary closures are positively related to both family structure (in particular, single motherhood) and unemployment and negatively related to financial assets. However, involuntary closure is not related to poverty per se, but rather is more widespread. Second, behavioral traits or cognitive ability may affect individual budgeting ability. We find evidence of a higher incidence of closures by the young, the old, and by poorly educated households. Using a sociological framework, we also consider how community structure might affect closure activity. We explore whether communities with greater social capital might have lower closure rates, whether due to lower rates of individual overdrafts, more lenient or flexible bank policies, or more financial support from friends and family. We find evidence that social capital, measured by voting turnout, is related to closure rates—a finding that holds even after controlling for differences in income, wealth, and other demographic and economic characteristics across counties. Bank policies also play a role in involuntary closure rates. Banks set rules about which debit applicants to accept; design products that condone or even encourage overdrafts, at a cost; and choose when and whether to close accounts. For example, many banks offer to honor the debits of a customer with insufficient funds with the expectation that the customer will pay back the amount of the posted checks plus a fee (often $25–35). 4 These interactions can be thought of as high-cost, short-term credit because of the “effective interest rate” for the short-term extension of credit. For example, a $30 fee for a $100 overage remediated in 2 weeks would correspond to an APR of 780%, were it considered interest. Moreover, many banks also offer “free checking” programs with overdraft privileges but no monthly maintenance fees. Such programs may attract less financially stable customers. As these customers provide a significant source of revenue through overdraft fees, banks may have little incentive to promote consumer self-discipline. 5 Involuntary account closure also reflects the bank’s decision to close the account. Banks set policies—or permit branch discretion—to determine when to close accounts. For example, banks do not typically close accounts simply because the customer bounced a single check. Involuntary account closure implies that the bank found the customer’s financial mismanagement to be so severe that it was no longer in the bank’s interests to extend that person banking privileges. These dynamics may affect closures in several ways. First, we posit that in more competitive banking markets, banks extend their pool of potential clients to include more risky customers. Second, we consider whether “local” banks—which know their customers better or have “soft” information on local markets (Agarwal and Hauswald, 2007)—are less likely to force closures than larger multi-market banks. As hypothesized, we find that bank competition and localness are related to closures. This line of inquiry is consistent with related research on bank credit decisions (Agarwal and Hauswald, 2007, Degryse and Ongena, 2005, DeYoung et al., 2008 and Hauswald and Marquez, 2006). Finally, banking activity takes place in the context of a broader financial service sector. In particular, the short-term, high-cost unsecured loans made available by payday lending could forestall closures or could exacerbate them by enabling unmanageable debt levels. Using both our national data and data from a regulatory shift in Georgia, we find that the presence of payday lending is positively related to closures. The remainder of the paper is divided into five sections. In Section 2, we describe our data and empirical methodology. Sections 3 and 4 examine the empirical determinants of closure activity. In Section 5, we discuss the implications of our work and additional research.

Involuntary debit account closure is a frequent occurrence in the United States, with over 30 million accounts closed over a 6-year period. The impact of involuntary account closure is a concern for policymakers and consumer advocates who argue that the “unbanked,” “underbanked,” or “self-banked” may pay high costs for basic services or have limited access to additional financial services, such as credit. Our paper provides the first look at the incidence of involuntary closures using county-level data for the United States taken from an industry-standard database maintained by ChexSystems. The hypotheses we use to explain the incidence of involuntary closuresare drawn from economics (at both the individual and bank level), psychology, and sociology. While one might think that involuntary closures are primarily driven by poverty, this turns out not to be the case. Our analysis of ChexSystems’ data reveals that differences in the rate of involuntary account closures across and within counties are only partially explained by negative shocks to income and proxies for populations where margins between income and expenses are thin (e.g. poverty rates and the presence of single mothers). Even after controlling for income, financial assets, poverty, family structure, and unemployment, we find that rates of involuntary account closure are explained by other measures related to consumers’ ability to budget and forecast (education and age); bank incentives (local versus multi-market banks and competition); community norms and social capital; and the availability of credit through payday lending. Furthermore, banking practices and bank characteristics can help reconcile the seemingly contradictory impacts of income levels and changes on closures. As with any multivariate analyses, one cannot rule out the possibility that the measured variables are capturing effects of omitted variables. However, even after using a relatively rich set of explanatory variables, it seems troubling to find such high rates of involuntary account closure for certain groups, including single mothers and particular racial populations. These results call out for further study. In the current environment, where regulation of consumer financial services is the subject of tremendous debate, our results—which are consistent with higher involuntary closures where banks are less local and markets more competitive—require careful scrutiny. One interpretation is benign: greater competition, especially from non-local banks, encourages banks to open debit accounts for more Americans. While some of these individuals may be incapable of managing these accounts, the extension of banking services is a positive social good. A second interpretation is less benign: banks’ extension of overdraft-based revenue models to financially vulnerable customers leads to more fee income for the banks, but raises the cost of basic banking and ultimately leads these individuals to be “bounced” out of the system. Similarly, our results about the interaction between banking and payday lending raise many questions. Both in a large cross-section and in our analysis of the Georgia payday lending ban, we find a positive relationship between the existence of payday lending and involuntary account closures. A benign interpretation might be that a checking account is necessary for someone to use payday lending; where there is less payday lending, there is less need for a bank account. However, this explanation would not account for the involuntary closure of accounts. A less benign implication would be that the presence of this high-cost, short-term credit adds to the over-extension of household budgets, and exacerbates the rate at which households overdraw their accounts. While the paper raises many new questions, we believe that it makes a number of contributions. First, while many financial economists are eager to study relatively advanced financial “mistakes” (e.g., refinancing mortgages or failing to participate in equity markets), we call attention to a level of financial mismanagement that is unfortunately endemic and poorly understood. Second, by adopting perspectives from many disciplines, we find that this household “failure” cannot be understood from a single lens. It seems related to family economics, behavioral and cognitive factors, community structure, banking competition and incentives, and competition from non-banks. Third, by documenting the populations that seem most subject to this behavior, our paper helps to pinpoint areas where basic financial education, automatic defaults, or other “nudges” might be directed.