Chapter 7 Bank Density, Population Density, and Economic Deprivation across the United States: Implications for Public Health Outcomes

In: Economics and Mathematical Modeling in Health-Related Research
Author:
Scott W. Hegerty
Search for other papers by Scott W. Hegerty in
Current site
Google Scholar
PubMed
Close
Open Access

Abstract

Because limited financial access has been shown to be associated with adverse public health outcomes in the United States, modeling this access and identifying geographic areas where it is deficient is essential. Recent research on the locations of bank branches has identified thresholds below which a given area can be considered to be a “banking desert.” Thus far, most analyses of the country as a whole have tended to focus on minimum distances from geographic areas to the nearest bank, while a recent density-based analysis focused only on the city of Chicago. As such, there is not yet a nationwide study of bank densities for the entire United States. This study calculates banks per square mile for US Census tracts over ten different ranges of population density. One main finding is that bank density is sensitive to the measurement radius used (for example, density in urban areas can be calculated as the number of banks within two miles, while some rural areas require a 20-mile radius). This study then compiles a set of lower 5- and 10-percent thresholds that might be used to identify “banking deserts” in various urban, suburban, and rural areas; these largely conform to the findings of previous analyses. Finally, adjusting for population density using regression residuals, this chapter examines whether an index of economic deprivation is significantly higher in the five percent of “desert” tracts than in the remaining 95 percent. The differences are largest – and highly significant – in the densest tracts in large urban areas.

1 Introduction

Although the use of online and mobile banking has increased markedly in recent years, a case can be made that “bank branches matter,” particularly for older or lower-income residents with limited mobility or technological fluency. A recent report by the Federal Reserve’s Board of Governors (2019) notes that more than half of rural counties lost banks between 2012 and 2017, and that residents face difficulties adjusting to these changes. Likewise, the Woodstock Institute (2018) notes the challenges faced by older residents in the Chicago area.

Geographic areas with relatively little financial access – so-called “banking deserts” – might instead be served by alternative (and sometimes predatory) financial service providers. Even when traditional banks are utilized, residents of areas with few banks often pay higher interest rates for limited credit (Ergungor, 2010; Nguyen, 2019). This is usually explained by a lack of information among lenders who do not understand the communities in which they are lending, or to high monitoring costs due to increased distance between borrowers and lenders (Degryse & Ongena, 2005).

In addition, “banking deserts” have been linked to social effects such as crime (Kubrin et al., 2011) or adverse public health outcomes (Eisenberg-Guyot et al., 2018). In particular, this latter study finds that being unbanked or relying on “fringe” loans are associated with higher propensities of having poor or fair health. In the United States in particular, poor financial and physical health are related; money issues can cause a person to experience high levels of stress, or an expensive hospital visit might bankrupt a patient with poor or no health insurance. Since this insurance is often provided as part of an employment compensation package, there is a strong connection between individual income and access to healthcare. The connection between financial and physical well-being, and especially the directions of causality, are worthy of future investigation.

It is important to note that the concept of “banking deserts” appears to be uniquely American – and, as is shown below, is often most evident in large central-city areas. One reason why is due to urban structure: while city centers outside the United States often constitute prime real estate, this country’s central cities (near to, but outside, the central business district) are often relatively low-income, with property values rising in the suburbs. Urban density gradients and income gradients often follow different paths in US cities compared to elsewhere. But, since poverty is concentrated in these areas, residents lack access to financial services; limited automobile access and reliance on underfunded public transportation might matter more than physical proximity to a bank branch, even if it is only a few miles away.

Second, many Americans on the edge of the formal economy rely heavily on cash payments, and often do not have personal bank accounts. In contrast, access to payment technology is often more universal in other parts of the world, and in some cases, innovations in mobile payment systems have helped foster financial inclusion among low-income communities. US residents (and especially those without citizenship status) might rely more heavily on “non-bank financial institutions” that cash third-party checks for a large fee. Others might heavily leverage personal networks or relationships with store owners (Puchalski, 2016).

While a focus here is on the US banking system and differences in access across density ranges, an additional issue involves the relative access of banks versus other types of service providers. While there is ample literature on “retail deserts” as well as areas with limited access to food and other essentials, more research needs to be done analyzing whether banks represent a high-need service (akin to healthy food) or are less commonly used (such as a department store). This would give insight regarding minimum distances and travel costs in “banking deserts.” In addition, banks, grocery stores, and department stores are often located near one another, due to zoning laws or suburban development patterns, a particular neighborhood might be a “desert” for all types of service simultaneously. This “clustering” likely drives the findings presented here, but is left to future research. As they stand, these results highlight the presence – and significance – of US central cities with high degrees of economic deprivation and limited bank access.

This study proceeds as follows: Section 2 outlines important literature regarding “banking deserts” and their implications. Section 3 describes methodology. Section 4 presents the results, and Section 5 concludes.

2 Literature Review

Previous literature on “banking deserts” use geospatial and/or statistical methods to identify areas where bank access is limited. This can be in absolute terms (no banks at all within a certain area), or relative (such as the percentage of tracts with the least access). Once such areas are identified, the causes of such phenomena (such as income or racial determinants), or the effects (such as increased crime or disorder) can be analyzed.

Much of the literature that finds gaps in bank service provision – the so-called “spatial void” hypothesis noted by Smith et al. (2008) – focuses on the payday loan operators and other “fringe banking” providers that seek to exploit these gaps and offer limited services at much higher costs to the consumer. Brennan et al. (2011), for example, find that poor neighborhoods in Winnipeg have been overserved by these non-bank financial institutions and underserved by banks and credit unions. Fowler et al. (2014), on the other hand, find no evidence of any “spatial void” in their study of US counties. Racial disparities are an important covariate with the lack of banking services (Wheatley, 2010; Cover et al., 2011) as are income and socioeconomic variables such as housing tenure (Hegerty, 2016; Dunham et al., 2018).

Using a distance-based measure of “banking deserts,” Kashian et al. (2018) examine more than 60,000 Census tracts from 2009 to 2015, controlling for population density by regressing the distance from each centroid to its nearest bank and examining the residuals for urban, rural, and suburban areas separately. They find that poverty is negatively related to bank proximity only in urban areas; this relationship is insignificant in suburban tracts and positive in rural ones.

Most recently, Hegerty (2020) conducts an analysis of bank locations in Chicago, focusing on bank counts within one and two miles of each block-group centroid. He finds “banking deserts” in roughly nine percent of the city, and estimates that these areas contain roughly 0.4 banks per square mile. These block groups are shown to both be poorer and to have fewer white residents than the city as a whole, and particularly in comparison to neighborhoods with large shares of bank branches. But the “rule of thumb” implied in the paper most likely only applies to large cities, so further empirical work could find bank densities for other locations.

Is it possible to refine this definition to create a nationally applicable threshold (or set of thresholds), and how might these cutoffs differ in large cities, smaller cities, and rural areas? Kashian et al. (2018) consider the lowest 5 percent of residuals from a regression on population density (their measure of “adjusted” bank density) and find cutoffs of 1.56 miles for urban areas, 4.28 miles for suburban areas, and 12.54 miles for rural areas. Assuming that this implies one bank on the edge of each circle with the given radii, this converts to 0.41 banks per square mile only for urban areas; the corresponding suburban and rural values are 0.05 and 0.006, respectively. If banks were uniformly distributed in rural areas, this latter value would imply only two banks in a 10-mile radius, or less than eight within a 20-mile radius.

Noting that the relationship between bank and population densities appears to be nonlinear, the current study controls for population density by performing separate regressions over 10 different density ranges. Adjusted bank densities are highly correlated with the unadjusted densities, so the latter are the primary focus here. Comparing these results with those of Hegerty (2020), the 80th and 90th percentiles – in which most Chicago tracts are located – are found to have bottom 5 and 10 percent thresholds that match both the earlier block-group-level analysis and a tract-level re-estimation conducted here. Many of the most rural areas have no banks even within 20 miles, but many rural and suburban areas have density thresholds in line with the results of Kashian et al. (2018). Comparing bank densities with the tract-level economic deprivation index of Hegerty (2019), deprivation is shown to have significantly higher in the 5 percent of tracts with the lowest bank densities than in the remaining 95 percent, but that these differences are largest at higher population densities.

3 Methodology

Bank data are taken from the FDIC’s Summary of Deposits database; they were current as of June 30, 2019. These provided data contained 87,931 data points with both XY coordinate data and address data in the lower 48 states and the District of Columbia. These data points, which were able to be plotted in Geographic Information Systems software, comprise more than 99.9 percent of the original dataset. To maintain a single method of geocoding, no attempt was made to add the additional bank locations to the database. In addition, while Hegerty (2020) notes that FDIC latitude and longitude data contain various inaccuracies – and some were found in the current study – these were taken as is, since they constituted a small percentage of the total and did not appear to be non-randomly distributed.

These points are plotted against the 71,593 census tracts in 48 states plus DC. These tracts are used for two reasons. First, their centroids serve as the basis for buffers of different radii, within which bank densities are calculated. Second, these densities are compared against an index of economic deprivation proposed by Hegerty (2019), which uses US Census data (2015 ACS 5-year estimates) to combine five socioeconomic variables into a single measure.

In this study, bank densities are calculated as the number of bank locations per square mile, within the given radii from each tract centroid. For example, a tract with 20 banks in a 2-mile radius (which has an area of 12.57 square miles) would have 1.59 banks per 1 square mile. If, as the radius increases, the number of nearby banks changes at a rate different from the circle’s area, this density could increase or decrease. It is therefore possible to look for some sort of “optimal” radius for future analyses, as well as to use different measures for rural and urban areas. While densities are calculated at quarter-mile increments from 0.25 to 20.00 miles, the primary measures of analysis are the 2-, 5-, 10-, and 20-mile radii.

To control for different types of urban, suburban, and rural areas, this paper examines and controls for population density. After plotting bank density versus (log) population density and finding a distinct nonlinear relationship, a procedure similar to that of Kashian et al. (2018) is applied, which uses regression residuals as a measure of adjusted bank density. Log population densities are split into 10 equal segments (with unequal numbers of tracts, however) and separate OLS regressions are performed for each. The resulting residuals serve as a measure of “adjusted” bank density that can be used to measure relationships with a number of socioeconomic variables. For both the adjusted and the unadjusted bank densities, this paper focuses on the 5 and 10 percent thresholds, within each decile, using the four major measurement radii.

Finally, economic deprivation is compared in these low-density “banking deserts” against deprivation in the remaining tracts. Mean deprivation in the bottom 5 percent and top 95 percent of tracts within each population density decile are calculated and tested for significant differences using standard t-tests. Overall, the differences are largest and most significant in more “urban” tracts, generally at the 70th percentile of log population density or higher.

4 Results

Figure 7.1 shows that, as expected, banks are clustered in urban areas. In the sparsely populated Western states, there are relatively few banks, even in tracts with higher population densities. Most likely, many areas will have zero banks per square mile, even within a large radius, but tracts with higher population densities and no nearby banks are more likely to be considered to be “banking deserts,” since these values are lower than expected.

Figure 7.1
Figure 7.1

Bank locations (2019) and population density for US census tracts

Note: choropleth map of population density with natural breaks: Light = low density; black = higher

Source: Census Data, 2015 American Community Survey 5-year estimates and FDIC Summary of Deposits database

Figure 7.2 shows how the choice of measurement radius affects the calculation of bank density. The highest median bank density is calculated when a radius of around 2 miles is applied – the average from 1.75 to 2.25 miles is almost exactly 0.500. This density declines as the measurement radius increases; most likely, the resulting circles exceed the size of nearby urbanized areas, so as the radius increases, the rate of new banks falling within the buffer is less than the increase in total area. The 5 and 10 percent quantiles stabilize for radii larger than 5 miles, however.

Figure 7.2
Figure 7.2

Bank density quantiles at different measurement radii (in miles)

Source: Author’s elaboration

Table 7.1 presents summary statistics for this sample of (unadjusted) banking densities at the major radii used for this study, as well as for the index of economic deprivation. Since the median exceeds the mean in all cases, the data are skewed right.

Table 7.1
Table 7.1

Summary statistics for bank densities and deprivation index

Figure 7.3 plots bank densities (at 2 miles) against log population density; this relationship is nonlinear even when various additional transformations (not shown here) are applied. The highest decile (100th percentile) of log population density is typically composed of tracts in New York City, and the lowest (10th percentile) contains tracts in states such as Idaho and Wyoming. For comparison purposes, Chicago tracts are depicted in grey; these are typically located in the 80th and 90th percentile ranges. Table 7.2 presents statistics on the relative size, population, area, and population density of each decile. The 70th and 80th percentiles (7th and 8th deciles) contain the largest shares of tracts and population in a very small land area. The 40th to 60th percentiles are relatively less dense and might represent more small-town or “suburban” areas. Since these segments are density-, not geographically-based, they might also represent residential parts of a major city. This study, however, does not focus on specific locations. In fact, as is shown for Chicago, one city’s component tracts might fall across multiple density segments.

Figure 7.3
Figure 7.3

Bank density versus population density (2-mile measurement radius)

Notes:

Vertical lines: separators between density ranges; OLS regressions conducted for 10 separate groups

Grey line: OLS regression lines within each segment

Grey dots: tracts located in the city of Chicago

Source: Author’s elaboration
Table 7.2
Table 7.2

Population, area, and population density by tract group

Table 7.3 shows the distributions of bank densities within each segment, as well as for the entire sample. Many tracts have zero banks per square mile; while this is especially true for the 2- and 5-mile radii, it also applies for the least-dense deciles when a 20-mile radius is applied. A two-mile measurement radius, therefore, might be useful for a city such as Chicago, but it fails to find even a single bank in tracts at the 60th log population density percentile and below. The same can be said for the 5-mile radius (with a large share of zero-density tracts at the 50th percentile and below). The 10-and 20-mile radius see the same effect for the 30th and 20th percentiles, respectively.

Table 7.3
Table 7.3

Bank density at lower quantiles within density deciles and at different radii

Clearly, the choice of distance range matters among urban, suburban, and rural areas when calculating bank densities. Population density also affects the thresholds that can be used when defining “banking deserts.” In Table 7.3, the 5 and 10 percent quantile values are shown to differ across population density deciles. Chicago’s range (the 80th and 90th percentiles) match the findings of Hegerty (2020), who indirectly calculated a density of roughly 0.4 banks per square mile, using 1- and 2-mile radii, for the bottom nine percent of tracts. These findings are also confirmed for the 781 tracts in the city of Chicago, which are presented in Table 7.4, particularly when a 1-mile radius is applied. For more suburban ranges, using a 20-mile measurement radius, these quantiles fall in a range from 0.005 to 0.015 banks per square mile (which translates to between roughly 6 and 20 banks within this very large circle), and also aligns with the results of Kashian et al. (2018).

Table 7.4
Table 7.4

Lower quantile bank densities for tracts located in Chicago (N = 781)

Table 7.5 summarizes the thresholds for the lowest bank densities in each of four density categories. The lowest threshold in urban areas is double the threshold for “less urban” areas with population densities below 3,200 per square mile. The lowest 5 percent of rural tracts have even lower bank densities. Converted into the number of banks within a given radius, these figures show (for example) that to find nine banks, one might need a circle with a 2-mile radius in an “urban” area; five miles in a suburban area, and 20 miles in a rural area. These numbers can serve as a basis when assessing whether specific neighborhoods or areas can be classified as such a “desert.”

Table 7.5
Table 7.5

Summary for “banking desert” thresholds for selected population densities

The relationship between population density and bank density becomes more pronounced as the former variable increases. This is shown via the slope coefficients from a bivariate OLS regressions, within each decile. Depicted visually in Figure 7.3 for the 2-mile radius, they are generally significant at the 40th percentile or greater regardless of the measurement radius. For larger cities, we expect that even the harshest “desert” is expected to have at least a few banks nearby.

Table 7.6 shows large correlations between the adjusted and unadjusted bank-density measures, with the highest Spearman coefficients found around the 70th population-density percentile. Because they will have very few zero values at any radius, while at the same time differentiating between an area with no bank access within a populated area and a similar area in a sparsely populated one, the adjusted bank density measures are preferred when examining associations with other socioeconomic variables.

Table 7.6
Table 7.6

Statistics for regression errors and “bank deserts” at different radii

These adjusted values are used to compare differences in socioeconomic deprivation between “banking deserts” and non-deserts. Depicted graphically in Figure 7.4, the gaps between the bottom 5 percent of tracts and the remaining 95 percent are largest at the 80th percentile and above. The image of a socioeconomically deprived neighborhood with limited banking access, therefore, might hold mainly for large cities as Chicago. The t-tests in Table 7.6 show that, while small, these differences in mean deprivation scores for most tracts and density deciles are indeed significant. While differences are only significant for the 70th percentile and above within a 2-mile radius, they are significant for the 30th percentile and above when larger radii are used. Overall, areas with relatively few banks have significantly higher levels of economic deprivation than do non-deserts, but both groups’ levels – as well as the differences between them – are only large in highly urban areas.

Figure 7.4
Figure 7.4

Mean deprivation values, “bank desert” vs. non-desert block groups

Source: Author’s elaboration

5 Conclusion

While “banking deserts” are often examined by researchers and advocates who wish to increase communities’ access to financial services and limit the reach of high-interest alternative providers, little has been done to provide an exact definition of such an area. Combining the approaches of a recent distance-based analysis and a city-level density-based approach, this study calculates bank densities (number of branches per square mile) for Census tracts across the lower 48 states and the District of Columbia. Because banks’ service areas often exceed tract boundaries, these densities are calculated using a variety of measurement radii for each tract and centroid. Calculated densities, however, are sensitive to the choice of radii – a choice of two miles gives the highest median density nationwide and works well for large cities, while larger radii of 5, 10, or 20 miles work well in less-dense areas.

Bank density is compared against population density, but because the relationship is nonlinear, comparisons are made within ten deciles based on log population density. Linear regression for each segment suggests that this relationship is strongest for the highest-density, most urban, tracts. Regression analysis is also used to create an “adjusted” bank density measure, which controls for population density within each decile.

In line with the work of Kashian et al. (2018) and Hegerty (2020), this study defines thresholds to define “banking deserts” and which are based on the lowest 5 and 10 percent of unadjusted banking densities within each decile. These correspond to roughly 0.24 to 0.80 banks per square mile in highly urban tracts, 0.013 to 0.115 banks per square mile in less-urban tracts, and 0.004 to 0.14 banks per square mile in many rural tracts. These thresholds can be used to measure and compare bank access in communities across the country.

Further analysis, using the tract-level measure of Hegerty (2019), compares socioeconomic deprivation in the bottom five percent of tracts, as measured for adjusted bank density, with the remaining 95 percent. While most deciles’ differences are significant, particularly when radii of five miles are larger are applied, deprivation scores are highest and group differences are largest in the densest tracts. This suggests that more suburban and rural “banking deserts” might not face the same hardships as do their urban counterparts. This finding, which nonetheless is in line with the results mentioned above, requires further investigation.

The results presented here have clear implications for public health, especially in large US cities. Areas defined as “banking deserts” have limited financial access, which, as noted by Eisenberg-Guyot et al. (2018) above, is associated with lowered self-reported health status. The specific mechanism through which this occurs would be a worthy future study. In the near term, community leaders and organizations could more effectively target their resources to these known areas of the city.

Three specific targets stand out as being in particular need of these resources. First, large US cities (particularly with population densities in the range of 5,000 to 10,000 per square mile) are shown here to have unique limitations in terms of financial access. This group of cities include some of the poorest urban centers in the United States, such as Buffalo, Cleveland, Detroit, Milwaukee, and St. Louis; these are also plagued by health issues such as high infant and maternal mortality rates and shortened life expectancies. A city-level approach could concentrate on areas in the density ranges shown to have the highest population densities and density levels.

Secondly, such an approach could be focused on dense, high-deprivation areas of cities nationwide, regardless of overall population or city-level population density. Even a wealthy area might have neighborhoods with high deprivation scores and limited financial access; likely, public health outcomes are worse in these areas as well. Concentrating investment in the local level might improve all three indicators in neighborhoods that are often ignored.

The third target involves investigating the nexus between these three indicators, particularly in terms of the direction(s) of causation. If financial health indeed helps improve physical health, then strengthening community ties to the banking system may directly lead to improved public health outcomes and reduced deprivation. Specific channels may include lower stress levels and improved mental health among residents, as well as access to loans for neighborhood improvement or even increased ability to pay for healthcare treatments. As a more indirect channel, if bank access is a proxy for community investment and neighborhood cohesion, then encouraging these connections will help the vitality of neighborhoods across the country. Even if causation primarily runs in the other direction – where unhealthy neighborhoods are considered not to be profitable locations for bank branches – helping to improve financial access might result in a greater degree of overall investment in highly deprived neighborhoods.

These findings point to four additional research directions. First, the impact of bank branch closings can be assessed by incorporating additional years into a study. While FDIC data go back as far as 1994, the quality of their geographic information are limited, and a large percentage of address locations would need to be geocoded by the researcher. Carefully ensuring the comparability of data over time would allow for an analysis of which socioeconomic variables are most closely related to these closings. Second, a finer breakdown of density ranges, whether in terms of more groups, varying thresholds, or a more even distribution of tracts within groups, might also be useful. Third, incorporating geography – especially including place-level information for cities and suburbs – might also enhance the results. Finally, additional socioeconomic variables, including racial makeup, can be included in a multivariate model. The current results, however, provide an interesting look into how exactly one can define a “banking desert.”

References

  • Board of Governors of the Federal Reserve System. (2019). Perspectives from Main Street: Bank Branch Access in Rural Communities. Washington, DC.

  • Brennan, M., McGregor, M., & Buckland, J. (2011). The changing structure of inner-city retail banking: Examining bank branch and payday loan outlet locations in Winnipeg, 1980–2009. Canadian Journal of Urban Research, 20(1), 132.

    • Search Google Scholar
    • Export Citation
  • Cover, J., Spring, A. F., & Garshick, R. (2011). Minorities on the margins? The spatial organization of fringe banking services. Journal of Urban Affairs, 33(3), 317344.

    • Search Google Scholar
    • Export Citation
  • Degryse, H., & Ongena, S. (2005). Distance, lending relationships, and competition. Journal of Finance, 60(1), 231266.

  • Dunham, I. M., Foster, A., Graves, S., & Masucci, M. (2018). Navigating the dual financial service system: Neighborhood-level predictors of access to brick-and-mortar financial services. The California Geographer, 57.

    • Search Google Scholar
    • Export Citation
  • Eisenberg-Guyot, J., Firth, C., Klawitter, M., & Hajat, A. (2018). From payday loans to pawnshops: fringe banking, the unbanked, and health. Health Affairs, 37(3), 429437.

    • Search Google Scholar
    • Export Citation
  • Ergungor, O. E. (2010). Bank branch presence and access to credit in low- to moderate- income neighborhoods. Journal of Money, Credit and Banking, 42(7), 13211349.

    • Search Google Scholar
    • Export Citation
  • Fowler, C. S., Cover, J. K., & Kleit, R. G. (2014). The geography of fringe banking. Journal of Regional Science, 54(4), 688710.

  • Hegerty, S. W. (2016). Commercial bank locations and “banking deserts”: A statistical analysis of Milwaukee and Buffalo. Annals of Regional Science, 56, 253271.

    • Search Google Scholar
    • Export Citation
  • Hegerty, S. W. (2019). A Comparison of tract-level, nationwide indices of economic deprivation. Economics Bulletin, 39(1), 256263.

  • Hegerty, S. W. (2020) “Banking Deserts,” Bank Branch Losses, and Neighborhood Socioeconomic Characteristics in the City of Chicago: A Spatial and Statistical Analysis, The Professional Geographer, 72(2), 194205, DOI: 10.1080/00330124.2019.1676801.

    • Search Google Scholar
    • Export Citation
  • Kashian, R., Tao, R., & Drago, R. (2018). Bank deserts in the USA and the Great Recession: Geography and demographics. Journal of Economic Studies, 45(4), 691709.

    • Search Google Scholar
    • Export Citation
  • Kubrin, C., Squires, G. D., Graves, S. M., & Ousey, G. (2011). Does fringe banking exacerbate neighborhood crime rates? Investigating the social ecology of payday lending. Criminology and Public Policy, 10, 437466.

    • Search Google Scholar
    • Export Citation
  • Nguyen, H.-L. Q. (2019). Are credit markets still local? Evidence from bank branch closings. American Economic Journal: Applied Economics, 11(2), 132.

    • Search Google Scholar
    • Export Citation
  • Puchalski, V. A. (2019). Credit at the corner store: An analysis of resource exchange among Detroit-Area urban poor. Sociological Forum, 31(4), 10401062.

    • Search Google Scholar
    • Export Citation
  • Smith, T. E., Smith, M. M., & Wackes, J. (2008). Alternative financial service providers and the spatial void hypothesis. Regional Science and Urban Economics, 38, 205227.

    • Search Google Scholar
    • Export Citation
  • Wheatley, W. P. (2010). Economic and regional determinants of the location of payday lenders and banking institutions in Mississippi: Reconsidering the role of race and other factors in firm location. The Review of Regional Studies, 40(1), 5369.

    • Search Google Scholar
    • Export Citation
  • Woodstock Institute. (2019). Digitally divided: Older adult banking vulnerabilities in the Chicago region. https://woodstockinst.org/.

    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 46 46 24
PDF Views & Downloads 46 46 19