Austin Landini et al., Can Bank Density Provide Insights on the Appropriation of Disaster Relief Funds? Evidence from the Paycheck Protection Program (2024)
Summary
A research article by Austin Landini of the University of Missouri, Kristopher Deming of the Air Force Academy and Stephan Weiler of Colorado State University, Can Bank Density Provide Insights on the Appropriation of Disaster Relief Funds? Evidence from the Paycheck Protection Program, published in The Review of Regional Studies, Vol. 54, No. 1, 2024, pages 1–26. Revisiting Deming and Weiler (2023) with geocoded Paycheck Protection Program data mapped across three lending waves in a pooled cross-section of 2,112 commuting-zone observations, the study tests how local bank branch density relates to PPP loans per small business and loan amounts per small business. It finds a statistically significant positive relationship that is largest in the final round of lending and in high-loan-density commuting zones.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Austin Landini et al., Can Bank Density Provide Insights on the Appropriation of Disaster Relief Funds? Evidence from the Paycheck Protection Program (2024).”
Original at rrs.scholasticahq.com. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/austin-landini-et-al-can-bank-density-provide-insights-on-the-appropri-b2311a2d33/
, SHA-256 70795389fdf8fecf8acd92785a5efbb9b4ef4a8f0502dde44ffd3d776a860b7c.Full text
(2024) 54, 1–26
Can Bank Density Provide Insights on the
Appropriation of Disaster Relief Funds? Evidence from
the Paycheck Protection Program∗
Austin Landinia , Kristopher Demingb , and Stephan Weilerc
a Division of Applied Social Science, University of Missouri, USA
b Office of Labor and Economic Analysis, Air Force Academy, USA
c Department of Economics and Regional Economic Development Institute,
Colorado State University, USA
Abstract: In response to the COVID-19 pandemic, Congress in conjunction with the Small Business
Administration (SBA) created the Paycheck Protection Program (PPP) to help small business owners retain
employees. We improve on previous literature exploring the relationship between bank density and PPP loan
dispersion by implementing precisely geocoded data to map loans to commuting zones. Revisiting results
in Deming and Weiler (2023), we confirm the statistically significant relationship between bank density and
PPP loans per small business. We consider three distinct periods of lending to analyze how changes in
the PPP loan program rules might have impacted dispersion. The coefficient in the regression of loans per
small business on bank density is largest in the final wave of lending, and in commuting zones with high
loan densities. This suggests that having additional bank branches in the commuting zone may have helped
facilitate rapid dispersion of relief funds to high demand areas during the program.
Keywords: banking; relationship lending; geography; CARES Act; regional
JEL Codes: H25, R12, R51
1. INTRODUCTION
The COVID-19 pandemic created the first official economic downturn since the Great Re-
cession. In response, Congress created the Paycheck Protection Program (PPP) to help
∗
The views expressed in this article are those of the authors and do not necessarily reflect the official policy
or position of the United States Air Force Academy, the Air Force, the Department of Defense, or the U.S.
Government. Approved for public release: distribution unlimited. PA number: USAFA-DF-2023-753.
The authors would like to thank the editor and referees of the Review of Regional Studies for their critical
feedback.
Corresponding Author: Austin Landini, E-mail: Austin.Landini@missouri.edu
(c) Southern Regional Science Association 2024
ISSN 1553-0892, 0048-49X (online)
www.srsa.org/rrs
2 The Review of Regional Studies 54(1)
small business owners retain employees during periods of reduced mobility and sales. The
PPP loans program allowed small business to obtain low-interest loans to cover payroll and
other expenses, while eliminating liability for the loan originators, existing Small Business
Administration (SBA) 7(a) lenders.
On March 27, 2020, Congress signed the CARES act into law, allocating $349 billion
for the PPP program. Applications opened on April 3rd of that year, and the following
day for independent contractors and self employed individuals. Within less than two weeks,
the inital allocation of funding had been tapped out leading to a temporary lull in the
program. On April 24th, Congress approved an additional $310 billion for the program
with a deadline date of June 30, ultimately extended to August 8. At the August deadline,
over $130 billion of allocated funds remained unclaimed and, after a hiatus in Fall 2020,
Congress again extended the program, with additional funding resuming in January 2021.
By the conclusion of the program in late May 2021, nearly $800 billion in funding had been
disbursed as PPP loans under the CARES act program.
The Paycheck Protection Program has already prompted extensive literature covering
optimal allocation theory (Joaquim and Netto, 2022), bank performance (Granja et al.,
2022), flows to minority communities (Fairlie and Fossen, 2022), business survival (Bartik
et al., 2020), the impact on employment, e.g., Joaquim and Netto (2020) and Kapinos (2021)
and on GDP (Autor et al., 2022). Yet the question of local banking relationships and access to
the PPP program funding is particularly important for this business support policy. Margins
on PPP loans were very small (Marsh et al., 2020). Banks thus had extra incentive to rely
on soft information about the borrower, which is more transparent in the presence of local
banking relationships.
Banking services vary regionally, creating potential inequality in the ability to access
credit. While many banking services can be transacted remotely via banking and a rising
fintech sector, there are still public and personal benefits derived from proximity to a bank
branch or headquarters (Deming and Weiler, 2023). Local banks are able to utilize relation-
ship lending, based on soft information gleaned through business or personal networks for
credit decisions, e.g., Petach et al. (2021) and Berger et al. (2022). For certain borrowers,
such as those located in rural areas or banking deserts within urban areas, this soft informa-
tion is opaque, generating a potential for imbalance in borrowing opportunity based solely
on geography, e.g., Akerlof (1970), Bunten et al. (2015), and Conroy et al. (2017). While the
notion of banking deserts is anecdotally rich, there are few independent empirical analyses of
the extent of the geographically based banking access problem. Previous literature suggests
that the inequality is most pronounced between urban and rural areas because the travel
time from rural areas to banking locations is often much longer than in urban areas, e.g.,
Kashian et al. (2015) and Morgan et al. (2018).
A few recent publications address the geospatial relationship between banking access and
relationship-based lending. Firms which receive PPP loans from lenders with whom they
have had a past relationship receive larger loans and more quickly (Rabetti, 2022), while
firms with personal ties to banks are more likely overall to receive PPP loans (Duchin et al.,
2022). Smaller community banks responded more quickly to PPP loan requests and lent
more intensively to small businesses than larger banks (James et al., 2021). Because small
firms usually must depend on financial intermediaries such as commercial banks to obtain
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 3
credit , relationship lending or the geospatial lack theoreof may be particularly important
for small businesses. Small businesses were particularly fragile during the COVID-19 crisis
(Bartik et al., 2020), meaning that for many, firm survival could have depended on access to
PPP funds. Borrowers located geographically near more active PPP lenders received credit
earlier (Glancy, 2021), suggesting a potential geographic inequality in access to PPP funds.
Previous literature addresses this potential inequality in access to important banking
programs. Relationship lending has the potential to enhance disadvantageous or discrimi-
natory banking practices. Smaller businesses in general had lower awareness of government
programs during COVID-19, were less likely to apply for PPP funding, and were less likely to
report having their application approved (Humphries et al., 2020). PPP funding flowed less
rapidly to minority communities (Fairlie and Fossen, 2022). Black-owned businesses were
less likely to receive PPP loans (Chernenko and Scharfstein, 2022) and received loans that
were smaller than White-owned businesses. This effect is marginally smaller in areas with
more bank competition and disappears in the later stages of PPP which allowed for entry
by fintechs and other non-traditional lenders (Atkins et al., 2022). Together, these results
suggest that further research is warranted into the geographic relationship between banking
locations and the dispersion of PPP loans. Because data on PPP loans were released with
geographic information at the ZIP code level, previous research has commonly approximated
loans from ZIP code to county. A small number of more recent papers have used GIS pro-
grams to geocode the addresses found in the PPP loan dataset to levels of analysis which
coincide more favorably with publicly available data, e.g., Borawski and Schweitzer (2021)
and Glancy (2021).
This paper makes three main contributions to the literature. First, we revisit previous
results from Deming and Weiler (2023) which find a significant positive relationship between
the density of banking locations and PPP loans, with the geocoded data update. In doing so,
we confirm the significant and positive relationship between bank density and PPP loans per
small business. Second, we extend upon previous literature by considering loans issued over
the entirety of the PPP program, and decomposing those loans into three distinct lending
periods based on a timeline of actions taken by the SBA. We find that while there is a sig-
nificant and positive relationship between bank and loan density across all loan periods, this
relationship is strongest in the final lending period. Third, this paper is novel in considering
the possibility that competitive pressures associated with bank location and the origination
fee structure of the loan program could have played a part in explaining geographic pat-
terns of high loan density. We show that the relationship between bank and loan density is
the most intense in commuting zones which received the largest number of loans per small
business. In light of literature suggesting that smaller banks originated larger than expected
shares of the loans, we suggest that loan origination fee structure might have driven this
trend in areas with high bank density, and therefore high competition.
The remainder of the paper is divided into the following sections: Section 2 describes our
data sources and how the data was matched to the appropriate geographical areas. Section
3 presents our findings on the relationship between bank density and PPP loan dispersion.
Within this section, we study 1) loans and loan amounts per small business, 2) changes in
lending patterns over time and 3) interquantile changes in the intensity of the relationship
between bank and loan density. Section 4 concludes.
©Southern Regional Science Association 2024.
4 The Review of Regional Studies 54(1)
2. DATA
Data on Paycheck Protection Program loans come from the Small Business Administration
(SBA) and were coded into Census Blocks, Tracts, Counties, and States by Geocodio and
made public by Glancy (2021). These data contain information on all individual loans
distributed through the PPP from the first phase in April of 2020 until the official end of
the program on May 31, 2021. For each loan, the date of approval, loan amount, business
address, reported number of jobs, and demographic characteristics of the business owner are
observed. However a large number of loans are missing information from the demographic
survey.
PPP loan data is analyzed in three waves, the first spanning from the introduction of the
program to the initial exhaustion of the funds on April 16, a second wave from the restart
of the program to the initial revised deadline of August 8, 2020, and a final wave spanning
to the end of the program. Analysis of the different waves of the PPP loan program will
allow for discussion of whether changes to the program, such as allowing fintechs and other
non-traditional lenders to originate the loans in the third wave, impacted the dispersion of
PPP loans.
Commuting Zones (CZs), first developed by Tolbert and Sizer (1996), and designated
by the U.S. Department of Agriculture Economic Research Service (ERS)1 are the primary
units of analysis. CZs offer two main advantages over other geographic delineations. First
CZs represent local economies better than other political boundaries by grouping counties
together which have strong commuting ties. Second, CZs contiguously cover the continental
United States meaning that all businesses which received a PPP loan will be retained in
the sample. Commuting Zones also lend themselves particularly well to studying the distri-
bution of PPP loans because small businesses (especially rural) may have to seek banking
services outside of their city or county due to a lack of banking access (Deming and Weiler,
2023). Using Commuting Zones as the focal unit of analysis will more accurately capture the
number of banks available to small businesses. PPP loan data at the Census County level is
aggregated to Commuting Zones by matching the geocoded county of business location to a
crosswalk provided by the ERS.
Commuting zones are geographically diverse and do not have identical populations or
economies. For this reason, we normalize our outcomes of interest by the number of small
businesses in each CZ. We focus on the number of small businesses in each CZ since the
Paycheck Protection Program was designed to provide economic relief for small businesses.
The Small Business Administration definition of a small business varies by industry, so when
normalizing covariates by the number of small businesses, we utilize a more general defini-
tion of a small business adopted by the Small Business Administration — a business with
fewer than 500 employees. To get the number of businesses with fewer than 500 employees,
we use data from the most recent County Business Patterns from the U.S. Census Bureau,
aggregated to the Commuting Zone level. Previous literature approximates loan data into
Commuting Zones based on ZIP code weighting. The geocoded PPP data allows for more
precise analysis of loan dispersion. We find, in a comparison between the ZIP code weighted
1
2000 designations
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 5
and precisely geocoded datasets, that the two are highly correlated. Weighting is a poten-
tially reasonable approach to converting ZIP code data to Commuting Zone level. However,
the precisely geocoded data does eliminate some zeros in the approximated data which were
likely erroneous. As a result, the implementation of GIS procedures to improve the accuracy
of geospatial estimation is an important addition to the literature on banking and PPP loans.
Our main source of commuting zone-level economic characteristics, such as population,
household income, demographic characteristics, level of education, and land area, is from
the American Community Survey 5-year estimates for 2015-2019. This source provides data
on these characteristics at the county level which are then aggregated up to the CZ level.
Data on county GDP comes from the Bureau of Economic Analysis Regional Economic
Accounts and summed up to CZ GDP. These data are for 2019 since that is the most recent
data available at the county level, and effectively set the stage for the 2020 pandemic that
followed. Following Liu and Volker (2020), data on the number of COVID-19 cases comes
from The New York Times which continuously tracked the number of daily COVID-19 cases
by county from the beginning of the pandemic. To accurately measure the impact of COVID-
19 cases and potential local lock-down measures in an area, we use the cumulative count
COVID-19 cases for each county on April 3, 2020, which is when the program opened for
small businesses. The county case count is then aggregated up to the commuting zone. We
get our data for labor market outcomes from the Bureau of Labor Statistics’ (BLS) Local
Area Unemployment Statistics. Since the BLS measures employment in first two weeks of
the month, we use the county data on employment from April 2020 since these data reflect
the reality of local labor markets when business establishments were deciding whether to
apply for a PPP loan.
We obtain our data on bank location from two sources: Federal Deposit Insurance Cor-
poration (FDIC) and the National Credit Union Administration (NCUA). We focus on both
banks and credit unions because both were authorized to provide PPP loans to small busi-
nesses. We obtain the location of all bank branches and headquarters from the FDIC’s Insti-
tutions and Locations database. This provides the addresses of all federally insured banks,
county of banks location, the service type of banks, and whether the bank is the main office
or a branch location. We create a measure of total banks by including all full-service banks,
both brick and mortar and retail locations, as well as permanent limited-service banks that
only accept deposits and payments. We include the latter type of banks to capture the
effect of banking hinterlands, or commuting zones without any main bank offices, on the
distribution of PPP loans. The county data is again aggregated up to the CZ level. The
NCUA Quarterly Call Report Data for the second quarter of 2020 are used to get a list of
all credit unions by location. Credit Unions are only listed by address without their county
of residence, so we follow a similar zip code to county procedure as with the PPP loan data
before we aggregate these to the commuting zone. The number of credit unions are combined
with the number of banks to create a total measure of bank concentration for each CZ.
Buyers of loans are most likely to choose banks that are either close to home or workplace.
In terms of the status of the local economy, we focus on the employment/population ratio as
an indicator of job opportunity available to the region’s population following the significant
downturn in March/April 2020, as well as the proportion of citizens with a higher-education
degree and per capita GDP. Amior and Manning underscore the employment/population ra-
©Southern Regional Science Association 2024.
6 The Review of Regional Studies 54(1)
Table 1: Summary Statistics
mean sd min max count
PPP Loans per Small Business 1.83 1.05 0.34 10.19 706
PPP Loan Amount per Small Business 84842.48 22485.98 12751.05 216613.64 706
Rounds 1+2 Loans per Small Business 0.68 0.20 0.18 1.91 706
Round 3 Loans per Small Business 1.15 0.89 0.12 8.76 706
Rounds 1+2 Loan Amount per Small Business 52574.19 13644.58 8425.66 134863.17 706
Round 3 Loan Amount per Small Business 32268.30 13926.76 4325.39 148088.13 706
Banks + C.U.s per 10k people 4.49 2.91 0.00 37.83 706
Bank Headquarters per 10k people 0.64 0.97 0.00 9.65 706
COVID-19 Cases per 10k people 2.81 5.76 0.00 88.29 706
Employment
P opulation
0.45 0.06 0.24 0.83 706
M edianHH.Inc.
M eanHH.Inc.
0.75 0.04 0.60 0.90 706
GDP per Capita (Thousands) 54.04 25.50 22.77 392.17 706
Population Density (Thousands/sq.mi) 0.01 0.03 0.00 0.75 704
Share of Pop. with at least a B.A. 0.24 0.08 0.07 0.56 706
Non-White Population Share 0.18 0.16 0.01 0.89 706
Small Businesses per 10k people 252.72 88.48 48.99 1097.14 706
Share of Small Businesses <10 Employees 0.75 0.05 0.62 1.00 706
Share of Small Businesses <50 Employees 0.96 0.02 0.92 1.00 706
tio as being a particular appropriate measure of economic opportunity (Amior and Manning,
2018), which we leverage in this work.
Table 1 presents summary statistics for each of the variables in our sample. The final
sample consists of 706 commuting zones and over 11.7 million individual loans. The third
wave of PPP loans allowed for small businesses to reapply for a second and in many cases a
third PPP loan. As a result, we observe that in the final wave, there are a large number of
loans originated, but at a lower average loan amount than in the initial waves.
To account for population differences across CZs, we normalize the number of banks,
credit unions, and bank headquarters by the number of small business in a CZ. We also
normalize COVID-19 cases and the number of small businesses by CZ population per ten
thousand in regional population, in the spirit of Amior and Manning (2018).
Figure 1 maps the geographic dispersion of PPP loans per SBA small business. Interest-
ingly, the highest concentrations of PPP loans per small business occurred throughout the
Great Plains, Midwestern and Southern states.
Throughout this work, we compare the density of PPP loans to the density of bank and
credit union locations per 10k population. Figure 2 maps the concentration of bank and
credit union branches across United States Commuting Zones. There appears to be a high
degree of spatial correlation between the areas with a high density of lenders and areas with
a high density of PPP loans. To account for potential spatial clustering, our models to follow
include both White (1980) robust standard errors (listed in parenthesis), and Conley (1999)
spatially robust errors [listed in brackets].
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 7
Figure 1: Density of PPP Loans per Small Business by County
Figure 2: Density of Bank Locations per 10K Population by County
©Southern Regional Science Association 2024.
8 The Review of Regional Studies 54(1)
3. BANK DENSITY AND PPP LOAN ALLOCATION
Our approach follows clues left by the handful of studies on PPP loan distribution. Ami-
ram and Rabetti (2020) and Li and Strahan (2021) find those establishments with existing
banking relationships tended to get loans first and in the largest amounts. We indirectly test
both propositions in the present work, in particular through the resource flows channeled to-
wards the smallest businesses which are less likely to have established banking relationships.
Granja et al. (2022) further find that those relationships tended to outweigh stated goals of
the program, namely targeting those areas and businesses in greatest need of loans due to
the pandemic.
Barrios et al. (2020) find that establishment payrolls closely predict PPP loan receipts,
indicating there may be a positive relationship between business size and loan disbursement.
We test this proposition empirically as well, but do not find significant effects among small
businesses of less than 10 or less than 50 employees. Finally, Fairlie and Fossen (2022)
find that early loans went mainly to non-minority applicants, while the later waves flowed
more to these marginalized populations. We follow their lead in testing the significance on
non-white shares of commuting zone populations on loan disbursement. Including the non-
white population share likely does not control for banking discrimination, but instead picks
up more broadly on forms of discrimination may both limit banking opportunities within
a Commuting Zones and cause economically marginalized groups to migrate elsewhere. To
minimize concerns about collider bias, we present results in Tables A1-A3 without covariates.
The inclusion of the non-white population share does not on its own alter the interpretation
of the main relationship of interest, between bank density and loan dispersion.
The implications of systematic informational asymmetries based on geography can funda-
mentally shift innovation and resources away from lagging regions, further entrenching their
economic struggles (Weiler, 2000). These geographic information asymmetries (GIA) are
most likely to be felt in business-to-business transactions built on the supplier’s understand-
ing of those demanding services. Small business lending may be particularly vulnerable to
GIA discrimination, given its reliance on credit scoring developed at a bank’s headquarters –
which may not be congruent with the realities of a rural economy – as well the past viability
of similar projects. The latter will be a particularly high hurdle for innovative projects that
have no track record in the focal economy. Rural areas tend to have thin informational mar-
kets due to lower establishment dynamism and thus fewer datapoints from which to extract
the viability of loans (Bunten et al., 2015).
Statistically, the perception of otherwise identical probability distribution of outcomes
in two regions will have lower variance in the market with thicker information through
greater past business experience. In contrast, the thin market featuring fewer datapoints
will lead to higher perceived variance of outcomes, heightening uncertainty and thus risk for
bankers (Weiler et al., 2006). These risks may deter bankers from making loans to those
companies without existing intensive relationships, leading to disproportionate flows towards
advantaged regions and businesses. These informational asymmetries may thus be a driving
force for systematic discrimination of PPP loans towards those thick-market regions that
have denser labor markets as measured by the employment/population ratio, established
banking networks, and larger establishments. Our empirical work tests these propositions.
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 9
This paper investigates whether the distribution of banks and banking hinterlands par-
tially determined the distribution of PPP loans. Since the PPP loans were distributed
through banks and a limited amount of initial funds, we hypothesize that regions with a
greater concentration of banks as well as bank headquarters received more PPP loans, and
that banking deserts and banking hinterlands were systematically disadvantaged in PPP
allocations. We take the geographic distribution of banks as being given a priori, and thus
exogenous to the analysis. A higher concentration of banks provides more opportunities
for small businesses to find a bank that had not already exhausted their allotment of PPP
funding.
We are specifically also interested in whether loans went to the small businesses that were
avowedly priorities for the program. To empirically test these hypotheses, we estimate the
following equation:
Yi = β0 + β1 Bank Concentrationi + β2 Bank HQi + β4 COVID-19 Casesi + ΓXi + ϵ (1)
Where Yi are the outcome variables of interest, in this case PPP loans per small business
and PPP loan amount per small business at the Commuting Zone level, X is a matrix of
Commuting Zone covariates, and Γ is a vector of coefficients.
Bank Concentration is the number of banks and credit unions per 10,000 people in CZ
i, Bank HQ is the number bank main offices per 10,000 people in CZi, COVID case count
at initiation of the CARES act is the commuting zone number of COVID-19 cases per
10,000 people from New York Times counts. The matrix X contains the share of non-
White population in CZ i, the CZ-level share of businesses that have fewer than 50 or fewer
than 10 employees out of all small businesses, the commuting zone-level employment-to-
population ratio for April 2020, a commuting zone-level measure of inequality measuring the
ratio of household median income to household mean income, the per capita GDP of CZ,
and population density measured as total population divided by land area in square miles.
i, and the share of the CZ population that has at least a bachelor’s degree in CZ i. ϵi is an
idiosyncratic error term.
Variance Inflation Factor (VIF) analyses indicate that there was no substantive multi-
collinearity among the explanatory variables. We fully acknowledge that the cross-sectional
dataset can offer only a limited view of PPP impacts, given that the data show only trans-
acted loans. The lack of application data would tend to upwardly bias the findings by
overstating the effect of local banks, making our results an upper-bound of the PPP effect.
Coefficients β1 and β2 are our main coefficients of interest. Consistent with our banking
desert and banking hinterland hypotheses, we expect both coefficients to be positive for all
outcomes of interest, as more loans flow to those areas with more banks and more bank
headquarters rather towards banking deserts and hinterlands.
We are also interested in a question of intensity. In the areas with the most PPP loans
per small business, is the relationship between of bank density and loan density different
than in areas which recieved fewer loans per business? To answer this question, we first look
at the impact of squared bank location density.
©Southern Regional Science Association 2024.
10 The Review of Regional Studies 54(1)
3.1. Loans and Loan Amounts per SBA Small Business
Table 2 presents results from an OLS model with both Huber-White robust and Conley
(1999) spatially robust standard errors. Columns 1 and 2 report models for the number of
PPP loans per small business with and without Commuting Zone controls, while Columns
3 and 4 report specifications with PPP loan amount per small business as the independent
variable. We find a positive and significant relationship between both independent variables
of interest: bank locations and bank headquarters, and both outcomes of interest: loans
and loan amount per small business. These results are robust to inverse distance spatial
weighting with a distance cutoff of two degrees latitude, or about 140 miles.
We find that an increase of one bank location (22% increase from the mean) per 10,000
people is associated with an increase of about .33 PPP loans per small business (18% increase
from the mean), and an increase in total loan amount of $4775 per small business (6% increase
from the mean) in the Commuting Zone. We are also interested in whether communities
at need as defined by exposure to COVID-19 received more loans. We find that COVID-19
case load was not associated with the number of loans in the Commuting Zone, but was
associated with a larger loan amount per small business.
In columns 2 and 4 of Table 2 we also consider the intensity of banking services in a
commuting zone. We add a squared banking density term to Equation 1 to further investigate
how the intensity of bank concentration relates to the distribution of PPP loans. This allows
us to measure whether there are increasing, constant, or diminishing returns to increased
banking intensity. A positive (negative) coefficient on the squared term would indicate
increasing (decreasing) returns to greater concentrations of banks while a zero or null effect
may indicate constant returns. In the presence of decreasing returns, we can also calculate
the level at which an increased concentration of banking services no longer associated with
an increase in PPP loans. We report the results of this inclusion in Table 2 and Table 3.
Table 2 shows that there is an association between areas with a greater concentration of
banks and credit unions and areas that received more PPP loans. We do find, by including the
square of the banking variable, that bank concentration exhibits diminishing returns. Once
the concentration of banks and credit unions is greater than 24.93 banks per 10,000 people,
which is around the 70th percentile, the number of PPP loans is negatively associated with
additional banks in the commuting zone. Even before that level, each additional increase
in the concentration of banks associated with smaller increases in the number PPP loans,
however, diminishing returns to additional banks do not seem to set in until higher percentiles
of density. This suggests that in low bank concentration areas, an additional bank is helpful
for small businesses applying for PPP loans, but are less helpful in areas with an abundant
number of banks.
The same pattern holds when examining the number of loans per small business across
the three waves of PPP funding in Table 3. Increases in banking concentration exhibit
diminishing returns for each of the three rounds of PPP funding, though the level at which
an additional bank becomes detrimental in obtaining a PPP loan are are different for each
round. The benefits of an additional bank reach their peak at 18.92 banks per 10,000 people
in round 1, 28.5 banks per 10,000 people in round 2, and 34.61 banks per 10,000 people
in the final round of funding. The changes in the amount of funding for each round may
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 11
Table 2: PPP Loans and PPP Loan Amount per Small Business
(1) (2) (3) (4)
PPP Loans PPP Loans PPP Amount ($K) PPP Amount ($K)
Banks + C.U.s per 10k people 0.1447 0.3299 1.0106 4.7754
(0.0269)*** (0.0323)*** (0.4586)** (0.8487)***
[0.0290]*** [0.0358]*** [0.4390]** [0.8155]***
Bank Headquarters per 10k people 0.4415 0.2693 6.1301 5.7286
(0.0908)*** (0.0697)*** (1.9084)*** (1.8772)***
[0.0975]*** [0.0760]*** [1.8080]*** [1.6960]***
(Banks + C.U.s per 10k)2 -0.0054 -0.1047
(0.0009)*** (0.0189)***
[0.0009]*** [0.0196]***
COVID-19 Cases per 10k people 0.0014 0.2905
(0.0046) (0.1585)*
[0.0054] [0.1754]*
Employment/Pop 4.6702 128.8527
(0.7634)*** (16.6494)***
[0.8241]*** [18.4694]***
Inequality -0.6198 25.3039
(0.8274) (20.6096)
[0.8358] [21.8599]
GDP per Capita in Thousands -0.0031 0.0925
(0.0011)*** (0.0713)
[0.0012]*** [0.0725]
Non-White Population Share 2.4094 48.5135
(0.2851)*** (6.0884)***
[0.3762]*** [7.0908]***
Small Businesses per 10k people -0.0040 -0.0210
(0.0007)*** (0.0158)
[0.0007]*** [0.0175]
Share of Small Businesses <10 Employees 2.4628 -120.5275
(2.3446) (48.4854)**
[2.4167] [46.1229]***
Share of Small Businesses <50 Employees 4.2518 -113.2127
(7.0712) (142.1779)
[7.6797] [148.8537]
Population Density -0.1073 16.0332
(0.4333) (31.9671)
[0.5005] [31.0523]
PPP Loans 0.0003 0.0002
(0.0001)*** (0.0001)**
[0.0001]*** [0.0001]***
Constant 0.8951 -6.5208 73.9047 173.5806
(0.0818)*** (5.3929) (1.7573)*** (108.4098)
[0.1200]*** [6.0306] [2.0878]*** [119.6468]
Observations 704 704 704 704
R2 0.525 0.693 0.190 0.495
F 107.34 66.46 20.63 51.01
Note: Huber-White standard errors in parentheses. Conley (1999) Standard errors are in brackets. A
cutoff distance of 2 degrees (222.28 km) was used for both latitude and longitude. ∗ p < 0.10, ∗ ∗ p <
0.05, and ∗ ∗ ∗ p < 0.01.
©Southern Regional Science Association 2024.
12 The Review of Regional Studies 54(1)
be responsible for the changing inflection point at which there is no longer a benefit to an
additional bank in the disbursement of PPP loans. With more funding available in the
second and third rounds, more banks would ease congestion and make it easier for more
small businesses to obtain help through the program. Despite only three commuting zones
being affected by any of these concentration thresholds, these result suggest that areas with
a higher concentration of banks received more PPP loans but that each additional bank had
less of an impact in increasing the total number of loans allocated to that place.
As a robustness check, we normalize our bank + credit union and headquarters density
measures per SBA small business as Appendix Table A4. Rescaling bank density per small
business may better capture local competition in the commuting zone’s supply of financial
capital. The number of banks + credit unions per small business (0.017) is very small
compared to per 10k population (4.49). As a result, the interpretation of a change in one
banking location per small business is less clean. However, we do confirm that the positive
and signficant relationship between bank, headquarters locations and loan density holds.
3.2. Timing Across PPP Loan Waves
Because we may be interested in the progression of the relationship between bank locations,
headquarters and PPP loan dispersion, Tables 3 and 4 present the results on PPP loans
and PPP loan amount per capita, broken down into each of the three waves of PPP loans
defined in the preceding section. For ease of reading, we present the model with controls in
all three periods in the main text. We also include in the Appendix the same models without
controls as Tables A1 and A2. With regard to PPP loans per small business, we find that the
positive and significant relationship with bank and credit union locations per 10,000 people
holds across all three loan periods, but is much stronger in the final wave, post- August
8, 2020. The same holds for bank headquarters locations with a positive and significant
coefficient across all three periods, strongest in the last. In the PPP loan amounts model,
we find that a positive and significant relationship with bank density holds across all three
lending periods. However, a positive and significant relationship between bank headquarters
and loan amounts holds only in the final wave of lending.
These results are closely aligned with a general trend of PPP lending across waves, which
was that initial periods saw relatively fewer loans issued of larger amounts, while later periods
included larger volumes of lower value loans. These trends in part reflect changes in eligibility
requirements for the loans. For example, nonemployer businesses which make up about 80%
of all businesses in the United States, were largely ineligible until the second round of lending.
As the pandemic continued from Summer 2020 into Spring 2021, additional emphasis was
placed on distributing relief funds to at-need areas as quickly as possible, which helps explain
the high loan densities in the final rounds of lending. Businesses were also eligible to re-apply
for a second and in some cases third round of loans in the later period, which could also
inflate loan density per small business.
To further isolate the potential role of bank density in explaining PPP loan density, we
pool the three waves together with Commuting Zone fixed effects as a repeated cross section
in Appendix Table A5. Rules around who could service the loans and which businesses
were eligible to apply and reapply shifted between the first and third waves, as did the
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 13
Figure 3: Graph of Quantile Regression Coefficients
pool of workers who were deemed as ‘essential’. To test differences across lending waves we
control for potential unobserved time invariant factors across Commuting Zones, or spurious
correlation with other local conditions, which could have been incorrectly attributed to bank
density in Tables 2-4. We find that the coefficient on both bank and headquarters density
in the loan density model is significantly higher in the final round of lending, confirming the
findings in Table 3. Likewise, in comparison to Table 4, we find that although loan amounts
were highest in the first round of lending, the impact of bank and headquarters density is
strongest in the final wave of lending.
3.3. Regression on Quantiles of Loan Dispersion
Returning again to the question of the intensity of the relationship between banking density
and loan density, we implement a quantile regression approach on quantiles of loans per small
business and loan amount per small business, estimated by minimizing an asymmetrically
weighted sum of absolute errors (Koenker and Hallock, 2001).
Figure 3 maps the coefficient in the regression equation (1) across all possible quantiles
of the outcome variables: loans per small business and loan amount per small business.
In Commuting Zones which received the highest density of loans per small business, the
coefficient in the loan density model is the largest. That is, as the density of PPP loans per
small business increases, the relationship between loans per small business and bank density,
as measured by a quantile regression framework, becomes more statistically differentiable
from zero.
The quantile regression methodology is useful for understanding relationships between
variables outside of their conditional means. To test how the relationship between banking
locations and PPP loans varies across quantiles of PPP loans, in Table 5 we present the
results of interquantile regressions at the (.6 vs .4), (.75 vs .25), and (.9 vs .1) quantiles.
©Southern Regional Science Association 2024.
14 The Review of Regional Studies 54(1)
Table 3: PPP Loans per Small Business by Round
(1) (2) (3)
Round 1 Loans Round 2 Loans Round 3 Loans
Banks + C.U.s per 10k people 0.0454 0.0285 0.2561
(0.0047)*** (0.0052)*** (0.0307)***
[0.0042]*** [0.0060]*** [0.0342]***
Bank Headquarters per 10k people 0.0139 0.0194 0.2360
(0.0094) (0.0084)** (0.0683)***
[0.0079]* [0.0074]*** [0.0765]***
(Banks + C.U.s per 10k)2 -0.0012 -0.0005 -0.0037
(0.0001)*** (0.0002)** (0.0009)***
[0.0001]*** [0.0002]** [0.0009]***
COVID-19 Cases per 10k people -0.0010 0.0017 0.0007
(0.0005)** (0.0007)** (0.0041)
[0.0005]** [0.0008]** [0.0048]
Employment/Pop 0.4540 0.6294 3.5868
(0.1027)*** (0.0824)*** (0.7383)***
[0.1069]*** [0.0916]*** [0.7897]***
Inequality 0.1748 -0.1382 -0.6564
(0.1106) (0.1086) (0.7686)
[0.1372] [0.1202] [0.7548]
GDP per Capita in Thousands -0.0005 0.0001 -0.0027
(0.0002)*** (0.0003) (0.0010)***
[0.0002]*** [0.0003] [0.0011]**
Non-White Population Share 0.0623 0.2307 2.1165
(0.0271)** (0.0319)*** (0.2622)***
[0.0329]* [0.0354]*** [0.3501]***
Small Businesses per 10k people -0.0001 -0.0003 -0.0036
(0.0001) (0.0001)*** (0.0007)***
[0.0001] [0.0001]*** [0.0007]***
Share of Small Businesses <10 Employees -0.4115 0.5900 2.2843
(0.1660)** (0.2466)** (2.2017)
[0.1729]** [0.2510]** [2.2530]
Share of Small Businesses <50 Employees 0.6788 -0.3737 3.9466
(0.4702) (0.7267) (6.5131)
[0.5015] [0.7571] [7.1006]
Population Density -0.3134 0.4393 -0.2332
(0.1771)* (0.2358)* (0.4008)
[0.1717]* [0.2305]* [0.4504]
Constant -0.5228 0.0096 -6.0076
(0.4152) (0.5551) (4.9337)
[0.4582] [0.5928] [5.5567]
Observations 704 704 704
R2 0.565 0.429 0.661
F 54.75 27.97 52.12
Note: Huber-White standard errors in parentheses. Conley (1999) Standard errors are in brackets. A
cutoff distance of 2 degrees (222.28 km) was used for both latitude and longitude. ∗ p < 0.10, ∗ ∗ p <
0.05, and ∗ ∗ ∗ p < 0.01.
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 15
Table 4: PPP Loan Amount per Small Business by Round
(1) (2) (3)
Round 1 Amount ($K) Round 2 Amount ($K) Round 3 Amount ($K)
Banks + C.U.s per 10k people 2.7781 -1.4122 3.2945
(0.4932)*** (0.2903)*** (0.5709)***
[0.4820]*** [0.3218]*** [0.6025]***
Bank Headquarters per 10k people 0.6309 0.6270 4.4323
(1.1302) (0.4635) (1.2457)***
[1.1179] [0.5223] [1.2384]***
(Banks + C.U.s per 10k)2 -0.0797 0.0349 -0.0571
(0.0121)*** (0.0132)*** (0.0141)***
[0.0118]*** [0.0137]** [0.0150]***
COVID-19 Cases per 10k people 0.0199 0.1287 0.1358
(0.0531) (0.0424)*** (0.0915)
[0.0596] [0.0471]*** [0.1022]
Employment/Pop 53.5796 15.7834 59.5934
(10.3787)*** (5.3623)*** (13.2768)***
[10.5724]*** [5.4558]*** [14.3362]***
Inequality 26.1679 -5.5050 1.6039
(12.2139)** (6.1777) (14.8493)
[13.5412]** [6.2411] [15.0256]
GDP per Capita in Thousands 0.0200 0.0531 0.0169
(0.0231) (0.0231)** (0.0472)
[02040] [0.0230]** [0.04789]
Non-White Population Share 6.4771 3.9525 37.9301
(3.0467)** (1.8084)** (4.3394)***
[3.4482]* [1.8888]** [5.4927]***
Small Businesses per 10k people 0.0092 -0.0080 -0.0227
(0.0097) (0.0052) (0.0115)**
[0.0098] [0.0053] [0.0123]*
Share of Small Businesses <10 Employees -104.5639 7.9850 -23.1913
(20.9340)*** (13.2155) (40.2835)
[20.5401]*** [12.5844] [39.7444]
Share of Small Businesses <50 Employees -191.1702 29.2026 43.8916
(61.6154)*** (41.3183) (109.4726)
[65.2340]*** [40.5638] [115.8543]
PPP Loans 0.0000 0.0001 0.0001**
(0.0000) (0.0000)*** (0.0000)
[0.0000] [0.0000]*** [0.0000]**
Population Density -1.3626 15.2125 3.3293
(6.5358) (18.7966) (11.2152)
[6.6232] [17.8955] [11.3369]
Constant 240.0579 -19.5853 -39.7466
(48.7813)*** (31.7068) (82.0903)
[53.3138]*** [31.5116] [90.2629]
Observations 704 704 704
R2 0.455 0.397 0.513
F 41.46 20.10 47.58
Note: Huber-White standard errors in parentheses. Conley (1999) Standard errors are in brackets. A
cutoff distance of 2 degrees (222.28 km) was used for both latitude and longitude. ∗ p < 0.10, ∗ ∗ p <
0.05, and ∗ ∗ ∗ p < 0.01.
©Southern Regional Science Association 2024.
16 The Review of Regional Studies 54(1)
These statistically significant differences are also illustrated in the left panel of Figure 3, but
do not hold in the bank headquarters regression.
When comparing the 60th vs 40th percentiles of PPP loans per small business, we note a
slightly stronger effect at the 60th percentile, significant with 10% confidence. As we widen
the gap in commuting zone loans per small business to look at the 75th vs 25th and 90th
vs 10th percentiles, we observe that the difference in the relationship between bank density
and loans grows. In commuting zones at the 90th percentile of loans per small business, each
additional bank or credit union location per 10K population would correspond to about .16
more loans per small business than at the 10th percentile. This suggests that the relationship
between bank density and loan density may be more meaningful in commuting zones with
high loan densities. We do not observe a similar pattern in the relationship between banking
headquarters and loan density.
Further research is warranted to determine why bank density was closely related to loans
per small business in the commuting zones with the most loans. One possible explanation
lies in the origination fee structure for the loans. At the beginning of the program, lenders
earned origination fees on loans in the amount of 5% of the loan up to $350K, 3% of loans
between $350K and $2M and 1% of loans larger than $2M. In December 2020, the servicing
fee structure was revised so that lenders earned the minimum of 50% of the loan or $2500
for loans under $50K, and 5% of loans between $50-350K, with the remaining fee structure
unchanged. Competition for servicing fees may have been a powerful incentive to rapidly
originate PPP loans, especially among small banks. Regulatory data shows differences be-
tween large and small banks’ response to the PPP program with large bank loan originations
are smaller than predicted based on operational characteristics and historical lending pat-
terns (Kupiec, 2021). The incentive of a reasonably high fee for even smaller loans is a
potential explanation for the spike in loan density per small business in the final lending
period. Additional forthcoming research by the authors explores incentives created by the
origination fees, and by discontinuities in the fee structure.
4. CONCLUSION
The Paycheck Protection Program was an important federal government response to the
COVID-19 outbreak of 2020. In the span of just over a year, almost $800 billion in funds
were allocated as part of an ongoing effort to thwart widespread unemployment. Previous
literature studies the dispersion of PPP loans in the context of bank size and health, and the
impact on communities and jobs. However, there has been to date limited academic study
of the geospatial dispersion of PPP loans.
This paper makes three main contributions to the literature. First, we revisit previous
results from Deming and Weiler (2023) which find a significant positive relationship between
the density of banking locations and PPP loans. These results are an improvement on past
work because we now use more accurately geocoded loans data which removes erroneous
zero or near zero values in a handful of rural Commuting Zones. We confirm that there is
a positive and statistically significant relationship between bank location density and PPP
loans per small business. An additional bank location per 10K population is associated with
an increase of .33 loans per small business, or about 18% from the mean.
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 17
Table 5: Interquantile Regressions of PPP Loan per Small Business
(1) (2) (3)
(.6 vs .4) (.75 vs .25) (.9 vs .1)
PPP Loans PPP Loans PPP Loans
Banks + C.U.s per 10k people 0.041* 0.133*** 0.159***
(0.023) (0.034) (0.033)
Bank Headquarters per 10k people -0.079 -0.102 0.061
(0.067) (0.076) (0.093)
COVID-19 Cases per 10k people 0.005 0.006 0.003
(0.005) (0.005) (0.006)
Employment/Pop 0.770 1.688* 3.516**
(0.494) (0.982) (1.432)
Inequality 0.109 -0.848 -1.491
(0.576) (0.716) (1.074)
GDP per Capita in Thousands 0.000 -0.000 -0.000
(0.001) (0.002) (0.004)
Non-White Population Share 0.542** 1.636*** 2.511***
(0.264) (0.457) (0.463)
Small Businesses per 10k people -0.001 -0.001** -0.002**
(0.000) (0.001) (0.001)
Share of Small Businesses <10 Employees 0.435 0.375 -0.102
(1.075) (2.082) (2.830)
Share of Small Businesses <50 Employees 2.835 8.690 22.311**
(3.476) (6.282) (9.449)
Population Density -0.419 0.188 2.623
(1.109) (2.163) (2.159)
Constant -3.169 -8.327* -20.577***
(2.870) (5.046) (7.347)
Observations 704 704 704
Huber-White standard errors in parentheses
* p < .1, ** p < .05, *** p < .01
©Southern Regional Science Association 2024.
18 The Review of Regional Studies 54(1)
Second, we expand upon previous literature by considering loans issued across three
waves of the PPP program. Loans were decomposed into periods reflecting guidance issued
by the SBA. A first lending period lasted from the initiation of the program for about two
weeks until funds ran out. A week later, a second lending period began following additional
congressional approval. When that period expired with remaining funds still unclaimed,
a third lending period began which expanded the pool of potential lenders. We find that
there is a significant and positive relationship between bank and loan density across all loan
periods, but strongest in the last. In a pooled cross section, we find that the response of
loan volume to bank density was largest in the final wave of lending.
Finally, this paper considers differences in the relationship between bank density and
loans per small business across all values of loan density. We show that the relationship
between bank and loan density is the most intense in commuting zones which received the
largest number of loans per small business. In light of literature suggesting that smaller
banks originated larger than expected shares of the loans, we suggest that further research
is warranted to determine whether the loan origination fee structure might have created
competition to originate larger volumes of loans among smaller banks.
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 19
REFERENCES
Akerlof, George A. (1970) “The Market for ‘Lemons’: Quality Uncertainty and the Market
Mechanism,” Market Failure or Success, 66.
Amior, Michael and Alan Manning. (2018) “The Persistence of Local Joblessness,” American
Economic Review, 108, 1942–1970.
Amiram, Dan and Daniel Rabetti. (2020) “The Relevance of Relationship Lending in Times
of Crisis,” Available at SSRN 3701587.
Atkins, Rachel, Lisa Cook, and Robert Seamans. (2022) “Discrimination in Lending? Evi-
dence from the Paycheck Protection Program,” Small Business Economics, 58, 843–865.
Autor, David, David Cho, Leland D Crane, Mita Goldar, Byron Lutz, Joshua Montes,
William B Peterman, David Ratner, Daniel Villar, and Ahu Yildirmaz. (2022) “An Eval-
uation of the Paycheck Protection Program Using Administrative Payroll Microdata,”
Journal of Public Economics, 211, 104664. http://doi.org/10.1016/j.jpubeco.2022.104664.
Barrios, John Manuel, Michael Minnis, William C Minnis, and Joost Sijthoff. (2020) “As-
sessing the Payroll Protection Program: A Framework and Preliminary Results,” Available
at SSRN 3600595.
Bartik, Alexander, Zoe Cullen, Edward L Glaeser, Michael Luca, Christopher Stanton,
and Aditya Sunderam. (2020) “The Targeting and Impact of Paycheck Protection Pro-
gram Loans to Small Businesses,” SSRN Electronic Journal. http://doi.org/10.2139/ssrn.
3676759.
Berger, Allen N, Christa H S Bouwman, Lars Norden, Raluca A Roman, Gregory F Udell,
and Teng Wang. (2022) “Is a Friend in Need a Friend Indeed? How Relationship Borrowers
Fare During the Covid-19 Crisis,” How Relationship Borrowers Fare during the COVID-19
Crisis (July 1, 2022). Kelley School of Business Research Paper.
Borawski, Garrett and Mark E Schweitzer. (2021) “How Well Did PPP Loans Reach Low-and
Moderate-Income Communities?,” Economic Commentary.
Bunten, Devin M., Stephan Weiler, Eric Thompson, and Sammy Zahran. (2015) “En-
trepreneurship, information, and growth,” Journal of Regional Science, 55, 560–584.
Chernenko, Sergey and David S Scharfstein. (2022) “Racial Disparities in the Paycheck
Protection Program,” National Bureau of Economic Research: .
Conley, T. G.. (1999) “GMM Estimation with Cross Sectional Dependence,” Journal of
Econometrics, 92(1), 1–45. http://doi.org/10.1016/S0304-4076(98)00084-0.
Conroy, Tessa, Sarah A Low, and Stephan Weiler. (2017) “Fueling Job Engines: Impacts of
Small Business Loans on Establishment Births in Metropolitan and Nonmetro Counties,”
Contemporary Economic Policy, 35, 578–595.
Deming, Kristopher and Stephan Weiler. (2023) “Banking Deserts and the Paycheck Pro-
tection Program,” Economic Development Quarterly, p. 08912424231152873.
Duchin, Ran, Xiumin Martin, Roni Michaely, and Hanmeng Wang. (2022) “Concierge Treat-
ment from Banks: Evidence from the Paycheck Protection Program,” Journal of Corporate
Finance, 72, 102124.
Fairlie, Robert and Frank M Fossen. (2022) “Did the Paycheck Protection Program and
Economic Injury Disaster Loan Program Get Disbursed to Minority Communities in the
Early Stages of COVID-19?,” Small Business Economics, 58, 829–842.
Glancy, David. (2021) “Bank Relationships and the Geography of PPP Lending,” Available
©Southern Regional Science Association 2024.
20 The Review of Regional Studies 54(1)
at SSRN 3839671.
Granja, João, Christos Makridis, Constantine Yannelis, and Eric Zwick. (2022) “Did the
Paycheck Protection Program Hit the Target?,” Journal of Financial Economics, 145,
725–761. http://doi.org/10.1016/j.jfineco.2022.05.006.
Humphries, John Eric, Christopher Neilson, and Gabriel Ulyssea. (2020) “The Evolving
Impacts of COVID-19 on Small Businesses Since the CARES Act,” Cowles Foundation
Discussion Paper.
James, Christopher M, Jing Lu, and Yangfan Sun. (2021) “Time Is Money: Real Effects of
Relationship Lending: Evidence from Community Bank Lending in the Paycheck Protec-
tion Program.,” Available at SSRN 3633339.
Joaquim, Gustavo and Felipe Netto. (2020) “Bank Incentives and the Impact of the Paycheck
Protection Program,” SSRN Electronic Journal. http://doi.org/10.2139/ssrn.3704518.
Joaquim, Gustavo and Felipe Netto. (2022) “Optimal Allocation of Relief Funds: The Case
of the Paycheck Protection Program,” SSRN Electronic Journal. http://doi.org/10.2139/
ssrn.3939109.
Kapinos, Pavel. (2021) “Paycheck Protection Program: County-Level Determinants and Ef-
fect on Unemployment,” Federal Reserve Bank of Dallas, Working Papers, 2021. http:
//doi.org/10.24149/wp2105.
Kashian, Russell D, Ran Tao, and Claudia Perez-Valdez. (2015) “Banking the unbanked:
Bank deserts in the United States,” Whitewater: University of Wisconsin.
Koenker, Roger and Kevin F Hallock. (2001) “Quantile regression,” Journal of economic
perspectives, 15(4), 143–156.
Kupiec, Paul H. (2021) “Did Prudent Risk Management Practices or Weak Customer De-
mand Reduce PPP Lending by the Largest Banks?,” Journal of Risk Management in
Financial Institutions, 14(2), 148–160.
Li, Lei and Philip E Strahan. (2021) “Who supplies PPP loans (and does it matter)? Banks,
relationships, and the COVID crisis,” Journal of Financial and Quantitative Analysis,
56(7), 2411–2438.
Liu, Haoyang and Desi Volker. (2020) “Where Have the Paycheck Protection Loans Gone
So Far?,” Federal Reserve Bank of New York.
Marsh, W Blake, Padma Sharma, et al.. (2020) “PPP Raised Community Bank Revenue
but Lowered Profitability,” Federal Reserve Bank of Kansas City, Economic Bulletin,
December.
Morgan, Donald P, Davy Perlman, and Maxim L Pinkovskiy. (2018) The ‘Banking Desert’
Mirage.
Petach, Luke, Stephan Weiler, and Tessa Conroy. (2021) “It’s a Wonderful Loan: Local
Financial Composition, Community Banks, and Economic Resilience,” Journal of Banking
& Finance, 126, 106077.
Rabetti, Daniel. (2022) “Non-Information Asymmetry Benefits of Relationship Lending,”
Available at SSRN 3701587.
Tolbert, Charles M and Molly Sizer. (1996) “US Commuting Zones and Labor Market Areas,”
Economic Research Service, Rural Economy Division.
Weiler, Stephan. (2000) “Information and Market Failure in Local Economic Development:
A New Role for Universities?,” Economic Development Quarterly, 14, 194–203.
Weiler, Stephan, Dana Hoag, and Chuen-mei Fan. (2006) “Prospecting for Economic Returns
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 21
to Research: Adding Informational Value at the Market Fringe,” Journal of Regional Sci-
ence, 46, 289–311.
White, Halbert. (1980) “A Heteroskedasticity-Consistent Covariance Matrix Estimator and
a Direct Test for Heteroskedasticity,” Econometrica: journal of the Econometric Society,
pp. 817–838.
©Southern Regional Science Association 2024.
22 The Review of Regional Studies 54(1)
A. APPENDIX A
Table A1: PPP Loans per Small Business by Round (No Controls)
(1) (2) (3)
R1 Loans R2 Loans Round 3 Loans
Banks + C.U.s per 10k people 0.019*** 0.011** 0.114***
(0.006) (0.005) (0.020)
Bank Headquarters per 10k people 0.033** 0.037*** 0.373***
(0.013) (0.010) (0.084)
Constant 0.193*** 0.310*** 0.398***
(0.018) (0.015) (0.060)
Observations 706 706 706
R2 0.420 0.238 0.482
F 71.155 62.021 102.735
Notes: Results of this Table compare to those in Table 3. We verify that the results presented are not
driven by the inclusion of Commuting Zone level covariates. Huber-White standard errors in parentheses.
* p < .1, ** p < .05, *** p < .01
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 23
Table A2: PPP Loan Amount per Small Business by Round (No Controls)
(1) (2) (3)
R1 Amount ($K) R2 Amount ($K) Round 3 Amount ($K)
Banks + C.U.s per 10k people 167.802 -755.354*** 1131.911***
(290.716) (171.047) (300.879)
Bank Headquarters per 10k people 30.080 178.494 5305.306***
(1285.556) (559.421) (1348.830)
Constant 36501.890*** 18658.598*** 23812.483***
(929.633) (658.227) (916.448)
Observations 706 706 706
R2 0.002 0.089 0.300
F 0.487 17.509 67.554
Notes: Results of this Table compare to those in Table 4. We verify that the results presented are not
driven by the inclusion of Commuting Zone level covariates. Huber-White standard errors in parentheses.
* p < .1, ** p < .05, *** p < .01
Table A3: Interquantile Regression of PPP Loans per Small Business (No
Controls)
(1) (2) (3)
(.6 vs .4) (.75 vs .25) (.9 vs.1)
PPP Loans PPP Loans PPP Loans
Banks + C.U.s per 10k people 0.055** 0.116*** 0.218***
(0.027) (0.037) (0.040)
Bank Headquarters per 10k people -0.058 -0.021 0.028
(0.064) (0.095) (0.138)
Constant 0.020 0.174 0.465***
(0.076) (0.114) (0.137)
Observations 706 706 706
Notes: Results of this Table compare to those in Table 5. We verify that the results presented are not
driven by the inclusion of Commuting Zone level covariates. Huber-White standard errors in parentheses.
* p < .1, ** p < .05, *** p < .01
©Southern Regional Science Association 2024.
24 The Review of Regional Studies 54(1)
Table A4: Bank Density Scaled per Small Business (compare to Table 2)
(1) (2) (3) (4)
PPP Loans PPP Loans PPP Loan Amount PPP Loan Amount
per Small Business per Small Business per Small Business per Small Business
Banks + C.U.s per Small Business 63.696*** 66.027*** 7383.165 6.74e+05***
(7.556) (6.374) (1.83e+05) (1.88e+05)
Bank Headquarters per Small Business 102.713*** 71.589*** 1.68e+06*** 1.70e+06***
(20.014) (16.070) (5.45e+05) (4.76e+05)
COVID-19 Cases per 10k people 0.001 294.606*
(0.005) (168.949)
Employment/Pop 6.830*** 1.57e+05***
(0.667) (18765.644)
Inequality -0.375 28092.444
(0.694) (20807.373)
GDP per Capita in Thousands -0.004*** 81.949
(0.001) (67.624)
Share of Pop. with at least a B.A. -2.325*** -3.16e+04**
(0.421) (14217.670)
Non-White Population Share 2.436*** 45931.434***
(0.266) (6113.676)
Small Businesses per 10k people 0.001 38.543**
(0.000) (16.611)
Share of Small Businesses <10 Employees 0.973 -1.43e+05***
(1.745) (44199.846)
Share of Small Businesses <50 Employees 7.102 -4.15e+04
(5.406) (1.28e+05)
Population Density 0.996 20884.519
(0.713) (36020.596)
Total CZ PPP Loans 0.077***
(0.026)
Constant 0.506*** -9.706** 80621.683*** 1.06e+05
(0.086) (4.277) (2349.256) (98590.996)
Observations 706 704 706 704
R2 0.553 0.703 0.066 0.478
F 182.005 73.388 11.584 41.102
Notes: A higher relative density of banks to small businesses may be a better representation of local
competition than banks scaled per population. In this table we verify that the rescaling doesn’t change
the direction of the relationship of interest. Huber-White standard errors in parentheses. * p < .1, **
p < .05, *** p < .01
©Southern Regional Science Association 2024.
LANDINI, DEMING & WEILER: BANK DENSITY AND PPP LOANS 25
Table A5: PPP Loans and PPP Loan Amount per Small Business
(1) (2)
PPP Loans PPP Loan Amount
per Small Business per Small Business
R1 × Banks + C.U.s per 10k people -0.132*** -0.914**
(0.021) (0.392)
R2 × Banks + C.U.s per 10k people -0.131*** -2.124***
(0.021) (0.491)
R1 × Bank Headquarters per 10k people -0.242*** -3.260**
(0.062) (1.587)
R2 × Bank Headquarters per 10k people -0.241*** -4.754***
(0.062) (1.218)
R1 × COVID-19 Cases per 10k people -0.005 -0.167***
(0.004) (0.059)
R2 × COVID-19 Cases per 10k people -0.002 -0.040
(0.003) (0.082)
R1 × Employment/Pop -4.066*** -16.169
(0.723) (16.966)
R2 × Inequality -4.124*** -65.643***
(0.725) (13.202)
R1 × GDP per Capita in Thousands 0.003*** 0.018
(0.001) (0.047)
R2 × GDP per Capita in Thousands 0.004*** 0.047
(0.001) (0.032)
R1 × Share of Pop. with at least a B.A. 2.169*** 27.290***
(0.375) (9.355)
R2 × Share of Pop. with at least a B.A. 2.381*** 40.236***
(0.375) (8.985)
R1 × Non-White Population Share -2.065*** -34.506***
(0.268) (5.243)
R2 × Non-White Population Share -1.904*** -31.894***
(0.260) (4.442)
R1 × Small Businesses per 10k people 0.003*** 0.021
(0.001) (0.015)
R2 × Small Businesses per 10k people 0.002*** -0.004
(0.001) (0.010)
R1 × Share of Small Businesses <10 Employees -2.705 -79.498
(2.222) (50.778)
R2 × Share of Small Businesses <10 Employees -1.919 27.357
(2.114) (34.055)
R1 × Share of Small Businesses <50 Employees -3.115 -213.697
(6.700) (140.501)
R2 × Share of Small Businesses <50 Employees -3.745 -23.949
(6.325) (93.822)
Continued on next page...
©Southern Regional Science Association 2024.
26 The Review of Regional Studies 54(1)
(1) (2)
PPP Loans per Small Business PPP Loan Amount per Small Business
R1 × Total CZ PPP Loans -0.001* -0.001***
(0.001) (0.001)
R2 × Total CZ PPP Loans -0.001 0.0001
(0.001) (0.001)
R1 × Population Density 0.399 -5.435
(0.700) (14.063)
R2 × Population Density 0.537 17.565
(0.492) (11.155)
R1 5.836 278.278***
(5.043) (101.770)
R2 5.911 22.609
(4.772) (70.294)
Constant 1.146*** 32.256***
(0.012) (0.227)
Observations 2112 2112
R2 0.866 0.811
F 116.491 147.220
Fixed Effects CZ CZ
Notes: To conduct this analysis, Commuting Zones were pooled into a repeat cross section with three
observations per CZ. Cross-sectional covariates are dropped as multicollinear. We demonstrate that the
relationship between bank density and loan density, loan amount is strongest in the final round of lending.
Huber-White standard errors in parentheses. * p < .1, ** p < .05, *** p < .01
©Southern Regional Science Association 2024.
File and source
- File
- landini-2024-can-bank-density-provide-insights-on-the-appropriation-of-disaster-relief-fun.pdf
- Size
- 1,004,905 bytes
- SHA-256
- 70795389fdf8fecf8acd92785a5efbb9b4ef4a8f0502dde44ffd3d776a860b7c
- Our copy
- landini-2024-can-bank-density-provide-insights-on-the-appropriation-of-disaster-relief-fun.pdf
- Original
- rrs.scholasticahq.com