The Politics of the Paycheck Protection Program (Review of Corporate Finance Studies)

Summary

Journal article “The Politics of the Paycheck Protection Program” by Deniz Igan, Thomas Lambert, Prachi Mishra and Eden Quxian Zhang, published in the Review of Corporate Finance Studies in 2025. The abstract asks whether partisanship influences loan allocation through the Paycheck Protection Program (PPP), using the 2020 Presidential campaign contributions made by lenders’ employees as a partisanship measure. It reports that partisan misalignment increases lending, particularly to small and first-time PPP borrowers and those in Republican areas. The paper says the sample consists of 2,551 lenders over 38 weeks, and that the PPP initially authorized $659 billion in forgivable loans and guarantees.

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

Issuer: Source documents
Document type: The Politics Of The Paycheck Protection Program
Case: The Politics Of The Paycheck Protection Program

Cite this document

“The Politics of the Paycheck Protection Program (Review of Corporate Finance Studies).” Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/the-politics-of-the-paycheck-protection-program/, SHA-256 3b11f1864766b90140cbbb22efc101761018fa5b0526834a511985ceb45861c1.

Full text

EUR Research Information Portal

The Politics of the Paycheck Protection Program

Published in:
Review of Corporate Finance Studies

Publication status and date:
Published: 01/11/2025

DOI (link to publisher):
10.1093/rcfs/cfaf013
10.1093/rcfs/cfaf013

Document Version
Publisher's PDF, also known as Version of record

Document License/Available under:
CC BY

Citation for the published version (APA):
Igan, D., Lambert, T., Mishra, P., & Zhang, E. Q. (2025). The Politics of the Paycheck Protection Program. Review of
Corporate Finance Studies, 14(4), 1083-1122. https://doi.org/10.1093/rcfs/cfaf013, https://doi.org/10.1093/rcfs/cfaf013

Link to publication on the EUR Research Information Portal

Terms and Conditions of Use
Except as permitted by the applicable copyright law, you may not reproduce or make this material available to any third party
without the prior written permission from the copyright holder(s). Copyright law allows the following uses of this material
without prior permission:

• you may download, save and print a copy of this material for your personal use only;
• you may share the EUR portal link to this material.

In case the material is published with an open access license (e.g. a Creative Commons (CC) license), other uses may be
allowed. Please check the terms and conditions of the specific license.

Take-down policy
If you believe that this material infringes your copyright and/or any other intellectual property rights, you may request its
removal by contacting us at the following email address: openaccess.library@eur.nl. Please provide us with all the relevant
information, including the reasons why you believe any of your rights have been infringed. In case of a legitimate complaint,
we will make the material inaccessible and/or remove it from the website.
Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
The Politics of the Paycheck Protection
Program
Deniz Igan
International Monetary Fund and CEPR, United States of America
Thomas Lambert
Rotterdam School of Management, Erasmus University, Netherlands
Prachi Mishra
Isaac Center for Public Policy, Ashoka University, India
Eden Quxian Zhang
Monash Business School, Monash University, Australia

Does partisanship influence loan allocation through the Paycheck Protection Program
(PPP)? We examine the 2020 Presidential campaign contributions made by lenders’
employees as a partisanship measure and leverage the PPP’s phased implementation
under both the Trump and the Biden administrations. We find that partisan misalignment
increases lending, particularly to small and first-time PPP borrowers, as well as those in
Republican areas. Misalignment is also associated with higher payroll coverage for small
businesses. Our findings are consistent with Republican-leaning lenders viewing the
PPP’s 2021 phase as a legacy policy of the prior administration, shedding new light on
the partisan-alignment phenomenon in finance. (JEL D72, G21, G28, G32, G38, H12,
H81)

Received: 10 March 2024; Editorial decision: 23 May 2025
Editor: Isil Erel
Authors have furnished an Internet Appendix, which is available on the Oxford
University Press Web site next to the link to the final published paper online.

The COVID-19 outbreak triggered an unprecedented economic freeze that
left millions of businesses in dire need of liquidity (Bartik et al. 2020a). The
government aid response around the world was immediate and unprece
dented. In the United States, as a centerpiece of the Coronavirus Aid,

We are extremely grateful to Sergey Chernenko (our discussant and referee), Stijn Claessens, Andrew Ellul and Isil
Erel (the editors), Nandini Gupta, and participants at the 2022 AEA Meeting and the 2024 RCFS Winter Conference
for their comments and suggestions. Lambert received the support of the Dutch Research Council (NWO) for this
project (Grant VI.Veni.191E.055). The views expressed here are those of the authors and do not necessarily reflect
those of the International Monetary Fund. Corresponding author: Thomas Lambert, t.lambert@rsm.nl

The Review of Corporate Finance Studies 00 (2025) 1–40
© The Author(s) 2025. Published by Oxford University Press on behalf of The Society for Financial Studies.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://
creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any
medium, provided the original work is properly cited.
https://doi.org/10.1093/rcfs/cfaf013 Advance Access publication July 1, 2025
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Relief, and Economic Security (CARES) Act of 2020, the Paycheck
Protection Program (PPP) aimed to provide financial assistance to businesses
that were hit hardest by the freeze. The PPP provided a temporary source of
liquidity for small businesses, initially authorizing $659 billion in forgivable
loans and guarantees. PPP’s first phase ran between April and August 2020.
As a follow-on response to the outbreak, in January 2021, the Consolidated
Appropriations Act (CAA) of 2021 included $284 billion in additional for
givable loans for a second PPP phase.
In this paper, we examine whether lenders’ partisanship plays a role in
shaping lending within the PPP, which was carried out during a period of
intense political polarization in the United States.1 Although the Small
Business Administration (SBA) issued loan guarantees and ultimately deter
mined loan forgiveness, PPP loans were processed and delivered through the
nation’s financial system. Without the deployment of private lenders, it would
have been impossible to distribute billions of dollars in loans to millions of
borrowers within a few months. While lenders were bound by the borrower
eligibility rules outlined in the PPP, they retained discretion in processing and
approving applications.2 Both their capacity—such as the ability to meet
capital and liquidity requirements set by federal regulators—and their will
ingness to engage in PPP lending were critical factors. This willingness may
have stemmed from a desire to maintain existing client relationships or,
potentially, from partisan motivations. For instance, lenders may derive utility
from supporting policies of an administration aligned with their political
leanings, or they may share beliefs or preferences with the administration
regarding the PPP’s objectives.3 Whether, and to what extent, partisan align
ment influenced PPP implementation is the focus of this paper.
Isolating the effect of partisanship on PPP fund allocation decisions is
empirically challenging for two main reasons. First, partisan alignment
between a lender and the ruling administration could correlate with other
(omitted) factors of proximity between lenders and the administration.
Second, the need for PPP funds or the severity of the pandemic may be
directly affected by changes in government policies or political uncertainty
surrounding the presidential election. We address these challenges by
1
In 2019, 82 percentage points separated Republicans’ (89%) and Democrats’ (7%) average job approval ratings
of for President Trump, which was the largest degree of political polarization in any presidential year measured
by Gallup until then (see Jones 2020). By 2021, a new record of 84 percentage points separated Republicans’
(8%) and Democrats’ (92%) average job approval ratings of President Biden. Coibion et al. (2020) report large-
scale survey evidence of the effects of political polarization on economic expectations: both Republicans and
Democrats expect their preferred candidate to win with a very high probability, and they expect the economy to
perform poorly when the opposing party’s candidate wins.
2
SBA guidance allowed all loans to be processed by all lenders under delegated authority and relying on
borrower certifications to determine borrower eligibility and the use of loan proceeds.
3
Although the bills had strong bipartisan support, this does not necessarily imply that lenders indeed supported
the PPP in a bipartisan manner. According to some polls, Democrats and Independents were more likely to
believe that the loans were being distributed unfairly, whereas Republicans were less likely to share this view
(see, e.g., Maine Democratic Party 2020).

2
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
leveraging the phased implementation of the PPP around the 2020 presiden
tial election. While both of the economic relief packages under the PPP were
enacted during the Trump administration, the disbursement of loans in the
program’s second phase occurred under the Biden administration. This shift
in partisan alignment between PPP lenders and the administration in charge
began in early 2021, following the close 2020 presidential election.
Therefore, the transition between administrations during the PPP’s rollout
provides a unique opportunity to identify whether partisan (mis)alignment
between lenders and the governing administration affected lending decisions.
The question we raise in this context is whether and how Republican-
leaning lenders behaved differently than their Democrat-leaning counterparts,
controlling for other observable differences between the two groups. Did
Republican-leaning lenders perceive the second phase under the Biden
administration, with whom they were not aligned, as a less appealing part
of their portfolio, leading to reduced enthusiasm in PPP lending compared to
Democrat-leaning lenders? This “contemporaneous-alignment” view suggests
a negative association between partisan misalignment and PPP lending. Such
an association may exist because lenders derive disutility from engaging with
an administration from the opposing camp or they disagree with the admin
istration in control about the objectives and risks of the program or the
broader economic outlook.4 Conversely, did Republican-leaning lenders
view the second phase of the PPP still as a policy of the Trump administra
tion, with which they were more aligned, and extend more loans than their
Democrat-leaning counterparts did? This “legacy-policy” view suggests a
positive association between partisan misalignment and PPP lending.
To establish our results, we construct a balanced panel data set by aggre
gating partisan PPP lenders’ lending activities to calendar weeks spanning
both phases of PPP. We obtain PPP data from the SBA. Our sample consists
of 2,551 lenders over 38 weeks (18 weeks in 2020 and 20 weeks in 2021) who
extend an average (median) of 42.9 (6) loans per week amounting to $3.6
million ($141 thousand). To measure lender partisanship, we adopt a
“bottom-up” (employee-level) approach using campaign contributions made
by a lender’s employees. Specifically, we use lender employees’ individual
contributions to political action committees (PACs) and candidates in the
2020 election cycle.5 This approach has the advantage of reflecting the under
lying partisan support of people that constitute the organization. For the
4
See Kempf and Tsoutsoura (2024) for more on the “partisan utility” and “partisan belief” mechanisms through
which partisan alignment may influence agents’ decisions.
5
Measuring partisanship using data from the 2020 cycle is ideal for this study for three reasons. First, our
identification strategy exploits the outcome of the 2020 election cycle as an exogenous shock; therefore, it is
most appropriate to use contributions made during this same cycle rather than earlier ones. Second, campaign
contributions are not necessarily persistent over time, so relying on data from previous cycles may not
accurately capture partisanship during the 2020 election. Third, individual contribution records from the
2020 cycle include a large number of small-donor contributions by employees, enabling us to examine the
detailed record of individual contribution data to identify the partisanship of lender employees.

3
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
election cycle we study, we can also include small individual contributions
(below $200) since ActBlue and WinRed (crowdfunding platforms for polit
ical donations) required identity disclosure. This process allows us to address
potential concerns about data scarcity on campaign contributions made by
individual employees. Most lenders lean Republican, with 67% of their
employees supporting the Republican Party and candidates. For robustness
purposes, we also measure lender partisanship using a “top-down” approach
using campaign contributions made by lenders as legal entities.
Approximately 70% of lenders contribute to the campaigns of Republicans.
In a difference-in-differences research design, we find that Republican-
leaning lenders engaged in greater PPP lending during the program’s second
phase—that is, under the Biden administration. The effect is economically
meaningful: a Republican-leaning lender in the 75th percentile of the partisan
ship distribution increases its weekly lending volume by 24.3% more in the
PPP’s second phase compared to a neutral lender in the 25th percentile and
increases its number of new loans by 9.8% more.
This positive relationship between partisan misalignment and PPP lending
is consistent with the “legacy-policy” view. Several explanations may be
driving this finding, although we cannot fully disentangle them. One possible
explanation is that Republican-leaning lenders viewed the PPP as a legacy of
the Trump administration and, motivated by a desire to preserve political
capital and maintain connections with their Republican co-partisans,
increased their lending even under a politically misaligned administration.
Another, nonmutually exclusive, explanation is that the effect of partisan
alignment on lending may depend on who ultimately bears the credit risk.
In contrast to previous study settings in which lenders assume this risk, PPP
loans are forgivable, thus placing the cost of default on the administration in
charge. This distinction suggests that the partisan alignment’s effect on lend
ing may hinge on whether the lender or the government assumes the credit
risk. When it is the lenders, partisanship may offer them benefits that offset
costs or shape their risk perceptions, thereby creating a divide between the
behaviors of politically aligned and misaligned lenders. When it is the gov
ernment, the cost-benefit calculation changes. Consequently, misaligned lend
ers may adopt lending strategies that are more aggressive.
We conduct additional tests to ensure our results genuinely reflect partisan
ship rather than other potential mechanisms that could produce a positive
correlation between partisan misalignment and PPP lending. We find that
the partisanship of rank-and-file employees—those with direct discretion in
granting loans—have a significant impact on the volume and number of PPP
loans. The relationship is also primarily driven by loans granted to small and
first-time borrowers under the program (“first-draw loans”). These are cases
in which the loan officers are more likely to have had leeway in verifying
eligibility and reviewing the documentation supplied in support of the
requested loan amount. Furthermore, our results indicate that partisan

4
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
misalignment during the program’s second phase is associated with a rela
tively higher allocation of PPP funds in counties that are typically considered
Republican strongholds. Republican-leaning lenders expanded their PPP
lending more significantly in counties with a larger fraction of white popu
lation, lower personal income per capita, and a workforce less suited to tele
working. Republican-leaning lenders also mainly directed loans to borrowers
in Republican-voting states. If anything, these important tests are in line with
these lenders seeing the PPP as a legacy of the Trump administration and
looking to maintain ties with their co-partisans.
Does partisan misalignment ultimately matter? To shed some light on this
question, we investigate the aggregate and real effects. First, by collapsing
our data set at the county level, we document that the effect of partisan
misalignment translates meaningfully at the county level: a county with
Republican-leaning lenders (with an average 78.4% Republican-leaning
employees at the 75th percentile) experiences a weekly PPP lending volume
growth rate that is 32.26% higher than a county with a “neutral” lender
(defined as a lender with 50.5% Republican-leaning employees at the 25th
percentile), and its number of new loans increases by 12% more countywide.
Second, with county-level weekly unemployment insurance (UI) data, we
present evidence that partisan misalignment is associated with higher PPP
payroll coverage for small businesses. This result underscores that partisan
misalignment has real effects that go beyond its direct effects on PPP loan
origination.
Our findings hold significant implications for both research and policy.
First, we contribute to the literature on political polarization and finance,
providing evidence that partisan misalignment may actually incentivize lend
ing activities within the PPP context rather than holding loans back. These
findings complement prior work on political polarization in banking by sug
gesting a new dimension to consider: who ultimately bears the risk may
matter for the partisan-alignment phenomenon. In the PPP context, it is the
government that assumes the credit risk, while banks make the decisions to
grant credit. Second, we show the importance of accounting for small donor
contributions by employees in measuring lender partisanship, which may
better capture the intensity of political engagement and influence within
organizations. Last, we add to the growing body of literature on politics
and banking by uncovering novel evidence of “more benign” effects of polit
ical economy factors on small business lending in the context of one of the
largest economic relief packages in recent history.

1. Literature Review
Our paper contributes to several strands of the economics and finance liter
ature. First, it is related to work on the political economy of government aid
programs. Several studies examine whether electoral politics affects the

5
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
spatial allocation of government funds: Fishback et al. (2003) focus on funds
under the New Deal in the 1930s; Boone et al. (2014), under the American
Recovery and Reinvestment Act of 2009; Duchin and Hackney (2021) and
Berger et al. (2023), under the first PPP phase in 2020; and Ha (2023), under
the second PPP phase in 2021. Unlike these papers that examine the aggregate
spatial allocation of funds across states or counties, our analysis operates at
the lender level. We focus specifically on the political motivation behind
lending decisions in a highly polarized political context, uncovering that
partisan misalignment between lenders and the administration positively
influences support for small businesses during both PPP phases.
Second, our paper adds to the literature on the relation between economic
behaviors and partisanship, including that of loan officers (Dagostino et al.
2023), credit analysts (Kempf and Tsoutsoura 2021), entrepreneurs
(Engelberg et al. 2023), professional money managers (Cassidy and
Vorsatz 2021), judges (Chen 2020), and central bankers (Ioannidou et al.
2023).6 One of the closest papers to our line of inquiry is Kempf et al.
(2023), who find that ideological alignment between US banks and foreign
governments is an important factor behind the allocation of cross-border
syndicated loans to corporations. In a similar spirit, we infer the partisan
leanings of lenders using their political contributions to candidates running
in the 2020 presidential election to identify whether partisan alignment shapes
PPP fund allocations. We expand on this study by measuring partisanship
through the individual contributions of lender employees to political candi
dates. This approach is crucial, as aggregated individual contributions con
tribute to the “bottom-up” partisanship of organizations. Additionally, our
measures of partisanship consider small, individual contributions (under
$200) because both major platforms for individual contributions, ActBlue
and WinRed, required identity disclosure during the election cycle we exam
ine (Bouton et al. 2022; Cage 2024). This inclusion represents an improve
ment over much of the existing literature on politics and finance, which often
overlooks these smaller contributions. Including them allows us to capture a
more comprehensive view of partisanship and further mitigate concerns about
data scarcity. Interestingly, our results also contrast with prior studies that
show that partisan misalignment negatively impacts lending and offer explan
ations centered on partisan utility or partisan beliefs (e.g., Dagostino et al.
2023; Kempf et al. 2023). Instead, our results indicate that partisan misalign
ment between lenders and the administration increases PPP lending, consis
tent with the idea that Republican-leaning lenders may have still viewed the
program’s second phase as part of the Trump administration’s agenda. As a
result, they may have increased their PPP lending to preserve or even enhance

6
See Kempf and Tsoutsoura (2024) for a survey of the literature on political polarization and finance.

6
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
their political capital under the new, misaligned administration. Our findings
thus complement these prior studies on political polarization and banking.
Third, our paper belongs to the broader literature on politics and credit.7
Within this literature, our paper is most closely related to studies documenting
the influence of banks’ political connections on loan renegotiations (Agarwal
et al. 2018), mortgage lending (Chavaz and Rose 2019), consumer lending
(Akey et al. 2021), and corporate lending (Lambert et al. 2023) in the United
States. Our paper documents the effects of politics on small business lending
(PPP loans), while contributing to our understanding of the mechanisms
behind the partisan-alignment phenomenon.
Last, our paper is connected to a rapidly growing literature on the deter
minants of PPP allocation (Balyuk et al. 2021; Cororaton and Rosen 2021; Li
and Strahan 2021; Granja et al. 2022; Chernenko and Scharfstein 2024) and
its impacts on the US economy (Bartik et al. 2020a, 2020b; Chetty et al. 2020;
Hanson et al. 2020; Humphries et al. 2020; Meier and Smith 2020; Bartlett
and Morse 2021; Autor et al. 2022b; Chodorow-Reich et al. 2022; Elenev
et al. 2022; Granja et al. 2022). We join this literature by presenting novel
evidence on the extent to which lenders’ partisan alignment with the admin
istration influences PPP fund allocation and produces real effects.

2. Background and Data
2.1 PPP timeline
Our empirical strategy leverages the phased implementation of the PPP
around the closely contested presidential election of 2020. Below, we outline
the timeline of the PPP under both the Trump and Biden administrations. For
a detailed description of the PPP rounds and their targeting, we refer to prior
work on the PPP (e.g., Autor et al. 2022a).

• Phase 1 (“Trump”). President Trump signed into law the establishment
of the PPP as part of the CARES Act on March 27, 2020. The program
launched just days later, on April 3, 2020, with an initial allocation of
$349 billion aimed at providing forgivable loans and guarantees to small
businesses impacted by the COVID-19 pandemic. Demand for the funds
was intense, and the initial allocation was exhausted by April 16, mark
ing the end of the first PPP round. In response, Congress approved an
additional $310 billion on April 24, 2020, launching the program’s
second round, which continued until August 8, 2020, when applications
were halted again.
• Phase 2 (“Biden”). The 2020 U.S. presidential election occurred on
November 3, 2020, resulting in Biden’s election as president a few
days later. The PPP received further support later in the year when, on
7
See Lambert and Volpin (2018) for a survey of the literature on political economy and (bank) finance.

7
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
December 27, 2020, the CAA was signed into law, granting an addi
tional $284 billion. Under the new administration, the program reopened
for a third round on January 11, 2021. Following President Biden’s
inauguration on January 20, the administration implemented new guide
lines on February 22 to expand access to smaller and minority-owned
businesses, aiming to make PPP support more equitable. The application
deadline was initially set for March 31, 2021, but was extended to May
31. However, the PPP’s funds were nearly depleted by May 4, with the
program officially ending on May 31, 2021.
Loans from the initial two PPP rounds were issued in phase 1 and available
to businesses that fit the PPP’s definition of a small business, which typically
required having fewer than 500 employees. The third round, initiated in phase
2, targeted businesses that had not previously received a PPP loan.
Additionally, it offered “second-draw” loans to businesses that had already
taken a PPP loan, had fewer than 300 employees, and could demonstrate a
significant revenue decline in 2020. Notably, approximately 75% of the fund
ing from this third round was allocated to these second-draw loans.

2.2 Variable definitions and sources
2.2.1 PPP lending. We obtain PPP data from the SBA, which provides
information on business and loan characteristics. The characteristics of the
businesses that received loans under the program are their name, location,
legal status (e.g., sole proprietorship or corporation), reported number of
employees, and industry classification (six-digit NAICS). The loan character
istics consist of the loan amount in dollars, the date when the loan was
approved, and the processing method indicating first- or second-draw loans.
We use loan-level data released by the SBA in January 2022.
The original PPP data set from the SBA provides a unique identifier for the
originating lender, lender names, and location information. To construct our
sample, we start with the whole population of PPP loans and lenders, and then
we exclude 21 loans for which the lender identifier is missing and 2,477 loans
with approval dates outside the three PPP rounds. Our final sample consists of
11,467,324 loans and 5,374 lenders. To identify the lender types, we man
ually match the PPP lenders based on their names and locations to the banks
and bank holding companies (BHCs) in Call Reports and FRY-9C from
2019–2021. We first check whether a PPP lender can be matched to a bank
included in Call Reports. If not, we then check whether the lender can be
matched to a BHC covered in FRY-9C. We categorize the lenders that have a
match in this step as “Banks.” For the unmatched lenders, we categorize the
lenders whose names include “Credit Union,” “Farm Credit,” “Agricultural

8
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 1
PPP lender sample
Year Number of Total loan Number of PPP
PPP loans amount ($m) lenders

All 11,467,324 100.00% 796,496.56 100.00% 5,374 100.00%
2020 5,136,366 525,765.08 5,298
2021 6,330,958 270,731.48 5,153
With employee contributions 9,151,484 79.80% 705,591.09 88.59% 2,548 47.41%
With corporate PAC contributions 2,903,213 25.32% 268,793.48 33.75% 59 1.10%
Banks 8,144,108 71.02% 719,592.16 90.34% 4,245 78.99%
2020 4,600,836 501,578.65 4,212
2021 3,543,272 218,013.51 4,148
With employee contributions 7,240,105 63.14% 662,723.55 83.20% 2,155 40.10%
With corporate PAC contributions 2,703,594 23.58% 264,790.28 33.24% 55 1.02%
Credit 405,326 3.53% 16,598.76 2.08% 958 17.83%
Unions
2020 212,931 10,511.75 940
2021 192,395 6,087.01 851
With employee contributions 239,618 2.09% 9,955.48 1.25% 338 6.29%
With corporate PAC contributions 1232 0.01% 47.89 0.01% 2 0.04%
Others 2,917,890 25.45% 60,305.64 7.57% 171 3.18%
2020 322,599 13,674.68 146
2021 2,595,291 46,630.96 154
With employee contributions 1,671,761 14.58% 32,912.06 4.13% 55 1.02%
With corporate PAC contributions 198,387 1.73% 3,955.32 0.50% 2 0.04%

The table reports the total numbers, total loan amounts, and total lender counts of the PPP loans covered in the
different samples. “All” represents the original PPP loan sample obtained from the SBA website, dropping
loans with missing lender identifiers or originating dates after the PPP end date. “Year” represents the loans
originated in the first PPP phase (in either 2020 or 2021). “With employee contributions” is the sample of the
lenders’ employees who made individual contributions to Republican or Democratic presidential and congres
sional candidates as well as party PACs in the 2020 election cycle. “With corporate PAC contributions” is the
sample of the lenders with valid campaign contributions from their affiliated corporate PACs to Republican- or
Democratic-party PACs in the 2020 election cycle. The coverages of the latter two samples—in terms of the
number of loans, total loan amounts, and covered lender—relative to the “All” sample are reported in the
percentage’s column. The statistics for three subsamples—“Banks,” “Credit Unions,” and “Others”—are also
included in the ^^^subsequent rows. These three subsamples include the PPP loans for which the lenders are
banks, credit unions, and other financial firms, respectively.

Credit,” “ACA,” “CU,” and “FCU,” as “Credit Unions.” All the rest, we
categorize as “Others.”8
Table 1 shows the breakdown of the PPP loan sample based on the lender
type. The overall PPP loan sample covers almost $800 billion in originated
loans. Banks serve as the major disseminators of PPP funds: they constitute
78.99% of all PPP lenders providing 71.02% of all loans and 90.34% of the
total dollar amount. Institutions other than banks and credit unions (“Others”)
stepped up significantly more in the second PPP phase, in 2021.
To construct our dependent variables, we use the aggregated PPP loan
dollar amount originated by a lender in a given period and the number of
new PPP loans originated by a lender in a given period. The dependent

8
The three most active PPP lenders categorized as “Others” are Prestamos CDFI; Harvest Small Business
Finance, LLC; and Capital Plus Financial, LLC. These three financial firms extended 45.6% of the PPP loans
in this category.

9
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
variables constructed are respectively labeled Value of loans and Number of
loans. The baseline sample consists of lender-week observations.

2.2.2 Lender partisanship. Individuals and corporations can make contri
butions toward the election of a candidate in a number of ways. They can
contribute directly to a candidate, to a political party, or to a PAC. A PAC is a
committee that raises and spends money to elect or defeat candidates. Most
PACs represent businesses (e.g., Bank of America PAC), associations (e.g.,
American Bankers Association PAC), or ideological causes (e.g., EMILY’s
List PAC). An organization’s PAC solicits money from its employees or
members and directs the money in PAC’s name to candidates and political
parties. Other types of PACs include “leadership PACs,” which allow poli
ticians to raise money apart from their own campaigns to help other politi
cians (e.g., Majority Cmte PAC). There are limits on the amount that can be
contributed. For instance, in the 2020 election cycle, an individual donor
could contribute $2,800 per election to a candidate and $5,000 per year to
a PAC, while a PAC could contribute up to $5,000 per election to a candidate
and up to $35,500 per year to a political party committee.9
To measure lenders’ partisanship, we obtain information on contributions
by individual donors and PACs to presidential and congressional political
campaigns during the 2020 election cycle.10 Employing contributions from
the 2020 cycle is consistent with our empirical strategy and, importantly,
allows the inclusion of small donor contributions in the analysis. Moreover,
partisanship may change over time, and contributions from previous cycles
may not correlate that strongly with the current cycle.11
We source data on campaign contributions from the Center for Responsive
Politics as part of its “Open Secrets” database. We capture the political sup
port for Republicans as the ratio of campaign contributions to Republican
PACs and Republican candidates to total campaign contributions. The time
span is the 2020 election cycle (2019–2020). We use both campaign contri
butions made by individuals that are employed by the lenders and campaign
contributions made by lenders as legal entities. This procedure means that we
9
Following a 2010 U.S. Court of Appeals decision, a new type of PAC was created, a “super PAC.” Super PACs
make no contributions to candidates or parties, but they can spend money directly on actions that specifically
target an election, such as running ads or sending mail to voters. There are no restrictions on the amount or
sources of the funds spent.
10
We include campaign committees for all party-affiliated or leadership PACs, strong ideological PACs, congres
sional candidate PACs, and candidate PACs for three candidates involved in the presidential election (Donald
Trump, Joseph Biden, and Kamala Harris; Mike Pence did not have a candidate campaign committee registered
in the 2020 election cycle).
11
Partisanship, although generally stable for individuals over time, can shift in response to major life events,
political realignments, or changes in salient issues (see, e.g., Page and Jones 1979; Gerber et al. 2010). For
firms, partisanship is less persistent, often varying across election cycles due to changes in executive leadership
or strategic responses to evolving policy environments. For instance, Hutton et al. (2014) find that Republican-
leaning CEOs implement distinct corporate policies, but these patterns tend to shift with leadership transitions.
Hassan et al. (2019) show that firms adjust their political donation behavior in response to electoral outcomes.

10
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
rely on both “bottom-up” (employee-level) and “top-down” (lender-level)
approaches to capture lender partisanship. Most studies use only one of these
approaches (e.g., Duchin et al. 2023; Kempf et al. 2023). Although our
preferred approach is the “bottom-up” one—given its better data coverage
and its closer reflection of the underlying partisanship of the individuals who
constitute the organization (i.e., its employees and leadership)—we also use
the “top-down” approach in our robustness checks to mitigate concerns
associated with relying on one measure taken in isolation.
Regarding the “bottom-up” approach, we use the detailed record of indi
vidual contributions data to identify lender employee partisanship. The indi
vidual contributions record for the 2020 election cycle has the advantage of
including a vast number of small donors (defined as those contributing less
than $200), since the major individual contribution platforms (ActBlue and
WinRed) started to request all donors disclose their identities, including their
employers. ActBlue was created in 2004 to help Democrats raise money and
now dominates Democratic fundraising, while WinRed was launched in 2019
on the Republican side (Bouton et al. 2022; Cage 2024). We have over 80
million valid records of individual contributions in the 2020 cycle.12 We run
fuzzy match commands to compare the employer and lender names using
various methods, including “big-ram,” and “token,” and manually read
through all possible matches.13 We can identify over half a million records
for which the donor’s employer is a PPP lender. We drop invalid or duplicate
records and contributions made to independent PACs (not political ideolog
ical) or third-party candidates. In total, we can match over 23,873 employees
working for 2,548 lenders, covering 47.41% of PPP lenders (Table 1). We
aggregate the contribution amounts to either Republican or Democratic can
didates and leadership or ideological PACs14 and calculate the fraction of the
contribution going to Republican candidates and PACs. Figure 1, panel A
shows the distribution of employee partisanship. Most individuals demon
strate one-sided support to either Republicans or Democrats. The 23,873
employees contribute on average $3,135 (median $500) to party PACs and
candidates, of which 45.3% (untabulated statistic) goes to Republican
candidates or Republican Party PACs.

12
Considering these small campaign contributions is critical to mitigate measurement concerns about data
scarcity when relying on partisanship measures aggregating at the organization level contributions from
employees.
13
Many banks share similar words in their names, which complicates the identification of specific entities. Self-
reported employer names by individual donors include unstandardized company names. Therefore, an indi
vidual employee may be matched to multiple banks in the same election cycle. We use the donor’s location
information and the lenders’ headquarters and branches to identify the correct matches. We also manually
search more than 3,000 potential employees’ profile in LinkedIn, news releases, and lender information
webpages for ambiguous matches. See Internet Appendix B for more detail on the manual matching process.
14
Popular party PAC recipients for employee contributions in the 2020 cycle include “Republican National
Cmte,” “Trump Make America Great Again Cmte,” and “Democratic Congressional Campaign Cmte.”

11
Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Figure 1
Review of Corporate Finance Studies / v 00 n 00 2025
Distribution of lender employee contributions to Republicans
12
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Next, we aggregate employee partisanship at the lender level. As most
individuals support only one party, we consider an employee to be
Republican-leaning if the fraction of her contribution amount going to
Republicans is > 50% of the total amount going to both Republicans and
Democrats. We then calculate the proportion of Republican-leaning employ
ees in the 2020 election cycle for each PPP lender (the variable is labelled
Employee contributions to Republicans). Figure 1, panel B also shows the
distribution of the fraction of Republican-leaning employees measured at the
lender level. A substantial fraction of the lenders is Republican leaning (with
the measure value close to one), while a number of lenders have diverse
employee partisanship. On average, a lender has 67% Republican-leaning
employees of all politically active employees. In Table 1, we report the
breakdown of contributing lenders with employee contributions in the 2020
election cycle. Notably, the “bottom-up” measure of partisanship allows us to
cover a majority of the PPP loan sample, accounting for > 9.1 million PPP
loans (79.80% of the overall sample) and covering > $705 billion in loan
volume (88.59% of the total loan amount).
For robustness, we also aggregate employee partisanship at the lender level
based on the employees’ contributed dollar amounts. We calculate each
employee’s contribution fraction to Republicans of the total amount to both
parties and then take the average across all individual contributors employed
by the same lender. We use the simple average (Employee contributions to
Republicans (average value)) or the weighted average based on the total
contributed dollar amount (Employee contributions to Republicans (weighted
value)). We obtain values for these two partisanship measures that are con
sistent with Employee contributions to Republicans.
In addition to the “bottom-up” approach, we construct a measure of parti
sanship based on the “top-down” approach for robustness purposes; that is,
we use lender contributions from the lender’s corporate PAC to Republican or
Democratic Party leadership and ideological PACs.15 We do not include
contributions to candidates since lenders generally do not contribute directly

Figure 1
Continued
This figure shows the density distributions of Republican-leaning employees in PPP lenders in the 2020 election
cycle. In panel A, the distribution plots the fraction of employee contributions to Republicans over total
contributions to either Democratic or Republicans at the individual employee level. The data cover 23,873
employees in 2,548 PPP lenders and are based on the employees’ individual contributions to various political
party and ideological PACs and candidates from 2019–2020. 100% indicates the employee contributes only to
Republican PACs and candidates, while 0% indicates contributions only to Democrats. In panel B, the histogram
shows the employee partisanship aggregated at the lender level and describes the fraction of Republican-leaning
employees for each lender. An employee is Republican leaning if her fraction of contributions to Republicans is
> 50%. Source: Authors’ calculation.
A. Distribution of Republican contribution fractions by lender employees (individual level)
B. Distribution of fractions of Republican-contributing employees (lender level)

15
Popular leadership PAC recipients for bank contributions in the 2020 election cycle include “Building
Leadership & Inspiring New Enterprise” and “Democratic Congressional Campaign Cmte.”

13
Review of Corporate Finance Studies / v 00 n 00 2025

to presidential candidates.16 Lenders tend to contribute to congressional can

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
didates in connection to their potential power in congressional subcommittees
or in influencing policies affecting their local areas rather than based on
partisanship. We manually compare all of the lender names (and their BHC
names if applicable) to the PAC names active in the 2020 cycle. In total, we
identify 59 PPP lenders contributing via their affiliated corporate PACs to 253
various party leadership PACs during the 2020 election cycle. Although the
number of matched lenders with corporate PAC contributions seems very
limited (only 1.10% of all lenders in Table 1), the sample covers some of
the most active PPP lenders, providing over a quarter of all PPP loans and
over a third of the total dollar volume. Table 1 also shows that 55 of 59
contributing lenders are banks. On average, a lender contributes $145,490
(median $37,100) to party PACs, of which 70% goes to Republican PACs.
We calculate the proportion of the contribution dollars that go to Republican
PACs of the total amount going to either Republican or Democratic PACs as
the first partisanship measure (labelled Lender contributions to Republicans).
If the measure takes the value of 100%, the lender does not contribute to
Democratic PACs at all and only to Republican PACs. Figure 2 shows the
distribution of the measure. Notably, the distribution is similar to the distri
bution of employee contributions reported in Figure 1 (panel B).17

2.2.3 Other determinants of PPP lending. Since many other factors are
likely related to PPP lending, we control for a host of other political and
institutional factors as well as bank and regional characteristics in our
analysis.
First, we address the potential confounding effect of lenders’ political
connections on their PPP lending behaviors by using a direct proxy of poten
tial change in political connections around 2020. We capture the exogenous
change in the connected senators’ political power by measuring the ascension
of Senate committee chairs in the 2020 elections. We hand collect the Senate
committee assignment information from the Congressional Directory for the
116th and 117th Congresses and cross-check the chairs and ranking members.
As Democrats gained majority control of the Senate in the 2020 elections, all
Senate committee chairs were rotated from Republican to Democratic
Senators (one exception is the Budget Committee chair, which shifted to
Independent Bernie Sanders). Following Akey et al. (2021), we include
16
In 2020, only one bank in our sample contributed to Trump, and none contributed to other presidential
candidates. In contrast, individuals more commonly contribute directly to presidential candidates than their
employer corporate PACs do.
17
For robustness purposes, we ensure that our results are not driven by a subset of individuals switching
partisanship in the 2020 election cycle or by small contributions made via WinRed or ActBlue. To address
this concern, we reconstruct our partisanship measures based on contributions from the 2018 electoral cycle
and excluding small contributions. We do not find any evidence suggesting that our results are sensitive to these
measurement concerns. We thank the referee for encouraging us to explore these potential issues.

14
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Figure 2
Distribution of lender contributions to Republicans
The figure shows the density distribution of the proportion of lender PAC contributions to Republican PACs of
total contributions to both Republican and Democratic PACs in the 2020 election cycle. The data cover 59 PPP
lenders and are based on their corporate PAC contributions to 253 various political party PACs from 2019–2020.
100% indicates the bank corporate PAC contributes only to Republican PACs, while 0% indicates contributions
only to Democratic PACs. Source: Authors’ calculation.

two turnover events in which the previous Democratic ranking member or
vice chair left the committee for the 117th Congress for relatively exogenous
reasons (Table 2).18 We validate every rotation by checking news coverage
and Ballotpedia. We construct a dummy variable, Powerful politician, indi
cating whether the PPP lender headquarter state is in the state whose senator
took a committee chair for exogenous reasons in year 2021.
Second, we control for the political environment at the lender headquarter
state. We collect the state governor party affiliations and the number of seats
held by the two parties in the state house and senate as reported by the
National Conference of State Legislatures. We construct a dummy variable,
Republican state, indicating whether both the governor and state legislative
control are Republican in the PPP lender headquarter state.19
Third, we control for how lenders handle PPP applications. Borrowers may
have switched to other lenders in 2021 if a lender did not treat them well in
the first phase in 2020. To capture this factor, we download the consumer

18
For completeness, Table IA1 in Internet Appendix A also provides a full list of the Senate committee
chairpersonships of the 117th Congress.
19
For most states, the executive and legislative bodies are controlled by the same party. In 20% of the cases, the
state control is recorded as “divided” when the governor does not belong to the same party as that controlling
the state legislature. This variable is stable over the years 2020 and 2021 except for New Hampshire and
Montana, which changed from Republican controlled to divided.

15
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 2
117th Congress Senate committee chair ascensions
Committee Name State Party Previous chair Reason

Judiciary Richard J. Illinois D Lindsey Previous ranking member
Durbin Graham, of stepped down
South Carolina
Indian Affairs Brian Schatz Hawaii D John Hoeven, Previous vice chair retired
of North
Dakota

The table reports the two exogenous Senate committee chair turnovers in which the previous ranking member
who was vice chair left the committee for relatively exogenous reasons. The two events are as follows. First,
Dick Durbin (D, Illinois) took over as the chair of the Judiciary Committee. The previous lead Democrat in the
committee, Dianne Feinstein, announced that she would not seek the chairmanship in the next Congress in
November 2020 and subsequently left the Judiciary Committee in 2021. Second, Brian Schatz (D, Hawaii)
became the chairman of the Committee on Indian Affairs. The previous Democratic vice chair, Tom Udall,
retired in 2020 and became U.S. Ambassador to New Zealand in 2021.

complaints record from the Consumer Financial Protection Bureau (CFPB)
from 2020 to July 2024 and match the company names to the PPP lenders. We
also search the consumers’ narratives of the issues and identify those related
to PPP if the narrative contains “ppp,” “paycheck loan,” “paycheck program,”
or “paycheck protection.” In total we have 446 PPP-related complaints
matched to 37 lenders. We thus construct a variable, Number of PPP com
plaints, being the total number of PPP-related complaints for each lender.20
Fourth, we control for bank characteristics. We obtain balance sheet and
income statement information from Call Reports and matched the information
to the previous calendar quarter. The controls include Bank size, Bank ROA,
Bank NPL, Bank Tier1 ratio, and Bank core deposit. Naturally, the controls
are available only for PPP lenders classified as “Banks.”
Last, we gather county-level data on sociodemographic conditions and on
vulnerability to pandemic-induced lockdowns and social distancing as well as
state-level data on political conditions. To proxy for racial and income dis
parities in PPP applications, we collect county-level racial, income, and pop
ulation characteristics from the Census Bureau21 and the Bureau of Economic
Analysis. The variable Minority county captures the fraction of the nonwhite
population in a given county. Income per capita enters as the log level of per-
capita income in thousands of dollars in a given county, while Population is
the county population (also in log). Furthermore, to proxy for the resilience of
an industry to a pandemic, we rely on teleworkability measures, which are
based on detailed survey information on occupations as classified by Dingel
and Neiman (2020). Their indices capture the share of jobs in a given industry

20
To address any reverse causality concerns arising from PPP complaints over all rounds, we construct an
alternative measure of the number of PPP complaints that covers only the first two PPP rounds in 2020.
The results remain robust (see Subsection 3.3).
21
We use County Characteristics Resident Population Estimates, 2019 and 2021 releases, reported by the U.S.
Census Bureau, Population Division.

16
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
that can be done at home and are aggregated at the two-digit industry level
using either the employment or the wage share. Industries with a higher share
of teleworkable jobs are likely to suffer less from the pandemic or associated
social-distancing measures. To properly evaluate the teleworkability of the
local workforce, we gather the employment numbers for each NAICS indus
try from the Bureau of Economic Analysis. Local workforce teleworkability is
the weighted average of the Occupational Information Network–derived
work-from-home score from Dingel and Neiman (2020), weighted by the
number of employed under each NAICS code in a given county. To further
capture the political environment at the state level, we construct two variables
measuring the political ideology of the citizens and the local government
sourced from Richard Fording’s website (https://rcfording.wordpress.com).
The variable State citizen liberal score is the revised 2016 citizen ideology
series following Berry et al. (1998), while the State government liberal score
is the 2017 state government ideology measure following Berry et al. (1998).
Higher (lower) values of these political ideology variables indicate more
liberal (conservative) values and positions in a state.

2.2.4 Employment. For our analysis on the real effects, we collect the
weekly UI claims from 14 state labor department websites, which we supple
ment with data from the Opportunity Insights Economic Tracker (https://
tracktherecovery.org). The country-week initial UI claims signal increases
in local unemployment in the timeliest possible way based on available sta
tistics. The dependent variable UI claims rate is the weekly number of initial
claims per 100 people in the 2019 county labor force. Initial claims data
reflect the number of initial claims approved by the state. The approved
claims measure is distinct from the number of claims filed as it includes
only claims that were ultimately accepted, not all claims submitted.

2.3 Summary statistics
Table 3 presents the summary statistics for the key variables on PPP lending,
lender partisanships, and lender and location characteristics for the main
analysis. For the sake of brevity, the summary statistics for all other variables
are relegated to Tables IA2 and IA3 in Internet Appendix A. We construct the
lender-week panel data set by aggregating PPP lenders’ lending activities to
calendar weeks. In this baseline sample, we keep only the lenders with non
missing partisanship measures using either the “bottom-up” or “top-down”
approach. We then transform the sample to a balanced panel data set by
replacing the loan amount and number with zeroes if the lender is not active
during that week. As a result, the panel data set consists of 38 weeks (18 in
2020 and 20 in 2021) and 2,551 lenders. Eventually, we have 96,938 lender-
week observations in the baseline sample, with an average (median) loan
amount of $3.6 million ($141 thousand) and 42.9 (6) loans in a week

17
18

Table 3
Summary statistics
Obs. Mean S.D. Perc. 25 Median Perc. 75

PPP lending
Value of loans ($thousand) 96,938 3,646.189 13,849.381 3.250 141.119 1,033.095
Number of loans 96,938 42.879 133.170 1 6 25
Value of loans (log 1þ) 96,938 9.784 5.880 8.087 11.857 13.848
Number of loans (log 1þ) 96,938 2.067 1.740 0.693 1.946 3.258
Lender Partisanship
Employee contributions to Republicans 96,824 0.669 0.398 0.333 1 1
Employee contributions to Republicans (average value) 96,824 0.669 0.396 0.333 0.957 1
Employee contributions to Republicans (weighted value) 96,824 0.679 0.419 0.200 0.998 1
Review of Corporate Finance Studies / v 00 n 00 2025

Number of contributing employees 96,824 5.002 14.554 1 2 3
Lender contributions to Republicans 2,242 0.700 0.346 0.432 0.912 1
Other determinants of PPP lending
Powerful politician 96,938 0.056 0.230
Republican state 96,900 0.458 0.498
Number of PPP complaints in CFPB 96,938 0.014 0.116 0 0 0
Number of PPP complaints (log 1þ) 96,938 0.010 0.081 0 0 0
Bank size (log 1þ) 81,630 13.412 1.464 12.432 13.203 14.131
Bank ROA (%) 81,630 2.190 1.988 0.922 1.542 3.234
Bank NPL (%) 81,529 0.807 0.984 0.189 0.503 1.021
Bank Tier1 ratio (%) 81,630 10.252 2.507 8.666 9.710 11.136
Bank core deposits (%) 81,579 61.683 18.541 50.773 64.937 75.162
Local workforce teleworkability 117,116 0.271 0.051 0.242 0.270 0.298

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Minority majority county 119,016 0.057 0.232
Income per capita ($thousand) 117,116 51.429 14.442 43.031 48.659 56.313
Population (log) 119,016 10.274 1.482 9.293 10.158 11.133
Employment
UI claims rate 48,093 0.861 0.981 0.288 0.502 0.970

The table presents the summary statistics for the key variables of the baseline panel sample (lender-week level) except for the county-level variables. The sample keeps only those PPP
lenders with PAC or employee contributions in the 2020 election cycle. All variables are winsorized at the 1st and 99th percentiles. Value of loans is the aggregated loan dollar amount
originated in a calendar week by a given lender. Number of loans is the number of new PPP loans originated in a week by a given lender. Value of loans (log 1þ) is the natural logarithm of
loan amount plus one. Number of loans (log 1þ) is the natural logarithm of number of new loans plus one. First draw loans are the PPP loans extended to new borrowers in 2020 or in 2021
with the processing method identified as “PPP.” Second draw loans include PPP loans extended in 2020 and those to previous borrowers in 2021 with the processing method identified as
“PPS.” Loans to small (big) borrowers are the PPP loans extended to borrowers with reported numbers of employees under (at least) 20. Employee contributions to Republicans is the
fraction of employees contributing more than half to Republican PACs or candidates in a given lender during the 2020 election cycle. Employee contributions to Republicans (average
value) is the simple average for a given lender of each employee’s contribution fraction to Republican candidates and PACs of their contribution to either party (fraction of total). Employee
The Politics of the Paycheck Protection Program

contributions to Republicans (weighted value) is the weighted average for a given lender of each employee’s contribution fraction to Republican candidates and PACs of their contribution
to either party, weighted by the individual’s total contribution amount. Number of contributing employees is the number of contributing employees in the 2020 election cycle in a lender.
Lender contributions to Republicans is the fraction of the total contributions made from a lender’s corporate PAC to Republican leadership PACs or party PACs of the total contributions to
either party in the 2020 election cycle. Powerful politician is a dummy indicator of whether the lender headquarter state is either Illinois or Hawaii. Republican state is a dummy variable
that takes the value of one if the lender headquarter is in a state controlled by Republican in 2020 and zero otherwise. Number of PPP complaints in CFPB is the number of consumer
complaints mentioning PPP since 2020 from the CFPB. Number of PPP complaints (log 1þ) is the natural logarithm of the number of PPP complaints plus one. Rank-and-file employee
(Executive) contributions to Republicans (number) is the fraction of the rank-and-file employees (executives) in a given lender contributing more than half of their total donations to
Republican PACs or candidates during the 2020 election cycle. Number of contributing rank-and-file employees (executives) is the number of contributing rank-and-file employees
(executives) in the 2020 election cycle in a lender. Bank size is the natural logarithm of one plus the bank’s total assets in the previous quarter. Bank ROA (%) is quarterly net income × 4 /
total assets × 100, lagged one quarter. Bank NPL (%) is loans past due 90 days or more and nonaccruals over total loans × 100, lagged one quarter. Bank Tier1 (%) is tier one capital over
total loans × 100, lagged one quarter. Bank core deposits (%) is the sum of the transactions deposits and all other savings deposits that are included in total nontransaction accounts,
excluding MMDAs, over total deposits × 100, lagged one quarter. The county-level characteristics are at the county-week level. Local workforce teleworkability is the weighted average of
the Occupational Information Network–derived work-from-home score from Dingel and Neiman (2020), weighted by the number of employments under each NAICS code for a given
county. Minority majority county is a dummy variable that takes a value of one if the nonwhite population in a given county is over 50% of the total population and zero otherwise. Income
per capita is county personal income per capita in $thousands. Population (log) is the natural logarithm of the county population. UI claims rate is the weekly number of UI initial claims at
the county level per 100 people in the 2019 labor force from Opportunity Insights Economic Tracker.

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
(Table 3). On average, five employees in a lender firm make campaign con
tributions, and 67% of such employees support Republicans. Similarly,
approximately 70% of total corporate PAC contributions to leadership
PACs are directed toward Republicans.

3. Results
3.1 Identification strategy
To examine the effect of lenders’ partisan alignment with the administration
on PPP lending, we run difference-in-differences regressions at the lender-
week level:
Yit ¼ βPostt × Political Shocki þ γXit þ μi þ μt þ εit : (1)
We index the lender by i and the week by t. Yit is either the total amount or
the total number of loans (in log) originated by lender i in week t. The data
span the two PPP phases (cf. Subsection 2.1): from 2020W14 to 2020W32 for
the first phase (“Trump”) and from 2021W2 to 2021W22 for the second phase
(“Biden”). Postt is a dummy variable that takes a value of one for the weeks
that belong to the second PPP phase and a value of zero otherwise.
Political Shocki is a variable measuring lenders’ partisanship during the
2020 presidential election. Xi;t is a vector of lender-level control variables.
μi and μt denote lender and week fixed effects, respectively. εi;t is the error
term. Standard errors are clustered at the lender headquarter state level and
are robust to heteroskedasticity. The parameter of interest is β, which captures
the effect of partisan misalignment on PPP lending. If partisan misalignment
between the lenders and the administration has a positive (negative) effect on
PPP lending, we expect β>0 (β<0).
The key identifying assumption in a difference-in-differences design is that
of parallel trends. Identification relies on the assumption that the outcome
would have behaved similarly across treated (Republican-leaning lenders)
and control (Democratic-leaning and neutral lenders) groups absent the treat
ment. In our setting, this assumption translates into maintaining that the PPP
lending would have evolved similarly across the treated and the control
groups if Trump emerged as the winner of the 2020 presidential election.
We provide evidence in support of this assumption in the next subsection.

3.2 Partisan alignment and PPP lending
Table 4 presents the difference-in-differences regression results from estimat
ing Equation (1). The dependent variable in columns (1) and (2) is the total
value of the PPP lending originated in a week, while that in columns (3) and
(4) is the number of PPP loans originated. We interact Postt with our main
independent variable of interest, Employee contributions to Republicans. It
captures the leaning of the lender employees toward Republican relative to

20
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 4
Partisan alignment and PPP lending
(1) (2) (3) (4)
Value of loans Number of loans

Post × Employee contributions to Republicans 0.8130 0.3646 0.2407 0.1469
(3.9192) (2.0272) (4.2029) (3.0268)
Post × Powerful politician 0.0760 −0.1712 0.0336 −0.0127
(0.4882) (−0.9263) (0.6274) (−0.2259)
Republican state −0.8615 −1.1012 −0.1759 −0.2372
(−1.4162) (−1.9524) (−0.9451) (−1.2813)
Post × Republican state 0.4582 0.3634 0.1108 0.1054
(2.2804) (2.0721) (1.3866) (1.4608)
Post × Number of PPP complaints −1.8228 1.1696 −0.4757 −0.1519
(−3.4060) (1.7896) (−2.3330) (−0.6369)
Bank size −1.6374 −0.6567
(−3.4237) (−4.2262)
Post × Bank size −0.5168 −0.0636
(−6.3490) (−2.9121)
Bank ROA 0.0329 0.0219
(1.1992) (2.8348)
Bank NPL −0.0300 −0.0345
(−0.4159) (−1.8872)
Bank tier1 ratio −0.1004 −0.0418
(−2.0070) (−2.9161)
Bank core deposits −0.0064 −0.0019
(−1.0750) (−1.3516)
Lender FE Y Y Y Y
Week FE Y Y Y Y
Adj. R-sq .546 .564 .705 .733
N 96,786 81,414 96,786 81,414

This table documents the effects of lenders’ partisanship on PPP lending. The difference-in-differences regres
sion is specified in Equation (1) in the lender-week panel data set. The dependent variables are Value of loans
(log 1þ) and Number of loans (log 1þ). The measure for lenders’ partisanship is based on employee con
tributions to Republican (fraction of Republican-leaning employees). All variables are defined in the note to
Table 3. t-statistics are in parentheses. Robust standard errors are clustered at the lender headquarter state level.
Variables are winsorized at the 1st and 99th percentiles.
p < .1; p < .05; p < .01.

Democratic candidates, ranging between zero and one; a higher value for this
variable indicates that the lender is more likely to lean Republican. We also
interact Postt with the following controls: Powerful politiciani, Republican
statei,t, Number of PPP complaintsi, and Bank sizeit. Interacting these controls
allows us to account for differential effects across these bank- or bank-state
characteristics.22 For instance, small firms and banks could have exclusive
windows that could have differential effects on small (Republican-leaning)
versus large (neutral or Democratic-leaning) banks. All of the regressions
include lender and week fixed effects, while columns (2) and (4) control
for further bank characteristics.
Regarding the results, the estimated coefficient on the interaction term of
interest β is positive and statistically significant at the 1% level in most
columns. These results suggest that when the Biden administration took office

22
In the regressions, the time invariant lender-level controls are absorbed by the lender fixed effects.

21
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
following the 2020 presidential election, the Republican-leaning lenders
engaged more actively in PPP lending during the program’s second phase.
The economic magnitude is also meaningful. Consider column (2). A
Republican-leaning lender (defined as having 100% Republican-leaning
employees at the 75th percentile) has an average weekly lending volume
growth rate in the second phase 24.3% (0.3646 × 0.667) higher than a
“neutral” bank does (defined as having 33.3% Republican-leaning employees
at the 25th percentile). In column (4), the average number of new loans is
9.8% (0.1469 × 0.667) higher. Overall, these results, showing a positive rela
tionship between partisan misalignment and PPP lending, support the idea
that Republican-leaning lenders viewed the PPP’s second phase as a legacy of
the Trump administration. As a result, they may have increased their PPP
lending to preserve or even enhance their political capital under the new
administration.23 However, these findings are inconsistent with a
“contemporaneous-alignment” view typically portrayed in the literature:
Democrat-leaning lenders are better aligned with the Biden administration
but they do not lend as vigorously as Republican-leaning lenders do despite
sharing the beliefs of their co-partisans in the government.
Regarding the controls, they generally display the expected sign across
most columns, although they are not always statistically significant at conven
tional levels. The interaction with Republican state is positive in sign and
statistically significant at the 5% level in columns (1) and (2) but fails to be
statistically significant at the 10% level in columns (3) and (4). This result
may suggest that there is more lending during the program’s second phase in
states where the lender’s headquarters is controlled by Republicans. The
lenders with a lower number of PPP-related CFPB complaints tend to lend
more, although the coefficient is not stable when the bank characteristics are
also controlled for. Smaller banks (that are typically Republican-leaning, see
below) also tend to lend more in the second phase, and this relationship
appears to be more stable than those with the other interaction variables.
Importantly, the inclusion of these controls does not change our results on
partisan misalignment.
Figure 3 plots the estimated coefficients for the total value of PPP loans
over the weeks covered by the program, supporting the validity of the parallel
trends assumption. We estimate the difference-in-differences regression from
Equation (1), replacing dummy variable Postt with a set of weekly dummy
variables. The base period is the final week of the second PPP, ending August
8, 2020. Each dot represents the coefficient on the interaction between
23
Another, nonmutually exclusive, explanation for our findings is that the lending effects of misalignment
between lenders and the government depend on who ultimately bears the credit risk. Unlike prior work, in
the context of the PPP—in which loans are forgivable—the government assumes the credit risk. This structure
may have incentivized Republican-leaning lenders to adopt lending strategies that were more aggressive,
potentially to affect an ideologically distant administration. Unfortunately, with the available data, we cannot
disentangle and directly test this explanation.

22
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Figure 3
Parallel trends
This figure shows the effects of lenders’ partisanship on PPP lending over the weeks during the PPP. The
difference-in-differences regression is specified in Equation (1) in the lender-week panel data set, where we
replace Post with a set of week dummy variables. The graphs plot the size of the coefficients of the interaction
terms between Employee contributions to Republicans and the weekly dummy variables as well as the 95%
confidence intervals. The vertical red line indicates the change in administration. The base level is set to be the
last week of the second PPP round, ending on August 8, 2020. We include Powerful politician, Republican state,
Number of PPP complaints, and their interaction terms with Post as well as lender and weekly dummy variables.
All variables are defined in the notes to Table 3. The robust standard errors are clustered at the lender head
quarter state level. The variables are winsorized at the 1st and 99th percentiles. Source: Authors’ calculation.

Employee contributions to Republicans and a weekly dummy, capturing the
relative effect on loan value. The lines indicate 95% confidence intervals. The
figure shows that the weekly loan amount by the Republican- versus the
Democratic-leaning and neutral lenders evolve in parallel prior to the presi
dential election. While the Republican-leaning lenders respond more strongly
at the onset of PPP, the lending patterns across the groups remain similar for
most of the second round. A noticeable increase in loan originations by the
Republican-leaning lenders emerges in the weeks following the election and
persists through the final month of the third round. We also arrive at a similar
conclusion of parallel pretrends when using the number of loans as the out
come variable and when controlling for bank characteristics (Figure IA1 in
Internet Appendix A).

3.3 Robustness
We assess the robustness of our findings through a series of additional checks,
all reported in the Internet Appendix A for brevity. We first verify that our
results hold when using alternative measures of lender partisanship, including

23
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
the “top-down approach” (Table IA4 and Figure IA2). We then show that our
results are robust to the use of Poisson regressions to account for the count-
like nature of the outcome variables (Table IA5, panels A and B, and Figure
IA3), to the potential nonlinearity in the effects of bank size (Table IA6,
panels A and B), to the inclusion of additional bank characteristics (C&I
loans, small business loans, and C&I commitments) interacted with Postt
(Table IA7, panel A), to the inclusion of time-invariant bank characteristics
interacted with Postt (Table IA7, panel B), to the use of alternative defini
tions of the dummy variable Powerful politician using the full list of Senate
committee chairmanships (Tables IA8, panels A and B), and to the use of
CFPB complaint data limited to the first two PPP rounds (Table IA9). Finally,
we address concerns about preexisting differences across the Republican-
leaning banks, on the one hand, and the Democratic-leaning and neutral
banks, on the other hand. We test for differences in means across several
bank characteristics (Table IA10, panels A and B) and implement a propen
sity score matching procedure based on these characteristics (Table IA11).
Our results remain robust: Republican-leaning lenders engaged in more PPP
lending following the 2020 presidential election and the transition to the
Biden administration, even within a matched sample of banks with similar
preelection characteristics.

3.4 Mechanisms and other explanations
Thus far, we have shown that partisan misalignment is associated with
increased lending in the PPP context, in contrast with prior studies on the
partisan-alignment relationship in finance. The positive effect of partisan
misalignment on PPP lending that we document is consistent with the idea
that Republican-leaning lenders, likely viewing the PPP’s 2021 phase as a
legacy policy, increased their lending to preserve political capital. Our base
line specification controls for bank characteristics, political and institutional
factors, and lender and time fixed effects, addressing some concerns about
confounding dynamics. Nonetheless, alternative explanations for this positive
relationship remain. We now turn to a discussion of potential mechanisms and
explore these further.

3.4.1 Loan officers versus executives. One potential mechanism through
which Republican-leaning lenders may have increased lending during the
program’s second phase is via loan officers, whose political views could
have influenced their loan approval decisions once Biden took office.
Alternatively, the mechanism could involve executives who are incentivized
to maintain relationships with key clients.
To test for the influence on marginal loans of loan officers, we first recon
struct our lenders’ partisanship measures based on lender employee contri
butions by breaking the employees down by job titles. We utilize the

24
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 5
Loan officers versus executives
(1) (2)
A. Rank and file employees Value of loans Number of loans

Post × Rank-and-file employee contributions to Republicans 0.2917 0.1293
(1.7345) (2.7853)
Controls Y Y
Lender FE Y Y
Week FE Y Y
Adj. R-sq .567 .735
N 67,911 67,911
B. Executives Value of loans Number of loans
Post × Executive contributions to Republicans 0.2053 0.0996
(1.0229) (1.7457)
Controls Y Y
Lender FE Y Y
Week FE Y Y
Adj. R-sq .575 .746
N 39,837 39,837

This table documents the effects of partisanship based on different sets of employees on PPP lending. The
difference-in-differences regression is specified in Equation (1). The dependent variables are Value of loans
(log 1þ) and Number of loans (log 1þ). In panel A, the measure for partisanship is the fraction of Republican-
leaning rank-and-file employees of all contributing rank-and-file employees. In panel B, the measure for
partisanship is the fraction of Republican-leaning lender executives. Executives include employees whose
titles are CXOs, president, chairman, board director, managing director, executive, or senior/executive vice
president; all the rest are rank-and-file employees. The set of control variables includes all control variables in
column 2 of Table 4. All variables are defined in the notes to Tables 3 and IA2. t-statistics are in parentheses.
Robust standard errors are clustered at the lender state level. Variables are winsorized at the 1st and 99th
percentiles.
p < .1; p < .05; p < .01.

occupation information from the individual contribution filings and supple
ment the information by manually looking up over 3,000 individuals with
ambiguous occupations and employers on LinkedIn, on bank information
webpages, or in press releases. We categorize a bank employee as an
“executive” if her title or reported occupation is bank executive, chairman,
president, board director, managing director, CEO, CFO, COO, CTO, CIO, or
senior/executive vice president. Otherwise, we categorize the employee as
“rank-and-file.” Of 23,873 contributing employees, 3,539 are bank executives
working for 1,255 PPP lenders and 20,334 rank-and-file employees working
for 2,098 lenders during the sample period. We recalculate the variable,
Employee contributions to Republicans, using only the rank-and-file employ
ees or the executives and reestimate Equation (1).
Table 5 presents the results. Panel A shows that the estimated coefficient on
the variable of interest, Rank-and-file employee contributions to Republicans,
is positive and statistically significant at conventional levels in both columns.
The economic effect is also of the same order of magnitude as it is in Table 4
(columns (2) and (4)). However, the estimated coefficient on the variable
Executive contributions to Republicans is also positive but lower in magni
tude and significant only at the 10% level in column (2). Nevertheless, these

25
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
results should be interpreted with caution, as panels A and B rely on different
samples. Another issue is that there are fewer executives than rank and-file
employees; therefore, executive contributions to Republicans could be meas
ured with greater noise than is involved with the rank-and-file employee
contributions, which could bias the estimated coefficients for executive con
tributions toward zero. We thus estimate the regressions in the small inter
section at which both the rank-and-file employee partisanship and executive
partisanship measures are nonmissing, which is roughly a third of the full
sample (see Table IA12 in Internet Appendix A). We also observe that rank-
and-file employee partisanship has a larger effect size on PPP lending than
executive partisanship does, although both lose statistical significance. Given
that the results are qualitatively consistent in both Tables 5 and IA12, one can
say that rank-and-file employees seem to play a more significant role in the
observed increases in PPP lending.
These findings suggest that the partisanship of those directly involved in
PPP loan origination—namely, loan officers—drives our main results, con
sistent with the influence of partisan alignment at the operational level rather
than top-down, lender-wide strategic considerations.

3.4.2 Small versus large borrowers. Chernenko et al. (2023) show that,
conditional on applying to a bank, PPP loan approval odds are strongly
positively correlated with a firm’s number of employees. Smaller firms, for
example, may have gaps or mistakes in their applications, which would create
scope for loan officers’ preferences and decisions to affect outcomes. To
investigate whether loan officers are more likely to be influenced by their
personal judgment when processing loan applications, we split our loan sam
ple based on borrower size and categorize those with less than 20 employees
as “small borrowers” and the rest as “big borrowers.” We then decompose our
original dependent variables for loans to small or big borrowers ($-amount
and number).
Table 6 presents the results. In panel A, we find that the effects of lenders’
partisanship on PPP lending to small borrowers are both statistically and
economically comparable to our baseline results in Table 4, columns (2)
and (4). By contrast, in panel B, we find that the effects of lenders’ partisan
ship on PPP lending to large borrowers are insignificant. Panel C reports the
z-statistics comparing the estimated coefficients from panels A and B, which
are (somewhat) significant in both columns. In Table IA13 in Internet
Appendix A, we verify (and confirm) the robustness of these results to alter
native cutoffs to define small borrowers.
Overall, these results indicate that lenders’ partisanship primarily influen
ces marginal loans to small borrowers, consistent with the idea that such loans
are more likely to be shaped by the personal discretion of loan officers.
Indeed, these loans typically have less standardized and timely information

26
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 6
Small versus large borrowers
(1) (2)
A. To small borrowers Value of loans Number of loans

Post × Employee contributions to Republicans 0.3677 0.1491
(2.1217) (3.1071)
Controls Y Y
Lender FE Y Y
Week FE Y Y
Adj. R-sq .558 .728
N 81,414 81,414
B. To big borrowers Value of loans Number of loans
Post × Employee contributions to Republicans 0.0204 −0.0156
(0.1592) (-0.9819)
Controls Y Y
Lender FE Y Y
Week FE Y Y
Adj. R-sq .534 .651
N 81,414 81,414
C. Coefficients in panels A and B compared
Z-stat. 1.612 3.258
P value .107 .001

This table documents the effects of partisanship on PPP lending to small and big borrowers. The difference-in-
differences regression is specified in Equation (1). In panel A, the dependent variables are Value of loans (log
1þ) and Number of loans (log 1þ), but these are restricted to loans to small borrowers with less than 20
employees. In panel B, the dependent variables include only loans to big borrowers with at least 20 employees.
The partisanship measure is based on employee contributions (fraction of Republican-leaning employees). The
set of control variables include all control variables in column 2 of Table 4. All variables are defined in the
notes to Tables 3 and IA2. t-statistics are in parentheses. Robust standard errors are clustered at the lender state
level. Variables are winsorized at the 1st and 99th percentiles. The symbols , , and denote statistical
significance at the 10%, 5%, and 1% levels, respectively. In panel C, z-statistics are the difference of the
coefficients in the same column from panels A and B / the square root of the sum of the variances of the two
estimated coefficients.

available, giving officers more leeway in interpreting eligibility criteria or
evaluating supporting documentation.

3.4.3 First-draw versus second-draw loans. As described in Subsection
3.1, the program’s second phase permitted second-draw loans. These loans
are largely predetermined, as they depend on the number of firms that
received a PPP loan in the first two rounds and are eligible for a second
draw. Firms qualifying for a second-draw loan are likely to be in urgent
need of liquidity. Moreover, they are familiar with the process and have
much of the necessary paperwork ready, having already been through the
process in phase 1. We therefore expect that it would be unusual for a lender
to deny a valid second-draw application from a firm previously approved by
the lender.
We address this possibility by decomposing the dependent variables on
PPP lending depending on whether a loan is first or second draw. The first

27
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 7
First- versus second-draw loans
(1) (2)
A. First-draw loans Value of loans Number of loans

Post ×Employee contributions to Republicans 0.7581 0.2453
(3.3974) (3.9382)
Controls Y Y
Lender FE Y Y
Week FE Y Y
Adj. R-sq .521 .691
N 81,414 81,414
B. Second-draw loans Value of loans Number of loans
Post ×Employee contributions to Republicans 0.1771 −0.0051
(1.0562) (-0.1382)
Controls Y Y
Lender FE Y Y
Week FE Y Y
Adj. R-sq .548 .728
N 81,414 81,414
C. Coefficients in panels A and B compared
Z-stat. 2.082 3.458
P value .037 <.001

This table documents the effects of partisanship on PPP lending in two different batches. The difference-in-
differences regression is specified in Equation (1). In panel A, the dependent variables are Value of loans (log
1þ) and Number of loans (log 1þ), but these are restricted to first-draw loans, including all loans originating in
2020 and first-draw loans only in 2021. In panel B, the dependent variables include only second-draw loans,
including all loans originating in 2020 and second-draw loans only in 2021. The partisanship measure is based
on employee contributions (fraction of Republican-leaning employees). The set of control variables includes all
control variables in column 2 of Table 4. All variables are defined in the notes to Tables 3 and IA2. t-statistics
are in parentheses. Robust standard errors are clustered at the lender state level. In panel C, z-statistics are the
difference of the coefficients in the same column from panels A and B, over the square root of the sum of the
variances of the two estimated coefficients. Variables are winsorized at the 1st and 99th percentiles.
p < .1; p < .05; p < .01.

set of these measures considers only borrowers’ first-draw loans, which
include all loans from 2020 and only first-draw loans from 2021 (according
to the processing method variable recorded by the SBA). These measures
capture lending to first-time borrowers. The second set of measures considers
only borrowers’ second-draw loans, which also include all loans from 2020
and only second-draw loans from 2021 (according to the processing method
variable recorded by the SBA). These measures capture lending to borrowers
in 2020 and repeated lending to the same set of borrowers in 2021.
Table 7 contains the results. In panel A, we find that the effects of lenders’
partisanship on PPP lending to first-draw borrowers are positive and statisti
cally significant at the 1% level. This result contrasts with panel B, where
lenders’ partisanship is not associated with PPP lending to second-draw bor
rowers. Panel C reports the estimated coefficients from panels A and B via z-
statistics, which are highly significant in both columns.
We also test whether the effects for first-draw loans are stronger for small
borrowers. To do so, we further decompose the dependent variables on PPP

28
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 8
Two-way split of PPP loans originated: First- versus second-draw loans to small versus large borrowers
(1) (2) (3) (4)
Processing method First-draw loans Second-draw loans

A. To small borrowers Value of loans Number of loans Value of loans Number of loans

Post × Employee contributions 0.7891 0.2434 0.1598 −0.0086
to Republicans (3.6425) (3.9714) (0.9948) (-0.2418)
Controls Y Y Y Y
Lender FE Y Y Y Y
Week FE Y Y Y Y
Adj. R-sq .516 .687 .537 .723
N 81,414 81,414 81,414 81,414
B. To big borrowers Value of loans Number of loans Value of loans Number of loans
Post × Employee contributions 0.0175 0.0093 0.0467 −0.0107
to Republicans (0.1341) (0.5037) (0.3804) (-0.7042)
Controls Y Y Y Y
Lender FE Y Y Y Y
Week FE Y Y Y Y
Adj. R-sq .532 .673 .534 .649
N 81,414 81,414 81,414 81,414

This table documents the effects of partisanship on PPP lending in two different batches and to two different set
of borrowers. The difference-in-differences regression is specified in Equation (1). In columns 1 and 2 of panel
A (B), the dependent variables are Value of loans (log 1þ) and Number of loans (log 1þ), but these are
restricted to first-draw loans to small (big) borrowers, including all loans originating in 2020 and first-draw
loans only in 2021 to borrowers with less than (at least) 20 employees. In columns 3 and 4 of panel A (B), the
dependent variables include only second-draw loans to small (big) borrowers, including all loans originating in
2020 and second-draw loans only in 2021 to borrowers with less than (at least) 20 employees. The partisanship
measure is based on employee contributions (fraction of Republican-leaning employees). The set of control
variables includes all control variables in column 2 of Table 4. All variables are defined in the notes to Tables 3
and IA2. t-statistics are in parentheses. Robust standard errors are clustered at the lender state level. Variables
are winsorized at the 1st and 99th percentiles.
p < .1; p < .05; p < .01.

lending on whether a loan is extended toward small borrowers (that is, those
with < 20 employees). Table 8 shows the results. As expected, in panel A,
columns (1) and (2), the effects of lenders’ partisanship on PPP lending to
first-draw and small borrowers are positive and statistically significant at the
1% level, while the effects of lenders’ partisanship are insignificant in panel B
and the remaining columns of panel A.
Together, these results indicate that lenders’ partisanship affects only
first-time, small borrowers. This result is consistent with the idea that only
first-draw loans are subject to personal discretion, and thus loan officer partisan
ship, while second-draw loans are unaffected by such bias. Importantly, the
significance of first-draw loans in our setting does not contradict a legacy-
policy-based explanation for our results.

3.4.4 Republican versus Democratic counties. Our findings thus far sug
gest that Republican-leaning lenders engage in greater PPP lending once Biden
came into power and they do so when loan officers have more discretion (small

29
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
and second-draw loans). We now ask where they direct the credit supply. Did
they lend more in Republican-leaning counties, which would also be suggestive
of legacy effects?
To answer this question, we collapse our data at the lender-county-week
level and fill the missing observations with zeros to form a balanced panel
data set. We use the PPP loan project county information from the SBA to
identify the loan location and match to 5-digit Federal Information Processing
Standards codes. This collapsing results in over 7 million observations. We
then run the following regression:
Yict ¼ βPostt × Political Shocki þ γXit þ μic þ μcτ þ μt þ εict ; (2)
where i denotes a lender, c denotes a county, and t denotes a week. Yict is
either the total amount of loans (in log) or the total number of loans (in log)
originated in county c by lender i in week t. μic represents lender × county
fixed effects and μcτ county × PPP round fixed effects, which absorb any
county-year-level sociodemographic controls. μt represents week fixed
effects. This specification allows us to mitigate unobservable demand-side
effects. The remaining variables are defined as before, and the remaining
Greek symbols are parameters to be estimated (the treatment effect being
given by β). We provide the summary statistics for the additional variables
at the lender-county-week level in Table IA14 in Internet Appendix A.
Table 9 presents the results of estimating Equation (2).24 Our main findings
carry through in this analysis. Using the estimated coefficient in column (2), a
Republican-leaning lender (defined as having 100% Republican leaning
employees at the 75th percentile) has an average weekly lending volume
growth rate in the second phase 5.9% (0.0801 × 0.733) greater than does a
“neutral” bank (defined as having 26.7% Republican-leaning employees at the
25th percentile) across counties. Its growth rate in average number of new
loans is 1% (0.0147 × 0.733) greater across counties (column (4)).
We further explore this idea by looking in Table 10 at the cross-sectional
heterogeneity of the partisanship-PPP lending relationship. In panel A, we
focus on county-level sociodemographic characteristics, while in panel B we
consider state-level characteristics of the political environment. In panel A,
the estimated coefficients on the triple interaction terms are negative and
statistically significant at the 1% level in all columns. These results indicate
that Republican-leaning lenders expanded their PPP lending more signifi
cantly during the program’s second phase, particularly in counties with a
larger fraction of white population, lower personal income per capita, and a
workforce less suited to teleworking. These counties often align with
Republican strongholds, which supports the notion that partisanship influ
enced the PPP lending patterns—a legacy reflecting the policies of the
24
In Table IA15 of Internet Appendix A, we use Poisson regressions to account for the count-like nature of the
outcome variables, which exhibit masses at zero. The results remain materially unchanged compared to those
reported in Table 9.

30
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 9
Republican versus Democratic counties
(1) (2) (3) (4)
Value of loans Number of loans

Post × Employee contributions to Republicans 0.5479 0.0801 0.0720 0.0147
(18.3709) (2.5570) (20.1410) (3.9262)
Post × Powerful politician (headquarter state) −0.3205 −0.2194 −0.0436 −0.0259
(-3.6003) (-2.4660) (-4.4683) (-2.5424)
Republican state (headquarter) 0.1019 0.1474 0.0152 0.0222
(0.5983) (0.9675) (0.6594) (1.0864)
Post × Republican state (headquarter) −0.3017 −0.2126 −0.0419 −0.0283
(-9.1133) (-6.7804) (-8.2937) (-6.6199)
Post × Number of PPP complaints −0.0021 0.1652 −0.0016 0.0199
(-0.1961) (8.2595) (-0.8880) (7.1442)
Bank size 0.4869 0.0686
(7.4635) (7.7426)
Post × Bank size −0.1027 −0.0126
(-12.0179) (-12.0595)
Bank ROA −0.0597 −0.0078
(-9.3516) (-7.7539)
Bank NPL 0.0892 0.0107
(4.7123) (4.1378)
Bank tier1 ratio 0.1185 0.0171
(18.2131) (15.3725)
Bank core deposits −0.0022 −0.0004
(-2.9289) (-3.7399)
Lender × county FE Y Y Y Y
County × PPP round FE Y Y Y Y
Week FE Y Y Y Y
Adj. R-sq .375 .413 .478 .527
N 7,373,900 6,048,597 7,373,900 6,048,597

This table documents the effects of partisanship on PPP lending at the lender-county level. The difference-in-
differences regression is specified in Equation (2) in the lender-county-week panel data set. The dependent
variables are Value of loans (log 1þ) and Number of loans (log 1þ). The partisanship measure is based on
employee contributions (fraction of Republican-leaning employees). All variables are defined in the note to
Table IA14. t-statistics are in parentheses. Robust standard errors are clustered at the state level. Variables are
winsorized at the 1st and 99th percentiles.
p < .1; p < .05; p < .01.

Trump administration.25 In panel B, the estimated coefficients on the triple
interaction terms display the expected signs and are statistically significant at
the 1% level across all columns. The results suggest that, irrespective of the

25
Although we examine in Table 10 the relationship between partisanship and PPP lending across counties in
terms of racial disparities, the variable Minority ratio is a relatively crude proxy for capturing such differences.
This situation raises the concern that our results could still be explained by the possibility that Republican-
leaning banks underlent to minority-owned firms in the first two rounds and merely increased their lending to
such firms in the third round. To address this concern, Table IA16 in Internet Appendix A uses measures of
county-level racial composition that are more refined (i.e., shares of Black, Asian, and Hispanic populations)
and yields similar results. Moreover, in Table IA17, we use borrower-level racial information from the original
PPP data. While racial or ethnic information is available for only approximately 30% of loans, our core results
hold. Following Howell et al. (2024)—who use a machine-learning approach to estimate the probability that a
business owner is a minority—we further draw on data generously provided by the referee for incorporated
borrowers in four states (i.e., Florida, Georgia, Louisiana, and North Carolina). Again, we find no evidence that
Republican-leaning lenders increased lending to minority-owned businesses in 2021 relative to their
Democratic counterparts (untabulated results). Together, these results indicate that our findings are unlikely
to be driven by Republican-leaning banks compensating in the third round for prior underlending to minority
borrowers.

31
32

Table 10
Republican versus Democratic counties: Further cross-sectional heterogeneity
A. Interactions with sociodemographic characteristics and teleworkability

(1) (2) (3) (4) (5) (6)
Value of loans Number of loans Value of loans Number of loans Value of loans Number of loans

Post × Employee contributions to Republicans 0.1740 0.0295 0.4202 0.0564 0.8009 0.1051
(3.9280) (5.3392) (6.7904) (6.1195) (9.6122) (8.1112)
Post × Employee contributions to Republicans × Minority ratio −0.4708 −0.0741
(−3.8138) (−4.4987)
Review of Corporate Finance Studies / v 00 n 00 2025

Post × Employee contributions to Republicans × Income per capita −0.0065 −0.0008
(−7.0339) (−5.4295)
Post × Employee contributions to Republicans × Local workforce teleworkability −2.3230 −0.2909
(−10.0989) (−7.6687)
Controls Y Y Y Y Y Y
Lender × county FE Y Y Y Y Y Y
County × PPP Round FE Y Y Y Y Y Y
Week FE Y Y Y Y Y Y
Adj. R-sq .412 .525 .412 .526 .412 .526
N 6,034,309 6,034,309 5,947,516 5,947,516 5,947,516 5,947,516

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
B. Interactions with state-level political conditions

(1) (2) (3) (4) (5) (6)
Value of loans Number of loans Value of loans Number of loans Value of loans Number of loans

Post × Employee contributions to Republicans −0.0170 0.0036 0.7288 0.0895 0.3561 0.0463
(−0.4418) (0.6961) (7.9120) (7.0145) (5.5628) (5.0368)
Post × Employee contributions to Republicans × Republican state 0.2132 0.0244
(4.0950) (3.5741)
Post × Employee contributions to Republicans × State citizen liberal score −0.0126 −0.0015
(−7.4289) (−5.9590)
Post × Employee contributions to Republicans × State gov. liberal score −0.0074 −0.0008
(−4.6112) (−3.5175)
The Politics of the Paycheck Protection Program

Controls Y Y Y Y Y Y
Lender × county FE Y Y Y Y Y Y
County × PPP Round FE Y Y Y Y Y Y
Week FE Y Y Y Y Y Y
Adj. R-sq .414 .527 .412 .525 .412 .525
N 6,048,597 6,048,597 6,021,782 6,021,782 6,021,782 6,021,782

This table documents the effects of partisanship on PPP lending at the lender-county level. The dependent variables are Value of loans (log 1þ) and Number of loans (log 1þ). The
partisanship measure is based on employee contributions (fraction of Republican-leaning employees). In panel A, the interaction terms are further interacted with three county-level
demographic characteristics. Minority ratio is 1 - white-only population / total population in 2019. Income per capita is county income per capita in $thousands in 2019. Local workforce
teleworkability is the weighted average of the Occupational Information Network–derived work-from-home score from Dingel and Neiman (2020), weighted by the number of employees
under each NAICS code for a given county in 2019. Panel B shows the cross-sectional variations across the ideological bias in the local state. Republican state is a dummy variable that
takes the value of one if the given county is in a state controlled by Republicans in 2020 and zero otherwise. State citizen liberal score is the revised 2016 citizen ideology series following
Berry et al. (1998). State government liberal score is the 2017 state government ideology measure following Berry et al. (1998). The set of control variables includes all control variables in
column 2 of Table 9. All other variables are defined in the note to Table IA14. t-statistics are in parentheses. Robust standard errors are clustered at the state level. Variables are winsorized
at the 1st and 99th percentiles.
p < .1; p < .05; p < .01.

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
measures used across columns, Republican-leaning lenders primarily directed
loans to borrowers in Republican states. Again, this result is consistent with a
legacy-policy view.

3.5 Aggregate and real effects of partisan misalignment
3.5.1 PPP lending effects at the county level. We now assess the aggre
gate effects of our results on PPP loan origination by examining whether the
effect of misalignment aggregates at the county level. With county-level
weekly data, we run the following regression:
Yct ¼ βPostt × Political Shockc þ γXct þ μc þ μt þ εct ; (3)
where subscripts, variables, and parameters are defined similarly as before.
The dependent variable Yct is the aggregated PPP lending originating in
county c in week t. Political Shockc is constructed at the county level.
Specifically, we focus on the employee contributions of lenders originating
PPP loans in each county. For each county c, we calculate the weighted
average of these lenders’ partisanship, using their PPP lending activity in
the county as weights. For lender i, let Political Shocki represents its employ
ees’ partisanship, and TotalLoanic represents the total dollar value of PPP
loans that lender i originated in county c. The total PPP loans originating in
P
county c is then given by TotalLoanc ¼ i TotalLoanic . We average all active
lenders’ partisanship in each county as follows:
P
ðPoliticalShocki × TotalLoanic Þ
Political Shockc ¼ i : (4)
TotalLoanc
Essentially, Political Shockc measures the average partisanship of PPP
lenders active in each county, weighted by their aggregated PPP loan volumes
in the county. However, this measure is time invariant as it calculates a
weighted average of the partisanship of all PPP lenders that have ever issued
loans in the county, assigning higher weights to lenders with a larger market
share in PPP lending.
Table 11 presents the results, for which the dependent variable is the aggregate
PPP lending within a county in each week. Our measure of lender partisanship at
the county-level is the time-invariant variable Political Shockc .26 In column (1),
a county with Republican-leaning lenders (defined as having an average 78.4%
Republican-leaning employees at the 75th percentile) experience a growth rate in
weekly PPP lending of 32.26% (1.1563 × 0.279) higher than that of a “neutral”
lender (defined as having an average 50.5% Republican-leaning employees at the
25th percentile). For the number of PPP loans originated in column (2), the effect
is 12%.
26
We do not use the moving average measure of lender partisanship as it uses the lagged dependent variable in
the calculation.

34
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 11
PPP lending effects at the county level
(1) (2)
Value of loans Number of loans

Post × Employee contributions to Republicans 1.1563 0.4327
(5.5174) (3.6381)
Post × Powerful politician −0.4716 0.0680
(−3.1154) (1.2093)
Republican state 0.0440 −0.2444
(0.2315) (−3.1568)
Post × Republican state 0.1283 0.1180
(0.7712) (1.6545)
Post × Number of PPP complaints −0.7685 −0.1637
(−9.5956) (−5.6932)
Teleworkability 1.0439 −0.1452
(0.7274) (−0.3329)
Minority majority county 0.4131 0.4267
(1.1889) (2.4292)
Income per capita −0.0242 −0.0033
(−1.4362) (−0.8239)
Population −7.0858 −4.4070
(−2.6103) (−2.5342)
County FE Y Y
Week FE Y Y
Adj. R-sq .626 .872
N 117,116 117,116

This table documents the effects of partisanship on PPP lending at the county-week level. The difference-in-
differences regression is specified in Equation (3) in the county-week panel data set. The dependent variables
Value of loans (log 1þ) and Number of loans (log 1þ) measure the aggregated PPP loans originating within a
given county in each week. The partisanship measures are based on employee contributions (fraction of
Republican-leaning employees). Employee contributions to Republicans is the weighted average of the fraction
of Republican-leaning employees, weighted by the total PPP volume originating with each lender in the given
county throughout the sample period. All variables are defined in the notes to Tables 3 and IA3. t-statistics are
in parentheses. Robust standard errors are clustered at the state level. Variables are winsorized at the 1st and
99th percentiles.
p < .1; p < .05; p < .01.

Interestingly, column (2) shows that the Republican states received fewer
loans during the first PPP phase in 2020, although column (1) does not reveal
significant differences in the aggregate value of PPP loans. At first glance, this
result challenges the presidential-particularism explanation offered by Ha
(2023), who focuses solely on the second PPP phase in 2021 and finds that
the Biden administration disproportionately directed lending toward
Democratic counties. However, by examining both phases of the program
and leveraging within-county variation—unlike Ha (2023)—our empirical
strategy uncovers a more nuanced pattern. Specifically, the initial gap
observed in 2020 narrows after the Biden administration took office in
2021, as reflected in the positive and marginal significance of the interaction
terms in column (2).
These results are important because they demonstrate that lenders whose
partisanship became misaligned with that of the administration during the
PPP’s second phase engaged more actively in the program, with these effects
being tangible at the county level.

35
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
3.5.2 Employment effects at the county level. The PPP was intended to
support small businesses and their employees. Our findings show that partisan
misalignment increased loan origination during the PPP’s second phase. But
what are the real effects of this misalignment? To what extent did it help save
more jobs? We address these questions by estimating the additional number
of workers who would rely on UI in the absence of partisan misalignment
during the PPP’s second phase.
With the county-level weekly UI data, we run Equation (3) with the dependent
variable Yct as the weekly number of UI initial claims rate (labeled UI claims
rate). To capture temporal changes, we also construct a measure using the
moving average of the partisanship of PPP lenders who issued loans in the
past four weeks within a given county. Let LoanValueics denote the total PPP
lending dollar value originating with lender i in county c during week s. For each
lender i, the total PPP lending dollar value over
P the past four weeks before week t
−4
in county c is calculated as TotalLoanict ¼ ts¼t − 1 LoanValueics . This equation
allows us to construct a time-varying county-level partisanship measure for PPP
lenders, capturing contributions by lender employees to Republicans, as a mov
ing average:
P
ðPoliticalShocki × TotalLoanict Þ
Political Shockct ¼ i : (5)
TotalLoanct
Table 12 shows the results of the effects of partisan misalignment on the UI
claims rate. Partisan misalignment is associated with lower unemployment. In
column (1), the estimated coefficient on the interaction term of interest is
negative in sign and statistically significant at conventional levels. A county
with Republican-leaning lenders (with average 78.4% Republican-leaning
employees at the 75th percentile) experienced a decrease in average weekly
UI rate from 2020–2021 of 0.27 (0.9839 × 0.279) larger than a “neutral”
bank (defined as having 50.5% Republican leaning employees at the 25th
percentile). This result is economically meaningful as it represents 31.5%
of the sample mean of the UI rate. In column (2), we use the moving average
measure of lender partisanship. The estimated coefficient on the interaction
term of interest is again negative in sign and statistically significant at conven
tional levels. The decrease in the UI rate in counties with Republican-leaning
lenders (defined as having 79.3% Republican-leaning employees at the 75th
percentile) is 0.2 (0.5824 × 0.345) larger than a “neutral” bank (defined as
having 44.8% Republican-leaning employees at the 25th percentile), which is
meaningful as it corresponds to 23.2% of the sample mean. In the last column,
we include county × PPP round fixed effects to absorb potential county-level
demand shocks. The result remains largely the same as in column (2).
Overall, this evidence using county-level, weekly UI data suggests that par
tisan misalignment is associated with higher PPP payroll coverage for small
businesses. This finding is remarkable because it underscores that partisan

36
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Table 12
Employment effects at the county level
(1) (2) (3)
UI claims rate

Post × Employee contributions to Republicans −0.9839
(−1.9538)
Employee contributions to Republicans, moving average 0.0767 0.7358
(1.1280) (2.5482)
Post × Employee contributions to Republicans, moving average −0.5824 −0.5472
(−2.0487) (−1.8935)
Post × Powerful politician 0.7401 0.6581
(3.1049) (2.8013)
Post × Republican state 0.0946 0.0855
(0.2550) (0.2169)
Post × Number of PPP complaints −0.1062 −0.0377
(−1.0540) (−0.3595)
Teleworkability 0.7510 0.7097
(0.8390) (0.7623)
Minority majority county −0.4479 −0.3514
(−2.2372) (−1.8790)
Income per capita 0.0162 0.0153
(1.1548) (1.0604)
Population −3.9495 −5.1068
(−0.5730) (−0.7199)
County FE Y Y N
County × PPP round FE N N Y
Week FE Y Y N
Adj. R-sq .665 .621 .748
N 48,074 44,026 43,650

This table documents the effects of partisanship on county unemployment at the county-week level. The
difference-in-differences regression is specified in Equation (3) in the county-week panel data set. The depend
ent variable is the UI claims rate, which is the number of initial claims per 100 people in the 2019 county labor
force. The partisanship measure is based on employee contributions (fraction of Republican-leaning employ
ees). In column 1, Employee contributions to Republicans is the weighted average of fractions of Republican-
leaning employees, weighted by the total PPP volume originating with each lender in the given county
throughout the sample period. In columns 2 and 3, Employee contributions to Republican, moving average
is the weighted average of the fraction of Republican-leaning employees, weighted by the total PPP volume
originating with each lender during the previous 4 weeks in the given county. All variables are defined in the
notes to Tables 3 and IA3. t-statistics are in parentheses. Robust standard errors are clustered at the state level.
Variables are winsorized at the 1st and 99th percentiles.
p < .1; p < .05; p < .01.

misalignment has real effects that extend beyond its direct effects on PPP loan
origination.

4. Conclusion
Initially authorized for $659 billion in business loans, the PPP ultimately
grew into an $800-billion program, an unprecedented scale in U.S. history
(Autor et al. 2022a). The program has faced significant criticism regarding the
allocation of funds. This paper represents the first systematic attempt to
examine whether partisan alignment influenced PPP aid allocation, a crucial
issue in the politically polarized context of its implementation.

37
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Our findings show that partisan alignment significantly affects PPP loan
allocation, revealing how political affiliations within financial institutions
shape lending decisions in government aid programs. Specifically, we exploit
the program’s phased rollout to demonstrate that partisan misalignment
between lenders and the administration resulted in increased loan origination
during the PPP’s second phase, particularly benefiting small and first-time
PPP borrowers as well as those in Republican-leaning areas. The effects of
partisan misalignment at the bank level are also evident at the aggregate
(county) level. We further show the real effects: partisan misalignment is
associated with higher payroll coverage for small businesses through PPP
loans.
Combined, our results suggest that politically misaligned lenders may
respond to legacy programs of a former administration with heightened
engagement under a new administration, potentially to maintain or bolster
their political capital. A complementary explanation, which we cannot fully
disentangle, relates to who ultimately bears the credit risk. Since PPP loans
are forgivable, the administration in charge bears the cost of defaults, allow
ing misaligned (Republican-leaning) lenders to adopt more aggressive lend
ing strategies. These findings paint a rather “benign” picture of political
polarization in finance, challenging the notion that partisan-misaligned behav
ior in financial institutions always leads to negative outcomes. Therefore, the
findings carry significant implications for the effectiveness of government aid
programs and provide new insights into the intersection of politics and
finance, particularly regarding the partisan-alignment phenomenon.

Code Availability
The replication code and data are available in the Harvard Dataverse at https://
dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/PG8CLJ

References
Agarwal, S., G. Amromin, I. Ben-David, and S. Dinc. 2018. The politics of foreclosures. Journal of Finance
73:2677–2717.

Akey, P., R. Z. Heimer, and S. Lewellen. 2021. Politicizing consumer credit. Journal of Financial Economics
139:627–55.

Autor, D., D. Cho, L. D. Crane, M. Goldar, B. Lutz, J. Montes, W. B. Peterman, R. Ratner, D. Villar, and A.
Yildirmaz. 2022a. The $800 billion Paycheck Protection Program: Where did the money go and why did it go
there? Journal of Economic Perspectives 36:55–80.

———. 2022b. An evaluation of the Paycheck Protection Program using administrative payroll microdata.
Journal of Public Economics 211:104664.

Balyuk, T., N. R. Prabhala, and M. Puri. 2021. Small bank financing and funding hesitancy in a crisis: Evidence
from the Paycheck Protection Program. Working Paper.

Bartik, A. W., M. Bertrand, Z. Cullen, E. L. Glaeser, M. Luca, and C. Stanton. 2020a. The impact of COVID-19
on small business outcomes and expectations. Proceedings of the National Academy of Sciences 117:17656–66.

38
The Politics of the Paycheck Protection Program

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Bartik, A. W., Z. B. Cullen, E. L. Glaeser, M. Luca, C. T. Stanton, and A. Sunderam. 2020b. The targeting and
impact of Paycheck Protection Program loans to small businesses. NBER Working Paper No. 27623.

Bartlett, R. P., and A. Morse. 2021. Small-business survival capabilities and fiscal programs: Evidence from
Oakland. Journal of Financial and Quantitative Analysis 56:2500–44.

Berger, A. N., M. U. Karakaplan, and R. A. Roman. 2023. Whose bailout is it anyway? The roles of politics in
PPP bailouts of small businesses vs. banks. Journal of Financial Intermediation 56:101044.

Berry, W. D., E. J. Ringquist, R. C. Fording, and R. L. Hanson. 1998. Measuring citizen and government
ideology in the American states, 1960–93. American Journal of Political Science 42:327–48.

Boone, C., A. Dube, and E. Kaplan. 2014. The political economy of discretionary spending: Evidence from the
American Recovery and Reinvestment Act. Brookings Papers on Economic Activity Spring:375–428.

Bouton, L., J. Cage, E. Dewitte, and V. Pons. 2022. Small campaign donors. NBER Working Paper No. 30050.

Cage, J. 2024. Political inequality. Annual Review of Economics 16:455–490.

Cassidy, W., and B. Vorsatz. 2021. Partisanship and portfolio choice: Evidence from mutual funds. Working
Paper.

Chavaz, M., and A. K. Rose. 2019. Political borders and bank lending in post-crisis America. Review of Finance
23:935–59.

Chen, D. 2020. Priming ideology: Why presidential elections affect US judges. Working Paper.

Chernenko, S., and D. Scharfstein. 2024. Racial disparities in the Paycheck Protection Program. Journal of
Financial Economics 160:103911.

Chernenko, S., N. Kaplan, A. Sarkar, and D. S. Scharfstein. 2023. Applications or approvals: What drives racial
disparities in the Paycheck Protection Program? Working Paper.

Chetty, R., J. N. Friedman, N. Hendren, M. Stepner, and the Opportunity Insights Team. 2020. The economic
impacts of COVID-19: Evidence from a new public database built from private sector data. NBER Working
Paper No. 27431.

Chodorow-Reich, G., O. Darmouni, S. Luck, and M. Plosser. 2022. Bank liquidity provision across the firm size
distribution. Journal of Financial Economics 144:908–32.

Coibion, O., Y. Gorodnichenko, and M. Weber. 2020. Political polarization and expected economic outcomes.
NBER Working Paper No. 28044.

Cororaton, A., and S. Rosen. 2021. Public firm borrowers of the US Paycheck Protection Program. Review of
Corporate Finance Studies 10:641–93.

Dagostino, R., J. Gao, and P. Ma. 2023. Partisanship in loan pricing. Journal of Financial Economics
150:103717.

Dingel, J. I., and B. Neiman. 2020. How many jobs can be done at home? Journal of Public Economics
189:104235.

Duchin, R., A. E. K. Farroukh, J. Harford, and T. Patel. 2023. The economic effects of political polarization:
Evidence from the real asset market. Working Paper.

Duchin, R., and J. Hackney. 2021. Buying the Vote? The economics of electoral politics and small-business
loans. Journal of Financial and Quantitative Analysis 56:2439–73.

Elenev, V., T. Landvoigt, and S. Van Nieuwerburgh. 2022. Can the COVID bailouts save the economy?
Economic Policy 37:277–330.

Engelberg, J., J. Guzman, R. Lu, and W. Mullins. 2023. Partisan entrepreneurship. Journal of Finance July 3.

Fishback, P. V., S. Kantor, and J. J. Wallis. 2003. Can the New Deal’s three Rs be rehabilitated? A program-by-
program, county-by-county analysis. Explorations in Economic History 40:278–307.

39
Review of Corporate Finance Studies / v 00 n 00 2025

Downloaded from https://academic.oup.com/rcfs/advance-article/doi/10.1093/rcfs/cfaf013/8180554 by Erasmus University Rotterdam user on 15 August 2025
Gerber, A. S., G. A. Huber, and E. Washington. 2010. Party affiliation, partisanship, and political beliefs: A field
experiment. American Political Science Review 104:720–44.

Granja, J., C. Makridis, C. Yannelis, and E. Zwick. 2022. Did the Paycheck Protection Program hit the target?
Journal of Financial Economics 145:725–61.

Ha, H. 2023. Presidential particularism, the Paycheck Protection Program PPP, and the Biden administration.
Presidential Studies Quarterly 53:543–69.

Hanson, S. G., J. C. Stein, A. Sunderam, and E. Zwick. 2020. Business credit programs in the pandemic era.
Brookings Papers on Economic Activity 2020:3–60.

Hassan, T. A., S. Hollander, L. Van Lent, and A. Tahoun. 2019. Firm-level political risk: Measurement and
effects. Quarterly Journal of Economics 134:2135–2202.

Howell, S. T., T. Kuchler, D. Snitkof, J. Stroebel, and J. Wong. 2024. Lender automation and racial disparities in
credit access. Journal of Finance 79:1457–1512.

Humphries, J. E., C. Neilson, and G. Ulyssea. 2020. The evolving impacts of COVID-19 on small businesses
since the CARES Act. Working Paper.

Hutton, I., D. Jiang, and A. Kumar. 2014. Corporate policies of Republican managers. Journal of Financial and
Quantitative Analysis 49:1279–1310.

Ioannidou, V., S. Kokas, T. Lambert, and A. Michaelides. 2023. (In)dependent central banks. CEPR Discussion
Paper No. 17802.

Jones, J. M. 2020. Trump third year sets new standard for party polarization. Gallup. Jan 21. https://news.gallup.
com/poll/283910/trump-third-year-sets-new-standard-party-polarization.aspx

Kempf, E., and M. Tsoutsoura. 2021. Partisan professionals: Evidence from credit rating analysts. Journal of
Finance 76:2805–56.

———. 2024. Political polarization and finance. Annual Review of Financial Economics 16:413–34.

Kempf, E., M. Luo, L. Sch€afer, and M. Tsoutsoura. 2023. Political ideology and international capital allocation.
Journal of Financial Economics 148:150–173.

Lambert, T., and P. Volpin. 2018. Endogenous political institutions and financial development, in Handbook of
Finance and Development, eds. T. Beck and R. Levine. London: Edward Elgar.

Lambert, T., W. Wagner, and E. Q. Zhang. 2023. Banks, political capital, and growth. Review of Corporate
Finance Studies 12:613–55.

Li, L., and P. E. Strahan. 2021. Who supplies PPP loans and does it matter? Banks, relationships, and the COVID
crisis. Journal of Financial and Quantitative Analysis 56:2411–38.

Maine Democratic Party. 2020. New poll: Plurality of voters think paycheck protection program funds were
distributed unfairly. May 7. MEDEMS. https://www.mainedems.org/media/new-poll-plurality-voters-think-
paycheck-protection-program-funds-were-distributed-unfairly

Meier, J.-M., and J. Smith. 2020. The COVID-19 bailouts. Working Paper.

Page, B. I., and C. C. Jones. 1979. Reciprocal effects of policy preferences, party loyalties and the vote. American
Political Science Review 73:1071–1089.

© The Author(s) 2025. Published by Oxford University Press on behalf of The Society for Financial Studies.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction
in any medium, provided the original work is properly cited.
Review of Corporate Finance Studies, 2025, 00, 1–40
https://doi.org/10.1093/rcfs/cfaf013
Article

File and source

File: the-politics-of-the-paycheck-protection-program.pdf
Size: 1,928,348 bytes
SHA-256: 3b11f1864766b90140cbbb22efc101761018fa5b0526834a511985ceb45861c1
Our copy: the-politics-of-the-paycheck-protection-program.pdf
Original: doi.org