Corporate Taxation and Bank Outcomes: Evidence from U.S. ...

Corporate Taxation and Bank Outcomes: Evidence from U.S. State Taxes

John Gallemore

University of Chicago

[email protected]

Michael Mayberry

University of Florida

[email protected]

Jaron Wilde

University of Iowa

[email protected]

September 2017

We appreciate helpful comments from Dan Collins, Cristi Gleason, Jeff Hoopes, Anya Kleymenova, Jon

Medrano, and workshop participants at University of Florida. John Gallemore gratefully acknowledges

financial support of the University of Chicago Booth School of Business and the Charles E. Merrill Faculty

Research Fund.

Corporate Taxation and Bank Outcomes: Evidence from U.S. State Taxes

Abstract: We examine the association between corporate taxation and bank outcomes, including lending

growth, leverage, liquid asset holdings, and risk-taking. To study these relations, we employ a sample of

single-state U.S. commercial banks and exploit cross-sectional and inter-temporal variation in U.S. state

income tax rates, a setting that allows us to better isolate the effect of taxation from other correlated factors.

Overall, we fail to find that changes in income tax rates impact bank activities on average. However, we

find compelling evidence that the tax rate has significant effects on specific types of banks and during

periods of economic downturn and credit risk uncertainty, highlighting potentially significant policy-

relevant implications for regulators and policy makers. Our collective findings demonstrate how corporate

income taxation affects bank outcomes, such as lending and leverage, which in turn affect the capital

available for both non-bank corporations and individuals

Keywords: corporate taxation, banks, lending, leverage, liquidity, risk-taking

JEL codes: G01, G21, G32, H25, H71

1

1. Introduction

With few exceptions, we know little about how corporate taxation affects banks (Hanlon and

Heitzman 2010). Understanding whether corporate taxation is associated with bank choices has important

policy implications because of the central role that banks play in spurring economic activity by providing

liquidity and capital for businesses and individuals. We follow prior research on corporate taxation (Doidge

and Dyck 2015; Beatty, Chamberlain, and Magliolo 1995; Collins, Shackelford, and Wahlen 1995) and

take a holistic approach to studying the effect of taxation on banks by examining several bank outcomes:

lending, leverage, liquidity, and risk-taking. Our study contributes to a more complete understanding of

how corporate taxation interacts with bank decisions.

Our empirical methodology exploits variation in bank income tax rates across U.S. states. This

setting offers several advantages for our study. First, bank income taxation differs across states and within

a state across time, providing necessary and plausibly exogenous variation for exploring the effect of

taxation on bank outcomes. Second, state corporate income taxes are an economically significant corporate

expense, accounting for approximately 21 percent of the total income taxes paid in Compustat (Heider and

Ljungqvist 2015). Prior research shows that state corporate income taxes have an economically meaningful

effect on corporate behavior for non-banks, including capital structure (Heider and Ljungqvist 2015) and

risk-taking (Ljungqvist, Zhang, and Zuo 2016). Third, our setting exploits within-country variation which,

compared to potential cross-country settings, has the benefit of a relatively homogenous regulatory and

economic environment. Fourth, the U.S. setting is relatively unique in terms of the availability of high-

quality banking data from commercial bank call reports, which allow us to examine specific bank outcomes

while controlling for other factors.

Most notably, U.S. bank regulatory filings reveal the states in which commercial banks have

deposits, allowing us to assemble a sample of single-state commercial banks, mitigating concerns common

in many state income tax studies regarding unobservable heterogeneity related to firms allocating income

2

across and having economic nexus in multiple states.1 Employing this banking setting strengthens our

ability to draw inferences regarding the effect of corporate income tax changes on bank outcomes. However,

our setting potentially suffers from endogeneity to the degree that state tax rates are correlated with state-

year factors that also impact bank outcomes. We include variables that control for state-year-level

macroeconomic conditions that simultaneously affect tax policy and bank choices such as lending and

liquidity, and include state-year fixed effects in certain specifications. Furthermore, to mitigate concerns

about concurrent changes in state banking regulation, we separately examine the effects of taxes on

federally-chartered banks which, while subject to state income tax, are exempt from state banking

regulation. While we cannot conclusively rule out correlated omitted variable concerns, we believe that our

design mitigates concerns about the most obvious factors.

Overall, we generally find little evidence of income taxation impacting banks on average. However,

we find compelling evidence that income taxation affects certain banks outcomes for particular types of

banks as well as during periods of economic downturn and credit risk uncertainty. In this way, our findings

are similar to concurrent research that examines the effect of corporate income taxation on economic

outcomes such as employment (Ljungqvist and Smolyansky 2016). The totality of our evidence points to a

nuanced relation between corporate taxation and bank outcomes, offering policy-relevant insights.

First, we examine how tax rates are correlated with loan growth. Understanding the determinants

of loan growth is important because banks are a critical source of capital for both individuals and

corporations. In particular, the banks in our sample are a key provider of small business lending (Keeton,

Harvey, and Willis 2003).2 Given that corporate taxation reduces the cash flows available for new lending

and reduces the after-tax rate of return on lending, higher tax rates could in turn constrain lending. On the

1 A clear drawback to this sample of banks is that they tend to be small on average. Our sample comprises 31

percent of the total loans and 29 percent of the total deposits in the U.S. banking sector. This feature of our sample

suggests that, while we cannot speak to the effect of corporate income taxation on large banks, our sample still

represents an economically important part of the U.S. banking system. We believe the economic relevance of the

sample coupled with the significant improvement in internal validity makes investigating these banks worthwhile. 2 Specifically, Berger, Miller, Petersen, Rajan, and Stein (2005) note that community banks enjoy competitive

advantages in small business lending due to their ability to utilize soft information.

3

other hand, banks may choose to increase lending (and potentially to riskier borrowers in particular) in

order to offset the after-tax rate of return effect of income taxation. We do not find evidence that income

tax changes are associated with loan growth on average, whether using a measure of traditional lending or

a measure that also captures unexercised loan commitments (Cornett, McNutt, Strahan, and Tehranian

2011). However, we find that tax rate changes are positively associated with loan growth during non-

recessionary periods and periods with low credit risk costs and uncertainty, and that the relation becomes

negative during recessionary periods and periods with high credit risk costs and uncertainty. These findings

suggest that during periods in which alternative methods of locating liquid assets are more costly or less

available, banks rely more heavily on operating cash flows to fund lending and are thus negatively affected

by income taxation. On the other hand, during normal periods higher tax rates can actually be associated

with greater lending.

Next, we study whether tax rates are associated with bank leverage. Prior research documents that

corporate taxation – through the deductibility of interest expense – encourages debt financing for non-

financial institutions.3 However, it is not clear that income taxation will have a similar effect on banks. For

example, a bank’s business model is predisposed to high leverage ratios because of its deposit-taking

function, and therefore the incremental incentive to increase leverage for tax benefits may be small if it

exists at all. Moreover, bank leverage is implicitly regulated given that banks are subject to minimum capital

requirements, a key difference from non-financial firms. While a few studies have examined the capital

structure of financial institutions, such studies either utilize a cross-country setting where correlated omitted

variables abound (Keen and de Mooij 2012; de Mooij, Keen, and Orihara 2013) or small-sample, within-

country one-off events that do not allow the researcher to investigate heterogeneity in the effect of taxation

on leverage across economic cycles and bank types (Schepens 2016). Moreover, these studies generally are

unable to speak to the specific leverage components that banks change in response to tax rate changes.

3 See Graham (2005) and Hanlon and Heitzman (2010) for thorough reviews of this literature.

4

We do not find evidence of an average effect of taxation on leverage. However, we do find that the

effect varies significantly in the cross-section. In particular, we find that association between tax rate

changes and leverage changes is increasing in the bank’s capital ratio, suggesting that better capitalized

banks have a greater ability to increase leverage in response to changes in taxation. We also provide

evidence that bank leverage appears to be positively associated with tax rate changes during normal periods,

but that the relation becomes either insignificant or negative during recessions and periods of uncertainty.

In terms of the components of leverage, we find that uninsured deposits and non-deposit forms of leverage,

which are generally more prone to runs than are insured deposits, tend to be the funding sources most

sensitive to tax rate changes. This finding suggests that tax rate increases may lead to increased funding

risk, even if it does not lead to increases in leverage on average.

Third, we examine how corporate taxation interacts with liquidity choices. Bank illiquidity played

a prominent role in the 2007-09 financial crisis (Cornett et al. 2011), and regulators have subsequently

focused heavily on reforming liquidity regulations (Allen 2014; Wall 2015; Diamond and Kashyap 2016).

Banks need sufficient liquidity on hand to satisfy funding providers’ needs (such as funds withdrawal), and

generally rely on a mix of operating cash flows and liquid asset holdings to manage liquidity risk. As

corporate taxation decreases the amount of pre-tax operating cash flows that are available to use for liquidity

management, increases in the corporate tax rate could lead banks to increase liquid asset holdings. We find

that on average, bank income taxes are not significantly associated with liquid asset holdings. However, we

again find that this effect varies in the cross-section. In particular, tax rate changes are positively

(negatively) associated with changes in liquid asset holdings during recessions and periods of higher and

more uncertain interbank lending costs (during normal periods). This finding suggests banks increase liquid

asset holdings in response to tax rate increases when alternative external sources of liquid funds are less

available or more costly.

Finally, we examine whether corporate taxation is associated with bank risk-taking. While prior

research has explored how corporate taxation affects risk-taking for non-banks (Ljungqvist et al. 2016;

Lester and Langenmayr 2017), this topic has not yet been explored within banks. Understanding the

5

determinants of bank risk-taking is important because prior research (Bernanke 1983; Keeley 1990;

Calomiris and Mason 1997, 2003b, 2003a) and past financial crises suggest that bank risk-taking

significantly affects economic stability. We find income tax rates exhibit little association with measures

of bank risk-taking on average. However, we again find that the effect varies in the cross-section. Tax rate

changes are positively (negatively) associated with changes in risk-taking during economic expansions

(recessions) and periods with relatively lower (higher) credit risk costs and uncertainty. One potential

explanation for this result is that during these non-crisis periods, banks adjust their risk-taking in response

to tax rate changes in an effort to compensate for changing tax burdens and maintain certain levels of after-

tax returns, consistent with our lending results. However, during periods of economic downturn and

uncertainty, bank risk-taking is negatively associated with tax rate changes, consistent with such periods

constraining banks’ ability to attract capital and recover from unprofitable investments. Along these lines,

we also find that the relation between tax rate changes and risk-taking is positive for more profitable banks.

Our primary contribution is to provide empirical evidence on the effect of corporate taxation on

bank operations. Although banks play a central role in economic activity worldwide, with a few exceptions,

prior studies on corporate taxation generally ignore banks, often removing financial institutions from their

samples (Hanlon and Heitzman 2010). One important exception is the stream of literature that examines

the effects of corporate taxation on bank leverage choices (Keen and de Mooij 2012; de Mooij et al. 2013;

Schepens 2016).4 Our study contributes to the literature on corporate taxation by providing evidence on

important bank decisions that are associated with corporate income taxation. We utilize a within-country

setting with numerous tax rate changes to provide better internal validity while maintaining a sufficient

sample size to test heterogeneity in the tax-leverage association across economic cycles and bank types.

Relatedly, we add to the literature on the effect of corporate taxation on banks’ various, and

sometimes-conflicting, objectives. Prior studies investigate the interaction between banks’ various

objectives, including earnings, regulatory capital, and taxes (Beatty et al. 1995; Collins et al. 1995; Hodder,

4 Another exception is Andries, Gallemore, and Jacob (2017), which shows that corporate taxation increases timely

loan loss recognition, which could improve bank stability through greater bank transparency.

6

McAnally, and Weaver 2003). We add to this literature in two ways: (1) our design allows us to better

identify the tax frictions that affect bank choices, and (2) we focus on various core banking activities,

including increasingly important but previously omitted objective, liquidity risk management, and show

how taxation interacts with them.

Finally, we extend the research examining the effects of state taxation on corporations. Prior

research suggests state taxation provides significant tax planning opportunities (Dyreng, Lindsey, and

Thornock 2013). Consistent with state taxes representing an economically large cost to corporations, extant

evidence suggests that changes in corporate income taxation have significant effects on capital structure

(Heider and Ljungqvist 2015) and risk-taking (Ljungqvist et al. 2016) for non-financial institutions. We

contribute evidence to this literature by shedding light on the nuanced effect that state tax rate changes have

on a host of economically important bank choices, including lending, leverage, liquidity, and risk-taking.

Understanding these bank choices is important for at least two reasons. First, banks play a central role in

providing liquidity and capital necessary to cultivate economic growth in non-banking sectors. Second,

bank risk taking and liquidity influence bank failure (Jin, Kanagaretnam, and Lobo 2013), which is costly

to taxpayers (Sidel 2011), impairs local personal income and employment growth, and increases poverty

rates (Kandrac 2014). Our results are therefore of interest to investors, banking regulators, and

policymakers, who collectively benefit from policies that foster bank health and activity.

2. Conceptual Framework

Recent research provides evidence that tax attributes such as rate and loss carryback provisions

influence corporate activities (e.g., Doidge and Dyck (2015), Heider and Ljungqvist (2015), Langenmayr

and Lester (2015), and Ljungqvist et al. (2016)). While recent work has improved researchers’

understanding of the role taxes play in non-financial firm policies, there is still much debate regarding how

taxes influence firm outcomes and decisions (e.g., Fama (2011), Myers, McConnell, Peterson, Soter, and

Stern (1998), Doidge and Dyck (2015). To date, there is far less evidence on the role taxes play in bank

decision-making because much of the research examining taxes and corporate decision-making omits

financial firms, ostensibly due their unique regulatory and operating environments.

7

The focus of this study is examining the effect of taxes on four economically important banking

activities: lending (loan growth), leverage (funding), liquidity management, and risk-taking. By

investigating several outcome variables together, we follow prior research that explores the effect of

taxation on multiple, interrelated firm outcomes. For example, Doidge and Dyck (2015) examines several

firm responses, including changes in investments, leverage, cash holdings, and corporate payouts, to

changes in a Canadian tax law that imposed an entity-level income tax on publicly traded income trust firms

which were previously exempted from the corporate income tax system. Several other studies explore the

trade-offs firms make between earnings, capital, and tax objectives (e.g., Beatty et al. (1995), Collins et al.

(1995)). The benefit of investigating several outcome variables is that we can provide a more complete

picture of the effects of corporate income taxation on banks.

2.1. Loan growth

A long literature highlights the effects of taxes on corporate investment activities (see Hanlon and

Heitzman (2010) for a review). Lending represents the fundamental investment feature of a bank’s business

model, as bank earnings derive primarily from the interest and fees generated on loans. Moreover, as

lending is a key source of capital for firms and entrepreneurs, understanding the role of corporate taxation

on bank lending is critical in understanding non-banks access to capital and financing.

It is unclear how changes in income taxation will influence lending outcomes. On the one hand,

increases in income tax rates mechanically reduce after-tax income, all else equal. Because increases in

income taxes divert operating cash flows to the government that banks would otherwise deploy to new

lending arrangements, increases in tax rates should be negatively associated with lending activity. However,

such an argument assumes that bank managers are unwilling to sacrifice loan quality or find alternative

sources of capital to fund loan activity in the face of increasing tax rates. Given the preeminence of lending

in banks’ business models, changes in tax rates may induce banks to compensate for the reduced operating

cash flows by changing funding sources or cutting other operating expenditures before sacrificing profitable

lending opportunities. Alternatively, if bank managers are focused on maintaining a targeted level of net

income, in spite of tax rate increases, then banks may actually increase lending activities in an effort to

8

compensate for higher tax costs. Such increases in lending arise as managers lower the minimum acceptable

quality of their loans, thereby charging higher interest rates on new lending and increasing pretax rates of

return. In summary, it an open empirical question whether and how tax rates impact bank lending.

2.2. Leverage

Prior research examines the role of taxes in corporate capital structure and financing decisions (see,

for example, Graham (2003) and Hanlon and Heitzman (2010)). Although the notion of interest tax shields

is intuitive, with interest deductions reducing after-tax costs of debt, the role taxes play in such decisions is

the subject of continued debate (e.g., Doidge and Dyck (2015), Fama (2011), Heider and Ljungqvist (2015),

Myers et al. (1998)). In a recent paper, Heider and Ljungqvist (2015) use changes in state corporate income

tax rates to investigate the effect of taxes on capital structure. They document that taxes have a first-order,

positive effect on non-financial firms’ leverage. While interest tax shields are intuitive in an industrial or

nonfinancial context, it is not clear whether the corporate tax system will similarly impact the banking

sector for at least two reasons. First, in contrast to non-financial firms, banks’ business models mandate the

funding of longer term investments with short-term demandable deposits, thereby inherently requiring

leverage.5 Indeed, despite the interest tax shield’s availability to banks and non-banks firms, bank leverage

is considerably higher on average compared with non-bank leverage. Second, unlike non-financial firms,

bank regulators set strict minimum capital requirements and closely monitor bank leverage. Deviations

from statutorily mandated leverage ratios can result in penalties and even bank receivership. Thus, prior

evidence on the relation between corporate income taxation and the leverage of non-financial firms does

not necessarily inform researchers on the role of taxation and bank leverage.

Moreover, while nearly all banks draw from short-term funding sources to fund long-term

investments (i.e., loans), the composition of this short-term funding dramatically varies across banks. Banks

have at their disposal various mixes of insured deposits, uninsured deposits (e.g., jumbo CDs), and short-

5 In the aftermath of the 2007-09 financial crisis, there has been a debate in the finance literature about the

optimality of high leverage for banks (see DeAngelo and Stulz (2015); Admati, DeMarzo, Hellwig, and Pfleiderer

(2013)).

9

term subordinated non-deposit debt. Each of these funding sources faces different adjustment costs and risk

profiles and thus could be differentially sensitive to tax rates. Given that non-deposit funding is typically

longer in duration, at a higher cost than deposits, and uninsured by the FDIC, we expect banks’ non-deposit

funding to be incrementally more sensitive to tax rate changes than deposit funding. If so, then to the extent

that short-term non-deposit debt is more subject to runs than deposits (and particularly insured deposits),

the corporate tax system could affect the susceptibility of banks to runs by funding providers.

2.3. Liquidity

All else equal, banks typically prefer to hold fewer liquid assets because they generate far less

income compared with illiquid assets, such as loans. Thus, a key element of bank operations is managing

liquidity risk, or the risk that a bank fails to have sufficient funds on hand when a funding provider

withdraws funds or a borrower draws down on a credit line. The 2007-09 financial crisis highlighted the

severity of bank and economy-wide spillover effects arising from insufficient liquidity. Indeed, some

suggest that the financial crisis was ultimately a liquidity crisis (Cochrane 2009), and regulators took

emergency actions during the crisis to improve banks’ liquidity (e.g., the FDIC’s Temporary Liquidity

Guarantee Program and the Federal Reserve’s creation of several liquidity facilities (Federal Reserve Bank

of St. Louis)). After the crisis, regulators across the globe responded with proposed regulations aimed at

mitigating future liquidity concerns.6

To date, research on bank liquidity risk has primarily focused on documenting the outcomes of

liquidity risk, especially during crisis periods. The heavy reliance of many financial institutions on short-

term funding in the lead-up to the 2007-09 financial crisis, coupled with the sharp increases in illiquidity in

many funding markets, led to runs by short-term funding providers during the crisis (Allen 2014; Ivashina

and Scharfstein 2010). Borrowers’ run on credit lines to access capital only compounded the scarcity of

liquidity (Ivashina and Scharfstein 2010) and banks’ ability to cope with these funding shocks varied with

6 For example, one part of Basel III requires banks to maintain certain liquidity coverage ratios (LCR), so that they

could survive a distressed funding scenario. Furthermore, Basel III includes provisions for a net stable funding

ratio (NSFR) which requires banks to address liquidity mismatches (Allen 2014).

10

their exposures (Cornett et al. 2011) and with explicit government support (Acharya and Mora 2015).

Moreover, liquidity shocks during this crisis also led banks to restrict lending (Ivashina and Scharfstein

2010; Cornett et al. 2011). Relatively few studies have explicitly investigated the determinants of liquidity

risk outside the financial crisis. Kashyap, Rajan, and Stein (2002) show that deposits and credit lines are

complements because they share the costly burden of liquid asset holdings. Similarly, Gatev, Schuermann,

and Strahan (2009) document that transaction deposits serve as a natural liquidity hedge for unused credit

line commitments. But there is little evidence on the role taxes play in banks’ liquidity risk management.

Corporate taxation should affect bank liquidity management through its negative effect on

operating cash flows. The income tax diverts a percentage of every dollar of operating cash flows to the

government, and thus only a portion of each dollar of operating cash flow is available to satisfy liquidity

needs. As the tax rate increases, the bank’s ability to rely on operating cash flows for liquidity needs

diminishes. Assuming operating cash flows and liquid asset holdings are substitutes, we expect a bank will

need to hold a higher proportion of liquid assets under higher income tax rates. That said, banks can manage

their liquidity risk using different methods, such as drawing on alternative funding sources, reducing

investment in illiquid assets, or reducing non-tax operating costs. Thus, whether tax rate changes influence

bank liquidity management is unclear.

2.4. Risk-taking

Prior research finds that taxes have a significant effect on firm risk-taking. For example, Ljungqvist

et al. (2016) use staggered changes in state corporate income tax rates to investigate the effect of tax on

non-financial firms’ risk-taking activities. They argue that taxes affect risk-taking asymmetrically, with the

government sharing in upside (profits) of risky projects, but not the downside of such projects. They show

that tax rate increases are associated with firms reducing risk, but do not find evidence that tax rate decreases

have a corresponding effect. Consistent with tax loss offsets allocating some of the downside risk to the

government, they find offsetting tax loss rules moderate the effect of taxes on risk-taking. Langenmayr and

Lester (2015) similarly show that risk-taking is positively associated with the length of tax loss periods and

that tax rates are positively (negatively) related to risk-taking for firms expecting to use losses (for firms

11

that are unable to utilize the losses). To our knowledge, no prior research has extended these studies to the

banking sector. The unique regulatory environment in which banks operate suggests that it is unclear

whether taxes influence bank risk-taking in ways similar to industrial firms.

Banks are subject to the same body of traditional regulators (e.g., auditors, investors, analysts,

financial and tax regulators, etc.). However, banks also are subject to intense and periodic scrutiny from

banking regulators, who are particularly focused on bank viability, liquidity, and capital adequacy. For

example, the Federal Deposit Insurance Corporation (FDIC) Risk Management Manual of Examination

Policies notes that “examination activities center on evaluating an institution’s capital, assets, management,

earnings, liquidity, and sensitivity to market risk” (Federal Deposit Insurance Corporation 2017). Moreover,

banks face greater costs for increases in risk, as their capital ratios and deposit insurance are risk weighted.

Thus, the stringent monitoring environment in which banks operate coupled with increased costs of risk

taking potentially mitigates tax-induced risk-taking.

Understanding whether taxation affects bank risk-taking has particular policy relevance. Excessive

bank risk-taking has been cited as a significant contributor to bank problems during the 2007-09 financial

crisis (Ellul and Yerramilli 2013; Kashyap 2010). Several studies have examined determinants of bank risk-

taking, such as monetary policy (Dell'Ariccia, Laeven, and Suarez 2017), CEO overconfidence (Ho, Huang,

Lin, and Yen 2016), and government assistance during the recent crisis (Duchin and Sosyura 2014).

However, to our knowledge, there is no empirical evidence on the association between income taxation and

bank risk-taking. Given the systematic importance of banks, in particular, to the larger economy, studying

specific bank risk consequences to corporate tax policy is important in its own right.

2.5 Heterogeneity in the Effect of Taxes on Bank Activities

Prior research investigating the effect of taxes on economic outcomes suggests that the impacts of

tax rate changes are often nuanced and context-specific. For example, Ljungqvist and Smolyansky (2016)

provide evidence that state corporate income tax rate cuts increase economic activity only during recessions,

when they have significant positive effects on employment and income. There are reasons to expect the

effect of corporate taxation on banking activities to also vary with economic conditions and firm attributes.

12

For instance, during recessions, funding providers may be more likely to run from the bank (Gertler and

Kiyotaki 2015). Therefore, the importance of internal funding (i.e. operating cash flows) for liquidity risk

management, and the negative effect of income taxation on liquidity risk, should be stronger during

recessions. Similarly, banks’ potential sensitivity during periods of deteriorating economic conditions may

influence how tax changes affect bank lending, given such conditions likely reduce their ability to draw on

alternative funding sources for investment. Finally, raising leverage in response to tax rate increases may

be more difficult during recessions and periods with high credit risk costs and uncertainty, and therefore

tax rate changes may only be positively associated with changes in leverage during normal periods.

Periods of economic uncertainty are also likely to impact the effect of taxes on bank activities.

Banks often rely on the interbank market in order to manage liquidity risk on a short-term basis. Banks’

ability to rely on the interbank market is straightforward and predictable when credit risk in the interbank

market is relatively low and stable. However, when there is high uncertainty in credit risk (e.g., high

volatility in the interbank market), banks may be unable to access the interbank market for liquidity risk

management purposes, and instead shift focus to operating cash flows and liquid asset holdings. Therefore,

tax rate changes may exhibit positive association with liquid asset holdings when interbank funding markets

are not reliable alternative sources of liquid assets. Such a setting might also lead the bank to also cut

lending, as more liquid assets are held internally rather than loaned out.

Similarly, bank characteristics may moderate the effect of taxation on bank outcomes. For example,

prior research suggests that large and small banks have different sources of funding and capitalization, and

are differentially affected by crisis periods (Beatty and Liao 2011; Cornett et al. 2011; Berger and Bouwman

2013). Therefore, larger banks may respond differentially to tax rate changes relative to smaller banks; for

example, richer information environments that reduce borrowing costs coupled with the ability to draw

from larger investor and depositor bases may attenuate larger banks’ sensitivity to income tax rate changes.

Given evidence that bank profitability is linked to earnings and capital management activities (Collins et

al. 1995), it is possible that bank profitability may be a factor that moderates for the relation between state

income tax rate changes and real bank activities. Although income tax rate changes likely have a greater

13

impact on the cash tax payments of more profitable banks, such profitability may also allow banks to

cushion capital reserves, access funding channels, and weather the changes in tax provisions.

Bank capitalization in particular likely plays an important role in the association between tax rate

changes and bank outcomes. For example, Beatty and Liao (2011) suggest that capital concerns (e.g., during

recessionary periods) affect lending activities and that such lending activities are related to firms’ loan loss

provisioning. Capital requirements impose limits on bank leverage, and thus, the association between tax

rate and leverage changes is likely increasing in bank capitalization. Additionally, poorly capitalized banks

may be at greater risk of a funding run (either from depositors, short-term funding providers, or credit line

holders), especially as funding providers use the capital ratio as a summary measure of bank health. Thus,

changes in tax rates are likely to have a differential effect on bank lending, leverage, liquidity, and risk-

taking activities depending on bank size, profitability, and capitalization.

3. Sample and Empirical Methodology

3.1. Setting

To test our hypotheses, we exploit variation in U.S. state bank income tax rates. We believe this

setting provides several advantages over alternative approaches (e.g., cross-country tax rate variation or

bank-level effective tax rates). First, the within-country design allows us to examine the effect of taxation

on bank outcomes in a setting where the regulatory and economic environment does not differ substantially

across banks. While the variation in corporate tax rates is likely larger in a cross-country setting than in the

U.S. state setting, a cross-country design will suffer to a greater extent from correlated omitted variable

concerns related to bank regulation or macroeconomic trends. Second, the U.S. setting allows us to better

identify the statutory tax rates for banks. As we explain in section 3.4, we conduct an extensive data

collection to gather the applicable state-level tax rates for banks. In a cross-country setting, it is more

difficult to accurately identify the specific tax rate applicable to banks, relative to the corporate tax rate

applicable to non-financial institutions. Third, it is easier to identify the applicable income tax rate(s) for

U.S. commercial banks than for banks in a cross-country. As we explain in section 3.3, U.S. bank regulatory

filings allow us to identify the states in which banks have material operations (i.e., deposits), and thus we

14

are able to accurately measure the applicable tax rate. In a cross-country setting, banks are more likely to

have material operations that cross borders, and we are unaware of data sources that allow researchers to

separate out the banks’ operations by country in order to determine the applicable tax rate. Fourth, statutory

tax rates are plausibly exogenous to any individual bank, providing an advantage over the use of firm-level

effective tax rates, which are likely determined in part by the banking outcomes we seek to study.7 Finally,

the U.S. setting offers us access to high quality quarterly regulatory filings, which allow us both to examine

a number of outcomes we would be unable to in an international setting (e.g., different leverage

components) as well as the ability to control for many bank-level factors.

3.2. Overview of identification strategy

Our identification strategy exploits the fact that state-level bank income tax rates change at different

times for different states. Therefore, our setting allows us to compare treated banks (those experiencing a

tax rate change in a particular period) to untreated banks (those that do not experience a tax rate change in

that period). The staggered nature of these tax rate changes provides an extensive set of counterfactuals,

which allow us to examine how bank outcomes would have evolved absent a tax rate change (Heider and

Ljungqvist 2015). These tax rate changes are plausibly exogenous to the individual banks in our sample.

Our sample is primarily composed of smaller single-state U.S. commercial banks, which we believe are

unlikely to exert strong influence over the design of the state corporate income tax system.

Of course, state tax rates are not randomly assigned to states, and are likely correlated with

concurrent macroeconomic and regulatory factors. Furthermore, our identification strategy relies on the

control (i.e., non-treated) banks being similar to the treated banks except for the tax rate change. We take

several steps to mitigate these correlated omitted variable concerns.

To address the former concern, we include state-quarter-level variables that likely capture

macroeconomic factors at the state level, such as state GDP growth, unemployment rates, and housing price

indices. We also include year-quarter fixed effects, which account for country-level macroeconomic and

7 Of course, one downside in using statutory tax rates is that they may be a noisy proxy for the marginal tax rate to

the extent that the bank engages in material tax planning efforts.

15

regulatory factors. In bank-level heterogeneity tests, we replace these fixed effects with state-year fixed

effects, which control for any state-level factor that affects all banks equally in a given year. In addition,

we compare the effects of tax rate changes on state-chartered banks versus those on federally-chartered

banks. As federally-chartered banks are subject to state tax rates, but not any state-level banking regulation,

concurrent changes in state banking regulations cannot explain any effects of tax rate changes on outcomes

in these banks.

To address the latter concern, we control for time-varying bank-level determinants of our outcome

variables, such as regulatory capital, size (e.g., assets), and pre-tax profitability. We also include bank fixed

effects to account for any time-invariant observable and unobservable factors at the bank level that may

affect changes in the outcomes we study. Our identification strategy therefore allows us to rule out a number

of alternative explanations that could explain our observed results. For a state-level omitted factor to be a

concern, it would need to be correlated with both changes in state tax rates and changes in bank outcomes,

but not correlated with the state-level macroeconomic variables or charter-type.8 That said, there are still

potentially correlated omitted variables that our design would not be able to account for, such as regulatory

actions that federal banking supervisors took that affected bank outcomes at federally-chartered banks

subject to an income tax rate change but that did not affect non-treated federally chartered banks.

3.3. Sample

State corporate income tax provisions generally require firms with economic activity (nexus) in the

given state to pay income taxes based on such factors as property, sales, and employee payroll. This feature

of state income taxation results in the difficulty of examining corporate state income tax activities and

effects for firms operating in multiple states, especially because researchers are generally unable to observe

the amount of income earned in different states. A key advantage of our setting is the ability to use bank

regulatory filings to identify the states in which each commercial bank operates (i.e., has deposits), thereby

8 That said, there are still potentially correlated omitted variables that our design would not be able to account for,

such as regulatory actions that federal banking supervisors took that affected bank outcomes at federally-chartered

banks subject to an income tax rate change but that did not affect non-treated federally chartered banks. However,

we believe that this is unlikely.

16

allowing us to assemble a sample of single-state commercial banks. Limiting our sample to single-state

banks strengthens our ability to draw inferences regarding the effect of state-level income tax changes on

bank outcomes. We find that 95.88 percent of commercial bank-years operate in only one state during our

sample period, suggesting that limiting our sample to single-state banks does not prevent us from retaining

the vast majority of commercial banks in our sample. Still, this restriction does come with a trade-off, in

that our sample will omit the largest commercial banks that operate in multiple U.S. states. In untabulated

analyses, we find that our sample covers 31 (29) percent of total U.S. commercial bank loans (deposits).

Our sample thus still represents an economically important percentage of U.S. banking activity.

We collect quarterly commercial bank data from Call Reports filed with bank regulators using the

Compustat Bank files. Table 1 presents our sample selection process. We begin with 591,392 single-state

bank-quarters from 1996-2013. Our sample period starts in 1996 due to the availability of data for several

control variables. Furthermore, the availability of the FDIC Summary of Deposits data begins around this

time period (1994), enabling us to identify single-state banks. Consistent with our focus on examining the

effect of changes in state income tax rates on bank outcomes, we remove 125,804 bank-quarters electing to

be treated as S-corporations for tax purposes.9 We remove bank-quarters with less than $25 million in assets

(52,150 observations), as very small banks are likely to face distinct operating decisions. Furthermore, we

remove 34,903 bank-quarters with asset growth exceeding 10 percent, as these banks have likely acquired

other institutions during the quarter. Our final sample consists of 351,767 bank-quarter observations from

11,860 unique banks.10

3.4. Statutory income tax rate data

We conduct an in-depth hand collection effort to assemble the historical state corporate tax rates

for financial institutions. We collect historical data from the following sources: the U.S. Multistate Tax

9 We remove bank-quarters of banks signifying S-corporation election on their quarterly call reports. Although

firms make the S-corporation election for Federal tax purposes, a majority of states recognize the Federal election

for state corporate income tax purposes under current corporate income tax law. 10 The sample for our measure of asset risk (ARISK), discussed in Section 3.5, is smaller than the other samples

because data on the risk weighting of assets does not become available until the first quarter of 2001. Thus, in

specifications involving ARISK, our sample is 220,733 bank-quarter observations.

17

Guide (CCH), individual state income tax codes, state revenue department websites, the Book of the States,

and the Significant Features of Fiscal Federalism. We cross-check all tax rates and settle disputes across

sources using original source code documents. Consistent with our focus on commercial banks, we collect

the historical income tax rates for financial institutions in each state. This focus on financial institutions

specifically is important given differences in how (whether) some states levy corporate income taxes on

financial institutions. Our final dataset consists of the historical state corporate income tax rates for financial

institutions for each of the 50 states and the District of Columbia.

Table 2 presents summary statistics for the statutory income tax rates for financial institutions

during our sample period (1996 – 2013). For each state, we report the average tax rate across the sample

period, as well as the number of unique banks and bank-quarters in each state. We find that Illinois, Texas,

and California (Hawaii, Alaska, and Rhode Island) have the most (fewest) commercial banks in our sample.

We also find considerable variation in state corporate income tax rates for financial institutions. The average

statutory income tax rates of bank-quarters operating in states levying corporate income taxes range from

1 percent (Maine) to 10.681 percent (California). Four states (Michigan, Washington, Nevada, and

Wyoming) levy no statutory corporate income tax rate on financial institutions during our sample period.

3.5. Regression design and variable definitions

We use variations of the following OLS regression model to test our hypotheses, where i indexes

banks, j indexes states, and t indexes time (year-quarters) (all variables are defined in the appendix):

𝛥𝑂𝑈𝑇𝐶𝑂𝑀𝐸𝑖,𝑡 = 𝛽1𝛥𝑅𝐴𝑇𝐸𝑗,𝑡 + ∑ 𝛽𝑚𝐵𝐴𝑁𝐾 𝐶𝑂𝑁𝑇𝑅𝑂𝐿𝑆𝑖,𝑡−1𝑚 + ∑ 𝛽𝑛𝑀𝐴𝐶𝑅𝑂 𝐶𝑂𝑁𝑇𝑅𝑂𝐿𝑆𝑗,𝑡𝑛 +

∑ 𝛼𝑖𝑖 + ∑ 𝛼𝑡𝑡 + 𝜀𝑖,𝑡 (1)

Our primary independent variable is ΔRATE, which equals the change in the income tax rate

applicable to banks in state j at time t. Because tax rates change annually, but our data is quarterly, we apply

the tax rate change to all quarters in the calendar year in which it is changed. Thus, we are agnostic as to

when the bank outcomes change in response to the tax rate change, as long as the change occurs at some

point during that year. The trade-off with this design is that if the bank’s response to the tax rate change

18

occurs all in one quarter, our design potentially inhibits our ability to pick up an effect. Because our

treatment affects all banks in a state similarly, we cluster standard errors at the state level (Petersen 2009).

Our regression design is akin to a difference-in-differences design (Roberts and Whited 2013).

Given that tax rate changes are staggered across states and time, our setting provides us with an extensive

set of counterfactuals. A key assumption to interpreting the results of our estimation as a causal effect is

the assumption of parallel trends; that is, that the treated and untreated banks would have experienced

similar outcomes absent the tax rate change. We examine the plausibility of this assumption in section 5.1.

We examine four groups of dependent variables (ΔOUTCOME): lending, leverage, liquidity, and

risk-taking. When exploring loan growth, we employ two proxies. The first, ΔLOANS, is defined as the

change in gross loans over the quarter, scaled by total assets at the beginning of the quarter. This variable

captures changes in term loans as well as changes in the drawn down portion of outstanding credit lines.

Alternatively, we use ΔCOMMIT, which captures changes in gross loans and outstanding loan commitments

over the quarter, scaled by total assets and outstanding loan commitments at the beginning of the quarter.

The difference between ΔLOANS and ΔCOMMIT is that the latter also captures changes in undrawn portion

of credit lines (Cornett et al. 2011).

When studying leverage, we employ several variables that capture both the overall leverage and

the components of leverage. We first use an overall measure of leverage, ΔLEV, which is defined as the

change in total liabilities over the quarter, scaled by total assets at the beginning of the quarter. We further

split this out into the change in deposit funding, ΔDEP, and the change in non-deposit funding, ΔNONDEP

(both scaled by total assets at the beginning of the quarter). Finally, we split the change in deposit funding

into both an insured (ΔINSDEP) and uninsured components (ΔUNDEP). To separate out the change in

deposits into the insured and uninsured components, we use call report data to estimate the uninsured

portion, and then measure the insured portion as total deposits minus our estimate of uninsured deposits.11

11 Only large banking institutions are required to provide estimates of uninsured deposits in their call report filings.

We estimate uninsured deposits by subtracting the number of uninsured accounts multiplied by the insurance limit

from the aggregate balance of deposits over the insurance limit. Following June 30, 2006, uninsured accounts are

disaggregated into retirement and nonretirement values. Before the second quarter of 2006, the deposit insurance

19

When studying liquidity, we examine changes in liquid asset holdings. Specifically, we follow

Cornett et al. (2011) and use the variable ΔLIQUID, which is the quarter-over-quarter change in the sum of

cash and cash equivalents, federal funds sold and securities purchased under agreements to resell, and U.S.

treasury securities. We scale this amount by total assets at the beginning of the quarter. When examining

risk-taking, we employ two variables. First, we exploit the fact that regulatory filings require banks to

classify assets based on risk. Specifically, banks must assign a zero, 20, 50, or 100 percent risk-weight to

each asset. Our risk-taking proxy is the change in 100-percent risk-weighted assets, scaled by total risk-

weighted assets at the beginning of the quarter (ΔARISK). Higher values of this measure are associated with

increased asset risk.

As discussed in section 3.2, we include two sets of control variables. First, we include time-varying,

bank-level variables that control for potential differences between treated and non-treated banks that may

be associated with changes in bank outcomes. Specifically, we control for bank regulatory capital using the

banks’ tier 1 capital scaled by risk-weighted assets, both measured as of the beginning of the quarter

(CAPITAL). Bank capital is a critical component of a bank’s balance sheet and a focal point for bank

regulators, and it likely determines the bank’s ability to issue new loans and increase leverage. In addition,

we control for bank size using the natural log of total assets at the beginning of quarter (SIZE), given

evidence that bank size is associated with lending and liquidity (Cornett et al. 2011). Finally, we include a

measure of pre-tax profitability, which we define as net income before income taxes but after adding back

the loan loss provision, scaled by assets at the beginning of the period (ROA). By adding back the loan loss

provision, our measure of profitability closely approximates a measure of operating cash flows (Altamuro

and Beatty 2010), which likely influences a bank’s ability to lend, satisfy funding providers’ claims, and

its need for liquid assets.12 Second, we include bank and year-quarter fixed effects. The former accounts for

limit on all accounts was $100,000. Following the fourth quarter of 2008, the deposit insurance limit on all

accounts was $250,000. Between the second quarter of 2006 and the fourth quarter of 2008, retirement accounts

were insured up to $250,000 while nonretirement accounts were insured up to $100,000. 12 We add back the loan loss provision because it is the primary non-cash expense included in pre-tax net income.

This approach allows us to compute a measure of operating cash flows, which are not disclosed in reports filed

with bank regulators (i.e., call reports do not include a statement of cash flows). There are other items disclosed

20

the fact that banks that experience tax rate changes may differ fundamentally (e.g., state-level political

environment or bank-level business strategy) from those that do not experience tax rate changes. The latter

accounts for nation-wide economic shocks that may be correlated with state-level tax rate changes and

changes in bank-level outcomes.

In several tests, we examine how the effect of tax rate changes on bank outcomes varies with

macroeconomic conditions. Our first macroeconomic proxy is RECESSION, an indicator variable equal to

one if the year-quarter is part of a recession as defined by the National Bureau of Economic Research as

the year 2001 as well as the fourth quarter of 2007 through the second quarter of 2009, and zero otherwise.

We also employ two measures that capture the state of the interbank lending market, which is commonly

used for short-term liquidity. Our second proxy is TED, the average daily spread between the difference

between the three-month London Inter-bank Offered Rate (LIBOR) and the three-month Treasury rate

during the quarter (otherwise known as the TED spread). This measure captures the amount of credit risk

in the interbank market; higher values indicate that short-term liquidity is more costly and difficult to come

by. Prior research shows that the TED spread increased dramatically during the 2007-09 financial crisis

(Cornett et al. 2011). Our third proxy is STED, the standard deviation of the daily TED spread during the

quarter. While TED captures the level of liquidity costs, STED captures the uncertainty regarding the cost

of short-term liquidity.

3.6. Descriptive Statistics

Table 3 reports summary statistics for our variables. The average state income tax rate in our sample

is 6.22 percent. Regarding tax rate changes, we have 68 rate changes in our sample, 75 percent of which

are decreases. The average rate increase is approximately 1 percent, and ranges from 0.2 percent to 2.2

percent. The average rate decrease is approximately 1 percent, and ranges from 0.03 percent to 8.5 percent.

In an untabulated analysis, we find that the majority (approximately 90 percent) of our observations do not

experience a tax rate change; we examine the robustness of our results to excluding states that never

in call reports that may contain non-cash expenses (e.g., facilities expense), but these items are often of immaterial

amounts and are grouped together with cash expenses (e.g., rent).

21

experience a tax rate change during our sample period (see section 5). The average change in loans and in

total credit commitments is approximately 1.2 percent of assets. Similarly, the average change in leverage

is approximately 1.1 percent of assets, with approximately 72 percent coming from insured deposits. On

average, banks are increasing risk-taking and decreasing asset liquidity. The average bank in our sample is

well capitalized, with a tier 1 ratio of 16 percent. As mentioned above, the average bank is relatively small,

with average (median) total assets of approximately $460 million ($124 million) (untabulated).

4. Results

4.1. Loan growth

Our first set of analyses examines the association between corporate income taxes and lending

activity. We report the results of these tests in Table 4. We first report the results of the parsimonious

specifications (i.e., without interaction terms used to assess cross-sectional variation) in columns 1 and 2.

Here, we do not find evidence consistent with bank income tax rates being associated with loan growth on

average, using a both measure of traditional lending (ΔLOANS) and a measure that also reflects unexercised

loan commitments (ΔCREDIT). The standard error on the ΔRATE coefficient in column 1 indicates that our

design had sufficient power to pick up a 0.057 (1.96 t-statistic/0.0291 standard error) percentage point

increase in loans in response to a 1 percentage point tax rate increase. Given that the median change in

loans is 0.919 percentage points, this suggests that our inability to find an effect on average is unlikely to

be due to low statistical power.

Turning to bank-level cross-sectional tests, we find little evidence that the tax rate-loan growth

association varies with the bank’s capitalization or size. However, in columns 7 and 8, we find the

coefficient on ΔRATE*ROA is positive and significant for the ΔLOANS (p < 0.05) and ΔCREDIT (p < 0.10)

specifications. Untabulated F-tests indicate that the relation between taxation and lending is insignificant at

low levels (25th percentile, median) of ROA but becomes positive and marginally statistically significant

at higher levels (75th percentile) (p-values of 0.06 and 0.11 for ΔLOANS and ΔCREDIT, respectively),

suggesting that changes in corporate income tax rates are not associated (positively) associated with lending

activities for less (more) profitable banks. Higher tax rates may encourage banks to increase lending in

22

order to maintain desired after-tax levels of profitability. More profitable banks can use internally generated

cash flows to fund this lending growth, whereas less profitable banks may be constrained in their ability to

respond to higher tax rates by increasing lending. Therefore, profitability appears a relevant dimension in

predicting whether banks respond to tax changes by changing lending growth.

In columns 9 and 10, we find that tax rate changes are negatively (positively) associated with loan

growth during recessions (during expansionary periods), suggesting that taxes discourage (encourage)

lending in these periods (p < 0.05). In terms of economic significance, a 1 percentage point increase in

ΔRATE is associated with a 0.04 percent increase (0.09 percent decrease) in leverage during non-recession

(recession) periods or periods characterized by low (high) credit risk costs and uncertainty. This suggests

economic conditions influence the association between tax rates and bank lending, consistent with evidence

that tax rate cuts can have unique effects on employment during recessions (Ljungqvist and Smolyansky

(2016). Specifically, the ability to access outside funds is more difficult during recessions (e.g., Beatty and

Liao (2011)) or during periods with high credit risk costs and uncertainty, and therefore banks may be more

reliant on after-tax cash flows to fund lending during these periods. As a result, banks may reduce lending

in response to tax rate increases during recessions. We also find evidence that uncertainty in economy-wide

credit risk moderates the relation between income tax rate changes and lending activity. Specifically, the

coefficients on ΔRATE*TED and on ΔRATE*STED (are negative and significant (p < 0.01) in columns 11

through 14, implying that the tax rate-lending association is decreasing in the level and volatility of

interbank funding costs. Collectively, these results suggest that during periods in which alternative methods

of finding liquid assets are more costly or less accessible, banks may rely more heavily on after-tax

operating cash flows to fund lending and are thus more negatively affected by income taxes.

4.2. Leverage

Our second set of analyses examines the association between income tax rates and bank leverage.

In Panel A of Table 5, we do not find evidence of an average effect of taxation on leverage, as the coefficient

on ΔRATE is positive but statistically insignificant. The standard error on the ΔRATE coefficient in column

1 indicates that our design had sufficient power to detect a 0.033 (1.96 t-statistic/0.0168 standard error)

23

percentage point increase in leverage in response to a 1 percentage point tax rate increase, indicating that

our inability to find an effect on average is not likely due to low statistical power.

However, we do find that the effect varies in the cross-section. Specifically, we find that the relation

between tax rate changes and leverage changes is increasing in the bank’s capital ratio (i.e., the coefficient

on ΔRATE*CAPITAL is positive and significant, p < 0.05). Untabulated F-tests show that the association

between ΔRATE and CAPITAL is insignificant at low capital levels (25th percentile and median of

CAPITAL), but is stronger at higher levels of CAPITAL (75th percentile, p-value = 0.11). This result suggests

that better capitalized banks have a greater ability to increase leverage in response to changes in taxation,

most likely due to some combination of better capitalized banking identifying more debtholders and facing

lower regulatory costs for increasing leverage. On the other hand, we find little evidence that the association

between tax rate changes and overall leverage varies with the bank’s profitability or size.

Consistent with the relation between tax rate changes and bank leverage varying with economic

conditions, we find that tax rate changes are negatively associated with leverage changes during recessions

(i.e., the coefficient on ΔRATE*RECESSION is negative and significant, p < 0.01), although an untabulated

F-test shows that the effect of ΔRATE on leverage during recessions is statistically insignificant (p-value of

0.14). Similarly, we find that the tax rate-leverage association is decreasing in the level of interbank funding

costs (i.e., the coefficient on ΔRATE*TED is negative and significant, p < 0.05) and volatility in interbank

funding costs (i.e., the coefficient on ΔRATE*STED is negative and significant, p < 0.05). This result may

reflect that the cost and/or scarcity of capital during these periods outweigh the benefits of an increased tax

deduction. In each of these specifications, the main effect on ΔRATE is positive and statistically significant,

suggesting that in normal (non-recessionary or low credit risk uncertainty) periods, bank leverage is

positively associated with tax rate changes, similar to the findings from prior research on non-financial

institutions. In terms of economic significance, a 1 percentage point increase in ΔRATE is associated with

a 0.037 percent increase in leverage during non-recession periods, which is very small. The fact that this

relation is economically smaller than is typically observed in non-financial institutions is perhaps not

24

surprising, given that leverage is a core element of a bank’s business model, and therefore bank leverage

choices may be less sensitive to tax incentives.

Unlike industrial firms, banking regulation requires banks to file call reports that provide more

information on funding sources, allowing us to examine specific leverage components that banks may

change in response to tax rate changes. We begin by examining changes in deposit (ΔDEP) versus non-

deposit (ΔNONDEP) leverage. As reported in Panel B of Table 5, we find that the relation between tax rate

changes and deposit (ΔDEP) funding changes is increasing in the bank’s capital ratio (i.e., the coefficient

on ΔRATE*CAPITAL is positive and significant, p < 0.05). However, untabulated F-tests suggest that the

overall tax rate-leverage association is not statistically significant at the median capital ratio (due to the

main effect on ΔRATE being negative). We find no evidence that the relation between tax rate changes and

non-deposit funding changes (ΔNONDEP) is associated with capitalization. Furthermore, we do not find

evidence that bank size significant moderates the association between tax rate changes and either deposit

or non-deposit funding in columns 5 and 6. However, we find the coefficient on ΔRATE*ROA is positive,

p < 0.10, (negative, p < 0.01) in the ΔDEP (ΔNONDEP) analysis in column 7 (8), suggesting that more

profitable banks are able to turn to depository funding sources in response to changes in taxation, but are

associated with less non-depository funding. Untabulated F-tests indicate that at high (low) levels of ROA,

the relation between taxation and deposit (non-deposit) funding is statistically positive (negative), p-values

= 0.085 and 0.020, respectively. These findings suggest that the bank’s overall leverage may not change in

response to tax rates, but the composition can depending on the bank’s capitalization or profitability.

Next, we investigate how the association between taxation and deposit or non-deposit funding

varies by macroeconomic conditions. First, we document that tax rate changes are negatively associated

with non-deposit funding changes during recessions, with an untabulated F-test indicating that the sum of

the coefficients on ΔRATE and ΔRATE*RECESSION is statistically significant with a p-value < 0.05. On

the other hand, tax rate changes are positively associated with non-deposit funding changes during non-

recessionary periods, with a 1 percentage point increase in the tax rate associated with a 0.02 percentage

point increase in non-deposit funding. Similarly, we find that the association between tax rates and non-

25

deposit funding is decreasing in the level of interbank funding costs (i.e., the coefficient on ΔRATE*TED

is negative and significant, p < 0.01) and volatility in interbank funding costs (i.e., the coefficient on

ΔRATE*STED is negative and significant, p < 0.01). We find no evidence that the association between

taxation and deposit funding is affected by macroeconomic conditions (i.e., the coefficients on

ΔRATE*RECESSION, ΔRATE*TED, and ΔRATE*STED are statistically insignificant). Since non-deposit

funding tends to be more expensive than deposits, these findings are consistent with banks being more

sensitive to the potential for higher income tax rate savings of non-depository funding.

Investigating these results further, we decompose the changes in deposit funding into changes in

insured (ΔINDEP) and uninsured deposits (ΔUNDEP). As reported in Panel C of Table 5, we generally fail

to evidence that income taxation is associated with insured deposit funding, either on average or in the

bank-level cross-sectional tests. While the coefficient on the interaction between ΔRATE and CAPITAL is

marginally statistically, an untabulated F-test shows that the overall association between ΔRATE and

insured deposit funding is statistically insignificant at various levels (25th, 50th, or 75th percentile) of

CAPITAL. On the other hand, we find that income tax rate changes are negatively associated with uninsured

(ΔUNDEP) deposit funding changes at low capital and size levels (untabulated F-test p-values < 0.05 and

< 0.10, respectively), while the negative association between taxation and ΔUNDEP diminishes as capital

and size increase (coefficients on ΔRATE*CAPITAL and ΔRATE*SIZE are positive and significant with p-

values < 0.01 and < 0.05, respectively). However, we find little evidence of changes in tax rates being

associated with changes in either type of deposit funding based on economic conditions (i.e., the

coefficients on ΔRATE*RECESSION, ΔRATE*TED, and ΔRATE*STED are insignificant in columns 7

through 14). Overall, the results of the leverage tests suggest that income taxation has nuanced effects on

bank funding, with significant variation based on bank characteristics and economic conditions.

4.3. Liquidity

Table 6 reports the results of the third set of analyses, which examine how income taxation impacts

banks’ liquidity choices. Consistent with the lending and leverage tests, we do not find evidence consistent

with bank income taxes having, on average, a significant effect on liquid asset holdings. The standard error

26

on the ΔRATE coefficient in column 1 indicates that our design had sufficient power to pick up a 0.064

(1.96 t-statistic/0.0326 standard error) percentage point increase in liquid asset holdings in response to a 1

percentage point tax rate increase, suggesting that our inability to find an effect on average is not likely due

to low statistical power. Furthermore, we find little evidence that the tax rate-liquidity association is

moderated by bank capitalization, size, or profitability.

However, similar to the lending and leverage tests, the results in Table 6 are consistent with the

associations between tax rate changes and liquidity management varying significantly with economic

conditions and uncertainty. In particular, we find that income tax rate changes are negatively associated

with liquid asset holdings during non-recessionary periods (coefficient on ΔRATE in column 5 is positive

and significant, p < 0.01), but positively associated with liquid asset holdings during recessions (untabulated

F-test of the sum of coefficients on ΔRATE and ΔRATE*RECESSION has a p-value < 0.01). In addition,

we find that the association between income taxation and liquid asset holdings is increasing in the level of

interbank funding costs (i.e., the coefficient on ΔRATE*TED is positive and significant, p < 0.01) and

volatility in interbank funding costs (i.e., the coefficient on ΔRATE*STED is positive and significant, p <

0.01). Overall, the results in Table 6 suggest that banks’ liquid asset holdings are positively associated with

tax rate changes when alternative external sources of liquid funds are more costly or scarce (e.g., during

recessions and periods with high credit risk costs and uncertainty). Since income taxation impacts the

amount of pre-tax operating cash flows that are available to use for liquidity management, during recessions

and periods with high credit-risk costs and uncertainty – when banks need liquid assets most but are least

able to find alternative sources of liquid assets – they respond to the negative impact of taxation on after-

tax cash flows by increasing liquid asset holdings. Relatedly, we find that banks decrease liquid asset

holdings in response to tax rates during non-recession periods and periods with relatively lower credit risk

costs and uncertainty.

4.4. Risk-taking

Our final set of analyses examine whether income taxation is associated with bank risk-taking. The

results of the risk-taking tests, reported in Table 7, are generally consistent with the lending, leverage, and

27

liquidity tests, and are consistent with the effect of tax changes varying in the cross-section with economic

conditions and uncertainty. We do not find evidence that bank income taxes significantly impact bank risk-

taking, on average. The standard errors on these coefficients indicates that our tests have the statistical

power to find or a 0.065 percentage point increase in ARISK in response to a 1 percentage point tax rate

increase, which is quite small compared to the sample median value of 0.583 percentage points.

Turning to the bank-level cross-sectional results, we find that the association between income

taxation and bank risk-taking is decreasing in bank size (the coefficient on ΔRATE*SIZE is negative and

significant, p < 0.05, for ΔARISK), but untabulated F-tests indicate that the overall effect of ΔRATE on

ΔARISK is not statistically significant at various levels (25th, 50th, and 75th percentiles) of SIZE. Similarly,

we find that the relation between taxation and risk-taking is increasing in bank profitability (the coefficient

on ΔRATE*ROA is positive and significant, p < 0.01); however, an untabulated F-test shows that the relation

between ΔRATE and risk-taking is only statistically significant overall at the 75th percentile of ΔARISK. We

find little evidence that bank capital moderates the association between changes in corporate income

taxation and bank risk.

Turning to the macroeconomic tests (columns 5 through 7), we find that corporate income tax rate

changes are negatively associated with ΔARISK during recessions, and the association between income

taxation and ΔARISK is decreasing in the level and volatility in interbank funding. Overall, the findings

suggest that income taxation is associated with lower (greater) risk-taking during recessionary (non-

recessionary) and high (non-high) credit risk uncertainty periods.

5. Robustness

5.1. Alternative timing

We next examine how our findings change when we alter the timing of the tax rate change variable.

In our primary tests, we measure the tax rate change contemporaneously with the outcome variables. This

assumes that banks do not respond early to tax rate changes. However, tax rate changes may be anticipated,

because they are often the result of political negotiations and lobbying efforts and often take effect

subsequent to enactment. Therefore, banks may begin to respond to tax rate changes before they actually

28

take effect. To address this concern, we re-estimate our regressions using the future tax rate change (i.e.,

the tax rate change for the next calendar year).

Examining future tax rate change specifications also allows us to investigate differences in bank

outcomes for treated (untreated) observations in the pre-tax-change period. Indeed, to interpret our

regression results as causal effects, one must assume that the treated and untreated banks would have

experienced similar outcomes but for the tax rate change (Roberts and Whited 2013). While this assumption

is not formally testable, failing to find evidence that the outcome variables differ between the treated and

control banks before the treatment would be consistent with this assumption.

The results of our forward-looking tax rate change specifications (untabulated) show that next

year’s tax rate change is insignificantly associated with lending and liquid asset holdings, consistent with

parallel trends between our treatment and control banks in the year before the tax rate change, and

inconsistent with banks anticipating the tax rate changes by changing lending and liquid asset holdings.

However, we find that there is a positive association between future tax rate changes and both leverage

(coefficient = 0.065, p-value < 0.01) and asset risk (coefficient = 0.047, p-value < 0.05). It is perhaps not

surprising that banks change leverage in anticipation of tax rate changes, as this allows banks to take full

advantage of the additional interest deduction. Furthermore, unlike non-financial institutions, banks are

able to adjust their leverage more quickly in response to tax rate changes.

5.2. Federal vs. state-chartered banks

One plausible omitted variable that could affect our inferences is bank regulatory or supervisory

changes that occur concurrently with tax rate changes. For example, concurrent with a state cutting its bank

income tax rate, the state banking regulator may decide to decrease its oversight of bank lending practices

in a way that spurs increased lending. To mitigate concerns about concurrent regulatory or supervisory

changes, we compare the effects of tax rate changes on state-chartered banks versus those on federally-

chartered banks. Because federally-chartered banks are subject to state tax rates, but are not subject to any

other state-level banking regulation, concurrent changes in state banking regulations cannot explain any

effects of tax rate changes on outcomes in federally chartered banks. In untabulated analyses, we verify that

29

our findings are generally consistent across both federally and state chartered banks (and generally

statistically weaker for state chartered banks), mitigating concerns about concurrent changes in state

banking regulation or supervision.

5.3. Other robustness tests

Given that we include bank fixed effects, which absorb any observable and unobservable state-

specific factors that are constant over our sample period, in all of our analyses, it is unlikely that our primary

variable ΔRATE is capturing fundamental differences between states that change bank income taxation and

those that do not. To confirm this assumption, we examine the robustness of our findings to removing states

that never experience a change during our sample period. Those states include those that never tax bank

income, such as Michigan, Nevada, Washington, and Wyoming, as well as other states that simply have a

consistent tax rate over the sample period.13 This screen results in substantial sample attrition, causing us

to lose approximately 47 percent of our sample. Nevertheless, in untabulated analyses, we confirm that our

inferences remain generally unchanged when removing these states from our sample.

Additionally, our primary analyses use a $25 million asset value threshold for sample inclusion, in

order to balance the desire to maximize sample size versus the desire to exclude very small banks, which

are likely subject to distinct operating and regulatory pressures. However, banks with assets between $25

and $75 million are still relatively small. In untabulated analyses, we verify that our primary inferences are

qualitatively unaffected if we increase the asset cut-off value to $75 million.

6. Conclusion

Understanding whether taxation is associated with bank choices is critical for policymakers because

banks function as vital liquidity providers for creditors and capital providers for businesses and individuals.

To explore this topic, we examine the role of state income taxation on single-state U.S. commercial banks.

While our identification strategy cannot conclusively rule out all correlated omitted variable concerns, we

13 The following states tax bank income but have no rate changes during our sample period: Alaska, Washington

DC, Florida, Georgia, Hawaii, Iowa, Indiana, Louisiana, Minnesota, Montana, Mississippi, Montana, Nebraska,

New Mexico, Oklahoma, Pennsylvania, South Carolina, South Dakota, Utah, Virginia, and Wisconsin.

30

believe our setting allows for a better identification of an effect of corporate taxation on banks relative to

alternatives methods employed in prior research (e.g., cross-country analyses). Our study provides some of

the first empirical evidence on the association between bank income taxation and various bank outcomes.

In doing so, we take an initial step in answering the call in Hanlon and Heitzman (2010) for more evidence

on the interaction between taxation and financial institutions.

Overall, we find little evidence of an average effect of state income taxation on banks across our

four outcome categories: lending, leverage, liquidity, and risk-taking. However, we do find that state

income taxation affects bank outcomes in specific situations – by certain bank characteristics (e.g.,

capitalization, size, and profitability) and in certain periods (e.g., during recessions and periods of high

credit risk costs and uncertainty). The overall set of findings suggests that the relation between corporate

taxation and bank outcomes is nuanced. We view our study as a first step towards a more complete

understanding of how corporate taxation interacts with bank decisions. We look forward to additional

research that examines the important topic of the interaction between taxation and financial institutions.

31

References

Acharya, V. V., and N. Mora. 2015. A Crisis of Banks as Liquidity Providers. Journal of Finance 70 (1):

1-43.

Admati, A., P. DeMarzo, M. Hellwig, and P. Pfleiderer. 2013. Fallacies, Irrelevant Facts, and Myths in

the Discussion of Capital Regulation: Why Bank Equity Is Not Socially Expensive. Working

paper. Stanford University.

Allen, F. 2014. How Should Bank Liquidity Be Regulated? Working paper. University of Pennsylvania.

Altamuro, J., and A. Beatty. 2010. How Does Internal Control Regulation Affect Financial Reporting?

Journal of Accounting & Economics 49 (1/2): 58-74.

Andries, K., J. Gallemore, and M. Jacob. 2017. The Effect of Corporate Taxation on Bank Transparency:

Evidence from Loan Loss Provisions. Journal of Accounting and Economics 63 (2–3): 307-328.

Beatty, A., S. L. Chamberlain, and J. Magliolo. 1995. Managing Financial Reports of Commercial Banks:

The Influence of Taxes, Regulatory Capital, and Earnings. Journal of Accounting Research 33

(2): 231-261.

Beatty, A., and S. Liao. 2011. Do Delays in Expected Loss Recognition Affect Banks' Willingness to

Lend? Journal of Accounting and Economics 52 (1): 1-20.

Berger, A. N., and C. H. S. Bouwman. 2013. How Does Capital Affect Bank Performance During

Financial Crises? Journal of Financial Economics 109 (1): 146-176.

Berger, A. N., N. H. Miller, M. A. Petersen, R. G. Rajan, and J. C. Stein. 2005. Does Function Follow

Organizational Form? Evidence from the Lending Practices of Large and Small Banks. Journal of

Financial Economics 76 (2): 237-269.

Bernanke, B. S. 1983. Nonmonetary Effects of the Financial Crisis in the Propagation of the Great

Depression. American Economic Review 73 (3): 257.

Calomiris, C. W., and J. R. Mason. 1997. Contagion and Bank Failures During the Great Depression: The

June 1932 Chicago Banking Panic. American Economic Review 87 (5): 863-883.

Calomiris, C. W., and J. R. Mason. 2003a. Consequences of Bank Distress During the Great Depression.

American Economic Review 93 (3): 937-947.

Calomiris, C. W., and J. R. Mason. 2003b. Fundamentals, Panics, and Bank Distress During the

Depression. American Economic Review 93 (5): 1615-1647.

Cochrane, J. H. 2009. Lessons from the Financial Crisis. Regulation 32 (4): 34-37.

Collins, J. H., D. A. Shackelford, and J. M. Wahlen. 1995. Bank Differences in the Coordination of

Regulatory Capital, Earnings, and Taxes. Journal of Accounting Research 33 (2): 263-291.

Cornett, M. M., J. J. McNutt, P. E. Strahan, and H. Tehranian. 2011. Liquidity Risk Management and

Credit Supply in the Financial Crisis. Journal of Financial Economics 101 (2): 297-312.

32

de Mooij, R., M. Keen, and M. Orihara. 2013. Taxation, Bank Leverage, and Financial Crises. Working

paper. International Monetary Fund.

DeAngelo, H., and R. M. Stulz. 2015. Liquid-Claim Production, Risk Management, and Bank Capital

Structure: Why High Leverage Is Optimal for Banks. Journal of Financial Economics 116 (2):

219-236.

Dell'Ariccia, G., L. U. C. Laeven, and G. A. Suarez. 2017. Bank Leverage and Monetary Policy's Risk-

Taking Channel: Evidence from the United States. Journal of Finance 72 (2): 613-654.

Diamond, D. W., and A. K. Kashyap. 2016. Liquidity Requirements, Liquidity Choice and Financial

Stability. Working paper. University of Chicago.

Doidge, C., and A. Dyck. 2015. Taxes and Corporate Policies: Evidence from a Quasi Natural

Experiment. Journal of Finance 70 (1): 45-89.

Duchin, R., and D. Sosyura. 2014. Safer Ratios, Riskier Portfolios: Banks׳ Response to Government Aid.

Journal of Financial Economics 113 (1): 1-28.

Dyreng, S. D., B. P. Lindsey, and J. R. Thornock. 2013. Exploring the Role Delaware Plays as a

Domestic Tax Haven. Journal of Financial Economics 108 (3): 751-772.

Ellul, A., and V. Yerramilli. 2013. Stronger Risk Controls, Lower Risk: Evidence from U.S. Bank

Holding Companies. Journal of Finance 68 (5): 1757-1803.

Fama, E. F. 2011. My Life in Finance. Annual Review of Financial Economics 3 (1): 1-15.

Federal Deposit Insurance Corporation. 2017. Federal Deposit Insurance Corporation Risk Management

Manual of Examination Policies. Accessed at

https://www.fdic.gov/regulations/safety/manual/section1-1.pdf.

Federal Reserve Bank of St. Louis. The Financial Crisis Timeline. Available from

https://www.stlouisfed.org/Financial-Crisis [last accessed February 15, 2017].

Gatev, E., T. Schuermann, and P. E. Strahan. 2009. Managing Bank Liquidity Risk: How Deposit-Loan

Synergies Vary with Market Conditions. Review of Financial Studies 22 (3): 995-1020.

Gertler, M., and N. Kiyotaki. 2015. Banking, Liquidity, and Bank Runs in an Infinite Horizon Economy†.

American Economic Review 105 (7): 2011-2043.

Graham, J. R. 2003. Taxes and Corporate Finance: A Review. Review of Financial Studies 16 (4): 1075-

1129.

Graham, J. R. 2005. A Review of Taxes and Corporate Finance. Foundations & Trends in Finance 1 (7):

573-691.

Hanlon, M., and S. Heitzman. 2010. A Review of Tax Research. Journal of Accounting and Economics

50 (2/3): 127-178.

Heider, F., and A. Ljungqvist. 2015. As Certain as Debt and Taxes: Estimating the Tax Sensitivity of

Leverage from State Tax Changes. Journal of Financial Economics 118 (3): 684-712.

33

Ho, P.-H., C.-W. Huang, C.-Y. Lin, and J.-F. Yen. 2016. Ceo Overconfidence and Financial Crisis:

Evidence from Bank Lending and Leverage. Journal of Financial Economics 120 (1): 194-209.

Hodder, L., M. L. McAnally, and C. D. Weaver. 2003. The Influence of Tax and Nontax Factors on

Banks' Choice of Organizational Form. Accounting Review 78 (1): 297.

Ivashina, V., and D. Scharfstein. 2010. Bank Lending During the Financial Crisis of 2008. Journal of

Financial Economics 97 (3): 319-338.

Jin, J. Y., K. Kanagaretnam, and G. J. Lobo. 2013. Unintended Consequences of the Increased Asset

Threshold for Fdicia Internal Controls: Evidence from U.S. Private Banks. Journal of Banking &

Finance 37 (12): 4879-4892.

Kandrac, J. 2014. Bank Failure, Relationship Lending, and Local Economic Performance. Working paper.

Federal Reserve Board.

Kashyap, A. K. 2010. Lessons from the Financial Crisis for Risk Management. Working paper.

University of Chicago.

Kashyap, A. K., R. Rajan, and J. C. Stein. 2002. Banks as Liquidity Providers: An Explanation for the

Coexistence of Lending and Deposit-Taking. Journal of Finance 57 (1): 33-73.

Keeley, M. C. 1990. Deposit Insurance, Risk, and Market Power in Banking. American Economic Review

80 (5): 1183.

Keen, M., and R. de Mooij. 2012. Debt, Taxes, and Banks. Working paper. International Monetary Fund.

Keeton, W., J. Harvey, and P. Willis. 2003. The Role of Community Banks in the U.S. Economy.

Economic Review (01612387) 88 (2): 15.

Langenmayr, D., and R. Lester. 2015. Taxation and Corporate Risk-Taking. Working paper. Catholic

University of Eichstaett-Ingolstadt and Stanford Graduate School of Business.

Lester, R., and D. Langenmayr. 2017. Taxation and Corporate Risk-Taking. The Accounting Review

Forthcoming.

Ljungqvist, A., and M. Smolyansky. 2016. To Cut or Not to Cut? On the Impact of Corporate Taxes on

Employment and Income. Working Papers -- U.S. Federal Reserve Board's Finance & Economic

Discussion Series: 1-60.

Ljungqvist, A., L. Zhang, and L. Zuo. 2016. Sharing Risk with the Government: How Taxes Affect

Corporate Risk Taking. Forthcoming,

Myers, S., J. McConnell, A. Peterson, D. Soter, and J. Stern. 1998. Vanderbilt University Roundtable on

the Capital Structure Puzzle. Journal of Applied Corporate Finance 11: 8-24.

Petersen, M. A. 2009. Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches.

Review of Financial Studies 22 (1): 435-480.

34

Roberts, M. R., and T. M. Whited. 2013. Endogeneity in Empirical Corporate Finance. In Handbook of

the Economics of Finance, Volume 2a: Corporate Finance, 1st Edition. Oxford, U.K.: Elsevier,

493-572.

Schepens, G. 2016. Taxes and Bank Capital Structure. Journal of Financial Economics 120 (3): 585-600.

Sidel, R. 2011. Fdic’s Tab for Failed U.S. Banks Nears $9 Billion. Wall Street Journal, March 17.

Wall, L. D. Federal Reserve Bank of Atlanta. 2015. Liquidity Regulation and Financial Stability.

Accessed at https://www.frbatlanta.org/cenfis/publications/notesfromthevault/1502.aspx.

35

Appendix: Variable definitions

Panel A: Bank Outcomes

Variable Definition

ΔLOANS Change in gross loans over quarter t, scaled by total assets at the end of quarter t-1.

The amount is converted to a percentage by multiplying by 100.

ΔCREDIT Change in gross loans and undrawn credit line commitments over quarter t, scaled by

total assets and undrawn commitments at the end of quarter t-1. The amount is

converted to a percentage by multiplying by 100.

ΔLEV Change in overall leverage, scaled by total assets at the end of quarter t-1. The

amount is converted to a percentage by multiplying by 100.

ΔDEP Change in total deposits, scaled by total assets at the end of quarter t-1. The amount is

converted to a percentage by multiplying by 100.

ΔINSDEP Change in the total deposits minus an estimate of uninsured deposits, scaled by total

assets at the end of quarter t-1. The amount is converted to a percentage by

multiplying by 100.

ΔUNDEP Change in an estimate of uninsured deposits, scaled by total assets at the end of

quarter t-1. The amount is converted to a percentage by multiplying by 100. We

estimate uninsured deposits by subtracting the number of uninsured accounts

multiplied by the insurance limit from the aggregate balance of deposits over the

insurance limit. Following June 30, 2006, uninsured accounts are disaggregated into

retirement and nonretirement values. Before the second quarter of 2006, the deposit

insurance limit on all accounts was $100,000. Following the fourth quarter of 2008,

the deposit insurance limit on all accounts was $250,000. Between the second quarter

of 2006 and the fourth quarter of 2008, retirement accounts were insured up to

$250,000 while nonretirement accounts were insured up to $100,000.

ΔNONDEP Change in non-deposit leverage, scaled by total assets at the end of quarter t-1. The

amount is converted to a percentage by multiplying by 100.

ΔLIQUID Change in liquid asset holdings over quarter t, scaled by total assets at the end of

quarter t-1. The amount is converted to a percentage by multiplying by 100. We

define liquid asset holdings as cash, federal funds sold and securities purchased under

agreements to resell, and U.S. treasury holdings

ΔARISK Change in the amount of assets that receive a 100 percent risk-weight, scaled by total

risk-weighted assets at the end of quarter t-1. The amount is converted to a

percentage by multiplying by 100.

36

Panel B: Independent Variables

Variable Definition

RATE The state's highest income tax rate applicable to banks

ΔRATE The top statutory state income tax rate applicable to banks at the end of year t, minus

the rate at the end of year t-1

CAPITAL Tier 1 regulatory capital ratio at the end of quarter t-1

SIZE The natural logarithm of total assets at the end of quarter t-1

ROA Pre-tax net income plus the loan loss provision for quarter t, scaled by total assets at

the end of quarter t-1

GDPGROWTH Change in state GDP over the year

UNEMP The unemployment rate for the state in year t

HPI The housing price index for the state in year t

Panel C: Macroeconomic Variables

Variable Definition

RECESSION An indicator variable equaling one for recession time periods, as defined by NBER.

Specifically, equaling one for all four quarters of 2001 and the time period from the

fourth quarter of 2007 through the second quarter of 2009

TED The average daily spread between the difference between the three-month London

Inter-bank Offered Rate (LIBOR) and the three-month Treasury rate during the

quarter

STED The standard deviation of the daily TED spread during the quarter

37

Table 1: Sample selection This table describes the sample selection process employed in our study, and presents the number of bank-years and

unique banks in our final sample.

Screen Observations

All bank-quarters with call reports from 1996-2013 591,392

- Less bank-quarters with Subchapter S election (125,804)

- Less bank-quarters with less than $25 million in assets (52,150)

- Less bank-quarters with asset growth exceeding 10% (34,903)

- Less bank-quarters with insufficient data for variables (26,768)

= Less bank-quarters with insufficient data for variables 351,767

Number of unique banks 11,860

38

Table 2: State tax rate information This table presents the average state bank income tax rate, the number of bank-quarters, and the number of unique banks for each state in our sample.

State Avg. Tax

Rate

# Bank-

Quarters

# Unique

Banks State

Avg. Tax

Rate

# Bank-

Quarters

# Unique

Banks

Alaska 9.40 376 11 Montana 6.75 3,179 99

Alabama 6.33 7,504 217 North Carolina 7.03 4,713 172

Arkansas 6.52 8,571 257 North Dakota 7.24 2,560 104

Arizona 7.30 1,790 93 Nebraska 3.81 8,171 272

California 10.87 14,730 549 New Hampshire 8.09 1,459 50

Colorado 4.72 6,808 249 New Jersey 9.07 5,262 189

Connecticut 8.69 3,162 104 New Mexico 7.60 1,769 79

District of Columbia 9.98 197 12 Nevada - 1,232 55

Delaware 1.76 1,324 54 New York 7.80 8,357 258

Florida 5.50 11,644 494 Ohio 5.25 11,538 355

Georgia 6.00 14,817 509 Oklahoma 6.00 7,380 307

Hawaii 7.92 352 13 Oregon 6.89 1,738 70

Iowa 5.00 12,728 474 Pennsylvania 1.25 12,659 344

Idaho 7.68 760 28 Rhode Island 9.00 418 17

Illinois 7.56 28,577 939 South Carolina 4.50 4,040 126

Indiana 8.50 8,144 263 South Dakota 6.00 2,791 100

Kansas 4.72 9,617 353 Tennessee 6.30 9,945 312

Kentucky 7.48 11,081 318 Texas 3.39 26,320 972

Louisiana 8.00 6,576 221 Utah 5.00 2,275 76

Massachusetts 10.45 11,485 247 Virginia 6.00 6,778 225

Maryland 7.34 3,778 132 Vermont 9.28 954 26

Maine 1.00 1,909 45 Washington - 3,874 143

Michigan - 9,073 239 Wisconsin 7.90 14,578 420

Minnesota 9.80 10,820 478 West Virginia 8.78 3,788 126

Missouri 7.00 14,464 480 Wyoming - 1,274 56

Mississippi 5.00 4,428 128

39

Table 3: Descriptive statistics This table presents the descriptive statistics for the primary variables employed in our study. Panel A contains the

bank outcome variables (dependent variables), and panel B contains the independent variables. All variables are

defined in the appendix. All continuous bank-level variables are winsorized at the 1st and 99th percentiles.

Panel A: Bank Outcomes

MEAN STD DEV P25 MED P75

ΔLOANS 1.110 2.883 -0.525 0.919 2.603

ΔCREDIT 1.192 3.175 -0.596 0.940 2.764

ΔLEV 1.081 3.523 -0.976 1.044 3.259

ΔDEP 0.975 3.459 -1.053 0.861 3.028

ΔINSDEP 0.787 2.887 -0.701 0.552 2.043

ΔUNDEP 0.171 3.071 -1.066 0.190 1.600

ΔNONDEP 0.104 1.913 -0.309 0.016 0.433

ΔLIQUID -0.222 3.571 -1.990 -0.113 1.660

ΔARISK 0.766 3.136 -0.798 0.583 2.257

Panel B: Independent Variables

MEAN STD DEV P25 MED P75

ΔRATE -0.047 0.506 0.000 0.000 0.000

CAPITAL 16.056 7.497 11.151 13.815 18.309

SIZE 11.867 1.092 11.051 11.714 12.491

ROA 0.972 0.730 0.449 0.869 1.407

GDPGROWTH 2.664 2.413 1.281 2.614 4.158

UNEMP 5.511 1.939 4.233 5.100 6.433

HPI 2.714 1.007 2.054 2.458 3.044

Panel C: State-level ΔRATE Descriptive Statistics

N MEAN P25 MED P75

Increases only 17 0.906 0.500 0.750 1.300

Decreases only 51 -1.046 -1.000 -0.500 -0.250

40

Table 4: Corporate taxation and bank loan growth This table presents the results of estimating equation 1 via OLS. The dependent variable is either ΔLOANS, the change in gross loans over the quarter scaled by

total assets at the beginning of the quarter, or ΔCREDIT, the change in gross loans and outstanding loan commitments over the quarter, scaled by total assets at the

beginning of the quarter. The primary independent variable is ΔRATE, the change in the state income tax rate on banks from year t-1 to year t in percentage points.

All regressions include the full set of control variables, bank fixed effects, and year-quarter fixed effects. All variables are defined in the appendix. Standard errors

presented below the coefficients are clustered at the state level. ***, **, * denote two-tailed statistical significance at the 1, 5, and 10 percent levels.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

DEP VAR: ΔLOANS ΔCREDIT ΔLOANS ΔCREDIT ΔLOANS ΔCREDIT ΔLOANS ΔCREDIT ΔLOANS ΔCREDIT ΔLOANS ΔCREDIT ΔLOANS ΔCREDIT

ΔRATE 0.0085 0.0010 -0.0504 -0.0442 0.2161 0.2396 -0.0451 -0.0517 0.0403*** 0.0306** 0.0635** 0.0423 0.0392* 0.0231

(0.0291) (0.0285) (0.0803) (0.0687) (0.1511) (0.1573) (0.0463) (0.0494) (0.0131) (0.0139) (0.0279) (0.0320) (0.0222) (0.0260)

ΔRATE*CAPITAL 0.0035 0.0027

(0.0035) (0.0030)

ΔRATE*SIZE -0.0175 -0.0201

(0.0141) (0.0127)

ΔRATE*ROA 0.0643** 0.0632*

(0.0264) (0.0317)

ΔRATE*RECESSION -0.1289*** -0.1198***

(0.0175) (0.0176)

ΔRATE*TED -0.0007*** -0.0006**

(0.0002) (0.0002)

ΔRATE*STED -0.0018*** -0.0013**

(0.0005) (0.0006)

CAPITAL 0.0444*** 0.0523*** 0.0446*** 0.0524*** 0.0444*** 0.0523*** 0.0444*** 0.0522*** 0.0444*** 0.0523*** 0.0444*** 0.0523*** 0.0444*** 0.0523***

(0.0072) (0.0082) (0.0072) (0.0081) (0.0072) (0.0081) (0.0072) (0.0082) (0.0072) (0.0081) (0.0072) (0.0082) (0.0072) (0.0082)

SIZE -1.1774*** -1.3293*** -1.1779*** -1.3296*** -1.1783*** -1.3303*** -1.1775*** -1.3293*** -1.1772*** -1.3291*** -1.1774*** -1.3292*** -1.1773*** -1.3292***

(0.0991) (0.1162) (0.0983) (0.1156) (0.0985) (0.1159) (0.0991) (0.1162) (0.0992) (0.1163) (0.0992) (0.1163) (0.0992) (0.1162)

ROA 0.2681*** 0.2475*** 0.2682*** 0.2477*** 0.2680*** 0.2475*** 0.2709*** 0.2504*** 0.2675*** 0.2470*** 0.2675*** 0.2471*** 0.2675*** 0.2472***

(0.0320) (0.0334) (0.0319) (0.0334) (0.0320) (0.0334) (0.0316) (0.0332) (0.0319) (0.0334) (0.0319) (0.0333) (0.0319) (0.0334)

GDPGROWTH 0.0370*** 0.0498*** 0.0370*** 0.0498*** 0.0369*** 0.0497*** 0.0368*** 0.0497*** 0.0371*** 0.0499*** 0.0372*** 0.0500*** 0.0371*** 0.0499***

(0.0112) (0.0135) (0.0112) (0.0135) (0.0112) (0.0135) (0.0112) (0.0135) (0.0112) (0.0135) (0.0112) (0.0135) (0.0112) (0.0135)

UNEMP -0.2517*** -0.2724*** -0.2518*** -0.2725*** -0.2515*** -0.2723*** -0.2516*** -0.2724*** -0.2490*** -0.2700*** -0.2501*** -0.2713*** -0.2502*** -0.2714***

(0.0432) (0.0462) (0.0433) (0.0462) (0.0432) (0.0462) (0.0432) (0.0461) (0.0427) (0.0456) (0.0429) (0.0459) (0.0430) (0.0460)

HPI 0.2779*** 0.2108*** 0.2775*** 0.2105*** 0.2783*** 0.2113*** 0.2769*** 0.2099*** 0.2801*** 0.2128*** 0.2796*** 0.2120*** 0.2788*** 0.2114***

(0.0571) (0.0691) (0.0570) (0.0689) (0.0570) (0.0690) (0.0568) (0.0688) (0.0562) (0.0680) (0.0572) (0.0691) (0.0570) (0.0689)

Bank fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Year-quarter fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Clustering State State State State State State State State State State State State State State

Observations 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767

Adjusted R-squared 0.201 0.176 0.201 0.176 0.201 0.176 0.201 0.176 0.201 0.176 0.201 0.176 0.201 0.176

41

Table 5: Corporate taxation and bank leverage This table presents the results of estimating equation 1 via OLS. In panel A, the dependent variable is ΔLEV, the

change in leverage over the quarter scaled by total assets at the beginning of the quarter. In panel B, the dependent

variable is ΔDEP, the change in deposits over the quarter scaled by total assets at the beginning of the quarter, or

ΔNONDEP, the change in non-deposit funding over the quarter scaled by total assets at the beginning of the quarter.

In panel C, the dependent variable is ΔINDEP, the change in insured deposits over the quarter scaled by total assets at

the beginning of the quarter, or ΔUNDEP, the change in uninsured deposits over the quarter scaled by total assets at

the beginning of the quarter. The primary independent variable is ΔRATE, the change in the state income tax rate on

banks from year t-1 to year t in percentage points. All regressions include the full set of control variables, bank fixed

effects, and year-quarter fixed effects. All variables are defined in the appendix. Standard errors presented below the

coefficients are clustered at the state level. ***, **, * denote two-tailed statistical significance at the 1, 5, and 10

percent levels.

Panel A: Overall leverage

(1) (2) (3) (4) (5) (6) (7)

DEP VAR: ΔLEV ΔLEV ΔLEV ΔLEV ΔLEV ΔLEV ΔLEV

ΔRATE 0.0221 -0.0874* 0.0588 0.0109 0.0367** 0.0511*** 0.0389**

(0.0168) (0.0450) (0.1051) (0.0267) (0.0153) (0.0183) (0.0156)

ΔRATE*CAPITAL 0.0065**

(0.0027)

ΔRATE*SIZE -0.0031

(0.0090)

ΔRATE*ROA 0.0135

(0.0238)

ΔRATE*RECESSION -0.0591***

(0.0184)

ΔRATE*TED -0.0004**

(0.0002)

ΔRATE*STED -0.0010**

(0.0004)

CAPITAL 0.0067 0.0071 0.0067 0.0067 0.0067 0.0067 0.0067

(0.0053) (0.0053) (0.0053) (0.0053) (0.0053) (0.0053) (0.0053)

SIZE -1.6921*** -1.6930*** -1.6923*** -1.6921*** -1.6920*** -1.6921*** -1.6920***

(0.0912) (0.0903) (0.0911) (0.0912) (0.0913) (0.0913) (0.0913)

ROA 0.8037*** 0.8040*** 0.8036*** 0.8043*** 0.8034*** 0.8034*** 0.8034***

(0.0354) (0.0353) (0.0354) (0.0353) (0.0355) (0.0355) (0.0355)

GDPGROWTH 0.0143* 0.0144* 0.0143* 0.0143* 0.0144* 0.0144* 0.0144*

(0.0083) (0.0083) (0.0083) (0.0083) (0.0083) (0.0083) (0.0083)

UNEMP -0.2505*** -0.2507*** -0.2505*** -0.2505*** -0.2493*** -0.2497*** -0.2497***

(0.0393) (0.0393) (0.0393) (0.0392) (0.0391) (0.0392) (0.0392)

HPI 0.0546 0.0539 0.0546 0.0544 0.0556 0.0555 0.0551

(0.0500) (0.0498) (0.0500) (0.0499) (0.0498) (0.0502) (0.0500)

Bank fixed effects Yes Yes Yes Yes Yes Yes Yes

Year-quarter fixed effects Yes Yes Yes Yes Yes Yes Yes

Clustering State State State State State State State

Observations 351,767 351,767 351,767 351,767 351,767 351,767 351,767

Adjusted R-squared 0.126 0.126 0.126 0.126 0.126 0.126 0.126

42

Table 5: Corporate taxation and bank leverage (continued)

Panel B: Deposit vs. non-deposit funding

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

DEP VAR: ΔDEP ΔNONDEP ΔDEP ΔNONDEP ΔDEP ΔNONDEP ΔDEP ΔNONDEP ΔDEP ΔNONDEP ΔDEP ΔNONDEP ΔDEP ΔNONDEP

ΔRATE 0.0115 0.0108 -0.0864*** 0.0031 -0.0436 0.0884 -0.0292 0.0387*** 0.0142 0.0225*** 0.0065 0.0436*** 0.0113 0.0273***

(0.0156) (0.0106) (0.0231) (0.0348) (0.1257) (0.0987) (0.0227) (0.0108) (0.0179) (0.0068) (0.0205) (0.0093) (0.0185) (0.0074)

ΔRATE*CAPITAL 0.0058*** 0.0005

(0.0018) (0.0015)

ΔRATE*SIZE 0.0046 -0.0065

(0.0108) (0.0091)

ΔRATE*ROA 0.0488* -0.0335***

(0.0258) (0.0081)

ΔRATE*RECESSION -0.0109 -0.0475***

(0.0246) (0.0134)

ΔRATE*TED 0.0001 -0.0004***

(0.0002) (0.0001)

ΔRATE*STED 0.0000 -0.0009***

(0.0005) (0.0002)

CAPITAL -0.0041 0.0104*** -0.0037 0.0105*** -0.0041 0.0104*** -0.0041 0.0104*** -0.0041 0.0104*** -0.0041 0.0104*** -0.0041 0.0104***

(0.0054) (0.0017) (0.0054) (0.0017) (0.0054) (0.0017) (0.0054) (0.0017) (0.0054) (0.0017) (0.0054) (0.0017) (0.0054) (0.0017)

SIZE -1.4862*** -0.1847*** -1.4871*** -0.1847*** -1.4860*** -0.1850*** -1.4863*** -0.1846*** -1.4862*** -0.1846*** -1.4862*** -0.1846*** -1.4862*** -0.1846***

(0.0734) (0.0229) (0.0726) (0.0227) (0.0737) (0.0226) (0.0735) (0.0228) (0.0734) (0.0229) (0.0734) (0.0229) (0.0734) (0.0229)

ROA 0.6743*** 0.1148*** 0.6746*** 0.1149*** 0.6743*** 0.1148*** 0.6765*** 0.1133*** 0.6743*** 0.1146*** 0.6743*** 0.1145*** 0.6743*** 0.1146***

(0.0347) (0.0141) (0.0347) (0.0141) (0.0347) (0.0141) (0.0344) (0.0141) (0.0347) (0.0141) (0.0347) (0.0141) (0.0347) (0.0141)

GDPGROWTH 0.0095 0.0040 0.0095 0.0040 0.0096 0.0040 0.0094 0.0041 0.0095 0.0040 0.0095 0.0041 0.0095 0.0041

(0.0087) (0.0034) (0.0087) (0.0034) (0.0087) (0.0034) (0.0088) (0.0034) (0.0087) (0.0034) (0.0088) (0.0034) (0.0087) (0.0034)

UNEMP -0.2381*** -0.0138 -0.2383*** -0.0139 -0.2381*** -0.0138 -0.2380*** -0.0139 -0.2379*** -0.0129 -0.2382*** -0.0129 -0.2381*** -0.0131

(0.0418) (0.0134) (0.0418) (0.0134) (0.0418) (0.0134) (0.0418) (0.0134) (0.0419) (0.0133) (0.0420) (0.0132) (0.0420) (0.0132)

HPI -0.0280 0.0812*** -0.0286 0.0812*** -0.0281 0.0814*** -0.0287 0.0817*** -0.0278 0.0820*** -0.0282 0.0822*** -0.0280 0.0817***

(0.0612) (0.0250) (0.0610) (0.0251) (0.0613) (0.0250) (0.0609) (0.0250) (0.0613) (0.0251) (0.0613) (0.0249) (0.0613) (0.0251)

Bank fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Year-quarter fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Clustering State State State State State State State State State State State State State State

Observations 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767

Adjusted R-squared 0.105 0.023 0.105 0.023 0.105 0.023 0.105 0.023 0.105 0.023 0.105 0.023 0.105 0.023

43

Table 5: Corporate taxation and bank leverage (continued)

Panel C: Insured vs. uninsured deposits (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)

DEP VAR: ΔINDEP ΔUNDEP ΔINDEP ΔUNDEP ΔINDEP ΔUNDEP ΔINDEP ΔUNDEP ΔINDEP ΔUNDEP ΔINDEP ΔUNDEP ΔINDEP ΔUNDEP

ΔRATE 0.0085 -0.0005 -0.0366* -0.0497*** 0.1552 -0.2243** -0.0230 -0.0117 0.0139 -0.0018 0.0124 -0.0077 0.0090 -0.0000

(0.0131) (0.0065) (0.0209) (0.0106) (0.1134) (0.1064) (0.0219) (0.0099) (0.0142) (0.0061) (0.0167) (0.0092) (0.0142) (0.0083)

ΔRATE*CAPITAL 0.0027* 0.0029***

(0.0015) (0.0006)

ΔRATE*SIZE -0.0124 0.0189**

(0.0098) (0.0091)

ΔRATE*ROA 0.0377 0.0134

(0.0229) (0.0103)

ΔRATE*RECESSION -0.0219 0.0054

(0.0219) (0.0163)

ΔRATE*TED -0.0001 0.0001

(0.0001) (0.0001)

ΔRATE*STED -0.0000 -0.0000

(0.0004) (0.0003)

CAPITAL -0.0026 -0.0018 -0.0024 -0.0016 -0.0025 -0.0018 -0.0026 -0.0018 -0.0026 -0.0018 -0.0026 -0.0018 -0.0026 -0.0018

(0.0052) (0.0018) (0.0052) (0.0018) (0.0052) (0.0018) (0.0052) (0.0018) (0.0052) (0.0018) (0.0052) (0.0018) (0.0052) (0.0018)

SIZE -0.9790*** -0.5137*** -0.9794*** -0.5141*** -0.9796*** -0.5128*** -0.9790*** -0.5138*** -0.9789*** -0.5137*** -0.9790*** -0.5137*** -0.9790*** -0.5137***

(0.0546) (0.0377) (0.0542) (0.0376) (0.0542) (0.0380) (0.0546) (0.0377) (0.0546) (0.0377) (0.0546) (0.0377) (0.0546) (0.0377)

ROA 0.2937*** 0.3889*** 0.2938*** 0.3891*** 0.2936*** 0.3890*** 0.2954*** 0.3895*** 0.2936*** 0.3890*** 0.2936*** 0.3890*** 0.2936*** 0.3889***

(0.0279) (0.0218) (0.0279) (0.0218) (0.0279) (0.0218) (0.0275) (0.0219) (0.0279) (0.0218) (0.0279) (0.0218) (0.0279) (0.0218)

GDPGROWTH 0.0042 0.0052 0.0042 0.0052 0.0041 0.0053 0.0041 0.0052 0.0042 0.0052 0.0042 0.0052 0.0042 0.0052

(0.0075) (0.0048) (0.0075) (0.0048) (0.0075) (0.0048) (0.0075) (0.0048) (0.0075) (0.0048) (0.0075) (0.0048) (0.0075) (0.0048)

UNEMP -0.1530*** -0.0841*** -0.1531*** -0.0842*** -0.1529*** -0.0843*** -0.1530*** -0.0841*** -0.1526*** -0.0842*** -0.1529*** -0.0843*** -0.1530*** -0.0841***

(0.0268) (0.0189) (0.0268) (0.0189) (0.0268) (0.0189) (0.0268) (0.0189) (0.0269) (0.0189) (0.0269) (0.0190) (0.0269) (0.0190)

HPI -0.0136 -0.0134 -0.0139 -0.0137 -0.0133 -0.0139 -0.0142 -0.0136 -0.0132 -0.0135 -0.0135 -0.0137 -0.0136 -0.0134

(0.0441) (0.0317) (0.0440) (0.0316) (0.0440) (0.0318) (0.0438) (0.0316) (0.0441) (0.0317) (0.0442) (0.0317) (0.0441) (0.0317)

Bank fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Year-quarter fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Clustering State State State State State State State State State State State State State State

Observations 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767 351,767

Adjusted R-squared 0.163 0.092 0.163 0.092 0.163 0.092 0.163 0.092 0.163 0.092 0.163 0.092 0.163 0.092

44

Table 6: Corporate taxation and bank liquidity

This table presents the results of estimating equation 1 via OLS. The dependent variable is ΔLIQUID, the change in

in the sum of cash and cash equivalents, federal funds sold and securities purchased under agreements to resell, and

U.S. treasury securities over the quarter, scaled by total assets at the beginning of the quarter. The primary independent

variable is ΔRATE, the change in the state income tax rate on banks from year t-1 to year t in percentage points. All

regressions include the full set of control variables, bank fixed effects, and year-quarter fixed effects. All variables are

defined in the appendix. Standard errors presented below the coefficients are clustered at the state level. ***, **, *

denote two-tailed statistical significance at the 1, 5, and 10 percent levels.

(1) (2) (3) (4) (5) (6) (7)

DEP VAR: ΔLIQ ΔLIQ ΔLIQ ΔLIQ ΔLIQ ΔLIQ ΔLIQ

ΔRATE 0.0143 -0.0055 -0.0950 0.0264 -0.0253** -0.0855*** -0.0367***

(0.0326) (0.0532) (0.0975) (0.0269) (0.0113) (0.0151) (0.0137)

ΔRATE*CAPITAL 0.0012

(0.0015)

ΔRATE*SIZE 0.0092

(0.0104)

ΔRATE*ROA -0.0146

(0.0195)

ΔRATE*RECESSION 0.1606***

(0.0179)

ΔRATE*TED 0.0013***

(0.0002)

ΔRATE*STED 0.0029***

(0.0004)

CAPITAL -0.0730*** -0.0729*** -0.0730*** -0.0730*** -0.0730*** -0.0730*** -0.0730***

(0.0058) (0.0058) (0.0058) (0.0058) (0.0058) (0.0058) (0.0058)

SIZE -0.5119*** -0.5121*** -0.5114*** -0.5119*** -0.5121*** -0.5120*** -0.5121***

(0.0594) (0.0594) (0.0592) (0.0593) (0.0593) (0.0593) (0.0593)

ROA 0.6115*** 0.6116*** 0.6116*** 0.6109*** 0.6122*** 0.6125*** 0.6124***

(0.0315) (0.0315) (0.0315) (0.0316) (0.0314) (0.0314) (0.0314)

GDPGROWTH -0.0228** -0.0228** -0.0227** -0.0227** -0.0229** -0.0231** -0.0230**

(0.0099) (0.0099) (0.0099) (0.0099) (0.0098) (0.0099) (0.0099)

UNEMP 0.0078 0.0077 0.0077 0.0077 0.0045 0.0049 0.0054

(0.0357) (0.0357) (0.0356) (0.0357) (0.0357) (0.0356) (0.0356)

HPI -0.1206*** -0.1207*** -0.1208*** -0.1203*** -0.1233*** -0.1236*** -0.1221***

(0.0365) (0.0364) (0.0365) (0.0366) (0.0360) (0.0363) (0.0363)

Bank fixed effects Yes Yes Yes Yes Yes Yes Yes

Year-quarter fixed effects Yes Yes Yes Yes Yes Yes Yes

Clustering State State State State State State State

Observations 351,767 351,767 351,767 351,767 351,767 351,767 351,767

Adjusted R-squared 0.0469 0.0469 0.0469 0.0469 0.0470 0.0470 0.0470

45

Table 7: Corporate taxation and bank risk-taking

This table presents the results of estimating equation 1 via OLS. The dependent variable is ΔARISK, the change in 100

percent risk-weighted assets scaled by total risk-weighted assets at the beginning of the quarter. The primary

independent variable is ΔRATE, the change in the state income tax rate on banks from year t-1 to year t in percentage

points. All regressions include the full set of control variables, bank fixed effects, and year-quarter fixed effects. All

variables are defined in the appendix. Standard errors presented below the coefficients are clustered at the state level.

***, **, * denote two-tailed statistical significance at the 1, 5, and 10 percent levels.

(1) (2) (3) (4) (5) (6) (7)

DEP VAR: ΔARISK ΔARISK ΔARISK ΔARISK ΔARISK ΔARISK ΔARISK

ΔRATE 0.0139 -0.0426 0.3368** -0.0419 0.0465*** 0.0500 0.0358

(0.0331) (0.0688) (0.1442) (0.0487) (0.0135) (0.0326) (0.0278)

ΔRATE*CAPITAL 0.0033

(0.0025)

ΔRATE*SIZE -0.0272**

(0.0129)

ΔRATE*ROA 0.0740***

(0.0246)

ΔRATE*RECESSION -0.1114***

(0.0168)

ΔRATE*TED -0.0005**

(0.0002)

ΔRATE*STED -0.0011**

(0.0004)

CAPITAL 0.1342*** 0.1344*** 0.1342*** 0.1341*** 0.1342*** 0.1342*** 0.1342***

(0.0081) (0.0081) (0.0081) (0.0082) (0.0081) (0.0081) (0.0081)

SIZE -1.0450*** -1.0461*** -1.0463*** -1.0457*** -1.0444*** -1.0446*** -1.0446***

(0.1140) (0.1134) (0.1136) (0.1139) (0.1141) (0.1141) (0.1140)

ROA 0.4788*** 0.4790*** 0.4787*** 0.4815*** 0.4778*** 0.4782*** 0.4781***

(0.0399) (0.0398) (0.0398) (0.0391) (0.0399) (0.0399) (0.0399)

GDPGROWTH 0.0430*** 0.0430*** 0.0430*** 0.0430*** 0.0431*** 0.0432*** 0.0431***

(0.0125) (0.0125) (0.0125) (0.0125) (0.0125) (0.0125) (0.0125)

UNEMP -0.2612*** -0.2612*** -0.2610*** -0.2612*** -0.2582*** -0.2600*** -0.2601***

(0.0483) (0.0483) (0.0483) (0.0482) (0.0484) (0.0483) (0.0483)

HPI 0.1348** 0.1347** 0.1350** 0.1340** 0.1401** 0.1363** 0.1360**

(0.0579) (0.0579) (0.0579) (0.0581) (0.0575) (0.0577) (0.0577)

Bank fixed effects Yes Yes Yes Yes Yes Yes Yes

Year-quarter fixed effects Yes Yes Yes Yes Yes Yes Yes

Clustering State State State State State State State

Observations 220,733 220,733 220,733 220,733 220,733 220,733 220,733

Adjusted R-squared 0.154 0.154 0.154 0.154 0.154 0.154 0.154