Banking deregulation and corporate tax avoidance

Bill B.Frncis,Ning Ren,Qing Wu,*

aLally School of Management,Rensselaer Polytechnic Institute,United States

bCollege of Management,Long Island University,Post,United States

Banking deregulation and corporate tax avoidance

Bill B.Francisa,Ning Renb,Qiang Wua,*

aLally School of Management,Rensselaer Polytechnic Institute,United States

bCollege of Management,Long Island University,Post,United States

ARTICLE INFO

Article history:

We investigate whether tax avoidance substitutes for external fi nancing.We exploit interstate banking deregulation as a quasi-external shock to examine whether fi rms engage in less tax avoidance after banking deregulation,because of cheaper and easier access to credit from banks.We fi nd no empirical evidence to support this substitutive relation,even for fi rms with higher fi nancial constraints or fi rms with higher external fi nancing dependence.

?2016 Sun Yat-sen University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

This study examines the substitutive relation between corporate tax avoidance and f i rms’use of debt.Corporate tax avoidance activities can increase a f i rm’s tax savings,and consequently decrease its reliance on external funding such as debt.The substitution between debt tax shields and non-debt tax shields is modeled theoretically(DeAngelo and Masulis,1980).However,empirical evidence on this issue is very mixed.Using a sample of 44 tax shelter f i rms,Graham and Tucker(2006)f i nd empirical evidence that is consistent with the substitutive relation between these two.However,Edwards et al.(2013)f i nd that only f i rms facing f i nancial constraints exhibit a higher level of tax avoidance.Further,Bradley et al.(1984)f i nd that non-debt tax shields and leverage are positively related,casting doubt on the existence of a signif i cant avoidance-debt-substitution ef f ect.

A major challenge in determining the empirical relation between tax avoidance and the use of debt is that both are endogenous in nature(Graham and Tucker,2006).We alleviate this concern by exploiting the staggered interstate banking deregulation events in the United States.The Interstate Banking and Branching Efficiency Act(IBBEA)was passed in 1994 and became ef f ective as of 1 June 1997.As described by Rice and Strahan(2010),during this period,states were allowed to erect up to four barriers to protect their local bank-ing industry from out-of-state competition.Prior studies f i nd that in states where the restrictions are relaxed, because of increased banking competition,f i rms have cheaper bank loans and easier access to credit(e.g.,Rice and Strahan,2010;Amore et al.,2013).1Using DealScan dataset,we examine how interstate banking deregulation a ff ects the costs and amounts of bank loans for US public fi rms.We fi nd that bank loan spreads are signi fi cantly reduced and bank loan amounts are signi fi cantly increased after banking deregulation.Therefore,interstate bank deregulation provides an ideal quasiexperiment to examine the substitutive relation between the use of credit and corporate tax avoidance.

Following Rice and Strahan(2010)and Cornaggia et al.(2015),we use RSindex to capture the degree of deregulation in dif f erent states and at dif f erent times.RSindex ranges from 0 to 4,indicating how many barriers the state erected after interstate banking deregulation.Firms in states that are open to competition(e.g., RSindex value is 0)have easier and cheaper access to bank loans than f i rms in less open states(e.g.,RSindex value is 4).

We construct tests using staggered interstate banking deregulation events as exogenous shocks to the credit supply and the cost of bank loans.If the substitutive relation between cash savings from tax avoidance and external f i nancing holds,we expect to observe a decrease in f i rms’tax avoidance practices after interstate deregulation when they have easier and cheaper access to external f i nancing(i.e.,bank loans).Therefore, we expect RSindex to be positively associated with tax avoidance.

Given that we are interested in broad tax avoidance strategies that could reduce the f i rm’s explicit taxes, following Dyreng et al.(2010),Hope et al.(2013)and Hasan et al.(2014),among others,we use the ef f ective tax rate(GAAP ETR),cash ef f ective tax rate(Cash ETR),discretionary book-tax dif f erence(Discretionary BT, as in Desai and Dharmapala(2006))and discretionary permanent book-tax dif f erence(DTAX,as in Frank et al.(2009)),as our measures of tax avoidance.

In our baseline model,we collect all available observations around interstate deregulation events(pooled sample)and control for standard determinants of tax avoidance.We f i nd that the coefficients on RSindex are not statistically signif i cant for all four measures of tax avoidance.Therefore,our results do not provide supportive evidence for the substitutive relation between tax avoidance practices and external f i nancing.We further constrain our sample to di ff erent event windows and fi nd similar results.

According to Edwards et al.(2013), fi rms with higher fi nancial constraints are more likely to exploit cash savings from tax avoidance practices.To examine whether the substitutive relation holds for fi rms with higher fi nancial constraints,we divide our sample into two2We also perform an interaction model,interacting RSindex with the dummy variable high_KZ score.The results are consistent with the subsample regressions.subgroups based on fi rms’ fi nancial constraint levels and perform the baseline model using these subsamples.We fi nd that interstate banking deregulation has no signi fi cant e ff ect on corporate tax avoidance even for fi rms that are facing higher fi nancial constraints.

We further examine whether the extent to which companies depend on external fi nance a ff ects the substitutive relation between tax avoidance and external fi nancing.We assume that fi rms with higher dependence on external fi nance are more likely to be a ff ected by interstate banking deregulation.The easier access to and lower cost of bank loans should make it easier for fi rms to access external funding,especially fi rms that are highly dependent on external fi nancing.We perform subsample tests based on the measure of fi rms’external fi nance dependence developed by Duchin et al.(2010).We fi nd that the coefficients on RSindex are not statistically signi fi cant,even for fi rms with higher dependence on external fi nancing.In summary,we fail to fi nd that interstate banking deregulation has a signi fi cant e ff ect on fi rms’tax avoidance behavior,even when fi rms are facing fi nancial constraints or highly dependent on external fi nancing.

This study contributes to a growing stream of literature that examines the determinants of tax avoidance. Previous studies fi nd very mixed results with regard to the relation between the use of debt and corporate tax avoidance.In our paper,we use banking deregulation as a natural experiment to better identify the e ff ect of external fi nancing shocks on tax avoidance behavior.Our empirical evidence fails to fi nd a signi fi cant substitutive relation between tax avoidance and the use of debt,even for fi rms with fi nancial constraints.Our paper sheds light on the debate in this research fi eld.Our paper also contributes to the banking literature that examines the real e ff ects of banking deregulation on corporate decision-making.

The rest of the paper proceeds as follows.Section 2 develops the hypothesis on the substitutive relation between cash savings from tax avoidance and external f i nance.Section 3 describes the sample.Section 4 presents the empirical results.Section 5 provides the subsample analyses and Section 6 explores the ef f ect of intrastate deregulation.Section 7 concludes.

2.Hypothesis development

Numerous studies focus on the ef f ects of f i rm-level characteristics on tax avoidance(e.g.,Chen and Chu, 2005;Desai and Dharmapala,2006;Wilson,2009;Armstrong et al.,2012;Hoi et al.,2013).However,many research questions remain unanswered.Maydew(2001)and Hanlon and Heitzman(2010)suggest that the theoretical and empirical tax research from the accounting,f i nance and economics f i elds should be integrated to provide a more in-depth perspective on this issue.In this paper,we examine whether cash savings from tax avoidance can be substituted for the use of debt.

Prior tax avoidance studies suggest that tax planning is not free.In fact,it can be very costly to build up complex tax avoidance strategies(Desai and Dharmapala,2006).Similar to raising funds from external sources,managers need to exploit unused tax strategies and consider the potential cost of cash savings from tax planning.According to Edwards et al.(2013),the implicit discount rate of funds is determined by(1)the amount of cash saved from tax planning,(2)the expected timing of eventual repayment(if at all),(3)potential penalties if caught by the tax authorities and(4)the cost of designing and implementing additional tax strategies.Not all of these costs apply to every tax strategy.For example,deferral strategies are similar to an interest-free loan obtained from the government,but f i rms eventually need to repay them.When f i rms use permanent tax avoidance strategies,they may avoid paying back the taxes if they are not sued by the tax authorities,but otherwise they will pay high penalties and interest on the tax owed to the tax authorities.3Andreoni(1992)models this tax avoidance behavior.This discussion does not include the costs associated with implementing tax planning strategies.Mills et al.(1998)estimate that for every$1 invested in general tax planning,although these costs are not trivial,f i rms have an average return of approximately$4.

DeAngelo and Masulis(1980)theoretically model the substitution between debt tax shields and non-debt tax shields.They argue that tax deductions from tax avoidance(e.g.,tax sheltering)are non-debt tax shields, and non-debt tax shields can be substituted for tax deductions from debt interest.DeAngelo and Masulis (1980)show that a f i rm has an optimal amount of total tax deductions.If a f i rm uses more non-debt tax shields,it will use fewer debt tax deductions.Graham and Tucker(2006)empirically test DeAngelo and Masulis’s(1980)theory.Using a sample of 44 tax shelter f i rms,Graham and Tucker(2006)f i nd evidence to support the substitutive relation between tax avoidance and the use of debt.However,other studies using larger samples f i nd very mixed results.For example,Edwards et al.(2013)f i nd that the substitutive relation only exists for f i rms facing f i nancial constraints.Bradley et al.(1984)f i nd the totally opposite result that nondebt tax shields and leverage are positively related.

Extant studies have examined the real ef f ects of banking deregulation.Jayaratne and Strahan(1996)suggest that bank deregulation signif i cantly increases the rates of real per capita growth in income and output.Black and Strahan(2002) fi nd that deregulation spurs entrepreneurship and helps small business and new business fl ourish.Morgan et al.(2004) fi nd that the state-level business cycle is generally less volatile after the interstate banking regulation and the associated fi nancial integration.A more related study by Rice and Strahan(2010) fi nds that after interstate banking deregulation,small fi rms have a lower cost of debt and easier access to bank loans in states that are more open to branching.Amore et al.(2013) fi nd that interstate bank deregulation is associated with an 8%increase in the total net loan supply.Recent studies use bank deregulation as an exogenous shock to the credit supply and examine how it a ff ects corporate decisions and outcomes(e.g.,Rice and Strahan,2010;Amore et al.,2013;Francis et al.,2014;Cornaggia et al.,2015).Based on these studies,we examine whether states’openness to branching directly a ff ects fi rms’tax avoidance behavior.The openness to branching is positively correlated with lower loan costs and easier access to credit;hence,we are able to use these staggered deregulation events to test whether cash savings from tax avoidance substitute for external fi nancing.

In sum,we predict that in states that are more open to branching after interstate deregulation,f i rms are more likely to reduce their tax avoidance practices because they have easier access to lower cost bank loans. We propose the following hypothesis:

H1:Tax avoidance is negatively correlated with states’openness to branching.

3.Sample and summary statistics

3.1.Sample

To investigate the ef f ect of banking deregulation on corporate tax avoidance,we obtain data from two sources.We obtain data on interstate banking deregulation from Rice and Strahan(2010)and f i nancial information from Standard&Poor’s Compustat.Following the tax avoidance literature,we exclude f i rms in the utility(SIC codes 4900-4949)and f i nance(SIC codes 6000-6999)industries.We merge the f i rm data from Compustat with the deregulation data if a f i rm is headquartered in the same state as the deregulation state. After dropping missing information,we f i nally have 48,013 f i rm-year observations for 7,374 unique f i rms in 50 states.Table 1 reports the sample distribution by f i scal year.The f i rm-year observations are relatively evenly distributed from 1987 to 2010.

3.2.Variables

We construct a variable named RSindex following Rice and Strahan(2010).As described in Rice and Strahan(2010),the IBBEA allowed states to erect out-of-state entry barriers from the time of enactment in 1994 until 1 June 1997.States could use any combination of the following four provisions to set their barriers to interstate branching:(i)a minimum age for the target institution;(ii)de novo interstate branching;(iii)theacquisition of individual branches;and(iv)a statewide deposit cap.RSindex represents the number of provisions a state sets on interstate branching.Therefore,RSindex ranges from 0 to 4.States that have an RSindex of 0 are the most open toward branching and vice versa.4See Rice and Strahan(2010)for a detailed discussion and information about interstate deregulation.

Table 1Sample distribution.

Table 2Summary statistics.

Hanlon and Heitzman(2010)state that‘‘if tax avoidance represents a continuum of tax planning strategies where something like municipal bond investments are at one end,then terms such as‘noncompliance,’‘evasion,’‘a(chǎn)ggressiveness,’and‘sheltering’would be closer to the other end of the continuum.”We are interested in all tax avoidance strategies that could reduce explicit taxes.Following studies such as Dyreng et al.(2010), Hope et al.(2013)and Hasan et al.(2014),we use four measures to capture this continuum.Two alternate tax rate measures,GAAP ETR and Cash ETR,are used to estimate broad tax avoidance practices(Dyreng et al., 2010;Hanlon and Heitzman,2010).GAAP ETR is the ratio of total tax expenses to pretax income for a f i rm in a given year.Cash ETR is the ratio of cash tax paid to pretax income for a f i rm in a given year.By def i nition,a higher GAAP ETR or Cash ETR value means less corporate tax avoidance.

Two measures of book-tax dif f erence are used to capture more aggressive tax planning strategies.Book-tax di ff erence is a reasonable measure of more aggressive tax avoidance.For instance,Mills et al.(1998) fi nd that fi rms with large book-tax di ff erences are more likely to be audited by the IRS and have larger proposed audit adjustments.Wilson(2009) fi nds that book-tax di ff erences are larger for fi rms accused of engaging in tax shelters than for a matched sample of non-accused fi rms.Our book-tax di ff erence measures are(1)the Desai and Dharmapala(2006)discretionary book-tax di ff erence(Discretionary BT)and(2)the Frank et al.(2009)permanent discretionary book-tax di ff erence(DTAX).Higher Discretionary BT or DTAX means more aggressive tax avoidance.

Following the literature,we control for a vector of fi rm characteristics that may a ff ect a fi rm’s tax avoidance practice.Variable de fi nitions are presented in the Appendix.

3.3.Summary statistics

In Table 2,Panel A reports the summary statistics of measures that capture tax avoidance practices.Due to the data requirement when constructing these measures,the sample sizes vary from 17,555 for DiscretionaryBT to 41,482 for GAAP ETR.The sample statistics for these tax avoidance measures are similar to those in the extant tax literature(e.g.,Dyreng et al.,2008;Hasan et al.,2014).

In Table 2,Panel B shows the summary statistics of RSindex and other control variables for the full sample. The mean of RSindex is about 2.80 with a standard deviation of 1.41,consistent with the study by Cornaggia et al.(2015).The statistics of the other control variables are in the range of those reported in previous studies (e.g.,Armstrong et al.,2012).

4.Empirical results

4.1.Baseline regression results

Following Rice and Strahan(2010)and Cornaggia et al.(2015),we use interstate banking deregulations as external shocks to the supply and the price of bank loans.The staggered multiple deregulation events alleviate the problem associated with a single shock design,and exclude the possibility of some omitted factors coinciding with the shock that could af f ect the dependent variable.We estimate the following model:

RSindex is the key independent variable,indicating the openness of the state toward out-of-state banking competition.Following the tax avoidance literature,we include relevant f i rm characteristics that may af f ect tax avoidance.We include the state f i xed ef f ect to control for omitted time-invariant state factors that might be correlated with tax avoidance practices such as the legal environment and the strictness of the tax authority.5Dif f erent regions may have dif f erent levels of tax enforcement.Instead of controlling for state f i xed ef f ects,in a robustness check we control for region f i xed ef f ects and our main results hold.We also control for year and industry f i xed ef f ects in the model.

Table 3 presents the results for the baseline OLS regression model in Eq.(1)and the adjusted standard errors for within-f i rm clustering and heteroscedasticity.The f i rst column represents the regression result for Cash ETR.The estimated coefficient is 0.0017 and is not statistically signif i cant,suggesting that interstate banking deregulation has no signif i cant ef f ect on f i rms’tax avoidance practices.The coefficients on the control variables are generally consistent with previous studies.

Column 2 shows the regression results for GAAP ETR.The results are similar to those in Column 1:the coefficient on RSindex is positive,but not signif i cantly correlated with GAAP ETR.We present the regression results for Discretionary BT and DTAX in Columns 3 and 4.Consistent with ef f ective tax rate-based tax avoidance measures,the coefficients on RSindex are not signif i cant,further suggesting that interstate banking deregulation has no signif i cant ef f ect on f i rms’tax avoidance practices.

We further include observations for a(-3,+3)-year window around interstate deregulation events and perform the baseline regression model.Table 4 presents the results for the(-3,+3)-year window sample.We exclude observations in the year of the interstate deregulations.Because by eliminating some potential policy or business environment changes in the state over the long run,we can obtain a relatively clear view of how interstate deregulation could change tax avoidance behavior.The results for Cash ETR,GAAP ETR,Discretionary BT and DTAX are presented separately in Columns 1 to 4.The results in these regressions are consistent with those in Table 3,indicating that there is no signif i cant ef f ect of banking deregulation on corporate tax avoidance.

4.2.Robustness checks

In this section,we perform a series of robustness tests to ensure that our baseline regression results hold. We discuss the purposes and results of these additional analyses below.

We use four dummy variables to represent the dif f erent values of RSindex.For example,dummy variable RSindex_0 equals 1 if the value of RSindex is 0;we also create RSindex_1,RSindex_3 and RSindex_4.We add these dummy variables to the baseline model instead of using RSindex.Therefore,the group with an RSindexvalue of 2 is considered as the reference group.Table 5 shows the regression results using this model.Only 2 of the 16 coefficients are signif i cant.Specif i cally,the coefficient on RSindex_1 is 0.023 and signif i cant at the 5% level when we use Cash ETR as the dependent variable,suggesting that f i rms in states where RSindex equals 1 pay more ef f ective cash tax than f i rms in states where RSindex equals 2.The results also show that the GAAP ETR is signif i cantly lower for f i rms in states where RSindex equals 3 than that for f i rms in states where RSindex equals 2.Overall,we f i nd no other signif i cant dif f erences for the other tax measures and groups.

Table 3Baseline regression.

To estimate the possibility of a nonlinear relation,we perform two nonlinear regression models using logRSindex and sqrRSindex,separately.logRSindex equals the natural logarithm of 1 plus RSindex.Table 6 presents the regression results.Consistent with the main f i ndings,interstate deregulation is not signif i cantly correlated with tax avoidance measures.sqrRSindex equals the square of RSindex.To perform the nonlinear model,we include both RSindex and sqrRSindex in the regressions.The results presented in Table 7 show that the coefficients on these two independent variables are not statistically signif i cant for the four tax avoidance measures,consistent with the main f i ndings.

To compare the states at the two ends of the spectrum of openness toward branching deregulation,we perform a test using a subsample of f i rms at the two extremes.More specif i cally,the subsample includes only states with an RSindex value of 0(totally open to branching)and 4(most restrictive to branching)up to 1 June 1997.We also include f i rm-year observations in these states before the IBBEA ef f ective date.BranchRes-trict is a dummy variable that equals 1 if the state implements four provisions by 1 June 1997,and 0 otherwise. After is a dummy variable that equals 1 when the f i scal year of the f i rm observation is after 1997.We drop the observations in the 1997 f i scal year to avoid contamination ef f ects.We are interested in the variable Restrict_-After,which is an interaction term between BranchRestrict and After.This research design is similar to a standard dif f erence-in-dif f erences method.

Table 4Baseline Regression:(-3,+3)window.

Table 8 presents the regression results for this dif f erence-in-dif f erences test.The results show that states that are restrictive to branching consistently pay lower GAAP ETR.Discretional BT is signif i cantly higher after the interstate deregulation event.For the interaction term Restrict_After,it is signif i cant at the 10%level when we use DTAX as the measure of tax avoidance,but it is not signif i cant for the three other measures of tax avoidance.The results in general suggest that there is no signif i cant relation between banking deregulation and tax avoidance practices.

It is possible that dif f erent regions may have dif f erent levels of tax enforcement.To mitigate this regional ef f ect,we perform a robustness check to examine whether our main results are af f ected by region.Specif i cally, following the United States Census Bureau def i nition,we divide states into four regions:Northeast,Midwest, South and West.We then add region as a f i xed ef f ect to our baseline model.We f i nd that our main results are unchanged after controlling for region.Finally,we test whether our results are driven by large states with more observations.Specif i cally,we drop observations from California,the state with the largest number of observations.We also drop observations from California,Texas and New York,the top three states in terms of the number of observations.Our main results hold for these robustness checks,suggesting that our results are not driven by certain large states.For brevity,the results of these robustness checks are not tabulated.

Table 5Robustness test 1.

To sum up,our robustness checks support that there is no evidence of a signif i cant relation between tax avoidance and bank deregulation.

5.Subsample tests

As discussed earlier,although the baseline tests show no signi fi cant results for RSindex,we assume that fi rms with high fi nancial constraints are more likely to be a ff ected by interstate banking deregulation,because easier access to bank loans after deregulation may relieve their fi nancial constraints and subsequently change their tax avoidance practices.In this section,we examine whether companies’ fi nancial constraints moderate the e ff ect of interstate deregulation on tax avoidance behavior.

Table 6Robustness test 2.

5.1.Financial constraints

If banking competition after interstate deregulation directly af f ects f i rms’access to bank loans,it may relax their f i nancial constraints and decrease their tax avoidance practices for cash saving purposes.We use Kaplan and Zingales’s(1997) fi nancial constraint index,KZscore,and divide the(-3,+3)-year window sample into fi ve subgroups according to the value of KZscore.We de fi ne fi rms in the top two quintile groups as highly fi nancially constrained fi rms and those in the bottom two quintile groups as low fi nancially constrained fi rms. The regression results for highly fi nancially constrained fi rms are presented in Table 9,Panel A.The coefficient on RSindex is-0.000174 and is not statistically signi fi cant when we use Cash ETR as the dependent variable. Similar results are found for all other three tax avoidance measures.The results suggest that bank deregulation has no signi fi cant e ff ect on tax avoidance practices even for fi nancially constrained fi rms.In Panel B,we perform regressions for low fi nancially constrained fi rms,and again the coefficients on RSindex are not signi fi cant for all tax avoidance measures.There are no signi fi cant di ff erences between these two groups.Therefore,we do not fi nd evidence to support the substitutive relation between tax avoidance and the use of debt,even for fi nancially constrained fi rms.

Table 7Robustness test 3.

5.2.External f i nance dependence

Our earlier tests show no signif i cant ef f ect of banking deregulation on tax avoidance practices,and thus do not support the use of debt as substitutive of the cash savings from tax avoidance practices.In this subsection, we examine whether companies that are highly dependent on external f i nancing sources substitute cash savings from tax avoidance with the use of debt.We expect that f i rms that are highly dependent on external f i nance decrease their tax avoidance,because they have easier access to credit after interstate deregulation.We use the measure of external f i nance dependence developed by Duchin et al.(2010)and construct a dummy variable, High_dependence,that equals 1 if the value of f i rm-year external f i nance dependence(EFD)is above the industry median(indicating higher EFD),and 0 otherwise.

In Table 10,Panel A shows the subsample regression analysis for high EFD f i rms.The coefficient on RSindex is 0.0080 and not statistically signif i cant when we use Cash ETR as the dependent variable.Similar results are found for the other three tax avoidance measures.Panel B reports the regression results for low EFD f i rms. Again,the coefficients on RSindex are not statistically signif i cant for all four tax avoidance measures.From these tests,we cannot conclude that these two subgroups are statistically and signif i cantly dif f erent from each other.The results indicate that external f i nancing dependence does not moderate the ef f ect of interstate banking deregulation on tax avoidance practices.

Table 8Robustness test 4.

6.Intrastate deregulation

The literature documents dif f erent ef f ects imposed by intrastate and interstate deregulations(e.g.,Chava et al.,2013).In this section,we perform an additional test on the ef f ect of intrastate deregulation on tax avoidance behavior.We assume that after intrastate deregulation,f i rms with severe f i nancial constraints may increase their tax avoidance savings,because they may face more restrictive screening technology by banks (Dick and Lehnert,2010).

We report the regression results in Table 11.Panel A presents the pooled sample regression results,which include all f i rm-year observations in the pre-and post-intrastate deregulation periods.The coefficients on Post_intra are not statistically signif i cant for all four measures of tax avoidance,suggesting that intrastate deregulation has no signif i cant ef f ect on f i rms’tax avoidance behavior.Panel B shows the regression results for a(-3,+3)-year window around intrastate deregulation events.Consistent with the pooled sample regres-sion results,the coefficients on Post_intra are not statistically signif i cant for all tax avoidance measures.In sum,we do f i nd no empirical evidence to support the ef f ect of intrastate deregulation on corporate tax avoidance.

Table 9Subsample regressions 1.

Table 9(continued)

Table 10Subsample regressions 2.

Table 10(continued)

Table 11Intrastate deregulation regressions.

Table 11(continued)

7.Conclusion

This paper examines the substitutive relation between tax avoidance practices and the use of debt.We use interstate banking deregulations as external shocks to the supply and the price of credit.In states that are more open toward branching,f i rms are more likely to enjoy lower loan costs and easier access to credit.If the substitutive relation holds,those f i rms in open states should make more use of debt and consequently engage in less tax avoidance.However,we do not f i nd empirical evidence to support this hypothesis.Further,we f i nd that f i rms do not signif i cantly change their tax avoidance behavior when they relax their f i nancial constraints. This study contributes to the current debate on whether a substitutive relation exists between tax avoidance savings and the use of external f i nancing.This paper also helps to understand the real ef f ect of banking deregulation on the real economy.

Although we do not observe a signif i cant ef f ect of bank deregulation on corporate tax avoidance,we are cautious in drawing the conclusion that there is no substitutive relation between corporate tax avoidance and the use of debt.It is possible that such an ef f ect may indeed exist,but we are simply unable to f i nd it empirically due to some limitations and caveats in our study.First,our four measures of tax avoidance in gen-eral capture the overall level of tax avoidance practices.It is possible that there is a substitutive relation between the use of debt and certain types of tax avoidance strategies,which are not directly measured in our study.Second,Graham and Tucker(2006)f i nd a substitutive relation between the usage of debt and more aggressive tax avoidance such as tax sheltering,while our four measures capture more broad tax avoidance. Third,another potential explanation for our main results is that tax avoidance behavior is permanent or not reversible.When f i rms are facing f i nancial constraints or are unable to meet their capital demand,they might exploit unused tax planning strategies and save cash for their capital demand.However,whether they reverse their tax avoidance strategies when their f i nancial constraints relax is unknown.If tax avoidance strategies are permanent or irreversible,it is not surprising that we do not observe a signif i cant change after banking deregulation.Fourth,although our research design is based on interstate banking deregulation as an external shock,our results could still be biased by some f i rm-level or region-level omitted variables.

Appendix A.Variable def i nitions

Amore,M.D.,Schneider,C.,Zˇaldokas,A.,2013.Credit supply and corporate innovation.J.Financ.Econ.109(3),835-855.

Andreoni,J.,1992.IRS as loan shark:tax compliance with borrowing constraints.J.Public Econo.49(1),35-46.

Armstrong,C.S.,Blouin,L.J.,Larcker,D.F.,2012.The incentives for tax planning.J.Account.Econo.53(1-2),391-411.

Black,S.,Strahan,P.,2002.Entrepreneurship and bank credit availability.J.Finan.57(6),2807-2833.

Bradley,M.,Jarrell,G.A.,Kim,E.,1984.On the existence of an optimal capital structure:theory and evidence.J.Finan.39(3),857-878.

Chava,S.,Oettl,A.,Subramanian,A.,Subramanian,K.V.,2013.Banking deregulation and innovation.J.Finan.Econ.109(3),759-774.

Chen,K.P.,Chu,C.C.,2005.Internal control vs external manipulation:a model of corporate income tax evasion.Rand J.Econ.36(1), 151-164.

Cornaggia,J.,Mao,Y.,Tian,X.,Wolfe,B.,2015.Does banking competition af f ect innovation?J.Finan.Econ.115(1),189-209.

DeAngelo,H.,Masulis,R.W.,1980.Optimal capital structure under corporate and personal taxation.J.Finan.Econ.8(1),3-27.

Desai,M.,Dharmapala,D.,2006.Corporate tax avoidance and high-powered incentives.J.Finan.Econ.79,145-179.

Dick,A.,Lehnert,A.,2010.Personal bankruptcy and credit market competition.J.Finan.65,655-686.

Duchin,R.,Ozbas,O.,Sensoy,B.,2010.Costly external f i nance,corporate investment,and the subprime mortgage credit crisis.J.Finan. Econ.97(3),418-435.

Dyreng,S.,Hanlon,M.,Maydew,E.L.,2008.Long-run corporate tax avoidance.Account.Rev.83(1),61-82.

Dyreng,S.D.,Hanlon,M.,Maydew,E.L.,2010.The ef f ects of executives on corporate tax avoidance.Account.Rev.85(4),1163-1189.

Edwards,A.,Schwab,C.M.,Shevlin,T.J.,2013.Financial constraints and the incentive for tax planning.Working Paper.

Francis,B.,Hasan,I.,Wang,H.,2014.Banking deregulation,consolidation,and corporate cash holdings:US evidence.J.Bank.Finan. 41,45-56.

Frank,M.M.,Lynch,L.,Rego,S.,2009.Tax reporting aggressiveness and its relation to aggressive f i nancial reporting.Account.Rev.84 (2),467-496.

Graham,J.,Tucker,A.,2006.Tax shelters and corporate debt policy.J.Finan.Econ.81(3),563-594.

Hanlon,M.,Heitzman,S.,2010.A review of tax research.J.Account.Econo.50(2-3),127-178.

Hasan,I.,Hoi,C.K.S.,Wu,Q.,Zhang,H.,2014.Beauty is in the eye of the beholder:the ef f ect of corporate tax avoidance on the cost of bank loans.J.Finan.Econ.113(1),109-130.

Hoi,C.,Wu,Q.,Zhang,H.,2013.Is corporate social responsibility(CSR)associated with tax avoidance?Evidence from irresponsible CSR activities.Account.Rev.88,2025-2059.

Hope,Ole-Kristian,Ma,Mark,Thomas,Wayne B.,2013.Tax avoidance and geographic earnings disclosure.J.Account.Econo.56,170-189.

Jayaratne,J.,Strahan,P.,1996.The f i nance-growth nexus:evidence from bank branch deregulation.Quart.J.Econ.111(3),639-670.

Kaplan,S.N.,Zingales,L.,1997.Do investment-cash f l ow sensitivities provide useful measures of f i nancing constraints?Quart.J.Econ. 112(1),169-215.

Maydew,E.L.,2001.Empirical tax research in accounting:a discussion.J.Account.Econo.31,389-403.

Mills,L.,Erickson,M.,Maydew,E.,1998.Investments in tax planning.J.Am.Taxation Assoc.20(1),1-20.

Morgan,D.,Rime,B.,Strahan,P.,2004.Bank integration and state business cycles.Quart.J.Econ.119(4),1855-1885.

Rice,T.,Strahan,P.E.,2010.Does credit competition af f ect small-f i rm f i nance?J.Finan.65(3),861-889.

Wilson,R.,2009.An examination of corporate tax shelter participants.Account.Rev.84,969-999.

1 March 2016

*Corresponding author.

E-mail addresses:francb@rpi.edu(B.B.Francis),ning.ren@liu.edu(N.Ren),wuq2@rpi.edu(Q.Wu).

http://dx.doi.org/10.1016/j.cjar.2016.09.004

1755-3091/?2016 Sun Yat-sen University.Production and hosting by Elsevier B.V.

This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Accepted 27 September 2016

Available online 24 November 2016

Tax avoidance

Interstate banking deregulation

External fi nancing