19
The importance of the internal information environment for tax avoidance $, $$ John Gallemore a , Eva Labro b,n a University of Chicago, USA b University of North Carolina, USA article info Article history: Received 12 April 2013 Received in revised form 25 July 2014 Accepted 8 September 2014 JEL classification: M41 Keywords: Internal information quality Management accounting Tax avoidance Tax risk abstract We show that firms' ability to avoid taxes is affected by the quality of their internal information environment, with lower effective tax rates (ETRs) for firms that have high internal information quality. The effect of internal information quality on tax avoidance is stronger for firms in which information is likely to play a more important role. For example, firms with greater coordination needs because of a dispersed geographical presence benefit more from high internal information quality. Similarly, firms operating in a more uncertain environment benefit more from the quality of their internal information in helping them to reduce ETRs. In addition, we provide evidence that high internal information quality allows firms to achieve lower ETRs without increasing the risk of their tax strategies (as measured by ETR volatility). Overall, our study contributes to the literature on tax avoidance by providing evidence that the internal information environ- ment of the firm is important for understanding its tax avoidance outcomes. & 2014 Elsevier B.V. All rights reserved. 1. Introduction In this paper, we argue that in order to understand firms' tax avoidance outcomes it is necessary to consider the role the quality of a firm's internal information environment plays in supporting such outcomes. We define internal information quality (hereafter IIQ) in terms of the accessibility, usefulness, reliability, accuracy, quantity, and signal-to-noise ratio of the data and knowledge collected, generated, and consumed within an organization. Decision theory has established that the quality of the information on which decisions are based affects the quality of those decisions and their outcomes. Hence, the firm's ability to avoid taxes is likely to be affected by the quality of the information on which tax planning decisions are based. Without good information, tax-reducing opportunities might be overlooked, coordination of tax planning across the Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/jae Journal of Accounting and Economics http://dx.doi.org/10.1016/j.jacceco.2014.09.005 0165-4101/& 2014 Elsevier B.V. All rights reserved. We thank Wayne Guay (the editor), Richard Sansing (the referee), Eddy Cardinaels (discussant at Tilburg Spring Camp), Ranjani Krishnan (discussant at GMARS conference), Petro Lisowsky (discussant at EAA conference), VG Narayanan (discussant at UNC Tax Symposium), Mary Margaret Frank, Thomas Hemmer, Margot Howard, Mark Lang, Ed Maydew, Jim Omartian, Doug Shackelford, Jake Thornock, Kelly Wentland, Hal White, and workshop participants at Harvard Business School, Tilburg Spring Camp, the University of North Carolina, the 2013 UNC Tax Symposium, the 2013 EAA conference, and the 2013 GMARS conference for comments. We thank Peter Iliev for sharing data on SOX public floats, and Karen Hennes, Andy Leone, and Brian Miller for making available their restatement classification data. Any errors are our own. John Gallemore gratefully acknowledges the financial support of the Deloitte Foundation. Eva Labro gratefully acknowledges the financial support of the Kenan-Flagler Business School and the Latané Fund. ☆☆ Data availability: from source identified in the text. n Correspondence to: Kenan-Flagler Business School, CB 3490, McColl Building, Chapel Hill, NC 27599, USA. Tel.: þ1 919 962 5747. E-mail address: [email protected] (E. Labro). Journal of Accounting and Economics ] (]]]]) ]]]]]] Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment for tax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

The importance of the internal information environment for ...taxdoctoralseminar.web.unc.edu/files/2016/02/Gallemore-Labro-1.pdf · The importance of the internal information environment

  • Upload
    vokiet

  • View
    215

  • Download
    0

Embed Size (px)

Citation preview

Contents lists available at ScienceDirect

Journal of Accounting and Economics

Journal of Accounting and Economics ] (]]]]) ]]]–]]]

http://d0165-41

☆ WeGMARSHemmeat HarvGMARSavailablFounda

☆☆ Dan CorrE-m

Pleastax a

journal homepage: www.elsevier.com/locate/jae

The importance of the internal information environmentfor tax avoidance$,$$

John Gallemore a, Eva Labro b,n

a University of Chicago, USAb University of North Carolina, USA

a r t i c l e i n f o

Article history:Received 12 April 2013Received in revised form25 July 2014Accepted 8 September 2014

JEL classification:M41

Keywords:Internal information qualityManagement accountingTax avoidanceTax risk

x.doi.org/10.1016/j.jacceco.2014.09.00501/& 2014 Elsevier B.V. All rights reserved.

thank Wayne Guay (the editor), Richard Sansconference), Petro Lisowsky (discussant atr, Margot Howard, Mark Lang, Ed Maydew, Jiard Business School, Tilburg Spring Camp, thconference for comments. We thank Peter Ile their restatement classification data. Anytion. Eva Labro gratefully acknowledges theta availability: from source identified in theespondence to: Kenan-Flagler Business Schoail address: [email protected] (E. Labro).

e cite this article as: Gallemore,voidance. Journal of Accounting an

a b s t r a c t

We show that firms' ability to avoid taxes is affected by the quality of their internalinformation environment, with lower effective tax rates (ETRs) for firms that have highinternal information quality. The effect of internal information quality on tax avoidance isstronger for firms in which information is likely to play a more important role. Forexample, firms with greater coordination needs because of a dispersed geographicalpresence benefit more from high internal information quality. Similarly, firms operating ina more uncertain environment benefit more from the quality of their internal informationin helping them to reduce ETRs. In addition, we provide evidence that high internalinformation quality allows firms to achieve lower ETRs without increasing the risk of theirtax strategies (as measured by ETR volatility). Overall, our study contributes to theliterature on tax avoidance by providing evidence that the internal information environ-ment of the firm is important for understanding its tax avoidance outcomes.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

In this paper, we argue that in order to understand firms' tax avoidance outcomes it is necessary to consider the role thequality of a firm's internal information environment plays in supporting such outcomes. We define internal informationquality (hereafter IIQ) in terms of the accessibility, usefulness, reliability, accuracy, quantity, and signal-to-noise ratio of thedata and knowledge collected, generated, and consumed within an organization. Decision theory has established thatthe quality of the information on which decisions are based affects the quality of those decisions and their outcomes. Hence,the firm's ability to avoid taxes is likely to be affected by the quality of the information on which tax planning decisions arebased. Without good information, tax-reducing opportunities might be overlooked, coordination of tax planning across the

ing (the referee), Eddy Cardinaels (discussant at Tilburg Spring Camp), Ranjani Krishnan (discussant atEAA conference), VG Narayanan (discussant at UNC Tax Symposium), Mary Margaret Frank, Thomasm Omartian, Doug Shackelford, Jake Thornock, Kelly Wentland, Hal White, and workshop participantse University of North Carolina, the 2013 UNC Tax Symposium, the 2013 EAA conference, and the 2013iev for sharing data on SOX public floats, and Karen Hennes, Andy Leone, and Brian Miller for makingerrors are our own. John Gallemore gratefully acknowledges the financial support of the Deloittefinancial support of the Kenan-Flagler Business School and the Latané Fund.text.ol, CB 3490, McColl Building, Chapel Hill, NC 27599, USA. Tel.: þ1 919 962 5747.

J., Labro, E., The importance of the internal information environment ford Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]2

different parts of the firm might be difficult, tax risk might be high, and the firm's documentation might not be acceptable tothe tax authorities. However, the internal information environment has been largely ignored in the tax avoidance literature.

Anecdotal evidence indeed suggests that internal information quality may play an important role in a firm's taxavoidance outcomes. For example, Bayer Group predicted that it would take several years to sort through thedocumentation problems with the IRS related to R&D tax credits dating back to 1998, because the company's accountingsystem could not readily provide the required information (McKinnon, 2012). Additionally, although globally coordinatedtransfer pricing schemes can allow multinational firms to avoid a substantial amount of taxes, a recent survey indicated thatonly 41 percent of multinationals prepared transfer pricing documentation concurrently on a globally coordinated basis. Theother 59 percent prepared documentation either for a single country (modified to the needs of other jurisdictions asappropriate), on an as-necessary country-by-country basis with limited coordination between countries, or not at all (Ernstand Young, 2010). Decision theory and these examples suggest that firms with higher IIQ might be better able to identifyand implement tax-reducing strategies. We empirically examine this hypothesis in our study.

We use four publicly available proxies of a firm's IIQ to examine the effect of IIQ on tax avoidance over the period 1994–2010: the speed with which management releases an earnings announcement after its fiscal year closing, the accuracy ofmanagement's earnings forecasts, the absence of Sarbanes-Oxley (SOX) Section 404 material weaknesses in internalcontrols, and the absence of restatements due to errors. With these IIQ proxies, and using the cash effective tax rate (CashETR) as the measure of tax avoidance, which we define as the reduction of explicit taxes according to Hanlon and Heitzman(2010), we find a statistically significant positive association between IIQ and tax avoidance.1 Specifically, we find that a one-standard-deviation increase in the continuous IIQ proxies (earnings announcement speed and management forecastaccuracy) is associated with a Cash ETR reduction of 1–2 percentage points, and that the absence of a restatement due toerrors (Section 404 material weakness) is associated with a Cash ETR reduction of approximately 2 (5) percentage points.These findings are consistent with the notion that IIQ is an economically important determinant of tax avoidance.

We then investigate cross-sectional differences in the effect of IIQ on tax avoidance that help explain the mechanismsthrough which IIQ affects tax avoidance. First, we predict—and find—that IIQ is more important in firms with moregeographically dispersed operations and hence greater coordination needs. A higher IIQ reduces information asymmetryand improves information coordination between the various business units, allowing for more effective tax planning.Second, we predict—and find—that firms operating in more uncertain environments benefit more from a higher IIQ. Forthese firms, IIQ can help identify tax opportunities, reduce doubts about the payoffs of particular tax planning strategies, andhelp with forecasting over a wide range of potential tax outcomes.

Finally, we examine the effect of IIQ on tax risk, which we define as the firm's uncertainty regarding its tax liability. Weexpect that high-IIQ firms are able to pursue more favorable tax avoidance outcomes without taking more risk, because abetter IIQ can support improved and sustained tax planning and documentation, which in turn can reduce the likelihoodthat particular tax strategies will be disallowed. Firms with lower tax risk are likely to experience less volatility in taxoutcomes. Thus, we use the standard deviation of Cash ETR over five years as a measure of tax risk (Guenther et al., 2012).We document that ETR volatility is substantially lower in high-IIQ environments, consistent with a higher IIQ leading tolower tax risk.

Our results suggest that high-IIQ firms achieve both increased tax avoidance and decreased tax risk. This finding shedslight on the difficulty of conceptually defining tax aggressiveness (Hanlon and Heitzman, 2010). Referring to the model byMills et al. (2010), both Frischmann et al. (2008) and Rego and Wilson (2012) define aggressive tax reporting as pursuingsignificant tax strategies with relatively weak supporting facts. Hence, the aggressiveness of a firm's tax position is a functionof its IIQ. Our findings indicate that when a firm has high-quality information to support its tax planning strategies, it is nottaking on more risk or behaving more aggressively when reducing its ETR.

Of course, it is likely that certain characteristics of the firm, such as the presence of foreign operations, are associatedwith both a higher IIQ and greater tax avoidance opportunities. Furthermore, both IIQ and tax avoidance are choices of thefirm, suggesting that endogeneity is a concern. We conduct a battery of additional tests that we believe mitigate concernsabout our results being partly attributable to correlated omitted variables and endogeneity. First, we use an extensive set ofcontrol variables and industry fixed effects in our analyses. Second, our inferences are unaffected if we employ first-difference, firm fixed-effect, and lead-lag specifications. Third, our findings hold in two analyses that exploit the shock tofirms' IIQ caused by the Sarbanes-Oxley Act of 2002. Finally, we show that our results are robust to employing a five-yearETR specification and additional control variables.

Our study aims to shed light on the role of an important aspect of the inner workings of a firm in tax avoidanceoutcomes, which is of interest to academics and non-academics alike. We document that IIQ is related to the ability toengage in tax avoidance, and we provide evidence that IIQ plays a stronger role in tax avoidance for firms with greatercoordination needs and uncertainty. Finally, we find support for the hypothesis that a high IIQ allows firms to engage in

1 We employ Cash ETR as our primary proxy for tax avoidance in order to capture the effect of IIQ on the reduction of explicit taxes. Prior research hassuggested that managers value both cash flows and book earnings when making tax planning decisions (Graham et al., 2011). While our main analyses usethe firm's Cash ETR as the empirical proxy for tax avoidance, we document in Section 6 that our findings are robust to employing the firm's book effectivetax rate (GAAP ETR). We list particular tax strategies that the firm is able to carry out because of high IIQ in Appendix A. Ideally, we would like to empiricallydocument the link between IIQ and such tax strategies but this is extremely difficult. Firms generally do not report specific tax avoidance strategies publicly(Lenter et al., 2003).

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 3

greater tax avoidance while incurring less tax risk. This suggests that academics, tax authorities, and practitioners shouldconsider IIQ when studying, identifying, or pursuing tax avoidance strategies and defining tax aggressiveness. Additionally,we respond to Maydew (2001) call to focus on the role of information and uncertainty when researching tax avoidance, aswell as Shackelford and Shevlin (2001)'s appeal to improve the understanding of taxation issues by establishing aconnection with the management accounting discipline's focus on the internal workings of the firm. Finally, we believe thatour results validate the IIQ measures and suggest that these measures can be used in future studies to serve as proxiesfor IIQ.

The remainder of the paper is organized as follows. Section 2 reviews the related literature and develops the hypotheses.Section 3 elaborates on the IIQ proxies. Section 4 discusses research design, the sample, and descriptive statistics. Section 5presents the results of the study. Section 6 reports on additional analyses, endogeneity tests, and robustness checks. Section 7concludes and presents avenues for future research.

2. Hypotheses development

2.1. Quality of the internal information environment

The management accounting literature has long argued that a high IIQ will lead to improved managerial decision-making(Horngren et al., 2012). A high IIQ improves decision making by providing management with real-time information aboutthe financial condition of the company and by eliminating barriers between accounting cycles (Brazel and Dang, 2008).High-IIQ environments exhibit centralized and standardized business transaction processing, short reporting cycle times,and integrated data across business units and geographical locations. A high-IIQ firm has timely access to accurateinformation as well as enhanced internal transparency. Information acquisition (finding the information) and integration(assessing its impact) are greater in high-IIQ environments (Hodge et al., 2004).2

While a high IIQ is likely to have a positive impact on any managerial decision-making, our focus is on the ability of IIQ toenhance the quality of tax-related decisions. We argue that a firm's ability to effectively avoid taxes is improved by a high-IIQ environment that makes information acquisition and integration easier. Information necessary for the tax function istypically dispersed across the firm, available only within systems tailored for financial and management reporting (Cranfordet al., 2012). Documentation and collection of information from disparate systems to support tax compliance can be verytime consuming. Mills et al. (1998) argue that firms have tax planning opportunities with regard to property, plant, andequipment, but that compliance and record-keeping costs to take advantage of these opportunities are high.3 We refer toAppendix A for further examples of tax strategies that are facilitated by high IIQ. Firms with a high-IIQ environment mightbe in a better position to deal more effectively with tax documentation processes and to more easily identify transactionsthat generate tax benefits. Hence, the high-IIQ firm might be able to avoid more taxes, which leads to our first hypothesis:

H1. Higher-quality internal information environments are associated with greater tax avoidance.

2.2. Improved coordination

Substantial information asymmetry can exist between different parties within a firm, with more specific informationavailable to particular business units and lower-level managers than is available to senior management or other businessunits (Bushman et al., 1995). In most firms, the tax department operates at the corporate level, regardless of the dispersion ofthe firm's operations or the number of different tax jurisdictions in which the firm files (Robinson et al., 2010). Firms withhighly dispersed operations not only face important information asymmetry issues but also potentially benefit from taxplanning opportunities. For example, interjurisdictional income-shifting for firms with geographically dispersed operationscan result in significant tax savings (Maydew, 2001), contingent on the availability of supporting compliance documentation.

Existing survey evidence speaks to the perceived usefulness of improved information in the context of disperseoperations or decentralization (although few papers look at whether organizational performance is actually improved, as wedo here; see Ittner and Larcker (2001)). Chenhall and Morris (1986) provide evidence that the usefulness of scope,aggregation, and integration of information is higher when a firm is characterized by disperse operations and organizationalinterdependence. Chapman and Kihn (2009) show that management's confidence in a firm's internal information is afunction of the extent of organizational integration of information across disparate business units. Hence, informationasymmetry issues created by disperse operations can be overcome by improving a firm's IIQ, which can reduce informationasymmetries not only among business units but also between the corporate tax department and the business units. A highIIQ can help identify tax avoidance opportunities in various jurisdictions and can reduce the cost of compliance anddocumentation across business units. Therefore, we expect to see the effect of IIQ on tax avoidance to be larger in firms thatbenefit more from improved coordination, which leads us to the following hypothesis:

2 Of course, managerial ability in information processing and analyzing also plays a role in the quality of decision making (Dyreng et al., 2010). InSection 6, we examine whether our results are driven by CEO or CFO ability or by recent CEO or CFO changes.

3 For example, whether to buy or lease buildings and equipment and the timing of asset acquisitions and disposals requires keeping track ofdepreciation methods and costs for each fixed asset, and using different depreciation methods for regular and AMT tax.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]4

H2. The beneficial effect of high internal information quality on tax avoidance is greater for firms that benefit more fromimproved coordination.

2.3. Reduced uncertainty

Uncertainty within the firm hinders the ability of managers to make decisions. Uncertainty can diminish a firm's taxplanning ability by making tax opportunities less apparent, by introducing substantial doubts about the payoffs of particulartax avoidance strategies, and by increasing the difficulty of forecasting the wide range of potential tax outcomes. Galbraith(1974) indicates that when there is greater uncertainty, more information must be processed by decision makers during taskexecution in order to achieve a given level of performance. One way in which firms can deal with uncertainty is to increasetheir capacity to process information by investing in internal information systems (Galbraith, 1974). Chenhall and Morris(1986) present survey evidence indicating that the perceived usefulness of information is higher in uncertain environments.Surprisingly little tax accounting research has looked at uncertainty (and its resolution by information) as a primary topic ofinterest (Maydew, 2001). A notable exception is Beck et al. (1996) who, in an experiment, find that subjects exposed to moreuncertainty about their tax liability are more likely to purchase the advice of experts who possess information that canresolve this uncertainty. Thus, information could play a more important role in tax avoidance for firms with moreuncertainty. This leads us to the following hypothesis:

H3. The beneficial effect of high internal information quality on tax avoidance is greater for firms that benefit more fromreduced uncertainty.

2.4. Tax risk

In their review paper, Hanlon and Heitzman (2010) discuss the difficulty of conceptually defining tax avoidance or taxaggressiveness and the notion that the degree of aggressiveness is in the eye of the beholder. We argue that informationavailable to the firm during the tax planning decision process is an important factor to consider when assessing theaggressiveness of a firm's tax strategies. To see this, imagine two firms, A and B, with an identical low ETR except that firm Ahas a high IIQ while firm B has a low IIQ. Firm A can achieve its low ETR by relying on high-quality information to support itsdecision making and its documentation for the tax authorities, thus assuming very little risk in the tax positions that ittakes. On the other hand, firm B's ability to convincingly document its tax strategies is negatively affected by its low IIQ. Thelikelihood that Firm B's tax avoidance choices will be disallowed by the tax authorities is high, and thus its tax risk is high.Hence, we argue that Firm B is more tax aggressive than Firm A. Referring back to the model by Mills et al. (2010), bothFrischmann et al. (2008) and Rego and Wilson (2012) define aggressive (or risky) tax reporting as engaging in significant taxpositions with relatively weak supporting facts, although neither Frischmann et al. (2008) nor Rego and Wilson (2012) controlfor the quality of information.

Therefore, we expect that firms with a higher IIQ have greater confidence in their information endowment (Chenhall,2008), better information to support their tax decision-making, and better documentation ability vis-à-vis the taxauthorities. All of these would affect both the level of tax avoidance and the risk assumed in pursuing those tax avoidancestrategies. Furthermore, a high IIQ could allow the firm to find more long-term tax avoidance opportunities with recurringand sustained benefits rather than pursuing risky one-time opportunities (Mills et al., 1998). Hence, we predict that high-IIQfirms will be able to achieve lower ETRs (as hypothesized in H1) while assuming only low levels of tax risk. The abovearguments lead us to the following hypothesis:

H4. Higher-quality internal information environments are associated with lower tax risk.

3. Internal information quality proxies

We employ four proxies for IIQ that are publicly observable: (1) earnings announcement speed (Earnings AnnouncementSpeed), measured as the number of days between the end of the fiscal year and the earnings announcement date, divided by365 and multiplied by negative one; (2) management forecast accuracy (Management Forecast Accuracy), measured as theabsolute value of management's last available estimate of earnings per share before fiscal year-end minus the firm's actualearnings per share, divided by the stock price at the end of the year and multiplied by negative one; (3) the absence ofmaterial weaknesses in controls (No Material Weaknesses), which is an indicator variable equal to zero if the firm reported aSOX Section 404 material weakness in the current fiscal year, and one otherwise; and (4) the absence of restatements causedby unintentional errors (No Error Restatement), which is an indicator variable equal to zero if the firm restated the currentfiscal year due to error, and one otherwise. For each proxy, higher values correspond to a higher IIQ. We discuss each proxybelow (see Appendix B for variable descriptions).

Our first proxy for IIQ is Earnings Announcement Speed. Jennings et al. (2012) argue that a high-quality accounting systemis capable of quickly integrating information from different parts of an organization, increasing the speed with which thebooks are closed. Additionally, increased accuracy caused by eliminating manual intervention, along with decreasedredundancy and rework and streamlined reporting, should also reduce the time between the fiscal period end and the

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 5

earnings announcement. Finally, firms with an automated financial closing process have typically received installationhelp from professionals and have been advised on how to improve their information acquisition processes. Forexample, Textron's 2003 news release about its adoption of Hyperion Business Performance Management softwareexplains that it has been working with Ernst & Young to implement the financial consolidation and reporting module,resulting in a shortening of its financial closing process by four days. Of course, there are strategic reasons for the timingof earnings announcements, such as news content. We include controls for known determinants of the IIQ proxies inSection 6.

Our second measure of IIQ is Management Forecast Accuracy. It has long been established that management hasincentives to forecast accurately (Williams, 1996). However, a high IIQ is a necessary condition for such forecasting ability(Jennings et al., 2012). Empirical evidence indicates that IIQ likely exhibits a positive correlation with management forecastaccuracy.4 Using survey data for small, privately held firms, Cassar and Gibson (2008) find that firms make more accuraterevenue forecasts if they have internal accounting reporting and budgeting processes in place. Dorantes et al. (2013) showthat firms that implemented an improvement to their information systems increased their forecast accuracy. Gong et al.(2009) document that common errors in management forecasts and accruals are caused by inaccuracies in the internalinformation available to managers.

Following Feng et al. (2009), our third proxy for IIQ is the absence of a SOX Section 404 material weakness in internalcontrols. Feng et al. (2009) argue that some material weaknesses result in erroneous internal management reports that areunlikely to be detected because they are not audited. Firms with material weaknesses are likely basing their decisions onstale financial information. Furthermore, business unit information can be untimely and inaccurately reported toheadquarters.5 Employing a similar IIQ argument, Goh and Kim (2013) find that firms with a Section 404 materialweakness have lower operational efficiency. Using hand-collected news releases, Masli et al. (2010) show that implementing(specific) IT solutions for internal control monitoring reduces the likelihood of a material weakness. Morris (2011) indicatesthat firms implementing enterprise resource planning (ERP) systems are less likely to report a material weakness. Finally,disclosures of material weakness typically trigger big investments in accounting systems, suggesting a previous lack of IIQ(Chasan, 2012).

Finally, we use the absence of restatements caused by unintentional errors as classified by the procedure of Henneset al. (2008) to proxy for IIQ.6 Such restatements are mostly the result of basic accounting or data errors (Hennes et al.,2008; Plumlee and Yohn, 2010). Examples include counting and pricing errors that misreport inventory and cost of sales;failure to record credit purchases; and unreliable procedures for rolling up amounts from segments and subsidiaries(Ashbaugh-Skaife et al., 2008). These unintentional errors also affect the information on which management relies fordecision making because they reflect a lack of accurate records and poorly designed and managed information systems(Hayes, 2013).

Using publicly available IIQ proxies overcomes stumbling blocks having to do with data availability, sample size, andgenerality issues, at the cost of construct validity. We address construct validity in multiple ways. First, as we have noted inthis section, extant literature suggests a positive correlation between IIQ and our proxies. Second, all proxies for IIQ arepositively correlated, with Pearson correlations ranging from 0.04 to 0.32 (reported later in Table 1).7 Third, one could arguethat the IIQ measures capture another construct, such as strategic disclosure behavior, managerial ability, or fundamentalvolatility. In Section 6, we show that our results on IIQ hold after controlling for these alternative explanations. Finally, usingthese proxies, we find that IIQ matters most where one might expect in our cross-sectional tests.

4 Predictions from theoretical models indicate a positive correlation between the precision of private information and the likelihood of disclosure (e.g.Verrecchia, 1990). To our knowledge, there are no theoretical models that link the precision of private information to the precision of the managementforecast. Information endowment models (e.g. Jung and Kwon, 1988) typically assume that there is either no private information (infinite imprecision) orperfect private information, and the disclosure decision is a binary one (yes or no disclosure). We are only able to observe the accuracy of the managementforecast for those firms that issue guidance. If managers issue guidance only if the quality of their disclosure is sufficiently high, this decreases variation inManagement Forecast Accuracy and reduces the power of our tests.

5 For example, Corporate Resource Services reports in its 2010 10-K filing that a “material weakness resulted from a lack of sufficient and effectivesupervisory review over the preparation and reconciliation of certain general ledger account balances to their underlying source documents at one of oursubsidiaries” and that as part of the remediation of this weakness they will “develop or address […] detailed financial reporting procedures to ensure thateach subsidiary provides timely, complete, and accurate information to the Company's headquarters.”

6 Restatement classification data are provided by Andy Leone (http://sbaleone.bus.miami.edu/). These data classify a restatement as caused by eitherunintentional error or intentional fraud. We use only those restatements that they identify as caused by errors. Note that “technical” restatements (e.g.,those caused by mergers or discontinued operations) are excluded in the classification by Hennes et al. (2008) since they do not imply an intentionalmanipulation, nor are they a sign of errors in the underlying information at the time of the financial statement release. We combined this dataset withrestatement data from Audit Analytics to identify the fiscal years that are being restated, with a match defined as an Audit Analytics restatement datewithin 7 two days of the Leone restatement date (43 percent of cases). To the extent that we cannot find a match in Audit Analytics, we estimate whichfiscal years are being restated by calculating the average time between the beginning of the restatement period and the end of the restatement period, aswell as between the end of the restatement period and the restatement date, for the matched observations. We then apply these averages to theunmatched observations to get an estimate of the fiscal years affected by the restatement. Inferences are qualitatively unaffected if we instead discard theunmatched observations. Reassuringly, when using restatements caused by intentional fraud, we do not find (or predict) the ETR effect we document withNo Error Restatement.

7 These correlations are consistent with prior research. For example, Feng et al. (2009) document a positive correlation between the absence ofmaterial weaknesses and management forecast accuracy, and Ashbaugh-Skaife et al. (2008) document a positive correlation between the presence ofmaterial weaknesses and restatements related to unintentional errors.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

Table 1Descriptive statistics and correlations.

Panel A: Descriptive statisticsVariables N Mean Std dev. 25th Pctl Median 75th Pctl

Cash ETR 38,223 0.26 0.21 0.10 0.25 0.36Cash ETR Volatility 24,359 0.13 0.10 0.06 0.10 0.16Earnings Announcement Speed 33,246 �0.12 0.05 �0.14 �0.11 �0.08Management Forecast Accuracy 9096 �0.01 0.03 �0.01 0.00 0.00No Material Weaknesses 11,606 0.94 0.23 1.00 1.00 1.00No Error Restatement 28,778 0.94 0.24 1.00 1.00 1.00Geographic Dispersion 34,929 0.23 0.26 0.00 0.09 0.47Restructure 38,223 0.16 0.36 0.00 0.00 0.00Volatility 38,223 �0.07 0.92 �0.69 �0.28 0.34Size 38,223 6.13 1.90 4.77 6.03 7.41PPE 38,223 0.30 0.23 0.11 0.23 0.44ΔPPE 38,223 0.03 0.08 0.00 0.01 0.05Leverage 38,223 0.18 0.17 0.02 0.15 0.29Intangibles 38,223 0.13 0.17 0.00 0.06 0.20R&D Expense 38,223 0.03 0.05 0.00 0.00 0.03NOL Dummy 38,223 0.28 0.45 0.00 0.00 1.00ΔNOL 38,223 0.00 0.05 0.00 0.00 0.00Extraordinary Items 38,223 0.00 0.01 0.00 0.00 0.00Foreign Income 38,223 0.01 0.03 0.00 0.00 0.01FI Dummy 38,223 0.39 0.49 0.00 0.00 1.00ROA 38,223 0.16 0.08 0.10 0.15 0.21MTB 38,223 2.76 2.64 1.35 2.06 3.27Sales Growth 38,223 0.16 0.29 0.02 0.10 0.23Age 38,223 2.65 0.85 2.08 2.64 3.40

Panel B: CorrelationsVariable 1 2 3 4 5 6 7 8 9

1 Cash ETR 0.27 �0.03 �0.15 �0.05 0.00 0.00 �0.03 �0.022 Cash ETR Volatility 0.12 �0.16 �0.16 �0.08 �0.04 0.01 0.04 0.103 Earnings Announcement Speed 0.00 �0.16 0.11 0.24 0.04 0.15 0.04 �0.094 Management Forecast Accuracy �0.05 �0.16 0.14 0.06 0.06 0.04 0.00 �0.185 No Material Weaknesses �0.02 �0.06 0.16 0.10 0.32 �0.02 �0.01 �0.026 No Error Restatement 0.02 �0.03 0.04 0.06 0.32 �0.03 �0.07 �0.017 Geographic Dispersion �0.02 �0.01 0.16 0.08 �0.02 �0.03 0.30 �0.028 Restructure �0.06 0.04 0.04 0.02 �0.01 �0.07 0.30 �0.069 Volatility �0.03 0.09 �0.09 �0.20 �0.02 �0.01 �0.02 �0.07

This table presents descriptive statistics for the variables used in our analyses. Panel A presents descriptive statistics for the sample and Panel B presents the Pearson (above diagonal) and Spearman (belowdiagonal) correlations. The sample in this table is composed of all observations with non-missing data for control variables, Cash ETR, and at least one IIQ proxy. The number of observations tabulated in Panel A isgreater than the number of observations used in subsequent tables since not all observations have non-missing data for each IIQ proxy. All variables are defined in Appendix B. Variables are based on amountsdenominated in millions of U.S. dollars. All continuous variables are winsorized at the 1st and 99th percentiles, except ETRs, which are winsorized at zero and one.

Pleasecite

this

articleas:

Gallem

ore,J.,

Labro,E.,

The

importan

ceof

the

intern

alinform

ationen

vironmen

tfor

taxavoid

ance.Jou

rnal

ofAccou

ntin

gan

dEcon

omics

(2014),http

://dx.d

oi.org/10.1016/j.jacceco.2014.09.005i

J.Gallem

ore,E.Labro/Journal

ofAccounting

andEconom

ics](]]]])

]]]–]]]

6

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 7

4. Research design, sample, and descriptive statistics

4.1. Research design

We employ the following regression specification to test our hypotheses:

Tax Avoidance Proxyi;t ¼ α0þα1IIQ i;tþ∑kαkControls i;k;tþ∑jαjIndustry Fixed Ef f ects i

þ∑lαlYear Fixed Ef f ects tþεi;t ð1Þ

We use Cash ETR as the tax avoidance proxy in our main tests. Cash ETR is defined as cash taxes paid divided by pre-tax income(adjusted for special items).8 When examining tax risk, we use five-year ETR volatility (Cash ETR Volatility) as the tax avoidanceproxy (Guenther et al., 2012). Cash ETR Volatility is measured by the standard deviation of Cash ETR over the five-year period endingin year t. We use the four different proxies for internal information quality with measurements as detailed in Section 3 (EarningsAnnouncement Speed, Management Forecast Accuracy, No Material Weaknesses, and No Error Restatement). Each of these variables isconstructed such that higher values indicate a better IIQ, so we expect a negative coefficient on each IIQ proxy.

We also include a number of control variables that prior research has shown to be associated with tax avoidanceoutcomes (Dyreng et al., 2008; Phillips, 2003; Rego andWilson, 2012): Size; two measures of property, plant, and equipment(PPE and ΔPPE); Leverage; Intangibles; R&D Expense; two measures of net operating losses, NOL Dummy and ΔNOL;Extraordinary Items; two measures of foreign operations, Foreign Income and FI Dummy; return on assets (ROA); market-to-book ratio (MTB); Sales Growth; and Age. To the extent that uncertainty drives both IIQ and tax avoidance, it could representa correlated omitted variable in our analysis. Prior literature has documented correlations between measures of uncertaintyand the IIQ proxies (Feng and Koch, 2010) and between measures of uncertainty and tax avoidance outcomes (Phillips,2003; Rego andWilson, 2012). To capture the general uncertainty regarding the firm's fundamentals, we include Volatility asan additional control variable, constructed as a factor of three variables capturing contemporaneous uncertainty: theabsolute change in sales scaled by lagged total assets; equity volatility over the current year; and the change in equityvolatility from year t�1 to year t. We examine the robustness of our results to additional volatility proxies in Section 6.Finally, we include industry and year fixed effects in order to capture differences in tax avoidance across industries and time.Including industry fixed effects also mitigates endogeneity concerns arising because IIQ and tax avoidance opportunities andabilities likely differ among industries.

4.2. Sample and descriptive statistics

We use financial accounting and segment data from Compustat, stock market data from CRSP, earnings announcementand management forecast data from IBES, Section 404 material weakness data from Audit Analytics, and restatement datafrom Andy Leone. Following Dyreng and Lindsey (2009), we set the following variables to zero if they are missing inCompustat: advertising expense, research and development expense, tax loss carryforwards, intangible assets, special items,and long-term debt. We also employ their method to correct for errors in foreign tax expense, foreign pre-tax income, pre-tax domestic income, total pre-tax income, federal current tax expense, and worldwide current tax expense.

Our sample period runs from 1994 to 2010. We begin our sample period in 1994 with the enactment of SFAS No. 109, tohave consistent accounting for income taxes throughout the sample. For a firm-year observation to enter our sample, it hasto have non-missing data for each control variable and at least one IIQ proxy. Two of the IIQ proxies (Earnings AnnouncementSpeed and Management Forecast Accuracy) are available throughout the entire time period (1994–2010), whereas No ErrorRestatement and No Material Weaknesses are available only from 1994 to 2005 and 2004 to 2010, respectively. The number offirm-years in our annual Cash ETR regressions range from 9,096 (using Management Forecast Accuracy) to 33,246 (usingEarnings Announcement Speed). The sample size is reduced when we investigate Cash ETR Volatility (which uses variablesmeasured over five-year periods). Finally, we exclude all financial firms (SIC codes 6000–6999) from the sample.

Panel A of Table 1 contains descriptive statistics for our sample. The number of observations tabulated in Panel A is greaterthan the number of observations used in subsequent tables because not all observations have non-missing data for each IIQproxy. The average Cash ETR is 26 percent and the five-year standard deviation is 13 percent. Panel B contains correlations for oursample variables. For brevity, we tabulate only the correlations for the tax avoidance proxies, the IIQ proxies, and thecoordination needs and uncertainty variables (we define the proxies for the latter two constructs in Section 5). Cash ETR and CashETR Volatility are generally negatively correlated with the IIQ proxies, consistent with better IIQ allowing for more effective taxavoidance. Lastly, all IIQ proxies are positively correlated, consistent with these variables capturing the same construct.

8 Consistent with prior tax avoidance research (Dyreng et al., 2008), we discard observations if Cash ETR's denominator (pre-tax income adjusted forspecial items) is negative, and winsorize Cash ETR at 0 and 1. We use Cash ETR as our primary measure of tax avoidance in order to avoid any mechanicalrelations between our dependent variable and the IIQ proxies, some of which are associated with accounting earnings (such as Management ForecastAccuracy). Prior research has shown that managers use tax expense and the valuation allowance to manage earnings (Dhaliwal et al., 2004). Cash effectivetax rates do not use tax expense in the numerator and are unaffected by changes in the valuation allowance or tax cushion (Dyreng et al., 2008). However,in robustness checks, we show in Section 6 that our inferences stay the same when we use tax avoidance proxies that have an accrual component (GAAPETR and book-tax difference). Hence, we argue that the firm's IIQ affects tax avoidance for both cash and book purposes.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

Table 2Internal information quality and tax avoidance.

IIQ Proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: Cash ETR Cash ETR Cash ETR Cash ETR

Variables pred Sign (1) (2) (3) (4)

IIQ - �0.233*** �0.843*** �0.054*** �0.015**(0.039) (0.146) (0.011) (0.007)

Size 0.003** �0.003 �0.001 0.001(0.001) (0.002) (0.002) (0.001)

PPE �0.068*** �0.003 �0.019 �0.085***(0.013) (0.024) (0.021) (0.013)

ΔPPE 0.009 �0.072 �0.071* 0.084***(0.020) (0.051) (0.041) (0.021)

Leverage �0.109*** �0.093*** �0.108*** �0.087***(0.014) (0.026) (0.020) (0.015)

Intangibles 0.023* 0.025 0.035* 0.023(0.013) (0.020) (0.018) (0.015)

R&D Expense �0.322*** �0.324*** �0.343*** �0.308***(0.038) (0.073) (0.067) (0.039)

NOL Dummy �0.035*** �0.021*** �0.025*** �0.048***(0.004) (0.006) (0.005) (0.004)

ΔNOL 0.090*** 0.131*** 0.098** 0.077**(0.028) (0.045) (0.038) (0.030)

Extraordinary Items 1.226*** 1.256*** 1.540*** 1.263***(0.153) (0.270) (0.346) (0.146)

Foreign Income �0.321*** �0.113 �0.336*** �0.228***(0.065) (0.091) (0.083) (0.079)

FI Dummy 0.029*** 0.019*** 0.032*** 0.028***(0.004) (0.006) (0.007) (0.005)

ROA 0.019 0.029 0.040 �0.024(0.022) (0.041) (0.035) (0.023)

MTB �0.004*** �0.004*** �0.002* �0.005***(0.001) (0.001) (0.001) (0.001)

Sales Growth �0.085*** �0.078*** �0.110*** �0.074***(0.006) (0.012) (0.011) (0.006)

Volatility 0.003* 0.002 0.012*** �0.005***(0.002) (0.003) (0.003) (0.002)

Age 0.003 0.001 0.008** 0.003(0.002) (0.004) (0.004) (0.002)

Industry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 33,246 9,096 11,606 28,778Adj. R-squared 0.097 0.105 0.104 0.094

This table presents the results of estimating Eq. (1) via OLS with Cash ETR as the dependent variable, where Cash ETR is measured as cash taxes paid dividedby pre-tax income adjusted for special items. Each column employs a different IIQ proxy (Earnings Announcement Speed, Management Forecast Accuracy, NoMaterial Weaknesses, No Error Restatement). All other variables are defined in Appendix B. Coefficients are presented with firm-clustered standard errors inparenthesis. ***, **, and * denote significance at a 1, 5, and 10 percent level for one-tailed tests where there are predictions and two-tailed tests otherwise.

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]8

5. Results

5.1. Internal information quality and tax avoidance

First, we examine whether firms that have a better IIQ are able to achieve more favorable tax avoidance outcomes(Hypothesis 1). To test this, we estimate Eq. (1) via ordinary least squares (OLS) with Cash ETR as the tax avoidance proxy.Table 2 contains the results of this analysis. Each column employs a different IIQ proxy (from left to right: EarningsAnnouncement Speed, Management Forecast Accuracy, No Material Weaknesses, and No Error Restatement) and contains thefull set of control variables as well as industry and year fixed effects (coefficients on fixed effects are not reported). Ingeneral, the coefficients on the control variables are statistically significant and match those found in prior research. Forexample, the coefficients on PPE, ΔPPE, Leverage, R&D Expense, NOL Dummy, and Foreign Income are all negative, consistentwith depreciation, interest expense, research and development, prior losses, and foreign income decreasing Cash ETRs.

Using each of the four proxies, we find a negative relation between IIQ and Cash ETR. The relation is statisticallysignificant, with three p-values significant at the 1 percent level and one at the 5 percent level.9 The effect of IIQ on Cash ETR

9 Statistical inferences are based on standard errors clustered by firm, and are made using one-tailed tests for variables with predictions, andtwo-tailed tests otherwise. Inferences are qualitatively unchanged if standard errors are clustered by both firm and year.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 9

is economically important: a one-standard-deviation increase in Earnings Announcement Speed or Management ForecastAccuracy is associated with a Cash ETR reduction of roughly 1–2 percentage points. The absence of a restatement associatedwith unintentional errors (a Section 404 material weakness) is associated with a Cash ETR reduction of about 2 (5)percentage points. The coefficient on No Material Weaknesses is in line with Bauer (2014) and consistent with a Section 404material weakness being indicative of a firm at the very extreme low end of the IIQ spectrum.10 Overall, these resultsindicate that firms with better-quality information have lower Cash ETRs, consistent with these firms being able to engage inmore effective tax avoidance.

The economic benefit of IIQ for tax avoidance is important. A one-standard-deviation increase in the continuous IIQproxies or the absence of a Section 404 material weakness or restatement due to unintentional error is associated with cashtax savings of between $2 and $10 million (based on mean pre-tax income of $192 million). Whether or not a firm increasesits IIQ to obtain these tax savings depends on the extent to which IIQ improvement is a decision variable and theconsideration of all costs and benefits, including those that go beyond (and may be more important than) tax avoidance.Potential barriers to improving IIQ include a lack of buy-in from lower levels (Major and Hopper, 2005), employeesinterpreting the change as a threat (Argyris and Kaplan, 1994), and a lack of support from departments that are bearing costsbut not reaping benefits (Burns and Scapens, 2000). Furthermore, improvement costs could be prohibitively high. Forexample, even though the costs of SOX section 404 compliance have been decreasing steadily over time, many executivesbelieve the costs of SOX compliance still outweigh the benefits (which presumably include additional tax avoidance benefitsthat could be achieved) (FinancialExecutivesInternational, 2007). Finally, firms might desire to maintain a minimum ETRlevel for reputational reasons and thus refrain from seeking to lower their ETRs by improving IIQ, although recent researchcasts doubt on damage to reputation as a significant cost of tax avoidance (Gallemore et al., 2014). Further research canconsider the extent to which firms can alter their IIQ and the role tax considerations play in this decision.

5.2. Internal information quality, coordination, and tax avoidance

Next, we test whether the beneficial effect of higher IIQ on tax avoidance is greater for firms that benefit more fromimproved coordination (Hypothesis 2). We use Geographic Dispersion as a measure of coordination needs, which wecalculate, similar to Bushman et al. (2004), by summing the squares of the ratio of firm sales in each geographic segment tototal firm sales, subtracting one, and then multiplying by negative one. For example, a firm with ten segments with equalsales would have a dispersion value of 0.9, while the firm with only one segment would have a dispersion value of zero. Totest our hypothesis, we add Geographic Dispersion and an interaction term between Geographic Dispersion and the IIQproxies to Eq. (1). For brevity, we report only the coefficients on the IIQ proxy, Geographic Dispersion, and theinteraction term.

While a firm's ETR can potentially both benefit (because of tax planning opportunities available fromworking in differenttax jurisdictions) as well as suffer (through information asymmetry) from geographic dispersion, the data suggest thatincreased coordination needs make tax avoidance harder. In Table 3, three of the four coefficients on Geographic Dispersionare significantly positive at the 1 percent level. The interaction effect between Geographic Dispersion and IIQ is negative in allfour regressions, and statistically significant in three. These results suggest that the positive effect of IIQ on tax avoidanceincreases as geographic dispersion increases, consistent with IIQ being more important for tax avoidance when coordinationneeds are greater (Hypothesis 2). A high IIQ is able to partially offset the negative effect of coordination needs on taxavoidance.11

5.3. Internal information quality, uncertainty, and tax avoidance

Table 4 examines the interactive effect of IIQ and uncertainty on tax avoidance (Hypothesis 3). We employ two measuresof uncertainty. First, Restructure is an indicator variable equal to one if the firm reports a restructuring expense during theyear, and zero otherwise. During a restructuring, the firm is undergoing a time of increased uncertainty with significantoperational changes (Morton and Neill, 2001). Lin and Yang (2006) similarly find that restructuring results in fundamentalchanges in a firm's organization, strategy, systems, or operations and produces uncertainty in virtually all aspects of theorganization, adding noise to its information environment. Our second measure of uncertainty is High Volatility, an indicatorvariable equal to one when the observation has a value of Volatility above the sample median, and zero otherwise.

Panel A (Panel B) contains the results of this estimation using Restructure (High Volatility) as the uncertainty proxy. Forbrevity, we report only the coefficients on the IIQ proxy, the uncertainty proxy, and the interaction term. Consistent with our

10 Note that Bauer (2014) primarily focuses on tax-related material weaknesses. In his 2004–2009 sample, he finds that firms with a tax-relatedmaterial weakness on average have a Cash ETR that is 4 percentage points higher. We believe that our finding of a higher Cash ETR increase of 5 percentagepoints as a consequence of any internal control material weakness suggests that the quality of the internal information environment as a whole isimportant to pursuing effective tax planning.

11 Results are similar, yet weaker when we use a measure of coordination needs that captures dispersion across the firm's business segments. Thecoefficient on the interaction term with Business Segment Dispersion is negative in three of the four regressions but only statistically significant withEarnings Announcement Speed as IIQ proxy. This is not surprising, since geographically segmented firms should have more tax planning opportunities thanfirms with high business segment dispersion.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

Table 3Internal information quality, coordination, and tax avoidance.

IIQ proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: Cash ETR Cash ETR Cash ETR Cash ETR

Variables Pred Sign (1) (2) (3) (4)

IIQ - �0.078* �0.753*** �0.023* �0.001(0.049) (0.190) (0.016) (0.010)

IIQnGeographic Dispersion - �0.705*** �0.528 �0.077** �0.068**(0.136) (0.692) (0.041) (0.032)

Geographic Dispersion �0.026 0.069*** 0.151*** 0.106***(0.018) (0.017) (0.041) (0.032)

Controls? Yes Yes Yes YesIndustry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 30,432 8,081 10,370 26,497Adj. R-squared 0.098 0.108 0.106 0.092

This table presents the results of estimating an OLS regression with Cash ETR as the dependent variable, where Cash ETR is measured as cash taxes paiddivided by pre-tax income adjusted for special items. Geographic Dispersion is the sum of the squares of (segment sales/total firm sales) minus one thenmultiplied by negative one. Each column employs a different IIQ proxy (Earnings Announcement Speed, Management Forecast Accuracy, No MaterialWeaknesses, No Error Restatement). All other variables are defined in Appendix B. Coefficients are presented with firm-clustered standard errors inparenthesis. ***, **, and * denote significance at a 1, 5, and 10 percent level for one-tailed tests where there are predictions and two-tailed tests otherwise.

Table 4Internal information quality, uncertainty, and tax avoidance.

Panel A: RestructuringIIQ proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: Cash ETR Cash ETR Cash ETR Cash ETR

Variables Pred Sign (1) (2) (3) (4)

IIQ - �0.203*** �0.693*** �0.049*** �0.010*(0.041) (0.162) (0.013) (0.007)

IIQnRestructure - �0.206** �0.531** �0.016 �0.028*(0.089) (0.278) (0.024) (0.019)

Restructure �0.023** �0.003 0.010 0.020(0.010) (0.005) (0.024) (0.019)

Control variables? Yes Yes Yes YesIndustry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 33,246 9,096 11,606 28,778Adj. R-squared 0.100 0.116 0.111 0.097

Panel B: High volatility

IIQ proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: Cash ETR Cash ETR Cash ETR Cash ETR

Variables Pred Sign (1) (2) (3) (4)

IIQ - �0.181*** �0.536*** �0.046*** �0.015**(0.046) (0.180) (0.015) (0.009)

IIQnHigh Volatility - �0.103** �0.463** �0.021 0.002(0.055) (0.225) (0.022) (0.012)

High Volatility �0.008 �0.005 0.032* �0.004(0.007) (0.005) (0.022) (0.012)

Control Variables? Yes Yes Yes YesIndustry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 33,246 9,096 11,606 28,778Adj. R-squared 0.097 0.106 0.104 0.093

This table presents the results of estimating an OLS regression with Cash ETR as the dependent variable, where Cash ETR is measured as cash taxes paiddivided by pre-tax income adjusted for special items. Panel A (B) uses Restructure (High Volatility) as the uncertainty proxy, where Restructure is an indicatorvariable equal to one if the firm reports a restructuring charge that year and zero otherwise, and High Volatility is an indicator equal to one when the factorof three variables capturing contemporaneous uncertainty (absolute change in sales scaled by lagged total assets, equity volatility over the current year, andchange in equity volatility over the prior year) is above the median and zero otherwise. Each column employs a different IIQ proxy (Earnings AnnouncementSpeed, Management Forecast Accuracy, No Material Weaknesses, No Error Restatement). All other variables are defined in Appendix B. Coefficients arepresented with firm-clustered standard errors in parenthesis. ***, **, and * denote significance at a 1, 5, and 10 percent level for one-tailed tests where thereare predictions and two-tailed tests otherwise.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]10

Table 5Internal information quality and tax risk.

IIQ proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: CashETRVol CashETRVol CashETRVol CashETRVol

Variables Pred Sign (1) (2) (3) (4)

IIQ5 - �0.238*** �0.608** �0.057** �0.024***(0.044) (0.318) (0.028) (0.009)

Size5 �0.001 �0.008*** �0.001 �0.007***(0.001) (0.003) (0.002) (0.001)

PPE5 0.040*** 0.042 0.027 0.023*(0.012) (0.031) (0.025) (0.014)

ΔPPE5 �0.077*** �0.065*** �0.068*** �0.072***(0.008) (0.021) (0.020) (0.009)

Leverage5 0.025** 0.028 0.012 0.044***(0.013) (0.033) (0.021) (0.014)

Intangibles5 �0.034*** �0.052** �0.073*** �0.004(0.012) (0.024) (0.018) (0.015)

R&D Expense5 0.082*** 0.024 0.016 0.098***(0.026) (0.058) (0.047) (0.031)

NOL Dummy5 0.001 �0.000 0.000 0.005(0.003) (0.006) (0.005) (0.004)

ΔNOL5 �0.002 0.006 0.003 �0.006(0.016) (0.035) (0.023) (0.020)

Extraordinary Items5 0.966*** 1.462** 1.281** 0.624***(0.210) (0.600) (0.543) (0.235)

Foreign Income5 �0.183*** �0.213* �0.322*** �0.008(0.058) (0.119) (0.094) (0.083)

FI Dummy5 0.014*** 0.010 0.003 0.012**(0.004) (0.008) (0.007) (0.005)

ROA5 �0.372*** �0.401*** �0.373*** �0.440***(0.026) (0.064) (0.044) (0.031)

MTB5 �0.001 �0.000 0.000 �0.001(0.001) (0.002) (0.001) (0.001)

Sales Growth5 �0.004*** �0.004 �0.005*** �0.003***(0.001) (0.003) (0.002) (0.001)

Volatility5 0.065*** 0.051*** 0.072*** 0.056***(0.005) (0.012) (0.010) (0.005)

Age5 0.005** 0.009* 0.002 0.005*(0.002) (0.005) (0.004) (0.003)

Industry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 14,155 2,308 2,882 10,478Adj. R-squared 0.204 0.208 0.214 0.216

This table presents the results of estimating Eq. (1) via OLS with Cash ETR Volatility as the dependent variable, where Cash ETR Volatility is measured as thestandard deviation of Cash ETR over the current five-year period. Each column employs a different IIQ proxy (Earnings Announcement Speed, ManagementForecast Accuracy, No Material Weaknesses, No Error Restatement). IIQ proxies are averaged over the five-year period ending in year t. All other variables aredefined in Appendix B, and are measured over the five-year period ending in year t. Coefficients are presented with firm-clustered standard errors inparenthesis. ***, **, and * denote significance at a 1, 5, and 10 percent level for one-tailed tests where there are predictions and two-tailed tests otherwise.

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 11

prediction that a high IIQ is more beneficial to firms that operate in high-uncertainty environments, the coefficient on theinteraction term in Panels A and B is negative in all but one regression, and significant at the 5 (10) percent level in four(one) regressions. This implies that firms with high uncertainty benefit more from a high IIQ.12

5.4. Internal information quality and tax risk

In Table 5, we investigate whether firms with high-quality information are able to minimize realized tax risk (Hypothesis 4).We use Cash ETR Volatility as our measure of tax risk. Measurement of each control variable is modified to match the five-yearperiod over which we calculate Cash ETR Volatility. For the IIQ proxies, we calculate the average over the five-year period, which

12 The coefficients on the main effects of Restructure in Panel A and High Volatility in Panel B are mostly insignificant. In our regressions, they representthe mean effect of these variables on tax avoidance. However, we expect that uncertainty can negatively affect tax avoidance for low-IIQ firms. To examinethis, we estimate these regressions replacing the continuous versions of Earnings Announcement Speed and Management Forecast Accuracy with rankedversions that range from zero (lowest quintile) to one (highest quintile). Using these ranked IIQ proxies, we find statistically significant positive coefficientson Restructure and High Volatility in both regressions, consistent with uncertainty negatively affecting tax avoidance in low-IIQ environments.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]12

results in sample attrition compared with the analyses in Tables 2–4.13 We find that all four of the IIQ proxies exhibit strongnegative associations with Cash ETR Volatility. A one-standard-deviation increase in Earnings Announcement Speed (ManagementForecast Accuracy) is associated with a reduction in Cash ETR Volatility of approximately 8 (9) percent of the sample mean, and theabsence of a Section 404material weakness (a restatement caused by error) is associated with a reduction of Cash ETR Volatility of45 (19) percent of the sample mean. The results in Table 5 indicate that firms with a better IIQ are able to achieve more favorabletax avoidance outcomes while also decreasing tax risk.

We also examine the association between the IIQ proxies and unrecognized tax benefits (UTB). UTB can be viewed as analternative measure of tax risk that captures the riskiness of a tax position at the time the position is taken. Note that wecannot run our analysis of UTB with the No Error Restatement proxy, because No Error Restatement is available only through2005 and UTB is available only from 2007 onward. Untabulated results show that the coefficients on our information proxiesare negative in two of the three regressions (Earnings Announcement Speed being the exception), and statistically significantwith Management Forecast Accuracy (one-tailed p-value of 0.05).

6. Additional results, endogeneity tests, and robustness

In this section, we employ a five-year ETR specification and examine the robustness of our primary findings to correlatedomitted variables and endogeneity issues, other tax avoidance proxies, and additional control variables. (Appendix Bprovides variable definitions and measurements.)

6.1. Long-run ETR

In our primary tests, we use a one-year version of Cash ETR as our dependent variable. Dyreng et al. (2008) argue thatlong-run measures of tax avoidance are superior to short-run measures in that they avoid variation in annual measuresdriven by timing differences and negative denominators. We investigate whether the results in Table 2 are robust to using along-run measure of tax avoidance. Table 6 contains the results of estimating Eq. (1) with the five-year cash effective tax rate(Cash ETR5) as the dependent variable. As in Table 5, we redefine the IIQ proxies and control variables to match the five-yearperiod over which the dependent variable is measured. The coefficients on the IIQ proxies are consistently negative andstatistically significant. A one-standard-deviation increase in the continuous IIQ proxies is associated with reductions inCash ETR5 of 1–2 percentage points, and the absence of a Section 404 material weakness (a restatement caused by error) inany year of the five-year period is associated with a reduction in the five-year cash ETR of 7 (5) percentage points.The results in Table 6 indicate that our primary results are both statistically and economically robust to using longer-runmeasures of tax avoidance.14

6.2. Correlated omitted variables and endogeneity

Our results are consistent with the notion that IIQ affects the firm's tax avoidance outcomes. However, other variablescould be correlated with both IIQ and tax avoidance opportunities. Furthermore, if both IIQ and tax avoidance are choices onthe part of the firm, our analyses could suffer from endogeneity. In this section, we attempt to rule out correlated omittedvariables and endogeneity as potential concerns.

First, since there are unobservable firm characteristics for which we cannot include controls in our analyses, we estimateour primary regression in first differences (Table 7). Employing a first-differenced specification mitigates the effect of firm-specific characteristics that are relatively constant over time. The coefficient on ΔIIQ in the changes model (with ΔCash ETRas the dependent variable) is negative for all four proxies and is statistically significant when usingΔEarnings AnnouncementSpeed, ΔManagement Forecast Accuracy, and ΔNo Error Restatement (although only weakly with the last). The insignificanceof the coefficient on ΔNo Material Weaknesses (and the weakness of the ΔNo Error Restatement coefficient) is likely causedby the lack of observed changes in these variables: only 7 (3) percent of observations with ΔNo Material Weaknesses (ΔNoError Restatement) have nonzero changes, compared with 91 (99) percent for ΔEarnings Announcement Speed (ΔManage-ment Forecast Accuracy). An alternative approach is to include firm fixed effects in the regression. Untabulated resultsindicate that the coefficients on all four IIQ proxies are negative and statistically significant when including firm fixed effectsinstead of industry fixed effects.15 The results mitigate concerns that the IIQ proxies are capturing a correlated omittedvariable that is constant for a firm across time.

13 Because five years of data are required, observations with non-missing values for Cash ETR Volatility are different from those with a missing value.For example, firms with a non-missing value for Cash ETR Volatility tend to be larger and more profitable. When we reestimate the analysis in Table 2separately within the subsamples of observations with and without a non-missing Cash ETR Volatility, we find that the associations between Cash ETR andthe IIQ proxies are consistently negative across both subsamples.

14 Dyreng et al. (2008) present interesting (but unexplained—see their footnote 21) results on the persistence of ETRs: low ETRs are more persistentthan high ETRs. Our finding of a significant negative association between IIQ and both the level and volatility in Cash ETR suggests that IIQ might partiallyexplain the higher persistence of low ETRs.

15 Reestimating Table 5 (Cash ETR volatility) and Table 6 (Long-run ETR) in first differences or with firm fixed effects, we find results consistent withthe reported results. We find support for our cross-sectional predictions (Tables 3 and 4) when including firm fixed effects (other than for High Volatility),but not when estimating these analyses in first differences. We speculate that the weaker cross-sectional results are caused by the stickiness in both the IIQ

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

Table 6Internal information quality and long-run tax avoidance.

IIQ proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: Cash ETR5 Cash ETR5 Cash ETR5 Cash ETR5

Variables Pred Sign (1) (2) (3) (4)

IIQ5 - �0.279*** �1.124*** �0.068** �0.045***(0.074) (0.412) (0.036) (0.015)

Size5 �0.003 �0.009** 0.001 �0.014***(0.002) (0.004) (0.003) (0.002)

PPE5 �0.037* �0.067 0.026 �0.075***(0.022) (0.052) (0.046) (0.022)

ΔPPE5 �0.084*** �0.085** �0.151*** �0.063***(0.014) (0.041) (0.033) (0.014)

Leverage5 �0.094*** �0.029 �0.087** �0.082***(0.025) (0.070) (0.039) (0.026)

Intangibles5 0.002 �0.034 0.006 0.028(0.020) (0.041) (0.029) (0.026)

R&D Expense5 �0.129*** �0.007 �0.118 �0.185***(0.045) (0.109) (0.078) (0.051)

NOL Dummy5 �0.027*** �0.010 �0.028*** �0.031***(0.005) (0.008) (0.008) (0.007)

ΔNOL5 0.041* 0.057 0.084** 0.072**(0.024) (0.059) (0.037) (0.031)

Extraordinary Items5 2.196*** 2.905*** 3.902*** 1.610***(0.331) (0.680) (0.937) (0.344)

Foreign Income5 �0.044 �0.246 �0.302* 0.178(0.106) (0.174) (0.156) (0.144)

FI Dummy5 0.006 0.014 0.009 0.009(0.006) (0.012) (0.011) (0.008)

ROA5 �0.070* 0.097 �0.036 �0.239***(0.041) (0.086) (0.077) (0.048)

MTB5 �0.004*** �0.004** �0.001 �0.004***(0.001) (0.002) (0.002) (0.001)

Sales Growth5 �0.009*** �0.008* �0.007** �0.008***(0.001) (0.004) (0.003) (0.001)

Volatility5 0.036*** 0.007 0.061*** 0.021***(0.007) (0.017) (0.014) (0.007)

Age5 0.006 0.009 0.007 0.006(0.004) (0.007) (0.007) (0.004)

Industry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 17,741 2,456 3,593 13,980Adj. R-squared 0.127 0.182 0.176 0.138

This table presents the results of estimating Eq. (1) via OLS with Cash ETR5 as the dependent variable, where Cash ETR5 is measured as cash taxes paid overthe current five-year period scaled by pre-tax income adjusted for special items over the same period. Each column employs a different IIQ proxy (EarningsAnnouncement Speed, Management Forecast Accuracy, No Material Weaknesses, No Error Restatement). IIQ proxies are averaged over the five-year periodending in year t. All other variables are defined in Appendix B, and are measured over the five-year period ending in year t. Coefficients are presented withfirm-clustered standard errors in parenthesis. ***, **, and * denote significance at a 1, 5, and 10 percent level for one-tailed tests where there are predictionsand two-tailed tests otherwise.

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 13

Second, we reestimate the analysis in Table 2 replacing the current year's Cash ETR with next year's Cash ETR, examiningwhether the proxies for the current IIQ can help predict tax avoidance in the subsequent year. Showing that this year's IIQhas predictive power for next year's Cash ETR mitigates concerns that the relation documented in Table 2 is simply capturingshocks to the business environment that affect the contemporaneous values for both the IIQ proxies as well as Cash ETR.Untabulated results show a significant negative relation between three of the four IIQ proxies (Management ForecastAccuracy being the exception) and next year's Cash ETR.

Third, we examine a shock to IIQ: the enactment of SOX. SOX required firms to assess the adequacy of their internalcontrols on financial reporting and disclose whether they had a material weakness. First, the act brought negative attentionto firms that had to initially disclose a Section 404 material weakness in 2004, which possibly forced them to focus onimproving their IIQ. We examine whether firms that consequently remedied the material weakness subsequently displaylower Cash ETRs compared with all other firms. We employ a differences-in-differences design: we regress Cash ETR on an

(footnote continued)proxies (the continuous variables exhibit an absolute change that is larger than the standard deviation of Earnings Announcement Speed (ManagementForecast Accuracy) in only 7 (8) percent of observations, and all IIQ proxies exhibit strong positive correlation between once and twice lagged variables) andthe cross-sectional predictors as well as considerable sample attrition when estimating a first-differenced specification.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

Table 7Changes in internal information quality and tax avoidance.

ΔIIQ proxy: Earnings announcement speed Management forecast accuracy No material weaknesses No error restatementDep Var: ΔCash ETR ΔCash ETR ΔCash ETR ΔCash ETR

Variables Pred Sign (1) (2) (3) (4)

ΔIIQ - �0.330*** �0.843*** �0.014 �0.015*(0.072) (0.193) (0.013) (0.009)

ΔSize 0.010 �0.024 0.004 0.011(0.008) (0.019) (0.018) (0.009)

ΔPPE �0.077* 0.130 �0.176* �0.067(0.046) (0.106) (0.092) (0.050)

ΔΔPPE 0.051** 0.028 0.005 0.091***(0.023) (0.072) (0.048) (0.025)

ΔLeverage �0.184*** �0.239*** �0.149*** �0.182***(0.026) (0.052) (0.046) (0.029)

ΔIntangibles �0.050 0.073 �0.054 �0.093**(0.033) (0.070) (0.063) (0.039)

ΔR&D Expense 0.234** �0.047 0.514 0.307**(0.116) (0.292) (0.319) (0.123)

ΔNOL Dummy �0.008 0.009 0.006 �0.019***(0.006) (0.011) (0.009) (0.007)

ΔΔNOL 0.055** 0.071* 0.073** 0.023(0.022) (0.037) (0.036) (0.023)

ΔExtraordinary Items 0.692*** 0.364 0.870*** 0.685***(0.141) (0.277) (0.300) (0.135)

ΔForeign Income �1.124*** �0.488** �1.034*** �1.222***(0.112) (0.198) (0.170) (0.140)

ΔFI Dummy 0.014* 0.008 0.006 0.013(0.008) (0.012) (0.015) (0.009)

ΔROA �0.791*** �0.875*** �0.728*** �0.870***(0.044) (0.098) (0.075) (0.047)

ΔMTB �0.000 �0.000 �0.002 0.002(0.001) (0.001) (0.002) (0.001)

ΔSales Growth �0.031*** �0.021 �0.050*** �0.027***(0.007) (0.018) (0.014) (0.007)

ΔVolatility 0.005*** 0.001 0.007** �0.001(0.002) (0.004) (0.003) (0.002)

ΔAge �0.010 0.005 0.005 -0.015(0.014) (0.040) (0.032) (0.015)

Industry FE? Yes Yes Yes YesYear FE? Yes Yes Yes Yes

Observations 26,606 5,954 8,615 22,533Adj. R-squared 0.063 0.066 0.062 0.061

This table presents the results of estimating a first-differenced version of Eq. (1) via OLS with ΔCash ETR as the dependent variable, where ΔCash ETR ismeasured as the change in the cash ETR in the current year minus the cash ETR in the previous year. Each column employs a change in a different ΔIIQproxy (ΔEarnings Announcement Speed, ΔManagement Forecast Accuracy, ΔNo Material Weaknesses, ΔNo Error Restatement), measured as the change in valuefrom year t�1 to t. All other variables are defined in Appendix B, and are also measured as the change in the variable from the year t�1 to t. Coefficients arepresented with firm-clustered standard errors in parenthesis. ***, **, and * denote significance at a 1, 5, and 10 percent level for one-tailed tests where thereare predictions and two-tailed tests otherwise.

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]14

indicator variable equal to one for firms that initially disclosed a Section 404 material weakness in 2004 but remedied itimmediately (that is, did not disclose a Section 404 material weakness after 2004), an indicator variable equal to one for theyears 2005 and 2006, and the interaction of these two variables. We run this test over the period 2002–2006 (whichcaptures two years before the initial disclosure, the year of initial disclosure, and two years afterward) and include the fullset of control variables and fixed effects. We find that the coefficient on the interaction term is negative and statisticallysignificant (–0.057, one-tailed p-value of 0.012). This result implies that firms that improved their IIQ after initially reportinga Section 404 material weakness upon the enactment of SOX experienced a decrease in Cash ETR relative to other firms,consistent with the improvement in IIQ enabling the firm to avoid more taxes.

Second, firms with public float (the portion of equity not held by management or large shareholders) under $75 millionin 2004 could delay the implementation of Section 404(a) until 2007, while those with public float equal to $75 million orgreater had to implement immediately. Thus, following Iliev (2010), we investigate whether otherwise similar firms justabove the $75 million public float threshold exhibit greater tax avoidance after the implementation of SOX compared withthose firms just below the threshold. We implement a regression discontinuity design to test this hypothesis: we regressCash ETR on an indicator variable equal to one if the firm has a public float equal to or greater than $75 million, an indicatorvariable equal to one if the fiscal year is 2004 through 2006, and the interaction of these two variables. We estimate thisregression for firm-years between 2002 and 2006, and include the full set of control variables and fixed effects. Similar to

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 15

Iliev (2010), we restrict the sample to firms with public floats between $50 million and $100 million. We drop firms thatvoluntarily adopted early and firms that failed to adopt when they should have. We find that the coefficient on theinteraction term is negative and statistically significant (�0.097, one-tailed p-value of 0.018). Again, this result suggests thatthat the improvements in firms' IIQ imposed by the SOX shock allowed them to experience an economically meaningfulreduction in their Cash ETR.

We believe that the consistency of our results when employing first-differences and lead-lag specifications, includingfirm fixed effects, and showing the effect of the shock to firms' IIQ caused by SOX on tax avoidance, suggest that correlatedomitted variables and endogeneity are not responsible for our results.

6.3. Alternate tax avoidance proxies

As discussed earlier, we employ Cash ETR as our primary measure of tax avoidance in order to avoid any mechanical relationsbetween our tax avoidance proxy and IIQ variables, some of which involve accounting earnings. Nevertheless, we investigate thesensitivity of our results to two alternative proxies for tax avoidance that have an accounting accrual aspect: the firm's bookeffective tax rate (GAAP ETR) and book-tax difference (BTD). Whenwe regress annual levels, five-year volatility, five-year levels, andannual changes of these alternate tax avoidance proxies on the five IIQ proxies, untabulated results show that 88 and 56 percent ofthe coefficients are statistically significant with the predicted sign on GAAP ETR and BTD, respectively. We also investigate the effectsof IIQ on firms with higher coordination needs and more uncertainty employing GAAP ETR and BTD as the tax planning proxies.When using GAAP ETR, we find similar results to those in Tables 3 and 4, with stronger results on Restructure (Table 4, Panel A) butweaker results on High Volatility (Table 4, Panel B) (results not tabulated). Results with BTD are weaker in Table 3 but similar inTable 4. We conclude that our inferences are robust to these alternate measures of tax avoidance, and that higher IIQ allowsmanagers to avoid more taxes for both cash and book purposes.

6.4. Alternate control variables

Although we already use an extensive set of controls, we also investigate the sensitivity of our results to severaladditional control variables. First, the IIQ measures could be correlated with the volatility of the firm's fundamentals (forexample, income stream stability), which likely affects tax avoidance outcomes. We include variables that capture thevolatility of the firm's fundamentals: Cash Flow Volatility, Stock Return Volatility, Sales Volatility, and Sales Growth Volatility.We also include Pre-Tax Income Volatility, which captures pre-tax earnings smoothness. All our results continue to hold,although the results of the High Volatility test weaken when controlling for the five other volatility measures.

Second, the IIQ proxies could be capturing managerial ability. Both IIQ and tax avoidance are under the purview of theCEO and the CFO. More capable CEOs or CFOs might manage the firm's IIQ more effectively, and Dyreng et al. (2010) showthat managers are an important determinant of tax avoidance. We include the natural logarithm of the CEO's (CFO's) totalcompensation to capture managerial ability. Although including these measures results in significant sample attrition(between 29 percent and 67 percent depending on the executive type and IIQ proxy), our inferences are similar in the levelsanalysis, with the sole exception being the uncertainty result, which weakens substantially (untabulated). Furthermore, ourchanges results (Table 7) are qualitatively unaffected when we include indicator variables that capture whether the CEO(CFO) changed in the current or prior year, or first-differenced versions of total compensation. Our tests suggest that the IIQproxies are not simply capturing managerial ability.16

Some of the IIQ proxies might be correlated with the quality of the firm's external information environment and thefirm's strategic disclosure choices, which in turn could be associated with tax avoidance. For example, firms that have highexternal transparency might be more comfortable adopting complex tax avoidance strategies.17 Furthermore, prior researchsuggests that the firm's disclosure strategy is correlated with its tax avoidance strategy (Frank et al., 2009). To control foraspects of the external information environment and strategic disclosure determinants that might be correlated with boththe IIQ proxies and tax avoidance, we include Analyst Forecast Dispersion, Bid-Ask Spread, Discretionary Accruals, EarningsSurprise, and Good News. Even though the inclusion of these variables leads to sample attrition ranging from 15 to 35 percentdepending on the IIQ proxy used, we find qualitatively similar results (the sole exception is the restructuring test result,which is significantly weaker).

Finally, there are other known determinants of the IIQ proxies that might also be correlated with tax avoidance. Severalof the controls already in our ETR regressions or added in this robustness section are known determinants of Earnings

16 Prior research documents a relation between corporate governance and tax avoidance (Desai et al., 2007) and shows that the relation between taxavoidance and firm value depends on corporate governance (Desai and Dharmapala, 2009). It could also be the case that better-governed firms have ahigher IIQ. Following Desai and Dharmapala (2009), we include the level of institutional ownership as a measure of corporate governance in our analyses.Our inferences are robust to the inclusion of institutional ownership in the regression analyses (results not tabulated). The primary exception is a weakeneduncertainty result.

17 Note, however, that prior research suggests the opposite sign on the relation between firm external transparency and tax avoidance: firms with hightransparency are less likely to be significant tax avoiders. For example, Frank et al. (2009) find that tax aggressive firms are also aggressive for financialreporting purposes. Desai and Dharmapala (2009) find that aggressive tax avoiders are more opaque so that managers can extract rents. Balakrishnan et al.(2011) find that aggressive tax planning increases the opacity of external information, due to the complexity of activities specifically designed to avoidtaxes. Kim et al. (2011) find that tax avoidance masks bad news hoarding activities.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]16

Announcement Speed and/or Management Forecast Accuracy.18 We further add Forecast Horizon (Baginski and Hassell, 1997;Chen and Matsumoto, 2006), Big 5 Auditor (Feng et al., 2009), predictability of earnings (Earnings Per Share Change) (Brazeland Dang, 2008; Feng et al., 2009), Unqualified Audit Opinion Indicator, and Time Trend (Brazel and Dang, 2008). Our resultsare robust to the inclusion of these variables.

7. Conclusions and further research

Our paper takes an important step toward a better understanding of tax avoidance by studying the role of the internalinformation environment in supporting more favorable tax outcomes. We provide evidence that a higher IIQ is associatedwith greater tax avoidance, and find that this effect is stronger for firms with greater coordination needs because ofgeographic dispersion and for firms with greater uncertainty. Finally, we find that a higher IIQ is associated with lower taxrisk. Understanding the quality of the information on which tax planning decisions are based allows us to disentangle therisk aspect from the tax avoidance measure, a necessary condition to understand and define what it means to exhibit taxaggressiveness.

We see many avenues for further research. First, using these proxies for internal information quality, subsequent studiescan further disentangle which variables affect tax risk and which variables affect tax avoidance conditional on such risk.Also, the market's perception of tax avoidance might be different for firms with high versus low IIQ. Second, recent researchhas started to look at how internal performance measurement and incentives affect tax outcomes (e.g. Armstrong et al.,2012; Phillips, 2003; Robinson et al., 2010). Future research can address how other aspects of managerial accounting affecttax strategies. For example, it is feasible that the quality of the budgeting process, the match between organizational andinformational design, and the precision with which performance measurement information is collected all affect the qualityof tax planning decisions. Finally, although our focus in this paper is on tax avoidance, IIQ affects many other aspects ofdecision making in a firm, such as procurement, capital investment decisions, production and project planning, product orservice mix choices, and customer mix selection. New insights can be generated into the effect of IIQ on the quality of thesedecisions and their outcomes.

Appendix A. Examples of tax strategies affected by IIQ

al.,chaincl

Pta

Transfer Pricing: By lowering the price of goods and services sold by parents and affiliates in high-tax jurisdictions andraising the price of purchases, income can be shifted to low-tax jurisdictions. Consistent documentation for transferpricing must be coordinated across jurisdictions because the type and detail of the information that the taxpayer shouldmaintain and submit ranges from very exhaustive in some countries to higher-level in others (OECD, 2013).

Research and development tax credits: In order to document and sustain an R&D tax credit position, a firm has to beable to (1) trace increasing R&D costs to particular “business components” (i.e., R&D projects) and (2) show that thenature and purpose of the costs meet the requirements of “qualified research expenditures.” Hence, it is not sufficient toknow overall firm R&D or increases in it.

Allocation of debt and earnings stripping: Companies can minimize taxes by borrowing more in high-tax regimes andless in low-tax regimes (without altering overall leverage). Firms must have coordinated information to sustain thesepositions across different tax jurisdictions, both internationally and in the U.S. (between the different states) and be ableto document that the true debt risk is assumed in the low-tax regime (Gravelle, 2013).

Allocation of intellectual property (IP): To document the allocation of IP (without altering the aggregate level of IP) inlow-tax regimes (both internationally as well as within the U.S.), information must be coordinated across regimes, andeconomic substance has to be documented in the chosen tax regime.

Other examples: advanced pricing agreements (APAs), cross-crediting and sourcing of foreign tax credits, contractmanufacturing, use of Check-the-Box, captive insurance companies.

Appendix B

See Table B1.

18 As Management Forecast Accuracy determinants, we already include Good News (Baginski and Hassell, 1997), Size (Baginski and Hassell, 1997; Feng et2009), Age, Volatility, organizational change (Sales Growth, Asset Growth, Leverage), complexity (Restructuring and Geographic Dispersion), financialllenges (ROA, Special Items, R&D Expense), and Analyst Forecast Dispersion (Feng et al., 2009). As Earnings Announcement Speed determinants, we alreadyude Size, Discretionary Accruals, Earnings Surprise, and Good News (Brazel and Dang, 2008).

lease cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment forx avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

Table B1Variable definitions.

Dependent variablesCash ETR Cash taxes paid, divided by pre-tax income adjusted for special itemsCash ETR Volatility Standard deviation of Cash ETR over five-year period ending in year t

Internal information quality (IIQ) variablesEarnings AnnouncementSpeed

Number of days between the end of the fiscal year and the firm's earnings announcement, divided by 365 and multiplied bynegative one

Management ForecastAccuracy

Absolute value of (management's last available estimate of EPS before year-end minus actual EPS) multiplied by negativeone, divided by year-end price

No Material Weaknesses Indicator variable equal to zero if firm reported a Section 404 material weakness in current fiscal year; one otherwiseNo Error Restatement Indicator variable equal to zero if the firm restated the current fiscal year due to unintentional error according to Hennes

et al. (2008); one otherwise

Mechanism and control variablesGeographic Dispersion Sum of the squares of (firm sales in each geographic segment / total firm sales) minus one, then multiplied by negative oneBusiness Dispersion Sum of the squares of (firm sales in each business segment / total firm sales) minus one, then multiplied by negative oneRestructure Indicator variable equal to one if the firm reported a restructuring expense in the current fiscal year; zero otherwiseVolatility Factor of three variables: absolute change in sales scaled by lagged total assets, equity volatility over the current year, and

change in equity volatility from year t-1 to year tHigh Volatility Indicator variable equal to one if Volatility is above the sample median; zero otherwiseSize Natural log of average total assetsPPE Average property, plant, and equipment, divided by average total assetsΔPPE PPE in year t minus PPE in year t-1, divided by average total assetsLeverage Average long-term debt, scaled by average total assetsIntangibles Average intangible assets, scaled by average total assetsR&D Expense Research & development expense, divided by average total assetsNOL Dummy Indicator variable equal to one if net operating loss carryforward at beginning of the year is positive; zero otherwiseΔNOL Current net operating loss carryforward minus last year's net operating loss carryforward, divided by average total assetsExtraordinary Items Extraordinary items, divided by average total assetsForeign Income Pre-tax foreign income, divided by average total assetsFI Dummy Indicator variable equal to one if pre-tax foreign income is nonzero; zero otherwiseROA Operating income before depreciation, divided by average total assetsMTB Average market value of equity, divided by average common equitySales Growth Sales in year t minus sales in year t-1, divided by sales in year t-1Age Log of number of years since first year on Compustat database

Alternate dependent variablesGAAP ETR Current tax expense divided by pre-tax incomeBTD Pre-tax book income minus estimated taxable income, scaled by average total assetsUTB Average unrecognized tax benefit, divided by average total assets

Additional control variablesCash Flow Volatility Standard deviation of operating cash flows over the five-year period ending in year t, scaled by median total assets over the

same periodStock Return Volatility Standard deviation of monthly stock returns over five-year period ending in year tSales Volatility Standard deviation of sales over five-year period ending in year t, scaled by median total assets over the same periodSales Growth Volatility Standard deviation of sales growth over five-year period ending in year tPre-Tax Income Volatility Standard deviation of pre-tax income over five-year period ending in year t, scaled by median total assets over the same

periodInstitutional Ownership Number of shares held by institutional investors scaled by total shares outstandingDiscretionary Accruals Performance-matched discretionary accruals as measured in Frank et al. (2009)Earnings Surprise Actual earnings per share minus the median analyst earnings forecast prior to the fiscal year endGood News Indicator variable equal to one if Earnings Surprise is positive; zero otherwiseAsset Growth Assets in year t minus sales in year t-1, divided by assets in year t-1Special Items Special items scaled by average total assetsAnalyst ForecastDispersion

Standard deviation of all outstanding analyst earnings forecast before the fiscal period end, scaled by the absolute value ofthe median analyst forecast

Bid-Ask Spread The average monthly bid-ask spread over the last twelve monthsForecast Horizon Days between management forecast date and the end of the forecasted periodBig 5 Auditor Indicator variable equal to one if firm uses a Big 5 auditor in year t; zero otherwiseEarnings Per ShareChange

The change in earnings per share from year t-1 to 1, scaled by earnings per share in year t-1

Unqualified AuditOpinion

Indicator variable equal to one if firm received an unqualified audit opinion in year t; zero otherwise

Time Trend Current fiscal year minus 1994Price Year-end stock priceInventory Average inventory scaled by average total assetsAdvertising Expense Advertising expense scaled by average total assets

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 17

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]]18

References

Argyris, C., Kaplan, R.S., 1994. Implementing new knowledge: the case of activity-based costing. Accounting Horizons 8 (3), 83–105.Armstrong, C.S., Blouin, J.L., Larcker, D.F., 2012. The incentives for tax planning. Journal of Accounting and Economics 53, 391–411.Ashbaugh-Skaife, H., Collins, D.W., Kinney, W.R., LaFond, R., 2008. The effect of SOX internal control deficiencies and their remediation on accrual quality.

The Accounting Review 83 (1), 217–250.Baginski, S.P., Hassell, J.M., 1997. Determinants of management forecast precision. The Accounting Review 72 (2), 303–312.Balakrishnan, K., Blouin, J.L., Guay, W., 2011. Does tax aggressiveness reduce financial reporting transparency? Working Paper.Bauer, A., 2014. Tax avoidance and the implications of weak internal controls. Working Paper.Beck, P.J., Davis, J.S., Jung, W.-O., 1996. Tax advice and reporting under uncertainty: theory and experimental evidence. Contemporary Accounting Research

13 (1), 49–80.Brazel, J.F., Dang, L., 2008. The effect of ERP system implementations on the management of earnings and earnings release dates. Journal of Information

Systems 22 (2), 1–21.Burns, J., Scapens, R.W., 2000. Conceptualizing management accounting change: an institutional framework. Management Accounting Research 11, 3–25.Bushman, R.M., Chen, Q., Engel, E., Smith, A., 2004. Financial accounting information, organizational complexity and corporate governance systems. Journal

of Accounting and Economics 37, 167–201.Bushman, R.M., Indjejikian, R.J., Smith, A., 1995. Aggregate performance measures in business unit manager compensation: the role of intrafirm

interdependencies. Journal of Accounting Research 33, 101–128.Cassar, G., Gibson, B., 2008. Budgets, internal reports, and manager forecast accuracay. Contemporary Accounting Research 25 (3), 707–737.Chapman, C.S., Kihn, L.-A., 2009. Information system integration, enabling control and performance. Accounting, Organizations and Society 34, 151–169.Chasan, E., 2012. CFO Journal: The Financial-Data Dilemma––Unify Systems all at Once Or Wait Until a Purchase Or Restructuring Requires it? Wall Street

Journal, Jul 24, 2012, Eastern edition.Chen, S., Matsumoto, D., 2006. Favorable versus unfavorable recommendations: the impact on analyst access to management-provided information. Journal

of Accounting Research 44 (4), 657–689.Chenhall, R.H., 2008. Accounting for the horizontal organization. In: Chapman, C.S., Hopwood, A., Shields, M.D. (Eds.), Handbook of Management

Accounting Research, vol. 3. , Elsevier, pp. 1207–1234.Chenhall, R.H., Morris, D., 1986. The impact of structure, environment, and interdependence on the perceived usefulness of management accounting

systems. The Accounting Review 61 (1), 16–35.Cranford, S., Murias, F., Schutzman, M., 2012. Elevating tax department productivity with an “efficiency dividend”. Corporate Tax Management.Desai, M.A., Dharmapala, D., 2009. Corporate tax avoidance and firm value. The Review of Economics and Statistics 91 (3), 537–546.Desai, M.A., Dyck, A., Zingales, L., 2007. Theft and taxes. Journal of Financial Economics 84, 591–623.Dhaliwal, D.S., Gleason, C.A., Mills, L.F., 2004. Last-chance earnings management: using the tax expense to meet analysts' forecasts. Contemporary

Accounting Research 21 (2), 431–459.Dorantes, C.-A., Li, C., Peters, G.F., Richardson, V.J., 2013. The effect of enterprise systems implementations on the firm information environment.

Contemporary Accounting Research 30 (4), 1427–1461.Dyreng, S.D., Hanlon, M., Maydew, E.L., 2008. Long-run corporate tax avoidance. The Accounting Review 83 (1), 61–82.Dyreng, S.D., Hanlon, M., Maydew, E.L., 2010. The effects of executives on corporate tax avoidance. The Accounting Review 85, 1163–1189.Dyreng, S.D., Lindsey, B.P., 2009. Using financial accounting data to examine the effect of foreign operations located in Tax Havens and other countries on U.S.

multinational firms' tax rates. Journal of Accounting Research 47 (5), 1283–1316.Ernst, Young, , 2010. Global Transfer Pricing Survey: Addressing the Challenges of Globalization. [Whitepaper].Feng, M., Koch, A.S., 2010. Once bitten, twice shy: the relation between outcomes of earnings guidance and management guidance strategy. The Accounting

Review 85 (6), 1951–1984.Feng, M., Li, C., McVay, S., 2009. Internal control and management guidance. Journal of Accounting and Economics 48, 190–209.FinancialExecutivesInternational, 2007. FEI Survey: Management Drives Sarbanes-Oxley Compliance Costs Down.Frank, M.M., Lynch, L.J., Rego, S.O., 2009. Tax reporting aggressiveness and its relation to aggressive financial reporting. The Accounting Review 84 (2),

467–496.Frischmann, P.J., Shevlin, T., Wilson, R., 2008. Economic consequences of increasing the conformity in accounting for uncertain tax benefits. Journal of

Accounting and Economics 46, 261–278.Galbraith, J.R., 1974. Organization design: an information processing view. Interfaces 4 (3), 28–36.Gallemore, J., Maydew, E.L., Thornock, J.R., 2014, The reputational costs of tax avoidance. Contemporary Accounting Research, http://dx.doi.org/10.1111/

1911-3846.12055.Goh, B.W., Kim, J.B., 2013. Internal control and operational efficiency. Working Paper.Gong, G., Li, L.Y., Xie, H., 2009. The association between management earnings forecast errors and accruals. The Accounting Review 84 (2), 497–530.Graham, J., Hanlon, M., Shevlin, T., 2011. Real effects of accounting rules: evidence from multinational firms' investment location and profit repatriation

decisions. Journal of Accounting Research 49 (1), 137–185.Gravelle, J.G., 2013. Tax Havens: International Tax Avoidance and Evasion. Congressional Research Service Report.Guenther, D.A., Matsunaga, S.R., Williams, B.M., 2012. Corporate tax aggressiveness and firm risk. Working Paper.Hanlon, M., Heitzman, S., 2010. A review of tax research. Journal of Accounting and Economics 50, 127–178.Hayes, L., 2013. Identifying unintentional error in restatement disclosures. Working Paper.Hennes, K., Leone, A., Miller, B., 2008. The importance of distinguishing errors from irregularities in restatement research: the case of restatements and

CEO/CFO turnover. The Accounting Review 83 (6), 1487–1519.Hodge, F.D., Kennedy, J.J., Mains, L.A., 2004. Does search-facilitating technology improve the transparency of financial reporting? The Accounting Review 79

(3), 687–703.Horngren, C.T., Datar, S.M., Foster, G., Rajan, M., Ittner, C.D., 2012. Cost Accounting: a Managerial Emphasis. Prentice Hall, Pearson.Iliev, P., 2010. The effect of SOX section 404: costs, earnings quality, and stock prices. Journal of Finance 65 (3), 1163–1196.Ittner, C.D., Larcker, D.F., 2001. Assessing empirical research in managerial accounting: a value-based management perspective. Journal of Accounting and

Economics 32, 349–410.Jennings, J., Stoumbos, R., Tanlu, L., 2012. The effect of organizational complexity on earnings forecasting behavior. Working Paper.Jung, W.-O., Kwon, Y.K., 1988. Disclosure when the market is unsure of information endowment of managers. Journal of Accounting Research 26 (1),

146–153.Kim, J.-B., Li, Y., Zhang, L., 2011. Corporate tax avoidance and stock price crash risk: firm-level analysis. Journal of Financial Economics 100, 639–662.Lenter, D., Shackelford, D.A., Slemrod, J., 2003. Public disclosure of corporate tax return information: accounting, economics, and legal perspectives.

National Tax Journal 56 (4), 803–830.Lin, B., Yang, R., 2006. The effect of repeat restructuring charges on analysts' forecast revisions and accuracy. Review of Quantitative Finance and Accounting

27, 267–283.Major, M., Hopper, T., 2005. Managers divided: implementing ABC in a Portuguese telecommunications company. Management Accounting Research 16,

205–229.Masli, A., Peters, G.F., Richardson, V.J., Sanchez, J.M., 2010. Examining the potential benefits of internal control monitoring technology. The Accounting

Review 85 (3), 1001–1034.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i

J. Gallemore, E. Labro / Journal of Accounting and Economics ] (]]]]) ]]]–]]] 19

Maydew, E.L., 2001. Empirical tax research in accounting: a discussion. Journal of Accounting and Economics 31, 389–403.McKinnon, John D., 2012. Firms pass up tax breaks, citing hassles, complexity. Wall Street Journal, Eastern edition.Mills, L.F., Erickson, M.M., Maydew, E.L., 1998. Investments in tax planning. Journal of the American Taxation Association 20 (1), 1–20.Mills, L.F., Robinson, L.A., Sansing, R.C., 2010. FIN 48 and tax compliance. The Accounting Review 85 (5), 1721–1742.Morris, J.J., 2011. The impact of enterprise resource planning (ERP) systems on the effectiveness of internal controls over financial reporting. Journal of

Information Systems 25 (1), 129–157.Morton, R.M., Neill, J.D., 2001. The value-relevance of current and forward-looking accounting information subsequent to a corporate restructuring.

Advances in Accounting 18, 195–220.OECD, 2013. White Paper on Transfer Pricing Documentation. Public Consultation (July 30).Phillips, J.D., 2003. Corporate tax-planning effectiveness: the role of compensation-based incentives. The Accounting Review 78 (3), 847–874.Plumlee, M., Yohn, T., 2010. An analysis of the underlying causes attributed to restatements. Accounting Horizons 24 (1), 41–64.Rego, S., Wilson, R., 2012. Equity risk incentives and corporate tax aggressiveness. Journal of Accounting Research 50 (3), 775–810.Robinson, J.R., Sikes, S.A., Weaver, C.D., 2010. Performance measurement of corporate tax departments. The Accounting Review 85 (3), 1035–1064.Shackelford, D.A., Shevlin, T., 2001. Empirical tax research in accounting. Journal of Accounting and Economics 31, 321–387.Verrecchia, R.E., 1990. Information quality and discretionary disclosure. Journal of Accounting and Economics 12, 365–380.Williams, P., 1996. The relation between prior earnings forecast by management and analyst response to a current management forecast. The Accounting

Review 71, 103–116.

Please cite this article as: Gallemore, J., Labro, E., The importance of the internal information environment fortax avoidance. Journal of Accounting and Economics (2014), http://dx.doi.org/10.1016/j.jacceco.2014.09.005i