on litigation risk and disclosure complexity: evidence from canadian firms cross-listed in the us

Available online at www.sciencedirect.com

ScienceDirect

ACCOUN-00694; No of Pages 29

On Litigation Risk and DisclosureComplexity: Evidence from Canadian

Firms Cross-Listed in the US☆

Stuart L. Gillana,1, Christine A. Panasianb,⁎

a Department of Finance, Terry College of Business, 452 Brooks Hall, University of Georgia; Athens, GA 30602,United States

b Department of Finance, Sobey School of Business, Saint Mary's University, Halifax, NS B3H 3C3, Canada

Received 7 June 2013

Abstract

We find that D&O insurance premiums for Canadian firms cross-listed in the US are more than twicethose of Canadian-only listed firms, and audit fees are approximately 50% higher.While this supports theview that both service-providers view the US as a more litigious environment, our findings also suggestthat these differentials for cross-listed firms reflect premia for both litigation risk and the complexity offirms' financial disclosures. In particular, we show that D&O liability insurers charge differently forcross-listed firms that have different levels of disclosure; while D&O premiums are significantly higherfor all cross-listed firms, they increase and then decrease with increased disclosure complexity. Incontrast, audit fees increase monotonically as the filing complexity increases. We also find that auditorsappear to price abnormal premiums charged by D&O insurers. Thus, audit fee premia for cross-listedfirms appear to capture aspects of both litigation risk and increased disclosure complexity.© 2014 University of Illinois. All rights reserved.

JEL classification: G15; G32; G34; G38; M423Keywords: Litigation; Disclosure; Audit fees; D&O liability insurance; Cross-listed

☆ Wewould like to thank Scott Bauguess, Jeff Coles, Marshall Geiger, Patrick Kelly, Darius Miller, Mike Stegemoller,Mike Vetsuypens, and seminar participants at Baylor University, Cal-State Fullerton, Oklahoma State, SouthernMethodist University, Texas Tech University, University of Alberta, University of Auckland, University of Canterbury,University of Otago, University of Richmond, and University of South Florida for helpful comments on earlier versions.We are especially grateful to RashadAbdel-Khalik (the Editor) and the anonymous referees for extensive and constructivesuggestions.⁎ Corresponding author. Tel.: +1 902 420 5726.

E-mail addresses: [email protected] (S.L. Gillan), [email protected] (C.A. Panasian).1 Tel.: +1 706 542 9501.

http://dx.doi.org/10.1016/j.intacc.2014.10.0030020-7063/© 2014 University of Illinois. All rights reserved.

Please cite this article as: Gillan, S.L., & Panasian, C.A., On Litigation Risk and Disclosure Complexity: Evidencefrom Canadian..., The International Journal of Accounting (2014), http://dx.doi.org/10.1016/j.intacc.2014.10.003

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.intacc.2014.10.003


2 S.L. Gillan, C.A. Panasian / The International Journal of Accounting xx (2014) xxx–xxx

1. Introduction

This study contributes to several areas of empirical accounting and finance literaturestudying aspects of litigation risk, disclosure complexity, and how such features arepriced by Directors' and Officers' (D&O) liability insurers and auditors. Of note, priorwork suggests the potential for differential audit effort, and possibly differentialexposure to litigation risk conditioned on cross-listing status and the firms' disclosurechoices.

Our findings suggest that audit fees and D&O premiums for cross-listed firms varywith firms' disclosures, indicating that both litigation risk and increased disclosurecomplexity drive the cross-listing premium in audit fees. We also confirm prior findingsthat audit fees and D&O insurance are larger for cross-listed firms, but find that thedifferences are much larger than those reported for earlier time periods (for our sample,audit fees are almost 50% higher and D&O insurance premiums more than double those ofnon-cross-listed firms).

Our focus on Canadian firms provides a unique experimental setting for studying theseissues as information pertaining to both D&O insurance purchases (premiums, coverage, anddeductibles) and audit fees is available. Moreover, during our sample period, under anagreement between United States (US) and Canadian regulators, the Multi-JurisdictionalDisclosure System (MJDS)-eligible Canadian firms are allowed to use their home countrydocuments when cross-listing and thus satisfy the US disclosure requirements without a reviewby the Securities and Exchange Commission (SEC). As a result, the firms we examine operatein different regulatory regimes (Canada or the US) and, as we discuss in further detail below,they also have different levels of disclosure complexity.

This, in turn, suggests variation in litigation risk based on these differences, andconsequently variation in the fees charged by auditors and D&O insurers. Our results aresupportive of those reported by Callaghan, Parkash, and Singhal (2008) in that we find thatCanadian cross-listed firms using the MJDS have lower audit fees than those not using theMJDS. However, we also find that audit fees and D&O premiums are generally higher for allcross-listed firms, and that the specific type of disclosure is important in audit fee pricing. Thesefindings also suggest that caution is warrantied when generalizing from studies of cross-listedfirms to the US environment. Thus, more generally our work contributes to the literatures onaudit fee pricing, disclosure complexity, and how these and D&O insurance relate to corporatelitigation risk and suggest caution when generalizing from.

The paper proceeds as follows. Section 1 provides background details and reviews theliterature on audit fees, D&O insurance, and cross-listing. Section 2 describes our data andsample selection, Section 3 presents univariate analyses, while Section 4 describes themultivariate approach and discusses the results. Section 5 concludes.

2. Background and hypothesis development

2.1. The role of D&O insurers and auditors

Empirical evidence supports the view that both insurers' and auditors' assets are at risk inthe event of shareholders suits. For example, in their September 2007 10-Q, the Sun-Times



3S.L. Gillan, C.A. Panasian / The International Journal of Accounting xx (2014) xxx–xxx

Media Group Inc. (formerly Hollinger International) reported a US $30 million settlement ofseveral class action suits that would be paid for entirely by the firm's D&O insurers. At thesame time, it was disclosed that the insurers had set aside an additional US $24.5 millionin escrow to fund the firm's legal defense costs and any other potential claims. Similarly,at YBM Magnex International, Deloitte and Touche, the firm's auditor, contributed to aCanadian $120 million lawsuit settlement.2

D&O insurers provide explicit protection to client firms by covering expenses anddamages in the event that the firm is sued, while auditors might also act as insurers forclaimants in the case of a client's audit failure (although in the US a proportionate liabilityrule limits auditors' potential exposure).3 Thus, both D&O insurers and auditors haveincentives to evaluate and price their clients' litigation risk, and such risk will be reflectedin the D&O premiums and audit fees charged.

Indeed, a growing body of work examines aspects of litigation risk in the context of D&Oinsurance and audit pricing. For example, focusing on a sample of US IPOs, Chalmers, Dann,and Harford (2002) argue that if managers take firms public when they are overvalued (basedon insider information), then ex-post performance will be poor and lawsuits more likely. Assuch, rational managers will protect themselves by purchasing more D&O insurance. Onfinding a link between insurance coverage and poor performance, the authors conclude thatthe insurance purchase decision is indicative of managerial opportunism. Of course, insurerswill attempt to price such actions. Moreover, Baker and Griffith (2007) interview D&Oprofessionals and conclude that insurers “…in addition to performing a basic financial analysisof the company focus a large part of their efforts in understanding the corporate governance ofthe prospective insured….”

In this context, Cao and Narayanamoorthy (2011) use D&O liability insurance premiumsfor US firms as a proxy for litigation risk and find that managers with bad news are morelikely to issue earnings warnings. Furthermore, Beck and Narayanamoorthy (2013) examinethe relation between ex-ante litigation risk and audit pricing, using D&O insurance premiumsas a proxy for a firm's litigation risk, and show that audit fees are positively associated withthe D&O premium even after controlling for D&O policy limits, firm size, and other riskfactors known to influence both the cost of D&O insurance and audit fees.

Other studies in this area focus on firms cross-listed in different markets with distinctlitigation environments. For example, Core (1997, 2000) reports that D&O premiums areapproximately 35% higher for his 1992–93 sample of Canadian firms that cross-list in theUS,4 5 In addition to increased exposure for insurers of cross-listed firms, there is thepossibility of increased risk for auditors. Of note, Seetharaman, Lynn, and Gul (2002) reportfees that are 21% higher for UK firms with a US cross-listing, while Choi, Kim, Liu, andSimunic (2009) find audit fees 15–40% higher, depending on the difference in the

2 International Accounting Bulletin, March 14 2002.3 Core (1997, 2000), Boyer (2006), Simunic (1980); Simunic and Stein (1996), Dye (1993), Chaney and

Philipich (2002), Pratt and Stice (1994), Mansi, Maxwell, and Miller (2004).4 The US is typically viewed as a more litigious environment, and class action lawsuits are a significant source of

litigation risk. In contrast, investor protections in Canada have historically been weak (see, for example Bhattacharya andDaouk (2002); King (2009); Heys & Berenblut (2011); Siegel (2005)).5 Gillan & Panasian (2010) document an association between measures of D&O insurance coverage and the

likelihood of securities litigation for cross-listed Canadian firms.




strength of the legal regime between the home country and the country where the firmcross-lists.

On balance, prior work suggests the potential for differential audit effort, and possiblydifferential exposure to litigation risk conditioned on both cross-listing status and the nature offirms' disclosure complexity. Our focus is on the extent to which these different disclosurecomplexity and regulatory regimes are priced by auditors and D&O insurers.6 Our first twohypotheses revisit these issues and focus on themagnitude of the cross-listing premium in bothD&O insurance and audit fees.

Hypothesis 1. D&O liability insurance premiums are higher for Canadian firms cross-listed inthe US than for the Canadian-only listed firms.

Hypothesis 2. Audit fees are higher for Canadian firms cross-listed in the US than for theCanadian-only listed firms.

2.2. Disclosure complexity and litigation risk

Of note, during our sample period, firms listing only in Canada had to comply with Canadiandisclosure requirements. If cross-listed in the US, one of several alternatives could apply. Underthe MJDS if firms were i) a foreign private issuer incorporated or organized under Canadianlaw, ii) had been subject to the continuous disclosure requirements in Canada for a period of atleast 12 months, and iii) had an aggregate public-float market value of equity (held bynon-affiliates) of at least US $75 million, they could file annual reports with the SEC on form40-F. However, they could also cross-list under the rules for foreign private issuers and file a20-F or a 10-K.7

Historically, the 40-F took the form of a “wrap-around” of the Canadian filing with limitedreconciliation to US Generally Accepted Accounting Principles (GAAP) and no SEC review.In contrast, a 20-F entailed adherence to home-country accounting standards, but moredetailed disclosures, greater reconciliation to US GAAP, and SEC review. A 10-K requiredfull compliance with US GAAP. Thus, at a broad level, our sample comprises firms that aresubject to different levels of disclosure complexity (a 40-F, 20-F, or 10-K), and differentregulatory regimes (Canada or the US).

Prior work has examined the association between disclosure and litigation risk along severaldimensions. Lev (1992) and Skinner (1994) contend that voluntary disclosures bymanagementcan deter litigation, while Francis, Philbrick, and Schipper (1994) suggest that such disclosuresincrease the likelihood of litigation. Field, Lowry, and Shu (2005) find that firms with higherlitigation risk are more likely to disclose early in order to preempt lawsuits, and that earlydisclosure reduces the likelihood of being sued. In contrast to Beck and Narayanamoorthy(2013); Wynn (2008) reports that Canadian cross-listed firms with excess D&O coverage

6 Other recent work, e.g., Wynn (2008), suggests that Canadian firms with more D&O insurance coverageprovide more timely and accurate information, while Chung and Wyn (2008) report that increased managerialliability coverage is associated with less conservative earnings, while increased litigation risk is associated withmore conservative earnings. We abstract from timeliness and quality of earnings issues in this paper.7 For example, a Canadian firm not meeting the MJDS requirements may still cross-list and file a 20-F.




are less likely to provide bad news forecasts, especially so for large firms with highlitigation risk. Nonetheless, these papers suggest a link between litigation risk and aspectsof disclosure, with a potential nonlinearity in D&O premiums given that some disclosuresmight attenuate lawsuits.

At the same time, it is generally viewed that increased audit complexity is associated withincreased audit effort and audit fees (Causholli, DeMartinis, Hay, &Knechel, 2010). Related tothis, Bédard et al. (2009) focus on audit fees for Australian and Canadian firms cross-listed inthe US, and suggest that both litigation risk and disclosure complexity are important in auditpricing, although they focus on audit fees and not D&O insurance. In a similar vein, Callaghanet al. (2008) report that Canadian firms cross-listed in the US under the MJDS firms paysignificantly lower audit fees than non-MJDS firms.

On balance, then, the literature suggests that cross-listing and disclosure complexity resultin differential exposure to litigation risk along with differential D&O premiums and auditfees. Our next hypotheses focus on disclosure complexity.

Hypothesis 3. D&O liability insurance premiums increase with disclosure complexity,although there is a potential nonlinearity.

Hypothesis 4. Audit fees increase monotonically with disclosure complexity.

Further, increased audit fees may reflect at least two distinct components: increasedaudit effort associated with the additional disclosures accompanying a cross-listing, and alitigation risk premium, e.g., Bédard et al. (2009). We attempt to distinguish between thesecomponents, while also exploring how the disclosure complexity may affect audit effort,audit fees, and D&O insurance premiums.

Hypothesis 5. Audit fees for cross-listed firms reflect both audit effort and a litigation riskpremium.

To examine this issue, we model audit fees as a function of standard covariates including ameasure of abnormal D&O insurance premiums (where we explicitly omit the cross-listingvariable when estimating abnormal premiums). The rationale is that the residual containspublic information about cross-listing plus the D&O insurers' private assessment of abnormallitigation risk. To the extent this measure captures incremental litigation risk, including thatassociated with cross-listing, it provides insights as to incremental audit fees associated withcross-listing (and the differential disclosure levels).

3. Data and sample selection

The initial dataset comprises the largest 350 Canadian companies by market capitalizationlisted on the Toronto Stock Exchange (TSX) in 2005, representing approximately 95% ofboth the TSX market value and the market value of firms represented on the CanadianCompustat database.8 We exclude financials, REITS, and mutual funds, as their accounting

8 The Toronto Stock Exchange Review, December 2005.




disclosures and governance structures are fundamentally different from those of industrialfirms. Our analysis requires that each Canadian firm has audit fee data, D&O insuranceinformation, and a US counterpart (matched by 2-digit NAICS and size). This results in asample of 210 observations.

Data on D&O liability insurance (including premiums, coverage limits, and deductibles) iscollected from proxy statements available on the System for Electronic Document Analysisand Retrieval (SEDAR). Unfortunately, D&O data is not widely available for US firms. Auditfees for the Canadian sample were collected from proxy statements and annual informationforms. For US firms, we obtain audit fee data from The Corporate Library andAudit Analytics,supplemented with hand-collection where necessary. Financial statements variables areobtained from Compustat. Unless specified otherwise, throughout the paper we report allvalues in Canadian dollars.9

While the literatures on D&O insurance and audit fees are distinct, there are common firmcharacteristics and empirical approaches across both literatures. Thus, we draw on research fromboth areas to develop a parsimonious empirical specification. In terms of firm characteristics,Mayers and Smith (1982) argue that litigation risk increases with the size of the firm and thescope of its operations. Firm size (Size) is consistently found to be positively related with feelevels, as predicted by Simunic (1980). The underlying rationale is that larger firms place ahigher demand on auditors' resources, and present larger potential loss exposures. Similarly,because the literature suggests that high levels of free cash flow (FCF) are indicative of potentialagency problems, particularly in the face of low growth opportunities (measured by Tobin's Q),we incorporate proxies for these variables into our analyses.10

The firm's financial condition is another characteristic that is associated with potentiallitigation and thus with D&O premiums and audit fees. Core (1997) notes that D&Oliability claims are often triggered by weak financial performance. Moreover, Kinney andMcDaniel (1989) note that companies with weak financial condition are more likely to use“window dressing” to disguise problems, thus making the auditor's work more complex,while Palmrose (1986) observes a relationship between bankruptcies and lawsuits againstauditors.

As proxies for financial condition we use the debt to equity ratio (Debt/Equity), return onassets (ROA), an indicator variable denoting whether or not firm declared a loss in the currentyear (Loss), and the quick ratio, defined as the ratio of current assets less inventories to currentliabilities (QRatio). We expect that Debt/Equity and Loss should have a positive relation withaudit fees, while ROA andQRatio are expected to be negatively related to fees, as low ROA andquick ratios are potentially indicative of financial problems. We also include the standarddeviation of the stock price for the eight quarters prior to being in the sample (see, for example,

9 For the US-matched sample, all Compustat data and audit fees are transformed to Canadian dollars, using theaverage Canadian–US exchange rate for the year.10 From a theoretical perspective, Tobin's Q is the ratio of the market value to the replacement value of assets.We use an empirical proxy common to the accounting and finance literature of the ratio of the market value of acompany's stock to its equity book value.




Rogers & Stocken, 2005; Johnson, Nelson, & Pritchard, 2006), and 2-digit NAICS industryindicators.

As Simunic (1980) notes, auditor lawsuits often involve problemswith audit complexity.Weinclude four additional proxies from the audit literature that control for complexity: theproportion of current to total assets defined as current assets to total assets (CA/TA); an indicatorvariable equal to one if the firm acquired or disposed of a subsidiary during the year analyzed,and zero otherwise (Acquisition); the total number of business and geographical segments(Segments); and an indicator equal to one if the firm has international operations (Intl.Operations; Simunic, (1980); Choi, Kim, Liu, and Simunic, (2008); Choi et al. (2009)). Finally,as a measure of information asymmetry, we include the standard deviation of quarterly stockreturns over the prior eight quarters (Standard Deviation).

While not the primary focus of our analysis, the literature also suggests that corporategovernance attributes may be associated with D&O premiums and audit fees (Holderness,1990; Core, 1997, 2000; Carcello, Hermanson, Neal, and Riley, 2002; Gillan and Panasian,2010; Hay, Knechel, and Wong, 2006). For example, Larcker and Richardson (2004) findthat firms with weaker governance structures (less independent boards, low institutionalholdings, high insider holdings) have the strongest negative association between the feespaid to auditors and accounting accruals, and conclude that governance is a key factor indetermining accrual choices. For our Canadian firms, we hand-collect governance datafrom proxy circulars (including board size, board independence, directors' and officers'ownership, and block ownership), while we rely on data from The Corporate Library forthe US sample.

4. Univariate analyses

Table 1, panel A, presents summary statistics for the insurance Premium, Coverage,Deductible, and the Ratio of premiums to coverage less deductible as reported in year 2005proxy statements. The first column reports means and medians (in parentheses) for theentire sample. Columns 2 and 3 report means and medians (in parentheses) for the twosub-groups based on whether or not firms are cross-listed in the US The last twocolumns report statistics (p-values in brackets []) for differences of means (t-test) andmedians (Wilcoxon Z-values) between our two sub-groups. We note that all D&Oinsurance variables are statistically significantly higher for cross-listed firms than forthose not cross-listed, broadly consistent with the view that the US is a higher litigationregime.

Panel B presents the same statistics for audit variables: Audit Fees, Non-Audit Fees,Total Fees, and the Ratio of Audit Fees to Total Assets. Once again, the tests of differencesin mean and medians show statistically higher levels of all audit variables for cross-listedfirms. In Panel C, we present summary statistics and difference tests between Canadianlisted firms and the cross-listed firms for all other business and governance variables usedin our study. We note that cross-listed firms are significantly larger firms than the pureCanadian sample, and have lower FCF and ROA levels. Further, cross-listed firms havehigher CA/TA ratios, larger number of business segments, and higher percentage of firmswith international operations, suggesting increased complexity from an auditingperspective. Focusing on the standard deviation of stock returns, we note that US



Table 1Directors' & officers' liability insurance statistics. Summary statistics for the insurance Premium, Coverage, andDeductible and the Ratio of premiums to coverage less deductible. We divide the 2005 sample firms into two groupsbased on US cross-listing status. The first column reports means (medians in parentheses) for the entire sample. Column2 and 3 report means (medians in parentheses) for the two sub-groups based on the firms' listing status: Cross-listed firmand Canadian firms. The last two columns report test-statistics [p-values in brackets] for differences of means (t-test) andmedians (Wilcoxon Z-values) between the above-mentioned sub-groups. The last row presents the total number oobservations in each group. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels respectively.

Full sample Cross-listed Canadian Difference tests

Mean(median)

Mean(median)

Mean(median)

t-Value WilcoxonZ

Panel A: D&O liability insuranceInsurance

Premium 754,715 1,067,897 315,539 4.352*** 7.530***(353,073) (687,202) (195,000) [0.000] [0.000]

Max. coverage 57,900,000 66,000,000 45,700,000 2.503** 2.447**(35,000,000) (41,900,000) (25,000,000) [0.013] [0.014]

Deductible 1,105,727 1,610,626 397,708 4.654*** 5.063***(350,000) (605,865) (250,000) [0.000] [0.000]

Ratio 0.016 0.022 0.008 4.267*** 8.720***(0.011) (0.014) (0.007) [0.000] [0.000]

No. observations 210 122 88

Panel B: audit feesAudit fees

Audit fees 1,363,526 1,859,928 668,544 2.073** 3.462***(595,419) (759,650) (360,527) [0.041] [0.000]

Non-audit fees 511,177 600,577 385,810 1.810* 0.085(188,000) (205,398) (157,000) [0.059] [0.986]

Total fees 2,124,807 2,764,883 1,227,230 2.115** 2.681***(947,700) (1,320,200) (756,315) [0.039] [0.007]

Audit fees/TA 0.0011 0.0014 0.0007 2.054*** 2.451***(0.0006) (0.0006) (0.0004) [0.009] [0.008]

No. observations 210 122 88

Panel C: business risk and governance variablesVariable

InsuranceSize 3074.19 4117.77 1610.79 4.483*** 3.292***

(1,167.54) (1,298.49) (931.80) [0.000] [0.001]Debt/Equity 0.188 0.175 0.206 0.998 0.422

(0.176) (0.157) (0.186) [0.318] [0.673]FCF 0.059 0.053 0.067 2.445** 1.764*

(0.078) (0.079) (0.078) [0.015] [0.077]ROA 0.024 0.001 0.056 4.686*** 4.456***

(0.047) (0.037) (0.059) [0.000] [0.000]Tobin's Q 1.440 1.402 1.494 1.185 1.089

(1.143) (1.178) (1.178) [0.236] [0.275]CA/TA 0.490 0.607 0.327 1.270 2.648***

(0.299) (0.306) (0.290) [0.205] [0.008]QRATIO 0.143 0.167 0.110 3.636*** 3.598***

(0.064) (0.074) (0.056) [0.000] [0.000]



ss

f


Table 1 (continued)

Full sample Cross-listed Canadian Difference tests

Mean(median)

Mean(median)

Mean(median)

t-Value WilcoxonZ

Panel C: business risk and governance variablesVariable

InsuranceLoss 0.225 0.311 0.103 4.860*** 4.682***

(0.000) (0.000) (0.000) [0.000] [0.000]Acquisition 0.105 0.098 0.115 0.342 0.343

(0.000) (0.000) (0.000) [0.732] [0.731]Segments 5.445 6.008 4.655 4.291*** 4.118***

(0.000) (6.000) (4.000) [0.000] [0.000]Intl. operations 0.560 0.639 0.448 4.592*** 4.441***

(1.000) (1.000) (0.000) [0.000] [0.000]Std. Dev. Stockreturn

71.608 91.519 43.685 2.984*** 5.830***(24.586) (33.039) (14.118) [0.003] [0.000]

Board Size 9.148 9.361 8.851 3.049*** 2.986***(0.157) (9.000) (9.000) [0.002] [0.002]

Board Independence 0.743 0.756 0.725 1.490 1.340(0.778) (0.789) (0.750) [0.137] [0.180]

Blocks 0.270 0.232 0.323 1.400 1.278(0.142) (0.116) (0.195) [0.162] [0.201]

D&O Own. 0.152 0.144 0.165 0.574 2.574***(0.023) (0.017) (0.031) [0.566] [0.010]

No. Observations 210 122 88


cross-listed firms have means and medians more than double those of the Canadiansample, consistent with results of prior studies. Regarding governance variables, we findthat cross-listed firms have larger boards and lower inside ownership than those notcross-listed.

Table 2 provides insights into how insurance premiums and audit fees vary with firmcharacteristics. Panel A reports the mean (median in parentheses) D&O liability insurancepremium by quartiles (from low to high) for each variable analyzed. In addition, we testfor the equality of premiums across the quartiles and report Spearman rank correlationsbetween D&O premiums and each variable. The median insurance premium differssignificantly across quartiles for all audit variables utilized: Audit Fees, Non-Audit Fees,Total Fees, and the Ratio of Audit Fees to Total Assets. With regard to business riskvariables, premiums differ across the quartiles based on the levels of Size, Debt/Equity,and Standard Deviation in a monotonic fashion. While premiums tend to decrease acrossthe quartiles for Tobin's Q, the changes are not strictly monotonic.

When focusing on the governance variables, we note that premiums increase with boardsize, and while they differ for D&O ownership, again, the increase is not strictly monotonic.On balance, the univariate analyses are consistent with prior work, suggesting that premiumsvary with firm characteristics.



Table 2Univariate Tests for Quartiles of D&O Premiums and Audit Fees. Four separate sub-samples are created, based onthe quartile values of a given variable. We report the mean (median in parentheses) of D&O Insurance Premiumsfor each quartile in Panel A and the mean (median in parentheses) Audit Fees for each quartile in Panel B. Forexample, for Size in the business risk variables, we divide the sample into four groups on the basis of firm size,with quartile 1 consisting of those observations in the smallest quartile of size, and quartile 4 consisting of thoseobservations in the largest quartile sub-sample of size. The mean and median values of D&O Premiums withineach of the size sub-samples are then computed. Size (market value of equity), Debt/Equity ratio (book value ofdebt divided by the book value of equity), ROA (return on assets), FCF (free cash flow) as measured by([operating income before depreciation − taxes − dividends] /Total Assets), Tobin's Q (market value of equitydivided by [book value of equity + long term debt + current assets]), QRATIO (quick ratio, computed as the ratioof current assets less inventories to current liabilities), CA/TA (proportion of current to total assets, computed asthe ratio of current assets to total assets), Loss (an indicator equal to one if firm incurred a loss in the prior yearand zero otherwise), Acquisition (an indicator variable equaling one if the firm has acquired or disposed of asubsidiary in the prior year), Std. Stock Price (standard deviation of the stock price calculated for the previouseight quarters), and Segments (the total number of business and geographic segments). Governance variablesinclude Board Size (the number of directors on the board), Board Independence (the percentage of independentdirectors on the board), Blocks (the total percentage of shares held by blockholders of 10% or more), and D&OOwn (ownership by directors and officers). The last two columns present the Kruskal–Wallis test for the equalityof premiums and Spearman rank correlations, respectively (p-values in brackets).

Quartile 1(Low)

Quartile 2 Quartile 3 Quartile 4(High)

Kruskal–Wallis[p-value]

Spearman corr.[p-value]

Panel A: mean value of D&O insurance premium by quartile ranking of explanatory variablesAudit variables

Audit Fees 279,041 365,427 789,781 1,703,160 84.139** 0.620***(153,404) (257,294) (554,000) (1,111,364) [0.012] [0.000]

Non-Audit Fees 326,901 446,421 684,603 1,231,939 23.857*** 0.327***(166,007) (166,007) (424,105) (564,750) [0.000] [0.000]

Total Fees 246,702 405,076 726,080 1,446,349 69.218*** 0.572***(129,602) (271,150) (450,447) (920,000) [0.000] [0.000]

Audit Fees/TA 869,758 873,449 399,752 890,916 10.853** −0.160**(564,087) (367,792) (257,800) (333,119) [0.012] [0.019]

Business risk variablesSize (millions) 296,011 314,268 613,200 1,499,650 73.441*** 0.565***

(177,444) (214,500) (350,000) (987,000) [0.000] [0.000]Debt/EquityRatio

360,860 705,951 772,450 1,109,936 14.546*** 0.256***(231,785) (318,684) (395,713) (480,894) [0.002] [0.000]

FCF 847,554 702,095 778,931 655,132 3.562 −0.013(281,070) (379,566) (424,105) (344,750) [0.312] [0.840]

ROA 996,280 745,300 565,509 735,563 6.475* −0.101(324,063) (745,300) (295,953) (339,500) [0.090] [0.141]

Tobin's Q 851,180 925,354 396,893 844,295 51.217*** −0.305***(574,174) (436,000) (214,500) (353,073) [0.001] [0.000]

CA/TA 405,924 452,000 287,408 285,000 5.567 0.043(387,962) (452,000) (284,000) (303,448) [0.135] [0.297]

QRATIO 658,617 762,585 775,958 816,598 33.614*** −0.058(320,502) (406,906) (462,350) (276,849) [0.000] [0.156]

Std. Dev. Stockreturn

239,003 411,496 648,405 1,277,386 81.985*** 0.320***(136,275) (216,000) (322,457) (782,566) [0.000] [0.000]

Governance variablesBoard size 220,989 559,228 403,248 1,228,009 49.790*** 0.470***

(129,602) (268,277) (286,197) (773,826) [0.000] [0.000]




Table 2 (continued)

Quartile 1(Low)

Quartile 2 Quartile 3 Quartile 4(High)

Kruskal–Wallis[p-value]

Spearman corr.[p-value]

BoardIndependence(%)

590,551 985,523 615,949 784,702 2.307 0.051(263,039) (476,788) (392,052) (537,279) [0.315] [0.455]

Blocks (%) 944,821 392,052 286,197 537,279 0.324 −0.009(493,996) (286,197) (157,560) (157,560) [0.850] [0.890]

D&O Own. 954,281 386,064 547,457 840,756 14.238*** −0.097(503,300) (285,599) (143,581) (384,772) [0.002] [0.157]

Panel B: mean value of Audit Fees by quartile ranking of explanatory variablesD&O insurance

Premiums 297,084 618,584 1,269,946 3,232,549 81.899*** 0.623***(202,673) (386,221) (946,192) (1,780,000) [0.000] [0.000]

Maximumcoverage

354,286 757,831 936,346 3,145,474 71.725*** 0.579***(181,300) (386,128) (725,500) (1,403,000) [0.000] [0.000]

Deductible 854,982 465,487 609,168 2,658,805 59.268*** 0.457***(338,000) (336,000) (473,000) (1,464,380) [0.000] [0.000]

Insurance ratio 664,735 1,054,621 1,645,569 2,113,983 4.684 0.142**(484,000) (395,000) (1,005,000) (716,800) [0.196] [0.038]

Business risk variablesSize (millions) 304,779 486,779 889,192 3,106,537 237.001*** 0.628***

(176,074) (346,600) (548,000) (1,984,000) [0.000] [0.000]Debt/EquityRatio

385,309 874,659 1,355,664 2,626,586 90.361*** 0.383***(268,759) (485,261) (760,750) (1,098,712) [0.000] [0.000]

FCF 1,739,318 1,792,288 1,100,941 776,932 33.648*** 0.069*(352,439) (1,496,478) (1,310,305) (761,063) [0.000] [0.090]

ROA 2,058,348 994,700 564,473 687,414 20.044*** 0.018(413,750) (1,507,163) (734,724) (354,000) [0.000] [0.662]

Tobin's Q 1,458,077 1,462,081 734,724 1,645,295 51.217*** −0.305***(1,022,850) (658,855) (364,000) (359,000) [0.001] [0.000]

CA/TA 1,870,181 972,994 1,731,007 860,372 5.567 0.043(751,000) (620,700) (689,000) (385,000) [0.135] [0.297]

QRATIO 1,240,555 1,467,711 1,199,967 1,554,705 33.614*** −0.058(772,500) (851,250) (516,631) (355,000) [0.000] [0.156]

Std. Dev. Stockreturn

437,689 666,523 934,494 2,437,513 81.985*** 0.320***(269,444) (431,788) (471,286) (1,129,500) [0.000] [0.000]

Governance variablesBoard Size 344,164 594,617 708,055 2,558,302 43.257*** 0.447***

(273,270) (336,000) (485,261) (1,347,500) [0.000] [0.000]BoardIndependence(%)

1,961,452 1,390,647 1,274,706 954,876 9.012** 0.109(1,088,889) (705,068) (343,616) (444,017) [0.029] [0.113]

Blocks (%) 1,220,454 1,730,405 1,333,891 1,611,739 0.458 0.015(526,076) (957,000) (697,066) (667,720) [0.795] [0.827]

D&O Own. 1,580,738 835,096 542,678 1,847,415 27.897*** 0.165**(691,000) (387,000) (348,264) (849,500) [0.000] [0.016]




Table 3 Panel AOLS regressions obtaining abnormal D&O insurance coverage.

Model: Ln(MaxCoverage) = α0 + β1Prior Litigation + β2log(Size) + β3Debt/Equity + β4ROA +β5FCF + β6Board Size + β7Board Independence + β8Blocks + β9D&OOwn + β10Dual Class +β11Duality + β12C EO Total + β13C EO Incentives + β14Severance + β15Agreements + ε.

Constant 11.089***(0.000)

Prior litigation 0.076*(0.082)

Size 0.364***(0.000)

Debt/equity 0.039**(0.027)

ROA −0.400(0.523)

FCF 0.434(0.487)

Board size 0.055**(0.032)

Board independence 0.774**(0.047)

Blocks (%) 0.482**(0.040)

D&O Own. −0.105(0.686)*

Dual class 0.063(0.539)

CEO duality 0.112(0.373)

CEO total 0.152(0.774)

CEO incentives −0.377(0.190)

Severance 0.184*(0.060)

Agreements 0.403***(0.003)

Observations 210Regression F-stat 19.73Regression p-value (0.000)Adj. R-squared (%) 57.47

The last four rows present the number of the observations, the regression F-stat, the overallregression p-value, and regression-adjusted R-squared value. ***, **, and * represent two-tailedstatistical significance at the 1%, 5%, and 10% levels respectively (p-values in brackets).For variable definitions, see Table A-I in Appendix A.



In Panel B, we report the mean (median in parentheses) audit fees by quartiles for each variable,along with corresponding tests of differences between groups. The mean (and median) auditfees differ significantly across quartiles for all insurance variables: premiums, deductibles, total



coverage, and the ratio of premiums to coverage less deductible. Similar to the analysis in PanelA, and the prior literature, audit fees vary with proxies for business risk and governancecharacteristics. In both panels, audit fees and D&O insurance premiums increase in concertwith each other, suggesting, at a broad level, some commonality in D&O insurance and auditpricing.

Overall, the univariate analyses are suggestive of both D&O premiums and audit feesthat vary with firm characteristics, litigation regime that a firm is exposed to, anddisclosure complexity. We now move to examine these issues in the context of amultivariate setting.11

5. Multivariate analyses

We first present multivariate analyses for D&O insurance and audit fees.We then integratethe two analyses by incorporating information from D&O pricing into the audit fee models todisentangle the different components of the audit fee premium (audit effort and increasedlitigation risk). Finally, we attempt to provide insights into these issues for US audit pricing byincorporating a matched sample of US firms into the analysis.

5.1. Overview

We use the following general specification as the basis for our multivariate work:

Ln D&OpremiumsorAudit Feesð Þ ¼ α0 þ β1US Listing þ β3log Sizeð Þ þ β4Debt=Equityþ β5ROAþ β6FCF þ b7Tobin

0sQþ β6CA=TAþ β7QRATIO þ β8Lossþ β9Acquisitionþ β10 St:Dev:Stock Priceþ β12Segmentsþ β12Intl:Operationsþ β13Board Sizeþ β14Board Independenceþ β15Blocksþ β16D&OOwnþ β17IndustryControlsþ ε ð1Þ

We note, however, that we vary this specification slightly throughout the analysis. Inparticular, we add measures of abnormal coverage and non-audit fees, respectively, for theD&O Premium and Audit Fee regressions. In addition, the integrated analysis of audit feesand D&O insurance introduces measures of abnormal insurance premiums (discussedfurther below). Further, in several specifications the US listing indicator is replaced withseparate indicators representing the cross-listed firms filing a 40-F, 20-F, 10-K, or theUS-matched firms to capture the differential disclosure complexity.

11 With respect to the matched-US sample, we present a series of summary statistics and univariate tests in TablesA-II and A-III in the Appendix A. We match the US sample based on the 2-digit NAICS code and size, keepingsize variation within $5 million. Where a match is not found within that size bracket, we go up to $10 million.Table A-II in the appendix presents summary statistics for the audit and business risk variable employed in theanalysis according to the regime in which the firm operates. Table A-III reports results of tests of differences inmeans and medians between all Canadian firms in our sample and their US matches, for all audit and business-riskvariables. We note here that Canadian firms pay significantly lower audit fees, non-audit related fees, and havelower ratios of audit fees to total assets than the respective US sample.



Table 3 Panel BOLS regressions of the log of D&O premiums and audit fees.

Stage 2 results

Model 1: Ln(D&O premiums) = α0 + β1US Listing + β2Abnormal Coverage + β3log(Size) + β4Debt/Equity +β5ROA +β6FCF + b7Tobin's Q + β8QRATIO + β9CA/TA + β10Loss + β11Acquisition + β12St. Dev. Stock Price + β13Segments +β14Intl. Operations + β15Board Size + β16Board Independence + β17Blocks + β18D&O Own + β19IndustryControls + ε.Model 2: Ln(D&O premiums) = α0 + β140 F + β220 F + β310 K + β4 Abnormal Coverage + β5log(Size) +β6Debt/Equity + β7ROA + β8FCF + b9Tobin's Q + β10QRATIO + β11CA/TA + β12Loss + β13Acquisition +β14St. Dev. Stock Price + β15Segments + β16Intl. Operations + β17Board Size + β18Board Independence +β19Blocks + β20D&O Own + β21Industry Controls + ε.Model 3: Ln(Audit Fees) = α0 + β1US Listing +β2Abnormal Coverage + β3log(Size) +β4Debt/Equity + β5ROA + β6FCF + b7Tobin's Q + β8QRATIO + β9CA/TA + β1Loss + β11Acquisition + β12St. Dev.Stock Price + β13Segments + β14Intl. Operations + β15Board Size + β16BoardIndependence + β17Blocks + β18D&O Own + β19Industry Controls + ε.Model 4: Ln(Audit Fees) = α0 + β140 F + β220 F + β310 K + β4Non-Audit Fees + β5log(Size) + β6Debt/Equity + β7ROA +β8FCF + b9Tobin's Q + β10QRATIO + β11CA/TA + β12Loss + β13Acquisition + β14 St. Dev. StockPrice +β15Segments + β16Intl. Operations + β17Board Size + β18Board Independence + β19Blocks + β20D&O Own +β21Industry Controls + ε.

Model 1 Model 2 Model 3 Model 4

D&O premium D&O premium Audit fees Audit fees

US listing Filing status US listing Filing status

Constant 6.423*** 6.161*** 8.211*** 7.727***(0.000) (0.000) (0.000) (0.000)

Regulatory regimeUS listing 0.751*** 0.391*** –

(0.000) (0.002) –40 F – 0.650*** 0.209

– (0.000) (0.116)20 F – 1.084*** 0.572**

– (0.000) (0.011)10 K – 0.892*** 0.885***

– (0.000) (0.000)Abnormal coverage 0.444*** 0.470***

(0.000) (0.000)Non-audit fees 0.062*** 0.065***

(0.002) (0.001)Business risk

Size 0.383*** 0.408*** 0.323*** 0.375***(0.000) (0.000) (0.000) (0.000)

Debt/equity 0.272 0.311 −0.054 0.028(0.535) (0.475) (0.902) (0.947)

ROA −2.303*** −2.138** −1.925** −1.946**(0.006) (0.011) (0.021) (0.017)

FCF 0.479 0.271 −0.433 −0.360(0.554) (0.739) (0.592) (0.652)

Tobin's Q −0.072 −0.071 −0.198 *** −0.208 ***(0.187) (0.187) (0.000) (0.000)

QRATIO −0.199 −0.134 −0.685* −0.501(0.584) (0.713) (0.060) (0.162)




Table 3 Panel B (continued)

Stage 2 results

Model 1: Ln(D&O premiums) = α0 + β1US Listing + β2Abnormal Coverage + β3log(Size) + β4Debt/Equity +β5ROA +β6FCF + b7Tobin's Q + β8QRATIO + β9CA/TA + β10Loss + β11Acquisition + β12St. Dev. Stock Price + β13Segments +β14Intl. Operations + β15Board Size + β16Board Independence + β17Blocks + β18D&O Own + β19IndustryControls + ε.Model 2: Ln(D&O premiums) = α0 + β140 F + β220 F + β310 K + β4 Abnormal Coverage + β5log(Size) +β6Debt/Equity + β7ROA + β8FCF + b9Tobin's Q + β10QRATIO + β11CA/TA + β12Loss + β13Acquisition +β14St. Dev. Stock Price + β15Segments + β16Intl. Operations + β17Board Size + β18Board Independence +β19Blocks + β20D&O Own + β21Industry Controls + ε.Model 3: Ln(Audit Fees) = α0 + β1US Listing +β2Abnormal Coverage + β3log(Size) +β4Debt/Equity + β5ROA + β6FCF + b7Tobin's Q + β8QRATIO + β9CA/TA + β1Loss + β11Acquisition + β12St. Dev.Stock Price + β13Segments + β14Intl. Operations + β15Board Size + β16BoardIndependence + β17Blocks + β18D&O Own + β19Industry Controls + ε.Model 4: Ln(Audit Fees) = α0 + β140 F + β220 F + β310 K + β4Non-Audit Fees + β5log(Size) + β6Debt/Equity + β7ROA +β8FCF + b9Tobin's Q + β10QRATIO + β11CA/TA + β12Loss + β13Acquisition + β14 St. Dev. StockPrice +β15Segments + β16Intl. Operations + β17Board Size + β18Board Independence + β19Blocks + β20D&O Own +β21Industry Controls + ε.

Model 1 Model 2 Model 3 Model 4

D&O premium D&O premium Audit fees Audit fees

US listing Filing status US listing Filing status

CA/TA 0.001 0.001 −0.015 −0.009(0.885) (0.713) (0.559) (0.716)

Loss −0.050 −0.020 −0.581*** −0.537***(0.791) (0.915) (0.002) (0.004)

Acquisition −0.018 −0.014 0.111 0.088(0.914) (0.930) (0.499) (0.579)

Std. dev. stock price 0.001*** 0.001*** 0.002*** 0.001***(0.006) (0.006) (0.001) (0.004)

Segments 0.016 0.011 0.072*** 0.059***(0.361) (0.532) (0.000) (0.001)

Intl. operations 0.043 0.037 0.112 0.109(0.705) (0.742) (0.320) (0.321)

Governance controlsBoard size 0.020 0.033 0.030 0.044

(0.525) (0.296) (0.338) (0.151)Board independence 0.929** 0.971** 0.461 0.621

(0.032) (0.025) (0.284) (0.141)Blocks (%) −0.672*** −0.643** −0.374 −0.305

(0.012) (0.016) (0.159) (0.238)D&O own. 0.731** 0.743** 0.522* 0.552*

(0.016) (0.014) (0.084) (0.060)Industry controls Yes Yes Yes Yes

Total no. obs. 210 210 210 210Reg. F stat. 12.11 11.82 13.65 13.96Reg. p-value (0.000) (0.000) (0.000) (0.000)Adj. R-squared (%) 65.78 66.35 67.65 70.00

The last four rows present the number of the observations, the regression F-stat, the overall regression p-value, andregression-adjusted R-squared value. ***, **, and * represent two-tailed statistical significance at the 1%, 5%, and10% levels respectively (p-values in brackets).For variable definitions, see Table A-I in Appendix A.




Table 4OLS regressions of the log of audit fees with D&O insurance control.

Panel A: stage 1 — determining abnormal premiums

Model: Ln(D&O premiums) = α0 + β1Abnormal Coverage + β2log(Size) + β3Debt/Equity +β4ROA + β5FCF +b6Tobin's Q + β7QRATIO + β8CA/TA + β9Loss + β10Acquisition + β11St.Dev. Stock Price + β12Segments + β13Intl. Operations + β14Board Size + β15BoardIndependence + β16Blocks + β17D&O Own + ε.

Constant 8.371***(0.000)

Abnormal coverage 0.459***(0.000)

Size 0.446***(0.000)

Debt/equity 0.436(0.300)

ROA −2.585***(0.005)

FCF 1.523*(0.073)

Tobin's Q −0.059(0.333)

QRATIO 0.536(0.182)

Loss 0.327(0.115)

Acquisition 0.033(0.861)

Std. dev. stock price 0.001**(0.015)

Segments 0.021(0.255)

Intl. operations 0.197(0.105)

Board size 0.030(0.374)

Board independence 0.776(0.108)

Blocks (%) −0.703**(0.015)

D&O Own. 0.735**(0.026)

Observations 210Regression F-stat 15.95Regression p-value (0.000)Adj. R-squared (%) 53.49



Prior work has also employed indicators for the presence of a high quality auditor or priorlawsuits. For example, Choi et al. (2008) find an audit fee premium for “Big-4” auditors. Bédardet al. (2009) report similar premiums for complexity between non-Big 4 and Big 4, but the latterappear to charge more for expected liability. However, for our Canadian sample, only four firms


Panel B: stage 2 regression results

Model 1: Ln(Audit Fees) = α0 + β1US Listing + β2Non-Audit Fees + β3AbnormalPremium + β4log(Size) + β5Debt/Equity + β6ROA + β7FCF + b8Tobin's Q +β9QRATIO + β10CA/TA + β11Loss + β12Acquisition + β13St. Dev. Stock Price +β14Segments + β15Intl. Operations + β16Board Size + β17Board Independence +β18Blocks + β19D&O Own + β20Industry Controls + ε.Model 2: Ln(Audit Fees) + β140 F + β220 F + β310 K + β4Non-Audit Fees + β5AbnormalPremium + β6log(Size) + β7Debt/Equity + β8ROA + β9FCF + b10Tobin's Q +β11QRATIO + β12CA/TA + β13Loss + β14Acquisition + β15St. Dev. Stock Price +β16Segments + β17Intl. Operations + β18Board Size + β19Board Independence +β20Blocks + β21D&O Own + β22Industry Controls + ε.

Model 1 Model 2

US listing Filing status

Abnormal insurance Abnormal insurance

Constant 8.632*** 8.143***(0.000) (0.000)

US listing 0.371*** –(0.003) –

40 F – 0.201– (0.124)

20 F – 0.501**– (0.023)

10 K – 0.860***– (0.000)

Non-audit fees 0.054*** 0.059***(0.007) (0.003)

Abnormal premium 0.182** 0.159**(0.017) (0.033)Model 1 Model 2

Business riskSize 0.334*** 0.377***

(0.000) (0.000)Debt/equity −0.002 0.061

(0.996) (0.883)ROA −1.821** −1.881**

(0.027) (0.020)FCF −0.499 −1.881

(0.524) (0.648)Tobin's Q −0.196*** −0.204***

(0.000) (0.000)QRATIO −0.604* −0.442

(0.093) (0.211)Loss −0.566*** −0.523***

(0.003) (0.004)Acquisition 0.135 0.108

(0.401) (0.494)Std. dev. stock price 0.002*** 0.002***

(0.001) (0.003)

(continued on next page)

Table 4 (continued)




Table 4 (continued)

Panel B: stage 2 regression results

Model 1: Ln(Audit Fees) = α0 + β1US Listing + β2Non-Audit Fees + β3AbnormalPremium + β4log(Size) + β5Debt/Equity + β6ROA + β7FCF + b8Tobin's Q +β9QRATIO + β10CA/TA + β11Loss + β12Acquisition + β13St. Dev. Stock Price +β14Segments + β15Intl. Operations + β16Board Size + β17Board Independence +β18Blocks + β19D&O Own + β20Industry Controls + ε.Model 2: Ln(Audit Fees) + β140 F + β220 F + β310 K + β4Non-Audit Fees + β5AbnormalPremium + β6log(Size) + β7Debt/Equity + β8ROA + β9FCF + b10Tobin's Q +β11QRATIO + β12CA/TA + β13Loss + β14Acquisition + β15St. Dev. Stock Price +β16Segments + β17Intl. Operations + β18Board Size + β19Board Independence +β20Blocks + β21D&O Own + β22Industry Controls + ε.

Model 1 Model 2

US listing Filing status

Abnormal insurance Abnormal insurance

Segments 0.069*** 0.057***(0.000) (0.001)

Intl. operations 0.105 (0.106)(0.344) (0.327)

Governance controlsBoard size 0.036 0.048

(0.242) (0.111)Board independence 0.417 0.580

(0.326) (0.164)Blocks (%) −0.339 −0.279

(0.195) (0.274)D&O own. 0.522** 0.551*

(0.080) (0.058)Industry controls Yes Yes

Observations 210 210Regression F-stat 13.65 13.96Regression p-value (0.000) (0.000)Adj. R-squared (%) 68.65 70.30

The last four rows present the number of the observations, the regression F-stat, theoverall regression p-value, and regression-adjusted R-squared value. ***, **, and *represent two-tailed statistical significance at the 1%, 5%, and 10% levels respectively(p-values in brackets).For variable definitions, see Table A-I in Appendix A.

Business risk

12 In robustness tests we find that adding controls for auditor quality or prior lawsuits does not qualitatively altethe results, thus we do not include these as covariates in our main analyses.



have non-Big 4 auditors; thus, we do not explicitly control for auditor quality.12 Finally, all ofour analyses incorporate industry fixed-effects at the 2-digit NAICS code level, and we estimaterobust standard errors.

We expect both D&O Premiums and Audit Fees to be significantly higherfor our cross-listed sub-sample and, furthermore, that the premiums will vary withthe disclosure complexity of the cross-listed firms. However, as discussed in our

r



hypotheses, while audit fees likely increase monotonically due to increaseddisclosure complexity, D&O premiums may exhibit a nonlinearity, as increaseddisclosure might deter litigation.

We present in Table 3, Panel A results of the first regression employed to calculate theAbnormal Premium used subsequently in Panel B. The model used is as follows:

Ln MaxCoverageð Þ ¼ α0 þ β1Prior Litigationþ β2log Sizeð Þ þ β3Debt=Equityþ β4ROA

þ β5FCF þ β6Board Sizeþ β7Board Independenceþ β8Blocksþ β9D&OOwnþ β10DualClassþ β11Dualityþ β12CEOTotal þ β13CEOIncentives þ β14Severanceþ β15Agreementsþ ε:

Table 3, Panel B presents results of our primary specifications for both D&O insurancepremiums and audit fees. The first two columns (Models 1 and 2 of Table 3, Panel B) focuson D&O insurance premiums. Model 1 emphasizes the US cross-listing indicator (whileincorporating the main control variables and industry indicators), while in Model 2 wereplace the US indictor with separate indicators if the cross-listed firm files a 40-F, 20-F, or10-K.

Following Core (1997), the D&O Premium analyses include a measure of abnormalinsurance coverage as an additional covariate. Specifically, we estimate an ordinary leastsquares regression of the maximum coverage less the deductible on business-risk andgovernance controls, and incorporate the residuals from this specification into thepremium regression. The rationale is that both coverage and deductibles affect thepremium charged, and thus must be accounted for to estimate abnormal premiums.

The last two columns of Table 3, Panel B report the audit fee regressions (Models 3 and4). Using a similar approach as in the D&O analyses, in Model 3 we incorporate anindicator variable for the cross-listed firms, while in Model 4 we use three indicators tocapture the filing status of each cross-listed firm. However, consistent with the prior auditliterature, in the audit fee analyses we incorporate fees paid for non-audit services.

We find that in Models 1 and 3 the US listing coefficient is positive and highly significant,confirming earlier findings that D&O premiums and audit fees are higher for cross-listed firms.From an economic perspective,Model 1 suggests that D&O premiums are approximately 112%higher for Canadian firms that cross-list in the US relative to those that do not.13 The results ofthe audit fee analysis in Model 3 (where we incorporate non-audit fees as an additionalcovariate) are consistent with both the D&O regressions and with prior work reporting higheraudit fees for cross-listed firms. The significant US listing coefficient of 0.391 equates to a 48%audit fee premium for cross-listed firms. This contrasts with the Seetharaman et al. (2002) auditfee premium of 15–22% for their 1996–1998 sample of cross-listed United Kingdom firms andChoi et al. (2009) an estimate of 15–40%, depending on the home and foreign-market.

In dollar terms, based on the mean D&O premium for firms that do not cross-list in theUS ($315,539 from Table 1, Panel A) this represents an additional insurance cost of

13 Recall that D&O premiums are logged; thus, the economic magnitude of the increase in fees or premia relativeto the base is given by e Coefficient −1.



Table 5OLS Regressions of the Log of Audit Fees for Canadian Firms and U.S. Matches.

Model 1: Ln(Audit Fees) = α0 + β1US Listing + β2US Firms + β3Non-Audit Fees + β4Abnormal Premium +β5log(Size) + β6Debt/Equity + β7ROA + β8FCF + b9Tobin's Q + β10QRATIO + β11CA/TA + β12Loss +β13Acquisition + β14St. Dev. Stock Price + β15Segments + β16Intl. Operations + β17Board Size + β18BoardIndependence + β19Blocks + β20D&O Own + β21Industry Controls + ε.Model 2: Ln(Audit Fees) = α0 + β140 F + β220 F + β310 K + β4US Firms + β5Non-Audit Fees + β6AbnormalPremium + β7log(Size) + β8Debt/Equity + β9ROA + β10FCF + b11Tobin's Q + β12QRATIO + β13CA/TA + β14Loss +β15Acquisition + β16St. Dev. Stock Price + β17Segments + β18Intl. Operations + β19Board Size + β20BoardIndependence + β21Blocks + β22D&O Own + β23Industry Controls + ε.

Model 1 Model 2

Business risk Business risk

US listing Segments

Constant 9.625*** 9.425***(0.000) (0.000)

Regulatory regimeUS listing 0.400*** –

(0.003) –40 F – 0.270*

– (0.055)20 F – 0.436*

– (0.096)10 K – 0.997***

– (0.000)US firms 1.467*** 1.507***

(0.000) (0.000)Non-audit fees 0.041*** 0.044***

(0.001) (0.000)Business risk

Size 0.267*** 0.281***(0.000) (0.000)

Debt/equity −0.163 −0.164(0.577) (0.570)

ROA −0.828 −0.834(0.143) (0.137)

FCF 0.213 0.262(0.697) (0.631)

Tobin's Q −0.165*** −0.169***(0.000) (0.000)

QRATIO 0.113 0.169(0.694) (0.554)

Loss −0.275* −0.268*(0.078) (0.082)

Acquisition 0.176 0.163(0.250) (0.280)

Std. dev. stock price 0.002*** 0.002***(0.000) (0.000)

Segments 0.046*** 0.039***(0.002) (0.010)

Intl. operations 0.071 0.079(0.456) (0.400)




Table 5 (continued)

Model 1 Model 2

Business risk Business risk

US listing Segments

Governance controlsBoard size 0.040 0.047*

(0.120) (0.064)Board independence 0.625 0.721**

(0.059) (0.029)Blocks (%) 0.070 0.112

(0.752) (0.611)D&O own. 0.286 0.325

(0.255) (0.191)Industry controls Yes YesObservations 420 420Regression F-stat 11.72 11.64Regression p-value (0.000) (0.000)R-squared (%) 49.73 50.84

The last four rows present the number of the observations, the regression F-stat, the overall regression p-value, andregression-adjusted R-squared value. ***, **, and * represent two-tailed statistical significance at the 1%, 5%, and10% levels respectively (p-values in brackets).For variable definitions, see Table A-I in Appendix A.



approximately $353,000. Similarly, using the mean value of audit fees for firms that do notcross-list ($668,544 from Table 1, Panel B), this equates to incremental audit fees ofapproximately $320,000.

In the second D&O regression, Model 2, we divide the US indictor into threegroups based on their US filing status. Each of the coefficients for the differentdisclosure levels is statistically significantly different from zero. The economicinterpretation is that D&O premiums are 92% higher for 40-F filers, 196% for 20-F,and 144% higher for 10-K filers relative to Canadian issuers that do not cross-list in theUS. Our tests of differences between the coefficients in Model 2 show that thecoefficients for the 20-F and 40-F filers are significantly different from each other atthe 6% level. The coefficient for a 10-K filing is significantly less than that for a 20-F,but does not differ significantly from that of a 40-F filing. Overall, D&O insurerscharge a premium for cross-listing, and the premium varies depending on the firm'sdisclosure complexity.

The corresponding audit fee analysis in Model 4 indicates that audit fees aresignificantly higher for 20-F and 10-K filers, but are not significant for the 40-F firms atconventional levels. Focusing on the coefficients, the magnitude of premiums tends toincrease monotonically with the disclosure regime: 23% for a 40-F filer, 77%, for a 20-Ffiler, and 142% for a 10-K filer. The coefficients for 40-F and 10-K filers in Model 4 differsignificantly from each other at the 1% level, those for 40-F and 20-F filers differ at the10% level, while the coefficients for 10-K and 20-F filers do not differ from each other.



Similar to the findings for D&O insurance, audit fees differ according to disclosurecomplexity, albeit in a monotonic fashion.

With respect to the control variables, there are several commonalities between the D&Oinsurance and audit fee pricing models: Larger firms and firms with a higher standard deviationof stock returns and those with higher director and officer ownership pay consistently higheraudit fees and D&O insurance premiums, while higher ROA is associated with lower fees andinsurance premiums.

However, there are also several differences between the audit and D&O pricingmodels. For the D&O analyses, increased board independence is associated with largerpremiums, while higher block ownership is associated with smaller premiums, but thesevariables do not load in the audit fee model. In contrast, audit fees are significantly lowerfor firms with higher Tobin's Q, suggesting auditors price this differential valuationeffect. Furthermore, firms with a prior year loss pay higher audit fees, but these factors(Tobin's Q, prior year loss) are not priced by insurers. Similar to the findings of Simunic(1980); Choi et al. (2008, 2009), auditors appear to price the number of segments,although D&O insurers do not.

Overall, our findings suggest that, while there are commonalities in the pricing of D&Oinsurance and audit fees, there are also differences. Further, as indicated by the differentloading for the different filing groups, there is some evidence of pricing differences based onthe nature of firms' overall disclosure practices. At issue, however, is the extent to whichhigher audit fees for cross-listed firms are indicative of increased litigation risk, increasedaudit effort, or other factors. We explore this question in more detail below.

5.2. Audit fees premiums and D&O insurance coverage

Given that there are differences in the way that D&O insurers and auditors price theirservices, we use these differences to further examine the link between D&O pricing and auditfees. To do so, we include the “abnormal” D&O premium as an additional covariate in theaudit fee regression model. We obtain the “abnormal” D&O Premium by taking the residualsfrom the regression of D&O Premiums on its economic determinants, while omitting the USList indicator. Thus, the residuals capture both the public information about cross-listing andthe D&O insurers' private assessment of firm risk. Table 4, Panel A displays the results of thisfirst stage model used to obtain the Abnormal Premium coefficients. The model use is thefollowing:

Ln D&Opremiumsð Þ ¼ α0 þ β1AbnormalCoverageþ β2log Sizeð Þ þ β3Debt=Equityþ β4ROAþ β5FCF þ b6Tobin’sQþ β7QRATIOþ β8CA=TAþ β9Lossþ β10Acquisitionþ β11 St: Dev:Stock Priceþ β12Segmentsþ β13Intl:Operationsþ β14Board Sizeþ β15Board Independenceþ β16Blocksþ β17D&OOwnþ ε:

The results of the second stage regression are reported in Table 4, Panel B and indicate thataudit fees are positively associated with the abnormal D&O premium. Thus, in viewing theD&O abnormal premium as measure of firm-specific litigation risk not captured by the otherexplanatory variables plus the insurers' assessment of the cross-listing premium (as in Table 3,Panel B, Model 1), it appears that auditors largely price this risk and the model-adjusted




R-square's increase only modestly. At the same time, the US cross-listing coefficient inModel 1 remains significantly positive, and at a magnitude similar to that of the earlieranalysis (0.371 here versus 0.391 in Table 3, Model 3). Of note, the coefficient of the US Listindicator can be seen as an elementary estimation of the lower bound of the auditors pricing ofaudit effort, given that the abnormal premium coefficient now captures the cross-listingpremium.

When focusing on differential disclosure complexity, the coefficients in Table 4, Model2 are also of a similar magnitude to the earlier analyses. The coefficient for the 40-F filersequates to an audit fee premium of approximately 22%, although it is now significant onlyat the 0.126 level. Given that these filings are a “wrap-around” of the Canadian financials,any premium is arguably attributable to increased litigation risk rather than increased effortper se. In contrast, the coefficients remain significant and markedly higher for firms filing a20-F or a 10-K. This suggests that the increased audit fees for cross-listed firms are notablyhigher for firms with greater disclosure complexity.14 The audit fee analysis incorporatesthe US-matched sample.

Our tests thus far provide evidence of increased D&O premiums and audit fees for firmscross-listing in the US, and that the increased fees are concentrated in firms with morecomplex disclosures. One question that arises is the extent to which this evidence has thepotential to provide insights as to the litigation risk component of audit fees for US firms.To explore this issue, we augment our analysis by adding a matched sample of US firms.Although we cannot include our proxy for litigation risk as measured by abnormal D&Opremiums (given that such data is not generally available for US firms), we use the samebasic empirical specifications as before, with an additional indicator variable for thosefirms that are in the US-matched sample.

The results in Table 5 are generally consistent with those of the earlier analyses. Model 1of Table 5, with all business and governance variables included, suggests an audit feepremium of 49% for the Canadian cross-listed firms (coefficient of 0.40) and 334% for USfirms (coefficient of 1.467), both relative to Canadian-only listed firms. When we divide thecross-listed sample based on the disclosure level together with the separate indicator for USfirms, we find premiums of 31%, 55%, and 171% for 40-F, 20-F, and 10-K filers,respectively, while US firms pay fees 351% higher than similar Canadian-only listed firms(Model 2).

Testing for differences in coefficients, we find that while 20-F and 40-F are notstatistically different from each other, both 40-F and 20-F differ significantly from thecoefficients for 10-K filers and from US firms. We also find a marginally significantdifference between the coefficients on the Canadian 10-K filers and the US firms. That is, itagain appears that increased disclosure complexity plays a large role in audit fees for firmslisted in the US and that caution is warranted when generalizing the results for cross-listedfirms to the US environment.

14 As in our previous model, the test of difference between coefficients reveals that only 10-K and 40-F filershave statistically different coefficients, suggesting differential audit effort for these two groups.




5.3. Endogeneity, selection biases, and robustness tests

Given that not all firms choose to purchase D&O insurance or cross-list, selectionbias is a potential problem for our analysis. For example, it is possible that firms withweaker governance and greater litigation risk or higher probability of distress are morelikely to carry D&O insurance, choose higher limits, and be charged higherpremiums.15

Gillan and Panasian (2010) examine potential selection bias in the insurance purchasedecision for a sample of firms with and without insurance coverage using a two-stageselection procedure developed by Heckman (1979). They model the probability of a firmpurchasing D&O insurance as a function of key business and governance variables that arelikely associated with the demand for liability insurance, consistent with the approach ofCore (2000). They then use a probit specification with a dichotomous dependent variable,taking a value of one for firms carrying D&O insurance and zero otherwise. In the secondstage, they focus on the price of insurance, as measured by the natural logarithm of theD&O liability insurance premium. As the coefficients for the inverse-Mills ratio areconsistently not significant, they conclude that there is no evidence indicating a selectionbias.

We also investigate the possibility of a self-selection bias associated with cross-listing.The issue rests in the fact that cross-listing systematically varies with firms characteristics:For example, it has been documented that cross-listed firms are larger, have moreinvestment opportunity costs, and higher financing needs than firms that chose not tocross-list (Leuz, Nanda, & Wysocki 2003). In order to address this issue, we re-estimateboth the premiums and audit fee regressions using a Heckman 2-stage methodology,where the first stage is the probability of cross-listing as a function of size, 2-digit NAICScodes, and the average growth rate in operating income over the last eight quarters.Coefficients are similar in magnitude with the results obtained in the OLS regressions and,furthermore, the mills ratios are insignificant, suggesting our results are robust to controlsfor self-selection.

Our results might also be affected by other endogeneity issues — for example, reversecausality or omitted variables. As noted by Baker and Griffith (2007), premiums are set inan independent manner by the insurers, and the contracts are periodically renegotiated aftera thorough evaluation of the firm's risk factors. Therefore, endogeneity issues in terms ofreverse causality are potentially less severe for premiums.

We cannot rule out that the insurers' pricing decision might be driven by the omittedfirm characteristics that influence a firm's choice to cross-list or purchase D&Oinsurance decisions. However, we note that the magnitude and direction of our keyvariables of interest accord with economic intuition and remain significant even afterincluding various controls for firm governance and risk characteristics. At the same

15 Boyer (2006) discusses the probability of purchasing insurance as a function of firm characteristics, but doesnot explicitly incorporate that into his modeling of the premium decision. Core (2000) reports that his results arerobust when using a two-stage model that accounts for selection issues.




time, given evidence that cross-listing is indicative of better governed firms (e.g., Leland Miller (2008)), it would suggest that these firms are potentially less, rather thanmore, risky.

We also estimate additional specifications to investigate potential nonlinearities, in which weadd squared terms for board size, board independence, block ownership, and D&O ownership.We find that none of the squared terms are significant and the presence of the squared term doesnot qualitatively affect the results. We also specify additional models that include piecewiseregressions to further analyze the theoretical curvilinear relationship of inside and blockownerships and agency costs in a firm. Again, the results are qualitatively similar.

Although we include an intercept control for size in out regressions, we know that size is amajor determinant for both D&O premiums and audit fees. To further explore the possibilitythat our results are driven primarily by a size effect, we address this in an additional test. Fromour overall sample, we size-match a sample of Canadian-only listed firms with Canadiancross-listed firms. The resulting subsample is 45 matched firms in each group. For these firms,we do a difference in means and medians for D&O insurance variables (premiums, coverage,and deductibles) and audit variable (audit fees and total fees). All of our tests show that evenwhen size is held constant across the cross-listed status, insurance and audit variables aresignificantly higher for cross-listed firms than for the non-cross-listed ones. This suggests thatour results are not primarily driven by size; rather, they are attributable to the litigation regimeand the firm's disclosure complexity.

6. Discussion and conclusions

Consistent with prior evidence, we find that audit fees and Directors' and Officers'(D&O) liability insurance premiums are higher for firms cross-listed in the US. However,for our 2005 Canadian sample, we find substantially higher premiums than those reportedfor earlier time periods. For D&O insurers, we find that premiums are more than doublefor US cross-listed firms, while audit fees are approximately 50% higher. The findingssuggest that both service-providers view the US as a more litigious environment and,although there are commonalities in the pricing of D&O insurance and audit fees, there arealso differences.

We also find that auditors and D&O liability insurers charge differently for cross-listedfirms depending on disclosure complexity; while D&O premiums are significantly higherfor all cross-listed firms, they increase and then decrease with increased disclosurecomplexity. In contrast, audit fees increase monotonically as the filing complexityincreases. We also find that auditors appear to price abnormal premiums charged by D&Oinsurers. Thus, audit fee premia for cross-listed firms appear to capture aspects of bothlitigation risk and increased disclosure complexity.

Finally, an audit fee analysis focusing on differences between Canadian cross-listedfirms and a matched sample of US counterparts also finds differences based ondisclosure complexity. This suggests caution when generalizing results from theanalyses of cross-listed firms to the US environment. It also suggests that a potentiallyfruitful area for ongoing work in this area is to examine in more depth the nature ofdisclosure complexity, its components, and how it relates to audit fee premia andlitigation risk.




Appendix A

Table A-IVariables Used in the Study.

Variable name

Table A-IISummary Statisticreport means andthree, and four div

Please cite this afrom Canadian...

Variable definition

Business risk variables
Size Natural log of a firm's Total Assets Debt/Equity Book value of debt divided by the book value of equity ROA Return on assets FCF Free Cash Flow measured as: (Operating income before depreciation − taxes − dividends) / Total
Assets
Tobin's Q Market value of equity divided by (book value of equity + long term debt + current assets) QRATIO Quick ratio, computed as the ratio of current assets less inventories to current liabilities Loss Indicator equal to one if firm incurred a loss in the prior year and zero otherwise Acquisition Indicator variable equaling one if the firm has acquired or disposed of a subsidiary in the prior year Std. Dev. Stock
Price
Standard deviation of the stock price calculated for the previous eight quarters
Segments
The total number of business and geographic segments Intl. Operations Indicator of whether or not a firm has operations outside the country
Governance Controls
Board size Measures the number of directors on the board Board
Independence
The percentage of independent directors on the board
Blocks (%)
Total percentage of shares held by blockholders with 10% or more D&O Own. Ownership percentage held by directors and officers
s Including US Match Sample. Summary statistics for the main variables used in the study. Wemedians (in parentheses) for our entire sample of Canadian firms in column one. Columns two,ide the entire sample based on the firms' listing status: 10-K, 20-F, 40-F, respectively, with their

corresponding means and 43 medians. The last column presents means and median for the matched US sampleused in the study. The last row presents the total number of observation in each sub-group.

Variable

rticle as: Gi, The Interna

Canadian mean(median)

llan, S.L., & Panasian,tional Journal of Acco

10 K mean(median)

C.A., On Litigatiounting (2014), http

20 F mean(median)

n Risk and Disclo://dx.doi.org/10.1

40 F mean(median)

sure Complexity:016/j.intacc.2014.1

US mean(median)

Audit variables
Audit Fees ('000) 842.10 3,735.90 1,085.98 1347.42 1843.09
(356.59)
(1273.70) (249.77) (732.20) (1043.24) Non-Audit Fees ('000) 453.61 552.72 405.70 545.53 336.13
(160.38)
(86.46) (134.50) (213.00) (119.24) Total Fees ('000) 1,480.60 4,660.02 1,665.69 2177.34 2065.484
(768.31)
(1599.34) (514.56) (1234.00) (1145.08) Audit Fees/TA 0.0007 2.6175 2.6728 0.9063 0.0016
(0.000)
(2.159) (1.395) (0.482) (0.001)
Size ('000) 1,623.45 3,878.37 901.21 4485.92 2,924.94
(952.80)
(734.09) (317.02) (1737.25) (1098.38) Debt/Equity 0.212 0.170 0.084 0.182 0.212
(0.206)
(0.104) (0.007) (0.165) (0.188)
Evidence0.003



Table A-II (continued)Table A-II (continued)

Variable

Table A-IIIUnivariate Tests Including USentire sample into two groups:parentheses) values for the entirthree, respectively. The last two

Please cite this article as: Gifrom Canadian..., The Intern

Canadian mean(median)

Match Sample. SummaCanadian firms (regardle sample are presentedcolumns present test s

llan, S.L., & Panasian,ational Journal of Acco

10 K mean(median)

ry statistics for theess of cross-listingin column one andtatistics and p-valu

C.A., On Litigatiounting (2014), http

20 F mean(median)

main variables usestatus) and US firmfor Canadian andes (in brackets) fo

n Risk and Disclo://dx.doi.org/10.1

40 F mean(median)

d in the study. Wes. The mean and

US firms in columr difference in me

sure Complexity:016/j.intacc.2014.1

US mean(median)

Business risk variablesROA
0.055 −0.009 −0.007 −0.003 0.024
(0.058)
(0.010) (0.004) (0.039) (0.049) FCF 0.066 0.028 0.103 0.041 0.407
(0.077)
(0.077) (0.093) (0.076) (0.383) Tobin's Q 0.107 0.171 0.178 0.171 41.109
(0.055)
(0.155) (0.105) (0.060) (18.757) QRATIO 0.320 0.379 2.576 0.349 32.237
(0.274)
(0.324) (0.344) (0.299) (16.381) CA/TA 54.218 108.727 53.927 79.985 0.090
(14.149)
(28.806) (12.778) (33.106) (0.084) Std. Dev. Stock Return 46.812 106.429 40.822 72.559 1.629
(11.161)
(37.995) (7.981) (30.300) (1.239) No. Obs. 88 90 14 18 210
Variable definitions: see table A-I in Appendix A.

divide themedian (inns two andans (t-tests)

and medians (Wilcoxon's Z) across groups. ***, **, and * represent significance at the 1%, 5%, and 10% levelsrespectively.

Variable
All firms Canadian US Statistics for tests of differencesbetween means and medians
Mean(median)

Mean(median)

Mean(median)

t-Value
Wilcoxon Z
Audit variables
Audit Fees ('000) 1582.888 1322.691 1843.085 1.802* 5.142***
(775.62)
(508.015) (1043.23) [0.072] [0.000] Non-Audit Fees (‘000) 416.999 497.866 336.131 −1.966** −2.160**
(155.26)
(191.28) (119.23) [0.049] [0.030] Total Fees ('000) 2063.40 2061.32 2065.48 0.504 2.162**
(1042.75)
(940.42) (1145.07) [0.991] [0.030] Audit Fees/TA 1341.93 1077.19 1606.67 2.934*** 6.131***
(763.24)
(469.55) (1056.93) [0.003] [0.000]
Size ('000) 2970.68 −3016.42 2924.93 −0.189 0.037
(1129.66)
(1160.94) (1098.37) [0.850] [0.970] Debt/Equity 0.200 0.187 0.212 1.427 1.129
(0.183)
(0.178) (0.188) [0.154] [0.258] ROA 0.022 0.020 0.024 0.233 0.933
(0.047)
(0.044) (0.049) [0.815] [0.351]
(continued on next page)

Evidence0.003



Table A-III (continued)Table A-III (continued)

Variable

Please cite this article as: Gillfrom Canadian..., The Interna

All firms

an, S.L., & Panastional Journal of A

Canadian

ian, C.A., On Litigccounting (2014)

US

ation Risk and Di, http://dx.doi.org/1

Statistics for tests of differencesbetween means and medians

Mean(median)

Mean(median)

Mean(median)

t-Value

sclosure Complexity0.1016/j.intacc.201

Wilcoxon Z

Business risk variablesQRATIO
0.150 0.145 0.155 0.550 1.350
(0.067)
(0.063) (0.071) [0.582] [0.177] CA/TA 0.445 0.482 0.407 0.533 2.985***
(0.339)
(0.295) (0.383) [0.594] [0.002] Std. Stock Return 56.401 71.692 41.109 −2.874*** −1.492
(20.115)
(21.473) (18.757) [0.004] [0.135] FCF 0.072 0.054 0.090 3.064*** 2.001**
(0.081)
(0.078) (0.084) [0.002] [0.045] Tobin's Q 2.084 1.630 2.538 −0.968 −6.380***
(0.845)
(0.003) (1.687) [0.333] [0.000] Total No. Obs. 420 210 210
Variable definitions: See Table A-I in Appendix A.

References

Baker, T., & Griffith, S. (2007). Predicting corporate governance risk: Evidence from the directors' & officers'liability insurance market. University of Chicago Law Review, 74, 487–544.

Beck, P. J., & Narayanamoorthy, Z. C. (2013). Litigation risk and audit fees: The impact of qualitative factors.(Available at SSRN 2292699).

Bédard, J., Carson, E., & Simnett, R. (2009). Legal regime and audit fees: Disentangling the effect of litigationrisk and audit complexity. Working paper.

Bhattacharya, U., & Daouk, H. (2002). The world price of insider trading. Journal of Finance, 57, 75–108.Boyer, M. (2006). Directors' and officers' insurance and shareholders' protection. Scientific Series. CIRANO.Callaghan, J. H., Parkash, M., & Singhal, R. (2008). The impact of the multi-jurisdiction disclosure system on

audit fees of cross-listed Canadian firms. The International Journal of Accounting, 43(2), 99–113.Cao, Z., & Narayanamoorthy, G. S. (2011). The effect of litigation risk on management earnings forecasts*.

Contemporary Accounting Research, 28(1), 125–173.Causholli, M., DeMartinis, M., Hay, D., & Knechel, W. (2010). Audit markets, fees and production: Towards and

integrated view of empirical audit research. Journal of Accounting Literature, 30, 167–215.Carcello, J. V., Hermanson, D. R., Neal, T. L., & Riley, R. A. (2002). Board characteristics and audit fees.

Contemporary Accounting Research, 3, 365–384.Chalmers, J., Dann, L., & Harford, J. (2002). Managerial opportunism? Evidence from directors' and officers'

insurance purchases. Journal of Finance, 57, 609–636.Chaney, P., & Philipich, K. (2002). Shredded reputation: The cost of audit failure. Journal of Accounting

Research, 40, 1221–1245.Choi, J. H., Kim, B., Liu, X., & Simunic, D. A. (2008). Audit pricing, legal liability regimes and big 4 premiums:

Theory and cross-country evidence. Contemporary Accounting Research, 1, 55–99.Choi, H., Kim, J., Liu, X., & Simunic, D. A. (2009). Cross-listing audit fee premiums: Theory and evidence.

Accounting Review, 84, 1429–1463.Chung, H. H., & Wyn, J. P. (2008). Managerial legal liability coverage and earnings conservatism. Journal of

Accounting and Economics, 46, 135–153.Core, J. E. (1997). On the corporate demand for directors' and officers' insurance. Journal of Risk and

Insurance, 64, 63–87.

: Evidence4.10.003

http://refhub.elsevier.com/S0020-7063(14)00093-4/rf0010






























Core, J. E. (2000). The directors' and officers' insurance premium: An outside assessment of the quality ofcorporate governance. Journal of Law, Economics, and Organization, 16(2), 449–477.

Dye, R. (1993). Auditing standards, legal liability, and auditor wealth. Journal of Political Economy, 101, 887–914.Field, L., Lowry, M., & Shu, S. (2005). Does disclosure deter or trigger litigation? Journal of Accounting &

Economics, 3, 487–507.Francis, J., Philbrick, D., & Schipper, K. (1994). Shareholder litigation and accounting disclosures. Journal of

Accounting Research, 2, 137–164.Gillan, S., & Panasian, C. (2010). On lawsuits and managerial opportunism: Evidence from the market for

directors' and officers' liability insurance. Working Paper. Texas Tech University.Hay, D., Knechel, R., & Wong, N. (2006). Audit Fees: A meta-analysis of the effect of Supply and Demand

Attributes. Contemporary Accounting Research, 23, 141–191.Heys, B. A., Berenblut, M. L., & Dwhytie, J. (2011). Trends in Canadian securities class actions: 2013 Update.Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153–161.Holderness, C. G. (1990). Liability insurers as corporate monitors. International Review of Law and Economics,

10, 115–129.Johnson, M., Nelson, K., & Pritchard, V. (2006). Do the merits matter more? The Impact of the Private Securities

Litigation Reform Act. Journal of Law, Economics and Organization Advances Access, 23, 1–26.King, M. R. (2009). Prebid run‐ups ahead of Canadian takeovers: How big is the problem? Financial

Management, 38(4), 699–726.Kinney, W., & McDaniel, L. (1989). Characteristics of firms correcting previously reported quarterly earnings.

Journal of Accounting & Economics, 6, 185–204.Larcker, D. A., & Richardson, S. A. (2004). Fees paid to audit firms, accrual choices, and corporate governance.

Journal of Accounting Research, 3, 625–658.Lel, U., & Miller, D. P. (2008). International cross-listing, firm performance and top management turnover: A test

of the bonding hypothesis. Journal of Finance, 63, 1897–1937.Leuz, C., Nanda, D., & Wysocki, P. D. (2003). Earnings management and investor protection: An international

comparison. Journal of Financial Economics, 69(3), 505–527.Lev, B. (1992 June 8). Why corporations need a strategy on disclosure. San Francisco Chronicle.Mansi, S. A., Maxwell, W. F., & Miller, D. P. (2004). Does auditor quality and tenure matter to investors?

Evidence from the bond market. Journal of Accounting Research, 42, 755–793.Mayers, D., & Smith, C. (1982). On the corporate demand for insurance. Journal of Business, 55, 281–296.Palmrose, Z. (1986). Audit fees and auditor size: Further evidence. Journal of Accounting Research, 24, 97–110.Pratt, J., & Stice, J. (1994). The effect of client characteristics on auditor litigation risk judgments, Required audit

evidence, and recommended audit fees. The Accounting Review, 69, 639–656.Rogers, J. L., & Stocken, P. C. (2005). Credibility of management forecasts. The Accounting Review, 80, 1233–1260.Seetharaman, A., Lynn, S., & Gul, F. (2002). Litigation risk and audit fees: Evidence from UK firms cross-listed

on US markets. Journal of Accounting and Economics, 33, 91–115.Siegel, J. (2005). Can foreign firms bond themselves effectively by renting US securities laws? Journal of

Financial Economics, 75, 319–359.Simunic, D. A. (1980). The pricing of audit services: Theory and evidence. Journal of Accounting Research, 18,

161–189.Simunic, D. A., & Stein, M. (1996). The impact of litigation risk on audit pricing: Review of the economic and

evidence. Auditing: A Journal of Practice & Theory, 15, 119–134.Skinner, D. J. (1994). Why firms voluntary disclose bad news. Journal of Accounting Research, 1, 38–60.Wynn, J. P. (2008). Legal liability coverage and voluntary disclosure. Accounting Review, 83, 1639–1670.
















































on litigation risk and disclosure complexity: evidence from canadian firms cross-listed in the us

Documents