The impact of the Multi-jurisdiction Disclosure System on audit fees of cross-listed Canadian firms

Transcription

1 Available online at The International Journal of Accounting 43 (2008) The impact of the Multi-jurisdiction Disclosure System on audit fees of cross-listed Canadian firms Joseph H. Callaghan, Mohinder Parkash, Rajeev Singhal Oakland University, United States Abstract The Multi-jurisdiction Disclosure System (MJDS), a treaty between Canada and the United States (U.S.), was intended to facilitate the cross-listing of a firm's securities in the neighboring country. Under this system, eligible Canadian companies are allowed to use home-country documents to meet U.S. disclosure requirements and these documents are generally not reviewed by the Securities and Exchange Commission (SEC). We posit that the single-reporting requirement and lower SEC scrutiny may result in lower audit fees for MJDS firms. Based on audit-fee disclosures mandated by the SEC rule-making authority granted by the Sarbanes Oxley Act of 2002, we find a negative association between audit fees paid by U.S. cross-listed Canadian companies and their use of the MJDS. This result suggests that the lower audit fees provide an economic incentive to use the MJDS. Thus, our study provides evidence that the implementation of the MJDS may help facilitate crossborder listings by reducing audit costs. Additionally, this study confirms, for Canadian firms, some of the audit-fee determinants reported in earlier studies University of Illinois. All rights reserved. Keywords: MJDS; Audit fees; Audit-fee determinants; Cross-listing; Cross-listed Canadian firms; Disclosure systems We gratefully acknowledge the improvements suggested by both anonymous reviewers and the journal coeditor, Jere R. Francis. We also appreciate the comments and suggestions received from the discussant and participants of the AAA conference at the 2005 National Meeting in San Francisco, CA. Any remaining errors are our own. Corresponding author. addresses: callagha@oakland.edu (J.H. Callaghan), parkash@oakland.edu (M. Parkash), Singhal@oakland.edu (R. Singhal) /$ University of Illinois. All rights reserved. doi: /j.intacc

2 100 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Introduction Canada and the U.S. implemented the Multi-Jurisdiction Disclosure System (MJDS) in 1991 to facilitate cross-listing of firms' securities in both countries (Securities and Exchange Commission, 1991, 1993). A motivation for implementing the MJDS was to reduce registration and reporting costs associated with cross-listings. Under the MJDS, eligible U.S. cross-listed Canadian firms can use Canadian documents to meet standard U. S. reporting requirements for foreign issuers. 1 MJDS documents are generally not subject to review by the Securities and Exchange Commission (SEC). We posit that the singlereporting requirement and lower SEC scrutiny may result in lower audit fees for MJDS firms. Regulators as well as researchers have questioned the benefits of the MJDS and there is debate about whether to continue with the system (see the Aircraft Carrier Release ; Houston and Jones, 1999). Based on a survey of Canadian firms, Houston and Jones (1999) conclude that managers perceive no significant benefits of the MJDS for Canadian firms. On the other hand, in a report to the Ontario Securities Commission (OSC), Puri and Sen (2003) find that using the MJDS to meet U.S. annual disclosure requirements results in some financial savings. However, Puri and Sen do not observe substantial savings in accounting and auditing fees for MJDS firms based on interviews with three Canadian MJDS issuers and senior partners at an international accounting firm. Canadian firms listed on U.S. exchanges are obligated to meet certain filing requirements specified by the SEC. Under the rule-making authority granted to it by the Sarbanes Oxley Act of 2002 (SOX), the SEC required cross-listed Canadian firms to disclose auditor-fee data. 2 These disclosure requirements present an opportunity to empirically investigate the nature and determinants of audit fees for cross-listed Canadian firms. The limited evidence for the determinants of audit fees for Canadian firms is both dated and based only on survey data (Chung & Lindsay, 1988; Anderson & Zeghal, 1994). In this study, we examine the determinants of audit fees based on fee data disclosed by Canadian companies listed in the U.S. Specifically, we examine whether using the MJDS results in lower audit fees while controlling for other known audit fee determinants. Our sample consists of 118 Canadian firms cross-listed in 2002 and 2003, yielding 195 firm-year observations. Of these, 78 are MJDS firms, with 134 firm-year observations, and 40 are non-mjds firms, with 61 firm-year observations. We find evidence that audit 1 With the MJDS, the U.S. permits Canadian companies to issue securities in the U.S. under Canadian rules, and Canada permits U.S. companies to issue securities in Canada under U.S. rules. Although the MJDS also applies to U.S. firms cross-listing their securities on Canadian exchanges, this research is limited to Canadian firms cross-listing in the U.S. Hereinafter, the MJDS refers to the system available to Canadian firms listing in the U.S. 2 The SEC Final Rule of 2000 (File No. S ) required these disclosures for SEC registrants filing proxy statements on or after February 5, The SEC Final Rule of 2002 (File No. S ) required firms not issuing proxy statements to include these disclosures in their annual filings included in Form 20-F and Form 40-F. For Final Rule on Revision of the Commission's Auditor Independence Requirements, see final/ htm and for Final Rule on Strengthening the Commission's Requirements Regarding Auditor Independence, see

3 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) fees paid by MJDS firms are significantly lower than those paid by non-mjds firms. This result implies that Canadian firms utilizing the MJDS obtain significant economic benefits in the form of lower audit fees. We also find that firm size, book-to-market ratio, reporting lag, and overall industry effect are significant in explaining audit fees. This paper makes several contributions to the extant literature. First, while SEC reporting and disclosure requirements are costly barriers for foreign firms considering entering the U.S. capital markets (Bhushan & Lessard, 1992), there is debate as to whether the MJDS is effective in facilitating cross-border listings. The lower audit fees for MJDS firms documented in this paper contributes to the debate and provides support for the continuation of the system. Second, the results of this study may be useful for firms that are considering cross-listing, for audit firms taking cross-listed firms as clients, and for regulatory authorities. Third, DeFond and Francis (2005) in a recent paper argue in favor of more research on the effects of alternative institutional arrangements on auditing. In this spirit, we analyze and confirm some of the underlying determinants of audit fees for a new institutional arrangement. Finally, this study confirms for Canadian firms some of the audit fee determinants reported in earlier studies. The remainder of the paper is organized as follows. Section 2 provides background and presents the research hypothesis. Section 3 describes the research design and presents the empirical results. Section 4 concludes the paper. 2. Background and research hypothesis The MJDS allows eligible issuers to satisfy registration and reporting requirements by providing the SEC with disclosure documents prepared under Canadian securities laws. At the time the SEC adopted the MJDS, Canada adopted a parallel MJDS for U.S. issuers. Together, the system provides that issuers in the U.S. and Canada are principally subject to the specific disclosure requirements of only their home country. A Canadian issuer is eligible to use the MJDS to make public offering of any security in the U.S. if it has a minimum public float of $75 million (USD) and a minimum 12-month reporting history in Canada. 3,4 Canadian firms utilizing the MJDS have to file Form 40-F with the SEC. Form 40-F is an integrated form used both as a registration statement and as an annual report by eligible Canadian issuers. It thus serves as a wraparound for the Canadian companies' public reports. Canadian issuers submit annual reports prepared according to Canadian Generally Accepted Accounting Principles (CGAAP) to the SEC. 5 However, some Canadian 3 The public float of specified securities is defined as the market value of those securities held by persons other than affiliates of the issuer. For this purpose, an affiliate of an issuer is anyone who beneficially owns, directly or indirectly, or exercises control or direction over, more than 10% of the outstanding equity shares of the issuer. The determination of an issuer's affiliates is made as of the end of such issuer's most recently completed fiscal year. See footnote 3 of Chifor (2001) for MJDS eligibility criteria. 4 There has been a recently proposed amendment to the MJDS that would increase the public-float requirement from $75 million to $250 million (Chifor, 2001). 5 Although MJDS issuers submit annual reports prepared according to Canadian GAAP to the SEC, they are still required to reconcile differences between Canadian GAAP and U.S. GAAP in Form 40-F.

4 102 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) companies eligible to use the MJDS choose not to do so. 6 A Canadian company ineligible to use the MJDS is required to file disclosure documents either as a foreign private issuer using Form 20-F or as a U.S. domestic issuer using Form 10-K, which is based on U.S. GAAP. Given these process and disclosure differences, these three regulatory possibilities may give rise to differences in audit fees. Simunic (1980) posits that audit fees are an increasing function of the level of audit effort faced by an auditor. Effort differences between MJDS and non-mjds audits may arise from differential Generally Accepted Accounting Standards (GAAS), GAAP, and time constraints to file annual reports with the SEC. Canadian issuers using Form 20-F and Form 40-F have to reconcile differences between Canadian GAAP and U.S. GAAP. The MJDS allows Canadian GAAP financial statements, including the reconciliation to U.S. GAAP, filed with Form 40-F to be audited in accordance with Canadian GAAS, whereas financial statements in Form 10-K and reconciliations in Form 20-F are audited according to U.S. GAAS (Puri & Sen, 2003). If the effort required to apply different GAASs varies, then we would expect to find differences in audit fees. The direction in effort of these differences is an empirical question. Puri and Sen (2003) argue that U.S. GAAS has different standards than Canadian GAAS as a result of different standards imposed by U.S. GAAS, there is an incremental cost associated in moving from a Canadian GAAS audit under MJDS to a U.S. GAAS audit If this argument holds, we would expect to find lower audit fees for MJDS firms. The preparation of financial statements and any reconciliation, whether according to Canadian GAAP or U.S. GAAP, is the responsibility of the filing firm's management. Anecdotal evidence suggests that preparation of financial statements according to U.S. GAAP may be cumbersome for Canadian firms. For example, Deloitte & Touche, LLP, an independent registered chartered accounting firm based in Toronto, Canada, notes in its 2004 report to the shareholders of Nortel Networks Corporation (a non-mjds firm) that management has recognized material weaknesses in the form of lack of sufficient personnel with appropriate knowledge, experience and training in U.S. GAAP and lack of sufficient analysis and documentation of application of U.S. GAAP to transaction 7 Such deficiencies may lead to increased audit work and higher fees for non-mjds Canadian issuers. Additionally, while the SEC allows MJDS firms 180 days from the end of their fiscal year to file Form 40-F, the time period allowed for Form 10-K filers is only 90 days. If filing in a shorter period requires additional audit effort or costs, Canadian non-mjds filers may have to pay an audit premium. Furthermore, the SEC has implemented Regulation G, amending filing rules and requiring public companies that disclose or release non-gaap financial measures to also include the most directly comparable GAAP financial measures in that disclosure or release. 8 6 Some Canadian firms eligible to use the MJDS have filed Form 10-K as a U.S. domestic issuer. Puri and Sen (2003) point out that this may be because such foreign issuers want to look like domestic firms. Houston and Jones (1999) note that (p. 239) firm files form 10-K because U.S. stockholder ownership hovers around 50%... it files form 10-K even in years that ownership dips below 50% when it could theoretically file other forms. Data limitations do not permit us to infer why eligible firms choose not to utilize the MJDS. 7 See annual report for the company at annual_report.pdf. 8 See description of regulation G at

5 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Firms using the MJDS are exempt from this requirement. If these additional disclosures require audit, there may be more audit work and fees for companies not using the MJDS. Simunic (1980) also posits that audit fees are an increasing function of expected litigation losses faced by auditors. Pratt and Stice (1994) point out that auditors assess the expected liability loss and may raise the level of effort to reduce it. It follows that audit-fee differences may be observed when these factors vary across clients for a given audit period. The SEC does not generally review Form 40-F, whereas both Form 20-F and Form 10-K are reviewed by the SEC. Pincus, Holders, and Mock (1998) report that the SEC obtains enforcement leads from different sources including reviews of SEC filings. 9 Bonner, Palmrose, and Young (1998) observe that enforcement actions by the SEC have resulted in higher litigation by investors. The difference in the level of SEC scrutiny may give rise to perceived reduction in litigation risk between a Canadian cross-listed firm using the MJDS versus one using an alternative disclosure mechanism. The reduction in expected auditor litigation loss with use of the MJDS may decrease audit fees. In summary, utilizing the MJDS may result in lower expected litigation loss and less audit effort leading to lower audit fees. This was not the findings of prior research, however; Houston and Jones (1999) utilize data from a survey of Canadian mangers and report no benefits of the MJDS for Canadian firms listing in the U.S. They observe no significant increase in the U.S. cross-listing by Canadian companies following the implementation of the MJDS and only a few of the firms that responded to their survey indicated that the MJDS affected their decision to list on the U.S. exchanges. On the other hand, in a report to the Ontario Securities Commission (OSC), Puri and Sen (2003) find that using the MJDS to meet U.S. annual disclosure requirements offers some financial savings. To do the cost benefit analysis of utilizing Form 40-F, instead of Form 20-F, they conducted extensive interviews with numerous relevant stakeholders, including issuers, security lawyers, senior public-accounting firm partners, and investment bankers. Additionally, Puri and Sen also gather limited information from interviews of three Canadian MJDS issuers and senior partners at an international accounting firm. Based on those interviews, they do not observe substantial savings in accounting and auditing fees by firms using the MJDS. Given the limitations inherent in prior research (such as using survey data), we want to offer a more rigorous test of the cost differential for MJDS firms. We formulate the following hypothesis (in the alternative form) to investigate whether there are any benefits to firms utilizing the MJDS through the reduction in audit fees after controlling for other known determinants of audit fees: H1. After controlling for other determinants of audit fees, there is a negative association between utilization of the MJDS and audit fees. The minimum public-float requirement creates a selection system in favor of larger firms. Any analysis of audit-fee differences between MJDS firms and non-mjds firms needs to mitigate the self-selection bias. We address this issue in Section 3.4 as part of additional analyses. 9 Palmrose and Scholz (2004) find significant association between restatements, in general, and litigation. Palmrose, Richardson and Scholz (2004) point out that The SEC sometimes requests a restatement after reviewing company filings.

6 104 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Research design and empirical results 3.1. Research design In this section we present an audit-fee model to test our hypothesis. The determinants of audit fees are drawn from Simunic (1984), Palmrose (1996), Craswell and Frances (1999), DeFond, Raghunandan, and Subramanyam (2002) and Whisenant, Sankaraguruswamy, and Raghunandan (2003) among others as follows: LNAUDIT ¼ a 0 þ a 1 LNNAF þ a 2 LNTA þ a 3 BIG4 þ a 4 ROA þ a 5 RETURN þ a 6 VOLATILITY þ a 7 LEV þ a 8 INVREC þ a 9 INSTIT PCT þ a 10 SPECIAL þ a 11 BM þ a 12 SQSEGS þ a 13 FOROPS þ a 14 EMPPLAN þ a 15 LAG þ a 16 INITIAL þ a 17 D2003 þ a 18 MJDS þ e ð1þ In Eq. (1), the independent variables represent agency costs, complexity of operations, size, risk, performance, and the characteristics of the auditor. Consistent with earlier research, we define the variables as follows: LNAUDIT the natural log of the audit fees ($ actual); 10 LNNAF the natural log of the sum of all nonaudit fees paid to the auditor ($ actual); LNTA the natural log of total assets ($ thousands); BIG4 an indicator variable equal to one when an auditor is a member of the Big 4, zero otherwise; ROA operating income divided by total assets; RETURN the firm's raw stock return over the fiscal year; VOLATILITY the variance of the residual from the market model over the current fiscal year; LEV total debt divided by total assets; INVREC inventory plus accounts receivables divided by total assets; INSTIT_PCT the percentage of institutional holdings; SPECIAL an indicator variable equal to the absolute value of negative special items divided by total assets, zero otherwise; BM the book-to-market ratio; SQSEGS the square root of number of segments; FOROPS an indicator variable equal to one if the firm has foreign operations as indicated by foreign currency adjustments to income, zero otherwise; EMPPLAN an indicator variable equal to one if the firm has a pension or post retirement plan, zero otherwise; LAG number of days between fiscal year-end and earnings announcement date; INITIAL an indicator variable equal to one if the audit engagement is the initial two years, zero otherwise; 10 If audit fees were reported in Canadian dollars, we converted audit fees to USD using prevailing currencyexchange rates on the year-end dates. All other data obtained for this study were expressed in USD. Hereinafter, all currency amounts are expressed in USD.

7 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) D2003 MJDS an indicator variable equal to one if the firm's current fiscal year is reported as 2003, zero otherwise; an indicator variable equal to 1 if the firm utilized MJDS, zero otherwise. MJDS is our variable of interest and a significantly negative coefficient (α 18 )will confirm our hypothesis of a negative association between utilization of the MJDS and audit fees after controlling for other known determinants of audit fees. Eq. (1) includes various control variables to minimize the possibility that the experimental variable proxies for some other effect. Prior research suggests that audit fees are positively related with client size (Simunic, 1980), therefore we include LTNA to control for firm size with a predicted positive sign. The extant literature also indicates that audit fees increase with audit complexity. As in other studies we control for complexity by including INVREC, SQSEGS, FOROPS, and EMPPLAN with predicted positive signs. The audit literature documents that audit fees increase with audit risk and decrease with firm performance. As suggested by Whisenant et al. (2003), we include ROA, RETURN, VOLATILITY, LEV, and BM to control for audit risk and firm performance. As longer reporting lags are associated with higher audit fees, we include LAG (Gul, 1999). We include LNNAF to control for the effect of nonaudit fees on audit fees (DeFond et al., 2002; Whisenant et al., 2003). BIG4 is included as earlier studies have reported fee premia paid to larger audit firms (Whisenant et al., 2003). As suggested by DeAngelo (1981), we include INITIAL to control for any discounting of audit fees because of low-balling at the time of initial engagement of auditors. Following DeFond et al. (2002) and Whisenant et al. (2003), we include INSTIT_PCT and SPECIAL with positive predicted signs. Since we are pooling observations across two periods, we include D2003 to control for any period effect Sample To construct our sample, we searched for auditor-fee data for all the 675 Canadian incorporated firms (1350 firm-years for 2002 and 2003) present in the active and research files of the Compustat database for the year We include data from 2002 and 2003 in our analysis to increase the sample size and to assess the stability of parameter estimates. We exclude observations with no Compustat data from our sample, thereby reducing the number of observations to 498 firm-years. We use two sources to obtain the auditor-fee data for our sample firms. First, we obtained auditor-fee data from the Compustat-provided audit and nonaudit-fee database. Then, we searched the SEC Edgar database for proxy statements and, Form 10-K and Form 40-F filings to obtain auditor-fee data for the remaining firms. We could obtain fee data for 268 firm-years. For our analysis, we obtain firm-level accounting data from Compustat, institutional holding data from Compact Disclosure, and market data from the CRSP database. Excluding firms that have data missing from these databases reduces the final sample to 195 firm-years. The final sample has 118 firms (78 MJDS and 40 non-mjds firms), yielding 134 firm-years for MJDS firms and 61 firm-years 11 The SEC Final Rule of 2002 requires U.S. cross-listed firms with fiscal year ending after December 15, 2003, to implement provisions of SOX, including disclosure of auditor-fee data for the two most recent years.

8 106 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Table 1 Sample selection from 2002 and 2003 Firm-years Firm-years in 2004 version of Compustat for 675 Canadian firms 1350 Less: Firm-years for inactive firms (852) Firm-years for active firms 498 Less: Firm-years with missing auditor-fee data (230) Firm-years with auditor-fee data 268 Less: Firm-years with missing institutional ownership data (31) Less: Firm-years with missing CRSP data (28) Less: Firm-years with missing Compustat data (14) Firm-years in final sample of 114 firms 195 Firm-years for MJDS firms 134 Firm-years non-mjds firms 61 Number of observations common to 2002 and fornon-mjds firms. Thus,asreported in Table 1, our final sample consists of 195 firm-years from fiscal years 2002 and 2003 for which we have complete accounting, institutional, and market data. There are 82 firms with complete data for both years. Table 2 Sample distribution of audit fees Industry MJDS Non-MJDS Difference Number of observations Median audit fees ($) Number of observations Median audit fees ($) p-value median test Panel A: Distribution of audit fees by disclosure system and industry Mining and construction , , Food 2 1,065,972 Textiles and printing 12 1,061,266 Chemicals 4 475, , Pharmaceuticals , , Extractive 9 689, , Durable manufacturers , , Transportation 14 1,348, , Utility 7 976,600 Financial 7 3,866, , Retail 2 189, , Services , , Computers 4 263, , Panel B: Audit fees paid by disclosure system and auditor type BIG , , Non-BIG , , Panel C: Audit fees paid by disclosure system and fiscal year , , , , Change in Median (%) ,, Significance at the 0.01, 0.05, and 0.10 levels, respectively, using two-tailed tests.

9 Table 3 Sample distributions of fees and control variables Variable MJDS Non-MJDS p-value difference test Mean Median Standard deviation Mean Median Standard deviation Mean Median Audit fees (AUDIT) $1,120,825 $561,489 $1,852,054 $783,950 $263,000 $1,680, Nonaudit fees (NAF) $986,333 $373,000 $2,076,348 $827,385 $167,506 $2,835, Natural log of AUDIT (LNAUDIT) Natural log of NAF (LNNAF) Total assets (TA) ($000) $5,444,070 $1,006,840 $12,499,906 $1,166,115 $190,748 $3,301, Natural log of TA (LNTA) Square root of segments (SQSEGS) Debt to assets (LEV) Inventory and receivable intensity (INVREC) Return on assets (ROA) 1.34% 4.76% 15.04% 4.20% 5.47% 25.98% Institutional ownership (INSTITPCT) 22.20% 20.27% 16.62% 18.94% 15.01% 17.57% Return volatility (VOLATILITY) Book-to-market (BM) Reporting lag in days (LAG) Fiscal year stock return (RETURN) 60.28% 43.99% % 29.57% 3.47% % First or second year audit (INITIAL) 6% 5% 0.76 Big 4 audit firm (BIG4) 93% 93% 0.82 Employee benefit plan (EMPPLAN) 51% 31% 0.01 Foreign operations (FOROPS) 62% 61% 0.87 Special items (SPECIAL) ,, Significance at the 0.01, 0.05, and 0.10 levels, respectively, using two-tailed tests. J.H. Callaghan et al. / The International Journal of Accounting 43 (2008)

10 108 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Table 2 shows the distribution of audit fees for MJDS and non-mjds sample firms by industry, auditor, and fiscal year. As reported in panel A of Table 2, two industries (durable manufacturers and mining and construction) have higher firm-year observations than other industries. The mining and construction industry has the highest number (25) of MJDS firm-year observations and relatively fewer non-mjds firm-year observations (4). Firms in two industries, textile and printing and utility, use the MJDS exclusively, while firms in one industry, Food, are all non-mjds. There are notable differences in median audit fees for MJDS and non-mjds observations in different industries. For three industries (retail, services and computers) median audit fees are higher for non-mjds observations than for MJDS observations, whereas for all the other industries the opposite is true. Panel B of Table 2 presents the composition of audit fees by type of auditor in different disclosure systems. Median audit fees for MJDS firms are significantly higher (p-value b 0.05) than median audit fees for non-mjds firms for both Big 4 and non-big 4 auditors. Panel C of Table 2 shows audit-fee composition by fiscal year for the two disclosure systems. Table 3 reports descriptive statistics for all variables of interest by disclosure system. The median audit fees paid to auditors by MJDS firms is $561,489, which is significantly higher than the $263,000 of audit fees paid by non-mjds firms. This may be attributable in part to size differences between MJDS and non-mjds firms. On average the total assets of MJDS firms are larger than those of non-mjds by a factor of 4.7. We also find significant differences across disclosure systems in nonaudit fees, inventory and receivable intensity, return volatility, reporting lag, fiscal year stock return, employee-benefit plans, and special items. The higher new financing for MJDS firms may be the result of the relative ease with which MJDS firms can issue equity and debt in U.S. financial markets in contrast to non- MJDS firms. The higher reporting lag for non-mjds firms may be an indication of the extra time needed by auditors for non-mjds firms to prepare additional disclosures. We find that INITIAL, BIG4, and FOROPS are similar for MJDS and non-mjds firms Regression results Table 4 presents the results of estimating Eq. (1) using ordinary least squares (OLS) regressions with pooled data for fiscal years 2002 and Pooling observations across time may lead to dependence among observations. To mitigate this potential effect, we include a dummy variable, D2003. We also separately estimate Eq. (1) for each year. To control for possible industry effects, we include 12 dummy variables to represent membership of 13 industry classifications (see panel A of Table 2). These results are included in Table 4. The tests of overall model fits result in F-statistics ranging from to 35.95, all statistically significant (p-value b 0.01), indicating that the variations in audit fees are adequately explained by the set of independent variables included in the tested models. We find adjusted-r 2 ranging from 0.82 to 0.85 for the three regressions estimated. The White (1980) test fails to reject the null hypothesis of homoskedasticity in the data. Therefore, we report standard t-statistics. To test the possibility of multicollinearity, we compute the variance inflation factor (VIF) for each of our variables. The highest VIF is less than 10. Kennedy (1992) indicates that VIFs of greater than 10 indicate a multicollinearity problem. Therefore, multicollinearity does not appear to be a problem with our data. We also test the structural stability of our model using the Chow test. We find

11 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Table 4 Audit-fee model Predicted sign Years 2002 and 2003 Year 2002 Year 2003 OLS estimate t-statistics OLS estimate t-statistics OLS estimate t-statistics INTERCEPT? LNNAF? LNTA SQSEGS LEV INVREC ROA? INSTIT_PCT VOLATILITY BM LAG RETURN INITIAL BIG EMPPLAN FOROPS SPECIAL D2003? MJDS MINE_CONS? FOOD? TEX_PRINT? CHEMICAL? PHARMA? EXTRACT? DUR_MAN? TRANSP? UTILITY? FIN? SERVICES? COMPUTER? Observations F-statistic p-value Adjusted R ,, Significance at the 0.01, 0.05, and 0.10 levels, respectively, based on one-tailed tests for signed predictions, two-tailed tests otherwise. no evidence of structural changes between 2002 and 2003 (p-value = 0.65). Our main variable of interest, MJDS, has a significantly negative coefficient (p-value b 0.05, onetailed) for the pooled data, as well as for individual years. This provides support for H1 indicating that the U.S. cross-listed Canadian companies utilizing the MJDS pay lower audit fees than non-mjds firms after controlling for other audit-fee determinants. The lower audit fees paid by MJDS firms provide an economic advantage for those firms availing themselves of this cross-listing mechanism.

12 110 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) We also find that the coefficients of LNNAF, LNTA, BM, and LAG are significantly (p-value b0.05, one- or two-tailed) different than zero. The coefficient of LNNAF is positively significant, using two-tailed tests, for pooled and individual year observations. This result is consistent with prior research on the relationship of nonaudit and audit fees, indicating possible knowledge spillovers (Simunic, 1984). The significantly positive coefficient on firm size is consistent with earlier Canadian studies that found higher audit fees for larger firms (Chung & Lindsay, 1988; Anderson & Zeghal, 1994). In contrast to the results reported in DeFond et al. (2002) and Whisenant et al. (2003), wefindthe coefficient on BM to be positively significant. The coefficient on LAG is positively significant, indicatinghigherauditfeesforlongerreporting lags, which is consistent with results reported in earlier studies (e.g., Gul, 1999). Even though MJDS firms are allowed a longer time period to meet their filing requirements, we observe shorter reporting lags for these firms (see Table 3), which may have an indirect effect on audit fees. The coefficient of INITIAL is negatively significant (p-value b0.10), indicating the low-balling discount effect predicted by DeAngelo (1981). We perform a partial F-test on the industry dummy variables and find significant industry effect (p-value b 0.01). We find significant coefficient only on EXTRACT, indicating that the extractive industry has significantly lower audit fees Additional analyses 12 The results in Table 3 indicate that firms utilizing the MJDS are significantly larger in size. To test for possible model misspecification with respect to firm size and endogeneity of MJDS choice, we perform three analyses. First, we calculate the correlation of the residuals obtained from Eq. (1) with firm size (LNTA). We find no correlation implying the independence of firm size and the error process. Second, to test for endogeneity of MJDS choice, we model MJDS as a function of firm size and industry membership. We use PROBIT regression to estimate the MJDS model and OLS regression for the audit-fee model (e.g., see DeFond et al., 2002). Then, we employ the Hausman (1978) test to check for endogeneity. We are unable to reject the null hypothesis of no endogeneity. Third, although the Hausman test fails to detect endogeneity of MJDS choice, we use a two-stage least-square estimation procedure to directly control for the possible misspecification. The untabulated results are qualitatively similar to the reported OLS results. Based on preceding analyses, our results are robust to possible model misspecification with respect to firm size and endogeneity of MJDS choice. The number of observations reported in Table 1 indicates that there are 82 firms in our sample with complete data for both years. In order to ensure that the reported results are not driven by differences in firm-year observations, we estimate Eq. (1) using only those observations having data for both periods. Our main result is qualitatively similar to the result reported in Table We use outlier detection and estimation methods of Huber (1973), Rousseeuw and Yohai (1984), Yohai (1987), and Rousseeuw and Van Driessen (submitted for publication). The four procedures resulted in identification of three, five, one, and 10 outliers, respectively. The resulting estimated regressions are qualitatively similar to OLS regressions reported in Table 4.

13 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) The information in panel A of Table 2 indicates that the audit-fee distribution across industries and disclosure systems warrants additional analyses to test the robustness of the results. First, mining and construction has the highest number (25) of MJDS firm-year observations and relatively fewer non-mjds firm-year observations (4). Second, two industries, textile and printing and utility, have firms that use the MJDS exclusively, while one industry, food, has all non-mjds. Third, for three industries (retail, services, and computers) median audit fees are higher for non-mjds observations than for MJDS observations, whereas for all the other industries the opposite is true. We re-estimate Eq. (1) by: (i) excluding observations from mining and construction industry, (ii) excluding observations from food, textile and printing, and utility industries, and (iii) excluding observations from retail, services, and computers industries. The untabulated results remain qualitatively unchanged from those previously reported. In Table 4, we observe a positive association between audit and nonaudit fees. Earlier research interpreted the observed association between audit and nonaudit fees as suggesting knowledge spillover between these services (Simunic, 1984; Palmrose, 1996; Davis, Ricchiute, & Trompeter, 1993; Bell, Landsman, & Shackelford, 2001). However, these inferences were based on single-equation estimation of audit-fee and nonaudit-fee models. Recent research proposes that audit and nonaudit fees may be simultaneously determined. Therefore, the relationship between audit and nonaudit fees may be biased when singleequation estimation is used (Antle, Gordon, Narayanamoorthy, & Zhou, 2002; Whisenant et al., 2003). 13 If simultaneously determined, both audit and nonaudit fees are endogenous to a system of equations. Treating audit and nonaudit fees as endogenous does not qualitatively change our results with respect to the effect of MJDS choice on audit fees Concluding remarks We have extended prior research on the determinants of audit fees to consider the influence of the MJDS on audit fees for U.S. cross-listed Canadian firms. The empirical results provide support for the assertion that cross-listed Canadian companies utilizing the MJDS pay lower audit fees than do non-mjds firms, after controlling for other known determinants of audit fees. This result implies that Canadian firms utilizing the MJDS obtain significant economic benefits in the form of lower audit fees. Foreign firms acknowledge that the SEC reporting and disclosure requirements are costly barriers to enter U.S. capital markets. The implementation of the MJDS by the SEC as a bilateral agreement 13 For U.S. data, Whisenant et al. (2003) report no association between audit and nonaudit fees when the system of audit and nonaudit fees equations is simultaneously estimated. In contrast, using audit and nonaudit data from the United Kingdom, Antle et al. (2002) find evidence of a significant association between audit and nonaudit fees. 14 The variables explaining nonaudit fees are identified from Parkash and Venable (1993), Firth (1997), DeFond et al. (2002), Frankel, Johnson, and Nelson (2002) and Whisenant et al. (2003) and are LNAUDIT, LNTA, BIG4, ROA, RETURN, LEV, INSTIT_PCT, SPECIAL, BM, SQSEGS, FOROPS, EMPPLAN, INITIAL, MERGER, FINANCE, SALES GROWTH, and MJDS. SALES GROWTH is growth rate in sales over the previous fiscal year, MERGER equals one if the firm acquired another firm during the current fiscal year, zero otherwise, and FINANCE equals one if the firm issues either equity or long-term debt in either the current or the subsequent fiscal year, zero otherwise.

14 112 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) with Canadian regulatory authorities was an important step in facilitating globalization of capital markets. By showing at least one advantage related to accounting and disclosure, our study provides support for the policy adopted by the SEC. Only Canadian firms crosslisted in the U.S. disclose auditor-fee data. Therefore, the results on determinants of audit fees should be interpreted with caution, as they may not apply to other Canadian firms. Finally, this paper does not address the issue of audit fees for U.S. firms cross-listed on Canadian exchanges. This paper does not specifically address the issue of all costs and benefits associated with MJDS choice. Future research is required on this issue. Future research could also profitably examine specific determinants of MJDS choice, the potential for similar agreements with other jurisdictions, and the influence of institutional arrangements on audit and accounting fees for cross-listed firms. The study of the effects on audit fees resulting from specific changes in institutional arrangements, like SOX and harmonization of international reporting standards, may also enhance our understanding of audit-fee determination. References Anderson, T., & Zeghal, D. (1994). The pricing of audit services: Further evidence from Canadian market. Accounting and Business Research, 24, Antle, R., Gordon, E., Narayanamoorthy, G., & Zhou, L. (2002). The joint determination of audit fees, non-audit fees, and abnormal accruals. Working Paper, Yale School of Management. Bell, T., Landsman, W., & Shackelford, D. (2001). Auditors' perceived business risk and audit fees: Analysis and evidence. Journal of Accounting Research, 39, Bhushan, R., & Lessard, D. (1992). Coping with international accounting diversity: Fund managers' views on disclosure, reconciliation, and harmonization. Journal of International Financial Management and Accounting, 4, Bonner, S., Palmrose, Z. -V., & Young, S. (1998). Fraud type and auditor litigation: An analysis of SEC Accounting and Auditing Enforcement Releases. The Accounting Review, 73, Chifor, G. (March 2001). Reviewing the current status of multijurisdiction disclosure system. Canadian Securities Law Report & Stock Exchange. Chung, D., & Lindsay, W. (1988). The pricing of audit services: The Canadian perspective. Contemporary Accounting Research, 5, Craswell, A., & Frances, J. (1999). Pricing of initial audit engagements. The Accounting Review, 74, Davis, L., Ricchiute, D., & Trompeter, G. (1993). Audit effort, audit fees, and the provision of non-audit services to audit clients. The Accounting Review, 68, DeAngelo, L. (1981). Auditor size and auditor quality. Journal of Accounting and Economics, 3, DeFond, M., & Francis, J. (2005). Audit research after Sarbanes Oxley.Auditing: A Journal of Practice & Theory, 24, 5 30 Supplement. DeFond, M., Raghunandan, K., & Subramanyam, K. (2002). Do nonaudit services fees impair auditor independence? Evidence from going concern audit opinions. Journal of Accounting Research, 40, 4, Firth, M. (1997). The provision of non-audit services by accounting firms to their audit clients. Contemporary Accounting Research, 14, Frankel, R., Johnson, M., & Nelson, K. (2002). The relation between auditors' fees for nonaudit services and earnings management.the Accounting Review, 77, Supplement. Gul, F. (1999). Audit prices, product differentiation and economic equilibrium. Auditing: A Journal of Practice & Theory, 18, Hausman, A. (1978). Specification tests in econometrics. Econometrica, 46, Houston, C., & Jones, R. (1999). The MultiJurisdictional disclosure system: Model for future cooperation? Journal of International Financial Management and Accounting, 10, Huber, P. J. (1973). Robust regression: Asymptotics, conjectures and Monte Carlo. Annals of Statistics, 1,

15 J.H. Callaghan et al. / The International Journal of Accounting 43 (2008) Kennedy, P. (1992). A guide to econometrics. Cambridge, MA: MIT Press. Palmrose, Z. -V. (1996). The effect of non-audit services on the pricing of audit services: Further evidence. Journal of Accounting Research, 24, Palmrose, Z. -V., & Scholz, S. (2004). The accounting causes and legal consequences of non-gaap reporting: Evidence from restatements. Contemporary Accounting Research, 21, Palmrose, Z. -V., Richardson, V., & Scholz, S. (2004). Determinants of market reactions to restatement announcements. Journal of Accounting & Economics, 37, Parkash, M., & Venable, C. (1993). Auditee incentives for auditor independence: The case of nonaudit services. The Accounting Review, 68, 1, Pincus, K., Holders, W., & Mock, T. (1998). Reducing the incidence of fraudulent financial reporting: The role of the Securities and Exchange Commission. Los Angels: SEC and Financial Reporting Institute of the University of Southern California. Pratt, J., & Stice, J. D. (October 1994). The effect of client characteristics on auditor litigation risk judgments, required audit evidence, and recommended audit fees. The Accounting Review, 69, Puri, P., & Sen, A. (2003). A cost benefit analysis of the multi-jurisdictional disclosure system. Submitted to Ontario Securities Commission. Toronto, Ontario M5S 2V6: Canada. Rousseeuw, P.J., & Van Driessen, K. (submitted for publication). Computing LTS regression for large data sets. Technical Report, University of Antwerp. Rousseeuw, P. J., & Yohai, V. (1984). Robust regression by means of S estimators. In J. Franke, W. Härdle, & R. D. Martin (Eds.), Robust and nonlinear time series analysislecture Notes in Statistics, vol. 26. (pp ) New York: Springer Verlag. Securities and Exchange Commission (SEC) (1991). Multijurisdictional disclosure and modifications to the current registration and reporting system for Canadian issuers. Securities Act Release No Washington, D.C.: U.S. Government Printing Office. Securities and Exchange Commission (SEC) (1993). Amendments to the multijurisdictional disclosue system for Canadian issuers. Securities Act Release No Washington, D.C.: U.S. Government Printing Office. Simunic, D. A. (Spring 1980). The pricing of audit services: Theory and evidence. Journal of Accounting Research, Simunic, D. A. (Autumn 1984). Auditing, consulting, and auditor independence. Journal of Accounting Research, Whisenant, S., Sankaraguruswamy, S., & Raghunandan, K. (2003). Evidence on the joint determination of audit and non-audit fees. Journal of Accounting Research, 4, White, H. (1980). A heteroskedasticity-consistent covariance matrix and a direct test for heteroskedasticity. Econometrica, 48, Yohai, V. J. (1987). High breakdown point and high efficiency robust estimates for regression. Annals of Statistics, 15,