Auditor s Reputation, Equity Offerings, and Firm Size: The Case of Arthur Andersen

Transcription

1 Auditor s Reputation, Equity Offerings, and Firm Size: The Case of Arthur Andersen Stephanie Yates Rauterkus Louisiana State University Kyojik Roy Song University of Louisiana at Lafayette First Draft: November 5, 2002 Current Draft: December 31, 2003 ABSTRACT We study the impact of Arthur Andersen s declining reputation on their clients. When Andersen clients issue seasoned equity, we find that the negative reaction to SEO announcements is two percent worse for SEOs audited by Andersen versus other Big Five firms. A median firm in our sample loses $31.4 million more than a non-andersen client. This result supports the argument that the certifying and monitoring role of auditors is valuable to clients. We do not find any unusual underpricing for SEOs or for IPO firms audited by Arthur Andersen. However, we do find that Andersen clients suffered significant value losses (approximately two percent) surrounding two key events: the admission of error by Andersen s CEO and the announcement of the criminal indictment against Andersen. We find that these results are driven by the large firms in our sample implying that only large firms stock is affected by the deteriorating reputation of Andersen. JEL classification: G30 Keywords: Reputation; Certification; Equity Offering; Firm Size Corresponding author: Kyojik Roy Song Dept. of Economics/Finance University of Louisiana at Lafayette P.O.Box Lafayette, LA sxk8306@louisiana.edu Phone: Fax: We thank participants the University of Michigan at Flint Research Workshop and the 2003 Midwest Finance Association Meetings for their helpful comments.

2 1. Introduction The impact of a service provider s reputation on its clients has been a topic of great interest in the literature. However, the impact of reputational losses of service providers has not received as much attention. Due to irregularities in their audit procedures with respect to Enron, Arthur Andersen recently suffered significant reputational losses that provide a unique opportunity to examine the impact of these losses on their clients. A recent paper published by Chaney and Philipich (2002) also investigates the impact of the Enron audit failure on auditor reputation and specifically the stock price of other Arthur Andersen clients. Our study differentiates itself from the recent work by Chaney and Philipich (2002) in two ways. First, we focus on the certifying and monitoring role of auditors. Myers and Majluf (1984) show that firms issuing equities experience an adverse selection problem due to asymmetric information between managers and investors about firm value. The stock prices of firms issuing equities tend to decline when they announce seasoned equity offerings (SEOs) or experience underpricing at initial public offerings (IPOs). Therefore, firms tend to employ the reputation of financial intermediaries to mitigate this problem in equity offerings. Slovin, Sushka, and Hudson (1990) find that the quality of auditing mitigates the negative stock price reaction to the announcement of a SEO. Also, Titman and Trueman (1986) and Datar, Feltham, and Hughes (1991) develop models in which audit quality reduces IPO underpricing, and Balvers, McDonald, and Miller (1988) and Beatty (1989) find that firms employing a Big Eight auditor experience less underpricing. Because Arthur Andersen loses its reputation over the period of the Enron saga, we expect that the firms audited by Arthur Andersen might be negatively 1

3 affected when they issue equities over the sample period as compared to other firms. Second, we focus on whether the firm size of Arthur Andersen clients affects the magnitude of the stock price decline of the clients. Chaney and Philipich (2002) find that the stock price of Arthur Andersen clients decreases around the Arthur Andersen events over the period, October 2001 to March 2002, due to the reputational losses of Arthur Andersen. They investigate the effect of the events on only large firms. Myers and Majluf (1984) imply that the certifying and monitoring role is of more importance to young firms with more growth opportunities. Accordingly, we can expect that small Andersen clients lose more value. However, Chaney and Philipich (2002) argue that a large firm s managers have the ability and the incentive to manage earnings. Therefore, if large firms have reports that are more complex and less transparent, the market discounts the quality of these reports. They expect that large Andersen clients lose more value. Small firms rely on the reputation of financial intermediaries because they tend to have more asymmetric information. Therefore, we need to investigate whether small firms or large firms are more affected by the events of Enron and Arthur Andersen. We first investigate the effect of the declining reputation of Arthur Andersen on the announcement effect of SEOs. We examine 39 SEOs audited by Arthur Andersen and 124 SEOs audited by other Big Five firms over the period, October 2001 to August We find that, when they announce a SEO, firms audited by Arthur Andersen see a decrease in stock prices that is two percent greater than that experienced by firms audited by other Big Five firms. The strong negative reaction to the SEO for the Andersen clients remains significant after controlling for other determinants of the stock price reaction to the announcements of SEOs. The result is statistically significant, and 2

4 is also economically significant. For a median firm in our sample, the firm loses $31.4 million more when it announces a SEO. In addition, we examine the difference of SEO underpricing between Arthur Andersen clients and other firms at offering dates, and we do not find any significant difference. The results mean that the effect of the reputational loss of an auditor on seasoned equity offering is fully reflected at the announcement of SEOs. Second, we examine IPO underpricing in a sample of firms audited by Arthur Andersen to investigate whether the effect of Arthur Andersen on an equity offering is present for IPO firms. The importance of this sample is also enhanced by the fact that the prospectuses that we reviewed cited the use of Arthur Andersen as an auditor as an additional risk factor in an IPO. We do not find, however, that these firms experience unusual underpricing. Third, we examine a portfolio of firms audited by Arthur Andersen in 2001 and find that it underperforms the market as the information related to Arthur Andersen is gradually revealed after Enron reports a third quarter loss on October 16, The underperformance of Andersen clients is consistent with Chaney and Philipich s (2002) findings. We find that Andersen clients start to lose their value after the first event date, October 16, However, Chaney and Philipich (2002) find that Andersen clients do not lose their value until January, The information on Andersen s auditing failure starts to be revealed in October Our results show that the market reflects the new information on Arthur Andersen right away. Then, we divide Andersen clients into large firms and small firms based on the median of market capitalization of all CRSP (Center for Research in Security Prices) listed firms. We find that a portfolio of 1 See Figure 1 in Chaney and Philipich (2002). 3

5 large firms underperforms the market, but a portfolio of small firms does not underperform the market. This result is consistent with Chaney and Philipich s (2002) argument that the market reaction might be related to the manager s incentive to manipulate the financial statements. The next section describes the recent history of Arthur Andersen and Enron, Section 3 reviews the relevant literature and develops hypotheses, Section 4 describes the data and presents the empirical results, and Section 5 concludes. 2. Enron and Arthur Andersen Enron began in 1985 due to a merger between Houston Natural Gas and InterNorth of Omaha, Nebraska and was designed to create the first nationwide natural gas pipeline system. During the 1990s, Enron entered into utilities trading and became number seven on Fortune magazine s list of the 500 largest firms in the United States. However, during most of the 1990s, Enron executives and directors created a series of off-book partnerships that they used to hide millions of dollars of debt while allowing the executives and directors to make substantial profits. All of these partnerships were approved by Enron s board of directors and reviewed by Arthur Andersen, the firm s external auditor. Arthur Andersen was founded in 1913 by a Northwestern University professor of the same name. In 2001, Andersen was the fifth largest auditing firm in the world (based on worldwide revenue), employing 85,000 people in 84 countries and with reported revenues of $9.3 billion ($4.3 billion in the United States alone) 2. The Appendix details the major events of related to Enron and Arthur 2 See Chaney and Philipich (2002). 4

6 Andersen s demise. After Enron reports a $638 million third quarter loss on October 16, 2001, a series of accounting irregularities related to Enron and Andersen is revealed to the market. Finally, Andersen is barred from conducting business after August Enron experiences the downward path of its stock price throughout 2001 from a high at the end of January of $80 per share to a low at the end of November of $0.26 per share. 3. Literature Review and Hypothesis Development One of the major issues in the literature has been the impact of asymmetric information on security offerings. The pecking order theory, developed by Myers (1984) and Myers and Majluf (1984), is based on the idea that insiders know more about their firm s value and future projects than outside investors, and maximize the wealth of existing shareholders. The insiders or managers avoid issuing equity when they believe the firm is undervalued. Consequently, firms tend to experience an adverse selection problem when they issue equities. That is, because investors have incomplete information, they will not be able to distinguish overvalued firms from undervalued firms. However, because they know that firms avoid issuing equities when they are undervalued, investors assume that firms that issue equities are overvalued. Therefore, firms employ financial intermediaries to mitigate the adverse selection problem in an equity offering. The certification and monitoring role by financial intermediaries at an IPO has been extensively analyzed in the literature 3. The role of auditing firms has been well recognized when firms go to external markets to raise capital. In the IPO market, the higher reputation of auditing firms tends to decrease IPO underpricing for their 3 For example, Beatty and Ritter (1986), Carter and Manaster (1990), and Carter, Dark, and Singh (1998) find that investment banks can certify IPOs. Also, Megginson and Weiss (1991) find that the presence of venture capital reduces IPO underpricing. 5

7 client firms. Titman and Trueman (1986) and Datar, Feltham, and Hughes (1991) develop models in which audit quality reduces IPO underpricing. Empirically, Balvers, McDonald, and Miller (1988) and Beatty (1989) find that firms employing a Big Eight auditor experience less underpricing. However, in the 1990s, it is hard to test the effect of audit quality on IPO underpricing because Big Five firms dominate the auditing market. Also, when firms go public, they tend to switch to Big Five firms. The deteriorating reputation of Arthur Andersen around the Enron saga is a good testing environment for the effect of auditing quality on IPO underpricing. Previous literature has documented negative announcement effects for firms issuing seasoned equity 4. The external monitors can mitigate the adverse selection problem associated with seasoned equity offerings. Slovin, Sushka, and Hudson (1990) find that the stock price reaction to the announcement of a seasoned equity offering is a positive function of the quantity of bank debt, the quality of the firm s investment banker, and the quality of the auditing firm. The result indicates that the deteriorating reputation of Arthur Andersen might exacerbate the negative announcement effects of their client firms. In the accounting literature, DeAngelo s (1981b) definition of audit(or) quality - - the probability that the auditor will both discover and report a breach in the client s accounting system has long been accepted as the industry standard. Further, audit firm size (i.e. Big Five versus non-big Five) is commonly used as a proxy for audit quality. The rationale for this standard is that larger audit firms supply higher quality because they have more to lose than smaller firms with respect to reputation. 5 In light 4 For example, see Asquith and Mullins (1986), Masulis and Kowar (1986), Korajczyk, Lucas, and McDonald (1991), and Choe, Masulis, and Nanda (1993). 5 See DeAngelo (1981b) 6

8 of recent events, a particularly interesting study by Reynolds (2000) finds that economic dependence (within a particular office) does not cause Big Five auditors to report more favorably. In fact, Reynolds findings support the notion that reputation protection dominates auditor behavior. In finance too, the literature ties quality to reputation. In fact, a firm s ability to repeatedly deliver goods and/or services of high quality is the cornerstone of the reputation building process (Klein and Leffler (1981); Shapiro (1983)). Additionally, numerous studies have examined the tradeoff between preserving reputation and the potential gains related to providing a low quality product and found that for firms that expect continued operations, the benefits do not outweigh the costs. 6 Our concern is more specific. We are interested in the effect of reputation when a professional service provider s reputation can impact the market value of their client and the client s stakeholders rely on the service provider to mitigate the asymmetric information problem that has been shown to lead to market failure. 7 Carter and Manaster (1990) find a negative relation between initial public offering (IPO) underpricing and underwriter reputation. More importantly, Carter and Manaster (1990) provide evidence that low dispersion issuers signal their low risk characteristics by engaging prestigious underwriters who, in order to preserve their high reputation, market only IPOs of low dispersion firms. Megginson and Weiss (1990) find that the presence of venture capitalists reduces IPO underpricing in a sample of firms in the period Similarly, the choice of auditor also provides firms with an opportunity to signal their value. Thus, the investment bank, the venture capitalist, and 6 See Maksimovic and Titman (1991) for a detailed theoretical treatment of this topic. 7 See Akerlof (1970) 7

9 the auditing firm (i.e. the service providers) assume both a certification and a monitoring role. The literature based on asymmetric information implies that the certifying and monitoring role is more important to small and young firms with more growth opportunities. It also implies that the reputation of an auditor is more important to IPO firms. However, Chaney and Philipich (2002) argue that large firms are often viewed as having the ability and the incentive to manage earnings. Therefore, if large firms have reports that are more complex and less transparent, the market discounts the quality of these reports. They predict that the deteriorating reputation of Arthur Andersen affects larger firms more severely. As noted previously, prior studies have typically considered the use of a Big Five audit firm as a proxy for a quality audit. 8 However, due to the signaling effect of engaging a Big Five audit firm, the vast majority of large, publicly traded firms in the United States engage Big Five firms. Until now, it has been difficult to test the differences in quality among audit firms and its impact on their clients due to the virtual oligopoly that these large audit firms enjoy and the minimal variation between them. With the series of events surrounding Arthur Andersen and their indictment for fraud in relation to a major audit client, we have the opportunity to examine a service provider that has experienced a reputational loss. The magnitude of this failure and the existence of other concurrent failures greatly reduce the probability that Arthur Andersen would both discover and report a breach in a client s accounting system. By definition, then, Arthur Andersen has exhibited a reduction in quality and therefore reputation. When 8 See Becker et al. (1998), DeAngelo (1981a), Francis and Wilson (1988) and Krishnan and Schauer (2000). 8

10 compared to prior empirical evidence suggesting that such an event is not expected, this loss raises many new empirical questions. We focus on the impact of auditor failure and the subsequent decline in the auditor s reputation on auditee stock performance. We test the relation between an auditing firm s (here, Arthur Andersen s) changing reputation and their client s stock returns. We categorize the clients in three ways: 1) firms issuing seasoned equity; 2) IPO firms; 3) auditing clients in Specifically, we divide 2001 auditing clients into small firms and large firms based on market capitalization to investigate the relation between firm size and auditee stock performance. If auditors reputation is a concern to investors, it may be appropriate to consider the retention of a disreputable auditor as a risk factor particularly for a firm issuing equity. Even some companies mention auditing by Arthur Andersen as a risk factor in the prospectus. In its SEC filing, Inveresk Research Group Inc. said if Arthur Andersen becomes unable to make required representations to us or for any other reason (including the loss of key members of our audit team from Arthur Andersen), Arthur Andersen is unable to provide audit-related services for us in a timely manner. Also, it said that certain investors, including significant funds and institutional investors, may choose not to hold or invest in securities of a company that does not have current financial reports available. We examine whether or not this risk is priced in the market at the time of IPOs. Also, we examine whether the deteriorating reputation of an auditor affects the reaction to the announcements of SEOs or SEO underpricing. In summary, we test three main null hypotheses based on the three categories of audit clients and their related circumstances described above. First, with respect to all firms audited by Arthur Andersen in 2000, we test the null hypothesis: We investigate 9

11 whether the declining reputation of Arthur Andersen affects stock prices of its clients around the announcement of a seasoned equity offering. Specifically we test: H 1 : The announcement effects for Andersen s SEO clients are not different from those for other Big Five s SEO clients. Second, we test the null hypothesis that issuers that engage Arthur Andersen as their auditor do not experience any unusual underpricing. Specifically we test: H 2 : The underpricing for Andersen s IPO clients is not different from that for other Big 5 s IPO clients. Third, we test the null hypothesis that large firms and small firms experience the same stock performance after their auditor lose its reputation. H 3 : The abnormal returns for the small firm portfolio surrounding each event equals those for the large firm portfolio. 4. Sample Selection and Empirical Results 4.1 SAMPLE SELECTION We use three data sets to test our hypotheses in this research. First, we obtain seasoned equity offering (SEO) data from Securities Data Corporation (SDC) Global New Issues Database. We obtain offer prices, proceeds, underwriters, auditors, offering dates, and filing dates for each offering from SDC. We find 280 SEOs over our sample period, October 16, 2001 to August 31, after removing the offerings from closedend funds and financial firms with SIC codes of We remove the offerings of financial firms because financial firms are strictly regulated by government agencies, and the certifying and monitoring role of an auditor is of less importance to them. We check the Business News in Lexis-Nexis to obtain the announcement dates of SEOs. If 9 We use October 16, 2001 and August 31, 2002 as cutoff dates. October 16, 2001 is the first date that the accounting irregularity of Enron is announced. Arthur Andersen stopped its operation in the end of August,

12 we do not find the announcement dates from this source, we use the filing date as the announcement date. We remove 70 firms because they announced the offerings before the starting date of our sample period, October 16, Then, we remove seven firms because they are audited by non-big Five auditors. Also, we remove 36 firms that only offer secondary shares. We match the remaining sample to CRSP to obtain daily return data, and remove three firms because the firms are not listed on CRSP. In this way, we are left with 163 seasoned equity offerings. Second, we also obtain IPO data from SDC. We obtain the data for 73 firms that went public from October 16, 2001 to August 31, 2002 after removing ADRs, closed-end funds, REITs, unit issues, and spinoffs. Out of 73 firms, Arthur Andersen audits 9 firms at the time of IPO. We get offer dates, offer prices, filing price ranges, proceeds, lead underwriter s name, initial returns, the names of the venture capitalists who invested in the firm and the auditor s name for each IPO from SDC. Because there are occasional inaccuracies in SDC IPO initial return data, we verify the initial returns using data from CRSP. To control for underwriter s reputation, we obtain each underwriter s rank from Jay Ritter s web site. Third, we obtain a sample of 1,222 firms audited by Arthur Andersen for the fiscal year of 2000 from Research Insight. We also obtain market value of equity and book value of assets for each firm from Research Insight. After removing 37 ADRs, 118 financial firms with SIC code of , 48 regulated firms with SIC code , and Enron, we are left with 1,018 industrial firms. We remove financial firms and regulated firms because the certifying and monitoring role of an auditor is limited to these firms. Then, we match those firms with the firms in CRSP, and lose 61 firms in the process. We remove the 173 firms that are not traded in the whole year of 11

13 2001 and are not traded by our last event date, March 15, Also, we check auditor changes in 2001 using Audit Trak 10. We remove 17 firms that dismiss Arthur Andersen before our first event date, October 15, Finally, we obtain 768 Arthur Andersen audit clients for the year, This approach is in contrast to Chaney and Philipich (2002) who use only firms included in the S&P This restriction reduces the size of their sample to 287 firms versus our 768. Out of 768 firms, 251 firms are traded on NYSE, 63 firms are traded on AMEX, and 454 firms are traded on NASDAQ. From CRSP, we obtain daily returns for each firm and CRSP value-weighted index returns for 2001 and EMPIRICAL RESULTS We compare 39 SEOs audited by Arthur Andersen and 124 SEOs audited by other Big Five firms. Panel A of TABLE 1 presents the comparison of SEO characteristics between the two sub-samples. It reports the mean offer price, the mean stock price run-up during the estimation period (day 170 to day 21), the mean proceeds, the mean market capitalization on the announcement date, the mean underwriter s ranks, and the mean difference test results for each variable between the two sub-samples. The mean offer price for SEOs audited by Arthur Andersen is $24.13 and the mean offer price for SEOs audited by other Big Five firms $21.71, but these mean offer prices are not significantly different. The stock prices of both samples increase about 40 percent over the estimation period, which means firms tend to issue equities when they are overvalued. The mean proceeds, the mean market capitalization, and the mean of underwriters reputation measures of both samples are not significantly different. 10 We thank Don Deis for providing us with the Audit Trak data. 12

14 We estimate two-day cumulative abnormal returns (CARs) around SEO announcements (over day 0 and day 1) using a single factor market model. We use the CRSP value-weighted index as a market index. Panel B of TABLE 1 reports the mean and median CARs of the two sub-samples, the mean difference test result, and the nonparametric median difference test result. As expected, the firms experience a negative stock price reaction when they announce seasoned equity offerings. The firms audited by Arthur Andersen experience more significant negative reaction ( percent), compared to the reaction ( percent) of other firms. The median CAR for SEOs audited by Arthur Andersen is 4.55 percent and the median CAR for SEOs audited by other Big Five firms is 2.51 percent. The chi-square test statistic (2.71) shows that the median CARs are marginally significantly different. Previous literature finds that stock prices decline two to three percent on average when firms announce SEOs 11. The firms audited by Andersen lose their value much more, compared to other normal SEOs. This result supports our argument that the deteriorating reputation of Arthur Andersen affects equity offerings of their clients negatively. [TABLE 1 about here] TABLE 2 presents the results of multivariate regressions examining two-day cumulative abnormal returns around SEO announcements after controlling for other determinants. The dependent variable is the two-day abnormal returns. Model 1 includes the log of proceeds, stock price run-up, CRSP value weighted index run-up over the day 170 to 21 (market run-up), a dummy variable for SEOs which includes secondary shares, and underwriter s reputation measure as controlling variables. We expect the coefficient on the log of proceeds to be positive because large offerings tend 11 For instance, see Masulis and Korwar (1986). 13

15 to suffer less from adverse selection. The coefficient is negative and insignificant which is not consistent with what we expect. Stock price run-up before the issuance is significantly negatively related to the reaction to the SEO announcement. This result is consistent with Choe et al. s (1993) timing argument that firms with greater stock price rises experience larger negative reaction because firms tend to issue stock when they are overvalued. Also, Choe et al. (1993) find that SEO firms experience less of a negative announcement reaction in up markets than in down markets. The coefficient on market run-up, which controls for market condition, is insignificantly positive. The insignificant result may come from our short sample period, which means the market condition does not change much over the short period. Even though we remove the SEOs consisting of only secondary shares, we have some SEOs that have a portion of secondary shares. The secondary shares might affect the stock price reaction due to inside information. The coefficient on the dummy variable for secondary shares, which controls for inside information, is positive and insignificant. This result is inconsistent with our expectation. Also, the coefficient on the underwriter s reputation measure is negative and insignificant. Our key variable, a dummy variable for auditing by Arthur Andersen, is significantly negative with a t-statistic of The dummy variable indicates that the stocks audited by Arthur Andersen lose two percent more compared to other firms when they announce seasoned equity offerings. In model 2, we use the log of market capitalization on the announcement date as a firm size measure instead of the log of proceeds. The results are qualitatively the same as in model 1. The coefficient on the log of market capitalization is positive and insignificant. [TABLE 2 about here] 14

16 In summary, univariate and multivariate test results support our hypothesis that the deteriorating reputation of Arthur Andersen exacerbates the negative reaction to SEO announcements compared to SEOs audited by other Big Five firms. The results show that the firms audited by Arthur Andersen experience a decrease in stock prices two percent more when they announce SEO compared to the firms audited by other Big Five firms. These results are statistically significant and are also economically significant. For a median firm in the sample, the firm loses $31.4 million more when it announces SEO. TABLE 3 presents univariate test results on SEO underpricing. Corwin (2003) finds that SEO underpricing is positively related to offer size, price uncertainty, and the magnitude of pre-offer returns. We examine whether the deteriorating reputation of Andersen affects the underpricing as well as the announcement reaction. The underpricing is measured as the return of offer price to close price on the offering date. We find that the underpricing of SEOs audited by Andersen is not significantly different from that of other firms. This result indicates that the effect of Andersen s deteriorating reputation on SEOs is fully reflected at the announcements of SEOs. [TABLE 3 about here] To investigate the effect of the declining reputation of Arthur Andersen on equity offering further, we examine IPO underpricing. TABLE 4 compares IPOs audited by Arthur Andersen with IPOs audited by other Big Five firms. Our IPO sample consists of 73 firms going public from October 16, 2001 to August 31, Out of 73 firms, Arthur Andersen audits 9 firms and other Big Five firms audit the other 64 firms. The mean initial return (the ratio of the first day closing price to the offer price) of IPOs audited by Arthur Andersen is percent, which is slightly higher 15

17 than that (11.62 percent) of other IPOs. The mean of offer prices for both samples is close to $15. The mean proceeds ($ million) of IPOs audited by Big Five firms appear to be much higher than those ($ million) of IPOs audited by Arthur Andersen, but the difference is not significant. Also, the mean proceeds of IPOs audited by Big Five firms are highly skewed to the right, and the median proceeds of both samples are not much different. The mean of the standard deviation of daily stock returns of IPOs audited by Arthur Andersen is significantly lower than that of other IPOs, which means that Arthur Andersen has audited less risky firms for the sample period. Even though the clients of Arthur Andersen are less risky, they experience slightly higher underpricing. The adjusted tombstone ranks, which measure underwriter s reputation, are similar for both samples. Basically, the univariate test results show that the two sub-samples have similar characteristics except the standard deviation of daily stock returns. [TABLE 4 about here] TABLE 5 reports multivariate regression results for the IPO sample. In model 1, initial returns are regressed on offer size (natural log of the proceeds), standard deviation of daily stock returns during day +21 to day +120 after IPO, underwriter s reputation (natural log of adjusted tombstone ranks), and a dummy variable for venture capital backed IPOs. The coefficient on offer size is positive and insignificant, which is not consistent with Beatty and Ritter (1986) who argue that a smaller offering is more speculative than a large offering. The coefficient on the risk measure, standard deviation of daily stock returns, is negative and insignificant, even though we expect a positive coefficient. The coefficient on underwriter s reputation is significantly negative, which means that prestigious underwriters tend to mitigate the adverse 16

18 selection problem of IPOs. The dummy variable for a venture capital backed IPO is positively significant, which is not consistent with Megginson and Weiss (1991). In model 2, we consider the partial adjustment phenomenon. Hanley (1993) argues that investment banks can have a strong bargaining position when there is high demand for the issue. We introduce the interaction term AboveRange*UnderwriterReputation, which indicates that the offer price is higher than the initial filing range, and underwriter s reputation measure. As expected, the coefficient on the interaction term is significantly positive. Also, we introduce the interaction term VCBacked*UnderwriterReputation for venture capital backed IPOs and underwriter reputation measure because venture capital backed IPOs tend to employ more prestigious investment banks. The coefficient on the interaction is insignificantly positive, and the coefficient on dummy variable for venture capital backed IPOs is now negative. In model 3, we add our key variable to the model -- a dummy variable which indicates whether an IPO is audited by Arthur Andersen. The coefficient on the dummy variable is positive and insignificant. We expect that the deteriorating reputation of auditor s reputation should have a negative effect on the IPOs of their client firms. However, we don t find any significant result in univariate and multivariate tests. We think that the results might come from our limited sample size, or small and young firms might not be affected by the reputational losses of Andersen. We have only nine IPOs audited by Arthur Andersen in the sample period. [TABLE 5 about here] To investigate the relation between firm size and excess returns for Andersen clients, we divide 768 clients into 310 small firms and 458 large firms using the median 17

19 market capitalization of all CRSP firms as of October 15, Then, we construct a small firm portfolio and a large firm portfolio to use event study methodology. Of the major Enron and Andersen-related events in Appendix that took place from October 2001 through March 2002, we consider those that could have the greatest impact on other Andersen clients. We also review newspaper coverage and the events used in Chaney and Philipich (2002) to ensure that we choose dates that receive the greatest reaction. Specifically, we examine November 8, 2001 when Enron revises its financial statements for the previous five years; December 12, 2001 when the CEO of Arthur Andersen testifies before congress; January 10, 2002 when Andersen announces that they shredded documents; February 3, 2002 when Andersen announces the creation of an independent oversight board and March 15, 2002 when the criminal indictment against Andersen is unsealed 12. We compare firm characteristics between small firms and large firms and present the results in TABLE 6. Firms in a small firm portfolio have much smaller market capitalization and total assets by construction. Specially, we are interested in the market to book ratio of equity and the research and development expense to sales ratio (R&D) 13. Myers and Majluf (1984) show that firms with high growth opportunities have more asymmetric information. These firms need to rely more on the certifying and monitoring role of auditors. We proxy growth opportunities using the market to book ratio of equity and the research and development expense to sales ratio. Interestingly, the mean growth opportunities are not significantly different between the two portfolios. 12 Compared to Chaney and Philipich (2002), we have one more event date, March 15, 2002 when the criminal indictment against Andersen is unsealed. 13 R&D expense to sales ratio is available for only 181 small firms and 246 large firms from Research Insight. 18

20 The leverage of firms in a large firm portfolio is marginally higher, and the mean earnings per share (EPS) are not significantly different. [TABLE 6 about here] We examine the stock price performance of 768 client firms around each event date to confirm Chaney and Philipich s (2002) results. We combine client firms into a single portfolio. Excess returns are the mathematical difference between observed returns and predicted returns. We use a single factor market model to predict marketconditional returns. In this model, portfolio returns are regressed on CRSP valueweighted index returns over a 150 day estimation period from t = -170 to 21. Because there are multiple events in our analysis, the estimation period differs across events. In unreported tests, we use the CRSP equally-weighted index returns as market returns, and obtain qualitatively similar results. Our results do not appear to be driven by the choice of market index. Because the event dates are common for all clients, there is cross-sectional correlation in their respective returns. Forming portfolios of all firms in the sample incorporates any cross-sectional correlation in returns. Chaney and Philipich (2002) recognize that contemporaneous correlation exists, but they try to correct for it by forming portfolios by industry. However, the events affect the firms across each industry and therefore, cross-sectional correlation continues to exist in their industry portfolios. We report two-day CARs (day 1 and day 0) and the corresponding z- statistics in TABLE Andersen clients lost their value by 2.02 percent and 1.75 percent on two event dates, December 12, 2001 (Andersen s CEO admits Andersen made an error at a Congressional testimony) and March 15, 2002 (Criminal indictment against Andersen is unsealed). The results are somewhat different from Chaney and 14 Degree of freedom of z-statistics depends on the number of days in the estimation period. 19

21 Philipich s (2002) results. They find that Andersen clients do not underperform the market until January 10, However, because the information on accounting irregularities related Enron and Andersen began to be revealed to the market starting October 16, 2001, it is reasonable that Andersen clients should start to lose their value before January 10, We report CARs over the period, October 16, 2001 to December 31, 2002 in Figure 1. We find that Andersen clients start to lose their value as the information on accounting irregularities is gradually revealed to market after October 16, 2001, which is different from Chaney and Philipich s (2002) results. It is likely that the different results come from our larger sample size and the event study methodology. [TABLE 7 and Figure 1 about here] We report two-day cumulative abnormal returns of a small firm portfolio and a large firm portfolio, and report difference test results around each event date in TABLE 8. We report CARs based on a single factor market model using the CRSP value weighted index returns or CRSP equally weighted index returns as market index returns. The results are qualitatively similar regardless of the choice of market index, which implies that the results are not driven by the size effect. Consistent with our previous results, we find the significant value loss on two event dates, December 12, 2001 (Andersen s CEO admits Andersen made an Error) and March 15, 2002 (Criminal indictment against Andersen is unsealed). Around December 12, 2001, large firms underperform the market significantly while small firms do not underperform the market. Also, the difference in the CARs between the two portfolios on the event date is statistically significant. These results are also similar around March 15, The results are consistent with Chaney and Philipich s (2002) argument that because large 20

22 firms financial statements tend to be more complex and less transparent, the market discounts the quality of these statements. [TABLE 8 about here] 5. Conclusion In this research, we examine the impact of Arthur Andersen s declining reputation on their clients. When Andersen clients issue seasoned equity, we find that the negative reaction to SEO announcements is two percent worse for SEOs audited by Andersen versus other Big Five firms over the sample period, October 2001 to August The result is statistically and economically significant. For a median firm in our sample, the firm loses $31.4 million more when it announces SEO compared to other firms. However, we do not find any unusual SEO underpricing for Andersen clients, which implies that the effect of the deteriorating reputation of Andersen is fully reflected at the announcements of SEO. We also do not find any unusual underpricing for IPO firms audited by Arthur Andersen. We find that a large firm portfolio audited by Andersen underperforms the market during the period from October 2001 to March 2002 while a small firm portfolio does not underperform the market, which is consistent with Chaney and Philipich s (2002) argument that the market discounts the quality of large firm s more complex and less transparent financial statements. The result that only large firms are affected by Andersen s deteriorating reputation is also consistent with the results from equity offering samples. The auditor s reputational losses affect seasoned equity issues that tend to be offered by large and established firms, but do not affect new equity issues that tend to be offered by small and young firms. 21

23 REFERENCES Akerlof, G The Market for 'Lemons': Quality Uncertainty and the Market Mechanism. The Quarterly Journal of Economics 84: Asquith, P., and D. W. Jr. Mullins Equity Issues and Offering Dilution. Journal of Financial Economics 15: Balvers, R., B. McDonald, and R. Miller Underpricing of New Issues and the Choice of Auditor as a Signal of Investment Banker Reputation. The Accounting Review 63: Beatty, R. P Auditor Reputation and the Pricing of Initial Public Offerings. The Accounting Review 64: Beatty, R. P., and J. R. Ritter Investment Banking, Reputation and the Underpricing of Initial Public Offerings. Journal of Financial Economics 15: Becker, C. L., M. L. Defond, J. Jiambalvo, and K. R. Subramanyam The Effect of Audit Quality on Earnings Management. Contemporary Accounting Research 15, no. 1: Carter, R., F. Dark, and A. Singh Underwriter Reputation, Initial Returns and the Long-Run Performance of IPO Stocks. Journal of Finance 53: Carter, R., and S. Manaster Initial Public Offerings and Underwriter Reputation. Journal of Finance 45, no. 4: Chaney, P. K., and K. L. Philipich Shredded Reputation: The Cost of Audit Failure. Journal of Accounting Research 40, no. 4: Choe, H., R. W. Masulis, and V. Nanda Common Stock Offerings Across the Business Cycle: Theory and Evidence. Journal of Empirical Finance 1: Corwin, S. A., 2003, The determinants of underpricing for seasoned equity offers, Journal of Finance 58, no. 5: Datar, S., G. Feltham, and J. Hughes The Role of Audits and Audit Quality in Valuing New Issues. Journal of Accounting and Economics 14: DeAngelo, L. E a. Auditor Independence, 'Low Balling', and Disclosure Regulation. Journal of Accounting and Economics 3: b. Auditor Size and Audit Quality. Journal of Accounting and Economics 3:

24 Francis, J. R., and E. R. Wilson Auditor Changes - a Joint Test of Theories Relating to Agency Costs and Auditor Differentiation. Accounting Review 63, no. 4: Hanley, K. W The Unerpricing of Initial Public Offerings and the Partial Adjustment Phenomenon. Journal of Financial Economics 34: Klein, B., and K. B. Leffler The Role of Market Forces in Assuring Contractual Performance. Journal of Political Economy 89, no. 4: Korajczyk, R. A., D. J. Lucas, and Robert L. McDonald The Effect of Information Releases on the Pricing and Timing of Equity Issues. Review of Financial Studies 4: Krishnan, J., and P. C. Schauer The Differentiation of Quality Among Auditors: Evidence From the Not-For-Profit Sector. Auditing: A Journal of Practice and Theory 19, no. 2: Maksimovic, V., and S. Titman Financial Policy and Reputation for Product Quality. Review of Financial Studies 4, no. 4: Masulis, R. W., and A. N. Korwar Seasoned Equity Offering: An Empirical Investigation. Journal of Financial Economics 15: Megginson, W. L., and K. A. Weiss Venture Capitalist Certification in Initial Public Offerings. Journal of Finance 46, no. 3: Myers, S. C The Capital Structure Puzzle. Journal of Finance 39, no. 3: Myers, S. C., and N. S. Majluf Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have. Journal of Financial Economics 13: Reynolds, J. K Does size matter? The influence of large clients on office-level auditor reporting decisions. Journal of Accounting and Economics 30, no. 3: 375. Shapiro, C Premiums for High Quality Products as Returns to Reputation. The Quarterly Journal of Economics 98: Titman, S. and B.Trueman Information Quality and the Valuation of New Issues, Journal of Accounting and Economics 8, White, H A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica 48:

25 TABLE 1 Univariate test results on SEO sample Panel A presents the number of seasoned equity offerings audited by Arthur Andersen and other Big Five firms during 10/16/2001-8/31/2002, the mean offer price, the mean stock price run-up during the estimation period (day 170 to 21), the mean proceeds (in million dollars), the mean market capitalization (in million dollars) on the announcement date, and the mean of adjusted tombstone ranks of underwriters. Also, it shows the test results about whether the means of each variable between twosubsamples are different from each other. Panel B presents the mean and median of two-day cumulative abnormal returns of SEOs audited by Arthur Andersen and other Big Five firms, and the mean difference test result and the nonparametric median difference test result. Panel A. Difference tests on SEO characteristics N Offer price Stock price run-up Proceeds Market capitalization Underwriter Reputation SEOs audited by Arthur Andersen SEOs audited by other Big 5 firms Difference tests (t-statistics) % % Panel B. Difference tests on two-day cumulative abnormal returns around announcement dates of SEOs N Two-day CARs Mean Median SEOs audited by Arthur Andersen SEOs audited by other Big 5 firms % % % % Difference tests t = -2.22** χ 2 = 2.71* *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels respectively. 24

26 TABLE 2 The effect of the deteriorating reputation of Arthur Andersen on abnormal returns of SEOs The table shows whether the deteriorating reputation of Arthur Andersen influences the announcement effects of SEOs over the period, 10/16/2001-8/30/2002. The dependent variable is the two-day cumulative abnormal returns (day 0 and day 1), which are measured by a single factor market model. Log of proceeds indicates the natural log of proceeds from the SEO, and log of market capitalization indicates the log of market capitalization on the announcement date. Stock price run-up is the holding period return of each stock during the estimation period (day 170 to 21), and market run-up is the holding period return of CRSP value-weighted index during the estimation period. Dummy for secondary shares is a qualitative variable indicating that the SEO includes some portion of secondary shares among equities offered. Underwriter s reputation is measured as the adjusted tombstone ranks. Dummy for auditing by Arthur Andersen is a qualitative variable indicating that SEO is audited by Arthur Andersen. T-statistics based on White s (1980) consistent covariance are given in parentheses. Expected Sign Model 1 Model 2 Intercept.15 (1.27).001 (.02) Log of proceeds (+) (-.82) Log of market capitalization (+).004 (.97) Stock price run-up (-) -.01** (-2.04) Market run-up (+).04 (.64) Dummy for secondary shares (-).0003 (.03) Underwriter s reputation (+) -03 (-1.10) -.01** (-2.01).02 (.42).0009 (.09) -.05 (-1.59) Dummy for auditing by Arthur Andersen (-) -.02** (-2.18) -.02** (-2.26) N Adjusted R *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels respectively. 25

27 TABLE 3 Univariate test results on SEO underpricing The table presents the univariate test results on seasoned equity offerings (SEOs) underpricing between SEOs audited by Arthur Andersen and SEOs audited by other Big Five firms over the period, October 16, 2001 to August 31, SEO underpricing is measured as the return of an offering price to the closing price on offering date. The table presents the number of SEOs, the mean and median underpricing, the mean difference test results, and the median difference test results. N SEO underpricing Mean Median SEOs audited by Arthur Andersen SEOs audited by other Big 5 firms % 1.61% % 1.89% Difference tests t = χ 2 =

28 TABLE 4 Univariate Test Results on IPO Sample This table shows the number of IPOs audited by Arthur Andersen and by other Big Five firms, the mean offering price, the mean proceeds from IPOs (in million dollars), the mean standard deviation of daily stock returns during +21 through +120 trading days after IPO, and the mean of adjusted tombstone ranks of underwriters. Also, the table presents t-test results showing whether the means of each variable between two sub-samples are different. IPOs audited by Arthur Andersen IPOs audited by other Big 5 firms Difference tests (t-statistics) N Initial returns Offer price Proceeds Standard deviation Underwriter reputation % % % % *** *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels respectively. 27