Equity value implications of the SEC Exchange Act Rule 13a-14: a litigation cost perspective

1. Introduction

The Securities and Exchange Commission (SEC), on 12 June 2002, proposed that chief executive officers (CEOs) and chief financial officers (CFOs) of firms must certify their financial reports under oath (SEC, 2002a). The SEC enacted the proposal on 27 June 2002 (Exchange Act Rule 13a-14, SEC, 2002b) with one notable difference from the previously released proposal. The new Act Rule 13a-14 required only firms with revenues in excess of $1.2 billion (hereafter LRGREVs) to submit their sworn certification statements to the SEC, and relieved firms with revenues of $1.2 billion or less (hereafter SMLREVs) from compliance with certification requirement. CEOs and CFOs of LRGREVs were required to file sworn statements certifying, among others, the accuracy of their financial reports. Although firms’ executives have always been required to sign off their financial reports, the Exchange Act Rule 13a-14 required them to specifically certify that (1) the executive officers have read the reports; (2) to the best of their knowledge the information in the report is true in all important respects as of the last day of the period covered by the report; and (3) the reports contain all information about the company which is important to a reasonable investor for making investment decisions. These new requirements, specially requiring executives to certify they have disclosed all information important to a reasonable investor for making investment decisions, can have costly implications in the context of shareholders’ class action lawsuits. Prior to Rule 13a-14, certifying financial reports which subsequently proved to be incorrect could have been defended by executives on the basis that they have not acted with requisite scienter. However, under Rule13a-14, senior executives no longer have recourse to this defense and plaintiffs could file claims based on the premise that certification represented a materially misleading assurance.

The SEC contention that the sworn testimonies of company executives will restore investors’ confidence in the firms’ financial reports has motivated studies of stock price implications of such regulatory changes (e.g., Bhattacharya et al., 2007; Chang et al., 2006; Griffin and Lont, 2005; Hirtle, 2006; Wilkinson and Clements, 2006). Prior studies that have focused on the economic consequences of Rule 13a-14 have, however, primarily focused on the stock price reactions surrounding the submission dates in August 2002, when the sworn testimonies were filed with the SEC with one study also examining 27 June 2002, when the SEC enacted Rule 13a-14. For example, Chang et al. (2006) examine stock price reactions of 590 LRGREVs in response to the submission of certification statements to the SEC in August 2002 and report positive abnormal returns on the submission dates. They attribute the results to the increased assurance provided to investors by the certification process. Similarly, Hirtle (2006) reports positive abnormal returns on the submission dates in August 2002, for a sample of 42 LRGREVs bank-holding companies. Bhattacharya et al. (2007), on the other hand, examine stock prices surrounding 6 June and 27 June. On 6 June 2002, the New York Stock Exchange (NYSE) proposed several changes to the corporate governance rules, including certification of financial reports by CEOs of NYSE-listed firms. It is possible that the finding of no results on 6 June 2002 by Bhattacharya et al. (2007) is due to the stock price effects of other corporate governance news announced simultaneously by the NYSE. Furthermore, because the 27 June ruling offered no surprise information for LRGREVs over and above what was already available on 12 June when the SEC proposal on certification was released, it is not surprising that prior studies have not observed any price effect on this date. Other studies such as Griffin and Lont (2005) and Wilkinson and Clements (2006) have examined the stock prices surrounding the submission dates in August 2002, and have not observed stock price reactions on the filing dates.

Our study is motivated by the potential role that litigation risk can play in the stock price behavior of firms exposed to the proposal and final ruling of the SEC Rule 13a-14. Prior studies provide evidence of litigation risk in various settings. For example, Jones and Wu (2010) have examined the effects of executive compensation and potential for earnings management on the incidence of shareholder class action lawsuits, suggesting that high earnings management increases settlement amounts for security class action lawsuits. Cheng et al. (2010) have investigated the effectiveness of securities litigation as a disciplining tool for institutional owners and find that securities class actions with institutional owners as lead plaintiffs are less likely to be dismissed and have larger monetary settlements. Gande and Lewis (2009) have examined the stock price reactions to shareholder-initiated class action lawsuits and observe that shareholders partially anticipate these lawsuits based on lawsuit filings against other firms in the same industry and capitalize part of these losses prior to a lawsuit filing date. Finally, Ferris et al. (2007) have examined board changes surrounding the filings of shareholder derivative lawsuits and find that derivative lawsuits are associated with significant improvements in the boards of directors and can serve as an effective corporate governance mechanism.

Our study examines both the dates 12 June and 27 June which are crucial in evaluating the investors’ perception of Rule 13a-14. This is because the first proposal released on 12 June did not specify which firms will be covered by the rule, which implied that the new rule would apply to all firms. The new proposal released on 27 June 2002, however, unexpectedly relieved firms with revenues of less than $1.2 billion. Examination of stock prices on both dates allows more specified tests of investors’ perception of Rule 13a-14. Also, the litigation cost aspect of the SEC ruling that we examine in this study has not been addressed in prior research.

We conduct a comprehensive search of the Wall Street Journal Index from the beginning of 2002 and find that the SEC proposal on 12 June 2002 (SEC, Press Release (2002-88), Exchange Act Rule 13a-14) was indeed the very first SEC announcement of Rule 13a-14 which was followed by the SEC order on 27 June 2002. Additionally, we found overwhelming evidence in the press that SEC Rule 13a-14 increases executives’ exposure to the shareholders’ litigation. Firm executives are covered by directors and officers (D&O) indemnity insurance, which is usually purchased and owned by the firm (Core, 1997; Parry and Parry, 1991). Accordingly, it is plausible to expect that any lawsuits against individual executives, if successful, will be borne by the firm’s insurer. Such lawsuits can consequently lead to an increase in the firm’s D&O insurance premium affecting the firm’s stock price. Even academics involved in authoring sections of the Sarbanes–Oxley Act (2002) believe that certification of financial reports by executives increase potential liabilities under civil and criminal prosecution. These statements look even more realistic when shareholders’ class action lawsuits grew in popularity in 2002 with a 31 percent increase in the number of firms facing shareholders’ litigation (WSJ, 2003, March 18). The threat of shareholders’ class action lawsuits was also known to the SEC when in June 1999 the SEC withdrew, due to an increased threat of litigation facing firms, one of the controversial requirements of the Blue Ribbon Committee that the audit committee members should express in writing that financial reports comply with generally accepted accounting principles (WSJ, 1999, 14 July). Hence, our examination of stock price behavior on 12 June and 27 June 2002 focuses on litigation cost implications for firms. Litigation costs have been examined in various settings, other than Rule 13a-14, including auditor decision-making (Geiger et al., 2006), equity offering (Li, 2009), and the use of forecast-based accounting (Ijiri, 2005). In addition to examining the litigation cost implications for the stock prices of LRGREVs, we investigate the comparative stock price reactions of SMLREVs. Evaluating the stock price reactions of SMLREVs on 27 June could also yield useful inferences about the litigation cost implications of the certification requirement, by expecting reversal abnormal returns for firms with $1.2 billion or less in revenues on 27 June 2002.

We provide evidence in support of increased costs due to the certification requirement, which is more pronounced for firms operating in the high-litigation-risk (hereafter HLR) industries. After controlling for other variables in a cross-sectional analysis, we find that the market value of firms operating in the HLR industries responds more significantly and negatively to the SEC’s proposal on 12 June 2002. Given that the SEC intended for all firms to comply with this requirement on 12 June 2002, we do not find, as expected, any differences between the market reaction of LRGREVs and SMLREVs in the HLR industries. Furthermore, we document that firms relieved from the certification requirement on 27 June 2002 (SMLREVs) experienced significant reversal (positive) abnormal returns relative to the abnormal returns of firms that remained under Rule 13a-14 (LRGREVs). In a more refined test we find that the market value of SMLREVs operating in the HLR industries reacts more positively to the news of the exemption than the market value of SMLREVs operating in the low-litigation-risk (hereafter LLR) Industries.

The remainder of this study is organized as follows. In Section 2 we provide a background on Exchange Act Rule 13a-14. Section 3 describes the study’s hypotheses. The sample selection is explained in Section 4. Portfolio return and cross-sectional models are discussed in Sections 5 and 6, respectively. Section 7 discusses descriptive statistics on abnormal returns and Section 8 reports the regression results. A sensitivity analysis is provided in Section 9 and Section 10 concludes the study.

2. The Exchange Act Rule 13a-14

The failure of high profile corporations due to accounting fraud have resulted in a decline in investor confidence which led to the President’s announcement of the ‘10-point plan’ on 7 March 2002 for improvement of corporate responsibility and the protection of shareholders’ interests (WSJ, 2002b, 7 March). One of the points in the plan referred to CEOs’ and CFOs’ responsibility to personally vouch for the accuracy and fairness of their firms’ public disclosures, including financial reports.

Following the president’s announcement, the SEC proposed Rule 13a-14 on 12 June 2002, which required a company’s CEO and CFO to certify the contents of the company’s quarterly and annual reports (SEC, 2002a). The proposed rule, consistent with a key provision of the President’s 10-point plan announced on 7 March 2002, required the CEO and CFO of a company to certify, with respect to the company’s quarterly and annual reports, that the executive officers have read the reports, to the best of their knowledge the information in the report is true in all important respects as of the last day of the period covered by the report, and the reports contain all information about the company which is important to a reasonable investor for making investment decisions. The SEC’s proposal on 12 June, however, did not specify which firms would have to comply with the certification requirement, implying that the rule would apply to all listed firms.

On 27 June 2002, pursuant to Section 21(a)(1) of the Securities Exchange Act, the SEC ordered (Exchange Act Rule 13a-14) that the CEOs and CFOs of firms with revenues in excess of $1.2 billion (LRGREVs) shall file a statement in writing, certifying the accuracy of their financial reports (SEC, 2002b). In contrast to prior expectation, the SEC Act relieved firms with revenues of $1.2 billion or less (SMLREVs) from complying with the certification requirement. Rule 13a-14 ordered the top executive and finance officers of 947 large firms (LRGREVs) to certify that their financial reports are accurate and provide all information needed by a reasonable investor for investment decisions. Due to variation in the fiscal-year-ends only 745 (out of the 947 firms identified by the SEC) had to comply with the due date of 14 August 2002, and the deadline for the remaining firms was set at 28 September 2002.

3. Hypotheses

Although the central objective of the certification requirement was to reduce the executives’ incentive to commit fraud and create more transparency, structure and accountability in the financial reporting process, critics have questioned the incremental value of certifying financial reports and have instead argued that the certification process is likely to instigate increased litigation (Dantzler et al., 2003).^1,2 For example, Feldman (2002) states that the certification provisions substantially enhance the civil, private, and governmental exposure of senior executives. However, the immediate source of litigation concern arises from the SEC’s decision to review the annual reports of all Fortune 500 companies, following the collapse of Enron. It seemed unlikely that the SEC would be able to complete its review prior to the 14 August deadline for the certification of financial reports, and to make matters worse, the SEC predicted a ’triple digit’ number of the scrutinized companies that would get comment letters that could potentially result in a restatement of earnings (WSJ, 2002a, 16 July). ³ David Smith, president of the American Society of Corporate Secretaries, claimed that this meant executives were certifying something that was still under review by the SEC without knowing the outcome of the review, which could lead to some serious legal liability issues. This view was also echoed by several members of legal firms including William S Lerach, a partner in the law firm Milberg Weiss Bershad Hynes & Lerach, who claimed that restatement by certifying firms is an admission that financial reports were materially false, which gives any securities case an enormous boost. According to a report by Huron Consulting Group LLC, a Chicago firm that specializes in financial advisory services, a rise in a securities fraud suits in 2002 coincides with an increase in financial restatements by public companies (WSJ, 2003, 18 March). Insurers of D&O coverage also argued that the certification requirement would open Pandora’s Box so as to increase litigation (Wojcik, 2002).

The SEC also acknowledged that a CEO or CFO providing a false certification could result in SEC enforcement actions for securities law violations and SEC and private actions for violating Section 10(b) of the Exchange Act and Rule 10b-5 (Bryan Cave LLP, 2002; Goodwin Procter LLP, 2003). ⁴ Moreover, Dantzler et al. (2003) argues that the likelihood of executives mounting successful defenses in such lawsuits has decreased following the introduction of the certification requirement. Prior to introduction of this, senior executive officers of a firm could easily argue that, even though they signed an inaccurate report, they did not act with requisite scienter to give rise to liability for securities fraud. However, with the introduction of the certification requirement senior executives no longer have recourse to this defense, and plaintiffs could file claims based on the premise that the certification represented a materially misleading ‘assurance’. ⁵ Field et al. (2005) argue that securities lawsuits are costly because they divert management time away from more productive activities, and involve substantial attorney fees and damages to the reputation of the company and its management, potentially affecting stock prices. In support, Gande and Lewis (2009) provide evidence that the effect of potential lawsuits is factored into the stock prices ahead of the lawsuit filing date. Furthermore, successful lawsuits against individual executives can affect stock prices through increased D&O insurance premiums. ⁶

The litigation cost hypothesis predicts that any negative or positive stock price effect will be more profound for those firms that are more exposed to class action law suits (for example, see Johnson et al., 2001 and Ali and Kallapur, 2002). Consistent with the litigation-cost argument we formulate the following hypothesis on the date the SEC’s proposal was released (12 June 2002). ⁷

Upon the release of Act Rule 13a-14 on 27 June 2002, the certification of financial reports was mandated for LRGREVs, while the remaining firms were exempted from compliance with the certification requirement. In the context of the litigation risk argument, we expect the release of the SEC rule on 27 June 2002 to be associated with positive abnormal returns for the portfolio of firms that were relieved from compliance with the SEC new rule (SMLREVs) because of a reversal of previously negative abnormal returns associated with the SEC proposal on 12 June 2002. Furthermore, we expect that, within the SMLREV portfolio, a sub-portfolio of firms in the HLR industries will experience an incremental positive stock price reaction over the sub-portfolio of firms operating in the LLR industries. ⁸

4. Data

Our initial sample consists of 8065 firms covered by the CRSP database, during the year 2002. We identify 7328 firms for which we can calculate abnormal returns in the three days surrounding 12 June and 27 June 2002. We then require the sample firms to have coverage on the Compustat industrial files, which reduces our sample to 6870 firms. Furthermore, we exclude 1819 firms with unavailable or missing Compustat data for the 2002 year. Finally, we eliminate 127 firms that had confounding news (e.g. earnings and dividend announcement, ex-dividend and other firm-specific news) disclosed in a 3-day period surrounding 12 June and 47 firms with confounding news in a 3-day period surrounding 27 June 2002. The final sample consists of 4668 firms for the analysis on 12 June 2002 and 4645 firms for the analysis on 27 June 2002. As discussed earlier, our analysis requires the partitioning of the sample firms into portfolios of LRGREVs, which were subject to the SEC Rule 13a-14 on 27 June 2002, and the SMLREVs which were relieved from the certification requirement on the ruling date (SMLREVs). Our final sample for 12 June consists of 620 (4048) firms in the portfolio of LRGREVs (SMLREVs), and for 27 June consists of 610 (4035) firms in the portfolio of LRGREVs (SMLREVs).

5. Methodology

5.1. Portfolio returns on 12 and 27 June 2002

Given clustering of events on 12 and 27 June, we employ, similar to prior research (e.g., Chang et al., 2006), a portfolio approach to overcome any potential cross-correlation problem as addressed in Bernard (1987). More specifically, we use the portfolio approach suggested by Sefcik and Thompson (1986) to test stock price reactions on 12 and 27 June: ⁹

A R p t = α p + β p R m t + ∑ j = 1 J = 2 γ p j D j t + ε p t

(1)

where:

AR_pt = the return on portfolio p,

α_p = the intercept term,

β_p = the market return slope coefficient,

R_mt = the market return on the CRSP equally weighted NYSE, AMEX and NASDAQ, depending on which stock exchange the firm is listed on,

γ_pj = the abnormal return of the j_th event days t₋₁,t₀,t₊₁, surrounding 12 June (or 27 June), with t₀ equal to the event day,

D_jt = a dummy variable coded as 1 if the observation falls in the three days surrounding 12 or 27 June, and coded 0 for all other days, and

ε_pt = the error term.

We estimate Equation (1) for various portfolios over a period of up to 250 days (but not less than 100 days) through to 10 days before each event. ¹⁰ We exclude five days surrounding 12 June 2002, when estimating the model for 27 June 2002. This procedure ensures that the estimates of the model parameters for testing 27 June are not influenced by the 12 June event.

We then partition the full sample into two portfolios of LRGREVs (required to provide sworn testimony) and SMLREVs (not required to provide sworn testimony) to examine the cumulative abnormal returns for each event date. Furthermore, we partition the full sample into two portfolios based on firm-litigation exposure (HLR and LLR firms) to test the abnormal returns of firms exposed to high-litigation versus those exposed to low-litigation risks. Finally, in a refined test, we create four portfolios of (1) SMLREVs operating in the HLR industry, (2) SMLREVs operating in the LLR industry, (3) LRGREVs operating in the HLR industry, and (4) LRGREVs operating in the LLR industry, to compare the abnormal returns given litigation cost effects.

5.2. Cross-sectional analysis

We employ two sets of multiple regression models to assess the variations in cumulative abnormal returns surrounding the SEC’s proposal and Rule 13a-14. The test variables in the models are REVS, LIT, and REVS×LIT and the remaining variables are used as controls and they are similar to those used in prior research (e.g., Chang et al., 2006). We use REVS as a categorical variable to distinguish between firms with revenues of $1.2 billion or less (REV = 1) that ultimately were relieved from the certification requirement on 27 June (SMLREVs) and firms with revenues of more than $1.2 billion (REV = 0) that remained subject to the certification requirement (LRGREVs). The second test variable (LIT) is also a categorical variable representing firms in the HLR industries (LIT = 1), relative to the LLR industries (LIT = 0). The third test variable (REVS×LIT) captures the effect of the SEC’s proposal and ruling for HLR and LLR firms with more (or less) than $1.2 billion revenues. Specific details on how these test variables are employed to test the hypotheses are discussed later in the paper. The cross-sectional models are as follows:

3 - D a y C A R 12 o r 3 - D a y C A R 27 = β 0 + β 1 R E V S + β 2 L I T + β 3 R E V S × L I T + β 4 S A L E S _ G R O W T H + β 5 A N D E R S E N + β 6 I N V E S T I G A T E + β 7 B I G _ N + β 8 M K T _ T O _ B K + β 9 L E V E R A G E + β 10 R E G U L A T E D + β 11 F I N A N C I A L + μ

(2)

where:

3-DayCAR12 = cumulative abnormal return of firm i in t₋₁,t₀,t₊₁ (t₀ = 12 June 2002), when the SEC released its initial proposal,

3-DayCAR27 = cumulative abnormal return of firm i in t₋₁,t₀,t₊₁ (t₀ = 27 June 2002), when the SEC released Rule 13a-14,

REVS = a categorical variable coded as 1 if firm i was not required to provide sworn testimony (revenues of $1.2 billion or less), as specified by the SEC on 27 June 2002, and 0 otherwise,

LIT = categorical variable coded as 1 if firm i’s SIC is 2833–2836, 3570–3577, 7370–7374, 3600–3674, 5200–5961, 8731–8734 (HLR industries), and 0 otherwise,

SALES_GROWTH = percentage change in sales revenue of firm i in 2001,

ANDERSEN = a categorical variable coded as 1 if firm i is a client of Arthur Andersen in 2001, and 0 otherwise,

INVESTIGATE = a categorical variable coded as 1 if firm i is under investigation by the SEC or Department of Justice, and 0 otherwise,

BIG_N = a categorical variable coded as 1 if firm i is a client of a big 5 auditor (excluding Andersen) in 2001, and 0 otherwise,

MKT_TO_BK = market to book equity ratio of firm i in 2001,

LEVERAGE = long-term debt to total assets ratio of firm i in 2001,

REGULATED = a categorical variable coded as 1 if firm i’s SIC is 4000–4999 (utility industry), and 0 otherwise, and

FINANCIAL = a categorical variable coded as 1 if firm i’s SIC is 6000–6999 (financial industry), and 0 otherwise.

6. Variable measurement

6.1. Dependent variable

Individual firm abnormal returns on 12 and 27 June 2002

For the purpose of regression analysis we require firm-specific abnormal returns to be used as the dependent variable. We determine firm-specific abnormal returns from t₋₁,t₀,t₊₁ surrounding 12 and 27 June by using the standard event study approach pioneered by Fama et al. (1969). The standard market model, as opposed to the portfolio return model used in the first part of this study, allows estimation of the abnormal return for each firm:

R i t = α 0 + β 1 R m t + ν i t

(3)

where

R_it = security return for firm i on day t (12 or 27 June 2002),

R_mt = market return on the CRSP equally weighted NYSE, AMEX and NASDAQ stocks on day t,

α₀ and β₀ = Ordinary Least Square (OLS) coefficients, and

v_it = the disturbance term (residual).

We estimate the market model for each event using daily and three-day stock and market return data over an estimation period similar to that defined earlier in the paper under the portfolio approach. Abnormal returns (AR_it) are then computed for each firm on an event date by using the equation

A R i t = R i t − ( a 0 + b 1 R m t ​ )

(4)

where

AR_it = abnormal return for firm i on day t, a₀ and b₁ are the OLS estimates of market model parameters for firm i.

Daily abnormal returns are accumulated to arrive at the three-day cumulative abnormal return as follows:

3 - D a y C A R i = 1 / N ∑ t = 1 3 A R i

(5)

3-DayCAR_i is then used as the dependent variable in the cross-sectional model (2).

7. Descriptive statistics

Table 1 presents descriptive statistics for the independent variables employed in the regression analysis.

OPEN IN VIEWER

Table 1.

Descriptive statistics of independent variables used in the regression analysis. The sample consists of 4668 firms.

	Mean	Median	Std. Dev.	Min	Max	HLR vs LLR #
LIT	0.2828	0.0000	0.4504	0.0000	1.0000	-
REVS	0.8672	1.0000	0.3394	0.0000	1.0000	-
SALES_GROWTH	0.1207	0.0262	0.5694	-0.7602	3.7644	sig at 5%
ANDERSEN	0.1699	0.0000	0.3756	0.0000	1.0000	-
INVESTIGATE	0.0002	0.0000	0.0146	0.0000	1.0000	-
BIG_N	0.8781	1.0000	0.3272	0.0000	1.0000	-
MKT_TO_BK	2.7373	1.7300	3.2450	0.0239	21.120	sig at 1%
LEVERAGE	0.1676	0.1019	0.1853	0.0000	0.9286	n/s
REGULATED	0.0932	0.0000	0.2907	0.0000	1.0000	-
FINANCIAL	0.1127	0.0000	0.3162	0.0000	1.0000	-

#

The significant levels are reported only between continuous variables across HLR and LLR firms.

LIT = categorical variable coded as 1 if firm i’s SIC is 2833–2836, 3570–3577, 7370–7374, 3600–3674, 5200–5961, 8731–8734 (HLR industries), and 0 otherwise,

REVS = a categorical variable coded as 1 if firm i was not required to provide sworn testimony (revenues of $1.2 billion or less), as specified by the SEC on 27 June 2002, and 0 otherwise,

SALES_GROWTH = percentage change in sales revenue of firm i in 2001,

ANDERSEN = a categorical variable coded as 1 if firm i is a client of Arthur Andersen in 2001, and 0 otherwise,

INVESTIGATE = a categorical variable coded as 1 if firm i is under investigation by the SEC or Department of Justice, and 0 otherwise,

BIG_N = a categorical variable coded as 1 if firm i is a client of a big 5 auditor (excluding Andersen) in 2001, and 0 otherwise,

MKT_TO_BK = market to book equity ratio of firm i in 2001,

LEVERAGE = long-term debt to total assets ratio of firm i in 2001,

REGULATED = a categorical variable coded as 1 if firm i’s SIC is 4000-4999 (utility industry), and 0 otherwise, and

FINANCIAL = a categorical variable coded as 1 if firm i’s SIC is 6000-6999 (financial industry), and 0 otherwise.

The summary statistics indicate that 28.28 percent of the sample firms operate in industries that have high exposure to litigation risk, and 86.72 percent of the sample firms are those exempted, by the SEC, from complying with the certification requirement. The mean sales growth of the sample firms is 12.07 percent. A large proportion of the sample firms (87.81 percent) employ a brand name auditor (BIG-N, excluding Andersen), while approximately 17 percent of the sample firms employed Arthur Andersen as their external auditor. The summary statistics also indicate that a very small proportion of the sample firms (0.02 percent) were under investigation by the SEC or Department of Justice. Market to book ratio and leverage variables have means of 2.7373 and 0.1676, respectively. Finally, approximately 9.32 (11.27) percent of the sample firms operate in the regulated (financial) industries.

Panel A of Table 2 reports the portfolio returns of SMLREVs (Portfolio A, REVS = 1) and LRGREVs (Portfolio B, REVS = 0) in the three days surrounding 12 June 2002.

OPEN IN VIEWER

Table 2.

Descriptive statistics on portfolio returns in the three days surrounding the SEC’s proposal on 12 June 2002

Panel A: Portfolio returns of full sample and sub-samples of small revenue firms not required to provide sworn testimony (REVS = 1) and large revenue firms required to provide sworn testimony (REVS = 0)
	Full sample	Portfolio [A] (REVS =1) (<$1.2 billion)		Portfolio [B] (REVS = 0) (>$1.2 billion)
3-DayCAR (p-value) [Firms]	−0.0036 (0.0001) [4668]	−0.0037 (0.0001) [4048]		−0.0033 (0.0001) [620]
Two-sample test of differences:
Portfolios [A] vs [B] (p-value = 0.6520)
Panel B: Portfolio returns of full sample and sub-samples of high litigation (LIT = 1) and low litigation (LIT = 0) risk firms
	Full sample	Portfolio [C] (LIT =1) HLR		Portfolio [D] (LIT =0) LLR
3-DayCAR (p-value) [Firms]	−0.0036 (0.0001) [4668]	−0.0057 (0.0001) [1320]		−0.0028 (0.0001) [3348]
Two-sample test of differences:
Portfolios [C] vs [D] (p-value = 0.0001)
Panel C: Portfolio returns of full sample and sub-samples of small revenue firms in the HLR (LLR) industries and portfolios of large revenue firms in the HLR (LLR) industries
	Full sample	Portfolio [E] (REVS = 1, LIT =1)	Portfolio [F] (REVS = 1, LIT =0)	Portfolio [G] (REVS = 0, LIT =1)	Portfolio [H] (REVS = 0, LIT =0)
3-DayCAR (p-value) [Firms]	−0.0036 (0.0001) [4668]	−0.0055 (0.0001) [1202]	−0.0029 (0.0001) [2846]	−0.0078 (0.0001) [118]	−0.0021 (0.0177) [502]
Two-sample test of differences:
Portfolios [E] vs [F] (p-value = 0.0265)
Portfolios [G] vs [H] (p-value = 0.0040)
Portfolios [E] vs [G] (p-value = 0.2613)
Portfolios [F] vs [H] (p-value = 0.5086)

Variable definitions:

All variables are defined in Table 1.

Consistent with the litigation cost argument, we find that the full-sample experiences significant negative abnormal returns (-0.36 percent), and that the results remain negative and significant for both SMLREV and LRGREV portfolios. Specifically, we find that SMLREV portfolio experiences significant abnormal returns of -0.37 percent while LRGREV portfolio experiences significant abnormal returns of -0.33 percent. These abnormal returns are not, as expected, significantly different from one another given that the proposal on 12 June did not specify which firms would be subject to the certification requirement at that point, increasing the likelihood of litigation risk for all firms.

Panel B of Table 2 reports the portfolio abnormal return for the HLR (Portfolio C, LIT = 1), and LLR (Portfolio D, LIT = 0) industries. While the statistics indicate that both portfolios experience significant negative abnormal returns, the results from the two-sample test indicate that the equity value effect of the SEC proposal for HLR portfolio is significantly more negative than the LLR portfolio at the 1 percent level. This finding supports the likelihood of increased litigation risk due to the SEC’s proposal. Finally, panel C of Table 2 presents the portfolio abnormal return of different combinations of revenue size and exposure to litigation risk. The results from panel C indicate that SMLREVs and LRGREVs, whether operating in the HLR or LLR industries, experience statistically significant negative abnormal returns. However, the results indicate that both the SMLREV and LRGREV portfolios experience significantly larger negative abnormal returns when they operate in the HLR industries (LIT = 1). In summary, the statistics presented in Table 2 indicate that the significant negative market reaction surrounding 12 June 2002 is largely driven by increased exposure to litigation risk. ¹¹

Table 3 reports statistics, similar to those reported in Table 2, based on the portfolio abnormal return surrounding 27 June 2002. Panel A of Table 3 reports the abnormal returns experienced by all firms and then by SMLREVs (Portfolio A, REVS = 1) and LRGREVs (Portfolio B, REVS = 0) when the SEC released its final ruling.

OPEN IN VIEWER

Table 3.

Descriptive statistics on portfolio returns on the announcement of the SEC’s ruling on 27 June 2002

Panel A: Portfolio returns of full sample and sub-samples of small revenue firms not required to provide sworn testimony (REVS = 1) and large revenue firms required to provide sworn testimony (REVS = 0)
	Full sample	Portfolio [A] (REVS =1) <$1.2 billion		Portfolio [B] (REVS = 0) >$1.2 billion
3-DayCAR (p-value) [Firms]	0.0040 (0.0001) [4645]	0.0051 (0.0001) [4035]		−0.0034 (0.0001) [610]
Two-sample test of differences:
Portfolios [A] vs [B] (p-value = 0.0001)
Panel B: Portfolio returns of full sample and sub-samples of high-litigation-risk firms (LIT = 1) and low litigation risk firms (LIT = 0)
	Full sample	Portfolio [C] (LIT =1) HLR		Portfolio [D] (LIT =0) LLR
3-DayCAR (p-value) [Firms]	0.0040 (0.0001) [4645]	0.0075 (0.0001) [1318]		0.0026 (0.0001) [3327]
Two-sample test of differences:
Portfolios [C] vs [D] (p-value = 0.0002)
Panel C: Portfolio returns of full sample and sub-samples of small revenue firms in the HLR (LLR) industries and portfolios of large revenue firms in the HLR (LLR) industries
	Full sample	Portfolio [E] (REVS = 1, LIT =1)	Portfolio [F] (REVS = 1, LIT =0)	Portfolio [G] (REVS = 0, LIT =1)	Portfolio [H] (REVS = 0, LIT =0)
3-DayCAR (p-value) [Firms]	0.0040 (0.0001) [4645]	0.0087 (0.0001) [1202]	0.0036 (0.0001) [2833]	−0.0056 (0.0014) [116]	−0.0029 (0.0072) [494]
Two-sample test of differences:
Portfolios [E] vs [F] (p-value = 0.0004)
Portfolios [G] vs [H] (p-value = 0.1966)
Portfolios [E] vs [G] (p-value = 0.0001)
Portfolios [F] vs [H] (p-value = 0.0001)

Variable definitions:

All variables are defined in Table 1.

On average, the full sample experiences positive abnormal returns of 0.40%. However, the positive market reaction is only prevalent for the portfolio of small-revenue firms (SMLREVs) that were exempted, on 27 June, from complying with the certification requirement. On the other hand, the LRGREV portfolio experiences significant negative abnormal returns. This further negative reaction by LRGREVs is likely to be due to the resolution of uncertainty about the enactment of the SEC Rule 13a-14. A two-sample test indicates a significant difference between the abnormal returns of SMLREVs and LRGREVs, at the 1 percent level. Panel B of Table 3 indicates that firms operating in the HLR industry (Portfolio C, LIT = 1) experience significantly more positive abnormal returns than firms operating in the LLR industry (portfolio D, LIT = 0) at the 1 percent level. Finally, panel C of Table 3 presents the abnormal returns of the sample firms based on different combinations of the two variables capturing firm revenue size and exposure to litigation risk. The summary statistics indicate that SMLREVs in the HLR industries experience significantly larger positive abnormal returns than LRGREVs in the HLR industries. We also find that SMLREVs in the LLR industries also experience significantly larger positive market reaction than LRGREVs in the LLR industries. Furthermore, within the portfolio of SMLREVs, the stock prices of the HLR sub-portfolio react more positively in comparison to the LLR sub-portfolio. Our daily results (on 27 June) are statistically equivalent to the three-day results reported in the paper. In summary, the statistics presented in Table 3 indicate that the significant positive market reaction of the sample firms is largely driven by the market reaction to news of SMLREVs being exempted from the certification requirement, and this positive reaction becomes more pronounced for the sub-portfolio of SMLREVs operating in the HLR industry. While the results reported in Tables 2 and 3 offer support for the hypotheses, these results could also be induced by firm characteristics which have not been controlled for when reporting these descriptive statistics. In the next section, we report the results from a cross-sectional regression test which will provide a more accurate test of the hypotheses.

8. Cross-sectional results

Table 4 reports the correlation matrix for the independent variables.

OPEN IN VIEWER

Table 4.

Correlation matrix of the independent variables used in regression analysis

	LIT	SREV	SALES_GROWTH	ANDERSEN	INVESTIGATE	BIG_N	MKT_TO_BK	LEVERAGE	REGULATED	FINANCIAL
LIT
REVS	0.0803
SALES_GROWTH	−0.0584	−0.0068
ANDERSEN	−0.0370	−0.0432	0.0061
INVESTIGATE	−0.0092	−0.0374	0.0063	0.0324
BIG-N	−0.0132	−0.1265	0.0304	0.1685	0.0055
MKT_TO_BK	0.1052	−0.0903	0.2252	0.0086	−0.0089	0.1015
LEVERAGE	−0.2834	−0.1901	0.0585	0.0687	0.0019	0.0832	−0.0819
REGULATED	−0.1784	−0.1047	0.0878	0.0846	−0.0047	0.0676	−0.0359	0.2285
FINANCIAL	−0.1952	0.0237	0.0574	−0.0151	−0.0052	0.0023	−0.0717	0.0486	−0.1142

Pearson (Spearman) correlations are shown in each cell. Correlations significant between 1% and 5% level are in bold

Variable definitions:

All variables are defined in Table 1.

On the whole, although there are a large number of significant correlation coefficients, the coefficients are not large enough to prohibit the use of a multivariate regression analysis. ¹²

Our first two litigation-cost hypotheses concern the stock price effects of the announcement of the SEC’s proposal on 12 June 2002. Given that the proposal’s wording implied that all firms would be required to comply with the new certification requirement, we expect a statistically equivalent negative equity value effect for all firms, regardless of whether they were SMLREVs or LRGREVs. That is, variable REVS would be insignificant. Because of the sensitivity of HLR firms to litigation risk, we expect the abnormal returns of HLR firms to be more negative than that of LLR firms. A significant negative coefficient on the variable that captures exposure to litigation risk (LIT) would provide empirical support for the first hypothesis.

Panel A of Table 5 presents the results from the first cross-sectional regression model (Model 1), where the cumulative abnormal return of the sample firms in the three days surrounding 12 June 2002 is regressed on independent variables with the exception of the interaction term, REVS×LIT.

OPEN IN VIEWER

Table 5.

Multivariate regression analysis of market reaction to SEC’s proposal on 12 June 2002 and Ruling 13a-14 on 27 June 2002

Model: 3 - D a y C A R 12 or 3 - D a y C A R 27 = β 0 + β 1 R E V S + β 2 L I T + β 3 R E V S × L I T + β 4 S A L E S _ G R O W T H + β 5 A N D E R S E N + β 6 I N V E S T I G A T E + β 7 B I G _ N + β 8 M K T _ T O _ B K + β 9 L E V E R A G E + β 10 R E G U L A T E D + β 11 F I N A N C I A L + μ
Panel A: Dependent variable (3-DayCAR12) is cumulative abnormal return surrounding 12 June 2002
	Predicted sign	Model 1		Predicted sign	Model 2
		Parameter estimate	p-value		Parameter estimate	p-value
Intercept		0.0017	0.3881		0.0021	0.3123
LIT	−	−0.0025	0.0101		−0.0044	0.0836
REVS	?	−0.0003	0.8335		−0.0007	0.6306
REVS× LIT				?	0.0022	0.5107
SALES_GROWTH		−0.0020	0.0016		−0.0020	0.0111
ANDERSEN		0.0037	0.0025		0.0036	0.0027
INVESTIGATE		0.0290	0.3426		0.0287	0.3486
BIG_N		−0.0040	0.0043		−0.0040	0.0044
MKT_TO_BK		−0.0004	0.0014		−0.0004	0.0015
LEVERAGE		−0.0013	0.6030		−0.0013	0.6075
REGULATED		0.0023	0.1527		0.0023	0.1611
FINANCIAL		−0.0001	0.9187		−0.0001	0.9264
Adjusted R2		0.0080			0.0079
Panel B: Dependent variable (3-DayCAR27) is cumulative abnormal return surrounding 27 June 2002
Intercept		−0.0003	0.8939		0.0010	0.7018
LIT	+	0.0036	0.0052		−0.0032	0.4128
REVS	+	0.0075	0.0000		0.0060	0.0006
REVS× LIT				+	0.0075	0.0314
SALES_GROWTH		0.0015	0.1191		0.0015	0.1267
ANDERSEN		−0.0008	0.5753		−0.0009	0.5410
INVESTIGATE		0.0103	0.7822		0.0090	0.8078
BIG_N		−0.0031	0.0722		−0.0031	0.0734
MKT_TO_BK		0.0002	0.2325		0.0002	0.2133
LEVERAGE		−0.0073	0.0205		−0.0072	0.0213
REGULATED		0.0031	0.1201		0.0029	0.1396
FINANCIAL		−0.0015	0.4014		−0.0014	0.4157
Adjusted R2		0.0103			0.0109

Variable definitions: All variables are defined in Table 1.

The results from the estimation of Model 1, reported in the second and third columns of panel A, indicate that the parameter estimate of LIT is negative (−0.0025) and significant at the 1 percent level, which is consistent with the first hypothesis. The results suggest that the abnormal returns of the LRGREV and SMLREV firms around the proposal date are not distinguishable from one another. As discussed earlier, this is expected because on 12 June all firms, regardless of their revenue size, were expected to comply with the certification requirement. Several of the control variables also yield significant parameter estimates. The parameter estimate for SALES_GROWTH is negative (−0.0020) and significant indicating that the release of the SEC’s proposal was more negatively received by shareholders of firms with high sales growth. On the other hand, the parameter estimate for ANDERSEN is positive (0.0037) and significant implying that, in comparison to clients of the other Big-N auditors, firms affiliated with Arthur Andersen were more positively affected by the SEC’s certification proposal. Arguably, the positive results for Andersen clients indicate a resolution of the prior conservative perception of the market about the quality of the financial reports of these firms.

Finally, the parameter estimates for the variables BIG_N and MKT_TO_BK are negative and significant implying that shareholders of firms with brand-name auditors and high growth firms considered that the SEC’s proposal would be associated with more costs.

We next replicate the regression analysis of 12 June (Model 2), after including the interaction term, REVS×LIT, as an additional independent variable. The parameter estimate of the interaction term would indicate whether the abnormal return of HLR firms is more/less pronounced for firms that were subsequently exempted by the SEC, on 27 June 2002, from complying with the certification requirement. Given that on 12 June 2002 the proposal did not distinguish firms that would be subject to the regulation, we expect an insignificant coefficient for the interaction term in Model 2.

The last two columns of panel A report the results from the estimation of Model 2. Consistent with the second hypothesis, the results indicate that the parameter estimate of the interaction term is insignificant. This indicates that on 12 June, even within the sub-sample of HLR firms, SMLREVs are not more/less adversely affected by the SEC’s proposal. The results for the remaining variables are similar to those reported for Model 1.

The two remaining hypotheses focus on the abnormal returns in response to the SEC’s release of Rule 13a-14, on 27 June 2002. Hence, the abnormal returns of the sample firms surrounding 27 June 2002 (CAR ₂₇) are employed as the dependent variable, to test these hypotheses. As proposed earlier, the third hypothesis conjectures that the market value of SMLREVs would be more positively affected by the release of Rule 13a-14. This is because, in contrast to SEC’s 12 June proposal, the 27 June order effectively exempted SMLREVs from complying with the certification requirement. A significant positive coefficient on the variable REVS, when the regression is run without the interaction term, REVS×LIT, would provide support for the third hypothesis. When the regression is replicated after including the interaction term, a significant positive parameter estimate on the interaction term would indicate that the positive market reaction for SMLREVs becomes more pronounced if firms had high exposure to litigation risk. This particular result would support the fourth and final hypothesis.

Panel B of Table 5 presents the results from the regression analysis employed to test the third and fourth hypotheses, where the three-day cumulative abnormal returns of the sample firms surrounding 27 June 2002 (CAR ₂₇) is regressed on the independent variables. Model 3 includes all independent variables with the exception of the variable that captures the interaction effect of the variables, LIT and REVS (REVS×LIT). The results reported in the second and third columns of panel B indicate that the parameter estimate of REVS is positive and significant at the 1 percent level. This finding supports the third hypothesis, suggesting that on 27 June 2002 shareholders of SMLREVs recovered from their previous loss incurred on 12 June 2002. We also find that stock prices of firms with high exposure to litigation risk react positively and significantly at the 1 percent level in the three days surrounding 27 June. Turning to the results for the remaining variables, the significant results for the variables BIG_N and LEVERAGE indicate that firms with Big-N auditors and higher leverage were also more likely to be negatively and significantly affected by the issuance of the SEC’s final ruling. The last two columns of panel B reports the results from the estimation of the cross-sectional Model 4, after including the interactive variable REVS×LIT. Consistent with our last hypothesis, the results indicate that the parameter estimate of the interaction term is positive and significant (at the 5 percent level), indicating that SMLREVs in the HLR industries experience a significant incremental increase in market value, over SMLREVs in the LLR industries. The results for the remaining variables are similar to those reported for Model 3.

Our results are consistent with the litigation cost hypotheses we offer in the study. It appears that a proper identification of the earliest events and designing the study in such a way that includes firms exempted under the SEC’s ruling on 27 June 2002 provides more conclusive results supporting the increased threat of litigation and hence reduced stock prices.

9. Sensitivity analysis

The results reported so far are based on three-day cumulative abnormal returns (dependent variable) surrounding each event. In a sensitivity analysis we replicate our tests using an augmented dependent variable which accumulates abnormal returns across two days (day -1 and day 0), and use return on equity (ROE) as an additional independent variable. We report the results in Table 6.

OPEN IN VIEWER

Table 6.

Multivariate regression analysis of market reaction to SEC’s proposal on 12 June, 2002 and ruling 13a-14 on 27 June 2002

Model: 2 - D a y C A R 12 or 2 - D a y C A R 27 = β 0 + β 1 R E V S + β 2 L I T + β 3 R E V S × L I T + β 4 S A L E S _ G R O W T H + β 5 A N D E R S E N + β 6 I N V E S T I G A T E + β 7 B I G _ N + β 8 M K T _ T O _ B K + β 9 L E V E R A G E + β 10 R E G U L A T E D + β 11 F I N A N C I A L + β 12 R O E + μ
Panel A: Dependent variable (2-DayCAR12) is cumulative abnormal return over day 0 (12 June ) and day -1 (11 June), 2002
	Predicted sign	Model 1		Predicted sign	Model 2
		Parameter estimate	p-value		Parameter estimate	p-value
Intercept		−0.0017	0.5693		−0.0018	0.5569
LIT	−	−0.0036	0.0277		−0.0030	0.5481
REVS	+	0.0014	0.4924		0.0016	0.4945
REVS× LIT				?	−0.0008	0.8838
SALES_GROWTH		−0.0027	0.0345		−0.0027	0.0349
ANDERSEN		0.0036	0.0597		0.0036	0.0592
INVESTIGATE		0.0081	0.8591		0.0082	0.8572
BIG_N		0.0000	0.9925		0.0000	0.9930
MKT_TO_BK		−0.0010	0.0000		−0.0010	0.0000
LEVERAGE		−0.0039	0.3203		−0.0040	0.3198
REGULATED		0.0077	0.0025		0.0078	0.0025
FINANCIAL		0.0006	0.7928		0.0006	0.7939
ROE		−0.0014	0.0534		−0.0014	0.0529
Adjusted R2		0.0082			0.0086
Panel B: Dependent variable (2-Day CAR27) is cumulative abnormal return over day 0 (27 June) and day -1 (26 June), 2002
Intercept		−0.0127	0.0005		−0.0106	0.0055
LIT	+	0.0050	0.0129		−0.0072	0.2306
REVS	+	0.0028	0.2757		0.0002	0.9429
REVS× LIT				+	0.0135	0.0320
SALES_GROWTH		0.0029	0.0493		0.0029	0.0532
ANDERSEN		0.0000	0.9876		−0.0001	0.9614
INVESTIGATE		−0.0192	0.7293		−0.0213	0.7016
BIG_N		0.0080	0.0024		0.0080	0.0024
MKT_TO_BK		0.0006	0.0198		0.0007	0.0155
LEVERAGE		−0.0154	0.0013		−0.0154	0.0014
REGULATED		−0.0019	0.5389		−0.0022	0.4793
FINANCIAL		0.0014	0.6151		0.0014	0.5967
ROE		0.0004	0.6268		0.0005	0.5610
Adjusted R2		0.0082			0.0091

Variable definitions: ROE = net income divided by total common equity. All remaining variables are defined in Table 1.

We find that the parameter estimate of LIT (under Model 1, second column of Panel A) is negative (−0.0036) and significant (p-value = 0.0227), which is consistent with our main results from testing the first hypothesis. The insignificant results for REVS is also consistent with our main results suggesting that abnormal returns of LRGREV and SMLREV firms are not distinguishable from one another, given that on 12 June all firms, regardless of their revenue size, were expected to comply with the certification requirement. Moving on to the results under Model 2, Panel A, we find that the parameter estimate of the interaction term, REVS×LIT, is not significant. This result is also consistent with our main results previously reported in the paper. Panel B of Table 6 presents the results from the regression analysis employed to test the third and fourth hypotheses, where the two-day cumulative abnormal returns of the sample firms surrounding 27 June 2002 (CAR ₂₇) are regressed on the independent variables. The results reported in the second and third columns of panel B indicate that the parameter estimate of REVS is positive but insignificant. This positive and insignificant result suggests that on 27 June 2002 shareholders of SMLREVs partially recovered from their previous loss incurred on 12 June 2002.

When the regression is replicated after including the interaction term, a significant positive parameter estimate on the interaction term would indicate that the positive market reaction for SMLREVs becomes more pronounced if firms had high exposure to litigation risk. This particular result would support the fourth and final hypothesis. We provide results consistent with our main results with the interactive variable coefficient being positive (0.0135) and significant (p-value = 0.0320).

On the whole, the sensitivity results confirm the main results that we have reported in the paper.

10. Conclusion

In this study, we employ a litigation-cost approach to examine the stock price reaction to the SEC’s proposal on 12 June 2002 and its enactment of Rule 13a-14 on 27 June 2002, which required CEOs and CFOs to certify the accuracy of their firm’s financial reports. Prior studies investigating stock price reactions to the submission of the certification statements in August 2002 have reported inconclusive results on whether stockholders benefit from the certification requirement. We argue that the investors’ perception of the new rule must be tested around the initial event dates of 12 and 27 June, rather than when firms file their submissions with the SEC in August 2002. The implied assumption by prior studies that examine the submission dates of certification statements in August is a delay in market reaction, which is inconsistent with market efficiency and investors’ rational behavior. Our analyses provide support for the litigation cost hypothesis by showing that firms operating in the HLR industries reacted more negatively to the SEC’s 12 June proposal. On 27 June 2002, when the SEC relieved firms with $1.2 billion or less in revenues from the certification requirement, these firms experienced significant positive abnormal returns, and within the relieved firms, the HLR firms reacted even more positively to the news of exemption from the certification requirement.