The Pricing of IPOs Post-Sarbanes-Oxley

Authors


  • We thank Arnold Cowan (the editor) and two anonymous referees for their valuable comments.

* Corresponding author: Walker College of Business, Appalachian State University, Boone, NC 28608; Phone: (828) 262-2893; Fax (828) 262-6612; E-mail: jjohnston@appstate.edu

Abstract

The Sarbanes-Oxley Act (SOX) imposes new requirements for firms going public. Many provisions of SOX should improve the transparency of U.S. firms going public and therefore reduce the uncertainty surrounding their valuation. We find that initial returns of initial public offerings (IPOs) in the United States have declined since SOX. Furthermore, the aftermarket performance of IPOs since SOX is significantly higher. While the expense of public reporting has increased in the United States because of SOX, the valuations of newly public firms at the time of the IPO are subject to less uncertainty and smaller aftermarket corrections.

1. Introduction

The Sarbanes-Oxley Act (SOX) has received considerable attention since being implemented in 2002. Much of the attention has focused on how SOX affects the internal control processes of publicly traded firms and increases the costs of internal controls and reporting. However, the potential impact of SOX on firms going public has been largely ignored. Because SOX affects the transparency not only of public firms but also those preparing to go public, SOX could alter the underpricing behavior of initial public offerings (IPOs).

Firms that are about to go public can be subject to asymmetric information, that is, investors may not be aware of as many details of the firm when compared to managers of the firm. The quality of information (or lack of it) can affect the degree of underpricing. Ibbotson (1975), Rock (1986), and others suggest that public firms need to underprice their shares to entice investors. Beatty and Ritter (1986) find that underpricing is higher for firms that have more uncertainty. Thus, firms that are subject to more asymmetric information require a greater degree of underpricing. SOX imposes many provisions intended to improve the transparency of firms that go public and to reduce the uncertainty surrounding their valuations. In particular, SOX requires that firms have audit committees and internal controls in place at least one year prior to going public. Their chief executive officers (CEOs) and chief financial officers (CFOs) must be involved and accountable for the internal controls process. To the extent that SOX can reduce the asymmetric information of firms going public, it should reduce the uncertainty surrounding valuations and therefore reduce the degree of underpricing that is required. The IPO process screens out some firms that cannot conform to SOX guidelines. In theory, firms that satisfy SOX guidelines leave less money on the table.

SOX could also affect the performance of IPOs in the period following the initial return. Aggarwal and Rivoli (1990) argue that firms can experience weak aftermarket performance if the initial returns at the time of the IPO are relatively high. That is, the initial returns of some IPOs are inflated as investors develop overly optimistic expectations because they do not have sufficient information. Aftermarket performance reflects a correction to the expectations as more information becomes available. To the extent that SOX can increase transparency and reduce the overly optimistic expectations of investors, it could reduce the initial returns and therefore prevent an aftermarket correction. Consequently, the aftermarket performance of IPOs could be more favorable (or less unfavorable) after SOX.

Our objective is to determine the impact of SOX on underpricing and aftermarket performance of IPOs in the United States. To assess the impact on underpricing, we measure the initial returns of IPOs before and after SOX. We also conduct a comparison of initial returns between U.S. IPOs and Canadian IPOs before and after SOX because Canadian IPOs are not subject to SOX rules. We find that initial returns of IPOs of U.S. firms are much lower in the post-SOX period than in the pre-SOX period, while the initial returns of Canadian IPOs (not subject to SOX) are not lower in the post-SOX period. We also conduct a cross-sectional analysis of U.S. IPO initial returns and find that the initial returns are lower after SOX, even after controlling for many factors. Overall, our results support our hypothesis that the degree of IPO underpricing is lower after SOX.

To assess the aftermarket performance of U.S. IPOs, we measure one-year abnormal returns of IPOs occurring before and after SOX. We find that the one-year abnormal returns following IPOs after SOX are significantly higher than IPOs before SOX. We also conduct a cross-sectional analysis and find that the one-year abnormal returns remain significantly higher after SOX even after controlling for other factors. In general, the results suggest that initial returns in the pre-SOX period are excessive and require corrections in the aftermarket. Conversely, initial returns and the aftermarket correction in the post-SOX period are not as pronounced. In the post-SOX period, firms are subject to more scrutiny and therefore face more formidable barriers to justify going public. Investors appear to rely on more substantive information when valuing the firm, such as the offer price set by the underwriter.

2. Background on the SOX

In response to financial reporting scandals, the SOX became law on July 30, 2002 (see Crutchley, Jensen and Marshall, 2007). The act requires firms to disclose detailed financial information. Second, it requires firms to establish a system of financial controls and frequently monitor their systems to determine whether they are working properly.

SOX also improves the reliability of audits by requiring that only outside board members of a firm be on the firm's audit committee. SOX prevents conflicts of interest between accounting firms and their audit clients and requires a regulatory review of audit firms every one to three years. The benefits of SOX are improved internal controls and governance of the audit process. Crutchley, Jensen and Marshall (2007) find that firms with higher percentages of outside directors and outside audit committee members are less likely to be involved in accounting scandals. Jain, Kim and Rezaee (2008) show that the quality of financial reports and market liquidity have improved since SOX. Akhigbe, Martin and Newman (2008) suggest that since SOX, firms are more willing to disclose negative information that they would prefer to withhold.

SOX has major implications for firms that are about to go public. First-time filers of a Securities Act registration statement are immediately subject to some provisions of SOX. The firm must compose an audit committee of independent directors, and one of the members must be a financial expert who has experience in auditing or analyzing financial statements. A quality control process must be established, and a process for internal communication of information is also required. Internal controls must be in place for one year before an IPO. Since it is a legal requirement that firms going public have internal controls and procedures in place, there is more legal compliance to ensure that the documentation satisfies the law.1 Furthermore, the executives, such as the CEO and CFO, must be involved in establishing the internal controls. There are other governance issues, such as a code of ethics, which must be articulated. In addition, there are disclosure requirements for firms that do not meet the normal listing requirements of U.S. stock exchanges. Firms are subject to delays if they pursue an IPO without complying to SOX requirements.

Because of SOX, the transparency of newly public firms arguably has increased. Consequently, the information asymmetry between the firm's managers and prospective investors should decrease. Thus, some firms avoid going public if they cannot justify the cost. Chhaochharia and Grinstein (2007) examine the announcement effects of SOX and other governance rules. They find that firms with a history of lower compliance experience positive and significant abnormal returns. For example, a portfolio of firms that have a history of financial restatements performs 8.5% above a matched portfolio.

SOX also requires changes to address conflicts of interest when analysts recommend equity securities in research reports. The SEC and the major stock exchanges have also initiated similar reforms. The reforms separate the analysts from the investment banking division and prohibit the analysts from participating in the solicitation of investment banking business.2

Linck, Netter and Yang (2009) find that since SOX, boards of directors are more independent and CEOs are less likely to be chairmen of the boards. Aggarwal and Williamson (2006) find that firms have actively implemented new corporate governance policies since SOX.

3. Reasons for IPO underpricing

This section summarizes several studies that identify factors that can affect underpricing. Since SOX can influence other factors, it can therefore influence underpricing as well. Information about a firm going public influences the degree of underpricing that is necessary to attract sufficient demand for the shares. Welch (1989) argues that investors have very limited knowledge about the value of shares for a firm going public, and that a firm is more willing to underprice its shares if it needs credibility to engage in additional stock offerings in the future. Allen and Faulhaber (1989) also suggest that credible issuers will underprice their IPOs so they can recoup the costs in subsequent offerings. Jegadeesh, Weinstein and Welch (1993) empirically confirm that firms that use greater underpricing during the IPO are more likely to engage in a seasoned offering in the future. However, if SOX can make IPOs more transparent, firms should not be forced to underprice the IPO to such a large degree.

Another argument used for underpricing is to protect against legal liability. Tinic (1988) finds that underpricing has been more pronounced since the Securities Act of 1933, which made firms and their respective underwriters more accountable for fraudulent behavior. Thus, a higher degree of underpricing is intended to avoid situations in which the IPO is initially overpriced (whether knowingly or not) by the issuer and underwriter. If SOX provisions allow for more transparency, fraudulent behavior could be prevented or reduced, and IPO shares should no longer have to require such a high premium to account for the possibility of fraudulent behavior.

Chemmanur (1993), Aggarwal, Krigman and Womack (2002), and Loughran and Ritter (2004) suggest that underpricing is intended to indirectly reimburse investment banks for providing analyst coverage. A sufficiently high degree of underpricing ensures more favorable returns to investors, which establishes credibility for the underwriter. Furthermore, analyst coverage that is commonly initiated by the underwriter at the end of the quiet period serves as an effective endorsement to the market, which can reduce information asymmetry. Yet, if SOX can establish credibility at the time of the IPO, investors will not need to rely on the implicit endorsement resulting from the initiation of analyst coverage.

4. Hypotheses

4.1. Effect of SOX on IPO initial returns

SOX can alter the effects of the factors discussed above. First, the provisions on the disclosure of financial information are more complete due to SOX. Second, the requirement to establish internal controls and a quality control process can increase the credibility of an issuer. Third, the CEO and CFO are held accountable for reporting and internal controls, which can increase the quality of information. Fourth, the provisions can improve the credibility of the auditing process, which can reduce uncertainty. Fifth, the provisions of SOX can reduce the potential conflicts of interest between underwriters and analysts, diminishing the influence that underpricing has on analyst coverage.

One can argue that because SOX requires more accountability for proper pricing of an IPO, it could lead to greater underpricing by the issuer and underwriter as protection against legal liabilities, similar to Tinic's (1988) argument about the Securities Act of 1933. However, the increased accountability required by SOX is closely tied to the disclosure of more information, which should allow for greater transparency.3

4.2. Impact of SOX on aftermarket performance

Aggarwal and Rivoli (1990), Ritter (1991), and Loughran and Ritter (1995) find that newly public firms experience weak aftermarket performance. Each of these studies suggests that irrational pricing at the time of the IPO could be the reason for the weak aftermarket performance. That is, the aftermarket performance reflects a downward correction to the excessively high price at the time of the IPO. Jain and Kini (1994) find that the financial reports of IPOs are more favorable at the time of the IPO than after the IPO. Put together, the studies cited here suggest that firms use window dressing at the time of the IPO, which could cause excessive optimism among investors and therefore a stock price correction over time in response to subsequent financial reporting.

Cohen, Dey and Lys (2008) find that earnings management declines following SOX, which improves the quality of the reported earnings of firms going public and reduces the uncertainty surrounding stock valuation. As a related point, SOX can serve as a screen for firms that go public. Firms that are more willing to be transparent would still pursue an IPO, while some firms that have something to hide can avoid the SOX provisions by remaining private. Thus, firms that pass the SOX screening process should exhibit relatively low risk, and there is less likelihood that investors would need to correct for over-optimism in the aftermarket. To the extent that SOX provisions reduce irrational pricing at the time of (or shortly after) the IPO, it could reduce or eliminate an aftermarket correction. We hypothesize that the aftermarket price performance following IPOs has improved since SOX.

5. Methods

5.1. Initial returns

We measure and test the difference of the initial returns of IPOs that occur before and after SOX. Initial returns are calculated as the percent change from the offer price to the price at the close of the first day of trading and are not market adjusted. We recognize that a change in initial returns since SOX could be caused by factors other than SOX. Therefore, we also include a sample of Canadian IPOs that are not subject to SOX guidelines. Our hypothesis of reduced uncertainty and lower initial returns since SOX only applies to U.S. firms. If the uncertainty surrounding IPO firms is reduced because of SOX, only the sample of U.S. IPOs should experience lower initial returns since SOX. Conversely, if the uncertainty surrounding IPO firms is reduced because of confounding conditions, the Canadian IPOs should experience lower initial returns as well. To conduct our comparison of initial returns between United States and Canadian IPOs, we match, with replacement, the Canadian sample to the U.S. sample based on time of the offering and size of the offering.

5.2. Aftermarket returns

We measure aftermarket performance of IPOs before and after SOX. The aftermarket returns begin the day after the offer date and are calculated as a one-year buy-and-hold abnormal return

image(1)

where Ri.t is the return of the IPO firm i on trading day t, and Rb,t is the return of the CRSP value-weighted market index over the same period. The abnormal returns are winsorized at three standard deviations to decrease the influence of outliers.4Cowan and Sergeant (2001) find that winsorized abnormal returns produce better-specified test statistics. We also apply a cross-sectional analysis to the sample of aftermarket returns and include a SOX dummy variable set equal to 1.0 following the passage of SOX. This analysis controls for other factors that could affect aftermarket performance.5

5.3. Cross-sectional analysis

We regress the initial and one-year aftermarket returns separately using the following model

image(2)

where Ri is the initial or aftermarket return and the independent variables are described below. We use White's (1980) correction to correct for heteroskedasticity.

We also apply the cross-sectional model separately to the pre-SOX and post-SOX periods. We then apply the cross-sectional model to aftermarket returns (in place of initial returns) over the entire sample period, as well as to the pre-SOX and post-SOX periods.

5.3.1. Sarbanes-Oxley Act

The SOX was enacted on July 30, 2002. We use a dummy variable equal to one if the firm goes public following the passage of SOX, zero otherwise.6

5.3.2. Underwriter prestige (UP)

According to Carter and Manaster (1990), prestigious underwriters are less likely to underwrite risky offerings and could be more skilled at pricing their offerings appropriately. Research by Johnson and Miller (1988), Carter and Manaster (1990), Michaely and Shaw (1994), and others find that underpricing is inversely related to UP. Our focus on UP centers on whether and how SOX alters its association with initial returns of IPOs. In other words, does SOX reduce the importance of the lead underwriter's prestige?

The UP variable is the Carter-Manaster rank used in Loughran and Ritter (2004) and is available on Ritter's website (http://bear.cba.ufl.edu/ritter/rank.pdf). For 2005 and 2006, we use the 2001–2004 rankings.

5.3.3. Venture capital (VC)

Firms that are backed by venture capitalists can benefit from VC expertise or monitoring (see Baker and Gompers, 2003). Thus, there should be less uncertainty surrounding value, which would be reflected in a lower degree of underpricing. Barry, Muscarella, Peavy and Vetsuypens (1990) empirically confirm this relation. However, VC firms are more likely to pursue investments that offer the potential for high returns, and the demand for shares backed by VC can be stronger, which could cause a more pronounced initial return unless additional shares are issued. Further, venture capitalists can relinquish their control over time (see Black and Gilson, 1998) and could cause downward price pressures in the aftermarket as they cash out (see Field and Hanka, 2001). The VC proxy is a dummy variable equal to one for firms with VC investment.

5.3.4. Size

Larger offerings should be subject to more scrutiny by the market and could exhibit less uncertainty than smaller offerings (see Jegadeesh, Weinstein and Welch, 1993; Michaely and Shaw, 1994). However, larger offerings could create more sentiment and demand on the first day of the offering, which could result in more pronounced initial returns. The log of gross proceeds raised in the IPO is used to measure the size of the offering.7

5.3.5. Hot IPO period (HOT)

When there is much favorable sentiment about IPOs, more firms are expected to come to market because they can more easily place their shares. Loughran and Ritter (2002) find a significant autocorrelation of initial returns. Similar to Bradley and Jordan (2002), we use an average of the initial returns for IPOs in the previous 30 calendar days.

5.3.6. Bubble

Initial returns of IPOs in 1999–2000 are unusually high and are followed by poor long-run performance. Loughran and Ritter (2004) find that initial returns increase dramatically during the internet bubble years of 1999–2000. To control for these unusual returns, we include a dummy variable equal to one if the IPO occurred during 1999–2000.

5.3.7. Auditor prestige (AP)

Michaely and Shaw (1995) find that lower-risk IPOs attract higher quality auditors. This certification could lower the initial underpricing and improve long-run returns. We control for AP with a dummy variable equal to one if the auditor is a Big 4 (or Big 5–8 in previous years).

5.3.8. Industry

Benveniste, Ljungqvist, Wilhelm and Yu (2003) find that IPOs in nascent industries, when compared to IPOs in mature industries, are more influenced by other IPOs within their industry. Similar to Benveniste, Ljungqvist, Wilhelm and Yu (2003), we control for the industry effect with a dummy variable if the firm is in any of the following three-digit Standard Industry Classification (SIC) codes: 283, 357, 366, 367, 381, 382, 383, 384, and 737.

5.3.9. Age

There could be more information available for older firms. Additional information decreases uncertainty and, therefore, underpricing. Loughran and Ritter (2004) use age as a proxy for risk. Age is measured as the log of the number of days from founding date to offer date. Founding dates from Ritter's website (http://bear.cba.ufl.edu/ritter/ipodata.htm) are used. The Thomson Financial SDC Platinum (SDC) database and prospectuses are used to find missing data.

5.3.10. Sales

Loughran and Ritter (2004) use sales to measure risk composition and find lower underpricing among IPOs of firms with higher sales. The log of sales for the fiscal year ending prior to the IPO date is used to measure the amount of sales.

5.3.11. Overhang (OH)

The initial returns are expected to be larger for IPOs in which the OH is larger. OH is measured as the number of shares retained by insiders divided by the number of shares offered. Insiders should be more willing to leave more money on the table if they retain more shares. Aggarwal, Krigman and Womack (2002) and Bradley and Jordan (2002) find that underpricing is positively correlated with the amount of inside ownership.

5.3.12. Number of uses of proceeds

A larger number of uses of proceeds could increase uncertainty. Beatty and Ritter (1986) detect a positive relation between the number of uses of proceeds and uncertainty. Once again, the SDC database is used with data from prospectuses to find missing data.

5.3.13. Discretionary current accruals (DCACC)

Firms can modify their earnings by adjusting the recognition of revenues and expenses. DCACC include short-term cash flows that can be adjusted by managers. Teoh, Wong and Rao (1998) determine that firms engaging in IPOs have high positive earnings and abnormally high accruals in the year of their IPO. However, these firms experience weak long-run earnings. One argument is that firms with high accruals are subject to more uncertainty, and therefore should exhibit higher initial returns. However, a counterargument is that the valuations of firms with high accruals are often discounted by investors, which results in lower initial returns for these firms. Regardless of the effect on initial returns, we expect that aftermarket performance will be weaker for firms with higher accruals. Teoh, Welch and Wong (1998) find that firms engaging in IPOs have relatively high accruals and experience weak long-run stock price performance. The results suggest that investors do not fully account for the accruals at the time of the IPO.

Following Teoh, Welch and Wong (1998), DCACC are calculated using the Jones (1991) model and are based on year-end financial statements prior to the offering date.

6. Data

The SDC database is the primary data source for our sample of IPOs from January 1, 1990 through December 31, 2006. We include only issues listed on the New York, American, and Nasdaq stock exchanges. After eliminating banks, savings and loans, closed-end funds, partnerships, depository shares, real estate investment trusts, unit issues, and spinoffs, we are left with a sample of 4,512 IPOs. The sample falls to 4,215 for the long-run analysis due to missing stock price information. Data from Jay Ritter's website (http://bear.cba.ufl.edu/ritter) are used to calculate UP and Age. An additional sample of Canadian IPOs for the same period is collected for comparison purposes. The initial sample is from SDC and the closing prices are from Datastream. After eliminating banks, investment funds, investment trusts, unit issues, dual-listed stocks, and issues smaller than C$1,000,000, we are left with a sample of 402 Canadian IPOs.

The total proceeds from IPOs in our sample over the entire period amount to approximately $365 billion. The average size generally increases through time, particularly from 1999 through 2001. The average size declines slightly following SOX. The number of IPOs declines substantially following the bubble period of 1999–2000.

Descriptive statistics appear in Table 1. UP generally increases over time. The mean rank increases from 7.11 in the pre-SOX period to 7.79 in the post-SOX period, and the difference between the two periods is significant. The proportion of IPOs backed by VC increases substantially during the three years prior to SOX and declines following SOX, but the difference between the mean proportion in the pre- versus post-SOX periods is not significant. Heesen (2005) suggests that SOX reduces the potential gains to VC that is invested with the intent of pursuing an IPO: “Any company that plans to enter the public domain through a stock offering or acquisition must follow the SOX rules as well. For small private companies, this translates into taking resources away from R&D and putting them toward compliance” (p. 46). The use of high prestige auditors declines by 5% following the passage of SOX, and the difference between pre- and post-SOX periods is significant. The average age of the firm is significantly higher following SOX while the average OH is not significantly different between periods.

Table 1. 
Sample statistics of IPOs: January 1, 1990–December 31, 2006
The sample contains 4,512 IPOs. SOX took effect on July 30, 2002. Number is the total number of companies in the sample that went public during the time period, Total proceeds is the total capital raised in the offering, Size is the mean amount raised by IPOs during that time period, UP is the mean Carter-Manaster rank, VC is the percentage of IPOs that had VC investment, AP is the percentage of IPOs that used a Big 4+ auditor, Age is the log of the number of days from founding to offer date, and OH is the average overhang (shares retained/shares offered). The p-values are calculated using a two-group test.
PeriodNumberTotal proceedsSizeUPVCAPAgeOH
1990–20064512365,348,400,00080,972,6067.1943.8%89.1%7.883.50
Pre-SOX3995271,791,600,00068,032,9417.1144.089.77.843.53
Post-SOX51793,556,800,000180,960,9287.7942.484.78.203.28
Difference  112,927,9870.68−1.6−5.00.36−0.25
p-value  0.000.000.470.000.000.13
1990872,839,200,00032,634,4837.6743.790.88.143.26
199122610,180,600,00045,046,9037.7250.991.28.213.14
199232316,629,300,00051,483,9017.0042.791.68.232.20
199341921,703,200,00051,797,6136.7842.091.28.072.10
199435914,398,600,00040,107,5216.1834.083.67.702.52
199541723,711,000,00056,860,9116.8242.587.57.712.84
199662132,775,700,00052,778,9056.7439.689.57.703.52
199743829,102,900,00066,444,9776.8128.185.67.824.09
199824616,060,900,00065,288,2117.0330.987.87.644.76
199941539,515,900,00095,219,0367.9562.494.07.545.74
200033544,647,200,000133,275,2248.2069.693.17.784.01
20016612,547,500,000190,113,6367.9256.197.08.303.65
2002 Pre-SOX437,679,600,000178,595,3498.0744.297.78.504.10
2002 Post-SOX121,954,200,000162,850,0007.8333.391.77.974.12
2003569,515,100,000169,912,5007.9142.996.48.393.74
200415625,099,100,000160,891,6677.9059.089.78.213.94
200515028,194,800,000187,965,3337.5929.383.38.293.14
200614328,793,600,000201,353,8467.8338.575.58.032.45

The number of issues for each business sector is listed in Table 2. Ritter (1984) demonstrates that the popularity of IPOs can shift in response to a particular sector. In the pre-and post-SOX periods, the manufacturing sector (SIC 2000–4000) has the most issues. The percentage of issues in the manufacturing sector increases in the post-SOX period. Conversely, the proportion of issues in the services sector (SIC 7000–9000) declines from 36% in the pre-SOX period to 23% in the post-SOX period.

Table 2. 
Industry distribution of sample U.S. IPOs, 1990–2006
IndustrySIC codePre-SOXPost-SOX
NumberPercentNumberPercent
Agriculture, forestry, and fishing    0–1,000100.25%10.19%
Mining1,000–1,500751.88234.45
Construction1,500–1,800370.9350.97
Manufacturing2,000–4,0001,53138.3222543.52
Transportation and utilities4,000–5,0003388.465711.03
Wholesale trade5,000–5,2001523.80132.51
Retail trade5,200–6,0002776.93356.77
Finance, insurance, and real estate6,000–6,8001263.15387.35
Services7,000–9,0001,44936.2712023.21

7. Results

7.1. Initial returns

We assess the initial returns before and after SOX in the U.S. sample (in Panel A) and the Canadian sample (in Panel B). Panel A of Table 3 shows that the mean initial return is 25.5% in the pre-SOX period versus 10.6% in the post-SOX period. The difference in mean initial returns in the pre-SOX period versus the post-SOX period is significant. The significance remains even after removing the initial returns of firms going public during the 1999–2000 bubble. When excluding bubble period initial returns, the mean initial return during the pre-SOX period is 15.9%. Also, the mean and median initial returns are significantly lower in the post-SOX period than in the pre-SOX period.

Table 3. 
Initial returns surrounding Sarbanes-Oxley
The two-tailed p-values for mean initial returns are calculated using a two-groups test and assume unequal variance. The two-tailed p-values for median initial returns are calculated using the Wilcoxon rank-sum test.
Panel A: Initial returns for U.S. sample
 Pre-SOX mean (median)Post-SOX mean (median)Difference test t-test p-value (Wilcoxon p-value)
Entire sample0.2550.1060.000
(0.100)(0.059)0.000
N3,995517 
Excluding 1999–2000 period0.1590.1060.000
(0.080)(0.059)0.000
N3,245517 
 
Panel B: Initial returns for Canadian sample
 Pre-SOX mean (median)Post-SOX mean (median)Difference test t-test p-value (Wilcoxon p-value)
 
Entire sample0.1600.2350.362
(0.026)(0.051)0.487
N30894 
Excluding 1999–2000 period0.1360.2350.219
(0.025)(0.051)0.413
N25994 
 
Panel C: U.S. and Canada paired comparison of initial returns
 U.S. mean (median)Canada mean (median)Difference test t-test p-value (Wilcoxon p-value)
 
Entire sample
 Pre-SOX initial returns0.2550.1810.000
(0.100)(0.044)0.000
 N3,9953,995 
 Post-SOX initial returns0.1060.3210.000
(0.059)(0.104)0.000
 N517517 
Excluding 1999–2000 period
 Pre-SOX initial returns0.1590.1590.975
(0.080)(0.022)0.000
 N3,2453,245 
 Post-SOX initial returns0.1060.3210.000
(0.059)(0.104)0.000
 N517517 

Next, we provide results for a similar analysis of Canadian IPOs not subject to SOX regulations. The results of the comparison (Table 3, Panel B) show no difference in initial returns for Canadian IPOs between the pre-SOX and post-SOX periods. We also conduct the comparison after removing Canadian IPOs during the bubble period, and there still is no difference between the initial returns in the pre-SOX versus post-SOX periods. Thus, the results suggest that Canadian IPO initial returns are not affected like the U.S. IPO initial returns following SOX.

Next, we match, with replacement, the Canadian sample to the U.S. sample based on time and size so we can compare initial returns of IPOs between United States and Canada. The Canadian IPO that is closest in size (based on proceeds raised) and issued within 90 days of the U.S. IPO is selected as the match. If there is no matching IPO within 90 days, the matching IPO is selected based on size and must be issued within 180 days. The match also has to be on the same side of the SOX date. The results in Panel C show that the initial returns of the U.S. sample are higher than the initial returns of the Canadian sample in the pre-SOX period but lower in the post-SOX period. When the 1999–2000 bubble is excluded, the initial returns of the two samples are very similar in the pre-SOX period, and there is no significant difference. Yet, the IPO initial returns in the U.S. are significantly lower than in Canada during the post-SOX period. The results support the notion that the reduction in IPO initial returns in the United States is attributed to SOX and is not due to a confounding event.

7.2. Aftermarket returns

Table 4 shows that the mean one-year abnormal return following the IPO is −8.5%, significant at the 0.01 level, during the pre-SOX period. The mean one-year abnormal return in the post-SOX period is 5.1% and not significant at the 0.10 level. The difference in mean and median aftermarket returns between the two groups is significant. The significance remains when the IPOs of firms that went public during the 1999–2000 bubble are removed from the sample.

Table 4. 
Long-run abnormal returns surrounding Sarbanes-Oxley
The sample contains 4,215 IPOs during January 1, 1990–December 31, 2006. Abnormal returns are calculated using the CRSP value-weighted index and are winzorized at three standard deviations. The p-values are calculated using the one-sample t-test and the Wilcoxon signed rank test for the mean and median, respectively. The difference test p-values are calculated using a two-group t-test and the Wilcoxon rank-sum test for the mean and median, respectively.
 Pre-SOXPost-SOXDifference test p-value
Panel A: Entire sample
One-year mean return−0.0850.0510.000
p-value0.0000.107 
One-year median return−0.231−0.0240.000
p-value0.0000.838 
N3,850365 
 
Panel B: Excludes IPOs from the 1999 to 2000 period
One-year mean return−0.0580.0510.001
p-value0.0000.107 
One-year median return−0.179−0.0240.000
p-value0.0000.838 
N3,139365 

The results support our hypothesis that the aftermarket performance of IPOs is less negative since the passage of SOX. Aggarwal and Rivoli (1990) and Loughran and Ritter (1995) suggest that poor aftermarket performance following an IPO reflects a correction to excessive initial returns. The combination of a less favorable initial return and a more favorable (or less unfavorable) aftermarket return since SOX implies that the initial pricing at the time of the IPO is more rational.

7.3. Cross-sectional analysis of initial returns

Table 5 reports regressions of initial returns for the entire sample period. All three models are highly significant, with more than 20% of the variation in the initial returns explained by each model. The results for the variables that are included in multiple models are qualitatively similar. Most importantly, the SOX variable is negative and significant in all models, which supports our hypothesis. The SOX coefficient in the full model (Model 3) is –0.057, implying that the initial return is estimated to be 5.7 percentage points less on average in IPOs since SOX, after controlling for other factors. The coefficient is also negative in the other two models, and slightly larger in absolute value.

Table 5. 
Cross-sectional regressions of initial returns
The dependent variable is the initial return, which is measured as the percent change from the offer price to the closing price of the first day of trading. SOX is a dummy variable equal to one after the effective date of Sarbanes-Oxley (July 30, 2002). UP is the Carter-Manaster rank. VC is a dummy variable equal to one if the firm had VC investment. Size is the log of the gross proceeds raised in the IPO. HOT is the average initial return for IPOs in the previous 30 days. Bubble is a dummy variable equal to one if the IPO occurred during 1999–2000. AP is a dummy variable equal to one if the auditor is one of the Big 4+. Industry is a dummy variable equal to one if the firm is in an emerging industry (determined by SIC code). Age is the log of the number of days from founding to offer date. Sales is the log of sales from the year prior to the offer date. OH is the number of shares retained divided by the number of shares offered. Uses is the number of uses of gross proceeds. DCACC represents discretionary current accruals, which is measured by the Jones model.
VariableModel 1Model 2Model 3
  1. ***, **, * indicate statistical significance at the 0.01, 0.05 and 0.10 level, respectively.

Intercept−0.478−0.296−0.428
−3.421***−2.025**−2.543**
SOX−0.076−0.070−0.057
−5.599***−5.093***−3.252***
UP−0.0000.000−0.001
−0.0260.102−0.161
VC0.0680.0630.064
4.740***4.364***3.863***
Size0.0330.0350.044
3.681***3.735***3.993***
HOT0.4190.4180.485
5.459***5.431***5.044***
Bubble0.2020.1970.164
4.179***4.080***2.868***
AP−0.071−0.074−0.080
−3.460***−3.514***−3.578***
Industry0.0950.0960.092
6.540***6.395***5.458***
Age −0.031−0.034
−5.776***−5.408***
Sales 0.0020.001
1.2320.798
OH  0.007
3.584***
Uses  −0.008
−1.910*
DCACC  −0.002
−0.252
Adj. R20.2030.2080.234
N4,5124,4233,721
F-statistic144.972***117.026***88.351***

The UP variable is not significant, suggesting that the initial return is not conditioned on the perceived credibility of the underwriter. The VC variable is positive and significant, which suggests that VC firms pursue opportunities where substantial returns are possible. The size variable is positive and significant, which suggests that initial returns are more favorable for larger offerings. The HOT variable is positive and significant, which supports the hypothesis that initial returns are higher when market sentiment is favorable. The bubble variable is positive and significant, which supports the hypothesis that initial returns are unusually high during the 1999–2000 bubble. The AP variable is negative and significant, which confirms that there is lower underpricing when the auditor is one of the big accounting firms. The industry variable is positive and significant, which suggests that initial returns are more favorable for emerging industries. The age variable is negative and significant, which confirms that older firms have lower underpricing. The sales variable is not significant, suggesting initial returns are not conditioned on the firm's level of sales. The OH variable is positive and significant, which supports the hypothesis that initial returns are higher when insiders retain more ownership. The uses variable is negative and significant at the 0.10 level, which offers weak evidence that IPOs with a higher number of uses have lower initial returns. The DCACC variable is not significant, which suggests that IPOs are not conditioned on current accruals.

7.4. Cross-sectional analysis of aftermarket returns

Table 6 shows the results from cross-sectional regressions of aftermarket returns. The SOX variable is positive and significant in all three models, which supports the hypothesis of a less pronounced downward correction in the aftermarket. The SOX coefficient estimated from the complete model (Model 3) is 0.160, which suggests that the long-run abnormal return is estimated to be 16% higher since SOX. This estimate is comparable to the estimated difference in long-run abnormal returns between the pre-SOX and post-SOX periods of 13.6% reported in Panel A of Table 4.

Table 6. 
Cross-sectional regressions of long-run abnormal returns
The dependent variable is the one-year market-adjusted buy-and-hold return. Independent variables are explained in Table 5. Hall's (1992) skewness-corrected transformed normal t-statistic is used.
VariableModel 1Model 2Model 3
  1. ***, **, * indicate statistical significance at the 0.01, 0.05 and 0.10 level, respectively.

Intercept0.041−0.0120.302
0.176−0.0491.150
SOX0.0910.0970.160
2.452**2.586***3.807***
UP0.0420.0350.031
5.778***4.687***3.816***
VC−0.049−0.030−0.030
−1.942*−1.189−1.071
Size−0.023−0.035−0.044
−1.531−2.219**−2.630***
HOT−0.198−0.202−0.283
−2.079**−2.115**−3.152***
Bubble−0.057−0.0310.046
−0.864−0.4710.692
AP0.0600.0600.026
1.5171.4740.570
Industry0.0290.0510.042
1.1752.048**1.563
Age 0.0050.001
0.5460.054
Sales 0.0150.014
4.893***4.310***
OH  0.014
4.557***
Uses  −0.029
−3.701***
DCACC  −0.003
−0.410
Adj. R20.0240.0280.051
N4,2154,1393,721
F-statistic13.724***13.062***15.500***

The UP variable is positive and significant, suggesting that the aftermarket performance of IPOs underwritten by prestigious investment banks is more favorable. The VC variable is negative and significant (at the 0.10 level) in Model 1 but not significant in the other models. The size variable is negative and significant in two of the models, which suggests the aftermarket performance is worse for large firms. The HOT variable is negative and significant, which suggests the aftermarket performance is worse for firms that go public during periods when initial returns are generally high. The industry variable is positive and significant in Model 2, offering weak evidence that IPOs in emerging industries perform better. The sales variable is positive and significant, which suggests the aftermarket performance is higher for firms that have higher sales levels prior to the IPO. The OH variable is positive and significant, which suggests the aftermarket performance is better for firms with high inside ownership. This favorable effect is in addition to the positive effect of high OH on the initial return found in Table 5. The uses variable is negative and significant, which suggests that aftermarket performance is worse for IPOs that list many uses of proceeds. The remaining variables in the model are not significant.

8. Summary

SOX imposes important reporting and control requirements on firms that go public, including an audit committee of independent directors, a financial expert on the audit committee who has experience in auditing or analyzing financial statements, internal controls, a quality control process, and a process to communicate information internally. Executives, such as the CEO and CFO, are expected to be involved in establishing the internal controls and are held accountable for reporting inaccuracies. While there are higher compliance reporting costs for firms going public, the requirements should increase the transparency and reduce asymmetric information between newly public firms and prospective investors. Consequently, firms face more scrutiny when attempting to go public.

Our objective is to determine the impact of SOX on the initial IPO returns at the time of an IPO and on aftermarket performance following an IPO. We find that the initial returns of IPOs are significantly lower in the post-SOX period than in the pre-SOX period on average, while initial returns of Canadian IPOs (not affected by SOX provisions) are unchanged since SOX. In addition, the aftermarket performance of IPOs is significantly higher in the post-SOX period than in the pre-SOX period. These results are supported even after controlling for other characteristics unique to the offerings. Firms leave less money on the table during IPOs since SOX.

The same events regarding fraudulent financial reporting that led to the implementation of SOX could have also caused investors to be more cautious. In addition, the SEC imposed new rules to prevent conflicts of interest in the investment banking business. The reforms deserve some of the credit for reducing the excessive hype (and therefore initial returns) at the time of an IPO and reducing the aftermarket correction following IPOs. Although SOX may need to share the credit with other reforms for increased investor scrutiny, our results imply that SOX allows underwriters and investors access to higher quality information when valuing an IPO.

Footnotes

  • 1

    Beyond the establishment of internal controls, the companies now also have to assess their internal controls and summarize their assessment publicly. As of June 2007, public firms are required to provide the following information starting with their second annual report: (1) a management report on their internal control, and (2) an auditor attestation of the management's internal control report.

  • 2

    The Global Settlement of April 2003 determined fines for ten investment banks and created new provisions to separate analysts and investment bankers. It was the result of investigations about investment bank conflicts of interest by the SEC, New York attorney general, New York Stock Exchange, National Association of Securities Dealers, and the North American Securities Administrators Association.

  • 3

    To the extent that the pricing process by the firm and underwriter capitalizes the higher cost of going public since SOX, the offer price should reflect the higher costs of going public. However, if the offer price does not capitalize the higher cost of going public since SOX, any change in underpricing since SOX could be partially attributed to an adjustment by investors. The higher cost of being public could reduce the valuation, and therefore reduce the degree of underpricing. This is especially true for smaller firms. We also test whether the impact of SOX is conditioned on the size of the firms engaged in IPOs in our cross-sectional analysis.

  • 4

    The significance of the results remains unchanged when the data are not winsorized.

  • 5

    Our use of a pricing model to test aftermarket returns presumes that any adverse effects of SOX on firms that are already public before SOX are immediately impounded in the stock prices of those firms. However, IPO firms are unique in that they do not have an investor following or a transparent valuation until the time of the IPO. Thus, the underpricing of IPOs that occur months or years after SOX can still be affected by SOX because of the new structure that allows for a smaller degree of asymmetric information for firms that go public. There is uncertainty surrounding the valuation of an IPO, regardless of whether the IPO occurs before or after SOX. Yet, to the extent that SOX guidelines affect the degree of uncertainty surrounding the valuation of IPO firms, they can affect the underpricing and therefore affect initial returns.

  • 6

    To investigate the possibility that the effects of SOX were realized prior to the date of passage, the analysis was repeated for the alternative event dates of October 1, 2001 (Enron failure) and January 1, 2002 (approximate SOX introduction). The results are similar.

  • 7

    If the pricing process by the firm and underwriter does not capitalize the higher cost of going public since SOX, any change in underpricing since SOX could also be attributed to an adjustment by investors. The higher cost of being public could reduce the valuation, especially for small firms.

Ancillary