Analysis of the financialiImpact of the Sarbanes-Oxley Act on shareholder return and financial distress

(1)

15 AUGUST 2014

Analysis of the Financial Impact of the Sarbanes-Oxley Act on

Shareholder Return and Financial Distress

EMIEL MADDOX Supervisor: R.I. Todorov University of Amsterdam MSc Business Economics, Finance

Master Thesis

ABSTRACT

In this paper a gap in the existing literature about the financial impact of the Sarbanes-Oxley Act will be filled using a difference-in-difference analysis of small and large firms. An explanation will be given of the possible impacts of Sarbanes-Oxley and the differences between small and large firms for the regulations. The overall effects of SOX on financial distress is not proven in this study, but small firms, compared to large, do have more financial distress after SOX. The empirical evidence also tells us on average all firms suffered from SOX introduction in terms of shareholder return. However there is no evidence that small firms had more disadvantages on this area.

(2)

2

THE SARBANES-OXLEY ACT (SOX) has been enacted in 2002 to increase the reliability of financial statements. Before the Enron and WorldCom scandals the resistance against the Act was strong. Thereafter the need for stricter regulations was seen as necessary by the House and Senate, proven by the clear majority of votes. The main creators of SOX are Senator Paul Sarbanes and Representative Michael Oxley, they gave their names to the Act. The aim of the Act is to increase reliability of financial reporting of U.S. publicly listed firms in three different ways. First top management has to individually certify the accuracy of financial information. Second penalties for fraudulent behaviour are larger. Lastly the overseeing role of the board of directors is increased as well as the independency of outside auditors reviewing the financial statements. The Act is mandatory for all publicly listed firms in the U.S. The regulations came into force in the firm’s first fiscal year ending after 15 November 2004.

Many other studies examine the influence of SOX on fraudulent activity of U.S. publicly listed firms. However besides the consequences of the Act on accounting, also the financial consequences are important to research. There can be positive and negative financial consequences of SOX regulations on firm performance. The positive effects relate to more reliability, while the negative effects are linked to the extra cost of regulations.

When fraud is restrained and heavier penalized, it would be expected that managers would take less risk with their fraudulent activities. The risk of being caught and the corresponding penalties are larger. So there will be a decrease in excessive risk taking by a firm. This causes an increase in the reliability of financial statements and lowers the risk of financial distress. An overall increase in financial performance will be the final effect due to the increase in transparency.

On the other hand stricter accounting regulations means firms are more restricted in their activities, which could result in a decrease of firms’ profits. Risk and profits go hand in hand. So if projects with manageable risks and high profits are not taken due to SOX, it will decrease the firm’s performance. For small firms stricter accounting regulations are costly. The costs consist of compliance costs as well as hiring external auditors. So except for the financial costs also underinvestment can be a cost of SOX. If these costs outweigh the benefits of more transparency and reliability in terms of shareholder returns will be researched in this study.

(4)

4

Also if these increased costs outweigh the benefits of less risky and fraudulent behaviour will be researched on a financial distress level.

As with most changes in corporate governance, the introduction of SOX will also come with a divergence in interests. Different stakeholders of the firm have different views on the new regulations, this is researched by Hochberg, Sapienza and Vissing-Jørgensen (2009). Corporations themselves lobbied mainly negatively for the SOX regulations. While investors have a positive view about the law. This difference can be explained that SOX effect both groups differently. For investors it increases transparency and reliability of the firm. On the other hand for corporations it means they are more restricted in both their bookkeeping and their operations. However it is interesting to study if this really is the case: do investors really benefit from tighter regulations? Is the intended increased transparency and reliability of SOX worth it compared to the increased effort it costs the firm to comply with the regulations.

To answer this question in a single paper is almost impossible, since the cost and benefit analysis would become very extensive. In this paper the focus is on two aspects of this problem. The most general question asked in this study is: What has been the impact of the Sarbanes-Oxley Act of 2002 on shareholder return of publicly listed U.S. firms, and is this different for small and large firms? By answering this question the view investors should have on tighter regulations (in this case SOX) is made clear. Also an insight of the effect of accounting regulations on firm value can be acquired. To maintain focussed on the reasons why SOX has been implemented, which is the scandals of 2002, another dependent variable in this study is the financial distress firms experience after the introduction of SOX. This is the second part of the research question: What has been the impact of the Sarbanes-Oxley Act of 2002 on financial distress of publicly listed U.S. firms, and is this different for small and large firms? With the answers on this question a contribution is made to existing literature by explaining if stricter regulations in the form of SOX can indeed decrease excessive risk taking and fraud to prevent scandals as those of 2002.

The aim of this study is to contribute to existing literature to get more insight in the effects of increased regulations on firms as well as the economy. This information is valuable and relevant for public authorities as the government, but also for firms and their stockholders. In previous studies (see Literature review)

The main issue to overcome is the lack of a control group, since the Act applies to all U.S. publicly listed firms. In existing literature this problem is tried to overcome in multiple

(5)

5

ways. This study deals with the problem in a unique way, by using a difference-in-difference approach. The dataset is divided in 2 groups and split up based on firm size. Which gives a difference in compliance to the law. Small firms are more influenced by the regulations, because the costs of compliance are relatively high (Chhaochharia and Grinstein, 2007; Engel, Hayes and Wang, 2007; Leuz, 2007). Furthermore small firms generally have more difficulties in finding qualified independent directors (Holmstrom and Kaplan, 2003). Compliance to the Act is greater for small firms. The other dummy variable in this diff-in-diff analysis is the introduction of SOX for this firm, to describe the ‘post-effect’. The interaction of both variables measures if the introduction of SOX has a different effect on small firms than it has on big firms. With this approach the financial consequences of SOX on the financial performance of U.S. publicly listed firms can be further examined to contribute to existing literature about governance and firm value.

In Section II the literature is examined. The first focus is on background information about SOX. Thereafter the main theories about corporate governance and firm performance are presented. Finally the focus will narrow down to empirical evidence of the effects of SOX on firm performance. In this part the most important theories in relation to this paper are given. To summarize for this section, the main theories and predictions in existing literature are given including the debates they contain. Further it will be explained if the empirical evidence is in line with the theories, and the relationship of both with existing and this research. Also background information about SOX is given in this part.

Section III describes the methodology used and the process of acquiring the necessary data. The hypotheses is presented and an explanation is given how to test the hypotheses with the difference-in-difference model. Furthermore in this Section the data including control variables needed are described. Finally I will use the hypotheses to explain the significance and signs of the coefficients of the analysis as I would expect them to be.

A more extensive insight about the data is given in Section IV, which presents the data and gives descriptive statistics. First the acquiring of the dataset is explained and the dataset is reviewed. To get a feeling of the data, summary statistics and correlations are given for important variables.

The results of this study are presented and evaluated in Section V. Apart from listing the signs and coefficients of the variables of the model, an economic meaning will be given to

(6)

6

them. The coefficients will be placed into context and the first implications of the results will be discussed and related to other papers.

Finally Section VI summarizes the full paper. It also describes the final conclusion about the results and if they are in line with my hypotheses and other related literature. Also the implications of the results and the relationship to other papers are further elaborated. Finally in this section there is also room for the limitations of this paper and directions for further research.

II. Related Literature

The introduction of the SOX law had its main cause in the scandals of 2000. Why the law was implemented and the effects of the law are described in section A. Section B treats the main theories about corporate governance and firm performance. The empirical evidence subject to SOX are examined in section C.

A. Background and provisions of the Sarbanes-Oxley Act

Before going more extensively in the academic literature a complete description of the SOX provisions will be given. Knowledge of the consequences and effects of SOX is useful before going further into the theories and debates as found in the literature. As earlier described the goal of SOX is to improve the quality of the auditing of US public firms. The articles of the act are mainly devoted to giving incentive to firms for increasing their internal controls. This increase in cost and effort should be undone by the long-term benefits of greater accountability and transparency to make the increased regulations worth it.

When SOX regulations was not introduced yet, laws against fraudulent activities and theft were not effective. It led to the corporate scandals of for example Enron and WorldCom around 2000. Financial restatements were rising in numbers during these years (U.S. General Accounting Office, 2002). This went together with a decline in investor confidence, which was not only widely described in the media, but also academically researched (Jain and Rezaee, 2006; Holmstrom and Kaplan, 2003) by measuring bid-ask spreads, trading activity and market depth.

The act is divided in eleven titles. Title I defines the Public Company Accounting Oversight Board (PCAOB). All public accounting companies have to register with the PCAOB to continue their accounting operations. The Board is established to achieve high standards in

(7)

7

the accounting industry. This would serve as protection for stakeholders and the general public against the greed and fraud experienced in the corporate scandals around 2000. The conditions to establish high standards in accounting is described in Title II. Its main focus is auditor independence by separating the interest of the company with the interest of auditors. Creation of a ‘safe’ environment for honest accounting is dealt with in Title III. Members of the audit committee cannot be directly employed by the company and cannot receive any fees or compensation from the company for their role as member. Also employees can anonymously submit complaints to the audit committee without the risk of repercussions. Finally the executive officers have personal responsibility for the soundness of the financial reports and have to design internal controls for the prevention of fraud. Title IV describes rules concerning certain financial instruments and transactions which were heavily used around the 2000s. Rules to achieve greater confidence in security analysts are dealt with in Title V. The provisions made available to support the commission and to enforce its authority are made in Title VI. Title VII describes some studies and articles. Accountability for corporate fraud is explained in Title VII. Finally Title IX-XI deal with white-collar fraud, corporate tax returns, and corporate fraud. These titles are used for making accountability policies for corporate officers and directors of securities issuers stricter. Also they establish more severe penalties for violating securities laws and corporate fraud.

A connection of these new regulations with this study is mostly found within the increased monitoring from independent outside auditors and increased penalties in the case of fraud. The relationship with shareholder return lies mostly within the costs of complying to the rules, but also positively in the fact that increased reliability and transparency can increase investor confidence.

B. Main theories about corporate governance and firm performance

As mentioned in the introduction the separation of ownership and control creates a conflict of interest between shareholders and directors. The corporate governance system of a firm tries to align these interests with the right incentives. SOX regulations creates incentives for increasing the transparency, accountability and reliability of the firm.

For example one of the goals of SOX is to increase the monitoring of sound financial statements by setting rules for auditing. Before the introduction of SOX companies had more freedom for choosing their level of monitoring. In the paper of Jensen and Meckling (1976) highly leveraged firms benefit more from monitoring. Highly leveraged firms are most likely

(8)

8

to be firms in financial distress (because of the inability to pay the debts) or large firms (who are generally more highly levered (Titman and Wessels (1988)). SOX should be beneficial for those two type of firms.

Another goal of SOX is increasing the independence of the board. CEO’s who had poor performance are more likely to get fired with a majority of outside directors (Weisbach, 1988). This would mean that poor performing CEO’s are replaced earlier by possible better performing ones. This should start increasing the shareholder returns again and should lower the possibility on bankruptcy. Also it is found that firms with independent boards act more in line with shareholder interests and make fewer value decreasing acquisitions (Byrd and Hyckman, 1992; Cotter, Shivdasani and Zenner, 1997). On the other hand independent directors’ personal wealth are less sensitive to firm performance. Which would cause them to have less incentive to monitor the firm closely. Possibly leading in worse performance by the manager. This tells us that board independency can have different effects and is not necessarily value increasing and/or decreasing the possibility for bankruptcy or financial distress.

The adjustment of governance standards, like SOX, adds additional requirements to minimum governance standards. These standards try to overcome the principal-agent problem described earlier. However there can be negative consequences to these standards. For some firms the cost of complying can be higher than the actual benefit. This would decrease the value of such a firm. In some cases this also may lead to a going private decision. This is confirmed in the academic studies of Zhang (2007) and Engel, Hayes and Wang (2007). The negative side of introducing regulations to which all firms need to comply with is that for some firms it may lead to a negative trade-off between costs and benefits. Changing governance is value enhancing and good governance depends on a wide variation of firm characteristics. However this is debated by Leuz (2007). In this study it is found that it is not implausible that ‘one-size-fits-all’ regulations could be costly for firms, but the evidence related to SOX is small and we should be cautious when interpreting the evidence.

Other studies about the corporate governance question, includes Yermack (1996) who does not find a relationship between board composition (one of SOX actions) and firm performance. Suggesting the increased oversight and independency of the board due to SOX regulations, would not affect firm performance. However Gompers, Ishii and Metrick (2003) find positive abnormal returns for firms with stronger shareholder rights, which is also an impact of SOX. This would suggest that SOX, which enacts shareholder protection, would have a positive abnormal return as well.

(9)

9

C. Empirical evidence of the effect of SOX on firm performance

In existing literature some papers already examine the financial effects of the introduction of SOX. The number of studies however is relatively small. Although some of them share similarities with this study, this paper will try to go beyond previous studies. Except a review of the important related literature, also the most important differences with the existing literature is given.

Chhaochharia and Grinstein (2007) find results which are comparable to the expectations of this study. However they examine the short-term effects. They compare less compliant (to SOX) to more compliant firms using an event study. They find a positive abnormal return for less compliant firms. However they do not examine the actual improvement in shareholder wealth in the long run. Iliev (2010) finds that Section 404 (independency of accountants) of SOX increased costs and lowered discretionary earnings for firms. Enlightening the costs of SOX.

Other comparable studies that do use a long-term approach, are Hochberg, Sapienza and Vissing-Jørgensen (2009) and Arping and Sautner (2012). They both use 2 different difference-in-difference approaches. Hochberg, Sapienza and Vissing-Jørgensen (2009) examine the effect of SOX on firm performance using a difference-in-difference method, but compared lobbyers (which lobby against the provisions of SOX) with non-lobbyers. They find negative relative returns for lobbyers. As lobbying and non-lobbying is self-selective, the approach in this study focuses on compliancy to SOX regulations like Chhaochharia and Grinstein (2007).

Arping and Sautner (2012) examine transparency (not firm performance) by comparing U.S. firms with cross-listed firms. The results of this study are that cross-listed firms experienced a relatively negative effect on transparency after SOX introduction. Examining cross-listed firms was also a possibility for the analysis in this paper. However the aim is to examine small and large firms, as it should be more effective to control for firm size effects than it is to control for global effects.

Small firms incur large costs of listing publicly (Engel, Hayes and Wang (2007); Leuz (2007)). Going-private transactions of small firms and the corresponding abnormal returns are positively significant to the introduction of SOX.

Studies with comparable research questions are Li, Pincus and Rego (2008) as well as Jain and Rezaee (2006). They both find a positive effect of the announcement of SOX on firm

(10)

10

value on the short term (event studies). Zhang (2007) finds a negative effect on the US market return using a cross-sectional event study with other markets. As they are event studies and focus on the short-term rather than the long term. This thesis will contribute to the literature in describing the long-term effect of SOX regulations on firm performance.

In this paper the impact of the Sarbanes-Oxley Act on financial performance of U.S. publicly listed firms is being researched. The relevancy of the study lies in the fact that the impact of stricter accounting regulations on firm performance are still a subject for research. The exact effects, especially in the long-term are still unknown. While the impact of accounting regulations on board composition and corporate governance have been widely studied. Also economic theory and empirical evidence remains unambiguous causing a need for more research. The paper fills the gap between papers describing that small firms have large costs of compliance and papers describing less compliant firms achieve better performances after SOX. By making a distinction between small and large firms, considering the effects of financial distress and using multiple difference-in-difference regressions to capture the long-term impact of SOX, this study will attempt to go beyond some previous studies.

III. Methodology & Data

Before going further on the data, first the hypotheses are given. Using the following four hypotheses a dataset and a methodology can be constructed to analyse the data. Also it gives a clear view on the direction of this study.

𝐻₁: 𝑇ℎ𝑒 𝑙𝑜𝑛𝑔 𝑡𝑒𝑟𝑚 (𝑝𝑜𝑠𝑡)𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 𝑆𝑂𝑋 𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑠ℎ𝑎𝑟𝑒ℎ𝑜𝑙𝑑𝑒𝑟 𝑟𝑒𝑡𝑢𝑟𝑛 (𝑇𝑆𝑅) 𝑖𝑠 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒

𝐻2: 𝐹𝑜𝑟 𝑠𝑚𝑎𝑙𝑙 𝑓𝑖𝑟𝑚𝑠 𝑡ℎ𝑒 𝐷𝑖𝐷 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑜𝑟 𝑜𝑓 𝑆𝑂𝑋 𝑜𝑛 𝑇𝑆𝑅 𝑖𝑠 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒

𝐻₃: 𝑇ℎ𝑒 𝑙𝑜𝑛𝑔 𝑡𝑒𝑟𝑚 (𝑝𝑜𝑠𝑡)𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 𝑆𝑂𝑋 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑑𝑖𝑠𝑡𝑟𝑒𝑠𝑠 𝑖𝑠 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒

𝐻₄: 𝐹𝑜𝑟 𝑠𝑚𝑎𝑙𝑙 𝑓𝑖𝑟𝑚𝑠 𝑡ℎ𝑒 𝐷𝑖𝐷 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑜𝑟 𝑜𝑓 𝑆𝑂𝑋 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑑𝑖𝑠𝑡𝑟𝑒𝑠𝑠 𝑖𝑠 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒

An explanation of how I acquire the data and set up a model to test these hypotheses is given in the remainder of this section.

(11)

11

From these hypotheses it is clear that data is needed on the SOX introduction date for firms, size of the firm, shareholder return and an indicator of financial distress. The indicator used for financial distress is the Z-score (Altman, 1984). It is necessary to use this parameter as financial distress itself is hard to quantify. To construct this important dependent variable data about assets, liabilities, retained earnings, working capital, operating income, sales and market capitalization is needed. Negative working capital is an indicator for firms that lack the assets to meet their short-term obligations. When companies have high retained earnings compared to their assets, it suggests a history of profitability and that they are able to cope with a bad year of losses. Operating income shows us the ability of a firm to make profits, before interest and tax are deducted. Further a ratio is added to give a market value dimension to the Z-score model, which isn’t based on fundamentals. This is the market capitalization divided by total liabilities. A durable market capitalization can be viewed as the investors’ confidence in a solid financial position of the company. Lastly sales divided by total assets shows us how efficiently the firm generates sales by using their assets. The formula is listed below:

Z-score = (Working Capital

Total Assets ×1.2) + ( Retained Earnings Total Assets ×1.4) + ( EBIT Total Assets×3.3) + (Market Capitalization Total Liabilities ×0.6) + ( Sales Total Assets×1.0)

Most of the other data is immediately available in WRDS. Stock returns for calculating the other dependent variable, TSR, are available in CRSP. Firm information, like market capitalization and sales for firm size and the Z-score, is available from CRSP/Compustat. Data from all non-financial publicly listed firms in the U.S. are used. This is gathered from the CRSP/Compustat database. The time frame is 2000-2012. Stock exchange codes of 0 and 1 are removed from the dataset, because these are foreign firms or not publicly traded firms.

The goal of this study is to extensively analyse the effect of SOX on firm performance in the long term. The basic methodology of a difference-in-difference study is used. The main idea is to use a difference-in-difference model with before/after the introduction of SOX as one of the dummy variables (𝐷𝑡) and small firms versus large firms as other dummy variable (𝐺𝑖).

The interaction between the two previous variables is described as 𝑋_𝑖𝑡 = 𝐺_𝑖 × 𝐷_𝑡. The dependent variables (𝑌_𝑖𝑡) used will consist of indicators of firm performance and stability like shareholder wealth, volatility in firm value and parameters of financial distress, included in separate regressions. Variables (𝑊𝑖𝑡) are added to measure individual characteristics prior to

(12)

12

the experiment. Fama-French data (SMB, HML, Rm-Rf and UMD) are an example and are also available in WRDS (Fama and French, 1993). The regression model will look like this:

𝑌_𝑖𝑡 = 𝛽₀+ 𝛽₁𝑋_𝑖𝑡+ 𝛽₂𝐷_𝑡+ 𝛽₃𝐺_𝑖+ 𝛽₄𝑊_𝑖𝑡+ 𝑢_𝑖𝑡

The other included control variables are firm age, leverage, audit fees and return on assets. They are used to measure the pre-treatment effects of firms and control for them. It is important that they determine the dependent variable and are correlated with the independent variable.

Firm age is used in various empirical studies, for example it is included in Hochberg, Sapienza and Vissing-Jørgensen (2007) as control variable. Firm age can also tell something about the complexity of the firm’s operations. An older firm is usually more developed and has more operations than a firm that just entered the market. Also older firms have acquired a reputation and have experience, which should make them less risky and decrease their chance on financial distress.

Leverage will also be used as a control variable. From the studies so far, it can be easily stated that large firms in the US have greater leverage ratios than smaller firms, so by using it, there is control for differences between the firms’ sizes. Leverage can be also thought of as a proxy for risk because the higher the leverage, the higher the risk.

Audit fees represent another control variable as they have a high influence on the firm’s cost and performance. As SOX is considering regulations about the auditing process, a control variable is needed to measure the cost of firms with the audit. Also audit fees are a fixed and exogenous cost, this means small firms would be at a disadvantage.

The paper of Hochberg, Sapienza and Vissing-Jørgensen (2009) tells us that firms who are characterized by high expected compliance costs or agency problems are likely to be more affected by SOX. This means firms with traditional free cash-flow problems, including firms with high profitability, are more likely to be affected and thus to lobby against SOX. Both theories would lead to a decrease in the mean return on assets after the introduction of SOX.

The expectations for this study are that both for large and small firms the introduction of SOX negatively affects their performance. This means a lower Z-score and lower total shareholder return. For small firms the costs of stricter accounting regulations of SOX

(13)

13

outweigh the benefits even more. This includes that small firms are predicted to have even bigger disadvantages. This means the diff-in-diff estimator for both Z-score and TSR are predicted to be significantly negative.

IV. Data and Descriptive Statistics

Most of the data from the methodology is available on the CRSP/Compustat merged database. The starting dataset consisted out of 57.298 observations or 8227 U.S. firms over the period of 2000-2012. Excluding the financial firms with SIC codes between 6000 and 6999. For adding the necessary control variables, the main dataset is merged with the Fama-French benchmark factors and with audit fee amounts from AuditAnalytics. Using this data the differences within large and small firms, before and after SOX introduction are analysed to find effects on probability on bankruptcy (Altman Z-score) and total shareholder return (TSR).

A. Acquiring the dataset

The remaining dataset consists of 34.899 observations or 5447 firms after dropping observations. This means less than 50% of the dataset had been dropped. The biggest amount (60%) of dropped observations were due to a market capitalization under $100M. Firms with a stock exchange code of 0 are not publicly traded and thus excluded. Other reasons to drop firms or observations include missing variables and observations which were not logical, like a negative amount of assets, CapEx or sales. Some of the necessary variables still have to be constructed. How they are constructed can be seen in Appendix 1. There are minor differences (0%-10%) in the summary statistics of both datasets. However standard deviations are relatively high, so there is not much evidence that dropping the variables has influenced the reliability of the dataset.

B. Summary statistics

Table 1A and 1Bgive you some feeling about the data. The table is split in two parts. In Table 1A you can see the summary statistics for small and large firms before SOX introduction and in Table 1B you can see the statistics for after SOX introduction. It is not possible to conclude anything out of these tables, but they help to understand the data better. However, even without statistical proof so far, some interesting information can be extracted out of these tables. The tables include the following (dependent) variables: Z-score and total shareholder return. Also the summary statistics of some of the control variables are found in

(14)

14

the tables. They also contain interesting information about the dataset, they include: firm age, leverage and audit fees.

TABLE 1.A

The Z-score is a measurement of probability on bankruptcy. From basic economic theory as well as the goal of SOX, it would be expected that the Z-score would increase after the introduction of SOX. SOX has been introduced to tighten the regulations concerning the reliability of financial statements. Giving less room for taking risk and fraudulent activities we would expect the probability of bankruptcy for firms will go down after the introduction of SOX, meaning an increase in the Z-score. However comparing Table 1.A with Table 1.B it can be seen that the Z-score decreased after the introduction of SOX. Firms on average performed worse on the characteristics that determine the Z-score, meaning they were more susceptible for bankruptcy. This can have multiple causes, ranging from the financial crisis to the cost of

Dependent var. Obs Mean Median St.Dev Min Max

TSR 4,747 0.11 0 0.62 -0.79 2.49

Z-Score 7,077 7.45 4.55 9.31 -5.88 41.83

Control var. Obs Mean Median St.Dev Min Max

Firm Age 7,077 8.88 6 8.72 0 42

Leverage 7,077 1.64 1.36 1.66 -6.25 13.95 Audit Fee ($) 6,325 384,799 251,000 437,158 0 6,625,260

RoA 7,077 -0.05 0.03 0.25 -0.88 0.37

Summary statistics of dependent and control variables after SOX introduction

Small firms

TSR 5,069 0.15 0.06 0.52 -0.79 2.49

Z-Score 6,705 3.83 2.83 4.29 -5.88 41.83

Firm Age 6,705 16.70 11 13.67 0 42

Leverage 6,705 2.47 2.06 2.20 -6.25 13.95 Audit Fee ($) 5,722 2,209,601 951,112 4,040,544 0 81,200,000

RoA 6,705 0.06 0.06 0.12 -0.88 0.37

(15)

15

complying with SOX regulations. From Altman (1984) we know that the Z-score does not work as good on small firms as it does on large firms. For small firms the score usually will be larger. Below a total asset amount of $1M the Z-score is problematic. However in this dataset there are no firms with total assets lower than $1M, suggesting that the method should be feasible. Also we filtered the dataset to exclude firms with market capitalization below $100M, this would filter the smallest firms as well. However if the mean and median Z-score will be higher or lower for small firms cannot be predicted very well in advance. From the Tables we can see that the Z-score for smaller firms is higher than that of larger firms.

TABLE 1.B

The expectations I have considering return on assets before and after SOX are also based on economic theory but also on the paper of Hochberg, Sapienza and Vissing-Jørgensen (2009). Economic theory would say that the costs of complying with SOX is reduced from your

TSR 10,464 0.07 0 0.53 -0.79 2.49

Z-Score 11,584 5.94 3.91 7.68 -5.88 41.83

Firm Age 11,584 11.56 9 10.19 0 50

Leverage 11,584 1.74 1.42 1.8 -6.25 13.95 Audit Fee ($) 11,419 874,290 705,000 798,116 0 28,200,000

RoA 11,584 0.00 0.05 0.22 -0.88 0.37

Summary statistics of dependent and control variables after SOX introduction

Small firms

TSR 10,193 0.10 0.04 0.48 -0.79 2.49

Z-Score 10,533 3.50 2.90 3.15 -5.88 41.83

Firm Age 10,534 19.84 15 15.13 0 50

Leverage 10,534 2.45 2.04 2.26 -6.25 13.95 Audit Fee ($) 10,534 4,580,833 2,433,805 6,608,681 0 90,200,000

RoA 10,533 0.07 0.07 0.10 -0.88 0.37

(16)

16

profit and thus it will decrease the returns on assets. The costs are relatively higher for small firms so it will reduce the returns on assets for small firms even more. Because the financial statements should become more reliable and transparent, economic theory would say that the standard deviation of the returns on assets should decrease, since the variation on earnings and assets caused by fraud would be less. The paper of Hochberg, Sapienza and Vissing-Jørgensen (2009) tells us that firms who are characterized by high expected compliance costs or agency problems are likely to be more affected by SOX. This means firms with traditional free cash-flow problems, including firms with high profitability, are more likely to be affected and thus to lobby against SOX. Both theories would lead to a decrease in the mean return on assets after the introduction of SOX. However in both Tables it is seen that in contradiction to the expectations the return on assets has increased after the introduction for both small and large firms. It might have increased because of the increase in ROA of some firms in the sample. For example, auditing firm’s ROA might have improved substantially. Another reason might be that just after the introduction of SOX, many managers were still conservative when writing their financial reports. They might have lowered the accruals which led to a higher ROA. The prediction of decreasing variations is right.

For total shareholders return, first of all the number of observations should be smaller. Which can be confirmed in the tables. This is because it is a variable using a lag, meaning for before and even for after the introduction of SOX some observations are missing. Further most importantly I would expect that consistent with SOX goals the variability of returns would decrease after the introduction of SOX. The regulations should make the financial statements more reliable and also the need for restatements smaller. More transparent financial statements would also mean that the stock prices will be more stable. Also for small firms we would expect higher variation in shareholders return, as there is more risk involved with smaller firms. Comparing both tables this predictions are confirmed. Less interesting in this case is the change in the average and median shareholder return. We know from basic knowledge that probably the financial crisis in 2007-2008 would cause this numbers to decrease, as can be seen in the tables.

Considering firm age it can be expected that after the introduction of SOX firms get older, because the tighter regulations decrease the chance on bankruptcy. However SOX carries relatively high compliance costs for small firms, so for small firms this effect can be expected to be smaller. From comparing the Tables this is confirmed as small firm’s age increased by

(17)

17

2.86 years and large firm’s age by 3.14 years. If this can be attributed to SOX is of course very unclear as there can be numerous other factors that cause the aging of firms.

Smaller firms sometimes have problems with debt financing. One of SOX goals is to decrease asymmetric information between stockholders and the firm. Because financial statements are now more reliable and transparent, debt financiers might now be more willing to issue debt to small firms. However larger firms are usually already more debt financed, they have established long-term relationships with financiers to easily acquire more debt. This combined with the credibility principle, which means firms can use leverage as a way to convince investors that the firm has good growth prospects, would suggest that larger firms could have less benefits from a high leverage. Thus the average leverage would decrease for large firms. The summary statistics in Table 1.A and 1.Bconfirm this, as small firms increased their leverage after SOX introduction and large firms decreased their leverage. Again if this really is caused by SOX cannot be told from this table.

As firms need to comply with SOX and its auditing regulations, audit fees should increase. They should increase relatively more for small firms as the auditing costs are more or less fixed. Also all firms should pay auditor fees after the introduction of SOX, since they have to be audited independently. Apparently 86 firms could avoid the regulations or did not pay auditor fees, causing less observations for auditor fees after SOX. In Tables 1.A and 1.B it can be seen that the fees increased tremendously after SOX introduction. Probably this is also because of price inflation. However indeed relatively auditor fees increased more for small firms than they did for large firms (127% compared to 107%).

C. Cross-correlation table

Also from researching the correlations of some of the variables interesting conclusions can be drawn. The most interesting variable to check for correlations is the dummy variable for the introduction of SOX. In the case of correlation it is less interesting to look at the size of the firms as it is not possible to control for the effects of bigger firms. The lack of control for variables is important to mention considering correlation tables. It should be clear that correlation does not imply causation. The correlations between variables of interest can be found in Table 2.

The first two interesting correlations are between SoxEffect and HML and SMB. The introduction of SOX is negatively correlated with the value and size premium. So after firms were under the regulations the premium for being a small and for being a high book-to-market

(18)

18

value firm decreased. Audit fees are as could be expected positively correlated with the introduction of SOX. After the introduction of the regulations firms started to pay more fees for auditing. The Z-score is negatively correlated and thus decreased after the introduction. This would be a bad sign as it would be expected that the Z-score increases because of the regulation as less firms get into financial distress. Also shareholder returns are negatively correlated with the introduction of SOX. This is as predicted.Further all the correlations related to Z-score and TSR are interesting to view, the correlations with the Fama-French factors are logical considering the Fama-French model. Also the correlations of the other control variables with the Z-score are interesting. For instance it is good and especially logical to find that leverage is negatively correlated with Z-score. A higher leverage would mean higher debt payments and this leads to increased chance on bankruptcy or a lower Z-score.

TABLE 2

V. Results

To analyse the effects of the introduction of SOX a difference-in-difference analysis has been done with a dummy variable for post-SOX and a dummy variable for small firms. Firm size as treatment effect is interesting as the compliance to SOX should differ between small and large firms. Small firms are much more influenced by the cost and other difficulties belonging to the regulations of SOX (Chhaochharia and Grinstein, 2007; Engel, Hayes and Wang, 2007; Holmstrom and Kaplan, 2003; Leuz, 2007). Using the difference-in-difference

Variables SoxEffect (Rm-Rf) HML SMB UMD Firm age RoA Leverage CapEx Audit fees Z-score TSR SoxEffect 1 (Rm-Rf) 0.0598 1 HML -0.4488 -0.0656 1 SMB -0.5419 0.1387 0.3325 1 UMD -0.0371 -0.5851 0.0063 -0.3532 1 Firm age 0.082 0.0153 -0.0588 -0.0424 -0.0099 1 RoA 0.0641 0.0481 -0.0457 -0.073 0.018 0.1653 1 Leverage 0.008 -0.0139 -0.0138 -0.0049 0.0032 0.0614 0.0021 1 CapEx 0.0563 -0.0093 -0.0349 -0.0347 0.0025 0.1555 0.0964 0.0542 1 Audit fees 0.1313 0.011 -0.0745 -0.0862 -0.0101 0.2252 0.0865 0.1085 0.5831 1 Z-score -0.0511 0.0274 0.0463 0.027 0.0044 -0.1018 0.2123 -0.1751 -0.1081 -0.1305 1 TSR -0.0505 0.3879 0.0138 0.1705 -0.2757 -0.0223 0.1279 -0.008 -0.0242 -0.0117 0.1103 1 Dark grey = Interesting correlations or lack of correlations

Light gray = Variables predicted to be correlated but are not Correlation table of variables of interest

(19)

19

method we can analyse both the overall (post-) effect of SOX as well as compare the effects it has on small firms with the (post*treatment-) effects it has on large firms.

The regression is done with three separate dependent variables. The first dependent variable is the total shareholder return (TSR). This is the share price in period n minus the share price in period n-1, divided by the share price of period n-1. Total shareholder return is adjusted for stock splits and dividends. This is an important outcome variable as it measures the increase in shareholders wealth, which a firm should strive for. The other dependent variable is the Z-score, which is a measurement of multiple ratios predicting the possibility on bankruptcy (see the Methodology section for the formula). This is interesting to research as it indicates if SOX regulations can indeed prevent financial distress in firms. If there would be a significant increase in the Z-score, this would mean SOX would increase the financial performance of firms.

Control variables are added to measure individual characteristics prior to the introduction of SOX. Fama-French control variables take care of market-specific events (Fama and French, 1993). They include small-minus-big (SMB), high-minus-low (HML), market risk factor (Rm-Rf) and up-minus-down (UMD) (Jegadeesh and Titman, 1993). The control variables firm age and return on assets are added based on the correlation they have with firm size, but also with the introduction of SOX (Hochberg, Sapienza and Vissing-Jørgensen, 2009). They also are determinants of shareholder return and the probability of bankruptcy (aged firms have different characteristics as young firms). Leverage is added based on Arping and Sautner (2012). It is a determinant of shareholders return and the probability of bankruptcy. High debt can be an indicator of financial distress. Also from Jensen and Meckling (1976), highly leveraged firms benefit more from the increase in monitoring from complying with the regulations of SOX. Audit fees represent another control variable as they are a major part of SOX compliance costs. As SOX is considering regulations about the auditing process, a control variable is needed to measure the cost of firms with the audit. Also audit fees are a fixed and exogenous cost, this means small firms would be at a disadvantage. It is a control variable widely used in papers studying SOX and it is included in most of the papers from Section II.C. The first column of Table 3 shows that there were 29383 observations in the regression model for the effects of SOX on total shareholder return. The regression has an R2, which is a measure for goodness of fit of the model, of 0.182. This means the model can explain 18.2% of the variation within TSR. Looking at the PostSox coefficient of -0.0292 and the information that it is significant at the 1% level, it can be concluded that SOX had a negative effect on TSR.

(20)

20

A decrease in shareholder return of almost 3% is also economically very significant. Imagine that the interest rate return of your saving accounts would go down by 3%, which is a very large effect. It cannot be proven however that small firms got hit harder as the coefficient is 0.0134, which would even suggest that small firms have gotten more benefits from SOX than large firms. This number is economically significant, but statistically it is not. So it cannot be proven that the effect is significantly different from 0.

TABLE 3

The regression on the dependent variable, Z-score, can be examined using Column 2 of Table 3. For this regression there are 33798 observations. This is more than the regression for TSR, because some return observations are missing since TSR is a lagged variable. The R2 equals 0.129, which means that 12.9% of the variation in Z-score is explained by this model. Table 3, Column 2, shows a positive insignificant coefficient for the post-SOX average Z-score at the 1% level. So the overall effect of the introduction of SOX is insignificantly positive. The

(1) (2)

Coeff. St. Err. t Coeff. St. Err. t

PostSOX -0.0292*** 0.0086 -3.42 0.1095 0.0834 1.31 SmallFirm -0.0423*** 0.0099 -4.27 3.617*** 0.2109 17.15 PostSOX * SmallFirm 0.0134 0.0112 1.21 -1.507*** 0.1980 -7.61 (Rm-Rf) 0.9487*** 0.0165 57.64 1.0521*** 0.1775 5.93 SMB 0.6706*** 0.0488 13.74 -1.2467*** 0.3822 -3.26 HML -0.0342 0.0367 -0.93 3.5904*** 0.3571 10.05 UMD -0.0909*** 0.0193 -4.71 0.4279*** 0.1286 3.33 Firm age -0.0019*** 0.0002 -10.20 -0.0434*** 0.0044 -9.82 Leverage -0.001 0.0018 -0.57 -0.4269*** 0.0250 -17.10 Audit fees ($M) -0.0016*** 0.0006 -2.91 -0.0703*** 0.0092 -7.61 RoA (%) 0.0035*** 0.0002 19.01 0.0799*** 0.0055 14.67

Firm fixed effects Yes Yes

Year fixed effects No No

Firm-Year fixed effects Yes Yes

N 29383 33798

adj. R-squared 0.182 0.129

The effects of introducing SOX on financial distress (Z-score) and total shareholder return (TSR)

This table gives insight in the determinants of a firm's Z-score (column 1) and TSR (column 2). PostSox is a dummy variable that takes the value 1 for fiscal years after compliance with SOX. SmallFirm is a dummy variable that equals 1 if the average log(sales)

of a firm was smaller than 6.5 during the period 2002-2010. Standard errors are clustered at firm level and robust. For an explanation of all variables see Appendix 1. Constants were added to the regressions but are not reported. *, **, and *** indicate a

significance at 10%, 5% and 1% respectively.

(21)

21

coefficient is also economically quite insignificant. A change in Z-score after SOX with the coefficient of 0.1095 is small compared to the values that Z-scores normally takes (see Table 1.A and 1.B). The coefficient of 0.1095, compared to the mean Z-score for small firms before and after SOX, is very small. Therefore it can be concluded that the Z-score does not significantly differ from 0. There is no proof for an overall effect of SOX on the Z-score. The same column shows a negative significant difference-in-difference estimator for the Z-score of firms. The coefficient of -1.507 is large enough to have economic significance (Table 1.A). Also the coefficient is statistically significant at the 1% level. This means that indeed small firms are more severely hit by the introduction of SOX and their Z-score, compared to large firms, has decreased significantly more. On average the decrease was 1.507 which is a large for a Z-score. This means that small firms had increased probability to bankruptcy compared to the larger companies. So SOX did not help on increasing Z-score numbers and for small firms it was even harmful for the Z-score.

VI. Conclusion and Discussion

This paper further introduced the topic of regulation and its possible effects on financial performance. The main interest of this study was to get more insight in shareholder returns and probability on bankruptcy (Z-score) for small and large firms. A difference-in-difference approach is used to compare firms who are more influenced by the SOX regulations (small firms) with firms who are less influenced (large firms). The lack of control group remains a point of concern. However firm size is a good criterion for defining the control group. So in this paper the U.S. publicly listed firms are split up in a portfolio with a small firm size and a portfolio with large companies. Most important reason to choose firm size as control group is that the effects of SOX are very different on small firms compared to large firms. For large firms complying with SOX has very little implications. Auditing costs are mostly fixed and it would not cause any problems to find independent directors for larger firms. For small firms complying with SOX is much more problematic.

The paper fills the gap between papers describing that small firms have large costs of compliance and papers describing less compliant firms achieve better performances after SOX. Some of the results are strong, but the effects of SOX regulation remains difficult to describe. The overall effects of SOX on financial distress is not proven in this study, but small firms, compared to large, do have more financial distress after SOX. The empirical evidence also tells us on average all firms suffered from SOX introduction in terms of shareholder return.

(22)

22

However there is no evidence that small firms had more disadvantages on this area. A possible explanation is that in small firms there is more space for ‘creative’ accounting, and SOX regulation increases the trust of investors in small firms. However this is suggestive and could be a possible future direction for research.

SOX regulation is costly and in this study there is evidence that it can even be costly in terms of total shareholder return. Also there is evidence given that it affects the probability of bankruptcy (Z-score) for small firms negatively. This is mostly in line with literature, although there is much debate. Most of the literature find evidence of worse financial performance when the governance structure of firms is changed. Which is in line with the results of this paper. Also the increased chance on bankruptcy of small firms is in line with most literature. Which mostly conclude that the costs of compliance outweigh the benefits of regulation for small firms.

Limitations of my study are mainly in the implications the diff-in-diff approach has on policy changes. As SOX consist of many aspects it is hard to separate the effect of interest. This problem is overcome by looking at the SOX regulation as a whole. However still the policy change can be endogenous. Through the control variables the effect of SOX only is isolated as much as possible. Also the period before SOX can be contaminated by expectations of the enforcement of the law.

Therefore it might be a good suggestion to be very careful as policy maker to further tighten accounting regulations. The cost of regulation possibly consists of much more factors than only financial ones and it could also contain the costs of firms getting more restricted in their activities for example. This could also be a venue for further research. Possibly in this case it was also important to give a signal that fraud has to be scaled down, as a way to deal with the scandals of 2000. However chances are great that firms are always going to find ways to work around the laws and it is not possible to endlessly tighten accounting regulations.

(23)

23

APPENDIX 1

Definition of variables

The definitions of variables used in this empirical analysis and their data sources are listed in the table below.

Variable Definition Source

Total Shareholder Return

The total return of a stock to an investor. It equals the capital gain + dividends and is adjusted for stock splits (winsorized, 0.01)

CRSP

Z-score Based on Altman (1984). It measures the probability of bankruptcy for a firm (winsorized, 0.01)

Self-constructed

SoxEffect A dummy variable indicating if SOX regulations were introduced for the firm at that particular time

Self-constructed

SmallFirm A dummy variable equal to 1 if the firm had log(sales) smaller than 6.5

Compustat Leverage Book value of total assets divided by the

book value of equity (winsorized, 0.01)

Compustat Audit Fee Audit fees paid by a company AuditAnalytics Return on Assets Pre-tax income divided by total assets of a

company (winsorized, 0.01)

Compustat Firm Age The current fiscal year minus the first

effective date of the firm according to the database

CRSP/Compustat

(Rm – Rf) The market risk premium Fama-French

Benchmark Factors Small-minus-Big The (size) premium for small firms Fama-French

Benchmark Factors High-minus-Low The (value) premium for high

book-to-market value firms

Fama-French Benchmark Factors Up-minus-Down The (momentum) premium for firms with

high past returns

Fama-French Benchmark Factors

(24)

24

REFERENCES

Altman, E.I., 1984, The Success of Business Failure Prediction Models: An International Survey, Journal of Banking and Finance 8 (2), 171-198.

Arping, S. and Z. Sautner, 2012, Did SOX Section 404 Make Firms Less Opaque? Evidence from Cross-Listed Firms, Contemporary Accounting Research, forthcoming.

Bebchuk, L., A. Cohen and A. Ferrell, 2009, What Matters in Corporate Governance?, Review of Financial Studies 22 (2), 783-827.

Byrd, J. and K. Hickman, 1992, Do Outside Directors Monitor Managers? Evidence from Tender Offer Bids, Journal of Financial Economics 32 (2), 195-207.

Chhaochharia, V., Y. Grinstein, 2007, Corporate Governance and Firm Value: The Impact of the 2002 Governance Rules, The Journal of Finance 62 (4), 1789-1825.

Cotter, J., A. Shivdasani and M. Zenner, 1997, Do Independent Directors Enhance Target Shareholder Wealth During Tender Offers?, Journal of Financial Economics 43 (2), 195-218.

Engel, E., R.M. Hayes and X. Wang, 2007, The Sarbanes-Oxley Act and firms’ going-private decisions, Journal of Accounting and Economics 44 (1), 116-145.

Fama, E.F. and K.R. French, 1993, Common risk factors in the returns on stocks and bonds, Journal of Financial Economics 33 (1), 3-56.

Gompers, P., J. Ishii and A. Metrick, 2003, Corporate Governance and Equity Prices, The Quarterly Journal of Economics 118 (1), 107-156.

Hochberg, Y., P. Sapienza and A. Vissing-Jørgensen, 2009, A Lobbying Approach to Evaluating the Sarbanes-Oxley Act of 2002, Journal of Accounting Research 47 (2), 519-583.

Holmstrom, B., S.N. Kaplan, 2003, The State of U.S. Corporate Governance: What’s Right and What’s Wrong?, Journal of Applied Corporate Finance 15 (3), 8-20.

Iliev, P., 2010, The Effect of SOX Section 404: Costs, Earnings, Quality, and Stock Prices, Journal of Finance 65 (3), 1163-1196.

Jain, P.K. and Z. Rezaee, 2006, The Sarbanes-Oxley Act of 2002 and Capital-Market Behavior: Early Evidence, Contemporary Accounting Research 23 (3), 629-654.

Jegadeesh, N., S. Titman, 1993, Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency, The Journal of Finance 48 (1), 65-91.

Jensen, M., and W. M. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs, and ownership structure, Journal of Financial Economics 3 (4), 305–60.

(25)

25

Leuz, C., 2007, Was the Sarbanes–Oxley Act of 2002 really this costly? A discussion of evidence from event returns and going-private decisions, Journal of Accounting and Economics 44 (1), 146-165.

Li, H., M. Pincus and S. Rego, 2008, Market Reaction to Events Surrounding the Sarbanes‐ Oxley Act of 2002 and Earnings Management, Journal of Law and Economics 51 (1), 111-134.

Titman, S. and R. Wessels, 1988, The Determinants of Capital Structure Choice. The Journal of Finance 43 (1), 1–19.

U.S. General Accounting Office, 2002, Financial Statement Restatements: Trends, Market Impacts, Regulatory Responses, and Remaining Challenges, GAO-03-138.

Weisbach, M., 1988, Outside Directors and CEO Turnover, Journal of Financial Economics 20 (1), 431-460.

Yermack, D., 1996, Higher market valuation of companies with a small board of directors, Journal of Financial Economics 40 (2), 185-211.

Zhang, I.X., 2007, Economic Consequences of the Sarbanes-Oxley Act of 2002, Journal of Accounting and Economics 44 (1), 74-115.

Analysis of the financialiImpact of the Sarbanes-Oxley Act on shareholder return and financial distress

Analysis of the Financial Impact of the Sarbanes-Oxley Act on

Shareholder Return and Financial Distress

Contents