An Analysis of the Underlying Causes Attributed to Restatements

Abstract

The dramatic increase in the number of restatements filed over the past years has been attributed to numerous causes, including the complexity of the accounting standards, internal control reviews, changes in materiality thresholds, the overly conservative nature of auditors, earnings management, increased transaction complexity, and the second-guessing of management judgments by a variety of interested parties. However, empirical evidence on the underlying causes of restatements has been lacking. This study provides such evidence by directly addressing these questions: (1) To what causes do companies attribute restatements? (2) To what characteristics of the accounting standards do companies attribute restatements? Relying on the restating companies' disclosures about restatements, we find that companies most often attribute restatements to basic internal company errors unrelated to any specific characteristic of the accounting standards. We also find that, for those restatements attributed to some characteristic of the accounting standards, the primary contributing factor is the lack of clarity in applying the standards and/or the proliferation of the literature because the original standard lacked clarity. These findings should interest standard setters and regulators addressing the proliferation of restatements and academics using restatements as proxies for constructs of interest in research. [PUBLICATION ABSTRACT]

Headnote

SYNOPSIS: The dramatic increase in the number of restatements filed over the past years has been attributed to numerous causes, including the complexity of the accounting standards, internal control reviews, changes in materiality thresholds, the overly conservative nature of auditors, earnings management, increased transaction complexity, and the second-guessing of management judgments by a variety of interested parties. However, empirical evidence on the underlying causes of restatements has been lacking. This study provides such evidence by directly addressing these questions: (1) To what causes do companies attribute restatements? (2) To what characteristics of the accounting standards do companies attribute restatements? Relying on the restating companies' disclosures about restatements, we find that companies most often attribute restatements to basic internal company errors unrelated to any specific characteristic of the accounting standards. We also find that, for those restatements attributed to some characteristic of the accounting standards, the primary contributing factor is the lack of clarity in applying the standards and/or the proliferation of the literature because the original standard lacked clarity. These findings should interest standard setters and regulators addressing the proliferation of restatements and academics using restatements as proxies for constructs of interest in research.

Keywords: restatement; restatement causes; accounting standards; accounting complexity.

Data Availability: The data are available from public sources.

JEL Classifications: K22; M41; M43; M49.

INTRODUCTION

Accounting restatements have been filed at record levels in the past few years: Glass Lewis & Co. (2006) documents that 1,538 restatements were filed in 2006, more than three times the 475 restatements filed in 2003. Although the rapid increase in the number of restatements filed in the United States is apparent, a significant debate remains regarding the underlying cause of this increase. Various groups have posited plausible reasons for restatements, including accounting complexity, second-guessing of management judgment, proliferation of accounting rules and implementation guidance, application of the Sarbanes-Oxley Act of 2002 (SOX) Section 404 requirements, transaction complexity, and earnings management. The various constituents have also suggested specific measures that could be taken to reduce restatements based on the causes suggested. However, measures proposed to curtail restatements based on one suggested cause may contradict measures proposed based on another suggested cause. This conflict highlights the necessity for first understanding the underlying cause of restatements before implementing measures to curtail them. To date, as the Advisory Committee on Improvements to Financial Reporting (SEC 2008) noted, definitive evidence on the drivers of restatements is lacking.

In this study, we analyze disclosures surrounding restatements filed from 2003 to 2006 to provide empirical evidence of the restating companies' explanations of the underlying cause of restatements. Relying on these disclosures, we classify each restatement as having been attributed to one of the following four causes: (1) an internal company error; (2) intentional manipulation; (3) transaction complexity; or (4) some characteristic of the accounting standards. For restatements attributed to some characteristic of the accounting standards, we also consider whether the companies' disclosures suggest that the restatement is most consistent with either (1) a lack of clarity in the standard and/or the proliferations of the accounting literature because of the lack of clarity in the original standard; (2) the use of judgment in applying the standard; or (3) the misapplication of detailed and complex rules. While restating companies may have strategic incentives regarding how they disclose restatements, we argue that these disclosures are informative and that classifying the restatements based on an analysis of those disclosures improves our understanding of the underlying causes of restatements and of the attributes of accounting standards that might contribute to them.

For a subset of our sample, we calculate the net income effect of the restatement to better understand whether materiality levels related to restatements have become more conservative over time and to compare with prior research. Finally, we examine whether a company's reporting of a material internal control weakness relates to the underlying cause of the restatement identified.

We document that the majority of restatements (57 percent) filed from 2003 to 2006 are attributed to internal company errors, inconsistent with the conventional wisdom that the complexity of the accounting standards drives most restatements. The second most commonly attributed cause of restatements filed during the four-year period is some characteristic of the accounting standards (37 percent). Of those restatements, 58 percent are related to a lack of clarity in the standard, and 37 percent are related to the use of judgment in applying the standard. Overall, we document that the mean net income effect of restatements is negative ((0.013), although a significant number of restatements have no net income effect (26 percent). The proportion of restatements for which the absolute net income effect is greater than 5 percent of total assets decreased over the four-year period, consistent with a change in the quantitative materiality threshold across time. Finally, we find that only 66 percent of the restatements attributed to manipulation and 44 percent attributed to internal errors are from companies that report material internal control weaknesses. This suggests deficiencies in the implementation of SOX internal control weakness attestation and reporting requirements.

Our results provide insights to regulators and standard setters regarding the causes to which companies attribute restatements, which might aid in designing and implementing initiatives to reduce the number of restatements. In addition, our finding that companies frequently attribute restatements to an internal company error is consistent with SOX reviews working as expected and suggests the number of restatements will decline as improved controls are established. The findings regarding restatements attributed to judgment in the standards are especially important to consider as the United States moves toward convergence with International Financial Reporting Standards, which are often considered more principles-based and reliant on judgment. Finally, the finding of inconsistencies between restatement filings attributed to basic internal company errors and manipulation and reported material internal control weaknesses suggests a need to address issues related to improving the attestation of internal controls and the reporting of material weaknesses.

Our results are also useful to academic research that employs restatements. Researchers may benefit by identifying a subset of restatements based on the attributed underlying cause that most closely relates to the construct of interest to ensure that the restatement sample is appropriate for the research question. Prior academic research that examines restatements either does not distinguish between different types of restatements or merely sorts restatements into unintentional errors or intentional manipulations (i.e., Hennes et al. 2008; Palmrose et al. 2004). This study considers a broader range of causes and factors related to restatements, distinguishing between those attributed to unintentional internal books and records deficiencies, misapplications of accounting standards because of their specific characteristics, transaction complexity, and intentional errors. The study also identifies the characteristics of accounting standards that companies attribute to restatements. This classification provides important insights relative to prior research that primarily focuses on whether the restatement can be classified as intentional or unintentional.

The remainder of this study is organized as follows. First, we provide background information about restatements and their suggested causes as motivation for our study. We then detail our data collection and classification process and provide descriptive statistics regarding restatement causes and contributing factors. In the next section we report the empirical findings related to the net income effect and the reporting of material internal control weaknesses. We end with a summary of the findings and some general conclusions.

BACKGROUND AND MOTIVATION

Clearly, the number of restatements filed within the United States between 2003 and 2006 increased, but the cause of this increase continues to be debated. The U.S. Chamber of Commerce, the Securities and Exchange Commission (SEC), and the Financial Accounting Standards Board (FASB) have each identified accounting complexity as a formidable problem; some argue that this complexity is a primary driver of restatements (Ciesielski and Weirich 2006). In response to this concern, the SEC formed the Advisory Committee on Improving Financial Reporting (ACIFR) whose aim is to reduce the complexity of financial reporting. The ACIFR has outlined plans to move toward more principles-based standards. They and many others argue that rules-based standards, with their array of confusing bright lines and exceptions, lead to accounting restatements, and suggest that moving to principles-based standards will decrease complexity and help reduce the number of restatements. However, others contend that restatements are driven by auditors' and regulators' unforeseeable reinterpretations of management judgments (Pozen 2007). They argue that the move toward more principles-based standards, with an increased reliance on management judgment, has increased the number of restatements as auditors and regulators second-guess those judgments.

Another reason stated for the increased number of restatements is the sheer volume of accounting standards. Dzinkowski (2007) and others suggest that companies struggle to find the paragraphs that apply to the transaction of interest as they sift through the thousands of pages of accounting standards. In addition, they argue that the SEC and FASB periodically change the interpretation of the standards and that the changes "are episodically announced through speeches, Staff Accounting Bulletins, and other outlets without any advanced notice or public comment" (Pozen 2007, 2).1 In response to these concerns, the FASB has "codified" the accounting standards to allow for logical and expeditious access to the rules related to a specific topic and has worked to limit the guidance issued for standards.

Still others attribute the recent increase in restatements to the implementation of SOX Section 404 internal control reviews, which has uncovered past errors, as expected. In contrast, some claim that the creation of the Public Company Accounting Oversight Board (PCAOB) has caused auditors to act too conservatively, requiring restatements for technical corrections "with dubious materiality" (Pozen 2007). They argue that auditors went from an "anything goes perspective in the late 1990s" to an extremely conservative perspective after the collapse of Arthur Andersen (Taub 2005). To address this issue, the FASB, SEC, and PCAOB encourage a communicative relationship between auditors and companies' managers.

Increasingly complex business transactions without a corresponding change in financial reporting have also been mentioned as a reason for increased restatements (Dzinkowski 2007). Finally, some argue that the increase is because of the rise in earnings management and companies' focus on meeting or beating earnings benchmarks. This, in turn, causes companies to misapply generally accepted accounting principles to meet earnings targets that eventually have to be restated once the earnings management is uncovered.

Ultimately, numerous explanations are proposed for restatements and their dramatic increase in recent years; various parties (e.g., FASB, SEC, and PCAOB) have taken steps to address the concerns based on these explanations. It is important, however, to distinguish among the proposed alternative explanations for accounting restatements using empirical evidence on the drivers of restatements to properly address the problem. Such evidence assists regulators and standard setters in focusing their efforts on initiatives that are more likely to curb the incidence of restatements.

Empirical evidence on restatement causes will benefit academics as well, who have made implicit or explicit assumptions about the causes of restatements. Restatements have been used to proxy for earnings management (Efendi et al. 2007; Lee et al. 2006; Desai, Krishnamurthy, and Venkataraman 2006; Desai, Hogan, and Wilkins 2006), for accruals quality (Doyle et al. 2007), for internal company errors from inexperienced financial executives (Aier et al. 2005), and for poor corporate governance (Srinivasan 2005). However, if restatements are not the appropriate proxy for the underlying construct of interest, interpretations of the findings from these studies may be misleading. Evidence on restatement causes will therefore facilitate the use of restatements as a proxy for the appropriate construct in academic research.

Prior research highlights the importance of distinguishing between restatements that are intentional or unintentional. For example, Palmrose et al. (2004) and Hennes et al. (2008) sort restatements into intentional and unintentional misstatements and document that the market reaction to restatements is significantly greater for restatements classified as intentional relative to those classified as unintentional. Hennes et al. (2008) also find that sorting restatements into these two groups increases the power of tests that rely on restatements as an indicator of deliberate misreporting.

As the objective of our study differs from that of prior work, we contribute significantly to the literature and note many differences in the sample and the analysis. First, we examine a much larger sample of restatements and include restatements that are not announced via 8-K reports, whereas Hennes et al. (2008) and Palmrose et al. (2004) examine only publicly announced restatements. Our sample includes 3,744 restatements from 2002 through 2006; Hennes et al. (2008) examine a sample of 630 restatements from approximately the same period. Plumlee and Yohn (2009) document that only approximately 67 percent of restatements were reported via an 8-K report during 2002 through 2006, suggesting that relying on 8-K reports to identify restatements will result in a much smaller sample of restatements with potentially very different characteristics. Second, Hennes et al. (2008) exclude restatements with no income effect (almost one-fourth of our sample) and eliminate observations where the investigation (their indicator of intentional misstatement) did not lead to a restatement. Finally, Hennes et al. (2008) employ only two categories (unintentional errors and intentional irregularities), whereas we consider a broader group of categories. We consider manipulation as only one of the causes to which companies may attribute a restatement and are also interested in how the standards and the characteristics of the standards relate to the incidence of restatements.

DATA AND DESCRIPTIVE STATISTICS

Our analysis includes restatements filed in 2003, 2004, 2005, and 2006, compiled by Glass Lewis & Co. The Glass Lewis & Co. data set includes restatements filed to correct accounting errors as defined by Accounting Principles Board Opinion No. 20; therefore, restatements are not included if they are because of a change in accounting principle, a change in estimate, the adoption of a new standard, a change in the discussions of results, a minor wording change, or a typographical error.2

The original Glass Lewis & Co. data set includes 4,070 restatement filings. For each restatement, we analyzed the corporate disclosures and outside news sources surrounding the restatement to ascertain the underlying restatement "cause."3 If no information regarding a cause was available, the restatement was excluded from the sample. When necessary, we gathered additional information from other company filings and the accounting literature to classify the restatement. Each co-author independently analyzed each restatement, identified differences in classifications, and reconciled them to arrive at the final classifications.4 We consulted extensively with the staff in the Office of the Chief Accountant at the SEC and with senior audit managers at Big 4 accounting firms for assistance in identifying the restatement "cause" and "contributing factor" classifications. Our final sample includes 3,744 restatements filed from 2003 through 2006.

We also employ what Glass Lewis & Co. labels the "restatement type" (the accounting issue to which the restatement relates) in our analysis. Glass Lewis & Co. identifies a minimum of one restatement type for each restatement filing; for 70 percent of the restatement filings a single restatement type is identified. When more than one restatement type is identified, we include only the primary type in our analysis.

Restatement Causes

For each restatement filing, we classify the underlying cause to which the company attributes the restatement into one of four categories: an internal company error (INTERNAL ERROR), an intentional manipulation (MANIPULATION), transaction complexity (COMPLEXITY), or some characteristic of the accounting standards (STANDARD). A restatement is classified as being caused by INTERNAL ERROR if the disclosures suggest that the error was because of a "books or records deficiency" or a simple misapplication of an accounting standard, and the company disclosures include no discussion that suggests these errors were intentional or because of any notable characteristic of the accounting standard or the transaction. A restatement is classified as being caused by MANIPULATION if the disclosures suggest that earnings manipulation was involved, if there is an SEC enforcement action or shareholder class action lawsuit related to the restatement, or if any news articles suggest that the restatement was associated with earnings manipulation. Essentially, if at the time of the restatement some public suggestion indicates that the error was intentional, it is classified as such. A restatement is classified as being caused by COMPLEXITY if the disclosures suggest that the transaction itself created difficulties in the accounting that caused the error. Finally, a restatement is classified as being caused by a STANDARD if the disclosures suggest that the error was caused by a misapplication of an accounting standard and some factor related to the accounting standard contributed to the restatement.

Table 1 provides descriptive statistics on the cause classifications. Panel A includes the number of restatements on an annual basis along with the proportion of those restatements by each restatement cause.5 We document that, overall, 57 percent of the restatements are attributed to an INTERNAL ERROR; most of restatements are attributed to books and records deficiencies and simple misapplications of generally accepted accounting principles. In addition, more than half of the restatements in each of the four years examined are attributed to this cause. STANDARD is the second most commonly attributed cause. In each year and across the sample as a whole, disclosures suggest that restatements are attributed to MANIPULATION and COMPLEXITY less than 8 percent of the time.6

Panel B of Table 1 begins with the number of restatements by restatement cause and then details the percentage of restatements by cause and type. Restatements attributed to INTERNAL ERROR are significantly more likely to be associated with Expense, Inventory, Liability/ Contingency, Misclassification, or Tax issues, while restatements attributed to MANIPULATION are significantly more likely to be associated with Reserve/Allowance or Revenue Recognition issues. Restatements attributed to COMPLEXITY are significantly more likely to be related to Equity, OCI, or Acquisition/Investment issues; restatements attributed to STANDARD are significantly more likely to be Equity, OCI, or Capital Asset issues. Across the sample period as a whole, restatements are most frequently related to Expense (21 percent), Equity (19 percent), Misclassification (16 percent), and Revenue Recognition (10 percent). In additional untabulated results, we find an increase (decrease) across the four-year period in the percentage of restatements related to Equity, OCI, and Acquisition/Investment (Expense, Reserves/Allowances, and Revenue Recognition).

STANDARD Restatement-Contributing Factors

For each restatement classified as STANDARD, we also identify a "contributing factor" based on the restatement disclosures. Specifically, we identify one of three possible contributing factors: (1) lack of clarity in the standard and/or the proliferation of the literature because the original standard lacks clarity (CLARITY); (2) the use of judgment in applying the standard (JUDGMENT); or (3) complications in applying detailed rules (RULES).

In Table 2 we analyze the contributing factors related to the standards for the restatements attributed to STANDARD (37 percent of our sample and 1,394 restatements). Within this set of restatements, CLARITY is the most common contributing factor, identified for 58 percent of the restatements. JUDGMENT is a contributing factor for 37 percent of the restatements, and RULES is a contributing factor for only 5 percent of the restatements. For the restatement sample as a whole (based on untabulated results), CLARITY is a contributing factor for 21 percent of the restatements; JUDGMENT is a contributing factor for 14 percent of the restatements; and RULES is a contributing factor for just under 2 percent of the restatements.

We document that companies consider JUDGMENT to be a contributing factor for significantly fewer restatements and CLARITY to be a contributing factor for significantly more restatements in 2005 and 2006 relative to the earlier years.7 The increased incidence of restatements attributed to RULES during 2005 can be directly traced to lease restatements as a result of the letter sent by the SEC to the AICPA, stating the regulator's stand on the accounting for leases in 2005. The decrease in restatements attributed to JUDGMENT across the sample period may suggest that auditors and regulators have been more tolerant with respect to management judgment over time, although it may also reflect a change in managers' use of judgment. The increase in restatements attributed to CLARITY provides support for the notion that accounting complexity via a lack of clarity in the standards or the proliferation of implementation guidance may have increased over time.

We also examine the relation between the contributing factor and the restatement type. We find that restatements where judgment is considered a contributing factor are significantly more likely to be related to Inventory, Reserve/Allowance, Revenue Recognition, Tax, Acquisition/ Investment, and Capital Assets than other restatement types. CLARITY is more likely to be associated with Expense, Equity, and Liability/Contingency issues.

INCOME EFFECTS AND INTERNAL CONTROL WEAKNESSES

Net Income Effect of Restatements

To address the stated concerns that companies have become more conservative in their decisions to restate, we examine changes in a materiality threshold based on the restatement net income effect using hand-collected data from the restated filings. The net income effect (EFFECT) is the difference between the restated net income and the originally reported net income, scaled by the company's total assets prior to the restatement filing. We also calculate the absolute value of the net income effect scaled by total assets (ABEFFECT).

The results of our analysis are reported in Table 3. Panel A reports EFFECT by year for the 1,306 restatement filings for which we have available data.8 Overall, we document a significantly negative mean net income effect. However, this effect is driven by the first year of our sample period (2003). Median values are never significantly different from zero. Consistent with this result, we find that ABEFFECT is significantly greater in 2003 (0.037) than in the other three years of our sample. We also document that the proportion of restatements with a net income effect that exceeds 5 percent of total assets (FIVE) is significantly higher in 2003 than in the other sample years; 17 percent of restatements filed in 2003 resulted in an income effect that exceeded 5 percent of total assets. In contrast, only 10 percent of the restatements filed in 2006 resulted in an income effect that exceeded 5 percent of total assets. As a comparison with our finding of a mean restatement income effect of (0.013, Palmrose et al. (2004) report a mean net income effect of (0.024 using restatements reported from 1995 to 1999. This is consistent with restatements filed in the more recent periods being less consequential, as the popular press suggests. These findings provide limited support for the notion that quantitative materiality thresholds for restatements may have fallen over time. The materiality threshold may have decreased because of factors such as increased auditor conservatism or the use of the iron curtain as well as the rollover method of determining materiality as required by Staff Accounting Bulletin (SAB) No. 108 in 2006. (See Keune and Johnstone [2009] for an analysis of the impact of this SAB.) Overall, we document that 20 percent of the restatements result in an increase in net income (POS); 26 percent have no net income effect (ZERO); and 54 percent of the restatements result in a decrease in net income (NEG).9

In Panel B we report the net income effect by restatement cause. If misapplications of accounting standards and internal errors are unintentional, we expect restatements attributed to STANDARDS, INTERNAL ERROR, and COMPLEXITY to be equally likely to have positive and negative effects on income. On the other hand, if companies tend to manipulate earnings upward, then we expect restatements attributed to MANIPULATION to be more likely to have a negative income effect. Consistent with expectations, MANIPULATION restatements are significantly more likely to decrease net income and significantly less likely to have no effect on net income. INTERNAL ERROR restatements are significantly more likely to have no income effect and significantly less likely to have a negative effect on net income than restatements with other causes. In contrast, STANDARD and COMPLEXITY restatements are significantly more likely to have a negative income effect and significantly less likely to have a no-income effect. This finding may suggest that companies that attribute the restatement to STANDARDS (COMPLEXITY) exploit the ambiguity within or the judgment allowed in the accounting standards (the complexity of the transaction) to their benefit. An analysis of the specific factor related to the standard that contributed to the restatement reported in Panel B provides insight into this issue. The proportion of restatements that result in material adjustments (FIVE) suggest that INTERNAL ERROR and STANDARD restatements are significantly more likely to result in material income effects.

Panel C reports the net income effect by contributing factor for the 501 restatements attributed to STANDARD for which we have income data. We expect that if auditors and regulators allow for discretion in the use of judgment then restatements attributed to JUDGMENT will have larger income statement effects. Consistent with this, we find that restatements where JUDGMENT is a contributing factor are significantly more likely to have an income effect greater than 5 percent of total assets than are other restatements. In addition, if managers exploit the use of judgment or the lack of clarity in the standards to manage earnings upward, then we expect JUDGMENT- and CLARITY-related restatements to result in more negative restatements. We find no evidence consistent with this expectation for the use of JUDGMENT, but we do find evidence consistent for CLARITY. Restatements with JUDGMENT (CLARITY) as a contributing factor are significantly more (less) likely to have no effect on net income and significantly less (more) likely to have a negative income effect. These results are consistent with auditors and regulators being more tolerant of misstatements associated with standards requiring management judgment than for other errors. In addition, the evidence is inconsistent with managers using the judgment allowed with respect to the standards to manage earnings upward but is consistent with managers using the lack of clarity in the standard to do so.

Reporting of Material Internal Control Weaknesses

Our final analysis, reported in Table 4, explores the relation between reported material internal control weakness and restatements. Although many believe that the filing of a restatement is tantamount to a material internal control weakness within the company, a large proportion of companies that file restatements do not indicate a material weakness. In Panel A, we document that only 46 percent of companies that file restatements indicate a material weakness. This proportion increases across time-56 (47) percent of restatement filers in 2005 (2006) report material weaknesses, whereas only 20 (41) percent do so in 2003 (2004). This is consistent with more companies uncovering errors as they apply the provisions of the SOX internal controls requirements.

Panel B reports the proportion of restating companies that report WEAKNESS by restatement cause. We expect companies with restatements attributed to INTERNAL ERRORS and MANIPULATION to be more likely to report material internal control weaknesses than companies with restatements attributed to STANDARDS and COMPLEXITY. We argue that errors related to a characteristic of the related accounting standard or to the complexity of the transaction are less likely to reflect a deficiency in a company's internal controls than errors related to books and records deficiencies, simple misapplications of accounting principles, or intentional manipulation. However, we find that companies are significantly more likely to report a WEAKNESS if the restatement is because of COMPLEXITY or MANIPULATION and significantly less likely to report a WEAKNESS if the restatement is because of INTERNAL ERROR. The result for MANIPULATION is as expected, such that companies with fraudulent reporting are likely to have severe internal control deficiencies. The COMPLEXITY result could be driven by the type of companies that engage in more complex transactions. Smaller and higher growth companies that engage in more complex transactions are perhaps more likely to have internal control weaknesses. The result for INTERNAL ERROR is a bit counterintuitive. By definition, a restatement caused by a books or records deficiency or a simple misapplication of the accounting principles (INTERNAL ERROR) would be more likely to have an internal control weakness. Our results are inconsistent with intuition, although they provide support for concerns expressed in Turner and Weirich (2006) that many restatements are issued by companies that report no deficiencies in internal controls.10 The Institute of Management Accountants (2008) argues that this discrepancy might be because of a lack of standards for internal control attestation. Our results are consistent with the concerns surrounding the inadequate reporting of material internal control weaknesses; internal control attestations either fail to uncover or companies are not reporting material weaknesses related to misstatements attributed to basic internal company errors.

Panel C reports the proportion of restating companies that report WEAKNESS, sorted by the contributing factor for restatements attributed to STANDARD. Because this analysis is limited to restatements attributed to STANDARD, our sample size is reduced. In this case, we document that companies with restatements related to RULES or CLARITY (JUDGMENT) are significantly more (less) likely to report a WEAKNESS than other companies. This result is intuitive, given that the questioning of judgment is less likely to reflect an internal control weakness than the misapplication of specific rules within a standard.

CONCLUSIONS

Although various individuals and groups have posited numerous explanations for both the number and increase in restatements over the past years, empirical evidence on the underlying causes of restatements has been lacking. This study analyzes the causes to which companies attribute restatements and the characteristics of the accounting standards that relate to restatements. We analyze company disclosures related to each restatement and identify the explanation of the underlying causes, as suggested by the restating company. We also examine the relation between the restatement cause and the effect of each restatement on net income and the reporting of internal control weakness.

We document that internal company error is the primary cause to which company disclosures attribute restatements, although a significant portion of restatements are attributed to accounting standards. Restatements attributed to a characteristic of the accounting standards are most often associated with a lack of clarity in applying the standards and/or the proliferation of the literature because the original standard lacked clarity. Judgment-related restatements are less common and declined over the sample period. We find some evidence that the proportion of restatements with signed net income effects that exceed 5 percent of total assets, a commonly used materiality threshold, decreased over the four-year period. Finally, we find an increase in the reporting of material internal control weaknesses by restating companies over the sample period, although many companies with restatements attributed to basic company errors fail to report material internal control weaknesses.

Overall, the results are inconsistent with claims that the underlying complexity of the accounting is the primary driver of restatements. Rather, the majority of restatements are attributed to basic books and record deficiencies within the company and to simple misapplications of generally accepted accounting standards. When the restatements are attributed to accounting standards, the lack of clarity in applying the standard and/or the proliferation of the literature (which many consider a significant source of accounting complexity) is the primary underlying factor.

Our results provide information useful to the FASB and the SEC as they seek to implement measures to reduce the number of restatements. The results suggest that improved internal financial reporting controls from the application of SOX provisions may reduce the number of restatements. Our results also provide useful insights for the SEC as it moves forward in responding to the proposals included in the ACIFR report, providing them with empirical evidence on the sources of "avoidable complexity." It may be that the FASB's codification project and the SEC's efforts to limit detailed implementation guidance will reduce the number of restatements attributed to this issue, as long as these efforts do not also reduce the clarity of the standards. The results also suggest that a move to more principles-based standards may increase the number of judgmentrelated restatements (which have declined over time), perhaps suggesting that auditors and regulators will continue to become more tolerant of management judgments. Our results also suggest that regulators may need to address the issue of internal control weakness attestation and reporting because many restatements attributed to books and records deficiencies and simple misapplications of accounting standards are not associated with companies that report material internal control weaknesses.

Finally, our results should interest academic researchers who use restatements to proxy for earnings management and other corporate events. While using restatements as a proxy for various constructs, researchers should consider that all restatements are not alike; it is important that researchers understand the underlying causes of restatements that they include in their samples when considering whether restatements serve as an appropriate proxy.

As a final note, we emphasize that we rely on company disclosures about restatements for our classifications and that companies may have strategic incentives regarding these disclosures. We highlight that the disclosures reviewed were extremely terse, provided few details about the restatement, and may have presented inaccurate or incomplete explanations regarding the underlying cause of the restatement. Our reliance and investors' likely reliance on restatement disclosures underscores the potential need for the SEC to more closely review the disclosures about and the related explanations provided for restatements. As regulators focus attention on strategies to curtail the number of restatements, they should perhaps also pay heed to restatement disclosure requirements.

Footnote

1 Examples include the lease accounting letter the SEC sent to the American Institute of Certified Public Accountants (AICPA) stating the regulator's stand on the accounting for leases and the investigation of Fannie Mae in which the SEC's stand on hedge accounting was made clear.

2 This definition of an error includes a mathematical mistake, a mistake in applying generally accepted accounting principles, an oversight or misuse of facts that existed at the time the financial statements were prepared, or a change from a nonaccepted accounting method to a generally accepted accounting principle.

3 We use company disclosures to ascertain the cause of the restatement primarily because the footnote disclosure is often the only place the information about the restatement is available. The SEC audits and reviews footnote disclosures, and prior research suggests that companies are forthcoming about disclosing bad news (Skinner 1994), which lends credibility to these disclosures. However, we do not make predictions about the strategic disclosure choices made by a company or posit reasons for specific disclosure choices, leaving this to future research (e.g., Plumlee and Yohn 2009). We argue that classification from the restating company's perspective is interesting and informative in its own right.

4 Although the process of determining the four classification categories and the classification process itself is designed to be objective, an inherently subjective part of the process exists. To mitigate this subjectivity, we (1) consulted with professionals, including senior audit managers and the staff at the SEC, to determine the four categories and contributing factors; (2) established the categories and factors based on discussion with the professionals, analysis of over 100 random disclosures, and reading of the prior literature related to restatements; and (3) employed two coders who independently categorized each disclosure. For about 2 percent of the restatements, the classifications between the two coders differed, requiring reconciliation.

5 We use a Fisher Exact Test because it provides a nonparametric test of whether the frequencies of two independent samples differ significantly across two mutually exclusive classes. In this case, we examine whether the frequency of restatements in (not in) the specified row category differs significantly across restatements in (not in) the specified column category. Therefore, the Fisher Exact Test provides a test for each 2) 2 contingency table.

6 We document fewer "intentional" or MANIPULATION-caused restatements than prior research. For example, Palmrose et al. (2004) and Hennes et al. (2008) classify approximately 21 percent and 25 percent of their restatement sample as being intentional, respectively. We attribute these differences to the samples examined. Palmrose et al. (2004) examine restatements during a period when restatements were less prevalent and when revenue recognition misstatements represented a greater proportion of the restatements. Hennes et al. (2008) examine a much smaller sample of restatements that were announced via 8-K reports, eliminate restatements with no net income effect, and eliminate restatements in which an investigation did not lead to a restatement. Consistent with our finding of fewer restatements classified as MANIPULATION than in prior research, we document that a large proportion of the restatements in our sample have no net income effect and that the market reaction to restatements during our sample period is frequently not different from zero.

7 Given the relatively short sample period (four years) we are not able to provide tests of the statistical significance of the time trends. Thus, although we discuss the trends in general terms throughout the paper, our inferences should be interpreted with caution.

8 Data were collected for restatement filings for the subset of firms for which CRSP and Compustat data are available. Because of the high cost of hand-collecting data, we limit our analysis to this subset of firms. These firms may differ from the sample as a whole, so our results should be interpreted with some caution.

9 In untabulated analyses, we find no significant difference in the frequency of POS and NEG effects between restatements that result in a material adjustment (FIVE) and those that do not.

10 Our results suggest that the reporting of an internal control weakness is statistically more likely for restatements attributed to COMPLEXITY or MANIPULATION. However, given the small proportion of our restatements attributed to these two causes (6 percent), this result should be interpreted with some caution.

References

REFERENCES

Aier, J., J. Comprix, M. Gunlock, and D. Lee. 2005. The financial expertise of CFOs and accounting restatements. Accounting Horizons 19 (3): 123-135.

Ciesielski, J., and T. Weirich. 2006. Reading the SEC's tea leaves. Journal of Corporate Accounting & Finance 17 (6): 43-55.

Desai, H., C. Hogan, and M. Wilkins. 2006. The reputational penalty for aggressive accounting: Earnings restatements and management turnover. The Accounting Review 81 (1): 83-112.

_____, S. Krishnamurthy, and K. Venkataraman. 2006. Do short sellers target firms with poor earnings quality? Evidence from earnings restatements. Review of Accounting Studies 11 (1): 71-90.

Doyle, J., W. Ge, and S. McVay. 2007. Accruals quality and internal control over financial reporting. The Accounting Review 82 (5): 1141-1170.

Dzinkowski, R. 2007. What's keeping corporate controllers up at night? Strategic Finance (February): 34- 38.

Efendi, J., A. Srivastava, and E. Swanson. 2007. Why do corporate managers misstate financial statements? The role of option compensation and other factors. Journal of Financial Economics 85: 667-708.

Glass Lewis & Co. 2006. Getting it wrong the first time. Restatements Trend Alert (March 2). Available at: http://www.glasslewis.com/downloads/Restatements2005Summary.pdf.

_____. 2007. The error of their ways. Yellow Card Trend Alert (February 27).

Hennes, K., A. Leone, and B. Miller. 2008. The importance of distinguishing errors from irregularities in restatement research: The case of restatements and CEO/CFO turnover. The Accounting Review 83: 1487-1519.

Institute of Management Accountants. 2008. Accounting control assessment standards: The missing piece of the restatement puzzle. Discussion paper (February). Available at: http://www.imanet.org/pdf/FGRC_02.05.FINAL.pdf.

Keune, M., and K. Johnstone. 2009. Staff Accounting Bulletin No. 108 disclosures: Descriptive evidence from the revelation of accounting misstatements. Accounting Horizons 23 (1): 19-53.

Lee, J., L. Li, and H. Yu. 2006. Performance, growth and earnings management. Review of Accounting Studies 11 (2-3): 305-334.

Palmrose, Z.-V., V. Richardson, and S. Scholtz. 2004. Determinants of market reactions to restatement announcements. Journal of Accounting and Economics 37: 59-89.

Plumlee, M., and T. Yohn. 2009. The influence of materiality and voluntary incentives on companies' decisions to announce accounting restatements through 8-K and/or amended filings. Working paper, The University of Utah and Indiana University.

Pozen, R. 2007. The SEC's fuzzy math. Wall Street Journal (March 23): A11.

Securities and Exchange Commission (SEC). 2008. Advisory Committee on Improvements to Financial Reporting. Release No. 33-8896; File No. 265-24. Available at: http://www.sec.gov/about/offices/oca/acifr/acifr-finalreport.pdf.

Skinner, D. 1994. Why firms voluntarily disclose bad news. Journal of Accounting Research 32 (Spring): 1-23.

Srinivasan, S. 2005. Consequences of financial reporting failure for outside directors: Evidence from accounting restatements and audit committee members. Journal of Accounting Research 43 (2): 291-334.

Taub, S. 2005. Remarks before the 2005 AICPA National Conference on Current SEC and PCAOB Developments, Washington, D.C., December 5.

Turner, L., and T. Weirich. 2006. A closer look at financial statement restatements: Analyzing the reasons behind the trend. The CPA Journal (December): 1-12.

Appendix

APPENDIX

EXAMPLES OF RESTATEMENT DISCLOSURES AND CLASSIFICATIONS

Cause-Internal Error

Example 1: From Ableauctions.com, Inc. 8-K with a Filing Date of March 30, 2006

Item 4.02(a). Non-Reliance on Previously Issued Financial Statements. On March 25, 2006, the Registrant's Chief Executive Officer and the Audit Committee of the Board of Directors concluded that the financial statements covering the fiscal year ended December 31, 2004 should no longer be relied upon because of certain errors in the financial statements.

During the year ended December 31, 2004, the Registrant recorded marketable securities at cost. During the preparation of its financial statements for the year ended December 31, 2005, the Registrant determined that the marketable securities should have been recorded at fair value. The effect of the necessary restatement is to increase the carrying value of marketable securities by $269,474 and to increase investment income by $269,474 for the fourth quarter of 2004.

Additionally, during the preparation of its financial statements for the year ended December 31, 2005, the Registrant determined that certain accounts receivable at December 31, 2004, which had been outstanding for over one year, should have been offset by an allowance for doubtful accounts. The Registrant is restating its 2004 financial statements to reflect a provision for bad debt related to these accounts. The effect of this restatement is to decrease the carrying value of accounts receivable by $200,524, increase bad debts expense by $192,531, and decrease accumulated other comprehensive income by $7,991.

Example 2: From E-Com Ventures, Inc. 10-K with a Filing Date of April 28, 2006

Subsequent to the issuance of the Company's consolidated financial statements for fiscal year 2004, the Company determined that it had incorrectly excluded the after-tax effects of temporary differences in the amount of approximately $3.0 million related to capital lease obligations of property and equipment and $2.4 million of Puerto Rican net operating loss carryforwards in the computation of its deferred income tax accounts. The error understated the related components of deferred tax assets, with an offsetting understated valuation allowance, as of January 29, 2005, of approximately $5.4 million. Since the Company had recorded a full valuation allowance related to its deferred tax assets as of January 29, 2005 and prior to fiscal year 2003, the error had no impact on the net deferred tax assets reflected in the balance sheet as of January 29, 2005 or on the provision for income taxes in the accompanying consolidated statements of operations for the years ended January 29, 2005 and January 31, 2004. However, the below disclosures give effect to the correction of the error from disclosures previously reported.

Cause-Manipulation

From America Service Group Inc. 8-K with a Filing Date of March 16, 2006

Restatement of Financial Statements. On March 15, 2006, the Company announced that the Audit Committee had reached certain conclusions with respect to findings of the investigation that would result in a restatement of the Company's consolidated financial results for fiscal years 2001 through 2004 and the first and second quarters of fiscal year 2005. The restatement reduces previously reported net income for these periods by $2.1 million, in the aggregate, and reduces previously reported retained earnings as of January 1, 2001 by $347,000.

Consistent with the conclusions reached by the Audit Committee, the Company has included in this filing the restated audited consolidated financial statements for the years ended December 31, 2003 and 2004, and restated selected financial data for the years ended December 31, 2001 and 2002 and certain of its unaudited condensed consolidated financial information for the first and second quarters of 2005.

Refer to Item 7, "Management's Discussion and Analysis of Financial Condition and Results of Operations" and Note 3 in the consolidated financial statements for further discussion of the restatement-related adjustments and schedules reconciling the various restatement-related adjustments. Refer to Item 1A, "Risk Factors," for further discussion of potential legal and other matters that could have a material adverse effect on the Company's financial condition and results of operations.

The Company did not amend its previously filed Annual Reports on Form 10-K or Quarterly Reports on Form 10-Q for the restatement. Accordingly, the financial statements and other information contained in such reports should no longer be relied upon.

Restatement of Financial Statements. On October 24, 2005, the Company announced that the Audit Committee of its Board of Directors had initiated an internal investigation into certain matters related to its subsidiary, SPP. The Audit Committee retained outside counsel who, in turn, engaged independent accountants with significant forensic experience to assist in the investigation. The Company voluntarily reported the issues being investigated to the staff of the SEC and, since that time, the Company has cooperated with the SEC in an informal inquiry it is conducting, as well as with the office of the United States Attorney for the Middle District of Tennessee, and intends to cooperate fully with all government inquiries. Although the investigation focused on a number of items, the investigation was primarily conducted to determine whether SPP provided pricing of pharmaceuticals in accordance with applicable contract terms and whether some of the accruals and reserves maintained by SPP were established and utilized in accordance with generally accepted accounting principles.

On March 15, 2006, the Company announced that the Audit Committee had concluded its investigation and reached certain conclusions with respect to findings of the investigation that would result in a restatement of the Company's consolidated financial results for fiscal years 2001 through 2004 and the first and second quarters of fiscal year 2005. The restatement reduces previously reported net income for these periods by $2.1 million, in the aggregate, and reduces previously reported retained earnings as of January 1, 2001 by $347,000.

Pricing Adjustment. In certain instances, SPP did not charge its customers (including PHS) in accordance with applicable contracts. The Audit Committee's investigation involved testing the amounts SPP charged for pharmaceuticals against SPP's purchase invoice cost or the applicable third-party reference price. The Audit Committee's investigation also estimated rebates received by SPP from manufacturers for the purchase of certain pharmaceuticals, savings that SPP realized in purchasing certain pharmaceuticals from alternative sources ("buy-in savings"), and the dollar amounts of returns to SPP of pharmaceutical products. The Company, in consultation with special pharmaceutical counsel, determined the requirements of each of PHS's or SPP's contracts with respect to the pricing of pharmaceuticals sold as well as any contractually required sharing of rebates, savings and credits for returns described above. Applying that information to the results of the testing, the Company then concluded that, in certain instances, SPP charged customers more than SPP's purchase invoice cost or the applicable third-party reference price. The Company further concluded that SPP failed to properly credit customers with discounts, rebates or buy-in savings and, in certain instances, failed to provide customers with contractually required credit for the return of pharmaceutical products. As a result, the Company intends to provide aggregate refunds to affected customers of approximately $3.7 million through June 2005, plus $0.6 million of interest calculated using the applicable federal rate (which ranged from 4 percent to 9 percent) during the period covered by the investigation. The Company also concluded that SPP charged some customers less than should have been charged under applicable contracts relative to SPP's purchase invoice cost or the applicable third-party reference price. Such amounts total approximately $5.9 million from September 2000 to June 2005; however, because collectability of such amounts is uncertain, these amounts have not been recognized as revenue.

Accruals and Reserves. The Company determined that accruals and reserves maintained by SPP were not established and utilized in accordance with generally accepted accounting principles as key members of SPP's senior management inappropriately established and used certain reserves during various periods over the last five years to more closely match SPP's reported earnings to its budgeted results. The effect of the adjustments related to this matter necessitated by the internal investigation has been determined by the Company to be an increase in previously reported pre-tax income of approximately $355,000, in the aggregate, since January 1, 2001.

Other. The Audit Committee's investigation identified certain other issues, relating to accruals for rebates and inventory valuation, which resulted in changes to the Company's previously reported financial results. To correct the identified issues, the Company determined that adjustments representing an increase in previously reported pre-tax income of $146,000, in the aggregate, since January 1, 2001 was needed.

Income Tax Adjustments. Income tax adjustments were recorded to correct the Company's income tax expense for the impact of the restatement adjustments discussed above.

Cause-Complexity

From Paradigm Oil and Gas, Inc. 10-KSB with a Filing Date of May 24, 2006

During January 2005, the Company paid Win Energy Corporation $298,631 (less a $50,000 deposit paid in December 2004) to acquire a 10 percent interest in the Todd Creek Property (see also Note 6 (a). The $50,000 deposit was paid through the advancement of the funds by a shareholder as the Company had no other funds at that time with which to acquire the interest in the Todd Creek Property. At December 31, 2004, with no funds available to complete the acquisition and with no apparent ability to derive a benefit from the acquisition, a decision was made for the Company to write off the $50,000 advance. Although the amount should have been capitalized pursuant to SFAS No. 19 it was written off as it appeared that there were no opportunities to raise the capital required to make the acquisition. Subsequent to year-end, the Company was, in fact, able to complete a private placement of securities and was able to complete the acquisition of the Todd Creek interest.

(a) Participation Proposal Agreements. On January 25, 2005, the Company closed two participation proposal agreements with Win Energy Corporation (hereafter, Win), an unrelated Calgary, Alberta-based private corporation. The Company acquired an interest in two exploration projects in Alberta, Canada for the total payments of $506,014.

Todd Creek Property. During January 2005, the Company paid Win $298,631 (less a $50,000 deposit paid in December 2004) to acquire a 10 percent interest in the Todd Creek Property (10-34-5-29W4) located in Alberta, Canada. On June 18, 2005, the Company received a payment of $147,258 from Win for the sale of 50 percent of the Company's 10 percent interest in the Todd Creek Property. The Todd Creek Property currently has no proven reserves.

Hillsprings Property. During January 2005, the Company paid Win $207,383 to acquire a 5 percent interest in the Hillsprings Property (10-34-5-29W4) located in Alberta, Canada at a cost of $207,383. The Company held an option to acquire an additional 5 percent interest by paying an additional $207,383 to Win, but the option expired on July 1, 2005 unexercised. The Hillsprings Property currently has no proven reserves.

(b) Farm-out and Option Agreement with Related Party.

Sawn Lake Property

On February 14, 2005 the Company entered into a farmout and Option Agreement with a private Alberta corporation and a related party for consideration of $152,423. The Company will farm-in to a 5 percent interest in a test well, and a similar interest in an additional option well in the Sawn Lake area located in Alberta, Canada. The Company will earn 100 percent of the farmout's interest (an undivided 10 percent interest in the drilling spacing unit) before payout, reverting to 50 percent of the farmout's interest (an undivided 5 percent interest) after payout. In order to earn its interest in the initial test well, total costs of the test well, estimated to be $173,200, up to the point of commercial oil sales are to be borne 100 percent by the Company in order to earn its undivided interest.

Other

The Company has an option to acquire an interest in a similar well located in Alberta, Canada. In order to acquire an interest, the Company must pay the total costs of a test well up to the point of obtaining commercial oil sales. The Company has made no payments and has taken no further action on the agreement as of the date of this report.

Cause-Standard

Example 1: Contributing Factor Judgment

From Allegro Biodiesel Corp. 8-K with a Filing Date of December 27, 2006. In connection with the acquisition of Vanguard Synfuels LLC (the "Acquisition"), on September 20, 2006, the Registrant issued (i) certain stock options; (ii) certain warrants for the purchase of the Registrant's Common Stock; (iii) Series J Convertible Preferred Stock (described below); and (iv) Series K Convertible Preferred Stock, all as described in the Registrant's Form 8-K dated September 20, 2006 (hereafter, the September 2006 Convertible Securities).

For purposes of accounting for the issuance of the September 2006 Convertible Securities, the Registrant estimated the fair value of the Common Stock underlying the September 2006 Convertible Securities as $0.7587 per share, rather than the closing price of $3.10 per share on the OTC Bulletin Board on September 18, 2006, which was the last date on which the Common Stock was traded on the OTC Bulletin Board prior to the issuance of the September 2006 Convertible Securities. The Registrant's use of the $0.7587 per share valuation was previously disclosed in the footnotes to the Registrant's unaudited, pro forma combined financial statements dated June 20, 2006, which were included in the Registrant's Form 8-K filed with the Securities and Exchange Commission (SEC) on September 26, 2009 and our previously filed Form 10-QSB for the threemonth period ended September 30, 2006, filed on November 14, 2006. In reaching this conclusion, the Board of Directors of the Registrant determined that the trading price of the Common Stock did not accurately reflect the fair market value of the Common Stock. The Board relied on a number of factors in making its determination, including the fact that, as of September 18, 2006, the date of the last trade prior to the Acquisition, the Registrant's liabilities exceeded its assets; the Registrant had no ongoing business and was a "shell company," as defined under the rules and regulations of the SEC; trading in the Registrant's Common Stock was extremely limited and sporadic in nature; and the Registrant had obtained a third-party valuation in August 2006 that stated that the value of the equity of the Registrant as a "shell company" was $0.42 on a fullydiluted basis.

To finance the Acquisition, the Registrant issued $28,500,000 of shares of its Series J Convertible Preferred Stock (hereafter, the Series J Preferred) with a conversion price of $0.7587 per share of underlying Common Stock. The Series J Preferred was the senior equity security of the Registrant, with a liquidation preference over the Common Stock, as well as other preferential rights, including an 8 percent preferential dividend.

Based on the above factors, the valuation of the Series J Preferred, and other factors, and given the lack of certainty involving the valuation of the Common Stock of the Registrant, the Board determined to set the value of the Common Stock underlying the September 2006 Convertible Securities as $0.7587, the same valuation as the shares of Common Stock underlying the Series J Preferred.

Subsequent to filing its Form 10-Q for the quarter ended September 30, 2006, the Registrant discussed the valuation of the September 2006 Convertible Securities with the staff of the SEC. Based on such discussions, the Registrant determined to utilize the $3.10 valuation for financial reporting purposes, rather than the $0.7587 share price. Accordingly, the Board concluded that the financial statements of the Registrant issued for the period ended September 30, 2006 should be restated to reflect a valuation of the shares of Common Stock underlying the September 2006 Convertible Securities at $3.10 per share instead of $0.7587 and should no longer be relied upon. In reaching this conclusion, the Board and the Audit Committee of the Board discussed the matters disclosed herein with the Registrant's independent accountants, McKennon, Wilson & Morgan, LLP.

The Registrant intends to file an amended Form 10-QSB/A for the quarterly period ended September 30, 2006 to reflect the restatement discussed above as soon as practicable.

Example 2: Contributing Factor Clarity

From Grant Life Sciences, Inc. 10-QSB with a Filing Date of August 14, 2006. During the year ended December 31, 2005, it was determined the correct application of accounting principles had not been applied in the 2005 accounting for convertible debentures and detachable warrants (see Note C).

In its original accounting for the debentures and detachable warrants, the Company recognized an embedded beneficial conversion feature present in the convertible note and allocated a portion of the proceeds equal to the intrinsic value of that feature to additional paid in capital. The Company has determined that the embedded conversion feature should have been accounted for in accordance with SFAS No. 133, EITF 98-5, EITF 00-19, EITF 00-27, and APB 14. Accordingly, the proceeds attributed to the common stock, convertible debt, and warrants have been restated to reflect the relative fair value method.

In accordance with SFAS No. 154, the necessary corrections to apply the accounting principles on the aforementioned transactions are currently reflected in the reported Consolidated Statements of Losses for the three and six months ended June 30, 2005 and the Consolidated Statement of Cash Flows for the six months ended June 30, 2005.