Family ownership structure and non-GAAP disclosures : the effect of family ownership structure on the quality of non-GAAP disclosures

(1)

1

Family ownership structure and non-GAAP disclosures

The effect of family ownership structure on the quality of non-GAAP disclosures

Name: Lynn Zhong

Student number: 11417382

Thesis supervisor: David Veenman Date: 25 June 2018

Word count: 19,476

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

2

Statement of Originality

This document is written by student Lynn Zhong who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1

Abstract

This study examines the relation between family ownership structure and the quality of non-GAAP disclosures. The largest and most recent hand collected dataset of Bentley et al. (2018), key variables data collected from Compustat and the family ownership dataset from Anderson’s website were used. From the 2004 to 2009 sample, I find that the overall quality of the total exclusions of non-family firms are of relatively high quality. The total exclusions of family firms appear to be of lower quality as the total exclusions are associated with future operating cash flows. When I decompose total exclusions into special item exclusions and other item exclusions, I find that the special item exclusions of non-family firms appear to have significant predictive power for future operating earnings and cash flows. Also, the special items exclusions of family firms are significantly associated with future operating cash flows. Thus, these exclusions of both family and non-family firms are of very low quality. This implies that managers have adapted their earnings management mechanisms by shifting recurring transactions (i.e. from other item exclusions) into special item exclusions. After when the SEC updated non-GAAP reporting regulation in 2010, I find that family firms provide high quality total exclusions, special item exclusions and other item exclusions. However, now, the other item exclusions of non-family firm have significant predictive power for future operating earnings and cash flows. Furthermore, I do not find significant difference in the magnitude of other item exclusions between family firms and non-family firms. In summary, I conclude that the overall quality of non-GAAP disclosures of family firms are higher compared to those of non-family firms. In addition, I conclude that non-family firms are more likely to engage in aggressive non-GAAP reporting compared to family firms.

Keywords: Family ownership structure; non-GAAP earnings; non-GAAP exclusions; Special items; Other exclusions.

(4)

2

1 Introduction

Accounting information is important for firms in acquiring resources on both equity and debt markets. Moreover, financial statement users appreciate high quality accounting information because it provides them with insider information and consequently reduces information asymmetry, increases transparency, and provides better contracting devices (Watts and Zimmerman, 1986). Financial statement users are able to make better economic decisions because high quality financial statements contain more reliable and useful information which give a better reflection of the underlying economic performance of a firm.

The aim of my thesis is to investigate whether family ownership structure affects the quality of non-GAAP earnings measures disclosed by firms. More specifically I will investigate whether family ownership structure leads to a higher non-GAAP disclosure quality. This leads to the following research question: Does family ownership structure have a positive effect on the quality of non-GAAP disclosures in the United States?

Healy and Palepu (2001) argue that managers have superior information about their firms’ performance relative to outside investors. Information about a firm’s current and future performance is communicated through the financial statement of a firm. In addition to the required GAAP disclosures, Securities and Exchange Commission allows managers to voluntarily disclose additional useful information through, for example, non-GAAP disclosures. Non-GAAP numbers are derived from adjustments to GAAP earnings numbers by managers and are not subject to mandatory audit. The number of firms which report non-GAAP earnings in addition to the required GAAP earnings are increasing and has become a common practice (Black and Christensen, 2009; Choi et al., 2007; Hitz, 2010). In the early 2000s around 300 firms in the S&P 500 report non-GAAP earnings (Baumker et al., 2014).

Whether non-GAAP disclosures are informative depends on the motives of the manager who reports the numbers. There is an ongoing debate in the literature about the added value of non-GAAP earnings. On the one hand, managers can use non-GAAP earnings to convey relevant information to users. On the other hand, managers can misuse the discretion in calculating non-GAAP earnings by creating more positive perception of their firms’ current and future performance. This could be a way to strategically meet earnings forecasts or to mislead investors by manipulating their perception of the firms’ performance. (Bhattacharya et al., 2004; Black and Christensen, 2009; Chen et al., 2012; Lougee and Marquardt, 2004; Walker and Louvari, 2003). Also, non-GAAP reporting is considered a relatively low-cost form

(7)

5

of perception management compared to accruals or real earnings management (Black and Christensen, 2009).

According to Doyle et al. (2003), non-GAAP disclosures are of high quality when the items excluded do not have predictive power for future firm performance. Existing research suggests that some stakeholders rely on non-GAAP earnings more than GAAP earnings for economic decision making (Bhattacharya et al., 2004; Doyle et al., 2003; Lougee and Marquardt 2004; Bowen et al., 2005; Bradshaw et al., 2017). Andersson and Hellman (2007) find strong reactions from unsophisticated investors and analysts relative to sophisticated ones. So, investigating the quality of non-GAAP earnings is relevant because such disclosures have direct impact on different stakeholders and non-GAAP reporting is at an all-time high.

Furthermore, family ownership plays a big role in the financial market. Family ownership is a common ownership structure in firms all over the world. For example, companies such as Nike, Volkswagen, Samsung, Oracle, Facebook and Walmart are only a few of the many family owned firms around the world (Stern, 2015). According to Anderson and Reeb (2003), about one-third of the S&P 500 are family owned. Prior studies show that the ownership structure of a firm have an effect on corporate disclosure quality. Therefore, I expect family ownership structure to have an effect on non-GAAP disclosure quality.

However, findings about the effect of family ownership on corporate disclosure quality are controversial (Fan and Wong, 2002; Sánchez-Ballesta and García-Meca, 2007). This controversy is caused by Type I and Type II agency conflicts. Type I agency conflict refers to the problem caused by separation of ownership and management and Type II agency conflict is the conflict between controlling and non-controlling shareholders (Gilson and Gordon, 2003). Thus, empirical studies provide inconsistent results about whether family ownership has a positive effect on financial reporting quality. Therefore, I find it interesting to investigate the extent to which family ownership structure affects non-GAAP disclosure quality.

By researching the relation between family ownership structure and the quality of GAAP disclosures, I add to the literature of family ownership and the discussion about non-GAAP reporting. Investigating the quality of non-non-GAAP disclosures is relevant because current literature shows that there is a difference between the financial reporting quality of family owned firms and non-family owned firms. Several studies examine the linkage between family ownership and financial reporting quality (e.g., Ali et al., 2007; Ghosh and Tang, 2015; Wang, 2006), but the results are inconclusive and controversial. There are several studies that

(8)

6

investigate corporate disclosure quality between family and non-family owned firms (Jiraporn and Dadalt, 2009; Wang, 2006; Zhao and Millet- Reyes, 2007). However, there is no existing study which explores the relation between family ownership structure and non-GAAP disclosure quality. Thus, my thesis will be the first study to examine this relation.

Following Whippel (2015) and similar to Doyle et al. (2003) and Kolev et al. (2008), I examine the quality of non-GAAP disclosures by using a dual approach. I regress the total exclusions, special item exclusions and other item exclusions with future operating earnings and cash flows. The dataset used for the regression analyses comprises of three different datasets. The first database is the largest and most recent hand collected dataset of Bentley et al. (2018) which identifies GAAP and non-GAAP reporting with at least 95% accuracy. The second dataset is collected from Compustat. Key variables such as total assets, Common/Ordinary Equity, Earnings Per Share (Diluted) were extracted from Compustat Fundamentals Quarterly database based on the unique GVKEYS from Bentley et al. (2018) dataset. Lastly, to proxy for family ownership, the hand collected dataset from Anderson’s website was used. After merging these datasets in Stata, a final sample of 9,721 firm quarters observations was identified.

The descriptive statistics show that the non-GAAP earnings per share of firms are more positive than their GAAP earnings per share on average. This confirms the finding of Baumker et al. (2014) and implies that the exclusions made by managers commonly result in a more positive non-GAAP earning compared to GAAP earnings, which could be a sign of aggressive GAAP reporting. Next, the results from the Pearson correlation matrix show that non-GAAP earnings are more permanent and more value-relevant than non-GAAP earnings. This finding is similar to that of Kolev et al. (2008) where they find that non-GAAP earnings are more positively associated with future firm performance than GAAP earnings. In addition, I find that non-GAAP exclusions have predictive power for future firm performance. In line with Doyle et al. (2003) the Pearson matrix shows significant positive correlation between total exclusions and future operating earnings and cash flows.

I examine the relation between family ownership structure and non-GAAP disclosure quality in two separate timelines. The first period is from 2004 to 2009, which is the time after the initial implementation of Regulation G. The second period ranges from 2010 to 2011, which is the time when the SEC updated non-GAAP reporting regulation. The reason why I look at these two timeframes separately is because the extant literature find that quality of non-GAAP disclosures has improved overall after Regulation G. However, firms were allowed to exclude

(9)

7

recurring items (i.e., other item exclusions) the update in 2010, even if they do not meet the previous requirements of “non-recurring, infrequent or unusual” (Webber et al., 2013). Other item exclusions are considered low quality in the existing literature. Therefore, I want to examine whether family firms and non-family firms report more other item exclusions after the update.

First, from the regression results of the period 2004 to 2009, I do not find conclusive evidence that family firms provide higher quality GAAP disclosures compared to non-family firms. While the total exclusions of non-family firms do not appear to have significant predictive power for future operating earnings, these exclusions have significant predictive power for future operating cash flows. In addition, the special items exclusions of family firms are significantly associated with future cash flows.

Next, examining the non-GAAP reporting behavior of non-family firms from the period 2004 to 2009, I find that the total exclusions of non-family firms are in overall of relatively high quality. These exclusions do not have significant predictive power for future operating earnings and cash flows. However, while the other item exclusions are not associated with future firm performance, I consistently find that the special item exclusions of non-family firms have significant predictive power for future operating earnings and cash flows.

In the 2010 to 2011 period, it appears that none of the exclusions made by family firms have significant predictive power for future operating earnings and cash flows. Thus, it can be concluded that the overall quality of family firm’s non-GAAP disclosures is of high quality and the transactions excluded are mainly of transitory and non-cash in nature. However, in contrast to the findings of the 2004 to 2009 period, now the other item exclusions of non-family firms are significantly associated with future operating earnings and cash flows. In addition, the total exclusions of non-family firms have predictive power for future operating cash flows as well. This indicates that non-family firms are more likely to engage in aggressive non-GAAP reporting.

Because other item exclusions are considered to be of low quality (Doyle et al,. 2003), I examined whether the magnitude of these exclusions is higher for non-family firms compared to family firms. Surprisingly, I do not find significant difference in the magnitude of other item exclusions between family and non-family firms.

In summary, I find that only in 1 of the total of 4 regressions for family firms, the total and special exclusions have predictive power for future firm performance. In contrast, in all 4

(10)

8

regressions for non-family firms, either total exclusions, special item exclusions or other item exclusions have predictive power for future earnings. Thus, it can be concluded that the overall quality of non-GAAP disclosures is higher for family firms compared to non-family firms. This evidence supports the findings of Wang (2006), which imply that family firms have stronger incentive to provide higher quality corporate disclosures as an effort in preventing litigations or damage to their firm’s reputation.

The results from this thesis contribute to the extant family ownership and non-GAAP literature in several ways. First, I am the first to examine the relation between family ownership structure and non-GAAP earnings. By providing new insight, I contribute to the extant family ownership literature and non-GAAP literature. I find that family firms provide higher quality non-GAAP disclosures compared to non-family firms. This finding is in line with the finding of Gilson and Gordon (2003), that the type I agency problem (i.e., alignment effect) is more severe in non-family firms in the United States. Furthermore, unlike prior research, my findings also imply that family ownership does not lead to a more severe type II agency problem (i.e., the entrenchment effect). The evidence of family firms providing higher quality non-GAAP disclosures than non-family firms is in line with family firms being more risk-adverse, conservative and long-term oriented (Lins et al., 2013).

Furthermore, my findings have implications for regulators. Although the overall quality of non-GAAP disclosures improved after the implementation of Regulation G in 2003, I find that after the update of non-GAAP reporting regulation in 2010, non-family firms demonstrate aggressive non-GAAP behavior. Hence, the concerns of financial regulators about the possible misleading motives of non-GAAP disclosures may be justified.

(11)

9

2 Literature review

2.1 Capital market, information asymmetry and corporate disclosure

Capital market

An ideal capital market is one where all available information for economic decision making is fully reflected in prices. Capital market refers to a financial market where debt and equity are bought or sold. For example, entities that need funding can acquire funding from investors. The allocation of ownership of the economy’s capital stock is considered the main role of the capital market. Improving the efficiency of transactions and decreasing the work that an entity has to do, such as searching, analyzing, making legal contracts and completing transfers is the fundamental role of an efficient capital market (Fama, 1970).

Information asymmetry

The demand for financial reporting originates from the information asymmetry problem. Information asymmetry occurs when the amount of information is unequally distributed between two parties in a business transaction (Chang et al., 2008). This leads to one party having more or better information, which creates an imbalance of power in business transactions.

In theory there are two types of information asymmetry: (1) adverse selection and (2) moral hazard. Adverse selection occurs when one party in a business transaction has an information advantage over the other parties (Scott, 2012). In this situation, the party with superior information could exploit their advantage for their own gain at the expense of other parties. Consequently, the party with less information will realize their disadvantage and thus invest less. However, the problem of adverse selection can be reduced by financial reporting, as this is a way to transfer private information to outsiders which in turn reduces information asymmetry (Scott, 2012). The second type of information asymmetry, moral hazard, occurs for example, when there is a separation between ownership and control between management and shareholders. In this instance, management is supposed to do its best to act in the best interest of the shareholders. However, as the shareholders cannot perfectly monitor the actions of the management, the management could use this as an opportunity to demonstrate shirking behavior.

In summary, the higher the degree of information asymmetry between a firm and the market is, the more difficult it becomes for investors to evaluate and predict the performance

(12)

10

of the firm (Chang et al., 2007). Moreover, Francis et al. (2005) find that information asymmetry lowers the transparency of accounting disclosures. In accordance, when information asymmetry is reduced, it results in better market efficiency and lower cost of capital (Bleck and Liu, 2007). These findings indicate that the quality of corporate disclosures is influenced by information asymmetry.

Corporate disclosures

Financial regulators like the International Accounting Standards Board (hereafter IASB), Financial Accounting Standards Board (FASB) and the Securities and Exchange Commission (SEC) oversee capital markets to protect stakeholders and investors against non-transparent or misleading disclosures resulting from information asymmetry and agency conflicts. The FASB enforces a set of rules referred to as the Generally Accepted Accounting Principles (GAAP) to ensure consistency in accounting practices and methods. In the United States the SEC requires publicly-traded company to comply with the GAAP guidelines when disclosing financial reports. These guidelines encompass the details, complexities, and legalities of business and corporate accounting. To ensure that firms comply with the GAAP rules, their GAAP numbers are audited (O'Sullivan and Sheffrin, 2003). The main aim of GAAP is to safeguard the consistency and quality in accounting practices across firms so that the users of the financial statements can easily compare the financial performance of firms.

In addition to mandatory disclosures, firms can publish voluntary information such as non-GAAP earnings, press releases, sustainability reports and management forecasts to communicate alternative information to stakeholders. Moreover, intermediaries such as financial analysts and the financial press also provide their own adjusted measures about a firm’s financial performance. These alternative measures are derived from GAAP earnings and adjusted to better value the ‘core’ earnings of a firm (Bradshaw and Sloan, 2002). Manager adjusted GAAP earnings are referred to as non-GAAP earnings in the literature while analyst adjusted earnings (i.e., analyst earnings forecast) are called “street” earnings. The most commonly used forecast data providers (FDP) earnings metric is earnings per share which can be on GAAP basis or non-GAAP basis. These earnings metrics can be extracted in databases like I/B/E/S and Compustat and are widely used to calculate a firm’s earnings surprise. Earnings surprise occurs when the realized earnings of a firm differs from the earnings forecasts made by analysts. It essentially estimates the ‘news’ at earnings announcements.

(13)

11

In the last decades it has become commonplace for firms to disclose non-GAAP earnings as a substitute of GAAP earnings. This raised concerns for the FASB because non-GAAP disclosures are not subject to mandatory audit and therefore its information content is not validated. According to Collins et al. (1997), GAAP earnings have become noisier and less informative about a firm’s underlying economic performance as the frequency of one-time items included by firms have increased. In the extant non-GAAP theory there are two main categories of items which managers and analysts exclude from GAAP earnings to arrive at their alternative earnings. First there are special items exclusions, which are one-time items such as litigation charges. Then there are other item exclusions, which relate to recurring expenses such as amortization. Other item exclusions are the remaining exclusions after special item exclusions (Whipple, 2015).

According to managers and analysts, the aim of providing non-GAAP earnings is to emphasize cash flows and to provide numbers which will help investors better understand a firm’s true performance (Doyle et al., 2003). Stakeholders such as analysts and investors often use other information than GAAP earnings for evaluation purposes. Usually they use GAAP earnings as a basis and exclude items which they consider to be transitory or non-cash. Managers and analysts claim that those items excluded do not reflect the firm’s ‘core’ performance properly (Doyle et al., 2003). The steady growth of non-GAAP metrics is a direct reflection of the demand for metrics other than GAAP-earnings. It also reflects that GAAP earnings are becoming less value relevant as investors consider these numbers to be more noisy than non-GAAP disclosures (Ribeiro, 2016). Thus, it is of importance to gain further knowledge about non-GAAP reporting as they are valued by investors and are used in valuation and investment decisions.

2.2 Non-GAAP disclosures

Non-GAAP earnings are an alternative measure of firm performance which is often used by managers in addition to GAAP earnings. Non-GAAP earnings disclosures have been ranked by managers as the most important voluntarily disclosed financial metric for external shareholders (Graham et al., 2005). Because managers have discretion in disclosing non-GAAP earnings, these earnings have a low comparability across firms (Bhattacharya et al., 2004; Marques, 2006). Typically, non-GAAP earnings depict adjusted-GAAP measure such as earnings before metric (EB). Earnings before interest and taxes is the most common metric of EB used.

(14)

12

Alternative financial measures have become common among large companies. According to Jagannath and Koller (2013), the 25 largest non-financial companies in the United States all reported non-GAAP earnings. Similarly, in the S&P 500 almost 90% of the firms report at least one or more non-GAAP earnings measures along with GAAP numbers. Firms often present non-GAAP earnings in addition to GAAP earnings to make their firm’s financial performance better understandable for stakeholders.

An example of non-GAAP earnings is Earnings before interest, taxes, depreciation and amortization (EBITDA). The formula for EBITDA is earnings before interest and tax + depreciation + amortization. EBITDA measures a firm’s performance independently of its tax environments and financing- and accounting decisions. It is calculated by adding back the non-cash expenses of depreciation and amortization to a firm's operating income. By doing this, the impact of non-operating decisions are excluded and the outcome of operating decisions are better revealed (Bhattacharya et al., 2004; Marques, 2006).

Existing literature shows contradicting evidence of a manager’s motives to disclose non-GAAP earnings. It appears that the motive can be informative or opportunistic (Bradshaw and Soliman, 2007; Beyer et al., 2010; Young, 2014). There is a debate about whether non-GAAP earnings add value (Bradshaw and Sloan, 2002; Hirshleifer and Teoh, 2003). Managers have superior insider information about the firm’s current and future performance which outsiders do not have (Healy and Palepu, 2001). Therefore, managers can voluntarily choose to communicate this information through non-GAAP earnings disclosures. But on the other hand, managers can also misuse this discretion by communicating information which are self-serving in nature (e.g., earnings-based compensation).

Moreover, Bhattacharya et al. (2004) claim that non-GAAP earnings are valued by investors. Expenses like stock-based compensation, amortization of intangible assets are aggregated into R&D expenses or cost of goods sold instead of separate line items. This makes it hard for investors to disaggregate those items. In this case non-GAAP disclosures can be helpful as it makes these items more salient. Consequently, investors can better disaggregate earnings into components which will help them value a firm’s future performance (Leung and Veenman, 2018). Lougee and Marquardt (2004) find that non-GAAP earnings are especially informative about future firm performance when GAAP earnings are less informative.

According to Johnson and Schwarts (2005), non-GAAP disclosures are informative and lead to improvement of the non-GAAP disclosure quality. This is because items considered to

(15)

13

be transitory or non-recurring are excluded, given that they do not accurately reflect a firms’ financial health or future performance. Bradshaw and Sloan (2002) and Bhattacharya et al. (2004) find the same results. By separating expenses like stock-based compensation and the amortization of acquired intangible assets, non-GAAP disclosures can be informative and useful for investors (Hirshleifer and Teoh, 2003).

However, Bhattacharya et al. (2004) and Bowen et al. (2005) argue that non-GAAP earnings disclosures are often used to manipulate the capital market. Prior financial research finds evidence that when the GAAP earnings of firms fail to meet forecast earnings benchmarks, these firms are more motivated to use non-GAAP earnings as a substitute of earnings management to meet the forecasted earnings (Dechow et al., 2003). As a result, the quality of non-GAAP disclosures could be negatively impacted by such motives.

2.3 Types of exclusions

Whether non-GAAP earnings are useful to stakeholders is debated in the non-GAAP literature. Adjusted numbers (e.g., non-GAAP earnings and street earnings) can be divided into special item exclusions and other item exclusions. Together they represent the total exclusions by managers or analysts.

Special item exclusions are related to non-recurring items which are transitory in nature. These items are typically one-time expenses such as litigation costs. Studies find that when non-GAAP earnings are informative, it is because of transitory item exclusions which is referred to as special item exclusions in the literature (Bradshaw and Sloan, 2002; Bhattacharya et al., 2004).

Other item exclusions relate to recurring items such as stock-based compensation, depreciation and amortization of intangible assets (Whipple, 2015). However, other items have features which are hard to identify. For example, impairment of special assets and recurring items that do appear frequently. Moreover, the magnitude of other item exclusions has increased tremendously in recent years. At least 78% firms that report non-GAAP earnings exclude other items in 2012 (Whipple, 2015). Other item exclusions are viewed as low quality in the extant non-GAAP studies and are considered as an indication of aggressive non-GAAP reporting (e.g., Black and Christensen, 2009; Brown et al., 2012; Black et al., 2018). Similarly, Doyle et al. (2003) find that other item exclusions have significant predictive power for future firm performance. This means that these exclusions could be based on opportunistic motives

(16)

14

like meeting strategic benchmarks. In contrast, Whippel (2015) find that recurring items could be informative because the usefulness of these items differ per firm.

Whipple (2015) examines the motives of other item exclusions by managers and analysts in the post-Regulation G period. Even though Regulation G seem to have reduced the opportunistic motives for disclosing non-GAAP earnings, it was not clear which incentives motivate managers and analysts to exclude other item exclusions. To get a deeper understanding of other item exclusions, he distinguishes between recurring other items and transitory other items in his research. Following a broad series of prior non-GAAP studies, he used I/B/E/S/ to identify non-GAAP earnings. In order to identify other item exclusions, he compared the difference of non-GAAP earnings between I/B/E/S and Compustat. Whippel (2015) also used hand collected data which entailed specific detailed transactions of other item exclusions. In contrast to previous non-GAAP studies he finds that the incentive to exclude other items is mainly informative as those exclusions relate to non-cash items which investors heavily discount in valuing firm performance. His results show that over 70% of other item exclusions can be attributed to one-time items (29%), stock-based compensation (22%) and amortization (21%) (Whipple, 2015). Thus, other item exclusions comprise of transitory and recurring items. In line with transitory other item exclusions being non-cash items, these items are not associated with future cash flows. In contrast, less common exclusions like investment gains and losses are positively associated with future cash flows.

Moreover, he finds that an important incentive to excluding other items is to have comparable actual performance calculations to the ex-ante non-GAAP earnings forecasts of analysts (Whippel, 2015). Thus, investors can compare ex ante performance forecasts to the ex post ones. Managers and analysts claim that other item exclusions are informative to investors because they are non-cash in nature and are not so relevant in evaluating firm performance. Even though the incentive to exclude other items appears to be mainly of informative nature, Whippel (2015) also find evidence that opportunism can still occur in certain situations where a negative GAAP surprise is changed to positive by other item exclusions.

Similarly, Black et al. (2018) examining non-GAAP disclosures from the period 2009 to 2014 find that the magnitude of non-GAAP disclosures has increased. Moreover, they find that this increase is caused by non-recurring exclusions, which almost doubled in size over their sample period. According to Johnson et al. (2011), the following one-time transactions can be classified as special item exclusions: merger and acquisition expenses, goodwill

(17)

15

impairment, extinguished debt charges, asset dispositions gains or losses, restructuring charges and litigation settlements.

2.4 Non-GAAP disclosure quality

A common way of measuring the quality of GAAP disclosures is to examine how non-GAAP exclusions are associated with future firm performance. For example, Doyle et al. (2003) argue that this way of measuring non-GAAP disclosure quality is in line with the claim of managers and analysts that excluded items do not reflect the ‘core’ earnings of a firm. According to this claim, non-GAAP exclusions should have zero association with future firm performance. To test this claim Doyle et al. (2003) regressed future cash flows on total non-GAAP exclusions to examine the quality of these exclusions. In addition, they decomposed total non-GAAP exclusions into special item exclusions and other item exclusions to examine the predictive power of these different types of exclusions as well.

However, Easton (2003) questioned whether using future cash flows to assess the quality of non-GAAP exclusions was desirable because current liabilities could have implications for future cash flows. Following this claim, Kolev et al. (2008) took a different approach by regressing future operating earnings on non-GAAP exclusions.

Some researchers criticized using future earnings as the dependent variable in regressions because it could lead to a mechanical relation between current non-GAAP exclusions and future firm performance (Black et al., 2018). Consequently, it would result in no association between current transitory exclusions (i.e., special item exclusions) but there would be a mechanical association between other item exclusions and future firm earnings.

Several researchers like Whipple (2015) and Leung and Veenman (2018) follow both the approaches of Doyle et al. (2003) and Kolev et al. (2008) by applying a dual approach. They test both the association between future cash flows and future earnings on non-GAAP exclusions. This is in accordance with the claims of managers that non-GAAP exclusions are non-cash in nature and thus do not reflect the ‘core’ earnings. A dual approach gives more insight into the different types of exclusions (i.e., special item exclusions and other item exclusions) made by managers.

Similarly, non-GAAP exclusions and the type of exclusions (i.e., special item exclusions and other item exclusions) are measures of non-GAAP disclosures quality. Doyle et al. (2003) and Kolev et al. (2008) find that special item exclusions represent high quality

(18)

16

exclusions as they are mostly not predictive of future firm performance. Other item exclusions are labeled as low quality as these exclusions have significant predictive power regarding future firm performance. Therefore, examining the type of non-GAAP exclusions can also provide a better understanding of the motives behind these exclusions.

In addition, the quality of non-GAAP exclusions can also be measured by examining how investors price non-GAAP earnings and by testing whether the motive of firms to provide non-GAAP disclosures is to meet strategic benchmarks. According to Black and Christensen (2009) and Doyle et al. (2013) firms that fail to meet GAAP earning targets can use non-GAAP earnings as a strategic tool to meet benchmarks. Therefore, such practice is considered low quality as it is evidence of aggressive non-GAAP reporting.

Most recently, Bradshaw et al. (2017) used a different measure to examine non-GAAP disclosure quality. They used a ‘consensus’ metric developed by Barron et al. (1998). They examined whether non-GAAP reporting leads to more common versus idiosyncratic belief in a firm's information environment.

2.5 Motives for non-GAAP exclusions

The motives for providing GAAP disclosures have been extensively researched in the non-GAAP literature. The motives can be divided in two categories: informative or opportunistic. There are different factors which could influence the motives of managers to disclose alternative information such as non-GAAP earnings.

For example, Leung and Veenman (2018) examined whether there is a difference in the incremental information of non-GAAP disclosures and GAAP earnings in loss firms for forecasting and valuation purposes. They argue that the demand for supplemental disclosures is higher for loss firms because there is an increase of uncertainty about the firm’s future and it is more difficult for investors to comprehend the valuation implications of losses. Loss components are hard to understand because the variation of the persistence of loss is very wide. Consequently, managers have the motive to inform investors better by providing additional information which in turn reduces the uncertainty and helps investors to understand the loss numbers better.

To examine the informativeness of non-GAAP and GAAP earnings in loss firms, Leung and Veenman (2018) tested the power of these earnings in predicting future operating earnings and cash flows. The testing was done for two categories: (1) loss firms that only reported GAAP

(19)

17

numbers and (2) loss firms that report non-GAAP numbers in addition to GAAP earnings. They find that the non-GAAP earnings disclosures in loss firms are highly informative about the nature of GAAP losses compared to firms that only provide GAAP earnings. This means that those earnings disclosures are significantly predictive of the future performance of the firm. Compared to profitable firms, the non-GAAP disclosures of loss firms are less strategic in nature.

Moreover, they also find that investors perceive these non-GAAP disclosures as informative compared to GAAP only loss firms (Leung and Veenman, 2018). Loss firms that disclose non-GAAP earnings have better future performance and are not overvalued by investors compared to GAAP only loss firms. However, their findings are not fully conclusive because they also find that non-GAAP disclosures reflect both informative and opportunistic motives. Leung and Veenman (2018) conclude that managers in loss firms disclose non-GAAP earnings that are GAAP losses which are converted to a non-GAAP profit, to signal the strength of cash flows underlying earnings instead of hiding poor firm performance. Overall, they find that the motive of loss firms to disclose non-GAAP earnings are of informative nature rather than for strategic reasons.

Furthermore, Leung and Veenman (2018) make a distinction in their research between non-GAAP earnings and street earnings. Non-GAAP earnings are adjusted GAAP earnings by managers while street earnings are analyst adjusted earnings for earnings forecasts purposes. According to Bentley et al. (2018), managers and analysts do not have the same motives for providing alternative earnings measures such as non-GAAP earnings. Therefore, they could exclude different items in calculating non-GAAP earnings. A commonly excluded recurring item is for example stock-based compensation. Managers exclude stock-based compensation transactions to highlight the differential nature of this item because the predictive ability of this item varies per firm (Barth et al., 2012). Furthermore, excluding this item is supposed to aid investors in understanding the implications of such items for forecasting and valuation.

2.6 Investors reactions to non-GAAP earnings

Due to the increasing volume of non-GAAP disclosures and the potential effects it could have on investor’s reactions, numerous research investigated this potential effect. For example, Elliott (2006) conducted an experiment examining how the appearance of voluntary disclosures affects non-professional investors and analyst reliance, judgement and capital allocation decisions. She finds evidence indicating that professional investors rely more on

(20)

non-18

GAAP disclosures when these disclosures are more emphasized than GAAP earnings in press releases. However, Elliott (2006) also find that the before mentioned effect is mitigated when non-GAAP earnings are reconciled with GAAP earnings. Likewise, Allee et al. (2007) find that less-sophisticated investor’s decisions are indeed affected by how non-GAAP earnings are presented in press releases. Unlike Elliott’s finding, they do not find that reconsolidation of non-GAAP earnings with GAAP earnings mitigates this effect. Furthermore, Frederickson and Miller (2004) experimental research find that unlike sophisticated investors, less-sophisticated investors do react to and are affected by non-GAAP earnings. Although sophisticated stakeholders are less susceptible to the influence of non-GAAP numbers, Andersson and Hellman (2007) find that analysts’ forecast can be influenced by non-GAAP earnings. Also, sell-side analysts base their forecast revisions on non-GAAP numbers (Bhattacharya et al., 2003)

Whether non-GAAP disclosures are informative or misleading is thus inconclusive. On the one hand, Bhattacharya et al. (2003) claim that investors find non-GAAP earnings more informative and more useful in reflecting the core business of a firm than GAAP earnings. On the other hand, Black et al. (2017) find that firms develop non-GAAP earnings measures to increase the informativeness of the earnings numbers.

2.7 Regulation G

Non-GAAP earnings are voluntary and not formally audited. Consequently, financial statement users and regulators have expressed concerns about the possible opportunistic motives of managers in disclosing non-GAAP earnings. In response to the highly publicized alleged misuse of pro-forma disclosures such as the WorldCom and Enron scandals, the United States Congress ordered the SEC to issue new guidelines governing the presentation of non-GAAP metrics. The aim of introducing these guidelines was to improve the quality and transparency of financial accounting information. Also, an important objective was to limit the opportunistic use of non-GAAP measures (Section 401(b) of the Sarbanes-Oxley Act of 2002).

As a result, the SEC released Regulation G (SEC 2003) on January 2003. Regulation G requires firms that disclose non-GAAP earnings in preliminary earnings announcements to (1) clearly reconcile non-GAAP earnings to GAAP earnings with equal emphasis on both figures, (2) to adequately label non-GAAP measure, (3) to classify non-recurring items only when it is characterized as such and (4) to present non-GAAP earnings in a non-misleading way (Heflin and Hsu, 2008).

(21)

19

One stream of literature suggests that Regulation G has had a positive impact on the quality of non-GAAP exclusions. According to Black et al. (2012), investors show more interest in non-GAAP earnings than GAAP earnings after Regulation G went into effect. Black et al. (2012) claim that investors consider non-GAAP disclosures more informative about the performance of firms than GAAP earnings. Kolev et al. (2008) find that non-GAAP exclusions are in overall more transitory in nature and are of informative motive. However, their findings also indicate that Regulation G might have led to unintended consequences because the quality of non-recurring items (i.e., special item exclusions) appear to have decreased in the post Regulation G period. In addition, Whipple (2015) find that even though recurring items (i.e., other item exclusions) remained common after Regulation G, the quality of these items increased and were of less misleading nature. This could be a demonstration of how managers adapted to the new rule. He further concludes that the overall quality of non-GAAP earnings was higher than before Regulation G was implemented.

In addition, Heflin and Hsu (2008) find that the number of managers using non-GAAP earnings to meet earnings benchmarks decreased after Regulation G. Furthermore, some studies show evidence of a decline in non-GAAP disclosures in the post regulation G period (Heflin and Hsu, 2008; Kolev et al., 2008). Overall, these studies show evidence that opportunistic motives for non-GAAP reporting have decreased following Regulation G. In accordance, several studies find that investors’ reaction to non-GAAP disclosures relative to GAAP disclosures is higher in the post Regulation G period (Black et al., 2012; Marques, 2006). This is an indication of increased confidence in non-GAAP reporting but it also means that investors are more sensitive to misleading non-GAAP disclosures (Black et al., 2012).

Since 2003 the SEC has updated non-GAAP reporting regulations twice, once in 2010 and another time in 2016. This shows that regulators are still concerned about the strategic use of non-GAAP disclosures and consider it as a risk to investors. Other studies suggest that the decline in non-GAAP reporting after the Regulation G implementation in 2003 was only temporary and that the increase of non-GAAP disclosures have been steady since 1998 (Black et al., 2012; Bentley et al., 2018).

2.8 Family ownership structure

In theory there are two main firm ownership structures, family ownership structure and non-family ownership structure (hereafter referred to as non-family firms and non-non-family firms respectively). A majority of firms globally are family owned (Burkart et al., 2003). Family

(22)

20

firms are firms that are controlled or managed by founding families or a next of kin. Founding families can have direct control over a firm through their shares and voting rights. They could also have indirect control through use of pyramid structures (Sacristán-Navarro and Gómez-Ansón, 2007).

Founding families are characterized as long-term investors with substantial common stocks. They have very concentrated but poorly diversified portfolios. Moreover, founding families typically play an active role in the management or board of directors (Wang, 2006). Within the S&P 1500 around 62% of founding families hold CEO positions. According to Cheng (2014), around 98,4% of family firms appoint at least one of their own as a member to the board of directors. Moreover, more than half of family firms appoint at least two family members and one fifth appoint three or more family members to a management role. By doing this, the founding families can ensure that their influence is reflected within the firm.

Publicly traded firms are commonly family owned (Burkart et al., 2003). According to Anderson and Reeb (2003), about one-third of the S&P 500 firms in the United States are family owned or family controlled. These families account for approximately 11% of their firms’ cash flow rights and 18% voting rights (Ali et al., 2007). In Western Europe around 44% of large firms are family owned (Faccio and Lang, 2002). The highest concentration of family ownership is in East Asian countries with one-thirds of firms being family owned (Claessens et al., 2000). Moreover, family owned firms can be found in a broad range of industries such as high-tech industries, retail, transportation and automobile (Chen et al., 2008).

According to Lins et al. (2013), family firms behave differently compared to non-family firms. Non-family firms mainly focus on maximizing shareholder value and are more willing to take risks. In contrast, family firms are mainly focused on the continuality of their firm and as result are more risk-adverse, more conservative and long-term oriented (Lins et al., 2013). This could imply that family firms’ non-GAAP disclosures could be of higher quality.

2.9 Agency problems and family ownership

The extant finance literature suggest that firm ownership structure could affect the quality of financial reporting and voluntary disclosures of firms. According to Ho and Wong (2001), listed family firms with family members on the board of directors have less transparent disclosures. Because controlling families are both insiders and shareholders of a firm, they have unrestrained access to private information and thus have less incentive to disclosure of such

(23)

21

information. This suggests that family ownership could impede the quality of financial reporting.

Studies exploring the effects of family ownership on financial reporting quality have controversial conclusions due to Type I and Type II agency problems. Type I agency problem is referred as the alignment effect and is caused by separation of ownership and management. Type II agency problem is the conflict between controlling and non-controlling shareholders and is referred to as the entrenchment effect (Gilson and Gordon, 2003).

The alignment effect implies that family firms put a lot of effort in preventing damage to their firm’s reputation and passing on the business. Therefore, family firms are less likely to engage in opportunistic behavior in corporate reporting. Subsequently, family firms have stronger incentive to report higher quality disclosures compared to non-family firms (Wang, 2006). Family firms in which the controlling family have large ownership of the firm, the managers and management are usually family members which are chosen by the controlling family (Claessens et al., 2000). Family managed boards in family firms have a long-term investment perspective and therefore have a strong incentive to monitor managers decisions (Wang, 2006). Usually family owned firms solve the problem of separation of ownership and management by placing a family member in a manager position. By doing this they gain a better control over a manager’s opportunistic behaviors than non-family shareholders are able to (Anderson et al., 2003). This results in a more aligned interest between the controlling shareholders and managers and consequently possibly less earnings management. Thus, the alignment effect could also have implications for the quality of non-GAAP disclosures. According to Wang (2006), contracting terms for family firms are less sensitive to the quality of financial information if contracting parties believe that family ownership improves the corporate governance of the firm. Hence, these parties rely less on financial information to monitor a firm and thus, in turn the demand for high quality financial information decreases.

The entrenchment effect portrays a negative relation between earnings management and family ownership (Sánchez-Ballesta and García-Meca, 2007; Wang, 2006). Consistent with the results of Wang (2006) and Anderson and Reeb (2003), Ali et al. (2007) find that the quality of financial reports and disclosures of family firms are better than those of non-family firms. They also find that accounting numbers of family firms contain less errors due to less managerial distortions due to the higher monitoring and higher management integrity in family firms. This indicates that family ownership could have a positive effect on the quality of non-GAAP disclosures as well. Furthermore, family firms have a lower cost of debt and better firm

(24)

22

performance than non-family firms (Anderson and Reeb, 2003). Family firms in the United States have a less severe type I agency problem (i.e., alignment effect) compared to non-family owned firms (Gilson and Gordon, 2003).

On the other hand, as the level of family shareholdings increases, it could encourage management entrenchment (Type II agency problem). Type II agency problem occurs when controlling families demonstrate self-serving behavior where they manipulate earnings numbers for their own gain at the expense of outside investors (Fan and Wong, 2002). The entrenchment effect increases information asymmetry and decreases corporate disclosure quality as the earnings are less informative to outside investors. Type II problem is considered to be the most serious agency problem in the literature (Claessens et al., 2002; Fan and Wong, 2002).

Fan and Wong (2002) find that the incentive of controlling families to manage earnings for self-serving purposes increases as the insider ownership of controlling families becomes larger. This could be due to ineffective monitoring by the board members due to the amount of family members holding these positions. This could result in inferior corporate governance. Consequently, earnings informativeness decreases as family ownership increases and information asymmetry becomes greater between families and outside stakeholders. The entrenchment effect denotes that family ownership has a positive relation with earnings management, lower quality and less transparent corporate disclosures (Fan and Wong, 2002; Sánchez-Ballesta and García-Meca, 2007). Consistent with a body of literature in which earnings management is used as a measure of opportunistic behavior in financial reporting, a lower disclosure quality equals a higher degree of earnings management. Thus, there is a significant difference between the effects of family firms and non-family firms on corporate disclosure quality. In summary, the results from existing family ownership studies examining whether family firms produce higher quality financial reports compared to non-family firms is not conclusive.

(25)

23

3 Hypothesis development

3.1 Family ownership structure and non-GAAP disclosure quality

The demand and supply of non-GAAP disclosures are at a growing rate in the United States and other countries. Consistent with Healy and Palepu (2001) I follow the assumption that managers have access to superior information relative to outsiders regarding their firms’ current and future performance. Frederickson and Miller (2004), Allee et al. (2007) and Andersson and Hellman (2007) find empirical evidence that less sophisticated investors and financial analysts react to non-GAAP earnings. Similarly, Bradshaw and Sloan (2002) argue that investors prefer to use non-GAAP earnings than GAAP earnings in valuations decisions. They also find evidence suggesting that stock prices reflect this behavior.

Existing family ownership research finds that family ownership structure affects the quality of GAAP disclosures. This finding may indicate that family ownership structure could also have an effect on non-GAAP disclosures. According to Ali et al. (2007), family firms have higher financial reporting quality compared to non-family firms. However, Chen et al. (2008) find that corporate disclosures by family firms are less transparent compared to those of non-family firms. When non-family members are managers within the non-family firm, the problem of separation of ownership and management is largely mitigated (Anderson et al., 2003). Moreover, because founding families have deep knowledge of their firm and directly monitor managers, opportunistic behavior is less likely to occur in family firms. This suggests that family firms’ corporate disclosures are more likely to be of higher quality compared to those of non-family firms (Ali et al., 2007)

From existing non-GAAP literature, it is not clear what the true intentions of managers are to disclose non-GAAP earnings. Managers can use the discretion in calculating non-GAAP numbers to reduce information asymmetry between a firm and its investors by providing relevant information about the firm’s current and future earnings (Bhattacharya et al., 2003; Lougee and Marquardt, 2004). However, some studies find that managers show opportunistic behavior in reporting non-GAAP disclosures to meet earnings benchmarks or to mislead investors (Black and Christensen, 2009; Doyle et al., 2013).

As discussed before, the quality of corporate disclosures between family firms and non-family firms depend on the severity of the Type I and Type II agency problems of the firm (Ali et al., 2007). Thus, whether family firm’s non-GAAP disclosures are of better quality compared

(26)

24

to those of non-family firms is an empirical question. It is therefore important and relevant to investigate the possible effects of the characteristics of family firms on non-GAAP disclosure practices. My thesis aims to find empirical evidence of whether family ownership structure has a positive effect on the quality of non-GAAP disclosures. Specifically, based on the inconclusive and controversial results of existing research, I expect that there will be a relation between family firms and non-GAAP disclosure quality. Prior family ownership research finds that the alignment effect is more severe in non-family firms compared to family firms. This is because the entrenchment effect is largely mitigated in family firms as these firms appoint family members on the management team to resolve this problem. Therefore, I predict that family firms are more likely to disclose higher quality GAAP earnings metrics than non-family firms. This leads to the following hypothesis.

H1 Family owned firms provide higher quality GAAP earnings compared to non-family owned firms

3.2 Magnitude of non-GAAP exclusions

The difference between GAAP earnings and pro forma earnings have been steadily increasing since 1850 according to Bradshaw and Sloan (2002). The magnitude of the difference between bottom-line GAAP earnings and pro forma earnings is around 20% wherein the pro forma earnings are more positive.

The Wall Street Journal detected in the second quarter of 2001, that more than half of the S&P 500 firms reported pro forma earnings which were adjusted from their GAAP earnings. The magnitude of the exclusions to arrive at these pro forma earnings was so significant that it led to every 60 cents of each dollar of pro forma earnings being the result of the exclusions made (Doyle et al., 2003).

Managers claim that by excluding items from GAAP earnings they are able to better inform investors about their firm’s current and future performance. However, financial regulators and the financial press remain skeptical and suggests that the motives of managers to provide using non-GAAP metrics could be to mislead investors. Doyle et al. (2003) find that when a firm’s difference between GAAP earnings and non-GAAP earnings is large, these firms are likely to have lower future cash flows and stock returns compared to firms with lower magnitude exclusions. Consistent with the claim of managers that they exclude items to reflect the core earnings of their firm, these excluded items should primarily be non-recurring and

(27)

25

non-cash in nature. Consequently, the financial markets and its stakeholders do not appreciate the predictive power of the excluded expenses as they are considered low quality.

Family ownership structure could have unique implications in terms of providing voluntary disclosures such as non-GAAP earnings. First, family owners typically have a large but undiversified equity holdings, hence their wealth depends heavily on the performance of their firm. Subsequently, they are more likely to make an assessment of the benefits and costs of voluntary disclosures relative to a non-family member.

Family firms have incentive to provide less voluntary disclosures due to their longer investment horizon, better alignment between managers and family owner, better monitoring of the manager and less information asymmetry. On the other hand, providing more voluntary disclosures could lead to lower cost of capital (Bleck and Liu, 2007). The benefits from this could give family firms incentives to provide more voluntary disclosures.

It is not clear whether the magnitude of non-GAAP disclosures is bigger in family firms or in non-family firms. Ali et al. (2007) examined the voluntary disclosure practices between family firms and non-family firms of the Standard and Poor's 500 firms from 1997 to 2002. They find that family firms provide less voluntary disclosures such as earnings forecasts and conference calls.

In accordance with family firms providing less voluntary disclosures and family firms taking much more effort in preventing litigations and reputation damage, I expect that family firms are less likely to engage in aggressive non-GAAP reporting. Thus, I expect that family firms provide less other item exclusions. The second hypothesis is as follows:

H2 The magnitude of other item exclusions is smaller for family firms relative to non-family firms

(28)

26

4 Research methodology

4.1 Data and sample

The aim of this thesis is to examine whether family ownership structure has a positive relation with the quality of non-GAAP disclosures. An overview of this thesis is illustrated in figure 1. The extant non-GAAP literature often uses hand collected data to examine empirical questions regarding non-GAAP disclosures. Not only is this very time consuming and costly, it also limits the size of the datasets.

Following Bentley et al. (2018), the dataset of Bentley et al. (2018) with the proxy for manager’s non-GAAP disclosures per firm was used to examine hypothesis 1 and 2. This dataset1 is collected through textual analysis and contains 146,121 quarterly observations of 7,090 firms for fiscal years spanning January 1, 2003 to December 31, 2016. A dummy variable is equal to 1 if a firm reports non-GAAP numbers in that quarter and otherwise 0. Moreover, this is the largest and most recent hand collected dataset which identifies GAAP and non-GAAP reporting with at least 95% accuracy. Thus, this eliminated the need to proxy for manager’s non-GAAP disclosures using databases such as I/B/E/S.

Although data is available regarding the performance metrics of managers’ non-GAAP disclosures on analyst forecast data providers (e.g., I/B/E/S). Bentley et al. (2018) empirically examined the difference between non-GAAP measures defined by managers and street earnings defined by analysts on the database I/B/E/S. They find that I/B/E/S emphasizes the higher quality non-GAAP earnings while systematically failing to capture the aggressive disclosures by managers. Thus, using I/B/E/S to proxy for manager’s aggressive non-GAAP exclusions will have a biased effect. Logically, the sample from Bentley et al. (2018) is a better alternative. Moreover, they find that the data of I/B/E/S substantially overlap with their dataset and that there are systematic differences. For instance, I/B/E/S excludes managers’ lower quality non-GAAP numbers and sometimes provides higher quality non-GAAP measures that managers do not explicitly disclose. In addition, Bhattacharya et al. (2003) finds evidence showing that the expense items excluded by managers are not always excluded by analysts. There is a significant difference of 4 cents per share between non-GAAP earnings disclosed in press releases and the numbers reported in I/B/E/S reflecting actual earnings. Thus, the

1_{This link contains most recent manager non-GAAP disclosure dataset of Bentley et al. (2018), with data from} 2003 through part of 2016. This dataset was made available through https://sites.google.com/view/kurthgee/data

(29)

27

numbers in I/B/E/S are not completely reliable as a source of manager-disclosed non-GAAP earnings per share.

Since the focus of this thesis is on the relation between non-GAAP disclosure quality and family ownership structure, the dataset from Bentley et al. (2018) is more suitable than data from I/B/E/S for example. Moreover, this dataset is very recent and one of the largest hand-collected dataset of non-GAAP disclosures (Bentley et al., 2018). By using more recent data the empirical results from my research are more valuable and relevant.

Furthermore, the results from existing literature provide evidence that managers use aggressive non-GAAP reporting as a substitute of accrual and real earnings management (Black and Christensen, 2009). This could have a negative effect on the quality of non-GAAP disclosures. Therefore, it is essential that the dataset I will use for my thesis reflects the aggressiveness of managers’ reporting choices. Bentley et al. (2018) argues that their dataset captures managers’ reporting choices more accurately. Therefore, I used the dataset of managers’ non-GAAP earnings disclosures of Bentley et al. (2018).

4.2 Data collection Compustat

The second data sample originates from Compustat. Key variables such as total assets, common/ordinary equity, earnings per share (diluted) were extracted from Compustat Fundamentals Quarterly database based on the unique GVKEY’s from Bentley et al. (2018) dataset. The GVKEY is a unique six-digit number key assigned to each company. These key variables are then used to calculate the dependent and control variables. The dataset from Compustat contains 260,831 firm quarter observations of 4,776 firms. All the duplicates in this sample were dropped. In table 1 is an overview of all the retrieved variables from Compustat. After all the relevant variables were collected, the non-GAAP dataset was merged with the dataset from Compustat in Stata.

TABLE 1 Key variables from Compustat

Description Compustat data item

Total assets atq

Common shares for diluted earnings per share cshfdq

(30)

28

Standard Industry Classification Code sic

Operating Activities - Net Cash Flow oancfy

Capital Expenditures capxy

Earnings Per Share from Operations opepsq

Earnings Per Share (Diluted) Excluding Extraordinary items epsfxq

Income Before Extraordinary Items ibq

Shareholders' Equity - Total seqq

Common Shares for Diluted EPS cshfdq

Price / Close / Quarter prccq

4.3 Family ownership dataset

Lastly, to proxy for family ownership the hand collected dataset from Anderson’s website2_was used. This dataset contains 16,200 observations of the top 2,000 largest firms in the United States from the period 2001 through 2010. This dataset contains a dummy variable that equals 1 when the family owns or has voting rights of 5% or larger stake. Furthermore, there is an indicator variable that equals 1 when the firm has a dual-class share structure.

To match the datasets, the final sample is limited from the year 2004 to 2011. I choose 2004 as the start date of my data sample as it is the year after when Regulation G was implemented by the SEC. Then I merged the first merged dataset with the family ownership dataset using Stata. Since the dataset of family ownership is only of 2,000 firms and consist of fiscal year 2001 to 2011, observations that did not match with the family ownership dataset were dropped.

The aim of my thesis is to examine family and family firms that disclose non-GAAP exclusions and therefore I dropped the dummy variable for manager non-non-GAAP reporting if it was equal to 0. In other words, firm quarters without non-GAAP exclusions were dropped from the final sample. To have a more homogeneous set of firms I excluded financial firms because as prior studies (e.g., Marques, 2006; Leung and Veenman, 2018) argue that the nature of these firms’ non-GAAP disclosures differs systematically from those of nonfinancial

(31)

29

firms due to factors such as different regulations. Table 2 presents the sample selection for my thesis.

TABLE 2 Sample selection quarterly data

Description

Number of observations Bentley et al. (2018) non-GAAP firm quarters 146,121

Number of firms 7,090

Firm quarter observations of key variables from Compustat 260,831

Family ownership firm-year dataset from Anderson’s website 16,200

Merge 1 Non-GAAP and Compustat 401,573

Successful match (282,057 observations dropped) 119,516 Merge 2 Merge above dataset with family dataset 119,516 Successful match (89,449 observations dropped) 30,067 Only non-GAAP firms (18,193 observations dropped) 11,874 Dropping financial firms (96 observations dropped)

Cleaning missing values (2.057 observations dropped)

Final dataset firm quarters observations 9,721

4.4 Libby boxes

The Libby boxes in figure 1 give an overview of the first hypothesis. It illustrates the first hypothesis which examines the effect of family ownership structure on the quality of non-GAAP disclosures.

FIGURE 1 Research Libby boxes

Independent variable (X) Dependent variable (Y)

Family firm = 1 Non-family firm = 0

Family ownership Non-GAAP disclosure quality

Association between non-GAAP exclusions and future firm performance

Control variables • Sales growth • Size • Earnings volatility • Loss • Book to Market • Age of the firm

Family ownership structure and non-GAAP disclosures : the effect of family ownership structure on the quality of non-GAAP disclosures