Earnings quality at public offerings : the economic function of financial statements with a growing number of shareholders

(1)

Amsterdam Business School

Earnings quality at public offerings

The economic function of financial statements with a growing number of

shareholders.

Student: Samuel Davies Student ID: 10071946 Date: 18/08/2014

Supervisor: dr. J.J.F. Van Raak University of Amsterdam

Amsterdam Business School

Faculty of Economics and Business

(2)

Preface

This is the master’s thesis I wrote as part of the Master Accountancy at the University of Amsterdam. I was triggered to write about seasoned public offerings in context to

accounting, after following recent topics in the media concerning companies that want to collect money by seasoned public offerings. The purpose of this thesis is to provide new insights and add to discussions in this field.

My thanks go firstly to my supervisor dr. J.J.F. Van Raak for his guidance during the graduation program. I would also like to thank my girlfriend Berber for her unconditional support. Of course I also want to thank my family and friends for their interest.

(3)

Abstract

More and more companies want to collect money by means of a seasoned public offering. This creates a growing difference between ownership and management, bringing about agency problems and information asymmetry.

Similarly a conflict of interest arises between managers and shareholders. To prevent this, shareholders hire an independent party (i.e. auditors) to monitor the activities of

managers (Anderson and Reeb, 2003). As the group of shareholders increases, the economic function of the financial statements also increases. Therefore, the growing number of

shareholders will insist on stricter control of the financial statements. Based on this theory the thesis proposes that the earnings quality will rise when there is an increase in the number of shareholders. Because the economic function of financial statements has grown, a stricter control is expected, and this should lead to higher earnings quality.

The results of the empirical study support the hypothesis. This implies that the quality of the earnings rises in proportion to the size of the public offering. A possible research question for further exploration is: What external factors play an increasing role as the group of shareholders increases?

(4)

1. Introduction

According to Kakebeeke, Dutch listed companies were financially better off in 2013 than in the previous 5 years (Kakebeeke, 2014). Due to higher profits and prudent investments, companies managed to reduce their debt for the first time in that period. Especially bank debts were repaid. Companies have repaid debts with the issuance of new shares (a seasoned equity offering) out of necessity, according to research by the Dutch Financial Times (Kakebeeke, 2014). A seasoned equity offering is defined as the issuance of shares by a company that is already publicly listed. Loughran and Ritter (1997) argue that most issuing companies profit from asymmetrical information when issuing shares. In response to this conclusion, the present research focuses on the question whether information asymmetry also impacts the earnings quality at the time of a seasoned equity offering.

In previous studies, the differences in earnings quality between public (listed) and private (non-listed) firms has been examined (including Ball and Shivakumar, 2005; Hope et al., 2013). These studies conclude that earnings quality is higher for public firms. A possible explanation is the much larger circle of stakeholders in public firms than in private ones; therefore the annual reports of public firms have a greater economic function.

Ball and Shivakumar (2008) followed up on their 2005 analysis, seeking to determine the effect of market and regulatory changes on the quality of financial statements. They accomplished this by focusing on IPOs. This motif is based on the hypothesis of Ball and Shivakumar (2005) that public firms have a higher reporting standard due to an increased demand and stricter regulation. Investors, lenders and other users of annual reports of public firms are more distanced than at a private firm, and therefore demand higher quality of financial statements to solve the information asymmetry.

In many studies a distinction is made between private and public firms. The studies by Ball and Shivakumar (2005) and Burgstahler et al. (2006) are considered pivotal in this research area because they have made significant contributions with new research models. The economic function of the annual report is considered important in both studies. According to the researchers, the economic function justifies the lower earnings quality in private firms compared to public firms. An explanation for these different economic functions is the degree of information asymmetry.

(6)

Within the framework of this research, information asymmetry is described as a situation in which there is a separation between management and ownership; the management has more information than the owners. If the number of stakeholders increases, for example with seasoned equity offerings, the information asymmetry will also increase.

Several studies (Ang et al., 2000 and Jensen and Meckling, 1976) indicate that the ownership structure has an effect on agency costs. Agency costs arise in a situation where there is information asymmetry (for example the separation between management and ownership) and consist of the sum of monitoring costs, bonding costs and residual loss (Jensen and Meckling, 1976). Monitoring costs are the costs that the principal (owner) makes to check the agent (manager), for example audit fees. Bonding costs are those arising from contracts between the principal and the agent to reduce information asymmetry, as is the case of a bonus contract. Residual losses are the costs incurred because the measures for monitoring and bonding do not eliminate all information asymmetry, therefore there are always cases where the agent makes decisions because they are beneficial to him or herself.

In the study of Ang et al. (2000) it is also acknowledged that agency costs increase when the capital interest of external shareholders (non-executive board members) increases. Ang et al. argued that, in a situation when one entity owns all the shares but appoints a director for everyday business, there are always agency costs. These costs will be low in the situation described above, because the owner has all the possibilities to check, and eventually fire, the director, but it is impossible for the owner to monitor everything the director does.

The objective of this research is to gain knowledge and understanding of the change of earnings quality at the time of a seasoned equity offering. The research question of this study is:

To what extent does earnings quality change when the group of stakeholders increases at public firms?

The underlying idea is that the more stakeholders there are, the more important it is that the annual report, including the profit figures, provides reliable information to the stakeholders.

This study contributes to the discussion on the applicability of the agency theory in relation to earnings quality. Previous research (Ball and Shivakumar, 2005) already showed that the earnings quality increases with an Initial Public Offering. In this study, evidence is

(7)

sought to the extent which earnings quality change when the growing number of stakeholders increases. And therefore, the applicability of the agency theory is tested.

There are several ways to measure earnings quality. Many studies about the topic have been published, and various measuring instruments are used. Dechow et al. (2010) combined these studies and provided an overview of many common instruments.

An important feature of earnings quality is timely loss recognition (TLR); this feature makes the annual report more useful for different purposes (Ball and Shivakumar, 2005). Ball and Shivakumar (2005) made a link between conservatism and TLR in their research. By using TLR as a measuring instrument, the degree of conservatism can be measured and the link to earnings quality can be made. For this, two models in the research of Ball and Shivakumar (2005) are used, namely a ‘time-series test in TLR’ and an ‘accruals-based test in TLR’. The Time-series model measures TLR by analysing result mutations. The accruals-based model reflects the extent to which TLR is visible in accruals. This model uses the relation between cash flow and accruals.

The empirical results of the time-series test indicate that in accordance with the theory of timely loss recognition, persistent gains are accounted persistent, when losses occur they are estimated to be higher, and are one-off. This asymmetric accountability between gains and losses are confirmed by the empirical results. The results of the

regression analyzes based on the accruals-based model found a highly significant outcome which implies that there is a positive correlation between the use of the recognition time when the issue of shares is growing stronger.

The effects found implies that the earnings quality improves and the economic function of the financial statements rises when the number of stakeholders increases.

2. Literature and hypothesis development

This chapter provides an overview of relevant economic literature in the field of information provision based on the agency theory. In section 2.1, agency theory is discussed in relation to the ownership structure. In section 2.2, earnings quality and accounting conservatism, as accounting subjects are discussed. Finally, section 2.3 summarizes the chapter.

(8)

2.1 Agency theory

2.1.1 Information asymmetry

Agency theory is based on the separation between management and ownership of a firm (Jensen and Meckling, 1976). In this theory, asymmetrical information is key. There is asymmetric information when different parties are unable to (fully) access the same information. The theory seeks to explain conflicts of interest between two main parties: the owner (principal) and the management (agent), based on an interpersonal relation. This means that the principal and the agent are seen as individuals, who each will try to maximize their own utility (Jensen and Meckling, 1976). Jensen and Meckling (1976) define the relation between agent and principal as follows: “a contract under which one or more persons (the

principal) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent” (p. 308)

Jensen and Meckling (1976) distinguish between two types of agency problems: ‘adverse selection’ and ‘moral hazard’. With ‘adverse selection’ (ex-ante) the agent has more information available than the principal when a contract is signed. With ‘moral hazard’

(ex-post) the agent is able to affect the outcome after signing a contract, unbeknownst to the

principal.

These agency problems generate costs, which arise from the possible solutions to solve information asymmetry (Chow, 1982). These costs are considered residual loss. The most important solution is to draw up a contract (Jensen and Meckling, 1976). Contracts ensure that the interests of the agent and the principal are aligned (Healy and Palepu, 2001).

However, agency theory does not only refer to a relation with one agent, but it can also be extended to an n number of agents. By extension, organisations can be seen as a nexus of contracts (Watts, 1992). Contractual relations are important for the existence of a firm (Jensen and Meckling, 1976; Watts and Zimmerman, 1986).

2.1.2 Agency theory in relation to ownership structure

Ali et al. (2007) mentioned two potential effects that explain agency theory in relation to the ownership structure: (a) Agent-principal problem of separation of management and ownership (Type I). This separation creates a conflict of interests between the management and owners. Also, (b) an agent-principal problem between majority and minority shareholders (Type II) in

(9)

which conflicts about the degree of control over activities arises. The following paragraphs discuss these two types of agency problems.

2.1.2.1 Agent-principal problem: separation of management and ownership (Type I)

As a result of the separation between management and ownership, there are fewer conflicts of interest between the agent and the principal at private firms than at public ones, (Type I agent-principal problem). The lower degree of separation between management and ownership thus reduce conflicts with regard to ‘moral hazard’ (Ali et al., 2007). Several characteristics reduce conflicts of interest between the agent and the principal. First, majority shareholders are inclined to maintain an undivided and concentrated share in the company; this means they have a stronger incentive to check managers. Majority shareholders often have better knowledge of the company’s activities in the industry in which it operates, so they are better able to directly monitor managers (Anderson and Reeb, 2003). Furthermore, with fewer shareholders there are less conflicts of interest because of the possibility to monitor the management more directly. This enables shareholders to assess managers more objectively (Anderson and Reeb, 2003). Type I agent-principal problems could lead to less manipulation of financial figures. However, certain factors may contribute to moderate the differences in type I agent-principal problems. Rewarding managers on the basis of objective performance measures aligns the interests of managers and shareholders.

2.1.2.2 Conflict between majority- and minority shareholders (Type II)

Other conflicts arise between the agent and principal as a result of the ratio between majority and minority shareholders (Type II agent-principal problems). Members of the Board of Directors or the Supervisory Board are less likely to be independent when there are majority shareholders (Anderson and Reeb, 2003; Anderson and Reeb, 2004). Type II agent-principal problems could lead to manipulation of financial figures.

Majority shareholders control the activities of a firm for a substantial part due to the concentrated shares they hold, the voting right and the position in the Board of Directors/Supervisory Board. This control allows majority shareholders the possibility to gain advantages at the expense of other shareholders (Anderson and Reeb, 2003). As the number of shareholders increases and no majority shareholders are present, the degree of control over activities decreases.

(10)

2.2 Ownership structure and earnings quality

In the preceding paragraphs, agency theory in relation to the ownership structure has been described. In this section the concept of earnings quality is looked at, and the influence of ownership structure on the earnings quality is examined.

2.2.1 Earnings quality

Earnings quality is a widely used concept in scientific studies. However, there is no single definition of the term (Schipper and Vincent, 2003). Because this study focuses on the relation between the economic role of the annual report and earning quality, the following definition, by Dechow and Schrand (2004), has been selected:

“We define high-quality earnings to be those that accurately reflect companies’ current operating performance” (p.5)

Other researchers often include a notion of measuring earnings quality in their definitions. They can therefore not be seen as pure ‘definitions’. When earnings quality is discussed in this research, they refer to the degree to which the reported results correspond with the actual performance of a firm. The definition of Dechow and Schrand (2004) includes an explanation that earnings quality is a measure of information asymmetry. The negative correlation between earnings quality and information asymmetry is thus described. This is consistent with the research question of this thesis.

2.2.2 Factors affecting earnings quality

In this section the factors that determine the earnings quality are introduced. Several studies have looked at the factors that affect the earnings quality. Ball and Shivakumar (2005), Burgstahler et al. (2006) and Hope et al. (2013), among others, have looked at this topic. However, their studies do not provide a comprehensive overview of these factors. Dechow et al. (2010) devoted their research to the instruments to measure earnigs quality, the factors that determine it, and its consequences. Previously, the definition of earnings quality by Dechow and Schrand (2004) was chosen; one of the six categories that Dechow et al. distinguish is relevant for this study:

Financial reporting practices: this category of factors mentioned by Dechow et al. (2010) is related to differences in methods of financial reporting. The researchers indicate that, in

(11)

general, the possibility of choices in reporting will lead to lower earnings quality. It is expected that managers will be more opportunistic in their choices than motivated to increase the information value of the annual report.

2.3 Accounting conservatism

Companies can present current results and expectations in future cash flows (Lafond en Roychowdhury, 2008). According to Watts (2003), conservatism is a means to facilitate this. Two definitions are included to explain this concept: Firstly, According to Basu (1997) conservatism means that bad news is incorporated in the results before good news. The value in the annual report is therefore always lower than the fair value. Then, Watts (2003) explains conservatism as profits being verified better than losses before they are both recognized.

Ball (2001) concludes that there is a relation between the ownership structure of a firm and conservatism. His argument is consistent with the research of Lafond and Roychowdhury (2008), that a manager without shares tends to postpone losses as much as possible. This creates a demand for conservatism. The conclusion of Wang (2006) is consistent as well. The more separation between managers and shareholders, the more conservatism is observed. The abovementioned arguments suggest that, as the number of shareholders increases, the tendency to conservatism increases as well.

2.3.1 Accounting conservatism and earnings quality

With conservatism the results are, in many cases, much higher than reported results. Earnings quality is optimal if the reported results are in accordance with the actual operational performance of the firm. This indicates a negative relation between conservatism and earnings quality, and is consistent with agency theory and the findings of Ball (2001) that managers without shares report to their own interests.

However, conservatism arises as a result of the behaviour of managers and is intended, among other things, to prevent that managers inflate results. This indicates a positive relation between conservatism and earnings quality. Scientific literature also points to a positive relation. Ball and Shivakumar (2005), for example, use the timeliness of taking losses as a measuring instrument for earnings quality. Also, the model used in the research of Basu (1997) sees conservatism as indicator of earnings quality.

(12)

This research assumes a positive relation between conservatism and earnings quality. When the number of shareholders increases, the demand for stricter controls and audits increases too, leading to improved use of conservatism and increased quality of the earnings.

2.4 Hypothesis development

On the basis of the literature review, it is expected that the earnings quality increases when there is more separation between ownership and management. Agency theory plays an important role in this process. The difference in earnings quality between private firms and public firms has already been studied (with an IPO as a trigger) by Ball and Shivakumar (2008). An important motivation for their research was to examine the differences between the market environment and regulations between private and public firms. Public firm investors, lenders and other users of annual reports are at a greater distance from the company than they would be in private firms. They are being confronted with higher information asymmetry, and therefore demand a higher quality of financial reporting to solve this problem. Therefore a seasoned public offering is used as a trigger event in this research, where the separation between ownership and management increases, which results in higher information asymmetry and a higher economic demand for solving the problem. Owners will focus more on controls to increase the quality of the financial statements. Changes in the market environment and regulations are not applicable here. Based on these premises, the following hypothesis is formulated, and tested in this study:

H1: Earnings quality is positively associated with the increase of outstanding shares.

3. Research method

In this chapter the research method is explained. Section 3.1 provides an overview of the data and sample selection. In section 3.2 the models used to test the hypothesis are described.

3.1 Data collection and sample selection

The data needed to perform the regression analysis requires annual report data of listed companies. The data from the annual reports are derived from the database COMPUSTAT North America. For the sake of relevance a selection was made of companies listed at the

(13)

the use of conservatism during these years, a selection was made for the years where t is a public offering, and IPOs are excluded. The 1% extreme outcomes per variable are excluded to distinguish any incorrect entries and to exclude them from the test results.

Companies where the balance total in one of the years changed by more than 30% compared to the previous year were excluded from the database, in accordance with Ball and Shivakumar (2005). It is likely that these companies were involved in a major acquisition, restructuring or substantial divestment. The final data is described in section 4.1.

3.2 Measuring earning quality

This section examines which instruments for earnings quality exist and which instruments are used for this study. This is done based on previous studies about earnings quality. One of these measures is also used in this research to measure earnings quality at public offerings.

3.2.1 Different measuring instruments of earnings quality

There are several ways to measure earnings quality. Many studies about the topic have been published, and various measuring instruments are used. Dechow et al. (2010) combined these studies and provided an overview of many common instruments. Their research will therefore be central to this section. Dechow et al. (2010) divide the different measuring instruments into three categories. Because conservatism concerns the timely recognition of revenues and losses—which together form a part of earnings—the category ‘properties of earnings’ is discussed next.

3.2.2 Timely loss recognition

An important feature of earnings quality is timely loss recognition (TLR); this feature makes the annual report more useful for different purposes (Ball and Shivakumar, 2005). The definition of conservatism by Basu (1997) more or less includes the concept of TLR:

“earnings reflects bad news more quickly than good news”. As an example of conservatism

Basu (1997) recalls the principle of cost-price or a lower market value of stocks. Also taking losses as soon as they are known and the recognition of an impairment loss on an asset, while at a higher market value a profit is not accounted for, are examples of TLR. Ball and Shivakumar (2005) made a link between conservatism and TLR in their research. Earlier in

(14)

the present paper a link was made between conservatism and earnings quality. By using TLR as a measuring instrument, the degree of conservatism can be measured and the link to earnings quality can be made.

The literature thus suggests that TLR is a good measuring instrument for conservatism.

3.3 Method

The previous section showed that earnings quality can be measured with TLR. For this, two models in the research of Ball and Shivakumar (2005) are used, namely a ‘time-series test in TLR’ and an ‘accruals-based test in TLR’. The time-series test by Ball and Shivakumar (2005) is based on the regression model of Basu (1997). The accruals-based test is a regression model developed by Ball and Shivakumar (2005). In other scientific articles these models are mentioned in relation to TLR (for instance Dechow et al., 2010 and Hope et al., 2013). This seems to indicate that these regression models are conventional and reliable, and can thus be used in this study as well.

3.3.1 Time-series test

This model measures TLR by analysing result mutations. Basu (1997) concludes from the conservatism principle that losses are taken in a timelier manner than profits. In addition, losses are often a one-time event, while profits have a more continuous character. If a profit has been taken at the appropriate time the possibility exists that a part of this has to be corrected in the subsequent year. In this case there is a negative relation between the result mutation in year t-1 and the result mutation in year t. Because losses are taken on time, it might be the case that in year t the losses were overestimated, causing a part to be corrected. In this case there is a negative relation between the result mutation in year t-1 and the result mutation in year t. The assumption of Basu (1997) is that this negative relationship in year t is more powerful if the result mutation in year t-1 is negative.

The regression model included in the research of Ball and Shivakumar (2005) is shown in equation 1, where the dummy used by them for public/private firms (DPR) was replaced by the relevant variable

∆OS

t.

(15)

(1) Time-series model

∆NI_t= α_{0 + α1}D∆_{NIt-1 + α2}∆NI_{t-1 + α3}D∆NI_t-1*∆NI_t-1+ α₄∆OSt + α₅∆OSt

*D∆NI_t-1+ α₆*∆OSt *∆NI_t-1+ α₇∆OSt *D∆NI_t-1*∆NI_t-1+ ε_t

Explanation model:

∆NIt = mutation of the result of fiscal year t-1 to t, scaled by dividing the mutation

by the balance total at the start of fiscal year t. D∆NIt - 1= dummy variable, = 1 if ∆NI t – 1 < 0; = 0 otherwise.

∆NIt – 1 = mutation of the result of fiscal year t-2 to t-1, scaled by dividing the mutation

by the balance total at the start of fiscal year t-1.

∆OS

t = continuous variable, the mutation % of the outstanding shares of fiscal year t-1 to t.

In the event that profits are taken at the moment they are realized, the expectation is

α₂= 0. If the profits are taken on time then the expectation is α₂< 0. From the expectation it is more likely that a negative result mutation in year t-1 will lead to a positive result mutation in year t than the opposite, therefore, the expectation is that α₃< 0. Consequently α₂ + α₃< 0.

Because it is expected that the degree of TLR varies according to height of the mutation of outstanding shares, the variable

∆OS

t was added. In this research α₇ is most important.

Because of the expectation that TLR is less applied in proportion to the magnitude of a public offering, it is assumed that α₇ < 0. In accordance with the study of Ball and Shivakumar (2005) there are no expectations for the coefficients α₀, α₁, α₄, α₅ and α₆.

3.3.2 Accruals-based test

Ball and Shivakumar (2005) indicate that the time-series model has two limitations. Firstly, the model is unable to distinguish temporary profit and loss components from ‘accidental’ errors in accruals (for example a counting error in the stocks) and certain types of earnings management (for example excess provisions that are repaired over time). These are in all cases of temporary negative accruals. In addition, the model can only identify temporary components, but not to what extent the accountability is timely or not. Therefore, Ball and Shivakumar (2005) developed an alternative model that reflects the extent to which TLR is

(16)

(2) visible in accruals. This model uses the relation between cash flow and accruals. According to Dechow et al. (1998) the relation between accruals and cash flow is negative.

Ball and Shivakumar (2005) argue, however, that there can be a positive relation between cash flow and accruals. The researchers expect a positive correlation between cash flow and accruals at the moment of a negative cash flow. The regression model included in the research of Ball and Shivakumar (2005) is presented in equation 2, with the relevant variable

∆OS

t.

Accruals-based model

ACCt = β0 + β1DCFO_{t + β2}CFO_{t + β3}DCFO_t*CFO_{t +β4}∆OSt + β₅∆OSt * DCFO_{t + β6}∆OSt *CFO_{t + β7 ∆OS}t * DCFO_{t *CFOt + νt} Explanation model:

ACCt = ∆Stocks +/+ ∆Accounts receivable +/+ (∆Other current assets -/- ∆Cash) -/-

∆ Accounts payable -/- (∆Other current liabilities -/- ∆Current liabilities to credit institutions) -/- Depreciation costs

DCFOt = dummy-variable,= 1 if CFOt < 0; = 0 otherwise.

CFOt= Operational cash flow, calculated as follows: result -/- ACCt.

∆OS

t = continuous variable, the mutation % of the outstanding shares of fiscal year t-1 to t.

For the calculation of the accruals (ACCt)an addition was made to the research of Ball

and Shivakumar (2005). The mutation in the cash and cash equivalents and the current liabilities to credit institutions is not included in the calculation, because this concerns the mutation of cash and thus no accruals (Burgstahler et al., 2006). In a study by Burgstahler et al. (2006) the mutation of cash and cash equivalents is also not included in the calculation of accruals. The study by Ball and Shivakumar (2005) does not indicate whether this was taken into account. Therefore, this addition was made on the basis of the research by Burgstahler et al. (2006).

Based on the view that there is a negative relation between accruals and cash flow (Dechow et al., 1998), it is expected that β₂< 0. Ball and Shivakumar (2005) expect a positive relation when there is a negative cash flow, so β₃> 0. In accordance with the hypothesis the

(17)

offering increases, thus more likely to recognize losses as transitory items; β₇ > 0. In accordance with the study of Ball and Shivakumar (2005), no predictions are made for the other coefficients.

3.3.3 Additional dummy analyses

In addition to the above mentioned models, an extra regression will be made with the dummy-variable PO instead of dummy-variable ∆OSt.This addition was made to test the effect of the earnings

quality in the year of a seasoned public offering compared to a year without a one. The dummy equals 1 when the outstanding shares mutation > 10%. For lower mutations (< 10%) there is no effect suspected on the information asymmetry between managers and shareholders. A similar outcome of the regression is suspected compared to the outcomes with the variable

∆OSt.

4. Empirical results

This chapter describes the statistical analyses conducted to answer the hypothesis. Section 4.1 provides a general overview of the data collected before the regression analyses. The results of the Time-test series are then presented in section 4.2 with the results of the accruals-based test.

4.1 Descriptive data

To provide an image of the data used and to assess the representativeness of the used data, on the next page is a table (Table 1) with the descriptive statistics presented.

(18)

Table 1 Summary statistics

N Minimum Maximum Mean Std. Deviation

Time-series model ∆NIt 79,618 -.1772 .2393 .0067 .0406 ∆NIt-1 79,618 -.2061 .2650 .0078 .0442 D∆NIt-1 79,618 0 1 .3309 .4705 ∆OSt 79,618 -.1778 .8057 .0561 .1405 PO 79,618 0 1 .1560 .3628 Accruals-based model ACCt 92,225 -.3930 .9187 .0354 .1460 CFOt 92,225 -2,705 5,449 118.5463 563.5684 DCFOt 92,225 0 1 .3294 .47 PO 92,225 0 1 .1977 .3983

The complete database as it was generated from COMPUSTAT contained 117,448 records of 5,738 companies who were listed at the New York Stock Exchange (NYSE) between January 1950 and May 2014.

For the generation of the variables, like they are mentioned in the paragraph described earlier, all the 117,448 records are used. Afterwards the 1% cut-off results for each variable are removed. The outcome of ‘Not Applicable' is deleted as well. Because the time-series model uses variables in which data from t-2 is necessary, it provided more variables with the outcome: 'Not applicable'. This explains in large part the difference in used records between Time-series model and Accruals-based model (79,618 records, respectively, and 92,225 records). The final data was for the Time-series model over 4,397 companies and for the Accruals-based model over 4,731 companies.

4.2 Results regression analyses

This section investigates whether public firms after a seasoned public offering are more likely to report loss components of income that are transitory, and whether their accrual behavior is consistent with timely loss recognition.

(19)

4.2.1 Results Time-series model

The time-series model is used to display the effect of the magnitude of a public offering on earnings quality. The results of the regression analysis are presented in the Table 2.

The results of model I that are presented in Table 2 relate to the time-series model that considers the continuous variable ∆OSt. With this variable, the effect of the magnitude of the

public offering can be measured. In accordance with the hypothesis, the positive correlation between earnings quality and the issue of shares is examined. When more shares are issued there is a larger group of shareholders, which implies a greater economic function and higher information asymmetry while the demands for stricter controls on the financial statements increases. This should result in higher earnings quality.

The predictions are based on agency theory and the reasoning and evidence from the study of Ball and Shivakumar (2005). This study predicts deferred recognition of economic gains, as "persistent" positive components of accounting income, the implication being α₂ = 0. Additionally, this study predicts that economic losses are recognized earlier than profits, as transitory income decreases, the implications being α₃ < 0 and α₂ + α₃ < 0.

No hypothesis is formed on the differences in recognition of income, consequently, there is no prediction is made for α₆. When the group of stakeholders increases it is more likely that economic losses are recognized less timely, prediction is therefore α₇> 0.

Table 2 reports the output results and can be read for the two time-series models, which were analyzed by means of a regression test.

(20)

Table 2 Outcomes Time-series model

Regression of change in earnings on lagged change in earnings for all firm-years.

Time-series Model I with the change in Outstanding Shares (∆OSt) as continuous variable.

∆NIt = α0 + α1D∆NIt-1 + α2∆NIt-1 + α3D∆NIt-1*∆NIt-1 + α4∆OSt+ α5∆OSt*D∆NIt-1 + α6∆OSt*∆NIt-1

+ α7∆OSt*D∆NIt-1*∆NIt-1 + εt

Time-series Model II with a dummy variable for the firm-years with a public offering (PO).

∆NIt = α0 + α1D∆NIt-1 + α2∆NIt-1 + α3D∆NIt-1*∆NIt-1 + α4PO+ α5PO*D∆NIt-1 + α6PO*∆NIt-1

+ α7PO*D∆NIt-1*∆NIt-1 + εt

Dependent variable is ∆NIt _{Model I} _{Model II}

Predicted Coefficient P Value Coefficient P Value

Intercept (α0) ? .0028 * .000 .0030 * .000 D∆NIt-1 (α1) ? -.0078 * .000 -.0078 * .000 ∆NIt-1(α2) 0 -.0184 * .000 -.0159 * .002 D∆NIt-1*∆NIt-1 (α3) - -.5032 * .000 -.5036 * .000 ∆OSt (α4) ? .0218 * .000 - - ∆OSt*D∆NIt-1 (α5) ? .0106 * .001 - - ∆OSt*∆NIt-1 (α6) ? .2328 * .000 - - ∆OSt*D∆NIt-1*∆NIt-1 (α7) - -.0868 .209 - - PO (α4) ? - - .0076 * .000 PO*D∆NIt-1 (α5) ? - - .0028 ** .013 PO*∆NIt-1 (α6) ? - - .0759 * .000 PO*D∆NIt-1*∆NIt-1 (α7) - - - -.0435 *** .072 Adj.-R-squared (%) 8.66 8.40 No. of obs 79,618 79,618

* Significant at the significant level of .01

** Significant at the significant level of .05

*** Significant at the significant level of .1

The regressions exclude extreme 1% on each side for the variables.

There is evidence that the earnings components are persistent recognized. The α₂ coefficient on lagged positive earnings changes is close to the outcome 0 (-.0184) and is significant (P value = 0.000). This indicates that income is recognized at the moment it is realized. The height of the result has nearly no effect on delayed or advanced recognition of the accounting profit. This is also confirmed by the sum of the coefficients α₂ + α₃ of - 0.522,

(21)

outcome is consistent with timely recognition of economic losses with a non-repeating character as a component of transitory income.

The incremental coefficient α₃ on lagged negative earnings changes is significantly negative, with a value of -0.5032, and is consistent with substantially more timely loss recognition than gain recognition.

Because the hypothesis of this study is about the impact of the magnitude in outstanding shares, I focus on the coefficient α₆ and particular coefficient α₇. Relatively, it is more likely that the transitory gains are recognized earlier at a larger magnitude of a public offering. Coefficient α₆ gives no support for earlier gain recognition when the magnitude of the public offering increases, a significant and positive outcome (0.233) is noted. With coefficient α₇support was found for the hypothesis. The negative result (-0.087) implies that the degree to which the group of shareholders increases, the less likely that companies consistently incorporate transitory losses of income. This indicates a higher earnings quality and supports the hypothesis. However, the negative outcome of α₇is not significant. Although no significant result is shown with model I, the hypotheses are also supported by the results of model II for coefficient α₇ (-0.043). The outcome of coefficient α₇ in Model II shows that there is a significant difference; it is less likely that companies consistently incorporate transitory losses in income in a year of public offering compared to a year without a public offering. Based on the results of the time-series test, the outcomes imply a higher earnings quality when the number of shareholders increases.

Overall, these results show a positive correlation between the magnitude of a public offering and earnings quality. This is consistent with the hypothesis.

4.2.2 Results Accruals-based model

Table 3 present the results of model I of the regression analysis from the Time-series model. The time-series model shows a relation between the accruals and cash flow levels of firm years with and without public offerings.

The role of accruals in mitigating negative serial correlation in cash flows predicts a negative coefficient for β₂. Asymmetric loss recognition predicts a positive coefficient for β₃. In accordance with the hypothesis, it is expected that firms with a smaller public offering are

(22)

less likely to recognize losses as transitory items, which implies that the asymmetry is lower. Therefore, for coefficient β₇a positive outcome is predicted.

Table 3 Outcomes Accruals-based model

Regression of accruals on cash from operations for all firm-years.

Accruals-based Model I with the change in Outstanding Shares (∆OSt) as continuous variable.

ACCt = β0 + β1DCFOt + β2CFOt + β3DCFOt*CFOt +β4 ∆OSt + β₅∆OS_t* DCFO_t

+ β6∆OSt*CFOt + β7 ∆OSt* DCFOt *CFOt + νt

Accruals-based Model II with a dummy variable for the firm-years with a public offering (PO).

ACCt = β0 + β1DCFOt + β2CFOt + β3DCFOt*CFOt + β4 PO + β₅PO*DCFO_t

+ β6PO*CFOt + β7 PO* DCFOt *CFOt + νt

Dependent variable is ACCt _{Model I} _{Model II}

Predicted Coefficient P Value Coefficient P Value

Intercept (β0) ? -.0231 * .000 -.0227 * .000 DCFOt (β1) ? .1630 * .000 .1589 * .000 CFOt (β2) - -.0000 * .000 -.0000 * .000 DCFOt*CFOt (β3) + .0000 * .000 -.0000 * .000 ∆OSt (β4) ? .0308 * .000 - - ∆OSt*DCFOt (β5) ? .1560 * .000 - - ∆OSt*CFOt (β6) ? -.0000 * .000 - - ∆OSt*DCFOt*CFOt (β7) + .0001 * .000 - - PO (β4) ? - - -.0094 * .000 PO*DCFOt*CFOt (β5) ? - - .0790 * .000 PO*CFOt (β6) ? - - -.0000 * .000 PO*DCFOt*CFOt (β7) + - - .0000 * .000 Adj.-R-squared (%) 36.94 37.20 No. of obs 92,225 92,225

* Significant at the significant level of .01

** Significant at the significant level of .05

*** Significant at the significant level of .1

The regressions exclude extreme 1% on each side for the variables.

For years without a public offering, β₂ is - .0000 and statistically significant at 0. This implies that on average non (0%) of the cash flows is offset by the accruals of years that these cash flows were positive. In accordance with the expectations, coefficient β₃ is

(23)

significantly positive ( 0.00002), indicating that negative cash flows are offset by accruals and therefore losses are recognized symmetrically.

In years with positive cash flows, firms with a lower magnitude of public offering do exhibit similar accruals behavior than firms with a higher magnitude of public offering (β₆is

-.0000 and statistically significant). Into years with cash losses: the incremental coefficient for

the height of mutation of outstanding shares in cash-loss years β₇is positive (.0001) and statistically highly significant (p-value = .000).

The results of Model II for coefficients β₂ and β₇are consistentwith Model I. The outcome of β₃coefficient implies that negative cash flows are not offset by accruals.

5. Conclusion

In this chapter, the conclusions are presented as well as some suggestions for future research, together with the limitations of this study. In Section 5.1, the conclusions from the results of the empirical research are described. In Section 5.2, the limitations and research agenda are displayed.

5.1 Conclusions

The study tries to address an answer for the central research question using literature and empirical research. The central research question is:

To what extent does earnings quality change when the group of stakeholders increases in public firms?

The literature shows that separation between management and ownership can lead to agency problems (Jensen and Meckling, 1976). A major cause is that information

asymmetry may arise because shareholders delegated decision-making powers to the management. Because the shareholders are further away from the business, they do not always know what is going on in the company.

Similarly a conflict of interest arises between managers and shareholders. To prevent this, the shareholders hire an independent party (i.e. auditors) to monitor the activities of managers (Anderson and Reeb, 2003). As the group of shareholders increases, the economic function of the financial statements also increases. Therefore, the growing number of

(24)

shareholders will increasingly insist on stricter control on the financial statements. Based on this theory the present research tried to confirm the hypothesis that the earnings quality will rise simultaneously with the increase of the group of shareholders. Because the economic function of financial statements has grown, a stricter control is expected which should lead to higher earnings quality.

The empirical results of the time-series test indicate that in accordance with the theory of timely loss recognition, persistent gains are accounted persistent, when losses occur they are estimated to be higher, and are one-off. This asymmetric accountability between gains and losses are confirmed by the empirical results.

In response to the research question the assumptions about the growth of earnings quality and the economic function to the extent of the increase of the group of shareholders is supported by the results of the regression analysis based on the time-series model. The results of the regression analyzes based on the accruals-based model are also supporting the hypothesis. A highly significant outcome implies that there is a positive correlation between the use of the recognition time when the issue of shares is growing stronger.

In previous research Ball and Shivakumar (2005) found evidence for higher earnings quality in public firms than in private ones. In that study, the separation between

management and ownership was examined. In this study, the magnitude of the separation of ownership and management was examined, similar effects were found, namely that the earnings quality improves and the economic function of the financial statements rises.

5.2 Limitations and future research

5.2.1 Limitations

Since companies made several public offerings during the period of 1950 and 2014, the progress in outstanding shares, and therefore in measuring the progress of earnings quality over the years, was a hindrance. By observing the magnitude of the seasoned public offerings in this study, all years without public offering are the benchmark years and the progress is not measured during these years. The complexity of measuring progress was a complex. The hurdle was solved by using a continuous variable for the magnitude of the mutation of outstanding shares.

(25)

This study looked at information asymmetry and agency problems, but several other factors may play a role in determining the earnings quality. These other factors are not examined. In addition, earnings quality has been affected by the continuous changes in the laws and regulations for accounting of financial statements. These changes affect the recognition of gains and losses. It could be the case that gains and losses were recognized differently in the accruals in different years. These effects are not included in this study either.

5.2.2 Future research

The research on the effect of increasing information asymmetry (through a growth of shareholders at a distance) on earnings quality provided several insights. Also, this study yielded new questions and suggestions for future research. A possible interesting research question is: What external factors play an increasing role as the group of shareholders increases? Another possibility is to examine the progress of outstanding shares and the correlation with quality earnings throughout the years with several seasoned public offerings.

(26)

Reference list

Ali, A., Chen, T., and Radhakrishnan, S. (2007). Corporate disclosures by family firms. Journal of Accounting and Economics, 84, 238-286.

Anderson, R., and Reeb, D. (2003). Founding-family ownership and firm performance: evidence from the Standard and Poor’s 500. Journal of Finance, 58(3), 1301-1328. Anderson, R., and Reeb, D. (2004). Board composition: Balancing family influence in

Standard and Poor’s 500 Firms. Journal of Finance, 49(2), 209-237.

Ang, J. S., Cole, R. A., and Lin, J. W. (2000). Agency costs and Ownership Structure. The Journal of Finance, 55(1), 81-106.

Ball, R. (2001). Infrastructure Requirements fora n Economically Efficient System of Public Financial Reporting and Disclosure. Brooking-Wharton Papers on Financial

Services, 0(0), 127-169.

Ball, R., and Shivakumar, L. (2005). Earnings quality in UK private firms: comparative loss recognition. Journal of Accounting and Economics, 39(1), 83-128.

Ball, R., and Shivakumar, L. (2008). Earnings quality at initial public offerings. Journal of Accounting and Economics,45(2-3), 324-349.

Basu, S. (1997). The conservatism principle and asymmetric timeliness of earnings. Journal of Accounting and Economics, 24(1), 3-37.

Burgstahler, D. C., Hail, L., and Leuz, C. (2006). The Importance of Reporting Incentives: Earnings Management in European Private and Public Firms. The Accounting Review, 81(5), 983-1016.

Chow, C. (1982). The demand for external auditing: Size, debt and ownership influences. The Accounting Review,57(2), 272-291.

DeAngelo, L. (1981). Auditor independence, ‘low balling’, and disclosure regulation. Journal of Accounting and Economics, 3(2), 113-127.

Dechow, P. M., Kothari, S. P., and Watts, R. L. (1998). The relation between earnings and cash flow. Journal of Accounting and Economics, 25(2), 133-168.

Dechow, P. M., and Schrand, C. (2004). Earnings Quality (3). Charlottesville, United States: Research Foundation of CFA Institute.

Dechow, P. M., Ge, W., and Schrand, C. (2010). Understanding Earnings Quality: A review of the proxies, their determinants and their consequences. Journal of Accounting and Economics, 50(2-3), 344-401

(27)

Healy, P., and Palepu, K. (2001). Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics, 31(1-3), 405-440.

Hope, O. K., Thomas, W. B., and Vyas, D. (2013). Financial Reporting Quality of U.S. Private and Public Firms. The Accounting Review, 88(5), 1715-1742.

Jensen, M. C., and Meckling, W. H. (1976). Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure. Journal of Financial Economics, 3(4), 305-360. Kakebeeke, P. (2014, June 20). Nederlandse bedrijven leunen minder op bankschulden.

Retrieved from http://fd.nl/beurs/212700-1406/nederlandse-bedrijven-leunen-minder-op-bankschulden

Lafond, R., and Rochowdhury, S. (2008). Managerial Ownership and Accounting

Conservatism, Journal of Accounting Research. Journal of Accounting Research, 46(1), 101-135.

Loughran, T., and Ritter, J. B. (1997). The operating performance of firms conducting seasoned equity offerings. The Journal of Finance, 52(5), 1823-1850.

Schipper, K., and Vincent, L. (2003). Earnings quality. Accounting Horizons, 17, 97-110.

Wang, D. (2006). Founding Family Ownership and Earnings Quality. Journal of Accounting Research, 44(3), 619-656.

Watts, R.L. (1992). Accounting choice theory and market-based research in accounting. The British Accounting Review, 24(3), 235-267.

Watts, R. L. (2003). Conservatism in Accounting Part I: Explanations and Implications. Accounting Horizons, 17(3), 207-221.

Watts, R.L., Zimmerman, J.L. (1986). Positive accounting theory, Prentice Hall, Englewood Cliffs, NJ.

Earnings quality at public offerings : the economic function of financial statements with a growing number of shareholders

Amsterdam Business School