The relationship between Financial Reporting Quality (FRQ) and Firm Ownership (Public Firm or Private Firm)

Master Thesis

The relationship between Financial

Reporting Quality (FRQ) and Firm

Ownership (Public Firm or Private Firm)

Student: Libing Qiu

Student Number: 10557237 Date: 23 June, 2014

Education: MSc Accountancy and Control (2013-2014), Accountancy track Institution: University of Amsterdam, Faculty of Economics and Business Supervisor: Dr. Bo Qin

Co-assessor: Dr. Alexandros Sikalidis Version: 1.0

Page 1 of 43

Table of Contents

1 Abstract ... 2

2 Introduction ... 3

2.1 Motivations ... 5

3 Literature Review and Hypothesis Development ... 7

3.1 Earnings Management ... 7 3.2 Conservatism... 9 3.3 Public Firms... 10 3.4 Private Firms ... 10 3.5 Hypotheses development ... 12 4 Methodology ... 16

4.1 Accrual Quality Measures ... 16

4.2 Conservatism Measures ... 18

5 Sample and Descriptive Statistics... 19

6 Results ... 25

7 Sensitivity analysis ... 29

7.1 Sensitivity analysis by controlling the year ... 29

7.2 Sensitivity analysis by controlling the industry. ... 33

8 Conclusion and Limitation ... 37

Page 2 of 43

1 Abstract

Although some debates still exist, the recent research concludes that public firms have higher financial reporting quality than private firms. This research, however, focuses on a new perspective to investigate the level of financial reporting quality between private and public firms, namely on comparing the financial reporting quality of the same firms but between their privately-hold and publicly-hold periods, instead of comparing different firms at the same time period, a topic that has been studied by many researchers. It is found that the accrual quality and reporting conservatism, proxies for the financial reporting quality, of the same firms are higher in the public holding phase than that in the private holding phase, consistent with the general belief that the financial reporting quality of public firms are higher than that of private firms.

Page 3 of 43

2

Introduction

During the last decades, there is an increasing focus on earnings management research. According to Walker (2013), in the last decade, among all the accounting research publications, between 7% and 10% of them are on Earnings Management research. In the other words, there are roughly 30 earnings management articles published in each of important accounting journals per year. Furthermore, it is worth noticing that the publishing rate was materially higher for second half of last decade than the first half. It is obvious that this topic is attracting more and more researchers’ attention.

Previously, research focus was emphasized on earnings management in public firms (Burgstahler et. al, 2006). And they always focus on the influence factors that may have impact on the incentive of applying earnings management. Public firms are always large, so even a small amount of them have significant influence on economy. Therefore there have been already a lot of researches on the accounting performance of these firms, according to which it seems that the public firms are more likely to use earnings management during their financial reporting.Within the previous researches, there are very few researches focusing on the Financial Reporting Quality (FRQ) of the private firms, because many people believe that their economic impacts are relatively small due to the fact that private firms are generally small comparing to public firms. For instance, Hope et al. (2011) included the private firm as well, however, their researches just used the private firm as an comparison sample for the public firm to present the influence factors of earnings management in public firms. Furthermore, the limited data resource of private firms is also a reason for that there has been much less public focusing on private firms.

However, with the economic globalization, it is noticeable that the influence of private firms on economy is increasing significantly, due to the rapid growing amount and the size of private firms. According to La Porta et al. (1999) even some of the world’s largest firms are privately held. Hope et al. (2013) highlighted the importance

Page 4 of 43

of private firms. They showed that according to the data obtained from Census Bureau of the U.S., there are 29 million privately held companies in the whole country, among which 7.6 million companies representing one-half of the nation’s GDP. All of these realities attract more and more research attention on private firms. Therefore, it is important to have equal treatment of private firms as public firms.

With the demand of firms’ development, more and more private firm register as public firms. As many researches (Pagano at.al, 1998) mentioned, becoming public firms reduce the overall cost of capital and offer the firms more solid standings when they have to negotiate financial affairs with banks e.g. interest costs on existing debt. The main motivations for private firms deciding to register to be public firms are to raise money for the firms’ future development and to share the risk with a larger group of shareholders. Therefore registering to be public firms offers the firms the opportunity to have their stock listed on a stock exchange. Going to be public firms and offering their stocks is a milestone for the former private firms’ future development.

On the other hand, there are public firms deregister to be private firms as well. These firms choose to go private in order to avoid the risk and cost of being public reporting firms. Most of the time, after seeking financing phase, when the firms are no longer interested in making acquisitions, they may decide to deregister. As mentioned by Jannat Thompson, a Wall Street lawyer, in 2005, there are many factors that trigger the public firms going private, while the two main reasons are that it may eliminate the cost of time and money in complying with the requirement of Sarbanes-Oxley (especially the legal and accounting fees), and that the firms may benefit from conducting business out of the public concern. This research will only focus on the private-to-public transaction because there is a limitation of data availability of firm deregistration (see Section 4). The research (Renneboog et al,2007) investigated the public-to-private transactions by collecting 181 public-to-private samples from 1997 to 2003 from the database of the Centre for Management Buyout Research (CMBOR), which is inaccessible from our university network. For the accessible databases in

Page 5 of 43

WRDS, on the other hand, there is no relevant information, which can be used to identify the year of public-to-private transactions (or deregistration).

According to Beuselinck and Manigart (2007), when the private equity investors have high equity stake in the firms, their accounting information quality is lower than that when the private equity investors have low equity stake in the firms. Hope et.al, 2013 present that public firms have higher FRQ than the private firms through analyzing the U.S private and public sample. Beatty and Harris (1999) and Beatty et al. (2002) have shown that there is more earnings management in public firms than private firms among U.S. banks. According to Leuz et al. (2003), under weak legal system, earnings quality is less transparent in both private and public firms. Focusing on private and public sample in 13 EU countries, Burgstahler et.al, (2004) have found that capital markets provide more incentives for the firms to hide economic performance.

Despite that all these prior researches emphasized on the comparison of private and public firms’ accounting information quality, this research focuses on a different perspective, which is the relationship between the change of a firm’s ownership structures ( from publicly holding to privately holding) and their FRQ. Therefore, this research will not make the comparison of the FRQ between private and public firms, but make the comparison of the FRQ before and after the registering of the firms.

2.1 Motivations

This research provides a new perspective to investigate the situation of FRQ in public holding and private holding structures. Previously, Hope et al. (2013) have compared the FRQ between public and private firms in U.S from 2001 to 2009. Burgstahler et al., (2004) have investigated on the same topic but focused on firms in 13 EU countries from 2001 to 2006. Paganor et.al, (1998) have analyzed Italian private firms to show the reason why firms choose to go public. However, this research will focus

Page 6 of 43

specifically on the level of FRQ changing in the firms that experienced the ownership transaction instead. To my best knowledge, there is no prior research investigating this topic, so this research will contribute to fill the gap in this area.

Additionally, with a relatively new database, which contains data on nearly 100,000 private firms, Hope et al. (2013) have found that in the U.S. public firms have significantly higher accrual quality than private firms, which is contrary to the earlier research findings (Givoly et al. 2010), claiming that private firms had higher accrual quality in the U.S. (consistent with the opportunistic behavior hypothesis). The two contrary results may be attributed to the limited number of private firm samples in the earlier research. Therefore, it is of importance to compare the accrual quality of the same firm. It is noticeable that there are a large amount of firms experienced private-to-public transactions. The motivation of this research is to investigate the accrual quality difference of the same firms before and after these transactions.

This research focuses on the private-to-public firms in UK. The author believe that using UK sample to study this topic will be quite interesting and will get valuable findings as well. Firstly, UK regulations for public and private firms are robust, and both public and private firm reporting are subject to equivalent regulations in UK. Ball and Shivakumar, (2004) mentioned three principal features of the UK financial reporting regulations: 1). UK Companies Act requires both public and private firms to generate annual financial reporting basing on the same accounting standards. 2). The UK regulations require that the financial statements of private firms should be audited. 3). UK public and private firms are both subject to the same tax law. Therefore, the results based on UK samples will be more reliable because the country regulations do not have different influences on public and private firms. Secondly, as mentioned by Renneboog et al. (2007) the amount of private firms that going public was increasing rapidly in the past years in UK. Therefore, it is worthy investigating the switch from private and public firms in UK because I can have a relatively large sample under same regulation environment.

Page 7 of 43

3

Literature Review and Hypothesis Development

3.1 Earnings Management

There are many prior researches that gave the definition of Earnings management. Walker (2013)’s definition is presented as that the earnings management is through managerial discretion of accounting choices, earnings reporting choices, and real economic choice to impact on the measures of earnings in the economic events.

There are various ways for managers to manage earnings through both real economic and accounting methods. Many prior researches present the incentives of managers to manage earnings. Firstly, managers have bonus incentive. Jensen and Murphy (2005) find that because CEOs’ compensation based on firms’ stock market price, they have incentive to manage earnings -- in order to gain more bonus managers may have strong incentive to manipulate their earnings to meet their targets. Lots of researchers have found that showing good performance is the most common incentive of managers in firms. Secondly, Management has inventive to manipulate stock price by over-reporting short term earnings. According to Louis (2004) even sophisticated investors will make wrong investment decisions because of earnings management. It is worth mentioning here that earnings management reduces firms’ long term value (Friebel and Guriey, 2005). According to Bebchuk and Gill (2003), Scheinkman and Xiong (2003), and Bolton et al, (2004) investors are rational so they understand that firms’ stock price does not always indicate firms value as it is not always rely on the most optimistic signals. Thirdly, firms may need to achieve tax optimization purpose (Misai 2003). In order to achieve this objective managers may under-reporting the earnings. By avoiding paying more taxes, the shareholders’ value will be increased. Suwardi (2013) present the empirical evidence of that manufacturing firms have the trend to do earnings management by lowering their current accruals.

Page 8 of 43

Also, a lot of prior researches have focused on the different incentives of public and private firms for earnings management. Burgstahler et al. (2004) find that in EU countries private firms have more incentives to manage earnings, and also in weak legal environment both public and private firms have more chance for the earnings management. On the other hand, according to Givoly et al. (2010), public firms have more incentives to show good results of the financial year, or they prefer not showing bad results to their stakeholders than the private firms.

There are lots of evidences indicating that different ownership structures offer different rights and opportunities for managers to manipulate earnings, so the results of earnings management in private and public firms are expect to be different, leading to different FRQ.

Jensen (2004) have shown that the equity overvalue may increase the agency problem between investors and managers. Most of the prior researches have used agency theory to investigate earnings management topic, because according to (Walker, 2013) agency theory is a standard approach, which most academic accountants used when they study the earnings management. Agency theory presents the conflict between firms’ executives and shareholders. When the interests of managers are aligning with the shareholders, the incentives of earnings management are to increase firms’ value. But when the interests of executives are not aligning with the shareholders, the incentives of earnings management are to meet executives’ own purposes.

According to Walker (2013), agency theory also emphasizes on the problems of information asymmetry between firms’ executives and outside investors. FRQ is important for investors to make relative decisions because if the managers do the earnings management for their own purposes, it is harmful for the outside investors to understand the true situation of the investing firms. According to (Pagano at.al, 1998), the main purpose that firms choosing to go public is to seeking more external financing. It is possible that the managers managing their earnings in order to make

Page 9 of 43

nice picture for the outside investors. The outside investors do not fully understand the business inside firms. The only way for them is through the disclosed accounting statements. It is, therefore, clear that the FRQ is really important for the investors to make good judgment of the firms’ economic performances. Therefore, it is crucial for the outside investors and other stakeholders to get a true economic performance of the firms through a high level FRQ.

3.2 Conservatism

Normative research on conservatism started in the early 20th century, stimulated by the introduction of corporate income taxes. However, academic researchers were unable to provide any definitive conclusions until 1997, when Basu translated the traditional accounting principle of “anticipate all losses but anticipate no gains” as “capturing accountants' tendency to require a higher degree of verification for recognizing good news than bad news in financial statements.” After conducting research on a large amount of observations in a time interval from 1963 to 1990, he found that the concurrent sensitivity of earnings to negative returns (a proxy of bad news) is two to six times larger than that to positive returns (a proxy of good news); the greater timeliness of earnings than cash flow is due to more timely recognition of 'bad news' through accruals (the timeliness of earnings over cash flows for good news is not increased); positive earnings changes tend to persist, whereas negative earnings changes show a marked tendency to reverse; positive earnings changes have higher earnings response coefficients (or abnormal return per unit currency of unexpected earnings) than negative earnings changes.

These findings have been developed as models to investigate the FRQ in private firms and public firms (Ball and Shivakumar, 2005; Hope et al. 2013). This research has applied the models and findings that related to conservatism measurement from Basu 1997 as well.

Page 10 of 43

3.3 Public Firms

A public firm is defined as a limited company whose shares may be purchased by the public and traded freely on the open market and whose share capital is not less than a statutory minimum. So public firms are mostly large firms, and have much more demands for the external financing for their future extension.

According to prior researches (Pagano at.al, 1998), the purpose of the firms going public is mainly to raise money to help the firms’ future development. (Balasubramaniam, 2009) presented the reasons that more and more firms going to be public firms to: 1). raise additional funds through the issuance of more stock; 2). gain possibility to offer securities in the acquisition of other firms. 3). gain possibility to supplying equity-based programs to potential employees to attract higher performance of employees; 4). get additional leverage when they obtain loans from other financial institutions; 5). get more attention of the mutual and hedge funds, market makers and institutional traders when their stock listed on an exchange. 6). increase firms’ credibility with the public.

It is clear that the level of FRQ of the firm that going public is quite important for the investor. When the FRQ is high, the investors will get more reliable information from firms’ accounting statements. However, if the FRQ is low, it is possible that the investors will make wrong investment decision.

3.4 Private Firms

A private firm is defined as a limited firm that does not issue shares for public subscription and whose owners do not enjoy an unrestricted right to transfer their shareholdings. The private firms are mostly small scare. Some of them only have simple ownership structure; some of them are even family-owned. However, Furthermore, because the properties of private firms change over time, the capital

Page 11 of 43

structures may become more complex; the distance between shareholders and executives become larger; their public concern about the accounting performances is also expected to be increasing. So the “demand” of private firms becomes stronger, and managers are more sensitive to the financial targets and performance pressures.

There are some researches that emphasized on the public-to-private transaction of the firms as well, for instance, Renneboog et.al, (2007), Ritter and Welch, (2004). People may feel not understandable for the purpose that the firms going private. When the pubic firms change their business strategy, e.g. no longer seek external financing, or do not want public concern, they choose to go back to private firms. It is important to mention that these firms are different from the traditional private firms: they are mostly large scared and have complex ownership structure as well. Obviously, the economic impact of this kind of firms is still huge even after they deregister as private firms. Steele, (2009) explained the reasons for the firms going back private: 1). Firms will reduce the cost of being public firms. 2). Firms will reduce the potential liability for directors when they follow Sarbanes-Oxley Act. 3). Firms may reduce the disclosure obligation for competitive business information. 4). Firms can be longer term focus instead of ‘managing for the next quarter’. 5). Firm may have more freedom on corporate governance flexibility, but not always to meet the requirements of security market regulation.

According to the prior researches such as Hope et.al, (2011) and Burgstahler et.al, (2006), the private firms have less incentive to manage earnings. However, in this research, the private firm samples are not the traditional ones because they are mostly large-scared and complex-structured so as to meet the need to become public holding. Most of them already have all the properties of public firms. Therefore, investigating the FRQ level that before register phases will be valuable for the users to understand how FRQ changes in different business phases.

Page 12 of 43

3.5 Hypotheses development

This research use accrual quality and conservatism as the proxies for the measurement of FRQ. They will be measured separately under both before and after private-to-public transactions.

According to the concept of “demand” hypothesis (Givoly et al. 2010) private firms have weak incentive to achieve high quality financial reporting because the capital providers and other stakeholders have internal access to obtain the financial information of the firms and there is weak demand due to the close relations between shareholder and executives. This hypothesis predicts that the FRQ is low in private firms.

On the other hand, according to opportunistic behavior hypothesis (Givoly et al. 2010), executives of private firms have weak incentives to manage earnings through manipulating financial reporting, because they normally do not have capital market pressures to meet earnings expectations and their personal wealth is not directly connected to firms’ value, resulting in a high accrual quality.

The two completely different hypotheses lead to a question: whether the FRQ is higher in public holding ownership or in private holding ownership. Hope et al. (2013) have empirically supported the demand hypothesis, while the even earlier research findings supported the opportunistic behavior hypothesis instead, according to Hope et al. (2013) itself. The different results may be due to the different amount and market size of the samples or different investment period, with the newer result to be more reliable.

Additionally, many prior studies, i.e. Reynolds and Francis (2000), have shown that firms that employing Big 4 auditors are positively related to high accrual quality because that Big 4 auditors have more resources and experience to conduct effective

Page 13 of 43

audit. It is generally agreed that there are much less possibilities for private firms to hire Big 4 auditors, because private firms are comparably small and family owned, so it is a relatively big burden for them to spend this cost. Public firms, on the other hand, have a large preference to hire Big 4 auditors to reduce uncertainty about their financial statements and to protect investors from future losses due to audit failure.

There are many prior researches that mentioned the public firms have more incentive to manage the earnings because the public firms have the requirements to meet more financial conditions.

Previously, Burgstahler and Dichev (1997) and Degeorge et al. (1999) have shown that firms turn to use accounting discretion to avoid reporting small loss, manipulating it to be small profit. They mentioned that the public firms may care more about debt financing, compensation, dividend smoothing, management buy-outs, and potential initial public offerings. Kanagaretnam et al. (2004) and Dechow et al. (2011) have shown that firms seeking for external financing are more likely to manipulate their financial reports, and this manipulation are mostly through upward earnings management and income smoothing. Some prior researches (Hope et al. 2013) believe that because private firms are largely manager or family owned, these firms do not face the capital market pressures for external financing, hence private firms are less likely than public firms to manage earnings.

According to these prior researches, it might also be true that the accrual quality is higher in private firms than in public firms, which is inconsistent with the result of Hope et al. (2013).

Therefore, the first hypothesis will be: through the transaction from private firm to public firm the accrual quality will become higher.

Page 14 of 43

The literature provides ample evidence that managers aggressively report earnings to meet earnings target (Roychowdhury 2006). This research will test whether managers will remain using conservative reporting method when they do not meet their financial target. Many researches (Hope et al. 2013) show that public firms face more capital market pressure than the private firms. So at that time the public firms may have more incentives to manage earnings, which lead to a low accounting conservatism.

Watts (2003) shows that four factors, contracts (debt and compensation contracts), litigation, taxation and regulation, may impact the conservatism. According to Khan and Watts (2009), these four factors vary as firms have different investment opportunity set (IOS). When the firms have growth options relative to fewer debt contracts (market-to-book ratio), they are more likely to have higher probability of litigation, lower taxable earnings and to be unregulated (Smith and Watts, 1992). In the prior literature like Easley at al. (2002), there is a net effect that the large firms have lower information asymmetry than the small firms, which show a lower contracting demand for reporting conservatism in large firms. The reason is that compared with private firms, the public firms are always large scaled, which brings them less incentives to report conservatively. They also find that the firms with more financial leverage have higher contracting demand to report conservatively. However, According to Hope et al. (2013), they explain that no matter the financial leverage is high or low, private firms always have less need to report conservatively because they mostly contract by communicating through private channels. Hope et al. (2013) have shown that when the financial leverage is low, the public firms are reporting less conservatively.

This research develops the second hypothesis basing on the results from these prior researches, meaning public firms reporting less conservatively than private firms. Therefore, it is predictable that through public to private transaction, firms reporting conservatism will be higher. In contrast, it is also predictable relatively that through

Page 15 of 43

private to public transaction, firms reporting conservatism will be lower.

Page 16 of 43

4

Methodology

This research will use a quantitative research method, with all the samples that located in the UK. The main database that will be used in this research is S&P Capital IQ's Compustat Global, provided byWRDS (Wharton Research Data Services), a database of more than 33,900 non-U.S. non-Canadian companies with annual data history from 1987. The advantages of using Compustat Global are: 1). it provides achieve accurate, comparable results with high quality, standardized data; 2). it has deep, historical company analysis with unrivaled historical fundamental data available back to 1950; 3). It has access to extensive market data, including historical daily and monthly prices, dividends, and corporate actions. This research is to focus on the annual accounting reports of firms in the UK. In the following part of the research paper, models for accrual quality and conservatism, which are planned to use in the research, will be explained.

4.1 Accrual Quality Measures

There are several popular methods to measure the accrual quality, such as the model basing on performance-adjusted discretionary accruals (Kothari et al. 2005), the model focusing on the deviation the relation between current accruals and past, present, and future cash flows (Dechow and Dichev 2002; Ball and Shivakumar 2006; Givoly et al. 2010), the model on revenue-accrual quality (McNichols and Stubben 2008; Stubben 2010), and the model on the ratio of the absolute value of accruals to cash flows (Burgstahler et al. 2006). In this research, I will only use the first model, developed Kothari et al. (2005), to measure the accrual quality.

The model of Kothari et al. (2005), utilizing performance-adjusted discretionary accruals, reads:

, 0 1(1 ,t 1) 2 , 3 , 4 , ,

i t i i t i t i t i t

TA =α α+ Assets ₋ + ∆α Rev +α PPE +α ROA +ε ,

where:

Page 17 of 43

,

i t

TA is the total accruals of firm i in year t, measured as the change in non-cash current assets minus the change in current non-interest-bearing liabilities excluding the current portion of long-term debt, minus depreciation and amortization, scaled by lagged total assets;

,t 1

i

Assets ₋ is the lagged total assets of firm i in year t;

,

i t

Rev

∆ is the annual change in revenues of firm i in year t scaled by lagged total assets;

,

i t

PPE is the property, plant, and equipment of firm i in year t, scaled by lagged total assets;

,

i t

ROA is the net income of firm i in year t scaled by average total assets. The residuals of Eq. (1) are used as a proxy for discretionary accruals, absolute values of which (AB_DA) are used as a proxy for the accrual quality, in a relation that the higher the AB_DA, the lower the accrual quality.

To test the H1, which is private to public transaction leads to higher accrual quality, I am planning to choose the sample firms which were previously private, and experienced a private-to-public transaction later to become public. The equation for testing H1 is:

, 0 1 , 2 , 3 , 4 , 5 ,

_ _{i t i t} _ _{i t i t} _ _{i t i t}

AB DA =β +β Public +β Log Assets +β ROE +β Std ROA +β Lev

+β₆Growth_{i t,} +β₇Op Cycle_ _{i t,} + +β₈Inv_{i t,} +β₉Loss_{i t,} +ε_{i t,}

where Publici,t is equal to 1 when the firm i is public in year t, and equal to 0 otherwise. Therefore, β₁ represents the differential accrual quality of the same firms before and after the private-to-public transactions. The other controls, which are all defined in Ashbaugh-Skaife et al. (2008), are: log of total assets (Log_Assets), return on equity (ROE), standard deviation of return on assets (Std_ROA), financial leverage (Lev), growth (Growth), operating cycle (Op_Cycle), one-year-ahead percentage change in common stock, preferred stock, and long-term debt (Capital_Need),

Page 18 of 43

inventory divided by total assets (Inv), and the cumulative proportion of years with losses (Loss). Because lower value of AB_DA represents the higher accrual quality, in order to fulfil H1, β₁ should be negative and significant.

4.2 Conservatism Measures

There are also several models to qualify the conditional conservatism, such as the two models defined in Ball and Shivakumar (2005). In this thesis, I will use one of the models, basing on the relation between accruals and cash flow, to test conservatism. The model is written in Eq. (3),

, 0 1 , 2 , 3 , , 4 ,

i t i t i t i t i t i t

TA =γ +γ OCF +γ DOCF +γ DOCF ×OCF +γ Private

5DOCFi t, Privatei t, 6OCFi t, Privatei t,

γ γ

+ × + ×

7DOCFi t, OCFi t, Privatei t, i t,

γ ε

+ × × +

where TAi,t is the total accruals for firm i in year t, OCFi,t is the operating cash flows, and DOCFi,t is an indicator variable, which is equal to 1 when OCFi,t is negative and equal to 0 when OCFi,t is positive. According to Basu (1997), conditional conservatism means that economic losses are more likely to be recognized on a timely basis and economic gains are more likely to be recognized when realized. Therefore, more conservative reporting results in γ₃ > . Consequently, if we expect that firms 0 would have more conditional conservatism after they transfer from private to public (supporting H2), the coefficient of the interaction term DOCF_{i t,} ×OCF_{i t,} ×Private_{i t,} (γ₇) would also be positive, and vice versa.

Page 19 of 43

5

Sample and Descriptive Statistics

In order to compare the accrual quality and conservatism of the same firms between its public period and private period, I need to know exactly which year each individual firm experienced its private-to-public or public to private transactions. It is relatively easy to identify the year of private-to-public transaction for each firm – we can simply set it to be the year of Initial Public Offering (IPO); this kind of information can be easily found in S&P Capital IQ's Compustat Global. On the other hand, I encountered a major problem with identifying the year of public-to-private transactions. There are some prior researches on the topic of public-to-private transactions of UK firms (Wright et al., 2000; Renneboog 2007). However, they all used the database of the Centre for Management Buyout Research (CMBOR), devoted to the study of private equity and buyouts. Unfortunately, this database is inaccessible by our university, which makes the study of public-to-private transactions impossible. Hence, although it is interesting to compare the accrual quality and conservatism of the same firms before and after the public-to-private transactions, which have never been done before, this work is technically impossible. Therefore, this thesis will only focus on comparing the accrual quality and conservatism of the same firms before and after the private-to-public transactions, namely to test H1 and H2 of this thesis.

I collected the financial information of UK firms from 1995 to 2014 (two decades), which is equal to an amount of 26,936 firm-year observations for 2871 firms. Unfortunately, only 476 (out of 2871) firms have their IPO information available in the database, which reduces the firm-year observations from 26,936 to 4,249 (see Table 1). Furthermore, since Compustat Global is a database focusing on the public firms, I could only investigate the accrual quality and conservatism of the firms within ± 2 years of the IPO – the annual reports 3 years prior to IPO are not available for most of the firms in this database. Because the selected period for data collection is 1995 – 2014, in order to compare the accrual quality and conservatism of firms pre-

Page 20 of 43

and post-IPO, IPO should be published in the range of 1996 – 2013, which further reduced the number of firms to be 452 and number of firm-year observations to be 3,968. (For convenience, in the following part of this thesis, I will refer the IPO year as Year 0, the financial year that 2 years before the IPO year as Year -2, the financial year that 1 years after the year of IPO as Year 1, etc.) Furthermore, because many parameters in the equations shown in Section 4 need to be scaled by lagged total assets, the 1-year advance total assets, the financial information of Year -2 cannot be used for regression, but only for calculating the dependent and independent parameters of Year -1. As a result, only the accrual quality and conservatism of firms in Year -1, 0 +1, and +2 can be compared, which means that the number of firm-year observations for pre-IPO (private) are around 1/3 of that for post-IPO (public). Because the annual reports of the firms in Year 0 were always published in the end of that year, the firms were actually public at that moment. In order to get rid of the problem resulted from the asymmetric number of the data samples for private and public firms, I made four regressions to compare the accrual quality and conservatism: [(Year) -1 vs. (Year) 1], [-1 vs. 1,2], [-1 vs. 0,1], and [-1 vs. 0,1,2]. The number of observations for each regression is 422, 720, 656, and 954, respectively. The information about sample selection is available in Table 1.

Table 1 Sample Selection

Set B Number of firms Number of firm-years

Observations available in database 2,871 26,936

Observations with IPO available 476 4,249

Observations with IPO from 1996 to 2013 452 3,968

[−1 vs. 1] 306 422

[−1 vs. 1, 2] 343 720

[−1 vs. 0, 1] 311 656

Page 21 of 43

It is worth mentioning that although the sample size of this research is much smaller than other research on the topic of FRQ, it is relatively larger than the sample size of the study of public-to-private transactions such as Renneboog et al. (2007), which has only 177 observations in total.

According to Kothari et al. (2005), the residuals from the regression of Eq. (1) are used as a proxy for discretionary accruals, while the absolute value of them (AB_DA) is a for accrual quality, in a relation that the higher of AB_DA the lower the accrual quality. The regression results of Eq. (1) for the four different sample sets are shown in Table 2, with all the coefficients significant to the 1 percent level.

Table 2

, 0 1(1 ,t 1) 2 , 3 , 4 , ,

i t i i t i t i t i t

TA =α α+ Assets ₋ + ∆α Rev +α PPE +α ROA +ε

Variables −1 vs. 1 −1 vs. 1, 2 −1 vs. 0, 1 −1 vs. 0, 1, 2 Intercept −0.03193*** [−3.25153] −0.02683*** [−4.28043] −0.04226*** [−3.64226] −0.03194 [−3.75518] 1/Assets −0.00243*** [−5.68812] −0.00169*** [−8.46794] −0.00136** [−2.16110] −0.00398*** [−15.0955] ΔRev −0.00683*** [−4.3219] −0.00755*** [−7.84474] −0.02774*** [−69.30853] −0.02733*** [−80.147] PPE −0.08825*** [−18.3315] −0.07914*** [−22.4012] −0.05236*** [−17.66710] −0.05238*** [−20.37] ROA −0.01033*** [−14.6664] −0.00900*** [−17.3385] −0.00774*** [−10.59305] −0.0085*** [−14.0011] # of firm-years _{422 720 656} 954 R2 0.54853 0.50562 0.95694 0.95203

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. t−statistics are in brackets.

Page 22 of 43

After obtaining all the coefficients of (1) for four different sample sets, [-1 vs. 1], [-1 vs. 1,2], [-1 vs. 0,1], and [-1 vs. 0,1,2], which are all listed in Table 2, the value of AB_DA for each firm-year observation is calculated. As a result, all the variables in Eq. (2) (for testing H1) are known. Table 3 shows the descriptive of sample set [-1 vs. 1] with a total number of 422 observations. The mean, median, standard deviations, and 25 and 75 percentile each variable are listed in column 3 to column 7. All the variables will be used in the accrual quality regression.

Table 3

Descriptive Statistics for [−1 vs. 1]

Variable n Mean Median Std. Dev. 25% 75%

AB_TA 422 0.05345 0.03321 0.08663 0.016955 0.059213 Log_Assets 422 1.42314 1.31439 0.97761 0.728658 2.034733 ROE 422 -0.22423 0.02718 6.34583 -0.41058 0.201463 Std_ROA 422 0.75662 0.11936 4.17819 0.042967 0.363817 Lev 422 3.07056 1.47392 43.20923 0.839876 2.587896 Growth 422 4.74465 0.34985 43.40116 0.073721 0.962251 Op_Cycle 422 407.0187 134.2077 1455.975 70.64264 227.0125 Inv 422 0.05951 0.01745 0.09170 0 0.082338 LOSS 422 0.49040 0.40000 0.41782 0 1

Table 4 shows the descriptive of sample set [-1 vs. 1, 2]. The mean, median, standard deviations, and 25 and 75 percentile each variable are listed in column 3 to column 7.

Table 4

Descriptive Statistics for [−1 vs. 1, 2]

Variable n Mean Median Std. Dev. 25% 75%

AB_TA 720 0.04728 0.02716 0.07818 0.015185 0.058582 Log_Assets 720 1.44829 1.36568 0.94231 0.797077 2.044473 ROE 720 −0.39197 0.00755 6.02345 −0.41133 0.18108 Std_ROA 720 1.34710 0.11543 14.60584 0.04266 0.353507

Page 23 of 43 Lev 720 3.47580 1.51587 35.62628 0.745461 2.287274 Growth 720 3.43808 0.30141 33.82932 0.062082 0.808382 Op_Cycle 720 469.5899 133.6509 1997.0733 71.63618 237.7863 Inv 720 0.06247 0.01829 0.10182 0 0.079647 LOSS 720 0.50588 0.50000 0.42107 0 1

Table 5 shows the descriptive of sample set [-1 vs. 0, 1]. The mean, median, standard deviations, and 25 and 75 percentile each variable are listed in column 3 to column 7.

Table 5

Descriptive Statistics for [−1 vs. 0, 1]

Variable n Mean Median Std. Dev. 25% 75%

AB_TA 656 0.08340 0.03844 0.30816 0.020047 0.068479 Log_Assets 656 1.44995 1.34604 0.95727 0.769672 2.058772 ROE 656 −0.13249 0.02853 5.23474 −0.3709 0.193512 Std_ROA 656 0.74309 0.11750 3.59584 0.042944 0.363817 Lev 656 2.52157 1.23857 34.81399 0.525391 2.221775 Growth 656 3.61350 0.35667 34.95312 0.082443 0.878158 Op_Cycle 656 443.5577 132.2965 2161.2170 71.48966 238.8846 Inv 656 0.05620 0.01572 0.08697 0 0.078592 LOSS 656 0.48176 0.40000 0.41622 0 1

Table 6 shows the descriptive of sample set [-1 vs. 0, 1, 2]. The mean, median, standard deviations, and 25 and 75 percentile each variable are listed in column 3 to column 7.

Table 6

Descriptive Statistics for [−1 vs. 0, 1, 2]

Variable n Mean Median Std. Dev. 25% 75%

AB_TA 954 0.07049 0.03107 0.26974 0.015185 0.058582 Log_Assets 954 1.46055 1.37081 0.93654 0.797077 2.044473 ROE 954 −0.28774 0.00970 5.33296 −0.41133 0.18108 Std_ROA 954 1.19296 0.11544 12.73567 0.04266 0.353507

Page 24 of 43 Lev 954 2.99889 1.39177 31.08286 0.745461 2.287274 Growth 954 2.98074 0.31799 29.49640 0.062082 0.808382 Op_Cycle 954 479.3676 133.1209 2298.1447 71.63618 237.7863 Inv 954 0.05946 0.01640 0.09654 0 0.079647 LOSS 954 0.49614 0.40000 0.41942 0 1

Page 25 of 43

6

Results

The result of accrual quality comparison for firms between its private period and public period are shown in Table 7.

Table 7

, 0 1 , 2 , 3 , 4 ,

_ _{i t i t} _ _{i t i t} _ _{i t}

AB DA =β +β Public +β Log Assets +β ROE +β Std ROA

+β₅Lev_{i t,} +β₆Growth_{i t,} +β₇Op Cycle_ _{i t,} + +β₈Inv_{i t,} +β₉Loss_{i t,} +ε_{i t,}

Variables [−1 vs. 1] [−1 vs. 1, 2] [−1 vs. 0, 1] [−1 vs. 0, 1, 2] Public indicator −0.02331** _[−2.36484] −0.02512*** _{[−3.06182] [}−0.02232 _{−0.74861] [}−0.02267 _−0.86808] Log_Assets 0.00337 [0.68403] 0.00358 [1.026902] 0.02818 [2.07010] 0.02000* [1.900521] ROE _[0.32215] 0.00022 _[−0.00005 _{−0.09838] [0.04886]} 0.00011 _[−0.00048 _−0.28189] Std_ROA _[0.595249] 0.00061 _[0.01069] 0.00000 0.01850*** _{[5.69306] [1.19905]} 0.00082 Lev −0.00005 [−0.52303] −0.00004 [−0.50465] −0.00014 [−0.42081] −0.00015 [−0.49952] Growth 0.00004 [0.45057] 0.00008 [0.923437] 0.00008 [0.25978] 0.00029 [0.99515] Op_Cycle _[−0.88726] 0.00000 _[−1.14997] 0.00000 _[−0.00000 _{−0.47671] [}−0.00000 _−0.22843] Inv _[−0.08971* _−1.91484] −0.05995** _{[−2.0751] [}−0.22250* _{−1.64881] [}−0.08973 _−0.97982] LOSS 0.00636 [0.54266] 0.015403* [1.93397] 0.06388** [2.01867] 0.06759*** [2.83187] # of firm−years 422 719 655 953 R2 0.0254 0.0262 0.0622 0.0146

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. T-statistics are given in brackets.

Page 26 of 43

For the comparison between Year -1 and Year 1, the coefficient β₁ is negative (-0.02331) and significant at the 5 percent level, indicating that, comparing to Year -1, when the firms are private, firms’ accrual quality in Year 1, when the firms are public, is increased. For the comparison between Year -1 and Year 1 & 2, the coefficient β₁ is negative (-0.02512) and significant at the 1 percent level, indicating that comparing to Year -1, firms’ averaged accrual quality of Year 1 and Year 2 is also increased. On the other hand, when Year 0 is also included in the regression, however, the coefficient β₁ becomes insignificant (see the last two columns of Table 7), demonstrating that although the firms in the year of IPO are already public, their accrual quality is not increased yet. It is worth noticing that unlike the other research such as Hope et al. (2013), who use the similar model to compare the accrual quality of different public and private firms and see most of the control variables significant in the regression, most of our control variables are insignificant (see Table 7). This may due to the fact that I am comparing the accrual quality of the same firms within a short period (4 years at maximum), and in principle the control parameters such as total assets (Log_Assets), growth (Growth), etc. should not vary significantly.

In conclusion, it is proved that after the private-to public transactions, the accrual quality of the same firms increase in the next financial year, which supports my first hypothesis.

Page 27 of 43

Similarly, in order to test H2, conservatism of the UK firms before and after their private-to-public transactions is compared by using Eq. (3). The results are listed in Table 8.

Table 8

, 0 1 , 2 , 3 , , 4 ,

i t i t i t i t i t i t

TA =γ +γ OCF +γ DOCF +γ DOCF ×OCF +γ Public

5DOCFi t, Publici t, 6OCFi t, Publici t,

γ γ

+ × + ×

7DOCFi t, OCFi t, Publici t, i t,

γ ε + × × + Variables −1 vs. 1 −1 vs. 1, 2 −1 vs. 0, 1 −1 vs. 0, 1, 2 OCF −0.22424*** [−3.84963] −0.22424*** [−4.75121] −0.21748 [−0.52468] −0.21748 [−0.62961] DOCF −0.0264 [−0.69005] −0.0264 [−0.85166] −0.02574 [−0.09825] −0.02574 [−0.1179] DOCF * OCF 0.251966*** [4.32384] 0.251966*** [5.33647] 0.245176 [0.59125] 0.24518 [0.70949] Public 0.020876 [0.64724] 0.025485 [1.04873] 0.024693 [0.11695] 0.026512 [0.15665] DOCF * Public −0.02965 [−0.67107] −0.02771 [−0.82558] −0.12367 [−0.42599] −0.0864 [−0.36948] OCF * Public −0.17171*** [−2.82928] −0.1611*** [−3.28381] −0.16634 [−0.38586] −0.15733 [−0.43884] DOCF * OCF * Public 0.144044** [2.37255] 0.133454*** [2.71914] 0.150188 [0.348264] 0.14126 [0.39386] # of firm-years 576 968 893 1285 R2 0.59820 0.57493 0.07519 0.07551

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. T-statistics are given in brackets.

Similar to the comparison of accrual quality in Table 7, the comparison of the conservatism of the same firms of Year -1 (private) to Year 1 (public) results in a

Page 28 of 43

positive (0.14404) and significant (to 5 percent level) coefficient γ₇, indicating that from Year -1 to Year 1, the conservatism of the same firms is increased. After including Year 2 in the comparison, the coefficient γ7 is still positive (0.13345) and

significant at 1 percent level, indicating that comparing to Year -1, firms’ averaged conservatism of Year 1 and Year 2 is also increased. Similar to the results of Table 7, when Year 0 is also included in the comparison of conservatism, however, the coefficient γ₇ becomes insignificant (see the last two columns of Table 8), demonstrating that the conservatism of the same firms is not increased from Year -1 to Year 0.

Therefore, it is proved that after the private-to public transactions, the conditional conservatism of the same firms increase in the next financial year, which supports my second hypothesis.

Page 29 of 43

7

Sensitivity analysis

Given the fact that the samples of this research include a relatively large amount of years and industries, I control the year and industry parameters separately to investigate the year and industry effects on my results. Due to the limited observations of this research, it is impossible to control the both parameters at the same time.

7.1 Sensitivity analysis by controlling the year

Since the firm-year observations used for regressions above are separated in a relatively long range (1996-2013), and since it is generally idea that year is also an influence parameter on the accrual quality and conservatism of firms, it is wise to run the sensitivity analysis by controlling the year. An analysis of the year-distribution of number of observations for [-1 vs. 1] indicates that most of the observations are located in a range of 2004 - 2008 (see Figure 1). Therefore, it is reasonable to control the year to be 2004 - 2008, so that the time period of the observations is not so large (5 years instead of 18) and the number of observations is still large enough for regression.

Conventionally, in other articles on the topic of accrual quality and conservatism, year dummies are set to pick up the year fixed effect (Chi et al. 2011). However, for this research, since I actually compare the accrual quality and conservatism of the same firms between different years, at some point of view I need to detect (instead of fix) the year effect. Therefore, it is not so straightforward to include the year dummies into the models in my research. Since I need to compare the accrual quality and conservatism of the same firms in a range of 3 or 4 years, and I limit the observations in a 5 year period (2004 – 2008), actually I already use the effective “year dummies” in an unobvious way.

Page 30 of 43

Figure 1: the year-distribution of the observations for [-1 vs. 1] regression.

The results of comparison of accrual quality for [-1 vs. 1], [-1 vs. 1, 2], [-1 vs. 0, 1], [-1 vs. 0, 1, 2] and after controlling the year are shown in Table 9. By comparing Table 9 to Table 7, I find that after controlling the year, the results for [-1 vs. 1], [-1 vs. 1, 2] are not changed. However, the coefficients of the public indicator for [-1 vs. 0, 1] and [-1 vs. 0, 1, 2] become significant (at 5% level), indicating that for firms experienced private-to-public transactions within 2004 - 2008, their accrual quality increased immediately during the year of IPO. This finding may also suggest that the accrual quality of firms during the IPO year is increased in general, but since the time range of the observations are too large (from 1996 to 2013), I could not successfully detect it. Generally speaking, the sensitivity analysis by controlling the year still supports my first hypothesis, which claims private to public transaction leads to higher accrual quality.

Table 9

, 0 1 , 2 , 3 , 4 ,

_ _{i t i t} _ _{i t i t} _ _{i t}

AB DA =β +β Public +β Log Assets +β ROE +β Std ROA

+β₅Lev_{i t,} +β₆Growth_{i t,} +β₇Op Cycle_ _{i t,} + +β₈Inv_{i t,} +β₉Loss_{i t,} +ε_{i t,}

Variables [−1 vs. 1] [−1 vs. 1, 2] [−1 vs. 0, 1] [−1 vs. 0, 1, 2] Public indicator −0.04642*** _[−3.60788] −0.04119*** _[−4.02346] −0.03151** _[−2.18176] −0.03998** _[−2.52387] 0 50 100 150 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 _More # of ob se rv at ion s year

Page 31 of 43 Log_Assets _[−0.33212] −0.00244 _[0.86303] 0.00415 _[−0.46152] −0.0035 _[0.60423] 0.004338 ROE 0.000293 [0.37965] 0.00006 [0.11064] 0.00046 [0.48050] 0.000315 [0.33670] Std_ROA 0.00315 [0.71080] −0.00001 [−0.04160] 0.00607 [1.52608] −0.00006 [−0.17166] Lev _[−0.59798] −0.00007 _[−0.65181] −0.00006 _[−0.44543] −0.00007 _[−0.52008] −0.00009 Growth _[0.86304] 0.000158 _[1.10929] 0.000164 _[1.30888] 0.00030 _[1.63529] 0.000412 Op_Cycle _[−0.59392] −0.00000 _[−0.74341] −0.00000 _[−0.00000 _{−1.01274] [−1.02118]} −0.00000 Inv _[−1.16286] −0.07451 _[−1.05967] −0.03871 _[−1.57499] −0.10459 _[−0.20189] −0.01138 LOSS 0.003241 [0.21384] 0.009276 [0.93325] 0.01151 [0.74137] 0.030652** [2.08109] # of firm-years 307 488 488 669 R2 0.05718 0.03967 0.03320 0.02046

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. T-statistics are given in brackets.

The comparison results of conservatism for [-1 vs. 1], [-1 vs. 1, 2], [-1 vs. 0, 1], [-1 vs. 0, 1, 2] after controlling the year are shown in Table 10. Similar to the previous paragraph, the coefficient of the interaction term DOCF_{i t,} ×OCF_{i t,} ×Public_{i t,} (γ ) is ₇ still positive and significant (at 5% level) for [-1 vs. 1], [-1 vs. 1, 2], respectively. For [-1 vs. 0, 1] and [-1 vs. 0, 1, 2], however, γ is changed from insignificant (in Table ₇ 8) to significant, which is consistent with the finding in the previous paragraph. As a result, one can still make the conclusion that the sensitivity analysis by controlling the

Page 32 of 43

year still supports my second hypothesis, namely private to public transaction leads to reporting more conservatively.

Table 10

, 0 1 , 2 , 3 , , 4 ,

i t i t i t i t i t i t

TA =γ +γ OCF +γ DOCF +γ DOCF ×OCF +γ Public

5DOCFi t, Publici t, 6OCFi t, Publici t,

γ γ

+ × + ×

7DOCFi t, OCFi t, Publici t, i t,

γ ε + × × + Variables −1 vs. 1 −1 vs. 1, 2 −1 vs. 0, 1 −1 vs. 0, 1, 2 OCF −0.13832 [−1.30190] −0.13832 [−1.58572] −0.13064 [−1.34720] −0.13064 [−1.54439] DOCF −0.04043 [−0.79614] −0.04043 [−0.96971] −0.0368 [−0.79111] −0.0368 [−0.9069] DOCF * OCF 0.16612* [1.66343] 0.16612** [1.99427] 0.15841* [1.63331] 0.15841* [1.87238] Public 0.028674 [0.66062] 0.035713 [1.04962] 0.04414 [1.15345] 0.045585 [1.39908] DOCF * Public −0.01191 [−0.20864] −0.01105 [−0.24711] −0.02165 [−0.42779] −0.01775 [−0.41186] OCF * Public −0.25822** [−2.39813] −0.25034*** [−2.83374] −0.25227*** [−2.56452] −0.24552*** [−2.86312] DOCF * OCF * Public 0.230477** [2.14020] 0.222617** [2.51956] 0.227318** [2.31056] 0.220581*** [2.57192] # of firm-years 425 668 667 910 R2 0.65544 0.64364 0.59898 0.59206

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. T-statistics are given in brackets.

Page 33 of 43

7.2 Sensitivity analysis by controlling the industry.

According to the Standard Industry Classification Code (SIC code) that is given in the Compustat Global database, my sample firms are from 160 different industry sectors. Therefore, it is of importance to perform the sensitivity analysis by controlling the industry parameter, which is defined as the first two digits of the SIC code (calculated by dividing the SIC code by 100, followed by keeping only the integer of the quotient). An additional analysis of the distribution (over the industry parameter) of number of observations for [-1 vs. 1] indicates that when the industry parameter is equal to 13, 28, 36, 38, 48, 73, or 87, the number of the observations are more than 20. Below, I perform the sensitivity analysis of accrual quality and conservatism by controlling the industry parameter to be 13, 28, 36, 38, 48, 73, and 87 only, compromising a limited number of industry sectors and an enough amount of observations for regression. Here, I also include the industry dummies (for the industry parameters to be 13, 28, 36, 48, 73 and 87, respectively) in the model to control for industry fixed effects. The results for the accrual quality and conservatism are shown in Table 11 and Table 12, respectively.

As shown in Table 11, the results of my sensitivity analysis on accrual quality are unchanged comparing to Table 7: the coefficient of Public indicator, β , is negative ₁ and significant for [-1 vs. 1] and [-1 vs. 1, 2], respectively. However, after including Year 0 in the comparison, namely for [-1 vs. 0, 1] and [-1 vs. 0, 1, 2], β becomes ₁ insignificant, indicating that the industry parameter has no obvious effect on the increase of accrual quality after the private-to-public transactions.

Page 34 of 43 Table 11

, 0 1 , 2 , 3 , 4 ,

_ _{i t i t} _ _{i t i t} _ _{i t}

AB DA =β +β Public +β Log Assets +β ROE +β Std ROA

5Levi t, 6Growthi t, 7Op Cycle_ i t, 8Invi t, 9Lossi t, i t,

β β β β β ε + + + + + + + Variables [−1 vs. 1] [−1 vs. 1, 2] [−1 vs. 0, 1] [−1 vs. 0, 1, 2] Public indicator −0.02502*** _[−2.72533] −0.01917** _{[−2.26519] [−0.62255]} −0.0294 _[−0.82480] −0.03187 Log_Assets −0.00242 [−0.49283] 0.00155 [0.40079] 0.04955** [2.12854] 0.03247* [1.93436] ROE _[−0.43667] −0.00022 _[−1.19702] −0.00051 _[−0.50042] −0.00142 _[−0.38447] −0.0008 Std_ROA _[0.31206] 0.000235 _[−0.14187] −0.00006 0.01560*** _[3.82295] 0.00491** _[2.49303] Lev _[−0.64483] −0.00005 _[−0.84466] −0.00006 _[−0.00046 _{−1.03003] [−1.06522]} −0.00039 Growth _[0.23684] 0.00002 _[0.49356] 0.00004 _[0.21108] 0.0001 _[0.64307] 0.00026 Op_Cycle _[−0.77837] −0.00000 _[−1.01783] −0.00000 _[−0.09776] −0.00000 0.00000 [0.10630] Inv _[−0.91264] −0.05349 _[−1.14486] −0.04572 _[−0.34637] −0.10128 _[−0.34439] −0.06297 LOSS 0.01123 [1.04920] 0.01494* [1.83441] 0.08161* [1.66994] 0.06263* [1.80155] Industry

dummies Yes Yes Yes Yes

# of firm-years 256 437 401 582

R2 0.06510 0.03269 0.09137 0.02720

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. T-statistics are given in brackets.

Page 35 of 43

The results of my sensitivity analysis on conservatism by controlling the industry parameter are shown in Table 12. Again, comparing to Table 8: the coefficient of of the interaction term DOCF_{i t,} ×OCF_{i t,} ×Public_{i t,} (γ ) is still positive and significant ₇ for [-1 vs. 1], [-1 vs. 1, 2], and is insignificant for [-1 vs. 0, 1, 2] (results unchanged). Surprisingly, γ is still positive and significant for [-1 vs. 0, 1]. This may due to the ₇ fact that by controlling the industry parameter, the year parameter is unavoidably controlled – for some industry sectors firms may trend to go public simultaneously in the range of 2004 – 2008.

Table 12

, 0 1 , 2 , 3 , , 4 ,

i t i t i t i t i t i t

TA =γ +γ OCF +γ DOCF +γ DOCF ×OCF +γ Public

5DOCFi t, Publici t, 6OCFi t, Publici t,

γ γ

+ × + ×

7DOCFi t, OCFi t, Publici t, i t,

γ ε + × × + Variables [−1 vs. 1] [−1 vs. 1, 2] [−1 vs. 0, 1] [−1 vs. 0, 1, 2] OCF −0.22603*** [−4.35275] −0.22053*** [−5.53155] −0.19802*** [−3.72044] −0.2296 [−0.62746] DOCF −0.02891 [−0.70077] −0.02887 [−0.90417] −0.03564 [−0.86577] −0.0556 [−0.19552] DOCF * OCF 0.23046*** [4.43571] 0.22524*** [5.64619] 0.20258*** [3.80395] 0.23549 [0.64312] Public 0.03375 [0.94803] 0.03883 [1.51954] 0.03536 [1.03391] 0.04459 [0.19670] DOCF * Public −0.02457 [−0.50340] −0.02301 [−0.66235] −0.03416 [−0.73854] 0.02916 [0.09557] OCF * Public −0.16883*** [−3.12388] −0.17377*** [−4.19453] −0.19148*** [−3.46529] −0.16842 [−0.44379] DOCF * OCF * Public 0.17442*** [3.22200] 0.17927*** [4.31977] 0.18767*** [3.39464] 0.26496 [0.69775] Industry

Page 36 of 43

# of firm-years 314 591 478 783

R2 0.73052 0.70118 0.62090 0.42087

*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. T-statistics are given in brackets.

Since it has been proved that audit choice is also an influence parameter for the accrual quality and conservatism, it would be good if I could also control the auditor parameter as well. Several kinds of auditor information are available in the Compustat North America, such as AU – Auditor (indicating the auditing firms that sample companies use) and AUOP -- Auditor Opinion, but none of this information is available in the Compustat Global database. Therefore it is impossible to control the Auditor as a parameter in this research.

Page 37 of 43

8

Conclusion and Limitation

Unlike most of prior researches, which always focused on comparing the financial reporting quality (FRQ) between different public and private firms, this study applies a novel method to compare the level of FRQ of the same firms but between the privately-hold and publicly-hold periods, which to the author’s knowledge has never been investigated before. The accrual quality and conservatism (as proxies for the FRQ) of the same firms within ± 2 years of their ownership transactions have been systematically compared. The results indicate that both the accrual quality and the conservatism of firms are higher during the publicly-hold period than the privately-hold period, which agrees with the other recent researches that the accrual quality and the conservatism of public firms are higher than those of (other) private firms. Since the work of self-comparison of FRQ has never been investigated before, the author believes that this research makes an initial but critical contribution to the research of the financial reporting quality.

This study is subject to some unavoidable limitations due to the insufficient information of selected database Compustat, a very common database that are used by many researchers for the financial reporting quality research. The initial proposal was to separate the data samples into two groups: a), the firms were firstly privately-owned and experienced a private-to-public transactions to be public later; b), the firms were at first public and experienced a public-to-private transactions to be private later. In this way, the FRQ before and after both the public-to-private and the private-to-public transactions could have been compared. However, due to the fact that the information of public-to-private transactions are only contained in the database of the Centre for Management Buyout Research (CMBOR), which is inaccessible by our university, the FRQ of the same firms before and after the public-to-private transactions cannot be investigated. Therefore, further research is suggested if the access to the CMBOR is guaranteed. Furthermore, even for the private-to-public transactions, the research samples for UK firms are very small – for

Page 38 of 43

a period of 20 years the useful firm-year observations are only around 3000, an order of magnitude smaller than the other researches. Last but not least, it is found that the Big 4 information for UK firms is not available in the Compustat, which limited the control of auditor choice in the models for regression. Therefore, it is suggested to extend this research by using other database which contains more observations for UK firms and full knowledge of auditor information. Also, same investigations in other locations would be very interesting.

Page 39 of 43

Reference

Ashbaugh-Skaife, H., D. Collins, W. Kinney, and R. LaFond. 2008. The effect of SOX internal control deficiencies on firm risk and cost of equity. Journal of Accounting Research 47: 1–43.

Balasubramaniam, K. (2009). What are the advantages and disadvantages for a

company going public?

http://www.investopedia.com/ask/answers/06/ipoadvantagedisadvantage.asp

Ball, R., and L. Shivakumar. 2005. Earnings quality in U.K. private firms: Comparative loss recognition timeliness. Journal of Accounting and Economics 39 (1): 83–128.

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

Brennan, M. J., & Franks, J. (1997). Underpricing, ownership and control in initial public offerings of equity securities in the UK. Journal of Financial Economics, 45(3), 391-413.

Burgstahler, D. C., Hail, L., & Leuz, C. (2006). The importance of reporting incentives: earnings management in European private and public firms. The accounting review, 81(5), 983-1

Business Dictionary, Why Does a Company Decide to Go Public?

http://www.businessdictionary.com/article/780/why-does-a-company-decide-to-go-pu blic/

Jeffrey Steele, (2009). Going Private.

http://www.mbbp.com/resources/business/private-company.html

Ball, R., and L. Shivakumar. 2006. The role of accruals in asymmetrically timely gain and loss recognition. Journal of Accounting Research 44 (2): 207–242.

Beuselinck, C., & Manigart, S. (2007). Financial reporting quality in private equity backed companies: The impact of ownership concentration. Small Business

Economics, 29(3), 261-274.

Burgstahler, D., and I. D. Dichev. 1997. Earnings Management to Avoid Earnings Decreases and Losses. Journal of Accounting & Economics 24: 99-126.

Burgstahler, D., L. Hail, and C. Leuz. 2006. The importance of reporting incentives: Earnings management in European private and public firms. The Accounting Review

Page 40 of 43 81 (5): 983–1016.

Chen, F., O.-K. Hope, Q. Li, and X. Wang. 2011. Financial reporting quality and investment efficiency of private firms in emerging markets. The Accounting Review 86 (4): 1255–1288.

Chi, W., Lisic, L. L., & Pevzner, M. (2011). Is enhanced audit quality associated with greater real earnings management?. Accounting Horizons, 25(2), 315-335.

Dechow, P., R. Sloan, and A. Sweeney. 1996. Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research 13: 1–36.

Dechow, P., and I. Dichev. 2002. The quality of accruals and earnings: The role of accrual estimation errors. The Accounting Review 77 (Supplement): 35–59.

Dechow, P., W. Ge, C. Larson, and R. Sloan. 2011. Predicting material accounting misstatements. Contemporary Accounting Research 28: 17–82.

Degeorge, F., J. Patel, and R. Zeckhauser. 1999. Earnings Management to Exceed Thresholds. Journal of Business 72: 1-33.

Francis, J., R. LaFond, P. Olsson, and K. Schipper. 2005. The market pricing of accrual quality. Journal of Accounting and Economics 39 (2): 295–327.

Givoly, D., C. Hayn, and S. P. Katz. 2010. Does public ownership of equity improve earnings quality? The Accounting Review 85 (1): 195–225.

Hope, O.-K., J. C. Langli, and W. B. Thomas. 2012. Agency conflicts and auditing in private firms. Accounting, Organizations, and Society 37 (7): 500–517.

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

Hribar, P., and C. Nichols. 2007. The use of unsigned earnings quality measures in tests of earnings management. Journal of Accounting Research 45 (5): 1017–1053. Kanagaretnam, K., G. Lobo, and D. Yang. 2004. Joint tests of signaling and income smoothing through bank loan loss provisions. Contemporary Accounting Research 21: 843–884.

Kothari, S. P., A. J. Leone, and C. E. Wasley. 2005. Performance matched discretionary accrual measures. Journal of Accounting and Economics 39 (1): 163–197.