The impact of institutional ownership on earnings quality : evidence from American listed companies in financial crisis 2008

(1)

The impact of institutional ownership

on earnings quality: evidence from

American listed companies in

financial crisis 2008

Name: Jin Luo Student number: 10602704

Version: draft Date:22/06/2014

(2)

Abstract

This study tests whether the institutional ownership has an impact on earnings quality of American listed companies in financial crisis 2008. Specifically, I question if the institutional investors will actively monitor the companies to report more transparent and fair financial statements. Based on my empirical research, I find that the firms with higher institutional ownership will present less discretionary accruals and less real activities manipulation in their financial information, thus improving their earnings quality in financial crisis. My findings are effective due to controlling different factors affecting institutional ownership and earnings quality, including various straightforward proxies to measure earnings quality, doing additional analysis with alternative measurement.

As a summary, my results are consistent with active monitoring hypothesis.

Keywords: institutional ownership; earnings quality; discretionary accruals; real

(3)

Table of Content

1 Introduction ··· 4

1.1 Background ··· 4

1.2 Research question ··· 6

2 Literature review and hypotheses ··· 8

2.1 Importance of institutional investors in corporate governance ··· 8

2.2 Agency problems to earnings quality ··· 9

2.3 Active monitoring hypothesis ··· 9

2.4 Private benefits hypothesis ··· 10

2.5 Earnings quality ··· 11

2.6 America stock market condition ··· 12

2.7 My hypothesis ··· 13

2.7.1 long-term profitability ··· 13

2.7.2 earnings quality usefulness ··· 14

2.7.3 the United States stock market needs ··· 14

3 Research design ··· 15

3.1 Data and sample selection ··· 15

3.2 Measurement of earnings quality and other variables ··· 17

3.2.1 Discretionary accruals or neutrality··· 17

3.2.2 Real activities manipulation or representational faithfulness ··· 18

3.3 Empirical models ··· 20 3.4 Control variables ··· 23 4 Results ··· 24 4.1 Descriptive statistics ··· 24 4.2 Regression results ··· 29 4.3 Additional analysis ··· 35

4.3.1 Analysis of an alternative measure of earnings quality ··· 35

(4)

1 Introduction

1.1 Background

As an essential theoretical concept, earnings quality has been a key point for financial

information users all the time. Dechow et al 2010 deﬁnes earnings quality from SFAC

NO.1 as follows: Higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker. Broadly to say, earnings quality should be decision usefulness. They reach no single conclusion on what earnings quality is, but provide a guide for us to understand and measure earnings quality better. They point out that quality of earnings is a function of a company’s fundamental performance. Statement of financial accounting concepts No.2 (FASB,1995) also discusses high earnings quality that makes information useful.

Regarding to institutional ownership, Bushee (1998) defines institutional investors as large investors such as bank trusts, insurance companies, mutual funds and pension funds that invest on behalf of others and manage at least $100 million in equity. They think that the presence of institutional investors may cause a change in the behavior of the firms in which they invest through their monitoring activities. To date,institutional investors own an ever increasing proportion of the outstanding shares of listed firms. They have the resources, expertise, and increasingly the power to effectively monitor the actions of management and to influence the decisions of firms. Daily et al (2003) also finds that institutional investors’ participation has emerged as an important force in corporate monitoring, serving a mechanism to protect minority shareholders’ interests.

The significant increase in the institutional ownership has led to a significant and powerful role in corporate governance. For example, in the UK, institutional investors own between 65 to 75 percent of the UK stock market, which suggests the prominent role that institutional shareholders can play as an agent for governance systems

(5)

(Mallin, 2003). In Australia, institutional ownership grew to about 49 percent of the listed equities by 1997 causing the Parliamentary Joint Committee to remove some legal barriers in order for institutional investors to be more involved in the corporate governance of their portfolio firms (Koh, 2003). In Asia, Chinese government also highlights the leading role of quickly increased institutional investors in promoting governance practices to ensure management priority in the best interests of the company at all times.

Most of academic researches focus on the effect of institutional ownership only to discretionary accruals in earnings management. Chung et al (2001) find that the presence of large institutional shareholdings inhibits managers from increasing or decreasing reported profits towards the managers’ desired level or range of profits. Michael and Maria(2011) also get some relationship between institutional ownership and earnings management which directly effects the earnings quality. Their result shows that monitoring by the largest institutional owners is negatively related to earnings management. These evidence are consistent with institutional investors monitoring and constraining the self-serving behavior of corporate managers. Clearly, decreased earnings management shoud improve the quality of earnings. Gerald and Jian (2001) suggest a significantly negative relationship between earnings management and disclosure quality and firms engage more in earnings management tend to have lower earnings quality.

However, fewer researchers directly test the effect of institutional ownership to earnings quality. Velury and Jenkins (2006) provide insights into the monitoring role of institutional investors by directly examining whether institutional ownership affects the quality of reported earnings. The results demonstrate significant evidence of a positive association between institutional ownership and earnings quality. In addition, the results indicate that concentrated institutional ownership may negatively affect earnings quality. Jiambalvo et al (2002) demonstrate that, on average, stock prices are more likely to reflect future earnings when institutional ownership is high. It partly reflects the positive association between earnings quality and investors monitoring role.

(6)

The financial crisis in 2007 and 2008, also recognized as the Global Financial Crisis is considered by a lot of economists the worst financial crisis since the Great Depression in the 1930s. In the United States, it brought the threat of complete collapse to large financial institutions, the emergent help to investment banks by national governments, followed by depress in stock markets around the world. Later the Financial Crisis Inquiry Commission argued that the cause of the financial crisis was national failure in financial regulation and supervision, lack of corporate governance and risk management in many important financial institutions, a combination of lack of transparency, excessive borrowings and risky investments from financial institutions. Furthermore, since the 1970s, the American Government policy has emphasized less regulation to encourage business, which brought less disclosure of information, less oversight of new investment activities undertaken by financial institutions and other evolving investment banks. Eventually, the American stock market experienced a pronounced decline which accelerated significantly in October 2008 while it peaked in October 2007 with the Dow Jones index exceeding 14,000 points. By April 2009, the Dow Jones average reached the lowest level around 6,600. During this period, those American institutional investors suffered a lot and until March 2011, about 165 failed institutional investors have received $9 billion to cover losses on unsuccessful investments from FDIC. As a whole, both the American institutional investors and listed companies experienced a great loss.

1.2 Research question

Based on the evidence from American listed companies in financial crisis 2008, does the institutional ownership have any impact on the earnings quality?

I want to test the direct relationship between institutional ownership and earnings quality of listed companies in the United states stock market. It can be separated into two questions, whether there exists a relationship between them and if so, how the institutional ownership effects the earnings quality. Especially in the most developed country America, I wonder if in the serious financial crisis 2008, the institutional investors would still keep a monitoring role in corporate governance, thus improving

(7)

earnings quality.

1.3 Motivation of my study

In the background, some literature notices the key role of institutional investors in influencing companies’ financial statements. Most of them pay attention to relationship between institutional ownership and discretionary accruals in earnings management. Further, their results reflect some clearly associations and earnings management eventually effects the quality of earnings. Earnings management has a lot in common with earnings quality. I think most would agree that highly managed earnings have low quality. However, the lack of discretionary accruals is not sufficient to guarantee high-quality earnings, because other factors also contribute to the quality of earnings. So analyzing the influence of institutional investors to final quality of earnings through discretionary accruals’ is indirect and only represents one dimension to evaluate. Now I plan to directly investigate the effect of institutional ownership to earnings quality. In this way, it can show a clear and direct image of institutional ownership and final earnings quality, thus helping outside information users to evaluate the impact conveniently and make useful decision.

Nowadays, most of evidence shows that in the both developed and developing countries the increased institutional investors can play a more important role in companies by monitoring the performance of management and limiting their opportunistic behaviors and manipulating their financial statements. Especially in the global leader America stock market, the power of institutional investors over earnings quality was unignorable and positive in most cases before the financial crisis. However, some evidence also reflects that some institutional investors would chase their own interests, thus impairing earnings quality. Concurrently, when the whole stock market and investment institutions faced a huge impact in financial crisis, i am curious about some factors which would change the role of institutional shareholders in normal period. Only limited literature researched the relationship during the sensitive period all over the world. To some extent, our study contributes to the literature by extending prior research into the effect of institutional ownership on

(8)

earnings quality in a special period.

Clearly, earnings quality directly decides the decision usefulness of outside investors which is the main objective of financial reporting. In the prior literature, many academic researchers emphasize the importance of earnings quality but usually only use discretionary accruals as an indicator for it. Actually, using the FASB conceptual framework as a basis, we can get various measures including persistence, accruals, smoothness, timeliness, loss avoidance, investor responsiveness, and external indicators such as restatements. Then I will use two most straightforward and representative measures.

2 Literature review and hypotheses

2.1 Importance of institutional investors in corporate governance

Some worldwide code emphasizes the importance of the institutional investors in corporate governance to protect minority shareholders interests. Various statements of principles, guides and codes (e.g. ICGN Statement of Principles on Institutional Shareholder Responsibilities 2007, UK Stewardship Code 2010) have been issued internationally in providing guidance to institutional investors in exercising their role. No matter what kind of political culture and business systems in western countries or eastern countries, there is always demand from investors worldwide for well-governed companies and investors willing to pay a high premium for them (McKinsey and Company, 2002).

On the other hand, some prior literature also emphasize the role of institutional investors to corporate governance. Claessen and Fan (2002) finds that the involvement of institutional investors may improve corporate governance practices by mitigating the problems associated with conflict between the controlling owners and minority shareholders in firms. Then good governance would bring enhancement of earnings quality. Gillan and Starks (2003) tested the role of institutional investors in different markets and the corporate governance of firms. They thought because of the

(9)

growth in institutional influence and ownership, institutional investors have played an important role in many markets worldwide. At the same time, since the institutional shareholders provided more information, it can bring reasonable corporate governance structure and effective monitoring to corporations. Xu and Wang (1999) searched evidence in china, concluding that the corporate governance performance of Chinese listed companies in shanghai stock market was significantly affected by institutional ownership.

Undoubtedly, corporate governance of high quality can bring the overall performance of a firm to a new level. As an important factor of financial performance, earnings quality would become better through institutional shareholders’ growing role in corporate governance framework.

2.2 Agency problems to earnings quality

As we all know, agency problem extensively exists in every company. Because the aim of owners and managers might not be the same all the time and the owners can hardly monitor the managers with asymmetric information, the interests of all shareholders would be impaired. The conflicts of interests originally come from the separation of ownership and management. When the managers own the private information on their effort, they might maximize their own benefits regardless of the companies’ interests. Once the bonus or compensation is related to earnings, the manager might manipulate the earnings through accruals or real activities to beat the benchmark, reducing the quality of earnings.

2.3 Active monitoring hypothesis

There are two contrast points about the institutional investors’ motivation. The active monitoring hypothesis thinks that institutional investors are willing to take an active part in the management and monitoring of investment because of the large amount of wealth they have invested and their great power and resource. A large numbers of prior studies have found evidence consistent with this hypothesis. In this case, prior literature can get an clearly positive relation between ownership and quality of

(10)

reported earnings. On one hand, they are regarded as the more intelligent investors than other common shareholders, who have advantages in inquiring and analyzing financial information, so they are able to manage invested firms to pursue long term equity benefits (Jiambalvo et al 2002). Hadani et al., 2011 suggests that those institutional investors could possibly monitor management behavior to improve earnings quality because of their position in the firm while it is also possible for them to order the managers to manage earnings for the sake of their own benefits. Karamanou and Vafeas, 2005 argues that institutional ownership serves as an important complement in disciplining management while Brous and Kini (1994) finds institutional ownership improves firm profitability and stock price performance. Jung and Kwon (2002) tests the association between corporate ownership and earnings informativeness. Their findings that institutional ownership makes the informativeness grow support this hypothesis. On the other hand, the institutional shareholders are recognized as long-term investors who are willing to monitor managers and themselves’ role for the whole company’s interests and development in a long horizon. Hadani et al. (2011) also finds that largest institutional owners constrain self-serving manipulations of accounting numbers. Koh (2007) finds that long-term institutional investors would reduce significant earnings management in the invested firms after it separates institutional investors to short-term or long-term by the investment horizons.

2.4 Private benefits hypothesis

Under the private benefits hypothesis, there is always a negative relationship between earnings quality and institutional ownership in empirical evidence. The large institutional investors are believed to pursue their own benefits by exploiting access to private information, which can be used for trading purposes. Leuz et al (2003) finds that in order to maintain the institutional investors’ private control advantage, they might manipulate financial performance from other shareholders by making use of earnings management. Then they give advice to set an internal link between earnings quality and corporate governance, avoiding the adverse impact. Velury and Jenkins

(11)

(2006) indicates that based on private trading motivation, the more concentrated the institutional ownership is, the more negative impact it have on earnings quality. Some researchers also indicate institutional investors and managers both enjoy the benefits of cooperation and alliance, resulting in less monitor and lower quality.

Some researchers also come up with an reason of short-term opportunistic investment of institutional shareholers. Koh 2003 regards institutional investors as short-term oriented shareholders who use their lager investment to chase self-interests through access to private information. Thus they will impair earnings quality and minority interests or individual investors. According to Porter 1992, institutional investors overly focus on the short-term earnings at the expense of long term equity value. If it stands, they would lose incentives to monitor management but would get private information for trading. Bushee (2001) gets evidence proving that large amount of abnormal returns always are connected to high level of short term institutional ownership with opportunistic trading purpose. Their findings would generate more concerns to opposite impact on earnings quality brought by those institutional investors focusing on quick trading benefits.

2.5 Earnings quality

There is large amount of prior academic research mentioning the importance of earnings quality to the interests of whole company and all the shareholders. Chan et al(2001) gets evidence supporting that the financial information of higher earnings quality will predict more accurate future stock returns. Actually, they observes a negative relation between the accrual( represent quality) and stock returns in the coming years. Bricker et al (1995) ‘s result shows a perspective of financial analyst: earnings of high quality would reduce the unstable earnings, avoid manager’s deliberate assertion in accounting practice, give a transparent description of allowance and off-balance-sheet items. Besides the above papers emphasize earnings quality’s importance to investors decision making, some researchers present its function in terms of other aspects. Dichev et al (2012) gives us a profound understanding of earnings quality by interviewing 169 listed companies’ CFO as well as several

(12)

standard setters. They regard high earnings quality as more sustainability in income, more consistency in accounting choices and less one time items. Briefly and accurately, they define earnings quality as a simple financial item setting a connection with inside decision making and outsider shareholders interaction.

To better understand earnings quality, Dechow et al 2010 deﬁne earnings quality

broadly to be decision usefulness and review the proxies, their determinants and their consequences. They reach no single conclusion on which is earnings quality’s best proxy, but provide us various methods and their advantages. Then we can choose from persistence, accruals, smoothness, timeliness, loss avoidance, investor responsiveness, and external indicators such as restatements and SEC enforcement releases to measure earnings quality. The Velury and Jenkins (2006) also discusses the components of earnings quality as being: predictive value, feedback value, neutrality, representational faithfulness, verifiability and timeliness. However, empirical researchers have

typically measured only discretionary accruals in theirstudy and ignore the remaining

dimensions.

2.6 America stock market condition

As the most developed and mature stock market in the world, earnings quality, corporate governance in American listed companies always remained a relatively good performance. However, in December 2001, the largest American energy provider Enron manipulated earnings to cover debt, then its bankrupt deeply hurt investors’ confidence. It was followed by shareholders’ skepticism to earnings quality (reliability and faithfulness) and extensive discussion on financial information. Later in 2002, a series of accounting scandals in big companies like Worldcom gave birth to the sarbanes–oxley act (SOX). The Securities and Exchange Commission (SEC) forced it into practice for guaranteeing earnings quality and rebuilding investors’ confidence. Chang et al (2002) first examines the effect of SOX on United States market and it shows the cost of capital declined while the trust to accounting grew. As we discussed in the above literature, there is still mixed relationship between institutional investors and earnings quality. Even though SOX pushes earnings quality

(13)

to a higher level, whether institutional investors will execute the monitoring role depends on specific condition.

When the financial crisis started, the interaction between institutional ownership, corporate governance and earnings quality would change. Erkens et al (2012) investigates the US firms’ financial performance in the financial crisis 2008 and observes firms with higher institutional ownership would suffer more stock loss because of their more aggressive risk appetite prior to the crisis.

2.7 My hypothesis

Based on the two opposed views, I still prefer to set the hypothesis supporting the active monitoring hypothesis because of a large number of empirical evidence in other countries. I assume that the institutional investors still have different kinds of motivation to improve quality of financial reporting especially in the largest and most mature stock market in the United States.

2.7.1 long-term profitability

Institutional investors do own the opportunity, resources, and ability to monitor, discipline, and influence managers of firms (Monks and Minow, 1995). Whether institutions use these powers is a part of their function, which origins from the size of their individual or total shares percentage. If their shareholdings are high, they are going to be less marketable (Maug, 1998) and probably held the shares for longer periods of time. In this circumstance, institution shareholders have more reason or greater incentives to obtain useful information, monitor management actions, and force better performance. In support of this, McConnell and Servaes (1990) reports a significant relationship in statistic results between the value of a firm and the percentage of share ownership held by institutional investors. When shareholdings are held for the benefits in the long term, institutions will be more concerned about the possible profitability of the companies and will become more responsible for the earnings quality, and hence monitoring managers' performance. Due to the direct impact on their own long term interests, Institutional investors want firm managers to

(14)

focus on future profitability rather than be pre-occupied with managing earnings on a year-by-year basis.

2.7.2 earnings quality usefulness

As we all know, undoubtedly financial statements remain an important source of information on the firm all the time. In order to plan and evaluate their investments, institutional investors must show great interests in using all kinds of information relevant and reliable to value. Assuming that institutional investors only pay attention to non-financial information instead of using and analyzing earnings information, then it would indicate that prices already reflect relevant information or earnings information has nothing to do with value. Actually, since perfect efficient market does not exist in practice and empirical evidence continues to document a fast market reaction to earnings announcements, we have reason to believe that earnings provide useful information in terms of a security pricing aspect. Further, institutional shareholders ( also known as more sophisticated investors) own better analysis ability to use the earnings information than common individual shareholders. In this case, institutional investors have the incentive and power to get earnings information of high quality.

2.7.3 the United States stock market needs

Before the crisis, the United States government put some new regulations including SOX act into public, hoping to improve the ownership structure and ensure the reliability and relevance of earnings information through encouraging the institutional investors’ participation as well as monitoring to corporate governance. In the center of financial crisis 2008, academic researchers, firms and authorities all focus more on the institutional ownership because they are the main player of the depression. Due to their high risk policy and high shareholdings prior to the financial crisis, they suffer the most and some global leading institutions even bankrupt. After the huge loss, they might reduce their investment scale and become more prudent about risk choice. The outside customers, suppliers, investors must lose their confidence in the suffered

(15)

market and would only cooperate with those corporations of really high earnings quality and better potential. In order to regain stock market confidence and minimize the loss of recession, I assume that institutional shareholders should try their best to show the customers or potential investors their active monitor to corporate governance. What is more, large amount of literature in the period from SOX to financial crisis provide evidence supporting the positive function of institutional shareholders in American stock market. It suggests that the fast development of institutional investors is beneficial for both the corporate governance system and the information environment in capital market. So we can clearly judge the needs of US stock market for institutional investors to monitor and improve earnings quality, especially in the crisis.

Considering other reasons mentioned in the active monitoring hypothesis, I believe institutional investors would have a positive effect on the earnings quality in American market.

Hypothesis: In the financial crisis 2008, for the United States listed companies, there is a positive relationship between institutional ownership and earnings quality.

3 Research design

3.1 Data and sample selection

In my thesis, i plan to investigate the relationship between institutional investors’ ownership which is my independent variable and earnings quality which is my dependent variable. In order to only investigate the relationship in financial crisis and ignore the following years’ impact, I would observe 2008 fiscal year. As the most representative companies for American stock market and main players in crisis, all the institutional investors of listed firms in S&P 1500( datastream series ‘LS&P15XX’) would be my choice, making the findings more effective and convincing. Regarding to the ;institutional ownership, many prior literature Velury U. & Jenkins D.S. (2006),

(16)

Chung et al (2002)downloaded it from compact disclosure database( disclosed by SEC in WRDS). Due to limited access to it, I download the percentage of shares owned by institutional investment or banks and ownership of pension funds from datastream, then add them together to replace the independent variable. Because in most cases, the annual reports show that the US institution shareholders consist of investment banks and pension funds. Since predicting earnings quality requires the data in prior year and the year before prior year, I get more financial data in 2007 and 2006 which i will explain in real earnings activities and discretionary accruals model . For one control variable corporate governance score, i obtain it from asset4-esg database. This database provides access to in-depth, objective and comparable esg data( environmental, social, governance) on over 3400 global companies. Furthermore, it uses different reliable channels such as financial statements, annual reports, academic articles and surveys to gather related corporate governance information. Several literature has used the corporate governance score in research and speak highly of the reliability and objectivity of the score which provides a standard evaluation to the overall corporate governance performance. Then I download all the other financial data from datastream which maintains the most widespread and useful financial information for all the American as well as global companies.

Originally, my initial sample consists of 1502 American listed companies on datastream and asset 4 database. Later according to some literature( Davidson et al, 2005, Peasnell et al, 2005, Jiambalvo et al 2002), all the financial industry firms such as insurance, commercial or investment banks, pension funds and unit trust companies might conduct completely different operation which cannot be represented by our discretionary accruals model and are monitored by the more strict regulations. Furthermore, these firms always act as institutional investors in other companies. I exclude them by using sic code ( financial institution 6000-6999). After deleting the observations including missing value for each variable, significant outliers, duplicate firm-year observations and negative total assets, there are 334 companies left. Most of the missing data come from corporate governance score, selling, general& advertising expenses. However, the role of corporate governance is a key factor to the association

(17)

of institutional investors and earnings quality, even though considering it would reduce my sample size a lot. Then i drop some errors when i calculate natural logarithm of market value of equity, then i get 335 listed companies. However, in the 335 companies, i find that there are 59 companies’ institutional ownership showing zero. It might origin from my limited access to compact D database. Finally, i get 276 listed companies.

3.2 Measurement of earnings quality and other variables

3.2.1 Discretionary accruals or neutrality

As the most popular measurement of earning management and quality, discretionary accrual is widely used and recognized in different kinds of prior research Kothari et al.2005, Jones 1991, Defond and Subramanyam 1998 and subramanyam 1996. Concurrently, it best represents the neutrality element of earnings quality which is defined by SFAC No.2 ‘the absence in reported information of bias intended to attain a predetermined result or to induce a particular mode of behavior’. Discretionary accruals are then modeled as a function of managers’ propensity to manage earnings and institutional investors’ actions to encourage or deter such earnings management. Managers usually use discretionary accruals to move underlying profits towards some desired level of earnings. With no doubt, this kind of behavior must influence the quality of earnings.

Similarly, i plan to use a cross-sectional version of the modified jones model in Defond and Subramanyam 1998 because of its less limits to data downloading and the superior specification. As the Warfield at al. 1995 and Klein 2002 said, the different direction of income accruals (more or less)can be chosen by management. So in my major model, I am going to use the absolute value of discretionary accruals. Using in the same way in Cohen et al (2008), my model estimating discretionary accruals looks as follows:

(18)

In the formula, TAit means the total accruals in firm year it which is the difference between net income before extraordinary items and net cash flows of operating activities. Assetsit-1 means the total assets in the prior year while the ∆REV means the

change in total revenues from the prior year. Concurrently, the PPEit represents the

gross value of property, plant and equipment. Then i use this model for running regression to get the coefficients, which i will use in the second formula as follows:

With the same coefficients, i can calculate the NAit normal accruals in firm year it.

Here the ∆ARit means the change in accounts receivables from prior year. Then i can

easily obtain the discretionary accruals from the difference between the total accruals and normal accruals. Once the results are the same as my active monitoring hypothesis(private benefits hypothesis), there will be a negative (positive) relationship between discretionary accruals and institutional ownership.

3.2.2 Real activities manipulation or representational faithfulness

Roychowdhury 2006 defines the real activities manipulation that in order to beat or meet specific earnings benchmarks, managers perform differently from normal business practice. Some existed evidence suggest that real activities manipulation is always connected to a significant decrease in following financial performance. As the important element for investors to evaluate the performance, earnings quality must be impaired. SFAC No.2 says representational faithfulness is the correspondence or agreement between a measure or description and the phenomenon that it purports to represent. When real activities deviate from normal practice, the agreement cannot be maintained.

Roychowdhury 2006, Zang 2012, Cohen and Zarowin 2010, Cohen et al.2008, Badertscher 2011 all show us the same proxy measuring it. There are four proxies to measure real activities manipulation: (1) abnormal levels of operating cash flows (AB_CFO), (2) abnormal production costs (AB_PROD), (3) abnormal discretionary

(19)

expenses (AB_EXP), and (4) a combined measure of real activities manipulation. For estimating each of the first three proxies, the formulas are as follows:

The CFOt means the net cash flows from operating activities in year t while the At-1

means the total assets in prior year to t. St represents the net sales and ∆S means the

change in net sales from prior year t-1. This model is the same as the one in Roychowdhury 2006 to measure the normal cash flows from operating activities. According to many existed academic evidence, managers might manipulate real activities like longer credit time to attain more sales targets, which result in less operating cash flows compared to net sales in current period.

The COGSt means the costs of goods sold in year t.

The ∆INVt means the changes in total inventory in year t while the ∆S t-1means the

sales change in prior year. According to the prior literature Roychowdhury 2006, Zang 2012, Cohen and Zarowin 2010, Cohen et al.2008, Badertscher 2011, they think the costs of production consist of costs of goods sold and change in total inventory in current year. Because the managers want to reduce the production costs, they would expand amount of production. Then the more extra production costs might cause more production costs compared to current sales. After combing the two elements, i estimate the below formula:

The PRODt represents the normal level of production costs.

The DISEXPt means the discretionary expenses in year t which is calculated by adding

R&D, Advertising, and SG&A expenses. When managers plan to decrease the discretionary expenses compared to the net sales, they might change the advertising

(20)

expenses, general expenses and R&D expenses.

Since the abnormal part is the difference between the expected normal amount and actual amount. Firstly, i use all the variable in the above formulas to run regression and get each coefficient for each variable. Secondly, i insert the actual amount of each variable in every firm to the models, calculating the actual amount. Thirdly, deducting the normal part from actual amount will give us the three real earings proxies.

According to my analysis above and the theory in Cohen et al.2008, the more real activities manipulation will bring less abnormal operating cash flows, discretionary expenses and more abnormal production costs. Then I will estimate the combined amount of real activities manipulation as the sum of AB_CFO ,AB_PROD, and AB_EXP. Due to different association, COMBINED_RAM=AB_CFO- AB_PROD+AB_EXP. If the results can prove my active monitoring hypothesis (private benefits hypothesis), there will be a positive (negative) relationship between institutional ownership and COMBINED_RAM.

3.3 Empirical models

After i download all the necessary initial data and modify them, i calculate them to get all the variables ready for regression. To examine the direct relationship between institutional ownership and earnings quality of listed companies in the United states stock market in financial crisis 2008, i develop the following models to test the most straightforward, representative measures of earnings quality:

t t 10 t 9 t 8 1 t 7 t 6 1 t 5 1 t 4 1 t 3 t 2 t 1 0 t

ε

CG_SCORE

a

GROWTH

a

RD_INT

a

LEV

a

BIG4

a

ROA

a

MB

a

SIZE

a

AM

COMBINED_R

a

INS

a

ABS_DA

+

=

− − − − t t 10 t 9 t 8 1 t 7 t 6 1 t 5 1 t 4 1 t 3 t 2 t 1 0 t

ε

CG_SCORE

a

GROWTH

a

RD_INT

a

LEV

a

BIG4

a

ROA

a

MB

a

SIZE

a

ABS_DA

a

INS

a

RAM_PROXY

+

=

− − − −

For each formula, the dependent variables RAM_PROXY and ABS_DA are the proxies for earnings quality while the only independent variable INS is institutional ownership. All the other variables are control variables which i will explain in

(21)

following part. The two models are developed from the Kim et al 2010, i replace their independent variable CSR with institutional ownership. Based on the original model and the specific condition for my thesis, I add the CG_SCORE and GROWTH as control variables.

TABLE 1

Variables

Definition Code(Datastream) Calculation

t

ABS_DA Abnormal discretionary accruals in firm year t. calculated by

modified jones model.

t

INS Institutional ownership in firm year t. Calculated by adding the

ownership of investment bank(noshic) and ownership of pension funds (noshpf)

AM

COMBINED_R _{Combined real activities manipulation. Calculated by}

COMBINED_RAM=AB_CFO- AB_PROD+AB_EXP

AB_CFO Abnormal cash flows from operating activities in year t. Residual

in model.

AB_PROD Abnormal production costs in year t. Residual in model.

AB_EXP Abnormal discretionary expenses in year t. Residual in model.

TAit Total accruals in firm year t. Earnings before extra items(wc1551)

minus operating cash flows (wc04860).

ASSETS/At Total assets in year t. (wc02999)

∆REVit Changes in revenues in year t. (wc01001) revenues in 2008minus

revenues in 2007

PPEit Gross property, plant and equipment in year t.(wc02301)

∆ARit Changes in accounts receivables in year t.(wc02051)Receivables in

2008 minus receivables in 2007.

St Net sales in year t.(wc01001)

(22)

2008-net sales in2007.

CFOt Net cash flows from operating activities.(wc04860)

COGSt Costs of goods sold in year t.(wc01051)

∆INVt Changes in total inventory in year t. (wc02101) inventory in 2008

minus inventory in 2007.

DISEXPt Discretionary expenses in year t. Sum of selling, general and

administrative expense and advertising expenses and R&D expenses.(wc01101) because in datastream, the S,G&A expense has included the other two kinds of expenses.

1 t

SIZE − natural logarithm of the market value of equity (MVE) in year

t-1(total shareholders equity wc03995).

1 t

MB − market-to-book equity ratio, measured as MVE/BVE, where BVE

is the book value of equity.( multiply book value per share wc05476 by common shares to calculate basic eps wc05192)

1 t

ROA − The ROA in the previous year, where ROA is measured as income

before extraordinary items(wc1551), scaled by lagged total assets.(wc02999)

t

BIG4 an indicator variable that takes a value of 1 if the firm is audited by

a Big 4 auditor, and 0 otherwise, auditor in year t(wc07800)

1 t

LEV − Leverage in year t-1, long-term debt (wc03251) scaled by total

assets.

t

RD_INT R&D intensity, (R&D expense wc01201/net sales wc01001), for

the year t

t

GROWTH Percentage changes in total assets in year t. Calculated by total assets in 2008-total assets in 2007/total assets in 2007(wc02999).

t

CG_SCORE The corporate governance score in year t. (cgvscore in asset4-esg

(23)

3.4 Control variables

I run multiple regression for the formulas in stata software. According to Kim et al 2010, companies always use real activities manipulation and discretionary accruals together to arrange their wanted earnings. To some extent, they are substitutes for each other. Based on Zang 2012’s results, there is a balance between the two earnings management tools. Companies would like to use the cheapest method which minimize their related costs. Just like Cohen et al 2008 did, for avoiding the interactive impact from the two proxies, i will insert real activities manipulation COMBINED_RAM as a control variable in first model and insert discretionary accruals as a control variable in second model. Then we can get the direct impact on each aspects of earnings quality from independent variable.

Now i will analyze the other control variables one by one, which might affect the earnings quality and institutional ownership if we don’t consider them. According to Roychowdhury 2006 findings, the size of a company as well as the growing chance would make changes to the earnings management results. Some academic evidence also shows the institutional ownership might remain different level in different firm size. Furthermore the growing chance and firm size both have systematic impact on the three parts of combined real earnings manipulation. So i introduce the first two control variables SIZE (company size) and MB( specific growing chance). Dechow et al 1995 suggest that abnormal accruals might have measurement error to overall performance. To separate the interruption to potential company performance, the return on assets ROA will be introduced in my regression models.

Some prior studies such as Francis et al. 1999 and Becker et al. 1998 show us the potential impact to earnings quality and overall firm performance from firms audited by the biggest four auditing firms. So i insert BIG4 as a control variable to minor the interaction. At the same time, the leverage in firms always drives managers to manipulate the earnings because once the debt grow without related growth in earnings, the outside investors, debtors would lose confidence in management or the

(24)

managers cannot get their related bonus (Teoh et al. 1998; Kim and Park 2005). Thus, i include the LEV in the regression models. As the McWilliams and Siegel (2000) says the more investment in research&development and advertising will bring more earnings to a company. To better measure the variable, i use net sales to divide R&D expense, getting a relative amount. However, too much data of the separate advertising expense in datastream is missing, i have to drop it.

Corporate governance performance in this case is an important factor or channel between the institutional ownership and earnings quality. Various prior studies has concluded the corporate governance would directly affect the earings quality, best practice always generate the highest quality. However, the institutional investors behavior and motivation will also affect corporate governance of invested firms. Kim et al 2010 highlights the positive function of corporate governance to earnings management. To isolate its both impact, i use the score as an overall evaluation to corporate governance in my models. By inserting all the control variables, i am able to avoid the correlated omission problem.

4 Results

(25)

TABLE 2

Sample description

Distribution of Firm-Year Observations by Industry

INDUSTRY SIC CODE

#OF

OBSERVATION % OF SAMPLE

Agriculture, Forestry and Fishing 0100-0999 0 0.00%

Mining 1000-1499 10 3.62%

Construction 1500-1799 4 1.45%

Manufacturing 2000-3999 171 61.96%

Transportation, Communications,

Electric, Gas and Sanitary service 4000-4999 9 3.26%

Wholesale Trade 5000-5199 12 4.35%

Retail Trade 5200-5999 33 11.96%

Finance, Insurance and Real Estate 6000-6999 0 0.00%

Services 7000-8999 37 13.41%

Public Administration 9100-9729 0 0.00%

Total 276 100%

In the table 2, I separate the firm-year observations by sic code industry distribution. Most of my companies come from manufacturing industry which weighs 61.96% of total sample. Then the second most companies do business in service industry with 13.41% of the sample while the third most firms involve in retail trade industry with 11.96% of the total. The other samples are evenly distributed in Mining; Transportation, Communications, Electric, Gas and Sanitary service; Wholesale Trade industry. Last point to mention is that all the financial firms have been dropped( sic code 6000-6999).

(26)

TABLE 3

Descriptive Statistics of Selected Variables

N Range Mini Max Mean

Std. Deviation Variance ABS_DA 276 0.717083 -0.51007 0.20701 -0.07137 0.084004 0.007 ABS_DA 276 0.509408 0.000665 0.510073 0.078549 0.07731 0.006 AB_CFO ₂₇₆ _{0.804394 -0.42724 0.377152 0.108669 0.086257} _0.007 AB_PROD ₂₇₆ _{1.408433 -0.95639 0.452043 -0.22491 0.197703} _0.039 AB_EXP ₂₇₆ _1.103111 _-0.3592 _{0.743913 0.095689 0.163749} _0.027 COMBINED_ RAM 276 2.720796 -0.91234 1.808454 0.429272 0.391137 0.153 INS 276 68 5 73 16.29 10.011 100.215 SIZE 276 8.322976 10.06049 18.38347 14.67083 1.15023 1.323 MB 276 3.19969 0.320031 3.519722 1.027305 0.209316 0.044 ROA 276 1.139748 -0.76454 0.375212 0.059322 0.117848 0.014 BIG4 276 1 0 1 1 0.06 0.004 LEV 276 0.520285 0.00E+00 0.520285 0.191075 0.130599 0.017 RD_INT 276 0.526571 0.00E+00 0.526571 0.041594 0.062425 0.004 GROWTH 276 4.207768 -0.74888 3.458884 0.046819 0.280195 0.079 CG_SCORE 276 70.2 26.47 96.67 79.2989 11.78359 138.853

The second ABS_DA is absolute value of the first ABS_DA

In the table 3, you can see the descriptive statistics of the sample after winsorizing all the variables from the range 1% to 99% of their distribution. Thus, we can drop the outliers and make the results more convincing. Clearly we can see the ABS_DA discretionary accruals to prior year total assets has a mean value of -0.07,which might

(27)

prove most of the firms chase income decreasing accruals. Further, it is a little higher than the mean value showed by Cohen et al 2008 and Klein 2002. However, the mean value of absolute value of abnormal accruals 0.08 is nearly the same as their prior studies. Then i check the mean value of three parts of real activities manipulation. The

AB_CFO and AB_EXP both has a positive mean value 0.108 and 0.096 respectively,

which implies more abnormal cash flows in operation and more discretionary expenses will bring less earnings. The AB_PROD shows a mean value of -0.225 which is at a higher level than the value -0.096 in kim et al 2010. As a whole, the combined real activities manipulation gets an average value of 0.429 which is more than the mean value 0.238 in kim et al 2012, indicating that the firms in my sample involve less in aggressive earnings management through real activities manipulation. Regarding to the institutional ownership, it ranges from 5 percent to 73 percent and on average it control 10% of all the outstanding common shares which is a little lower than the mean ownership of 25 percent in Chung et al 2002. Maybe the great loss in financial crisis drives the institutional investors to control less.

Now let us look at the control variables. The mean value of SIZE is 14.67 twice as the mean value 6.995 in kim et al 2012, which indicates the increase in market value of total equity before the financial crisis. Then the MB market to book value and ROA show their mean value 0.209 and 0.117 respectively, remaining the similar level as Kim paper. One point deserves attention is that nearly all the companies are audited by one of the big 4 audit firms, implying better earnings quality than others. Thus, controlling it must be a wise choice for our results. On average, these American listed companies have long term debt 19 percent of their total assets and might show lower profitability than the peers in kim et al 2012. Furthermore, they invest 4.15 percent of net sales amount to research and development. The mean value of growth in total assets 28 percent shows their high growing speed before crisis, maybe because of their high risk policy. Regarding to their overall corporate governance performance, an average score of 79 is similar as the mean value in the previous years before the crisis.

(28)

TABLE 4 Correlations among All the Variables

ABS_DA ABS_DA AB_CFO AB_PRODAB_EXP

COMBINE

D_RAM INS SIZE MB ROA BIG4 LEV RD_INT GROWTH

CG_SCO RE ABS_DA 1 -.886** -.145* .101 -.198** -.166** -.036 .064 -.027 .646** -.020 .043 -.325** .074 .037 ABS_DA -.886** 1 .033 -.036 .162** .094 .004 -.066 .011 -.619** .028 -.044 .315** -.041 -.076 AB_CFO -.145 * .033 1 -.604** .187** .604** -.042 .125* .007 .611** .018 -.252** .010 .080 .064 AB_PROD .101 -.036 -.604 ** 1 -.830** -.986** -.082 -.029 .000 -.358** .007 .223** -.154* -.030 -.071 AB_EXP -.198 ** .162** .187** -.830** 1 .880** .145* -.037 .022 .001 -.031 -.224** .346** .037 .006 COMBINED _RAM -.166** .094 .604** -.986** .880** 1 .093 .026 .011 .316** -.012 -.262** .225** .048 .052 INS -.036 .004 -.042 -.082 .145* .093 1 -.197** -.062 -.065 -.016 -.058 .102 -.007 -.131* SIZE .064 -.066 .125* -.029 -.037 .026 -.197** 1 .027 .137* .010 -.174** -.086 .116 .220** MB -.027 .011 .007 .000 .022 .011 -.062 .027 1 -.012 .000 .063 -.019 .223** -.038 ROA .646** -.619** .611** -.358** .001 .316** -.065 .137* -.012 1 -.007 -.202** -.244** .221** .054 BIG4 -.020 .028 .018 .007 -.031 -.012 -.016 .010 .000 -.007 1 .054 .040 -.008 -.017 LEV .043 -.044 -.252** .223** -.224** -.262** -.058 -.174** .063 -.202** .054 1 -.265** .003 .080 RD_INT -.325** .315** .010 -.154* .346** .225** .102 -.086 -.019 -.244** .040 -.265** 1 -.009 -.060 GROWTH .074 -.041 .080 -.030 .037 .048 -.007 .116 .223** .221** -.008 .003 -.009 1 -.122* CG_SCORE .037 -.076 .064 -.071 .006 .052 -.131* .220** -.038 .054 -.017 .080 -.060 -.122* 1

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

(29)

Table 4 presents the Pearson bivariate correlation coefficients for all the variables used in my models. With the variable definitions in table 1, i can find the ABS_DA has a negative coefficient -.036 with INS, implying the institutional shareholders would reduce discretionary accruals and improve earnings quality. The same condition appears to COMBINED_RAM, which is 0.093 positively correlated with INS, reflecting the positive impact to real activities manipulation from institutional investors. Another point consistent with active monitoring hypothesis is the significant positive 0.145 between AB_EXP and INS. Furthermore, the INS and

AB_PROD would move in an opposite direction, also the same as my hypothesis. In

addition, there is a positive relationship 0.094 between COMBINED_RAM and

ABS_DA, meaning that the two earnings management method are substitutes in terms

of firms. However, some of these coefficients are not that significant but in the right directions, supporting my hypothesis to some extent. The negative coefficient between the ABS_DA and BIG4 reflects the positive improvement to earnings quality form the big 4 auditors. At the same time, there exists a significant negative correlation between the ABS_DA and ROA, meaning the earnings management would negatively affect their financial performance. This is also supported by the significant positive coefficient of COMBINED_RAM and ROA. The ABS_DA, CG_SCORE have

an significant negative coefficient -0.76 while the CG_SCORE and

COMBINED_RAM own an significant positive coefficient 0.052. Clearly we can see

the positive interaction between high earnings quality and best corporate governance practice.

Among all the correlation coefficients, only the relationship between

COMBINED_RAM , AB_EXP, AB_PROD and AB_CFO is significantly higher than

the others. Because they all represent the same measure and can be easily replaced by each other. Thus, these highly correlated proxies will not be dropped and has no impact on my results.

(30)

TABLE 5

Multiple Regression of Discretionary Accruals on Institutional Ownership

*,**,*** implies statistical significance at the 0.10, 0.05 and 0.01 levels, respectively, based on two-tailed tests.

Variables definition are shown in TABLE 1.

According to Gow et al 2010, i improve the standard error by a two-dimensional

ABS_DA Coefficient T-stat P>t P-stat Std. Err. INS -0.0007 -2.02** 0.044** 0.0003 COMBINED_RAM 0.0592 6.15*** 0*** 0.0096 SIZE -0.0006 -0.2 0.84 0.0031 MB -0.0093 -0.57 0.571 0.0164 ROA -0.4951 -14.74*** 0*** 0.0336 BIG4 0.0376 0.68 0.498 0.0554 LEV -0.0667 -2.32** 0.021** 0.0287 RD_INT 0.0483 0.79 0.431 0.0612 CG_SCORE -0.0002 -0.78 0.434 0.0003 GROWTH 0.0314 2.46** 0.015** 0.0128 _cons 0.1029 1.37 0.172 0.0752 Source SS df MS Number of obs = 276 F( 10, 265) 27.73 Model 0.8404 10 .084040363 Prob > F 0 Residual 0.8032 265 .003031034 R-squared 0.5113 Adj R-squared 0.4929

(31)

cluster at firm year level.

TABLE 5 show us the results of multiple regression for the association between the institutional ownership and absolute value of discretionary accruals. The Prob > F=0 and R-squared=0.51 reflect the robustness and strength of my regression model. Undoubtedly, i get a significant negative coefficient -0.007(p value <0.05) between the ABS_DA and INS from my sample, supporting my active monitoring hypothesis. When institutional ownership increase by 1, the discretionary accruals will reduce 0.007 with related improvement in earnings quality. Although it is not a high coefficient, it still reflects the positive action by institutional investors. Further, there is a really significant association between the ABS_DA and

COMBINED_RAM, specifically, 0.06( p<0.01), indicating firms usually largely use

one of the two choices but not both of them. This is consistent with Cohen et al 2008, Graham et al 2005, Zang 2012, in which they find the managers always try to

trade-off between discretionary accruals and real activities manipulation in earnings management. The ABS_DA is also significantly negatively -0.4951(p<0.01)connected to ROA, suggesting the listed firms with better financial performance will not engage too much in accruals management. Concurrently, the higher leverage might bring less discretionary accruals. I assume that firms obtaining more debt might not care the earnings performance that much because their leverage has already show some bad signals. Last but not least, the significant positive coefficient 0.0314(p<0.05) of

GROWTH and ABS_DA suggests that the companies which grow fast in assets might

involve more in accrual-based earnings management. In my opinion, the firms want to give a better answer on earnings when they acquire more assets than their peers, needing further tests.

In a word, my first model regression results are consistent with active monitoring hypothesis. Even in financial crisis 2008, the American listed companies with higher institutional ownership might involve less in discretionary accruals, thus providing earnings report of higher quality.

(32)

TABLE 6

Multiple regression of Real Activities Manipulation on Ins Ownership

According to Gow et al 2010, i improve the standard error by a two-dimensional cluster at firm year level.

COMBINED

_RAM Coefficients t-stat P>t p-stat Std. Err.

INS 0.0045 2.22** 0.028** 0.0021 ABS_DA 2.1109 6.15*** 0.000*** 0.3431 SIZE -0.0037 -0.20 0.844 0.0187 MB 0.0906 0.93 0.355 0.0977 ROA 2.1240 8.96*** 0.000*** 0.2370 BIG4 -0.1468 -0.44 0.658 0.3310 LEV -0.1702 -0.98 0.326 0.1730 RD_INT 1.4275 4.02*** 0.000*** 0.3554 CG_SCORE 0.0026 1.45 0.147 0.0018 GROWTH -0.1020 -1.33 0.186 0.0769 _cons -0.0561 -0.12 0.901 0.4509

Source SS df MS Number of obs = 276.0000

F( 10, 265) 12.4300

Model 13.4355 10 1.3435504 Prob > F 0.0000

Residual 28.6362 265 .10806116 R-squared 0.3193

Adj R-squared 0.2937

(33)

TABLE 7

Regression of Three Parts of Real Activities Manipulation

AB_CFO

AB_EXP

AB_PROD

Coef. T-stat P>t Coef. T-stat P>t Coef. T-stat P>t

INS 0.0003 0.96 0.336 0.0020 2.11** 0.036 -0.0022 -2.13** 0.034** ABS_DA 0.7578 14.45*** 0*** 0.3551 2.25** 0.025 -0.9980 -5.66*** 0*** SIZE 0.0033 1.14 0.256 -0.0049 -0.56 0.573 0.0021 0.22 0.829 MB 0.0188 1.26 0.21 0.0336 0.75 0.456 -0.0382 -0.76 0.447 ROA 0.7880 21.75*** 0*** 0.2234 2.05** 0.042 -1.1126 -9.14*** 0*** BIG4 0.0049 0.1 0.923 -0.0997 -0.65 0.514 0.0520 0.31 0.76 LEV 0.0111 0.42 0.675 -0.1397 -1.75* 0.081* 0.0416 0.47 0.64 RD_INT 0.0912 1.68* 0.095* 0.7697 4.7 0 -0.5667 -3.1** 0.002** CG_SCORE 0.0003 1.08 0.283 0.0009 1.06 0.289 -0.0014 -1.56 0.12 GROWTH -0.0427 -3.64*** 0*** 0.0082 0.23 0.817 0.0675 1.71* 0.089* _cons -0.1015 -1.47 0.142 0.0836 0.4 0.688 0.0382 0.16 0.869

(34)

TABLE 6 and TABLE 7 represent the regression results of real activities

manipulation(combined and separate) on the institutional ownership. Again the Prob > F=0 and R-squared=0.319 show the robustness and strength of my regression model. Furthermore, the low level standard errors make my results more convincing. Clearly, in TABLE 6 you can find the COMBINED_RAM is significantly and positively associated with the INS, specifically, 0.0045(p<0.05), which is consistent with active monitoring hypothesis again. Because in my case, higher value of COMBINED_RAM means less aggressive real activities manipulation as well as better earnings quality. In

TABLE 7 first two parts of real earnings management,

AB_CFO, AB_EXP

are

0.0003 and 0.0020(p<0.05) positively connected to the INS. Concurrently, the

AB_PROD

has a negative relationship with INS, significantly at -0.0022(p<0.05), supporting my prior expectation. Further, the association between COMBINED_RAM and ABS_DA still maintain a pretty significant positive level, 2.1109(p<0.01),

consistent with the substitutes findings of Cohen et al 2008. The coefficients of the three parts of COMBINED_RAM and ABS_DA all have the expected directions and pretty high significance, further proving the prior academic evidence.

Regarding to the other control variables, the listed firms with higher ROA will not engage more in the real earnings management because their coefficient is significantly positive(p<0.01). Combining the result with the association in TABLE 5, I can

observe that the good performance in returns on assets will always prevent the earnings management in listed firms. The significant positive coefficient of RD_INT to COMBINED_RAM also reflects the fact that more investment on research & development usually bring higher earnings quality in terms of real activities

manipulation. Last point deserving our attention is that the CG_SCORE is positively associated with COMBINED_RAM. This result still implies the important role of corporate governance in guaranteeing earnings quality, especially after the SOX Act. In TABLE 7, the relation of separate parts and ROA, RD_INT, CG_SCORE are all in the same direction and at same significance level as COMBINED_RAM. However, the

(35)

companies with quick increase in total assets still engage more into the operation cash flows management, the same as the findings in TABLE 5. Specifically, this kind of companies have lower earnings quality due to their desire to better performance. In summary, the results of regression of real activities manipulation on institutional ownership also prove my active monitoring hypothesis.

4.3 Additional analysis

4.3.1 Analysis of an alternative measure of earnings quality

To ensure robustness of my findings, I will use an alternative measure to represent earnings quality and then use similar model to run regression on institutional

ownership. Based on the model in Penman 2001 and Abdelghany 2005, i will choose a simple measure which is the ratio of cash flows from operation divided by the net income. It describes the closeness of cash and has a negative association with earnings quality. Later I use the first model but replace discretionary accruals with this ratio and add ABS_DA as a control variable. Then I get TABLE 8 showing the regression results.

As you can see in TABLE 8, the slightly changed model still shows robustness and strength. Even though the coefficient of the alternative earnings ratio and INS is really small and insignificant, its negative direction can reflect the positive function of institutional shareholders to earnings quality to some degree. Maybe this alternative measure is kind of simple and cannot represent the earnings quality that well as the above two proxies. This time in terms of this proxy, the firms with more assets growth would announce earnings of higher quality.

To conclude, the additional analysis cannot give many support to my findings, but also cannot weaken my results above.

(36)

TABLE 8

Multiple Regression of Earnings Quality on Institutional ownership

Alternative Measure Coefficients T-stat P>t Std. Err.

INS -0.000001 0 0.998 0.0006 COMBINED_RAM 0.003916 0.21 0.832 0.0185 ABS_DA 0.813600 7.38*** 0*** 0.1103 SIZE 0.016797 2.99*** 0.003*** 0.0056 MB 0.009302 0.32 0.752 0.0294 ROA 0.951230 11.69*** 0*** 0.0813 BIG4 0.047963 0.48 0.63 0.0996 LEV 0.061968 1.19 0.236 0.0521 RD_INT -0.030207 -0.27 0.784 0.1101 CG_SCORE 0.000452 0.84 0.4 0.0005 GROWTH -0.088466 -3.81*** 0*** 0.0232 _cons -0.334031 -2.46** 0.014** 0.1356

Source SS df MS Number of obs =276

F( 11, 264) =20.21

Model 2.1718 11 0.1974 Prob > F =0

Residual 2.5792 264 0.0098 R-squared =0.4571

(37)

5 Conclusion and Limitation

In my thesis, based on the special background of financial crisis 2008 and mixed prior academic evidence on this topic, I come up with the question “does the institutional ownership in American listed companies have any impact on the earnings quality?”. According to the prior literature, there are two opposite opinions about the

association.The active monitoring hypothesis regards institutional shareholders as long-term, positive acting investors while the private benefits hypothesis regards them as short-term, private interests chasing investors. After considering the various factors, I set my hypothesis supporting the active monitoring hypothesis due to the long-term profitability, earnings quality usefulness and American stock market needs.

Basically from the models in Kim et al 2012, I develop my two models for

discretionary accruals and real activities manipulation to run regression. Finally, my results significantly prove that firms with higher institutional ownership would show less discretionary accruals and real activities manipulation. Clearly saying, the institutional owners in the United States listed companies tend to improve earnings quality, supporting my hypothesis.

However, some limitations do exist in my research. Firstly, because of adding corporate governance score as a control variable, I drop a lot of firms with missing data from my initial sample. Thus, the limited number of observations would weaken my findings. Secondly, due to limited access to Compact D database, I get the

institutional ownership by adding ownership of institutional investment, banks and pension funds, which cannot include all kinds of institutional shareholders. Thirdly, my models might not perfectly describe the relationship and contain all the possible control variables. The test will be more robust if I can add ownership of management or controlling shareholder. Lastly, my additional analysis cannot give a strong support to my findings, which needs further additional tests. I hope these limitations can be improved by future research addressing this topic.

(38)

Reference list

Brous PA, Kini O. (1994). The valuation effects of equity issues and the level of institutional ownership: evidence from analysts' earnings forecasts. Finance Management , 33–46.

BJ Bushee. (1998). The influence of institutional investors on myopic R&D investment behavior. Account Rev, 73, 305-334.

BJ Bushee,(2001). Do Institutional Investors Prefer Near-Term Earnings over Long-Run Value.Contemporary Accounting Research,18,207-246.

C Leuz, D Nanda, PD Wysocki.(2003).Earnings management and investor protection: an international comparison, Journal of financial economics,69.505-527.

Daily, C.M., Dalton, D.R., & Cannella Jr., A.A. (2003). Corporate governance: Decades of dialogue and data. Academy of Management Review, 28 (3), 371-382. David H. Erkens, Mingyi Hung , Pedro Matos. (2012). Corporate governance in the 2007-2008 financial crisis:Evidence from financial institutions worldwide, Journal of Corporate Finance,18,389-411.

Gunny, K. A.(2010). The relation between earnings management using real activities manipulation and future performance: evidence from meeting earnings benchmarks.Contemporary Accounting Research , 27 (3), 855-888.

Hadani, M., Goranova, M., & Khan, R. (2011). Institutional investors, shareholder activism and earnings management. Journal of Business Research, 64, 1352-1360. Hill,C.W.& Jones,T.M.(1992).Stakeholder-Agency theory. Journal of Management Studies , 29 (2), 131-154.

Hsihui Chang, Guy D. Fernando, Woody Liao,(2009) Sarbanes-Oxley Act, perceived earnings quality and cost of capital, Review of Accounting and Finance, 8, 216 - 231. Ilia D. Dichev, John R. Graham, Campbell R. Harvey, Shiva Rajgopal.(2013) Earnings quality: Evidence from the field,Journal of Accounting and Economics 56 1-33.