1
Upward earnings management and a firm’s long-run stock
performance following an initial public offering
Arjen de Graaf
Student number 1547798
Master’s Thesis
University of Groningen
Faculty of Economics and Business
MSc Business Administration, specialization Finance
Version date: 24 Augustus 2012
2
ABSTRACT
In this thesis, I investigate whether firms deliberately manipulate their earnings upwards when going public, in order to mislead investors and gain at their expense. Moreover, I investigate whether this manipulation is offset in the subsequent years by a downward adjustment in earnings. Finally, in order to investigate whether investors are able to detect earnings manipulation during the IPO year, I test whether pre-IPO earnings management has a negative effect on a firm’s long-run stock performance, using accounting accruals as a proxy for earnings management. If investors are not capable of detecting earnings management during the IPO year, they will not anticipate on the subsequent downward adjustments in earnings after the IPO. This will then result in stock underperformance. Results of this study indicate that IPO firms on average do not engage in upward earnings manipulation in the year of the IPO. I have found evidence that pre-IPO upward earnings management leads to post-IPO downward earnings management in the second, third and fourth year after the IPO. I have found some evidence that pre-IPO earnings management has a positive effect on post-IPO performance in the first year after the IPO, and a negative effect in the second year after the IPO, depending on the measurement for earnings management which is used. Therefore, results of this study indicate that investors are not widely misled by firms going public, which engage in upward earnings management.
Keywords: earnings management, earnings manipulation, initial public offerings; long-run stock performance
3
TABLE OF CONTENTS
1. INTRODUCTION ... 4
2. LITERATURE REVIEW ... 7
2.1 What is earnings management? ... 7
2.2 The relationship between earnings management and initial public offerings ... 9
2.3 Other factors influencing a firm’s performance after an equity offering ... 11
2.4 Hypotheses development ... 12
3. METHODOLOGY & DATA ... 14
3.1 Measures for earnings management ... 14
3.2 Methods used in this research: The Modified Jones model (1991) ... 15
3.3 The regression model to determine whether earnings management in a given year is negatively related to future earnings management ... 19
3.4 The regression model for the researched relationship ... 20
3.5 Sample selection and Data ... 22
4. RESULTS ... 30
4.1 Results of the regression analysis of the Modified Jones (1991) model ... 30
4.2 Results of the time-distribution of accrual measures ... 33
4.3 Results of the regression analysis of the effects of accruals in a given year on future accruals ... 38
4.4 Results of the time-distribution of post-IPO stock performance ... 43
4.5 Results of the Univariate regressions of post-IPO stock performance on accrual measures ... 45
4.6 Results of the Multivariate regressions of post-IPO stock performance on accrual measures ... 50
5. SUMMARY, CONCLUSIONS AND LIMITATIONS ... 61
5.1 Conclusions ... 61
5.2 Limitations ... 64
4
1. INTRODUCTION
Initial public offerings (IPOs) are major corporate events and play an economic role in facilitating economic innovation. Often there is only limited information available of a firm when it goes public, which puts the issuer at an information advantage and the potential investor at an information disadvantage. This may lead to incentives for the issuer to mislead the investor and make him- or herself better off at the expense of the investor. The empirical question of whether firms manage earnings in IPOs is therefore important both in itself and because of its bearing on capital markets in general, as stated by Teoh et al. (1998a).
When a firm goes public there is usually only little information available about the firm. Since investors are usually unfamiliar with the firm at hand, they have to rely on the prospectus which is drawn up by the issuing firm’s underwriter, which for the largest part consists of the firm’s financial statements for a number of years. DuCharme et al. (2001) state that the firm’s managers have considerable opportunities to manage the earnings which are reported in these financial statements, in such a way that investors may overestimate the firm’s prospects. Healy et al. (1999) mention three different categories of motives managers could have to manipulate the firm’s earnings, namely: Capital Market Motivations (I), Contracting Motivations (II), and Regulatory Motivations (III), which I will elaborate on further in the literature section.
As mentioned by several authors, any manipulation of earnings in one year has to be adjusted at some point in the future (Teoh et al., 1998a; Roosenboom et al., 2003; Chan et al. 2006). In the long-run, managers cannot inflate the firm’s net-income compared to the firm’s operational cash flows. Therefore, any positive income adjustments in the current year will be offset with negative income adjustments in some future year. This means that current upward manipulation of earnings should have a negative effect on future earnings. If investors fail to recognize upward earnings manipulation, they will overvalue the firm. When the firm is forced to adjust its earnings downward and investors have a better understanding of the firm’s outlook, this will lead to a realization that the firm had been overvalued. This will subsequently lead to a decrease in the firm’s stock value (Roosenboom et al., 2003; Chan et al., 2006). According to Teoh et al. (1998a) the larger the upward manipulation has been before the IPO, the larger the impact will be on the negative stock price adjustment.
5
already been a lot of research to examine this relationship. However, research concerning the relationship between earnings management and IPOs has focused on periods before large accounting scandals, such as Enron, Parmalat and Worldcom of 2001-2003. These scandals, in turn, led to the implementation of the Sarbanes-Oxley Act in the United States in 2002. In this study, I focus on the period after these events, excluding the financial crisis years of 2008 onwards. I expect investors to be more aware of potential earnings management by offering firms, and to be able to accurately reflect this in the firm’s stock prices. My goal with this research is to identify whether investors are still being misled by firms going public as has been proven in previous research, since the events that have taken place should have made investors more cautious and aware of earnings management. My research question for this thesis is:
“Will earnings management before an equity offering lead to a lower stock performance in the years after the offering?”
In my thesis, I will focus on IPOs and determine whether or not earnings before the equity offering has an influence on the firm’s long-run performance, as measured by its buy-and-hold return. My sample will consist of all non-financial firms which have made an equity offering between 2001 and 2007 in the United States.
A general used proxy for earnings management are accounting accruals. As will be explained in the methodology section in more detail, accruals are the difference between a firm’s net income and its operational cash flows. As first proposed by Jones (1991), these total accruals can be divided into a part which can be explained by economic factors, called normal or non-discretionary accruals, and into a part which cannot be explained by economic factors, called abnormal or discretionary accruals. Since discretionary accruals cannot be explained by economic factors, this part of accruals are assumed to be the results of earnings management. Following previous literature, I distinguish between non-discretionary and discretionary accruals by using the Modified Jones (1991) model.
6
earnings management does not have a significant effect on a firm’s stock price in the years after the IPO.
7
2. LITERATURE
In this chapter I will give an overview of the previous research on earnings management. In the first section, I describe what earnings management is. In the second section, I describe the relationship between earnings management and initial public offerings (IPOs). In the third section, I will list which other factors may also be of influence on equity offerings and the subsequent firm performance. Finally, in section four, I will describe my hypotheses for this thesis.
2.1 What is earnings management?
Before I can answer the question of what is already written in the literature about the relationship between earnings management and initial public offerings, it is desirable to have a definition of what earnings management actually is. In a theoretical paper in which Healy et al. (1999, p. 368) review the existing literature concerning earnings management, the authors define earnings management as:
‘‘Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers.”
8
The definition of earnings management by Healy et al. (1999) supports the vision of Beatty et al. (1999). Healy et al. (1999) state that the goal of earnings management is to mislead certain shareholders about the firm’s underlying economic performance. According to the authors, this will happen when managers believe that at least some of the firm’s shareholders will not notice this deception.
Healy et al. (1999) also give an overview of the relevant studies concerning earnings management and the empirical proof that earnings management occurs. They also show that there are several motives why firm management would engage in earnings management. They divide these motives into three categories, namely Capital Market Motivations (I), Contracting Motivations (II), and Regulatory Motivations (III).
They define Capital Market Motivations as incentives managers within a firm have to manipulate the firm’s earnings with regards to the firm’s stock performance. Since investors and analysts make heavy use of a firm’s financial reporting in their decisions, managers are motivated to manipulate the firm’s earnings upwards. In doing so, the managers attempt to influence the short term stock performance. An example of a situation in which this explicitly happens is during initial public offerings (IPOs), as the managers will try to get an offer price as high as possible. In section 2.2 I will go into more detail about the relationship between earnings management and IPOs. Another important motive for earnings management in relation to the capital markets is the incentive managers have to make sure their firm does not underperform with respect to the expectations of investors and analysts. Would the firm not live up to these expectations, the managers know that this will have a negative effect on the firm’s stock performance. In contrast, when the firm performs above expectations, this will have a positive effect on the stock performance. Since the manager’s evaluation and remuneration is often linked to the firm’s stock performance, managers will almost always have strong incentives to make sure the firm’s stock price performs well.
9
bonus or enlarged job security. Another type of Contracting Motivations described by Healy et al. (1999) is related to lending contracts. These contracts are designed to prevent firm management to undertake actions which are beneficial to shareholders and harmful to debt holders. These contracts could also stimulate earnings management, since debt covenants which are included in the contract are also often linked to the firm’s reported earnings.
With regards to the third category, Regulatory Motivations, Healy et al. (1999) state that managers can also engage in earnings management to avoid certain regulations. These motivations are often related to downward manipulation of a firm’s earnings. There can be certain earnings thresholds above which firms would need to comply with new rules, such as stricter accounting standards or environmental regulations. As long as the firm manages to stay below these thresholds these ‘extra’ regulations would not apply to them.
Several authors use accruals as a measure for earnings management, in the methodology section I will explain how accruals can be calculated.
2.2 The relationship between earnings management and initial public offerings
The IPO process gives firms several incentives to engage in earnings management. First of all, as also explained in the previous section, the information asymmetry between the investor and the firm is very big around the time of the IPO. The investor has to rely almost solely on the information in the prospectus which is offered by the firm, while the firm has a lot of opportunities to change its accounting choices in a way which could be beneficial for the offering price during the IPO. In addition, there are few reliable independent sources of information about the firm (Ritter, 1991; Lerner, 1994; Teoh et al., 1998a; Loughran et al., 1995 and 2000; Baker et al., 2000; Fan, 2007). Second, firms will know that high reported earnings will raise stock prices and that a higher stock price will be desirable when the firm is selling equity (Teoh et al., 1998a). Third, the issuer believes that its financial statement information will affect its IPO offering price. Consequently, the possibility of influencing investor response and the initial offer price through accounting choices may provide the issuer with the incentive to do so (Friedlan et al, 1994; Aharony et al, 1993).
10
more pronounced for small firms with large financial leverage. In an empirical study of U.S. firms going public between 1975 and 1984, Ritter (1991) shows that investors are periodically overoptimistic about the earnings potential of young growth companies, which would give these companies additional incentives to manage their earnings opportunistically in such a period leading up to an IPO. Ahmad-Zaluki et al. (2011) show in an empirical study of earnings management in Malaysian IPOs between 1990 and 2000, that older companies and those audited by a prestigious auditing firm (such as the Big 4 auditing firms) exhibit lower levels of earnings management. Furthermore, they also show that high levels of ownership concentration, as is common in Malaysia, points towards a higher level of post-IPO involvement, which diminishes the incentives for a firm to engage in earnings management. In a theoretical study Chaney et al. (1995) differentiate between two types of firms, high-value and low-high-value firms. Firm high-value is based on the ability to generate economic earnings and investors subsequently expect high-value firms to generate high economic earnings. They state that corporate taxation will create a trade-off for firms between the benefit of being identified as a high-value firm and the additional tax liability from over-reporting earnings. As such, investors should be able to realize that only true high-value firms are willing to pay the additional taxes associated with the inflated earnings. However, they also state that the same line of reasoning will not necessarily apply to the managers working at these firms, since they may be motivated to have their firms identified as high-value which could result in a higher compensation for the manager in question. In another empirical study of the effects of earnings management by IPO firms in the U.S. market between 1987 and 1997, Fan (2007) shows that another important factor whether or not to engage in earnings management before an IPO is ownership retention. She states that income-increasing earnings management is arguably costly to the firm, even though the manager might inflate earnings in a single period, over the firm’s life cycle, total reported earnings must equal total cash flows. Ownership retention is a clear signal to the market that the firm’s current owners believe that the offering price is justified. She also states that riskier firms tend to manipulate earnings more often and will less often make use of ownership retention.
11
companies (Ritter, 1991). Ahmad-Zaluki et al (2011) even go as far to say that earnings management might even be a temporary phenomenon which should only occur during periods of economic stress. When looking at the individual firm which may be interested in using earnings management surrounding an IPO, not all researchers agree when this will take place. Friedlan (1994) and Teoh et al. (1998a and 1998b) show in different empirical studies that earnings management takes place in the year before the IPO, as well as the year during which the IPO takes place. The empirical results of both Aharony et al (1993) and Roosenboom et al. (2003) do not support this. They come to the conclusion that there is not any earnings management prior to an IPO, though Roosenboom et al. (2003) do agree with the notion that earnings are being manipulated during the IPO year.
Finally, when looking at the effects earnings management has on the firm’s performance post-IPO, empirical literature shows that IPO-firms perform significantly below their industry standards or similar non-issuing firms. The fact that a firm engaged in earnings management prior to an IPO is a good predictor for low future earnings. The pre-IPO earnings management is positively related to the initial firm value directly after the IPO has taken place, which gives entrepreneurs the incentive to opportunistically manipulate earnings. This earnings management is negatively related to the firm’s long-term performance (Teoh et al., 1998a and 1998b; DuCharme et al., 2001). Teoh et al. (1998a) state that IPO firms tend to have high positive earnings management in the year of the IPO on average. This is followed by poor earnings in the long-run and negative or downward earnings management. Based on these results it is expected for an IPO firm to engage in upward earnings management before an IPO takes place, which will have a negative influence on the firm’s earnings in the long run.
2.3 Other factors influencing a firm’s performance after an initial public offering
12
greater ‘‘arm’s length’’ than in a private company, and consequently demand higher quality reporting to resolve the information asymmetry. They mention multiple reasons why it would not be beneficial for a firm to manipulate earnings prior to an IPO, such as: there would be a risk of regulatory action following the IPO; poor reporting quality would increase a firm’s cost of capital, which would be very worrisome if the firm needs external financing; poor reporting quality would create a negative reputational effect, something which a new firm’s managers, board members and auditors will try to prevent. According to Ball et al. (2008) it seems more likely that firms will enhance their financial reporting quality around the time of an IPO than the alternative hypothesis that managers opportunistically inflate earnings.
In an empirical study concerning Malaysian IPOs between 1990 and 2000, Ahmad-Zaluki et al. (2007) find evidence that the findings in the Western world do not necessarily hold in a developing economy. According to the authors, the question of whether or not post-IPO underperformance can be explained by pre-IPO earnings management will to a large extent depend on the methods used to examine a firm’s performance. The authors have used event-time cumulative abnormal returns (CARs) and buy-and-hold returns as return measures, and market benchmarks, equal-weighting schemes, and value-weighting schemes as benchmarks. Depending on the combination which is used IPO firms in Malaysia either overperformed or underperformed compared to the benchmark. In an empirical study of U.S. IPOs between 1935 and 1972, Gompers et al. (2003) support this notion that whether or not a firm is underperforming in the long-run depends on the method used to determine the performance. In a different study, Ahmad-Zaluki et al., (2011) also show that being audited by a reputable auditor (such as one of the Big 4) can also result in a better post-IPO performance. Chang et al. (2010) show in an empirical study that the reputation of the underwriter who facilitates the initial public offering also results in a better post-IPO performance.
2.4 Hypotheses development
13
H1: Firms will report relatively high earnings in both the year before the IPO and also
the year an IPO takes place
Roosenboom et al. (2003) state that given the fact that managers are forced to reverse earnings management in subsequent years, it is not likely that poor-quality firms with inflated earnings will be able to sustain these high earnings indefinitely. This means in the long run accruals should be zero, since the sum of all earnings should equal the sum of all cash flows over an extended timeframe. To be able to determine whether this is also the case for IPO companies, my second hypothesis is:
H2: Relatively high earnings in a given year will have a negative influence on
earnings in future years
Finally, it is important to answer the main research question whether relatively high earnings actually has a negative impact on a firm’s future performance. Several authors claim that earnings management, as measured by discretionary accruals, predicts the variation in a firm’s long-run stock performance in the years after the IPO. They state that investors are misled by the high reported earnings in the IPO year and therefore fail to compensate offer prices for ‘aggressive’ earnings management firms compared to the ‘conservative’ earnings management firms (Teoh et al., 1998a; Roosenboom et al, 2003). Since earnings management will have to reverse in subsequent years, it is expected that poor-quality firms which have inflated their earnings pre-IPO will be more likely to report lower earnings compared to their high-quality counterparts. The authors state that for poor-quality firms which have engaged in earnings management, the cash flows will not be sufficient to mitigate the impact of reversing accruals. This will finally result in investors being able to detect the earnings manipulation and subsequently adjust the poor-quality firm’s stock prices downwards. Thus my third hypothesis is:
H3: The level of earnings in the year before the IPO is negatively related to a firm’s
14
3. METHODOLOGY & DATA
This chapter is organized as follows. Section 1 describes which measures will be used for earnings management. In section 2, I explain how the Modified Jones (1991) model works, which I use to measure earnings management. In section 3, I present the regression model used to determine whether earnings management in a given year is negatively related to future earnings management. In section 4, I present the regression model used for measuring the relationship between earnings management before an IPO and the long-run stock performance. Finally, in section 5, I describe the data and sample used for this thesis.
3.1 Measures for earnings management
Most studies on earnings management have relied on accrual-based measures as a proxy for earnings management (e.g. Teoh et al., 1998a; Teoh et al., 1998b; Teoh et al., 1998c; DuCharme et al., 2001; Roosenboom et al., 2003; Ahmad-Zaluki et al., 2011). Total accruals are calculated as the difference between net income and operational cash flows in a given year (Jones, 1991; Aharony et al. 1993; Cotten, 2008):
TACC = NI – CFO
(1)Where:
TACC = Total accruals NI = Net income
CFO = Cash flows from operations
15
investors. An important example which is offered by Chan et al. (2006) of how managers might inflate earnings is overstating the ending level of inventory and thus underestimating the costs of goods sold. This would result in higher earnings, without altering the cash flow situation, causing the total accruals to increase. The increased accruals in one period will have consequences for future periods, since at a certain point in time these effects will have to be reversed. In the same example of Chan et al. (2006) regarding the overstating of inventory a potential impact could be an increase in inventory write-downs in subsequent years. This would have the opposite effect, resulting in a lower net income, while having no effects on the operational cash flow in the given year, resulting in lower or even negative accruals.
Although large positive accruals could be an indication for earnings management, several authors show that total accruals by itself do not necessarily have to be the result of earnings management (e.g. Jones, 1991; Teoh et al., 1998a; Teoh et al. 1998b; DuCharme et al., 2001; Chan et al., 2006). The literature states that total accruals can be split into two parts, namely non-discretionary or normal accruals and discretionary or abnormal accruals. Non-discretionary accruals can be explained by the market environment and the firm’s performance, whereas discretionary accruals cannot be explained by these factors. These authors state that discretionary accruals indicate that managers of a firm have manipulated the earnings. In section 3.3, I will elaborate further on how both non-discretionary and discretionary accruals can be estimated.
3.2 Methods used in this research: The Modified Jones model (1991)
non-16
discretionary accruals, (i.e. the amount of accruals which can be explained by economic factors) by estimating the coefficients
β
1, β
2 andβ
3 in using the following equation:TACC
it(1/A
it-1)
= β
1(1/A
it-1) + β
2(∆REV
it- ∆REC
it) (1/A
it-1)
+ β
3PPE
it(1/A
it-1) +
it(2)
Where:
TACC = Total accruals
ΔREV = Change in revenue compared to the previous year
ΔREC = Change in accounts receivable compared to the previous year PPE = Property, plant and equipment
A = Total assets
The reasoning behind the Modified Jones (1991) model is as follows: total accruals can be explained by three factors, namely the change in revenues compared to the previous year, the property, plant and equipment and earnings management in the form of discretionary accruals.
Jones (1991) has included the change in revenue in her model based on the assumption that changes in revenue influence total accruals, because changes in revenue also lead to changes in working capital (accounts payable, accounts receivable, inventory) which are part of total accruals. These changes in total accruals are not a result of manipulation caused by management. Therefore, to determine the non-discretionary accruals, the economic conditions of the organization have to be taken into account. This means that total accruals will either increase or remain the same, and therefore the expected sign of
β
2 is positive or neutral. She also states that revenue controls for17
The original Jones (1991) model does not contain the Accounts receivable term. The reason Dechow et al. (1995) have proposed to include this term in the model is that in the original Jones (1991) model it is assumed that a firm’s management is not able to manipulate the revenues, which would mean that all changes in accruals related to a firm’s revenue would be non-discretionary. In the Modified Jones (1991) model however, it is implicitly assumed that all changes in accounts receivables in the event period result from earnings management. This is based on the reasoning that it is easier to manage earnings by exercising discretion over the recognition of revenue on credit sales than it is to manage earnings by exercising discretion over the recognition of revenue on cash sales. If this modification is successful, then the estimate of earnings management should no longer be biased toward zero in samples where earnings management has taken place through the management of revenues. Therefore the authors suggest that all changes in accounts receivables are the result of earnings management and should for this reason be subtracted from the changes in revenue. An increase in accounts receivables should lead to an increase in net income, while it does not have an impact on operational cash flows. This means that the total accruals will increase.
Jones (1991) has included the property, plant and equipment in her model based on the assumption that an increase in property plant and equipment should lead to an increase in depreciation expenses. This in turn leads to a decrease in net income, but has no impact on operational cash flows. Thus, an increase in depreciation expense leads to lower total accruals, but is not caused by manipulation of management. The expected sign of
β
3 is therefore negative. Jones (1991) scales all variables in the accruals expectations model by lagged assets to reduce heteroscedasticity. She assumes in her model that the lagged assets are positively related with the variance of the disturbance term.
The second step in the Modified Jones (1991) model is to use the estimated values for
β
1, β
2 andβ
3 for a complete sample. These values can subsequently be used as the non-discretionary18
DACC
it(1/A
it-1) = TACC
it(1/A
it-1) - (β
1(1/A
it-1) +
β
2(∆REV
it- ∆REC
it) (1/A
it-1) + β
3PPE
it(1/A
it-1)
(3)Where:
β
1, β
2, β
3 = Estimated coefficients from equation (2)Note that DACC in equation 3 is the same as the error term in equation 2. The error term is subtracted from the non-discretionary part of equation 2 (the first, second and third terms) and is equal to the non-discretionary part of the equation.
Following Teoh et al. (1998a, 1998b), Roosenboom et al. (2003) and DuCharme et al. (2004), I will also use discretionary current accruals as a proxy for earnings management. The discretionary current accruals focus on the short-term rather than on both the short- and the long-term. Since managers are likely to have greater discretion over short-term rather than over long-term accruals, it is worthwhile to evaluate the two components separately (Teoh et al., 1998a, 1998b). To calculate the non-discretionary current accruals part of total current accruals, I will follow the same procedure as in equation 2. The regression for the total current accruals is as follows:
TCA
it(1/A
it-1)
= β
1(1/A
it-1) + β
2(∆REV
it- ∆REC
it) (1/A
it-1)
+
it(4)
19
TCA = Total current accruals
As can be seen from equation 4, the term for the property, plant and equipment has been dropped from the model. The reason for this is that property, plant and equipment has a long-term effect, while changes in revenues and accounts receivables have a short-long-term effect. In the same manner as with equation 3, the estimates for
β
1, β
2 andβ
3 are used to calculate the non-discretionary accruals for the individual firms. The expected value for TCA is the same as for TACC. Both are calculated using equation 1. These non-discretionary current accruals are subtracted from the total current accruals to calculate the firm’s discretionary current accruals, using the following equation:DCA
it(1/A
it-1) = TCA
it(1/A
it-1) - (β
1(1/A
it-1) +
β
2(∆REV
it- ∆REC
it) (1/A
it-1)
(5)
Where:
DCA = Discretionary current accruals
β
1, β
2, β
3 = Estimated coefficients from equation (4)3.3 The regression model to determine whether earnings management in a given year is negatively related to future earnings management
As I have described in the literature review, all accruals have to be reversed at some time in the future. This means that it is expected that accruals in a given year have a negative impact on accruals in future years, leading to lower earnings in those years. To be able to determine whether this is also the case for the sample used in this thesis, I use the following equations:
20
DACC
it+x(1/A
it+(x-1)) = β
0+ β
1DACC
it(1/A
it-1) +
it (8)DCA
it+x(1/A
it+(x-1)) = β
0+ β
1DCA
it(1/A
it-1) +
it (9)As described in the literature, accruals should always reverse at some point in the future. This means that positive earnings manipulation in a certain year should lead to negative accruals in subsequent years, resulting in lower earnings. My expectation is that for all three equations the value for
β
1 will be negative. I will focus on the four years following the IPO.3.4 Relationship between earnings management and long-term stock performance
Once the discretionary accruals have been calculated for every firm in the sample, the relationship between earnings management and the long-run stock performance can be determined. Following authors such as Teoh et al. (1998a) and Roosenboom (2003), I will use the annualized market-adjusted buy-and-hold return as the measure for a firm’s long-run stock performance. I will use the following equation to calculate firm’s annualized market-adjusted buy-and-hold return.
AR
it= (((P
it+ DIV
it) – P
t-1)/P
t-1) - R
m (10)Where:
AR = The annualized market adjusted buy-and-hold return P = The firm’s stock price
DIV = The firm’s annualized dividend payout Rm = The market return
21
performance. With the multivariate regression, I will also take several other factors into account, because according to the literature there are multiple variables which could influence the long-run stock performance. Several of these factors will be included in the regression. For the univariate analysis I use the following equation:
AR
it+x= β
0+ β
1Accrual
it+
it (11)Where:
Accrual = Can be either TACC, DACC or DCA
For the multivariate analysis, I use the following equation:
AR
it+x= β
0+ β
1Accrual
it+ β
2(∑ Lev
it+x)/x + β
3(∑ MV
it+x)/x+
β
4(∑ BTM
it+x)/x +
it (12)Where:
Lev = The firm’s leverage ratio
MV = The equity offering firm’s logged market capitalization BTM = The equity offering firm’s book-to-market ratio
22
Lev is included as a control variable because previous literature has indicated that a higher leverage ratio is expected to have a negative impact on the firm’s return (Ahmad-Zaluki, 2011). Therefore, the expected sign for
β
2 is negative.MV and BTM are included as control variables because Brav et al. (1997) show that smaller firms with a low book-to-market ratio underperform compared to large firms with a high book-to-market ratio.
3.5 Sample selection and data
The data for all U.S. firms is obtained from Datastream. This database consists of information of both financial and non-financial data of companies worldwide.
My sample consists of all firms which made an IPO during 2001 and 2007 in the United States on either the New York Stock Exchange (NYSE) or the NASDAQ. I have chosen a period after the Internet bubble of 1995 and 2000, and before the financial crisis of 2008. A second important reason to take this period for my sample is that it is after the scandals of Enron, Parmalat and Worldcom have taken place. To determine the effects of the IPO on the post-offering period, data for the crisis years 2008-2011 are included as well. For a firm to be included in the final sample, financial data needs to be available in order to calculate the accruals measures as described in the previous sections.
Because the current models for estimating accruals are unsuitable for financial firms, and financial firms have to comply with very different regulations than other firms, all financial firms are excluded from the sample. The fiscal year in which the offering takes place will be called year 0, the year before the offering year -1, the year after the offering year 1, etc. Only those firms for which financial data was available that was needed to calculate the accrual measures were included in the final sample. The final sample consists of 1,562 firms, as can be seen in table 1.
23
valuation, revenue, and the book-to-market ratio. I have created a subsample for the firms listed on the NYSE and also for the firms listed on the NASDAQ.
NYSE firms
For the NYSE IPO sample, the mean of the total assets is $463.4 million, while the median is only $80.41 million. For the NYSE non-IPO sample, the mean of the total assets is $1,412 million, but the median only $157.7 million. The mean market capitalization for the IPO sample is $454.6 million, while the median only is $46.84 million. This can be compared to a mean and median for the non-IPO sample of $6,350 million and $1,392 million, respectively. On average, the NYSE IPO firms are a lot smaller based on market capitalization compared to the non-IPO firms listed on NYSE. The mean and median of the revenue of the IPO sample are $1,874 million and $491.4 million, respectively, compared to a mean and mean for the non-IPO sample of $4,476 million and $983.1 million, respectively. It is interesting to see that based on revenue, the difference in size between the IPO and non-IPO firms is a lot smaller. Finally, looking at the mean and median of the book-to-market ratios, for the IPO sample the values are 3.659 and 2.140 respectively, compared to a mean and median book-to-market ratio for the non-IPO sample of 6.786 and 2.050, respectively.
NASDAQ firms
24
Panel C reports the time-distribution of the IPO firms in the sample. There is a clear concentration of IPOs in the calendar year 2003, while most other years are fairly balanced. The IPO year of 2003 represents 32.52% of the total sample.
25
TABLE 1
Selection process for sample firms going public from 2001 through 2007 on either the New York Stock Exchange or the NASDAQ
Non-financial U.S. common stock IPOs 2001-2007 3257
Less accrual data unavailable in IPO year 1695
IPO’s with available data 1562
Of which are listed on the NYSE 835
Of which are listed on the NASDAQ 727
26
TABLE 2
Sample Characteristics
The sample consists of 1562 IPO firms listed on either the New York Stock Exchange or the NASDAQ and going public in the period 2001-2007. For the IPO sample, Panel A presents characteristics of total assets (A), market capitalization (MV), revenue and the book-to-market ratio (BTM) measured at the end of fiscal year 0. For the non-IPO sample, total book assets (TA), market capitalization (MV), revenue, and the book-to-market ratio (BTM) are reported, measured for all non-IPO listed firms in the years 2001-2007. The distribution of the sample by IPO calendar year is reported in panel B, and the distribution of the sample by industry is reported in panel C.
Panel A: Firm characteristics of IPO sample in issue year
TA ($m) MV ($m) Revenue ($m) BTM
IPO firms NYSE:
Mean 463.4 454.6 1,874 3.659
Median 80.41 46.84 491.4 2.140
Maximum 5,545 51,440 64,279 414.5
Minimum 0.000 0.000 0.000 -265.3
N 828 301 828 643
IPO firms NASDAQ:
27
Table 2 continued
Panel B: Firm characteristics of non-IPO sample
TA ($m) MV ($m) Revenue ($m) BTM
Non-IPO firms NYSE:
Mean 1,412 6,350 4,476 6.786
Median 157.7 1,392 983.1 2.050
Maximum 294,244 504,240 378,611 10,90
Minimum 0.000 0.000 0.000 -1,201
N 2,462 2,117 2,496 2,103
Non-IPO firms NASDAQ:
28
Table 2 continued
Panel C: Time distribution of IPO sample
IPO Calendar year N % Cumulative Cum. %
29
Table 2 continued
Panel D: Industry distribution of IPO sample
Sector N % Cumulative Cum. %
Aerospace and Defense 21 1.34% 21 1.34%
Alternative Energy 15 0.96% 36 2.30%
Automobiles and Parts 15 0.96% 51 3.27%
Beverages 7 0.45% 58 3.71%
Chemicals 23 1.47% 81 5.19%
Construction and Materials 28 1.79% 109 6.98%
Electricity 13 0.83% 122 7.81%
Electronic and Electrical Equipment 32 2.05% 154 9.86% Equity Investment Instruments 38 2.43% 192 12.29% Fixed Line Telecommunications 24 1.54% 216 13.83%
Food and Drug Retailers 4 0.26% 220 14.08%
Food Producers 27 1.73% 247 15.81%
Forestry and Paper 6 0.38% 253 16.20%
Gas, Water and Multi-utilities 3 0.19% 256 16.39%
General Industrials 6 0.38% 262 16.77%
General Retailers 81 5.19% 343 21.96%
Health Care Equipment and Services 135 8.64% 478 30.60% Household Goods and Home Construction 17 1.09% 495 31.69%
Industrial Engineering 24 1.54% 519 33.23%
Industrial Metals and Mining 28 1.79% 547 35.02%
Industrial Transportation 62 3.97% 609 38.99%
Leisure Goods 11 0.70% 620 39.69%
Media 52 3.33% 672 43.02%
Mining 83 5.31% 755 48.34%
Mobile Telecommunications 18 1.15% 773 49.49%
Oil and Gas Producers 86 5.51% 859 54.99%
Oil Equipment and Services 64 4.10% 923 59.09%
Personal Goods 23 1.47% 946 60.56%
Pharmaceuticals and Biotechnology 150 9.60% 1,096 70.17% Real Estate Investment and Services 24 1.54% 1,120 71.70% Real Estate Investment Trusts 65 4.16% 1,185 75.86% Software and Computer Services 123 7.87% 1,308 83.74%
Support Services 73 4.67% 1,381 88.41%
Technology Hardware and Equipment 110 7.04% 1,491 95.45%
Travel and Leisure 71 4.55% 1,562 100%
30
4. RESULTS
In this chapter the results will be explained and presented. This chapter is organized as follows: in section 1, I will present the result of the regression analysis of the Modified Jones (1991) model. In section 2, the time-distribution of the post-IPO stock performance will be shown. In section 3, I will present the results of the regression analysis on the relationship between accruals in a given year and future accruals. In section 4, the time-distribution of post-IPO stock performance will be presented. In section 5, I will present the results of the Univariate regressions of post-IPO stock performance on accrual measures. Finally, in section 6, the results of the Multivariate regressions of post-IPO stock performance on accrual measures will be given.
4.1 Results of the regression analysis of the Modified Jones (1991) model
In chapter 3, the regression model has been discussed which will be used for this thesis, namely the Modified Jones (1991) model. For all regression analyses that I have done in this thesis I have used EViews 7. To be able to determine the discretionary accruals in equation 3 and the discretionary current accruals in equation 5, I first have to estimate the coefficients
β
1,
β
2, β
3 of the regression model in equations 2 and 4. These coefficients have been estimatedfor a combined sample of all IPO firms which went public on either the NYSE or the NASDAQ for the years 2001-2007. These results are shown in table 3. The regression outcomes vary significantly over the different fiscal years. The expected sign of
β
2 is positive, because a growth in a firm’s revenue should result in a higher TACC. The expected sign ofβ
3is negative, because a growth in a company’s power, plant and equipment should lead to a lower TACC, see also the methodology section for a more in depth explanation.
When looking at the results for the coefficient of the change in revenue minus the change in accounts receivables (the
β
2 coefficient), we can notice that for both the non-discretionaryaccruals (Panel A) and the non-discretionary current accruals (Panel B) the values are positive, which is the expected sign. The results are also significant at a 1% significance level. When looking at the results for the coefficient of the property, plant and equipment (the
β
331
Table 3
Accrual model coefficients estimated for all U.S. firms which have had an IPO in the years 2001 through 2007
Results of ordinary least squares regression of the modified Jones model and an extended version of the modified Jones used to calculate the accrual measures. TACC and TCA denote the total accruals and the total current accruals, respectively. A, REV, REC and PPE denote total assets, revenues, receivables and property, plant and equipment, respectively.
Panel A: Modified Jones (1991) model Non-discretionary accruals
TACC
it(1/A
it-1) = β
1(1/A
it-1) + β
2(∆REV
it- ∆REC
it) (1/A
it-1) + β
3PPE
it(1/A
it-1)
32
Table 3 continued
Panel B: Modified Jones (1991) model Non-discretionary current accruals
TCA
it(1/A
it-1) = β
1(1/A
it-1) + β
2(∆REV
it- ∆REC
it) (1/A
it-1)
β
1 Coefficient 2,931*** (T-statistic) (11.63)β
2 Coefficient 9.459*** (T-statistic) (152.2) N 7,225 Adjusted R2 0.81333
4.2 Results of the time-distribution of accrual measures
In Table 4, the time distribution of the accrual measures are shown from one year before the IPO has taken place to four years after the IPO. The results of total accruals are presented in panel A, the results of discretionary accruals are presented in panel B, and the results of discretionary current accruals are presented in panel C. I use a two tailed student t-test to determine if the mean is significantly different from zero.
Panel A shows that for the NYSE sample the total accruals are on average negative for all years relative to the IPO-year. In all years the mean of the total accruals is significantly different from zero. The median for total accruals is also negative in all years. For the NASDAQ sample the total accruals are on average negative for all years relative to the IPO-year. In all of these years the total accruals are significantly different from zero. The median for the total accruals is also negative in all years.
Panel B shows that for the NYSE sample the discretionary accruals are on average negative for all years relative to the IPO-year. In all years the mean of the discretionary accruals is significantly different from zero. The median for discretionary accruals is also negative in all years. For the NASDAQ sample the discretionary accruals are on average negative for all years relative to the IPO-year In all of these years the discretionary accruals are significantly different from zero. The median for the total accruals is also negative in all years.
Panel C shows that for the NYSE sample, the discretionary current accruals are on average negative for all years relative to the IPO-year In all years the mean of the discretionary current accruals is significantly different from zero. The median for discretionary current accruals is also negative in all years. For the NASDAQ sample the discretionary current accruals are on average negative for all years relative to the IPO-year In all of these years the discretionary current accruals are significantly different from zero. The median for the total accruals is also negative in all years.
34
35
Table 4
Time-distribution of accrual measures from years -1 to 4 relative to the IPO-year (year 0). Panel A shows the values for Total accruals, Panel B shows the values for Discretionary accruals, and panel C shows the values for Discretionary current accruals.
Panel A: Total Accruals
T -1 0 1 2 3 4
IPO firms NYSE:
Mean -0.080 -0.068 -0.064 -0.065 -0.075 -0.070
Median -0.064 -0.059 -0.054 -0.048 -0.053 -0.053
N 732 777 756 714 662 596
Mean p-value 0.000 0.000 0.000 0.000 0.000 0.000
IPO firms NASDAQ:
Mean -0.139 -0.070 -0.095 -0.088 -0.106 -0.111
Median -0.108 -0.065 -0.066 -0.076 -0.088 -0.080
N 563 618 670 629 589 545
Mean p-value 0.000 0.000 0.000 0.000 0.000 0.000
36
Table 4 continued
Panel B: Discretionary accruals
T -1 0 1 2 3 4
IPO firms NYSE:
Mean -0.044 -0.028 -0.033 -0.045 -0.045 -0.030
Median -0.017 -0.013 -0.022 -0.029 -0.021 -0.014
N 492 686 695 679 639 573
Mean p-value 0.000 0.002 0.000 0.000 0.000 0.000
IPO firms NASDAQ:
Mean -0.186 -0.124 -0.127 -0.108 -0.106 -0.086
Median -0.142 -0.113 -0.092 -0.085 -0.088 -0.069
N 505 591 652 615 565 532
Mean p-value 0.000 0.000 0.000 0.000 0.000 0.000
37
Table 4 continued
Panel C: Discretionary current accruals
T -1 0 1 2 3 4
IPO firms NYSE:
Mean -0.587 -0.574 -0.878 -0.780 -0.718 -0.619
Median -0.327 -0.345 -0.597 -0.453 -0.483 -0.376
N 639 703 686 655 618 561
Mean p-value 0.000 0.000 0.000 0.000 0.000 0.000
IPO firms NASDAQ:
Mean -1.104 -1.260 -1.223 -0.923 -0.773 -0.832
Median -0.879 -1.113 -1.018 -0.799 -0.627 -0.619
N 409 542 595 564 526 503
Mean p-value 0.000 0.000 0.000 0.000 0.000 0.000
38
4.3 Results of the regression analysis of the effects of accruals in a given year on future accruals
The results of the regression analysis of the effects of accruals in the IPO year on future accruals are shown in table 5. In panel A, the results of total accruals are presented, in panel B the results of discretionary accruals are presented, and finally in panel C the results of discretionary current accruals are presented. As explained in the literature and methodology sections, the expected value for
β
1 is negative, because it is expected that accruals should always be reversed at some point in the future.Panel A shows that the for the NYSE IPO sample total accruals in the years 1 to 4 after the IPO are all positively related to total accruals in the IPO year. The total accruals in the first and fourth year are not significantly related to total accruals in the IPO year, while the total accruals in the second year are significantly related at a 10% significance level and the total accruals in the third year are significantly related at a 1% significance level. For the NASDAQ IPO sample the third year is negatively and significantly related to the total accruals of the IPO year at a 1% significance level. The total accruals in the first, second and third year are positively related to the total accruals in the IPO year, of these the first and fourth year are statistically significant at a 1% level and the second year is not statistically significant.
39
accruals in the IPO year. Of these the second year is insignificant, the third year is statistically significant at a 10% level and the fourth year is statistically significant at a 1% level.
40
Table 5
Regression analysis of effects of current accruals on future accruals
In this table the results of the ordinary least squares (OLS) of future accruals ate the time of the IPO are shown. TACC and DACC stand for total accruals and discretionary accruals respectively. In panel A the total accruals are shown, in panel B the total accruals scaled by lagged total assets are shown, in panel C the discretionary accruals are shown, and finally in panel D the discretionary accruals scaled by lagged total assets are shown.
Panel A: Total accruals
TACC
it+x(1/A
it) = β
0+ β
1TACC
it(1/A
it-1)
X 1 2 3 4
IPO firms NYSE:
β
0 Coefficient -1.528 -0.173*** -0.076*** -0.646 (T-statistic) (-1.592) (-3.300) (-10.28) (-1.594)β
1 Coefficient 0.016 0.010* 0.004*** 0.021 (T-statistic) (0.153) (1.800) (4.668) -0.288 N 738 691 638 823 Adjusted R2 -0.001 0.003 0.032 -0.002IPO firms NASDAQ:
41
Table 5 continued
Panel B: Discretionary accruals
DACC
it+x(1/A
it) = β
0+ β
1DACC
it(1/A
it-1)
X 1 2 3 4
IPO firms NYSE:
β
0 Coefficient -0.326 -0.098*** -0.038* -0.698 (T-statistic) (-1.568) (-3.393) (-1.829) (-1.517)β
1 Coefficient 0.177*** -0.006*** 0.001 0.003 (T-statistic) (13.16) (-3.058) (0.658) (0.115) N 658 611 566 502 Adjusted R2 0.208 0.014 -0.001 -0.002IPO firms NASDAQ:
42
Table 5 continued
Panel C: Discretionary current accruals
DCA
it+x(1/A
it) = β
0+ β
1DCA
it(1/A
it-1)
X 1 2 3 4
IPO firms NYSE:
β
0 Coefficient -0.259 -1.193*** -0.875*** -1.385*** (T-statistic) (-0.465) (-7.056) (-6.518) (-2.956)β
1 Coefficient 1.436*** 0.178*** 0.025 0.022 (T-statistic) (16.48) (6.821) (1.238) (0.315) N 723 676 620 554 Adjusted R2 0.273 0.063 0.001 -0.002IPO firms NASDAQ:
β
0 Coefficient -2.047*** -20.48 -1.156*** -0.845*** (T-statistic) (-6.529) (-0.941) (-7.046) (-6.432)β
1 Coefficient 0.062*** 0.024 0.002* -0.002*** (T-statistic) (24.54) (0.192) (1.745) (-2.705) N 629 581 529 482 Adjusted R2 0.489 -0.002 0.004 0.01343
4.4 Results of the time-distribution of post-IPO stock performance
Earnings management on itself does not have to have a negative influence on the long run stock performance of an IPO firm. If investors are able to correctly interpret the financial statements of a firm going public, they should be able to account for any earnings manipulation by lowering the offer price for the IPO firm or would otherwise discount the stock price to reflect for this. As explained in the methodology section, to determine the impact on the IPO firm’s long-run stock performance I will use the annualized market-adjusted buy-and-hold return, which equals the net return of a stock including all dividend payouts. In table 6 I present the results of the time-distribution of IPO firms for the annualized market-adjusted return in the four years after the IPO.
For the NYSE IPO sample the mean is positive in all four years and in the first and second year, the mean is also statistically significant at a 1% level, while in the third and fourth year the mean is not significantly different from zero. This means that firms going public on the NYSE outperform the non-IPO firms on the NYSE in the first and second year after the IPO. For the NASDAQ IPO sample the mean is also positive in all four years. In the first and third year the mean is statistically significant at a 1%, in the second year the mean is statistically significant at a 5% level, in the fourth year the mean is not statistically different from zero. This means that firms going public on the NASDAQ outperform the non-IPO firms on the NYSE in the first, second and third year after the IPO.
44
Table 6
Time-distribution of post-IPO stock performance, from years 1 to 4 relative to the IPO-year
In this table the time-distribution of the stock performance in the years after the IPO are presented, for the years 1-4 relative to the IPO-year year (year 0). The stock performance is determined using the annualized market-adjusted buy-and-hold return, which equals the net return of a stock including the dividend payouts in the estimated period. The mean p-value has been calculated using a two-tailed student t-test.
1 2 3 4
IPO firms NYSE:
Mean 0.095 0.085 0.016 0.017
Median -0.101 -0.074 -0.138 -0.051
N 795 772 738 686
Mean p-value 0.007 0.008 0.485 0.485
IPO firms NASDAQ:
Mean 0.138 0.311 0.196 0.040
Median -0.062 -0.114 -0.112 -0.096
N 692 655 614 564
Mean p-value 0.001 0.041 0.001 0.169
45
4.5 Results of the Univariate regressions of post-IPO stock performance on accrual measures
In table 7 I present the results of the univariate regression analysis of annualized market-adjusted buy-and-hold returns on the accruals measures. I do so to further test the influence of earnings management before an IPO on the long-run stock performance of firms. I will look at the effects of the total accruals, the discretionary accruals and the discretionary current accruals in the IPO year on the annualized market-adjusted buy-and-hold return in the first, second, third and fourth year after the IPO. In table 7 ‘x’ stands for the number of years after the IPO. From here on I will use the term abnormal return to indicate the annualized market-adjusted buy-and-hold returns.
Panel A shows that for the NYSE IPO sample the abnormal return in the first year is negatively related to the total accruals in the IPO year, but not statistically significant. The abnormal return in the second, third and fourth year is positively related to the total accruals in the IPO year, but also not statistically different from zero. For the NASDAQ IPO sample the abnormal return in all four years are positively related to the total accruals in the IPO year, however none of the relationships are statistically different from zero.
Panel B shows that for the NYSE IPO sample the abnormal return in the first and fourth year are negatively related to the discretionary accruals in the IPO year, but not statistically significant. The third and fourth years are positively related to the discretionary accruals in the IPO year, but also not statistically significant. For the NASDAQ IPO sample the abnormal return in second year is negatively related to the discretionary accruals in the IPO year. The abnormal return in the first, third and fourth year are positively related to the discretionary accruals in the IPO year, of which the third year is statistically significant at a 1% level, while the first and fourth year are not significant.
46
current accruals in the IPO year, although none of them are statistically significant and all coefficients are very small.
47
Table 7
Univariate regressions of post-IPO stock performance on accrual measures
AR
it+x= β
0+ β
1Accrual
it+
itResults of ordinary least squares regression of annual buy-and-hold stock returns on different accrual measures. AR denotes the annual market-adjusted buy-and-hold stock returns. TACC, DACC, and DCA denote the total accruals, total discretionary accruals, and discretionary current accruals in the IPO year, respectively.
Panel A: Total accruals
X 1 2 3 4
IPO firms NYSE:
β
0 Coefficient 0.084** 0.069** 0.011 0.016 (T-statistic) (2.349) (2.282) (0.496) (0.855)β
1 Coefficient -0.001 0.000 0.003 0.005 (T-statistic) (-0.373) (0.149) (1.101) (1.321) N 759 736 704 654 Adjusted R2 -0.001 -0.001 0.000 0.001IPO firms NASDAQ:
48
Table 7 continued
Panel B: Discretionary accruals
X 1 2 3 4
IPO firms NYSE:
β
0 Coefficient 0.075** 0.062* 0.000 0.01 (T-statistic) (2.073) (1.893) (0.023) (0.517)β
1 Coefficient -0.003 0.002 0.000 -0.001 (T-statistic) (-1.166) (0.854) (-0.326) (-0.786) N 685 661 632 585 Adjusted R2 0.001 0.000 -0.001 -0.001IPO firms NASDAQ:
49
Table 7 continued
Panel C: Discretionary current accruals
X 1 2 3 4
IPO firms NYSE:
β
0 Coefficient 0.099*** 0.065** 0.003 0.003 (T-statistic) (2.620) (2.055) (0.143) (0.146)β
1 Coefficient 0.006 -0.005 -0.002 -0.007* (T-statistic) (1.013) (-1.059) (-0.546) (-2.582) N 743 720 690 643 Adjusted R2 0.000 0.000 -0.001 0.009IPO firms NASDAQ:
β
0 Coefficient 0.117*** 0.312* 0.143** 0.033 (T-statistic) (3.252) (1.874) (2.387) (1.066)β
1 Coefficient 0.000 0.000 -0.001* 0.000 (T-statistic) (0.208) (0.111) (-1.884) (0.296) N 636 596 552 498 Adjusted R2 -0.002 -0.002 0.005 -0.00250
4.6 Results of the Multivariate regressions of post-IPO stock performance on accrual measures
In table 8 I present the results of the multivariate regression analysis of annual market-adjusted buy-and-hold return on the accruals measures. I do so to fully test the influence of earnings management before an IPO on the long-run stock performance of firms, while also controlling for other factors which may influence the annual buy-and-hold return, namely the firm’s leverage, the firm’s market capitalization, and the firm’s book-to-market ratio. I will look at the effects of the total accruals, the discretionary accruals and the discretionary current accruals in the IPO year on the market-adjusted buy-and-hold return in the first, second, third and fourth year after the IPO. In table 8 ‘x’ stands for the number of years after the IPO. Panel A shows the regression outcomes for the NYSE IPO sample of the abnormal return on total accruals and the control variables. The abnormal return in all four years is negatively related to the total accruals in the IPO year, of these years the second year is statistically significant at a 1% level, while the years 1, 3 and 4 are not significant. When looking at the control variables, Leverage has a negative effect on the abnormal return in the first and fourth year, of which the first year is statistically significant at a 1% level, while the fourth year is not significant. The effects of Leverage in the second and third year on the abnormal return are both positive and not significant. Market capitalization has a negative effect on the abnormal return in the first and second year, and a positive effect in the third and fourth year. Of these results none of the coefficients are statistically significant. The book-to-market coefficient is positive in the first, second and fourth year, of which the first year coefficient is statistically significant at 1% level, while the second and fourth year are not statistically relevant. In the third year the coefficient is negative and not significant. The adjusted R2 of the regressions is 0.020 in the first year, 0.026 in the second year, 0.000 in the third year and -0.003 in the fourth year. These results indicate that earnings management in the form of total accruals in the IPO year has a negative impact on the long-run stock performance of NYSE firms in the second year after the IPO. In the other years after the IPO I have not found any effect total accruals have on a firm’s performance. This means that the IPO firms on the NYSE do underperform compared to their non-IPO counterparts in the second year after going public.
