1 | P a g e
Master Thesis:
Did earnings management benefit insiders in firms’ going private
transactions?
Di Wu
August, 2015
Thesis supervisor: dr. L. Zou
MSc Business Economics
University of Amsterdam
2 | P a g e
Statement of Originality
This document is written by Di Wu who declares to take full
responsibility for the contents of this document.
I declare that the text and the work presented in this document is
original and that no sources other than those mentioned in the text
and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the
supervision of completion of the work, not for the contents.
3 | P a g e
Abstract
Managers have the flexibility to use judgement in financial reporting and in structuring transactions to alter financial reports to either mislead stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on reported accounting numbers. This is when earnings management is applied. In previous literature, downward earnings management has been found in going private transactions as managers have incentive to understate earnings in an attempt to acquire a firm at a lower price. However, there is no conclusion whether those downward earnings management benefited insiders. This paper examines whether earnings management benefited insiders in going private transactions with a sample of 214 firms in LBOs and 60 firms in MBOs during 2004-2014. I isolate leveraged buyouts (LBOs) from management buyouts (MBOs) because managers anticipating LBOs have the incentive to overstate earnings in an attempt to secure external financing. Jones’ (1991) cross-sectional model is used to compute discretionary current accruals (DCA) -which is the proxy for earnings management-and the result indicates downward earnings management is present with negative mean/median of DCA in the year preceding the public announcement of going private transactions. Simultaneously, negative mean abnormal returns have been found during event period (-365,-60) using different benchmarks (CRSP value-weighted market-adjusted return, Fama-French return, matching firms return). The relation between downward earnings management and negative abnormal return has been tested and confirmed within different DCA quantiles, by sample partitions and event-time cross-sectional regression. Those results are consistent and indicate that downward earnings management effectively influences stock price and the extent of downward earnings management predicts the magnitude of negative abnormal returns. The difference between MBOs and LBOs is significant in the extent of earnings management and the magnitude of abnormal return. And firms in LBOs show less negative DCA and abnormal returns for various specifications on average. In summary, those results in this paper provide evidence supporting that downward earnings management was applied by most of the firms and insiders benefited from those downward earnings management in going private transactions. Hence, downward earnings management is useful to insiders in going private transactions at the risk of manipulating financial data and the theory of semi-strong efficiency of capital market is strengthened.
4 | P a g e
Table of Contents
1. Introduction
2. Literature Review
3. Data and Model Design
4. Empirical Result
5. Robustness Check
6. Summary
7. References
8. Statistic Tables
5 | P a g e
Did earnings management benefit insiders in firms’ going private
transactions?
1. Introduction
Earnings management is a widespread problem and has recently received considerable attention by regulators and the popular press (Biao Xie, Wallace N. Davidson III and Peter J. DaDalt 2002). Under both Generally Accepted Accounting Principles (GAAP) and International Financial Reporting Standards (IFRS), managers have the flexibility to use judgement in financial reporting and in structuring transactions. This can be used to alter financial reports to either mislead stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on reported accounting numbers (Healy and Wahlen, 1998). This is when earnings management happens. Table 1 summarizes how earnings management is realized by the choice of accounting methods, application of accounting methods, and timing of asset acquisitions and dispositions.
There is no universal measurement on earnings management but previous studies use discretionary accruals to measure earnings management. Downward earnings management is measured by negative discretionary accruals while upward earning management is measured by positive discretionary accruals. Reported earnings consist of cash flows from operations and accounting adjustments called accruals. Under both Generally Accepted Accounting Principles (GAAP) and International Financial Reporting Standards (IFRS), firms use accrual accounting which ‘‘attempts to record the financial effects on an entity of transactions and other events and circumstances that have cash consequences for the entity in the periods in which those transactions, events, and consequences occur rather than only in the period in which cash is received or paid by the entity.”(FASB 1985, SFAC No. 6, paragraph 139). The nature of accrual accounting gives managers a great deal of discretion in determining the actual earnings a firm reports in any given period, which is measured by discretionary accruals. Management has considerable control over the timing of actual expense items and can also alter the timing of recognition of revenues and expenses. However, it doesn’t mean violation to any principles and standards (Teoh et al., 1998a). In this paper, I follow Jones’ (1991) model to decompose discretionary accruals from total accruals and use discretionary current accruals as the proxy for earnings management as entrepreneurs have more discretion over short-term than long-term accruals (see, e.g., Guenther,1994).
In most cases, when managers are compensated both directly and indirectly depending on a firm’s earnings performance, they will have self-interest to give the appearance of better performance through upward earnings management. In contrast, when managers’ benefit is opposite to firms’ earnings performance, they tend to give the appearance of worse performance through downward earnings management. The latter cases have been found in going private transactions in which managers have clear incentive to understate earnings in
6 | P a g e an attempt to acquire the firm at a lower price (DeAngelo 1986, Perry and Williams 1994 and Wu 1997). In those studies, downward earnings management has been found in the year prior to going private transactions. Their studies offer convincing evidence of downward earnings management in the year preceding the public announcement of management’s intention to bid for the control of the company.
However, there is no empirical evidence showing that managers benefited from downward earnings management or downward earnings management effectively lowered firms’ stock price in going private transactions. If, as I hypothesize, investors are unable to understand fully the extent to which firms engage in earnings management by adjusting either the past or the future, lower reported earnings would translate directly to a lower stock price, which is the basis of a purchasing price. If downward earnings management existed and stock price was lower accordingly in going-private transactions, downward earnings management is meaningful to managers and they can gain from lower purchase price. At the same time, it shows that the capital market is semi-strong efficient as share prices adjust to publicly available new information such as earnings. In opposite, if downward earnings management existed but stock price was not lower accordingly or there is no relation between downward earnings management and lower stock price in going private transactions, then downward earnings management is meaningless to managers as they can’t gain from lower purchase price. If so, it can support that capital market’s strong-form efficiency as share prices reflect all public information (earnings in financial report) and private information (the extent of discretionary accruals). This is the rationale behind my investigating whether insiders benefited from downward earnings management in going private transactions.
Previous studies look at the incentive of lowering purchase price in going private transactions. This paper breaks down going private transactions into management buyouts (MBOs) and leveraged buyouts (LBOs) and looks at both the incentives of lowering purchase price and both the incentives of lowering purchase price and securing external financing. Both MBOs and LBOs are initiated by management and intended to restructure corporate ownership by replacing the entire public stock interest with full equity ownership by an incumbent management group. Consequently, insiders have a strong incentive to apply downward earnings management to benefit themselves through a lower purchase price. That’s why previous studies mix LBOs with MBOs together when talk about going private transactions (DeAngelo 1986, Perry and Williams 1994 and Wu 1997). However, an LBO is fundamentally different from an MBO. Firms involved in MBOs are self-funded with internal free cash flows whereas firms involved in LBOs are significantly externally funded and need external financing from banks or other institutions to complete transactions. In order to secure external financing, managers have an incentive to manage earnings upwards to enhance prospective external financers’ perceptions of firms’ value. Thus, managers face conflicting incentives in anticipation of LBOs because they face different stakeholders while publishing their financial statements. In summary, managers in MBO transactions have the motivation to release less
7 | P a g e favorable earnings reports to equity market participants prior to an MBO in an attempt to reduce the purchase price. However, managers anticipating an LBO transaction also need to consider the assessment of the bank and other external financing institutions. Managers anticipating an LBO transaction intend to report a more favorable earnings to get a favorable loan with lower financing cost compared to MBOs. The tradeoff between purchase cost and external financing cost depends on the extent in which they rely on external financing (Fish and Louis 2008).
If earnings management in LBOs is still downward but less so than in MBOs, the incentive of securing external financing is countervailing and tempers the incentive of lowering purchase price. Inversely, if earnings management in LBOs is not downward but upward, then the incentive of securing external financing dominates the one of lowering purchase price. Hence, the difference in earnings management before MBOs and LBOs may lead to a deeper understanding of the capital market. This provides me with a reason to differentiate between MBOs and LBOs.
This paper examines whether earnings management benefited insiders in firms’ going private transactions. The sample consists of 214 LBO firms and 60 MBO firms with complete information in COMPUSTAT and CRSP. The transactions have taken place between 2004 and 2014. I develop strict criteria for sample selection with lots of hand collection and validation. Those criteria are valuable to further research on going private transitions. From the firm statistics, I observe big differences in firm size, management ownership, and market-to-book ratio between MBOs and LBOs, which is consistent with previous studies (Fish and Louis 2008). I define the earliest date to announce a transaction as the event time day 0 and the fiscal year of announcement as Year 0. The measurement of earnings management I refer to is discretionary current accruals (DCA) from the financial report preceding the public announcement of the transaction which reflects the result of whole-year earnings management and starts at least 365 days before the announcement. I refer to the stock abnormal returns during (-365,-60) relative to various benchmarks as the measurement of stock return performance. If earnings management is effective, we will observe influences on subsequent abnormal returns.
Discretionary current accruals (DCA) - which are under the control of managers and proxy for earnings management- are decomposed from total accruals following Jones’ (1991) cross-sectional peer regression. I find that these DCAs were dominantly negative as 62% sample firms applied negative DCA in the year preceding the public announcement of going private transactions. The mean DCA is -2.33% of lagged total assets for firms in MBOs and -1.54% for LBOs. Similar to the mean DCA, the median DCA of LBOs is less negative than that of MBOs. Therefore it can be concluded that incentive of securing external financing is countervailing in LBOs and tempers the incentive of lowering purchase price.
8 | P a g e The paper documents negative abnormal returns -which are computed from expected returns based on different benchmarks (CRSP value-weighted market-adjusted return, Fama-French return, matching firms return) - during the event time (-365,-60) both for the mean cumulative abnormal return (CAR) and the mean buy and hold abnormal return (BHAR). The negative abnormal return on average is economically large and significant. For instance, on a CRSP value-weighted market-adjusted return measurement, the mean CAR for portfolio in MBOs is -17.32% and the mean BHAR for portfolio in MBOs is -15.21% both at 1% significance level. Consistent with the findings on the DCA, portfolios with pre-LBO firms have less negative abnormal returns with a CAR of -15.36% and a BHAR of -13.49% , both at the 1% significance level. Also, firms in LBOs show less negative abnormal return across every quantile (from most aggressive quantile to most conservative quantile) and specification (CRSP value-weighted market-adjusted return, Fama-French return, matching firms return). This result supports the hypothesis that both firms in MBOs and LBOs underperform in the year preceding the transaction but firms in LBOs earn less negative abnormal returns than MBOs.
I also find the discretionary current accruals in fiscal Year -1 are good predictors of subsequent stock return performance during (-365,-60) in a wide variety of specifications. Downward earnings management effectively lowered stock prices, and the extent of earnings management predicted the magnitude of stock abnormal performance. Depending on benchmarks, firms in MBOs/LBOs that are ranked in the lowest quantile (conservative firms) with lowest pre-announcement DCA earn a cumulative abnormal return of approximately 15 to 20 percentage less than the cumulative abnormal return of sample firms that are ranked in the highest quantile (aggressive firms). The equivalent buy-and-hold return differential between those two quantiles is 16 to 37 percent. This differential performance is robust to a battery of test specifications and controls (eg. firm size, book-to-market ratio, holding period, expected return benchmark). The event-time cross-sectional regression suggests the incremental influence of DCAs on buy-and-hold abnormal returns is significant which implies investors heavily rely on earnings in firms’ valuation. However, a Wald-test tells indifference between LBOs and MBOs in influence power of DCA on stock abnormal returns.
The main contribution of this study is: first, even though previous studies have proven downward earnings management existed in going private transactions, there is no empirical research on whether earnings management actually benefited insiders. The result in this paper indicates downward earnings management is meaningful to firms in anticipating MBOs and LBOs in terms of a lower purchase price even though they are at the risk of manipulating financial data. This result also supports that capital market is semi-form efficient rather than strong-form efficient as share prices only adjust to new public information rather than both new public and private information.
9 | P a g e Second, previous studies mixed MBOs and LBOs when researching going private transactions. However, LBOs are fundamentally different from MBOs because managers also have the incentive of securing external financing and intend to manage earnings upwardly. In this study, I isolate LBOs from MBOs and examine the difference of earnings management and abnormal returns between them. The result shows a clearer picture for managers’ incentives in going private transactions. Less downward earnings management and less negative abnormal returns found in LBOs suggest that the conflicting incentives may temper each other and mitigate the wealth loss for public stockholders to some extent.
Last but not least, there is no universal definition on going private transactions, nor are there standard rules on sample selections. This problem led to mixed results for earnings management in previous studies. For instance, DeAngelo included 14 firms (out of 64) whose hostile takeover offers had been announced before MBO offers and management might have possibly taken MBO as the last resort. In this paper, I developed strict criteria with lots of hand collection when constructing the sample. I make sure all transactions were actively initiated by managers and successfully completed by delisting from stock exchange. The criteria, which are explained in greater detail later in this paper, provide guidance for further studies on going private transactions.
The remaining of this paper is arranged as follows. Section 2 discusses the current literature on the topic. Section 3 describes data and model design. Section 4 reports empirical results. Section 5 presents robustness checks. Section 6 summarizes this study.
2. Literature Review
In going private transactions, the entire offering price to public shareholders consists of pre-offer market price and premium. DeAngelo et al. (1984) found that management paid 56 percent average premium above the pre-offer stock price in cash going –private proposals while Bradley and Jarrell found 49-56 percent average premiums in the inter-firm cash tender offer. The pre-offer stock price plays a significant role in determining overall offering price.
Open market stock prices do not by themselves satisfy the judicial concept of fair value, in part because of the potential for insider-managers to control the flow of information to the capital market (see, e.g., Brudney and Chirelstein [1974] and Chazen [1981]). The fairness of the offering price is validated by court and investment bank. Table 2 illustrates the alternative approaches used by the investment bank in valuation. Comparable acquisition approach emphasizes the importance of pre-offer market price again as it determines the basis of premium. Also, comparable firm’s approach is to capitalize the current firm’s per-share earnings at an average price/earnings ratio for comparable publicly-traded firms ( DeAngelo 1990). The reason DCF is not the sole approach in valuation by investment bank is because of
10 | P a g e inside management assumption and its sensitivity to cash flow ( DeAngelo ‘1990). At the same time, the court also relies on earnings to evaluate the fairness of offering price. (Perry and Williams 1994). Hence, management has a strong incentive to lower earrings prior to going private transaction with the abnormal lower stock price at the same time.
Bevis Longstreth (1983) raised the issue management is dealing with itself in a management buyout because they act on both sides of the transactions in a management buyout. The fiduciary capacity requires them to seek for the best deal for shareholders, but its proprietary capacity favors them to purchase the company at a price beneficial to management. Longstreth (1986) pointed out a variety of techniques for managers to adversely affect the open-market stock price. For example, they can report upward-biased expense estimates, defer revenue recognition, accelerate expense recognition, take direct write-offs, over accrue expected losses, expense items that they would otherwise capitalize, etc. Managers can also choose simply to delay profitable investments until after the acquisition.
When present earning management prior to going private transactions, different measurements were used in previous literatures. DeAngelo (1986) developed an empirical approach that uses firm’s discretionary accrual as a proxy for earning management. DeAngelo’s methodology estimates the extent of such manipulation as the total accrual change compared to benchmark year. DeAngelo’s study employs a firm-specific time series benchmark assuming that the non-discretionary accrual doesn’t change materially across time. Susan E. Perrya, Thomas H. Williams (1994) followed Jones’ (1991) model that used a cross-sectional regression-based expectations model and incorporated the economic activities of the firm during the test year. Following Jones’ (1991) model, Susan E. Perrya, Thomas H. Williams included two explanatory variables - changes in revenue and the level of gross property, plant, and equipment. These two variables control the change in non-discretionary accrual due to economic activities. Jones’ (1991) model requires a control sample group based on industry (proxied by SIC) and size (proxied by sales) in the period preceding to transactions.
Y. Woony Wu (1997) used a variety of measurements to validate earnings management in going private transactions. Those measurements include change in income, change in industry-adjusted income and Jones’ (1991) model. Different from the pooled measurement of earning management, Wright and Guan (2004) returned to the basic and examined from accounting policy choices.
To validate the motivation of earnings management, all the previous literature examined the significant level of earnings management using different measurements. Every measurement has its limitation and econometrical problems. Jones’ (1991) is recommended by most of the literatures because it controls for the industry’s and firms’ average situation. I argue that the
11 | P a g e ending goal of earnings management by managers in the transactions they involved is to benefit themselves by getting a favorable offer price through influencing open market price. So if we find the negative abnormal returns during the year preceding to going private transactions as well as the relation between downward earnings management and negative abnormal returns, then the evidence for downward earnings manipulation is strengthened and validated. Thus, test on the relation between earnings management and abnormal return could be a complementary evidence for managers’ incentive to take downward earnings management. Furthermore, study on capital market reaction to earnings management are the way to test the efficiency of the market. If capital market is semi-strong-form efficient, share prices only adjust to the publicly new information and managers could earn on private information. If capital market is strong-form efficient, share prices adjust to both publicly and privately new information and managers are not able to earn on private information.
Due to different measurements with different assumptions, empirical results are mixed among pervious researches. DeAngelo (1986)’s study investigated the accounting decision made by managers of 64 corporations who had the proposal to delist from the public by going private during 1973-1982. The result presented in DeAngelo (1986)’s paper didn’t favor the hypothesis that downward earnings management existed before the management buyout and understatement on earnings was not found in the test. The only plausible explanation DeAngelo pointed out is that earning was pretty important to attract scrutiny by public stock holders and their financial advisors. Earning manipulation could result in large personal wealth loss through allegations of fraud and federal securities law violations. However, Perry and Williams (1994) found empirical support in terms of earnings management preceding management buyout. Perry and Williams’s analysis of a sample of 175 management buyouts (with proposal) during 1981-1988 favored the hypothesis that downward manipulation of discretionary accruals existed in the year preceding the public announcement with the intention to bid for control of the company. After careful comparison with DeAngelo(1986)’s study, Perry and Williams suggested that the principle difference is from the sample instead of the measurement method. For instance, DeAngelo included 14 firms (out of 64) whose hostile takeover offers had been announced before MBO offers and management might passively take MBO as the last resort. In that case, earnings might be inflated to dissuade their shareholders from supporting the takeover Easterwood (1997). Wu (1997) isolated the cases of third-party takeovers and controlled industry effect in the sample. Wu examined the 87 management buyout cases during 1980-1987 and found incomes significantly dropped right in the year before the MBO announcement. Wu (1997)’s study pointed out there was a limitation on different earning management measurements. Wu (1997)’s study initially studied on the abnormal return during the period of [-400,-60] before the announcement and found CAR [-400,-60] was significantly negative. This result is consistent with downward earnings management preceding MBO. To confirm the relationship between earnings management and negative CAR [-400,-60], Wu (1997) ran leaner regression without any control variables and the result confirmed the leaner relationship between earnings
12 | P a g e management (proxied by standardized change in income) and CARs preceding management buyout. However, Wu (1997) didn’t convince us there were no omitted variables in the simple learner regression between earnings management and abnormal return. So the result was not convincible and solid.
Another issue is there is a systematic difference between pure management buyout (MBO) and leveraged buyout (LBO) in going private transactions. The difference is ignored by previous studies when research on earnings management. In a pure management buyout, as part of the investment group, the incumbent management seeks the complete control of the surviving firm by internal financing without external financing. In leveraged buyout, LBO firms obtain external resource by leveraging current assets to exchange additional financing. The resource could be from secured bank loans that may be syndicated. Furthermore, subsequent external debt financing could be attained from institutions such as venture capital, pension funds or insurance companies or through public offerings. After raising the cash, the investment group purchases the firm depends on the transaction type. For pure management buyout, the transaction could be implemented through asset sale, merger, the tender offer and reverse stock split. For leveraged buyout, it is typically structured as either asset sales or mergers (HARRY DEANGELO and LINDA DEA NGELO, 1984).
Evidence shows both firms before pure management buyout and leveraged buyout are significantly undervalued (Mittoo and Ng, 2014). However, they are significantly different before going private. Management ownership is significantly higher in the corporations before pure management buyout than the one in the firms before leveraged buyout. LBO firms are considerately larger than pure MBO companies in terms of size and revenues. MBO tends to have higher cash/transaction ratio which not only means higher internal financing capability but also presents the lower portion to be purchased in the transaction compared to leveraged buyout.
Previous studies conclude that managers have the incentive to manipulate earning downward to lower the purchase price in management buyout transactions. But Fisher and Louis (2008) hypothesizes managers also considers the financing cost from an external resource in leveraged buyout. Higher earnings presents better performance and would decrease external financing cost. Managers involved in leveraged buyout will trade off between the gain from purchasing cost reduction and the expense from borrowing external money. Dividing the LBO firms into different quantiles according to the extend firms relies on external financing, Fisher and Louis (2008) finds firms involved in leveraged buyout have earnings less negative than pure MBO. Previous studies suggest managers would like to reduce the market perception of firm’s value to reduce the purchase cost. The result by Fisher and Louis (2008) presents the conflicting incentive existed in external financing. They find firms used less negative accruals when more external financing was relied on. Also, as fixed assets could be used as collaterals, higher fixed assets ratio could offset the impact from external financing. The positive effects
13 | P a g e of external financing on earnings management decrease as the amount of fixed assets increases. Fisher and Louis (2008)
To explore how management ownership and control in firms play a significant role in determining offering price, we find evidence support studying on premium (the positive premium from market for going private) from Mittoo and Ng (2014). They argued that ownership structure was likely to influence both the decision to going private and the premium paid in MBO/LBO. They used a two-stage regression model in a sample of 321 US going private transactions during 2000-2011. The empirical result revealed ownership structure (proxied by the percentage of voting control held by directors or family members of directors) influenced the decision to going private as well as the premium paid to public shareholders. In Mittoo and Ng (2014)’s study, the premiums (measured as the final offer price-trading price x day before the announcement scaled by the price) differed significantly between leveraged buyout and pure management buyout. For the three-day premium, the mean of MBO firms is 9.65% higher than the mean of LBO significant at an average level. This conclusion is consistent with our hypothesis that management has strong However, no significant difference is observed for the CARs of 1,3,5,10 days’ windows between LBO and MBO.
This empirical result implies, even though management has different bargaining power on offering price (open market price plus premium), the reaction to going private from the market is the same. Recall that the positive abnormal return around the announcement is because of productive gains which are believed to be realized after MBO. The same reaction from the market means public shareholders evaluate the gain from LBO and MBO the same. Then I argue the information for LBO and MBO is only different in the transaction type but not different in the extent of transparency. Then I believe no matter how different the earning is distorted by earning management between LBO and MBO, the magnitude of market reaction to one unit change of earnings management is the same for LBO and MBO as earnings information is transparent to shareholders at the same level. Thus, I hypothesize there is difference of the relation between downward earnings management and negative abnormal returns between MBOs and LBOs.
Hence, consistent with previous study, I have hypotheses as below:
1) Downward earnings management are applied by firms in both leveraged buyouts (LBOs) and management buyouts (MBOs). On average, firms in LBOs report less extent of downward earnings management than firms in MBOs.
2) As the result of downward earnings management, abnormal returns before the public announcement of transactions are negative for both LBOs and MBOs on average. On average, firms in LBOs earn less negative abnormal returns than firms in MBOs.
14 | P a g e 3) The relation between downward earnings management and stock return
underperformance exists in both LBOs and MBOs and the extent of downward earnings management predicts the magnitude of stock return underperformance. The effect of earnings management on stock return underperformance is indifferent between LBOs and MBOs.
3. Data and Research Design
3.1 MBO/LBO Sample
There is no universal definition and criterion for going private transaction as well as for MBOs/LBOs. Previous studies use different methods and sources to construct their samples, including announcements of going private on the press, classification by other parties without a clear statement on criterion and SEC filling data. Our study is clearly targeting on cases where strong management incentive exists, so we rule out cases in which managers are passively involved like hostile takeovers. At the same time, to separate our study from going dark, we ensure that all sample companies were successfully deregistered from filing requirement. Also, we need to validate the role of external financing in the transaction to isolate leverage buyouts from management buyouts.
With substantial hand collection and validation, our sample during 2004-2014 is constructed following procedures as below:
1) All SEC Form 13E-3 followed by SEC Form 15 from January 2004 to December 2014 from Thomson One. SEC rule 13E-3 requires firms to file for the procedures in going private transactions, and Form 15 indicates the completion of deregistration.
2) Validate all sample companies ever existed and have been successfully deregistered from the stock market.
3) To validate manager’s incentive in those transactions, the Wall Street Journal and company proxy report confirm the incumbent management made an offer and had an equity interest in the newly private company.
4) To rule out passive involvement from managers, we confirm all those MBO proposals are before third-party takeover proposals.
5) To isolate LBO from MBO, we validate whether external financing is involved in going private proposal through filing 13e-3 and proxy statement.
The original sample constitutes of 350 successfully going private transactions. After we validate the financial information in COMPUSTAT and stock information in CRSP, 274 firms are remained as the sample. Table 3 provides the statistics summary for the 274 sample by transaction type.
15 | P a g e Panel A shows the distribution of sample firms across industries. The presentation across major industry groups is reasonably broad. Sample firms summary statistics are reported in Panel B. Both from assets and revenue, LBO firms are larger than MBO firms. The average assets are $450.22 million for MBO firms but $878.21 million for LBO firms. The mean revenue is $398.58 for pure MBO firms but $999.12 for LBO firms. Both the mean difference in assets and revenue are significantly different from zero at 5% significant level. As to the management ownership, the percentage of voting shares held by management in MBO is significantly higher than LBO. In the firms involved in MBO activities, management and the family members of management averagely hold over 50% of voting shares whereas the ones in LBO only hold less than 28% on average. This character implies management could have stronger power and capacity in managing earnings in MBO firms compared to LBO firms. For leveraged situation, there is no visible difference on Debt/Asset ratio and Long-term debt/equity ratio. This result is a little surprising to me as I hypothesize LBO firms might have lower debt/equity ratio. The ROA equals to the net income scaled by total assets. Both MBO and LBO firms are obviously negative in the fiscal year prior to the announcement. However, LBO firms are less negative than MBO in average. This result is consistent with our hypothesis that LBO firms have less downward earning management in MBO transactions. And when I look at the market to book ratio, I find LBO firms have the higher market to book ratio compared with MBO firms. And the variance is different from zero at 1% significant level. This implies LBO firms may need more external financing due to the higher market price per asset.
3.2 Identify the even time.
There is no standardized answer to when management starts to manipulate earnings before MBOs and LBOs. Previous studies argued that managers often planed going private transactions for a year or more years preceding the date of the public offer. Also, there is the case that MBOs was planned over a much shorter time interval after stock market crash Perrya and Williams (1994). Hence, I argue that managers may manipulate the earnings at least one year before MBOs/LBOs no matter how long they planned the transactions. So I believe, earnings preceding the announcement of MBOs is the most reliable data for testing earnings management in going private transactions. Then I define the earliest date to announce MBOs/LBOs is our event time day 0 and the fiscal year of announcement is Year 0. The measurement of earnings management I refer to is discretionary current accruals from the fiscal financial report preceding the public announcement of transactions which reflects the result of whole-year earnings management and starts at least 365 days before announcement. I refer to stock abnormal returns relative to a variety of benchmarks during (-365,-60) as the measurement of stock return performance. If earnings management is effective, we would observe influences on subsequent abnormal returns.
16 | P a g e Reported earnings consist of cash flows from operations and accounting adjustments called accruals. Among total accruals, some accrual adjustments are appropriate and necessary based on the business conditions but some accruals are discretionary by insiders with the desire for a lower short-term stock price in our case. What we refer to as earnings management are discretionary accruals. However, discretionary accruals are very difficult to be inferred from total accruals even through non-discretionary accruals could be expected by investors. Thus, we need a model to decompose accruals into two components, one that is dictated by firm and industry conditions and one that is presumed to be managed by the entrepreneur.
I first decompose total accruals into current accruals and long-term accruals. As entrepreneurs have more discretion over short-term than long-term accruals (see, e.g., Guenther (1994), I believe short-term accruals is more valuable to investigate insiders earnings management preceding transactions. Following Teoh, Wong, and Rao (1998), we use an extension of the cross-sectional Jones’ (1991) model for this purpose. In essence, total current accruals could be calculated using financial statements information and non-discretionary current accruals are expected accruals from a cross-sectional Jones’ (1991) model and the discretionary current accruals are the residuals of regression. Hence, downward earnings management is measured by negative DCA while upward earnings management is measured by positive DCA.
Total accruals (AC) consist of discretionary current accruals (DCA), nondiscretionary current accruals (NDCA), discretionary long-term accruals (DLA), and nondiscretionary long-term accruals (NDLA). Total accruals are measured using COMPUSTAT annual item numbers in parentheses as follows:
𝐴𝐶𝑡 = 𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒𝑡(172) − 𝐶𝑎𝑠ℎ 𝐹𝑙𝑜𝑤 𝑓𝑟𝑜𝑚 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 (308) (A1)
Current accruals (CA) are defined as the difference between the change in noncash current assets and the change in operating current liabilities as follows:
𝐶𝐴𝑡= ∆[𝑎𝑐𝑐𝑜𝑢𝑛𝑡𝑠 𝑟𝑒𝑐𝑒𝑖𝑣𝑎𝑏𝑙𝑒 𝑡(2) + 𝑖𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦𝑡(3) + 𝑜𝑡ℎ𝑒𝑟 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑎𝑠𝑠𝑒𝑡𝑠 𝑡(68)] −
∆[𝑎𝑐𝑐𝑜𝑢𝑛𝑡𝑠 𝑝𝑎𝑦𝑎𝑏𝑙𝑒 𝑡(70) + 𝑡𝑎𝑥 𝑝𝑎𝑦𝑏𝑙𝑒 𝑡(71) + 𝑜𝑡ℎ𝑟𝑒 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 𝑡(72) (A2)
Nondiscretionary current accruals (NCA) are the expected adjustments in a given firm and industry condition. Expected accruals for a firms before LBOs/MBOs in the year preceding announcements are estimated from a cross-sectional regression in that year on the changes in sales (12) scaled by total assets of last year using an estimation sample of all two-digit SIC code peers. We run following cross-sectional OLS regression to come out the estimated parameters of NCA: 𝐶𝐴𝑗,𝑡 𝑇𝐴𝑗,𝑡−1 =𝛼0( 1 𝑇𝐴𝑗,𝑡−1) + 𝛼1( ∆𝑆𝑎𝑙𝑒𝑠𝑗,𝑡
17 | P a g e Where, ∆Sales is the change in sales from last year and TA is the total assets (6). With the estimated 𝛼0 and 𝛼1, the expected current accruals or nondiscretionary current accruals are calculated as follows:
𝑁𝐷𝐶𝐴𝑖,𝑡 =𝛼̂(0 𝑇𝐴1
𝑗,𝑡−1) + 𝛼̂(1
∆𝑆𝑎𝑙𝑒𝑠𝑗,𝑡
𝑇𝐴𝑗,𝑡−1 ) (A4)
With total current accruals (CA) and nondiscretionary accruals (NDCA), discretionary current accruals are the residuals of regression and calculated as follows:
𝐷𝐶𝐴𝑖,𝑡 = 𝑇𝐴𝐶𝐴𝑗,𝑡
𝑗,𝑡−1− 𝑁𝐷𝐶𝐴𝑖,𝑡 (A5)
To obtain discretionary long-term accruals, nondiscretionary and discretionary total accruals are fist estimated in a similar manner as for current accruals. However, the difference is total accruals (AC) is used as the dependent variable and the regression includes gross property, plant and PPE as additional explanatory variables as follows:
𝐴𝐶𝑗,𝑡 𝑇𝐴𝑗,𝑡−1 =β0( 1 𝑇𝐴𝑗,𝑡−1) + β1( ∆𝑆𝑎𝑙𝑒𝑠𝑗,𝑡 𝑇𝐴𝑗,𝑡−1 ) + β2( 𝑃𝑃𝐸𝑗,𝑡 𝑇𝐴𝑗,𝑡−1) +𝜀𝑗,𝑡 (A6)
Accordingly, nondiscretionary total accruals are calculated as follows:
𝑁𝐷𝑇𝐴𝐶𝑖,𝑡=β̂0(𝑇𝐴1 𝑗,𝑡−1) + β̂1( ∆𝑆𝑎𝑙𝑒𝑠𝑗,𝑡 𝑇𝐴𝑗,𝑡−1 ) + β̂2( 𝑃𝑃𝐸𝑗,𝑡 𝑇𝐴𝑗,𝑡−1) (A7)
And discretionary total accruals are calculated as follows:
𝐷𝑇𝐴𝑖,𝑡 = 𝑇𝐴𝐴𝐶𝑗,𝑡
𝑗,𝑡−1− 𝑁𝐷𝑇𝐴𝐶𝑖,𝑡 (A8)
The difference between discretionary total accruals and discretionary current accruals is termed discretionary long-term accruals, and the difference between nondiscretionary total accruals and nondiscretionary current accruals is termed nondiscretionary long-term accruals.
3.4 Abnormal Returns
Previous studies have examined the positive abnormal return during the announcement of MBOs and they find positive stock reactions starting from 2 months prior to the announcement due to information leakage (Kaplan 1989 and Kieschnick 1989). I apply this method by extending the event window from days around announcement to 1 year earlier before the announcement. To isolate the positive reaction to MBO announcement, I define the event window to end 2 month prior to announcement. Hence, the daily event window is day (-365,-60).
18 | P a g e Buy and hold returns are frequently used by studies on stock performance due to long-term performance is more relevant to investors. However, they are problematic because their distribution is skewed and initial small difference could be exaggerated through compounding. At the same time, overlap in time periods introduces cross-correlation problems (Fama, 1998). Therefore, I repot both cumulative abnormal returns and buy and hold abnormal returns in this paper using a variety of benchmarks.
The fist benchmark is CRSP value-weighted market-adjusted return. The Model used to calculate daily CARs is the conventional market model and the benchmark index is CRSP value-weighted market index. The estimation period ends 3 years prior to event day 0, and the estimation length is 120 days to 365 days.
The parameters of the market model are estimated as follows:
𝑟𝑖,𝑡 = 𝛼 + 𝛽𝑅𝑚,𝑡+ 𝜀𝑖,𝑡 (B1)
Where, ri,t is the daily return on the stock in day t and Rm,t is the daily return on the CRSP value-weighted index in day t. After estimating α and β using OLS, we calculate expected returns from day -365 to day -60 as follows:
𝐸(𝑟𝑖,𝑡) = 𝛼̂ + 𝛽̂𝑅𝑚,𝑡 Where, t= (-365,-60) (B2)
The second benchmark is Fama-French Return. Fama-French returns are expected returns regressed on three Fama-French Factors across estimation periods over risk-free rates. As the same, the estimation period ends before three years prior to event day (0), and the estimation length is 120 days to 365 days.
𝑟𝑖,𝑡 = 𝑅𝑓,𝑡+ 𝛾1(𝑅𝑚,𝑡 − 𝑅𝑓,𝑡) + 𝛾2 𝑅𝑠𝑚𝑏,𝑡 + 𝛾3 𝑅ℎ𝑚𝑙,𝑡+ 𝜀𝑖,𝑡 (B3)
Where, 𝑟𝑖,𝑡 is the daily return on the stock in day t , 𝑅𝑓,𝑡 is the one-month Treasury bill (risk-free) rate of return, 𝑅𝑚,𝑡 is the daily return on CRSP value weighted index, 𝑅𝑠𝑚𝑏,𝑡 is the daily return on a small-capitalization portfolio minus a large-capitalization portfolio, 𝑅ℎ𝑚𝑙,𝑡 is the daily return on a high book-to market portfolio minus a low book-to-market portfolio.
The expected return for each month on stock is computed from day -365 to day -60 using the estimated coefficients from factor regression as follows:
19 | P a g e The third benchmark I use is the performance of matched firm. Each firm is matched with firms without going private proposals from CRSP based on industry (SIC code), market capitalization by end of month -14.
Finally, mean cumulative abnormal return (the CAR on the portfolio of sample firms) is calculated by averaging the residuals of actual returns minus expected returns, and sum over the length of the event window time (-365,-60) as follows:
𝑀𝐶𝐴𝑅𝑇 = ∑ [∑𝑁𝑖=1(𝑟𝑖,𝑡− 𝐸(𝑟𝑖,𝑡))
𝑁 ]
𝑇
𝑡=0 (B5)
Accordingly, mean buy and hold abnormal return (the buy and hold abnormal return on the portfolio of sample firms) is calculated by subtracting the compounding of holding-period returns from the compounding of benchmark index returns during event window for each firm and averaging the sum for all firms.
𝑀𝐵𝐻𝑅𝑇 = ∑𝑁𝑖=1[∏𝑡=0𝑇 (1+𝑟𝑖,𝑡)−∏𝑇𝑡=0(1+𝐸(𝑟𝑖,𝑡))]
𝑁 (B6) Both in equation (B5) and (B6), ri,t is daily return and E(ri,t) is daily expected return using different benchmark: CRSP value weighed market-adjusted index, Fama French return or matching firm return. T is the event window from day -365 to day -60.
3.5 Event-Time Cross-Sectional Regression
To examine the incremental influence of discretionary current accruals on stock return abnormal performance, I follow Ritter (1991, Table X)’s regression on firms’ underperformance and add DCA in the equation as below:
𝐵𝐻𝐴𝑅𝑖,𝑡 = 𝛼 + 𝛽0𝐷𝐶𝐴𝑖,𝑡+ 𝛽1𝑀𝑘𝑡𝑅𝑒𝑡𝑖,𝑡+ 𝛽2 𝐿𝑜𝑔(𝑀𝑉𝑖,𝑡−1) + 𝛽3 𝐿𝑜𝑔 (𝑀𝑉𝑖,𝑡−1 𝐵𝑉𝑖,𝑡−1) + 𝛽4 𝐿𝑜𝑔(1 + 𝐴𝑔𝑒𝑖,𝑡−1) + 𝛽5 ∆𝐶𝑎𝑝𝐸𝑥𝑝 𝑖, 𝑡−1+ 𝛽6 ∆𝑁𝑒𝑡Income 𝑖, 𝑡−1 + 𝜀𝑖,𝑡
Where dependent variable is CRSP value-weighted market-adjusted buy and hold abnormal return compounded during event window time (-425,-60) for each firm. The control variables include dummy variables for year and industry; buy and hold return for CRSP value-weighted market return during event window time (-365,-60); the log of firm’s market capitalization by end of fiscal Year -1;the log market to book by end of fiscal Year -1;log of firm’s age by end of fiscal Year -1 ; the change of capital expenditure between fiscal Year -2 and fiscal Year -1 scaled
20 | P a g e by total assets in fiscal Year -2 and the change of net income between fiscal Year -2 and fiscal Year -1 scaled by total assets in fiscal Year -2.
I include market return because correlation between individual and market returns are expected; I include market value and market to book ratio because there is a tendency that offerings with the highest previous returns do worst in the latter years may be a manifestation of a desire by issuers to avoid future lawsuit. And for low-capitalization stocks, there is a negative correlation between past and subsequent abnormal returns on individual stock as the evidence of market overreaction (DeBondt 1985, Thaler 1987). Age is included considering investors’ different reaction to younger firms and established firms. The change in net income and capital expenditure predicts the growth of the firm. In Ritter (1991)’s paper, those control variables are proven to be correlated with long-term performance of stock return. Although such regression ignores the contemporaneous correlations among variables, and can lead to biased standard errors (but not biased coefficient estimates), Ritter’s results have been supported in other studies. Furthermore, this specification provides regression coefficients that allow an easy interpretation of the economic significance.
4. Empirical Result
Table4 reports the time-series distribution of discretionary current accruals from fiscal Year -5 to Year -1 for MBOs and LBOs. Both the mean and median DCA don’t show a monotonic trend from Year -5 to Year -1. However, the median and mean DCA are significantly negative in Year -1 even though negative DCAs are also found from Year -5 to Year -2.
Firms adopting positive DCA are dominant from Year -5 to Year -2 for both MBOs and LBOs. However, the majority of 61% of firms apply negative DCA in Year -1. Hence, DCA in the fiscal year preceding the announcement of transactions is most typical for analyzing downward earnings management. This result is consistent with Perry and Williams 1994 and Wu 1997, who show that negative earnings management exit in the year preceding going private transactions. The mean of DCA is -2.33% of lagged total assets for MBOs and -15.4% for LBOs and the Wilcoxon-Mann-Whitney test shows the difference is significant at 1% level. At the same time, the median of DCA for MBOs is more negative than that for LBOs as -1.94% and -1.23% respectively. The difference of median DCA between MBOs and LBOs is significant at the 1% level. The difference is consistent with Fisher and Louis (2008), whose result shows upward earnings management in LBOs compared to pure MBOs. The proportion of firms applying negative DCA shows the similar result for MBOs and LBOs as 62% and 63% respectively, which implies the difference of mean DCA is not attributed to the portion of firms applying negative DCA but the magnitude of negative DCA applied by firms. Hence, I conclude downward earnings management exist both for MBOs and LBOs in the year preceding the announcement of transactions. Firms applying downward earnings
21 | P a g e management are found in 62% of the cases both for MBOs and LBOs. However, the extent of earnings management in LBOs is less downward than in MBOs. This difference is attributed to the magnitude of negative DCA applied by firms but not the percentage of firms applying negative DCA.
At various points throughout the paper, I rely on a quantile classification of firms to avoid linear parameterization of regressions. I sort 214 LBO firms and 60 MBO firms by their assets-scaled discretionary current accruals into four quantiles. I label the quantile of going private firms with the lowest DCA as “conservative” firm and the one with highest DCA as “aggressive” firm (Higher discretionary current accruals increase reporting earnings). Take MBO firms as an example, the DCA of conservative quantile is less than -3.32%, the second quantile has accruals of -3.32% to -1.94%, the third quantile has accruals of -1.94% to 0.43% and the DCA of aggressive quantile exceeds 0.4%. Similar for LBO, the quantiles are -2.13%,-1.23% and 0.61%.
Table 5 reports sample characteristics within these quantiles. There is significantly higher cross-sectional variation within the conservative and aggressive quantiles than within two middle quantiles both for MBOs and LBOs. The overall sample deviation is around 19% for both MBOs and LBOs. No visible systematic patterns between DCA and firms’ performance have been found if look at revenue, market to book ratio and return on assets. However, insider voting power is stronger within conservative and aggressive quantiles than within two middle quantiles. For instance, for MBO firms, the voting shares percentage held by insiders is above 50% for both conservative and aggressive quantiles but not exceeding 40% for middle quantiles. This result implies the magnitude of DCAs may rely on the voting power of insiders. This result is consistent with Usha and Dennis (2014), whose result argue that ownership structure is likely to influence the choice of going private and premium paid.
Our key objective is to evaluate the extent to which earnings management has an incremental influence on the abnormal stock returns in going private transactions. Different expected returns may produce different results. Also methods to compute returns differs between cumulative abnormal returns and buy-and-hold abnormal returns varies as mentioned in section 3. Hence, I adopt an agnostic approach which repots out both CAR and BHAR based on alternative benchmarks (CRSP value weighted market return, Fama-French return and matched firm return). The result is reported in Table 6 and all tests indicate that discretionary current accruals reliably predict pre-announcement abnormal returns and negative abnormal returns is associated with negative DCAs both for MBOs and LBOs.
In table 6, I observe a generally monotonic trend for abnormal returns across quantiles, so we only report the difference between aggressive and conservative quantiles. On a BH measurement, the difference is somewhat larger. The first row reports the raw return by quantile. In a cumulative way, the conservative portfolio underperforms the aggressive
22 | P a g e portfolio by 16% for MBOs and 30% for LBOs. In a compounding way, the conservative portfolio underperforms the aggressive portfolio by 24% for MBOs and 33% for LBOs. Comparison between MBOs and LBOs indicates the mean cumulative raw return for LBOs is 1% higher than that for MBOs and the mean compounding raw return for LBOs is 6% higher than that for MBOs. However, this difference of raw return doesn’t tell the discrepancy from earnings management. Hence, we need to look at the abnormal returns on alternative benchmarks.
In table 6, the second row illustrates CRSP value-weighted market-adjusted return by quantile. On a CAR measurement, the portfolio with conservative DCA underperforms the one with aggressive DCA by 20% for MBOs and 17% for LBOs on average. On a BHAR measurement, the portfolio with conservative DCA underperforms the one with aggressive DCA by 27% for MBOs and 19% for LBOs on average. On average, portfolio in MBOs underperforms LBOs by around 2% both on measurements of CAR and BHAR.
The third row in table 6 shows adjusted returns using Fama-French (1993) benchmark. Similarly, conservative portfolio underperform aggressive portfolio. Compared with CRSP value-weighted market-adjusted return, the difference between aggressive and conservative portfolios is larger on a CAR measurement compared with the one on a BHAR measurement. The difference of abnormal return between MBOs and LBOs turns to be around 1% for both CAR and BHAR.
The forth row of table 6 reports the adjusted returns using matched firms. This match suggests the difference of CAR between conservative and aggressive firms is around 19% for MBOs and LBOs on average and the difference of BHAR is above 25% for total going-private sample firms. And the difference of abnormal returns between MBOs and LBOs is 1.3% on CAR and 0.2% on BHAR on average.
Those results suggests firms with conservative DCA underperforms firms with aggressive DCA both for MBOs and LBOs and the performance differential between aggressive and conservative earnings managers seems large and economically significant regardless of the benchmark used. Hence, downward earnings management is effective in lowering stock price. At the same time, firms in MBOs underperform those in LBOs both on average and within quantile. Hence, this underperformance is not led by individual extreme values but the magnitude of each DCA.
To validate whether firms with conservative DCA underperform those with aggressive DCA by different groups, we have sample partition. As the result is the same for MBOs, I only report LBOs as an example. Table 7 reports the CRSP value-weighed market-adjusted abnormal returns for LBOs by different group and by quantile. Regardless of the firm size, performance
23 | P a g e and period of the firm, firms with conservative earnings management underperforms those with aggressive earnings management.
In sum, even if estimations on the extent of going-private firms’ underperformance are sensitive to abnormal return computation, the difference in performance between firms employing different degrees of earnings management is the same significantly large. More conservative firms underperform more aggressive firms by a margin that is economically significant. At the same time, firms in LBOs outperform firms in MBOs with small but also significant difference.
To examine the incremental influence of downward earnings management on pre-announcement stock return underperformance, we utilize the cross-sectional underperformance regression model from Ritter (1991, Table X). The pre-announcement buy-and-hold return during time (-365,-60) is regressed on the DCA in fiscal year -1, the control variables include dummy variables for year and industry; buy and hold return for CRSP value-weighted market return during event time window (-365,-60) ; the log of firm’s market capitalization by end of fiscal Year -1;the log market to book by end of fiscal Year(-1);log of firm’s age by end of fiscal Year -1 ; the change of capital expenditure between fiscal Year -2 and fiscal Year -1 scaled by total assets in fiscal Year -2 and the change of net income between fiscal Year -2 and fiscal Year -1 scaled by total assets in fiscal Year -2.
I include market return because correlation between individual and market returns are expected; I include market value and market to book ratio because there is a tendency for offerings with the highest previous returns to do worst in the latter years may be a manifestation of a desire by issuers to avoid future lawsuit. And for low-capitalization stocks, there is a negative correlation between past and subsequent abnormal returns on individual stock as the evidence of market overreaction (DeBondt 1985, Thaler 1987). Age is included considering investors’ different reaction to younger firms and established firms. The change in net income and capital expenditure predicts the growth of the firm. In Ritter (1991)’s paper, those control variables are proven to be correlated with long-term under performance of stock return. Although such regression ignores the contemporaneous correlations among variables, and can lead to biased standard errors (but not biased coefficient estimates), Ritter’s results have been supported in other studies. Furthermore, this specification provides regression coefficients that allow an easy interpretation of the economic significance.
Table 8 reports out the cross-sectional regression result of abnormal return on DCA for both MBOs and LBOs. DCA (discretionary current accruals) has a positive estimated coefficient of 2.31 for MBOs and 1.98 for LBOs at 5% significance level. This result indicates that firms with high earnings management to boost earnings in the year preceding announcement show greater outperformance whereas firms with low earnings management to reduce earnings in the year prior to announcement predicts greater underperformance. The extent of
24 | P a g e upward/downward earnings management predicts the magnitude of stock performance at significant level. Hence, investors heavily rely on earnings in valuation before going private transactions and earnings play an important role in predicting stock returns.
To identify the different influence power of earnings management to stock performance between MBOs and LBOs, we apply the Wald test comparing the coefficients of DCA. The P-value of the Wald test is 0.53, which implies the difference of coefficient of DCA between MBOs and LBOs is not significant at any level. Hence, the incremental influence of downward earnings management proxy on pre-announcement stock return underperformance is indifferent between MBOs and LBOs.
5. Robustness check
One can observe the buyout is the last attempt to save the company and the related jobs when the firms have been struggling economically. Those instances may raise the question whether observed negative discretionary accruals from calculation reflect continuously declining performance rather than earnings management while the negative abnormal returns also reflects this situation. In Jones’ (1991) model, we incorporate sales of current period in calculation of expected accruals to partially control the impact from financial situation. We examine the issue more carefully by analyzing the financial performance of firms in the year preceding announcement of transactions. Table 9 reports the changes of financial performance (earnings, cash flows and revenues) for MBOs and LBOs in the year preceding the announcement of transactions.
Changes for each variable are calculated as the percentage change from the previous year, scaled by beginning total assets. The change in cash flow is significantly positive both for MBOs and LBOs by 2.1% and 1.2% respectively. The change in revenue is also significantly positive for both by 15.1% and 17.6% respectively. However, the change in earnings is negative for both MBOs and LBOs by -1.3% and -0.6% respectively. As total accruals equal to the difference between net income and operating cash flow, the change of total accruals are inferred to be significantly negative. Those data indicates MBOs/LBOs firms’ negative accrual management in fiscal Year -1 is masking the unusually high operating cash flow and intends to depress the earnings below the previous year’s level. From the financial data, we conclude the sample firms in going private don’t appear to be in financial trouble on average and the declining financial performance is not an alternative explanation for the observed negative earnings management and underperformance.
25 | P a g e Prior studies have suggested that managers would like to reduce equity market perception of a firm’s value to reduce the purchase price in going private transactions. Consistent with the incentive, downward earnings management has been found in the year preceding going private transactions. Management has considerable control over the timing of actual expense items and can also alter the timing of recognition of revenues and expenses. However, whether insiders benefited from downward earnings management depends on whether the offering price reflects only public new information or both public and private new information. There is no empirical research documenting whether downward earnings management benefited inside managers through lower stock price before purchasing price is finalized. If downward earnings management didn’t effectively influence stock price so that insiders didn’t benefit from lower purchasing price, downward earnings management is meaningless to managers at the risk of manipulating financial data. At the same time, the theory of strong-form efficiency of capital market would be strengthened in this case.
The analysis in this paper is broken down into management buyouts (MBOs) and leveraged buyouts (LBOs). I expect differences between MBOs and LBOs both in earnings management and stock returns because managers anticipating LBOs face conflicting incentives. Both MBOs and LBOs are going private transactions initiated by insider managers. However, the completion of MBOs relies on internal financing but that of LBOs relies on external financing. Hence, managers anticipating LBOs not only have the incentive to reduce equity market perception of their firms’ value to reduce the purchase price but also have the incentive to increase equity market perception of their firms’ value to secure external financing. Hence, I hypothesize less downward earnings management and less lower offering price would be found in LBOs compared to MBOs. If earnings management in LBOs is still downward but less so than that in MBOs, then the incentive of securing external financing is countervailing and tempers the incentive of lowering purchase price. Inversely, if earnings management in LBOs is not downward but upward, then the incentive of securing external financing dominates the one of lowering purchase price.
This paper examines whether insiders benefited from earnings management in firms’ going private transactions. The sample firms consists of 214 LBO firms and 60 MBO firms during 2004-2014 with complete information in COMPUSTAT and CRSP. I adopt strict criteria for the sample selection with lots of hand collection and validation. Those criteria are valuable to further research on going private transactions. From the firm statistics, we observe big differences on firm size, management ownership, and market to book ratio between MBOs and LBOs.
First, I find mean and median discretionary current accruals -which are under the control of managers and proxy for earnings management- to be negative in the fiscal year preceding the announcement of going private. A majority of 60% of firms apply negative DCA in the fiscal Year -1.The mean DCA for MBOs is -2.33% of lagged total assets and the mean DCA for LBOs
26 | P a g e is 1.54%. The median DCA for MBOs is 1.94% for MBOs and the median DCA for LBOs is -1.23%. This result is consistent with our hypothesis that firms in LBOs have less downward earnings management than firms in MBOs in attempt to attract external financing. Hence, in LBOs, incentive of securing external financing is countervailing and tempers the incentive of lowering purchase price.
The paper documents negative abnormal returns -which are computed from expected returns based on different benchmarks (CRSP value-weighted market-adjusted, Fama-French adjusted, matching firm adjusted)- during the event time (-365,-60) both for mean cumulative abnormal return (CAR) and mean buy and hold abnormal return (BHAR). The negative abnormal return on average is economically large and significant. For instance, on a CRSP value-weighted market-adjusted return measurement, the mean CAR for portfolio in MBOs is -17.32% and the mean BHAR of a portfolio with pre-MBO firms is -15.21% at the 1% significance level. Consistent with DCA, portfolio in LBOs have less negative abnormal returns with -15.36% for CAR and -13.49% for BHAR on average at the 1% significance level. This result is consistent with our hypothesis that both firms in MBOs and LBOs underperform in the year preceding transactions but firms in LBOs have less negative abnormal returns than MBOs. This result has been confirmed in other specifications (Fama-French adjusted, matching firm adjusted).
This paper also examines how the extent of earnings management predicts the magnitude of underperformance of firms in going private transactions. Firms have been classified into four groups according to DCA quantile. There is no visible systematic pattern between DCA quantile and financial performance. Abnormal returns across DCA quantiles is monotonic and decreases from most aggressive (most positive) quantile to most conservative (most negative) quantile. The difference in abnormal return between two extreme quantiles is economically large and significant for each specification. For instance, on the measurement of CRSP value-weighted market-adjusted return, the difference in CAR between most aggressive quantile and most conservative quantile is 20% for MBOs and 17% for LBOs. The difference in BHAR between those two quantiles is 27% and 19% respectively. The difference is smaller for LBOs and I interpret the reason as lower management voting power in firms of LBOs which determines the ability to manipulate earnings. This result is robust when I observe the monotonic decrease of abnormal return from aggressive quantile to conservative quantile within all sub groups by firm size, market to book ratio and sub period. This relation has been carefully analyzed in cross-sectional regression of stock performance on DCA for both MBOs and LBOs. DCA (discretionary current accruals) have a positive estimated coefficient of 2.31 for MBOs and 1.98 for LBOs at 5% significance level. This result indicates that firms applying downward earnings management could effectively influence stock return to underperform and the extent of downward earnings management predicts the magnitude of stock return underperformance at a significant level.
