Do managers manipulate earnings prior to management buyouts?

    Do Managers Manipulate Earnings Prior to Management Buyouts? Yaping Mao, Luc Renneboog PII: DOI: Reference:

S0929-1199(15)00095-4 doi: 10.1016/j.jcorpfin.2015.08.005 CORFIN 949

To appear in:

Journal of Corporate Finance

Received date: Revised date: Accepted date:

20 July 2014 8 August 2015 12 August 2015

Please cite this article as: Mao, Yaping, Renneboog, Luc, Do Managers Manipulate Earnings Prior to Management Buyouts?, Journal of Corporate Finance (2015), doi: 10.1016/j.jcorpfin.2015.08.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Do Managers Manipulate Earnings Prior to Management Buyouts? Yaping Mao1

T

School of Business, Aalto University

RI P

Luc Renneboog2

SC

CentER, Tilburg University and ECGI

Abstract

MA NU

To address the question as to whether managers intending to purchase their company by means of a levered buyout transaction manipulate earnings in order to buy their firm on the cheap, we study different types of earnings management prior to the transaction: accrual management, real earnings management, and asset reserves revaluation. To identify the management engagement incentives, we contrast earnings management in management buyouts (MBOs) with that in (i) institutional buyouts (IBOs) and (ii) nonbuyout firms. We find: (i) strong negative earnings management via both accrual

ED

and real earnings activities in MBOs supporting the above management engagement incentive, (ii) modest negative accrual management and insignificant real earnings manipulation in IBOs, and (iii) positive earnings management in non-buyout firms. Asset revaluation in MBOs is not a frequently

PT

used channel. We do not find evidence that a high external borrowing need in the levered transactions mitigates the downward earnings manipulation in MBOs. The implementation of the revised UK

CE

Corporate Governance Code of 2003 has somewhat reduced the degree of both accrual earnings and real management in MBOs, but increased the relative use of real management as this may be harder to

AC

detect.

Keywords: accounting manipulation, earnings management, leveraged buyout, management buyout, MBO, institutional buyout, IBO, asset reserves revaluation, corporate governance regulation. JEL codes: G30, G32, M41.

1 Department of Accounting, Aalto University, P.O. Box 21210, FI-00076 Aalto, Finland, email: yapping.mao@ aalto.fin; phone: +358 50 411 8971. 2 Department of Finance and CentER, Tilburg University, P.O. Box 90153, 5000 LE Tilburg, the Netherlands,

1

ACCEPTED MANUSCRIPT and European Corporate Governance Institute (ECGI), email: [email protected], phone: +31 13 4668210,

AC

CE

PT

ED

MA NU

SC

RI P

T

fax: +31 13 466 2875.

2

ACCEPTED MANUSCRIPT Do Managers Manipulate Earnings

T

Prior to Management Buyouts?

RI P

1. Introduction

Prior to management buyouts (MBOs), managers have an incentive to deflate

SC

the reported earnings numbers by accounting manipulation in the hope of lowering the stock price. If they succeed, they could be able to acquire (part of) the company on the cheap. It is

MA NU

important to note that accounting manipulation in a buyout transaction may have severe consequences for the shareholders who sell out in the transaction: if the earnings distortion is reflected in the stock price, the wealth loss of shareholders is irreversible when the company goes private subsequent to the buyout. As manipulation of financial statements prior to US MBOs has occasionally been detected in the academic literature over the past 20 years (DeAngelo, 1986; Perry and Williams, 1994; Fisher and Louis, 2008; Li, Qian and Zhu,

ED

2013), we wonder whether accounting manipulation has occurred/still occurs in the second most important buyout market, namely that of the UK which is subject to different

PT

accounting practices and corporate governance regulation and enforcement (Wright, Shaw and Guan, 2006).1,2 We focus on a period that commences with the start of the second buyout

CE

wave: 1997-onwards, which also coincides with tightened corporate governance regulation (Guo, Hotchkiss and Song, 2011) and enhanced reporting integrity (Botsari and Meeks, 2008).

AC

We investigate whether earnings manipulation occurs preceding buyouts and, if

it does, how manipulation takes place by studying the channels of accrual management and real earnings management. Whereas accrual-based earnings management activities have no cash flow consequences, real earnings management refers to managerial activities that deviate from normal business practices and thus affect cash flows. We apply a performance-based modified Jones model and also advance an industry and industry-size adjusted buyoutspecific approach to capture the abnormal accounting numbers. In addition to the 1

For instance, for asset reserves revaluation in the UK, companies can choose between: (i) the historical cost model; (ii) the revaluation model. However, in the US, companies follow historical cost model to measure the value of fixed assets. 2 In the UK, there have been frequent developments in corporate governance during the second buyout wave. The Combined Code of 1998 comprising the Cadbury and Greenburg principles intended to address (amongst other issues) concerns about board representation and director compensation. The Turnbull report (Internal control: Guidance for directors on the combined code (1999)) stresses the directors’ responsibilities to keep good “internal controls” in their companies, and organizing effective audits and checks to ensure the quality of financial reporting and catch any fraud before it becomes a problem. The revised Corporate Governance Code of 2003 further emphasizes the independence of the directors and the audit committee for internal control. 3

ACCEPTED MANUSCRIPT aforementioned two types of earnings manipulation, we further test asset reserves revaluation manipulation on the balance sheet as an additional means of accounting manipulation that may influence the debt capacity in a buyout setting. We contrast MBOs to institutional

T

leveraged buyouts (IBOs) where the incumbent management will subsequently no further be

RI P

involved (as in Hafzalla, 20093) and to non-buyout firms.

We then investigate why earnings management takes place by analyzing two types of incentives. On the one hand, managers may opt to present lower earnings if they are

SC

likely to participate in a prospective buyout transaction and will subsequently stay within the company. Negative earnings manipulation or earnings understatement can be induced by

MA NU

these management engagement incentives. On the other hand, managers’ incentive to misrepresent the earnings and asset reserves may affect the financing capacity of the buyout target. Low earnings (cash flow) numbers influences the amount of external borrowing a firm can attract and the terms of debt contracts. Thus, managers who prepare a corporate sale by means of a leveraged buyout (LBO) and who anticipate that they will not be involved in the

ED

firm either as managers or equity holders may manipulate earnings upwards or at least less downwards (than in the case of MBOs). We call this is the external financing incentive.

PT

In general, outside the context of LBOs, there are various other reasons to manage earnings. For instance, in relation to a firm’s investment opportunities, Linck, Netter,

CE

and Shu (2013) document that by managing earnings upwards, a financially constrained company with investment opportunities can credibly signal positive prospects which can help raise external financing. Furthermore, related to the focus of this paper, is a large literature on

AC

managerial remuneration contracting and self-dealing (e.g. Bebchuk and Fried, 2003) which states that managers have strong incentives to manipulate earnings upwards if they have an impact on share prices which augments the value of their equity-based compensation. Our contributions to the literature are the following: First, whereas the past literature on earnings management in the context of buyouts is limited to income statement manipulation, we study multiple manipulation techniques: in addition to earnings manipulation via accrual management, we also search for evidence of real manipulation of earnings by means of sales and production choices, and of balance sheet statement manipulation by means of asset revaluation preceding the buyout transaction. Second, in terms of identification of the managerial engagement incentives relative to other incentives for earnings manipulation, we distinguish among MBOs, IBOs in which the pre-transaction 3

Alternative definitions are possible, but this one enables us to focus the distinction between the incentives to manage earnings downwards or upwards. 4

ACCEPTED MANUSCRIPT management is subsequently no longer involved in a management role or as an equity holder, and non-buyout firms. In MBOs, managerial incentives to manipulate earnings downwards may dominate, whereas IBOs upward manipulation may be exerted in order to relax financial

T

constraints such that the firm is able to increase its leverage and the incumbent management

RI P

can sell out. Third, most studies have examined a sample belonging to the first MBO wave of the 1980s or the period until the bursting of the high-tech bubble. As corporate governance has been improved in the recent decade - for instance, in 2003, the revised Combined Code

SC

on Corporate Governance (currently called the UK Corporate Governance Code) was implemented to enhance internal control and reporting quality – we examine whether or not

MA NU

earnings management is affected by the change in soft law regulation.4 We report the following findings: first, in contrast to non-buyout firms for which we find evidence of upwards earnings manipulation, buyout firms apply downward earnings management by means of both accrual management and real earnings management. These results are robust to various approaches to calculate earnings management: industry-

ED

and industry-size adjustments to abnormal earnings, and abnormal earnings based on performance matching.

PT

Second, for MBOs we find evidence of significantly more negative earnings manipulation by means of accrual management than for IBOs (in which the pre-transaction

CE

management will no longer be involved). Likewise, downward real earnings manipulation (through production costs and sales revenues) prevails in MBOs. This suggests that purposeful manipulation is applied in MBOs in order to drive the pre-transaction share price

AC

down such that the firm can be bought at a discount. We demonstrate that negative earnings manipulation in MBOs is timed to occur in the year prior to the deal announcement. Third, while the managerial engagement incentives are evident, we do not find

support for the external financing incentive because the negative earnings management in MBOs is not mitigated for MBOs with very high leverage and upward manipulation in IBOs (who need to borrow against earnings) is limited. The lack of support for the external

4

Wright et al. (2006) study earnings manipulation in MBOs from 1997-2002, but our study differs in that (i) we analyze earnings manipulation in both MBOs and IBOs which enables us to identify the managerial engagement incentives, (ii) we cover the complete second wave of buyouts from 1997 until the financial crisis when the use of LBOs crashed following a credit crunch, (iii) we can study the impact of the change in corporate governance regulation relevant to accounting quality, (iv) we not only study accrual management but also other earnings management channels (real earnings management, asset reserves revaluation), (v) we not only use the modified Jones’ model without performance correction, we show that controlling for performance is important to derive conclusions on earnings manipulation. 5

ACCEPTED MANUSCRIPT financing incentive may be explained by the fact that during the second buyout wave it was relatively easy to attract external funds given low interest rates and a strong growth in the high yield bond market.

T

Fourth, the revised Corporate Governance Code of 2003 has had a significant

RI P

impact on both accrual and real earnings manipulation: Accrual management does indeed decline since 2003. In contrast, real earnings manipulation techniques are more frequently used since the tightening of the corporate governance regulation, which may be induced by

SC

the fact that real earnings manipulation techniques are more difficult to detect. This finding is consistent with some recent US evidence: after the adoption of the Sarbanes Oxley regulation,

MA NU

companies shifted from accrual management to real earnings management (Cohen, Dey and Lys, 2008). Overall, our findings imply that corporate governance changes have been effective to curb earnings management in management buyout transactions. The rest of the paper is organized as follows. In the next section, we review the literature and develop the hypotheses. Section 3 describes how accounting management is

ED

measured and explains the empirical setup. Section 4 reports the sample selection criteria and discusses the descriptive statistics. The empirical results and robustness analyses are set out

PT

in Section 5. Section 6 concludes.

CE

2. Literature Overview and Hypotheses If managers foresee that their firm will be taken private in a leveraged

AC

transaction, their accounting choices prior to the transaction may be affected by the anticipated type LBO, e.g. an MBO or an IBO (Dechow, Ge, and Schrand, 2010). In this section, we will relate different types of LBOs to types of earnings management. In addition to earnings management, e.g. accruals manipulation or real earnings management which are both forms of income statement manipulation, they can also manipulate the balance sheet. The latter comprises asset reserves revaluation and this revaluation of tangible assets is then reflected in the recorded increments/decrements in the equity account, and leads to a change in debt-to-equity ratio. 2.1. Accounting Manipulation 2.1.1. Earnings Manipulation In the context of surging MBO activity in the US over the 1980s, virtually every buyout proposal was contested by shareholders claiming that they were cheated (Longstreth, 6

ACCEPTED MANUSCRIPT 1984). Even though recommendations by investment banks and approval by independent directors were sought to evaluate the fairness of buyout transactions, doubts about accounting manipulation remained. DeAngelo (1986) did not detect accrual manipulation preceding US

T

MBOs, but Perry and Williams (1994) who worked with a larger sample and utilized a

RI P

regression-based model to capture discretionary accruals more accurately, did document downward accrual management. Wu (1997) showed that on average, earnings manipulation prior to MBOs decreased the acquisition price by 18.6%, and Ang, Hutton and Majadillas

SC

(2014) confirm that downward earnings manipulation is especially prominent in firms where managers hold strong equity positions subsequent to the buyout. While managers may have

MA NU

valid personal reasons to manipulate earnings downwards (buying the company on the cheap), in some cases, they also have incentives to manipulate earnings upwards: namely, ex ante financial constraints. For instance, Fisher and Louis (2008) argue that managers overstate their earnings to get favorable debt contract terms in highly levered buyouts, and Linck et al. (2013) show that financially constrained firms with valuable investment projects use positive

ED

discretionary accruals to ease their financial constraints.

PT

Managerial Engagement Incentives (H1) Managers can understate the earnings numbers in MBOs (which keep the

CE

management team on board subsequent to the leveraged transaction) for the following reasons: First, managers are (co-)acquirers of MBO targets such that earnings manipulation resulting in a lower purchase price leads to self-dealing. Second, in the case of an MBO (IBO)

AC

in which a small (large) part of the equity is provided by financial sponsors, the sponsors usually send a “love letter” to win the support of the management which comprises an invitation to the current management team for further discussions and the intention to employ them after sealing the deal and reserve an equity stake (Das and Chon, 2011). Jackson (2013) even mentions that private equity firms go as far as bribing CEOs to push the deals through. So, managers in the buyout companies have incentives to facilitate the transaction, although the management’s personal benefits will be larger in MBOs in which the ownership is not shared with external parties. Third, in IBOs maintaining the incumbent management team, equity ratchets are often offered to managers which gradually increases their post-transaction ownership stake in order to motivate them to achieve strong periodic performance and good

7

ACCEPTED MANUSCRIPT exit returns 5 (Renneboog, Simons and Wright, 2007; Yates and Hinchliffe, 2010). As operating performance and returns are compared with pre-buyout performance, managers in those IBOs can reap additional benefits from earnings understatement in terms of post-buyout

T

bonus and exit premium. In IBOs in which the pre-transaction management was not retained,

RI P

the above arguments to manipulate earnings downwards may not apply (or apply to a lesser extent). Hence, we postulate the managerial engagement hypothesis: Prior to MBOs, earnings are manipulated downwards by both accrual management and real earnings

SC

management. Moreover, earnings are manipulated downwards to a larger extent in MBOs than in IBOs (H1). We expect that earnings manipulation will occur in the year prior to the

MA NU

transaction, which we call the manipulation year.

The implicit assumption underlying this hypothesis is that market participants cannot differentiate between earnings arising from business activities and manipulated earnings. In general, Bradshaw, Richardson and Sloan (2001) find that even sophisticated professionals, such as auditors and financial analysts, fail to detect accrual anomalies.

ED

Likewise, Bhojraj and Swaminathan (2007) show that creditors, such as bond investors also do not correctly price accruals. Hence, the possibility of detecting manipulation seems rather

PT

low. Moreover, if manipulation is found out, managers could more easily justify downward manipulation than upward manipulation by referring to the principle of accounting

CE

conservatism.

External Financing Incentives (H2)

AC

Buyout transactions largely rely on external financing, a combination of senior

loans, subordinated loans, and high-yield bonds. A typical buyout is traditionally financed with 60 to 90 percent of debt (Kaplan and Strömberg, 2009). Using the world-wide buyout transactions from 1986 to 2008, Axelson, Jenkinson, and Strömberg (2013) report that public-to-private buyouts are the most highly levered, with 73% of transaction capital raised from debt of various forms. Ample evidence points out that the debt financier is prone to use earnings numbers to predict future cash flows and make credit decisions (Palepu, Healy and Berbard, 2000). In the accounting literature, Dechow, Sloan and Sweeney (1996) report that attracting

5

A ratchet is an incentive mechanism which either offers managers a modest equity stake if managers meet exante specified performance targets after buyouts (Renneboog et al., 2007) and/or entitles managers to receive a higher proportion of the exit proceeds if an exit is achieved beyond a particular ‘hurdle’ return rate for investors (Yates and Hinchliffe, 2010). 8

ACCEPTED MANUSCRIPT external financing is an important motivation for earnings manipulation. Furthermore, Linck et al. (2013) point out that positive discretionary accrual management can be used to emit as a signal about a company’s prospects and thus can help to ease its financial constraints. In a

T

buyout setting, consistent with the view that earnings are a key determinant of the optimal

RI P

capital structure or how much a firm could possibly borrow, external financing concerns suggest that managers show less negative earnings numbers. Indeed, Fischer and Louis (2008) find that managers in need of much leverage to finance an MBO are more likely to apply less

SC

negative abnormal accrual management. Hence, the external financing incentive can be formulated as: Earnings management is negatively related to the external financing needs of

MA NU

a leveraged buyout. (H2).

However, two caveats arise. First, there are alternatives to earnings management to maximize the leverage a firm can sustain: a company may resort to asset revaluation manipulation (see below). So, if a firm revalues its assets and manipulates its earnings downwards, it could at the same time try to reduce the equity value of the firm while

ED

maximizing its leverage potential. Second, Axelson et al. (2013) argue that economy-wide debt market conditions, rather than firm characteristics explain an LBO’s debt capacity better.

PT

Likewise, Leary (2009) emphasizes the importance of the supply side effect of leverage, and Shivdasani and Wang (2011) confirm that the increase in buyout transactions from 2004 to

CE

2007 was fueled by the fast growth in collateralized debt obligations (CDOs). Therefore, the mitigating role of the external financing incentive in downward earnings manipulation may

AC

not apply in times of credit market booms.

2.1.2. Asset Revaluation (H3) While the literature on accounting manipulation prior to MBOs traditionally

concentrates on earnings management (income statement manipulation), balance sheet manipulation through asset revaluation may also occur, especially in the context of buyouts. This technique could enable a target company to attract more debt to finance the deal and to receive more favorable debt contracts. Since the implementation of FRS3 in 1993, companies are encouraged to revalue fixed assets on the ground that they provide useful and value-relevant information.6 The difference between an asset’s old carrying value and its revaluation is credited to a revaluation reserve account on the balance sheet. The new asset revaluation practice has the 6

This practice is consistent with the EU’s adoption of IFRS in 2005. Under IAS 16, companies can choose between: (i) the historical cost model; (2) the revaluation model. 9

ACCEPTED MANUSCRIPT following implications: (i) If assets are revalued upwards (downwards), it increases (decreases) the equity amount via the revaluation reserve account and thus lowers (boosts) the debt-to-equity ratio. (ii) If assets are revalued upwards, it will lower gains from a future

T

asset disposal, as the inflated carrying value will serve as the benchmark value. Meanwhile,

RI P

the upward revaluation increases the depreciation charges. (3) If assets are revalued downwards, the net revaluation decrement is expensed on the current income statement. Revaluations are discretionary in nature, because managers can decide whether, when, and

SC

what amounts of assets are revalued in financial statements (Lin and Peasnell, 2000). At first glance, in LBOs, managers have an incentive to revalue assets upwards

MA NU

in order to show a lower leverage ratio. This argument is consistent with Easton, Eddey and Harris’s (1993) survey in that a key motivation to revalue assets is indeed debt contract considerations. However, upward asset revaluation induces future costs. First, upward manipulation increases depreciation and decreases net income in the near future. The inflated assets will lower the gains from future asset sales, which will also exert a negative impact on

ED

earnings and performance. Second, the accumulated assets revaluation reserves exhaust companies’ possibilities to further use this manipulation tool subsequent to the buyout as the

PT

amount of upward revaluation is not unlimited. It is also noteworthy that upward revaluation is also costly, as valuation fees are paid to independent valuators to certify the revaluation.

CE

Therefore, weighing future costs against current benefits is imperative. Hence, we expect that: assets are revalued upwards in LBOs in order to increase the debt capacity, and occurs to a

AC

larger degree in IBOs than in MBOs (H3). 3. Accounting Manipulation Proxies and Empirical Models 3.1. Earnings Management Proxies Managers use accrual discretion to present specific earnings numbers and hence influence equity valuation. The presented bottom-line results can also be influenced by real earnings management. Hence, we will investigate both types of earnings management.

3.1.1. Accrual Management Proxies To measure discretionary (abnormal or manipulated) accruals, regression-based models have been developed for which Dechow et al. (1995) and Balatbat and Lim (2003) demonstrate that the modified-Jones model performs best. Still, Kothari, Leone and Wasley (2005) are concerned that ignoring the financial performance in those regression models leads 10

ACCEPTED MANUSCRIPT to spurious results, which is in particular relevant for LBOs. Therefore, we adopt two approaches: First, we directly add an additional performance control variable to our accrual model in order to exclude abnormal accruals resulting from mean reversion in the

T

performance (or performance momentum). Furthermore, as abnormal accruals after

RI P

performance adjustment may still comprise accruals arising from common manipulation incentives (e.g. compensation incentives or meeting analysts’ forecasts) or random effects induced by other events, we further adjust the abnormal accruals for (i) industry average

SC

abnormal accruals or (ii) average abnormal accruals in the same size group within the same industry7. Second, we use a performance-matched approach whereby we match the buyout

MA NU

target with a non-buyout company of the same two-digit SIC code and with the closest performance in the year of the buyout. The remaining part of the abnormal accruals captures the industry-adjusted buyout-specific manipulation.

The Performance-Adjusted Modified -Jones Regression Model (PAMJ)

ED

To measure the PAMJ model, we estimate the discretionary accruals for each year using all non-buyout firm-year observations with the same two-digit SIC code. Peasnell,

PT

Pope and Young (2000) state that the cross-sectional model is more able to capture the magnitudes of accrual management. Our primary expectations model for estimating non-

CE

discretionary accruals is as follows: TAACi,t (ΔSalesi,t - ΔReceivablesi,t ) PPEi,t 1 = β0 [ ] + β1 [ ] + β2 [ ] + β3 ROAi,t +εi,t (1) Assetsi,t-1 Assetsi,t-1 Assetsi,t-1 Assetsi,t-1

AC

where, for fiscal year t and firm i, TAAC stands for the total accruals defined as TAACi,t = EBXIi,t - OCFi,t , the difference between Earnings Before Extraordinary Items (EBXI) and Cash Flow from Operations (OCF) 8. ∆Salei,t and ∆Receivablesi,t stand for changes in sales and receivables, respectively. PPEi,t is gross Property, Plant and Equipment and Assetsi,t-1 represent the total book value of assets. Kothari et al. (2005) demonstrates that using contemporary ROAi,t produces less miss-specified tests relative to lagged ROA

i,t-1.

All

variables, except ROAi,t, are scaled by lagged total assets to mitigate heteroskedasticity in residuals. 7

For each year and each two-digit SIC code industry, we divide the control observations portfolio into terciles by ranking firms according to their total assets. We then match the buyout company with the non-buyout companies based on the same size tercile in the same year and the same two-digit SIC code. 8 Hribar and Collins (2002) state that accrual estimates calculated from balance sheets can be contaminated by measurement error and therefore prefer accruals calculated from cash flow statements. Ball and Shivakumar (2008) confirm that the balance sheet approach is biased to upward earnings management and the amount of discretionary accrual is overestimated. 11

ACCEPTED MANUSCRIPT The predicted raw abnormal total accruals RAW_ABN_TAACi,t are the difference between observed total accruals and normal total accruals calculated from the parameter estimates of equation (1).

T

To remove the non-event specific abnormal accruals, we subtract the mean

RI P

abnormal accruals of the control observations (firms in the same year and with the same twodigit SIC code) from the raw abnormal accruals, which yields the industry-adjusted buyoutspecific abnormal accruals Madj_ABN_TAACi,t:

SC

Madj_ABN_TAACi,t = RAW_ABN_TAACi,t - Mean_ABN_TAACi,t

(2)

For our robust tests, we will also subtract the mean abnormal accruals of the

label it as MadjSize_ABN_TAACi,t.

MA NU

control observations in the same size group within an industry from RAW_ABN_TAACi,t and

The Performance-Matched Modified -Jones Regression Model (PMMJ) An alternative approach to control for performance consists of adjusting the

ED

estimated abnormal accruals by subtracting the estimated abnormal accruals of a performance-matched non-buyout company. While the notation remains the same as above,

PT

we first estimate the expectations model without a performance regressor. (3)

CE

TAACi,t (ΔSalesi,t - ΔReceivablesi,t ) PPEi,t 1 = β0 [ ] + β1 [ ] + β2 [ ] +εi,t Assetsi,t-1 Assetsi,t-1 Assetsi,t-1 Assetsi,t-1

The raw abnormal accruals is the difference between actual accruals and

AC

estimated normal accruals obtained from equation (3). We then select for each firm in the buyout year a matched firm from the non-buyout companies with the same two-digit SIC code and with the closest ROAi,t. The difference between the buyout sample and the control observation comprises the industry-adjusted buyout-specific abnormal accruals ABN_TAACi,t. 3.1.2. Real Earnings Management Proxies We test two common types of real earnings manipulation: sales and production manipulation. Sales manipulation occurs when managers (temporarily) influence earnings and thus the bottom line earnings numbers by changing the sales price or/and credit terms. In a buyout context, managers attempt to lower the sales and thus the earnings by imposing a sales price premium or/and offering less lenient credit terms. For instance, by temporarily reducing lenient credit terms, customers may delay their purchases in the current period. Consequently, the sales decline and the earnings are deflated, but given the tightening of the 12

ACCEPTED MANUSCRIPT credit terms, the collection of current period’s sales increases which boosts the cash inflow. All in all, the effect of sales manipulation is expected to result in a higher level of operating cash flows.

T

Prior to the buyout, managers can slow down production in order to reduce net

RI P

earnings. On the one hand, by producing fewer units, the fixed costs are spread over a small number of units and the fixed cost per unit augments and, since the production is below its optimal scale, the marginal cost per unit rises as well. Hence, the total cost per unit increases,

SC

which implies higher reported cost of goods and lower operating margins. On the other hand, the other production and holding costs for inventory decline. As a result, the total production

MA NU

costs, the sum of the cost of goods and changes in inventory, can be reduced as the decline in the latter is expected to dominate the increase in the former (Roychowdhury, 2006) which leads to a low ratio of production costs to sales. In addition, operating cash flows can be higher than normal, given sales levels.

We use both cross-sectional performance-adjusted and performance-matched

ED

methods to derive industry-adjusted buyout-specific real earnings management proxies as

PT

reflected in sales and production.

Sales Manipulation

CE

Our expectations model is formulated as follows: OCFi,t Salesi,t ΔSalesi,t 1 = β0 [ ] + β1 [ ] + β2 [ ] + β3 ROAi,t +εi,t (4) Assetsi,t -1 Assetsi,t -1 Assetsi,t -1 Assetsi,t -1

AC

with all the variables as defined above. The raw abnormal operating cash flows RAW_ABN_OCFi,t are the actual operating cash flows minus the normal operating cash flows estimated from the above equation. To remove the non-event specific abnormal cash flows, we subtract the mean abnormal operating cash flows of the control firms (of the same year and with the same two-digit SIC code) from the raw abnormal operating cash flows, which yields the industry-adjusted event-specific abnormal operating cash flows Madj_ABN_OCFi,t. As before, we also use two alternative calculations: we subtract the mean abnormal operating cash flows of the control firms in the same size group within the same industry from RAW_ABN_OCFi,t and label it MadjSize_ABN_OCFi,t. We also use a performance-matched approach: a matched firm is selected by a non-buyout company in the same two-digit SIC code and year with the closest ROAi,t. Raw abnormal operating cash flows are calculated for both the sample and the control observations and the resulting difference is the buyout-specific abnormal operating cash flows ABN_OCFi,t. 13

ACCEPTED MANUSCRIPT

Production Manipulation We take the following production cost expectation model as our basis: (5)

RI P

T

PRODi,t Salesi,t ΔSalesi,t ΔSalesi,t-1 1 = β0 [ ] + β1 [ ] + β2 [ ] ++β3 [ ] + β4 ROAi,t +εi,t Assetsi,t-1 Assetsi,t-1 Assetsi,t-1 Assetsi,t-1 Assetsi,t-1

where, for fiscal year t and firm i, PRODi,t is the production costs and equals the sum of the Cost of Goods (COGSi,t) and the change in Inventory (∆INVENTORYi,t). The raw abnormal

SC

production costs is calculated from the above expectations model. As before, to remove the non-event specific abnormal production costs, we subtract the mean abnormal production

MA NU

costs of the control firms (of the same year and with the same two-digit SIC code) from the raw production costs and yield the industry-adjusted event-specific abnormal production costs Madj_ABN_PRODi,t. We also use two alternative measures: (1) Industry-size adjusted abnormal production costs MadjSize_ABN_PRODi,t, and (2) Performance-matched abnormal production costs ABN_PRODi,t.

ED

3.2. Asset Revaluation Manipulation Proxy Asset revaluation is calculated as the change in revaluation reserves on the

PT

balance sheet (Black, Sellers and Manly, 1998; Cheng and Lin, 2009). Asset revaluation reserves’ reduction (inflation) in the manipulation year implies downward (upward)

CE

revaluation. As revaluations are industry-specific, we further subtract the industry’s average revaluation or the average revaluation by the same size group within the same industry from

AC

the raw asset revaluation numbers to capture the industry-adjusted buyout-specific abnormal revaluation.

As changes in asset reserves may reflect transfers among different reserve

accounts, we collect detailed information on reserves revaluation from the footnotes of annual reports and record the frequency of four types of revaluation: (i) “No change” indicates that the asset reserves revaluation remain the same in both the manipulation and the year before the manipulation year; (ii) “Upward revaluation” indicates that there are overstated revaluation activities in the manipulation year; (iii) “Downward revaluation” captures the opposite case, and (iv) “Transfer” refers to the change in reserves revaluation arising from a transfer between the reserves revaluation account and other reserves accounts9.

9

For instance, Usborne plc underwent buyout in 1998. The 1997 (1996) annual report showed £32000 (£84000) in the revaluation reserves account. The decline in revaluation reserves by £52000 was not due to revaluation, but arose from transferring out of revaluation reserves account to the P&L reserves account. Although a 14

ACCEPTED MANUSCRIPT 3.3. The Incentives Underlying Earnings Management To analyze the underlying incentives of earnings management, we take the above proxies based on accrual, sales, or production manipulation and relate them to our

T

variables of interest: (1) the choice of the buyout type (MBO versus IBO) and (2) external

RI P

financing need. We also control for a set of firm and industry characteristics. Given that the realized MBO is endogenously determined (Fidrmuc et al. 2013), we adopt a two-stage instrument variable method. The first stage regression models the buyout choice and the

SC

predicted buyout choice will be included in the second stage regression as an explanatory variable of the degree of earnings manipulation. The MBO versus IBO choice in year t-110 is

MA NU

a function of the variables at year t-2:

Dum _ MBOi ,t =β0 + β1 Management Owni,t -2 + β2 Non - Executive Owni,t -2 +β3 Largest Owner Institi,t -2 + β4 Independent Diretorsi,t -2 + β5 Board Sizei,t -2 +β6 Analystsi,t -2 + β7 LSE Listingi,t -2 (6) +β8 MTBi,t -2 + β9 ROA i,t - 2+ β10 Cash to Assetsi,t -2 + β11 Debt to Assetsi,t -2 + β12 Sizei,t -2 +YearFixedeffects + IndustryFixedeffects + εi,t

ED

where the dependent variable is the realized buyout type (Dum_MBOi,t which equals one for an MBO and zero for an IBO). Management Owni,t-2 and Non-Executive Owni,t-2 are the

PT

respective percentages of equity held by the management team and the non-executive directors. Largest Owner Institi,t-2 equals one when the largest shareholder in the buyout

CE

company is an institutional investor, and zero otherwise. Independent Directorsi,t-2, is the number of independent directors divided by board size. Board Sizei,t-2 is the number of board

AC

members. Analystsi,t-2 is the number of financial analysts following the buyout company. LSE Listingi,t-2 equals one in case of a listing on the London Stock Exchange (LSE), and zero in case of a listing on the Alternative Investment Market (AIM). We include managerial ownership because a key reason for buyouts in offer documents is “to simplify the management structure to bring it more in line with companies’ prospects”. Another reason to do a buyout is “to remove costs associated with a listing” as companies with illiquid stocks are not able to attract sufficient investor recognition and the listing costs may therefore outweigh the benefits. Illiquidity is often related to high ownership concentration which implies that shareholders intending to dispose of their shares may have little alternative than to sell to the management or a buyout sponsor (Fidrmuc et al., 2013). revaluation decrease could be noted, the sum of the revaluation reserves account and P&L reserves account remained the same and the equity account on the balance sheet was is not influenced by such transfers. 10 Note that we use here t as the buyout year, t-1 as the manipulation year and t-2 as one year preceding the manipulation year. This notion is different from section 3.1 where t is the manipulation year and t-1 is the year preceding the buyouts. 15

ACCEPTED MANUSCRIPT Therefore, we expect that low visibility (proxied by analyst following and type of market listing) positively correlates to MBOs. The board needs to issue an independent evaluation of possible buyout choices and make a recommendation to investors. Therefore, a more

T

independent board and a stronger ownership stake held by the non-executive directors may

RI P

imply less collusion with the management, which may reduce the probability of an MBO. Lastly, we also include the cash balance, leverage ratio, and company size as control variables in the first stage regression.

SC

In the second stage, we replace the MBO dummy by the predicted MBO from the first-stage regression:

MA NU

Abnormali ,t 1 = β0 + β1 Pred_Dum_MBOi,t + β2 Dum_External Financingi,t +β3 NOAi,t -2 (INVRECi,t -2 ) +YearFixedeffects+ IndustryFixedeffects+εi,t

(7)

The dependent variable Abnormali,t-1 is MadjSize_ABN_TAACi,t-1, MadjSize_ABN_OCFi,t-1, and

MadjSize_ABN_PRODi,t-1

which

are

the

abnormal

accruals/operating

cash

flows/production costs of the buyout companies adjusted for the mean accruals/operating

The

ED

cash flow/production costs of the same size group within the same two-digit industry. management

engagement

incentive

variable

is

proxied

by

PT

Pred_Dum_MBOi,t and estimated in the first stage. We expect a negative coefficient because managers may manipulate earnings downwards prior to MBOs and subsequently benefit from

CE

a low purchase price (relative to IBOs). The variable Dum_External Financingi,t proxies for the external financing incentive and equals one when the target raises external funds at the

AC

transaction above the median and zero otherwise. We expect a positive sign, as external financing needs may mitigate downward manipulation. To alleviate the concern that we use an-ex post external financing needs proxy,

we also employ ex-ante measures of financial constraints: First, we use the size-age (SA) index, proposed by Hadlock and Pierce (2010) and calculated as (-0.737*Size + 0.043*Size2 0.040*Age), where size is the natural log of total assets. Companies belonging to the top 30 percent of SA Index are considered as being constrained. Second, we use the net leverage proxy introduced by Linck et al. (2013), which is measured as net debt (the sum of long term and short term debt minus excess cash) divided by the sum of net debt and shareholders’ equity. Companies ranking in the top 50 percent of positive net debt are considered financially constrained. Obviously, the degree of accounting manipulation is necessarily limited over time (Barton and Simko, 2002). The internal manipulation capacity is captured by the net 16

ACCEPTED MANUSCRIPT operating assets (NOAi,t-2), which is equity minus cash and marketable securities plus total debt (at the beginning of the year), divided by total sales (of the previous year). The level of the stock of inventories and receivables (INVRECi,t-2) captures the managerial flexibility to

T

manipulate real activities earnings (Roychowdhury, 2006). All the aforementioned accounting

RI P

variables are lagged and the definitions are presented in the Appendix. The independent variables are winsorized at 5th and 95th percentiles to control for outliers.

SC

4. Data Description

MA NU

4.1. Data Source and Sample Selection This study comprises all completed UK buyouts that occurred in the period from 1997 to 2007. The period corresponds with the second wave of LBOs in the UK, which picked up in 1997 and slowed down over the course of time and then fell abruptly with the emergence of the financial crises starting at the end of 2007. The transactions are retrieved

ED

from the database of the Mergers and Acquisitions of the Security Data Company’s online database (SDC), Venture Expert of Thomson One, Zephyr of Bureau van Dijk, Centre for Management Buyout Research (CMBOR), and Capital IQ. All deal information has been

PT

cross-checked by means of these datasets. To identify whether at least one member of the current management team participates in the transaction and stays in the firm subsequent to

CE

the buyout (our definition of an MBO), we gather the deal’s details from the above datasets as well as from the news releases in the Factiva, LexisNexis, Google news, and the offer

AC

documents. The accounting data is mainly obtained from Worldscope, but we complement missing information by means of annual reports. Corporate governance proxies are handcollected from annual reports and external financing information are gathered from offer documents (Thomson One). We collect a total of 407 buyout transactions and retain 16311 public-to-private transactions which satisfy the following criteria:  We retain 353 whole-company public-to-private buyouts: 14 private-to-private buyouts and 32 divisional buyouts are dropped for reasons of data limitations. Eight companies 11

This is not a small sample in the light of US MBO research: DeAngelo (1986) studies 64 MBOs (1973-1982). Perry and Williams (1994) examine 175 MBOs (1981-1988), and Fischer and Louis (2008) have 138 observations (1985-2005). Ang et al. (2010) study 179 MBOs (1997-2008). Li et al. (2013) collect 323 MBOs (1985-2005). These US studies (except Li et al., 2013) only require a minimum of five observations for their cross-sectional regressions, but we adopt a stricter rule for our cross-sectional regressions (minimum 10 observations, also adopted by Linck et al. 2013). For the UK, Wright et al. (2006) analyze 92 MBOs for the period 1997-2002. 17

ACCEPTED MANUSCRIPT that still remained public companies were also not included in the final database.  Transactions without all relevant accounting information in Worldscope reduced the sample to 299 buyouts.

T

 We excluded the financial services industry (SIC codes 6000-7000) and the regulated

RI P

industries (SIC codes 4400-5000), which reduced the sample to 233.

 Insufficient data to compute manipulation proxies reduced the sample to 191 (10 transactions with lacking SIC codes and 32 with incomplete financial information).

SC

 The inability to find a matching control firm leaves a sample of 173.  We dropped 10 observations, because we required at least 10 observations in each two-

MA NU

digit SIC industry per year to ensure the statistic power in the cross-sectional regressions. So, finally, we end up with a sample of 163 observations12. 4.2. Data Description Panel A of Table 1 shows the distribution of buyouts over time: the number of the buyouts has risen since 1997 and peaked around 1999-2000, consistent with Wright,

ED

Amess, Weir and Girma’s (2009) evidence that UK buyouts reached a new record in 2000 with total value of Euro 38.4 billion. Following the stock market downturn of early 2000, the

PT

buyout market rebounded in late 2002 and 2003. Our sample consists of companies from a wide business spectrum with most buyouts occurring in business services, retailing, and

CE

manufacturing industries (see Panel B of Table 1). In addition, buyouts in high-tech industry account for almost 14% of the total transactions. This trend is in line with Kaplan and

AC

Strömberg’s (2009) view that the industry scope of buyouts is broadening beyond the mature, high cash flow, high debt capacity type of industries. [Insert Table 1 about here]

The total assets of the average sample firm equal GBP 171.3 million in the year prior to the buyout (Panel A of Table 2). In two thirds of our buyout sample, at least one incumbent manager is involved in the transaction and stays on subsequent to the buyout. MBOs are relatively smaller, less levered, growing faster, and more cash-rich than IBOs

12

Ecker, Francis, Olsson and Schipper (2013) address sample attrition caused by cross-sectional methods to detect discretionary accruals and suggest a practical solution based on size matching. Our sample attrition rate is 14.66% ((191-163)/191), which is modest and far below the attrition rate for non-US data pointed out in Ecker et al.’s paper (minimally 32%). We resort to a cross-sectional method to measure earnings manipulation because, according to Ecker et al. (2013), when the sample attrition is not a big concern, a cross-sectional approach is preferred because of its a higher explanatory power. To complement this approach, we also use a matching procedure based on performance, as well as matching based on both performance and growth rate. 18

ACCEPTED MANUSCRIPT (Panel B). MBOs are associated with large ex ante equity stakes held by managers (17.3% versus only 6.0% in IBOs) and the management is more frequently the largest shareholder. Institutional ownership concentration does not differ between MBOs and IBOs. IBOs have a

T

higher proportion of independent directors than MBOs (47.82% versus 43.68%) and are

RI P

followed by twice as many analysts13. [Insert Table 2 about here]

SC

5. Results

MA NU

5.1. Measuring Earnings Manipulation We first address the question as to how managers manipulate earnings numbers (if at all) by means of accrual management and/or real earnings activities. Next, we test whether external financing incentives affect (or balance) managerial incentives to manipulate. Finally, we provide evidence on asset reserves revaluation.

ED

5.1.1. Managerial Engagement Incentives (H1) Accrual Management

PT

From Panel A of Table 3, we observe that buyout companies have negative total raw accruals amounting to -3% (RAW_ABN_TAACt-1). This degree of downward accrual

CE

management is in line with the US literature (Perry and Williams, 1994; Fisher and Louis, 2008). Both MBOs and IBOs are subject to negative accrual management (-3% and -2%,

AC

respectively, but the difference is not significant). When we adjust the raw abnormal accruals for the industry-mean total accruals (Madj_ABN_TAACt-1) or for the mean of the same industry size group (MadjSize_ABN_TAACt-1), we can draw two conclusions: First, the abnormal accrual figures become significantly more negative: for all buyout companies, the accruals are -12% lower than the average firm in the same industry and than the average firm from the same industry and within a similar size category. Non-buyout companies are generally subject to positive accrual management (by 9% of the assets) 14. This finding is unsurprising, because positive manipulation can affect managers’ bonuses and the likelihood of meeting or beating analysts’ expectations which may trigger a positive market reaction. 13

The correlation between all independent variables is small with exception of a positive correlation of 0.6 between the number of analysts following the firm and firm size. We run both tests: (i) include these two variables simultaneously, (ii) these two variables are not simultaneously included in our models. 14 We find the positive earnings manipulation for the industry peers, because on average, these public firms have an incentive to manipulate earnings upwards. At any point in time, some companies manipulate earnings upwards although this cannot be a persistent strategy for all companies. 19

ACCEPTED MANUSCRIPT Second, the difference in industry-adjusted abnormal accruals of MBOs and IBOs is striking: downward accrual management is twice as high in MBOs (-15%) than in IBOs (-7%). A similar difference is presented for an industry-size correction.15

T

When we use a performance-matched modified-Jones regression model, which

RI P

controls for the effect of performance on accruals by assigning to each target a non-buyout counterpart from the same industry and a performance profile that is similar in the manipulation year. The difference in abnormal accruals of the buyout targets and that of

SC

control companies yields peer-controlled abnormal accruals (ABN_TAACt-1). The approach yields very similar results: for both MBOs and IBOs, the downward accruals manipulation is

MA NU

significantly negative, but the manipulation in MBOs is even much larger (about eight times) than in IBOs.

In sum, in spite of the improved corporate governance over the past 15 years, downward earnings management preceding buyouts still frequently takes place, as persistently indicated by four types of accrual proxies. Moreover, MBOs are associated with

ED

larger deflated accrual manipulation than IBOs. The abnormal accruals of MBOs account for approximately 30% 16 of reduced earnings and are thus not only statistically but also

PT

economically significant. The fact that IBOs (which do not involve the pre-transaction management) have still applied some limited degree of earnings management may be

CE

explained by ex- ante probability that the incumbent management may still be involved in the post-LBO firm. The findings of this subsection strongly support the managerial engagement

AC

hypothesis (H1) in that managers deflate earnings numbers by means of accrual management. [Insert Table 3 about here]

Real Earnings Management We turn to real earnings management and focus on sales and production

manipulation. Panel B of Table 3 indicates that the abnormal operating cash flows are positive for both MBO and IBO targets, which is in line with the prediction that managers will delay sales to depress net income by using real earnings management. For instance, a reduction in lenient credit terms will decrease the sales volumes and therefore lead to low earnings numbers, but will increase the collection of current sales’ receipts and thus raise the

15

When we base our tests on the differences between the medians for the MBOs and IBOs, we find that the results are similar for results reported in this section. Tables are available upon request. 16 The mean values of abnormal accruals and net income are £ -1.9 million and £ 4.4 million respectively, which is in relative terms 30% (1.9/(4.4+1.9)) of earnings. 20

ACCEPTED MANUSCRIPT level of OCF. We observe that sales manipulation is carried out in MBOs (the four proxies are statistically significantly different from zero), but the evidence that this also occurs in IBOs is weak. This finding also supports Hypothesis 1, as managers seem to manipulate earnings

T

downwards by delaying sales. One further point regarding our industry-adjusted buyout-

RI P

specific approaches needs to be made: since both the industry-mean adjusted OCF and the same industry-size group adjusted OCF are lower than the raw OCF, it implies that the industry peers (the non-LBO firms) engage in negative sales manipulation to boost earnings

SC

numbers. This is consistent with the motive of positive accrual management used by the industry peers for increasing the bonus or meeting/beating analyst forecast.

MA NU

In relation to production manipulation, Panel C of Table 3 further supports hypothesis 1, because negative production manipulation occurs prior to buyouts leading to lower earnings figures. This implies that managers can slow down production to manage earnings downwards. We also show that MBOs are related to significant under-production manipulation, whereas production manipulation in IBOs does not occur according to the

ED

industry-adjusted buyout-specific and the matching-adjusted approaches. MBO targets decrease production while industry competitors increase production to inflate the earnings

PT

numbers, which is consistent with the role of positive accrual management and negative sales manipulation.

CE

In sum, in addition to the downward accrual management, we have presented further evidence that suggests that negative real earnings management is applied preceding MBOs (but not IBOs). Both types of earnings management support Hypothesis 1.

AC

Since accrual management and real earnings management may be correlated,

we report the correlation matrix in Table 4. Abnormal accruals and abnormal cash flows are significantly negatively correlated, which implies that companies are engaging in accrual management and real earnings management at the same time. Likewise, the negative correlation between abnormal cash flows and abnormal production costs suggests that both types of real earnings management are initiated by the average MBO. In addition, the positive operating cash flows may also capture the effect of under-production, as the under-production reduces production costs and cash outflows and thus leading to higher OCF. [Insert Table 4 about here] 5.1.2. External Financing Incentives (H2) The Level of Leverage and Accrual Manipulation 21

ACCEPTED MANUSCRIPT We relate external financing needs to accounting manipulation and test the difference in accrual manipulation between buyouts with high and low external financing needs. Buyouts with little external financing needs face fewer financing restrictions and have

T

fewer incentives to manipulate earnings upwards. We do not obtain any significant

RI P

differences for the subsamples of MBOs and IBOs. We also use ex-ante financial constraints proxies such as the SA index and the net leverage measure (see section 3.3) and find no significant difference between financially constrained and unconstrained companies in terms

SC

of sales manipulation.17 Thus, the univariate tests cast doubt on the manipulation incentive

MA NU

induced by external financing needs.

Asset Revaluation Manipulation (H3) Asset reserves revaluation can be used to influence the pre-buyout leverage ratio and thus increase the debt capacity for the buyout companies. Given that asset revaluation is industry-specific (in industries with high capital intensity, firms can revalue more), we control for industry effects by adjusting the raw figures for: (i) the industry mean,

ED

(ii) the mean of the industry-size group, and (iii) peer-effects by employing a matched control sample of non-buyout firms. The revaluations with these three adjustments consistently show

PT

that managers do not manipulate the value of the assets through revaluations in MBOs, but do so in case of IBOs (Panel A of Table 5). The reason is that managers who anticipate that they

CE

will not be involved in the post-buyout phase will not bear the future losses arising from upward revaluation (for instance, higher depreciation and lower profit on asset sales).

AC

Therefore, they are more likely to revalue the assets upwards which reduces the debt-toequity ratio and in turn increases the debt capacity of the unlevered transaction. In terms of the importance of revaluations, a caveat is at place: when we dig deeper into the details on the asset revaluation reserves as reported in the annual reports, we find that in 70.3% of the MBOs managers do not revaluate assets (Panel B of Table 5). In the vast majority of these cases (87%), only an upward manipulation is possible as the asset revaluation reserves were already at zero prior to the buyout. In short, in Panel A of Table 5, we show in absolute terms, IBOs are associated with more upward revaluation than MBOs, which is consistent with Hypothesis 3. However, it should be noted that the evidence is not very strong, for neither MBOs nor IBOs frequently revalue their assets. The findings on revaluations do not support Hypothesis 2 (the external 17

The exception is more negative accrual manipulation for firms with lower ex-ante financial constraints (as measured by the SA index and net leverage ratio), but for these measures we do not find real earnings manipulation. The tables are not shown for reasons of conciseness but are available upon request. 22

ACCEPTED MANUSCRIPT financing incentive) because upward revaluations are rarely used to increase the borrowing capacity by ex-ante reducing the debt-to-equity ratio. The finding is line with that of Alexson et al. (2013) that there is no relationship between the leverage ratio and the buyout debt

T

financing. The reason may be that when credit markets are booming, revaluations are not

RI P

necessary to increase debt capacity. Indeed, we observe that during the second buyout wave it was easier to attract external funds, considering the growth in the high yield bond market by more than 600% since 1997.

SC

[Insert Table 5 about here]

MA NU

5.1.3. Robustness Tests

First, it is possible that the management has made the manipulation decision not in the year or months prior to the buyout transaction but at an earlier time. Therefore, we measure all accounting manipulation proxies at a time preceding the transaction by more than one year (the fiscal year is then ending 13 to 24 months prior to the buyout). Overall, we

ED

hardly find any significant results for the year prior to what we rightfully call the manipulation year. If accounting manipulation or asset reserves revaluation takes place, it

PT

occurs immediately preceding the buyout transaction. Second, we examine whether the enactment of the revised UK Corporate

CE

Governance Code of 2003 reduces the degree of accounting manipulation. Following the introduction of Sarbanes-Oxley (SOX) Act of 2002, the “Combined Code” of 1998 was

AC

revised in 2003 to improve financial reporting quality by enhancing the accountability of the board of directors and the audit committees. Independent boards (Kleiv, 2002; Cornett et al., 2008) and the quality of audit committees (Abbott et al., 2004; Vafeas, 2005) do indeed significantly affect reporting quality. We partition the sample period into two subperiods: 1997-2003 and 2004-2007. From Panel A of Table 6, we observe that active accrual manipulation was larger before the change in corporate governance regulation (1997-2003), although there is still evidence of manipulation subsequent to 2003. After the change in the corporate governance regime, manipulation techniques (related to production costs and asset revaluations) are more frequently used, which may be induced by the fact that these methods are more difficult to detect (Graham, Harvey and Rajgopal, 2005). This finding is also consistent with US evidence: since the adoption of SOX, companies shift from accrual management to real earnings management (Cohen et al., 2008). When we redo the above tests only for the sample of MBOs, the above findings are upheld. 23

ACCEPTED MANUSCRIPT [Insert Table 6 about here]

Third, we also perform a time-series approach to estimate abnormal accruals as

T

in e.g. Perry and Williams (1994) and Hafzalla (2009). For each individual buyout company,

RI P

we run a time-series regression using company data over a six year period ending in the year prior to the manipulation year to measure the normal accruals. The limitation of this method is that a sufficiently long time series (we take at least six years) of accounting numbers prior

SC

to the manipulation period should be available for each firm in order to estimate the parameter coefficients. Although this approach reduces the sample size to 72 observations,

MA NU

we still find negative accrual management proceeding MBOs. We also conduct a robustness check by adjusting abnormal accruals for both performance and sales growth (Pungaliya and Vijh, 2009) and obtain similar results.

5.2. The Incentives Underlying Earnings Manipulation We now turn to the reasons why the management engagement incentive

ED

dominates and study whether the external financing incentive mitigates downward

PT

manipulation.

5.2.1. Managerial Engagement Incentives versus External Financing Incentives

CE

It is important to note that when we relate the earnings manipulation variables to the independent variable MBO versus IBO (a dummy variable which equals 1 in case of an

AC

MBO and zero otherwise), the type of transaction measures the ex-ante probability of management engagement with error. In some companies, the management may contemplate an MBO, but the transaction may end up to be an IBO (or vice versa) which imposes a bias on the resulting coefficients from the regression models. To address this concern, we make use of a two-stage instrumental variable method. The first-stage equation models the MBO/IBO choice and the second equation explains the degree of accounting manipulation. In other words, we test whether or not managers manipulate earnings when they perceive the buyout type. As suggested by Berry (2011), an OLS model is preferred in the first stage even for a dummy variable, the reason being that only an OLS estimation produces first stage residuals that are uncorrelated with the covariates and fitted values. As a robustness check, we will also employ a probit model for the first stage estimation following Wooldridge (2002). The choice of our instrumental variables (IVs) for the MBO/IBO depends on 24

ACCEPTED MANUSCRIPT the relative ex ante ownership concentration held by management and non-executive directors, and the firm size, which influences the ability to take over the company by the management, which usually faces personal financial constraints. Panel A of Table 7

T

demonstrates that these IVs are significantly related to the MBO decision. The higher is the

RI P

manager’s equity investment in the target company, the higher the probability of an MBO. When the level of non-executive ownership is higher and the target firm is larger, the company is more likely to undergo an IBO. The Hausman endogeneity test rejects the null

SC

hypothesis that the realized buyout type is exogenous. A p-value of 0.2 from the overindentifying restriction test indicates that at least one of the IVs is exogenous. To test the

MA NU

relevance of the IVs, the F-statistics are required to be larger than 10 to avoid weak IVs; our F-test amounts to 18.2 which implies that our IVs are characterized by a sufficiently large correlation with the endogenous regressor.

The main finding of the second stage is that the predicted MBO proxy is significantly negatively related to the abnormal accruals (Models 1a and 1b, Panel B of Table

ED

7) and positively to abnormal operating cash flows (Models 2a and 2b). Whereas the latter can reflect both sales and production manipulation, the use of sales manipulation is most

PT

likely because a separate test on production cost manipulation does not yield significant results (Models 3a and 3b). The findings of Models 1a,b and 2a,b support Hypothesis 1 in

CE

that managers are more prone to participate in accounting manipulation in order to obtain a lower purchase price via both accrual and real earnings manipulation. In case of an MBO, the mean abnormal accruals (standardized by total assets) are 17.7% lower than the accruals of

AC

firms of the same size group and within the same industry (Model 1a). When we use both the (realized) financing needs variable (Model 1a, Panel B of

Table 7) and the ex- ante financial constraints proxies 18 (Model 1b) to test the external financing incentive, we show that this incentive does not emerge as a mitigating force for either accrual or real earnings manipulation. The reason for its insignificance may be that over the period 1997 to 2007 a fast-growing high-yield bond market emerged (the GBP 5.4 billion high-yield bond market of 1997, soared to GBP 32 billion in 2007). Axelson et al. (2013) argue that the main robust predictor of buyout leverage consists of the credit market

18

In addition, following Linck et al.’s (2013) approach, companies belonging to the top (bottom) 30 percent of SA Index are considered as being constrained (unconstrained). Two dummies are generated accordingly: Dum_Contrained indicates the financially constrained companies and Dum_Unconstrained indicates the financially unconstrained companies. When we include both dummies in the regression (using the group of companies whose SA is between 30 to 70% of the sample SA as the base), the coefficient on financial constraints remain insignificant. 25

ACCEPTED MANUSCRIPT conditions of the high-yield bond market rather than the standard determinants of capital structure, such as leverage and earnings. Thus, our Hypothesis 2 on the external financing incentive is not upheld. The inactive asset revaluation frequency presented in Panel B of

RI P

[Insert Table 7 about here]

T

Table 5 is in line with this finding.

5.2.2. Robustness Tests

SC

To verify the results of the above subsection, we perform three robustness tests. First, as an alternative dependent variable for accrual management, we use the

MA NU

performance-matched abnormal accruals (see Section 3). The perceived MBO probability still has a significantly negative impact on accrual manipulation (-0.147 in Model 1 of Table 8), which supports hypothesis 1. When we use either the raw abnormal operating cash flow (OCF) as a proxy for sales manipulation, the perceived MBO remains positive and statistically significant (0.069 in Model 2 of Table 8), which also supports hypothesis 1.

ED

These results do not change when we use industry mean adjusted OCF. Second, we use two alternative estimation approaches. In the first stage, we use

PT

a probit model (rather than OLS) to predict the MBO likelihood and then use this predicted value as a regressor in the second stage. We confirm that the management engagement

CE

incentive plays a crucial role in negative accrual manipulation (Model 3 of Table 8). We also apply a GMM IV approach and obtain a coefficient for the predicted MBO (-0.180 in Model 4) which is very similar to that of the two-stage approach (-0.177). As the standard errors are

method.

AC

close, there is almost no efficiency gain from a GMM approach relative to a two-staged

Third, we explore the effect of the enactment of the revised UK Corporate

Governance Code of 2003 on both accrual and real earnings activity manipulation. Model 5 of Table 8 shows that the implementation of the revised Code (as captured by the interaction term) reduces the magnitude of manipulation in the case of an MBO. This suggests that the revised Code has improved the reporting quality of a buyout company and hence reduced the potential manipulation opportunities. Model (6) estimates the effect of the revised Code on sales manipulation. Likewise, since the Code’s revision, the real earnings manipulation in predicted MBOs is reduced as well (the coefficients of the MBO dummy and the interaction term are identical, although the p-value of the latter is only 11%). Taking these two pieces of evidence on accrual and real earning manipulations together, we argue that even soft law such 26

ACCEPTED MANUSCRIPT as the revised Code enhances the reporting integrity of companies during the MBO event, which could therefore lead to a fairer and more transparent representation of earnings and cash flows.

T

[Insert Table 8 about here]

RI P

5.3. Discussion We elaborate on three issues: the impact of external financing on accounting

SC

manipulation, the market for corporate control, and the endogeneity issue. We had posited the need for external financing as a mitigating incentive to downward earnings management (if much external financing is needed, downward earnings

MA NU

manipulation could be lower). However, downward earnings manipulation can lower the buyout price and hence also the need for external borrowing to finance the transaction (in an absolute sense) may be lower. We test the relationship between the magnitude of earnings manipulation and the probability and the amount of external borrowing. We find that buyouts with abnormal accruals below the sample median are more likely to use external financing

ED

than buyouts with accruals above the median – a difference that is statistically significant at the 5% level. We find that buyouts with abnormal accruals below the median borrow more

PT

(both in absolute and in relative terms to total assets) although the difference is not significant. These two tests do not support of the alternative proposition on external financing in that

CE

negative earnings management is not negatively related to external financing needs. We find some support for Axelson et al.’s (2013) proposition that the degree of external financing in

AC

buyout transactions relies more on credit market conditions. Indeed, in ‘hot’ LBO years (1999 and 2000), the degree of both negative accrual manipulation and external borrowing (per total assets, one year before manipulation year) are higher.19 Downward earnings manipulation may make takeover targets look (initially) cheap and induce the emergence of multiple bidders. Hence, the price effect of negative earnings management could be undone by competitive bidding. We check bid information from the databases listed in Section 4.1 and also refer to Mergerstat reports which comprise the context on each buyout. Only seven companies (four MBOs and three IBOs) receive competitive bidding; excluding these companies does not affect our findings. In relation to the endogeneity issue, there can be three sources of endogeneity: simultaneity, unobservable heterogeneity, and the possibility that current values of

19

Tables available upon request. 27

ACCEPTED MANUSCRIPT governance variables are a function of past firm performance. To address simultaneity, (i) we use lagged independent variables to test the manipulation timing (the years before the manipulation year); (ii) to verify that the causality goes from the buyout decision to earnings

T

management in regressions, we estimate the realized buyout type dummy variable on

RI P

different proxies for earnings manipulation in addition to factors influencing the buyout choice. From untabulated results, we do not find any significant impact of earnings management on the choice of buyout type. Moreover, the insignificant relation between

SC

accounting manipulation two years before the buyouts and the buyout decision also suggests that the buyout decision causes earnings manipulation and not the other way around

MA NU

(consistent with the first robustness test in Section 5.1.3.). Eliminating unobservable heterogeneity is difficult as we do not have a panel data structure. To address concerns that previous firm performance or accrual performance affect the governance variables (in our case, ownership structure), we use performance– adjusted or performance-matched dependent variables in the regression. We further conduct a

ED

series of tests: (i) we add the abnormal accruals one year before the manipulation year as an additional control variable. We do not find the past accrual management affects the impact of

PT

governance variables; (ii) we control for performance lags from two periods before the manipulation year, similar to Wintoki, Linck and Netter’s (2012) dynamic OLS model (their

CE

Table 9). We find that our main results are upheld and the lagged performance measures are statistically insignificant. As an additional test, we test how strongly ownership structure relates to the past performance. We regress ownership structure variables at the manipulation

AC

year on the lagged performance and firm characteristics. We do not obtain significant result of past performance on ownership structure. 6. Conclusions We have investigated whether accounting manipulation occurred prior to leveraged buyout transactions in the UK during the second buyout wave of 1997 to 2007 (at which point the buyout market collapsed following the banking crisis). We distinguished among earnings manipulation through accrual management, real earnings management by means of sales and production, and asset reserves revaluation. We examined whether (and how) earnings manipulation takes place in LBOs but also why by distinguishing between (i) managerial engagement incentives to manipulate earnings negatively in order to drive the share price down and buy the firm on the cheap in an MBO transaction, and (ii) financing 28

ACCEPTED MANUSCRIPT incentives to manipulate earnings upwards (in case of an IBO) or less downwards (in case of an MBO) which is motived by the need to increase the debt capacity of the firm in order to facilitate an LBO transaction. We have defined IBOs as LBO transactions in which the pre-

T

transaction management is no longer involved as either managers or equity holders.

RI P

We find that buyout targets engage in negative earnings manipulation in the year prior to the buyout transaction (the manipulation year), through both accrual management and real earnings management, whereas the average non-buyout firm engages in

SC

positive earnings manipulation. The latter finding can be explained by the fact that managers intend to reach analysts’ forecasts and targets expected by the market, which may have a

MA NU

positive impact on the share price and hence directly benefits management whose remuneration package may consist of equity-based pay. Furthermore, firms can attempt to reduce financial constraints which enable them to increase their borrowing ability for e.g. large capital expenditures (Linck et al, 2013).

When we dissect the sample of buyouts into MBOs and IBOs, we demonstrate

ED

that MBOs are associated with significantly more - twice as much in terms of industry-size adjusted accruals - downwards accrual manipulation than IBOs. This is not unexpected

PT

because a management team contemplating an MBO may try using earnings understatement to negatively influence the acquisition price. Although we would not expect this type of

CE

behavior in IBOs with no ex post managerial involvement, we still find some evidence (albeit at a lower level) of negative earnings manipulation which reflects the ex-ante uncertainty about what the type of buyout transaction will eventually arise. It should be noted that the

AC

degree of downward accrual manipulation is also economically significant: the abnormal accruals of MBOs account for approximately 30% of reduced earnings. Likewise, we show that negative real earnings manipulation by adjusting sales and production takes place in MBOs (but not in IBOs) as reflected in operating cash flows and production costs. All in all, our findings on earnings manipulation strongly support the managerial engagement hypothesis for MBOs. The external financing incentive which we expected to be visible in the mitigation of downwards earnings manipulation in MBOs or in upward manipulation in IBOs is not supported by our data. This may not surprise in a context of lenient credit market conditions (abundant availability of capital, also stimulated by a growing high-yield bond market). We expect positive manipulation of asset reserves revaluation for IBOs (for external financing reasons) and negative manipulation for MBOs (the managerial engagement 29

ACCEPTED MANUSCRIPT hypothesis), and find that this is indeed the case although only a small number of buyout firms do actually revalue reserves indicates that asset revaluation is not an important channel for accounting manipulation. Finally, we also examine whether earnings manipulation is

T

affected by changes in regulation, more specifically by the revised UK Corporate Governance

RI P

Code of 2003, and conclude that the enhanced soft corporate governance law regulation does

AC

CE

PT

ED

MA NU

SC

indeed curb both accrual and real earnings manipulation in MBOs.

30

ACCEPTED MANUSCRIPT References Axelson, A., Jenkinson, T., Strömberg, P., Weisbach, M.S., 2013, Borrow cheap, buy high? The determinants of leverage and pricing in buyouts, Journal of Finance 68: 2223-2267.

T

Abbott, L., Parker, S., Peters, G., 2004, Audit committee characteristics and restatements, Auditing 23: 69-87.

from the UK, Journal of Accounting and Economics 26:149–178.

RI P

Aboody, D., Barth, M., Kasznik, R., 1999, Revaluations of fixed assets and future firm performance: evidence

Ang, J., Hutton, I., Majadillas, M.A., 2014, Managerial divestment in leveraged buyouts, European Financial

SC

Management 20: 462-493.

Balatbat, M.C.A., Lim C.Y., 2003, Earnings management and share market performance: Further evidence from equity carve-outs, working paper.

Economics 45: 324-349.

MA NU

Ball, B., Shivakumar, L., 2008, Earnings quality at initial public offerings, Journal of Accounting and

Barton, J., Simko, P.J., 2002, The balance sheet as earnings management constraint, The Accounting Review 77: 1-27.

Bebchuk, L. and Fried, J. 2003, Executive compensation as an agency problem, Journal of Economic

Perspectives 17, 71-92.

ED

Berry, 2011, http://harrisschool.uchicago.edu/Blogs/EITM/wp-content/uploads/2011/06/IV-Lecture-2.pdf. Bhojraj, S., Swaminathan, B., 2007, How does the corporate bond market value capital investments and accruals? Review of Accounting Studies 14: 31–62.

PT

Black, E.L., Sellers, K., Manly, T.S., 1998, Earnings management using asset sales: An international study of countries allowing noncurrent asset revaluation, Journal of Business Finance and Accounting 25: 1207-

CE

1317.

Botsari, A., Meeks, G., 2008, Do acquirers manage earnings prior to a share for share bid? Journal of Business Finance & Accounting 35: 633-670.

AC

Bradshaw, M.T., Richardson, S.A., Sloan, R.G., 2001, Do analysts and auditors use information in accruals? Journal of Accounting Research 39: 45-74. Cheng, C.S., Lin, W.J., 2009, When do firms revalue their assets upwards? Evidence from the UK, International Journal of Accounting and Information Management 17: 166-188. Cohen, D., Dey, A., Lys, T., 2008, Real and Accrual-based Earnings Management in the Pre- and Post-Sarbanes Oxley Periods, The Accounting Review 82: 757-787. Cohen, D., Zarowin, P., 2010, Accrual-based and real earnings management activities around seasoned equity offerings, Journal of Accounting and Economics 50: 2-19. Cornett, M.M., Marcus, A.J., Tehranian, H., 2008, Corporate governance and pay-for-performance: The impact of earnings management, Journal of Accounting and Economics 87: 357-373. Das, A., Chon, G., 2011, “Love letters” for a takeover private equity firms’ new way to seduce companies starts with an Email, The Wall Street Journal. DeAngelo, L., 1986, Accounting numbers as market valuation substitutes: a study of management buyouts of public stockholders, The Accounting Review 61: 400-420. Dechow, P., 1992, Accounting earnings and cash flows as measures of firm performance: The role of accounting 31

ACCEPTED MANUSCRIPT accruals, working paper. Dechow, P., Dichev, I., 2002, The quality of accruals and earnings: the role of accrual estimation errors, The Accounting Review 77: 35–59. Dechow, P., Sloan, R.G., Sweeney, A.P., 1995, Detecting earnings management, The Accounting Review 70:

T

193225.

RI P

Dechow, P., Sloan, R.G., Sweeney, A.P., 1996, Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC, Contemporary Accounting Research 13: 1-36. Dechow, P., Ge, W.L., Schrand, C., 2010, Understanding earnings quality: A review of the proxies, their

SC

determinants and their consequences, Journal of Accounting and Economics 50: 344-401. Easton, P., Eddey, P., Harris, T., An investigation of revaluation of tangible long-lived assets, Journal of Accounting Research 40: 105-134.

MA NU

Ecker, F., Francis, J., Olsson, Per, Schipper, K., 2013, Estimation sample selection for discretionary accruals models, Journal of Accounting and Economics 56: 190-211.

Erickson, M., Wang, S.W., 1999, Earnings management by acquiring firms in stock for stock mergers, Journal of Accounting and Economics 31:1-38.

Fischer, P.E., Louis, H., 2008, Financial reporting and conflicting managerial incentives: the case of management buyouts, Management Science 54: 1700-1714.

ED

Fidrmuc, J., Palandri, A., Roosenboom, P.G.J., Van Dijk, D.J.C., 2013, When do managers seek private equity backing in public-to-private transactions? Review of Finance17:1099-1139. Graham, J.R., Harvey, C.R., Rajgopal, S., 2005, The economic implications of corporate financial reporting,

PT

Journal of Accounting and Economics 40: 3-73. Guo, S.R., Hotchkiss, E.S., Song, W.H., 2011, Do buyouts (Still) create value, Journal of Finance 66: 479-517.

CE

Hadlock, C.J., Pierce, J.R., 2010, New evidence on measuring financial constraints: moving beyond the KZ index, Review of Financial Studies 23: 1909-1940. Hafzalla, N.M., 2009, Managerial incentives for discretionary disclosure: evidence from management leveraged

AC

buyouts, Review of Accounting Studies 14: 507-533. Hribar, P., Collins, D., 2002, Errors in estimating accruals: Implications for empirical research, Journal of Accounting Research 40: 105-134. Jackson, R.J. , 2013, Private Equity and Executive Compensation, UCLA Law Review 60, 638-677. Jones, J., 1991, Earnings management during import relief investigations, Journal of Accounting Research 29: 193-228. Kaplan, S.N., Strömberg, P., 2009, Leveraged buyouts and private equity, Journal of Economic Perspectives 23:121–46. Kasznik, R., 1999, On the association between voluntary disclosure and earnings management, Journal of Accounting Research 37:57–81. Klein, A., 2002, Audit committee, board of director characteristics, and earnings management, Journal of Accounting and Economics 33: 375-400. Kothari, S.P., Leone, A.J., Wasley, C.E., 2005, Performance matched discretionary accrual measures, Journal of Accounting and Economics 39: 163-197. Leary, M.T., 2009, Bank loan supply, lender choice and corporate capital structure, Journal of Finance 64: 32

ACCEPTED MANUSCRIPT 1143-1185. Li. X., Qian, J., Zhu, L., 2013. Earnings-reducing activities before management buyouts, working paper. Linck, J.S., Netter, J., Shu, T., 2013. Can managers use discretionary accruals to ease financial constraints? Evidence from discretionary accruals prior to investment? The Accounting Review 88: 2117-2143.

T

Lin, Y.C., Peasnell, K.V., 2000, Fixed asset revaluation and equity depletion in the UK, Journal of Business

RI P

Finance and Accounting 27: 359-394.

Longstreth, B., 1984, Management buyouts: Are public shareholders getting a fair deal? (Speech by SEC Commissioner Bevis Longstreth to the International Bar Association, Toronto, October 6, 1983), Fed. Sec. L.

SC

Rep. (CCH) Sec. 83.

Palepu, K., Healy, P., Berbard, V., 2000, Business analysis and Valuation: using financial statements, 2 nd edition, Southwestern Publishing, Cincinnati.

MA NU

Peasnell, K.V., Pope, P.F., Young, S., 2000, Detecting earnings management using cross-sectional abnormal accruals models, Accounting and Business Research 30: 313-326. Perry, S., Williams, T., 1994, Earnings management preceding management buyout offers, Journal of Accounting and Economics 18: 157-179.

Renneboog, L., Simons, T., Wright, M., 2007, Why do public firms go private in the UK? The impact of private equity investors, incentive realignment and undervaluation, Journal of Corporate Finance 13: 591-628.

and Economics 42: 335-370.

ED

Roychowdhury, S., 2006, Earnings management through real activities manipulation, Journal of Accounting

Shivdasani, A., Wang, Y.H., 2011, Did structured credit fuel the LBO Boom? Journal of Finance 66: 1291-

PT

1328.

Yates, G. and Hinchliffe, M., 2010, A practical guide to private equity transactions, Cambridge University Press.

CE

Vafeas, N., 2005, Audit committees, boards, and the quality of reported earnings, Contemporary Accounting Research 22: 1093-1122.

Wintoki, M.B., Linck, J.S. , Netter J.M., 2012, Endogeneity and the dynamics of internal corporate governance,

AC

Journal of Financial Economics 105: 581-606. Wooldridge, J., 2002, Econometric analysis of cross section and panel data, MIT Press. Wright, M., Amess, K., Weir C., Girma, S., 2009, Private equity and corporate governance: Retrospect and prospect, Corporate Governance: An International Review 17: 353-375. Wright, C.J., Shaw, J.R., Guan, L.M., 2006, Corporate governance and investor protection: Earnings management in the U.K. and U.S., Journal of International Accounting Research 5: 25-40. Wu, Y.U., 1997, Management buyouts and earnings management, Journal of Accounting and Finance 12: 373389.

33

ACCEPTED MANUSCRIPT Appendix: Definition of Variables

Table 1 defines the variables and presents the data sources.

Definition

RI P

Variables

T

Panel A: Dependent variables

CE

PT

ED

MA NU

SC

First stage regression dependent variable Dum_MBO Dummy variable equals 1 in case of an MBO (one or more members of the pre-transaction management team participates in the buyout and subsequently stay in the firm as important shareholders), and 0 in case of an IBO (institutional buyout). Second stage regression dependent variables ABN_TAAC Matched Abnormal Accruals: raw abnormal accruals minus abnormal accruals of matched control observations selected from non-buyout companies with the same two-digit SIC code and in the same year and with the closest ROA. MadjSize_ABN_TAAC Industry-Size Mean Adjusted Abnormal Accruals: raw abnormal accruals minus mean abnormal accruals of the control observations for the same year and with the same size group at the same two-digit SIC code. MadjSize_ABN_OCF Industry-Size Mean Adjusted Abnormal Operating Cash Flow: raw abnormal operating cash flow minus mean abnormal operating cash flow of control observations for the same year and with the same size group at the same two-digit SIC code. MadjSize_ABN_PROD Industry-Size Mean Adjusted Abnormal Production Cost: raw abnormal production costs minus mean abnormal production costs of control observations for the same year and with the same size group at the same two-digit SIC code. RAW_ABN_OCF Raw Abnormal Operating Cash Flow.

Source SDC, Capital IQ, Zephyr, Venture Expert and news release.

Calculations with Worldscope data

Calculations with Worldscope data

Calculations with Worldscope data

Calculations with Worldscope data

Calculations with Worldscope data

Variables

AC

Panel B: First stage regression independent variables

Analysts Board Size Cash to Assets

Debt to Assets Independent Directors Largest Owner Instit LSE Listing

Management Own MTB Non-Executive Own ROA SIZE

Definition

Source

Number of financial analysts following pre-buyout target. Number of directors on the board. Cash and marketable securities divided by total assets of (pre-buyout) target. Total debt divided by total assets of (pre-buyout) target. Proportion of independent directors on the board. Dummy variable equals 1 if an institutional investor is the largest shareholder in pre-buyout target and 0 otherwise. Dummy variable equals 1 when the pre-buyout target listed on the London Stock Exchange, and 0 when listed on the Alternative Investment Market. Ownership stake (%) held by management in pre-buyout target. Market-to-book value of (pre-buyout) target. Ownership stake (%) held by non-executives in pre-buyout target. Return on assets of (pre-buyout) target. Logarithm of total assets of (pre-buyout) target.

DataStream Annual report Worldscope Worldscope Annual report Annual report DataStream

Annual report Worldscope Annual reports Worldscope Worldscope 34

ACCEPTED MANUSCRIPT

Panel C: Second stage regression independent variables Definition

Dum_External Financing

Dummy equals 1 if the target raises external funds at the transaction above the median and 0 otherwise.

Source

INVREC

Sum of inventories and receivables, divided by total assets. Net operating assets: Sum of shareholders’ equity minus cash and marketable securities and plus total debt, divided by total sales. Predicted MBO obtained from the first stage regression (of 2SLS model). SA index is calculated as (-0.737*Size + 0.043*Size2 0.040*Age), where size is the natural log of total assets (in millions). We winsorize book value at 4.5 billion dollar (adjusted to GBP) and age at 37 years (Hadlock and Pierce, 2010). Companies belonging to the top 30 percent of SA Index are considered constrained (based a group of companies excluding utilities and financial firms from 1990 to 2007). Therefore, we create a dummy variable equals 1 if the company is financially constrained and 0 otherwise. Dummy variable equals 1 if the buyout year is after the implementation of the revised Corporate Governance Code in 2003 and 0 otherwise.

RI P

T

Variables

SC

NOA

MA NU

Pred_Dum_MBO

ED

SA index

Worldscope

Calculation with Worldscope data

CMBOR, SDC, Capital IQ, Zephyr, and Venture Expert

AC

CE

PT

YearCode

SDC, Capital IQ, Zephyr, Venture Expert and offer documents. Worldscope

35

ACCEPTED MANUSCRIPT Table 1: Sample Description

RI P

T

This table reports the distributions of UK buyouts by year (panel A) and by industry (panel B) over the period 1997 to 2007. The full sample includes all buyouts, and is split into MBOs (LBOs with at least one member of the pre-transaction management team participating in the buyout and remaining in the post-buyout firm) and IBOs (LBO transactions without post-buyout involvement of the pre-transaction management team). The industries are classified based on the Fama-French 10 industry classification. The financial services industry and the utilities’ sector are excluded. We further divide Fama-French’s “Others” category into the business service industry and construction industry, such that we end up with nine industry categories. Sources: CMBOR, SDC, Venture Expert, Zephyr and Capital IQ.

Full sample

MBOs

%

Number

1997

4

2.45

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

19 36 22 11 13 17 8 10 13 10

11.66 22.09 13.50 6.75 8.00 10.43 4.91 6.13 7.98 6.13

Total

163

CE

PT

ED

Number

100.00

IBOs

%

Number

%

MA NU

Year

SC

Panel A: Distribution of buyouts over time

3

2.78

1

1.82

15 29 14 9 8 10 5 5 6 4

13.89 26.85 12.96 8.33 7.41 9.26 4.63 4.63 5.56 3.78

4 7 8 2 5 7 3 5 7 6

7.27 12.73 14.55 3.64 9.09 12.73 5.45 9.09 12.73 10.91

108

100.00

55

100.00

Panel B: Distribution of buyouts across industries

AC

Industry

Consumer Non-durables Consumer Durables Manufacturing High-Tech Wholesale & Retail Healthcare & Drugs Business Services Construction Total

Full sample

MBOs

IBOs

Number

%

Number

%

Number

%

17 6 27 22 42 4 36 9 163

10.43 3.68 16.56 13.50 25.77 2.45 22.09 5.52 100

12 4 22 14 25 3 20 8 108

11.11 3.70 20.37 12.96 23.15 2.78 18.52 7.41 100

5 2 5 8 17 1 16 1 55

9.09 3.64 9.09 14.55 30.91 1.82 29.09 1.82 100

36

ACCEPTED MANUSCRIPT

Table 2: Descriptive Statistics for Buyout Companies

151.26 29.83 76.55 1.43 3.27 62 3.14 22.25 4.69 0.32 0.12 1.00 6.50 50.00 3.00 1.00

1.84 1.13 0.40 -1.95 -117.77 0.00 -92.05 0.00 0.00

NU S

345.45 129.24 287.24 1.86 1.37 273.31 9.57 23.19 7.99 5.96 8.52 0.51 6.85 47.82 4.98 0.87

291.47 246.70 284.128 2.04 15.81 296.52 72.85 17.78 13.15

MA

62.82 31.48 46.00 1.30 4.18 32.08 3.32 16.64 4.98

Min

D

171.34 121.65 147.85 1.92 2.39 125.81 17.23 19.32 10.23

S.D.

TE

Medium

IBOs (55) 434.13 281.89 435.2 1.82 12.45 502.69 33.37 19.89 10.02 10.95 19.60 0.50 1.98 12.18 5.41 0.34

CE P

Panel B: Information by types of buyouts Total assets t-1 (GBPmil) Total market value t-1(GBPmil) Deal equityvalue(GBPmil) MTB t-1 ROA-1(%) External financing t (GBPmil) Sales growth t-1(%) Debt to equity t-1 (%) Cash to assets t-1(%) Management Own t-2 (%) Non-Executive Own t-2(%) Largest Owner Instit t-2 Board Size t-2 Independent Directors t-2(%) Analysts t-2 LSE Listing t-2

Mean

AC

Accounting information Panel A: Information on All firms (163) Total assetst-1 (GBPmil) Total market value t-1(GBPmil) Deal equity value t (GBPmil) MTBt-1 ROA t-1(%) External financing t (GBPmil) Sales growth t-1(%) Debt to equity t-1 (%) Cash to assets t-1(%)

CR I

PT

The table summarizes statistics on buyout firm-specific characteristics. t is the buyout year and t-1 is the manipulation year (one year before the buyout) and t-2 is one year prior to the manipulation year. Variables are defined in Appendix. ***, ** and * stand for statistical significance at the 1%, 5% and 10% level, respectively.

10.08 1.2 7.45 -1.88 -27.51 0 -26.92 0 0 0.00 0.00 0.00 4.00 25.00 0.00 0.00

Max

Mean

Medium

82.67 117.85 78.15 1.95 3.51 65.54 21.83 17.34 11.37 17.29 3.57 0.43 6.11 43.68 2.78 0.84

49.71 31.75 35.95 1.23 4.54 29.70 3.47 13.57 5.94 7.97 0.21 0.00 6.00 46.43 1.00 1.00

S.D.

Min

Max

Diff

1.84 1.13 0.4 -1.95 -30.16 0.00 -51.28 0.00 0.00 0.00 0.00 0.00 3.00 0.00 0.00 0.00

466.9 1600 699 9.61 32.87 485.00 466.99 74.04 76.97 85.83 55.54 1.00 10.00 80.00 14.00 1.00

262.79*** 11.39 209.09*** -0.09 -3.33 207.78*** 14.63* 5.84** -3.37** -11.33*** 4.96** -8.32 0.74** 2.25** 2.20*** 0.03

1600.00 1800.00 2121.50 9.61 55.08 2850.00 466.99 84.23 76.97 1600 1800 2121.50 8.66 19.92 2850 154.23 84.23 44.08 44.82 25.43 1.00 12.00 75.00 19.00 1.00

MBOs (108) 99.97 228.41 109.98 2.15 10.75 99.75 83.84 16.35 14.40 20.71 8.81 0.50 1.51 14.13 2.98 0.37

37

ACCEPTED MANUSCRIPT Table 3: Earnings Management The table reports discretionary accruals, abnormal operating cash flows and abnormal production costs one year before the buyouts. The raw discretionary accruals are obtained the following expectations model: TAACi,t (ΔSalesi,t - ΔReceivablesi,t ) PPEi,t 1 (1) =β [ ]+β [ ]+β [ ] + β ROA +ε 0

Assetsi,t -1

1

2

Assetsi,t -1

Assetsi,t -1

3

i,t

i,t

T

Assetsi,t -1

MA NU

SC

RI P

where, for fiscal year t and firm i, TAAC is the total accruals defined as TAACi,t = EBXIi,t-OCFi,t, (Earnings before Extraordinary Items (EBXI) minus Cash Flow from Operating activities (OCF)). ∆Salei,t stands for the change in Sales, ∆Receivablesi,t is the change in Receivables, and PPEi,t is the gross Property, Plant and Equipment. Assetsi,t-1 represents the book value of Total Assets. Performance is measured by ROAi,t. All variables (except ROA) are scaled by lagged total assets to mitigate heteroskedasticity in residuals. The industry mean-adjusted abnormal accruals (Madj_ABN_TAACt-1) are calculated by subtracting the mean abnormal accruals of the control observations in the same year and within the same two-digit SIC code from the raw abnormal accruals (RAW_ABN_TAACt-1). Industry-size mean adjusted abnormal accruals (MadjSize_ABN_TAACt-1) are calculated by subtracting the mean abnormal accruals of the control observations falling in the same industry-size group from the raw abnormal accruals. The matching-adjusted abnormal accruals (ABN_TAACt-1) consist of the difference in abnormal accruals between the sample firms and the control firms. We match each target buyout with a non-buyout control company with the closest ROAi,t and in the same two-digit SIC code and the same year. The raw abnormal operating cash flows (RAW_ABN_OCFt-1) are measured by the difference between actual total operating cash flows and the estimated cash flows from an expectation model of which the results are presented in Panel B: OCFi,t Salesi,t ΔSalesi,t 1 = β0 [ ] + β1 [ ] + β2 [ ] + β3 ROAi,t + εi,t Assetsi,t -1 Assetsi,t -1 Assetsi,t -1 Assetsi,t -1

CE

PT

ED

(4) The industry-mean adjusted abnormal cash flows (Madj_ABN_OCFt-1) are calculated by subtracting the mean abnormal operating cash flows of the control observations (from the same year and within the same two-digit SIC code) from the raw abnormal cash flows. Industry-size mean adjusted abnormal cash flows (MadjSize_ABN_OCFt-1) are obtained by subtracting the mean abnormal cash flows of the control observations falling in the same industry-size group from the raw abnormal cash flows. Matching-adjusted abnormal operating cash flows (ABN_OCFt-1) consist of the difference in abnormal operating cash flows between the sample buyouts and control firms. We match each target buyout with a non-buyout control company with the closest ROAi,t and in the same two-digit SIC code and year. The raw abnormal production costs (RAW_ABN_PRODt-1) are measured by the difference between actual total production costs and the estimated production costs from an expectation model of which the results are presented in panel C: PRODi,t Salesi,t ΔSalesi,t ΔSalesi,t -1 1 = β0 [ ] + β1 [ ] + β2 [ ] + β3 [ ] + β4 ROAi,t +εi,t Assetsi,t -1 Assetsi,t -1 Assetsi,t -1 Assetsi,t -1 Assetsi,t -1

AC

(5) The Industry-mean adjusted abnormal production costs (Madj_ABN_PRODt-1) are calculated by subtracting the mean abnormal production costs of the control firms (within the same two-digit SIC code and of the same year) from the raw abnormal production costs. The industry-size mean adjusted abnormal production costs (MadjSize_ABN_PRODt-1) are calculated by subtracting the mean abnormal production costs of the control firms (falling in the same industry-size group as the target firms) from the raw abnormal production costs of the target buyouts. The matching-adjusted abnormal production costs (ABN_PRODt-1) consist of the difference in abnormal production costs between the sample firms and the control firms. We match each target buyout with a non-buyout control company with the closest ROAi,t and in the same two-digit SIC code and year. ***, ** and * stand for statistical significance at the 1%, 5% and 10% level, respectively.

38

ACCEPTED MANUSCRIPT

Panel A. Abnormal accruals Abnormal accruals

Total

MBO

IBO

RAW_ABN_TAACt-1 Madj_ABN_TAACt-1

-0.03 -0.12***

-0.03 -0.14***

-0.02 -0.07***

-0.01 -0.07***

MadjSize_ABN_TAACt-1

-0.12***

-0.15***

-0.07***

-0.07***

ABN_TAACt-1

-0.06***

-0.08***

-0.01***

-0.07***

Nr. of observations

163

108

55

T

***

SC

RI P

***

Diff

***

Abnormal operating CF

Total

MBO

IBO

Diff

***

**

**

0.01 0.00 0.01 0.02

0.03 0.02*** 0.02*** 0.02*** 163

0.03 0.02*** 0.02*** 0.03*** 108

0.02 0.02* 0.01 0.01 55

ED

RAW_ABN_OCFt-1 Madj_ABN_OCFt-1 MadjSize_ABN_OCFt-1 ABN_OCFt-1 Nr. of observations

MA NU

Panel B. Abnormal operating cash flows

PT

Panel C. Abnormal production costs Total

MBO

IBO

Diff

RAW_ABN_PRODt-1 Madj_ABN_PRODt-1 MadjSize_ABN_PRODt-1 ABN_PRODt-1 Nr. of observations

-0.06** -0.03* -0.03 -0.06 159

-0.07** -0.04** -0.04* -0.06* 104

-0.04 0.01 0.00 -0.02 55

-0.03 -0.02 -0.03 -0.04

AC

CE

Abnormal production costs

39

ACCEPTED MANUSCRIPT Table 4: Correlation Matrix for Earnings Management Proxies

RI P

T

This table shows the Pearson correlations between the accrual and real earnings management proxies. MadjSize_ABN_TAACt-1 is industry-size mean adjusted abnormal accruals (obtained by subtracting the mean abnormal accruals of the control firms of similar size in the same year and within the same two-digit SIC code) from the raw abnormal accruals. MadjSize_ABN_OCFt-1 is the industry-size mean adjusted abnormal operating cash flows. MadjSize_ABN_PRODt-1 is the industry-size mean adjusted abnormal production costs. Panel A shows the matrix based on all buyouts. Panel B (C) shows the matrix for MBOs (IBOs). ***, ** and * stand for statistical significance at the 1%, 5% and 10% level, respectively.

Panel A. Correlation matrix for all buyouts MadjSize_ABN_OCFt-1

MadjSize_ABN_PRODt-1

1 -0.18*

1

SC

MadjSize_ABN_TAACt-1 1 -0.49*** -0.05

MA NU

All LBOs (163) MadjSize_ABN_TAACt-1 MadjSize_ABN_OCFt-1 MadjSize_ABN_PRODt-1

Panel B. Correlation matrix for all MBOs MBOs (108)

MadjSize_ABN_TAACt-1

MadjSize_ABN_OCFt-1

MadjSize_ABN_PRODt-1

1 -0.53*** -0.06

1 -0.18*

1

PT

ED

MadjSize_ABN_TAACt-1 MadjSize_ABN_OCFt-1 MadjSize_ABN_PRODt-1

Panel C. Correlation matrix for all IBOs

MadjSize_ABN_TAACt-1

CE

IBOs (55)

1 -0.42** -0.07

MadjSize_ABN_OCFt-1

MadjSize_ABN_PRODt-1

1 -0.18

1

AC

MadjSize_ABN_TAACt-1 MadjSize_ABN_OCFt-1 MadjSize_ABN_PRODt-1

40

ACCEPTED MANUSCRIPT Table 5: Asset Revaluation

MA NU

SC

RI P

T

The raw abnormal asset revaluation (RAW_ABN_REVALUEt-1) in the manipulation year is measured as the change in asset revaluation reserves scaled by current total assets. We then subtract the industry average of the revaluation amount from the raw asset revaluation in order to obtain the industry mean-adjusted abnormal revaluation (Madj_ABN_REVALUEt-1). Industry-size mean adjusted abnormal asset revaluation (MadjSize_ABN_REVALUEt-1) is calculated by subtracting the mean asset revaluation of the control firms (falling in the same industry-size group) from the raw asset revaluation. ROA-matched asset revaluation (ABN_REVALUEt-1) is measured as the difference in asset revaluation between sample and control firms. The control firms are non-buyout companies with the same two-digit SIC code and the ROAi,t (considered in the same year as the sample firm) that is closest to the buyout target. In Panel B, “No change” signifies that the asset revaluation reserves remain the same in both the manipulation year and the year before. “Upward revaluation” indicates that there is an increase in revaluation activities from one year before the manipulation year to the next year, while “Downward revaluation” captures the opposite case. “Transfer” reflects that the change in revaluation reserves is arising from transferring in or transferring out between revaluation reserves account and other reserves accounts. ***, ** and * stand for statistical significance at the 1%, 5% and 10% level, respectively.

Panel A. Abnormal revaluation Abnormal revaluation

Total

MBO

IBO

Diff

0.001 0.002 0.004* 0.005**

-0.001 -0.000 0.001 0.004

0.006** 0.008** 0.010* 0.010*

-0.007** -0.008* -0.009 -0.006

156

103

53

PT

ED

RAW_ABN_REVALUEt-1 Madj_ABN_REVALUEt-1 MadjSize_ABN_REVALUEt-1 ABN_REVALUEt-1 Nr. of observations

Panel B. Detailed information on the asset revaluation reserves Total

MBO

IBO

No change (%) Upward revaluation (%) Downward revaluation (%) Transfer (%) Nr. of observations

69.28 4.58 5.88 20.26

70.30 3.96 3.96 21.78

67.31 5.77 9.62 17.31

153

101

52

AC

CE

Abnormal revaluation

41

ACCEPTED MANUSCRIPT Table 6: Earnings Manipulation Before and After Regulatory Change

T

This table assesses the impact of the enactment of the revised UK Corporate Governance Code of 2003 on accounting manipulation. We divide the sample period into two subperiods: 1997-2003 and 2004-2007. Abnormal accruals, abnormal operation cash flows (OCF), abnormal production costs (PROD), abnormal assets revaluations are calculated similarly as in table 3 and 5, with variables lagged by two years. ***, ** and * stand for statistical significance at the 1%, 5% and 10% level.

RAW_ABN_TAACt-1 Madj_ABN_TAACt-1 MadjSize_ABN_TAACt-1 ABN_TAACt-1 Nr. of observations

-0.03** -0.12*** -0.12*** -0.06*** 163

19972003

SC

Total

-0.03*** -0.14*** -0.15*** -0.08*** 122

MA NU

Abnormal accruals

RI P

Panel A. Abnormal accruals

20042007

-0.03*** -0.04** -0.02 -0.00 41

Diff 0.00 -0.11*** -0.14*** -0.07***

Panel B. Abnormal operating cash flows Total

ED

Abnormal operating CF

CE

PT

RAW_ABN_OCFt-1 Madj_ABN_OCFt-1 MadjSize_ABN_OCFt-1 ABN_OCFt-1 Nr. of observations

0.03*** 0.02*** 0.02*** 0.02*** 163

19972003

20042007

0.02** 0.01** 0.02** 0.02** 122

0.04** 0.03** 0.03** 0.01 41

Diff -0.02 -0.02 -0.02 0.01

Panel C. Abnormal production costs Total

RAW_ABN_PRODt-1 Madj_ABN_PRODt-1 MadjSize_ABN_PRODt-1 ABN_PRODt-1 Nr. of observations

-0.06** -0.03* -0.03 -0.06 159

AC

Abnormal production costs

19972003 -0.02

2004-0.14 2007**

-0.00 -0.00 -0.05 118

-0.10** -0.09** -0.05* 41

Diff -0.11** -0.09** -0.08** 0.00

Panel D. Abnormal revaluation Abnormal revaluation

Total

RAW_ABN_REVALUEt-1 Madj_ABN_REVALUEt-1 MadjSize_ABN_REVALUEt-1 ABN_REVALUEt-1 Nr. of observations

0.001 0.002 0.004* 0.005** 156

1997-0.001 2003***

20040.007 2007**

-0.001 -0.001 0.003

0.011** 0.019** 0.014*

116

40

Diff -0.008* -0.012** -0.020** -0.013*

42

ACCEPTED MANUSCRIPT

Table 7: Analysis of the Incentives Underlying Earnings Manipulation (2SLS approach)

MA NU

SC

RI P

T

The first stage dependent variable is Dum_MBO, which indicates whether the buyout is an MBO (Dum_MBO=1) or an IBO (Dum_MBO=0). The IVs are Management Own (equity share owned by managers in the pre-buyout target), Non-Executive Own (equity share held by non-executive directors) and Size (logarithm of pre-buyout total assets). The second stage dependent variable is Industry-size mean adjusted abnormal accruals (MadjSize_ABN_TAACt-1) /operating cash flow (MadjSize_ABN_OCFt-1) /production costs (MadjSize_ABN_PRODt-1). Pred_Dum_MBO, is the predicted type of buyouts (from stage 1). Dum_ External Financing equals one if the target raises external funds at the transaction above the median, and zero otherwise. Dum_SA equals one if the company is financially constrained and zero otherwise. SA index is calculated as (0.737*Size + 0.043*Size2 - 0.040*Age), where size is the natural log of total assets (in millions). We winsorize book value at 4.5 billion dollar and age at 37 years (Hadlock and Pierce, 2010). Companies belong to the top 30 percent of SA Index are considered constrained (based a group of companies excluding utilities and financial firms from 1990 to 2007). For accrual management, the internal manipulation capacity is captured by the net operating assets (NOA) position (sum of equity minus cash and marketable securities, plus total debt, standardized by total sales). The level of the stock of inventories and receivables (INVREC) captures the flexibility of managers to manipulate real activities. ***, ** and * stand for statistical significance at the 1%, 5% and 10% level, respectively.

Panel A: First stage: The buyout type Model 1: First stage

Management Own i, t-2

0.516*** (0.182) -0.491*** (0.274) -0.105*** (0.028) Yes Yes 0.208 0.005 0.202 18.206

ED

Dep.Var. Dum_MBO Non-Executive Own i, t-2

PT

Size i, t-2

AC

CE

Year Fixed effects Industry Fixed effects Adjusted R2 Tests of endogeneity (p value) Test of overidentifying restrictions (p value) Robust F

43

ACCEPTED MANUSCRIPT

NOA i, t-2

0.018 (0.062)

0.012 (0.035) 0.007 (0.065)

TE

0.031 (0.101) Yes Yes 150

CE P

Year Fixed effects Industry Fixed effects Observations

0.013 (0.091) Yes Yes 158

MadjSize_ ABN_PRODt-1 (3a) 0.051 (0.115) -0.096* (0.058)

0.020 (0.023)

-0.051 (0.043) -0,059 (0.063) Yes Yes 150

MadjSize_ ABN_PRODt-1 (3b) 0.011 (0.139)

-0.093 (0.090)

0.131 (0.137) -0.016 (0.171) Yes Yes 155

0.132 (0.135) -0.112 (0.188) Yes Yes 150

AC

Constants

-0.063 (0.043) -0.046 (0.061) Yes Yes 158

D

INVREC i, t-2

MadjSize_ ABN_OCFt-1 (2b) 0.080* (0.043)

CR I

MadjSize_ ABN_OCFt-1 (2a) 0.082** (0.035) -0.006 (0.020)

NU S

Dum_External Financingi, t Dum_SAi, t-2

MadjSize_ ABN_TAACt-1 (1b) -0.181** (0.063)

MA

Pred_Dum_MBOi, t-2

MadjSize_ ABN_TAACt-1 (1a) -0.177** (0.055) 0.011 (0.029)

PT

Panel B: Second stage: Incentives underlying earnings manipulation

44

ACCEPTED MANUSCRIPT

Table 8: Robustness Tests on the Incentives Underlying Earnings Manipulation

NU S

CR I

PT

This table provides the robustness tests for the second stage regressions of Panel B, Table 8. The dependent variable in Model (1) is ROA matched abnormal accruals (ABN_TAACt-1). The dependent variable in Model (2) is raw abnormal operating cash flows (RAW_ABN_OCFt-1). Model (3) uses the probit model in the first stage. Model (4) conducts the second stage by means of a GMM approach. Pred_Dum_MBO, is the predicted type of buyouts. Models (5) and (6) further investigate the change in earnings management behavior after the enactment of the revised UK Corporate Governance Code (Code) in 2003. YearGovCode equals one if the buyout took place after 2003, zero otherwise. For model (1)- (6), the first stage IVs are Management Own, Non-Executive Own and Size and their interaction terms with YearCode. The other variables are the same as in Table 7. ***, ** and * stand for statistical significance at the 1%, 5% and 10% level.

Second stages: Incentives underlying accounting manipulation discretion

-0.147** (0.053)

0.069** (0.034)

-0.039 (0.035) -0.093 (0.076)

-0.004 (0.020)

MadjSize_ ABN_TAACt-1 (Probit) (3) -0.121** (0.048)

MA

RAW_ABN_OCFt-1 (2)

Constants Year Fixed effects Industry Fixed effects Observations

0.214 (0.100) Yes Yes 158

-0.060 (0.040) -0.044 (0.057) Yes Yes 158

AC

INVREC i, t-2

CE P

TE

Year GovCode Pred_Dum_MBOi, t-2 *Year GovCode Dum_External Financingi, t NOA i, t-2

MadjSize_ ABN_TAACt-1 (GMM) (4) -0.180*** (0.055)

D

Pred_Dum_MBOi, t-2

ABN_TAACt-1 (1)

-0.001 (0.029) 0.040 (0.061)

-0.039 (0.086) Yes Yes 158

0.008 (0.029) 0.005 (0.060)

0.315 (0.090) Yes Yes 158

MadjSize_ ABN_TAACt-1 (Gov Code) (5) -0.235*** (0.074) -0.074 (0.084) 0.244** (0.113) 0.001 (0.031) 0.009 (0.061)

0.065 (0.105) Yes Yes 158

MadjSize_ ABN_OCFt-1 (Gov Code) (6) 0.114** (0.045) 0.096* (0.059) -0.114 (0.072) -0.003 (0.020)

-0.054 (0.044) -0.077 (0.070) Yes Yes 158

45

ACCEPTED MANUSCRIPT Do Managers Manipulate Earnings Prior to Management Buyouts?

RI P

SC

MA NU

ED



PT

 

CE



Prior to an MBO, negative earnings management (understating earnings prior to the deal) occurs. MBOs are subject to more negative manipulation than institutional buyouts (IBOs). Earnings manipulation occurs both by accrual and real earnings management. In non-buyout firms, positive earnings management happens because it affects managers’ bonuses. Competing incentives affecting degree and size of earnings manipulation are studied via an IV approach.

AC



T

Highlights

46