Executive compensation and goodwill recognition under IFRS: Evidence from European mergers

Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Contents lists available at SciVerse ScienceDirect

Journal of International Accounting, Auditing and Taxation

Executive compensation and goodwill recognition under IFRS: Evidence from European mergers Dominic Detzen ∗ , Henning Zülch Both at HHL Leipzig Graduate School of Management, Jahnallee 59, 04109 Leipzig, Germany

a b s t r a c t Based on principal agent theory we posit that managers account for a business combination opportunistically by recognizing goodwill in excess of its economic determinants. We examine the relationship between CEOs’ short-term cash bonuses and the amount of goodwill recognized in IFRS acquisitions. We find that with increasing cash bonus intensity managers recognize more goodwill. More detailed analysis indicates that this relationship is not a linear one. Instead, there seems to be a corridor in which CEOs are susceptible to the incentive given by bonus payments. In particular, the relationship seems to be fulfilled only for CEOs whose cash bonus is between 150% and 200% of their base salary prior to the acquisition. Our findings have an implication for companies that bonus caps should be introduced to limit CEOs’ bonuses to a given percentage of their base salary. By doing so, they may re-align shareholders’ and managers’ interests and avoid an increased impairment risk in the future. © 2012 Elsevier Inc. All rights reserved.

1. Introduction Principal agent theory (Fama, 1980; Jensen & Meckling, 1976) suggests that the separation of ownership and control creates a fundamental problem in organizations since interests of shareholders, who want to maximize the value of their equity, diverge from those of managers, who want to be paid more while working less. To re-align these interests, organizations aim to employ optimal pay-for-performance mechanisms. A company’s accounting performance as disclosed in audited financial reports plays an important role when determining the level of executive compensation. However, the flexibility – or discretion – in accounting standards introduces the problem of earnings management. Johnson and Revsine (1988) indicate two perspectives in this regard. One emphasizes that accounting reports can be used by managers to communicate their private information on future cash flows. Consequently, managers use discretionary choices to increase the information content of accounting. The second perspective is based on the view that managers are opportunistic and use discretionary accounting choices for their personal wealth. Thus, accounting decisions may be heavily influenced by executive compensation arrangements. This latter explanation has led to a large body of research on managers’ accounting choices as a result of incentives given by compensation contracts.1 We base our research on this perspective, which is also termed the bonus plan hypothesis (Watts & Zimmerman, 1986). More specifically, our study examines whether managers’ compensation provides an incentive for them to use their discretion potential regarding the recognition of goodwill. Managers’ compensation typically consists of different components: a base salary, a short-term (and sometimes a midterm) bonus as well as a long-term incentive. While the latter component is typically paid in share options, the others are paid in cash. Accounting research indicates that compensation mixes that are more weighted toward accounting-based pay are associated with higher future performance (Larcker, Richardson, & Tuna 2007). Managers’ short-term cash bonus

∗ Corresponding author. E-mail address: [email protected] (D. Detzen). 1 See Pavlik et al. (1993) for an early review of executive compensation issues. Bodolica and Spraggon (2009) provide a more recent review of research on the link between executive compensation and mergers and acquisitions, while Fields, Lys and Vincent (2001) review research on accounting choice. 1061-9518/$ – see front matter © 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.intaccaudtax.2012.07.002

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

107

constitutes accounting-based pay and may provide a larger incentive when making accounting decisions. We thus examine the relationship between a manager’s short-term cash compensation and the amount of goodwill recognized in an acquisition. Based on Gordon’s (1964) proposition, we expect that managers choose accounting procedures that maximize their own welfare. Managers whose compensation package depends heavily on cash bonuses would thus recognize more goodwill in order to avoid amortization charges and increase their company’s earnings as well as their bonuses. This reasoning suggests a linear relationship between bonus payments and the recognition of goodwill. However, Healy (1985) points out that managers do not always choose income increasing accounting policies to increase their variable compensation. Instead, he argues that below and above certain thresholds, changes in earnings do not matter for the determination of cash bonuses, which is why managers are not expected to choose accounting policies opportunistically outside a certain corridor. Adapting Healy’s (1985) framework to the recognition of goodwill, we expect that bonus intensity is only to a certain extent an incentive to recognize more goodwill. Below and/or above certain thresholds, cash bonuses are not expected to influence goodwill recognition since managers do not expect additional gains from higher goodwill. Mergers and acquisitions are worth studying from an accounting perspective since they are unique corporate events that often have a considerable impact on an acquirer’s financial position and on the public’s perception of the merging companies. We contribute to the literature by providing insights into the impact of compensation arrangements and accounting choice on the accounting for business combinations. While Shalev, Zhang, and Zhang (2011) examine executive compensation and goodwill recognition under US GAAP, ours is the first study to analyze these issues in an environment that is based on International Financial Reporting Standards (IFRS). IFRS 3 Business Combinations provides in-depth guidance on how to account for mergers and acquisitions. However, the standard includes large areas of discretion, which foremost relate to the identification and valuation of intangible assets. For these assets, managers apply valuation techniques that are based mainly on level 3 inputs. Consequently, they obtain an advantage over other parties such as investors due to informational asymmetry (Landsman, 2007). Discretion regarding the valuation of assets and informational asymmetry culminates in the amount of goodwill the acquirer recognizes since goodwill remains as a residual after deducting the fair value of acquired net assets from the price paid for the target. Thus, discretionary valuation of intangible assets translates into discretion regarding the recognition of goodwill, which is an item that is all the more interesting to managers due to its exceptional accounting treatment. Since goodwill is not amortized but merely tested for impairment (at least) yearly, managers avoid otherwise necessary depreciation and amortization charges by recognizing more goodwill. Consequently, more goodwill would, ceteris paribus, result in higher net income. Managers may wish to make use of the flexibility regarding goodwill depending on their accounting policies. In addition, they may want to avoid amortization charges for their personal benefit given the importance of accounting earnings for determining executive compensation. Our paper examines whether or not CEOs tend to increase the amount of goodwill recognized in acquisitions if their pre-acquisition cash compensation depends largely on bonuses, possibly to further increase their remuneration. We test our hypotheses by examining 123 mergers and acquisitions that were completed by Stoxx Europe 600 companies between 2005 and 2008. We find that the amount of goodwill recognized is heavily influenced by economic factors concerning the target company and expected synergies to be created by the merger. We also find that managers recognize more goodwill the more their compensation package depends on short-term bonuses. Examining bonus intensity more closely, we find that the relationship between cash bonus and goodwill applies for managers whose bonus element is between 150% and 200% of their base salaries. For CEOs whose bonuses are above 200% of their base salaries, we do not find a significant relationship between cash bonus and amount of goodwill recognized. In these cases, managers may choose to refrain from influencing the recognition of goodwill because they may not expect additional benefits from higher goodwill. Further analysis shows that for CEOs whose bonuses were above 150% of their base salaries prior to the acquisition, bonuses increase in the year of the acquisition but decrease significantly in subsequent years. The initial increase in bonus may be interpreted as a reward to the CEO for conducting the acquisition. Based on our results, we recommend that companies introduce bonus caps, which are common features of compensation arrangements and which represent upper limits to the amount of bonus a manager can earn. If a bonus plan does not include such a cap, managers seem to choose incomeincreasing accounting policies and recognize more goodwill, increasing future impairment risk. Our results are in line with Arya, Glover, and Mittendorf (2007) who find that bonus caps help align shareholders’ and managers’ interests by making CEOs more risk-averse and more focused on longer-term objectives. Other sub-sample analyses provide somewhat weaker evidence and suggest that CEOs’ cash bonus is a determinant of goodwill recognition especially for non-financial companies and acquirers residing outside the United Kingdom. These results seem to be in line with prior research that finds an increased earnings management culture for countries in continental Europe (e.g. Lang, Smith Raedy, & Wilson, 2006; Leuz, Nanda, & Wysocki, 2003). Our research adds to this literature by suggesting that compensation concerns may be an additional reason for country differences regarding earnings management. 2. Theoretical framework 2.1. Agency theory, accounting choice and goodwill recognition In a principal agent setting, accounting takes an essential role due to its stewardship function, which helps fulfill principals’ demand for information to control agents’ performance (Gjesdal, 1981). Absent perfect information on managers’ behavior,

108

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

reported accounting earnings serve as a proxy for assessing managers’ performance in a period. Thus, managers are rewarded depending on their companies’ results. This pay-for-performance mechanism helps re-align principals’ and agents’ interests by sharing the risk of varying returns on the company’s investments and by making managers interested in the outcome of their work. Managers are given an incentive to work more by the prospect of earning more money. As a result, there is a close relationship between managers’ compensation and – given the stewardship function of accounting – companies’ performance as disclosed in financial reports. Flexibility in accounting standards may hinder capital providers from truthfully evaluating managers’ efforts. While it can be argued that managers use their discretion to choose the most beneficial reporting alternative and to communicate private information (Demski, Patell, & Wolfson, 1984), there is also a downside to flexible accounting standards. By behaving opportunistically, managers can use the discretion given to them to enhance earnings figures, emphasize various line items or manipulate accounting numbers. The overall goal of managing financial information may be to ensure that users perceive information favorably. Another motivation is provided by agency theory and accounting choice, which argues that opportunistic managers employ their discretion to portray themselves as good executives and to pursue higher bonuses. The relation between managerial discretion and executive compensation has led to a large body of literature on accounting choice, which aims to explain why discretion is allowed in bonus contracts (e.g. Dye & Verrecchia, 1995; Evans & Sridhar, 1996). It is argued that compensation contracts can neither be specified so precisely ex ante nor renegotiated without costs ex post that accounting choices do not affect bonus payments (Holthausen & Leftwich, 1983). Thus, high contracting costs prevent companies from excluding flexibility in compensation contracts and accounting choices remain. It has also been suggested that managerial discretion contributes to the alignment of managers’ and shareholders’ interests, regardless of how it is used. Fields et al. (2001) explain this reasoning by arguing that higher accounting earnings increase managers’ compensation levels but at the same time increase share prices. As a result, both shareholders and managers may benefit and compensation contracts that allow discretion may be considered efficient. However, it needs to be pointed out that such reasoning implies short-term orientation on behalf of both shareholders and managers, and it remains unclear whether discretion ultimately benefits both parties. Overall, we may conclude that financial reporting and accounting choice remain vulnerable to executives’ pursuit of higher bonuses (e.g. Balsam, 1998). We use this framework in the context of the accounting for business combinations and examine whether managers behave opportunistically when applying IFRS 3. The theoretical background to our study is given by IFRS 3 (2004). While in 2008 the IASB issued a revised IFRS 3, our article examines business combinations that were completed under IFRS 3 (2004) to ensure an adequate number of observations. By limiting our research to the years 2005–2008, we avoid sample contamination by the financial crisis. One of the main features of IFRS 3 (2004) is the purchase method, which is the core principle of the standard. According to this method, one of the combining companies has to be identified as the acquirer, which is the entity that obtains control of another entity or entities. As a second step, the acquirer needs to measure the cost of the combination, which is determined as the sum of the fair values of assets given, liabilities assumed and equity instruments issued in exchange for the control of the target as well as any costs that are directly attributable to the business combination. Finally, the acquirer has to allocate the cost of the business combination by identifying and measuring the target’s assets and liabilities at their acquisition date fair values. Any difference between the cost of the business combination and the acquirer’s interest in the net fair value of identifiable assets and liabilities is recognized as goodwill. While this description may indicate comprehensive and detailed rules for the accounting for business combinations, acquirers face a complex process that results in subjective values, particularly for intangible assets. For example, Watts (2003) argues that SFAS No. 141 – which is, in effect, similar to IFRS 3 – requires managers to recognize unverifiable values of intangible assets. He criticizes that these unverifiable estimates of future cash flows “depend on assumptions about the future that experts cannot agree upon.” Therefore, he concludes, both value of a firm and corresponding goodwill are not objective measures but highly subjective and open to manipulation. This situation seems to be increasingly so, the less reliant and the less verifiable the employed valuation techniques are. Landsman (2007) addresses these issues and points out that the use of level 3 fair value estimates required by both US and international accounting rules results in informational asymmetry. Accordingly, managers may manipulate model inputs for their own benefit. Landsman’s (2007) findings are especially relevant in the context of measuring acquired intangible assets since these typically require the application of level 3 valuation techniques. We argue that acquisitions accounted for under IFRS 3 produce unverifiable accounting figures, which adds high monitoring costs to the theoretical framework of accounting choice (Evans & Sridhar, 1996; Holthausen & Leftwich, 1983). Acquirers’ subjectivity as a result of discretion applied when measuring assets acquired culminates in goodwill recognized in the acquisition. The residual often makes up a considerable share of the purchase price and remains an accounting figure that is not intuitive for investors, especially when considering that goodwill is a residual instead of a measured figure. Several components may be represented by the line item “goodwill”:

(1) The fair value of the “going concern” element of the acquiree. The going concern element represents the ability of the acquiree to earn a higher rate of return on an assembled collection of net assets than would be expected from those net assets separately.

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

109

(2) The fair value of the expected synergies and other benefits from combining the acquiree’s net assets with those of the acquirer. Those synergies and other benefits are unique to each business combination, and different combinations produce different synergies and, hence, different values. (3) Overpayments by the acquirer. (4) Errors in measuring and recognizing the fair value of either the cost of the business combination or the acquiree’s identifiable assets, liabilities or contingent liabilities, or a requirement in an accounting standard to measure those identifiable items at an amount that is not fair value. (IFRS 3.BC130 (2004)) This summary of goodwill components indicates that it is difficult for investors to evaluate and interpret the line item. On the one hand, some components, such as (1) and (2), may be welcomed by investors, whereas others, such as (3) and (4), will most likely be disapproved of by shareholders. Overall, goodwill remains a vague accounting figure that presents manipulation potential and a discretionary accounting choice for managers. 2.2. Development of hypotheses In a principal agent setting, opportunistic managers may be interested in recognizing goodwill since they may strive to use the residual figure to blur the stewardship function of accounting. Goodwill is not amortized on a straight-line basis but tested for impairment annually (or more often in the case of triggering events).2 With the purchase price remaining constant, recognizing more goodwill will reduce acquirer’s amortization charges and, ceteris paribus, increase a company’s earnings. If executives are inclined to manage earnings, they may be willing to recognize a large amount of goodwill to avoid amortization charges. By doing so, they may not only increase current and future earnings but also obtain personal benefits since they avoid being rewarded according to the company’s “true” accounting income. Thus, they act in the sense of Gordon’s (1964) and Watts and Zimmerman’s (1986) propositions and choose accounting policies to maximize their own welfare. While most research has focused on the subsequent accounting for and impairment of goodwill (e.g. Beatty & Weber, 2006; Hayn & Hughes, 2006; Li, Shroff, Venkataraman, & Zhang, 2011), there is a growing body of literature on the application of SFAS No. 141 and IFRS 3, respectively (e.g. Hamberg, Paananen, & Novak, 2011; Shalev et al., 2011). The general notion of these studies is that acquirers use their discretion potential opportunistically. Shalev et al. (2011) examine how compensation structures affect the accounting for business combinations in the U.S. They find that bonuses that are tied to earnings are an incentive for CEOs to recognize more goodwill. This relationship is diminished if CEOs’ bonuses are tied to cash-flow-based measures. Hamberg et al. (2011) examine a Swedish sample and find that the amount of goodwill recognized in acquisitions has increased substantially after Swedish firms adopted IFRS 3. They find that companies with large amounts of goodwill yield abnormal returns and conclude that investors view the increase in earnings due to lower or no goodwill amortization charges as an indication of higher future cash flows. Based on the framework provided by agency theory and the discretion potential inherent in the accounting for business combinations, we posit that managers use discretionary accounting choices concerning the valuation of intangible assets and recognition of goodwill opportunistically. They will choose to recognize a higher amount of goodwill the more they expect to profit personally from higher earnings. Thus, they are increasingly encouraged to recognize more goodwill if their compensation package depends heavily on bonus payments. Consequently, we define cash bonus intensity as the ratio of short-term cash bonuses to total cash compensation measured prior to an acquisition. We base our research on short-term cash bonuses because recent research suggests that these components may provide a larger incentive for managers. For example, Larcker et al. (2007) find that cash bonuses are associated with future company performance. Armstrong, Jagolinzer, and Larcker (2010) do not find evidence that equity-based holdings provide incentives to manipulate accounting reports. Besides basing our research on these recent findings, we argue that short-term cash bonuses can be influenced more easily by making decisions opportunistically. Goodwill recognition is a decision that primarily has short-term effects, namely avoiding amortization charges in the short run. We may assume that a change in amortization directly impacts short-term bonuses, whereas it influences long-term compensation only indirectly. Due to vesting periods, the latter component of compensation contracts is to a large extent dependent on the company’s future performance and future share prices. Thus, managers may be more likely to use their discretion concerning goodwill opportunistically the more their compensation is short-term oriented. Accordingly, we formulate the following hypothesis: H1. In acquisitions, managers recognize more goodwill if their compensation package is based on high cash bonus intensity. Healy (1985) was among the first to criticize the linear relationship between executive compensation and earnings management. He points out that there may only be a certain corridor in which bonuses provide an incentive for executives to manage their companies’ earnings. He provides evidence that, outside of this corridor, managers do not choose income increasing accounting policies since they do not expect further benefits from such policies. Accordingly, managers do not seem to behave opportunistically above and below certain thresholds. While being a powerful framework to explain

2 Note that acquirers may also assign indefinite useful lives to some intangible assets (such as brands). We do not expect the effect of these intangibles to be different from goodwill. Thus, in the following, we subsume indefinite-life intangibles under the term goodwill (see also footnote 3).

110

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

managers’ behavior, Healy’s (1985) findings could not be confirmed entirely by later research (e.g. Gaver, Gaver, & Austin, 1995; Holthausen, Larcker, & Sloan, 1995). We adapt Healy’s (1985) framework and argue that cash bonus intensity is only to a certain extent an incentive to recognize more goodwill. Below and/or above certain thresholds, managers do not consider the recognition of more goodwill as an adequate procedure to gain personally from taking accounting choices opportunistically. This result may either be due to a limited influence of the effect or due to the fact that once reaching a certain level of goodwill, the longer-term consequences, i.e. the risk of future goodwill impairment, dominate acquirers’ actions. Consequently, we formulate our second hypothesis as follows: H2. Cash bonus intensity influences managers’ choice of goodwill recognition only below and/or above certain thresholds. 3. Data and research design 3.1. Data We start our data sampling process by collecting all acquisitions completed by Stoxx Europe 600 companies between January 1, 2005 and December 31, 2008 from CorpfinWorldwide. We refine our search by limiting the results to those that include the terms “acquisition” in the “Deal Type” segment and “public” in the “Target–Ownership” segment. After processing all 600 companies included in the index, we come up with 315 acquisitions. With the help of the database, we collect announcement and acquisition dates as well as the two-digit SIC codes. Since the announcement dates are the basis for calculating abnormal changes in market values, we cross-check the dates via the press release sections of acquirers’ web pages. We then analyze the annual reports of the acquiring companies and, first, exclude those companies that do not prepare their financial statements according to IFRS. While the European Union has required application of IFRS since 2005, member states could apply transitional provisions that required application of IFRS for financial years that started on or after January 2007. In addition, we hand-collect accounting information on the acquisitions, including the total consideration paid, goodwill recognized as well as the fair value of the net assets acquired. After excluding transactions not accounted for according to IFRS and transactions with insufficient accounting information, our sample is reduced to 219 transactions. We collect financial data from Datastream to compute several variables. For some companies, Datastream does not include the necessary information and, as a result, we eliminate an additional 25 observations. Due to missing executive compensation data which we also hand-collect from annual reports, our sample is reduced by an additional 71 observations. Our final sample includes 123 transactions. Table 1 provides an overview of the sampling process. Table 2 shows the distribution of our sample. Panel A shows a disaggregation of the sample by years in which the acquisitions were completed. There is a slight bias toward the years 2007 and 2008 with 39 transactions completed in each year, which reflects the general economic trend of our sample. When examining the data segmented by country of acquirer as shown in Panel B, we observe that almost 45% of the transactions were completed by acquirers residing in the UK. While France, the Netherlands, Germany and Sweden are also strongly represented in the sample, there does not seem to be a strong bias toward these countries. Panel B also shows institutional clusters to which the countries can be assigned. The clusters are based on Leuz (2010) and show that cluster 2 is slightly under-represented in the sample (16% of all transactions). The remaining transactions are almost evenly distributed between clusters 1 and 3. Panel C shows a breakdown of the data by origin of the target. We observe a strong representation of the Anglo-Saxon countries with the United States (50 transactions), the United Kingdom (27), and Canada (10) being countries with more than ten targets each. Finally, Panel D shows the industries of both acquirer and target companies. Most acquirers in our sample were from the manufacturing industry followed by finance, insurance, and real estate and the services industry. We take the biases discussed above into account and consider differences in the macroeconomic environment as well as the regulatory framework by including both year and acquirers’ country cluster dummies in our analyses. 3.2. Depreciation and amortization effects on cash bonuses Our study examines whether managers’ bonus concerns present an incentive for them to recognize more goodwill. Since recognizing a high level of goodwill lowers amortization charges in subsequent years, we posit that, ceteris paribus, a company’s earnings as well as managers’ bonuses will increase. The hypothesis hinges on the assumption that depreciation Table 1 Sample selection. Transactions of Stoxx Europe 600 companies between 2005 and 2008 on CorpfinWorldwide

315

Non-IFRS transaction and/or acquisition information not available Financial information not available on Datastream Executive compensation data not available

−96 −25 −71

Total number of acquisitions used in the study

123

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

111

Table 2 Sample distribution. Panel A: Years Year

Number of acquisitions

Percentage

2005 2006 2007 2008

17 28 39 39

13.8 22.8 31.7 31.7

Total

123

100.0

Panel B: Number of acquisitions per country of acquirer and institutional cluster Cluster

Country

Cluster 1

United Kingdom

55 55

44.7 44.7

Cluster 2

Belgium Finland Netherlands

3 3 14 20

2.4 2.4 11.4 16.2

Cluster 3

France Germany Italy Norway Spain Sweden Switzerland

15 9 6 1 4 9 4 48

12.2 7.3 4.9 0.8 3.3 7.3 3.3 39.1

123

100.0

Number of acquisitions Percentage

Total Panel C: Country of target Country Australia Belgium Canada Colombia Denmark France Germany India Italy Netherlands Norway Pakistan Poland Russia South Africa Spain Sweden Switzerland Taiwan United Kingdom United States Total

Number of acquisitions

Percentage

1 2 10 2 1 3 4 1 1 6 2 1 1 2 2 2 2 2 1 27 50

0.8 1.6 8.1 1.6 0.8 2.4 3.3 0.8 0.8 4.9 1.6 0.8 0.8 1.6 1.6 1.6 1.6 1.6 0.8 22.0 40.7

123

100.0

Panel D: SIC codes Description

Two-digit SIC code

Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communications, electric, gas, and sanitary services Wholesale trade Retail trade

01–09 10–14 15–17 20–39 40–49

Number of acquirers 2 9 4 53 13

Percentage 1.6 7.3 3.3 43.1 10.6

Number of targets 0 7 3 41 10

Percentage 0.0 5.7 2.4 33.3 8.1

50–51 52–59

2 1

1.6 0.8

4 3

3.3 2.4

112

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Table 2 (Continued ) Panel D: SIC codes Description

Two-digit SIC code

Finance, insurance, and real Estate Services Public administration

60–67 70–89 91–99

Total

Number of acquirers

Percentage

Number of targets

Percentage

21 18 0

17.1 14.6 0.0

21 33 1

17.1 26.8 0.8

123

100.0

123

100.0

Note. Table 2 depicts the distribution of our sample. Panel A reports the number of acquisitions per year. Panel B shows the number of observations per country of acquirer. Institutional clusters are based on Leuz (2010). Panel C shows the number of acquisitions per country of target. Finally, Panel D reports the distribution of our sample according to the two-digit SIC codes of both acquirer and target.

and amortization charges influence managers’ bonuses. Prior to analyzing our two hypotheses, we examine whether changes in the level of depreciation and amortization have an impact on bonuses. We run the following regression based on Cheng (2004) and Shalev et al. (2011): Bonusi,t = ˛ + ˇ1 Returni,t + ˇ2 RoAi,t + ˇ3 DAi,t + ˇ4–6 Yri,t + ˇ7–8 Clusteri,t + ˇ9–15 Industryi,t + εi,t

(1)

The dependent variable Bonus is the year-to-year change in the level of CEO’s short-term cash bonus divided by CEO’s total cash compensation. Return is measured as the one-year buy-and-hold-return of the company’s stock, while RoA is computed as the year-to-year change in a company’s return on assets. The numerator in the return on asset ratio is defined as net income plus depreciation and amortization expenses in order to separate the effect of income from the one of depreciation and amortization charges. We expect both Return and RoA to be positively associated with the change in the level of bonus. The variable of interest DA is computed as the year-to-year change in depreciation and amortization charges divided by total assets. DA is expected to be negatively correlated with Bonus, implying that an increase in the level of depreciation and amortization has a negative impact on the level of bonus. We include year as well as industry dummies in our regression model. In addition, we include cluster dummies that are based on Leuz (2010) and that represent the institutional cluster of the country in which the acquirer resides. 3.3. Economic model To test our hypotheses, we first develop a model that includes only economic determinants of goodwill. By examining this model, we aim to ensure that we capture factors that determine economically the level of goodwill recognized in an acquisition. Specifically, our economic model is run as follows: GWi,t = ˛ + ˇ1 Synergyi,t + ˇ2 BTMi,t−1 + ˇ3 Indi,t + ˇ4 Stocki,t + ˇ5 Sizei,t−1 + ˇ6−8 Yri,t + ˇ9−10 Clusteri,t + εi,t

(2)

Our dependent variable GW is computed as the amount of goodwill and indefinite-life intangible assets recognized in the acquisition divided by the total purchase price. While goodwill is typically the largest intangible recognized in the transaction, indefinite-life intangibles provide a similar incentive for acquirers since they also are not amortized on a straight-line basis.3 Synergy represents one of the economic determinants of goodwill and corresponds to the fair value of the expected synergies from the combination. The computation of the variable is based on Bradley, Desai, and Kim (1988). Accordingly, total synergistic gains of a business combination are defined as the sum of the changes in stockholders’ wealth of target and acquirer. The authors posit that these changes in market values are entirely due to investors’ expectations concerning the synergies to be created by the merging companies. We calculate Synergy as the logarithm of positive changes in target’s and acquirer’s market value after the merger announcement divided by total consideration. To have increasing values of Synergy correspond to an increased level of goodwill recognized, we multiply the ratio by −1.4 The changes in market values are estimated as the sum of abnormal changes in market value over a three-day window around the merger announcement. Normal returns were estimated based on a market model using a 200-day window prior to the announcement day. By totaling abnormal changes in market values of both target and acquirer, we calculate a proxy for the synergies that investors anticipate from the combination of the two companies. BTM corresponds to the fair value of the going-concern element and describes the ability of the target to earn a higher rate of return on its assembled assets. We argue that investors are aware of the element, which is why the difference between a company’s market value and book value of equity is expected to correspond to the degree of the going-concern element. Consequently, we compute BTM as the target’s book to market ratio of equity. We compute the ratio one quarter prior to

3 We focus on the similar accounting treatment of indefinite-life intangible assets and goodwill, although we are aware that these items may signal different information. To check for robustness, we reran the regressions without indefinite-life intangibles. Results remain similar. 4 Note that our sample also contains observations for which the changes in market values divided by the purchase price is greater than one. For these cases, multiplying the logarithm by minus one yields a negative value, although investors seem to expect high synergies. We are aware that this effect limits the interpretation of our variable. Our results hold when we exclude these observations.

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

113

the merger announcement in order to obtain a variable that is not contaminated by merger effects. We multiply the ratio by −1 in order to have increasing values of BTM correspond to increasing values of goodwill. Both Ind and Stock proxy for the effect of a possible overpayment by the acquirer. If a company is acquired at a price that is higher than its value, any overpayment will remain as goodwill. Morck, Shleifer, and Vishny (1990) show that acquirers are more likely to overpay if they operate in an industry that differs from the target’s. Consequently, we compute Ind as a dummy variable that takes the value of one if the merging companies operate in the same two-digit SIC code. We expect Ind to be negatively correlated with GW since a value of one on Ind indicates that the acquirer is less likely to overpay, which would result in lower goodwill. Similarly, Myers and Majluf (1984) find that acquirers are more likely to settle the purchase price using stock if they are overvalued. Consequently, acquirers might be more generous in negotiations and pay a higher price for the target. Stock is a dummy variable that takes the value of one if the acquirer pays the purchase price at least partially with shares. We expect a positive coefficient on Stock since settling the purchase price with shares indicates a higher likelihood of overpayment which, ceteris paribus, results in more goodwill. While our sample includes mainly large acquirers, targets differ to a larger extent in size. Size effects might thus influence our results and we include Size in our regression, which is computed as the ratio of total purchase price to acquirer’s total pre-acquisition assets. While proxying for potential effects of targets’ size differences, the variable also represents the impact that the acquisition has on the acquirer’s balance sheet. If the price paid to acquire a target is fairly large, the acquisition has a larger impact on the financial position of the acquirer and managers may examine the acquisition more closely to avoid overpayment. Consequently, Size would be negatively associated with goodwill. In addition, we include year and country dummies in our regression. While Yr corresponds to the years in our sample, Cluster represents dummies for the acquirers’ institutional cluster according to Leuz (2010), who groups countries into five clusters according to regulatory characteristics and financial reporting practices. In order to limit the number of controlling variables included in our regression, we do not employ a separate dummy for each country of the acquirer. Results hold when replacing cluster by country dummies. We do not have an expectation for the values of the institutional dummies. 3.4. Bonus model As a next step, we test our hypotheses and build a model that includes compensation variables. The model is based on the economic model as discussed above. More specifically, the model is as follows: GWi,t = ˛ + ˇ1 Synergyi,t + ˇ2 BTMi,t−1 + ˇ3 Indi,t + ˇ4 Stocki,t + ˇ5 Sizei,t−1 + ˇ6–8 Yri,t + ˇ9–10 Clusteri,t + ˇ11 Bonusi,t−1 + ˇ12 DxBonusi,t−1 + ˇ13 Interesti,t−1 + εi,t

(3)

To test hypothesis H1, we expand the economic model by including two additional variables: Bonus and Interest. Bonus is our key independent variable and is calculated as the average of CEO’s short-term cash bonus paid in the two years prior to the acquisition divided by CEO’s total cash compensation. We use the two-year-average to take into account that bonuses may fluctuate. Consequently, an average figure better represents what CEOs expect to receive and seems a better proxy for the incentive presented by short-term bonuses. Bonus is measured prior to the acquisition to proxy for CEOs’ compensation concerns and their expectations of post-acquisition compensation. In addition, pre-acquisition compensation is not contaminated by potential acquisition effects. While we argue that a high level of cash bonus provides an incentive for managers to recognize more goodwill, we need to take mitigating factors into account. We include Interest which is the logarithm of the value of shares held by management in the year prior to the acquisition. We argue that the variable is a proxy for management’s wealth and, consequently, for management’s willingness to manipulate accounting numbers. If a CEO has a high equity stake in a company, his marginal wealth increase is lower compared to a situation in which he owns only a small number of a company’s shares. This reasoning is in line with Benston (1985) who finds that changes in managers’ gains and losses from changes in the values of their stockholdings exceed the compensation they receive. Consequently, a high equity stake is considered to serve as a mitigating factor for short-term opportunistic accounting behavior. Note that we do not include variables in the model that capture future impairment risk. While recognizing a high level of goodwill may increase the risk of having to record impairment in subsequent years, Dechow, Huson, and Sloan (1994) and Gaver and Gaver (1998) show that cash compensation is typically shielded from non-recurring losses such as impairment. More recent research also indicates that cash bonuses are at least partially shielded from extraordinary charges (Adut, Cready, & Lopez, 2003) and are typically not punished for poor firm performance (Shaw & Zhang, 2010). In a second step, we test H2 by including a further variable, DxBonus, which is an interaction term of Bonus and a dummy variable D, which takes the value of one if managers cash bonus is below or above certain thresholds. Healy (1985) uses variables defined in the actual bonus contracts to come up with lower and upper limits for his hypothesis. Since we do not have access to the contracts, we use a more inductive approach and test for thresholds on an exploratory basis. More specifically, we define possible thresholds based on the following equation: s(x) = c + d(x)c

(4)

114

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Table 3 Descriptive statistics. Variable

N

Mean

Std. deviation

Median

Maximum

Minimum

GW Bonus Synergy BTM Interest Size

123 123 123 123 123 123

0.6036 0.5036 0.7967 −0.4736 12.8612 0.2432

0.3019 0.1663 1.3448 0.3442 3.4242 0.5322

0.6323 0.5183 0.1633 −0.4348 13.5349 0.0496

1.2570 0.8859 4.7357 0.2137 16.9450 3.9426

0.0000 0.0000 −1.6432 −2.2727 0.0000 0.0002

Note. Table 3 shows descriptive statistics for the variables included in the regression models. For an explanation of the variables, please refer to Appendix A.

where s(x) is a CEO’s total cash compensation for the period, which depends on the company’s results x, c is the CEO’s fixed compensation for the period, and d(x)c is the CEO’s cash bonus that depends on the outcome x. D(x) constitutes the percentage amount of cash bonus that is granted to a CEO in relation to his or her fixed salary. Often, d(x) is limited to a certain value, i.e. a bonus cap is introduced to provide an upper bound to the CEO’s bonus. Since disclosure on actual bonus caps is insufficient, we cannot assume that lack of disclosure on bonus caps corresponds to nonexistence of such an upper limit. Instead, we use the limited number of bonus caps disclosed in annual reports as indicators to come up with possible values for d(x), which are the following: 100%, 130%, 150%, and 200%. While we cannot assume that a company has not installed a bonus cap, if Bonus is below the respective values for d(x), we can reasonably infer from Bonus being higher than one of the values that a company has not installed the respective cap. 4. Results 4.1. Descriptives and correlation analysis Table 3 reports descriptive statistics for our data. The mean of GW, i.e. the amount of goodwill and indefinite-life intangible assets recognized divided by total consideration, is 60% for our sample, while the maximum is 126% and the minimum 0%. For the executive compensation, our descriptives show that the mean ratio of CEO’s short-term cash bonus to total compensation paid in cash is 50% with maximum and minimum values of 89% and 0% respectively. Synergy takes, on average, a value of 80% while the standard deviation is 134%. Maximum and minimum values are 474% and −164% respectively. BTM has a mean value of −47% with a standard deviation of 34% and a maximum value of 21% while the minimum is −227%. Note that the negative values are due to the fact that we multiply book-to-market ratios by −1 to have increasing values of BTM correspond to increasing values of GW. The descriptives for the shares held by management show that Interest has a mean value of 12.86 and a standard deviation of 3.42, while the maximum is 16.95 and the minimum 0. Finally, Size indicates the relation of the purchase price to acquirer’s total pre-acquisition assets and takes on average the value of 24.32%. Maximum and minimum values are 394% and 0.02%, respectively. Table 4 presents a correlation analysis for our dataset consisting of the variables used in our models. Our key variable, Bonus, is positively correlated with goodwill, GW, which may be seen as an indication in favor of our hypothesis. As expected, the economic determinants of goodwill, Synergy and BTM, are positively correlated with GW. The industry dummy, Ind, is Table 4 Correlation analysis. GW GW Bonus Synergy BTM Ind Stock Interest Size

1 0.1741* (1.9453) 0.2080** (2.3390) 0.3664*** (4.3314) −0.3027*** (−3.4940) 0.1018 (1.1260) −0.1821** (−2.0377) −0.0263 (−0.2891)

Bonus

Synergy

BTM

Ind

Stock

Interest

Size

1 0.1010 (1.1171) 0.0318 (0.3503) 0.0006 (0.0070) 0.1287 (1.4275) 0.0399 (0.4396) −0.1249 (−1.3846)

1 −0.0619 (−0.6820) 0.1088 (1.2043) 0.1964** (2.2038) −0.1616* (−1.8015) 0.251*** (2.8522)

1 −0.1641* (1.8296) 0.0148 (0.1634) −0.0225 (−0.2481) 0.0003 (0.0033)

1 0.2672*** (3.0500) 0.0930 (1.0270) 0.2291 (2.5893)

1 −0.0711 (−0.7847) 0.1914** (2.1445)

1 0.0108 (0.1184)

Note. Table 4 depicts the Pearson pair-wise correlation for the variables included in the regression models. T-statistics are reported in parentheses. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

1

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

115

Table 5 Implicit assumption. Variable Intercept Return RoA DA 2006 2007 2008 Cluster1 Cluster2 AFF Const Man Min Serv TCEGS WT

Predicted + + −

Coefficient 0.0083 0.1181*** −0.1069 −1.4662** 0.0352* 0.0302 0.0567** −0.0012 −0.0318* −0.0417 −0.0249 −0.0437 −0.0687 −0.0339 −0.0506 −0.1140*

N Adj. R2

Std. error 0.0602 0.0229 0.1816 0.6525 0.0187 0.0193 0.0254 0.0137 0.0186 0.0775 0.0644 0.0587 0.0614 0.0597 0.0603 0.068

t-Statistics 0.14 5.16 −0.59 −2.25 1.89 1.57 2.23 −0.09 −1.71 −0.54 −0.39 −0.74 −1.12 −0.57 −0.84 −1.68

266 0.126

Note. Table 5 reports the results of the following regression: Bonusi,t = ˛ + ˇ1 Returni,t + ˇ2 RoAi,t + ˇ3 DAi,t + ˇ4–6 Yri,t + ˇ7–8 Clusteri,t + ˇ9–15 Industryi,t + εi,t . All variables are defined in Appendix A. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

significantly negatively correlated with the dependent variable, which seems to confirm the findings of Morck et al. (1990). Consequently, Ind can be assumed to capture overpayment effects since a value of one, i.e. acquirer and target being in the same industry, would result in a lower amount of goodwill. Similarly, we find that Stock is positively correlated with GW which seems to indicate that acquirers that settle the purchase price at least partially with shares overpay (Myers & Majluf, 1984). While Bonus is positively correlated with GW, we find that Interest is negatively correlated with our dependent variable, which indicates that wealth effects as measured by the amount of closely held shares exist. The correlation suggests that wealthier CEOs are less likely to recognize high goodwill. Finally, Size is only weakly negatively correlated with GW, suggesting a weak relationship between the size of the target compared to the acquirer and goodwill recognized in the acquisition.

4.2. Depreciation and amortization effects on cash bonuses Prior to testing our hypotheses, we need to assess whether the assumption on which our reasoning hinges is fulfilled. Our hypotheses argue that avoiding amortization charges positively influences CEOs’ bonuses. Thus, we examine the effect of changes in depreciation and amortization on changes in the level of CEO cash bonuses. We run our regression model (1) for all acquirers in our sample based on firm years from 2005 to 2008. We exclude financial services companies (SIC codes 60–67) and years in which a firm’s CEO changed, which leaves us with a final sample of 266 firm years. Table 5 presents the results of our regression. As expected, stock returns are positively associated with positive changes of CEOs’ bonuses, which is not surprising since a company’s share price is typically correlated with the company’s results, which in turn is the basis for CEOs’ compensation. However, we find that the coefficient on RoA is insignificant, which at first is counter-intuitive since a change in return on assets should have an impact on the change in the level of bonuses. The insignificant coefficient may either be explained by a misspecification of the model or by the fact that in the years leading up to the financial crisis, the relation between companies’ performance as measured by return on assets and CEOs’ cash bonuses was weak. This reasoning may be confirmed by the significant year dummies, which indicate that bonuses fluctuated more than economic factors may explain. An alternative explanation could be that the economic effect of changes in return on assets is already captured by stock returns, which are highly significant in the model. Nonetheless, our results are in line with prior research (e.g. Ely, 1991), which provides mixed evidence on the relation between RoA and executive compensation. In addition, results are in line with Shaw and Zhang (2010), who find that cash compensation is not punished for poor firm performance. Concerning our variable of interest, we observe that changes in depreciation and amortization charges are associated with the level of CEOs’ cash bonuses. The significant negative coefficient indicates that an increase in amortization is associated with a decrease in cash bonus. This result confirms the assumption inherent in our hypotheses, namely that avoiding amortization charges by recognizing more goodwill may help CEOs avoid decreases in cash bonuses.

116

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

4.3. Economic model Our economic model consists of the economic determinants of goodwill. We include 123 observations in the regression and obtain an adjusted R2 of 23.4%. The regression was run using White-adjusted standard errors. While Synergy, BTM and Ind are significant at the 1% level, Stock and Size are insignificant. In addition, year and cluster dummies are insignificant. The significant coefficients on the economic determinants indicate that model (2) captures two important factors of goodwill recognition. On the one hand, Synergy represents the positive changes in acquirers’ and targets’ market value and thus the amount of synergies to be created by the merger as expected by the combining companies’ investors. The coefficient indicates that for every one-unit increase in Synergy, on average, 5.38% more goodwill and/or indefinite-life intangible assets are recognized.5 In line with the findings of Bradley et al. (1988), our results indicate that investors seem to anticipate the amount of synergies an acquisition generates. In addition, BTM is the relation between the target’s book value of equity to its market value of equity, multiplied by −1. The larger the difference between market and book value, the larger is the target’s ability to generate a higher rate of return on a collection of assets than expected from the assets separately. An increasing fair value of the going concern element, for which BTM is a proxy, implies that the acquirer recognizes more goodwill. As expected, BTM is an important factor of goodwill recognition and we receive a positive coefficient of 0.2864 on the variable. Ind is a proxy for the likelihood of acquirer’s overpayment in that it indicates whether the combining companies operate in the same industry. Our findings are in line with Morck et al. (1990) who find that being in the same industry as the target results in a lower likelihood of overpaying and thus in recognizing less goodwill. Other than that, we do not obtain significant variables for the recognition of goodwill. An acquirer’s mode of payment does not seem to be a factor since the coefficient on Stock is insignificant. In addition, there do not seem to be size effects and results suggest that the size of a target in relation to acquirer’s pre-acquisition total assets does not have an impact on goodwill. Similarly, neither year nor country dummies yield significant coefficients. Overall, model (2) indicates that we capture the economic determinants of goodwill recognition. Table 6 shows the results for our regressions based on the economic, bonus and threshold models. 4.4. Bonus model The bonus model extends the economic model by including the compensation variables, Bonus and Interest. We run model (3) based on 123 observations, using White-adjusted standard errors, and obtain an adjusted R2 of 26.5%. While signs and values of the economic determinants remain largely constant, Bonus has a positive coefficient of 0.2832, which is significant at the 5% level. The coefficient on Bonus indicates that for each one-unit increase in the ratio of cash bonus to total cash compensation the amount of goodwill and/or indefinite-life intangible assets recognized increases by 28.32%. We find weaker evidence that CEOs’ wealth levels, as represented by their stock ownership, is a mitigating factor for goodwill recognition. Interest is significant at the 10% level and thus seems to be a marginal determinant of goodwill recognition. Nevertheless, the higher a CEO’s interest in his or her company, the lower is the amount of goodwill they aim to recognize in excess of economic factors. The results of the bonus model confirm our hypothesis H1. A high level of short-term cash bonus suggests that the acquirer on average recognizes more goodwill. CEOs seem to anticipate this effect since the pre-acquisition average of their cash bonuses is a determinant of the amount of goodwill and indefinite-life intangible assets recognized in the acquisition. 4.5. Threshold model In the threshold model, we include an interaction term of Bonus and a dummy variable that takes the value of one if Bonus is above a certain threshold. As indicated above, there is not sufficient disclosure on bonus caps actually used by companies. Instead, we test different thresholds that serve as indicators and are taken from annual reports of those companies that provide disclosure on their use of bonus caps. We test possible effects of the following thresholds: 100%, 130%, 150%, and 200%. D takes the value of one if Bonus is above the respective threshold. We proceed by iteratively testing these thresholds and correspondingly adjusting the dummy on the interaction term. In results not tabulated, we do not find that the interaction term DxBonus is significant for the threshold levels of 100%, 130% and 200%. However, if we include the interaction term based on a dummy that takes the value of one if CEOs’ cash bonuses are above 150%, we obtain significant results. In fact, we are able to limit the significant coefficient to those companies whose CEOs earn bonuses that are between 150% and 200% of their base salary. For these observations, we find that the interaction term is significant at the 1% level. Since Bonus is insignificant once we include the interaction term and results seem to be driven exclusively by observations for which D takes the value of one, we exclude the Bonus variable for further analyses and reduce the threshold model to a regression that includes only DxBonus.6 Table 6 shows the results for our final model in which Interest is significant at the 5% level, while the coefficient remains largely the same.

5

Results hold when excluding observations for which changes in stockholders’ wealth divided by the purchase price is greater than one (see footnote 4). Regressions shown in the following were also run for the complete threshold model, i.e. the one that includes both Bonus and DxBonus. Results, which are untabulated, remain largely the same, although we lose some significance – probably due to multicollinearity. 6

Variable

Predicted

Economic model Coefficient

Intercept Synergy BTM Ind Stock Size 2006 2007 2008 Cluster1 Cluster2 Bonus DxBonus Interest N Adj. R2

Bonus model t-Statistics

Coefficient

t-Statistics

Coefficient

11.06 3.36 3.66 −3.18 1.47 −0.91 −1.35 −1.27 −0.91 −0.64 −0.75

0.1426 0.0158 0.0821 0.0556 0.0708 0.0340 0.0919 0.0688 0.0826 0.0557 0.0683 0.1249

6.73 2.75 3.38 −3.28 1.07 −0.27 −1.66 −1.61 −1.09 −1.29 −1.27 2.27

1.0518*** 0.0516*** 0.2639*** −0.1771*** 0.0872 −0.0129 −0.1292 −0.0747 −0.0653 −0.0810 −0.0721

0.1289 0.0158 0.0800 0.0551 0.0691 0.0294 0.0905 0.0679 0.0839 0.0568 0.0640

8.16 3.26 3.3 −3.22 1.26 −0.44 −1.43 −1.1 −0.78 −1.43 −1.13

+

0.9599*** 0.0435*** 0.2779*** −0.1823*** 0.0760 −0.0094 −0.1527* −0.1104 −0.0901 −0.0719 −0.0871 0.2832**

−

−0.0144*

0.0072

−2.02

0.2875*** −0.0163**

0.0963 0.0072

2.99 −2.28

+ + − + ±

0.8721*** 0.0538*** 0.2864*** −0.1875*** 0.0977 −0.0284 −0.1204 −0.0870 −0.0776 −0.0336 −0.0513

Std. error

Threshold model

0.0789 0.0160 0.0782 0.0590 0.0663 0.0309 0.0893 0.0685 0.0850 0.0525 0.0683

123 0.234

Std. error

123 0.265

Std. error

t-Statistics

123 0.298

Note. Table 6 depicts the results of the following regression: GWi,t = ˛ + ˇ1 Synergyi,t + ˇ2 BTMi,t−1 + ˇ3 Indi,t + ˇ4 Stocki,t + ˇ5 Sizei,t−1 + ˇ6–8 Yri,t + ˇ9–10 Clusteri,t + ˇ11 Bonusi,t−1 + ˇ12 DxBonusi,t−1 + ˇ13 Interesti,t−1 + εi,t . All variables are defined in Appendix A. The regressions were run using White-adjusted standard errors. The bonus model expands the economic model by including Bonus and Interest, while the threshold model adds the interaction term DxBonus. D takes the value of one if a CEO’s cash bonus is between 150% and 200% of their base salary. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Table 6 Results.

117

118

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Our results confirm our second hypothesis. Based on Healy’s (1985) research framework, we find that there is a certain corridor in which managers’ compensation represents an incentive for them to take discretionary accounting choices opportunistically. More specifically, this situation only seems to be the case if CEOs’ bonuses make up between 150% and 200% of fixed salaries. In that case, managers seem to be willing to recognize about 28.75% more goodwill with each unit increase of their cash bonus in relation to total cash compensation. Our results indicate that a bonus cap at 150% of base salaries helps to re-align managers’ and shareholders’ interests by making CEOs choose a middle-ground solution in that they do not recognize excessive goodwill. Bonus caps can make managers more risk-averse and focused on longer-term objectives (Arya et al., 2007). The incentive to recognize more goodwill seemingly ceases to exist once managers’ bonuses are above 200% of their base salaries. This finding may be due to the small number of observations in our sample for which the condition is fulfilled. We may also suggest that managers refrain from taking opportunistic action with regards to goodwill once their cash bonus is above the upper threshold. Managers may regard their high prior bonuses as exceptional and do not feel they can sustain the level of cash bonus. Alternatively, they do not influence the recognition of goodwill because they do not consider more goodwill as advantageous in their situation. They may not expect additional gains and instead might be aware that more goodwill also brings about higher risk of future impairment. 5. Further analyses 5.1. Subsample analyses Since our sample is relatively heterogeneous, we conduct subsample analyses to verify that our overall results hold. In addition, we want to assess whether results are driven by certain observations. Specifically, we examine more closely non-financial companies as well as country differences. Typically, financial institutions, such as banks and insurance companies, are excluded from studies since financial statements of these companies differ to a large extent from the ones of other companies. For the purpose of our study, we do not consider the differences between the accounting of financial and non-financial companies to be severe because our regression models rely only to a limited extent on accounting figures that are presented differently by financial institutions. Nevertheless, we re-run the model (3) regression for a sample for which we exclude those acquiring companies operating in the two-digit SIC codes 60–67. This procedure reduces our sample to 102 observations. We re-run the model using White-adjusted standard errors and observe that overall results hold, while the level of significance reduces for Synergy, BTM and Ind. The variables are significant at the 5% level only, which may be due to the lower number of observations included in the regression. In addition, we observe that Stock is significant at the 10% level. The coefficient on Stock has increased considerably compared to the full sample, weakly indicating that for non-financial companies, stock payments result in a higher purchase price and consequently in higher goodwill. This finding is again in line with Myers and Majluf (1984). The subsample analysis reveals another difference that relates to our key variable DxBonus. While still being significant at the 1% level, its coefficient has increased from 0.2875 to 0.3467, which may indicate that CEOs of non-financial companies are more exposed to the incentive given by cash bonuses. Cluster1 is significant at the 10% level and has a negative sign, which indicates that acquirers residing in the UK are not as susceptible to the incentive provided by their bonuses as acquirers from other European countries. Adjusted R2 for the subsample analysis is 27.9%, which is slightly higher compared to results of the full sample. Table 7 shows the results of our subsample analyses. We extend our previous analysis that suggests that acquiring companies may differ concerning goodwill recognition depending on their origin. Leuz et al. (2003) and Lang et al. (2006) show that continental European countries are more prone to earnings management than Anglo-Saxon companies. Adapting their research to our study, we run two additional regressions that analyze whether UK acquirers approach the recognition of goodwill differently from non-UK acquirers. As indicated in the descriptives section, UK acquirers make up the largest portion of our observations. Consequently, we suspect that acquirers from the UK drive our results, although we obtained an insignificant coefficient for the Cluster1 dummy for the full sample. Based on the subsample of 55 UK acquirers, we obtain an adjusted R2 of 18%, which is considerably lower than in the main results. We find that the model yields significant coefficients for Ind and Stock only, which are significant at the 5% level and the 10% level, respectively. Economic determinants of goodwill are insignificant as is the key variable in our model DxBonus. These results may be due entirely to the small number of observations included. Nevertheless, they suggest weakly that either other variables drive goodwill recognition in the UK or that cash bonuses are not a determinant of goodwill for the observations included. To compare these results, we run a regression based on non-UK companies, i.e. using the remaining 68 observations. Using White-adjusted standard errors, we obtain an adjusted R2 of 50% as well as significant coefficients for Synergy (5% level), BTM (1% level) and Ind (10% level). We observe that the 2008 dummy is significant at the 1% level. The negative coefficient on 2008 indicates that non-UK acquirers treated the recognition of goodwill considerably different in this year. This result may be explained by suggesting that acquirers anticipated the upcoming crisis and wanted to mitigate the risk of goodwill impairment in subsequent years. In addition, we find that the variable of interest, DxBonus, is significant at the 1% level. The coefficient of 0.4638 is larger than in the base model, suggesting that non-UK acquirers are more susceptible to cash bonus incentives. The Jarque–Bera

Variable

Predicted

No financial companies Coefficient

Intercept Synergy BTM Ind Stock Size 2006 2007 2008 Cluster1 Cluster2 DxBonus Interest N Adj. R2

+ + − + ±

+ −

0.9817*** 0.0462** 0.2255** −0.1462** 0.1666* −0.0315 −0.1054 −0.0124 −0.0286 −0.1134* −0.0798 0.3467*** −0.0147*

Std. error 0.1370 0.0182 0.1031 0.0636 0.0893 0.0361 0.1188 0.0785 0.0921 0.0631 0.0726 0.1009 0.0081 102 0.279

UK acquirers only

Non-UK acquirers only

t-Statistics

Coefficient

Std. error

t-Statistics

Coefficient

Std. error

7.16 2.54 2.19 −2.30 1.87 −0.87 −0.89 −0.16 −0.31 −1.80 −1.10 3.44 −1.82

0.8254*** 0.0404 0.0164 −0.2393** 0.1558* −0.0189 −0.1526 −0.0439 0.1240

0.2222 0.0317 0.1255 0.0895 0.0910 0.0903 0.1187 0.1099 0.1186

3.71 1.27 0.13 −2.67 1.71 −0.21 −1.29 −0.40 1.05

1.0311*** 0.0436** 0.4518*** −0.1047* −0.0446 0.0002 −0.0905 −0.1014 −0.2183***

0.1292 0.0197 0.0961 0.0562 0.0842 0.0312 0.0940 0.0636 0.0770

7.98 2.22 4.70 −1.86 −0.53 0.00 −0.96 −1.59 −2.83

0.2192 −0.0160

0.1575 0.0158

1.39 −1.01

−0.0389 0.4638*** −0.0074

0.0641 0.1313 0.0074

−0.61 3.53 −1.00

55 0.180

t-Statistics

68 0.500

Note. Table 7 depicts the results of our regression model for subsamples. The first column shows the regression results based on a sample that excludes financial companies (SIC codes 60–67). The second column reports the results for acquisitions that were completed by companies residing in the United Kingdom. The third column shows the results for the non-UK subsample that excludes acquirers from the UK. Regressions (1) and (3) were run based on White-adjusted standard errors. Residuals for the Non-UK subsamples may not be normally distributed, since the Jarque–Bera test yields a p-value of 0.049. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Table 7 Subsample analysis.

119

120

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

test shows that residuals are not normally distributed (p-value of 0.049) for this subsample, which again may be due to the small number of observations included. Nevertheless, the results of these models are in line with the findings of Leuz et al. (2003) and Lang et al. (2006) and provide somewhat weak evidence that non-UK acquirers are more reluctant to recognize higher goodwill due to compensation considerations than acquirers residing in the UK.7 In fact, the results of our subsample analyses, although weak, may be seen as adding to this literature, which concentrated on earnings management measures and examined differences between countries based on regulatory variables. Our study provides motivation for prior findings by suggesting that compensation concerns are a reason for the differences found. 5.2. Robustness tests Our previous results suggest that acquirers whose CEOs receive a high level of cash bonus recognize more goodwill in an acquisition. In order to test the robustness of these results, we conduct further analyses. As a first step, we split the observations into three portfolios based on CEOs’ bonus intensity and compare the amount of goodwill recognized in the respective acquisitions. We create a low portfolio that includes acquisitions completed by acquirers whose CEOs’ bonuses make up less than 100% of the base salary. The mid portfolio is based on acquirers with cash bonuses between 100% and 150% of the base salary, while the upper portfolio includes all acquirers whose CEOs earn more than 150% of their base salaries as cash bonuses.8 Using the three portfolios, we examine whether acquirers recognize different amounts of goodwill in acquisitions. Results are shown in Panel A of Table 8. A joint test of the portfolios indicates that the portfolios indeed differ with regards to the amount of goodwill and indefinite-life intangible assets recognized. Conducting a more detailed analysis, we observe that the difference between the amount of goodwill recognized by companies in the low versus mid portfolio is not statistically significant. There is a statistically significant difference between both the low and the upper and between the mid and the upper portfolio. These results support our prior findings that CEOs of companies in the upper portfolio are susceptible to the incentive provided by their cash bonuses, while CEOs of companies in the lower and in the mid portfolio are not. The comparison between the low and the upper portfolio yields statistical significance at the 5% level, while the values obtained for the comparison of the mid and the upper portfolio are significant at the 1% level. Overall, this analysis confirms the results of our regressions and provides additional support for our hypotheses. We next examine whether the hypothesized relation between bonuses and goodwill pays off for CEOs. In our analysis, we assume that CEOs act opportunistically based on their anticipation of either retaining the same bonus levels or even increasing cash bonuses in years subsequent to the acquisition. Consequently, we examine if CEOs’ expectations are valid. To do so, we compare the two-year pre-acquisition average of CEOs’ cash bonuses to total CEO cash compensation, i.e. the Bonus variable, to the same ratio in the year of the acquisition and in each of the two subsequent years. The comparison is conducted by examining whether the difference between pre-acquisition and subsequent bonus is statistically different from zero. Since bonuses are not normally distributed, we conduct the analysis based on medians. We proceed by first examining the entire sample and then observations that were below the 150% threshold prior to the acquisition. Finally, we examine bonus differences for acquirers in the corridor of 150–200%, i.e. for observations in which D takes the value of one in our regression. We are aware that the results of this analysis do not provide causality for the link between goodwill and post-acquisition bonus. However, we see the analysis as an indicator of whether or not CEOs’ anticipations are valid. Panel B of Table 8 shows the results of the bonus comparisons. For the entire sample as well as for the two subsamples, we observe that CEOs earn a higher bonus in the year of the acquisition. This increase in compensation may be considered a reward for conducting the acquisition. While the values of both Wilcoxon-test and van der Waerden-test show a statistical significance at the 1% level, the sign test indicates somewhat weaker significance for the subsamples. When comparing the median bonus differences between the two subsamples, we find somewhat weak evidence (statistical significance at the 10% level and at the 5% level, respectively) that CEOs who were in the 150% to 200% corridor pre-acquisition receive a higher bonus increase in the year of the acquisition (median difference of 0.0601) than CEOs whose pre-acquisition bonus was lower than 150% of their base salary (median difference of 0.0401). We also compare pre-acquisition bonuses to bonuses earned in the year after the acquisition. Results are shown in Panel C of Table 8. The analysis does not yield statistical significance for the median comparisons, indicating bonuses return to preacquisition levels. However, we find somewhat weak evidence (significance at the 10% level and at the 5% level, respectively) that changes in bonuses are different between the two subsamples. Considering the positive median of observations below the thresholds (value of 0.0244) and the negative median of the observations in the corridor (value of −0.0219), we might argue that there is a general–more positive–notion for CEOs who were below the threshold pre-acquisition and who are not assumed to have recognized goodwill opportunistically. Panel D of Table 8 confirms this trend. Results for the entire sample suggest that bonuses are slightly lower in the second year after the acquisition compared to pre-acquisition levels. The medians of the subsamples indicate that results

7 We also run a regression that includes acquisitions completed by companies from the Eurozone. These results are very similar to the ones for the non-UK subsample, which is why they are not tabulated. 8 Note that the upper portfolio is not completely in line with the definition of D in our final threshold model. Due to the fact that the number of acquisitions above the 200% threshold is not very large, we subsume these observations in the upper portfolio.

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

121

Table 8 Robustness tests. Panel A: Portfolio comparison Method

Joint test Value

t-Test Satterthwaite–Welch t-test ANOVA F-test Welch F-test

4.7040** 5.0425***

Low vs. Mid

Low vs. Upp

Mid vs. Upp

p-Value

Value

p-Value

Value

p-Value

Value

p-Value

0.0108 0.0088

0.7566 0.7630 0.5724 0.5822

0.4514 0.4477 0.4514 0.4477

−2.3726** −2.4184** 5.6292** 5.8486**

0.0199 0.0179 0.0199 0.0179

−3.0245*** −3.0286*** 9.1475*** 9.1727***

0.0035 0.0034 0.0035 0.0034

Panel B: Bonus comparison (in t) Method

Total Sample

Bonus < 150%

150% < Bonus < 200%

Value

p-Value

Value

p-Value

Value

p-Value

Sign Wilcoxon signed rank van der Waerden

3.5218*** 4.0643*** 4.2419***

0.0004 0.0000 0.0000

2.4518** 2.7982*** 2.9076***

0.0142 0.0051 0.0036

1.7650* 2.7685*** 2.8208***

0.0776 0.0056 0.0048

Sample median

0.0325

0.0401

0.0601

Equality of medians between subsamples Method

Value

p-value

Wilcoxon/Mann–Whitney Wilcoxon/Mann–Whitney (tie–adj.) van der Waerden

1.7275* 1.7276* 4.2734**

0.0841 0.0841 0.0387

Panel C: Bonus comparison (in t + 1) Method

Sign Wilcoxon signed rank van der Waerden Sample median

Total sample

Bonus < 150%

150% < Bonus < 200%

Value

p-Value

Value

p-Value

Value

p-Value

1.4910 0.0929 −0.3769

0.1360 0.9259 0.7063

2.5440** 0.8715 0.4869

0.0110 0.3835 0.6263

0.5883 1.6001 −1.7594*

0.5563 0.1096 0.0785

0.0091

−0.0219

0.0244

Equality of medians between subsamples Method

Value

p-Value

Wilcoxon/Mann–Whitney Wilcoxon/Mann–Whitney (tie–adj.) van der Waerden

1.9381* 1.9382* 4.1885**

0.0526 0.0526 0.0407

Panel D: Bonus comparison (in t + 2) Method

Total sample

Bonus < 150%

150% < Bonus < 200%

Value

p-Value

Value

p-Value

Value

p-Value

Sign Wilcoxon signed rank van der Waerden

1.3979 1.9607** −2.2539**

0.1621 0.0499 0.0242

0.3162 0.2052 −0.5365

0.7518 0.8374 0.5916

2.1573** 2.9209*** −2.9555***

0.0310 0.0035 0.0031

Sample median

−0.0087

−0.0267

0.0012

Equality of medians between subsamples Method

Value

p-Value

Wilcoxon/Mann–Whitney Wilcoxon/Mann–Whitney (tie–adj.) van der Waerden

2.1394** 2.1395** 5.1951**

0.0324 0.0324 0.0227

Note. Table 8 depicts the results of robustness tests. Panel A reports results of a portfolio comparison. Based on CEO’s bonuses, we split the sample into three different portfolios. Low denotes the portfolio of acquirers whose CEOs’ bonuses are less than 100% of the base salary. Mid is the portfolio of acquirers with CEOs whose bonus is between 100% and 150% of base salary, while Upp is the upper portfolio that consists of acquirers with bonuses above 150% of the base salary. The first column shows a joint test of the average amount of goodwill recognized in the portfolios. The other columns depict the results of a comparison between two portfolios (low vs. mid; low vs. upp; mid vs. upp). Panels B through D show the results of bonus comparisons for the year of the acquisition (Panel B), one year after the acquisition (Panel C) and two years after the acquisition (Panel D). We compare the difference between the two-year pre-acquisition average of cash bonus to total cash compensation and post-acquisition cash bonus over total cash compensation. Since bonuses are not normally distributed, we examine whether medians are different from zero. First, a test is run for the sample as a whole (shown in the first column). The second and third column show the results of tests for the subsamples of pre-acquisition bonus less than 150% of base salary and pre-acquisition bonus between 150% and 200% of base salary, respectively. In addition, we compare the bonus differences between the two subsamples. Results are shown in the lower table of the respective panel. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

122

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

are driven by those observations within the 150–200% corridor prior to the acquisition. While bonuses of CEOs below the threshold have returned to their pre-acquisition levels (median value of 0.0012), the in-corridor subsample shows that postacquisition bonuses are significantly lower than pre-acquisition bonuses (median value of −0.0267). A comparison between the subsamples confirms the results, differences being significant at the 5% level. Overall, we conclude that CEOs who are assumed to take discretionary actions regarding the recognition of goodwill are rewarded in the year of the acquisition, probably for conducting the acquisition. In subsequent years their bonuses decrease significantly, possibly consuming the benefits of the higher rewards earned in the year of the acquisition. On the other hand, CEOs who do not recognize goodwill opportunistically may serve as a control sample. An analysis of their bonuses shows that they too are rewarded for the acquisition, however somewhat lower than CEOs who are within the 150–200% bonus corridor pre-acquisition. In subsequent years, their bonuses return to pre-acquisition levels and remain more or less constant, which suggests more consistent bonuses and, in turn, smoother earnings for the acquiring companies. Further robustness tests are conducted on the regression results. Foremost, we aim to assess whether future impairment risk plays a role in CEOs considerations on goodwill. If executives are willing to recognize higher goodwill because of their cash bonuses, they need to consider that impairment charges may also have an impact on their positions. While Dechow et al. (1994) and Gaver and Gaver (1998) show that cash compensation is shielded from non-recurring losses such as impairment, we want to determine if this situation applies to our sample. Since impairment tests include a considerable amount of discretion potential as well, we argue that future impairment risk is less likely to be considered by CEOs if they expect to have large discretion potential concerning impairment tests in the future. In order to capture this discretion potential, we refer to Ramanna (2008) who identified three main factors as proxies: the number and size of reporting units, reporting-units’ fair-value-to-book-value ratios and unverifiability of net assets. Similar to Ramanna (2008), we compute Seg as the product of the logarithm of the number of segments and the logarithm of total firm sales (ln(segments)xln(sales)). While the number of segments proxies for the number of reporting units, firm sales represent the size of the reporting units, which taken together correspond to the size of the firm. It is argued that an increasing number of segments provides the acquirer with more choice of units to which to allocate goodwill. In addition, higher sales increase the size of the segments and, therefore, the flexibility to hide impairments. The second discretion variable, reporting-units’ fair-value-to-book-value ratios, is based on the rationale that, when conducting an impairment test, companies compare the fair value of a reporting unit to its book value. An increasing ratio of fair value to book value decreases the likelihood of impairment and increases discretion potential. As data on reporting units are not available, we use a firm-wide market-to-book-value ratio, MTBV. Finally, Ramanna (2008) argues that the fair value of some assets can be verified more easily than the fair value of others. With an increasing amount of unverifiable net assets, a company’s flexibility in estimating the fair values of net assets and goodwill increases. Consequently, discretion in conducting the impairment test increases. We follow Ramanna’s (2008) computation of verifiable net assets (VNA) as the ratio of [Cash + Investments − Debt − Preferred Equity]–[Assets − Liabilities]. In addition, we use his modifications and multiply VNA by −1 to obtain a proxy for a company’s unverifiable net assets, UNA. Also, we calculate Mod UNA as |(1−|UNA|)|. Larger values of UNA and Mod UNA correspond to more flexibility in estimating the fair value of goodwill and in managing impairment losses. All variables that are based on Ramanna (2008) are computed post-acquisition. We include these variables in the regression jointly and one by one. We find that overall results hold while both MTBV and any of the VNA variables are insignificant. Only Seg is significant at the 10% level. When including only Seg in the regression, the coefficient on the variable is negative and takes a value of −0.0092. Although all other discretion variables are insignificant, the significant coefficient on Seg may be seen as a weak indicator that CEOs consider how much discretion they can exercise in impairment tests. Contrary to our expectations, the coefficient is negative. This result may be explained by the fact that, according to Ramanna (2008), Seg does not necessarily capture future impairment discretion potential but is also a proxy for the size of a company. Seg may be seen as a variation of our Size variable and weakly suggests that large acquirers tend to recognize slightly less goodwill on an acquisition. Other robustness tests deal with the extent to which CEOs are susceptible to incentives given by their bonuses. Our hypothesized relationship may be more pronounced if, for example, managers’ bonuses are entirely based on earnings. By contrast, managers whose bonuses are determined by reference to cash flow figures may be less prone to manage goodwill. The relationship may also be impacted if bonuses are based on accounting figures that capture impairment charges (e.g. EBITDA). We construct two dummy variables that aim to capture the extent to which executives are exposed to the incentive. First, CF takes the value of one if acquirers’ compensation contracts as disclosed in the annual reports refer to cash flow figures, and zero otherwise. Second, Adj takes the value of one if CEOs’ bonuses are determined by reference to adjusted earnings figures, and zero otherwise. Adjusted earnings figures typically exclude extraordinary items such as impairment charges. Since not all acquirers provide sufficient disclosure in their remuneration reports, our sample is reduced to 108 observations. Results show that neither CF nor Adj is significant, which may be due either to the poor disclosure level in acquirers’ remuneration reports or to the fact that referring to adjusted earnings figures does not impact managers’ opportunistic recognition behavior. We compute two additional variables based on managers’ age and tenure. Zhang and Zhang (2007) argue that older managers and those with a long tenure are more entrenched than their younger counterparts or managers who joined a company recently. They posit that entrenched managers are, on the one hand, less likely to suffer from reporting goodwill impairment and more likely to have a say on their compensation contracts. On the other hand, entrenched managers have

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

123

more to lose if they are fired. Following this reasoning, we do not have a prediction for the values of Age and Tenure. When including the two variables in our regression either jointly or separately, we do not obtain significant coefficients on the variables. Overall results hold. 6. Discussion of results Our results confirm our hypotheses and suggest that CEOs whose pre-acquisition bonuses make up between 150% and 200% of their base salaries recognize goodwill in excess of its economic determinants. We argue that CEOs expect to receive higher cash bonuses since goodwill is not amortized on a straight-line basis. Our study provides further evidence for Gordon’s (1964) proposition that managers make accounting decisions opportunistically in order to maximize their own welfare. We cannot confirm that the relationship between cash bonus and goodwill recognition is a linear one. Instead, by building on Healy’s (1985) framework, we find a corridor in which a CEO’s cash bonus is an incentive to recognize more goodwill. Outside this corridor, our results do not suggest that management behaves opportunistically due to cash bonuses. Our findings suggest that companies should incorporate a bonus cap in their compensation arrangement. Such a cap would limit CEOs’ cash bonuses to a certain percentage of base salary. In particular, our study suggests that a bonus cap at 150% of CEOs’ base salaries could avoid opportunistic accounting for acquisitions and excessive goodwill.9 Our results are in line with the study by Arya, Glover, and Mittendorf (2007), who make a case for introducing bonus caps to compensation contracts. They find that such caps induce a longer-term focus in CEOs’ behavior and encourage executives to prefer a middle-ground solution, thus being more risk-averse. Overall, they suggest that bonus caps help align shareholders’ and managers’ interests. Our research is framed by principal agent theory, which suggests that shareholders’ and managers’ interests conflict and need to be realigned. Such realignment is done by compensation agreements that monetarily incentivize managers to perform their jobs effectively. Since the compensation arrangements make reference to accounting figures, discretion in applying accounting standards introduces another problem (Johnson & Revsine, 1988). Managers may engage in opportunistic behavior and exploit discretion potential in order to maximize their own wealth. In our case, the recognition of goodwill is considered a discretionary accounting choice that managers are willing to influence if their compensation relies heavily on cash bonuses. Thus, our approach is in line with accounting choice literature that examines one-time events such as accounting procedure changes (e.g. Healy, Kang, & Palepu, 1987) or, as in our case, mergers and acquisitions. Whittington (1986) suggests that principal agent considerations potentially introduce adverse selection problems since “unregulated information may well show the most unscrupulous (rather than the best) management in the most favorable light.” The accounting for business combinations and the recognition of goodwill may qualify for being loosely regulated since acquirers can exercise much discretion when identifying and measuring assets acquired and liabilities assumed. Unscrupulous managers may exploit the discretion inherent in IFRS 3 to recognize a high level of goodwill in order to be viewed more favorably. When examining bonuses in the year of the acquisition, we argue that the most unscrupulous management, i.e. CEOs who seem to recognize goodwill opportunistically, profit the most because their bonus increases are significantly higher than those of other CEOs. In subsequent years opportunistic managers’ bonuses decrease significantly below their pre-acquisition levels. “Honest” managers, by contrast, do not experience bonus decreases, thus receiving more sustainable and consistent compensation. Overall, our analysis suggests that an adverse selection problem may exist in the year of the acquisition. However, when examining bonuses in the following years, it seems that managers suffer from their opportunistic behavior in earlier periods and receive lower bonuses. Thus, the existence of adverse selection problems can be rejected. While agency theory provides an appropriate framework for our analysis, other interpretations of our results also may be possible. In particular, it might be argued that our research design does not capture managers’ expectations of higher bonuses but proxies for CEO hubris. Roll (1986) and Hayward and Hambrick (1997) explain M&A activities by managers’ hubris and relate executive compensation to premiums paid in acquisitions. Based on this reasoning, it might be argued that the more pronounced CEOs’ hubris is, the more likely the acquirer overpays. Consequently, any overpayment would result in a higher amount of goodwill to be recognized. Similarly, it could be argued that managers’ bonuses are an indicator for CEOs’ power in their respective companies (Finkelstein, 1992). The managerial power hypothesis suggests that the more powerful CEOs are the more they “exert their own wills.” In the case of a merger or acquisition, this hypothesis would suggest that powerful managers want to further increase their power and conduct acquisitions at any price. Any overpayment by the acquirers results in higher goodwill. Both the hubris and managerial power hypotheses seem to provide an additional explanation for the motives of mergers and acquisitions. However, as Pavlik, Scott, and Tiessen (1993) note, the theories lack a dynamic element and merely provide an explanation for the status quo. In our case, the theories may explain why acquisitions were conducted and why companies may have overpaid and need to recognize excessive goodwill. They may explain motives for mergers and acquisitions, per se, but do not help explain motives for the accounting for an acquisition and the possibility of voluntarily overstating goodwill. Principal agent theory provides the dynamic element the other hypotheses lack and helps explain the increase in CEOs’ bonuses in the year of the acquisition and the subsequent decrease. Thus, the social theories seem to be somewhat shortsighted for our research hypotheses. Nevertheless, we acknowledge that they provide an additional explanation for the recognition of goodwill. They do not change the general notion of our paper.

9

We are aware that companies may want to recognize high amounts of goodwill for various other reasons (e.g. financial covenants).

124

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Finally, we acknowledge that our research design has certain limitations that seem to be inherent in most of the accounting choice literature (e.g. Fields et al., 2001). Mergers and acquisitions are complex phenomena that are difficult to reproduce in a regression model. We admit that there might be important variables that we omitted either for practical reasons, such as non-disclosure by companies, or because they are not observable. Other limitations relate to the measurement of our Bonus variable. On the one hand, compensation agreements could include an acquisition clause that provides special provisions for CEOs’ bonuses in the event of an acquisition. However, such a clause would not explain the significant results we obtain. On the other hand, it could be argued that an acquirer’s CEO has nothing to do with the accounting for an acquisition or cannot exert any influence over the recognition of goodwill. Instead, a CFO would most likely handle all accounting issues and be the person ultimately responsible for the accounting. Our regressions could be spurious in that they do not capture the underlying economic relationship adequately. In that case, our results could be explained by the fact that CEOs’ and CFOs’ payment structures typically do not differ much and have similar effects on the recognition of goodwill.

7. Conclusions Based on agency theory, we examine the relation between CEOs’ bonuses and the recognition of goodwill in European IFRS mergers. We find that the more CEOs’ cash compensation packages depend on cash bonuses, the more goodwill is recognized in the acquisition. We explain our results by managers taking discretionary accounting choices opportunistically to exploit the fact that their compensation contracts rely on companies’ accounting results. By making use of the flexibility that IFRS 3 offers regarding the recognition of goodwill, managers are assumed to act in the way Gordon (1964) proposes and choose accounting procedures to maximize their own benefits. While we are able to confirm our first hypothesis that suggests a linear relationship between cash bonuses and goodwill recognition, we find that the relation seems to hold for a certain number of observations only. Specifically, we limit our findings to a corridor in which CEOs seemingly are susceptible to the incentive their cash bonuses provide. If cash bonuses are between 150% and 200% of CEOs base salaries, we find that managers recognize more goodwill. We are able to confirm this result by building portfolios based on CEOs’ cash bonuses. Acquirers in the upper portfolio, which is based on acquirers whose CEOs earn more than 150% of their salaries in cash bonuses, are associated with significantly higher goodwill than companies in the other portfolios. Our findings are in line with Healy (1985), who provides evidence for managers taking discretionary accounting choices within certain thresholds. Our research has implications for companies that use compensation arrangements. These agreements often include bonus caps that define an upper limit to the amount of cash bonus a manager can earn. Since we observe only opportunistic behavior of managers whose cash bonuses are above 150% of their base salary, we conclude that companies should install a bonus cap at that level. Subsample analyses indicate that non-financial companies and acquirers residing outside the UK are prone to recognizing goodwill in excess of its economic determinants. These findings are in line with prior literature (Lang et al., 2006; Leuz et al., 2003), which shows that earnings management is more common in continental European countries. We extend this literature by providing somewhat weak evidence that the countries in our sample seem to differ with regard to the susceptibility of CEOs to the incentives given by their cash bonuses. Our study presents an additional explanation for the differences in earnings management by suggesting that cash bonuses incentivize executives in a different way. When analyzing whether or not managers correctly anticipate that a higher level of goodwill will result in higher bonuses, we find that in the year of the acquisition, cash bonuses of CEOs who recognize goodwill opportunistically increase to a larger extent than cash bonuses of other CEOs. We interpret this finding by suggesting goodwill-adjusting CEOs receive a larger reward for conducting the acquisition than non-adjusting CEOs. Further analysis shows that, in subsequent years, cash bonuses of opportunistic CEOs decrease below pre-acquisition levels, while other CEOs’ bonuses return to pre-acquisition levels. This result suggests that cash bonuses of non-adjusting CEOs are more consistent. Given the link between bonuses and accounting performance, the results indicate that earnings of companies managed by non-adjusting CEOs are more sustainable. Our study contributes to research on accounting choice and executive compensation issues in an IFRS environment. In addition, we add to the literature on the recognition of goodwill, which so far has been examined only scarcely for IFRS companies. We are aware of the limitations of our study and suggest that future research could examine the relationship between executive compensation and goodwill based on a larger sample. Such research could combine questions of goodwill recognition with questions of goodwill impairment and could examine the extent to which short-term recognition decisions and longer-term impairment concerns are driven by economic incentives.

Acknowledgments We gratefully acknowledge the valuable and constructive comments by Sebastian Hoffmann. We also thank Kathleen Sinning (the editor), the reviewers and participants of the 34th Annual Meeting of the European Accounting Association (EAA) and the 7th Workshop on European Financial Reporting (EUFIN). We thank Gregor Kaczmarek and Tobias Kretzschmar for their research assistance.

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

125

Appendix A. Variable definitions Variable

Definition

Bonus Return RoA DA Yr Cluster

Year-to-year change in CEO’s short-term cash bonus divided by CEO’s total cash compensation One year buy-and-hold-return of company’s stock Year-to-year change in return on assets (return being defined as net income plus depreciation and amortization charges) Year-to-year change in depreciation and amortization expenses divided by total assets Year dummy Cluster dummy based on Leuz (2010) (Cluster 1: United Kingdom; Cluster 2: Belgium, Finland, Netherlands; Cluster 3: France, Germany, Italy, Norway, Spain, Sweden, Switzerland) Industry dummy according to SIC classification (AFF = agriculture, forestry, fishing; Const = construction; Man = manufacturing; Min = Mining; Serv = services; TCEGS = transportation, communications, electric, gas, and sanitary services; WT = wholesale trade) Goodwill plus indefinite-life intangible assets divided by total purchase consideration Logarithm of positive changes in market value of combining companies after merger announcement (multiplied by −1); Change is computed as sum of abnormal changes in market value over three-day window divided by total consideration Target’s book to market ratio one quarter prior to merger announcement (multiplied by −1) Dummy that is one if merging companies operate in same two-digit SIC code Dummy that is one if the acquirer partially paid the acquisition price with shares Total purchase consideration to acquirer’s total assets prior to the acquisition Two-year average of CEO’s short-term cash bonus to CEO’s total cash compensation (calculated pre-acquistion) D is a dummy that takes the value of one ifBonus is below/above certain thresholds Logarithm of value of shares held by management in the year prior to the acquisition

Industry GW Synergy BTM Ind Stock Size Bonus DxBonus Interest

References Adut, D., Cready, W. H., & Lopez, T. J. (2003). Restructuring charges and CEO cash compensation: A reexamination. The Accounting Review, 78(1), 169–192. Armstrong, C. S., Jagolinzer, A. D., & Larcker, D. F. (2010). Chief executive officer equity incentives and accounting irregularities. Journal of Accounting Research, 48(2), 225–221. Arya, A., Glover, J., & Mittendorf, B. (2007). Aligning incentives by capping bonuses. In R. Antle, F. Gjesdal, & P. J. Liang (Eds.), Essays in accounting theory in honour of Joel S. Demski (pp. 169–182). New York: Springer. Balsam, S. (1998). Discretionary accounting choices and CEO compensation. Contemporary Accounting Research, 15(3), 229–252. Bradley, M., Desai, A., & Kim, E. H. (1988). Synergistic gains from corporate acquisitions and their division between the stockholders of target and acquiring firms. Journal of Financial Economics, 21(1), 3–40. Beatty, A., & Weber, J. (2006). Accounting discretion in fair value estimates: An examination of SFAS 142 goodwill impairments. Journal of Accounting Research, 44(2), 257–288. Benston, G. (1985). The self-serving management hypothesis – Some evidence. Journal of Accounting and Economics, 7(1–3), 67–84. Bodolica, V., & Spraggon, M. (2009). Merger and acquisition transactions and executive compensation: A review of the empirical evidence. Academy of Management Annals, 3(1), 109–181. Cheng, S. (2004). R&D expenditures and CEO compensation. The Accounting Review, 79(2), 305–328. Dechow, P. M., Huson, M. R., & Sloan, R. G. (1994). The effect of restructuring charges on executives’ cash compensation. The Accounting Review, 69(1), 138–156. Demski, J. S., Patell, J. M., & Wolfson, M. A. (1984). Decentralized choice of monitoring systems. The Accounting Review, 59(1), 16–34. Dye, R. A., & Verrecchia, R. E. (1995). Discretion vs. uniformity: Choices among GAAP. The Accounting Review, 70(3), 389–415. Ely, K. M. (1991). Interindustry differences in the relation between compensation and firm performance variables. Journal of Accounting Research, 29(1), 37–58. Evans, J. H., & Sridhar, S. S. (1996). Multiple control systems, accrual accounting, and earnings management. Journal of Accounting Research, 34(1), 45–65. Fama, E. F. (1980). Agency problems and the theory of the firm. Journal of Political Economy, 88(2), 288–307. Fields, T. D., Lys, T. Z., & Vincent, L. (2001). Empirical research on accounting choice. Journal of Accounting and Economics, 31(1–3), 255–307. Finkelstein, S. (1992). Power in top management teams: Dimensions, measurement, and validation. Academy of Management Journal, 35(3), 505–538. Gaver, J. J., & Gaver, K. M. (1998). The relation between nonrecurring accounting transactions and CEO cash compensation. The Accounting Review, 73(2), 235–253. Gaver, J. J., Gaver, K. M., & Austin, J. R. (1995). Additional evidence on bonus plans and income management. Journal of Accounting and Economics, 19(1), 3–28. Gjesdal, F. (1981). Accounting for stewardship. Journal of Accounting Research, 19(1), 208–231. Gordon, M. J. (1964). Postulates, principles and research in accounting. The Accounting Review, 39(2), 251–263. Hamberg, M., Paananen, M., & Novak, J. (2011). The adoption of IFRS 3: The effects of managerial discretion and stock market reactions. European Accounting Review, 20(2), 263–288. Hayn, C., & Hughes, P. J. (2006). Leading indicators of goodwill impairment. Journal of Accounting, Auditing and Finance, 21(3), 223–265. Hayward, M. L. A., & Hambrick, D. C. (1997). Explaining premiums paid for large acquisitions: Evidence of CEO hubris. Administrative Science Quarterly, 42(1), 103–127. Healy, P. M. (1985). The effect of bonus schemes on accounting decisions. Journal of Accounting and Economics, 7(1–3), 85–107. Healy, P. M., Kang, S. H., & Palepu, K. G. (1987). The effect of accounting procedure changes on CEOs’ cash salary and bonus compensation. Journal of Accounting and Economics, 9(1), 7–34. Holthausen, R. W., Larcker, D. F., & Sloan, R. G. (1995). Annual bonus schemes and the manipulation of earnings. Journal of Accounting and Economics, 19(1), 29–74. Holthausen, R. W., & Leftwich, R. W. (1983). The economic consequences of accounting choice – Implications of costly contracting and monitoring. Journal of Accounting and Economics, 5, 77–117. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305–360. Johnson, W. B., & Revsine, L. (1988). Financial reporting standards, agency costs, and shareholder intervention. Journal of Accounting Literature, 7, 95–124. Landsman, W. R. (2007). Is fair value accounting information relevant and reliable? Evidence from capital market research. Accounting and Business Research, 2007, Special Issue, 19–30. Lang, M., Smith Raedy, J., & Wilson, W. (2006). Earnings management and cross listing: Are reconciled earnings comparable to U.S. earnings? Journal of Accounting and Economics, 42(1–2), 255–283.

126

D. Detzen, H. Zülch / Journal of International Accounting, Auditing and Taxation 21 (2012) 106–126

Larcker, D. F., Richardson, S. A., & Tuna, I. (2007). Corporate governance, accounting outcomes, and organizational performance? The Accounting Review, 82(4), 963–1008. Leuz, C. (2010). Different approaches to corporate reporting regulation: How jurisdictions differ and why. Accounting and Business Research, 40(3), 229–256. Leuz, C., Nanda, D., & Wysocki, P. D. (2003). Earnings management and investor protection: An international comparison. Journal of Financial Economics, 69(3), 505–527. Li, Z., Shroff, P. K., Venkataraman, R., & Zhang, I. X. (2011). Causes and consequences of goodwill impairment losses. Review of Accounting Studies, 16(4), 745–778. Morck, R., Shleifer, A., & Vishny, R. W. (1990). Do managerial objectives drive bad acquisitions? Journal of Finance, 45(1), 31–48. Myers, S. C., & Majluf, N. S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 13(2), 187–221. Pavlik, E. L., Scott, T. W., & Tiessen, P. (1993). Executive compensation: Issues and research. Journal of Accounting Literature, 12, 131–189. Ramanna, K. (2008). The implications of unverifiable fair-value accounting: Evidence from the political economy of goodwill accounting. Journal of Accounting and Economics, 45(2–3), 253–281. Roll, R. (1986). The hubris hypothesis of corporate takeover. Journal of Business, 59(2), 197–216. Shalev, R., Zhang, I., & Zhang, Y. (2011). CEO compensation and fair value accounting: Evidence from purchase price allocation. Working Paper. Shaw, K. W., & Zhang, M. H. (2010). Is CEO cash compensation punished for poor firm performance? The Accounting Review, 85(3), 1065–1093. Watts, R. L. (2003). Conservatism in accounting–Part I: Explanations and implications. Accounting Horizons, 17(3), 207–221. Watts, R. L., & Zimmerman, J. L. (1986). Positive accounting theory. Englewood Cliffs: Prentice-Hall. Whittington, G. (1986). Financial accounting theory: An over-view. British Accounting Review, 18(2), 4–41. Zhang, I., & Zhang, Y. (2007). Accounting discretion and purchase price allocation after acquisitions. AAA 2007 Financial Accounting and Reporting Section (FARS) Meeting Paper, HKUST Business School Research Paper No. 07-04.