Predicting Goodwill Impairment by Acquisition Characteristics
By Ewout Bastiaannet June 2013
University of Groningen
Faculty of Economics and Business
MSc International Financial Management
Student number: 1629182
Supervisor:
Dr. Ing. N. Brunia
ABSTRACT
2 1. Introduction
In August 2011, HP announced the takeover of British IT company Autonomy for around $12 billion. A little over a year later, HP announced a goodwill impairment of approximately $8.8 billion relating to the same acquisition. The reason for the impairment was that HP had significantly overpaid for Autonomy1. On the announcement date, investors estimate the expected synergies arising on the merger or acquisition. Synergies are the value of the business combination after the deal minus the stand-alone values of the firms before the deal. If investors are right on average, weighted average announcement returns of the acquirer and target are the expected synergies. When the announcement return for acquirer is positive, it did not overpay for the synergies arising. However, when synergies exist and the returns for the acquirer are negative, the acquirer has overpaid for the synergies according to investors. Goodwill impairment is a consequence of overpayment for the target or unrealized synergies. In takeovers, often, but not necessarily, a part of the deal value is capitalized as goodwill by the acquirer. Goodwill is the difference between the purchase price and the estimated fair market value of the net assets acquired and will appear on the asset side of the balance sheet. It represents the expected synergies. Therefore, goodwill impairment is a sign of a value destroying acquisition. However, not every value destroying acquisition necessarily leads to goodwill impairment because post acquisition events can influence future cash flows in a positive way. For example positive net present value investments.
The phenomenon of goodwill impairment is mostly studied in relation to the financial reporting system. Often, different accounting regimes are compared. Another widely investigated topic is the effect of the impairment announcement on the stock return. Bartov et al. (1998) and Hirschey and Richardson (2002) find that announcements of goodwill impairments result in lower stock prices. However, the abnormal returns are relatively small compared to the amount of goodwill impaired. The authors suggest that the stock price declines before goodwill impairment announcement shows that investors anticipate goodwill impairment. Other authors focus more on the prediction of goodwill impairment. Hayn and Hughes (2005) find that characteristics of the acquisition are the most powerful predictors of goodwill impairment compared to post acquisition performance. The acquisition characteristics are indicators of overpayment. Also Li et al. (2011) uses indicators of overpayment to explain goodwill impairment.
1
3
This paper will research if goodwill impairment can be predicted by characteristics of the acquisition. Just like Li et al. (2011) and Hugh and Haynes (2005) do, indicators of overpayment are used. This paper adds to literature by studying European acquirers. Previous studies study samples of US acquirers. Since Europe has a different governance system than the US, it is interesting to see if the results differ. Corporate governance can have a tempering effect on agency problems and, therefore, value destroying acquisitions. Furthermore, we include a dummy variable for cross-border deals to see if cross-border deals have a larger probability to lead to goodwill impairment.
The dataset composed for this research includes 158 acquisitions in the time period 2002-2009. The acquirers are listed firms from the U.K., France, Germany, Spain, Italy and The Netherlands. The acquisitions are traced through time to see if they have led to a goodwill impairment. In 36 cases it did. A probit/logit regression tests whether the characteristics of the original deal can predict the future impairments. The characteristics included in the regression are proxies for overpayment by the acquirer and unrealized synergies and are known on the announcement date of the acquisition. Not only tests this paper for prediction of unrealized synergies, it will also empirically test the prediction of goodwill impairment for European companies for the first time.
The results show that indicators for overpayment in general do not have an influence on the probability of goodwill impairment. Only the fact whether an acquisition is domestic or cross-border can predict goodwill impairment; cross-cross-border acquisitions have a larger probability to result in goodwill impairment. The other indicators for overpayment show no significant results. This is remarkable since past studies (in the US.) did find support for the predictive value of overpayment indicators.
The outcome of this study holds relevant information for managers. First of all, the finding that cross-border acquisitions have a higher probability of goodwill impairment can help managers in making decisions regarding these deals. With this information they know that they should be careful in assessing the value of the target and the expected synergies. Furthermore, although investors might expect a negative NPV on acquisition date, this does not necessarily lead to goodwill impairment.
4 2. Literature review
This section will elaborate on the literature relevant for this paper and explain the theoretical background. We start off with a description of goodwill and relevant accounting rules for business combinations.
2.1 Goodwill
IFRS requires that assets are reported against fair value. Goodwill is calculated as the difference between the purchase price and the fair value of the net acquired assets. The fact that goodwill is calculated this way makes it a residual post of the acquisition. Assuming the price paid is the correct value for the target plus synergies expected, goodwill arising on acquisition represents the expected synergies. However, if the cost of acquisition are higher than the value of the target plus expected synergies, the acquirer overpaid for the target. This overpayment is recorded on the balance sheets in the form of goodwill. Before IFRS, goodwill was amortized over a fixed number of years. In IFRS, goodwill is tested for impairment annually. The test consist of an assessment to determine if it is “more likely than not” that the fair value of goodwill exceeds the carrying value. This is usually done per segment or Cash-Generating Unit (CGU). If there is a large probability that the carrying amount cannot be retrieved and is too high, there will be tested for impairment. The fair value of goodwill is determined using the discounted cash flow (DCF) method. The fair value is also called the recoverable amount. Besides the annual test, companies also have to test for impairment if a triggering event has occurred. A triggering event is an event that significantly might change the future cash flows arising from the asset. Examples are lawsuits against the company, or a shift in market demand. Four indicators of goodwill impairment given by the Financial Accounting Standard Board are (1) payment of a significant premium, (2) presence of multiple bidders or an auction-like situation, (3) a significant amount of goodwill relative to the acquisition price, and (4) use of the acquiring firm‟s stock as the primary mode of consideration (Hayn and Hughes, 2005). We will discuss why these factors are indicators of goodwill impairment below.
2.2 Motives for acquisitions
5
acquiring and the target firm. As long as the value of the target before acquisition plus the synergies are lower than the purchase price2, the NPV will be positive for the acquirer. The target will have a positive NPV when the purchase price per share is higher than the current price per share of the firm. It is assumed that investors are rational and value shares correctly on average. The positive NPV of the acquirer and target will result in positive announcement returns for the respective firms. In this scenario, on average, synergies will arise and no overpayment takes place. There will be no overpayment, because that would lead to a negative NPV for the acquirer and thus would the acquiring firm not engage in the deal. Although an ideal scenario is sketched above, it does not mean that every acquisition will have a positive NPV automatically. On average the NPV will be positive, however, deals with a negative NPV can also occur. Due to, for example, information asymmetry, managers can make estimation errors which lead to negative NPV‟s. Value destruction of mergers and acquisition for the acquiring has two sources; overpayment and unrealized synergies. Overpayment is known at the announcement date because the purchase price, expected synergies and the value of the target (before the announcement) are all known by then. Unrealized synergies can‟t be known at the announcement date. We can conclude that negative announcement returns for the acquiring firm signals overpayment and should therefore have a higher probability of goodwill impairment, assumed that investors are right on average.
2.2.1 Hubris hypothesis and the winner’s curse
In Roll‟s Hubris Hypothesis (Roll, 1986) the prediction of total gains (bidder and target together) on mergers and acquisition is non-positive. Takeovers are merely a product of misevaluations of the target firm‟s value by managers.Roll (1986) assumes efficient markets. Also labor markets and product markets are efficient. Therefore, the stock of a company is correctly valued and the merger or acquisition does not offer any probabilities for synergies3. Managers who identify a potential target will only make a bid if the current market price is lower than the value calculated by the acquiring management. However, in the hypothesis the market is efficient and the current market price holds all available information. The management of the bidder should acknowledge the error made and revise the valuation
2
Normally there will also be transaction cost in mergers and acquisitions. We will leave them out of the discussion in this paper by assuming they are zero.
3
6
downward. The negligence of managers to do this, because they think their valuation is superior to the valuation of the market or because they do think that synergies exist, is classified as hubris behavior. Also when management does not consider the possibility of making a measurement or estimation error, this is considered as hubris behavior by Roll (1986). In other words, managers overestimate their own abilities, either by expecting synergies although they don‟t exist or by thinking their valuation is superior to the market. Valuations below the current market price of the target are not observed because the market price will function as a lower bound of the selling price for the target management. So the downward valuation errors made by the acquirer will not lead to a bid. Berkovitch and Narayanan (1993) state that mergers and acquisition motivated by the hubris hypothesis will lead to a transfer of wealth from the acquiring firm to the target firm. Therefore the announcement returns of the acquiring and target firm are negatively correlated. The amount of wealth transferred is the overpayment. Even if synergies do exist for some business combinations, the hubris hypothesis can be applied. In this case, the synergies as well as the current value of the target firm can be overestimated by hubris behavior. This will still lead to positive acquirer announcement return and positive target announcement returns. Berkovitch and Narayanan (1993) find a negative correlation in returns between target and acquirer in deals where the total gains are positive. This implies that there are synergies but the acquirer overpays for the target. Berkovitch and Narayanan (1993) devote this overpayment to hubris behavior. Hayward and Hambrick (1997) find evidence for hubris by specifically measuring hubris with four indicators4. They find that the indictors are associated with negative announcement returns for the acquiring firm and high deal premiums. If we assume that hubris behavior and synergies do both exist at the same time, we can say that synergies are likely to be overestimated by acquiring managers. Since premiums paid over the share price of the acquisition represent (part of) the synergies, it follows that deals with high premiums are more likely to be overpaid.
Roll‟s theory (1986) is an application of the winners curse theory on mergers and acquisitions (Gilberto and Varaiya, 1989). The winner‟s curse theory is originally based on sealed bid auction (Gilberto and Varaiya, 1989). The difference is that in sealed bid auctions it is assumed that the price of the item (the target in the case of mergers and acquisitions) is unknown ex ante, whereas in the hubris hypothesis the item already has an established market
4
7
price. This market price is considered the first bid by Roll (1986) but when the market wins the auction (no merger or acquisition) we do not consider it an event.
2.2.2 Behavioral theories
8
with relatively lower positive errors (less overvalued). Rhodes-Kropf et al. do not research the returns of stock deals. Gu and Lev (2011) do, and find positive relationship between overpayment of synergies and overvaluation of the acquiring firm as well. Fu, et al. (2013) also find these relationships. They even find proof for negative synergies in the post-acquisition period next to significant premiums paid. The performance of overvalued firms who engage in mergers and acquisitions is compared to firms who do not engage in mergers or acquisitions. The firms who do merge or acquire, significantly underperform the firms who do not engage in mergers and acquisition.
In contrast, Savor and Lu (2009) claim that the theory described by Schleifer and Vishny (2003) is indeed value creating for acquirers. In addition, it is noted that exactly the highly valued firms have an incentive to engage in mergers and acquisitions and, therefore, it is only logical that these deals lead to the highest negative abnormal returns. Savor and Lu (2009) conclude that unrealized acquirer-target combinations in which the acquirer is overvalued, would have earned higher returns than the acquirer did by itself, even without any synergies.
2.2.3 Agency theory
9
target firm is the result. Grinstein and Hribar (2004) find clear evidence of agency problems in M&A deals. They find evidence for the fact that the CEO compensation for completed M&A deals (mostly cash bonuses) is not correlated with the return for shareholders. In their study, they focus on CEO power to measure agency problems. Deals made by acquiring firms with higher CEO power have lower abnormal announcement returns. The abnormal returns of the deals made by CEO‟s with high power in the first two days after the announcement, are approximately three times as low as other deals. In other words, CEO‟s with high power and more potential agency problems, tend to overpay more than less powerful CEO‟s. Another finding by Grinstein and Hribar (2004), is that most of the CEO‟s are not rewarded by the value creation of a merger or acquisition. Rather, they are rewarded for extra effort, increasing the size/revenues of the firm and leadership shown during the deal among others. In the survey they conduct, only 22 out of 64 firms mention value enhancement of the acquiring firm as a reason to reward their CEO. With these results, Grinstein and Hribar (2004) show that managers have personal incentives to engage in mergers and acquisitions, no matter the returns for shareholders. We should note that agency problems do not necessarily lead to value destructive deals. However, shareholder value maximization is not the primary goal of the managers and this enhances the probability of value destruction.
2.3 Deal characteristics
This research will test whether certain acquisition characteristics can lead to a higher probability of goodwill impairments. Section 2.1 concludes that a negative NPV deal might lead to goodwill impairment. We also established that overpayment and unrealized synergies are the causes of goodwill impairment. The overpayment is estimated by the market on the announcement date, unrealized synergies will only show after a longer period. Since the overpayment for the target and/or synergies is directly observed by the announcement returns of the acquirer, our first hypothesis is:
H1: Acquisitions in which the announcement returns for the acquirer are more negative are more likely to result in goodwill impairment
post-10
11
impairment. It will be tested whether higher premiums lead to a larger probability of goodwill impairment:
H2: Acquisitions with a higher premium paid are more likely to result in goodwill impairment
The hubris hypothesis (Roll, 1986) is based upon the theory of the winner‟s curse and the winner‟s curse is based upon auction theory. Because multiple bidders will lead to hubris behavior the third hypothesis is:
H3: Acquisitions made in the presence of multiple bidders or auction-like situations are more likely to result in goodwill impairment
The market for corporate takeover is competitive (Ruback, 1983). This means that there is competition between the bidders for a target firm. The Winning curse hypothesis implicates that the winning bidder is the one that overvalues the target the most (Gilberto and Varaiya, 1989). The winners are „cursed‟ because they did not recognize this possibility and have now overpaid for the target (Gilberto and Varaiyna, 1989). This can happen both in an auction-like situation or when multiple firms are bidding for the target. Gilberto and Varaiya (1989) found empirical results supporting this hypothesis. Both Hayn and Hughes (2005) and Li et al. (2011) do not find significant results for a larger probability of goodwill impairment for deals with multiple bidders. Perhaps the finding of Hayn and Hughes (2005) is influenced by the low percentage of deals with multiple bidders (12.1%). The percentage of deals with multiple bidders is not presented in Li et al. (2011).
The behavioral theories suggest that mergers and acquisitions are motivated by overvaluation of the target. Section 2.2.2 discusses how overvaluation can lead to high premiums and possible overpayment. While studying large loss deals, Moeller et al. (2005) finds that high Tobin‟s Q-ratios5
can partially explain the variance in the abnormal returns of acquirers. The relationship between the returns and Tobin‟s Q ratio is negative. However, as Savor and Lu (2009) suggest, this can be the correction of the overvaluation. Since the highest overvalued firms are likely to see the largest decline in stock price to reach their fundamental valuation,
12
their returns are likely to be the most negative. The firms that are overvalued the most, have the most incentive to engage in mergers and acquisitions, whixch could explain the findings of Moeller et al. (2005). If this is true, it is not the merger or acquisition per se that destroys the value. However, since the acquisition is accounted for on the day that the stock is still overvalued, and thus likely to be overpaid, the probability on goodwill impairment rises significantly when overvalued stock is used as method of payment. In the behavioral theory the acquisition needs to be paid by stock of the acquirer. Jensen (2005), however, argues that a high market-to-book ratio (his measurement of overvaluation) leads to agency problems. Jensen (2005) argues that managers try to let the market believe that the high valuations are correct. They try to do this, among other things, by acquiring firms to boost revenues and size. Grinstein and Hribar (2004) find results that are similar to this. As discussed, agency problems can lead to overpayment. Han et al. (1998) find results that show that overvalued firms tend to overpay, regardless of the method of payment. In their study, Han et al. (1998) test both the overpayment by the acquirer and the effect of the convergence to the fundamental value separately. Therefore, the following hypothesis is tested:
H4: Acquisitions in which the acquirer is more overvalued are more likely to result in goodwill impairment
H5: Acquisitions in which stock is the sole method of payment are more likely to result in goodwill impairment than cash transactions
13
have a larger probability to impair goodwill after acquisition, regardless the method of payment.
Another indicator of overpayment is the relatedness of the industry of the target to the industry of the acquirer. Morck et al. (1990) find that unrelated (diversifying) acquisitions lead to higher premiums because of agency problems. Three reasons are given: (1) Managers want to diversify for their own benefit when their human capital is not diversified and use the merger or acquisition for this, (2) When wealth maximization dictates shrinkage or liquidation, managers may try to enter new lines of business to assure survival and continuity of the firm, and (3) when poor performance of the firm threatens the position of the manager, he might try to enter new lines of business in which he might be better. Li et al (2011) find highly significant results for the fact that unrelated mergers and acquisitions have a higher probability to lead to goodwill impairment. This leads to the sixth hypothesis:
H6: Unrelated acquisitions are more likely to result in goodwill impairment than related acquisitions
Furthermore, Hayn and Hughes (2005) find highly significant results for the positive relationship between the goodwill capitalized on the acquisition and the probability of goodwill impairment. Hayn and Hughes (2005) suggest that, in the case of a considerable allocation of the purchase price to goodwill, goodwill is assessed more subjectively than other assets. They see this as an indicator for possible overpayment of the target firm. Hence, the goodwill has a higher probability to be impaired.
H7: Acquisitions in which a large part of the purchase price is allocated to goodwill are more likely to result in goodwill impairment
14
overpayment to hubris behavior of acquiring managers. The cross-border factor has never been tested in relation to goodwill impairment by the knowledge of the author. The last hypothesis is:
H8: Cross-border acquisitions are more likely to result in goodwill impairment than related acquisitions
Table 1 provides an overview of the hypotheses.
Table 1.
Overview of the hypotheses
# Hypothesis1 Acquisitions in which the announcement returns for the acquirer are more negative are
more likely to result in goodwill impairment
2 Acquisitions with a higher premium paid are more likely to result in goodwill impairment 3 Acquisitions made in the presence of multiple bidders or auction-like situations are more
likely to result in goodwill impairment
4 Acquisitions in which the acquirer is more overvalued are more likely to result in goodwill impairment
5 Acquisitions in which stock is the sole method of payment are more likely to result in goodwill impairment than cash transactions
6 Unrelated acquisitions are more likely to result in goodwill impairment than related acquisitions 7 Acquisitions in which a large part of the purchase price is allocated to goodwill are more likely to
result in goodwill impairment
8 Cross-border acquisitions are more likely to result in goodwill impairment than related acquisitions
15 3. Data and Methodology
Previous studies linking indicators of overpayment and goodwill impairment are only tested in the US. Therefore, this study focuses on European acquirers. As we pointed out at the end of section 2, corporate governance can have influence on agency problems and hubris behavior by managers. Europe is known for having a stronger corporate governance system in use than the US. The so-called two-tier board system creates an extra supervision for executives. Therefore, there might be less incentive to overpay in mergers and acquisitions. The dataset composed for this study consists of completed acquisitions where the acquiring firm is from the U.K., France, Germany, Spain, Italy and The Netherlands. These countries were selected based on GDP; these six countries have the largest GDP of the EU in 2012. Initially UK acquirers were left out of the sample because they have the same corporate governance system as the US. However, this sample selection yields too few deals to research. Therefore, it is decided to include the UK in the sample as well, notwithstanding the corporate governance system.
We consider acquisitions that are completed between 2002-2009 and both the acquiring and target firms had to be listed. In 2002, the EU agreed that IFRS would apply for listed companies. IFRS is required for listed companies in the EU since 2005. Before 2005, many firms already adapted IFRS. Goodwill impairment was also used before IFRS by firms who used the purchase method of accounting for business combinations. The database of Zephyr counts 468 deals that meet the requirements. From this selection only deals which acquired a stake in the target of at least 90% were considered. Also the deal value had to be at least $10 million. This left 210 acquisitions. These acquisitions are followed through time to see if the goodwill arising on the acquisition is impaired. The amount of goodwill that arises on the deal was gathered from annual reports of the acquirer. It is important to know the amount of goodwill on a specific acquisition and not combined goodwill of multiple acquisitions. The transaction needs to be accounted for using the purchase method instead of the pooling of interests6. The acquisitions accounted for by the pooling of interest method were neglected. If it can‟t be traced if the acquisition gives rise to goodwill, or how much goodwill, the deal is also removed from the sample. The search strategy is reported in Table 2.
6
16 Table 2. Search strategy
Criteria Number of firms left
Initial sample (Geographic, time selection and listed) 268
Acquired stake >= 90% 224
Minimum deal value $ 10 million 210
Traceable Goodwill 158
In 158 deals the goodwill arising on acquisition and any impairment could be traced. From these 158 deals, of 36 acquisitions the goodwill is impaired (22.8%). Because the sample is truncated we can only say that the goodwill is not impaired yet in the other cases. We see a peek of impairments in the years 2008 and 2009 in table 3. This is quite remarkable. Of course in late 2007 the mortgage crisis occurred in the US and asset prices fell. As a result, the economic outlook changed for many firms. Such a change in economic outlook can change the fair value of goodwill and leads to impairments. Most acquisitions were done by UK firms, namely 92 (58%). Also the targets were mostly from the UK: 72 (46%). For more details about the countries of targets and acquirers, see table 4. The average time from acquisition to goodwill impairment was three years. Only once, goodwill is impaired in the same year as the acquisition. The longest time between an acquisition and the goodwill impairment was six years. Figure 1 shows the distribution of time from acquisition to goodwill impairment.
Figure 1.
The number of years between the acquisition completion and subsequent goodwill impairment
The X-axis displays the number of years, Y-axis is number of firms that impaired after X years. Only acquisition that resulted in goodwill impairment are included (N=36).
17 Table 3.
Number of acquisitions and goodwill impairments per year
This table shows the number of acquisitions per year and if any goodwill of those acquisitions is impaired after the acquisition. Also the number of firms of our sample that impaired goodwill is shown per year. Only the first goodwill impairment per acquisition is observed. So every observation of goodwill observation is from a unique business combination.
Year Number of acquisitions Number of acquisitions resulting in goodwill impairment Goodwill impairments Percentage goodwill impairments of total goodwill impairments in sample 2002 5 0 (0.00%) 0 0 2003 8 3 (37.5%) 0 0 2004 10 1 (10.0%) 0 0 2005 23 3 (13.0%) 1 2.78 2006 27 9 (33.3%) 1 2.78 2007 45 15 (33.3%) 0 0 2008 24 3 (12.5%) 10 27.78 2009 16 2 (12.5%) 9 25.00 2010 - - 3 8.33 2011 - - 7 19.44 2012 - - 5 13.89 Total 158 36 (22.8%) 36 100
18 Table 4.
Number of acquisitions and subsequent goodwill impairment per country The number of acquirers and targets are displayed per country. Furthermore, the table shows how many deals per country resulted in goodwill impairment. The percentages are the deals resulting in goodwill impairment of total deals per country
Acquirers Targets
Country Deals Deals resulting in
Goodwill Impairment Deals
Deals resulting in Goodwill Impairment UK 92 (58.2%) 13 (14.1%) 72 (45.6%) 13 (18.1%) France 21 (13.3%) 7 (33.3%) 7 (4.4%) 1 (14.29%) Germany 15 (9.5%) 5 (33.3%) 1 (0.6%) 0 The Netherlands 12 (7.6%) 3 (25%) 4 (2.5%) 0 Spain 9 (5.7%) 3 (33.3%) 5 (4%) 1 (20%) Italy 9 (5.7%) 5 (55.6%) 3 (1.9%) 1 (3.3%) Australia - 3 (1.9%) 0 Canada - 4 (2.5%) 0 Switzerland - 1 (0.6%) 1 (100%) Denmark - 2 (1.3%) 1 (50%) Israel - 1 (0.6%) 1 (100%) India - 1 (0.6%) 0 Jordan - 1 (0.6%) 0 Panama - 1 (0.6%) 0 US - 51 (32.2%) 16 (31.4%) Virgin Islands (British) - 1 (1.9%) 1 (100%) Total 158 (100%) 36 (22.8%) 158(100%) 36 (22.8%)
The bid premium is the percentage of price paid per share over the closing price of the last day before the announcement. These premiums are collected from the Zephyr database. To determine if there were multiple bidders to acquire the target or an auction-like situation was present the Zephyr deal rationale is consulted. If the deal rationale did not clearly confirm that the acquirer was the only bidder additional information was analyzed on the internet. Only trustable sources like Bloomberg and Reuters are considered. In the event of multiple bidders or an auction-like situation the variable has value 1. If this is not the case, the value is zero.
19
market-to-book ratio is collected from the Datastream database (field code: MTBV). It is calculated by dividing the market capitalization of the outstanding stock by the book value of the common equity of the firm. Both the nominator and denominator are based on the closing rate of the day before. Moeller, et al. (2005) find that acquirers with low book-to-market ratios7 make the largest losses on acquisitions.
Method of payment is a dummy variable that takes value 1 if the deal is paid with 100% stock and 0 otherwise. The data on the method of payment is from the Zephyr database.
The relatedness of the firm variable is a dummy variable as well. It has value 1 if the target and bidder are unrelated and 0 otherwise. Related is in this case defined as having the same four digit primary SIC-code from the Zephyr database. This approach is the same as Gu and Lev (2011).
From the annual report of the acquirer the amount of goodwill arising on acquisition was collected. This goodwill is stated as a percentage of the deal value reported in Zephyr. The deal value in Zephyr is the consideration paid for the actual stake acquired.
Cross-border is a dummy variable to determine if the target is foreign or domestic to the acquirer. It has value 0 if the target is from the same country as the acquirer in the Zephyr database. Cross-border acquisitions are coded 1.
Table 5 presents an overview of all variables and their description.
In table 6 the descriptive statistics are presented. We can see that only the variable Cross-border has significant different distributions when we compare the impairment group with the non-impairment group. As we expected, the sub-sample that did impair goodwill are cross-border deals more often. This means that goodwill arising on cross-cross-border deals is significantly impaired more often than domestic deals. Furthermore, the samples are only significantly different concerning the amount of goodwill that arises on the acquisition. Acquirers that impair goodwill record a significant higher level of goodwill relative to the deal value than the acquirers who did not impair goodwill. Based on the statistics we can conclude that there is no significant difference between acquisitions of which the goodwill is impaired and acquisitions of which the goodwill is not impaired regarding the announcement returns of the acquirer, the relatedness, whether there were multiple bidders, method of payment, premium paid, and market-to-book ratios of the acquirer.
7
20 Table 5.
Overview of variables and sources
Variable Description Source
Impairment Dummy variable; 1 if goodwill on the acquisition is impaired and 0 otherwise
Annual reports, Datastream Announcement returns Returns on the announcement day of the acquisition of the acquirer Datastream Relatedness Dummy variable; 1 if target and acquirer operate in different 4-digit
SIC-code
Zephyr Cross-border Dummy variable; 1 if target is from different country than acquirer Zephyr Overvaluation Market-to-book ratio of the target on the last trading day before
announcement
Datastream Premium Bid premium over the share price of the target on the last trading
day before the announcement of the deal
Zephyr Goodwill Percentage of the deal value that is capitalized as goodwill by the
acquirer
Annual report, Zephyr Multiple Bidders Dummy variable; 1 if other firms besides acquirer made a bid for
the target or if there was an auction-like situation
Zephyr, news sources Method of payment Dummy variable; 1 if method of payment is only stock, 0 otherwise Zephyr
The hypotheses will be tested with a regression. The model used to predict goodwill impairment based on the variables mentioned is shown in equation (1).
(1)
To test the hypotheses we will use a Probit/Logit regression. Both types of regression have very similar outcomes (Brooks, 2008). Therefore we only use the procedure and outcome of the Logit regression. Because the dependent variable is a dummy (0 for no impairment, 1 for impairment) the Linear Probability Model (LPM) is not the best estimator because it is has no boundaries (Brooks, 2008). However, the probability on goodwill impairment can‟t be higher than 1 or lower than 0. Therefore equation (1) is transformed to a logistic function. The logistic function F of random variable, γ, would be:
(2)
21 Table 6
Descriptive statistics
Left you will find the variables used to test the hypotheses. The variables are split up in two sub-samples. A group that did impair goodwill, and the other group did not impair goodwill. The groups are called subsample 1 and 0 respectively. The row both displays the descriptive statistics for the total sample. Difference calculates the difference between the two sub-samples and displays the test statistics under T-test and Mann-W-U. The T-T-test and Mann-Whitney-U T-test are used to calculate if the difference between the two sub-samples is significant. ***, **, and * means significant at the 1%, 5%, and 10% level for a 2-tailed test respectively.
Subsample Mean Median
22
Where is the exponential under the logit approach. We can now replace F with P to get equation (3).
(3)
Where Pᵢ is again the probability that IMPAIRMENT = 1. In this way, the probability of goodwill impairment will never reach 1 and never fall below 0. Although the formula is not a linear one, the original causality is modeled as a linear function, namely (1). To test for a non-linear effect of the non-dummy variables the sample is cut in five parts. The five parts are ranging from low to high values for the variables. For each of the five „sub-samples‟ the average score on goodwill impairment is calculated (ranging from 0 to 1). Plots of the averages are presented in figure 2.
Figure 2.
Plots of the average score on goodwill impairment for the five subsamples.
From figure 2 we can conclude that there is no reason to assume non-linearity. We can thus continue with the Logit model from (3). We will first regress each variable individually to see if they have influence on the probability. After this, we regress the full-model including all of the variables. In this way we can see if variables have influence on each other. When the significance or coefficient changes dramatically when other variables are added, there is a high chance on multicollinearity. Since all of our variables are indicators for overpayment, it
23
is not strange if they correlate. Our first test of collinearity is presented in table 7. In this table we see that the highest correlation is between the market-to-book ratio (0.32) and the method of payment. Higher market-to-book ratios tend to pay in shares more often. The correlations are not extremely high and therefore, we do not assume multicollinearity.
Table 7.
Correlations between independent variables.
24 4. Results
The regression results for predicting goodwill impairment are presented in table 8. It shows that cross-border acquisitions have a higher probability to be impaired. This result is the only significant result from the regression. In the full-model we see that also the market-to-book value has a significant test statistic but this result is a large difference from the single variable regression. We therefore conclude that the significance is caused by the addition of other variables due to multicollinearity. From the correlation matrix we can see that the market-to-book ratio has relatively high correlations with the cross-border variable and the method of payment variable.
The positive relationship of cross-border deals with goodwill impairment is as in the hypothesis. Therefore we accept Hypothesis 8. Regarding the other hypotheses, five of them show a relationship opposite as expected, but the results are not significant. The other two hypotheses do show the expected direction. However, they are not significant. Therefore all hypotheses except for hypothesis 8 are rejected.
When we insert the coefficients of the preferred model in equation (3) we get a chance of 15% on goodwill impairment for domestic acquisitions. Cross-border deals double the chance on goodwill impairment: 31%.
These findings are different from the findings of Hayn and Hughes (2005), Gu and Lev (2011) and Li et al (2011). Hayn and Hughes (2005) find that the premium paid, goodwill arising on acquisition, stock as method of payment, announcement returns of the acquirer all enlarge the probability of goodwill impairment. As this study they also don‟t find significant results for the presence of multiple bidders.
Gu and Lev (2011) find that overvalued stock and using this stock as method of payment increases the probability of goodwill impairment. Also Li et al. (2011) find this in their study, although they both measure the overvaluation different than this study. Lit et al.‟s (2011) way is most similar; they also use the market-to-book ratio, but they use a dummy variable for the firms with scores in the highest quintile. Furthermore, Li et al (2011) find comparable results as Hayn and Hughes (2005); the premium paid, method of payment, and goodwill arising on acquisition all have a significant influence on the probability of goodwill impairment. Next to this, Li et al. (2011) also find that unrelated acquisitions have a higher chance on impairment than related acquisitions.
25
on mergers and acquisition by US acquirers. As mentioned in previous sections, Europe has a different corporate governance system than the US. Gu and Lev (2011) find that effective corporate governance tempers the incentives of mangers to overpay for the target.
The timing of the studies is also different. Hayn and Hughes use a sample from the late 80‟s and 90‟s whereas Gu and Lev (2011) and Li et al. (2011) use a sample of 1990-2006 and 1996-2006 respectively. After 2006 the acquisitions of the samples are no longer tracked and might have been impaired afterwards. The economic events in late 2007 and after are thus not influencing the samples of these three studies. In this sample, most impairments are done after 2007. As we discussed in section 2, economic deterioration or improvement of events might cause expected goodwill impairments to not be impaired and vice versa.
26
Table 8. Regression Results
This table presents the results of the regression. The Variables are both regressed individually as well as all together in the full model. On the right side, the preferred model is presented. This model only consists of the variable Cross-border. The numbers shown are the coefficient with the T-statistic below. The significance of the T-statistic is
shown with ***,**, and * for significance at the 1%, 5%, and 10% level respectively
Variables only Full model
Preferred model Variable Announcement returns Premium Multiple bidders Overvaluation Method of payment Relatedness Goodwill Cross-border All
27 5. Conclusion
This paper analyzes if indicators of overpayment have an influence on the probability of goodwill impairment. Here for, acquisition characteristics were used as indicators of overpayment. Goodwill impairment can occur due to overpayment of the target by the acquirer or unrealized synergies. It occurs when the carrying value of goodwill exceeds the fair value of goodwill. Investors have an expectation of the value of synergies at the announcement date. When the weighted average returns of the acquirer and target are positive on the announcement date, investors believe that synergies exist. When the announcement returns for the acquirer are also positive, the acquirer did not pay too much for the synergies according to investors. When the acquirer has negative announcement returns, it overpaid for the expected synergies. But the announcement returns are just an expectation of the success of the merger or acquisition. Unexpected events might occur and have an influence on future cash flows, which can either be positive or negative. Negative events might lead to impairment of goodwill which was not expected by investors on the announcement date. On the other hand, positive events or economic situations might prevent the fair value of goodwill to become lower than the carrying value. As long as the NPV of the merger or acquisition is positive for acquirers, goodwill does not need to be impaired. In this study, results show that the expected NPV by investors on the announcement date, measured by the announcement returns of the acquirer, has no influence on the probability of goodwill impairment.
This paper hypothesized that seven other variables, known at acquisition date, have an influence on the probability of goodwill impairment. The variables are the relatedness of the industries of the target and the acquirer, if it is a cross-border or domestic target, the valuation of the acquirer, the premium paid, the amount of goodwill capitalized on acquisition, whether multiple bidders or an auction-like process was present, and if stock was the method of payment.
A sample of deals was assembled, of which the acquirers were from several European countries. The deals were completed between 2002 and 2009. The final sample consists of 158 deals of which 36 resulted in goodwill impairment.
28
integration of the target into the acquiring firm might be difficult. Therefore, the synergies expected at forehand might not be realized.
The other variables do not have any predictive power. This is not in line with previous studies who do find predictive power. Especially proxies for overpayment tend to be good predictors. Firms that impaired goodwill, did pay a higher premium on average, however, this is not significantly higher than firms who did not impaired goodwill. Although a positive relationship is expected, firms that impaired goodwill have a significantly lower valuation at the date of acquisition than firms who did not impair goodwill. However, the valuation of the firm does not have any predictive power in the regression.
The R-squared of the preferred model (only containing the cross-border variable) is low. Therefore, there must be other variables that can explain the phenomenon. Since acquisition characteristics are weak predictors, we argue that post acquisition events must be better able to explain goodwill impairment.
For future research, we suggest to investigate possible events, such as the mortgage crisis in late 2007 which led to economic deterioration, and their influence on goodwill impairments. Such events can have a strong effect on future cash flows and, therefore, the fair value of goodwill. We also suggest to investigate the effect of corporate governance on goodwill impairments, since this study finds different results than similar previous studies. The main difference between these studies is the sample. Where this study focuses on acquirers from Europe, previous studies focused solely on acquirers from the US. Furthermore, the mergers and acquisition were completed in a different time period and are not observed after 2006. This study does observe the deals after 2006. In fact, most goodwill impairments are done after 2007. We suggest that this is the main reason for the different findings.
Managers seem to act irrational in this study. Although the weighted average returns of acquirer and target are positive, the average return for the acquirer is negative (-0,58%) . This means that on the announcement date, investors expect a negative NPV for the acquisition. For investors and managers this paper implies that, although the acquisition characteristics studied often lead to negative announcement returns, they do not automatically lead to goodwill impairment.
29 References
Amihud, Y., and Lev, B., 1981, Risk Reduction as a Managerial Motive for Conglomerate
Mergers. The Bell Journal of Economics 12: 605-617.
Bartov, E., Lindahl, F., and Ricks W., 1998, Stock price behavior around
announcements of write-offs. Review of Accounting Studies 3: 327-346.
Berger, P., Ofek, E., 1995, Diversification’s effect on firm value. Journal of Financial Economics 37, 39–66.
Berkovitch E., Narayanan M., 1993, Motives for takeovers: an empirical
Investigation. The Journal of Financial and Quantitative Analysis 28: 347-362.
Brooks, C., 2008, “Introductory econometrics for finance”, Book, Cambridge university Press.
Bruner, R., 2004, “Applied mergers and acquisitions”. New York, NY: John Wiley & Sons.
Choi, W., Kwon, S., and Lobo, G., 2000, Market Valuation of Intangible Assets. Journal of Business Research 49: 35-45.
Conn, R., Cosh, A., Guest, P., and Hughes, A., 2005, The Impact on UK Acquirers of
Domestic, Cross-border, Public and Private Acquisitions. Journal of Business Finance &
Accounting 32: 815–870.
Denis, D., Denis, D., and Yost, K., 2002, Global Diversification, Industrial Diversification,
and Firm Value. The Journal of Finance 57: 1951–1979.
Giliberto S. M., and N. P. Varaiya, 1989, The winner's curse and bidder competition in
30
Grinstein, Y., Hribar, P., 2004. CEO Compensation and Incentives: Evidence from M&A
Bonuses. Journal of Financial Economics 73: 119-143.
Gu, F., and Lev, B., 2011. Overpriced shares, ill-advised acquisitions, and goodwill impairment. The Accounting Review 86: 1995-2022.
Han, K., Suk, D., and Sung, H., 1998. The evidence of bidders‟ overpayment in takeovers: the valuation ratios approach. The financial Review 33: 55-68
Hayn, C., and Hughes, P., 2006. Leading indicators of goodwill impairment. Journal of
Accounting, Auditing & Finance 21: 223-265.
Hayward, M.L.A., Hambrick, D.C., 1997. Explaining the premiums paid for large acquisitions: evidence of CEO hubris. Administrative Science Quarterly 42: 103-127.
Healy, P., Krishna, P., and Ruback, R., 1992. Does Corporate performance increase after mergers? Journal of Financial Economics 31 131-175.
Henning, S., Lewis, B., Wayne, S., 2000. Valuation of the components of purchased goodwill.
Journal of Accounting Research 38: 375-386.
Hirschey, M., Richardson, V.J., 2002. Information content of accounting goodwill Numbers. Journal of Accounting and Public Policy 21: 173-191.
Hofstede, G., 1991. “Cultures and organizations: Software of the mind” London: McGraw-Hill.
Jensen, M., 1986. Agency cost of free cash flow, corporate finance and takeovers. American
Economic Review Papers and Proceedings 76: 323-329.
Jensen, M., 2005. Agency cost of overvalued bidders. Financial Management 34: 5-19.
31
Moeller, S., Schinglemann, F.P., Stulz, R.M., 2005. Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave. Journal of Finance 60: 757-782.
Morck, R., Schleifer, A., Vishny, R.W., 1990, Do managerial objectives drive bad acquisitions? The Journal of Finance 45:31-48.
Rhodes-Kropf, M., Robinson, D., and Vishwanathan, S., 2005. Valuation waves and merger activity: The empirical evidence. Journal of Financial Economics 77: 561-603.
Roll, R., 1986, The Hubris hypothesis of corporate takeovers. Journal of business 59: 197-216.
Ruback, R., 1983. Assessing Competition in the Market for Corporate Acquisitions. Journal
of Financial Economics 11: 141–153.
Savor, P., and Lu, Q., 2009. Do stock mergers create value for acquirers? Journal of Finance 63: 1061-1097
Seth, A., Song, K., and Pettit, R., 2000. Synergy, managerialism, or hubris? An empirical examination of motives for foreign acquisitions of U.S. firms. Journal of International
Business Studies 31: 387-405.
Shleifer, A., and Vishny, R.W., 1989. Management entrenchment: the case of manager specific Investments. Journal of Financial Economic 25: 123-139.
Shleifer, A., and Vishny, R.W., 2003. Stock market driven acquisitions. Journal of Financial
