Strategic Loss Making: The long con?

(1)

Strategic Loss Making: The long con?

M. M. Romkes <

Business Administration

First supervisor: K. J. McCarthy

Second supervisor:

Strategic Loss Making: The long con?

10/12/2012

Master Thesis

M. M. Romkes <marielaromkes@gmail.com>

s2030527

Business Administration – Strategy Innovation

Rijksuniversiteit Groningen

First supervisor: K. J. McCarthy <k.j.mccarthy@rug.nl

Second supervisor: P.M.M. de Faria <p.m.m.de.faria@rug.nl>

Strategic Loss Making: The long con?

k.j.mccarthy@rug.nl>

(2)

2

ABSTRACT

This study sheds a different light on the interpretation of acquisition performance. Where most scholars focus on the acquirer’s performance only, this thesis proposes to include the performance of acquirer’s competition too. It enables practitioners and researchers to distinguish between value enhancing, value destroying, collusion and strategic loss making acquisitions (SLM). Existing literature explains that 50%-70% of the mergers fail; however, this can be biased since literature does not distinct between value destroying and SLM. The model demonstrates that acquirer’s performance can affect the competitions’ performance negatively as well as positively. Results indicate that rivals’ performance follow the performance of the acquirer in the same direction. Therefore a negative performance of the acquirer can deliberately hurt its rivals (SLM). Firms which engage in SLM are more likely to be weak (inefficient & less profitable) with increased market power.

(3)

3

1. INTRODUCTION

After fifty years of research on merger and acquisitions (hereafter M&A, merger or acquisition), no evidence has been found as to, why so many acquisitions fail (Carthwright, 2005; Clayton, 2010). Clayton (2010) argues that most variables tested, fail to explain why some acquisitions fail and why some succeed. More than seventy per cent of all acquisitions fail. Even the less sceptical researchers believe that fifty per cent of all acquisitions fail. This raises two questions: Are M&A performance measurement techniques missing something? Or do managers have unobserved motives for failing acquisitions? Current literature establishes three levels on which M&A performance is measured: (1) the task level (integration process); (2) transaction level (actual value in costs efficiencies and revenue growth generated) and (3) firm level (variation in firm performance) (Zollo & Meier, 2008). None of these levels measures M&A performance on industry level. A measurement level that is not concentrated on the acquirer alone, but also on its key industry players.

This study suggests that M&A performance should be measured on an industry level for several reasons. First of all, firm performance is measured quite often on industry level, since it is well established that the interdependence of an industry forces companies to react to strategic movement of other firms (Chen, 1996; Penrose, 1959 in Wernerfelt, 1984; Porter, 1985; Venkatraman & Ramanujam, 1986). This interdependence also forces rivals to react to M&A. Therefore measuring M&A performance should also include rivals’ performance. Secondly, current literature cannot distinguish between collusive synergies and efficiency based synergies (Clougherty and Duso, 2011). An increase in acquirer’s performance might indicate an efficiency gain. However, if the competition’s performance is also increased, this should be indicated as a collusive synergy instead of an efficiency gain. This also holds for value destroying acquisitions. There is no distinction made between the situation where the acquirer’s performance decreases and of its rivals’ increases (value destroying) and where both, the acquirer’s and its rivals’ performance decreases (hereafter strategic loss making or SLM). Until now, both types of acquisitions have been seen as value destroying acquisitions. This might explain the high failure rate of 50-70%, since it is based on two different types of acquisitions. Since the lack of an industry level measurement, SLM is not defined in the literature. Therefore this study will define and investigate indicators of SLM. Moreover, this research will investigate if unobserved motives for failing acquisitions are hidden in SLM acquisitions.

(6)

6

indicates how rivals’ performance reacts on the acquirer’s performance. Thereafter a logistic regression is executed to reveal indicators of SLM.

Findings of this study indicate that all four acquirer situations exist (value enhancing, collusion, value destroying and SLM). Secondly, results show that the acquirer’s performance significantly influences rivals’ performance. Both findings indicate that measuring M&A performance on industry level is not obsolete. Results also show that indicators such as efficiency, profitability, market value and R&D significantly predict SLM.

These findings contribute to existing literature in three ways. Firstly it argues the importance of measuring M&A performance on an industry level. Secondly, it enables one to distinct between four types of acquisitions and finally it defines SLM acquisitions.

The structure of the rest of this study is as follows. The industry level merger taxonomy and SLM are described in section two. This is followed by a description of the model and the methodology in section three. The results are presented in section four. Section five discusses and concludes the paper.

(7)

7

2. M&A & INDUSTRY LEVEL

Conceptually, a merger can have four outcomes (1) the acquirer’s performance increases and of its rivals’ decreases; (2) the acquirer’s performance decreases and of its rivals’ increases; (3) the acquirer’s performance increases and so does the performance of the competition and (4) both the acquirer’s performance and of the competition decreases.

Table 1: Conceptual situations

Situation Performance acquirer Performance Rivals

1. Positive Negative

2. Negative Positive

3. Positive Positive

4. Negative Negative

2.1 Merger Taxonomy

Situation 1: Value enhancing

(8)

8 Situation 2: Value destroying

These deals take place when acquirer’s performance decreases and rivals’ performance increases. A well-known value destroying antecedent is compensation. According to the literature CEO’s with higher compensation carry out bad acquisitions. Secondly managerial hubris and political games are also value destroying antecedents (Haleblian et al., 2009; Pettigrew 1977). Too much of an ego and confidence can result in bad decisions and in empire building (Mueller, 1969; Jemison and Sitkin, 1986). Target defensive tactics can also be an antecedent of value destroying (Haleblian et al., 2009). Current literature defines value destroying acquisitions as acquisitions where the acquirer’s performance is decreased. Hence, merging firms that do not incorporate advantages still have to deal with integration difficulties (Larsson and Finkelstein, 1999). This integration process decreases efficiency and competiveness. Moreover, rivals can exploit this opportunity by pursuing aggressive marketing in order to destruct more value of the acquiring firm (Clougherty & Duso, 2011; Ghemawat and Ghadar 2000). Therefore there is a difference between situation (2) and situation (4). Value destroying acquisitions (situation 2) enables rivals to exploit troubling acquirers. Current literature defines value destroying acquisitions as a decreased firm performance of the acquirer. However this definition does not distinguish between situations (2) and (4). Therefore value destroying acquisitions are acquisitions which decrease acquirer performance and increase rival performance.

Situation 3: Collusion

(9)

9 Situation 4: SLM

These deals occur when both, acquirer and rivals decrease, performance. However, existing literature cannot distinguish between value destroying acquisitions (situation 2) and SLM (situation 4). Therefore there is limited literature about this last concept. Clougherty and Duso (2011) proposed that in situation (2) rivals exploit unprofitable acquirers, but not in situation (4). The following chapter proposes that rivals are not able to exploit unprofitable acquirers due to the strategic acquisition. The following chapter will discuss situation (4) further as strategic loss making (SLM). It investigates motives and characteristics of SLM acquisitions.

Concluding, there are four situations (value enhancing, value destroying, collusion and SLM). It is assumed that the acquirer’s performance, whether negative or positive, force rivals to react. Therefore the following hypothesis is constructed.

H1: Acquirer’s performance predicts its rivals’ performance

Table 2: M&A taxonomy

Acquirer increases performance Acquirer decreases performance

Rivals increase performance Collusion effect Value destroying

Rivals decrease performance Value enhancing Strategic loss making

2.2 Strategic Loss Making (SLM)

Strategic loss making makes sense in two ways. Firstly, a firm might effectuate M&A in order to create enough market power to predate and force rivals out of the industry. Secondly M&A might stimulate predation in order to make a very expensive and profitable target possible to take over.

M&A may increase industry concentration (reduce competitors) and therefore market power. Market power enables firm to set prices instead of following prices of others. Due to interdependence in the industry, rivals might follow the increased prices of price setters. This may result in collusion if competitors increase prices too and protect their market at the expense of customers. So, collusion is profitable for both the acquirer and rivals. However, some mergers might not aim for collusion.

Creating a collusion is not easy. First of all some forms of collusions are forbidden by anti-trust laws (Motta, 2007). For example, three Dutch beer brewers were fined by the European Commission because of increased prices due to collusion (Pleus, 2007).

(10)

10

store, and increases its price from $1 to $2. If one player is following this increase and the other player increases its price only to $1.90, the second firm will attract more buyers. According to Motta (2007) there are two main ingredients necessary for collusion: (1) participants of collusion must be able to note deviation from collusion in a timely way; (2) participants must be able to punish deviation from collusion. Furthermore explicit collusions are not always legal and tacit collusion is challenging (Motta, 2007). Tacit collusion is collusion without explicit communication between players. To set the right collusion price is difficult without communication. When setting the price too high, others do not follow and you lose market share. If one firm believes that the collusion price is lower than other players do, it might be interpreted as deviation from the collusion. Furthermore collusion is not a competitive advantage, since both, rivals and mergers profit from the merger.

(11)

11

McGee (1911) has some criticism on predatory pricing. First of all, McGee (1911) claims that the large market share of the predator will incorporate higher losses than rivals. However this is only true when predators only discriminate price in the rival market. However, when price discrimination on its total market share it reduces cost of predation. Secondly McGee (1911) argues that rivals re-enter the industry if the predator increase the prices again. However, re-entering the industry is costly (McGee, 1911). Exiting the industry leaves the firm with sunk costs. Moreover re-entering the industry requires employees and plants, which were fired and closed. The third point McGee (1911) made that it is never tested if predators have deep pockets with a lot of resources and preys don’t. McGee (1911) proposes that preys can explain the situation to its creditors and receive a loan until predation ends. Motta (2007) argues that due to imperfect information and uncertainty predators make rivals and investors believe that it will not generate high profits. Therefore lending money is risky. The last point made by McGee (1911) is that merging rivals is more feasible than predation. However, acquiring rivals is forbidden by law for dominant firms. Secondly, it may stimulate new entrants, because they want to be bought by the incumbent firm at a profit (Motta, 2007). Moreover Telser (1966) and Yamey (1972) argue that merger and predation are not mutually exclusive (Motta, 2007). Predation might drive the price of a rival down and can be taken over at a lower price.

2.3 SLM Model

(12)

12

This market power enables firms to collude, predate and strengthen entry barriers further or increase industry concentration (Bello et al., 2009 & Jacquemin 1972). Due to the concentration and interdependence (hence, actions have a direct effect on rivals) in an oligopoly, it stimulates conditions for increasing or sustaining market power and therefore collusion and predation. So, characteristics, of an oligopolistic market structure, are suitable conditions for SLM. Therefore it is expected that firms which engage in SLM are incumbent firms which are older and larger than their entrants (Buzzel , 1975; Mukhopadhyaya, Roy and Raychudhuri, 2012). Moreover, according to Gottesman (2004) and Motta (2007), predators need ‘deep pockets’. Large firms with ‘deep pockets’ have easier access to funding and resources than small preys with small pockets. Therefore the following hypotheses are constructed.

H2: Firm size predicts the likelihood of SLM H3: Firm age predicts the likelihood of SLM

According to Motta (2007) the idea that competitive pressure forces firms to be the most efficient firm, is a common argument. An oligopoly is to a much lesser extent exposed to competitive pressure compared to perfect competition. In highly competitive pressured markets, firms which are not efficient enough, will not survive and shut down. Also Liebenstein (1966) argues that the ‘quiet life’ comes with inefficiency. So entry barriers need to remain high to prevent efficient rivals from entering the oligopoly. Therefore, inefficiency is a reason to predate to keep rivals out of the industry or preclude further entry. Kreps and Wilson (1982) proved that weak and inefficient incumbents indeed fight entry by a creating a strong reputation. Efficient incumbents will always fight entry by lowering prices to force rivals and entrants out of the industry. However, this is not predation per se, because the incumbent could be efficient enough (Motta, 2007). So firms which are not efficient enough will fight entrants using predatory pricing. This is made possible by the reputation model of predation. Due to imperfect information, rivals, entrants and investors do not know if the incumbent firm is strong or weak (Selten 1975, Kreps & Wilson 1982). Therefore a reputation of being strong is preferable. Uncertainty of rivals, investors or entrants, discourage further entry because they do not know if entering or staying in the industry generates high profits. Inefficient and weak incumbents will exploit the uncertainty of rivals or entrants by creating a reputation of being strong and efficient, and start predatory pricing. Therefore it is very likely that inefficiency incumbents tend to strategic loss making. The following hypotheses are constructed.

(13)

13

H5: Profitability predicts the likelihood of SLM

High market value predicts future growth (Rajan & Zingales, 1998; Tan, Cheah, Johnson, Sung & Chuah, 2012). Market value consists of: (1) capitalized rents due monopoly power; (2) value of rents due to scare factors of production and (3) present value of the firms existing capital (Smirlock, Gilligan and Marshall, 2001). A firm in an oligopoly which is protected by entry barriers would experience market capitalization of monopoly rents (Chen, Hite & Cheng 1989). Due to the increase in concentration, entry barriers and the reputation of being strong and profitable of SLM acquirers, capitalized rents due to monopoly power are expected to increase. Therefore it is expected that market value will increase during SLM.

H6: Market value predicts the likelihood of SLM

Lastly, R&D is also assumed to increase. Accumulation of knowledge is an in house activity, because creating tacit knowledge is mostly firm specific (Smulders & Klundert, 1995). According to Smulders and Klunder (1995), heavily competing firms end up with limited capacity to innovation. Therefore less competition may stimulate innovation. Fontana, Nuvolari Shimizu and Vezulli (2012) argue that innovation by incumbents is more likely in Schumpeter Mark II industries. Schumpeter Mark II industries are industries with a few big players, highly consolidated and with high entry barriers. So, large incumbents make use of this entry barriers and introducing innovations in order to capture the value more easily (Fontana, Nuvolari Shimizu and Vezulli, 2012; Sonenshine, 2011). Also several other authors argue that R&D expenditure is increased for firms with higher marker power (Acs , Audretsch, Braunerhjelm & Carlsson 2012; Bucci, 2003; Sonenshine, 2011). Acs and others (2012) found significant evidence that increased market power increases R&D intensity. He argued that market power eases the exploitation of profits from their innovations. SLM increases concentration and market power. Therefore it is assumed that R&D intensity increases during SLM. When rivals are forced out of the industry, the incumbent has enough market power to increase prices again and make use of is generated knowledge which is now easier to capture the value of. Moreover, it enables firms to recoup for the short term losses generated during predation. The following hypothesis is constructed.

(14)

14

3. DATA, SAMPLE & METHODOLOGY

3.1 Sample selection

The pharmaceutical industry is selected, because of its high entry barriers and the power is concentrated on the top (18 companies account for 76% of the total market share).

3.1.1 Sample

Two databases are created, a narrow and a broad one. The data is employed from the Thomson Reuter’s SDC merger database, we refine the narrow sample to include: (1) the acquisition is announced between January 1, 1990 and December 31, 2011: (2) the acquirer is operating in the pharmaceutical industry (SIC 2833:2866) (3) the acquirer is one of the top twenty pharmaceuticals in the world, based on revenue: (4) the acquirer status code is public and (5) the daily share price data (return index) is available in DataStream. These requirements yield a sample of 659 acquisitions and 18 companies. Additional (financial) information of the acquiring firms is employed from DataStream. The broad sample includes the same refinements; however, all mergers and acquisitions in the US, instead of the top twenty pharmaceuticals in the world, are included. The broad sample yields 2317 acquisitions.

3.2 Firm performance

Firm performance is measured as the stock reaction of a firm to a particular event, acquisitions and mergers in this case. This event study is a useful tool for competitive analysis of mergers and acquisitions; especially because it is assumed that the effect on an acquisition or merger is directly reflected in the stock price (Duso, Gugler & Yurtoglu, 2006). Therefore, the standard event study of Brown and Warner (1985) will be used to calculate the stock reaction of the firm to a merger or acquisition. This stock reaction is also known as the abnormal return (AR) and is defined as firm performance in this study.

(15)

15

abnormal return caused by the acquisition. The abnormal return (AR) is the difference between the realized return (R) and the expected return (ER). See following equation:

= − (1)

In equation (1) the i is the security of the firm and t is the time period. In this thesis the return index is used for the realized return (R) and is calculated as follows:

= (2)

Where being the closing price of the firm on the event date and being the closing price one day after the event. The realized return (R) is directly employed from DataStream. In order to calculate the abnormal return (AR), first the normal return needs to be estimated. The estimated normal return (ER) will show how the return would look like without the acquisition. The estimated normal return will be based on 190 previous trading days (estimation window: starts 250 days before the event until 60 days before the event. The Fama-French model is used to calculate the estimated return, see following equation:

= + + (3)

Where being the intercept coefficient estimated by OLS (ordinary least squares) and being the slope coefficient. is the global S&P 500 composition and the error of OLS. The estimated and are used to calculate the estimated normal returns for all the firms in the event window. As is shown in equation (1) this expected return is then subtracted from the realized return. This is shown in equation (4), which is the standard event study market model according to Brown and Warner (1985):

= − + (4)

The abnormal return is calculated for each single day in the event window (-21 days and + 1 day). Not all acquisitions showed a significant β when estimated, but because the return is still abnormal all acquisitions in the sample are used. The last step is to define the cumulative abnormal return (CAR). Which is done by adding all abnormal returns in the event window, see equation (5).

(16)

16

Hypothesis 1 will be tested on the narrow and broad sample and will be tested as follows. Multiple dummies are created for the firm performance of the rivals (dependent variable) and for the performance of the acquirer (independent variable).

Independent variables

The independent variable of the narrow sample is the performance of the acquirer. This is calculated as described in the previous chapter, the cumulative abnormal return. For the narrow sample, two dummies are created: (1) Acquirer performance positive (1 = positive CAR, 0 = negative CAR); (2) Acquirer performance negative (1 = negative CAR, 0 = positive CAR).

In the broad sample, the performance of the acquirer is also calculated as described in the previous chapter. However, the event window is one day before the announcement and one day after the announcement instead of 21 days before the announcement and 1 day after the announcement. Three dummies are created: (1) Acquirer performance positive (1 = positive CAR, 0 = negative CAR); (2) Acquirer performance negative (1 = negative CAR, 0 = positive CAR); (3) Acquirer performance unchanged (1 = if the CAR increases less than 0.1 or decreases less than -0.1, 0 = any other case).

Dependent variable

The dependent variable of the narrow sample is the performance of the competition of the acquirers. This is also calculated as described in the previous chapter. However, the sample consists of 18 companies. So during an acquisition of a firm, the other 17 firms are seen as the competition. The CAR of these 17 companies is calculated with the same event day as the acquirer. This makes it possible to measure the change of stock of each individual competitor during an acquisition of another firm. Also here two dummies are created: (1) Competition performance positive (>50% of competition positive CAR = 1, any other case = 0); (2) Competition performance negative (>50% of competition negative CAR = 1, any other case = 0).

(17)

17

(2) Competition index performance negative (1 = negative CAR, 0 = positive CAR); (3) Competition index unchanged performance (1 = unchanged CAR, 0 = any other case).

Table 3: M&A taxonomy

Comp. positive Comp. negative Comp unchanged

Acquirer unchanged 1. 2. 3.

Acquirer positive 4. Collusion 5. Value enhancing 6.

Acquirer negative 7. Value destroying 8. Strategic Loss Making 9.

3.4 Strategic Loss Making (SLM)

To test hypothesis 2-7 we create a number of indicator variables:

(1) SLM. A dummy is created based on the variables of the merger taxonomy (sample 1). 1 = when both, the acquirer and the competition have negative firm performance, any other case (value enhancing, value destroying and collusion) = 0.

(2 ) Efficiency. Two variables are created to test efficiency. The accounts receivable turnover measures how efficient a firm uses its assets. A dummy for accounts receivable is created (an accounts receivable lower than the median of the sample = 1, higher than the median = 0). The second variable is asset turnover. It measures how efficient a firm uses its assets in generating money.

(3) Profitability. Free cash flow (FCF) and earnings before interest (EBIT) are used (are defined as in DataStream). Two dummies are created. If EBIT or FCF is lower than the median of the sample = 1, higher than the median = 0.

(4) Firm size. Also here two variables are created. The first variable is the number of employees. A dummy is created (more employees than the median of the sample = 1, lower than the median = 0). The second variable is the sales per employee. Also here a dummy is created (sales per employee lower than the median of the sample = 1, higher than the median = 0).

(18)

18

capitalization. Market capitalization is the market value of the outstanding shares. Also here a dummy is created (above average = 1, below average = 0).

(6) R&D. A dummy is created for R&D intensity. R&D investments ($) higher than the average of the sample = 1 and a R&D lower than the sample = 0.

(7) Firm age. A dummy is created for firm age. Firms older than the median of the sample = 1 and younger = 0.

3.5 Model specification

Both models (merger taxonomy & SLM) use a binary dependent variable. Therefore I use the binary logistic regression model (Dikova, Sahib & Witteloostuijn, 2010).

3.5.1 Merger Taxonomy

The logistic model for the merger taxonomy is expressed as:

!" # $% %"#& '('("$ = 1/ 1 + + (6)

The performance of the competition in equation (6) can be negative as well as positive. In equation (6) the Z value is the regression between the independent variable and the coefficient. The Z value is estimated in equation (7).

, = _-+ !" # $% %./( + (7)

In this equation, _- represents the intercept and ₀ the coefficient. is the error and i refers to an acquisition of the sample.

3.5.2 SLM

The logistic model for SLM is expressed as:

1' ' 2(% 3"44 5 6($2 = 1/ 1 + + (8)

(19)

19

In equation (9), the _- represents the intercept and ₀ the coefficient. is the error and i refers to an acquisition of the sample.

, = -+ %%"/$'4 % (7 89 + : 44 ' '/ $"7 + ; < 4ℎ <9"> + ? @A + B C/#8 "! #&9"D 4 + E 1 9 4 & #&9"D +

F ""6 '" # 6 ' '(" + G A"8($H4 I + J 5 6 ' % &(' 9(K '("$ +

- 4 %ℎ $L L 7 9"&# $' + (9)

3.5.3 Multicollinearity

Multicollinearity means that two independent variables have a high correlation. Multicollinearity will bias the relation between the independent and independent variables. The merger taxonomy is tested with one independent variable, so multicollinearity cannot exist. However, the indicator variables of SLM are tested for multicollinearity. A variance inflation factor (VIF) of 10 requires further investigation. However, no variables have a VIF larger than 2, therefore there is no sign of multicollinearity.

3.5.4 Outliers and Influential observations

Due to predicting residuals, one can estimate if observations are outliers and do not fit the model. Observations with high predicting residuals do not fit the model well. Also some observations are more influential than others. Observations with a high Cook’s Distance are more influential than other observations. The variables for both, the merger taxonomy and SLM, are tested for outliers and influential observations. In the merger taxonomy model there was one observation which was an outlier and in the SLM model four outliers were detected. Since the outliers are so minimal, these observations will not bias the outcomes and are therefore not deleted. In the merger taxonomy model there are five observations with a high Cook’s Distance and in the SLM model only three. Also these observations are not deleted, since the expected effect will be minimal.

3.5.5 Goodness-of-fit

(20)

20 3.5.6 Specification error

(21)

21

4 EMPIRICAL RESULTS

4.1.1 Sample statistics

Of the narrow sample 27.91% are classified as SLM acquisitions. The sample consists of 27.91% of collusion. Value enhancing is represented by 21.63% and value destroying by 22.55%.

In the broad sample 7.91% is represented by SLM. 7.82% is collusion. Value enhancing represents 5.24% and value destroying 5.07%. 20.75% is represented by acquisitions where the acquirer’s performance and the competitor’s performance are unchanged. An Acquisition where the acquirer’s performance is unchanged and the competitors’ performance is positive is 4.28% and where the competitor’s performance is negative is 4.89%. Lastly, of 23.50% of the acquisitions, the acquirer’s performance is positive and the competition’s performance is unchanged. 20.58% is the acquirer’s performance is negative and the competition’s performance unchanged.

(22)

22 4.1.2 Results Merger Taxonomy

The following table presents the analyzed results of the merger taxonomy. Table 5: Merger taxonomy

Univariate Narrow Sample Broad Sample

Value enhancing (0.6173)*** (.6809)*** -3.07 -3.25 .4299 .1434 Value destroying (0.6409)*** (.8170)* -2.84 -1.68 .4441 .1510 Collusion (1.5603)*** (1.6093)*** 2.84 4.24 .5549 .2139

Strategic Loss Making (1.6200)*** (1.7593)***

3.07 5.09

.5498 .2357

*** p < 0.01, ** p < 0.05, * p < 0.1

(23)

23 Table 6: Merger taxonomy (broad sample)

Univariate Comp. positive Comp. negative Comp unchanged

Acquirer unchanged (1.01569) (.8572467) (1.4055)***

0.17 -1.26 3.50

.6428 .1623 .6944

Acquirer positive Collusion*** Value enhancing*** (1.0156)

4.24 -3.25 0.17

.2139 .1434 .6428

Acquirer negative Value destroying* Strategic Loss Making*** (.8391)*

-1.68 5.09 -1.97

.1510 .2357 .6133

*** p < 0.01, ** p < 0.05, * p < 0.1

Table 6 shows, that also in this sample, the probability of SLM and collusion is higher than value enhancing and value destroying. Secondly, the acquirer’s positive performance increases the likelihood that rivals’ performance is also positive. This also holds if the performance of the acquirer is negative. Moreover, if the acquirer’s performance does not change, there is a significant likelihood that rivals’ performance also not changes. This reinforces the findings found in table 6; rivals’ performance follows the acquirer’s performance in the same direction. Taken together these results support hypothesis 1. Acquirer’s performance predicts the performance of its rival’s. Hence: (1) increased (decreased) performance of acquirers increases the likelihood of increased (decreased) performance of its rivals; (2) increased (decreased) performance of acquirer decreases the likelihood of decreased (increased) performance of its rivals. These findings are also robust. If the acquirer’s performance is negative the likelihood that 80% of the competition’s performance is also negative increases with 4.33. This is also true for collusion, value enhancing and value destroying (see table 6.1, appendix 7.2).

4.2 Strategic loss making

(24)

24 4.2.1 Model 1: Efficiency

Table 8: Model 1 Efficiency – SLM

Variables Univariate Multivariate

Accounts receivable (0.104) (0.007)*** 1.3288 1.697372 1.63 2.71 7.45*** Asset Turnover (0.047)** (0.003)*** 0.4660 .2850868 -1.99 -2.94 8.79*** Prob > χ2 (0.0033)*** Pseudo R2 0.0147 *** p < 0.01, ** p < 0.05, * p < 0.1

Table 8 reveals that in the univariate model, high asset turnover decrease the likelihood of strategic loss making. An increase in asset turnover decreases the odds of Strategic Loss Making. In the multivariate model both, accounts receivable and asset turnover, are significantly important. The asset turnover odds ratio decreases further from 0.45 to 0.29 and accounts receivable becomes significant at α = 0.01. The odds ratio increases from 1.33 to 1.77. Secondly, the model is also highly significant (0.0033). The likelihood ratio test shows a highly significant likelihood ratio for accounts receivable and asset turnover as well (7.45 & 8). This indicates that dropping the variables from the model leads to a poorer fit of the model. The following table shows whether the efficiency model also predicts collusion, value enhancing and value destroying acquisitions.

Table 9: Multivariate Efficiency models

Variables Collusion Value enhancing Value destroying

Accounts receivable (.9002799) (.6624173)* (.9055727)

Asset Turnover (1.134259) (1.357336) (2.562115)**

Prob > χ2 (0.8590) (0.1467) (0.1044)

*** p < 0.01, ** p < 0.05, * p < 0.1

(25)

25

to table 8 and table 9, efficiency is a significant predictor for Strategic Loss Making, but not for collusion, value enhancing and value destroying acquisitions. Therefore hypothesis 4 is accepted.

4.2.2 Model 2: Profitability Table 10: Model 2 Profitability – SLM

Free Cash Flow (0.254) (0.056)*

.8141321 1.87923 -1.14 1.91 3.83* EBIT (0.015)** (0.002)*** 1.534091 2.800512 2.43 3.11 10.53*** Prob > χ2 (0.0026)*** Pseudo R2 0.0164 *** p < 0.01, ** p < 0.05, * p < 0.1

Table 10 reveals the relation between profitability and Strategic Loss making. In the univariate model, a decrease in earnings (EBIT) increases the probability of Strategic Loss Making. In the multivariate model, if free cash flow is kept constant, EBIT becomes more important and increases the likelihood of Strategic Loss Making. The odds ratio increases from 1.54 to 2.50. Also, when EBIT is kept constant Free Cash Flow, becomes a predictor of Strategic Loss Making. Moreover, the model is highly significant (α = 0.01). Also the likelihood ratio test shows that both variables cannot be omitted from the model (3.83 & 10.53). The following table compares the multivariate profitability model against collusion, value enhancing and value destroying acquisitions.

Table 11: Multivariate Profitability models

Free Cash Flow (1.13294) (1.037145) (.499954)**

EBIT (.8952922) (.6949237) (.5307677)**

Prob > χ2 (0.4687) (0.1548) (0.0894)**

(26)

26

In table 11 we can see that low profitability does not predict collusion or value enhancing acquisitions. Furthermore, low profitability has a significant predictor effect on value destroying acquisitions. However, low profitability decreases the probability of value destroying acquisitions instead of increases the probability.

Table 12: Wald test – Comparing SLM profitability model

Profitability Model χ2 Prob > χ2

Value destroying 13.94 0.0030***

*** p < 0.01, ** p < 0.05, * p < 0.1

In the table above the Wald test is executed in order to examine if the profitability model is significantly different between value destroying and SLM acquisitions. The p value is significant at α = 0.01. This implicates that in 99% of the cases a low profitability increases the likelihood of SLM and decreases the likelihood of value destroying. Therefore we conclude that low profitability predicts SLM and hypothesis 5 is accepted.

4.2.3 MODEL 3: combination model 1&2

In the following table, a model is constructed that shows the relation between weak (inefficient and low profitable) firms and Strategic Loss Making.

Table 13: Model 3

Variables SLM Collusion Value enhancing Value destroying

Asset turnover (.2624382)*** (1.725) (1.351526) (2.129196)

Accounts receivable (1.558273)** (.8865471) (.6976211) (.9029882)

Free cash flow (1.500341) (1.264191) (1.014491) (.5892308) EBIT (2.621964)*** (.9223253) (.6924482) (.5521748)**

Prob > χ2 (0.0004)*** (0.6090) (0.1851) (0.1468)

Pseudo R2 0.0285 0.0038 0.0101 0.0105

*** p < 0.01, ** p < 0.05, * p < 0.1

(27)

27

likelihood ratio test later in the chapter to investigate if free cash flow lead to a significant improvement in the fit of the model.

4.2.4 Model 4 Firm Size Table 14: Model 4 Firm Size – SLM

Number of employees (0.442) (0.537) 1.143506 1.114653

0.77 0.62

0.38 Sales per employee (dummy: median) ¹ (0.001)*** (0.001)***

1.810974 1.802362 3.34 3.31 11.18*** Prob > χ2 (0.0028)*** Pseudo R2 0.0152 *** p < 0.01, ** p < 0.05, * p < 0.1

Table 14 indicates that number of employees is not significant. However, low sales per employee increase the likelihood of Strategic Loss Making. If the number of sales per employee is kept constant the number of employees is still not a predictor factor for Strategic Loss Making. Also the likelihood ratio test is not significant (0.38). This means that leaving number of employees out of the model does not lead to a significant poorer fit. However, this variable will not be omitted from the model. This variable can have a significant effect in the combined or full model. The firm size model is highly significant (0.0028).

Table 15: Multivariate Firm Size models

Number of employees (1.432124)** (.9762662) (.6544656)**

Sales per employee (.8012163) (.6706862)** (.8839697)

Prob > χ2 (0.0619)* (0.1095) (0.0598)*

*** p < 0.01, ** p < 0.05, * p < 0.1

(28)

28

sales per employee is significant for value enhancing. Since the odds ratio is below 1, it means that when the sales per employee decrease the likelihood of value enhancing also decreases. However, the model for value enhancing is not significant. The model for value destroying is significant, but here the odds ratio decreases when the number of employees increases. In the following table the Wald test will be executed in order to determine if the size model significant differs from collusion and value destroying, compared against SLM.

Table 16: Wald test – Comparing SLM Firm Size Model

Firm Size Model χ2 Prob > χ2

Collusion 8.27 0.0408**

*** p < 0.01, ** p < 0.05, * p < 0.1

The table above shows that both models (collusion and value destroying) are significant different from SLM. This means that when the number of employees is kept constant and the sales per employee decreases the likelihood of SLM increases. This situation has no significant effect on collusion, value destroying or value enhancing. Therefore we conclude that size on its own does not predict SLM. However large firms, with low sales, do predict SLM. Therefore hypothesis 2 is rejected.

4.2.5 MODEL 5: combination model 3&4 Table 17: Model 5

Asset turnover (.1612261)*** (1.687684) (1.576336) (3.181288)**

Accounts receivable (1.63297)** (.8407871) (.6914235) (.9684255) Free cash flow (1.628419) (1.012009) (.915702) (.6678383)

EBIT (2.236054)** (1.077792) (.7836421) (.5340101)

Number of employees (1.180589) (1.456664) (1.046788) (.6007667)* Sales per employee (2.26332)*** (.7143117) (.6514821) (.7330594)**

Prob > χ2 (0.0000)*** (0.4014) (0.1670) (0.0166)**

Pseudo R2 0.0516 0.0086 0.0148 0.0240

*** p < 0.01, ** p < 0.05, * p < 0.1

(29)

29

the likelihood of SLM, they decrease the likelihood of value destroying acquisitions at the same time. The following table shows the results of the Wald test.

Table 18: Wald test – Comparing SLM model 5

Model 2 χ2 Prob > χ2

*** p < 0.01, ** p < 0.05, * p < 0.1

The Wald test of table 18 shows that the where the variables increase or decrease the likelihood of SLM the variables have a significant opposite effect on the likelihood of value destroying. Since number of employees is still not a significant predictor for SLM we will do the likelihood ratio test later in the chapter to investigate if dropping the variable leads to a significant poorer fit of the full model. However profitability, efficiency and size have a significant different effect on SLM than on value enhancing, value destroying and collusion.

4.2.6 Model 6: Market value Table 19: Model 6 High market value – SLM

Tobin’s Q (0.026)** (0.001)***

1.48219 2.56936

2.22 3.36

12.27*** Book to market ratio (0.918) (0.013)**

(30)

30

In the univariate model, only the Tobin’s Q is significant. So, overvalued firms increase the likelihood of SLM. When keeping book to market ratio and market capitalization constant, the likelihood ratio of Tobin’s Q increases further (2.57). Also book to market ratio becomes significant in the multivariate model. The p value of the model is also significant. The likelihood ratio shows that dropping the Tobin’s Q or Book to market ratio leads to a significant poorer fit of the model (12.27 & 6.64). Dropping the market capitalization does not lead to a significant poorer effect. However, this variable will not be omitted since it can have a significant influence on the combined or full model. The following table’s results indicate if the model is also significant for collusion, value enhancing and value destroying acquisitions.

Table 20: Multivariate market value model

Tobin’s Q (.600115)* (.7930375) (.8355161)

Book to market ratio (.6428337)** (1.083334) (.7371535)

Market capitalization (.184503)*** (1.747341) (1.443886)

Prob > χ2 (0.0006) *** (0.1535) (0.5392)

Pseudo R2 (0.0223) (0.0077) (0.0031)

*** p < 0.01, ** p < 0.05, * p < 0.1

Table 20 shows that the market value model is also significant for collusion. However, the odds ratio below one indicates that overvalued firms decrease the likelihood of collusion. In order to conclude that overvalued firms increase the likelihood of SLM and decrease the likelihood of collusion, the results of the Wald test are shown in the following table.

Table 21: Wald test – Comparing SLM High market value model

High market value Model χ2 Prob > χ2

Collusion 19.15 0.0007***

*** p < 0.01, ** p < 0.05, * p < 0.1

(31)

31 4.2.7 MODEL 7: combination model 5 & 6 Table 22: Model 7

Asset turnover (.1210787)*** (1.805939) (2.225667) (2.940472)*

Accounts receivable (1.58973)** (.7759489) (.7576171) (.9838053)

Free cash flow (1.792869) (1.225998) (.8034666) (.5997079)

EBIT (2.800459)*** (.9141811) (.7856148) (.5209064)* Number of employees (1.33783) (1.111779) (1.157777) (.6319175)*

Sales per employee (2.202351)*** (.7812902) (.6390708)* (.7301577)

Tobin’s Q (3.105596)*** (.6312432)* (.7382292) (.7314873)

Market capitalization (2.132256)* (.6312432)*** (1.285158) (1.31507) Book to market ratio (2.104217)** (.5916785)* (1.150581) (.7707533)

Prob > χ2 (0.0000)*** (0.0070)*** (0.2268) (0.0459) *

Pseudo R2 0.0774 0.0315 0.0191 0.0266

*** p < 0.01, ** p < 0.05, * p < 0.1

In this model market capitalization becomes significant. As one can see this model is significant for SLM, collusion and value destroying. However, the variables have an opposite effect on the likelihood on value enhancing and value destroying, compared to SLM. For example, a high market capitalization increases the likelihood of SLM, but decreases the likelihood of value enhancing and value destroying. The results of the Wald test are shown in the following table.

Model 7 χ2 Prob > χ

Collusion 42.98 0.0000***

*** p < 0.01, ** p < 0.05, * p < 0.1

(32)

32 4.2.8 Model 8: Research & Development Table 24: Model 8 Research & Development – SLM

Research & Development (0.117)

.489954

-1.57

Prob > χ2 (0.0925)*

Pseudo R2 0.0037

*** p < 0.01, ** p < 0.05, * p < 0.1

It was expected that an increase in R&D intensity increases the likelihood of SLM. However the opposite effect is observed. An increase in R&D decreases the likelihood of SLM. However, the P value is not significant (0.117), which means that R&D is in the univariate model not a predictor for SLM. Therefore hypothesis 7 is rejected.

Table 25: Univariate Research & Development models

Research & Development (.2996255)** (3.066203)*** (1.513675)

Prob > χ2 (0.0097) *** (0.0019)*** (0.2785)

Pseudo R2 (0.0087) (0.0143) (0.0017)

*** p < 0.01, ** p < 0.05, * p < 0.1

Table 25 shows that R&D intensity is a better predictor for collusion and value enhancing acquisitions than for SLM. For value enhancing acquisitions the p value is significant at α = 0.01 level and for collusion α = 0.05. This means that an increase in R&D increases the likelihood of value enhancing acquisitions and decreases the likelihood of collusion.

4.2.9 Model 9: Age Table 26: Model 9 Age – SLM

Age 1.076437 1.076437 .9171457 .9440548

*** p < 0.01, ** p < 0.05, * p < 0.1

(33)

33 4.2.10 The Full Model

Table 27: Full Model 9

Asset turnover (.139795*** (1.776482) (2.136868) (2.845771)

Accounts receivable (1.465382)* (.7801682) (.8091764) (.9992078)

Free cash flow (1.812137 (1.22758) (.8096355) (.5983003)

EBIT (2.722811)*** (.9154108) (.8104062) (.522536)* Number of employees (1.260544) (1.106639) (1.191383) (.6345893)*

Sales per employee (2.049321*** (.7857536) (.6781985) (.7416578)

Tobin’s Q (3.008749*** (.6342571)* (.7506429) (.7372031)

Market capitalization (5.207644** (.1679398)** (.4338641) (1.098192) Book to market ratio (2.164342** (.5901127)** (1.121758) (.7637585)

R&D (.2525116)* (1.049753) (3.954432)* (1.276508)

Prob > χ2 (0.0000)*** (0.0129)** (0.1280) (0.0694)*

Pseudo R2 (0.0805) (0.0313) (0.0246) (0.0267)

*** p < 0.01, ** p < 0.05, * p < 0.1

In the last model R&D becomes significant at α = 0.1 level, keeping other variables constant. This indicates that for every dollar increase in R&D the likelihood of SLM decreases with 0.2525. The Model is significant at α = 0.01 for SLM. The Model is also significant for collusion and value destroying. The variables of the model, which are significant for collusion or value destroying, have the opposite effect on SLM. Over valuation of firms increases the likelihood of SLM and decreases the likelihood of collusion significantly. Secondly, low earning increases the likelihood of SLM and decreases the likelihood of value destroying acquisitions. Finally, every additional employee decreases the likelihood of value destroying acquisitions. Number of employees is not a significant predictor for SLM. However the relation indicates that for every additional employee the probability of SLM increases. The following table observes if the model has a significant different effect on collusion and value destroying compared to SLM.

Model 3 χ2 Prob > χ2

Collusion 47.97 0.0000***

(34)

34

Table 28 shows that the model indeed has a significant different effect on collusion and value destroying. This means that indeed a high market value increase the likelihood of SLM and on the same time decreases the likelihood of collusion. Therefore we can conclude that efficiency, profitability, Firm size, market value & R&D have a significant different effect on SLM than on collusion, value enhancing and value destroying. Age is not included in this model, since it had no significant effect on one of the merger situations.

4.2.11 Shaping the model

Table 29 shows the odds ratios and their significance of the full model. In the following table the p values and the pseudo R2 of all previous models are shown.

Table 29: Comparing all models – SLM

Model Prob > χ2 Pseudo R2

Model 1 Efficiency (0.0033)*** 0.0147

Model 2 Profitability (0.0026)*** 0.0164

Model 3 (combination model 1&2) (0.0004)*** 0.0285

Model 4 Firm size (0.0028)*** 0.0152

Model 5 (combination model 3&4) (0.0000)*** 0.0516 Model 6 High market value (0.0049)*** 0.0168

Model 7 (combination model 5&6) (0.0000)*** 0.0774

Model 8 R&D (0.0925)* 0.0037

Model 9 Full Model (0.0000)*** 0.0805

*** p < 0.01, ** p < 0.05, * p < 0.1

(35)

35

shows the results of the likelihood ratio test. The results of the likelihood ratio test shows if the combined models are also significantly better than the other models.

Table 30: Likelihood ratio test – Comparing models

Model3 Model 5 Model 7 Model 9

Model 1 (13.00)*** (29.79)*** (48.51)*** (50.93)* Model 2 (8.77)** (25.56)*** (44.28)*** (46.68)* Model 3 (16.79)*** (35.51)*** (38.28)*** Model 4 (21.37)*** (40.10)*** (42.63)*** Model 5 (18.72)*** (21.71)*** Model 6 (42.33)*** (44.68)*** Model 7 (3.22)* Model 8 (56.06)*** *** p < 0.01, ** p < 0.05, * p < 0.1

(36)

36 Table 31: Likelihood ratio test Model 9 – SLM

Dropped Variable LR χ2 Prob > χ2

Asset turnover 10.44 (0.0012)***

Accounts receivable 2.83 (0.0928)*

Free cash flow 2.26 (0.1327)

EBIT 8.13 (0.0044)***

Number of employees 0.79 (0.3756)

Sales per employee 9.28 (0.0023)***

Tobin’s Q 14.17 (0.0002)***

Market capitalization 6.46 (0.0110)**

Book to market ratio 5.75 (0.0165)**

R&D 3.22 (0.0727)*

Free cash flow & number of employees 4.50 (0.1052)

*** p < 0.01, ** p < 0.05, * p < 0.1

The table above indicates that dropping number of employees and free cash flow from model 9 does not lead to a significant poorer fit of the model. This also explains why the linktest for the profitability and firm size model was not passed. However, dropping other variables from the model will lead to a significant poorer effect. Therefore, only number of employees and free cash flow will be omitted from the final model (model 13). In the following table the fit of the final model will be compared to collusion, value enhancing and value destroying.

Table 32: Final Model 13

Asset turnover (.1032368)*** (1.693458) (2.233628) (3.935879) Accounts receivable (1.456681)* (.7781317) (.8293015) (.9483765)

EBIT (1.492509)* (.7697641) (.8470397) (.9361772)

Sales per employee (1.881655)*** (.8159449) (.7379451) (.7443959) Tobin’s Q (2.601065)*** (.624565)* (.8157866) (.7927801)

Market capitalization (5.775379)** (.1531458)** (.4288203) (1.023394)

Book to market ratio (1.798701)** (.6155709)* (1.026667) (.8819745)

R&D (.1869241)* (.9847797) (4.665673)** (1.386585) Prob > χ2 (0.0000)*** (0.0044)*** (0.0362)** (0.3202)

Pseudo R2 0.0698 0.0290 0.0246 0.0134

(37)

37

The full model is also significant for collusion and value enhancing, the Wald test will be presented in the following table. The Wald test shows if the variables of model 13 have a significant different effect on Collusion and Value enhancing than on SLM.

Table 33: Wald test – Comparing SLM Model 13

Model 13 χ2 Prob > χ2

Collusion 46.15 (0.0000)***

Value enhancing 39.60 (0.0000)***

*** p < 0.01, ** p < 0.05, * p < 0.1

(38)

38

5 DISCUSSION

The aim of this study was to shed a different light on the interpretation of acquisition performance. Where most scholars focus on the acquirer’s performance only, this thesis proposes to include the performance of acquirer’s competition too. This industry level measurement technique revealed a merger taxonomy of four situations: (1) value enhancing acquisitions; (2) value destroying acquisitions; (3) collusion and (4) strategic loss making. This merger taxonomy made it possible to distinguish between value enhancing acquisitions (1) & collusions (3) and value destroying acquisitions (2) & SLM (4). This was not possible in existing literature. One of the most important findings is that the acquirer’s performance predicts the performance of its industry players. So there is a significant relation between the acquirer and its competition. This is also what Porters Five Forces argues; firms react on actions of their industry players (McGee et al., 2012). Results argue that the competition’s performance follows the acquirer’s performance in the same direction. In other words, if the acquirer’s performance increases (decreases) the competition’s performance also increases (decreases). This also holds if the acquirer’s performance does not change. As is argued before, it is well established that at least 50% of all acquisitions fail. However, this figure is based on the acquirer’s performance only. Both, value destroying and SLM are indicated by a decrease in the acquirer’s performance. When focusing on value destroying and SLM together (thus on only acquirer’s negative performance) they also represent 51.46% of the sample. However, when interpreting acquirer’s performance on both, the acquirer and its competition, only 22.55% represents value destroying acquisitions and 27.91% SLM. Also the probability of value destroying (44%) is lower than SLM (55%). Since there is a significant relation between the acquirer’s performance and its competitor’s performance, focusing on only the acquirer’s performance can bias the results. Therefore, this study urges that acquirer’s performance should be measured on both, the acquirer’s performance and its industry player’s performance. Therefore an industry level would improve M&A performance literature. The second part of this study was to focus on SLM. Due to the absence of measuring M&A performance on industry level, SLM was not defined in the literature. SLM is a strategic acquisition which is not unfavourable per se. When inefficient incumbents notice threats from efficient entrants, they could increase market power by acquisition and drive price down instead of up. This destroys value for the industry, which makes the industry less attractive and forces rivals and entrants out of the industry. Also it can drive prices down of expensive targets (rivals) which try to further entry the industry.

(39)

39

increases. The likelihood of SLM also increases when the profitability of the acquirer is lower than its competitors. Inefficient and low profit firms predict SLM better than inefficiency or low profit firms individually. This indicates that weak firms engage in SLM acquisitions are more likely than efficient and profitable firms. An explanation could be efficient firms are not threatened by new efficient entrants so much as inefficient incumbents. Moreover, efficient firms might be so efficient that they can lower the price without hurting themselves and increase market share. However, benefiting from an oligopoly and thus decreased competitive pressure, might accompany inefficiency. Although results are not significant, they indicate that efficient and profitable firms predict value enhancing acquisitions or collusion. One could argue that weak firms do not predict SLM, but value destroying acquisitions in general. However the model is only significant for SLM and not for value destroying acquisitions. The number of employees is not a significant predictor for SLM. An explanation could be that the sample exists of only large firms. When testing the number of employees for SLM in the second sample (all pharmaceutical acquirers in the US), a significant relation is found. For every additional employee the likelihood of SLM increases. However, the variable sales per employee, is very significant. Large companies with low sales, increase the likelihood of SLM significantly. However, low sales per employee also reinforces that efficiency and profitability are important predictors of SLM. So there is some minor support that large companies predict SLM. The age of the firm was also not a significant predictor of SLM. Also in this case, the sample exists of only incumbent firms. Therefore this could be an explanation why age was not significant.

The most important predictor is overvalued firms. Firms which have a high Tobin’s Q and Market to book ratio increase the likelihood of SLM. This means that the market values firms more than the actual value in the books. Moreover weak firms (inefficient and low profitable) which are also overvalued predict SLM significantly better than only weak firms. This means that the market expects further growth for these firms. It indicates that the firm is able to earn monopoly rents. Another reason that market value increases, is that firms create a reputation of being strong, thus profitable and efficient, while they are not. This reputation is necessary to make others believe that they are strong. Due to imperfect information and the low price, it forces rivals out of the industry, since they miss the economies of scale and scope. Secondly they might not have the resources to keep the price that low for as long as the predation lasts. Moreover it also gives a sign to other potential entrants that entering is not profitable.

(40)

40

SLM in order to be able increase prices again, the R&D intensity decreased. An explanation could be that weak firms, which engage in predatory pricing, have no money left to increase the R&D intensity. Another explanation could be that when rivals are forced out of the industry they increase prices again and then heavily increase R&D intensity and make use of the increased appropriability.

Acquisitions by firms that are inefficient, low on profit, overvalued and have a lower R&D intensity than its competitors, have a significant increased likelihood of engaging in SLM. Secondly, it significantly decreases the likelihood of collusion or value enhancing acquisitions. Moreover, it has no significant effect on value destroying acquisitions.

5.1 Managerial Implications

Acquisition performance can be best interpreted on both the acquirer’s performance and its competitor’s performance, since the acquirer’s performance has a significant effect on the performance of the competition. The competitor’s performance is more likely to increase (decrease) if the acquirer’s performance is also increased (decreased). This is an important trade off when making strategic decisions about acquisitions. Especially for weak (inefficient and low on profit) incumbents, who are threatened by rivals.

Investing all resources in a value enhancing acquisition might not be the best solution, since it increases the likelihood that competitors profit from the created value too. Hence, the probability of value enhancing (43% & 14%) is smaller than collusion (55% & 21%). As discussed in the literature, rivals try to create a collusion (also increase price) and try to profit from the value created. Also deciding to increase efficiency (value enhancing), because rivals pose a threat since they are more profitable or efficient (stronger), can have the adverse effect. The threatening rival can also profit from the value created, because the increased value by the acquirer can lead to an uninvited collusion. This does not decrease the threat of the more efficient rival.

(41)

41

SLM can be a better solution than collusion or value enhancing. First of all rivals are not given the opportunity to profit from value created due to acquisition. Secondly, investing resources in a value enhancing acquisition that increases the likelihood of collusion might be undesirable. Especially, for already threatened weak incumbents, since collusion does not always decrease the threat of rivals. Moreover, a collusion can eventually lead to a price war, when rivals try to deviate from the collusion. Therefore, investing those resources in SLM, instead of a more expensive value enhancing acquisition, is a more effective technique to fight efficient and profitable rivals and force them out of the industry. Managers need to take in consideration that large efficient and profitable firms will not be forced out the industry by SLM. Also weak incumbents with a lot of resources might be able to overcome the predation. SLM is only effective for rivals which do not have the same amount of resources as the acquiring firm and thus are not able to overcome the predation.

Victims of predatory pricing should persuade investors or funders for a loan. SLM acquirers are not the most efficient or profitable firms. This means that the predation cannot last for a very long time. A loan can help rivals to overcome the predation. Predation by SLM confirms that incumbents feel threatened by rivals, since weak incumbents are more likely to engage in SLM. However, due to imperfect information, rivals never know for sure if an incumbent predator is weak instead of strong. Therefore lending money to a prey might be a very risky investment.

5.2 Limitations & Future research

This study argues that a negative acquirer’s performance has significant negative effect on the competitor’s performance. This study assumes that this effect can be achieved by an increased market power (since it enables firms to set prices above or below an industry price and which explains a decrease of firm performance of the competition) and by predatory pricing (since it sacrifices short term profits, which could explain the decrease in the acquirer’s firm performance). However, the increased market value assumes that there is some increase in market power. However there is a debate if market value measures monopoly rents. According to (Chen et al., 1989) Tobin’s q is a better measurement for profitability than account rates of return. Since it uses present values that reflect long run expectations. Therefore Tobin’s Q is subject to measurement error. Moreover it should be further investigated if the concentration of firms increases during SLM.

(42)

42

an increase in market power and predatory pricing. This is also the case for forcing rivals out of the industry. It is not tested if rivals indeed leave the industry due to the predatory pricing. Therefore further research should also investigate if SLM indeed forces rivals out of the industry. Thirdly, there is minor evidence that large incumbents are more likely to engage in SLM. However, an explanation could be that the sample exists of only large incumbents. More research should point out if large incumbents increase the likelihood of SLM.

Lastly it is assumed that the effect on an acquisition or merger is directly reflected in the stock price (Duso, Gugler & Yurtoglu, 2006). However this could also be stock market inflation. However the Fama-French model is used to control for this. Existing literature does not concern with this issue.

5.3 Conclusion

M&A performance should also be measured on industry level. It enables practitioners and researchers to distinguish between value enhancing, value destroying, collusion and SLM acquisitions. Especially between value destroying and SLM acquisitions. Existing literature explains that 50%-70% of the mergers fail; however, this can be biased since literature does not distinct between value destroying and SLM. Acquirer’s performance can affect the competition’s performance negatively as well as positively. It is more likely that the performance of rival’s follows the performance of the acquirers. Acquirers can choose if they hurt the industry or consolidate the industry. Firms that are weak (inefficient & less profitable) and have increased market power and market value are more likely to engage in SLM. Rivals need to be aware that SLM will affect the industry negatively.

Strategic Loss Making: The long con?

Strategic Loss Making: The long con?

Strategic Loss Making: The long con?

Strategic Loss Making: The long con?

ABSTRACT

TABLE OF CONTENTS

1. INTRODUCTION

2. M&A & INDUSTRY LEVEL

3. DATA, SAMPLE & METHODOLOGY

4 EMPIRICAL RESULTS

5 DISCUSSION