Inorganic Growth Strategies in Private Equity: Empirical Evidence on Add-on Acquisitions

Transcription

1 Inorganic Growth Strategies in Private Equity: Empirical Evidence on Add-on Acquisitions Benjamin Hammer, Alexander Knauer, Magnus Pflücke, Bernhard Schwetzler Abstract We study add-on acquisitions of portfolio firms based on a world-wide data sample covering 9,548 buyouts between 1997 and We find that add-ons have become increasingly relevant, but activity is concentrated among a few PE sponsors. Further, our results suggest that synergies are important when pursuing add-on acquisitions. We show that portfolio firms initiate acquisitions early to realize synergies until the exit. Lastly we identify determinants that increase the likelihood of add-ons for a particular buyout. Our findings help academia and practice to understand inorganic growth in buyouts, which is important given the aging of the industry. This Version: 28 June 2014 Keywords: Mergers and acquisitions, leveraged buyouts, buy and build, bolt-on JEL classification: G23, G30, G34, G39 We gratefully acknowledge comments from April Knill and participants of the Midwest Finance Association Annual Meeting We would also like to thank the audience of a workshop hosted by Accuracy Corporate Financial Advisory and participants of WHU Campus for Private Equity 2014, Alpbach Finanzforum 2013 as well as the HHL doctoral colloquium in November We further thank Robert Friesel for excellent research assistance and Markus Brendel for providing econometric support. We are grateful to BofA Merrill Lynch for providing us with high-yield index data. This study also benefitted from valuable comments of Daniel Nothelfer. All remaining errors and omissions remain our own. All HHL Leipzig Graduate School of Management, Jahnallee 59, Leipzig (Germany). Correspondence concerning this article should be addressed to Benjamin Hammer, Benjamin.Hammer@hhl.de 1

2 1 Introduction Since the early days of the industry, the PE business model has undergone significant change. Kaplan (1997) argues that many corporations have adopted traditional governance improvement techniques of leveraged buyouts (LBOs) hence reducing the value potential from resolving agency costs for PE sponsors. Gompers and Lerner (2000) further show that high returns and improved access to capital have led to more money chasing fewer deals thus also reducing the value potential from traditional arbitrage. In such a commoditizing industry (Sensoy et al., 2013) where traditional value generation strategies have lost effectiveness (Achleitner et al., 2012), PE sponsors have been increasingly forced to seek for new sources of differentiation. 1 Research adapts to this new reality by revisiting the sources of PE value creation. This may apply less to long known value drivers such as agency cost reduction (e.g. Jensen, 1989), financial engineering (e.g. Guo et al., 2011) or multiple arbitrage (e.g. Achleitner et al., 2011) but rather to operating value drivers that have been frequently examined in recent years. Among the most recently discussed operating value drivers are those that foster portfolio firm 2 growth. While several studies have found that, for example, employment, asset and sales growth rates of PE owned portfolio firms exceed those of non-pe owned ones (Boucly et al., 2011) or that industries with PE involvement grow more rapidly in terms of production, value added and employment than those without (Bernstein et al., 2010), very few have distinguished between organic and inorganic sources of such growth. This is especially surprising given anecdotal evidence on the importance of inorganic sources in terms of add-on acquisitions 3 (e.g. Achleitner et al., 2012). Among the very few studies that deal with add-on acquisitions are Nikoskelainen and Wright (2007) who study 321 exited UK buyouts between 1995 and Although primarily focusing on the performance impact of different governance mechanisms, they also find buyouts with add- 1 Support for this was also given by Henry Kravis at the SuperReturn International 2009: You have to accept the fact that transactions will be smaller and have far less leverage - that is a fact. That means that all of us have to adapt. We have to change the way we will do business. If we do not, we will be left out. 2 Unless otherwise indicated, we use the term portfolio firm to refer to a firm that is owned by a PE sponsor. 3 We use the terms add-on, add-on acquisition, intermediate M&A events and inorganic growth (strategies) interchangeably throughout this paper. 2

3 on acquisitions to outperform those without in terms of internal rate of return (IRR). Chapman and Klein (2010) report similar findings when examining performance drivers of 288 US and UK small and middle-market buyouts between 1984 and They even find consolidating add-on acquisitions to be the single most important driver of buyout IRR. Acharya et al. (2013), finally, study 395 Western European buyouts entered between 1991 and Despite addressing the performance impact of different general partner backgrounds in first instance, they also find buyouts with M&A events to have above sector margin and multiple improvements. Moreover, their results suggest that PE sponsors with a background in finance adopt inorganic strategies more successfully than those without. What all these studies have in common is that they either treat add-on acquisitions of portfolio firms as a mere side aspect or link them to deal IRR only. Yet none of these studies explores add-on acquisitions themselves in more detail. Our work aims at closing this research gap by providing empirical evidence for add-on acquisitions of portfolio firms. For so doing, we study three guiding research issues: (i) What is the relevance of add-on acquisitions (Add-on relevance)? (ii) Why and how do PE sponsors execute add-on acquisitions (Add-on strategies)? (iii) What are the determinants of add-on acquisition strategies (Add-on determinants)? To answer research issue (i), we analyze the growth of add-on acquisitions over time in absolute and relative terms. We also examine the type of PE sponsors accounting for most of the add-on activity. Then, for answering research issue (ii), we study operating synergies as motives for add-on acquisitions. We do so by analyzing whether PE portfolio firms carry out add-ons in similar industries. Moreover, we examine whether the first add-on is performed early during the holding period for realizing synergies before exit. We finally investigate research issue (iii) by studying buyout and PE sponsor characteristics as well as economic conditions as potential determinants for add-ons. Basis for our study is a comprehensive data sample that is comparable in size and scope to Strömberg (2008): it contains 9,548 majority-stake buyouts covering the period 1997 to Our data sample is not limited to any world-region, nor is it limited to any deal sizes or types. Furthermore, our data sample is unique in the sense that it contains detailed information on both the buyout and add-on level. 3

4 First, our findings suggest that add-on acquisitions are cyclical and a feature of comparatively few buyouts. In particular, we find that add-on acquisitions are exposed to the same boom and bust cycles as the overall PE market although add-on activity has dropped less during the global financial crisis. We also find that approximately 26% of all portfolio firms conduct at least one add-on. Considering the PE sponsors that backed these deals, we show that only 12% of the PE sponsors account for 80% of all add-on acquisitions and that activity predominantly stems from large and successful PE sponsors. Thus, our findings suggest that particularly such PE sponsors have continuous access to the M&A market and exploit inorganic growth strategies systematically. Secondly, our findings provide insight into the way PE sponsors execute add-on acquisitions. We find approximately 8 out of 10 portfolio companies to perform add-ons in the same industry as they are operating in, indicating that PE sponsors pursue add-ons for exploiting operating synergies and reducing complexity. Furthermore, we find that portfolio companies perform the first add-on after on average 1.5 years, which equals 35% of the holding period in our data sample. The time to the first add-on even reduces in case of several add-ons. We also find weak evidence for decreasing time lags between subsequent add-ons in case of multi-acquirer portfolio firms. We attribute both the early exercise of the first add-on and the declining time lags between subsequent ones to the aim of PE sponsors to realize synergies before exit. Thirdly, we find evidence for systematic determinants at buyout entry that increase the likelihood of add-on acquisitions during the holding period. We consider this to provide preliminary evidence for inorganic growth strategies are deterministic and planned components of the investment case rather than the result of opportunistic behavior during the holding period. Our results show that public-to-private and financial buyouts are significantly more likely to conduct add-on acquisitions indicating that inorganic growth is an important source of value creation in such buyout types. We also find large buyouts and those financed by large and top-performing PE sponsors to be more likely to conduct add-ons suggesting that track-record and sufficient resources provide access to inorganic value creation strategies. This is furthermore supported by the finding that beneficial debt market conditions raise add-on likeliness too. 4

5 Our findings contribute to theory and practice alike. Theoretically, we shed light on relevance, strategies and determinants of add-on acquisitions thus explaining a hitherto unexplored but evidentially important driver of deal IRR (e.g. Chapman and Klein, 2010). Our findings also allow for a deeper and more differentiated insight into portfolio firm growth as we expand evidence on inorganic growth strategies. We therefore add to the growing research stream that revisits the value creation strategies of PE sponsors. Practically, our work equips both general partners and limited partners with empirical insight into the mechanisms of add-on acquisitions thus supporting decision-making on fund and invetsor level. In total, due to the exceptional size and scope of our data sample, our work constitutes one of the most comprehensive studies of the global PE market in recent years also covering the latest financial crisis. We proceed as follows. In section two we provide theoretical considerations and define our hypotheses. We explain the data sample in section three before presenting and discussing the main results in section four. In section five we conclude. 2 Related theory Merger and acquisition (M&A) activity of PE backed portfolio firms is all but a new phenomenon to academic research, but focus has long prevailed on divestitures rather than on add-on acquisitions. Most of the divestiture studies still trace back to the 1980s and 90s and the era of the Corporate Raiders (e.g. Easterwood et al., 1989; Seth and Easterwood, 1993). Since then, however, the market environment for LBOs has changed dramatically and it seems as if add-on acquisitions have superseded divestitures as the predominant source of M&A activity in PE buyouts 4. Theory on add-on acquisitions is still scarce though and likely to differ from existing findings on divestiture activity and motives. Theoretical questions on the relevance, motives, timing and determinants of add-on acquisitions have therefore been unanswered so far. Hereafter we provide the theoretical background for these questions. 4 See Nikoskelainen and Wright (2007), Chapman and Klein (2010), Guo et al. (2011) or Achleitner et al. (2012) for examples for the frequent use of add-on acquisitions. 5

6 2.1 Add-on relevance There are at least three arguments for why add-on acquisitions should have become more relevant over time. First, operational value drivers have played an important role in the strongly competitive environment of the second buyout wave (Achleitner et al., 2011). Since several studies have shown that add-on acquisitions are an important feature of such operational value drivers (e.g. Nikoskelainen and Wright, 2007; Chapman and Klein, 2010; Guo et al., 2011), it is reasonable to assume that add-on acquisitions have also been more frequently employed during the second buyout wave. The second argument is related to that: if the increasing competition and commoditization of the traditional LBO business model have forced PE firms to seek for new value creation strategies, then it is likely that PE firms have relied on measures they are familiar with first. Since PE firms are themselves frequent acquirers, add-on acquisitions might have become a popular source of value generation. Supporting this argumentation, thirdly, several studies have shown that add-on acquisitions are an important source of value generation (e.g. Nikoskelainen and Wright, 2007; Chapman and Klein, 2010). Hypothesis 1: The extent to which buyouts make use of add-ons, has grown over time. 2.2 Add-on strategies Add-on motives Add-on acquisitions are essentially mergers and acquisitions (M&A) executed by PE backed portfolio firms. The motives behind them might therefore be similar to M&A motives in general. Existing M&A literature suggests that synergies are the most important motive for conducting mergers and acquisitions (Healy et al., 1997; Maquieira et al., 1998; Mukherjee et al., 2004) 5. These synergies are often differentiated into operating (e.g. lower costs, higher revenues) and 5 A thorough review of the motives of M&A is provided in Trautwein (1990). 6

7 financial ones (e.g. reduced cost of capital) (Nielsen et al., 1973), of which both could a priori apply to PE backed buyouts. 6 However, several arguments indicate that operating synergies are the more common motive behind add-on acquisitions. Operating synergies are often achieved through cost reduction and revenue increase, both being evidentially important sources of operational value in PE buyouts (Achleitner et al., 2011; Guo et al., 2011). Given the fact that operational value drivers have become more important (Achleitner et al., 2011), operating synergies should have become a more important motive too. Furthermore, Maquieira et al. (1998) find that cost reduction and revenue increase can be best achieved in related industries 7. Portfolio firms might therefore frequently conduct add-on acquisitions in the same industries they are operating in. 8 Hypothesis 2.1: operating synergies. Portfolio firms perform industry-related add-on acquisitions to exploit Add-on timing The buy low, sell high nature of PE buyouts and limited holding period of typically 3 to 5 years (Strömberg, 2008) pressure PE firms to realize value early and fast. This pressure may be even greater in case of sophisticated and time-consuming value creation strategies, such as the achievement of operating synergies through add-on acquisitions. For such synergies to be realized and reflected in the exit price, PE firms may need to ensure that there is enough time for integration of the acquired add-on. We therefore expect that add-on acquisitions are performed 6 An example for the pursuit of financial synergies are roll-up strategies, where competitors are acquired at lower multiples and integrated into the portfolio firm at a higher consolidated multiple. 7 The degree of relatedness between two firms is often measured by comparing the firms primary industries (Singh and Montgomery, 1987; Maquieira et al., 1998). 8 Note that this is not to say there would not be situations where add-on acquisitions are the result of completely other motives. For example, we cannot rule out that add-on activity accelerates at a rather late stage of the holding period in case of unfavorable exit opportunities or prices. Misaligned incentives may compel PE firms to exploit moral hazard and enforce any kind of acquisition, value creating or not, to enhance exit likeliness by all means. Nevertheless, there is a priori no obvious reason for why such a motivation should be more common than the pursuit of operating synergies. 7

8 rather early during the holding period. Pressure to realize synergies early may furthermore be a function of the number of add-on acquisitions that the portfolio firm conducts. We therefore expect that portfolio firms with several add-on acquisitions initiate the first add-on earlier than those without. Hypothesis 2.2: Portfolio firms perform the first add-on acquisition early during the holding period. Portfolio firms with several add-ons initiate the first add-on earlier than those without. Finally, existing literature indicates that PE firms and their portfolio companies benefit from learning effects when conducting several add-on acquisitions during the holding period. Aktas et al. (2009) and Aktas et al. (2013), for example, state that serial acquirers learn along their acquisition experiences. Learning effects may improve bidding strategies and enhance acquisition pace and success, thus potentially decreasing the time between deals within a transaction sequence. On the basis of that, we formulate the following hypothesis: Hypothesis 2.3: Portfolio firms with more than one single add-on during the holding period benefit from a learning effect resulting in decreasing time lags between subsequent acquisitions. 2.3 Add-on determinants Whether add-on acquisitions are improvised during the holding period or deterministic components of the investment case is an open question. Exposure to M&A market conditions and associated cycles may favor the improvisation argument, while the pressure to realize operating synergies early and fast during a limited holding period may benefit the determination argument. Prior research rather points at the latter (Acharya et al., 2013). To verify whether add-on acquisitions are planned components of the investment case, we examine whether there are systematic determinants at entry, e.g. at the time of the buyout, that raise the likelihood of conducting add-on acquisitions during the holding period. A priori, there are several theoretical arguments suggesting that systematic determinants do exist. 8

9 The first set of arguments concerns the status of the portfolio company before the buyout. One important feature of a portfolio firm s pre-buyout status is the former owner and thus the associated entry channel 9. As any entry channel is likely to have distinct needs and value creation strategies, we expect add-on acquisitions to fit best when (i) growth in general is important for value generation and (ii) inorganic growth has not yet or only partly been achieved by the previous owner. These criteria may especially apply to secondary (or higher) buyouts (Financial) since the preceding PE firm might have preferred easy and fast-to-achieve value creation measures first, such as organic growth or traditional sources of PE value creation (e.g. reduction of agency deficiencies; more efficient use of resources etc.). We do not expect add-on acquisitions to play an important role in buyouts motivated by restructuring needs. Add-ons may therefore be less likely in case of Receivership buyouts (Robbie et al., 1993). 10 Another important feature of a portfolio firm s pre-buyout status is its size. From a motivational point of view, small, mid-sized and large buyouts might all have incentives to conduct add-on acquisitions. Small and mid-sized portfolio firms may rely on add-on acquisitions to reach a higher multiple valuation at exit (e.g. Chapman and Klein, 2010), while large portfolio firms may accelerate growth via add-on acquisitions to strengthen their market position in a fragmented industry or to expand into international markets (e.g. Achleitner et al., 2012). Beyond motivation, however, we expect the availability of sufficient resources and capabilities to be distinctive for conducting add-on acquisitions. We therefore expect that the entry conditions of large buyouts increase a portfolio firm s likeliness to conduct add-on acquisitions. Hypothesis 3.1: Add-on acquisitions are more likely for financial and large buyouts, but less likely for receivership buyouts. 9 See Appendix 1 for a concrete definition of the specific entry channels. 10 For other entry channels, argumentation is less clear-cut and may depend on the specific buyout motivation. A frequent motivation for Public-to-private buyouts, for example, is the aim to exploit advantages of privatelyheld firms, such as no disclosure rules and lengthy processes of shareholder approvals, both of which could drastically ease the process of conducting acquisitions for inorganic growth. In some cases they might also be motivated by restructuring needs though (e.g. Muscarella and Vetsuypens, 1990; Liebeskind et al., 1992). Contrary motivations may also apply to Divisional buyouts, e.g. they could be motivated by the aim to relieve growth constraints or by the need to refocus and restructure (e.g. Seth and Easterwood, 1993). 9

10 The second set of arguments for systematic add-on determinants at entry relates to characteristics of the sponsoring PE firms. Since PE sponsors significantly shape the future strategy of the portfolio firm (Alperovych et al., 2013), they might play an important role for the decision to pursue inorganic growth strategies. It is therefore necessary to control for different PE sponsor characteristics. At least three PE sponsor characteristics seem related to a portfolio firm s likeliness to conduct add-on acquisitions: size, performance and experience. Large and reputable PE sponsors, firstly, should have the necessary skill set, track-record, and deal network providing them with access to better deals and the entire M&A market (e.g. Wright et al., 1996; Kaplan and Schoar, 2005; Cressy et al., 2007). Larger PE sponsors can also afford to set up an own operations team, which might ease add-on acquisitions and their integration process (Hemptinne and Hoflack, 2009). 11 Second, we expect successful PE sponsors to rely on add-on acquisitions more frequently since existing research suggests that buyouts with inorganic growth strategies create higher returns than those without (e.g. Chapman and Klein, 2010; Acharya et al., 2013). We finally expect that experience has a positive impact upon add-on likeliness too, since experienced PE sponsors typically excel in identifying favorable market conditions and reacting fast to it (Ljungqvist et al., 2007). They might therefore be able to execute add-on acquisitions before their unexperienced peers thus leaving them few opportunities to conduct similar acquisitions. 12 Hypothesis 3.2: Add-on acquisitions are more likely for large and top performing as well as for experienced PE-sponsors. 11 Based on a recent INSEAD study, 52% of the 25 largest PE sponsors in 2009 had either already established large in-house operations teams, or have been about to develop them. Hemptinne and Hoflack (2009) argue that small PE sponsors can rarely afford operating partners or in-house operations teams. 12 Syndication behavior might be another important entry determinant. The effect on add-on likeliness is unclear though. On the one hand, syndicates might benefit inorganic growth strategies since they are associated with a large deal network (Lockett et al., 2010), diverse skill set (Brander et al., 2002), and risk reduction (e.g. Lerner, 1994; Cumming, 2006). On the other hand, researchers also associate syndicated buyouts with severe agency costs mainly due to free-riding conflicts and information asymmetries (Cumming, 2006; Meuleman et al., 2009). Such free-riding conflicts could indeed be large in case of work-intensive and time-consuming add-on acquisitions. 10

11 The third set of arguments for systematic add-on determinants at entry traces back to industry and overall market conditions. Concerning industry conditions, we expect the degree of fragmentation to be an important driver of add-on acquisitions. Add-ons might occur in both, fragmented and concentrated industries but for different reasons. In very fragmented industries where the majority of firms is typically of small and medium size without significant differences in market share, add-ons might be motivated by roll-up strategies, e.g. as the acquisition of an add-on allows for a higher consolidated multiple. In concentrated industries where few large firms have dominant market share, we expect add-ons to be rather motivated by economies of scale to exert cost advantages. 13 Concerning overall market conditions, we expect money supply and interest rate risk, debt market conditions, as well as GDP growth to influence the likeliness for conducting add-on acquisitions. We proxy money supply and interest rate risk with the LIBOR and debt market conditions with the high yield spread 14. Favorable LIBOR and high yield spread conditions at entry are associated with lower financing costs and may help PE sponsors to negotiate a more ambitious financing package for the buyout (e.g. Axelson et al., 2013). Since the granted financing at entry and associated costs are decisive for a portfolio firms s capacity to conduct add-on acquisitions during the holding period, we expect the add-on likeliness to be higher in case of a low LIBOR and small high yield spread. Finally, we use the GDP growth in the year of the buyout as a proxy for the strength of the overall economy. Given the cyclicality of M&A activity (Mitchell and Mulherin, 1996), a favorable economic environment in terms of GDP growth might increase add-on likeliness too. Hypothesis 3.3: Add-on acquisitions are more likely for buyouts within very fragmented and very concentrated industries and when the economic conditions at entry are favorable. 13 We also introduce the average industry sales growth as a control variable for the specifics of an industry. The impact upon add-on likeliness is unclear though. On the one hand, portfolio firms may pursue inorganic growth strategies in mature industries with overall low sales growth to compensate for lack of organic growth potential. On the other hand, add-on acquisitions might also be a reasonable strategy in industries with already high sales growth, as they allow the PE sponsors to grow the portfolio firm even faster. 14 We use the option adjusted spread of the BofA Merrill Lynch Global High Yield Index. 11

12 3 Data 3.1 Sources and sample construction process We follow a multi-step construction process in order to set up the final data sample. In the first step, we select all buyouts which have been completed 15 between 1 January 1997 and 31 December 2010 using Bureau van Dijk s Zephyr deal database (Zephyr). 16 We include Institutional Buyouts (IBO), Management Buyouts (MBO) or Management Buyins (MBI) for which the financing of the buyout is classified as either private equity or leveraged buyout. By doing this we ensure that we solely include financial sponsor backed buyouts. 17 Similar to Kaplan and Strömberg (2009) we do not include venture capital or other deals, such as private investments in public equity. Our data sample is limited to majority stake buyouts. In the second step, we check each buyout for its status as of 31 December For buyouts where we cannot identify an exit transaction in Zephyr, we use various press sources and PE sponsor websites for determining the status of the portfolio firm. We only keep buyouts for which we can identify the complete buyout cycle, i.e. both the exit channels and exit dates, or which are still within the portfolio of the PE sponsor. The final data sample comprises 9,548 buyouts on a world-wide basis over an entire sample period between 1 January 1997 and 31 December In a third step, we use Zephyr to identify M&A events that have been directly conducted by portfolio firms during PE ownership. Because PE firms occasionally acquire the target on behalf of their portfolio firms and subsequently merge these entities, we check the PE sponsors 15 We do not include deals that were announced but not yet completed. Because Zephyr does not always report the deal status accurately, we manually checked the status of all announced buyouts and included cases in which publicly available information indicates that the buyout was actually completed. 16 Zephyr covers approximately one million completed and rumoured deals of corporate M&A, private equity, venture capital and initial public offerings. In recent years, Zephyr has gained a growing importance among academic PE researchers (e.g. Tykvová and Borell, 2012; Wang, 2012; Knauer et al., 2014). 17 Our methodology is similar to the sampling process of Tykvová and Borell (2012). 12

13 websites (whenever possible) and Zephyr for such indirect add-on acquisitions. 18 We match each M&A event, directly or indirectly executed by the portfolio firm, with the respective buyout of this firm. Thus, our data sample incorporates the entire buyout cycle and all intermediate M&A events initiated during the holding period. In total, we end up with 4,937 add-on acquisitions and 859 divestitures for the 9,548 buyouts contained in our data sample. To our very best knowledge, this data sample is one of the most comprehensive and up-to-date data samples in academic PE research. It covers the last fifteen years of global buyout activity, including the years of the global financial crisis, and all intermediate add-on acquisitions and divestitures at the portfolio firm level. 3.2 Sample selection issues We are aware that our data sample may be subject to data limitations and sample selection issues, as it is commonly the case in PE research. First, similar to Strömberg (2008), information on deal values is not available for about 60% of the buyouts. In order to control for the size effects of buyouts, we estimate imputed deal values for all buyouts with missing deal values using a Heckman (1979) maximum likelihood estimation. 19 Secondly, because our data sample is based on Zephyr, we might slightly overestimate the share of European buyouts compared to buyouts from other world-regions. This may particulary apply to the earlier years of the data sample period. While Zephyr starts to cover deals mainly from 1997 onwards, it focused predominately on European deals in its first years (e.g. Wang, 2012). 20 We regard this potential sample selection bias as rather weak though, especially because the global coverage of Zephyr 18 In rare cases the PE sponsor independently acquires several firms before merging them into a new entity. Thus, whenever possible, we review the PE sponsor s website for such cases. Then we use Zephyr to identify all independent buyouts. We finally assign those buyouts as add-on acquisitions to the portfolio firm initially acquired by the PE sponsor. 19 We follow the methodology of Strömberg (2008). We convert all deal values to their 2012 EUR values to ensure comparability. For each buyout, where the deal value is missing, we use the imputed deal value. The methodology is outlined in Appendix A similar sample selection issue is, for example, discussed by Strömberg (2008) and Kaplan and Strömberg (2009), who based their data sample on the Capital IQ database, for which they report a potential bias for LBO deals outside the US. 13

14 has consistently improved over time. Thirdly, similar to Strömberg (2008), we might miss to include buyouts because of classification issues. This may be the case if Zephyr classifies buyouts as acquisition instead of IBO, MBO, or MBI, for example. We try to correct for potential classification errors whenever possible, although we cannot rule out still missing some of these buyouts. Fourthly, for the sake of consistency, we only include add-on acquisitions and divestitures which can be identified via Zephyr. 21 However, we believe that we include a fairly high fraction of the accurate intermediate M&A activity because the coverage of Zephyr has improved over time and Zephyr is regularly back-filling and updating its database. 3.3 Descriptive statistics Our final data sample covers 9,548 buyouts over an entire sample period ranging from 1997 to In Table 1 we present selected descriptive statistics. Insert Table 1 about here. Nearly 90% of the buyouts were performed after the dotcom bubble of 2000 and 47% were unrealized as of 31 December During the early recovery and the boom period of the buyout market ( ), we cover 5,252 buyouts that are representing 55% of all deals in our data sample. 23 The median (average) holding period for realized buyouts is 3.6 (3.9) years, which is in line with other studies (e.g. Strömberg, 2008; Achleitner et al., 2011). Our data sample is well distributed among entry and exit channels. The main entry channels were private-to-private (43%), divisional (28%) and financial (20%) buyouts, whereas the main exit channels (for realized deals) were trade sale (47%) and financial buyouts (36%). Moreover, 10% 21 Additionally, we might miss to include some intermediate M&A events, if for example the acquirer s and the portfolio firm s name differ, or if the deal status in Zephyr is classified as not completed. Besides the reported name in Zephyr we additionally matched intermediate M&A events on the basis of the Zephyr company identification number. 22 Unless otherwise indicated, we report information for the whole data sample without distinguishing between realized and unrealized buyouts. 23 The fact that most deals in our data sample relate to 2003 onwards, provides evidence for the increasing importance of buyouts as suggested by Strömberg (2008). 14

15 of all realized buyouts ended up in receivership, which is slightly higher than the 6% reported by Strömberg (2008) and the 9% in Achleitner et al. (2011). Our data sample is not restricted to a specific region or country. The majority of the buyouts were performed in the US, UK and Europe (93%), which were the most important buyout markets during that time (Strömberg, 2008). 24 The industry distribution indicates that our data sample is not biased towards specific industries. The top 5 (top 10) industries 25 account for 37% (60%) of the observations. Buyouts were predominantly performed in the business services (11%), construction (7%) and computer software (6%) industries. 26 With regard to the size of the buyouts, the median (average) imputed deal value was EUR 107 million (EUR 241 million). With regard to intermediate M&A events, our data sample comprises 4,937 add-ons and 859 divestitures. A substantial part of the add-on acquisitions was performed during the boom period between 2005 and 2007 (34%). These acquisitions were primarily conducted in the key buyout world-regions the US, the UK and Europe account for 92% of all add-ons in our data sample. The top five (ten) add-on industries account for approximately 46% (68%) of the observations. There is no industry which represents more than 15% of the observations. This indicates that there is no concentration on specific industries. Add-ons in our data sample primarily relate to the business services (15%), computer software (10%) and wholesale (8%) industries. 4 Results and discussion 4.1 Add-on relevance Hereafter we attempt to provide evidence for Hypothesis 1. We split the analysis in two parts. First, we analyze the portfolio firm level in terms of add-ons per portfolio firm and control for 24 Thereof, the most important countries include the US (36%), the UK (20%) and France (10%). 25 We extend the FF 30 industry classification breaking down the services segment into personal and business services. 26 While the relatively high share of computer software deals may seem surprising, this is in line with Strömberg (2008). 15

16 difference related to the sponsoring PE firms. Second, we analyze the timely development of add-ons in absolute and relative terms. Regarding the portfolio firm level, we find 26% of all portfolio firms to perform at least one add-on. 27 Given the growing importance of operating value drivers (Achleitner et al., 2011) and the value creation potential of add-on acquisitions (Guo et al., 2011), this figure seems surprisingly low at first hand. One potential explanation for this rather low percentage could be, that the market for add-ons is restricted to few PE sponsors that have the necessary skill set and resources to conduct add-on acquisitions. We therefore examine the number of PE firms involved into the majority of add-on acquisitions with focus on buyouts backed by PE sponsors ranked among the Top 50 of the annual PEI ranking (PEI 50), to proxy for size, and the Top 20 of the HEC DowJones Private Equity Performance Ranking (HEC-DowJones), to proxy for performance. We find 12% of the PE sponsors to account for 80% of the add-on acquisitions 28 and that PE sponsors among the PEI 50 or Top 20 of the HEC-DowJones frequently employ add-on acquisitions. While such PE sponsors represent approximately 3% of the observations in our data sample, they account for a remarkable 29% of the add-on acquisitions. Our findings suggest that few PE firms make up the market for add-on acquisitions and that especially large and top-performing PE sponsors benefit from the improved access to M&A deals (Wright et al., 1996; Kaplan and Schoar, 2005; Cressy et al., 2007). Regarding the timely development of add-on acquisitions, we find that add-ons grew by on average 32% every year between 2000 and 2007 (see figure 1). Despite declining in 2008 and 2009, add-ons almost recovered to the level of 2006 in the years following the crisis. Insert Figure 1 about here. The absolute numbers suggest that add-on acquisitions have been growing over time. Such 27 This is somewhat higher than in Nikoskelainen and Wright (2007), who find 20% of the portfolio firms to perform add-ons, but considerably lower than in Chapman and Klein (2010) and Guo et al. (2011), who find 56% and 50% of the portfolio firms to employed add-ons, respectively. These differences are not surprising though, given the considerably smaller sample sizes in Nikoskelainen and Wright (2007), Chapman and Klein (2010), and Guo et al. (2011). 28 For syndicated buyouts we allocate an equal share of each add-on acquisition to the involved PE sponsors. 16

17 growth could merely be the result of an increase in the overall buyout market though. To control for this, we also present the number of add-ons in relation to active buyouts in Figure Insert Figure 2 about here. We find that add-ons remained approximately constant as a percentage of active buyouts between 1998 and In the boom period of 2004 to 2007, the relative percentage of add-on acquisitions rose from 9% to a remarkable 14%, before dropping to 7% in light of the Global Financial Crisis. Relative importance of add-ons rebounded to (10%) in These developments suggest that PE buyouts make relatively more frequent use of add-ons in times of boom periods and relatively less in times of bust periods. To compare the cyclicality of add-on acquisitions with the overall buyout market, we also examine the scaled development of both (see figure 3). We observe that buyouts and add-on acquisitions tend to move along similar cycles. A Pearson correlation of 0.9 for the time period between 1997 and 2010 (statistically significant at the 1% level) 31 provides statistical evidence for this. The correlation between add-on and overall buyout market is less pronounced during the times of the Global Financial Crisis though: in 2008 and 2009, buyouts declined to 68% and 34% of their respective base levels whereas add-ons declined to only 86% and 54%. Insert Figure 3 about here. To test for significant differences in the average number of add-ons per buyout across different years, we also employ a one-tailed t-test. We find a statistically significant increase of the average number of add-ons per deal from 0.34 in 1997 to 0.54 in 2007 (at the 1% level), and from 0.29 in 2009 to 0.39 in 2010 (at the 5% level). These results provide further evidence for the growing relative importance of add-on acquisitions over time. 29 We define that buyouts are active if they were entered but not exited in each of the respective years. 30 In 2012, the relative percentage of add-ons declined to 6%. However, note that our data sample covers buyouts entered until 31 December Therefore, this decline might not be representative for the overall market. 31 Our data sample documents buyout entries until 31 December Therefore, the we calculate the correlation coefficient for the time period 1997 to

18 4.2 Add-on strategies Add-on motives Hereafter we attempt to provide evidence for Hypothesis 2.1. We examine how often portfolio firms conduct add-on acquisitions in the same industries as they are operating in, to explore the operating synergy motive. We rely on two alternative industry classifications, e.g. SIC divisional codes and extended FF 30, to ensure robustness. 32 We furthermore control for differences between realized and unrealized buyouts by testing for equality of distributions. The results are presented in Table 2. Insert Table 2 about here. We find almost 8 out of 10 portfolio firms to perform add-on acquisitions in the same SIC industry as they are operating in. Using the much more detailed industry breakdown of the extended FF30 classification, we still find approximately 6 out of 10 portfolio firms to do so. For both industry classifications, the results are robust to the number of add-ons per buyout and the buyout status, e.g. to whether the buyout is realized or not. Our findings suggest that the majority of add-ons are related acquisitions. This finding has several implications. First, the realization of operating synergies seems to be a more common motive than the realization of financial synergies. As this holds true independent of the number of add-ons and the buyout status, operating synergies also seem to be the prevalent motive at any given stage of the holding period. In other words, there is no indication that PE firms desperately perform acquisitions at the end of the holding period without having an operating synergy motive. Our findings furthermore imply that portfolio firms and associated PE sponsors predominantly conduct add-ons in industries they have experience and expertise in. This finding 32 The aggregation level of the industry classification might be important to the result, because the higher the degree of aggregation, the higher the probability to find a match between the portfolio firm and the add-on industry. We control for this effect by also analyzing the rather fine-grained extended FF 30 classification, although the comparison of SIC codes is more common in existing literature (e.g. Maquieira et al., 1998). 18

19 is in line with the recent tendency of PE firms to organize teams around industries and hire professionals with a specific industry focus (Kaplan and Strömberg, 2009) Add-on timing Hereafter we discuss results related to Hypotheses 2.2 and 2.3. We analyze the difference between the date of the entry and the first add-on in absolute terms and relative to the holding period. We differentiate between buyouts with only one add-on and with several add-ons to examine differences in pressure to realize acquisitions early during the holding period. We finally measure the gaps between subsequent add-ons and analyze their developments over time to explore potential learning curve effects. We find the average (median) portfolio firm to perform the first add-on acquisition 541 (416) days after the PE sponsor s entry. 33 This suggests that it takes, on average, more than a year to initiate inorganic growth by means of add-ons. On a holding period-weighted basis, the average (median) portfolio firm completed the first add-on acquisition at 35% (30%) of the holding period. Both findings suggest that exercising add-ons at an early point in the holding period is crucial for benefitting from synergy value at exit. 34 The pressure to exercise add-on acquisitions early might be greater in case of buyouts with several add-ons because integration might otherwise unfavorably prolong the holding period. We therefore split the sample of realized buyouts into two sub-samples. We present the results in Figure 4. Insert Figure 4 about here. 33 For the sake of consistency we only consider realized buyouts in the subsequent analyses. Note that the results are not comparable to the figures presented in Table 3, where we present time lags for portfolio firms with several add-on acquisitions. 34 Interestingly, the timely distribution of the add-on acquisitions is rather widespread. Some acquisitions are even directly performed at entry or close to exit. Performing add-ons immediately at entry could put some buyouts at an advantage in auction bidding processes, because the possibility to incorporate add-ons into the investment case may add a flexibility value. Such additional value may allow for bidding at higher prices, which increases the probability of winning in auction bidding. Performing add-ons close to exit could be the result of unexpectedly well exit conditions that may compel a PE sponsor to sell the portfolio company before originally intended. 19

20 We find portfolio firms with only one holding period acquisition to conduct this add-on at, on average (median), 41% (38%) of the holding period and portfolio firms with several holding period acquisitions to conduct the first add-on at 27% (22%) of the holding period. Both, the averages and the medians are different at the 1% level of significance. 35 Thus, our findings prove that multi-acquirer portfolio firms initiate add-ons much earlier than single acquirer portfolio firms, most likely for avoiding unfavorable prolongation of the holding period. The statistically significant differences across the two sub-samples also suggest that add-on acquisitions are rather a result of planned buy and build strategies at entry than of improvised or opportunistic behavior. To check for learning-curve effects of serial acquirers, we also examine whether time lags between subsequent add-on acquisitions decline in case of portfolio firms with several add-ons (Hypothesis 2.3 ). The results are presented in Table Insert Table 3 about here. Results are mixed regarding the development of time lags between subsequent acquisitions. For most of the add-ons, we find a declining tendency that is especially pronounced for the first acquisitions. Beginning with the time lag between the second and third and ending with the time lag between the fifth and the sixth add-on, there seems to be a plateau effect. A statistically significant decline in time lags is only obvious for the second (at the 1%) and the fifth (at the 5%) add-on. Evidence for consistent learning curve effects through serial acquisitions is therefore rather weak, potentially because of differences in add-on size. 4.3 Add-on determinants In this section, we address findings related to Hypothesis 3.1 to Hypothesis 3.3. For figuring out entry determinants of add-on likeliness, we deploy various multivariate logit regression analyses 35 We employ a t-test assuming unequal variances (Wilcoxon-Mann-Whitney test) to see whether there are differences in the average (median) timing of the first add-on between portfolio firms with a single add-on and those with several acquisitions. 36 We only consider realized buyouts in this analysis. 20

21 on different model specifications 37. We split the analyses into two parts: we firstly examine the dependent variable Add-on activity, which is a binary variable that takes the value of one if the portfolio firm conducted at least one add-on and a value of zero otherwise. 38 We check for robustness to varying industry characteristics and economic conditions. We secondly examine the dependent variable Multiple Add-on activity, which is a binary variable that takes the value of one if the portfolio firm conducted more than two add-ons and a value of zero otherwise. For examining the dependent variable Add-on activity, we provide three alternative model specifications that are depicted in table 4. All specifications show the marginal effects of a logit estimation for the entire buyout sample on the likelihood to conduct add-on acquisitions. We control for buyout and PE sponsor characteristics as well as for buyout world-region, year, and industry. 39 Insert Table 4 about here. Specification (1) includes only buyout characteristics, e.g. the entry type of the buyout. 40 The marginal effects of Public-to-private and Financial buyouts show a significantly positive effect (at the 1% level). Thus, formerly publicly listed and formerly PE owned portfolio firms are significantly more likely to conduct add-on acquisitions compared to Private-to-Private buyouts. 41 The opposite holds true for Receivership buyouts, which are significantly less likely to conduct add-on acquisitions (at the 5% level). These findings have several implications. First, they provide evidence for different use of value creation strategies in primary and secondary (or 37 Appendix 1 contains a detailed explanation of the dependent and independent variables. 38 We include both realized and unrealized buyouts. For every unrealized buyout, we observe M&A activity for at least two years after the entry. This is important to avoid bias, since the analysis of the realized sample shows that the first add-on acquisition is on average conducted during the first 1.5 of the holding period. To ensure robustness, we run the same logit regression on the sub-sample of realized buyouts, finding that the results remain generally robust. The results are available upon request. 39 Buyouts are assigned to world-regions following the methodology of Strömberg (2008). We build year groups of buyouts according the overall buyout activity and the average buyout deal value at entry to ensure the highest similarity within these groups. As to industry classification scheme, we use an extended version of the FF-30 industry classification. 40 We use the entry channel Private-to-Private as the omitted category for the dummy variable. 41 For example, Financial buyouts are about 10.8% more likely to get involved in subsequent M&A activity, all else being equal. 21

22 higher) buyouts. Our results particularly indicate that inorganic growth strategies are more likely for secondary (or higher) buyouts, potentially because they are subordinated to easier and faster-to-achieve value creation levers or because the primary buyout has set organizational preconditions that make the use of inorganic growth strategies more feasible. Our findings also suggest that Public-to-private buyouts might frequently be motivated by relieving growth constraints. In specification (2), we add PE sponsor characteristics to control for a PE firm s size (PE - PEI50 ), performance (PE - HEC-DowJones), experience (PE - experience) and syndication behavior (PE - Syndicated). Findings from specification (1) remain robust to the addition of these variables. We find portfolio firms backed by either a large and reputable, top-performing or experienced PE sponsor to be significantly more likely to conduct add-on acquisitions (at the 1% level and 5% level, respectively) compared to buyouts without such a sponsor. Thus, our findings indicate that PE firms and their characteristics significantly affect strategic decisionmaking of the portfolio firm and access to specific value creation strategies. Our results particularly show that size, track-record and experience of the PE sponsor are crucial to whether a portfolio firm will pursue inorganic growth during the holding period and thus, to whether this value creation strategy is available at all. We extend the analyses by adding the Log Deal Value (imputed) in specification (3). Due to the fact that especially large PE sponsors perform large buyouts, we exclude the PE - PEI50 dummy variable to avoid multicollinearity. 42 Previous findings remain robust and significant to the addition of the Log Deal Value (imputed). Consistent with our hypotheses, large buyouts are more likely to conduct add-on acquisitions. This finding indicates that inorganic growth strategies rely on a large platform (the portfolio company) that has sufficient capacities and resources to integrate one or several add-ons. In any specification, we control for the buyout world region (see Table 4). This is important because add-on activity might be dependent to the existence of a vital M&A market. Our 42 Log Deal Value (imputed) and PE - PEI50 might simultaneously control for the size effect of the buyout. The correlation coefficient between both variables is

23 results show that especially US, European, and UK buyouts are significantly more likely to conduct inorganic growth strategies. We attribute this to the maturity of the buyout industry in these world-regions. 43 To check for robustness to varying market characteristics, we also report findings without industry and time fixed effects in Table 5. We add the Herfindahl-Hirschman-Index (HHI) 44, as a measure of industry concentration, and the average industry sales growth to our logit regressions. In terms of economic environment, we add the GDP growth, LIBOR, and Debt market conditions. Insert Table 5 about here. The variables controlling for the buyout and the PE sponsor characteristics remain robust and significant compared to our previous logit estimations. In addition, we find that add-on acquisitions are significantly (at the 10% and 1% level) more likely within very fragmented (second quartile HHI score) and rather concentrated industries (fourth quartile HHI score). Our results also show that add-ons are significantly more likely for industries with high average sales growth during the last three years before the buyout (at the 10% and 5% level, depending on the specification) and in case debt market conditions are beneficial (at the 1% level) 45. Thus, our findings suggest that add-ons might create value differently depending on the fragmentation of the industry. In concentrated industries, price pressure might require cost effectiveness, which can be best-achieved through purchasing power, cost synergies and economies of scale - all of which are important motivations for consolidating acquisitions; in fragmented industries, add-ons might create value through rolling up a competitor for a higher consolidated multiple. Our results furthermore indicate that add-ons are done to benefit from 43 In addition, we examined the likeliness to conduct add-ons during the peak of the buyout boom era between 2007 and 2008, and in the aftermaths of the financial crisis from 2009 to We find that add-ons are significantly less likely for buyouts entered during these periods. This cannot be observed in the table as the results have been summarized. The full results are available upon request. 44 A detailed explanation is provided in Appendix Buyouts that are conducted during favorable debt markets are about 3.6 to 3.9% more likely to pursue subsequent add-on acquisitions, all else being equal. 23

24 growing industries and that sufficient and beneficial debt financing constitutes an important precondition for inorganic growth strategies. For examining the dependent variable Multiple Add-on activity, we provide the same three alternative model specifications, as depicted in table 6. Differentiation between Add-on activity and Multiple Add-on activity might be important for many reasons, but especially because section 4.3 has shown that inorganic growth strategies are executed differently in case of several add-ons. It is therefore worth examining whether entry determinants are different too. Insert Table 6 about here. We find the impact of entry determinants upon Multiple Add-on activity to be somewhat different to the impact upon Add-on activity and to be more sensitive to alternative model specifications. For example, specification (1) of Table 6 shows that Financial, and Public-toprivate buyouts are significantly more likely to conduct multiple add-ons compared to Privateto-private buyouts (at the 1% level). 46 While this is also the case in specification (2) where we add the PE sponsor characteristics (although at the 5% level only), we cannot find a statistically significant impact in specification (3) where we add Log Deal Value (imputed). We attribute this to the buyout size being potentially more important in Financial and Public-to-private buyouts. Regarding the PE sponsor characteristics, we find the PE - HEC-DowJones dummy variable to be statistically significant at any specification (at the 5% and 10% level). Thus, our findings indicate that especially top-performing PE sponsors are able to incorporate several add-on acquisitions into their investment case, potentially because only such sponsors have the necessary deal flow and capability to identify a number of suitable add-ons at entry and execute them during the holding period. 46 The marginal effects for Privatization buyouts have to be the interpreted with caution, because our data sample only contains 27 Privatization buyouts. 24

25 5 Conclusions While there is consensus that portfolio firm growth has become important for value creation in PE buyouts (e.g. Nikoskelainen and Wright, 2007; Chapman and Klein, 2010; Guo et al., 2011), this study is the first examining inorganic sources of such growth in large scale and detail. Our results show that academic attention to inorganic growth strategies has been overdue because add-on acquisitions have become an important and increasingly relevant feature of the PE market. However, this increasingly relevant feature is not open to all PE firms in the market: we find that only few large and successful PE sponsors account for the vast majority of 4,973 add-ons. The results of this study also provide reasons for this. We find preliminary evidence for the incorporation of add-ons into the investment case already at entry and show that subsequent execution of add-ons is usually done in related industries, most likely to exploit operating synergies, and at a rather early point of the holding period, most likely to benefit from these synergies before exit. Identifying a number of synergy providing add-ons already at entry and performing them early and fast during the holding period, however, requires huge deal network, financial resources and execution capability, all of which might put few large and successful PE firms at an advantage. In that sense, our study also indicates that the gap between the best and the rest might become bigger in a maturing PE market. This study furthermore shows that inorganic growth strategies, despite their growing importance, are not generally suitable to every PE buyout. We find beneficial debt market conditions, a large deal size as well as involvement of a large PE sponsor to increase the likeliness for add-ons during the holding period. This is clearly indicating that the availability of sufficient financial resources is an important precondition for inorganic growth strategies. Additionally, we find systematic differences for add-on likeliness across industries and entry channels. Differences across industries suggest that competitive environment and nature of the business affect motivation for add-ons, e.g. whether cost effectiveness and economies of scale are important or rolling up competitors for a higher consolidated multiple. Differences across entry channels furthermore suggest that the type of value potential left by the previous 25

26 owner determines the feasibility of inorganic growth strategies. That add-ons are significantly more likely for financial buyouts provides further evidence for the different use of value creation strategies in primary and secondary (or higher) buyouts. We believe that this finding indicates that inorganic growth strategies are not only return yielding, but also difficult to realize and time-consuming thus being potentially subordinated to easier and faster-to-achieve value creation strategies. From every point of view, the results of this study provide evidence for how sophisticated value creation in PE buyouts has become in a maturing PE market. This study also provides valuable ground for future research. At least four topics seem worth studying. First, future research should address the impact of add-on size upon acquisition timing, holding period length and learning effects. A second interesting research avenue concerns the impact of opportunistic behavior upon add-on acquisitions. Although we find preliminary evidence for the incorporation of inorganic growth strategies into the investment case at entry, we do not have access to internal documents regarding the intended strategy. Future research should therefore analyze whether suddenly appearing acquisition options during the holding period annul the intended strategy. Third, it seems worth comparing the economic efficiency of inorganic growth strategies to other value creation strategies, e.g. as add-on acquisitions might cause additional direct costs such as advisor fees, which reduce net return, and as they might require additional GP attention and involvement, which create opportunity costs. Finally, future research should examine whether large and successful PE firms have advantages in auction bidding because they are able to identify future add-ons and incorporate them into their investment case. 26

27 Table 1: Descriptive statistics PANEL A: Distribution of buyouts by entry and exit year Entries Exits a Time period No. in % Time period No. in % % % % % , % % , % , % , % , % , % % , % Total 9, % 5, % PANEL B: Distribution of buyouts by entry and exit types Exit Entry channel No. in % channel a No. in % Private-to-private 4, % Trade sale 2, % Divisional 2, % Financial 1, % Financial 1, % Receivership % Public-to-private % IPO % Receivership % Privatization % Total 9, % 5, % PANEL C: Distribution of buyouts by industry and world-region Top 5 industries No. in % World-regions No. in % Business services 1, % US 3, % Construction % Central Europe 3, % Computer software % UK 1, % Fabr. prod % Canada % Wholesale % Eastern Europe % Other 6, % RoW % Total 9, % 9, % Note. Units in this table refer to the number of observations unless otherwise stated. a Our sample also includes unrealized buyouts, i.e. buyouts that were not exited as at 31 December This table shows the distribution of buyouts differentiating between entry and exit years (Panel A), entry and exit types (Panel B) as well as between industries and world-regions (Panel C). 27

28 Figure 1: Absolute development of add-on acquisitions This figure illustrates the development of add-ons between 1997 and The y-axis represents the number of add-ons per year. 28

29 Figure 2: Relative development of add-on acquisitions This figure illustrates the development of add-ons as a percentage of active buyouts between 1997 and Buyouts are defined as active if they were entered but not exited in each of the years under consideration. The y-axis shows add-ons as a percentage of active buyouts. The x-axis represents the years for active buyouts and add-on acquisitions. 29

30 Figure 3: Indexed development of buyouts and add-on acquisitions This figure illustrates the scaled development of buyouts and add-ons between 1997 and The y-axis shows the index points, whereas the x-axis represents the respective years. 30

31 PANEL A: Average matches of SIC divisional codes Table 2: Industry relatedness between portfolio firms and addons Un- Deals with Realized realized Diff. (x) add-ons Obs. a (1) (2) Total (1)-(2) 1 1, more than Total 2, PANEL B: Average matches of extended FF 30 industries Un- Deals with (x) Realized realized Diff. add-ons Obs. a (1) (2) Total (1)-(2) 1 1, more than Total 2, Note. Units in this table refer to the percentage of portfolio firms with add-ons in the same industry (e.g is equivalent to 60%) unless otherwise stated. Differences are statistically significant at the 10% (*), 5% (**), and 1% (***) levels. a This table considers realized and unrealized buyouts with at least one add-on acquisition during the holding period. This table shows the consistency of platform and add-on industries using the SIC divisional codes as well as the FF 30 classification. We test for equality between realized and unrealized buyouts employing a two-sided t-test with equal variances. 31

32 Figure 4: Holding-period weighted timing of the first add-on acquisition (a) Timing for buyouts with a single add-on acquisition (b) Timing for buyouts with more than one add-on acquisition This figure illustrates the timing of the first add-on for realized buyouts. Timing is calculated as the difference between the date of the first add-on and the buyout entry date. The resulting time lag is then divided by the holding period of the buyout. In Panel A we show the distribution for buyouts with a single add-on acquisition. In Panel B we present buyouts with several acquisitions. 32