Do Mergers and Acquisitions Among SMEs Affect the Performance of Acquiring Firms? Spyros Arvanitis*, Tobias Stucki**

Abstract This articles investigates the post-merger performance effects of mergers and acquisitions (M&As) drawing upon a representative sample of all Swiss M&As occurring between 2006– 2008 where the majority took place between small- and medium sized firms (SMEs). Using this data, we were able to investigate the impact of M&As on differing measures of economic performance and also, on innovation for the post-merger period of 2008-2010. We found positive statistically significant performance effects of M&As for three of the five performance indicators - including the measure for innovation performance. Furthermore, there is some evidence which suggests that growth is a principal motive for external mergers whilst efficiency is influential upon internal mergers.

Keywords: mergers and acquisitions; economic performance; innovation performance JEL classification: L20; O31 ________________________ * ETH Zurich, KOF Swiss Economic Institute, CH-8092 Zurich, Phone: +41 44 632 51 68, Email: [email protected] ** ETH Zurich, KOF Swiss Economic Institute, CH-8092 Zurich, Phone: +41 44 632 63 07, Email: [email protected]

1 1

Introduction

There is a broad empirical literature on the determinants, as well as the economic performance of, mergers and acquisitions (M&As); (Agrawal and Jaffe 2000; Kaplan 2000; Martynova and Renneboog 2008; Gugler et al. 2012). These studies largely rely on stock market-based measures of performance (Andrade et al. 2001; King et al. 2004; and Meglio and Risberg 2011). As such the performance of M&As is almost exclusively analyzed for large firms, however, the bulk of M&As are between SMEs; thus, it is unclear whether the implications of these studies are valid for smaller firms (Weitzel and McCarthy 2011). SMEs are different from large firms in several aspects; so for example, they have often a simpler governance structure as the operational manager is also the owner or main shareholder. This difference would directly mitigate the agency problem which normally confronts M&As (Jensen 1986). Moreover, there is a discernibly greater coordination problem in larger firms (Williamson 1975); these are further attenuated in the case of restructuring - as in a merger with a comparably large enterprise. Accordingly, the performance effects of M&As might be expected to be greater for SMEs than for the larger organizations upon which that most previous M&A studies focus. There are further factors that rather restrain the synergy effects of mergers in the case of smaller firms; for example, the fact that such firms would not be able to fully exploit the potential of economies of scale or market power. Furthermore, large firms may have more M&A experience, be more open to other business cultures than owner-led SMEs and/or bring more financial resources and so be able to acquire larger targets. From a theoretical stance, the net effect of all these factors is unclear and thus, clarification depends upon empirical investigation; because of the existing differences between large firms and SMEs, it is unclear whether the results of previous studies with respect to the M&A performance are relevant and applicable to SMEs. Thus, there is a case to investigate the performance effects of M&As for SMEs - and more precisely, M&As in which both parties are SMEs.

2 In this study, we analyze the impact of M&As on performance using a unique dataset for a representative sample of M&As in Switzerland, of which most are SMEs. The sample includes all Swiss M&As that took place in the period 2006– 2008; on average, the sample firms employed 320 people – with 60% having less than 100 and 86% less than 500 employees. Accordingly, we are able to draw conclusions regarding the total number of M&As in the respective period of time and in addition, our survey provides detailed information on the performance and characteristics of M&As. This allows us to investigate the impact of M&As on different measures of economic performance and also, upon innovation. Despite the fact that M&A scholars commonly depict M&As as complex and multidimensional, most studies focus only on few performance indicators (Meglio and Risberg 2011). Furthermore, only a limited number of M&A studies focus on the consequences of M&A on technological activities (see Veugelers 2005; Schulz 2007 for a review of this literature). Given the increasing importance of innovation activities as a driver of growth greater insights into such on the effects of M&A are required (Cassiman et al. 2005; Cloodt et al. 2006). To explore these issues and inform our empirical analysis, we use a matching framework that accounts for endogenous selection. The paper is organized as follows. Section Two presents related theoretical and empirical literature and the research questions. Section Three provides a short descriptive analysis of the data used in the paper. In Section Four, the specification of the empirical models is presented. Section Five deals with the estimation results and Section Six concludes.

2 2.1

Related Theoretical and Empirical Literature Conceptual Background

The controversy in existing literature over the effects of M&A on the post-merger performance of firms has not only an empirical basis, but also reflects conceptual differences. Thus, the literature analyzing the performance of M&As can be divided in two broad branches: the value-increasing efficient-market approach and the value-decreasing agency approach. According to the former,

3 mergers occur because of the possibility of exploiting synergies (between the acquiring firm and the target firm), which in turn enhance the performance of the merged firm (Hitt et al. 2001). Such synergies can be driven by increased market power, economies of scale or financial synergies (Nguyen et al. 2012). The value-decreasing approach explains the performance failure of mergers through the existence of informational and agency problems between management and owners (for an overview of this literature see Parvinen and Tikkanen 2007). Such agency problems arise when self-serving managers pursue excessive growth to promote personal interests (Morck et al. 1990; Wright et al. 2002), or when they make specific investments to enhance the dependency of the firm on their skills even though the acquisitions reduce the value of the firm (Shleifer and Vishny 1989). Furthermore, due to managerial hubris, M&As may negatively affect performance; even if managers aim to increase value, they may over estimate their abilities to create synergies and overpay for targets (Roll 1986; Ben-David et al. 2013). With respect to M&As among SMEs, agency problems that occur with the separation of ownership and control are of less importance given the likely amalgamation of these functions (Schulze et al. 2001). Even where ownership and control are separated, the information asymmetries that cause agency problems are likely to be lesser in SMEs given the proximity of the managerial team. So, reflecting Weitzel and McCarthy (2011) we suggest that value decreasing motives should be of lower relevance for SMEs. There are a number of arguments which suggest that the value-increasing effects are constrained: first, SMEs have fewer resources or knowledge to identify targets with potential for synergies. Second, owner-led SMEs are probably less open to other cultures, for example, with respect to work-related values and management styles. Hence, the willingness to use synergies may be limited. Third, many SMEs lack relevant M&A experience that might enhance their ability to successfully use generated synergies (McDonald et al. 2008). As argued above however, the net effect of all these factors is theoretically unclear and requires empirical clarification.

4 2.2

Empirical Literature

Economic Performance Previous studies use various measures to investigate the impact of M&A on corporate performance such as: market measures (for example, stock price performance), accounting measures (profitability and sales ratios) or mixed measures (for example, operating cash flow and measures of stock price reaction). In a survey of empirical literature on post-merger performance effects in the 1980s and 1990s, Andrade et al. (2001) identified three main outcomes.1 First, the general conclusion from studies that analyzed total factor efficiency and other productivity measures using plant-level input and output data of US firms was that mergers are positively related to productivity improvements at the plant - but not at the firm level (Schoar 2002). Second, a series of in-depth case studies of a small number of mergers commissioned by the NBER could not generate substantial insights into exactly how mergers create value (Kaplan 2000). Third, studies based on stock-price performance yielded abnormal (low) returns for acquiring firms, for which different authors offer divergent explanations (Mitchell and Stafford 2000). In recent surveys of empirical literature linking corporate performance to M&A, Tuch and O’Sullivan (2007) and Ismail et al. (2011) concluded that the effect of mergers and acquisitions on the “abnormal” returns for corporate stock (that is those returns that were significantly above or below the industry average) were mostly inconclusive. Evidence using accounting performance measures was mixed: for example, a negative impact was often found for post-merger sales, but a positive effect for post-merger profitability. More specifically, Tuch and O’Sullivan (2007) reported mostly negative or insignificant effects for short-run event studies (up to three months) and overwhelmingly negative effects for long-run event studies (up to five years). Further, varying results for different industries are evident. Thus, there are still many open questions to be explored in this field.

1

An assessment of the performance effects of M&As in earlier studies is found in Caves (1987).

5 There is a dispute in the literature regarding factors that affect post-merger performance. A series of factors that correspond to characteristics of the merger and/or the acquiring and the acquired firm have been investigated as to their influence on performance: mood of the acquisition (friendly or hostile), method of payment (cash or stock), relative size of acquiring and acquired firm, industrial relatedness of acquiring and acquired firm, pre-merger performance and cross-border versus domestic M&A. To the best of our knowledge however, no empirical study has analyzed the impact of M&As on the economic performance for SMEs. However, some empirical evidence can be found in studies that analyze the impact of firm size on M&A performance however, the average firm size in these studies is significantly larger than in our sample. In addition, recent research exploring Swiss micro data has analyzed the determinants of M&A performance of SMEs but from the perspective of the acquiring firms themselves (Arvanitis and Stucki 2013). In a comprehensive study based on a sample of 12,023 acquisitions by public firms in the U.S.A between 1980 and 200, Moeller et al. (2004) found that the announcement return for acquirers is roughly two percentage points higher for small acquirers - irrespective of the form of financing- and whether the acquired firm is public or private. Further, the size effect is robust to firm and deal characteristics and is not reversed over time. A serious limitation of this study is that it refers only to public acquiring firms. Gugler et al. (2003) in a study based on a large panel of about 45,000 global mergers and acquisitions (about half in the U.S.A) between 1981 and 1998 investigated the performance effects of M&A by sales and profitability. The authors conclude that “one might expect mergers between small firms to be more likely to increase efficiency by creating economies of scale and scope, while mergers between large firms would be more likely to increase market power” (p. 646). These conjectures were supported by findings that sales increased following profitable mergers between small firms and decreased in the case of large firms.

6 To summarize, the empirical evidence is quite ambiguous. While a tendency for negative longterm effects on post-merger performance was found in stock price-based studies, the impacts tend to be positive for smaller firms. Accordingly, we formulate the following research question: Research question 1: Do M&As of SMEs positively affect economic performance? Innovation performance Whilst some studies analyze the impact of innovation activity on the probability of being a merger target (Blonigen and Taylor 2000, Frey and Hussinger 2006, Hussinger 2010), or have investigated the drivers behind the M&A impact on innovation performance (Ahuja and Katilia 2001, Cassiman et al. 2005), the literature exploring the effects of M&A on post-merger innovation performance is rather scarce (Veugelers 2005 and Schulz 2007). There is some analysis conducted upon industrylevel; so for example, Bertrand and Zuniga (2006) explore the relationship between the impact of industry M&A intensity and R&D intensity for OECD countries from 1990 to 1999. On an aggregate level they found no significant impacts but they did, however, find evidence that M&A intensity stimulates an industry’s R&D in low-tech industries but reduces related investment in medium- and high-tech industries. Further light is cast on the relationship between innovation activity and M&As in a recent study based on a large patent-merger US dataset over the period 1984-2006; Bena and Li (2012) show, first, that more innovative companies are more likely to be acquirers and axiomatically, less innovative firms are more likely to be acquired. Second, they find that the existence of technological overlaps has a positive and significant effect on merger formation; third, mergers are more likely between firms with either technological or product market synergies, but less likely when both are present. Finally, it was noted that acquirers with prior technological linkage with the target firms generate higher innovation output after merger. Hence, technological affinity, as well as technological complementarities, seems to be important factors for successful mergers from the point of view of innovative performance.

7 Phillips and Zhdanov (2013) investigated the differences between small and large firms as to the impact of M&A on innovation performance based on data for 11,288 U.S. firms during 1984 2006. The authors showed that incentives to conduct R&D increase with industry acquisition activity and are greater for small firms. At firm level, Danzon et al. (2007) analyzed the impact of M&A on R&D activities of firms in the pharmaceutical/biotechnology industry for the period 19882000; they found that large firms were not significantly different from non-merging firms with respect to R&D expenses in the three years following the merger. Small firms (enterprise value below $1 billion) experienced lower R&D expenses, suggesting that post-merger integration may absorb the financial resources that are required to invest in R&D. Finally, Stiebale (2013) investigated the impact of cross-border acquisitions on domestic R&D expenditures of the acquiring firms based on data for German firms with up to € 500 million annual turnover. After cross-border acquisition, acquiring firms exhibited a higher rate of domestic R&D expenditures than before but this was not the case for domestic acquisitions and green-field direct investments. As neither the theoretical or empirical literature offers a specific hypothesis for the impact of M&A of SMEs on innovation performance, we assume that innovation performance is similarly affected as economic performance. Thus, we formulate the following research question as starting point for our study: Research question 2: Do M&As of SMEs positively affect innovation performance?

3 3.1

Description of the Data Construction of the Data Set

The sample we use in this study refers to the cohort of Swiss M&As that occurred between 2006 – 2008 which was registered by the Swiss Federal Statistical Office and originally contained 2048 firms. We checked the original data in a multi-step process; first, the changes in the firm structure of acquiring firms were verified using the information of the Swiss Commercial Register. A further verification whether these (legal) changes corresponded to real M&A activities was attained

8 through specific questions addressed to the acquiring firms. From the original sample, 413 firms were excluded because (a) the registered M&As were only legal adaptations to already established economic relations; (b) they were non-profit organizations; or (c) they were firms with less than one full-time employee. Further, 237 firms had already left the market in 2011 and could not be contacted. After these adjustments, 1398 firms remained that corresponded to our definition of M&A as, ‘the partial or full merger or acquisition of firms that are legally independent from each other’. This definition covers both external M&As and M&As within the same group of firms (internal M&As).2 Information on these firms was collected in a postal survey on the “M&A of the Swiss Economy” undertaken in spring 2011.3 The survey yielded information on M&A characteristics (number of M&A per acquiring firm, relative size, method of payment, relatedness, motives, obstacles, etc.). In addition, we collected information on innovative activities and some basic characteristics of the firm (employment, firm age, industry affiliation, etc.). Our questionnaire also included information on the development of quantitative measures of economic performance and innovation (sales, value added, investment, innovation expenditures) after the M&A, specifically for the period 2008-2010. The survey yielded data for 405 enterprises, a response rate of 29%, this was satisfactory given the very demanding questionnaire and that not all ‘wrong’ M&As could be identified. Depending on the number of missing values of the model variables between 119 (innovation model) and 275 (employment model) observations could be used to econometrically analyze the impact of M&As on the development of performance. To deal with endogenous selection into M&A a control group is required. The control group of non-merged firms comes from the KOF enterprise panel that is based on a sample of some 6500 firms taken from the Business Register of the Federal Statistical Office. The sample, which covers 2

To make the results comparable with previous studies that focused on external M&As, we alternatively tested our models only for external M&As (see Table A.7). 3 The questionnaire is available in German and French on www.kof.ethz.ch/en/surveys/.

9 the manufacturing industry, construction and the commercial area of the service sector, is stratified according to sectors and sector-specific variables and is adjusted regularly. Information on these firms has been collected in a postal survey on innovation activities undertaken in fall 2011. The survey also included information on M&A activities of the firms; this allowed us to clean up our control group. To avoid performance measures being affected by reorganizations of the firms after 2008, we excluded all those with M&A and/or outsourcing activities4 after this date. Furthermore, the survey included information on the same basic characteristics of the firms that is also available in the M&A survey (sales, value added, employment, firm age, industry affiliation, etc.), which allowed us to specify a selection equation.5 Unfortunately, the innovation survey included information on quantitative measures (sales, value added, investment, innovation expenditures) only for 2010. To attain dynamic information on these quantitative measures for the control group, we matched the information of the innovation survey with the information of a survey on the “Internationalization of the Swiss Economy” based on the same sample of firms and undertaken in spring 2010. This survey included information on the objective measures for 2008, allowing us to construct comparable measures on the development of the objective variables as available in the M&A survey. Due to the fact that some of the firms did not answer both surveys and missing values for some of the model variables, the sample size of the control group of non-merged firms that could be effectively used in our estimations diminished significantly. Depending on the model, between 300 (innovation model) and 776 (value added model) observations could be used. This seems to be satisfactory as the number of firms in the control group for each model is at least 2.5 times the number in the treatment group.

4

This information came from our surveys and included M&A and outsourcing activities that made a difference of more than 10% of previous total sales. 5 The questionnaire is available in German, French and Italian on www.kof.ethz.ch/en/surveys/.

10 3.2

Characteristics of the M&As

Our data is based on the entire population of Swiss M&As in the period 2006–2008. Accordingly, the characteristics of the M&As should offer a description of the average M&As; detailed description of the collected data is presented in Table 1. Most of the M&As took place in the service sector (65%), 27% in the manufacturing sector, the remaining 8% in the construction sector. In the service sector as well as in the manufacturing sector, M&As were equally distributed among the high-tech and the low-tech sub-sector. M&As were relatively equally distributed among industries. Only the industries ‘wholesale’, ‘banks, insurance’ and ‘business services’ had a share of more than 10% of the total sample. The acquiring firms were mostly small firms; 60% of the firms had less than 100 employees and only 14% of the firms employed more than 500 employees. The relative size of the target to acquirer was mostly small; the sales of the target firm made less than 5% of the acquirer’s sales in nearly 30% of all M&As. Only in 1% of all transactions was the target firm was larger than the acquiring firm.

4 4.1

Econometric framework Potential econometric problems

To be able to make a statement on the impact of M&As on performance, we have to overcome two problems (see Egger and Hahn 2010 for a detailed description of these problems in the context of the assessment of the performance effects of M&As). The first is a missing data problem. We have only one observable outcome per firm, either with treatment or without treatment. Thus, it is difficult to assess how a firm with an M&A would have performed without an M&A. The second problem is that self-selection into treatment is usually evident in merger operations; before firms undergo a merger managers usually analyze the costs and benefits of such an operation. Accordingly, selection into treatment should be strongly related to the expected benefits of such an

11 M&A.6 This makes it difficult to identify merger-generated performance effects, for example, it is not advisable to take the mean outcome of non-treated firms as an approximation. There are different solutions to overcome these problems. We apply the matching approach as it seems to be well suited for our data.7 The basic idea of this approach is to compare the average outcome of the treated firms with average outcomes of structurally similar firms that are not treated. To ensure that the matching approach identifies and consistently estimates the treatment effect of interest, two key assumptions are required. First, the ‘conditional independence assumption’ (CIA) implies that, given a vector of observed variables which are not affected by treatment, assignment to treatment is independent of the outcomes. Second, the ‘common support condition’ (CSC) ensures that firms with the same vector of observed firm-specific variables have a positive probability of belonging to both treatment and control group. Due to the selection into treatment, M&As are complex activities that are quite difficult to model satisfactorily. Accordingly, CIA is a strong assumption. Nevertheless, the two key assumptions should not be violated in our case. Due to the matching of different data sets we have a large group of non-merged firms (see Section 3.1), what should ensure that good matches are found for the treated firms. Furthermore, the data set includes detailed information on the internal firm characteristics as well as the external market conditions. This allows us to control in detail for factors that may influence the treatment status and the outcome variables but are not affected themselves by treatment. Nevertheless, we have to state that our selection model is a simplification of the selection process and probably does not control for all drivers of M&As. This problem could be further reduced by combining the cross-section matching framework with a difference-in6

The selection situation is more complex than it is implied in our matching approach. At a first stage a firm has to decide whether it is willing to enter the acquisition market and in second stage, conditional on having chosen to enter the takeover market, it has to choose the firm to be acquired (for example, Hall 1988). In our case we confine our sample to firms which actually made acquisitions because any further information on the first-stage decision of entering the acquisition market is missing. 7 An alternative solution to estimate causal effects would have been to apply an instrumental variable (IV) approach. In contrast to the matching approach, IV estimation can provide consistent results even in the presence of hidden bias. However, this typically comes at the costs of a reduced precision of the estimates and introduces new uncertainty from its reliance on additional untestable assumptions (DiPrete and Gangl 2004).

12 difference (DiD) estimator. The DiD approach uses pre-treatment information to control for the selection on time-invariant un-observables (see Lechner 2010 for an in depth discussion of this procedure). Unfortunately, we do not dispose of information for the pre-acquisition performance of the firms in our sample. However, reviewing the M&A literature shows that this is a common problem (for example, Egger and Hahn 2010). For SMEs it is even more difficult to attain information on pre-acquisition performance. Such information would require balance sheet data for the acquiring and the acquired firm that is usually only available for firms quoted on the stock exchange.

4.2

Implementing the Matching Approach

Due to the high dimensionality of the covariate vector that explains selection into treatment, we use the propensity score matching (PSM) approach introduced by Rosenbaum and Rubin (1983). The idea of this procedure is to match firms based on balancing scores, i.e. on the probability of treatment given a vector of observed characteristics (for a detailed description of this approach see Caliendo and Kopeinig 2008). To evaluate these propensity scores, a selection equation is estimated in a first step. In a second step, firms from the treatment group and firms from the control group are matched whereby the outcome of a treated firm are contrasted with the outcomes of comparison group members. We compare the outcomes for four different performance measures.

Estimating the propensity scores As M&A is a binary treatment, probit models are used to estimate the propensity scores.8 To support the CIA, a broad set of variables is tested as explanatory variables in the selection equation. A first group of variables describes the general firm characteristics (firm age; independency of the firm; education level of the employees; export share; firm size), a second controls for external 8

Alternatively, we also tested logit models. Both models yielded similar results.

13 market conditions (development of the demand; intensity of price competition; intensity of nonprice competition). In addition, we control in detail for regional aspects and industry affiliation. The information for these variables comes from the surveys that are described in Section 3.1. A limitation is that the information mostly refers to the post M&A period. However, as we expect that local market conditions, as well as the structure of the firm, have remained unchanged for the great majority, this should only marginally affect our results.

Matching procedure To capture different effects of M&As, we use different measures of firm performance: gross investment, firm size, sales, value added per employee and sales of innovative products.9 The last variable measures a firm’s innovation performance. To reduce the impact of time-constant unobserved effects, difference in performance over the years 2008 and 2010, for which the respective information is available, is analyzed. If returns of M&As (that took place in the period 2006-2008) were primarily generated before 2008, this might lead to an underestimation of the performance effect of M&As. Nevertheless, we assume that the development of the variables for the period 2008-2010 is a good approximation for the development of post-merger performance.10 For the matching method, we chose the nearest-neighbor matching algorithm without replacement because the distribution of the propensity scores in the treatment, and the control group, is comparable (see the respective graphics in the Tables A.1 to A.5 in the appendix) and the control group is sufficiently large to identify good matches. Finally, we discarded treatment observations where the propensity score was higher than the maximum, or less than the minimum, of that of the controls.

9

Previous literature often used profitability to measure M&A performance. Unfortunately, we do not have such information in our survey. 10 Some robustness tests with respect to this assumption can be found at the end of Section 5.2 where the impact of M&As is separately analyzed for different distances of time between M&A and performance measurement.

14 5

Estimation Results

5.1

Selection Model

Reflecting previous research which uses a matching framework to investigate the performance effects of M&As (see, for example, Egger and Hahn 2010 and Park and Sonenshine 2012), the selection equation includes different variables that describe the firm structure and the external market conditions (see Table 2 for detailed information on variable definition and measurement). These variables are also in accordance with theoretical and empirical findings on the determinants of a firm’s decision to engage with M&As. To reduce the variance of the estimates, potential explanatory variables were discarded if they did not have significant impact; hence, sales share of exports and the intensity of non-price competition were dropped. Due to missing values for the different performance variables, the number of observations used to identify the treatment effect can vary substantially. To increase the matching quality, selection models are estimated for the same set of observations that could also be used to identify the treatment effect. Accordingly, we estimate for each target variable a separate selection model. The results of the final selection models are presented in Table 3. The quality of the selection model is satisfactory for a model based on firm-level survey data. Pseudo-R2 varies between 0.19 and 0.24, log-likelihood values between 131.77 and 252.95. Firm age shows a U-shaped relationship to the probability of treatment (that is, the probability of being involved in an M&A). Thus, young and old firms show a higher treatment probability than middleaged firms. Other attributes of firms that correlate significantly with the treatment propensity are independency and firm size. Furthermore, the M&A probability is affected by external market conditions. A positive demand development and low intensity of price competition11, both increase

11

The negative effect of price competition may be partially driven by reverse causality. However, given the fact that most of the firms in our sample are SMEs and only one third of the firms were direct competitors (this information comes from a specific survey question), we do not expect that the whole effect is driven by reverse causality.

15 the probability of being involved in an M&A. Due to the varying number of missing values for the different performance variables there are some small differences between the models. As mentioned above, propensity score matching is only a valid procedure if the CIA and CSC hold constant. Separate tests on these two assumptions for the different models are presented in Tables A.1 to A.5 in the appendix. As a first step, the success of the matching was tested. Propensity score matching requires that the treatment group and the control group are similar in each aspect. To check this assumption, we first tested whether the mean value of each model variable after matching was the same in the treatment group and in the control group. Based on ttests, the null hypothesis - that the conditions of the two groups do not differ after matching - could not be rejected at the 10% significance level for any explanatory variables in the four models. Furthermore, likelihood-ratio tests indicated joint insignificance of all right-hand variables after matching. In the next step, we graphically checked the CSC. The graphics showed that after dropping treatment observations whose propensity score was higher than the maximum, or less than the minimum, propensity score of the controls (‘off support’ observations), there was an overlap of the propensity scores of the treated and untreated firms for each model. Thus, we can assume that common support is evident.

5.2

Treatment Effect

5.2.1 Basic results Estimations for the average treatment effect for the treated (ATT) are presented in Table 4. Based on our sample of representative M&As in Switzerland, we find that M&A activity only affects specific performance measures. With respect to economic performance, we find that firms with M&As show, on average, significantly larger growth rates of sales and value added per employee. With respect to growth of investment and the number of employees however, no statistically significant differences are apparent. Accordingly, the empirical findings only partially support research question one (as formulated in Section Two). In general, it seems that the value increasing

16 effects of M&As do stimulate production output without affecting employment and investment expenditures. This may be interpreted that the synergy effects of M&As enhance labour productivity, even without additional investment beyond that which comparable firms without M&As have achieved in the same time period. Evidence that mergers of SMEs achieve productivity gains seems to be in accordance with profit-maximizing behavior, which is expected to be prioritized when management and owner functions are not separated. It is a stylized fact that managers, who are not tightly controlled by owners, as it is often the case - particularly in large firms with broadly distributed ownership would rather maximize revenues. Accordingly, it is expected that manager-led firms tend to grow faster than owner-led ones (Czarnitzki and Kraft 2004). Therefore, the small size of the firms in our sample may, at least partially, explain why M&A activities did not affect firm growth in terms of increasing employment, but only productivity. With respect to innovation performance, we identify positively statistically significant larger growth rates of innovative sales for firms with M&As than for those without such activities. Accordingly, the answer to the second research question, ‘Do M&As of SMEs positively affect innovation performance?’ is affirmative. Given the increasing importance of innovation activities as a driver of growth, not only for larger firms but also for SMEs, this is an interesting result as it indicates that M&As between SMEs also exploit synergies that stimulate innovation output.

5.2.2 Robustness tests Entropy balancing Although matching methods are often used in contemporary empirical research, there are some concerns regarding efficacy; the main concern being that balancing checking has to be manually undertaken. As improving balance on some covariates can decrease balance on others, it is challenging to find the correct model specification (Ho et al. 2007). An alternative procedure is entropy balancing, a method based on a reweighting scheme that calibrates unit weights to achieve

17 covariate balance on observables. As entropy balancing directly reweights units appropriately to achieve balance, and at the same time keeps the weights as close as possible to the base weights, it always improves upon the balance obtained by conventional matching procedures and obviates the need for continual checking for characteristics included in the specified balance constraints (Hainmueller 2012). To test the robustness of our results, we alternatively used an entropy balancing procedure based on the same election equation as before; the respective estimation results are presented in Table A.6. The results show that the significance and even the size of the effects are unaffected. Estimation results only for external M&As In this study, we did not distinguish between internal and external acquisitions. As external M&As may be seen as “genuine”, we tested the robustness of our findings in separate estimates only for such M&As. If the firms belonged to the same group of companies before the transaction occurred, it can be expected that the group would have taken advantage of potential synergies between the firms to some extent prior to the merger. Accordingly, the potential for additional synergies would be smaller for internal acquisitions. Thus, we expect larger performance effects for external M&A than those observed for the whole sample. The separate estimation results for external M&As are presented in Table A.7; the results show that our previous findings are robust with respect to the growth rates of sales and innovative sales. In contrast to the results for all mergers, growth of value added per employee is, however, not significantly affected by external M&As. In combination with the findings that in addition to sales growth, employee growth is close to statistical significance, indicates that external M&As do affect firm growth, whereas the positive productivity effect found for all mergers can be traced back to internal M&As. A possible interpretation for this difference could be that growth is a primary incentive for external mergers, whilst efficiency is a key motive for internal mergers. Time distance between M&A and performance measurement

18 The extant literature finds that time distance between M&A and performance measurement could affect the related effects (Meglio and Risberg 2010; Tuch and O’Sullivan 2007). Accordingly, our main results may be driven by the fact that the average time difference in our sample is either too short or conversely, too long. The M&As in our sample took place in different points of time between 2006 and 2008, so that the time distance between M&A and performance measurement is varying. This allows us to separately estimate the treatment effects for the sub-samples of early M&As (before mid 2007) and late M&As (post mid 2007). The results are found in Table A.7. The results show that the effect on the growth of innovation performance does not seem to be affected by the time distance of measurement. While the effect is slightly smaller for early M&As, the effect is statistically significant and positive for both sub-categories. With respect to economic performance measures, the results indicate that economic performance is only affected in an early stage. While we can again identify a statistically significant and positive effect for the two variables measuring growth of sales and value added per employee for the firms with M&As post mid 2007, the effect is clearly insignificant for the firms with M&As prior to mid 2007. Accordingly, it seems that the impact of M&As on innovation is more persistent than that upon economic performance.

6

Summary and conclusions

In this article, we investigated the actual performance effects of M&As based on a representative sample that includes all Swiss M&As which occurred between 2006–2008. To capture different aspects of performance, we used the development of five different performance measures: (a) investment expenditures, (b) number of employees, (c) sales, (d) value added per employee and (e) sales of innovative products per employee. To deal with potential endogeneity problems a matching procedure was applied. We found statistically significant performance effects of M&As for three of the five performance measures: sales growth, growth of value added per employee and sales of innovative products. These results are consistent within an alternative matching-method (entropy balance). No effect

19 could be found for growth in employment growth and gross investment. Thus, the tendency for long-term (one to five years) negative effects on post-merger performance noted in many stock price-based studies (Tuch and O’Sullivan 2007) is not confirmed here. On the contrary, the value increasing aspect seems to dominate the M&A process for SMEs leading to discernible positive performance effects. The fact that at the same time, no effects were found for employment and investment could be interpreted that the synergy effects of the M&As enhanced average labour productivity of the acquiring firm, even without additional investment expenditure. The results for the sub-samples of early M&As (prior to mid 2007) and late M&As (post mid 2007) indicated that effects on economic performance could primarily be traced back to the later M&As. However, the positive impact of M&A on innovation performance was found for both subsamples. This suggests that there needs to be some time distance between the structure change (caused by the M&A) and the post-merger performance measurement for discernible effects to become evident, but also that the innovation effect is of a permanent nature. Further, there is some evidence that suggests that growth and hence, expansion of activities, is a principal incentive for external mergers, while efficiency is the main motive for internal mergers. The study has a number of limitations that should be addressed in future research. First, we do not dispose of information for the pre-acquisition performance of the firms in our sample. Such information would allow us to combine the cross-section matching framework with a difference-indifference estimator and thus to further reduce a potential omitted variable bias. This would increase the credibility of the results. Second, it would be interesting to test, whether our results also hold for other countries. Third, due to the relatively small sample of M&As it is hardly possible to identify the drivers of M&A performance in the current framework. A larger sample of M&A would allow us to analyze the impact of M&As for different sub-samples and thus to get some indications about the drivers of M&A performance.

20 References Agrawal, A., and J.F. Jaffe (2000). ‘The Post-merger Performance Puzzle’, in Advances in Mergers and Acquisitions. Eds. C. Cooper and A. Gregory. CT: JAI Press, 1-42. Ahuja, G. and R. Katila (2001). ‚Technological acquisitions and the innovation performance of acquiring firms’, Strategic Management Journal, 22, 197-220. Andrade, G., Mitchell, M., and E. Stafford (2001). ‘New Evidence and Perspectives on Mergers’, Journal of Economic Perspectives, 15(2), 103-120. Arvanitis, S. and T. Stucki (2013). ‘How Do Swiss Small- and Medium-sized firms Assess the Performance Impact of Mergers and Acquisitions?’, Small Business Economics, forthcoming. Ben-David, I., Graham, J.R. and C.R. Harvey (2013). ‘Managerial Miscalibration’, The Quarterly Journal of Economics, forthcoming. Bena, J. and K. Li (2012). ‘Corporate Innovations and Mergers and Acquisitions’, American Finance Association 2012 Chicago Meetings Paper, Chicago. Bertrand, O. and M.P. Zuniga (2006). ‘R&D and M&A: Are Cross-Border M&A different? An Investigation in OECD Countries’, International Journal of Industrial Organization, 24, 401423. Blonigen, B.A. and C.T. Taylor (2000). ‘R&D intensity and acquisitions in high-technology industries’, The Journal of Industrial Economics, 48(1), 47-70. Caliendo, M. and S. Kopeinig (2008). ‘Some Practical Guidance for the Implementation of Propensity Score Matching, Journal of Economic Surveys, 22(1), 31–72. Cassiman, B., Colombo, M.G., Garrone, P., and R. Veugelers (2005). ‘The Impact of M&A on the R&D process: An Empirical Analysis of the Role of Technological- and Market-relatedness’, Research Policy, 34(2), 195–220. Caves, R.E. (1987). Effects of Mergers and Acquisitions on the Economy: An Industrial Organization Perspective’, in The Merger Boom, Proceedings of a Conference, Conference

21 Series No. 31, Melvin Village, New Hamshire. Eds. L.E. Browne and E.S. Rosengren. Boston: Federal Reserve Bank Boston. Cloodt, M., Hagedoorn, J., and H. Van Kranenburg (2006). ‘Mergers and Acquisitions: Their Effect on the Innovative Performance of Companies in High-tech Industries’, Research Policy, 35, 642-654. Czarnitzki, D. and K. Kraft (2004). ‘Firm leadership and innovative performance: Evidence from seven EU countries’, Small Business Economics, 22(5), 325–332. Danzon, P. M., Epstein, A., and S. Nicholson (2007). ‘Mergers and acquisitions in the pharmaceutical and biotech industries’, Managerial and Decision Economics, 28(4‐5), 307328. DiPrete, T.A., and M. Gangl (2004). ‘Assessing Bias in the Estimation of Causal Effects: Rosenbaum Bounds on Matching Estimators and Instrumental Variables Estimation with Imperfect Instruments’, Sociological Methodology, 34(1), 271–310. Egger, P., and F.R. Hahn (2010). ‘Endogenous Bank Mergers and Their Impact on Banking Performance: Some Evidence from Austria’, International Journal of Industrial Organization, 28(2), 155-166. Frey, R. and K. Hussinger (2006). ‘The role of technology in M&A: a firm level comparison of cross-border and domestic deals’, ZEW discussion paper 06-069. Gugler, K., Mueler, D.C., Yurtoglu, B.B., and C. Zulehrer (2003). ‘The Effects of Mergers: An International Comparison’, International Journal of Industrial Organization, 21, 625-653. Gugler, K, Mueller, D.C., and M. Weichselbauer (2012). ‘The Determinants of Merger Waves: An International Perspective’, International Journal of Industrial Organization, 30, 1-15. Hainmueller, J. (2012). ‘Entropy Balancing: A Multivariate Reweighting Method to Produce Balanced Samples in Observational Studies’, Political Analysis, 20(1), 25–46.

22 Hall, B.H. (1988). ‘The Effect of Takeover Activity on Corporate Research and Development, in Corporate Takeovers: Causes and Consequences. Ed. A.J. Auerback. Chicago: University of Chicago Press. Hitt, M.A., Harrison, J.S., and R.D. Ireland (2001). Mergers and Acquisitions: A Guide to Creating Value for Shareholders. Oxford: Oxford University Press. Ho, D., Imai, K., King, G., and E. Stuart (2007). ‘Matching as Nonparametric preprocessing for Reducing Model Dependence in Parametric Causal Inference’, Political Analysis, 15, 199–236. Hussinger, K. (2010). ‘On the Importance of Technological Relatedness: SMEs versus Large Acquisition Targets’, Technovation, 30, 57-64. Jensen, M. (1986). ‘Agency Costs of Free Cash Flow, Corporate Finance and Takeovers’. American Economic Review, 76(2), 323–329. Ismail, T.H., Abdou, A.A., and R.M. Annis (2011). ‘Review of Literature Linking Corporate Performance to Mergers and Acquisitions’, Review of Financial and Accounting Studies, 1, 89104. Kaplan, S.N. (2000). Mergers and Productivity. Chicago: University of Chicago Press. King, D.R., Dalton, D.R., Daily, C.M. and J.G., Covin (2004). ‘Meta-analyses of Post-acquisition Performance: Indications of Unidentified Moderators’, Strategic Management Journal, 25(2), 187–200. Lechner, M. (2010). ‘The Estimation of Causal Effects by Difference-in-Difference Methods’. Foundations and Trends in Econometrics, 4(3), 165-224. Martynova, M., and L. Renneboog (2008). ‘A Century of Corporate Takeovers: What Have We Learned and Where Do We Stand?’, Journal of Banking and Finance, 32, 2148-2177. McDonald, M.L., Westphal, J.D., and M.E. Graebner (2008). ‘What do they know? The effects of outside director acquisition experience on firm acquisition performance’, Strategic Management Journal, 29(11), 1155–1177.

23 Meglio, O., and A. Risberg (2011). ‘The (Mis)Measurement of M&A Performance – A Systematic Narrative Literature Review’, Scandinavian Journal of Management, 27(4), 418-433. Mitchell, M.L. and E. Stafford (2000). ‘Managerial Decisions and Long Term Stock Price Performance’, Journal of Business, 73, 287-329. Moeller, S.B., Schlingemann, F.P., and R.M. Stulz, R.M. (2004). ‘Firm Size and the Gains from Acquisitions’, Journal of Financial Economics, 73, 201-228. Morck, R., Shleifer, A., and R. Vishny (1990). ‘Do Managerial Objectives Drive Bad Acquisitions?’, Journal of Finance, 45, 31–48. Nguyen, H.T., Yung, K., and Q. Sun (2012). ‘Motives for Mergers and Acquisitions: Ex-Post Market Evidence from the US’, Journal of Business Finance & Accounting, 39(9-10), 13571375. Park, W.G. and R. Sonenshine (2012). ‘Impact of Horizontal Mergers on Research and Development and Patenting: Evidence from Merger Challenges in the U.S’, Journal of Industry, Competition and Trade, 12, 143-167. Parvinen, P. and H. Tikkanen (2007). ‘Incentive asymmetries in the mergers and acquisitions process’, Journal of Management Studies, 44, 759–87. Phillips, G. M. and A. Zhdanov (2013). ‘R&D and the Incentives from Merger and Acquisition Activity’, Review of Financial Studies, 26(1), 34-78. Roll, R. (1986). The Hubris Hypothesis of Corporate Takeovers, Journal of Business, 59, 197-216. Rosenbaum, P.R. and D.B. Rubin (1983). ‘The Central Role of the Propensity Score in Observational Studies for Causal Effects’, Biometrika, 70(1), 41–55. Shleifer, A. and R.W. Vishny (1989). ‘Management Entrenchment: The Case of Manager-specific Investments’, Journal of Financial Economics, 25(1), 123-139. Schoar, A. (2002). ‘Effects of Corporate Diversification on Productivity’, Journal of Finance, 57(3), 2379-2403.

24 Schulz, N. (2007). ‘Review of the Literature on the Impact of Mergers on Innovation’, ZEW Discussion Paper No. 07-61, Mannheim. Schulze, W.S., Lubatkin, M.H., Dino, R.N., and A.K. Buchholtz (2001). ‘Agency Relationships in Family Firms: Theory and Evidence’, Organizational Science, 12(2), 99–116. Stiebale, J. (2013). ‘The Impact of Cross-border Mergers and Acquisitions on the Acquirers’ R&D – Firm-level Evidence’, International Journal of Industrial Organization, 31, 307-321. Tuch, C. and N. O’Sullivan (2007). ‘The Impact of Acquisitions on Firm Performance: A Review of the Evidence’, International Journal of Management Reviews, 9(2), 141-170. Veugelers, R. (2005). ‘M&A and Innovation: A Literature Review’, in Mergers and Acquisitions: The Innovation Impact. Eds. B. Cassiman and M. Colombo. Cheltenham: Edward Elgar, 37-62. Weitzel, U. and K.J. McCarthy (2011). ‘Theory and Evidence on Mergers and Acquisitions by Small and Medium Enterprises’, International Journal of Entrepreneurship and Innovation Management, 14(2/3), 248-275. Williamson, O.E. (1975). Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press. Wright, P., Kroll, M., Lado, A. and B. Van Ness (2002). ‘The structure of ownership and corporate acquisition strategies’, Strategic Management Journal, 23(1), 41–53.

25 Table 1 Sample Composition by Sector, Industry and Firm Size

Industry/sector Manufacturing - High-tech manufacturing - Low-tech manufacturing Food, beverage, tobacco Textiles Clothing, leather Wood processing Paper Printing Chemicals Plastics, rubber Glass, stone, clay Metal Metalworking Machinery Electrical machinery Electronics, instruments Watches Vehicles Other manufacturing Energy Construction Services - Knowledge-intensive services - Traditional services Wholesale trade Retail trade Hotels, catering Transport, telecommunication Banks, insurance Real estate, leasing, computer services Computer services Business services Personal services Education Health, veterinary and social work Sewage and refuse disposal, sanitation and similar activities Recreational, cultural and sporting activities Total

Number 106 51 55 9 3 0 3 3 16 8 2 4 0 9 23 6 8 3 4 3 2 30 257 129 128 57 18 7 24 60 14 21 44 2 2 0 2 6 393

Percentage 27% 13% 14% 2% 1% 0% 1% 1% 4% 2% 1% 1% 0% 2% 6% 2% 2% 1% 1% 1% 1% 8% 65% 33% 32% 15% 5% 2% 6% 15% 4% 5% 11% 1% 1% 0% 1% 2% 100%

Firm size Number Percentage 1-9 employees 53 13% 10-19 employees 32 8% 20-49 employees 88 22% 50-99 employees 66 16% 100-499 employees 106 26% 500 and more employees 58 14% Total 403 100% Notes: Firm size information is based on information for the year 2010. Due to different response rates, the number of observations differs between variables.

26

Table 2 Variable Definition and Measurement Variable Δ(investments) Δ(number of employees) Δ(sales) Δ(value added per employee) Δ(innovative sales) firm_age independent tertiary_share firm_size demand_development price_competition Region dummies Industry dummies

Definition/measurement Change of gross investment between 2008 and 2010, ln Change of the number of employee between 2008 and 2010, ln (Number of employees measured in full time equivalents)

Change of turnover between 2008 and 2010, ln Change of the value added per employee between 2008 and 2010, ln (Value added is defined as sales less the cost of intermediate materials and services)

Change of the sales of innovative products (new product and significant modifications of existing products) between 2008 and 2010, ln Firm age (in years) and square of firm age Firm is not part of a group of companies yes/no Share of employees with a tertiary-level degree Number of employees (2010) (Dummy variables for two firm size classes: (a) 10-99; (b) more than 99; reference group: ‘less than 10’)

Development of a firm’s specific product demand (transformation of a five-level ordinal variable (level 1: ‘strong decrease’; level 5: ‘strong increase’) to a binary variable (value 1: levels 4 and 5; value 0: levels 1, 2 and 3 of the original five-level variable))

Intensity of price competition (five-level ordinal variable: level 1: 'very weak'; level 5: 'very strong')

Dummies for six regions (Espace Midland; Northwestern Switzerland; Zurich; Eastern Switzerland; Central Switzerland; Ticino; reference region: Lac Léman)

Dummies for 25 industries

27 Table 3 Selection Equations for the Different Target Variables Selection equation Firm characteristics firm_age firm_age*firm_age independent tertiary_share firm_size_10-99 firm_size_>99

(1)

(3)

(4)

(5)

-0.016*** -0.016*** -0.015*** -0.014*** -0.019*** (0.003) (0.003) (0.003) (0.003) (0.005) 0.000*** 0.000*** 0.000*** 0.000*** 0.000*** (0.000) (0.000) (0.000) (0.000) (0.000) 0.336*** 0.314*** 0.359*** 0.366*** 0.211 (0.104) (0.100) (0.103) (0.105) (0.166) 0.002 0.003 0.003 0.003 -0.000 (0.002) (0.002) (0.002) (0.002) (0.004) -0.386** -0.439*** -0.377** -0.442** -0.219 (0.183) (0.168) (0.182) (0.186) (0.379) -0.670*** -0.728*** -0.721*** -0.755*** -0.677* (0.189) (0.176) (0.191) (0.194) (0.388)

Market conditions demand_development 0.622*** (0.097) price_competition -0.109** (0.049) Control variables Region dummies yes Industry dummies yes N Log-likelihood

(2)

0.628*** (0.093) -0.106** (0.047)

0.646*** (0.098) -0.104** (0.049)

0.632*** (0.099) -0.104** (0.051)

0.824*** (0.152) -0.214*** (0.079)

yes yes

yes yes

yes yes

yes yes

1028 1144 1042 1015 437 233.45*** 252.95*** 223.83*** 212.52*** 131.77***

0.20 0.20 0.19 0.19 0.24 Pseudo R2 Notes: See Table 2 for the variable definitions; standard errors are in brackets under the coefficients; ***, **, * denote statistical significance at the 1%, 5% and 10% test level, respectively.

Table 4 Treatment Effect for the Treated (ATT) Δ(value added Target variable Δ(investments) Δ(number of employees) Δ(sales) per employee) Δ(innovative sales) ATT t-stat N treated N untreated N off support

-0.093 -0.36 256 765 7

0.001 0.02 275 863 6

Notes: Stata psmatch2 procedure is used.

0.059 2.03 250 787 5

0.051 1.91 233 776 6

3.900 5.16 119 303 15

28

Appendix

29 Table A.1 Test Balancing Property and Common Support of Model (1) Variable Firm characteristics Firm_age Firm_age*firm_age independent tertiary_share Firm_size_10-99 Firm_size_>99

Sample

Treated

Control

t

p>|t|

46.485 47.369 3834 3926.4 .40586 .39914 30.918 30.13 .52301 .52361 .3431 .35193

61.381 49.339 5263.6 3917 .28222 .3691 23.632 27.602 .46778 .50644 .48454 .39485

-5.13 -0.53 -2.75 0.01 3.63 0.67 4.41 1.08 1.49 0.37 -3.87 -0.96

0.000 0.594 0.006 0.989 0.000 0.506 0.000 0.281 0.135 0.712 0.000 0.339

.30284 .50644 4.0322 3.8541

7.68 1.11 -2.98 -0.46

0.000 0.266 0.003 0.647

Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched

Market conditions demand_development Unmatched .56904 Matched .55794 price_competition Unmatched 3.8159 Matched 3.8112 Summary of the distribution of the abs(bias) Mean abs(bias) Unmatched 13.9 Matched 4.4 LR chi2

Unmatched 212.52*** Matched 12.88 Notes: Furthermore, control variables for region and industry affiliation were included. All these variables are not significantly different for the matched treated and control units.

0

.2

.4 .6 Propensity Score Untreated Treated: Off support

.8 Treated: On support

1

30 Table A.2 Test Balancing Property and Common Support of Model (2) Variable Firm characteristics firm_age firm_age*firm_age independent tertiary_share firm_size_10-99 firm_size_>99

Sample

Treated

Control

t

p>|t|

Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched

44.555 44.185 3571.5 3280.2 .37722 .37455 31.517 30.972 .5089 .51273 .34164 .34545

61.313 48.909 5236.8 3973.8 .27115 .36364 23.749 28.335 .46813 .49455 .48088 .41091

-6.29 -1.45 -3.55 -1.31 3.39 0.26 4.94 1.20 1.19 0.43 -4.10 -1.58

0.000 0.148 0.000 0.190 0.001 0.791 0.000 0.231 0.235 0.670 0.000 0.114

.30359 .51273 4.0197 3.88

8.02 1.02 -3.44 -1.14

0.000 0.306 0.001 0.256

Market conditions demand_development Unmatched .56228 Matched .55636 price_competition Unmatched 3.7865 Matched 3.7818 Summary of the distribution of the abs(bias) Mean abs(bias) Unmatched 14.0 Matched 4.8 LR chi2

Unmatched 252.95*** Matched 22.12 Notes: Furthermore, control variables for region and industry affiliation were included. All these variables are not significantly different for the matched treated and control units.

0

.2

.4 .6 Propensity Score Untreated Treated: Off support

.8 Treated: On support

1

31 Table A.3 Test Balancing Property and Common Support of Model (3) Variable Firm characteristics firm_age firm_age*firm_age independent tertiary_share firm_size_10-99 firm_size_>99

Sample

Treated

Control

t

p>|t|

Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched

45.761 46.444 3722 3791.8 .40392 .4 30.558 29.981 .53725 .536 .33333 .34

61.432 51.948 5264.5 4259.3 .28081 .372 23.54 26.576 .4676 .512 .48285 .384

-5.57 -1.54 -3.07 -0.70 3.71 0.64 4.36 1.53 1.94 0.54 -4.21 -1.02

0.000 0.125 0.002 0.485 0.000 0.521 0.000 0.127 0.053 0.592 0.000 0.307

.30114 .548 4.0267 3.788

7.93 0.27 -3.13 0.13

0.000 0.788 0.002 0.896

Market conditions demand_development Unmatched .56863 Matched .56 price_competition Unmatched 3.8039 Matched 3.8 Summary of the distribution of the abs(bias) Mean abs(bias) Unmatched 13.7 Matched 4.9 LR chi2

Unmatched 223.83*** Matched 18.01 Notes: Furthermore, control variables for region and industry affiliation were included. All these variables are not significantly different for the matched treated and control units.

0

.2

.4 .6 Propensity Score Untreated Treated: Off support

.8 Treated: On support

1

32 Table A.4 Test Balancing Property and Common Support of Model (4) Variable Firm characteristics firm_age firm_age*firm_age independent tertiary_share firm_size_10-99 firm_size_>99

Sample

Treated

Control

t

p>|t|

Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched

46.485 47.369 3834 3926.4 .40586 .39914 30.918 30.13 .52301 .52361 .3431 .35193

61.381 49.339 5263.6 3917 .28222 .3691 23.632 27.602 .46778 .50644 .48454 .39485

-5.13 -0.53 -2.75 0.01 3.63 0.67 4.41 1.08 1.49 0.37 -3.87 -0.96

0.000 0.594 0.006 0.989 0.000 0.506 0.000 0.281 0.135 0.712 0.000 0.339

.30284 .50644 4.0322 3.8541

7.68 1.11 -2.98 -0.46

0.000 0.266 0.003 0.647

Market conditions demand_development Unmatched .56904 Matched .55794 price_competition Unmatched 3.8159 Matched 3.8112 Summary of the distribution of the abs(bias) Mean abs(bias) Unmatched 13.9 Matched 4.4 LR chi2

Unmatched 212.52*** Matched 12.88 Notes: Furthermore, control variables for region and industry affiliation were included. All these variables are not significantly different for the matched treated and control units.

0

.2

.4 .6 Propensity Score Untreated Treated: Off support

.8 Treated: On support

1

33 Table A.5 Test Balancing Property and Common Support of Model (5) Variable Firm characteristics firm_age firm_age*firm_age independent tertiary_share firm_size_10-99 firm_size_>99

Sample

Treated

Control

t

p>|t|

Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched Unmatched Matched

45.052 45.37 3887.2 3387.7 .39552 .38655 31.212 30.104 .52985 .52101 .3806 .41176

60.036 48.571 4967.9 3754.2 .33003 .35294 26.224 28.696 .37624 .5042 .59736 .45378

-3.70 -0.67 -1.38 -0.46 1.32 0.54 2.09 0.45 3.02 0.26 -4.26 -0.65

0.000 0.506 0.167 0.649 0.187 0.593 0.037 0.654 0.003 0.796 0.000 0.515

.35644 .63025 4.1089 3.7983

6.21 0.27 -3.66 -0.39

0.000 0.788 0.000 0.694

Market conditions demand_development Unmatched .66418 Matched .64706 price_competition Unmatched 3.7612 Matched 3.7479 Summary of the distribution of the abs(bias) Mean abs(bias) Unmatched 14.6 Matched 4.5 LR chi2

Unmatched 131.77*** Matched 8.43 Notes: Furthermore, control variables for region and industry affiliation were included. All these variables are not significantly different for the matched treated and control units.

0

.2

.4 .6 Propensity Score Untreated Treated: Off support

.8 Treated: On support

1

34 Table A.6 Treatment Effect for the Treated (ATT) based on entropy balancing Δ(value added Target variable Δ(investments) Δ(number of employees) Δ(sales) per employee) Δ(innovative sales) ATT 0.014 0.035 0.057 0.035 3.435 t-stat 0.05 1.01 1.68 1.44 4.35 N treated 256 275 250 233 119 N untreated 765 863 787 776 303

Notes: Stata ebalance procedure is used.

Table A.7 ATT for sub-samples a) External M&As

Target variable ATT t-stat N treated N untreated N off support

Δ(value Δ(number added of per Δ(innovative Δ(investments) employees) Δ(sales) employee) sales) 3.96 0.28 0.07 0.09 0.03 3.38 0.63 1.56 1.73 0.75 99 103 95 85 49 750 831 771 730 285 4 3 5 4 8

b) Early M&As (before middle of 2007)

Target variable ATT t-stat N treated N untreated N off support

Δ(investments) -0.20 -0.46 92 743 1

Δ(value Δ(number added of per Δ(innovative employees) Δ(sales) employee) sales) -0.04 -0.02 0.03 2.93 -0.90 -0.47 0.99 2.65 95 89 85 41 841 769 758 265 2 2 3 6

c) Late M&As (after middle of 2007)

Target variable ATT t-stat N treated N untreated N off support

Δ(investments) 0.27 0.79 163 754 7

Δ(value Δ(number added of per Δ(innovative employees) Δ(sales) employee) sales) 0.03 0.06 0.06 3.61 0.91 1.77 1.70 3.64 176 155 142 69 850 777 766 301 8 9 9 18