Psychic Distance and Serendipitous Value Creation in
Technology Acquisitions: An Empirical Analysis
Master thesis MSc Business Administration – Strategic Innovation Management
January 2017
Supervisor: dr. K. J. McCarthy
Co-assessor: prof. dr. J. Surroca
Sam Krouwer
S2731975
Boterdiep 56a
9712 LR Groningen
samkrouwer@student.rug.nl
Psychic Distance and Serendipitous Value Creation in Technology
Acquisitions: An Empirical Analysis
Abstract - Mergers and acquisitions are commonly used means for firms to obtain resources and
capabilities; however, failure rates are severe. Therefore, the avoidance of failure in M&A receives much attention in literature. For example, many scholars try to explain performance differences by looking at psychic distance or relatedness between the acquirer and target, but these studies produce inconclusive results. So, how about deals that exceed performance? We address this question by examining the role of psychic distance in relation to expectation exceeding value creation in M&A, which is referred to as ‘serendipitous value’. Serendipitous value is a concept introduced in 2004 and frequently used in later studies, but it is never measured on a large scale. Hence, we developed an operationalization of serendipitous value and estimated to whether the probability of its occurrence is influenced by four dimensions of psychic distance (i.e. business, cultural, geographical, and formal institutional distance). These estimations are based on a sample of 2,644 technology acquisitions announced between 2000 and 2015. Our findings suggest that business distance does not matter for the creation of serendipitous value. However, cultural distance positively influences the likelihood of creating serendipitous value, while geographical distance produces negative effects. Finally, formal institutional distance shows mixed effects across its sub-dimensions. In sum, we conclude that serendipitous value occurs more often than one might think and that the explanatory power of psychic distance can only be observed when looking at individual dimensions.
Keywords: M&A, Technology Acquisition, Psychic Distance, Serendipitous Value
INTRODUCTION
Firms increasingly conduct mergers and acquisitions (M&A) to grow their organization and gain access to valuable resources and capabilities. While M&A is a relatively fast method for acquiring resources compared to developing them internally, many acquisitions still fail to create value (Harrison, Hitt, Hoskisson, & Ireland, 2001). This ‘performance paradox’ has proven to be of great interest for both managers and scholars.
Graebner (2004) addresses this issue by distinguishing two separate forms of acquisition performance. The expected value of a deal reflects the primary objective of the acquiring firm to conduct the deal and to what extend this objective is achieved. Serendipitous value (hereafter SV) entails the fruits of a deal that were not anticipated by the buyer prior to the deal. While the former has received much academic attention (e.g. Hagedoorn & Duysters, 2002), the latter form of value is relatively underexplored. Organizational serendipity is undervalued by scholars because business theory prefers control, order, and predictability. Also, organizational serendipity is often confused with dumb luck. However, serendipity is distinct from dumb luck, since it is a combination of luck, effort, alertness, and flexibility (Cunha, Clegg, & Mendonça, 2010), which implies that it can be controlled and predicted to some extent.
SV can take various forms, such as new product ideas, new processes, and unexpectedly useful technologies (Graebner, 2004). It is said that sources of SV can be recognized by sophisticated buyers; however, they cannot predict the form in which it will occur (Graebner, Eisenhardt, & Roundy, 2010). Hence, the primary mechanism through which SV occurs is information asymmetry between the acquirer and target since it makes it difficult to predict the true value of the target’s resources prior to the deal, especially in technology deals, since the potential value of the target’s assets resides within tacit and socially complex knowledge (Coff, 1999; Gerbaud & York, 2007).
Information asymmetry, in turn, is fueled by the distance that exists between the acquirer and the target, as it determines to what extent the recourses and capabilities of both firms are related (Evans & Mavondo, 2002; Nooteboom, Van Haverbeke, Duysters, Gilsing, & Van den Oord, 2007). Distance can take various forms, such as distance in terms of industries, geography, or national culture. These forms of distance influence the value and transferability of resources as well as the ability to combine resources.
A frequently used concept in international business literature is psychic distance, which is often seen as a liability in doing business abroad, as it makes international business complex and costly (Cai & Sevilir, 2012; Estrin, Baghdasaryan, & Meyer, 2009). Psychic distance is introduced by the Uppsala school as “the sum of factors that complicate the free flow of information” (Johanson & Wiedersheim-Paul, 1975: 307). These factors could be, for example, geographical distance, technological diversity, language barriers, regulatory diversity, or cultural distance.
buyer and seller in four dimensions: (1) business; (2) national culture; (3) geography; and (4) formal institutions. For certain dimensions, distance between acquirer and target can be a source of information asymmetry, which is a necessary ingredient for SV creation (Graebner, 2004). Hence, we expect that the demarcation line between psychic distance being an asset versus being an obstacle might, for certain dimensions, correspond with the distinction between expected value and SV creation in technology acquisitions. Therefore, the goal of this study is to consider: What is the influence of psychic distance on
the likelihood of creating serendipitous value in technology acquisitions?
To do so, we constructed a proxy for SV and tested four hypotheses regarding the dimensions of psychic distance on a sample of 2,644 technology acquisitions. The findings suggest that SV is a phenomenon that commonly occurs in technology deals and that the effect of psychic distance differs across its dimensions. Cultural distance and formal institutional distance in terms of regulatory efficiency increase the likelihood of creating SV, while geographical distance and formal institutional distance in terms of government size decrease the probability that a technology deal creates SV.
The current study contributes to the existing literature in three ways. First, the current study shows that the effect of psychic distance cannot be observed in an aggregated way. Also, it improves our knowledge about which dimensions remain a liability and which are an asset in relation to SV, thereby contributing to international business literature on the ‘psychic distance paradox’ (e.g. Dikova, 2009; O'Grady & Lane, 1996). Second, the concept of SV is frequently used in conceptual papers (e.g. Cunha et al., 2010; Graebner, 2004; Weber et al., 2011), but we are the first that make an attempt at measuring it on a large scale and prove its existence. Although our proxy has some disadvantages, such as the difficulty to isolate the effect of the deal, it could be a starting point for future work to develop and improve ways to measure SV. Third, our findings raise questions about the weight M&A literature attaches to the event study as a performance metric for M&A. The cumulative abnormal return is often argued to equal value creation, while in fact it only equals stock market expectations (Papadakis & Thanos, 2010). In addition, we contribute to managerial practice in two ways. First, by giving insights in what could support and hamper the likelihood of finding SV in technology acquisitions, as it is important for managers to acquire those targets that could enhance the knowledge base of their organizations (Vermeulen & Barkema, 2001). Second, our study could be an eye-opener for managers that see psychic distance solely as a hurdle. This is important because psychic distance impacts performance through managerial decision-making (Azar & Drogendijk, 2014).
THEORY AND HYPOTHESES
General background
Serendipitous value. Mergers and acquisitions are important means for firms to obtain necessary
resources to create value. The value created from those resources can be either positive or negative. When the resources of the target appear to be more valuable than expected prior to the deal and the acquirer is able to appropriate it, we speak of SV. SV is defined by Graebner’s (2004: 756) as: “value that was not anticipated by the acquirer prior to the transaction”. The concept reflects a form of value resulting from an acquisition, beyond expected value. SV can occur in various forms, such as new product development processes, strategic renewal, or unexpected recombination of technology (Graebner et al., 2010).
For such ‘happy accidents’ to occur, there must be a certain level of uncertainty about the value of the target’s resources (Carayannis, 2008). This information asymmetry is the mechanism through which SV can occur. In turn, SV particularly occurs when a deal involves transfer of tacit knowledge. Hence, SV predominantly occurs in technology intensive acquisitions (Graebner, 2004), because, in general, those deals involve more uncertainty about the true value of the seller’s resources (Gerbaud & York, 2007). In addition, it is said that the true value of knowledge resources can only be assessed when those resources are employed (Wiklund & Shepherd, 2009).
Cunha (2005) argues that serendipitous discovery requires leaning and analysis. While SV is unexpected in nature, it requires deliberate exploration efforts. Through exposure to seemingly unrelated ideas and practices a firm could establish connections between knowledge that did not exist before (i.e. bisociation). Hence, for an opportunity to create SV to exist, it is necessary that the buyer and the seller are distant from each other in terms of ideas, knowledge, and practices. This distance, in turn, is a source of information asymmetry. In addition, differences in terms of technological knowledge (i.e. cognitive distance) are argued to be curvilinear related to innovation (Nooteboom et al., 2007). The upward slope of innovation performance is the result of the novelty value of the counterpart if firms are cognitively distinct. In turn, the downward slope is the effect of absorptive capacity, which requires some overlap in knowledge bases. Hence, a certain level of distance is necessary to create significant value.
Psychic distance. Psychic distance is a concept from international business literature that is used
to describe the internationalization process of a firm, as it entails differences between a firm’s home country and a foreign market. Johanson and Vahlne (1977) developed a model of the internationalization process of a firm. Their work on internationalization, together with the work of some other Scandinavian scholars (e.g. Johanson & Wiedersheim-Paul, 1975) is often referred to as the Uppsala school.
1997). Johanson and Vahlne (1977) argue that internationalization is a process of incremental steps to obtain knowledge about foreign markets and overcome psychic distance. In other words, it is designed as a gradual process of small investments to learn about the local circumstances to decrease uncertainty involved with market development. Johanson and Wiedersheim-Paul (1975: 307) define psychic distance as “the sum of the factors that complicate the free flow of information.” These factors are, for example, geographical distance, language barriers, and cultural distance. However, we adopt a slightly different definition, namely: “the distance between the home market and a foreign market, resulting from the perception of both cultural and business differences” (Evans & Mavondo, 2002: 517). This definition is, in our opinion, more accurate, as it points to the perception of the decision maker rather than treating psychic distance as an absolute fact. Also, Evans and Mavondo (2002) claim that the true influence of psychic distance cannot be properly observed when analyzed on an aggregated level. Following their example, we divide psychic distance in four dimensions: (1) business distance; (2) cultural distance; (3)
geographical distance; and (4) formal institutional distance.
Most studies on psychic distance conclude that it impedes firm performance. For example, Hutzschenreuter, Kleindienst, and Lange (2014) find that geographical, governance, and cultural distance negatively impact firm performance in international expansion. In the context of M&A, literature generally does not use the terminology of psychic distance; however, the different dimensions of psychic distance are often studied. For example, Datta and Puia (1995), found that acquisitions with great cultural distance create less shareholder value than acquisitions involving culturally related firms. Additionally, McCarthy and Aalbers (2015) found that geographical distance negatively impacts post-acquisition innovation performance in technology acquisitions. However, they also propose a negative effect of cultural and institutional distance on innovation performance, but did not find any effect for those dimensions.
While the majority of international business and M&A literature treats psychic distance as an obstacle, some propose that it can be an asset in specific situations. Ambos and Håkanson (2014) state that the predominant idea of distance being an ‘obstacle’ might overlook the potential benefits of being different. For example, O'Grady and Lane (1996) refer to a ‘psychic distance paradox’ as their findings indicate that investments in a neighboring country does not lead to better performance than entering a more distinct country. Additionally, Evans, and Mavondo (2002) find that psychic distance in terms of culture and business differences leads to better firm performance for international retailers.
positively impacts M&A performance in two ways: (1) prior to the deal as awareness of cultural differences leads to stricter target selection criteria; and (2) post deal through diversity in organizational strength. Zhu, Xia, and Makino (2015) found that when institutional distances and language differences increase, it is likely that IT service M&As create more value than manufacturing M&As. Their reasoning behind these findings is that distinct national environments shape processes, routines and resources in idiosyncratic ways. This is of value for IT service firms as they need novel ideas and knowledge to create value, while it complicates the integration of manufacturing firms as differences harm the ability to achieve standardization and efficiency synergies. This indicates that the role of distance in M&A is context specific; therefore, we examine how it relates to SV creation in technology deals.
Hypotheses development
To find sources of value that yield unexpected opportunities, there must be sufficient degree of novelty in the resources and capabilities (Nooteboom et al., 2007) of the target compared to those of the buyer. In addition, the idiosyncrasy of a firm’s resources is partly shaped by the characteristics of the industry (McGahan & Porter, 1999), the national culture (Morosini et al., 1998), and host country regulations (Meyer, Estrin, Bhaumik, & Peng, 2009). In contrast, distance increases complexity in M&A, which increases transaction and integration costs (Hutzschenreuter et al., 2014), and decreases financial and innovation performance (Chao & Kumar, 2010; McCarthy & Aalbers, 2015). Based on the divergent ideas about the role of distance, we strongly believe that the effect of psychic distance is not as unambiguous as early international business literature suggests (e.g. Johanson & Wiedersheim-Paul, 1975). Hence, we treat psychic distance as a multidimensional construct and developed hypotheses for business, cultural, geographical, and formal institutional distance by drawing on diverse literatures.
Business distance. Business distance is defined as the extent to which acquirer and target can be
expected to be dissimilar. Business distance between acquirer and target is reflected in the similarities of their respective industries, since the greater the difference between industries of two firms, the greater the likelihood that they are dissimilar in terms of resources, processes, and capabilities (Haleblian & Finkelstein, 1999).
business distance to be different in technology acquisitions, because of their exploratory nature (Cunha, 2005). Other studies find a curvilinear relation with financial performance because low levels of difference yield no opportunities to learn and high levels come with high costs of complex integration (Pehrsson, 2006). This is in line with the logic of novelty value as described by Nooteboom et al. (2007). The novelty value of a target’s knowledge base is greater when they have distinct processes, capabilities, and routines (i.e. cognitive distance). However, greater cognitive distance impedes the firm’s ability to absorb new sources of knowledge. Therefore, Nooteboom et al. (2007) argue that there is a certain optimal level of distance. In addition, knowledge resources are heterogeneously distributed among firms, which is partly caused by the business the firm is in (DeCarolis & Deeds, 1999). For instance, the knowledge base of a biotechnology firm is formed by primary research, while a pharmaceutical firm’s knowledge base is more related to the application and commercialization of technologies (Pisano, 1994).
Following DeCarolis and Deeds (1999) we assume that the knowledge bases of the acquirer and target are substantially different when business distance is great. In addition, a completely distinct knowledge base has a lot of novelty value, but is hard to internalize (Nooteboom et al., 2007). Based on this logic, we expect that, like cognitive distance, a certain optimal business distance exists in relation to SV. Therefore:
Hypothesis 1: A moderate level of business distance yields the highest likelihood of creating serendipitous value
Cultural distance. Another dimension of psychic distance is cultural distance between the
acquiring and target firm. Cultural distance is defined as the degree to which normative rules differ between the home countries of the acquirer and target (Morosini et al., 1998).
M&A research is inconclusive regarding the role of cultural distance for performance. Some studies argue that cultural distance is a bottleneck for cross-border M&A because hampers effective integration to achieve the proposed synergies (Zaheer, Schomaker, & Genc, 2003). Also, the complexity of integration impedes the acquirer’s ability to create intended economies of scale and scope (Cartwright, 2002). In contrast, research that focuses on acquisitions as a means of exploration find that cultural distance yields potential for learning. For example, Morosini et al. (1998) find that national culture shapes a firm’s resources and capabilities in an idiosyncratic way, which creates value for overseas counterparts. This mechanism increases the cross-border acquisition performance. Also, Cloodt, Hagedoorn, and van Kranenburg (2006) find a positive relation between cultural distance and post-acquisition innovation performance, because cultural distance forces the acquiring firm to critically asses its innovation strategy. Another mechanism through which acquisition performance can be enhanced is the application of better target selection criteria when the acquirer perceives great cultural differences (Chakrabarti et al., 2009).
While the results regarding the role of cultural distance seem to be opposing, it may be dependent on whether the deal is exploratory or exploitative in nature. Nielsen and Gudergan (2012) find that cultural distance decreases performance of exploitative alliances. In contrast, they find that cultural distance does not matter for exploratory alliances. Additionally, M&A studies that find a negative effect of cultural distance on M&A performance focus mainly on exploitative motivated deals (e.g. Catwright, 2002; Zaheer et al., 2003), while the studies that find a positive influence of cultural distance look primarily at exploratory motivated acquisitions (e.g. Cloodt et al., 2009; Morosini et al., 1998). Therefore, similar to the alliance context we expect that the nature of the deal explains the differences regarding the role of cultural distance.
Zooming in on the phenomenon of this study, one can conclude that it is rather exploratory, because deliberate searching behavior is a prerequisite for SV to occur (Cunha, 2005). In addition, technology acquisitions are generally focused on exploration while non-technology deals tend to have exploitative motivations (Puranam, Singh, & Zollo, 2006). As the former is the context of this study, we expect that cultural distance increases the likelihood of creating SV, because it increases opportunities for learning. This logic leads to the following hypothesis:
Hypothesis 2: Cultural distance positively influences the likelihood of creating serendipitous value
Geographical distance. The third dimension of psychic distance that we investigate is
geographical distance. Geographical distance is defined as the distance between the cities in which the
contracting (Cai & Sevilir, 2012; Chakrabarti & Mitchell, 2013). In addition, it is argued that close geographic proximity of acquirer and target eases the post-acquisition integration phase (Graebner, 2004; Hutzschenreuter et al., 2014; Krug & Hegarty, 1997). Next to these economic arguments, geographic distance hampers effective knowledge transfer between acquirer and target, especially when it concerns tacit or socially complex knowledge (Ambos & Ambos, 2009; Bertrand & Zuniga, 2006). The inability to properly internalize the tacit knowledge-based resources of the target impedes the acquiring firm’s ability to find unexpected sources of value. McCarthy and Aalbers (2015) underline this logic, since they found that a 1,000 kilometer increase in distance between acquirer and target in technology deals decreases the patent production by one. Furthermore, learning is a prerequisite for SV to occur and geographical distance strongly impedes the ability to acquire knowledge (Torre, 2008). Therefore, we expect that the negative effects of geographical distance will affect the probability for creating SV. Hence:
Hypothesis 3: Geographical distance negatively influences the likelihood of creating serendipitous value
Formal institutional distance. The last dimension of psychic distance that is studied is formal
institutional distance. Based on North (1990), we define formal institutional distance as the difference in
the structure of formal regulations of the home countries of the acquiring and target firm that govern economic exchanges. In the context of cross-border M&A it is said that distinct laws in the host country hamper replication of existing business practices to that country (Estrin et al., 2009). These authors also found that greater formal institutional distance increases the choice for a greenfield investment as the value of potential targets is decreased because of local restrictions on reorganizing that business. Additionally, Dikova, Sahib, and Van Witteloostuijn (2010) argue that institutional distance adds environmental complexity that increases transaction costs, therefore decreasing the likelihood and pace of deal closure. Also, formal institutional distance is said to increase the so called “liability of foreignness”, as uncertainty about prices, tax rates, and intellectual property rights increases the risk for opportunistic behavior, therefore increasing transaction costs (Eden & Miller, 2004). In addition, transaction costs are expected to be even higher when the desired resources involve tacit knowledge, as their transferability is uncertain and contracts cannot be effectively written and monitored (Meyer et al., 2009).
There is not much research that argues and find a positive impact of formal institutional distance on performance. Most of the international business literature takes a transaction cost perspective, by arguing that differences in rules that govern business increase complexity. This complexity impacts the entry mode of the firm (Estrin et al., 2009), hampers progress in the pre-deal phase (Dikova et al., 2010), and impedes firm performance (Chao & Kumar, 2010). In addition, some studies describe how greater formal institutional distance requires more learning to adjust to local regulations (e.g. Evans & Mavondo, 2002; North, 1990); however, there is not much written on the impact of that process on performance and innovation. Therefore, we expect the complexity logic to be the best reflection of reality in the context of SV creation in M&A. Therefore, we hypothesize:
Hypothesis 4: Formal institutional distance negatively influences the likelihood of creating serendipitous value
METHODS
As suggested by Van Aken, Berends, and Van der Bij (2012), we adopt a theory testing approach since the business phenomenon (SV) is theoretically conceptualized. However, the existence of SV is, to the best of our knowledge, not yet examined on a large scale.
Research setting
We only include public acquirers, since the current study is dependent on information about the acquirer to predict its probability to create SV. The sample is not restricted to a certain region, as we expect that serendipity can be found everywhere. We retrieve the M&A information from Thomson Reuters SDC Platinum M&A database. Other data sources (e.g. Thomson Reuters Datastream, Bureau van Dijk Orbis, the Hofstede Centre, Bing Maps, Index of Economic Freedom) are used as well to obtain necessary data on the acquiring and the target firm. For clarity, an overview of the used data sources and measures per variable is provided in appendix B.
Data collection and sample
We have included all deals in SDC Platinum from the 54 selected industries (see appendix A) by public acquirers that were announced between January 2000 and November 2015 in the sample. We used a relatively large timeframe to ensure inclusion of enough deals that outperform expectations. Only deals before November 2015 were included, because we needed data on the value of the acquiring firm one year after the deal for the proxy of realized performance. Furthermore, we only included transactions that (1) are fully completed; (2) are conducted by public acquirers; (3) had a deal value of more than USD 10 million; (4) concerned the purchase of 100% of the target’s shares; and (5) were not repurchases of own stocks, spinoffs, or recapitalizations. After applying these criteria, the dataset consists of 4,672 deals.
Acquisition data was extracted from the SDC Platinum M&A database as it provides detailed information (e.g. deal value, SIC-codes, SEDOL-codes, payment method, location of the firms, etc.). By using the SEDOL-codes (i.e. Stock Exchange Daily Official List), most financial data was obtained through Thomson Reuters Datastream. However, the SEDOL-codes of some of the acquirers were missing. After manually searching for the missing codes1, 204 codes were still missing. After eliminating
those deals from the sample, the sample size remained 4,468.
We used an event study as a building block for the measurement of the dependent variable, therefore we needed the Total Return Index in the estimation window and event window for all the indices the acquiring companies are listed on. However, for 48 deals, the Total Return Index was missing for the reference market in the event window. Therefore, these deals are eliminated from the sample. Furthermore, we had to drop an additional 231 deals since the available stock price data did not cover the entire event and estimation window used for the event study. Additionally, we dropped 63 deals of which data on all six cultural dimensions was lacking. Then an additional 1,482 deals were removed from the sample, because all deals of which the acquiring firm has closed more deals in the previous or subsequent year were removed. This was necessary, as it interferes with the accuracy of our measurement of SV. After these refinements, the final sample consists of 2,644 deals.
Since the validity of the model depends on the sample selection (Heckman, 1977), we tested for selection biases. The t-tests comparing the means of the deals included and excluded indicate that no significant differences are found for deal value (! = 1.02, (. ).), percentage of cross-border deals (! = −0.41, (. ).), geographical distance (! = −0.73, (. ).), cultural distance (! = −1.48, (. ).), and prior performance (! = −0.60, (. ).).
Measurements
Dependent variable. While some scholars conceptualize serendipity in value creation (e.g.
Cunha, 2005; Graebner, 2004), we did not find any paper that quantified it. Therefore, we attempted to construct a proxy for the dependent variable in this study serendipitous value. In contrast to expected value, SV in M&A refers to value that was not anticipated prior to the transaction (Graebner, 2004). Also, as SV can take various forms, it is difficult to find a proxy that reflects all potential forms in which it can occur. However, all forms of SV have one thing in common: its occurrence cannot be predicted up front.
We used stock prices and market values of the acquiring firm to build the proxy for SV for two reasons. First, it is even more difficult for the stock market to predict sources of SV prior to the deal (Graebner et al., 2010; Healy & Palepu, 2001), so the value they attach to the firm reflects the unpredictability of SV. Second, the stock market is expected to react to most forms of SV when they occur, such as new patents (Hirschey & Richardson, 2004), renewed supply chain management processes (Filbeck, Gorman, Greenlee, & Speh, 2005), or enhanced innovation capabilities (Annavarjula & Mohan, 2009). Furthermore, SV reflects additional value that is created through an acquisition beyond expectations (Graebner, 2004). In other words, SV is a positive difference between realized value and expected value. However, as SV is never quantified before, we treat it with caution and create different definitions by zooming in at the best performing deals to explore patterns.
of SV. This allows us to test the probability that SV, according to these definitions, occurs. This is preferred since we cannot perfectly isolate the effect of the transaction, thus estimating the magnitude of additional value would lead to inaccurate conclusions. The seven steps are further specified below.
First, we defined a relatively large event window of eleven trading days (see figure 1), as it takes time for the announcement to reach the relevant stock markets in such a global sample (Kroll, Walters, & Wright, 2008). The window covers five trading days before (123), the announcement date2 (5), and five
trading days after (13) to capture market reactions on rumors and the announcement itself. For the estimation window, we used 250 trading days (approximately one year) prior (12633) to the event window (123) (MacKinlay,1997). For robustness checking we calculated the CAR for a 3-day event window (i.e. 127− 17) and a 250-day estimation window (12637− 127) (Chatterjee, 1986).
Second, we estimated the normal return 8 9: for firm ;, assuming that the acquisition did not
take place. This is done according to equation 1, using the intercept coefficient <=, slope coefficient >=,
the period ! market return R@, and an error term A=: = 0 (MacKinlay, 1997).
8 9: = <=+ >=9@: + A=: (1) Third, we calculated the abnormal return for firm ; and announcement date 5 according to equation 2, wherein C9=D is the abnormal return, 9=D the actual return, and 8 9: the normal return.
C9=D = 9=D− 8 9: (2)
Fourth, the cumulative abnormal return EC9= 123, 13 is calculated by the summation of the abnormal returns in the period (123 to 13).
EC9= 123, 13 = C9=D (3)
Fifth, the expected value 8F of firm ; at the event date is calculated following equation 4. GFHI=JI= is the average market value (i.e. number of outstanding shares times share price) of firm ; over the month prior to the event window. A monthly average (21 trading days 126K to 123) market value is
used since it fluctuates per day. For robustness checking 8F is also calculated for a quarterly (63 trading days 12KN to 123) and a weekly average market value (5 trading days 127O to 123).
8F= = EC9=GFHI=JI= + GFHI=JI= (4)
2 Note that for 114 acquisitions that were announced on a non-trading day, we used the next trading day as event date.
( estimation window ] ( event window
]
12633 123 5 = 0 13Step six concerns the comparison of realized value and expected value, which results in the amount of additional value created (CF=). Realized value (9F=) of the acquirer is measured by the market value one year after the acquisition was announced. A one year timeframe is used since it is argued that the effect of institutional differences on M&A performance is best observed after one year of post-deal integration (Morosini, & Singh, 1994). Again, for the sake of accuracy we used the average market value over a month prior to the date one year after the announcement. In addition, for robustness checking weekly and quarterly averages are calculated. Then additional value is calculated as follows:
CF= = 9F=− 8F= (5)
Finally, dummies are created for five different definitions of SV based on the outcome of step six. For the main analysis, we marked the deals using the CF= based on a 11-day CAR and monthly average market values. The ‘expectation exceedance’ dummy is the most lenient definition in which SV represents a positive difference between realized and expected value (i.e. CF=> 0). This definition is a bit
volatile since it could include deals that outperform expectations as a result of just luck or factors beyond the deal. Therefore, we subsequently zoom in at the top 25, top 10, top 5, and top 1 percent best performing deals in terms of additional value (CF=). In this way, we aim to recognize patterns in the
results across these definition, which will increase the reliability of the conclusions. The five definitions of SV in terms of their range in the distribution of CF= are summarized in table 1. This process is repeated to mark the deals for robustness checking using the CF=’s for the weekly and quarterly based measures. Similarly, we marked the deals for the CF= based on a 3-day CAR for each time frame.
Independent variables. In this study, we test the effect of four psychic distance dimensions on
the probability of creating SV in technology acquisitions. The first independent variable is Business
distance which reflects how distinct the target and acquirer are in terms of product markets. This form of
same one-digit SIC code; ‘2’ for firms sharing the same two-digit SIC code; ‘3’ for the same three-digit SIC code, and ‘4’ for the same four-digit SIC code. Business distance is great for deals coded with a 0 or 1 are; moderate for deals coded with a 2; and deals coded with a 3 and 4 are related.
The second independent variable is Cultural distance, which reflects how distinct the acquirer and target firm are in terms of national culture. For measuring cultural distance, we could use Hofstede’s index as well as the GLOBE (Global Leadership and Organizational Behavior Effectiveness) index. Both have their advantages and disadvantages (Shi & Wang, 2011); however, we use the Hofstede index, because it covers more countries included in the sample. The initial model on national culture consisted of four dimensions: power distance, uncertainty avoidance, masculinity, and individualism (Hofstede & Hofstede, 2001). More recently, Hofstede added two more dimensions: long-term orientation and indulgence (Hofstede, 2011). Despite the lack of data for the latest two dimensions for 63 deals, we used this six-dimensional measure as we prefer precision over sample size. There was no dataset available that contains scores on all six dimensions, so we have built our own dataset for all countries for which the scores are available.3 Following Kogut and Singh (1988) EQ
R2S is calculated for each deal following
equation 6, in which TR= and TS= are the scores on the ;th dimension of the acquirer and target, F= is the
variance of the index of the ;th dimension, and ( is the number of dimensions (i.e. six).
EQR2S = {(TR=−
K
=W7 TS=)6/F=}
( (6)
Third, Geographical distance reflect the distance between acquiring and target firm in 1,000 kilometers. To measure this type of distance, city and country information on the acquirer and target is retrieved through SDC Platinum4. Based on the city and country of the firms, the latitudes and longitudes
are identified by using Bing Maps. With the latitude and longitude of each acquirer and target we calculated the ZQR2S with the Haversine formula (Robusto, 1957):
ZQR2S = 29 arcsin sin6
ab!S− ab!R
2 + cos(ab!S) cos(ab!R) sin6
ac(dS− ac(dR
2 1000 (7)
in which 9 = 6,371, reflecting the earth’s mean radius in kilometers, ab!R and ab!S are the latitudes of
the acquirer and target respectively, and ac(dR and ac(dS are the longitudes of both firms.
Finally, Formal institutional distance reflects the difference in regulations that stipulate how business is done in the home countries of the acquiring and target firm. Following the example of
3 Country scores are retrieved from Hofstede Centre through https://geert-hofstede.com/countries.html. In addition, we used the scores of France for firms from Monaco and the scores of Greece for firms from Cypress. For other missing country scores, we searched for academic papers that contain scores for those countries. In this way, we found the scores for all six dimensions of Puerto Rico (De Mooij, & Beniflah, 2016).
influential studies (e.g. De Beule, Elia & Piscitello, 2014; Gubbi et al., 2010), we operationalize formal institutional distance by using the dimensions of the Index of Economic Freedom, published by the Heritage Foundation (Miller & Kim, 2016). We used nine items5 of the index categorized in four categorical dimensions, which are briefly explained in table 2. To calculate the distance for each dimension, the scores of both firms’ home countries for each item in the year prior to the deal are retrieved from the database of the Heritage Foundation.
eQR2S= = {(eRf−
g
=W7 eSf)6/Ff}
( (8)
Equation 8 is used to calculate the distance for each categorical dimension (De Beule et al., 2014). We calculate distance scores eQR2S for each categorical dimension ; based on the score on the hth item. Ff reflects the variance of the index of item h, ( is the number of items in dimension ;.
Other variables. Next to the analysis of the main model described above, we did an additional
analysis to verify the effect of well-known M&A performance predictors. The variables used are generally included as controls in M&A literature; therefore, we did not build hypotheses. Even though controlling in statistical sense does not exist for the specific model used for binary dependent variables (Hoetker, 2007), it is still informative to know their impact on the likelihood of creating SV. Also, it tells us to what extent our measure deviates from commonly used M&A performance measures.
First, we included two deal level variables in our model. Method of payment is known to influence performance, as it is suggested that cash-financed deals outperform stock-financed deals (Rau & Vermaelen, 1998). This is measured the ratio of cash opposed to other payment methods (Faccio & Masulis, 2005). Another deal level variable that is included is the number of advisors used by the acquirer and seller. The number of advisors hired by both parties should reduce the information asymmetry (Boeh, 2011) and increases the ability of the acquirer to value the assets of the target. It is therefore reasonable to believe that the number of advisors negatively influences the probability of creating SV. We include two measures for the number of advisers: (1) the total number of financial advisors of the acquirer and target, and (2) the total number of legal advisors of the acquirer and target (Boeh, 2011).
Also, five firm level variables are included that generally influence value creation in M&A. First,
acquisition experience of the acquirer is likely to impact SV as experienced acquirers are better at
integrating the valuable resources of the target (King et al., 2004) and therefore at appropriating sources of SV. Acquisition experience is measured as the number of deals a firm has performed between 2000 and the event date available in SDC Platinum (King et al. 2004). Second, prior performance of the acquirer is expected to increase the created value after the deal. Therefore, we include the return on assets (ROA) of the acquiring firm in the year prior to the deal (Finkelstein & Haleblian, 2002). Third, absolute size of the
acquirer is likely to impact the creation of SV, since large firms have generally more slack resources (e.g.
human resources, cash, or managerial time), which helps support value creation (Bruton, Oviatt, & White, 1994). Size of the acquirer is measured by the number of employees at the end of the year prior to the deal. Fourth, R&D intensity of the acquirer is included as it is likely to impact innovative output of the firm, which is a form of value creation (Hitt, Hoskisson, & Kim, 1997). An acquiring firm that invests heavily in R&D is probably better able to create and appropriate value from the resources of the target. Following Hitt, Hoskisson, Ireland & Harrison (1991) we measure R&D intensity as a ratio of R&D expenditures relative to the total sales of the acquiring firm in the year prior to the deal. The final firm level variable is financial slack of the acquirer, which refers to financial reserves (i.e. unabsorbed slack) or to additional lending capacity (i.e. potential slack). Both types of slack can enable the firm to proceed with value creating projects that do not seem highly valuable at first sight (Greve, 2003). Therefore, it is likely that projects with potential to create SV have a higher probability to survive the selection process when the firm has financial slack. Unabsorbed slack is measured by the cash of the acquirer divided by the total assets at the end of the year prior to the event (Kiymaz & Baker, 2008). Similarly, potential slack (or financial leverage) is measured by the net debt (i.e. the sum of short- and long-term debt minus cash) of the year prior to the event divided by the total assets of the same year (Melicher & Rush, 1974).
combination of information. Therefore, these outliers could be the result of measurement errors (Osborne & Overbay, 2004). Hence, outliers at the 0.5% level were removed for: the underlying variable of the dependents (i.e. additional value); prior performance; R&D intensity; and both financial slack measures. This slightly changed the distributions of these variables; however, it did not lead to different results.
Analysis
The aim of current study is to find factors that predict the likelihood of conducting an acquisition that creates SV. Thereby, our dependent variables for each definition of SV take the value of either 1 (deal created SV) or a 0 (deal did not create SV), which points at the use of a probit or logit model. In practice, both models give highly similar results (Chambers & Cox, 1967). However, probit models are generally more appropriate when the binary dependent variable represents an underlying normal distribution, while logit models are more appropriate when the dependent represents an underlying equal distribution (Garson, 2012). In our case the underlying distributions of all dependent variables (i.e. additional value based weekly, monthly, and quarterly averages) are rather normally distributed. Therefore, using the probit model is most appropriate in our case. For reliability reasons, we estimated both regressions and found that they yield similar results, hence we feel confident in reporting the probit results only. To be able to draw meaningful conclusions from our analysis we interpret and report the average marginal effects of the variables in the model instead of the probit coefficients (Hoetker, 2007). The variables for the additional analyses do not cover all 2,644 deals, thus a smaller sample of 1,274 deals is used. For this sample, we repeated the process of creating the SV definitions as the distribution of AV is changed.
RESULTS
Value creation in technology deals
When taking a closer look at the data, we see that from all deals for which the CAR could be calculated, only 52.6 percent were expected to create value. For the remaining 47.4 percent, the CAR is either negative or zero. Figure 2 shows the technology acquisition activity in the period 2000 to 2015. We see that technology deal activity shows severe drops in 2000 to 2002, and 2007 to 2009 followed by periods of increase. In addition, it is rather informative in the context of this study to know whether the deals are expected to create value. The bars in the graph in figure 2 present the share of the deals that yield a positive versus negative CAR, which reflects whether the stock market expects a deal to succeed or fail in creating value (Papadakis & Thanos, 2010). Overall, it seems that in the periods of increase in deal activity the share of deals that are expected to create value exceeds the share of deals that are expected to fail to create value.
Comparing the expectations of the stock market with the number of deals that created more value than expected leads to interesting results as well. Figure 3 presents how many deals from the final sample of this study are: (1) expected to create value, but fail to do so; (2) not expected to create value and failed to do so; (3) expected to create value and surpass expectations; or (4) not expected to yield value, but exceeded expectations. Whether a deal exceeds expectations or not is determined by the most conservative measure of SV, which will be different for stricter measures; therefore, the graph should be
interpreted with care. As failure rates in M&A are said to be high (Harrison et al., 2001), it is no surprise that the bars in the graph show that a large share of the deals in our sample fail to realize value. However, in certain years (e.g. 2005, 2009, and 2014) we see serious spikes, which point at a sufficient number of technology deals that performed better than expected. Interestingly, the dotted line tells us that quite a few deals are undervalued by the stock market. In sum, the two line-diagrams point at the existence of SV creation in technology acquisitions according to our measure.
Table 4 shows the number and proportion of deals creating SV that include an acquirer and target from distinct industries, nations, or both. All five definitions of SV are included, thus, from SV defined as expectation exceedance to the stricter definitions (i.e. best performing 25, 10, 5, and 1 percent of the sample in terms of additional value). For the first four definitions, we see that around 42.5% to 50% of the deals that create SV stay within the same country and industry. Additionally, depending on the definition, 20.8% to 34.6% of the SV creating deals are done by acquirers that cross borders, but stay within the same industry. The share of SV creating deals staying within the country, but crossing industries varies between 20.5% and 22.4% across the first four definitions. In addition, only one of the deals creating SV according to the top 1 definition happened inter-industry and within the same nation. Across all definitions between 11.4% and 15.1% of the SV creating transactions are international and inter-industry. For all four categories the proportions are rather consistent for the expectation exceedance definition to
the top 5 definition. The proportions differ substantially only for the top 1 percent in terms of value creation, which is caused by the small number of deals that create SV according to that definition (i.e. 26 acquisitions). In sum, almost half of the SV creating deals are between firms that are the same in terms of home country and industry, while the majority crosses borders, industries or both.
Main results – Distance and serendipitous value creation
Table 5 shows the probit estimates for the monthly average measure of SV to test the hypotheses. All independent variables are included in all five models, only the dependent variable is different in each model. Model 1 is the probit model in which we attempt to estimate the likelihood that a deal creates SV, in which SV is defined as a positive additional value (i.e. realized value is greater than expected value). In model 2 the dependent variable is SV defined as the best performing quartile of the sample. Similarly, model 3, 4, and 5 contain the probability estimates for a deal being amongst the 10, 5, and 1 percent best performing deals in terms additional value. As SV is not measured before, we take a cautious approach and compare the different models in an attempt to recognize patterns in the effects of the independent variables.
fit slightly increases for the models using stricter definitions of SV. Rather than interpreting the results per model, we attempt to explain per variable what their respective impact is per definitions and to describe patterns.
First, business distance is generally not significantly related to the probability of creating SV. Only on the 2-digit SIC code level is business distance barely significantly related to the likelihood of finding SV for the top 25 (p < 0.1) and top 10 (p <0.1) measures with marginal effects of 6.2% and 4.7% respectively. However, the relationship is only marginally significant and far from significant in the other models. Also, the signs of the effect differ along the models, which is another indication that no patterns exist. Hence, hypothesis 1 is not supported.
Second, cultural distance does not influence the probability of creating SV in model 1 where SV is defined as expectation exceedance. Cultural distance also does not influence SV as defined as the quartile best performing deals. However, cultural distance does matter for the stricter measures, as it is a highly significant (p < 0.01) positive predictor of the likelihood of creating SV defined as the top 10, 5, and 1 percent performing deals. Also, it appears that the probability decreases when we adapt a stricter definition of what serendipity is (i.e. 2.3%, 2%, and 0.9% respectively). Therefore, there is enough evidence, at least for three stricter measures, to support hypothesis 2.
Third, in line with most of the literature, geographical distance seems to be an obstacle in creating value for the first two models. Contrary to cultural distance, geographical distance does not matter for the stricter definitions, but it significantly (p < 0.05) decreases the likelihood of creating SV if it is defined as expectation exceedance or as best performing quartile. The marginal effects indicate that for those measures an increase in geographical distance of 1,000 kilometers between acquirer and target decreases the probability of creating SV with 0.7% and 0.6% respectively. While the marginal effects are not strikingly high, we do find support for hypothesis 3 for the two most cautious measures.
Additional analyses – M&A performance predictors and serendipity
SV creation in M&A has never been tested on a large scale before. Therefore, to provide more complete insights in what influence the occurrence of SV, we tested the effect of some generally accepted controls in M&A performance literature. Table 6 shows the results of this additional analysis and is structured in the same way as the main analysis. The likelihood-ratio chi square test is significant (p < 0.01) across all measures and the model fit, based on the McFadden pseudo R squared, is again increasing when the SV definition is gets stricter.
Two deal-level variables are examined. First, when the proportion of the deal value that is paid in cash increases with 10%, the likelihood of creation SV decreases with 1.1% for SV as expectation exceedance (p < 0.01), 1% for SV as top 25 (p < 0.01), 0.7% for SV as top 10 (p < 0.01), 0.3% for SV as top 5 (p < 0.05). However, method of payment does not significantly affect the probability of finding SV defined as the best performing percent. Second, the number of advisors yields mixed results. While not significant for model 6, we find that when a deal is guided by one additional financial advisor, the likelihood of finding SV increases with 2.1%, 1.9%, 1.6%, and 1.1% for SV defined as the top 25 (p < 0.1), top 10 (p < 0.05), top 5 (p < 0.01), and top 1 (p < 0.01) performing deals. Contrary, the number of legal advisors does not matter for the likelihood of creating SV, since it is only significant for the top 25 definition (p < 0.05), but it is not consistent across the models since the signs change.
Robustness
Different timeframes. As no prior research operationalized the concept of SV, we did a first
from table 5. For reasons of space the standard errors are not presented; however, they are highly consistent to those for the monthly average measure.
Business distance is also not significantly related to SV creation across models 11 to 15. Also, the marginal significant effect we found for business distance on two-digit SIC level in model 2 is insignificant for the week and quarter measure, which underpins our initial conclusion that hypothesis 1 is not supported. For cultural distance a positive significant effect is found for the three stricter models and the marginal effects are similar to the primary results across the measures. Similarly, geographical distance decreases the likelihood of finding SV for the expectation exceedance and top 25 measure with 0.5 to 0.9 percent per 1,000 kilometers. Finally, formal institutional distance is also very consistent across the weekly and quarterly measures compared to the main models. The effects of formal institutional distance in terms of rule of law and market openness are again insignificant. In turn, government size is negatively related to the occurrence of SV for the same four models as shown in table 5. Furthermore, regulatory efficiency positively affects the likelihood of finding SV across all definitions. There is a slight difference in the level of significance for the latter dimension, but the marginal effects are very similar.
Additional event study. There is much debate in M&A literature about the appropriate length of
the event window. Event windows longer than 41 trading days are argued to lead to false conclusions, as it is difficult to control for other events (Rani, Yadav, Jain, 2016). However, for shorter windows there is no consensus about the best way to go. Hence, M&A performance literature generally tests the robustness of the results for different windows (e.g. Chatterjee, 1986).
To check the robustness of the inferences done in this study, we performed an additional event study for calculating the proxy for SV. As the initial event window of this study is relatively long (11-day) compared to most event studies on M&A, we also calculated the CAR for a 3-day event window (!"# to !#). Following the same steps described in the methods we defined the five definitions based on weekly, monthly, and quarterly average market values. These timeframes are also slightly different, since we used the average market value prior to the event window (i.e. 5 trading days !"% to !"#, 21 trading
days !"(# to !"#, and 63 trading days !"%) to !"#). The dummies for marking the deals that create SV are created in accordance to table 1 in the method section.
The table in appendix D shows the marginal effects of the probit estimates for the 3-day CAR dependents. Similar to the previous robustness check, the results for a week, month, and quarter based measures are compared. Again, we did not report the standard errors for reasons of space and because they are highly similar to those in the primary analyses.
as the signs change throughout the models and only a few effects are marginally significant (p < 0.1). Second, cultural distance is highly positive and significantly (p < 0.01 and p < 0.05) related to the likelihood of creating SV for three strictest definitions. Besides significance, the marginal effects are also similar to the main model of this study, since they decrease from 2.2% in model 18 to 1.7% in model 19 and 0.5% in model 20. Third, the influence of geographical distance is similar to the main model in terms of marginal effects and significance level for the expectation exceedance and top 25 definitions of SV. However, the effect of geographical distance on the likelihood of creating SV defined as the top 5% best performing deals is also negative and significant for the monthly (p < 0.05), and quarterly (p < 0.1) based measures. This might be an indication that geographical distance also matters for the stricter measures; however, this is not found in the primary analysis. Finally, results for the dimensions of formal institutional distance are basically the same as the primary analysis. Rule of law and market openness do not matter, because they are only barely significant for some of the SV definitions and the signs change across the models. Government size seems to impede the probability of creating SV for all definitions except top 25. However, significance levels for the 3-day CAR measures are slightly lower for some definitions than in the initial models. The last dimension, regulatory efficiency, is again a positive and significant predictor of the likelihood of creating SV. The marginal effects are slightly different, but the patterns across the models remain the same. The effect of regulatory efficiency is insignificant only for the monthly average measure; however, it is (marginally) significant for the week (p < 0.1) and quarter based measures (p < 0.05).
Inference. The checks above indicate that the results presented earlier in this section are robust.
Since there is no evidence to doubt the results, we stick to the initial conclusions regarding the proposed model. In sum, hypothesis 1 regarding business distance is not supported; hypothesis 2 regarding cultural distance is supported for the three stricter definitions of SV; hypothesis 3 regarding geographical distance is supported for the two most lenient definitions of SV; and there is mixed support for hypothesis 4, since two formal institutional distance dimensions are insignificant and the other two show significant but contradicting effects.
DISCUSSION AND CONCLUSION
study. The current study has a twofold objective, as we attempt to test the existence of SV in technology acquisitions on a large scale and assess the role of psychic distance dimensions for creating SV.
Main insights. The findings suggest that SV occurs more often than one might think because
54% of the deals outperform expectation. While around 45% of those deals are between firms from the same country and industry, the lion’s share of the deals that exceed performance expectations are between a buyer and seller from distinct countries, industries, or both. This indicates, contrary to what many international business scholars believe, that distance between acquirer and target is not necessarily an obstacle.
Many studies adopt the idea that psychic distance impedes performance as it makes information flows costly and complex (e.g. Hutzschenreuter et al., 2014; Johanson and Wiedersheim-Paul, 1975). However, our findings indicate that the effect of psychic distance is not as clear as these studies suggest. The main finding of this study is that the effect of psychic distance on SV creation can only be observed when disaggregated into different dimensions, because each dimension has a different effect. Similar to what Evans and Mavondo (2002) found, some dimensions appear to be irrelevant or impede value creation, while other dimensions increase the likelihood of creating SV. In other words, some dimensions create a valuable form of information asymmetry that stimulates SV creation.
Many scholars argue that being distinct in terms of business negatively affects M&A performance, because similarities in terms of business facilitate a strategic fit that enables the firm to create efficiency and cost synergies (e.g. Healy, Palepu, & Ruback, 1992; Nicholson & Salaber, 2013). In contrast, studies that are more focused on organizational learning and innovation find that business distance can lead to competitive advantages, since it decreases organizational inertia and creates opportunities for learning (Pehrsson, 2006). Choosing a middle course between these two streams, we proposed a certain optimal business distance that has most impact in the probability of creating SV. However, we did not find any effect of business distance regarding the creation of SV. A reason for this non-finding might be that business distance is just not relevant in the context of serendipitous sources of value. As most of the literature suggests, related deals outperform unrelated deals, since relatedness enables possibilities for economies of scale and scope (Hagedoorn & Duysters, 2002). These arguments are mainly exploitative M&A motives (Angwin, 2007), while SV and technology acquisitions are both exploratory in nature (Cunha, 2005; Puranam et al. 2006). Therefore, the explanatory power of business distance might be dependent on the nature of the deal and context.
information on the firms’ resources and capabilities, as we solely look at technology deals. More specifically, as the primary SIC code only reflects the core product market of a firm (Flanagan & O'Shaughnessy, 2003), it does not capture information on all the potentially valuable resources that reside in secondary activities of technology firms, especially since tacit and socially complex knowledge resources are most likely to yield serendipitous sources of value (Graebner, 2004).
Our findings suggest that cultural distance matters in creating SV for the best performing deals. This finding is in line with Morosini et al. (1998), meaning that acquiring a culturally distinct target enables the acquirer to access a diverse set of knowledge, resources and routines that yield sources for SV. Hence, contrary to what most internationalization and some cross-border M&A studies show (e.g. Catwright, 2002; Johanson & Wiedersheim-Paul, 1975; Zaheer et al., 2003), the cultural dimension of psychic distance is found to be an asset rather than a liability regarding SV in technology acquisitions.
Initially, there were no reasons to believe that psychic distance in terms of geography is likely to create SV. The findings support this as geographical distance between acquirer and target negatively influences the probability of creating SV when it is defined as ‘expectation exceedance’ or 25 percent best performing deals in terms of additional value. While the marginal effects are not striking, it is in line with the literature. Tacit and socially complex knowledge require frequent face-to-face interaction for successful transfer (Bresman, Birkinshaw, & Nobel, 2010; Roberts, 2000). Therefore, being physically distant lowers the probability of finding sources of SV.
Interestingly, similar to psychic distance, we find that the explanatory power of formal institutional distance is not consistent across its dimensions. Two of the formal institutional distance dimensions are insignificantly related to the probability of finding SV, while the other two are contradicting in the sign of the effect. While Gubbi et al. (2010) found a positive performance impact of formal institutional distance by using an aggregated measure, our findings suggest that finer formal institutional distance measures are more appropriate.
First, distance in terms of freedom of property rights and freedom from corruption (i.e. rule of law) do not matter for the likelihood of SV to occur. As differences in institutional regime increase uncertainty about the ability to appropriate value (Teece, 1986), there is likely a negative effect of rule of law distance on SV creation; however, we do not find such effects. An explanation might be that this uncertainty only matters for acquirers when the appropriability regime of the target’s country is weaker than the acquirer’s home country, which makes sense since Gubbi et al. (2010) find that formal institutional distance has a positive performance impact from outward acquisitions from an emerging country into a developed country.
of government size (i.e. fiscal freedom and government spending) both reflect the tax rates that firms and individuals face (Miller & Kim, 2016). This dimension of formal institutional distance leads to uncertainty about the tax environment, which is argued to impede financial success (Davies, 2004).
Third, formal institutional distance dimension market openness does not matter for SV creation. An explanation might be that the effect is dependent on whether the difference is in favor of the acquirer or target firm. An open environment in terms of trade, investment, and financial freedom is said to positively affect the entrepreneurial and innovative possibilities (Miller & Kim, 2016). Therefore, it might be that SV creation is stimulated when the target is in a more open host country, while it is less likely to occur when the target is located in a very restrictive country. For instance, when distance in terms of market openness is great, emerging market and developed market firms tend to differ in their investment behavior, because market openness is generally greater in developed countries (De Beule et al., 2014).
The last dimension of formal institutional distance, regulatory efficiency, appeared to be the most influential variable in the model as its effect is positive and significant across all definitions of SV with substantial marginal effects. This finding is remarkable since most of the literature argue that differences in regulations increase uncertainty and risks for opportunism and consequently increasing transaction costs (e.g. Eden & Miller, 2004; Meyer et al., 2009). However, our findings contradict transaction costs arguments, as we find that differences in terms of business and monetary regulations increase the probability of serendipitous sources of value to occur. This effect might be explained by the idea that if a technology firm experiences uncertainty about the target’s home county regulations, they tend to be more cautious and consciously attempt to obtain more knowledge about the local market (Evans & Mavondo, 2002, Dikova, 2009). Thus, the information asymmetry resulting from differences in regulatory efficiency of the respective home countries of the acquirer and target increases the urge for the acquirer to deliberately search for local market knowledge. In turn, this deliberate searching behavior can lead to serendipitous discovery (Cunha, 2005).
Additional insights. We also investigated the relation of variables that are generally included as
controls in M&A literature. Interestingly, the findings suggest that the effect of some of these variables cannot be generalized to the context of SV creation in technology acquisitions. Below the counterintuitive findings are briefly discussed.
