Categorization of mergers and acquisitions using transaction network features

https://doi.org/10.1016/j.ribaf.2021.101421Get rights and content

Highlights

  • Propose a novel M&A categorization method for corporate strategies from the linkage of transactions data and M&A data of Japan.

  • Use two features to categorize M&A: betweenness centrality and shortest path length.

  • Using the above two features, find several meaningful concentration areas and provide trends of transactions and M&A in these areas.

Abstract

Mergers and acquisitions (M&A) have occurred among tens of thousands of companies. Categorization of M&A is important to both corporate strategy and academic research. Previous research largely uses case studies and econometric data analysis to classify the motivations and types of M&A. Here, we propose understanding M&A using large-scale data to generate more applicable and generalized results. We use transaction relationships from transaction networks to better understand M&A. Based on detailed pre-analysis, including matching M&A and transaction data from Japan and clustering of transaction networks, we select several M&A observation perspectives. We use two features of transaction networks to categorize M&A cases: betweenness centrality and shortest path length. Betweenness centrality provides a view of the overall business situation from a macro perspective, and shortest path length helps to understand neighboring business environments from a micro perspective. We find several meaningful areas of concentration based on their betweenness centrality values and shortest path lengths. Finally, we re-examine M&A cases in each area, summarizing the trends identified using this categorization method. This study contributes to the M&A literature because it advances quantitative categorization of M&A cases.

Introduction

Mergers and acquisitions (M&A) are usually considered inorganic corporate growth. M&A transactions require considerable investment and resources, resulting in a vast body of M&A research and performance evaluation (e.g., Rao and Mishra, 2020). This field includes diverse approaches and perspectives, including empirical studies, case studies, and stock event studies. Novel methods have been applied in this field in recent years, especially in technology-related M&A research on technology fusion and R&D investment. Numerous review papers summarize this field from various perspectives. Reddy (2014) reviewed several industry-related M&A topics. Ferreira et al. (2014) examine how M&A research topics have changed over the past few decades, pointing out future research directions. Shi et al. (2012) summarize current research and propose a research agenda. In their review article, Shi et al. recommend examining company relationships in future M&A research.

Industrial classification is complex. Code-based classification methods include Standard Industrial Classification (SIC), North American Industry Classification System (NAICS), and Japan Standard Industrial Classification (JSIC). However, because industries are dynamic, there is always some bias when using these code-based classification methods.

Before conducting research, M&A must be categorized, taking into account M&A complexity and corporate strategy. M&A categorization has been studied for decades, and different categorization methods exist, generating diverging insights and results (Lewellen, 1971, Harford et al., 2009). Lubatkin (1983) implement three ways to categorize M&A: the Federal Trade Commission method, the M&A experience, and M&A firm size. Amit et al. (1989) and Bower (2001) focus on M&A motivation for categorization. Recently, Brueller et al. (2014) and Rompotis (2015) summarize several M&A categorization methods that reflect current business trends worldwide, such as those in technology and talent M&A and leveraged buyouts. However, there are no uniform standards for categorization, which is usually performed by business analysts, requiring significant human resources. However, categorizing M&A using data and artificial intelligence methods, such as machine learning and network analysis, would be beneficial in terms of lowering human capital costs.

Supply chains and transaction networks have been studied independently in business research for several decades. Campbell (1974) uses industrial input–output data to identify industrial clusters. Sammarra and Biggiero (2008) analyze business relationships using information flows inside transaction networks. Borgatti and Li (2009) apply network analysis to supply chain networks by converting concepts from the former to the latter, along with theoretical explanations. Zuo et al. (2016) propose that most companies can identify partners from transactions. M&A, as a special form of business partnership, should have some common features with business partner selection. Durach et al. (2017) summarize supply chain management research using a systematic literature review.

Despite the importance of M&A and transaction networks in business research, few studies examine these topics. Of those that have, most do so from a macro perspective. These works discuss several important aspects of the whole business environment, but do not provide information on individual behavior. The pioneering Harford et al. (2019) examines individual M&A using empirical methods, contributing several useful findings.

The purpose of the present study is to categorize M&A using transaction networks, providing insight into the mechanisms behind the relationship between them. We first match M&A data and transaction data for Japan. We then use a combination of the ratio of betweenness centrality and shortest path length indices in a network analysis to categorize M&A. Betweenness centrality shows information flows in business surroundings, whereas shortest path length reflects their distances. Several related investigations are employed to provide multiple perspectives. Rather than using a complex industrial classification, we suggest using the shortest path length orbit to integrate vertical and horizontal M&A for neighboring industries. To integrate further industries, we summarize their exploratory business behavior. This represents a novel method in the literature. Categorizing M&A in this way provides new insights for M&A research and business practices.

We use data on Japan for several reasons. First, Western countries, such as the United States, have already received significant attention. Thus, research based on Asia-Pacific markets expands the literature. Second, businesses operate differently in Japan than in Western countries, with the former in a process of “Westernization.” Therefore, studying Japan provides insights for Asian developing countries. Third, the data and social systems in Japan are mature enough to provide accurate research results. Fourth, we use transaction network data to investigate whether the “keiretsu” (a set of companies with loose connections) phenomenon affects M&A.

This paper proceeds as follows. Section 2 summarizes the literature. Section 3 presents the data. Section 4 discusses our methodology, and Section 5 sets forth our results, which are then discussed in Section 6. Finally, in Section 7, we discuss our conclusions.

Section snippets

Previous literature

Transaction networks and supply chains are used to investigate business relationships; studies apply various analytical methods, including network science (Barabási et al., 2016). In addition to several examples in the Introduction above, dynamic transaction networks are used to select business partners (Low, 1997). Kajikawa et al. (2012) analyze and extract cluster information. The Japanese concept of a “keiretsu,” discussed in Aoki and Patrick (1995) and Hoshi and Kashyap (2004), can also be

Data

We use data from TOKYO SHOKO RESEARCH, LTD. (TSR, TOKYO SHOKO RESEARCH, 2020) and Bureau van Dijk (BvD), a Moody's Analytics company (BvD, a). TSR provides business information inside Japan. We use TSR company information files and TSR company relationship files. The former are basic company information databases, including general information on company profiles, stakeholders, suppliers, customers, industrial classifications (in code form), product lines, financial statements, main banks, and

Database matching

The company identification number in the BvD data (BvD ID number) comprises two parts. The first two characters correspond to the two-digit ISO country code where the company is incorporated (e.g., “JP” denotes Japan). The rest of the BvD ID number can be 1) a national ID or tax number (prioritized by BvD), 2) an internal identifier of the information provider, or 3) created by BvD itself (BvD, a).

In addition to the TSR company code, the TSR company information data contain company names,

Results for data preprocessing

We create 7,975 company record matches, for which we confirm an accuracy of 94%. After matching the M&A deal information, we have 2,016 M&A entries for analysis. We use the Leiden method for clustering, yielding 248 clusters for the maximum connected component. Table 3 shows the numbers of nodes from the largest clusters, with a large decrease between clusters 22 and 23. Hence, we take the largest 23 clusters as our research targets (see Appendix A for details on the clustering). We conduct the

Interpretation of M&A categories

A popular way of categorizing M&A is as horizontal or vertical M&A, as argued by Lubatkin (1983) and Rompotis (2015). However, this method ignores M&A between the two types, due to diversification of corporate business and inaccurate dynamic changes in industrial classification. Indeed, “non-horizontal” or “non-vertical” M&A cases emerge. For example, integration of an electricity company and an electrical appliance company involves different industries, but the two companies are actually

Conclusion

This study uses TSR and BvD data to categorize M&A in Japan using transaction relationships. Following the literature, we combine betweenness centrality and shortest path length as our measures of investigation; this is a novel method. We classify M&A cases according to different shortest path lengths, and find several concentration areas. We further investigate M&A cases in these areas and summarize their main trends and characteristics. We propose integrating horizontal and vertical M&A in

Authors’ contribution

Bohua Shao: conceptualization, methodology, software, validation, writing; Kimitaka Asatani: conceptualization, methodology, supervision; Hajime Sasaki: conceptualization, resources, data curation; Ichiro Sakata: resources, methodology, data curation, supervision.

References (48)

  • Y. Zuo et al.

    Extraction of business relationships in supply networks using statistical learning theory

    Heliyon

    (2016)
  • M. Abdollahian et al.

    Towards trade equalisation: a network perspective on trade and income convergence across the twentieth century

    New Polit. Econ.

    (2014)
  • K.R. Ahern et al.

    The importance of industry links in merger waves

    J. Finance

    (2014)
  • A. Almazan et al.

    Financial structure, acquisition opportunities, and firm locations

    J. Finance

    (2010)
  • R. Amit et al.

    A classification of mergers and acquisitions by motives: analysis of market responses

    Contemp. Account. Res.

    (1989)
  • M. Aoki et al.

    The Japanese Main Bank System: Its Relevance for Developing and Transforming Economies

    (1995)
  • A.L. Barabási

    Network Science

    (2016)
  • P. Bick et al.

    Does distance matter in mergers and acquisitions?

    J. Financ. Res.

    (2017)
  • V.D. Blondel et al.

    Fast unfolding of communities in large networks

    J. Stat. Mech.: Theory Exp.

    (2008)
  • S.P. Borgatti et al.

    On social network analysis in a supply chain context

    J. Supply Chain Manag.

    (2009)
  • J. Bower

    Not all M&As are alike – and that matters

    Harv. Business Rev.

    (2001)
  • N.N. Brueller et al.

    How do different types of mergers and acquisitions facilitate strategic agility?

    Calif. Manag. Rev.

    (2014)
  • BvD, a. Bureau van Dijk. A Moody's Analytics Company. https://www.bvdinfo.com/ (Accessed 25 May...
  • BvD, b. Orbis. https://www.bvdinfo.com/en-gb/our-products/data/international/orbis (Accessed 25 May...
  • Cited by (5)

    • Multilayer network analysis of investor sentiment and stock returns

      2022, Research in International Business and Finance
      Citation Excerpt :

      Research on the above aspects remains relatively scarce. Due to transactions, equity relationship, business contracts and so on, the relationships among firms are becoming increasingly relevant (Chen et al., 2016; Shao et al., 2021; Wang et al., 2021a). A market consisting of many interacting firms is like a complex network, in which each firm is a node of the network, and the relationships between firms compose the edges of the network.

    • Evolution of the Merger Market and Acquisitions in Ecuador

      2022, Iberian Conference on Information Systems and Technologies, CISTI

    We thank John W. Goodell (Editor-in-Chief), an anonymous referee, and anonymous Editors for their useful comments and suggestions.

    View full text