Academic journal article Management International Review

Mode of International Entry: The Advantages of Multilevel Methods

Academic journal article Management International Review

Mode of International Entry: The Advantages of Multilevel Methods

Article excerpt

Abstract and Key Results

* International strategy empirical research on the mode of entry has typically overlooked the multilevel nature of this question and relied on non-multilevel quantitative methods. This creates important conceptual and statistical limitations. We examine such drawbacks by explaining the multilevel nature of this research question and the necessity to use multilevel methods.

* As an illustration, we develop a multilevel model and run a multilevel Bernoulli analysis to analyze the determinants of modes of entry, using a dataset on Japanese Foreign Direct Investment. Its results are compared to those of the dominant statistical method used in International Management for this topic: logistic regression.

* Research on mode of international entry has a clear conceptual and empirical multilevel dimension. Non-multilevel quantitative methods limit the conceptual development of this research and have negative statistical consequences that pose a risk for the validity and robustness of the results. In contrast, multilevel quantitative methods provide benefits when incorporating them for research on the selection of an entry mode. This has important methodological implications for future quantitative research on this topic.

Key Words

Entry mode, Bernoulli analysis, multilevel methods, FDI


The selection of an appropriate mode of entry in foreign markets is a critical issue in International Business. Research on this issue has proliferated (for a review, see Reus/Ritchie III 2004). Conceptual articles have proposed multidimensional and hierarchical models of market entry modes (Pan/Tse 2000). Empirical articles on the choice between an international joint venture (IJV) and a wholly owned subsidiary (WOS) for foreign direct investment (FDI) have investigated numerous predictors (e.g., Brouthers 2002, Guillen 2003, Harzing 2002, Lu 2002, Pak/Park 2004, Yiu/Makino 2002). According to these authors, a decision to implement a WOS or a JV can be explained by FDI-level variables, depending on the FDI's country and industry, but also by parent firm-level variables such as its international experience, size, or advertising intensity. Similarly, from their extensive review of the IJV literature, Reus and Ritchie III (2004, p. 369) consider three levels of analysis for IJV: interpartner, parent, and environmental.

Therefore, decisions about the mode of entry for FDI appear to be a multilevel phenomenon: explanations will come from FDI-level variables, parent firm-level variables and the interactions of both. Yet, studies sometimes mix or confuse these levels of analysis. Usually, their findings are derived from non-multilevel analyses such as logistic regressions. They implicitly present multilevel conceptual models with variables measured at two different levels (parent firm, FDI) but ignore this multilevel dimension in their data analysis method. As a consequence, there is a lack of actual multi-level research and such type of study must be developed (Reus/Ritchie III 2004).

This omission can have important consequences for our understanding of the determinants of the entry mode decisions as well as for the validity and robustness of the empirical evidence presented in the entry mode literature. Firstly, it constrains the conceptual development of more comprehensive models looking at level-1, level-2 variables, and their interactions (Bryk/Raudenbush 1992, Reus/Ritchie III 2004). Secondly, it involves potential statistical problems of disaggregation, intra-class correlation, and misestimated precision (Barcikowksi 1981, Bliese/Hanges 2004, Bryk/Raudenbush 1992, Hox 1997). Because a parent firm can have multiple FDIs, these FDIs cannot be considered as independent and the data structure is hierarchical. This situation may create methodological problems well recognized in the multilevel models literature (e.g., Bliese/Hanges 2004). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.