Explaining the Strategic Groups-Firm Performance Relationship: A Multilevel Approach Applied to Small and Medium-Sized Hotel Companies in Spain

Article excerpt

One of the main research questions in the field of strategic management is why firms obtain different performance levels. This paper answers this question from the strategic groups approach. This paper analyzes the linkage between strategic groups and firm performance offering a multilevel analysis about the relative importance of intergroup and intragroup performance differences based on the use of hierarchical linear models. The results show that intragroup differences explain firm performance better than intergroup differences.

Introduction

One of the main research concerns in the field of strategic management is the reason firms achieve different levels of performance (Rumelt, Schendel, and Teece 1994). Although the traditional industrial organization approach points to industry structure as the main determining factor for firm performance (Bain 1959; Scherer 1970), the resource-based view asserts that a firm's resources are the most relevant factors (Barney 1991; Peteraf 1993; Wernerfelt 1984). Another level of analysis--strategic groups--appears between industry and firm. Strategic groups are sets of companies within an industry that pursue mutually similar strategies (Porter 1979).

The impact of group membership on firm performance has been a central topic in the research on strategic groups (Cool and Schendel 1987; McGee and Thomas 1986; Peng, Tan, and Tong 2004; Thomas and Venkatraman 1988). Some studies identify significant performance differences between strategic groups (Leask and Parker 2007; Neill and Rose 2006; Reger and Huff 1993), but others do not reach any conclusive results (Amel and Rhoades 1988; Cool and Schendel 1987; Fiegenbaum and Thomas 1990; Lawless 1989; Olusoga, Mokwa, and Noble 1995; Wiggins and Ruefli 1995; Ziiniga-Vicente, de la Fuente-Sabate, and Suarez-Gonzalez 2004). There is also conflicting evidence with regard to performance differences among members of the same strategic group. These differences have been analyzed less than intergroup differences. Some studies have found significant intragroup performance differences (Claver-Cortes, Molina-Azorin, and Pereira-Moliner 2006; Cool and Schendel 1988; Lawless, Bergh, and Wilsted 1989), whereas others have found no conclusive results (Athanassopoulos 2003). Therefore, empirical research does not provide clear evidence of the group-performance relationship. This lack of agreement is one of the most prominent shortcomings of this line of research (Barney and Hoskisson 1990).

Our study offers a couple of contributions to address this shortcoming. First, we provide fresh insights into the relationship between strategic groups and performance, comparing differences in performance across groups (intergroup differences) with differences in performance among firms within each group (intragroup differences). From this comparison, we determine what kind of difference better explains firm performance, and we contribute to the literature by offering possible reasons for the non-conclusive results in the research into the relationship between strategic groups and performance, collecting conflicting approaches in the literature on strategic groups to propose the hypotheses.

Second, we offer an empirical contribution by introducing the multilevel analysis approach into the strategic groups literature, applying hierarchical linear models (HLM), which have hardly been used in the study of strategic groups (McNamara, Deephouse, and Luce 2003; Short etal. 2007). Strategic groups and firms are dependent levels of analysis because firms are nested into groups. Strategic groups are formed by firms, and consequently, a multilevel or hierarchical structure exists. Traditional statistical tests, such as regression analysis or analysis of covariance, lean heavily on the assumption of independence of the observations. If the independence assumption is violated (and in multilevel data, this is almost always the case), the estimates of the standard errors of conventional statistical tests are much too small or underestimated, and this results in many spuriously significant results (Hox 2002; Rasbash et al. …