Detecting Industry Substructure of NASDAQ Electronics Firms via Stock Return Method

While there are a number of theories suggesting why industries might be subdivided into subgroups or strategic groups in the field of strategic management, the empirical research on the strategic group concept which uses the cluster/F statistic method for discovery and testing for significance has been criticized as artifactual. Further, while it is crucial for the meaningful subgroups, the choice of clustering variables seems subjective and arbitrary. As an alternative to the conventional methods, the stock return method was proposed in the literature, which uses equity return data. Although insightful, the previous work only demonstrates that the groups found by the stock return method have good ‘face validity.’ In the paper, we explore to obtain statistical evidence by examining whether the subgroups identified by analyzing the movements of stock returns are statistically different in terms of exogenous variables which are in nature independent of stock returns. For a sample of 94 NASDAQ electronics firms which we chose to use due to the availability of exogenous variables, we conduct a canonical discriminant analysis using 67 independent taxonomic variables. We find that the clusters derived from the method show statistically significant difference each other with respect to the exogenous taxonomic variables. We conclude that the stock return method could detect non- artifactual strategic groups within an industry in an objective way.