Modeling Competitive Group Dynamics: A Bayesian Hidden Markov Model Approach

Open Access
Kani, Amirali
Graduate Program:
Business Administration
Doctor of Philosophy
Document Type:
Date of Defense:
June 07, 2018
Committee Members:
  • Duncan King-Hoi Fong, Dissertation Advisor/Co-Advisor
  • Duncan King-Hoi Fong, Committee Chair/Co-Chair
  • Wayne S. DeSarbo, Committee Member
  • John Richard Howell, Committee Member
  • David Russell Hunter, Outside Member
  • Dynamic Segmentation
  • Competitive Groups
  • Marketing Strategy
  • Dynamic Analysis
  • Hidden Markov Models
The concept of competitive groups has emerged as an important perspective for understanding the sources of intra-industry competitive heterogeneity. Three sources identified in the literature involve managerial strategy, corporate performance and the impact heterogeneity among the firms in an industry. In this dissertation, a new hidden Markov model is developed to assess dynamics in deriving competitive group memberships arising from changes in competitive market structure over time. Contrary to existing methods, the proposed approach allows some observed “entry” and “exit” states to model firms’ entry into and exit from the market, along with their transitions between various competitive groups during a specified study period. The proposed model is illustrated with a banking dataset created from the COMPUSTAT database and other publicly available financial reports. The dataset includes some 288 banks from the tri-state area of NY-OH-PA that have been publicly traded for at least one year during the observation period from 1995 to 2015. I set up the model in a Bayesian framework, and devise a reversible-jump Markov chain Monte Carlo estimation procedure to determine the number of latent competitive groups and uncover the characteristics of each derived group. Through comparison with some five alternative benchmark models that only account for some of the possible sources of competitive heterogeneity or completely ignore firms’ entry and exit information, I show the benefits and better performance of the proposed model in uncovering dynamics in the competitive market structure. In particular, with the inclusion of firms’ entry and exit information, the proposed model is able to provide a probabilistic prediction of the future trend in the competitive market structure based on various counterfactual conditions.