Simulating competitive collocation strategies with deep reinforcement learning

This paper introduces a simulation model based on deep reinforcement learning to study competitive collocation strategies of two internationalizing firms. Drawing on the strategic interaction and competitive collocation literature in IB, we explore the factors determining the performance-optimizing degree of collocation in market-seeking internationalization for a duopoly in product markets. Our simulation suggests that a high degree of collocation between the two firms is optimal in many circumstances. Whereas mid-to-high levels of collocation are driven by low competitive intensity, low entry barriers, low home-market size, and high market-size variation, complete collocation emerges under high competitive intensity, low first-mover advantage, and smaller home-market size. High entry barriers, combined with low variation in market size and strong liabilities of foreignness, drive the opposite case of complete mutual avoidance. A key takeaway from the simulation is a focus on the competing firm in global expansion: in many cases, it becomes key to limit the rival’s growth options and to enter key markets before access becomes more restricted. We discuss the possibilities of using computer-simulation-based approaches to theorizing on internationalization.