Renovation of Oligopoly Theory in The Light of Algorithmic Strategic Decision-Making: A Case Study for Turkish Cement Industry

Abstract

Oligopolistic markets, characterized by a few dominant firms, rely on interdependency theory to explain how companies’ strategies are influenced by their competitors. In the digital economy, data and computational capabilities have become crucial for competitive advantage with price-setting algorithms exemplifying this trend. Despite their benefits, these algorithms pose certain risks of anti-competitive pricing behaviors. In the thesis, we have analyzed industry-specific equilibriums for the perfect competition, Bertrand competition, real situation, and monopoly. In addition to the market’s traditional assumptions, we simulated its algorithmic equilibrium. With the assumption of a uniform algorithm that makes decisions for all firms, the algorithmic equilibrium approached to the monopolistic equilibrium. We also focused on algorithmic competition between two firms in a duopoly. In this scenario, we let companies use their own algorithms, which employed the same Q-learning technology, to price their products on a daily basis. Algorithmic simulations are performed for 36 different cases, with varying cross-price elasticity levels, marginal cost structures, capacity constraints, and demand models. After simulating the environment, we have drawn several propositions. First, the use of a uniform algorithm and allowing it to make pricing decisions for both members of the marketplace approaches to the monopolistic equilibrium. Second, increasing cross-price elasticity promotes competition by reducing the impact of capacity advantages. However, this effect is limited and depends on the capacity level of the inferior firm. Third, compared to the equal case, different MC structures result in greater industry profits with lower output. Additionally, they diminish the potential competitive effects of high cross-price elasticities. Fourth, equal capacity constraints favor collusion. The harmful consequence of collusion is further increased when equal capacity constraint is paired with the different marginal cost structures. Fifth and the last, capacity advantages are better leveraged when the marginal costs differ and/or demand elasticity is low. We believe that with these analyses, we might get a little closer to understanding the nature of algorithms.