Competition in the Digital Economy: Collusion by AI Pricing Algorithms
|Author||Wiroy Shin, Yeolyong Sung, Yangshin Park, Hoonsik Yang, Minji Kang, and Jin Park||Date||2018.12.17||Page|
We study collusive outcomes in the digital economy where firms commonly employ pricing algorithms. In the paper, we explore definitions and characteristics of algorithmic collusion from the competition policy perspective, and we discuss economic research on collusion and various screening methods. We also show that Q-learning algorithms coordinate their pricing decisions conducting simulations on prisoner’s dilemma environments. Based on such analyses, we provide logic and economic intuition of the algorithmic collusion, and we propose policy suggestions for Korea Fair Trade Commission.
We present economic reasoning on why the competition authority should regulate algorithmic collusion. Additionally, we discuss importance of an economic optimal policy framework on algorithmic collusion regulations. Finally, we examine specific policy topics relating to algorithmic collusion: risks in leniency programs of Korea, attempts of ban on firms’ information exchanges, and optimal policy choices that Korea FTC can choose under current resource and legal constraintscreating effective screening schemes and regulating algorithmic collusion cases where human communications are involved during the initiation phase of algorithmic collusion.