From imitation to collusion: a replication

Abstract

In oligopoly, imitating the most successful competitor yields very competitive outcomes. This theoretical prediction has been confirmed experimentally by a number of studies. A recent paper by Friedman et al. (J Econ Theory 155:185-205, 2015) qualifies those results in an interesting way: While they replicate the very competitive results for the first 25-50 periods, they show that when using a much longer time horizon of 1200 periods, results slowly turn to more and more collusive outcomes. We replicate their result for duopolies. However, with 4 firms, none of our oligopolies becomes permanently collusive. Instead, the average quantity always stays above the Cournot-Nash equilibrium quantity. Thus, it seems that ''four remain many'' even with 1200 periods. Keywords Imitation Á Experiment Á Cournot oligopoly Á Walrasian outcome JEL Classification C91 Á C72 Á D74 Electronic supplementary material The online version of this article (

An Evolutionary Cournot Oligopoly Model with Imitators and Perfect Foresight Best Responders.

“…For example, in Huck et al, 1999(see also Offerman et al, 2002and Bigoni and Fort, 2013, the authors show that players boost more competitive outcomes provided they are supplied by information on disaggregated quantities and related profits of competitors. Fairly high competitive levels are also observed in the experimental Cournot oligopolies considered in Apesteguia, Huck, and Oechssler, 2007, Huck, Normann, and Oechssler, 2004, Apesteguia, Huck, Oechssler, and Weidenholzer, 2010, Oechssler, Roomets, and Roth, 2016, Friedman, Huck, Oprea, and Weidenholzer, 2015. Such a circumstance provides indication about the presence of a relevant component of imitative behavior, an interpretation which is supported by the theoretical contribution by Vega-Redondo, 1997, where the author shows that the competitive equilibrium emerges under the so called "imitate the best" rule, exploited by quantity setting agents.…”

“…In particular, in the left panel, a dynamic scenario is shown, where β = 1, C = 1 and N = 12, at which the stationary state E * 2 Fundamental parameters are fixed at the values usually adopted within experimental oligopolies with an underlying linear structure (see e.g. Huck et al, 1999or Oechssler et al, 2016 As a consequence, most of the dynamics of the system is due to wide variations of the relative fraction of imitators, taking place above the floor 1/2. Then, the loss of stability of E * can be thought as a transition from an initial scenario, where each player produces following a given rule in time, towards a new scenario where players produce at the same level alternating the adoption of either rule or the other.…”

Multiple Behavioral Rules in Cournot Oligopolies

“…Friedman et al (2015) showed that duopolies and triopolies initially moved toward competitive outcomes, but participants gradually learned to coordinate on quantities near collusion. Oechssler et al (2016) replicated this result for duopolies, but found quantities above the Cournot one for tetrapolies and concluded that "four remain many even with 1200 periods." 1 1 Other experiments have examined the role of imitation in different settings.…”

“…In our experiment, participants interacted in 4-player Cournot oligopolies (tetrapolies). We made this choice because previous results have shown that convergence to Walrasian outcomes, which is compatible with imitative behavior, occurs more frequently with tetrapolies than with triopolies (Huck et al, 2004;Oechssler et al, 2016). We conducted four sessions with 32 participants each for a total of N = 128 (82 females; median age 22 years) at the Cologne Laboratory for Economic Research (CLER).…”

An evolutionary Cournot oligopoly model with imitators and perfect foresight best responders

“…() (see also Bigoni & Fort, ; Offerman et al., ), the authors show that players boost more competitive outcomes provided they are supplied by information on disaggregated quantities and related profits of competitors. Fairly high competitive levels are also observed in the experimental Cournot oligopolies considered in Apesteguia, Huck, and Oechssler (), Huck, Normann, and Oechssler (), Apesteguia, Huck, Oechssler, and Weidenholzer (), Oechssler, Roomets, and Roth (), Friedman, Huck, Oprea, and Weidenholzer (). Such a circumstance provides indication about the presence of a relevant component of imitative behavior, an interpretation which is supported by the theoretical contribution by Vega‐Redondo (), where the author shows that the competitive equilibrium emerges under the so called “imitate the best” rule, exploited by quantity setting agents.…”