Theses Doctoral

Stochasticity in Games: Theory and Experiment

Friedman, Evan Kyle

A large literature has documented a pattern of stochastic, or random, choice in individual decision making. In games, in which payoffs depend on beliefs over opponents’ behavior, another potentially important source of stochasticity is in the beliefs themselves. Hence, there may be both “noisy actions” and “noisy beliefs”. This dissertation explores the equilibrium implications of both types of noise in normal form games. Theory is developed to understand the effects of noisy beliefs, and the model is compared to the canonical model of noisy actions. Predictions—and assumptions—are tested using existing and novel experimental data.

Chapter 1 introduces noisy belief equilibrium (NBE) for normal form games, a model that injects “noisy beliefs” into an otherwise standard equilibrium framework. Axioms restrict the belief distributions to be unbiased with respect to and responsive to changes in the opponents’ behavior. We compare NBE to an axiomatic form of quantal response equilibrium (QRE) in which players have correct beliefs over their opponents’ behavior, but take “noisy actions”. We show that NBE generates similar predictions as QRE such as the “own-payoff effect”, and yet is more consistent with the empirically documented effects of changes in payoff magnitude. Unlike QRE, NBE is a refinement of rationalizability and invariant to affine transformations of payoffs.

Chapter 2, joint with Jeremy Ward, studies an equilibrium model in which there is both “noisy actions” and “noisy beliefs”. The model primitives are an action-map, which determines a distribution of actions given beliefs, and a belief-map, which determines a distribution of beliefs given opponents’ behavior. These are restricted to satisfy the axioms of QRE and NBE, respectively, which are simply stochastic generalizations of “best response” and “correct beliefs”. In our laboratory experiment, we collect actions data and elicit beliefs for each game within a family of asymmetric 2-player games. These games have systematically varied payoffs, allowing us to “trace out” both the action- and belief-maps. We find that, while both sources of noise are important in explaining observed behaviors, there are systematic violations of the axioms. In particular, although all subjects observe and play the same games, subjects in different roles have qualitatively different belief biases. To explain this, we argue that the player role itself induces a higher degree of strategic sophistication in the player who faces more asymmetric payoffs. This is confirmed by structural estimates.

Chapter 3 considers logit QRE (LQRE), the common parametric form of QRE; and we endogenize its precision parameter "lambda", which controls the degree of “noisy actions”. In the first stage of an endogenous quantal response equilibrium (EQRE), each player chooses her precision optimally subject to costs, taking as given other players’ (second-stage) behavior. In the second stage, the distribution of players’ actions is a heterogenous LQRE given the profile of first-stage precision choices. EQRE satisfies a modified version of the regularity axioms, nests LQRE as a limiting case for a sequence of cost functions, and admits analogues of classic results for LQRE such as those for equilibrium selection. We show how EQRE differs from LQRE using the family of generalized matching pennies games.


  • thumnail for Friedman_columbia_0054D_15967.pdf Friedman_columbia_0054D_15967.pdf application/pdf 3.83 MB Download File

More About This Work

Academic Units
Thesis Advisors
Casella, Alessandra M.
Ph.D., Columbia University
Published Here
August 4, 2020