Abstract
This paper illustrates techniques for assessing the dynamic stability of games where a continuum of types might be present by re-analyzing two models under incomplete information, the Lohman et al. (Unpublished manuscript, 2001) public goods game and the Kopel et al. (J Econ Dyn Control 48:394–409, 2014) Cournot duopoly game. The evolution of continuous types follows either replicator dynamics Oechssler and Riedel (Econ Theory 17:141–162, 2001; J Econ Theory 107:223–252 2002) or gradient dynamics Friedman and Ostrov (J Math Econ 46:691–707, 2010; J Econ Theory 148:743–777, 2013). The techniques rely on a system of partial differential equations. Numerical solutions obtained through replicator and gradient dynamics highlight the differences and the similarities that arise under both approaches. In the public goods game, the dynamic system affects the stationary distribution of types while in the Cournot duopoly model, the types evolve to a single mass point regardless of the dynamics used. Lastly, these techniques allow us to endogenize the distribution of player types.
Similar content being viewed by others
Notes
Similar ideas have been applied to the evolution of preferences in a discrete case in [21].
Königstein and Müller [13] also study a variant of the standard Cournot duopoly model. In their model, there is a continuum of types whose subjective profits are dependent on both the monetary profits and consumer welfare. The firms have perfect information about the counterparty type. Using static analysis, they find that firms that consider consumer welfare are evolutionary stable.
Ritzberger and Weibull [22] use the term sign-preserving dynamics when the growth rate of strategy i is positive (negative) and the payoff of strategy i is greater than (less than) the average payoff in the population. Rabanal and Friedman [21] point out that gradient dynamics follow the aforementioned sign-preserving dynamics when there are only two possible choices. In this case, given a particular type c, the dynamics become a replicator.
This method is also known as semidiscretization, since only one dimension (space) is discretized. For example, see [10].
This scheme is called upwind, because it follows the direction that gradient term flows. For a description of the scheme applied to a continuous action space, see Appendix B of [4].
This assumption simplifies the presentation of the model. KLS14 analyze exhaustively both complements and substitutes in their static analysis, but make similar assumptions to ease the exposition of their dynamics.
As mentioned earlier, I adopt gradient dynamics since they are a familiar specification that are a reasonable approximation for dynamics arising from a wide variety of evolutionary processes. KLS14 work with best-replies, which are equivalent to gradient dynamics at \(\hbox {d}\pi _i/\hbox {d}q_{i}=0\).
Defining the parameters \(c=0,\) \(\theta _{A}=\frac{3-a}{2}-a^{2}+a\sqrt{a^{2}+a-2}\), and \(\theta _{B}=\frac{1}{2}(\sqrt{9a^{2}-6a-3}-3a+3)\), they find the following
-
When \(a>1\) and \(0<\theta <\theta _{A}\), the model admits two equilibria: (i) all firms are type S, producing \(a/(3-2\theta )\), which is locally stable, and (ii) all firms are type P, producing 1 / 3, which is a saddle point.
-
When \(a>1\) and \(\theta _{A}<\theta <\theta _{B}\), the model admits two corner equilibria, and an inner equilibrium. Both corners equilibria are saddle points, and the inner equilibrium is defined as coexistence of both types. The type S produces \((a(\theta (2\theta -3)+2)-(\theta -3)\theta -2-2\sqrt{\psi })/\theta ^{2}\), where \(\psi =(a-1)(\theta -1)^{2}(a((\theta -1)\theta +1)+\theta -1).\) The type P responds as \(h(q)=(2-\theta )q/2(1-\theta )+(1-a-\theta )/(2(1-\theta )).\)
-
When \(a>1\) and \(\theta _{B}<\theta \le 1\), or when \(a=1\) and \(0<\theta \le 1\), the model admits two corner equilibria. All firms are type P, which is locally stable, or all firms are type S, which is a saddle point.
-
To ease the exposition of the paper, I do not rewrite most of the equations presented by KLS14 that describe firm production and firm types share.
The gradient term for the type S firm evaluated at the optimal share and firm’s best reply is \(\frac{4 \theta (15 \theta ^3 - 7 \theta ^4 - 48 \sqrt{\theta ^2 + \theta ^4} + \theta (48 + 2 \sqrt{\theta ^2 + \theta ^4}) + \theta ^2 (-6 + 7 \sqrt{\theta ^2 + \theta ^4}))}{5 (\theta - \theta ^2 + \sqrt{\theta ^2 + \theta ^4})^3}\).
I confirm their results by replacing \(\theta ^*\) and \(a=1+\theta ^*\) in the expressions that prove their Proposition 4. Out of the three families of equilibria described in footnote 11, only the first family survives in the long run.
See footnote 8.
In this numerical solution, I use a larger number of iterations compared to the public goods game since the initial distribution of types (uniform) is quite different compared to the stationary distribution.
References
Cheung M-W (2014) Pairwise comparison dynamics for games with continuous strategy space. J Econ Theory 153:344–375
Ely JC, Sandholm W (2005) Evolution in Bayesian games I: theory. Games Econ Behav 53:83–109
Friedman D, Ostrov DN (2010) Gradient dynamics in population games: some basic results. J Math Econ 46:691–707
Friedman D, Ostrov DN (2013) Evolutionary dynamics over continuous action spaces for population games that arise from symmetric two-player games. J Econ Theory 148:743–777
Friedman D, Singh N (2009) Equilibrium vengeance. Games Econ Behav 66:813–829
Güth W, Yaari M (1992) An evolutionary approach to explaining reciprocal behavior. In: Witt U (ed) Explaining process and change, approaches to evolutionary economics. The University of Michigan Press, Ann Arbor
Heifetz A, Shannon C, Spiegel Y (2007) The dynamic evolution of preferences. Econ Theory 32(2):251–286
Hofbauer J, Weibull J (1996) Evolutionary selection against dominated strategies. J Econ Theory 71:558–573
Hofbauer J, Oechssler J, Riedel F (2009) Brown–von Neumann–Nash dynamics: the continuous strategy case. Games Econ Behav 65:406–429
Iserles A (1996) A first course in the numerical analysis of differential equations. Cambridge University Press, Cambridge
Kopel M, Lamantia F, Szidarovszky F (2014) Evolutionary competition in a mixed market with socially concerned firms. J Econ Dyn Control 48:394–409
Kopel M, Brand B (2012) Socially responsible firms and endogenous choice of strategic incentives. Econ Model 29(3):982–989
Königstein M, Müller W (2001) Why firms should care for customers. Econ Lett 72(1):47–52
Lahkar R, Riedel F (2015) The logit dynamic for games with continuous strategy sets. Games Econ Behav. doi:10.1016/j.geb.2015.03.009
LeVeque RJ (2005) Numerical methods for conservation laws 2nd edn. Lectures in Mathematics ETH Zurich, Birkhauser Verlag, Basel
Lohman S, Oechssler J, Warneryd K (2001). Evolution and the social dilemma. Unpublished manuscript
Oechssler J, Riedel F (2001) Evolutionary dynamics on infinite strategy spaces. Econ Theory 17:141–162
Oechssler J, Riedel F (2002) On the dynamic foundation of evolutionary stability in continuous models. J Econ Theory 107:223–252
Ok R, Vega-Redondo F (2001) On the evolution of individualistic preferences: an incomplete information scenario. J Econ Theory 97:231–254
Perkins S, Leslie DS (2014) Stochastic fictitious play with continuous action sets. J Econ Theory 152:179–213
Rabanal JP, Friedman D (2014) Incomplete information, dynamic stability and the evolution of preferences: two examples. Dyn Games Appl 4:448–467
Ritzberger K, Weibull JW (1995) Evolutionary selection in normal-form games. Econometrica 63(6):1371–1399
Sandholm W (2010) Population games and evolutionary dynamics. MIT Press, Cambridge
Weibull W (1997) Evolutionary game theory. MIT Press, Cambridge
Acknowledgments
I am grateful to two anonymous referees of this journal, Mark Cheung, Dan Friedman, Michael Kopel and Olga Rabanal for their comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Rabanal, J.P. On the Evolution of Continuous Types Under Replicator and Gradient Dynamics: Two Examples. Dyn Games Appl 7, 76–92 (2017). https://doi.org/10.1007/s13235-015-0164-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13235-015-0164-0