How long to Pareto efficiency?
We consider uncoupled dynamics (each player knows only his own payoff function) that reach outcomes that are Pareto efficient and individually rational. We show that in the worst case the number of periods it takes to reach these outcomes must be exponential in the number of players and hence the same number of periods it takes to reach Nash equilibria. For social welfare maximizing outcomes we provide a tight bound on the minimal number of steps required for reaching such an outcome by uncoupled dynamics.
- Arieli I, Babichenko Y (2011) Average testing and the efficient boundary. Discussion Paper 567, Center for the Study of Rationality, Hebrew University.Google Scholar
- Conitzer V, Sandholm T (2004) Communication complexity as a lower bound for learning in games. In: Brody CE (ed.) Machine Learning, Proceedings of the Twenty-First International Conference (ICML 2004), Banff, July 2004, ACM, New York, pp 185-192.Google Scholar
- Kushilevitz E, Nisan N (1997) Communication complexity. Cambridge University Press, CambridgeGoogle Scholar
- Marden JR, Young HP, Pao LY (2011) Achieving Pareto optimality through distributed learning, Economics series working papers, 557. University of Oxford, OxfordGoogle Scholar
- Pradelski B, Young HP (2010) Efficiency and equilibrium in trial and error learning, Economics series working papers, 480. University of Oxford, OxfordGoogle Scholar