This article describes an experiment in a Kydland/Prescott type of environment with cheap talk. Individual evolutionary learning (IEL) acts as a policy maker that makes inflation announcements and decides on actual inflation rates. IEL evolves a set of strategies based on the evaluation of their counterfactual payoffs measured in terms of disutility of inflation and unemployment. Two types of private agents make inflation forecasts. Type 1 agents are automated and they set their forecast equal to the announced inflation rate. Type 2 agents are human subjects who submit their inflation forecast and are rewarded based on their forecast error. The fraction of each type evolves over time based on their performance. Experimental economies result in outcomes that are better than the Nash equilibrium. This article is the first to use an automated policy maker that changes and adapts its rules over time in response to the environment in which human subjects make choices.
I would like to thank Andriy Baranskyy and Shiqu Zhou for excellent research assistance. The support from the CIGI/INET grants program is gratefully acknowledged, grant #5533 July 22, 2014.
Arifovic, J. (2014), "Evolving Better Strategies for Central Bank Communication: Evidence from the Laboratory", Experiments in Macroeconomics (Research in Experimental Economics, Vol. 17), Emerald Group Publishing Limited, pp. 229-258. https://doi.org/10.1108/S0193-230620140000017007Download as .RIS
Emerald Group Publishing Limited
Copyright © 2014 Emerald Group Publishing Limited