Joshua F. Boitnott Contact Information: Simon Fraser University Web: https://sites.google.com/site/joshuafboitnott/ Department of Economics Phone: (306)715-2760 8888 University Drive E-mail : jfb@sfu.ca Burnaby, BC Canada V5A 1S6 Education: PhD Economics Simon Fraser University 2009-Present Dissertation: Three Essays on Learning Algorithms and Experimental Economics Primary Supervisor: Jasmina Arifovic Expected Completion Date: June 2016 MA Economics Simon Fraser University 2008-2009 BA Economics Simon Fraser University 2005-2008 Research and Teaching Fields: Learning Algorithms, Public Economics, Experimental Economics, Development Economics Conference Presentations: 2014 Canadian Economics Association “Learning Correlated Equilibria: An Evolutionary Approach” Vancouver, BC Teaching Experience: 2013 Instructor, SFU The World Economy 2013 Tutor Marker, SFU Macroeconomic Theory & Policy (MA course) 2008-2013 Teaching Assistant, SFU Principle of Microeconomics, Microeconomic Theory I: Competitive Behavior Principles of Macroeconomics, The World Economy (x3), Economics and Government, Environmental Economics, Canadian Macroeconomic Policy, Money and Banking (x2), Introduction to Economic Concepts and Issues, Economic History of Canada, Economic Development (x2), Economics of Popular Culture, Public Economics: Taxation, Public Economics: The Role of Government Research Experience: 2012 Research Assistant to Jasmina Arifovic and John Ledyard (Cal Tech) 2013 Research Associate with Centre for Researchin Adaptive Behaviour in Economics (CRABE) Volunteer Experience: 2013 Provincial Judge, Odyssey of the Mind 2006 - 2011 Provincial Head Judge, Odyssey of the Mind Computer Skills: Advanced: Basic: MatLab, Excel, LaTeX , Word, PowerPoint STATA, R, HTML Personal Information: Citizenship: U.S.A. Permanent Resident: Canada Languages: Fluent English, Basic Spanish Port Moody, BC Various Locations, BC Research Papers: “The Behavior of Learning Algorithms in Experimental Games with Congestion,” Job Market Paper Abstract:Learning algorithms represent an important step in modeling the dynamic behavior of human subjects observed in experimental settings, yet many existing algorithms have limited application outside of a single game or across different experimental settings when parameter values are fixed. Furthermore, while previous studies have investigated algorithmic agent performance relative to experimental data, they have relied upon data from a single source. In this study, we use a horse-race approach to compare the performance of four learning algorithms (Impulse-Matching Learning, Self-Tuning Experience Weighted Attractions, and Individual Evolutionary Learning(x2)) across five different sets of experimental data generated by multiple researchers where the subjects face congestion. We examine the ability of each algorithm to replicate experimental data independent of the training data set used to determine the initial parameter values via the root mean squared error for each algorithm relative to the data produced by the human subjects and the ability of the algorithms to match multiple experimental equilibria despite adjusting the payoffs, the number of players, and the number of repetitions played. We determined that the Individual Evolutionary Learning model with a modified set of assumptions has the closest fit of the regularities seen in the experimental data when human subjects were given full information measured by the average mean squared distance, but not in experiments with limited information. “Learning Correlated Equilibria: An Evolutionary Approach,” joint with Jasmina Arifovic and John Duffy (UC Irvine) Abstract: Correlated equilibrium (Aumann 1974, 1987) is an important generalization of the Nash equilibrium concept for multiplayer non-cooperative games. In a correlated equilibrium, players rationally condition their strategies on realizations of a common external randomization device and, as a consequence are able to achieve payoffs that Pareto dominate any of the game’s Nash equilibria. In this paper we explore whether such correlated equilibria can be learned over time using evolutionary learning model where agents do not start with any knowledge of the distribution of random draws made by the external randomization device. Furthermore, we validate our learning algorithm findings by comparing the limiting behavior of simulations of our algorithm with both the correlated equilibrium of the game and the behavior of human subjects that play that same game. Our results suggest that our evolutionary learning model is capable of learning the correlated equilibria of these games in a manner that approximates well the learning behavior of human subjects. “Beliefs, Learning, and the Linear Public Goods Game: Adjusting the Individual Evolutionary Learning with Other Regarding Preferences Model,” (In Progress) References: Jasmina Arifovic Simon Fraser University E-mail : arifovic@sfu.ca (778)782-5603 Erik Kimbrough Simon Fraser University E-mail : ekimbrough@gmail.com (778)782-9383 John Duffy University of California, Irvine E-mail : duffy@uci.edu (949)824-8341