Aggregate Dynamics in a Large Virtual Economy: Prices and Real Activity in Team Fortress 2* - eScholarship

Aggregate Dynamics in a Large Virtual Economy: Prices and Real Activity in Team Fortress 2 ∗ Matt Baumer † Curtis Kephart ‡ University of California Santa Cruz March 23, 2015 Abstract We examine economic activity in a large virtual economy which was designed to allow de- centralized barter as the sole exchange institution. We find that a small subset of goods emerges endogenously which act of media of exchange. Our analysis includes estimation of spot exchange rates between these numerous money goods and we develop methods which allows us to price all goods and track inflation. We then calculate nominal growth and its components. We find that per-captita real wealth is an increasing component of nominal growth. Separately, we find evidence of a certain form of nominal price ridigity– the price of an item commonly priced in a particular money good tends to move with the exchange rate of that money good. We also find that announcements made by the economic planners can induce speculation leading to localized asset price bubbles. Keywords: Virtual economy, barter system, national income accounting, wealth measurement, price level, inflation. JEL Codes: E01 E31 P44 Introduction The object of our study is the virtual economy of Team Fortress 2 (TF2) developed and overseen by Valve Software. This economy and others like it hold great potential for researchers: millions of users engaging in billions of economic transactions involving thousands of different types of goods; the game designers are near-omnipotent social planners able to create and destroy goods The authors would like to thank the valuable assistance provided to us by Valve Corporation, including Brandon Reinhart, Kyle Davis, and particularly Kristian Miller for his boundless support and deep insight into the Valve virtual economy. The research was partially funded by Valve Corporation, the University of California Santa Cruz, and LEEPS Lab. Valve Corporation additionally provided technology grants and access to anonymized proprietary data. The authors have agreements in place that entitle Valve Corporation to review our work before it being submitted for publication. The authors would also like to thank Danny Oberhaus at Statistics New Zealand and Anya Stockburger at the Bureau of Labor Statistics for offering their expertise in the development of our price index. Matthew Jee, Sergio Ortiz, and Matthew Browne provided excellent computer programming support. Economics Department, University of California Santa Cruz, CA 95064, mbaumer@ucsc.edu Economics Department, University of California Santa Cruz, CA 95064, Curtis.Kephart@googlemail.com