Learning by Trading in a Macro-Economic Forecasting Game

Macroeconomic forecasts are used extensively in industry and government even though the historical accuracy and reliability is questionable. Moreover, professional forecasters lack a test environment in which they can test their forecasting ability. We design a play-money market game for economic variables that aggregates macro-economic information. We analyse participation and learning in such an online game. In our platform learning occurs on three levels. First, participants learn how to trade in a continuous double auction just as in stock markets. Second, they learn about their own macroeconomic forecasting ability in comparison to their peers. Third, by following market forecasts they learn about the current state of the economy. We show that the game successfully aggregates macroeconomic information as forecast errors fall over the prediction horizon. The game-generated forecasts compare well to the Bloombergsurvey forecasts, the industry standard.