A macroeconomic forecasting market

Macro-economic forecasts are used extensively in industry and government even though the historical accuracy and reliability is disputed. Modern information systems facilitate participatory, crowd-sourced processes that harness the collective intelligence. One instantiation of such wisdom of the crowds are prediction markets which have proven to successfully forecast the outcome of elections, sport events and product sales. Consequently we specifically design a prediction market for macro-economic variables in Germany. The proposed market design differs significantly from previous ones. It solves some of the known problems such as low liquidity and partition-dependence framing effects. The market acts as a mechanism not only to aggregate dispersed information but also to aggregate individual forecasts. It does so by incentivizing participation and rewards early, precise forecasts. Moreover, the market-platform is yet alone in aggregating these forecasts continuously and for a long time horizon. Analyzing the market-generated forecasts, we find that forecast accuracy improves constantly over time and that generated forecasts performed well in comparison to the Bloomberg-survey forecasts. From an individual perspective, market participants interact in a repeated decision-making environment closely resembling decision-making in financial markets. We analyze the impact of cognition, risk-aversion and confidence on trading activity and success.