Aggregation mechanisms for crowd predictions

Area of Science:

• Decision Sciences
• Behavioral Economics
• Information Aggregation

Background:

• The "wisdom of crowds" effect suggests aggregated forecasts outperform individual experts.
• Effective aggregation mechanisms are crucial for leveraging collective intelligence.
• Asymmetric information structures are common in real-world forecasting scenarios.

Purpose of the Study:

• To compare the efficacy of various aggregation mechanisms in an asymmetric information setting.
• To determine which aggregation method best harnesses the "wisdom of crowds" for prediction.
• To analyze how information asymmetry impacts forecasting accuracy and excess returns.

Main Methods:

• A controlled experiment was conducted to compare aggregation methods.
• Methods evaluated included arithmetic/geometric means, medians, continuous double auction markets, and sealed bid-offer call markets.
• Participants possessed different subsets of information to estimate an asset's value.

Main Results:

• Continuous double auction market prices significantly outperformed all other aggregation methods.
• The best-informed participants were able to generate excess returns.
• Less informed participants' forecasts did not significantly contribute to the aggregate's accuracy.

Conclusions:

• Continuous double auction markets provide a superior mechanism for aggregating dispersed information.
• Market-based aggregation effectively utilizes collective intelligence, even with asymmetric information.
• Understanding information distribution is key to optimizing forecasting and market efficiency.