Post on 12-Jul-2020
Prediction Markets and Beliefs about Climate: Results from Agent-Based SimulationsJonathan M. Gilligan,1 John J. Nay,2 Martin Van der Linden3
1. Earth & Environmental Sciences, Vanderbilt U., Nashville TN, USA; 2. School of Engineering, Vanderbilt U., Nashville TN, USA; 3. Economics, Vanderbilt U., Nashville TN, USA
PA23B-2191
Further Reading: Contact: https://my.vanderbilt.edu/jonathangilligan
jonathan.gilligan@vanderbilt.edu
Premise• Many people unconvinced by scientific evidence of climate
change.• Cultural cognition:
• Ideological opposition to trusting climate scientists.• Trust in free markets
• J. Annan and others: • Bets and prediction markets• “Put your money where your mouth is.”• Good for testing sincerity of doubters.
• But can prediction markets change minds of sincere doubters?MP Vandenbergh, KT Raimi, & JM Gilligan. UCLA Law Rev. 61, 1962 (2014).
Model• Global Temperature:
• Start with historical temperature record.• Project future climates under two alternate theories:
• Temperature proportional to log(CO2) or Total Solar Irradiance (conventional science vs. popular alternative among doubters)
• Add AR(1) noise (best ARMA noise model for historical record).• Traders:
• Traders believe temperature depends on CO2 or TSI• Parameterize trader ideology (resistance to changing mind)
and tolerance for investment risk.• Traders compare their profits to others in their social network
• Network characterized by # edges (connections) and segmentation (are traders with different initial beliefs connected?)
• Prediction Market:• Continuous-Double Auction (typical of large stock exchange)• Climate futures:
• Bet on temperature six years in future.• During six-year period, traders buy and sell futures.• Every year: Bayesian updating of traders personal predictions for
future temperature based on current year’s temperature.• At end of six-year period, winners collect money• Traders revise beliefs about climate models,
based on ideology and beliefs of top earners in their social network.
Trader Network
Results:• Traders beliefs converge
toward true model• Convergence is slow• Depends on
network topology and trader ideology
Discussion:• Proof of concept, not a rigorous analysis
of prediction markets• Prediction markets might help
persuade doubters who do not trust scientists.
• However, persuasion might be too slow for stabilization targets.
• Echo-chamber (segmented social networks).
• This simulation uses very simple belief model.
• Next steps:• True Bayesian belief model• Multiple influences (neighbors and share
prices)• Compare psychological and economic
theories of belief.• Compare different types of securities.
Historical Simulation:• Use actual temperatures 1880-2015.• Betting from 1931–2014.• As warming signal begins to show
up in 1980s, traders begin to converge toward log(CO2) model.
CO2: IAMC RCP database (8.5 shown); TSI: VM Velasco Herrera, B Mendoza,
& G Velasco Herrera, New Astronomy 34, 221 (2015)