A market mechanism for synthesizing predictions of physical climate risks

The usefulness of predictions of physical climate risks to the financial sector is now appreciated but climate forecasting can also learn from the ability of financial markets to aggregate distributed information and expertise.  

CRUCIAL is an initiative that uses “prediction markets” — markets designed to discover and synthesize information rather than transfer assets or risks — to elicit and aggregate expert judgements about climate-related risks. Teams of expert participants, from academia and the private sector, are allocated credits which they can use to trade contracts tied to climate-related outcomes. The prices of these contracts can be interpreted as probabilities that evolve in real time as new information becomes available to participants.

Using prediction markets to aggregate climate forecasts means that the users of the forecasts do not have to select a single provider. This is an important feature because, for longer horizon forecasts, providers cannot demonstrate their competence with a statistically meaningful track record of accurate predictions. Instead, prediction markets directly reward forecasters for the contributions they make to improving the accuracy of collective forecasts.

CRUCIAL’s platform has been used to run markets that predict seasonal temperatures and rainfall, crop yields, El Niño-Southern Oscillation and Atlantic hurricane activity for horizons of up to 18 months ahead. These pilot markets produced forecasts that were consistent with good probabilistic calibration (reliability). CRUCIAL plans further markets with longer prediction horizons.

In a world where historic statistics of climate risks are not necessarily a good indication of future risks, prediction markets provide a mechanism which can combine information from historical data, climate models, and more tacit forms of expertise into quantitative probabilistic forecasts. Prediction markets have the potential to become a new type of scientific institution for synthesizing, summarizing and disseminating diverse climate expertise and different modelling approaches. Prediction markets can also be used to allocate funding for climate forecasting more efficiently than peer-reviewed grants. Such markets could allow experts from many different disciplines and both academia and the private sector to contribute effectively to the generation of probabilistic predictions of physical climate risks.