Prediction markets are often highly effective means of bringing together diverse perspectives, and have been shown to outperform individuals and teams in a wide range of tasks. A recent study from the University of Lancaster suggests they could also be effective in helping us assess future climate risks.
The authors argue that while organizations generally accept that they have to take into account the risks climate change poses to their strategy, they often lack the forward-looking information required to make strategic decisions.
“The institutional arrangements under which climate-risk information is currently provided mirrors the incentive problems and conflicts of interest that prevailed in the credit-rating industry prior to the 2007/8 financial crisis,” the researchers explain.
“In order to make sense of emissions scenarios and to support planning and decision-making, organizations have a pressing need for this type of forward-looking expert risk information.”
Understanding climate risk
The researchers suggest that fully understanding climate risk requires expertise from a diverse range of areas, including economics, political science, and policy, as well as local knowledge from each country.
“Prediction markets incentivize and reward participants with distinct expertise and information to come forward—and they offer a level playing field for experts from these complementary fields of expertise,” they continue.
“If providers of climate forecasts are paid upfront irrespective of accuracy, you don’t need to be an economist to spot the problem with that arrangement.”
Risk assessment
The authors believe that prediction markets can overcome some of the shortfalls in how forward-looking risk information is currently provided, which is obviously going to become even more important in the years ahead.
Prediction markets are designed so that incentives are given to those with vital bits of information to disclose that information. They then aggregate the information via the buying and selling of contracts that provide a fixed payoff should that event occur.
An outcome of interest—such as average CO2 concentration in the year 2040, for example—is partitioned into intervals. Expert participants compare the results of their own modeling with the prices of these intervals, and purchase or sell claims on these intervals if their model suggests the price is too low or too high.
Regulatory obstacles
The researchers believe that these kinds of long-range markets haven’t been developed to date due to regulatory obstacles. They’re confident, however, that the markets can be developed in such a way as to overcome these obstacles by ensuring that the “pay-to-play” aspect of the markets, which ensures that the losses of less-well-informed participants help to fund the winnings of the better-informed participants, are stripped out.
Instead, we can design markets so that research funding is distributed to modelers and experts in a way that is more consistent with the principles of effective altruism. In other words, an initial stake could be provided by a sponsor, with this then distributed to participants based on the information they contribute to the market.
They also propose selectively recruiting participants into the markets to ensure that sufficient diversity is included, with a range of expertise to ensure they are able to aggregate diverse sources of information.
