Focus on uncertain climate cliff-edge invites disaster

Scott Barrett from the Earth Institute at Columbia University says existing climate negotiation strategies have failed. Credit: Earth Institute

Scott Barrett from the Earth Institute at Columbia University says existing climate negotiation strategies have failed. Credit: Earth Institute

Countries trying to agree climate deals based on avoiding dangerous ‘cliff-edge’ limits will not work. That’s according to Scott Barrett and Astrid Dannenberg from the Earth Institute at Columbia University in New York, who’ve modelled the negotiations in a simple game. They found that uncertainty about the threshold at which their actions hurt everyone made players ignore the deals they made almost every time.  “We found that if the threshold for catastrophic climate change were certain, Mother Nature would essentially enforce an agreement to avoid the threshold,” Scott told me. “The fear of falling off of a cliff disciplined everyone to do what was needed, collectively, to avoid falling off. When the threshold was uncertain, this effect collapsed. Uncertainty about the impact made no difference at all.” And with science unable to be completely certain where the limit is, Scott and Astrid think that how climate deals are put together must change.

Scott got the idea that knowing where the cliff-edge is important because of the two different ways climate can affect the world. The first is ‘gradual’, where small emissions add a little to greenhouse gas levels in the air and raise global temperatures slightly. The second is ‘abrupt and catastrophic’, where small emissions push part of the world’s climate over a limit into a new state. “The second kinds of change are more important to welfare,” he said. “The effects tend to be uniformly negative – think of 5 meters of sea level rise, or the possibility of warm temperatures releasing methane stored in tundra, causing a rapid rise in temperature. The effects also occur quickly, meaning that adaptation will be harder. For both reasons it seemed to me that countries would have a stronger reason to prevent this kind of climate change than the gradual kind.”

Pushing the limit

Astrid Dannenberg from the Earth Institute at Columbia University recruited 400 students from the University of Magdeburg, Germany, for a game to test the effect of uncertainty on negotiations. Credit: Earth Institute

Astrid Dannenberg from the Earth Institute at Columbia University recruited 400 students from the University of Magdeburg, Germany, for a game to test the effect of uncertainty on negotiations. Credit: Earth Institute

Other researchers had looked at what abrupt climate change meant for cooperation in climate deals. However they assumed the limit where it happened was certain, and that the sides in the deal couldn’t communicate. “We knew both assumptions were important and completely at odds with the situation we face,” Scott said. To fix this, Astrid and Scott got 400 students at the University of Magdeburg, Germany, to play games that mimicked, but did not mention, climate negotiations.

The students were split into groups of 10, playing and communicating anonymously through computers. Astrid and Scott gave them each €31 spread across three different accounts. The first had €1 they could spend on €0.10 poker chips, reflecting easy steps to fight climate change. The second they had €10 they could spend on €1 poker chips, which reflect more difficult measures. That meant they could buy up to 20 chips for €11, with the first ten much cheaper than the second. In the third account they had €20 that could not be spent at all. At the end of the game, they went away with the amount of money left in their accounts, plus or minus an amount that depended on how many chips the whole group had bought. Each student gained €0.05 per chip the group bought, representing the benefits of fighting gradual climate change. However, unless the chip total reached a “threshold” they would have between €10 and €20 taken away from them, reflecting the impact of catastrophic change.

There were three stages to the game. First each group of 10 students negotiated what everyone would put in. “Specifically, we let them propose how much everyone ought to do and pledge how much each of them will do,” Scott said. “We suspected that this would be important, and it was.” Then they decided what they would actually pay, and lastly they were told what everyone had paid, and what the outcome of their choices would be. To test the effects of uncertainty in the limit and overall impact, the groups were randomly assigned to four different scenarios. In one, the students knew the exact size of the fine and threshold chip total, and in another they only knew a range for both. In the final two scenarios they knew one exactly and a range for the other.

Defeating cheating

When the threshold was fixed beforehand at 150 chips, almost all groups avoided catastrophic change, with students giving almost exactly the 15 chips they pledged on average. When the threshold was an unknown random number between 100 and 200 chips, students proposed around 167 chips per group. However students actually pledged slightly less, around 15 chips per player, then each student actually only gave around 8 chips each on average. That meant none of these groups avoided catastrophe – though a few reduced the risk slightly, they were still below the random threshold. “They proposed less than needed to eliminate the risk of catastrophe, they pledged less than they proposed, and they contributed less than they pledged,” Astrid noted.

Probability of catastrophe by treatment. Blue shows catastrophe is avoided, red that it cannot be avoided, and the other colours have a high chance of catastrophe, shown at the side of the graph. Catastrophe was avoided 8 of 10 times in the Certainty treatment and 10 of 10 times under Impact Uncertainty (I-Uncertainty). In contrast, the probability of catastrophe was reduced below 100% (to 93%) by only 1 of 10 groups under Threshold Uncertainty (T-Uncertainty) and by only 3 of 10 groups (to 91, 86, and 80%, respectively) under Impact-and-Threshold Uncertainty (IT-Uncertainty). In the four cases where the probability of catastrophe was reduced below 100%, the spinning wheel that decided the actual threshold determined that the threshold was crossed every time. Copyright: National Academy of Sciences

Probability of catastrophe by treatment. Blue shows catastrophe is avoided, red that it cannot be avoided, and the other colours have a high chance of catastrophe, shown at the side of the graph. Catastrophe was avoided 8 of 10 times in the Certainty treatment and 10 of 10 times under Impact Uncertainty (I-Uncertainty). In contrast, the probability of catastrophe was reduced below 100% (to 93%) by only 1 of 10 groups under Threshold Uncertainty (T-Uncertainty) and by only 3 of 10 groups (to 91, 86, and 80%, respectively) under Impact-and-Threshold Uncertainty (IT-Uncertainty). In the four cases where the probability of catastrophe was reduced below 100%, the spinning wheel that decided the actual threshold determined that the threshold was crossed every time. Copyright: National Academy of Sciences

Though this behaviour could be related to other international deals, like on fisheries, space junk, or antibiotic use, it seems to have a clear parallel to climate change. Scott pointed out that historical climate records show polar ice sheets have disappeared in the past when CO2 levels in the air were between 350 and 550 parts per million (ppm). Today, levels are over 390 ppm. “Our model can be interpreted as saying that everyone knows they should keep concentrations below 350.” he said. “However, they propose something like 400, knowing that the individual incentives to deliver 350 are so weak that proposing that would lack credibility. Then, when they make individual pledges, if you add them up they come to about 450. Finally, when they make their contributions they go beyond 550, guaranteeing that the threshold is crossed.”

To avoid this situation, Scott and Astrid suggest that measures like trade restrictions could be used against countries who don’t live up to their climate pledges, as was done successfully in the deal to fight ozone layer damage. “We can all agree that the negotiations have failed,” Scott said. “What we need, however, is a proper diagnosis of the reasons. Our research offers a diagnosis. We can’t prove it’s correct, but it’s for someone else to show that it’s wrong. Only when we know the diagnosis can we prescribe a treatment. Our research suggests that climate negotiations need to be framed differently. The same approach has been tried for more than 20 years. It’s time to try a different approach.”


Scenario Mean average overall chip proposal  Mean average student chip pledge  Mean average group contribution  Range of group contribution
Certainty 151.9 (1.57) 14.7 (0.51) 150.9 (7.69) 136–159
Impact Uncertainty 149.1 (4.98) 14.4 (0.80) 155.5 (2.92) 152–161
Threshold Uncertainty  166.3 (9.85) 15.8 (1.69) 77.2 (16.67) 55–107
Impact-and-Threshold Uncertainty 167.0 (10.40) 15.5 (2.07) 79.9 (26.90) 44–120

Under threshold uncertainty, players propose that more be contributed, compared with the threshold certainty treatments, but end up contributing less. Variability in proposals, pledges, and especially contributions is also greater for the threshold uncertainty treatments. Standard deviations are given in brackets.

Journal Reference

Scott Barrett and Astrid Dannenberg (2012). Climate negotiations under scientific uncertainty PNAS Proceedings of the National Academy of Sciences of the United States of America DOI: 10.1073/pnas.1208417109

About these ads

3 Responses to “Focus on uncertain climate cliff-edge invites disaster”

  1. Another Week of GW News, October 14, 2012 – A Few Things Ill Considered Says:

    […] 2012/10/13: SimpleC: Focus on uncertain climate cliff-edge invites disaster […]

  2. » Climate Negotiations and the Prisoner’s Dilemma Pike Research Says:

    […] causing global warming.  None of these negotiations have made much progress.  And, according to a study by environmental scientists Scott Barrett and Astrid Dannenberg of the Earth Institute of Columbia University in New York, published in the October 2012 […]

  3. Can we trust climate models? « Simple Climate Says:

    […] 4.1°C higher than 100 years previously by the end of the 21st century. You might like a narrower uncertainty range – and it’s possible scientists might be able to give us one. But Xuefeng Cui stressed […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 196 other followers

%d bloggers like this: