Climate Change and the risk of catastrophe

HSBC Private Bank (UK) Limited - December 2009

Imagine you have flu. You learn from your doctor that, if you spend three days in bed, you will certainly recover. If you carry on as normal, you will also very probably recover just as quickly. However, there is a small chance - one in a hundred - that complications will set in and you will die. What should you do? Obviously, you should spend those days in bed. True, it is most likely that you would be better off carrying on as normal, since it is most likely that you will recover anyway, and you will not waste three days in bed. But carrying on as normal carries a small risk of death that is quite enough to outweigh the wasted days.

The moral of this story is that, when you face uncertainty, what matters is not what is most likely to happen but the expected value of what will happen. ‘Expected value’ is the technical name for a sort of weighted average. To find the expected value of an act, such as carrying on as normal, take the weighted average value of each of the act’s possible outcomes, weighting each by its probability. Suppose (just an illustration) that the value of living a day normally is 10 units, and the value of living a day stuck in bed is 9 units. If you go to bed for three days, you lose 3 units. If you carry on as normal, you probably lose nothing, but you have a 1/100 chance of losing the rest of your life. If you lose, say, 10,000 days, you lose 100,000 units of value. The expected value of your loss is 100,000 units weighted by the 1/100 probability. That is an expected loss of 1,000 units. It is far better to go to bed and lose only 3 units.

In making predictions about climate change, the scientific community has generally concentrated on what is most likely to happen. Suppose we reduce emissions of greenhouse gases enough to hold their concentration in the atmosphere at a level equivalent to 550 parts per million of carbon dioxide. Evidence presented by the Intergovernmental Panel on Climate Change suggests the atmosphere is likely to warm in the long run by two or three degrees. That amount of warming seems manageable. So if you consider only what is most likely to happen, 550 ppm may seem a reasonable target to aim at.

However, the same evidence also suggests there is real chance - perhaps one or two in a hundred - that 550 ppm of carbon dioxide is enough to warm the climate by more than ten degrees. The last time the earth was ten degrees hotter was about 20 million years ago, long before there were human beings. We have no idea what might happen if temperatures reach those levels. There could be an extraordinary catastrophe. Civilization might collapse; humanity might even become extinct, or be reduced to a few struggling survivors. Those seem to be such dreadfully bad possibilities that, even multiplied by the small probability of their occurring, they would dominate the calculation of expected value. You would not take such a risk with your own life. The expected value of loss resulting from 550 ppm of carbon dioxide may depend much less on what is likely to happen than on what is unlikely to happen. 550 ppm may be far too dangerous a level.

This sets us two intellectual challenges. The first is for science. From science we need, not so much a judgement of what is likely to happen, as a judgement about the risk of extreme global warming. This gives science an even harder task than it has accepted already. Predicting what is likely to happen is hard enough, but data for estimating unlikely possibilities are especially hard to come by. Moreover, scientists face a problem caused by the powerful political forces that are ranged against them. We spend lots of money to reduce the minute risk of dying on an aeroplane, but a scientist who recommends reducing greenhouse gas emissions on the basis of a small risk of catastrophe can easily be branded as a crank. It is not surprising that scientists generally do not take on this task.

The second challenge is for moral philosophy. Since a catastrophe is possible, to judge the expected value of what may happen, we need to judge the value - in this case the badness - of a catastrophe. Values are a matter for moral philosophy. Our response to climate change is a moral matter.

The present rich of the world are being asked to make sacrifices for the sake of the future poor. Why should we do that? The only reason is that it is our moral responsibility. The analytical methods of philosophy can help us think about what our moral responsibility to future people really is.

They cannot at present deliver definitive answers. Just as the science of climate change contains a great many uncertainties and much room for debate, so does the moral philosophy of climate change. How bad would a catastrophe be? Various bad things might happen: the loss of civilization, the destruction of nature, the end of the human species, the early deaths of billions of people, and others. Setting a value on any of these losses is extremely difficult, and we have no consensus about how to do it.

We have made some progress. One part of a catastrophe’s badness is that, if it occurs, billions of people will die before their time. We need to judge the value of all those lives destroyed. It is not only a catastrophe that raises the question of the value of people’s lives. Climate change is already causing deaths through natural disasters and heat waves. For instance, it probably contributed to the 2003 heat wave in Europe, which killed 35,000 people. So we urgently need to know how bad it is for a person to die early.

That seems implausible to many people. But the real objection is a theoretical one. Just as a harmless extinction harms no one, so it also benefits no one. The person-affecting view therefore implies that a harmless extinction is no worse than humanity’s continuing existence, and no better either. That seems to mean it is equally as good. Now compare three possibilities: A, a harmless extinction; B, no extinction, and everyone lives happily ever after; C, no extinction (and the same people live as in B), and everyone lives less happily ever after. The person-affecting view tells us that A is equally as good as B, and also that A is equally as good as C. It follows that B is equally as good as C. But that is plainly false: B, where everyone lives happily, is better than C, where everyone lives less happily. The person-affecting view implies this falsehood. Therefore it is false itself.

That is a hard but approachable question. The value of human life is already a recognized element in public policy. For example, the World Health Organization has developed a measure of the ‘the burden of disease’, designed to measure the harm that is done to a population of people by diseases. It includes the harm that is suffered by those whom disease kills. Roughly, if a disease kills you, the WHO reckons the harm you suffer to be the number of years of life you lose, adjusted for the level of health you would have enjoyed during those years. This is a broadly plausible idea. The WHO already uses it to measure the burden of climate change, taking account of all the killing that climate change will do.

The possibility of catastrophe raises a much harder question. If humanity is extinguished or its population crashes, vast numbers of people who would otherwise have lived in future ages, will not come into existence at all. The absence of all the future of humanity seems to be another terrible part of the badness of a catastrophe. But that raises a puzzle. The nonexistence of a person harms nobody, since there is nobody who does not exist. How can nonexistence be a bad thing if it harms nobody?

Some philosophers insist it cannot be; nothing can be bad unless it harms someone. Their view - called the ‘person-affecting’ view - means that the extinction of humanity would be bad only in so far as people are killed or suffer in the course of it. Imagine a process of extinction that kills no one: imagine women just stop conceiving children, and the last people are somehow able to live normally to old age without the support of a younger generation. (Moral philosophers often test their theories with thought experiments like this.) Call this a ‘harmless extinction’. A harmless extinction harms nobody, so on the person-affecting view it would be no worse than humanity’s continuing existence.

This argument is not conclusive; the person-affecting view can still be defended. I believe the defence fails, but that is still a matter of debate. If it fails, we have to recognize that the nonexistence of future people is bad (or possibly good). We need to judge how bad (or good) it is in quantitative terms. This remains one of the hardest and most debated problems in practical philosophy. We are far from meeting the challenge to moral philosophy.

Until the two challenges are met, we face an incalculable risk of a catastrophe that seems dreadfully bad, but whose badness we cannot judge. Knowing how to respond to this condition of ignorance is itself a challenge. You might think we should simply minimize the risk, on precautionary grounds. But however much of our resources we devoted to reducing the risk posed by climate change, we could always find some way of reducing it further, at a further cost. The risk would be truly minimized only when we were left with resources for nothing else. Then humanity would have a better chance of survival, but humans would lead miserable lives. This cannot be the right response. There has to be a balance of costs and benefits.

At present, the world is not doing enough to limit climate change, even ignoring the risk of catastrophe. So the practical challenge at present is the political one of persuading governments and people to do more. But once we start to respond properly to climate change, those two intellectual challenges must be met.