Article 2 of the UN Framework Convention on Climate Change (UNFCCC) states very simply that we need to stabilise greenhouse gas concentrations at levels below those which would cause dangerous anthropogenic perturbation of the climate system. It goes on to talk specifically about allowing eco-systems to adapt naturally, that agricultural production should be unaffected, and that we need to respond in a way which is sustainable from an economic standpoint. I think this last point is extremely important. Embedded in this is that we have to look at human health issues and water issues. The problem is that the word "dangerous" in Article 2 is a very subjective word, it is not something for the scientific community to debate. What we need to do as scientists is present to governments what would happen at different levels of greenhouse gas concentrations, and they, the governments of the world, have to decide at what level they want to stabilise.

Just to put a couple of numbers in perspective, pre-industrial levels of carbon dioxide were 280 parts per million (ppm), while today they are 350 ppm. What is a safe level? Do we need to stabilise at today's level, or twice today's level, or three times today's level? The European Union has argued that we need to stabilise at 550 ppm, that would be double the pre-industrial number and about another 200 ppm above today. The point is, it is not a scientific debate. It is a political debate about what is safe and therefore what is not dangerous.

What is the greenhouse effect? The reason the earth is habitable is that we already have greenhouse gases in the earth's atmosphere, primarily water vapour, carbon dioxide and a few other minor gases. Without these gases in the atmosphere we would live in a very frozen world, 28—33°C colder than it is today. The issue at the moment is that we are putting more greenhouse gases into the atmosphere, primarily carbon dioxide and methane, mainly due to energy use, although methane also comes from agriculture. This is stopping the world's infra-red radiation from being emitted out into space. It is trapped like a thermal blanket, and the energy is re-radiated back to the earth's surface. Hence, the more greenhouse gases we put into the earth's atmosphere the more we see an increase in the earth's temperature.

I believe, as does the scientific community, that we have already seen changes in the climate due to the increasing greenhouse gas concentration in the atmosphere. This includes changes in temperature as well as precipitation, and it will have adverse effects on ecological systems, human health and socio-economic systems such as agriculture. What are some of the key findings of the Intergovernmental Panel on Climate Change (IPCC)? First, we are changing the atmospheric concentrations of greenhouse gases, primarily carbon dioxide and methane. In the case of carbon dioxide, this is mainly due to the use of fossil fuels and also land use change, currently known as deforestation, in the tropics. The concentration of aerosols is also increasing, also due to the combustion of fossil fuels. Sulphur dioxide forms sulphate aerosols, and locally (but not globally) there is a small cooling effect due to these. Overall though, the greenhouse effect of carbon dioxide and methane far overwhelms the small cooling effect of sulphur dioxide.

There is absolutely no question the earth's climate has warmed over the last 100 years. The average temperature of the earth's surface has gone up 0.5°C, the land surface 0.8°C. If we try and explain the observed changes in temperature we cannot explain them by natural variability alone. They are broadly consistent with what we would expect to occur due to the greenhouse effect. That is to say, due to the way we humans have increased the atmospheric concentrations of CO₂.

But looking back at the last hundred years is not the key issue. What will happen in the future if we do not control our emissions of greenhouse gases? The emissions of greenhouse gases are dependent on many factors, and you almost need a crystal ball to see what is going to happen in 100 years time. But basically emissions are going to be dependent upon population growth, economic growth, the availability of energy, the price of energy and the types of energy that we use. We have come up with some plausible scenarios, and we believe that over the next 100 years there will be a significant increase in carbon dioxide concentrations.

If we do not take action to try and limit carbon dioxide emissions we would then project the climate would increase between 1 and 3.5°C over the next 100 years. There would be significant changes in precipitation patterns and an increase in sea level. Even the lower end of our projection, 1°C in 100 years, is still a rate of change larger than anything we have observed in the last 10 000 years.

The last point that we need to remember, and it is a very key point for policy makers, is that the atmospheric residence time of carbon dioxide is well over 100 years. It does not have a simple life time, actually it has a multiple lifetime, but an average life time could be viewed as 100 years. If we wait for cause and effect to be established between the emissions of greenhouse gases and a changed climate, and we do not like the impact of that changed climate on ecological systems and human health, we could not reverse the damage, not in decades, not in centuries, in some cases even in millennia. This is the reason why we cannot wait for perfect knowledge.

I would like to say at this stage that there are significant uncertainties in our understanding of climate change. But what the policy making world has to do is consider what is the weight of evidence, to what degree do they want to take precautionary action and start to move away from the intensive use of fossil fuels. The key issues are human health (e.g. heat stress mortality, vector borne diseases), sea level rise and the displacement of people and loss of land, agricultural production and forestry, and ecological systems (e.g. the forests and coral reefs of the world).

A key point is that most greenhouse gases have been emitted into the atmosphere due to actions in the developed world, i.e. the OECD countries and the former Soviet Union. But the countries that are most vulnerable to climate change are the developing countries, primarily in the tropics and sub-tropics. The good news in the view of the IPCC, and this is an IPCC view of the world which is shared by the World Bank, but which was generated by the scientific community, is that we believe there is a wide range of cost effective technology in both energy supply and energy demand that means we can start to address this particular issue.

We also believe that we need to have international agreement on this. At the moment, the UNFCCC is only a framework protocol and it primarily addresses the industrialised world and not the developing countries. That is primarily because the greenhouse gases have been produced by the industrialised countries, and it is a recognition that the industrialised countries have the technology, and the economic and the institutional capability to deal with this. At some stage, however, developing countries will also have to be part of the deal, otherwise we cannot stabilise greenhouse gases and hence stabilise the earth's climate.

There is a whole series of actions which we would call a "no regret" strategy. One of the key environmental challenges facing developing countries like China today is that when they burn low quality coal they see significant emissions of particulates and sulphur dioxide, and therefore they have to worry about the human health issues of particulates, and the human health and ecological issues associated with acid deposition and sulphate aerosols. The argument we would use inside the World Bank and also IPCC is that China and other countries need to get to grips with the local and regional air pollution issues and hence have to move away from the way they produce and use fossil fuels today.

Looking at the data showing atmospheric concentrations of carbon dioxide and methane, we can see that carbon dioxide concentration has increased 30% since the industrial revolution, as mentioned above, primarily due to the combustion of fossil fuels and deforestation. Methane concentration has more than doubled. One of the sources of methane is leakage from natural gas pipelines and leakage from coal mines, although there are other major sources such as rice growing and cattle.

If you look at CO₂ concentration over the last 160 000 years, you can see that it has changed between about 180 ppm in glacial periods to about 280 ppm in inter-glacial periods. We live in an inter-glacial period now. So CO₂ has gone up and down over the last 160 000 years, we know this because we can accurately measure CO₂ that is trapped in the air bubbles in ice in Greenland and Antarctica. Temperature changes correlate perfectly with these variations. The same is true of methane concentrations, that is, as the atmospheric concentrations of methane and carbon dioxide go up, temperature goes up, and vice versa. So there is a strong correlation between the atmospheric concentration of greenhouse gases and temperature.

From pre-industrial levels of CO₂ concentration of 280 ppm, we project they will go up to at least 500 if not 1000 ppm over the next 100 years. This is an incredible rate of change relative to anything we have seen in the pre-historic records. It is almost impossible to believe that temperature will not rise to some degree, definitely not linearly, but to some degree. What do we believe will happen?

I will use the example of the United States, but this is the type of impact we would expect to see around the world. We would see warmer temperatures, more rainfall in winter, more severe droughts, and also we would see heavier precipitation. In the USA we would define that as more than 2 inches of rain in a 24 hour period. In addition, day to day variations in temperature would be minimised.

Have we seen any of these effects so far? The answer is we have seen all of these, except we cannot demonstrate the drought factors very well. We have seen higher temperatures, we have seen more precipitation in winter, we have seen more extreme rainfall events, that is why we have seen over the last decade or so more floods and more droughts. Can we prove that it is due to global warming? No. Is it consistent with global warming theory? Yes. That is what I mean by saying that we have to consider the weight of scientific evidence.

Temperature measurements have been made in many places all over the world, and in all but one or two places it has become warmer. The average increase for the land area is 0.8°C warmer over the last 100 years. Precipitation has changed, with an increase throughout much of North America and in northern Siberia. It has become much drier throughout Africa and parts of Southeast Asia in the last 100 years. The surface temperature average over the globe for the last 100 years has increased about 0.5 to 0.6°C, if you average the land and the ocean. Theoretical models show that this is consistent with a change in temperature of between 2 and 3°C in a world where CO₂ concentrations have doubled. The medium line is around 3 to 3.5°C, that is to say, if CO₂ concentration doubles at equilibrium, it would take a number of years to come to equilibrium, actually more than a century, and you would see about a 3°C temperature rise.

So there is a broad consistency. But the data that is most convincing in saying that we have got an understanding is where we simulate including the effect of both carbon dioxide (which warms the world) and sulphur dioxide (which cools it). If you simulate carbon dioxide effects only this shows that it should get a lot warmer in the two hemispheres fairly equally. Observations do not show that. However, including sulphur dioxide shows an inter-hemispheric asymmetry and lower temperature rise, just like the observations.

Taking into account all observed changes in atmospheric composition led the IPCC to say in 1997 that we believe there is now a human induced effect on the earth's climate system. One of the arguments has been that this will lead to more hurricanes, tornadoes and cyclones. The hurricane record for North America shows no observable change in hurricanes. However, this is one of the things that has really got the insurance and the reinsurance industry very nervous in North America. The theoretical models for the next 100 years are ambiguous on this point, some models say there will be an increase in hurricanes others say there will not be. What we are sure about is that all models show that there will be a change in temperature and a change in precipitation, but there is an inconsistency between the models on whether there will or will not be a change in hurricane frequency.

So to summarise, there could be a 3.5°C temperature change by 2100, and a 15 to 95 cm increase in sea level. These are not certainties because it is hard to project the increase in carbon dioxide and other greenhouse gases, there are uncertainties about population, economic growth and technology development. Also we are not quite convinced that we know precisely the so-called climate sensitivity. That is why we have given a range of 1 to 3.5°C, if the atmospheric concentration of carbon dioxide rises from 350 ppm today to almost 700 ppm by 2100. This is our best estimate of what will happen, and under this scenario the central estimate is a 2°C increase in global mean temperature.

Human health could be affected by heat stress mortality, especially in the major cities of the world. In 1996 in Chicago we had high temperatures for five days, and 600 people died. It tended to be the poorer people. In America, the effect was not due to the high daytime temperatures, it was the high night time temperatures. People could not acclimatise to high temperatures at night with very high relative humidity. In 1998, in Texas over 150 people died in about a two week period. I cannot say that this is due to global warming, but when you have sustained high temperatures one would expect an increase in heat stress mortality.

But much more important is the issue of vector borne diseases, such as malaria. There are roughly 300 million cases of malaria per year, of which 2 million result in death. This is basically due to poor health conditions and poor health care, primarily in Africa. But in the tropics, in a world with doubled CO₂ concentrations, with a sustained temperature increase of 4°C we would expect those numbers to go up by 25%, assuming we do not find a cure for malaria in the meantime.

The next serious issue we would have to consider is coastal areas. As mentioned above, we are projecting an increase of sea level of 15—95 cm. This would lead to a significant loss of land if we do not build sea wall protection. In Egypt one might see 1% land loss, in the Netherlands up to 6%. But Bangladesh is projecting a 17% loss of land, and some small island states could lose 80% or more of their land. Most of the Maldives, which comprises 1200 small islands in the Indian Ocean, would disappear with a 1 m sea level rise. There is no question that it would lead to the displacement of tens of millions of people. Bangladesh is one of the poorest areas in the world with a per capital income of under US$500 a year. It has such a complex deltaic system, such a poor infrastructure, such a poor financial situation, that there is no way it is going to be able to build adequate sea wall defences. The country would also lose something like 50% of its rice production, which is the major agricultural crop.

Eco-systems such as coral reefs and forests are the source of much food, fibre and medicine, but even more important are the things that do not trade in the market place. These include clean air, clean water, pollination services, soil generation and maintenance. Thus we underestimate the true value to society of most of these so-called ecological goods and services, which is why we are seeing rapid deforestation throughout most of the tropics. We as a society have not adequately valued those forests, which is way beyond the value of the timber. In fact, what the World Bank is trying to do is create markets to buy and sell ecological goods and services in various countries, both in the tropics and the sub tropics.

The whole of the east coast of the United States has the potential for forest at present. But according to our projections in 2100 the southern half of the east coast of the United States would have no potential for forest, it would only be grass land. Even in the forest area, trees would no longer be able to survive the temperature and precipitation regimes. With respect to agriculture, global production of food may not be affected by climate change. The negative effects of global warming could be offset by the carbon fertilisation effect of CO₂. The basic problem would be a major redistribution of where the food would grow, a significant decrease in the tropics and sub tropics, as much as a 30—40% in Africa and Latin America, offset by a significant increase in North America and in Siberia.

If you believe the world trade system works perfectly, we do not have to worry. Unfortunately we have hunger and famine today, which shows that the world trade system for food does not work very well. More unfortunately, the places where we have famine and hunger today are the places where we projecting a significant decrease in agricultural productivity in the future.

With water supply, we have a very hard time predicting with real accuracy changes in precipitation at the local level and changes in run off. But what we can say with some certainty is that arid and semi arid areas will become more arid. Today there are already 19 countries in the world that are defined as water stressed or water scarce. By 2025, independent of climate change, we project that number will double to about 38 countries. Climate change will exacerbate that problem in most of Africa, the Middle East and India, and even parts of the United States will not be immune to the problems.

On the east coast of America only a couple of per cent of the rain that falls is actually used. Thus we get vast amounts of rain on the east coast of America compared to what we consume. But in parts of Colorado they use 90% of the rain that falls, i.e. the consumption verses rainfall is 90%. If indeed there were to be a significant drying of the US Midwest, which is projected by some models, water stress problems in the western half of the United States would become quite severe.

What are the social and economic costs? We have a hard enough time as scientists projecting the changes accurately. When you try to put it into an economic framework it becomes even harder. We believe the damage costs are somewhere between US$5 and US$125 per tonne of carbon emitted, that is the social cost due to changes in health, agriculture, etc. We have not even really included the effects on the ecological systems, so it is a very, very significant potential change in economic output.

So those people who say that getting to grips with climate change, i.e. mitigating emissions, will cost money may be right. There is also a cost of inaction, there is a cost to human health, a cost to agriculture, and these translate into economic costs as well. So where do we need to go from here? Very simply, there are several greenhouse gases, of which CO₂ is the most important and methane is the second, which are most important in terms of emissions from human activities. Thus, we have to worry about deforestation, energy use, agriculture and industrial processes.

The IPCC came to a serious of conclusions on these. First, we believe there are technical options which, at little or no cost, can start to make in-roads into the climate change issue. Second, we realise that energy systems will turnover at least once or twice by the year 2100, and the challenge is changing our energy systems without the premature retirement of capital stock. It would cost a huge amount of money to suddenly change energy systems, the challenge is to integrate climate change and other environmental issues into our everyday energy systems and everyday development decisions. We should use the fact that there is going to be a turnover in capital stock and use that to our advantage. We should not have a strategy that results in premature retirement of capital stock.

We need to look at both supply and demand. Technology alone will not do it, we have clearly got to look at policy measures on the mitigation options, clearly we have to look at the supply side. The IPCC believes you need a very wide portfolio of options, including fuel switching from coal to gas, and increased efficiency of power plants. Many of the power plants in developing countries have 25—30% efficiency when it is possible to get to 45% now. We could also sequester the carbon dioxide, but it is relatively expensive today. Nuclear power is an option. I am not going to discuss that in front of a group of experts, you know better than I the public perception which is what you have to overcome on costs, safety issues, proliferation of nuclear materials. There are also renewable energies such as modern biomass, solar, wind and hydro.

So there needs to be a portfolio of options that are considered by governments and by industry. Our view is that no single option is likely to work, it would have to be a portfolio. On the demand side we would have to look at transportation, buildings and industry. There is also potential in looking at agricultural and forest systems. Clearly we have to look at the policy framework, we have to look at energy pricing, we have to remove subsidies. There are too many subsidies in transportation, energy systems and agriculture that promote the emission of greenhouse gases. We need to get rid of subsidies to level the playing field.

We should also, as set out in the Kyoto protocol, look at trade with emission permits, both project based emissions and emission rights trading. At the World Bank we believe that such a trading system will be the key to success for Kyoto. There are voluntary programmes, we need to look at education and other incentives to build up the market. But a key point to level the playing field, we would argue, is that you have to internalise the environmental externalities into the price of energy. If you internalise the externalities into the price of coal, then suddenly modern renewable energies and nuclear power become much more cost effective. It is bad economics not to internalise an externality when that is the true social cost, and that is of course a classical mantra of the World Bank.

There are major implementation issues, as I have mentioned above, and there are uncertainties in the science. However, some of the damage would be truly irreversible, and we have to consider that there are very long lag times between emissions and effects, amounting to decades, centuries or even millennia. It is a global problem, it cannot be solved in just the OECD alone, but there is an equity issue and there is a moral issue as to should the developed world take the lead, recognising that developing countries also have to take action later.

There would be winners and losers. I do not like that, but in some parts of the world agricultural production will increase, while in other parts it will decrease. The effects would be most severe in Africa and Latin America, small island states could be eliminated, there could be significant increases in vector borne diseases, again primarily in the tropics. That is what leads to the statement that developing countries are more vulnerable that developed.

There are real problems, such as what is the right stabilisation level, what is the right pathway to stabilisation, and what are the right choices of technology and policies. That is exactly what the governments were wrestling with in Kyoto, how to put the UNFCCC into operation and make the necessary choices. I have already described how large this challenge is. What could the global emissions of carbon dioxide be?

To give you a quick number for emissions which could stabilise CO₂ concentration at 550 ppm, which is what the European Union has argued for, you would never be able to go much above about 9 gigatonnes per year, compared to 7 gigatonnes today. In a more-or-less unconstrained world, CO₂ emissions would be about 20 gigatonnes per year by 2100. So to stabilise at 550 ppm one has to start to reduce global emissions almost immediately and within 100 years to have them lower than today's level. The challenge is how do we meet the energy needs of the world, alleviate poverty, and maintain lifestyles in North America and Europe, without causing a significant increase in greenhouse gas emissions.