Global Warming: A True Story for My Grandchildren William Nordhaus ...
Global Warming:<br />A True Story for My Grandchildren<br />William Nordhaus<br />Yale University<br />Draft for discussion purposes only<br />Copyright William Nordhaus<br />July 2010<br />Table of Contents<br />Preface<br />Chapter 1. The Science of Global Warming<br />Chapter 2. Impacts of Climate Change on Human and Natural Systems <br />Chapter 3. Slowing Climate Change: Science, Economics, Politics<br />Preface<br />If you read the newspaper or scan the daily blogs, you are virtually certain to find a story about global warming. Here is a sample from diverse sources: <br />“The last decade was the warmest on record.” <br />“Polar bears could disappear within a century.” <br />“Leaked emails are evidence of collusion among scientists.”<br />“Global warming claims a hoax.”<br />“The Greenland Ice Sheet has experienced record melting.”<br /> “More Americans think global warming is exaggerated.”<br />What is the non-specialist to make of all these conflicting stories? How can anyone sort through the contending voices without a lifetime of study? And how should concerns about global warming fit into other societal concerns like persistent unemployment, soaring public debt, low-intensity wars, religious conflicts, AIDS, and nuclear proliferation?<br />The purpose of this small book is to put global warming in a perspective so that the concerned citizens of the world can understand it. In these pages, I discuss the problem from the beginning, where warming originates, to the end, where societies can take steps to deal with the dangers of global warming. The book is written for non-specialists. It is for people who follow daily events, perhaps have studied a bit of science or economics, and are comfortable with numbers and logic. Most of all, it is for people who are interested. <br />What is my perspective? My professional training is as an economist working in a research and teaching university. I have taught and written in many areas of economics, particularly environmental economics and macroeconomics. I am co-author of a long-running textbook in introductory economics now in its 19th edition, and that has given me a special appreciation for young people who are struggling with new concepts. <br />I have also studied and written on global warming for more than three decades. When I first started worrying about the subject, it was a zero in economics. Absolute zero. Occasionally, along the way, it has been a joke among my colleagues, as if I was analyzing the macroeconomics of alien societies. But today economists and other policy analysts take global warming very seriously as a global threat.<br />Those who have read some of my earlier work sometimes ask me, “Didn’t you think that global warming was a minor issue? Have you changed your mind?” Here is what I would say: When I first started working on global warming, it was not a minor issue. It wasn’t an issue at all! I did not remotely expect that scientists and economists would today be devoting so much research to the question or that it would dominate scientific discussions.<br />But the evidence has accumulated, and the prospects of warming are grim. When asked if I have changed my mind, I am reminded of the answer that the great economist John Maynard Keynes gave to the same question in the middle of the Great Depression: “When the facts change, I change my mind. Pray, sir, what do you do?” If we never changed our minds, we would still be worshipping trolls and wandering around in bearskins.<br />I write this book particularly for young people, and I dedicate it to my grandchildren. They will inherit this world and are likely to live through the 21st century. The globe they will inhabit at century’s end will differ greatly from today’s world. The state of our planet will depend on the steps we take in the interim about many things, but those to slow global warming are perhaps the most momentous for the natural world.<br />The story of global warming is a fascinating one. It is as frightening as a Grimm’s fairy tale. But it is a true story, and for that reason even scarier. And more interesting. So to you who are just beginning the journey, I wish you, bon voyage.<br />New Haven<br />July 2010<br />The Science of Global Warming<br />The world is so full of a number of things,<br />I’m sure we should all be as happy as kings.<br />Robert Louis Stevenson, A Child's Garden of Verses<br />First Words to Our Grandchildren: A Tale of Two Lakes<br />As parents and grandparents, we have an awesome responsibility to those who will inherit this small, lovely, and often contentious planet. We feel this most tangibly when we cradle a tiny and helpless baby in our arms. When we think about how long is the road that baby will travel, how much the baby is dependent on parents and family and friends, on the education, security, and public goods provided by schools, teachers, and the institutions of our countries. The self-made man and the laisser-faire billionaire would not survive a month without the intricate support systems of human societies.<br />Although the earth looks huge and impervious to human activities, in fact life on earth is also a fragile and contingent system. We do not know whether the living systems that have evolved on earth are unique in the universe, but it seems a highly unlikely that the human and other life forms on Earth are found anywhere else. The drama that is life on earth will play only once. <br />I begin with the story of a small pond. In the summers, my family for four generations has enjoyed a small salt lagoon in southern Rhode Island called Quonochontaug Pond.  The picture on the cover of this book shows the night view of Quonnie under a full moon in the middle of summer. Quonnie has lived an exciting life. Fifteen thousand years ago, during the last glacial maximum, this area was buried under a mountain of ice. The pond was one of the string of coastal estuaries left as the glaciers retreated. It is home or touching point for piping plover, least terns, horseshoe crabs, and multicolored jellyfish. On the ocean side of the pond is a long barrier beach that has been free of houses since the 1938 hurricane tore through the area with winds of 110 miles an hour and demolished all the human structures other than a few posts and stone towers.<br />Quonnie is a vulnerable spot, subject to abuse from many quarters. Developers, motor vehicles, hurricanes and nor’easters, septic systems, landscapers, pesticides, oil tankers, and motor boats all beat upon the fragile system. Conservationists, ecologists, and coastal authorities beat back. In recent years, it has been a standoff between the forces of preservation and the forces of degradation. Unchecked global warming will change the ecosystems in unknown ways, and unchecked sea-level rise will eventually inundate the entire area.<br />What will Quonnie look like in a century? Will it look like the inlets of more southerly U.S. waters like those around Chesapeake Bay? Or will the combination of warming, increasing carbonization and acidification, and other human insults turn it into a dead salt marsh? I hope that we will take the steps to slow and reverse the warming and carbonization and protect Quonnie and other spots that are beloved or sacred to people who have visited them for generations.<br />If we look elsewhere, we can see how fragile lakes can be. One of the most disastrous example is the Aral Sea, in Central Asia. This was once the fourth largest lake in the world. But over the last four decades it has shrunk from 26,000 square miles to about one-tenth that size. (See Figure I-1.) What were the reasons? Nothing dramatic like global warming or war. The reason was primarily bad planning driven by bad incentives. It occurred primarily because the centrally planned “socialist” Soviet Union – not some runaway capitalist system -- decided to divert the rivers that feed the lake for irrigation of marginal lands.  Like a child starved of nutrition, the lake is slowly dying.<br /> <br />Figure STYLEREF 1 s I SEQ Figure * ARABIC s 1 1. Satellite photos shows the shrinkage of the Aral Sea. <br />____________________<br />This tale of two lakes tells the story of this book in the simplest way. We humans control the future of our planet. The earth has many enemies – unbridled political power, unchecked market forces, war, ignorance, wooden-headedness, and poverty. But, through a combination of careful planning, good institutions, and appropriate channeling of market forces, we can preserve the unique heritage around us. <br />This small book is about only one of the issues that we must address to preserve our world – global warming. Humans have been contributing to a warmer globe for centuries. But this 21st century is a critical period in which we must curb the unchecked growth in greenhouse gases, particularly those that come from fossil fuels. If we have not achieved a substantial reduction in these gases by 2100, then the environmental future of the earth is grim.<br />I write this book for my grandchildren because they will be living on this world in the late 21st century. Statistically, unless humans do some dreadfully stupid things, at least one of my grandchildren will be alive in 2100 to see what path human societies have taken over the coming century. Sometime in the 2030s, perhaps they will read this book and judge whether it served as an appropriate appraisal of the state of the art in 2010. Look online to see what remains of the Aral Sea. Judge whether your grandparents and parents have continued to preserve beautiful places like Quonochontaug Pond. And then, after you, my grandchildren, have learned the lessons of both history and science, it will be your job to continue to protect this wonderful planet.<br />A Summary of the Drama<br />Before beginning, I will summarize what follows. The story comes in three parts. Figure 1 is a schematic that shows the circular flow of climate change and a visual summary of the plot of this drama. Here, in the first part, we discuss the science of climate change. We examine the economic roots of global warming – how economic growth has led to rising carbon dioxide (CO2) emissions. This is the box at the upper left of Figure 1. We will see in this chapter that human activities are definitely leading to changes in atmospheric chemistry. This is the arrow from the upper left box to the upper right box, which is the climate system. The present chapter discusses these two boxes and the linkage between them. <br />The second act of our drama considers the impacts of climate change on human and natural systems. This is actually the most difficult task of all because of the evolving nature of human societies and technologies. So this second part will discuss primarily the two boxes on the right side and the link between them.<br /> The final act considers policies and politics: How can we bend the trajectory of global warming, and what are the most effective tools to do so? Even though this area is the most controversial and divisive, from a scientific point of view it turns out to be a straightforward issue. So this part discusses the lower left-hand box in Figure 1. <br />The dashed line indicates that climate-change policies close the loop. By taking policies to limit emissions, we can slow climate change. But that last little arrow is dashed to indicate that this is not an automatic link that will occur without countries taking strong steps. If we continue along with current path of virtually no policies, then the little dashed arrow will dissolve, and the globe will continue on the dangerous path of unrestrained global warming.<br />Figure I2. The circular flow of global warming<br />The discussion that follows can be seen as a circular flow. It starts at the upper left with the economic determinants of the emissions of greenhouse gases like CO2. These emissions lead to climate change, represented by the box at the upper right. As climate changes, there are impacts on human and natural system, represented by the box at the lower left. Human societies can take measures to reduce emissions as show in the box at the lower left. These would then close the link as shown by the dashed contingent arrow back to the starting box. The present chapter discusses the first two boxes and the linkages between them.<br />__________________________________________<br />Introduction to the Science of Global Warming<br />We are barraged with terrifying stories about the future prospects if global warming is unchecked. We hear about melting icecaps, extinct species, flooded subway systems, hordes of environmental refugees, dying oceans, and threats to national security. How secure are these projections? Are the projections based on the best science? Or is this all a vast left-wing conspiracy to deindustrialize the world?<br /> <br />A very short summary of climate science<br />Before we start on our journey, it will be useful to have a very short summary of the basic science of climate change. We will expand and refine this summary in the pages to come, but it will be important to look at the map before we start the journey.<br />The underlying premise of this book is that global warming is a serious, perhaps even a grave societal issue. The underlying scientific basis of global warming is well established. The core problem is that the burning of fossil (carbon) fuels such as coal, oil, and natural gas leads to emissions of carbon dioxide (CO2). <br />Gases like CO2, which are called greenhouse gases (GHGs), accumulate in the atmosphere and stay there for a long time. Higher concentrations of GHGs lead to surface warming of the land and oceans. The mechanism by which GHGs lead to warming can be understood as follows. The sun warms the earth with “hot” or short-wave radiation. Most of the radiation is sent back into space, but on its return voyage it is “warm” or long-wave radiation. <br />We are fortunate that most atmospheric gases (especially water, but also CO2) absorb more warm radiation than hot radiation. This process is like a blanket on a cold winter’s night, which keeps us warm. As a result, the earth is about 33 °C warmer than it would be without our normal blanket of greenhouse gases. However, we are adding more blankets to the atmosphere in additional CO2. We are thereby increasing the average temperature on the earth’s surface. Increasing the atmospheric composition of CO2 by what seems a tiny fraction (from about 0.28 parts per thousand to 0.56 parts per thousand) is projected to increase average temperature by around 3 °C. The reason for this huge impact is the CO2 intercepts warm radiation in a very powerful fashion.<br />While the exact future pace and extent of warming is highly uncertain, particularly beyond the next few decades, there can be little scientific doubt that the world has embarked on a major series of geophysical changes that are unprecedented for the last few thousand years. Scientists have detected the early symptoms of this syndrome clearly in several areas: The emissions and atmospheric concentrations of greenhouse gases are rising; there are signs of rapidly increasing average surface temperatures; and scientists have detected diagnostic signals –such as greater high-latitude warming – that are central predictions of this particular type of warming. Over the longer run, this produces profound and potentially dangerous changes in many earth systems and consequently to biological and human activities that are sensitive to the climate.<br />This in a few words is the syndrome that we will discuss in the pages that follow.<br />Why are carbon dioxide emissions rising?<br />Our journey begins, and will also end, with the daily activities of humans around the world. Because I am an American living in an urban environment, I will use that as an example, but it could equally well involve an Iowa farmer, a German automotive worker, a Chinese mechanic, an Indian farmer, or an Indonesian weaver.<br />The story starts with my decision to drive to a talk I am giving in upstate Connecticut. I have no realistic alternative, so I take my car up and back. There is much involved, but I will focus on the energy used. The trip is about 100 miles, and my car gets about 20 miles per gallon, so I consume 5 gallons of gasoline. In an ideal situation, with the gasoline burning cleanly, this will produce about 100 pounds of CO2, which will come out the tail pipe and go into the atmosphere. I can’t see, hear, or smell it – and indeed a few years ago I would not even have known about it. It seems unlikely to have much effect on the world, so I probably will ignore it.<br />But there are 6 billion people around the world making similar decisions many times, every day, every year, for decades and centuries. Suppose that everyone on earth does the equivalent of my drive twice a week (through heating or lighting or cooking or making steel or planting corn). Then this would add up to about 30 billion tons each year, which is what total CO2 emissions were in 2009. In a sense, virtually everything we do has some CO2 buried in the production process. You might think that riding your bicycle is “carbon-free.” But there is a little carbon in the bicycle, and quite a bit involved in making the road or sidewalk. <br />We can examine the trend in the carbon intensity of economic activity by looking at the CO2-GDP ratio for the United States. We have reasonably good data going back 100 years, and the intensity is shown in Figure 2.  This is quite a fascinating picture. I note one technical detail which will be used often in this book. The scale on the diagram is a “ratio” or “logarithmic” scale. This is a diagram in which equal vertical distances are equal proportions, so the vertical distance from 200 to 400 is the same as 400 to 800. This is extremely convenient because it means that a straight line (or one with a constant slope) has a constant rate of growth or decline. If you look at Figure 2, you see that the carbon intensity of the U.S. economy increased until around 1910 (this was the first age of coal). Since then, the CO2-GDP ratio has fallen at an average annual rate of 1.6 percent. There were wiggles along the way, but the long-term trend is clear. <br />The decline is sometimes called “decarbonization.” The reasons for the decarbonization are many, but it involves two main factors. One is that we use less energy per unit of output (say because our motors have become more efficient) and secondly because the most rapidly growing sectors (such as electronics and health care) use less energy per unit output than the average. <br />Figure STYLEREF 1 s I SEQ Figure * ARABIC s 1 3. Carbon intensity of U.S. economy, 1900-2008<br />―――――――――――――――――――――――――――――――――――――――<br />While the carbon intensity of production is declining, it is not declining fast enough to reduce total CO2 emissions, for either the world or for the U.S. If we take the period since World War II, real output has grown at 3.0 percent per year and the carbon intensity has declined at 1.6 percent per year, which means that CO2 emissions have grown at 1.4 percent per year. We do not have such data of similar quality for the world as a whole, but our estimates are that over the last half century, global output grew at an average rate of 3.6 percent per year, the rate of decarbonization was 1.1 percent per year, and CO2 emissions grew at 2.5 percent per year. The growth in CO2 emissions has actually been slightly above trend in recent years, primarily because of the rapid growth in developing countries like China. Figure 3 shows the long-term trend in total carbon emissions. <br />Figure I4. Global CO2 emissions, 1900-2006<br />―――――――――――――――――――――――――――――――――――――――<br />So in a nutshell, here is the problem: Countries around the world are growing rapidly (aside from some poor performers and putting aside recessions as painful but temporary setbacks). They use carbon-based resources like coal and oil to fuel their economies. The efficiency of energy use has improved over time, but the rate at which the efficiency is improving has been insufficient to bend down the emissions curve. <br />Are humans changing the atmosphere in significant ways?<br />We now move from economic activity to the geosciences. Is it true – indeed is it possible – that human activities are significant enough to change the global climate? After all, humans are but a tiny part of global activity. The answer here is unambiguous. Thanks to the foresight of a few scientists, we began monitoring atmospheric carbon dioxide in 1957 in Hawaii. Figure 4 shows a plot of monthly observations through the end of 2009. Over the half-century, atmospheric concentrations have risen by more than 20 percent.<br />Figure I5. Atmospheric concentrations of carbon dioxide measured in Hawaii, 1957-2009 <br />―――――――――――――――――――――――――――――――――――――――<br />How does this relate to human activity? We can compare the increase in atmospheric CO2 with estimates of total emissions of CO2 from industrial activity, shown in Figure 5. We see that somewhat more than half of emissions remain in the atmosphere. <br />Where is the rest? To answer this question, we need to make our first use of computerized models. (I defer the discussion of modeling until later and just use the results here.) This question is subject to vigorous scientific debate, but most models find that in the long run most of the non-atmospheric CO2 goes into the oceans, eventually diffusing into the deepest parts, but that process takes place very slowly.  Models of the carbon cycle developed by scientists estimate that around two-thirds of cumulative CO2 emissions over the next century will be in the atmosphere at the end of that period.  <br />Figure I6. The figure shows the fraction of each year’s CO2 emissions remaining in the atmosphere. On average, about 55 percent of industrial emissions since 1970 are still airborne<br />―――――――――――――――――――――――――――――――――――――――<br />The implication of the modeling results is that the residence time of CO2 in the atmosphere is very long. For this we can look at computer models as well as to statistical studies. A rough estimate is that if we emit 1 ton of CO2 today, about 40 percent of that will remain in the atmosphere after a century.  This has very important implications for how we think about climate change. The long residence time means that the effects of our activities today cast a very long shadow into the future. They do not just wash away in a few days or months like many other kinds of pollution. This long residence time will come back to haunt us when we consider the problem of discounting in our chapter on economics. <br />The Use of Models to Project Future Climate Change<br />In order to understand the scientific and policy issues, we need to pause to examine how scientists use their tools to understand future climate change. Up to now, we have been looking at historical trends. But climate change is about the future, not the past. We need to make projections – we hope accurate projections – about what climate change will be over the coming decades, perhaps even for centuries. <br />A projection is a conditional or “If…, then…” statement. In other words, it states, “If a given set of input events occur, and we use model X, then we calculate that the following output events will occur.” In the case at hand, the input events are things like a path of CO2 emissions and the earth’s geography. The model might be one developed by scientists at the Goddard Institute for Space Studies (GISS) in the U.S. And the output events might be time paths for regional daily temperature and precipitation, sea level rise, and sea ice.<br />We cannot do those calculations in our heads (at least, I cannot), so all these calculations are done with computerized models – generally very large and complicated computerized models. What is a model? You can think of it as akin to an architect’s model of a building. Figure 6 shows a model of the iconic house designed by American architect Frank Lloyd Wright along with a picture of the actual building. <br />Figure I7. A model of “Falling Water” and a picture of the actual house designed by Frank Lloyd Wright<br />A good model should capture the essence of the question at hand without overwhelming the user with unnecessary clutter. In economics, we build models of output and incomes, for example, to help the government forecast its revenues and spending and provide an informed basis on what is happening to, say, the government debt. In the area of climate change, we build models to estimate future emissions of CO2, the impact on atmospheric CO2 as well as the pace and extent and even regional dimensions of changes in climate. Other scientists estimate the impact on agricultural output, on sea level, on the extent of malarial mosquitoes, and on snowpack for water runoff. <br />In addition to very detailed models of specific areas (such as atmospheric chemistry, ocean chemistry, agricultural response, economic growth, and the like), there are also “integrated assessment models” or IAMs. IAMs link together the different components of the global warming syndrome using several stripped-down modules. I will often in this exposition rely on IAMs that we have developed at Yale known as the DICE/RICE family of models. These models combine in one large computerized package the end-to-end process from economic growth through emissions and climate change to impacts on the economy and finally include certain policies to slow climate change. The advantage of the IAM approach is that it can consider the entire process; the disadvantage is that it has to simplify drastically some of the processes that are captured in greater detail in the more complete models. Figure 7 uses the architectural analogy of an IAM. When well done, they are as elegant as a building sketch by Frank Gehry. <br />Figure STYLEREF 1 s I SEQ Figure * ARABIC s 1 8. Integrated Assessment Models are small-scale representations and are similar to a Frank Gehry sketch, shown left, which was an early model for the Walt Disney Concert Hall, right<br />_________________________<br />Two parts of a climate-change projection: emissions paths and climate modeling<br />With this background, we can now explain how scientists project future climate change. It is necessarily a two-step procedure. The first step is to estimate the inputs into the models, and the second step is to construct models and use them to project future climate.<br />Emissions projections: techniques<br />The first step is to develop a set of projections for the inputs into the models. These are primarily paths of emissions of CO2 and other greenhouse gases (GHGs). While CO2 is the most important of the gases, others can make a significant contribution. I will focus primarily on CO2 to keep the discussion manageable, but the complete treatment will also include other gases. However, when looking at actual projections, I use “CO2 equivalent,” or CO2-e, which adds together the contributions of all the GHGs.<br />Most climate models have relied on a standardized set of projections known as the SRES emissions scenarios, which were ones prepared in a Special Report on Emissions Scenarios by the IPCC. These are essentially “stories” rather than projections of future output and emissions paths. One has rapid population growth; another has rapid technological change; one is eco-friendly. They are interesting scenarios rather than attempting to be the most accurate projections. <br />Why would scientists want to use “interesting” rather than accurate projections? According to the SRES report, the reason is this: “However, many physical and social systems are poorly understood, and information on the relevant variables is so incomplete that they can be appreciated only through intuition and are best communicated by images and stories.”  This is definitely not the way that economists would recommend constructing long-term projections. We have no idea of whether these are good guesses or wild fantasies. We have no way of judging whether they span a reasonable range of possible future outcomes. <br />So, here is a first warning about projections about standard projections of future climate change: They are based on emissions scenarios that are not based on standard statistical techniques that are used in the social sciences. They do serve one very useful function: They allow climate scientists to focus on a range of standardized emissions scenarios to test and compare their climate and other natural-science models. But these standardized test runs should not be confused with best-practice projections or forecasts.<br />Statisticians and econometricians have studied techniques for making projections for many years. The modern approach is to use a combination of demographic, economic, and technological data; then estimate relationships using the historical data and technological or scientific constraints. From these, we can obtain a statistically based projection of future trends. This is the approach we use in constructing our RICE/DICE models. It can be reproduced and easily updated, but the main advantage is transparency. Unlike the SRES scenarios, we can understand the construction of statistically based projections.<br />In the discussion of emissions, I will rely on economic models rather than the SRES scenarios. Several modelers under the aegis of the Energy Modeling Forum (EMF) have compared their results on the “baseline” or unconstrained emissions of CO2. This result is extremely important because it gives us an idea of what the world will look like if we take minimal action to control global warming. Basically, this projection is the result of trends of decarbonization like those shown above along with economic growth.<br />Using the statistical approach also allows us to judge the uncertainties associated with the models. This is done using a technique called Monte Carlo simulation. This is like spinning a roulette wheel 10,000 times to see how often different numbers or colors come up. Figure 5 shows our estimates from the 2009 model along with four of the standard SRES scenarios. It is important to note that these are uncontrolled emissions runs. That is, they show what would occur if no policies are taken to slow global warming.<br />Emissions projections: results<br />We begin with the results of an important comparison study of different Integrated Assessment Models undertaken in the EMF-22 project. This study included modeling teams from around the world: 6 groups from Asia and Australia; 8 from Western Europe; and 5 from North America. A subset of these provide results for a no-policy run, for which Figure 8 shows projections of emissions for CO2 through 2100.  The 11 lines without markers are from the EMF study, while the line with the circles is the projection of the RICE-2010 model.  Two important points emerge from this figure. First, all models project a continued growth in CO2 emissions. The range of growth rates is between 0.5% and 1.7% per year over the 2000-2100 period. Even though these seem to be small growth rates, they imply that emissions in 2100 will be 1.6 to 5.4 times higher than 2000. The problem is not going to disappear or be solved by standard market forces.<br />The second feature is that future emissions are highly uncertain. Because of inertia of economic and technological systems, the near-term results are relatively certain. But the range of estimates grows over time as the uncertainties about population, technology, and particularly energy systems compound in the coming years.<br />We can also look at the uncertainty studies, or Monte Carlo estimates of the kind discussed above. Using the RICE-2010 model, we estimate that there is a 10 percent chance that the emissions will be greater than 116, and a 10 percent chance that they will be less than 46 billion tons of CO2 per year. So the estimated 10-90 percentile range from the RICE-2010 model is somewhat smaller than the range of model uncertainties shown in Figure 8. <br />Although I don’t think the SRES scenarios are useful for scientific analysis, we can compare the results just discussed with the standard scenarios. It turns out that two of the scenarios (the A2 and the B1) are outside the 10-90 percentile range, and indeed are outside the range of any of the models of the EMF study (A1B and B2) are closer to the middle of the projections, although they appear to grow too rapidly in the early decades and stagnate in the later decades. <br />Figure I9. Alternative projections for CO2 emissions<br />The lines show 11 models surveyed by the EMF-22 project. The RICE-2010 with the circles is a slightly later projection from the Yale RICE-2010 model.<br />―――――――――――――――――――――――――――――――――――――――<br />Return a moment to the uncertainties about future CO2 emissions. Where does this come from? A careful analysis finds that most of the uncertainty lies in our uncertainty about future economic growth. Will the world continue to robust economic growth of the period from 1950 to 2005? Or will it stagnate with slow technological change, recurrent financial crises and perhaps depressions, spreading pandemics, and occasional widespread wars? These are essentially unknowable. For this reason, the uncertainty about future emissions paths shown in Figure 8 is unlikely to be significantly narrowed in the next few years.<br />Climate models<br />The second part of the task of projecting future climate change is to take the emissions paths shown above along with other important input data and put these into climate models. Climate models, or what are technically known as atmospheric-oceanic general circulation models (AOGCMs), are essentially translations of physics and geography into computerized computational form. <br />So, to understand the AOGCMs, we need to understand the basic atmospheric science underlying the equations. We sketched the basics of climate science above, and will extend it slightly to describe climate models. Recall that hot or short-wave solar radiation comes through the atmosphere and warms the surface. The surface returns an equal amount of heat to the atmosphere. Part of the return is long-wave (warm) radiation. The atmosphere is a natural “greenhouse” in which gases such as water and CO2 absorb some of the outgoing radiation. If there were no greenhouse gases, the earth’s surface temperature would be −19 °C (cold like the moon), whereas the actual temperature is 14 °C. A useful but imperfect analog is that greenhouse gases are like a blanket in winter, allowing someone under the blanket to retain the heat and stay warm.<br />Scientists calculated, as we see, the natural greenhouse effect warms the earth by about 33 °C relative to what would happen with no atmosphere. The enhanced greenhouse effect is what happens when even more greenhouse gases, such as CO2 are added. The current quantity of greenhouse gases absorb some but not all of the outgoing long-wave radiation. As more and more gases are added, an increasing fraction is absorbed, and the new equilibrium comes at a higher temperature. Some saturation occurs, however, so adding doubling CO2 might raise the temperature of 3 °C, but adding the same quantity again would add only 1.8 °C. So there is a kind of diminishing returns to the enhanced greenhouse effect. <br />Because of the complexities of these processes, the best current models have quite different projections of climate change. Figure 9 shows the model recent model comparison prepared for the IPCC. Each model ran an identical experiment: They compared a run with no CO2 increase with a run that doubled atmospheric CO2 over 70 years and then held CO2 at that doubled level. The models calculated a “transient response,” which is the temperature increase at the time of doubling; and an “equilibrium response,” which is the long-run temperature increase when all adjustments have taken place.  Our RICE model calculations suggest that CO2-e (the CO2 equivalent of all greenhouse gases) will double around 2050. So, the blue lines in Figure 6 show what the estimated temperature response would be around 2050 according to the best-guess emissions paths. The average of the models is 1.8 °C, which compares with the actual increase of around 0.75 °C for the last 100 years.<br />Figure I10. Temperature response of 18 models in IPCC Fourth Assessment Report<br />The 18 models were run with an experiment in which CO2 concentrations doubled over a 70 year period. The red bars show the transient response, which is the global mean temperature increase at the end of the 70 years; the average of the models for this experiment was 1.8 °C. The blue bars show the response in the long run (usually 200 to 300 years); the average temperature increase for the equilibrium was 3.2 °C.  <br />―――――――――――――――――――――――――――――――――――――――<br />Figure 9 also shows the equilibrium temperature-sensitivity coefficient. The average long-run impact of the models is 3.2 °C, or almost twice as much as the transient response. The transition to the equilibrium proceeds very slowly over 2 or 3 centuries.  This large difference reflects the great inertia in the climate system due to the fact that the oceans warm very slowly. This slow response is another part of the difficulty of dealing with climate change. Like smoking cigarettes, it may take a long time to see the effects. If there is any happy note in this, it is that if the concentrations of CO2 are reversed relatively quickly, then temperature will also come down because the oceans have not yet warmed.<br />Many non-scientists look at the divergence among models and wonder why these uncertainties cannot be resolved. One is reminded of the joke, “If you ask five economists you will get six answers.” This is not even a joke with climate models because the same research group gets different estimates in different models as the models get refined.<br />The reason for the difference can be explained using an example from economics. When the Obama administration proposed its stimulus package, it proposed about $250 billion of government purchases. Economists desired to estimate the impact of this on the economy, and to do so they use a multiplier analysis. There are both amplifying and damping second-round effects of government spending. The powerful amplifying effects (or positive feedbacks) occur because each dollar of spending generates additional income, and consumers tend to spend some fraction of the additional income. However, there may be forces that dampen the response (negative feedbacks). For example, if output rises, then interest rates may rise, reducing stock and bonds prices and reducing investment. While there are uncertainties here, a standard estimate is that each $1 of purchases generates $1½ of additional GDP, indicating that the multiplier is around 1½.<br />A similar set of forces operate on the climate. Climatologists have estimated that if there were no feedback effects, the global warming from doubling CO2 would be 1.2 °C. But there are very strong multipliers at work in climate change. The estimated climate multiplier ranges leads to an amplified total impact of 1.8 to 4.4 °C for the models shown in Figure 9. <br />Why is the uncertainty be so large? There are some factors that amplify the simplest effect and others that diminish it. For example, if a warmer earth melts snow and ice, this means the earth reflects less sunlight and this amplifies the greenhouse effect through what is called the “albedo effect.” The most important amplification comes because higher temperatures lead to increased water vapor in the atmosphere, and water is a powerful greenhouse gas. The biggest problem for modelers has been clouds. Clouds pose difficulties because they both cool and warm – they cool when they reflect sunlight and they warm because they trap warmth. It turns out that modeling clouds formation is extremely difficult, and this produces a substantial amount of the difference among models.<br />The climate models are extremely detailed and produce a fantastic array of results, which can be assessed and used for studies of impacts. One of the most important set of results is the impact on temperature by region. Figure 10 takes four well-known models from the array and shown their estimated change in temperature by latitude at the end of the 21st century. These projections use a standardized scenario that is reasonably close to our economic projections (SRES scenario 1AB). <br />A few features stand out. First is the feature that the temperature increases are much larger at the poles, particularly the Arctic region. This is due largely to the melting of polar ice in summer. Second, the differences across models are particularly striking in this comparison. The models have only modest disagreements in tropical regions but it turns out that modeling high latitudes (particularly the behavior of ice and snow) is extremely difficult. The differences in model projections are a long-standing feature of climate models and have not disappeared with improved models, better resolution, and faster computers.<br />However, even with the disagreement among models, we should not lose sight of the central feature, which is that all major modeling groups (in both the full group of Figure 6 and the regional detail of Figure 7) show major climate change over the 21st century. These are the cutting edge of modern climate science, and the basic message cannot be obscured by the discrepancies.<br />Figure I11. Estimated temperature change averaged by latitude for four models, 2080- 2099<br />―――――――――――――――――――――――――――――――――――――――<br />In addition, to put a human face on these changes, we show the distribution of human populations by latitude as the solid blue line in Figure 10. It is not generally appreciated that almost half of human populations live between latitude 25 N (approximately Miami and Hong Kong) and latitude 42 N (Boston and Rome). As can be seen in Figure 10, there is likely to be substantial climate change in this region, particularly in the northern end of this range. There is much more that can be learned from climate models, particularly regarding impacts, but that will wait until the next chapter.<br />Temperature projections from integrated models for uncontrolled paths<br />Next, we can put the different components together to project climate change over the coming decades. For these estimates, we calculate the path with no climate-change policies. In other words, we assume that no policies are taken – such as limiting emissions or taxing carbon fuels – to slow the growth in CO2 and other greenhouse gas emissions. This is probably not a realistic assumption, and in my view definitely not a undesirable one, but it gives us a kind of “worst case policy assumption,” that is, a case where countries just sit on their hands and let the dice roll.<br />Recall that the standard projections from climate models come from arbitrary emissions trajectories rather than from models based on mainstream economic, demographic, and technological analysis. We cannot, therefore, look to standard projections from the IPCC to get an accurate picture. <br />Instead, the approach here will be to look at the Integrated Assessment Models – models that combine climate and economic models to construct what might be called a “combined best estimate” of climate change over the coming years. Even this is a difficult task because models use different assumptions. So I will present a slightly simplified approach. I will take the CO2 concentrations from different models shown in Figure 8. I combine these with estimates of non-CO2 greenhouse gases from the GISS model. And I will run them through the Yale RICE-2010 climate model. These RICE model has a climate module that assumes a standard temperature sensitivity coefficient of 3.2 °C per CO2 doubling. The runs apply the RICE climate module to the CO2 concentrations of the different models.<br />Figure 11 shows the results of these estimates.  This picture provides a good overview of different future climate changes as seen by multiple modeling groups around the world. The average change in temperature in 2100 is projected to be 3.5 °C above the 1900 average. The average of the models is very close to the projection of the RICE-2010 model, shown as the heavier line with the circle. <br />The spread among the models is discouragingly large, however, ranging from 2.8 °C to 4.3 °C in 2100. We must emphasize that the spread among the models is generated only by the differences in CO2 concentrations of the models. These in turn are driven by uncertainties in global and regional output growth, decarbonization, and the carbon cycle. Alas, the true uncertainty is even larger. To get the complete picture, we would need to add differences in other greenhouse-gas emissions, alternative land-use patterns, and alternative climate models. <br /> <br />Figure I12. Projected global mean temperature increase from 10 integrated assessment models<br />―――――――――――――――――――――――――――――――――――――――<br />I noted above that the standard projections of climate change from the IPCC and many climate models are based on unrealistic and poorly constructed economic models. How do these projections compare with the economically based models shown in Figure 10? Most of the standard scenarios track the economic models until the middle of the 21st century. They begin to turn down relative to the economic projections after 2050, and most seriously underestimate temperature trends for 2100 and beyond. For example, the average of the models shown in Figure 11 projects temperature increase relative to 1900 of 3.5 °C, where middle scenario B1 projects 1.8 °C. While projections beyond that are subject to even higher uncertainties, the RICE model projection for 2200 is 4.1 °C, while the scenario B1 projection is 2.1 °C. <br />The main reason why the economic projections show much higher temperature trajectories appears to be because they are much more “optimistic” about economic growth. As we noted at the beginning of this chapter, if there are no climate-change policies, climate change is a race between economic growth and technological decarbonization. Most economic models project continuing growth in the decades, and that leads to continued strong growth in CO2 emissions in the absence of policies to curb emissions.<br />Are we likely to encounter an unusually rapid climate change?<br />You might wonder whether we are making a mountain out of a bump in the road. Climate change is part of earth’s history, from the warm periods of the dinosaurs to the cold periods when New England lay under a mile of ice. Is this time really different?<br />The answer is yes and no. It is true that climate changes of similar magnitudes have occurred, and some of them appear to have occurred extremely rapidly. During a period known as the Younger Dryas about 12,000 years ago, the earth seems to have gone into one-third of an ice age in a few decades. Similar periods of abrupt climate change appears to have occurred in earlier periods, although the reasons are not well understood.<br />But there is one major concern about the pace of uncontrolled climate change that seems likely over the next century and beyond. Climatologists have concluded that no climate changes of the speed and scope we are witnessing have occurred through the course of human civilizations (say, the last 5000 years). While we do not have reliable records akin to the instrumental records compiled in Figure 13, there are proxy records that can be gathered from sources such as ice cores, tree rings, pollen of plants, and bore holes in the ground. The best guess is that the rate of global climate change we face over the next century will be about ten times as rapid as any change experienced during the last five millennia. So, while not unprecedented on the scale of geological time, it is unprecedented during the era of human settlements.<br />Further Downstream Impacts<br />We have emphasized the effects of human activities on temperature. At first blush, a change of 2 or 3 °C was not that alarming. After all, we experience that much change from 9 a.m. to 10 a.m. on the average day. Moreover, the climate changes envisioned are not large relative to the changes that individuals and groups have undergone through human migrations. People have moved from Moscow to Texas, from southern Italy to New England. Many people today move from snowbelt to sunbelt to enjoy the warmer lifestyle with an average temperature (in Phoenix) that is 12 °C warmer than where they left (say, Boston).<br />However, this comparison ignores the real problems raised by climate change. The problems are not a simple rise in mean temperature but the accompanying physical, biological, and economic impacts – and particularly the question of thresholds and non-linear responses. This is easily illustrated by the example of driving on wet roads. Consider what happens when the road surface goes from 1 degree above freezing to 1 degree below freezing. You go from slippery to deadly conditions.<br />In this final section, I will discuss the issue of critical thresholds in the earth systems that may be imperiled by the various aspects of climate change. This territory is much less well understood: We leave the world of relatively (and I emphasize that word) well understood implications of our emissions of CO2 and other greenhouse gases and turn to much more complex and imperfectly understood implications. <br />Tipping points and safe operating spaces<br />Scientists believe Planet Earth has experienced an unusually stable climate for almost 10,000 years (see Figure 16). This is the period during which human settlements, written languages, cities, and human civilizations as we know them emerged. <br />However, the combination of larger populations, economic expansion, and new technologies are changing the earth’s climate, ecosystems, land use, and water flows in ever-larger ways. Scientists have asked whether we are changing the earth system outside the biophysical environment within which human civilizations have developed and thrived.  <br />Figure I13. Estimated temperature over last 150,000 years from ice core proxy. Note the stability of the temperature during the last 10,000 years <br />===============================================<br />One set of concerns is that further climate change will trigger “tipping points” in the earth system. A tipping point comes when a system sharp discontinuity in behavior. We are familiar with tipping points and sharp discontinuities from our daily lives. Some familiar tipping points are seen when a tree is uprooted when the stress is too large or when a levy is ruptured when water is too high, as occurred with Hurricane Katrina. Financial systems display tipping points when people lose confidence in banks and cause a “run on the bank,” which can lead to severe financial crises like those of 2007-2008. Sometimes, there is a “good equilibrium” (banks healthy) and a “bad equilibrium” (banks failing).<br />Figure 17 illustrates a tipping point using with ball in a double-bottomed bowl. The height of the bowl represents the health of the economy or ecosystem. We start in a good equilibrium in (a). Then stresses push on the right side of the bowl. At first, the ball moves only a little. Then, once the tipping point is reached, the ball races to the bottom of the second curve in (c). Here, we have multiple locally stable equilibria. So once the ball is in the second curve, in (d), even though the stresses are removed, the ball is stuck in the bad equilibrium. <br />Figure I14. Illustration of how stresses can change system slowly until tipping point is reached, after which there are rapid and potentially catastrophic changes.<br />____________________________________________<br />Many earth scientists have warned that we are approaching or have passed important tipping points in the earth’s systems. One study, shown in Table 1, illustrates some important potential tipping points, time scales, and impacts.  This shows 14 potential tipping points of global or continental scale that were identified by a working group of scientists. I have sorted these be the time scale on which they operate. In addition, I have added a column indicating what I believe to be the level of concern, from least to most serious. (I risk being misunderstood here. Least concern is still great concern, but we need to set some kind of priorities in focusing our policies.) <br />We can consider two major dimensions here. First, is how soon we will cross the threshold, which is related to both the warming trigger and the degree of inertia in the system. Second, is the level of concern. The areas of most urgency are the three-star issues that we may cross in the next hundred years. There are two in this category: the disappearance of a substantial part of the Amazon rain forest and the reversal of the Atlantic thermohaline circulation. I believe that we should add ocean carbonization to this list (a topic discussed in the next chapter), although that has just appeared on the radar screen of scientists in the last decade.<br />One interesting feature of this table is that it appears that there are no two-star or three-star tipping points with time horizon less than 300 years until climate change reaches 3 °C. At 3 °C, we encounter the bottom end of the tipping range for several systems: Sahara/Sahel and West Amazon monsoon, Amazon rain forest, Boreal forest, Atlantic thermohaline circulation, El Nino–Southern Oscillation, West Antarctic ice sheet, and we have passed the bottom end of the range for the Greenland Ice Sheet. We have not yet passed the 3 °C threshold. <br />The research on tipping points and safe boundaries for operating human societies is in its infancy. We have already found new potential tipping elements since the paper discussed in Table 1 was published, while others have receded in the timing or level of concern. The following chapters will discuss how to stay within these and other suggested limits. But the work on thresholds is a sober reminder of the potential discontinuities that can occur in complex systems.<br /> <br />Table 1. Tipping Points in the Earth System<br />___________________________________________________<br />Should we take climate models and projections seriously?<br />Unless you are a specialist in the geosciences, you may well find the discussion complicated and even impenetrable. So before moving to the final section, I step back and consider the overall reliability of the climate models, data, and projections. Is this just some conspiracy of propaganda? Or is it real?<br />How can we deal with the possibility of a hoax? In my younger days, I studied the philosophy of Bishop Berkeley, who was a radical “idealist.” His idealism is not that of the protesters battling the police at a meeting of the World Trade Organization. Rather it was the idea that we can never really be sure that something is real because there is no absolute reference point from which to judge reality. <br />Some people claim similarly that the entire pageant of climate change science is a hoax, a grand conspiracy, a kind of scientific Protocol of the Elders of Science, designed to extract money from a gullible and frightened public. At some deep level, perhaps we are just an audience in a theatre where the gods play some elaborate movie. But that would apply to everything around us, not just climate science. So we have to judge climate science by the same standards that we use in judging the reality of drug safety, measuring gross domestic product, breaking a leg, and getting hit by a truck.<br />So in this discussion, I address three aspects of global warming science that have come under attack. I begin with a discussion of bloginess. I then examine the question of the reliability of the temperature data that are used to measure long-term trends. I conclude with a review of the question of whether historical trends provide empirical support for the climate models. <br />A measure of bloginess<br />One informal way to judge the seriousness of an issue is to see where it gets aired. Serious scientific questions are discussed in scholarly journals, while issues that arise from political groups or talk radio drift into the web from blogs and similar media. Discussions that arise as if from nowhere have a high degree of what I will call “bloginess.”<br />Here is a simple test: When I searched “climate change hoax” on Google, I got 312,000 results. When I searched it on Google Scholar (which requires some element of publication in a recognized source ), I landed only 4 results. Let’s call the ratio of overall web citations to Scholar web citations a measure of bloginess. For climate change hoax, the bloginess index is about 80,000. In fact, climate change hoax has a bloginess index that is close to that of “Lady Gaga.” For a serious term, “double CO2,” the ratio was 39, For a technical term, “radiative forcing,” the bloginess index was 8. Figure 13 shows the bloginess of the different terms. So, looking at its level of bloginess suggests that climate change hoax is mainly a creature of the media and blogs. <br />Figure STYLEREF 1 s I SEQ Figure * ARABIC s 1 15. Bloginess for different climate change terms<br />Bloginess indicates the relative importance of a term on the web relative to its citations in the scholarly literature.<br />―――――――――――――――――――――――――――――――――――――――<br />The reliability of the global temperature measures<br />Most people have seen graphs that show the rising trend of global temperatures over the last century, and perhaps also reconstructions to earlier centuries. These were recently questioned in the “Climategate” affair. This occurred when the web organization known as “Wikileaks” released thousands of emails of climate researchers. Critics claimed that climate scientists were cooking the books and falsifying the historical record. (The present author was not an author of any leaked materials.) What should the outsider think of this debate? <br />The issues involved here are very familiar to economists. Most economic magnitudes are “aggregates” of underlying source data. If you have ever read about the unemployment rate, the gross domestic product, or the consumer price index – each of these is aggregated from a variety of data from surveys or administrative data. Estimating the global mean temperature raises many of the same issues as does estimating these economic variables. How do we aggregate over the subunits? How reliable are the data? What do we do about missing observations?<br />My experience across a broad variety of fields tells me that the major problems with scientific data are the questions just raised and rather than fraudulent creation of data on a grand scale. Whether it is Watergate or Madoffgate, or less well known scientific frauds, there are sufficient checks and balances in an open society to root out fraud. In economics, the major fraudulent data were (and perhaps are) produced by authoritarian governments, for example the economic data from the former Soviet Union. There are two symptoms of data fraud. The first is that researchers do not make their underlying data and methods freely available. If I had to point to one major set of economic data that is suspicious on this count, it is the Chinese national output data, which are completely opaque to outsiders. The other symptom is unexplained inconsistencies between different data sources.<br />In climate science, perhaps the best check against fraud and error is the presence of multiple research teams competing to produce the best model or data. One of the reasons that the climate modeling efforts have credibility is that there are so many groups racing to produce the best model. Similarly, there are multiple research teams producing global climate data. Figure 13 shows, for example, the results of three estimates of global mean temperature over the last century-plus. It is visually obvious that the three reconstructions move quite closely. <br />Figure I16. Global temperature reconstructions from three centers<br />――――――――――――――――――――――――――――――<br />“Let me do it myself”<br />The distinguished physicist Richard Feynman once said “What I cannot create, I do not understand.” Children understand this when they say to parents who want to cut their food or steady their bicycle, “Let me do it myself.” Children and Feynman are saying that the only way to fully master something is to do it or reproduce it yourself.<br />How can we possibly reproduce any of the data or models that are produced by armies of climate scientists? This is a particularly daunting task here because of the complexity of the constructs and the sheet size of the data sets. Nevertheless, like the child trying to learn to ride a bicycle, I decided to see if I could reproduce the complicated indexes of global average land temperatures. <br />With the help of my colleague Xi Chen, I took annual data on average temperature for 23,019 grid cells around the world from the U.S. National Climate Data Center (NCDC). I used data only for the period 1980 – 2008 because of the sheer size of the calculations. These data exclude most of the high latitude regions. A further question is whether the data underlying the reconstruction can be verified. We undertook a spot check of a few stations around the world (in the United States, Ghana, Morocco) and found that the gridded historical data matched the stations we looked at. Finally, we used our data on average land area by grid cell from the Yale G-Econ project to calculate a series on global mean temperature. <br />Figure 14 shows our calculations and the published estimates from the NCDC and the Hadley Center in Britain. These move very closely together, although the Yale reconstruction has a slightly higher time trend than the other two series. <br />The conclusion on the historical time series is that they pass the usual tests of scientific scrutiny. As with all empirical estimates, they are subject to a variety of errors in estimation and construction. But they can be used for further analysis subject to the appropriate caveats about all similar aggregate indexes, whether from economics or from other areas.<br />Figure I17. Land surface temperature from NCDC, Hadley, and Yale reconstruction<br />――――――――――――――――――――――――――――――<br />Are the climate models consistent with recent observational data?<br />Another related question concerns the relationship between the increase in global temperature and human activities. This has been one of the most contentious issues debated in the IPCC, and critics of global warming have consistently questioned whether the warming is due to rising CO2 and other greenhouse gases or instead to natural changes such as due to the sun, random variation, and the like. Look at Figure 13, which shows rising temperatures particularly since around 1980. Can we separate human intervention from background noise?<br />A useful way to examine this question is to ask what a temperature path would look like if the climate models are correct and we have the observed trends in CO2 and other greenhouse gases. For this question, we use estimates of both CO2 only and the CO2 equivalent of all greenhouse gases and other influences such as volcanoes. We take these estimates and put these into a simplified climate model and then compare the results with the actual trend in global temperatures. We normalize all the series so that they are equal to 0 in 1900. <br />Figure I18. Actual and model temperature projections<br />――――――――――――――――――――――――――――――<br />The results are shown in Figure 15. The wiggly line shows the actual temperature trend. The smooth line at the top shows what the models would predict with the influence of CO2 only, while the bottom almost-smooth line shows the estimates with all estimated forcings. The results here are similar to the more complete experiments of scientists reported in the scholarly literature. One striking feature is that the projection with all gases is far below the projection with CO2 only. This fact comes primarily because of the estimated cooling effects of aerosols, which come from a wide variety of sources such as burning fossil fuels and biomass.<br />Three (?) features emerge from the graph. The first is that the temperature trend is very noisy. This arises from the complex dynamics of the earth-atmosphere-ocean system. As an economist, I recognize this feature of complex systems, for we see a similar volatility in stock markets, foreign-exchange markets, and in output and incomes. Neither economists nor climatologists can fully explain these erratic short-term fluctuations, but they do complicate interpretation of the long-term trends. <br />Second, it is clear that global temperatures are rising, and this upward trend is highly significant. There are several ways to test this. One way is to perform a regression analysis. This is a technique for estimating the best fit of a line, but it also allows us to estimate the reliability of the estimate. If we estimate the time trend using regression analysis, we find that the coefficient is positive and highly statistically significant.  This point is intuitively obvious from Figure 9, but it will also survive statistical scrutiny. There is simply no doubt that the indexes are rising.<br />Third, it is clear that global temperatures are correlated with the predictions from the calibrated climate model using all estimated influences (the bottom smooth line in Figure 15). The rise in mean temperature from 1950-59 to 2000-09 was 0.60 °C for the actual series and 0.45 °C for the prediction from the climate model. We can also test the association with a regression analysis. If we use the predicted temperature as an independent variable and actual temperature as the dependent variable, we find a highly significant coefficient. Indeed, the regression suggests that the model is slightly underpredicting climate change. <br />We can use a statistical analysis to determine the temperature change is just a time trend plus noise. For this purpose, we do another regression analysis. This adds a time trend to the basic equation from the last paragraph. The time trend The analysis suggests that CO2 is statistically significant, and indeed the time trend disappears when the greenhouse gases are included. Again, examination of Figure 2 shows this intuitively. The temperature trend moves around erratically in the first part of the period and then moves up steadily in the last half century. <br />Can we therefore conclude that humans are causing global warming? Statisticians know that association does not prove causation. Attributing causes is especially difficult when we cannot do controlled experiments. But scientists who look at the weight of evidence in many different areas along with graphs like that in Figure 15 conclude that it is causal. The Fourth Assessment Report of the IPCC concluded, “Most of the observed increase in global average temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations.” <br />Critics continue to attack these and similar conclusions using reasons that are sometimes serious and sometimes ridiculous. One argument is that scientists are not really 100 percent sure that global warming will occur. That is true. But a good scientist is not 100 percent sure of any empirical phenomenon. This was explained by Richard Feynman in a way that is humorous but very deep:<br />Some years ago I had a conversation with a layman about flying saucers — because I am scientific I know all about flying saucers! I said “I don't think there are flying saucers.” So my antagonist said, “Is it impossible that there are flying saucers? Can you prove that it's impossible?” <br />“No”, I said, “I can't prove it's impossible. It's just very unlikely”. At that he said, “You are very unscientific. If you can't prove it impossible then how can you say that it's unlikely?” But that is the way that is scientific. It is scientific only to say what is more likely and what less likely, and not to be proving all the time the possible and impossible. To define what I mean, I might have said to him, “Listen, I mean that from my knowledge of the world that I see around me, I think that it is much more likely that the reports of flying saucers are the results of the known irrational characteristics of terrestrial intelligence than of the unknown rational efforts of extra-terrestrial intelligence.” <br />This lovely story is a reminder about how good science proceeds – both in the natural sciences and in the social sciences like economics: A cool head at the service of a warm heart.<br />Endnotes: These notes are for the curious or specialists who would like to know where the statements in the text are from or their statistical support. <br />