11.2: Quantifying the Economic Impacts of Climate Change

History of climate damages

The first evidence of reflection on an impact of climate on human/economic activity goes back to Pythagoras’ disciple Parmenides, who divided the world into five zones: one torrid, two temperate, and two frigid. The torrid zones (which we call the tropics today) he thought were too hot to be inhabited. Aristotle later agreed with this view, for both the torrid and the frigid zones (the poles). He believed that the only areas on Earth habitable by humans were located between the tropics and the Arctic and Antarctic Circles—the area where he lived. The French philosopher Montesquieu took a much more direct and controversial line on the causal relationship between climate and human ability, suggesting that humans from colder climates were physically superior, braver, more honest, and more clever. This clearly incorrect perspective on the cross-sectional influence of climate on human well-being lacked any sort of empirical basis and led to a long discussion around environmental determinism.

Economic damages from climate change— why is this so difficult?

The emergence of climate change as a field of study in the physical sciences in the late 1970s quickly led social scientists to think about what the possible consequences of a changing climate could be on economic sectors. This is a difficult problem, to say the least. Let’s think about what one would like to know in order to make good policy. One would like to get an estimate of the damage a ton of CO₂ (or another greenhouse gas) causes after it has been emitted. Sounds simple, right? But this is where the heavy hand of physics presents the invisible hand of the marketplace with quite a challenge, which has to do with time and space.

First, CO₂, for example, is a long-lived gas. Once emitted, it stays in the atmosphere for hundreds and possibly thousands of years. Hence it continues to produce warming for a long time. That tank of gas you burned through getting home for the holidays was turned into CO₂ molecules that your great-great-great-[…]-great-great grandchildren will feel the consequences of. This means in order to figure out what the damage of that ton emitted is, you need to figure out the consequences for the economy today, tomorrow, and the next few hundred years. This means you will have to put a dollar amount on the damages experienced by people (and critters and plants) living at the end of the century and beyond. We are again back to the problem of having to project the future state of the climate system and economy out hundreds of years.

The second challenge stems from the fact that the vast majority of greenhouse gases are global pollutants, as we discussed above. This means that the exact location of emissions is irrelevant to the damages they cause. Furthermore, one has to calculate the damages across the entire planet—not just at the point of emissions.

To summarize, so far the challenge seems pretty steep. You have to calculate global damages for the next few hundred years. But it gets harder when you contemplate the broad array of economic sectors that can be affected by climate change. The most obvious sector that will be affected by climate change is agriculture. Crops and animals largely live outside and exposed to the weather. If it gets hotter and drier, most plants and animals do not do as well. It has been shown, for example, that crop yields drop significantly if the number of days with temperatures above 30°C rises, as most climate models predict. Another sector affected by climate change is the energy sector. When it gets hotter, people who have air conditioners turn them on and increase their energy consumption—often significantly. They also heat less in the winter, which is a good thing.

But you can probably already see it. Quantifying what will happen to crops across the world as well as energy use in developed and developing countries everywhere over the next few centuries is a daunting task. If you take into account that there are many other aspects of human society that are affected by the climate, this task becomes even trickier. It has been shown that mortality, morbidity, crime, conflict, productivity, water consumption, migration, spread of disease vectors, air pollution, happiness, cognitive performance, reproductive ability, and suicide are all affected by climate—worldwide. The studies showing this are not just telling stories but are using actually observed data to establish statistical links between weather/climate and these outcomes.

The damage function

As mentioned above, the literature examining this function linking climate to economic outcomes has exploded over the past decade. So let’s take a look and examine what these studies do in practice. The next section draws heavily on an article I published in the Journal of Economic Perspectives in 2018 (listed in the Supplementary Readings section; it is a great, free resource written for a general audience).

Here is how economists have recently attempted to quantify the link between changes in climate and the consequent damages (and in a few cases the benefits). The mathematical relationship used to map changes in climate into damages is something fittingly called a damage function. This deserves some elaboration using an example. When you leave your apartment in the morning, you encounter the day’s weather. If you live in sunny San Diego (like Professor Ramanathan, who is the brains and soul behind getting this book over the finish line), the weather you encounter is likely sunny and a pleasant 22°C. If you live in Northern Bavaria as I used to do, in the winter you will encounter a day close to −10°C with thick clouds. You can think of climate as the average weather over a long period of time. There are many different measures of climate that may be relevant to you. For example, you may be interested in what the average weather (climate) in a location is like in the summer. This is essentially how we pick vacation destinations! We look at average temperatures during the season we intend to visit a location. There is no guarantee that the weather when we actually travel will be what we anticipated, though. Often you will have traveled to the beach, expecting sunshine, and you just got unlucky and encountered rain and fog. This is weather.

So, if there is climate change, the average weather you will encounter will shift. For those of you who remember your statistics training, it is not just the average weather that shifts, but the whole distribution (which in some places can be approximated by what is commonly referred to as a bell curve). For example, the average temperature may increase, but at the same time, the variability of weather may also increase. A shift in the average weather and possibly in its variability will lead to a higher frequency of “extreme events” such as heat waves and droughts. The issue is that these changes in weather—driven by climate change—affect outcomes of interest in unexpected ways. What I mean is that a 1°C increase in temperatures when it’s cool outside could have a much smaller impact on, for example, crop yields and energy consumption than a 1°C increase when it’s hot outside. Some important food crops, for example, have been shown to react very negatively to temperature increases above 30°C but do not care much if it gets a little bit warmer at 20°C.

Hence, what we are interested in is how individuals/crops/animals/ plants respond to weather when the average weather (that is, climate) has changed in the long run. This response is likely different from the old response when the average weather had not changed. Let’s use an example close to my heart to help us clarify our thinking. Historically, a really hot day in the San Francisco Bay Area would lead to increases in ice cream consumption and lots of whining. Since San Franciscans historically knew that these hot days were extremely rare, almost no houses or apartment buildings had air conditioners. If, however, San Franciscans learn that there is such a thing as climate change, and that their summers will resemble Palermo’s unpleasant hot summers on average in the future, many will go ahead and install air conditioners.

Hence how the San Francisco hipsters react to a hot day after the climate has changed is different from how they would have reacted to the same hot day before climate change—because of the installation of “new gadgets,” that is, air conditioners. The future under climate change will hence likely result in higher electricity consumption due to the installation of additional air conditioners, which consumers will pay money to install. Complaining will likely increase as well, since we Californians are a whiny bunch. But what about the people that already had air conditioners? The rational thing for them to do is to run their air conditioners more frequently. So what we end up with is air conditioners new and old being run more frequently using electricity, which costs consumers money and results in higher emissions of greenhouse gases.

While qualitatively the above example makes a great deal of sense (at least to me), we need to quantify the impacts—meaning putting actual numbers to the problem of how much electricity consumption will change. This is done by statistically estimating the damage functions discussed above. The trick is that these damage functions need to be calculated for all (or at least for the most important) sectors sensitive to weather under climate change. These damage functions are key to making smart policy decisions and allow us to identify the sectors most vulnerable to a changing climate. So what do these damage functions used in policy analysis of the economic impacts of climate change look like—at least in our perfect ivory tower world? Figure 11.2.1 helps to fix ideas.

Figure 11.2.1 Mapping weather into impacts—the importance of accounting for adaptation. From Auffhammer 2018.

The top left panel of Figure 11.2.1 shows weather generated in a setting before climate change has occurred (light gray line) and one where the climate has changed (dark gray line). Here, we are only looking at changes in temperature. The post-climate-change temperature is warmer but also more variable. The top right panel of Figure 11.2.1 displays two damage functions (the smooth curves) that map weather into an outcome, in this case temperature into household electricity consumption (measured in kWh). The damage functions, as has been confirmed in many empirical settings, are highly nonlinear—they are not straight lines. When it is cold and temperatures rise, electricity consumption falls, as people heat less. When it is warm and temperature rises, electricity consumption increases, as people air-condition their homes.

Back to our San Francisco example, this response without any adaptation (the solid line) is relatively shallow, as few people have air conditioners. When the climate changes to become like that of Palermo, we assume that people eventually learn about this and will adapt. In this example they will do so by buying and using air conditioners, which changes the damage function to the dotted line. The response, especially at higher temperatures, is now much steeper, resulting in stronger post-adaptation increases in electricity consumption on a 1°C warmer day when it’s warm outside.

The impact of this can be seen in the bottom panel of Figure 11.2.1 very clearly. The solid light-gray series of electricity consumption shows consumption under the pre-climate-change weather with the no-adaptation response function. If the climate changes and we use the flatter (and wrong) pre-climate damage function, projected electricity consumption is the gray solid line. This is clearly incorrect, as one is using the right weather but the wrong response function. The correct response function is the dotted parabola, which results in the dark dotted time series of electricity consumption in the bottom panel. It is much higher and much more variable than the no-adaptation prediction. One way to think about this is by simulating weather impacts “with and without a climate adaptation response.” Allowing for adaptation, in this example, leads to substantial changes in electricity consumption, which the damage function seeks to incorporate. This seems straightforward in the case of electricity consumption, since we know the likely adaptation technology and can observe how people use it in hotter areas. This is much more difficult for other sectors. Trying to estimate how crops will adapt, conflict will change, and species and disease vectors will adapt to climate change is very difficult. By that I mean you should contemplate working on these topics in your research!

Table 11.2.1 Coverage of the damage function literature, showing what we do and do not know about damage functions for different sectors of the economy and beyond

Sector	Plausibly Causal Estimates	Adaptation Addressed	Global Coverage
Agriculture	Yes	Yes	Yes
Forestry	No	No	No
Species loss	No	No	No
Sea-level rise	Yes	Yes	No
Energy	Yes	Yes	No
Human amenity	Yes	Maybe	No
Morbidity and mortality	Yes	Yes	Yes
Migration	Yes	No	No
Crime and conflict	Yes	No	Maybe
Productivity	Yes	No	No
Water consumption	No	No	No
Pollution	Yes	Maybe	No
Storms	Yes	Yes	No

Source: Auffhammer, M. 2018. Quantifying economic damages from climate change. Journal of Economic Perspectives 32(4), 33–52. Table 1. https://doi.org/10.1257/jep.32.4.33.

Environmental economists have been preoccupied with developing statistical methods to estimate such damage functions from observed data on outcomes of interest. There are a number of great review papers, which discuss methods and results in greater detail and are included in the Supplementary Readings section. However, it is instructive to summarize where we are in terms of our understanding of damage functions and what is missing. Table 11.2.1 shows an overview of sectors and what we know about economic damages from climate change.

So much work to do!

What is obvious from Table 11.2.1 is that we know next to nothing about a number of important sectors. First off, the literature putting a value on the damages to so-called non-market goods is small to nonexistent. Non-market goods are things that improve our well-being but that are not typically traded in markets, such as biodiversity and clean air. An example helps to organize thoughts. If climate change wipes out a species, say the bald eagle, is there an economic loss/damage? The answer is clearly yes. But what is the value of a bald eagle? Eagles are not traded in markets (unlike chickens—think Chicken McNuggets!). But just because you cannot buy an eagle in a store does not mean it does not have value.

Economists have developed a number of methods to value non-market commodities. One is to simply ask people how much they would be willing to pay in order to ensure that bald eagles are preserved—this method is called contingent valuation. You can then take these numbers, add them up across all people, and come up with a valuation. Often these numbers end up being unrealistically big. If you asked me what I would be willing to pay to preserve the bald eagle, I would probably say $200. If you asked me to pay up, I may have conveniently forgotten my wallet. Economists have developed methods to account for these issues, and contingent valuation has been used to assess damages from oil spills, for example. There are also other methods to value non-market commodities, such as the travel cost method that looks at how much people spend to go see a national park for example. You can in certain settings use this number as an approximation of the value people place on said national park. So yes, we need much more research on the value of biodiversity and non-market commodities, including exotic things like the nitrogen cycle.

The other thing that we know very little about is the damages incurred by extreme events. What is the damage caused by the West Antarctic Ice Sheet melting? What is the damage caused by the thermohaline conveyer belt, which is the ocean circulation that gives Europe its gentle climate, shutting down? These are events that we have not experienced in human history, so it is hard to determine what the damage from such an event would be. This is where we turn to “experts.” We would call up world experts in ice sheet dynamics and sea level rise and ask how much sea level rise would be caused by the melting of the West Antarctic Ice Sheet. We would then talk to people who understand urban economies and come up with estimates of what it would cost to either protect or move certain coastal communities. Here we would have to rely on the assumption that experts actually know what they are talking about. In order to improve the quality of these expert assessments, we tend to not ask a single expert, but dozens or hundreds of them and average across their answers. That said, the economic damage from extreme events is hard to estimate, and much work needs to be done by credible academics to push the envelope of our understanding.

The social cost of carbon

Figure 11.2.2 The social cost of carbon—what is needed to produce a number in one model? A lot. From Rose, Diaz, and Blanford 2017.

Let’s assume for a moment that you have done an amazing job (high fives!) and obtained a set of credible damage functions that satisfy the criteria set out above. What do you do with them? What you would want to do is calculate a number called the social cost of carbon (SCC). The social cost of carbon is maybe the most important number you have never heard of. The social cost of carbon is an estimate of the present value of the stream of global damages from one additional ton of CO₂ emitted at a point in time. In short, it represents the damage your ton of CO₂ will do to all sectors everywhere over its lifetime.

In order to calculate this number, the literature has employed what are called integrated assessment models, which integrate simple models of the economic and climate system, as illustrated in Figure 11.2.2. These models start with assumptions (sometimes referred to as socioeconomic scenarios) about the evolution of global, and in some cases regional, income and population over the next 300 (!) years. The models then translate economic activity into emissions of greenhouse gases, most notably CO₂, but in some cases other GHGs such as methane. (Methane is a short-lived but more potent greenhouse gas than CO₂. It comes from natural gas wells, fermentation processes, and the back end of cattle.) These 300-year time paths of emissions are then fed into a very simple model of the global climate system, which translates emissions into surface temperature, precipitation, and sea level rise. These outputs are then fed to your amazing damage functions, which map the emissions path into economic damages. For example, a hotter state of Georgia due to climate change will likely use more electricity to cool the indoor environment. This is considered an economic damage. In order to calculate the effect that higher emissions have on outcomes of interest across many sectors of the economy, the integrated assessment model is run with and without one additional ton of CO₂. The time path of the difference in damages relative to the baseline represents the damages from that one ton for each year over the next 300 years. The stream of damages is then converted into a present value. This dollar amount is called the social cost of carbon and is measured in US dollars.

Box 11.2.1  What Is Discounting?

Discounting translates the value of future consumption into current-day dollars. You may value the consumption of a basket of goods valued at $10 that you get today more highly than the consumption of the same basket valued at $10 some 20 years from now. If you place a higher value on today’s consumption than future consumption, and we want to figure out the value of your consumption stream over a long time period, we need to translate the value of that stream of future consumption into current-day value. This is called discounting. The discount rate is your personal “interest rate,” and it reflects the relative value you place on current versus future consumption. The higher the discount rate, the less value you place on future consumption. This concept is central in evaluating the benefits and costs from doing something versus doing nothing about climate change, since the costs of doing something are largely incurred in the near term, but the benefits (avoided damages) come much later in time.

Some integrated assessment models are global and treat the world as a single region (for example, DICE by 2018 Nobel Laureate William Nordhaus), while others break out the world into very large regions(for example, PAGE by Chris Hope; FUND by David Anthoff and Richard Tol). In the case of models with regional resolution, damages are then aggregated across regions to calculate the global social cost of carbon. This number represents the damages caused globally over time by one additional ton of CO₂ emissions at a single point in time.

FIGURE 11.2.3 Sample of social cost of carbon (SCC) estimates used in federal rule making for three administrations. Estimates are in 2007 dollars for emissions of a ton of CO₂ in 2010. DOE-Department of Energy; EPA-Environmental Protection Agency; IWG-Interagency Working Group; NHTSA-National Highway Traffic Safety Administration. The black diamond indicates the "central estimate," if one was identified. The gray bars indicate selected upper and lower bounds used in regulatory analyses. From Auffhammer 2018.

While there is not one official integrated assessment model that rules them all, the US federal government has attempted to estimate the social cost of carbon going back to the George W. Bush administration. Figure 11.2.3 shows a set of values used by the three last administrations in federal rule making. For comparability, the graphic shows values for 1 ton of CO₂ emitted in the year 2010 valued in 2007 US dollars.

In the early years of the Obama administration, the Interagency Working Group* embarked on an effort to calculate an official social cost of carbon. The approach adopted, which is described in detail in Greenstone, Kopits, and Wolverton (2013), was to essentially embark on the effort described by Figure 11.2.2: feed three integrated assessment models with a set of harmonized assumptions regarding the evolution of the economy and population, account for uncertainty, and provide a statistical distribution of the social cost of carbon across models. The most frequently cited number for the SCC was $42 per ton emitted in 2020 as measured in 2007. There were several updates to the social cost of carbon calculation, and the final available estimates are given in Table 11.2.2.

Table 11.2.2 Social cost of carbon estimates by the Interagency Working Group

Year	Discount Rate and Statistic
	5% Average	3% Average	2.5% Average
2015	$11	$36	$56
2020	$12	$42	$62
2025	$14	$46	$68
2030	$16	$50	$73
2035	$18	$55	$78
2040	$21	$60	$84
2045	$23	$64	$89
2050	$26	$69	$95

Table 11.2.2 displays the global SCC estimates using three different discount rates for emissions between 2015 out until the year 2050. Two things stand out from this table. First, columns 2–4 display the average SCC across simulations using three different discount rates. A higher discount rate (5%) puts a lower value on future damages and hence results in a lower SCC. A lower discount rate places a relatively higher value on future damages and hence results in a higher SCC.

Second, one notices that for any chosen discount rate, the SCC is higher the later emissions are made. For example, 1 ton of CO₂ emitted in 2020 using the 3% discount rate results in a $42 per ton SCC. A ton emitted in 2050, using the same discount rate, has an SCC of $69. This increase occurs for two reasons. First, as time goes on, the stock of CO₂ in the atmosphere is higher, as CO₂ accumulates over time. Hence, each additional ton emitted at a later point in time arrives in an atmosphere with a higher stock of CO₂ in it, adding additional warming into a more stressed system and leading to higher damages. Second, for some of the integrated assessment models used, damages are a function of income (for example, GDP). As the world grows richer over time, later emissions arrive in a wealthier world, resulting in higher damages. An easy way to think about this is, for example, higher incomes result in more valuable infrastructure, which may be negatively affected by changes in climate.

There is much work to do in order to properly quantify the damages from climate change, and the economic literature on the social cost of carbon is a good literature to follow. One specific effort, which is pushing the frontier of this literature, is the Climate Impact Lab. It is driven by a collaboration of the University of California, Berkeley; the University of Chicago; Rutgers University; and the Rhodium Group. They have an extensive website documenting their research at www.impactlab.org.

*The IWG was composed of members from the president’s Council of Economic Advisers, Council on Environmental Quality, Department of Agriculture, Department of Commerce, Department of Energy, Department of the Interior, Department of Transportation, Department of the Treasury, Environmental Protection Agency, National Economic Council, Office of Management and Budget, and the Office of Science and Technology Policy. It was disbanded by President Trump.