I thought I knew the answer, but now it’s not so clear.
U.S. households and businesses use a whopping 11.5 quadrillion BTUs of energy annually for heating and cooling, about one-third of all residential and commercial energy use. How will this be impacted by global warming?
When it comes to electricity, the answer is fairly unambiguous. A growing body of evidence shows that global warming will increase U.S. electricity consumption. My colleague Max Auffhammer, together with Energy Institute alumni Patrick Baylis and Catherine Hausman, for example, have shown that global warming will have large impacts on U.S. summer peak load electricity consumption.
But, of course, global warming also means less energy used for heating. In the United States, most people live above 30° latitude or even above 40° latitude, so heating is at least as important as cooling. In fact, Americans currently use twice as much energy for heating as they do for cooling.
Still, if you had asked me a month ago, I would have said that that the net effect on U.S. energy consumption was probably an increase. I informally polled my colleagues about this, and most had the same view. Now that I’ve looked more closely at this, however, I’m not at all sure.
Sixty Years of Weather
The figure below plots population-weighted average U.S. annual cooling degree days (CDDs), a widely used measure of cooling demand calculated as the sum of daily mean temperatures above 65°F. Since the 1950s, the number of CDDs experienced on average in the United States has increased 30%.
But while CDDs have been going up, heating degree days (HDDs) have been going down. The figure below plots population-weighted average U.S. annual HDDs, calculated as the sum of daily mean temperatures below 65°F. HDDs have fallen steadily since the 1950s, from more than 5000 annually to about 4000 today.
The change in HDDs is smaller in percentage terms, but larger in absolute terms. Consequently, the sum of annual CDDs and HDDs has been steadily decreasing, down 12% since 1950. So, on average, U.S. households and businesses are experiencing less total days with extreme weather.
This Trend Will Likely Continue
According to a 2015 paper by Yana Petri and Ken Caldeira, about two-thirds of the U.S. will experience a decrease in CDDs+HDDs by end-of-century.
Impacts of Global Warming on Residential Heating and Cooling Degree-Days in the United States” (2015) Scientific Reports. This image is licensed under a Creative Commons Attribution 4.0 International License.
The geographic pattern makes sense. Southern cities like Houston, Atlanta, and Miami, which already experience more CDDs than HDDs, will see a net increase. While Northern cities like Seattle, Chicago, and New York, will see a net decrease, with decreases in HDDs outpacing increases in CDDs.
What Does This Mean for Energy?
Exactly what this means for net U.S. energy consumption is not clear. Americans will use less energy for heating, and more energy for cooling, but the changes need not be proportional. This is true whether we are talking about energy consumption (in BTUs), or energy expenditures (in $$), or, even carbon dioxide emissions (in tons).
How many BTUs of energy are used per HDD or CDD depends on how much people heat and cool. Over time people are heating and cooling more for a given CDD or HDD. Homes and businesses are larger so more total space is heated and cooled, and there is less tolerance for discomfort. How many BTUs are used also depends on the type of technology used for heating and cooling (e.g., natural gas furnace, electric heater, heat pump, air conditioner, etc.), as well as the energy efficiency of these technologies. More energy-efficient air conditioners, for example, mean that fewer BTUs are used per CDD. Predicting how these technologies are going to change over time is very difficult.
The piece of this puzzle which has received the most attention is electricity. The Third National Climate Assessment (NCA) provides a review of the existing literature, concluding that global warming will increase net U.S. electricity consumption, and, in particular, peak summer loads. This is consistent with several recent papers by Max Auffhammer and co-authors (here, here, and here).
A couple of papers have looked at a broader definition of energy. For example, a paper by economists Olivier Deschenes and Michael Greenstone published in 2011 forecasts that total U.S. residential energy consumption will increase 11% by end-of-century due to climate change. While their paper examines the residential sector only, their measure of energy is comprehensive and includes electricity, natural gas, and other energy sources.
Like most of this literature, the Deschenes and Greenstone study holds the stock of technologies fixed, and thus does not capture long-run adaptations like people switching from natural gas furnaces to electric heating. An important exception is this paper by economists Erin Mansur, Robert Mendelsohn, and Wendy Morrison. They model technology choices in the U.S. residential and commercial sectors, and forecast that energy expenditures will increase by 10% by end-of-century. This is a forecast about expenditures not BTUs so is hard to directly compare to other studies.
Other studies predict small decreases, or ambiguous impacts on total U.S. energy consumption. I think the reason these studies have reached such mixed results, is that, at least for the next several decades, the increase in cooling demand is likely to be similar in size to the decrease in heating demand. As a result, small differences in modeling assumptions can flip the sign for the net impact. The different studies have used different climate models, for example, with differing predicted impacts for HDDs and CDDs. There is huge uncertainty in estimates like these, particularly for forecasts many decades into the future. This is particularly true for carbon dioxide emissions. Alternative heating and cooling technologies can have very different carbon intensities, so changes in energy use need not result in proportional carbon dioxide impacts.
Where People Live
Another takeaway I have from these studies is that where people live matters. The annual average CDDs and HDDs above are “population-weighted”, meaning that for each year this is a weighted average, where the weights are given by populations in that year. The figure below shows U.S. population density in 2010.
Source: U.S. population density based on 2010 Census. Image licensed under creative commons.
More people today live in warm parts of the U.S. than they did in 1950, so this pushes up the population-weighted CDDs and pushes down the population-weighted HDDs. The DOE has looked at this and concluded that about one-third of the change over time in CDDs and HDDs is due to population changes.
Currently the two fastest growing states are Texas and Florida. The more people who live in Southern states like these, the more air conditioning matters and the more likely that the net impact will be to increase U.S. energy consumption. It will be interesting to see how U.S. migration patterns change in the future. Warmer weather makes Northern states relatively more attractive, which could lead to Northern migration.
Global warming is already impacting energy demand. Even if the net change for U.S. energy consumption is ambiguous, the regional impacts are unambiguous. Florida, for example, is clearly poised for continued increases in energy consumption, while states in New England, for example, will continue to see decreases. How all this adds up nationwide is a deceptively hard problem.
But keep in mind that this is just the United States. For much of the rest of the world, global warming will unambiguously increase energy consumption. About 3 billion people live in the tropics, between -23° and +23° latitude, where heating is basically non-existent. Sales of air conditioners are booming around the world, especially in places like India, which is poised to increase electricity consumption dramatically over the next several decades.
Moreover, energy consumption is just one sector. The IPCC outlines dozens of sectors that are being impacted by climate change: agriculture, human health, ocean acidification, hurricane intensification, species loss, etc. Understanding and quantifying the impacts in all these sectors continues to be an important priority for natural and social scientists.
Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.
Suggested citation: Davis, Lucas. “Will Global Warming Increase or Decrease U.S. Energy Consumption?” Energy Institute Blog, UC Berkeley, October 3, 2017,