Green New Deal: How much does fixing climate change cost the US?
A real-world argument with numbers you can check for yourself.
Swirling around the Green New Deal debate there is a cacophony of people who’d like to have us believe preventing climate change will cost too much. I’ve seen some astronomical numbers in the media like $100 Trillion, which are largely unsupported and are no doubt related to how ill-defined the GND is so far. This is partly an artifact of the Green New Deal including components that address important issues like employment and healthcare, which are not related to climate change, and should be solved in separate legislation. So what, then, is a reasonable and defensible cost estimate for a GND that focuses specifically on solving climate change and decarbonizing the economy?
There are some wonkish academic studies that pad around this issue, but I hope here to build a real-world, easy-enough-to-understand-on-the-second-reading kind of argument that shows it’s achievable with numbers you can check for yourself.
What no one who is trying to lampoon efforts at a Green New Deal will admit, is that once you have spent [the associated infrastructure costs], we’ll have much lower, very predictable, costs of energy for the 20–30 years all of those new machines will last.
Here, as usual, I’ll focus on the energy side of climate change. That’s about 80% of the problem. The other 20% is agriculture, industry, waste and land use — dominated by the methane emissions of the meat sector, cement and steel production, and landfill emissions. So as not to let perfection be the enemy of good we are only going to look at the energy side of the equation. We’ll address this tougher 20% in future pieces.
We’ll find that the price tag for complete decarbonization is reasonable, and doesn’t rely on a miracle cure. It is doable starting now, and most likely will not cost us money but in fact, save us money. We’ll see it is a capital cost challenge in the short term with a long-term payback as we won’t have to supply expensive fossil fuels to the system in the future.
How do we decarbonize?
The short version of how we decarbonize is through massive electrification — of nearly all transportation and of nearly all heat for buildings and industry. That electricity will come from wind, solar, hydroelectric, and nuclear. The details of the balance of those technologies will be determined by largely local factors that include climate (local annual temperatures), local renewable resources (solar, wind, biomass, hydro), politics (of nuclear for example), and even cultural issues like preferred architecture and urban design preferences.
If we can agree on the assertion above, then we can let the market decide the balance of the solution and avoid a needless and counterproductive debate about which there will be more of. This strategy allows for a miracle in carbon sequestration or fusion or something even more incredible to emerge, but we aren’t dangerously betting on it. Right now the sensible money is on very large amounts of solar and wind, both of which have had a precipitous cost drop since 2000. Nuclear’s price hasn’t fallen and is notoriously difficult to calculate because it’s unclear how the security expenses associated with the fuels and wastes fits in an all-accounts ledger. Even so, it’s a safe bet that we’ll do more nuclear than we do today, it will become cheaper, and we’ll grow more accustomed to it as we more responsibly deal with the waste.
By adopting such a hybrid nuclear/renewables program, the huge amount of wasted energy in finding, processing, transporting, and producing electricity or powering mobility with fossil fuels is eliminated, allowing for a seemingly miraculous lowering of the primary US energy use by more than half. However, such a reduction is no miracle and is quite easily understood once you understand the inherent efficiency advantage of electrical machinery driven by non-carbon sources. You can read a high-level explanation in Green New Deal: The enormous opportunity in shooting for the moon.
Massive electrification is a strategy that is deeply connected to both the supply side and the demand side of the energy question and highlights the interconnection of the two. The supply side is where the energy comes from: traditionally fuels for vehicles, fuels for electricity, fuels for industry, and fuels for buildings, both commercial and residential. The demand side is how we use the energy, in driving, air conditioning, making steel, heating water, and buildings. When we change the demand side to be made up of more efficient electrical machines (e.g. electric vehicles replacing internal combustion engines, electric heat pumps replacing furnaces) we get an efficiency win on the demand side, but we create a need for more electricity on the supply side. We have to meet the new supply side with carbon-free technologies.
So what does it cost?
2.1 Transportation (demand side):
There are 263M vehicles in the US. The only difference in cost between electric and gasoline vehicles is the cost of the battery. Otherwise, they have four wheels, a motor (electric or internal combustion), seats and a steering wheel, and all the little luxuries and cup-holders. Let’s assume they all will be electric. They will use about 300Wh/mile, and they will all have a 250-mile range. They will each need a (300x250) Wh battery. That’s 75,000 Wh or 75 kWh. Today’s batteries are roughly $200/kWh. By 2025 it is widely believed that they will cost out at $100–150/kWh. At $125kWh that pencils as (75x125) = $9,375. Let's call it $10,000 per vehicle. Multiply by 263 million vehicles, and it pencils out at $2.6 Trillion dollars.
Incidentally, the total energy storage capacity of those vehicles will be around 20 TWh¹. That means our cars and trucks will be able to store 10 hours of all of our total energy supply in their collective battery. Consequently, we should start understanding the role of our vehicles as part of our energy infrastructure. Using a fraction of their capacity will be enough by itself to offset a lot of the “variability” of wind and solar. At its heart, this is what a “smart” grid will let us do. It doesn’t have to be that smart; actually, it just needs to be sensible².
(1) 263 (million vehicles) X 300 (Wh/mile) X 250 (miles range) X $125/kWh = $2.6 Trillion.
2.2 Heating and Cooling (demand side):
Cooling is all done today using refrigerators (and their close cousins, air conditioners) running on electricity. Space and water heating is nearly all done with natural gas. We won’t capture the distributed emissions from natural gas in homes, so sequestration is off the table. We need to electrify that heat. Heat pumps (also cousins of refrigerators) can routinely do this now, and lower the primary energy required to heat a house by 3X. Heat pumps are the cooling mechanism in refrigerators, but they can also create heat (they actually work by creating heat and cooling simultaneously). There are 128,000,000 homes in the US with an average footprint of 2,012 sq.ft. Converting to radiant hydronic heat (considered by all experts to be not only the most efficient but the most comfortable and least allergenic) costs around $8/sq.ft including the cost of installing the heat pumps. That’s $16,000 per home or 2 Trillion dollars for the entire US housing stock. Incidentally, you can store a lot of energy in hot water (and in cold). If we added two days’ worth of thermal storage to each home (something the size of 2 large water heaters, or a hot tub), we would create another grid-scale battery of approximately 6 TWh (50kWh*120,000,000 homes). Like our cars, this is an enormous load that we can shift and use as a battery to balance a renewables-heavy grid. This thermal storage will be cheaper than batteries at an estimated 1–2c/kWh and is rolled into the $8/sq.ft. estimated above.
(2) 128 (million homes) X 2012 (average square footage) X $8 (per sq.ft. for hydronic heat pump heating plus storage) ~ $2 Trillion.
There are 5.6 million commercial buildings that cover 87bN square feet. There are advantages at scale of heating and cooling these buildings relative to residential buildings, and not all commercial buildings need to be heated or cooled. For economy-scale estimation purposes, we can ballpark that these installations will be cheaper than residential at something closer to $4/sq.ft. That pencils out at about $350bN ($0.35T) and another 1–2TWh storage opportunity.
(3) 87 (billion square feet of commercial buildings) X $4/sq.ft ~ $0.35 Trillion
2.3 Solar on every roof (supply side):
Australia and south-east Asia have demonstrated that the cheapest electricity available is the solar-generated electricity on your roof. This is because there are no transmission and distribution costs. The electricity is generated where you use it. Residential solar in Australia is routinely being installed at $1.30/W. Consequently, Australians are paying 6–7c/kWh for their solar electricity and installations are booming. For context, the average price of electricity in the US is ~13.8c/kWh. Largely because of soft costs (permitting, inspections, overhead), the average installed cost in the US is closer to $3.25/W. Let’s assume that we put a 15kW nominal (that means about 3 real all day kW³) systems on every roof⁴. That’s enough to cover the load for the electric cars⁵, the electric heat, and all of the other electrical loads under each roof. If you don’t own your home or it can’t go on that roof, the cost of nearby community solar (over car parks, on commercial buildings, in adjacent fields) will be similar. We use the $1.30/W Australian cost, assuming America can do at least that well. 1.30 X 128,000,000 X 15,000 = 2.5 Trillion. This will meet about 0.4–0.5TW or approximately ¼ of all the American power supply⁶.
(4) $1.30 (per installed Watt) X 128 (million homes) X 15 (kW) ~ $2.5 Trillion
2.4 A battery in every home (supply and demand):
Imagine that we already have installed a copious amount of storage capacity in the cars and trucks and water and space heaters of the above clean future. People get anxious about electricity black-outs the same way we get range anxiety with EVs, or nervous about our cellphone batteries lasting through the day. To address this, we’ll add 24 hours of storage for the electrical load of each home in addition to all of the above. This amounts to a 30kWh battery on average. Using the same $125/kWh costs as above, this is a $3,750 battery. 128 million of them will cost $450 bN or $0.48 T. This is yet another 3.6TWh of storage for the whole grid.
(5) 30 (kWh) X $125 (dollars per kWh) X 128,000,000 ~$0.5 Trillion
2.5 The rest of the electricity (supply side):
The US currently uses about 3.5TW of primary energy of all kinds. As we’ve discussed in Green New Deal: The enormous opportunity in shooting for the moon, we need less than half of that if we sign up for massive electrification. That means we need about 1.75TW. We’ll round that up to 2TW to account for growth and so that no-one can accuse us of underbidding. We already get 0.5TW of that from the solar on every roof as described above. The other 1.5TW we need to generate without also generating carbon emissions. Realistically the lion’s share will come from wind, solar, nuclear, and hydroelectric. Wind and solar both cost around $5/all-day-watt installed. This is a number that roughly factors in their $/W cost and their capacity factors.
Having experts agree on the true cost of nuclear power is extraordinarily difficult. Do we include security and waste storage costs? And then it gets more complicated from there. I have visited multiple nuclear plants, including a memorable visit at the invitation of Duke Energy (I was on a technical advisory board) to Catawba Nuclear Station in South Carolina. I remember the operator saying that they were the lowest-cost electricity on their network, and they were delivering at 2c/kWh with nearly 100% uptime. This number is the operational cost and doesn’t include the financed capital cost, but nevertheless, it is safe to say that nuclear may be quite cost-effective at scale. Whether it can be as cheap as solar and wind will be a debate for the ages. All of that is to say that for the purposes of this piece, let’s assume that nuclear will be installed at the same cost as solar and wind. $5/all-day-W. We need 1.5TW. That’s 7.5Terra$ or $7.5 Trillion.
(6) 1.5 (TerraWatts) at $5 (/ W) = $7.5 Trillion
2.6 Transmission and distribution (supply and demand):
We have to move all of that electricity around. The grid currently delivers close to 0.5TW. We now will have closer to 1.5TW running over it. We know from the current crop of wind and solar installations that about 10% of the project cost is the cost of transmission and interconnection fees from the remote industrial sites to the grid. We can thus assume a ballpark 10% of 7.5 Trillion dollars = $0.75 T for the new grid capacity to be bought to our homes and businesses. Almost certainly this number is conservative (high) as much more of the grid will be used bi-directionally, and therefore more efficiently, with the generation closer to the consumer than the average industrial wind or solar plant.
(7) 0.1 (10%) of 7.5 (trillion dollars of new capacity) ~0.75 Trillion dollars
3. What we have and haven’t included in the above numbers:
We have generated enough energy above to meet all of the energy demand of the US (and in fact allow for the total energy equivalent to increase by 15%). We have implicitly included a few days’ worth of storage which is enough to deal with most of the intermittency problems of solar and wind. We have replaced every car and retrofitted every home and commercial building so they can be carbon free. We have generated all of the energy with low and no- carbon sources.
What we haven’t done is include the cost of retrofitting industry to use electrical heat instead of natural gas. We haven’t factored in all of the energy in manufacturing. This will require some upgrades, probably estimable at 20–25% of the cost of all of the other components above⁷. As we previously mentioned, we are ignoring agriculture and land-use contributions of carbon at this point. We didn’t include the capital conversion costs to biofuels as substitutes for aviation fuels and some of the industrial heat and winter storage solutions.
4. So how much does it cost?
(I rounded up everything to the nearest significant digit. Too many significant digits really bothers me with its pretense at accuracy.)
$2.6 Trillion for electrifying all cars and trucks.
$2 Trillion for electrifying residential water and space heat.
$0.35 Trillion for electrifying commercial water and space heat.
$2.5 Trillion for solar on every rooftop, to supply nearly a quarter of our load.
$0.5 Trillion for a battery in every home.
$7.5 Trillion to build the rest of the grid.
$0.75 Trillion for the new transmission lines.
16.2 Trillion dollars in total.
Let’s add a little bit more into our budget to deal with the harder problems in the industrial sector like cement and cost-effective carbon-neutral biofuels.
OK, maybe $20 Trillion.
I don’t know how anyone can defend a price tag of $100 Trillion.
It’s worth mentioning that the country that wins the race to build the infrastructure to manufacture all of these systems will have an advantage in the global economy to come.
$20 Trillion still seems like an impossibly large amount of money. Remember, however, that every machine has a lifetime and a replacement cost. We can use the opportunity of the replacement of each of those machines as not only a chance to upgrade our lives and decarbonize, but as an effective discount on the cost of this project. The 100 Trillion dollar estimate assumes you will still be buying everything you do today, as well as everything you do tomorrow, whereas you’ll only be buying one of those things. If you are replacing your roof shingles anyway, installing solar will be cheaper at that moment. If you are replacing your furnace and hot water heater anyway, those costs offset the upgrade to zero-carbon heat pump technology. When you are going to purchase a new car anyway, the only real difference between an electric vehicle and one with a fossil fuel engine is the price of the battery.
Keep in mind that this spending will happen over decades. Cars and trucks last 13 years on average. HVAC systems 12–20. Asphalt shingle roofs 30 years. Residential solar 20–25 years. Industrial solar and wind 20–30 years. Nuclear 40–50 years. Grid transmission lines 25–50 years. Fossil-fuel-fired electricity generation plants last 30–50 years. By taking advantage of, and accounting for, the implied discount of getting the clean technology at the moment we’d be replacing the old technology anyway, the cost question gets a lot easier… Probably by half⁸.
So that brings it back to maybe $10 Trillion.
We could go twice as fast and decarbonize by 2030, only limited by political friction and materials supply and scale-up, and that would make the cost go up a little bit, but our climate outcome would be better.
It’s worth mentioning that the country that wins the race to build the infrastructure to manufacture all of these systems will have an advantage in the global economy to come. After scaling up its industry for WW2 production, America had an enormously productive economic engine with which to produce the durable goods for the rest of the world through the 1970s. Employment remained high and the economy boomed. A Green New Deal that similarly advantaged American industry would lead not just to a short-term win for decarbonizing the US economy, but a prolonged and profitable advantage as it used its industrial might and scale to export those technologies to the rest of the world, which will be playing climate technology catch-up.
We spend $1.25 Trillion per year on social security, unemployment, and labor.We spend $630bN per year on the military. $60bN on housing and community, $50bN on transportation. $1 Trillion on healthcare. And that’s only pieces of government spending. We spend a further $2 Trillion on private healthcare. The US construction industry spends close to $10 Trillion a year on goods and services. It’s not beyond reason to sign up to make the country more resilient, cleaner and healthier, safer, and more employed, by committing to this high-level plan to solve climate change.
We should remember that this plan requires a huge number of construction jobs, installation jobs, maintenance jobs, manufacturing jobs, and all of the jobs that support all of those jobs with accounting, food, components, and more. Many of these jobs will be necessarily domestic, and policies could make that even more likely, so a huge amount of that 10 Trillion dollar cost will actually be recognized as trillions of dollars of income for Americans everywhere.
What isn’t said enough, and what no-one who is trying to lampoon efforts at a Green New Deal will admit, is that once you have spent this $10 Trillion, we’ll have much lower, and very predictable, costs of energy for the 20–30 years all of those new machines will last. Not to mention that even AAA now believes electric cars have close to the lowest costs of ownership. If we could purchase the solar cells at zero interest, and install them for the Australian cost of $1.30 per W, each W will produce around 39 kWh in its lifetime at a cost of 3.3c/kWh⁹. This number shouldn’t surprise you; Sunfolding, an industrial solar company that I helped build, is routinely bidding on projects that are writing long-term electricity contracts to utilities at under 3c/kWh. These low costs will also be true of our wind turbines, and nuclear plants, which will quietly be supporting our cleaner, faster, quieter electric vehicles and more comfortable electrified homes. This supports an argument that US-government-backed low-interest financing should be key to a sensible GND. It’s akin to the US government financing and manufacturing buildup for WW2 and the reason we invented Fannie Mae and Freddie Mac in the first place. Honestly, it’s all going to come down to the interest rate.
The average American home pays $2000 for gasoline and $1945 for heat and electricity. That’s near enough to $4000 per household per year. That’s 0.5 Trillion dollars a year our households spend on energy. That’s half a trillion a year we won’t have to spend over the next 20 years (a total of $10 Trillion!) given our investments in capital-intensive but low-operational-cost clean technologies. Commercial and industrial businesses will see a similar saving.
A different way of stating this is that it is insane and tactical scaremongering to even quote a “cost” of solving climate change as one up-front number (whether it be $20 Trillion or $100 Trillion). We’ll obviously invent mechanisms to finance that, and the question comes down to whether the finance costs and interest will be lower than the ongoing fuel costs of our current infrastructure. It should also be mentioned that, like the manufacturing build up of WW2, it will likely be an enormous source of good jobs and income in all neighborhoods¹⁰. I’m strongly going to bet it will be cheaper to avoid fossil fuels and immediately go all-in on electrification.
Throughout this article, we have tried to build the argument from the home outwards. That is because we also have a choice of how tightly we integrate households into the future of grid and of infrastructure. If we put more of that infrastructure closer to everyone’s homes, the households will benefit more and the transmission costs will be substantially lower. Obviously, utilities will argue a different case with their legacy of monopoly on electricity. I doubt they will win the cost argument, but I’ll give them the opportunity.
To talk about it as one up-front cost is to misunderstand the problem. Yes, it’s hand-wavy, but this project will practically pay for itself. In the next column, we’ll go into detail about how it already does, and in fact, the only thing we really care about in thinking about the cost of solving climate change is the interest rate at which we finance it. Low-carbon technologies are the substitution of up-front capital costs weighed against fossil fuel’s ongoing fuel costs. It may not sound sexy, but stay tuned… we’ll show that we might just solve climate change with an interest rate.
I say ‘we’ a lot in this series of articles referring to the wonderful set of people who are providing me with voluntary feedback, suggestions, and editing help. Thanks, Arwen Griffith, Tim O’Reilly, Keith Pasko, Jim McBride, Jason Wexler, Pete Lynn, Patti Lord, Vince Romanin, and everyone else I’m forgetting.
¹ The total energy economy today uses around 3.5TW. The future energy economy will be as little as half that because an electrified energy economy is more efficient, so let’s call it 2TW. 20TWh (TerraWatt hours) is 2TW for 10 hours.
² This might be me over-reacting to the likes of IBM and the existing utilities that want to make this smart grid sound like a really hard problem when it isn’t. It’s really just a matter of having circuit boards that route our zero-carbon energy to our water heaters, cars, and other loads while taking advantage of the inherent energy storage capacity and load shifting possibilities of each. Switches that follow relatively simple rules like “heat the water heater now because the sun is shining and the cars aren’t home.”
³ Solar panels, inconveniently, are rated for their noon-time output. The sun outputs 1000 W/m² at noon, although the daily insolation in the US sits between about 200 and 300 W/m². Today’s solar cells capture about 20% of that. Consequently, you get 40–60 all-day W/m² of solar cell. This is near enough to 4–6W/ft².
⁴ For reference, a 15kW system would require about 300 sq ft of roof.
⁵ The cars won’t always be at home when the sun is shining, but we are assuming a future grid where your home is providing energy to the grid when it can and pulling energy from the grid when it can’t.
⁶ The power supply of the electrified future which will be half of what it is today.
⁷ Given this is roughly the proportion of an electrified industrial sector.
⁸ This is probably the most important aspect of drafting real climate policy or GNDs. Do we try to implement faster than the natural replacement rate (good for better climate outcome) or not?
⁹ $1.30 X 225/1000 (avg insolation/noon nominal) X 0.8 (efficiency) X 25 (years) X 8760 (hours in a year).
¹⁰ It is a reasonable argument to say that the US manufacturing buildup for WW2 was a bigger generator of jobs (and jobs that sustained) than the New Deal.