About $40 per ton of CO2. Or maybe over $100. Or maybe thousands of dollars. Or maybe, just maybe, all those greenhouse gasses (GHGs) are good for the world and we should be subsidizing fossil fuel extraction after all.
The social cost of carbon is a critical metric for socially responsible investors. And all investors need to incorporate carbon costs into their analysis given its growing importance to policymakers. But, the lack of consensus on the right approach requires us to sort through model mechanics and political motivations to find a useful set of estimates.
What Is the Social Cost of Carbon?
The social cost of carbon reflects the present value of all the economic benefits and costs related to climate change. It is typically represented as the marginal cost (or benefit) to society of an additional metric ton of carbon or carbon dioxide emitted by human activity. There is a consensus that the global effects of climate change are negative above 1° Celsius, although a small degree of global warming may be positive on balance for some regions. Because the world is already about a degree warmer than before industrialization, the debate is simplified to how much additional warming will harm global welfare.
To estimate this cost, economists have constructed models that relate economic activity to GHG emissions, GHG emissions to global temperature changes, and global temperature changes to economic damages. The present value of the economic losses are used to compute the social cost of carbon.
These models are complex and produce estimates of the social cost of carbon that are sensitive to a handful of assumptions. Experts don’t yet agree on the sensitivity of the climate to a given level of atmospheric GHGs. Nor do they agree on the level of damage caused by warming or even the nature of the relationship between temperature and economic losses (though most believe damages will increase exponentially with incremental warming). Perhaps the most important source of estimate variability is the discount rate used. Since most models covers several decades or centuries, the rate at which future damages are discounted carries an unusually heavy weight.
Whose Discount Rate Is It Anyway?
Multi-generational models don’t lend themselves to using discount rates observable in financial markets. The best risk-free rate proxy, the 30-year U.S. Treasury Inflation Protected Security (TIPS), is inadequate for evaluating policies that will affect my unborn grandchildren, much less their grandchildren.
Social scientists often rely on the Ramsey Rule, which states the appropriate discount rate should incorporate people’s preference to consume today rather than in the future and their desire for smooth consumption over time. The first concept is formally known as the pure rate of time preference, while the second is called the relative aversion to inter temporal inequality.
There is a healthy debate on the proper pure rate of time preference to use. But, if you believe our great-grandchildren’s welfare is as important as your own, the right answer is zero. The well-known Stern Review makes the case that the possible extinction of humanity is the only ethically appropriate reason to use a positive pure rate of time preference in formulating a discount rate.
Using zero for the pure rate of time preference is NOT the same thing as using a zero percent discount rate. The discount rate will still be a positive number determined by risk aversion and economic growth estimates.
Stand on the Shoulders of Giants
Dozens of published studies provide estimates of the social cost of carbon. Two more recent meta-analyses of these studies offer a solid footing for investors trying to find a credible estimate. Wang et. al (2019) develops a linear regression equation using 58 studies that provides an estimate of the social cost of carbon based on chosen parameters. And Tol (2018) summarizes the findings of 22 studies that includes segregating estimates based upon the chosen pure rate of time preference.
I estimate a social cost of carbon of $128 per ton of CO2 using the Wang regression equation. The key inputs I used are zero for the pure rate of time preference, and a desire for estimates derived from peer-reviewed studies with detailed modeling of uncertainty — models with uncertainty tend to boost the estimated social cost of carbon.
Tol’s mean estimate of the social cost of carbon is $185 using pure rate of time preference of zero. Because the Wang regression equation is linear, while the affect of varying discount rates is non-linear, this may partly explain while the Tol estimate is higher. Yet, the Wang model has its own advantages, such as ability to specify a preference for more sophisticated models of a certain flavor. I intend to use both figures in modeling the potential carbon costs of individual companies.
Socially conscious investors can use academic research to derive a social cost of carbon for their own analysis. But the $128 and $185 figures I intend to use are MUCH higher than what politicians around the world are now using. Investors whose interest is limited to the monetary impacts of carbon policy need another approach.
U.S. government agencies have been using an estimate of the social cost of carbon to evaluate its rulemaking since 2008. From 2010-2017, an interagency working group published unified estimates for the entire federal government. The last update in 2016 pegged the 2020 social cost of CO2 at $42 (in 2007 dollars) using a 3% discount rate. The $42 figure is a mean estimate of simulations using three different models — the 95th percentile of the distribution of estimates at a 3% discount rate is $123 for 2020. Other estimates were published at 2.5% ($62 for 2020) and 5% discount rates ($12 for 2020).
The published figures appear weighed down by political resistance, despite the inclusion of academic input. Because carbon-intensive industries are lobbying aggressively against any price on carbon, bureaucrats are reluctant to put forth cost estimates that risk the ire of campaign donors or threaten to scare the voting public. Experts generally support a discount rate for carbon cost estimation in the range of 1-3% — the median discount rate supported is closer to 2%. That $40-some dollar figure treated as central estimate would be more like a floor number among a broader group of social scientists. And let’s not waste time on the “estimates” emanating from the U.S. executive branch since early 2017.
Europe is well ahead of the U.S. on tackling climate change with carbon policy, but political resistance has slowed progress there as well. The European Union Emissions Trading Scheme (EU ETS) has been the world’s largest trading market for CO2 emissions since its launch in 2005. Prices for CO2 emissions allowances languished for years following the financial crisis but have since rebounded to nearly $30/ton of CO2 as the bloc has agreed to reduce the allowance surplus. The EU ETS covers about half of the area’s emissions — road transport and residential and commercial heating are not included.
Carbon taxes and emissions trading scheme prices in most countries are still below $30/ton with few exceptions (the most progressive nation is Sweden with a tax of approximately $130/ton). Even progressive California sports a carbon emissions trading price of just $15/ton of CO2.
How should investors model potential government carbon policy? The approach of governments is typically two-fold. First, start the effective carbon price at a low level to limit the economic shock and build popular support. Second, step up the price level over time to meet emissions targets or reflect a reasonable estimate of the social cost of incremental emissions.
For a specific company, the current year effective carbon price will be the sales-weighted average price across the countries where the company does business. In the long-run I expect most carbon pricing schemes implemented will be targeting a carbon price that reflects its estimated damages — so carbon prices could eventually be well over $100/ton. The challenge is figuring out how to interpolate between now and then. For modeling purposes, I’m assuming the long-run begins in the year 2030.
Why 2030? The 2030 Agenda for Sustainable Development was launched by the United Nations in 2015 and is probably the closest we have to a global vision of economic development. National policies affecting progress toward the agenda’s major goals (including mitigating the effects of climate change) will be under a global spotlight. And if emissions haven’t peaked by about 2030, the likely failure of keeping global warming below 2° Celsius will be clear. Such failure makes aggressive carbon policy even more likely post 2030.
The Most Important Number for Socially Responsible Investors
Estimates of the social cost of carbon are a critical ingredient for socially responsible capital allocation. My best guess is that the damage avoided by emitting one fewer ton of CO2 is well over $100. Long-term investors interested only in government imposed carbon taxes or emissions trading schemes are still advised to model the likelihood that current carbon policy and actual carbon costs will converge over the next decade plus.