Our society’s prevailing economic zeitgeist assumes that everything has a price, and that both costs and prices can be objectively calculated, or at least agreed upon by parties involved in the transaction.  There are some big problems with this proposition.

Externalized costs are involuntary transactions — those on the receiving end of the externalities have not agreed to the deal.  Putting a price on carbon can theoretically remedy this failure in the context of climate change.  In practice it’s much more complicated, because our energy markets are not particularly efficient (as we pointed out in our Colorado carbon fee proposal, and as the ACEEE has documented well), and because there are many subsidies (some explicit, others structural) that confound the integration of externalized costs into our energy prices.

The global pricing of energy and climate externalities is obviously a huge challenge that we need to address, and despite our ongoing failure to reduce emissions, there’s been a pretty robust discussion about externalities.  As our understanding of climate change and its potentially catastrophic economic consequences have matured, our estimates of these costs have been revised, usually upwards.  We acknowledge the fact that these costs exist, even if we’re politically unwilling to do much about them.

Unfortunately — and surprisingly to most people — it turns out that understanding how the climate is going to change and what the economic impacts of those changes will be is not enough information to calculate the social cost of carbon.

Two Elephants in the Room

There are two other variables that powerfully influence any estimate of carbon and energy costs, even after you’ve accounted for climate externalities — in fact, they’re especially powerful after you’ve accounted for externalities.  The influence these variables have over our price and cost estimates is so large that today our choice of their values is likely more important than further refinements to our understanding of the economic impacts of climate change.  The two variables I’m talking about are risk tolerance and time preference:

Risk tolerance conveys how confident we need to be in the expected outcome to feel comfortable with it.

Time preference expresses the degree to which we value short term rewards over long term pain.

In conversation, I like to paraphrase these as: How lucky do you feel, punk? and How quickly would you like to end the world?  The answers to those questions are inherently subjective.  They represent philosophical propositions, cultural norms, and personal preferences.  They’re measurable only insofar as we can discern what values people hold based on their words and actions.  And based on the actions of our global civilization so far, it looks like we are a bunch of terminally ill compulsive gamblers playing Russian roulette with about four bullets.

Risk tolerance is important because the future is uncertain, and the further out we look, the more uncertain it becomes.  Our understanding of most complex systems is not deterministic — we can’t say “if we take this action then we will get that outcome” — not with the climate, not with the economy, and certainly not with their interaction.  Even with the best of models, our understanding is stochastic — we know there’s a range of possible outcomes; we can get at what that range looks like, what types of outcomes are more or less likely than others, and how sensitive those outcomes are to different actions we might take.  This isn’t necessarily indicative of a flaw in our models — many systems are not completely predictable.  No matter how rigorously we model the future — whether it’s coal production costs, the amount of sea level rise expected over the next century, or the economic consequences of that sea level rise on the city of Miami — there will always be a range of plausible outcomes.  Where we choose to aim within that range has a huge impact on our estimation of the price of action.  Are we happy with a 50% chance of abandoning Miami by 2050, or would we rather have a 5% chance?  Do we want to assume the “true cost” of carbon is represented by the median estimate of the expected damages, or do we want to be more conservative and use the 95th percentile?  These are judgement calls.

Time preference is important because many of the costs associated with our energy system are spread out over many years.  Our commitments to burning coal or natural gas last for decades, because that’s how long the power plants and pipelines last, and that’s how long it takes for the capital investments in new extraction to be paid off.  We’ll be absorbing the costs of climate change for far longer — potentially millennia.  If our civilization continues to place a high value on benefits in the present even when they come at the expense of the future,  then costs that are 500 or 50, or even just 20 years out can be rendered utterly inconsequential in our deliberations.  Time preference is a big part of why we “discount” future costs when we’re trying to compare them to present day costs.  If you’re not familiar with discounting, and you care about climate change, you absolutely must go read this piece by Dave Roberts at Grist.

Application to the Social Cost of Carbon

Late in 2013 the White House released a revised estimate of the Social Cost of Carbon. The headline number widely reported was $37/ton (of CO2 equivalent, emitted in 2015).  This revision made news because it was almost double the previous OMB estimate of $21/ton! The increase in cost was attributed largely to improvements in how sea level rise and its economic impacts were modeled — a change in our estimation of the externalized costs.

However, if you actually go read the most recent OMB white paper, you’ll see that the range of carbon costs they report is much larger than the difference between the old and new estimates.  For CO2 emitted in 2015, the costs vary by about a factor of ten, from $11 to $109/ton, with $37 being in the middle of the distribution.  The low end of the range ($11/ton) assumes a 5% discount rate and uses the average expected cost.  The high end of the range ($109/ton) assumes a 3% discount rate and looks at 95th percentile costs instead of average costs.  The only differences between these two estimates are the intangible time preference (discount rate) and risk aversion that are assumed on the public’s behalf.

Social Cost of Carbon | OMBDistribution of Social Cost of Carbon estimates for emissions in the year 2020 (slightly different from the $11/$37/$109 numbers above because the emissions year is 2020, instead of 2015).

After the earlier $21 estimate was published, some folks at the NRDC made the following arguments in favor of using a lower discount rate:

  1. Disrupting the climate will hamper economic productivity.
    This will likely cause economic growth rates to deviate below their historical trajectories. If worse-case climate risks materialize, climate change could even reverse economic growth. In that instance, people in the future would be poorer than people today, not wealthier.
  2. Past economic growth is artificially inflated.
    Historical market returns do not take into account pollution externalities resulting from production, such as the liquidation of natural capital, public health damages, or other potentially negative social impacts related to economic production, such as inequality. They therefore tend to overestimate the impact growth has on real social welfare.
  3. Inter-generational discounting is ethically dubious.
    In large part, an individual’s time preference reflects the rate at which they are willing to borrow from their future income in order to consume more today.  When that “borrowing” takes place across generations, the borrowers (in the present) and the lenders (in the future) are different people, and the lenders are unable to consent to the loan.  That makes it a lot more like stealing.
  4. The benefits of economic growth are unevenly distributed.
    Even if income grows under a changing climate, it is unlikely that the people most harmed by climate change will be the primary beneficiaries of that growth.  If you doubt this, just look at the reaction of the developed nations (who are the primary beneficiaries of historical fossil fueled industrialization) to requests from the developing nations for financial support in adapting to and mitigating climate change resulting from those historical emissions.
  5. Money isn’t everything.
    Not all damages can be monetized.  What is the dollar value of a billion person-years worth of pain and suffering, or the eradication of entire coastal cities and low-lying nations?  Thousands of years of lost cultural heritage?  How much is it worth to avoid a mass extinction event, or the destabilization of global agricultural production?  Unabated climate change is economically non-marginal in the extreme.  It extends well beyond the boundaries of “the economy” and into the realm of existential risk to civilization.  Many would argue that it is worth avoiding at literally any financial cost.

Some of these criticisms can be addressed numerically.  You can imagine estimating what fraction of historical economic “growth” is actually attributable to the liquidation of natural capital, and correcting the discount rate accordingly.  Addressing the absurdity of “monetizing” the continuation of civilization is much closer to writing a philosophical tract.

The NRDC authors did their best to perform an alternative analysis, accounting for the numerical differences of opinion.  Instead of the OMB’s discount rates of 2.5%, 3% and 5% they used 1.0%, 1.5% and 2.0%.  Unsurprisingly the result was much higher estimates for the social cost of carbon, with central estimates ranging from $55 to $266/ton (compared to the old $21 central estimate).  The 95th percentile cost estimate for the 1% discount rate case was $758/ton!

All of the above criticisms can be similarly leveled at the revised OMB analysis, which also uses discount rates ranging from 2.5% to 5%.  Applying the NRDC discount rates would result in central estimates of the social costs of carbon grossly in the range $100/ton to $500/ton.  Using a 1% discount along with the more conservative 95th percentile, you’d get a cost more like $1500/ton.

So while the headlines highlighted OMB’s “big shift” from $21 to $37, I think the real story here is that if we have the courage to re-examine the value judgements that underlie these estimates, then we find that those judgements wield far more influence on the cost estimates than any revision of the science.  With a 5% discount rate and a 50% chance of failure, we get a price of $11.  With a 1% discount rate and a 5% chance of failure the cost is more than 100 times as high.  It’s also worth noting that private corporations routinely use discount rates of 6-8% (for debt financing) and 10-20% (for equity), which renders the long-term impacts of their actions completely irrelevant (to them).