New Study (Part I of II): Feds Underestimate Costs of Carbon Pollution, Low-Balling Climate Change's Impact on our Children and Grandchildren
This post was authored by Laurie Johnson, NRDC Alum.
The U.S. government is significantly underestimating the negative impacts climate change will impose on our children and grandchildren, according to a new study published today in the Journal of Environmental Studies and Sciences. In the article, my coauthor Chris Hope (Judge Business School, University of Cambridge) and I show how regulatory agencies are using a faulty model to estimate carbon pollution damages that all but ignores the economic damages climate change will inflict on future generations.
After valuing these costs more completely, our study finds that carbon pollution will impose damages between 2.6 to more than 12 times higher than the government’s main estimate. The government put the value of the damage caused by carbon pollution at $21 per ton of CO2, whereas our revised estimates place the cost between $55 and $266 per ton. Importantly, even these revised estimates may be too low: they correspond to what might happen if future temperature increases are in the middle of scientists’ projections—not any of the worse-case scenarios they warn us about. Additionally, the model left out many damages that couldn't be quantified.
The government’s approach matters a lot because its numbers are used to decide what actions should be taken now to fight climate disruption, such as the choice of what type of electric power we should build, which would change if more accurate estimates were used.
We incorporate the estimated climate costs (ours and the government’s) into the cost of electricity generation, concluding that the real price of fossil fuel generation is much higher than cleaner energy sources after properly accounting for pollution damages.
While natural gas might appear to be the cheapest generation option for new power plants, using renewable sources or advanced fossil fuel generation with pollution capture technology is more cost effective in the long run than building new conventional plants using either natural gas or coal, with real economic benefits accruing quickly and increasing over time.
In supplementary analysis to the paper, I also find that it would be cost effective to replace some of our existing power plants with cleaner sources, rather than continuing to operate them.
This blog distills some of the technical details in the study, for readers interested in the underlying analysis and its assumptions. First I outline the major flaw in the government’s approach that we critique, namely how it “discounts” future climate damages. In this section I include some necessary technical background on discounting and its rationale, and how our discounting differs with that of the government. Next, I give five fundamental reasons behind our critique of the government’s discounting.
The companion blog linked to above gives the comparison of electricity generation costs with and without consideration of pollution damages, under our assumptions versus the government’s.
Overview: flaws in the Government’s approach
Discounting future climate damages
In February of 2010, an interagency committee of the U.S. government published its first estimates of the “social cost of carbon” (SCC), a monetized value of the future damage caused per ton of carbon dioxide (CO2) emitted today. The committee was established to produce a uniform estimate of the SCC for use by regulatory agencies, such as the Department of Transportation and the Environmental Protection Agency, in carrying out energy efficiency and environmental protection statutes relating to carbon pollution and climate change.
Our study faults the joint agency report for using unjustifiably high “discount rates” to translate future climate damages into “present values.”
Discounting is a standard practice in economics when calculating costs and benefits over time. Because money today is more valuable than the same amount of money in the future due to interest and economic growth, future dollars are not directly comparable to dollars today. Economists therefore discount future income and cost streams to express everything into one standardized present value. For short periods of time, discounting often makes sense. However, over extremely long time periods, it can be quite problematic (as discussed further below).
The discount rate effectively determines a multiplier between 0 and 1 that is applied to future costs and benefits. This multiplier (referred to as the “discount factor” in the economics literature) gets smaller the higher the discount rate and the farther out in time. For example, the Administration’s lowest discount rate of 2.5 percent per year would value $100,000 worth of climate damages happening thirty years from now at approximately $48,000 (and at one hundred years roughly $8,500). Alternatively, using the highest discount rate of 5 percent, $100,000 in thirty years would be approximately $23,000 today (and at 100 years it would be $760).
For the reasons we give below, the government’s discount rates are unjustifiably high. We therefore re-estimated its model using (lower) “inter” generational discount rates permitted in official guidelines from the Office of Management and Budget when multiple generations are affected by a regulation.
Discounting: how it works and its rationale
Discounting mimics interest yields in bank accounts, but in reverse: if you could get a real (i.e. inflation-adjusted) return of 3 percent per year, $100,000 in thirty years would be equivalent to approximately $41,200 today. Why is that? It is simply compound interest running in the opposite direction: if you deposited $41,200 into a bank account today it would grow to approximately $100,000 in thirty years. You’d have a real net gain of $58,800. Economists thus say that $100,000 in thirty years should be valued today at $41,200. If we are looking instead at one hundred years, $100,000 in one hundred years is only $5,200 today ($5,200 deposited today would grow to $100,000 in one hundred years at 3 percent).
Discounting makes sense when applied to costs and benefits that accrue to one individual in his or her lifetime: most people would prefer to spend money today than to wait and spend it later (we are impatient creatures and like instant gratification); those with money to spare like to invest it so that it will grow to something larger in the future.
But discounting costs and benefits that accrue to different people over long time periods, such as the costs associated with climate change, runs into both a logical and ethical complication: a person today enjoying the instant gratification of consuming goods whose production generates climate pollution is not the same person harmed by the associated carbon emissions.
Nevertheless, two ethical arguments have been made for inter-generational discounting. The first runs as follows: people in the future will have more income as a result of economic growth, and can therefore withstand some climate damages without being any worse off than people today. Assuming all things can be measured and compared in monetary units, if real income grows at 3 percent per year, society can bear some climate damages with no net loss. Using our example above, if a person earning $41,200 today saw his or her income grow to $100,000 in thirty years, she would be almost $59,000 richer by then. By this logic, she could suffer up to $59,000 in climate damages and be no worse off.
The second rationale is less obvious: if the expected “return” (i.e. reduced climate damages) from emissions reduction is less than the return you could earn on markets, people in the future might be better off had you instead invested in traditional market assets rather than emissions mitigation.
For example, if our hypothetical $41,200 were invested in mitigation that resulted in avoided climate damages of $64,400 over a period of thirty years, that investment would yield a net benefit of $23,200 ($64,400 in avoided damages minus $41,200 spent on reducing emissions). This is an effective return of 1.5 percent per year ($41,200 growing at 1.5 percent per year over thirty years increases to $64,400). However, if the money were invested in an alternative asset yielding a 3 percent return, a future society would end up with over $102,000, which would more than offset the $64,400 in economic damages.
In effect, under these assumptions, the mitigation investment option would leave the future society with $37,600 ($102,000 minus $64,400) less in the future than it could have had. Of course, embedded in this example is the assumption that the economic damage caused by unmitigated climate disruption is small, so the assumption determines the answer. As we discuss below, since we do not know that such damages will be small (indeed current knowledge does not allow us to rule out very large damages), sound policy choices must examine cases where they are large.
Our discount rates versus those used by the Administration
To value future climate damages, the government examined discount rates of 2.5, 3 and 5 percent, specifying the estimate corresponding to the 3 percent rate as its primary one. Rates of 3 and 5 percent are typically applied to intra-generational costs and benefits that affect primarily the current generation. However, for inter-generational benefit-cost analysis, the OMB guidelines permit agencies to use rates ranging from 1 to 3 percent, based upon a review of the economics literature (see p. 36 of OMB Circular A-4 guidelines). Nonetheless, the government’s analysis excluded values from the inter-generational range lower than 2.5 percent.
To illustrate the importance of considering lower discount rates, we re-ran the government’s model with discount rates of 1, 1.5, and 2 percent (though for reasons discussed below, even these may be too high).
For comparison, at 1, 3 percent and 5 percent, $100,000 worth of climate damages thirty years from now has a present value of $74,200, $41,200 and $23,100, respectively (one hundred years from now, the present value would be $37,000, $5,200, and $760, respectively).
Five reasons the government’s discounting is flawed
The agencies’ discount rates of 2.5, 3 and 5 percent are based upon historically observed returns on various market assets. But these rates overstate, and potentially by a large degree, how well people may fare in the future, and the risks they are willing to assume.
Our paper makes three points in this regard (I add two further ones below):
1. An increasingly disrupted climate may hamper economic productivity, causing 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.
If people in the future could speak to us about climate, and if more children alive today understood the dangers of climate change, they might object to our banking on income growth to clean up the mess we are leaving them. They might insist we adopt a precautionary framework, where the objective is to minimize risk rather than maximize returns relative to alternative market investments.
A precautionary framework is similar to insurance markets, where one pays to insure against low probability, but catastrophic, events (e.g. auto, fire, the death of a parent with young children). Contrary to expecting a positive return comparable to markets, when people invest in insurance they never expect (or hope) to “cash in” on it.
The case for a precautionary framework with climate change is particularly strong relative to one for risks for which we regularly take out insurance. The magnitude of the catastrophic events associated with climate change is worse, and their probability of occurring significantly higher. Rather than being an isolated random event happening to just one individual, climate change could radically alter the ability of everyone to live safely on the planet. In light of this, trying to minimize risks would seem a more prudent approach than trying to maximize returns—and the framework people in the future would make us choose if they could.
For this reason, many economists and policy makers argue for much lower discount rates than those used by the Administration. Indeed, some recommend against discounting altogether, advocating instead an emissions reduction target corresponding to a maximum level of risk we would choose to impose on ourselves.
2. Private investors (and hence market returns) do not take into account pollution externalities resulting from production, such as the depreciation of natural capital (e.g., loss of natural habitats to development and pollution) and 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. According to a range of estimates cited in the article, negative externalities suggest real historical growth rates that are 0.2 to 3.2 percentage points lower than official statistics.
3. A lower discount rate should be used across generations than within generations for another ethical reason.
I touched upon this point in the previous section, but it merits special attention. Part of an asset’s return has nothing to do with expected income growth; it’s simply the rate at which people are willing to borrow from their future income in order to consume more today, and how much lenders require to delay the use of their assets by lending it them. Economists call this aspect of the discount rate the “pure rate of time preference.” The problem is that it refers to trade-offs a person is willing to make about her own income, not hers now versus someone else’s later. Apart from being unethical, time preference discounting is illogical for climate change damages: many of the people benefitting from emitting CO2 today are not the same people as those that will be harmed by it in the future.
The famous economist, Frank Ramsey, who identified time preference discounting as part of interest rate savings theory in 1928, said that would be “ethically indefensible” (emphasis added) to apply a positive rate of time preference across different generations. A common estimate in the economics literature for the time preference component of discounting climate damages is 1.5 percent.
Taking 2) and 3) together, adjusting the government’s discount rates downward by subtracting out negative production externalities and time preference discounting easily suggests a 0 percent discount rate, potentially even negative. The midpoint (1.7 percent) of the production externalities estimates, together with the average rate of time preference discounting (1.5 percent), would shave off 3.2 percentage points from the discount rate.
Two final points
Though not discussed in the article, two additional arguments for low discount rates can be made:
4. Even if income grows under a changing climate, it is unlikely that the people most harmed by climate change will be the recipients of that growth.
Recent income growth trends do not bode well for people who will be the most heavily impacted by climate disruption, the poor and middle class. In the last four decades, income gains have been highly skewed toward the wealthy. According to new data from the Congressional Budget Office, the bottom 99 percent of the income distribution saw average income (after taxes, and government income and health insurance transfers) increases of only 1.3 percent per year. In contrast, the top 1 percent averaged 3.2 percent per year.
5. Money isn’t everything.
Many of the monetized damage estimates in the government’s model (to say nothing of damages not included) do not capture pain and suffering that accompanies climate destruction, or the intrinsic value we attach to nature. For example, the monetized value of properties and infrastructure lost to extreme weather excludes multiple personal harms associated with their destruction, including lost loved ones, lost jobs and the important health benefits attached to them, lost cultural heritage, sudden homelessness, and the destruction of precious ecosystems and entire communities.
 The discount factor formula is given by 1/[(1+r)^t], where r is the discount rate expressed in decimal point (e.g. a discount rate of 3 percent would be equal to .03), and t is the number of years between the present and when the future cost or benefit occurs.
 We think of only “real” interest rates when discounting, because if there were inflation, you would just add the inflation rate to the discount rate, and there would be no difference in the present value. For example, if you expected 1 percent inflation on top of a 3 percent return, an asset would grow by 4 percent per year. While it would grow 1 percent more per year, it would also be discounted back by that much more. In this example, $42,000 would grow to $102,000 in thirty years; discounting $102,000 back by fifty years at 4 percent gives you $42,000.
 It also had a “worse-case” estimate using the 3 percent rate. For each discount rate, the model was run repeatedly, each time randomly picking an assumed temperature increase from a range of potential increases predicted by climate science models, which the model then translates into economic damages. The government’s worse-case estimate is the 95th percentile value using the 3 percent discount rate model runs, and corresponds to the value below the highest five percent of all the estimates. The primary estimates given using 2.5, 3 and 5 percent are simply the average of the estimates over all runs. For comparison, the government’s worse-case value is close to the average value we obtain using a 2 percent discount rate, $65/ton of CO2 versus $62/ton.
 Note that OMB guidelines specify two standard rates of 3 and 7 percent, regularly applied to benefits and costs that occur within the current generation. However, the government used 5 percent as its upper value rather than 7 percent for climate damages, for the reason that they are expected to primarily and directly affect consumption rather than the allocation of capital. A rate of 7 percent corresponds to returns on capital investments (see p. 33 of OMB Circular A-4 and our paper for further explanation).
 This is especially the case for the poor. See, for example, this analysis by a researcher at DePaul University on the effects of Hurricane Katrina. Simo, Gloria (2008). Poverty in New Orleans: Before and after Katrina. Vincentian Heritage Journal 28(2), article 22.
 Calculations made from Supplemental Table 7. Note that CBO revised its method for imputing in-kind income from government-provided health care in a way that significantly increased income for low income households, mostly to the bottom 20 percent of households. It presented results using its previous methodology in Table 4 of the main report. Using data from that table, the bottom 99 percent of households had an average annual increase of 1.1 percent per year rather than 1.3 percent.