*The social cost of carbon (SCC) is a very important number due to the influence it can have on US federal regulations and other climate polices. Its sensitivity to the discount rate—a measure of a society’s preference for valuing benefits more when those benefits are received sooner rather than later—calls for careful attention to how we choose the discount rate that feeds into new estimates of the SCC. New research from Resources from the Future aims to improve the SCC by carefully, and empirically, determining the most appropriate discount rate.*

The discount rate is a tool that is used to compare the value of future impacts to those experienced today; it’s important for understanding the social cost of carbon (SCC). The basic idea of discounting is simple: for a variety of reasons, people generally tend to value benefits (e.g., income, consumption) more when those benefits are received sooner rather than later. The discount rate reflects the strength of this preference, and the discount rate can have huge implications for how we evaluate impacts that occur far in the future.

In the context of climate change, the SCC is well known to be sensitive to the value of the discount rate. A higher discount rate indicates a higher value placed on immediate benefits relative to delayed benefits received in the future. For example, while the Biden administration’s interim $51-per-ton SCC estimate is based on a 3% discount rate, changing that rate to 2% more than doubles that SCC estimate to $121 per ton. By the same token, the Trump administration used a much higher 7% discount rate to arrive at a much lower SCC estimate of about $6 per ton. The sensitivity of the SCC to the discount rate owes to the fact that the impacts of carbon dioxide emissions today are very long lasting, since those carbon dioxide molecules remain in the atmosphere for hundreds of years, and the warming effect of those molecules lasts even longer.

The SCC is important for US government climate policy, because higher values motivate stronger regulatory action. Formally, the SCC appears prominently in the “benefits” column of the benefit-cost analyses routinely performed by the federal government for all major regulations, such as vehicle fuel economy standards or regulations that require power plants to reduce their emissions.

Historically, the discount rates used for federal benefit-cost analyses have been guided by a policy issued by the Office of Management and Budget (OMB) in 2003 called Circular A-4. The policy recommends using two discount rates for regulatory analysis: 3% and 7%. However, recent research has found that that these two rates are no longer appropriate, for several reasons.

First, the 3% rate was based on a calculation of average real interest rates for US Treasuries that spanned the three decades leading to 2003. Since then, market interest rates have fallen dramatically. As a result, a contemporary update to that calculation, but spanning the last three decades or so, would result in a lower discount rate of around 2%.

Second, it’s important to explicitly consider uncertainty about appropriate discount rates in the future. As demonstrated by economist Martin Weitzman, uncertainty about future discount rates suggests using lower and lower rates for longer and longer time horizons, which is particularly relevant for the very long time horizons considered in the SCC. In our recent research, which we describe in more detail below, we show how economic analysis can incorporate these uncertainties explicitly into the estimation of the SCC.

The 3% rate was based on a calculation of average real interest rates for US Treasuries that spanned the three decades leading to 2003. Since then, market interest rates have fallen dramatically.

Third, the underlying 7% rate is founded on some extreme assumptions that are especially inappropriate for discounting effects over time horizons that are relevant for climate change. The original motivation for the 7% rate in the OMB’s Circular A-4 was that the costs of regulations could apply to capital investment rather than consumption, with investment having a higher rate of return due to tax distortions. Rather than addressing this cost issue by adjusting the discount rate in benefit calculations, the “shadow price of capital” approach, long recognized as conceptually correct by the OMB, can be implemented. Recent public comments from Pizer show that multiplying regulatory costs by a shadow price of capital of 1.2 would reflect a case of all costs falling on investment. The possibility of some or all *benefits* falling on investment is absent from the current OMB discounting approach.

Now that the federal Interagency Working Group (IWG) on the SCC has been reestablished, the time is right to revisit the US government’s approach to discounting in the context of SCC estimation. Perhaps recognizing the shortcomings of the current approach to discounting, the Biden administration has issued an order that directs the OMB to update the regulatory review procedure, which includes revisions to Circular A-4. We suggest tackling three specific aspects in the revisions:

- Focus discounting on an updated, lower consumption rate of interest.
- For long-term decisions such as climate change and the SCC, incorporate uncertainty in the discount rate by using a discounting framework that links the discount rate to future economic growth.
- Apply a shadow price of capital to investment impacts instead of discounting benefits at an investment rate of return. In a simplified, conservative sensitivity case, apply a 1.2 shadow price of capital to regulatory costs.

### Discounting Advances in Detail

The arguments summarized above suggest that the OMB should consider revising its central consumption discount rate. Indeed, a landmark 2017 report from the National Academies of Sciences, Engineering, and Medicine (NASEM) recognizes the need for revisions and issued a series of recommendations to the IWG for future updates to the SCC.

Beyond the question of the appropriate *level* of the discount rate, one of the issues recognized in the NASEM report involves deeper economic questions about how to consider *uncertainty* about the discount rate, and whether future discount rates should depend on how future economic growth unfolds. As discussed in an accompanying article of this issue of *Resources*, recent work by Resources for the Future (RFF) and external collaborators at the University of California, Berkeley; the University of Washington; Harvard; and Princeton has made it possible for analysts to use fully probabilistic, detailed projections of population, economic growth, emissions, and the climate system in estimating the SCC.

These new probabilistic socioeconomic projections—we call them RFF socioeconomic projections—represent a major improvement over the simpler socioeconomic scenarios used in the past, but the use of these kinds of projections also raises new conceptual questions. For example, classic economic theory going back nearly a century shows that one should use a higher discount rate when economic growth is higher, and a lower rate when growth is lower. The intuition is this: Society naturally values a dollar’s worth of avoided future climate impacts more if that impact is on a relatively *poor* society in the future. On the other hand, a dollar’s worth of future climate impacts has less value if it falls on a relatively *rich* future society. Indeed, part of the rationale underlying the discounting of future outcomes is that the society is likely to grow wealthier over time. When uncertainty exists about the future path of income growth (as discussed in an accompanying article in this magazine), corresponding uncertainty exists about how much we should discount future impacts.

The relationship between uncertainty in economic growth and the discount rate is highlighted in the NASEM report, which recommends that updates to the SCC incorporate this relationship. The recommendation stands in contrast to the US government’s existing approach to the SCC, which features no such link because of its use of a fixed discount rate, such as the constant 3% value. But our work argues that the calculation of the SCC should feature the following two things: include a probabilistic set of socioeconomic projections, and apply a discount rate that’s contingent on economic growth in each projection—what we call “stochastic growth discounting.”

While the NASEM report strongly recommends that the SCC should account for this relationship between economic growth and the discount rate, it was not clear at the time exactly how the strength of that relationship would be determined. For example, given two projections—one with 1% growth and another with 2% growth—how much higher should the discount rate be in the second projection? To say the same thing in technical language: What is the mathematical relationship between the discount rate and growth?

Without getting into too much technical detail, this relationship between the discount rate (𝑟) and the economic growth rate (𝑔) typically depends on the two parameter values in the equation 𝑟 = 𝑎 + 𝑏𝑔, where 𝑎 reflects a constant element to how much we discount impacts on society over time, and 𝑏 reflects how much we discount future impacts because society has grown wealthier.

The question comes down to choosing those two values. One potential approach is to choose them *prescriptively* based on one’s ethical beliefs or on surveys of the economics profession. An alternative approach is to choose them *descriptively*; for example, by choosing values such that the discount rate calculated by the equation above matches observed market interest rates—a method embraced by prominent climate economist William Nordhaus.

Each of these methods faces potential problems for the US government’s calculation of the SCC. First, the US government largely has embraced descriptive over prescriptive approaches: recall that the OMB’s Circular A-4 determined its 3% and 7% discount rates descriptively by referring to observed market rates. But under the descriptive approach typically used by the US government, many different pairs of parameter values can be tuned to match any given interest market rate. For example, if growth is 𝑔 = 2%, both 1% + (1 x 2%) and 0% + (1.5 x 2%) yield a rate of 3%. Clearly, the simple method of matching a target market rate is insufficient; more information is needed to narrow down the parameter values descriptively.

This is where our research comes in, detailed in a recent paper, with implications that were first highlighted by climate economist Martin Weitzman. Weitzman noted that if uncertainty exists about the “right” discount rate, then we should use a lower and lower rate for impacts occurring further and further into the future. Exactly how much lower depends on how uncertain we are about the appropriate discount rate—the more future uncertainty, the lower the rate.

Returning to the equation 𝑟 = 𝑎 + 𝑏𝑔, uncertainty in the discount rate is closely tied to uncertainty in economic growth, at a rate determined by the value 𝑏: higher or lower values of 𝑏 imply higher or lower degrees of uncertainty in the discount rate. Recent research has presented evidence for uncertainty in both the left-hand side of this equation (interest rate uncertainty) and the right-hand side (growth uncertainty).

In particular, we calculate the parameter values that reconcile these two sources of evidence. For example, the 𝑎 and 𝑏 values of 0.2% and 1.24, respectively, when applied to the growth data from the RFF socioeconomic projections in a forthcoming Brookings paper, delivers a particular term structure of discount rates: a near-term effective discount rate that starts at 2% and reflects the uncertainty about that rate going into the future. The shape of this term structure is consistent with the behavior of interest rates. This leads to an effective discount rate (the technical term is the “certainty-equivalent rate”) that declines over time (Figure 1) and stays consistent with empirically estimated term structures from recent macrofinance research. Note that in application, however, the specific rate used for discounting future impacts would depend on a particular scenario, rather than being set as a fixed declining path.

This new work provides a set of descriptively driven discounting parameters that the US government can use, and it is directly responsive to the NASEM recommendations on discounting. In addition, our new work and our forthcoming article in the *Brookings Papers on Economic Activity* series demonstrate that incorporating this empirically driven relationship between economic growth and the discount rate is crucial when estimating the SCC amid uncertain economic growth.

All told, our proposed approach to discounting includes three components:

- Focus discounting on an updated, lower consumption rate of interest.
- For long-term decisions such as climate change and the SCC, incorporate uncertainty in the discount rate by using a discounting framework that links the discount rate to future economic growth.
- Apply a shadow price of capital to investment impacts instead of discounting benefits at an investment rate of return. In a simplified, conservative sensitivity case, apply a 1.2 shadow price of capital to regulatory costs.

This last sensitivity analysis, the third item in the list above, does not relate specifically to the SCC, and would be implemented generally in benefit-cost analysis. By using a shadow price of capital approach, benefit-cost analysis can address concerns about capital impacts, hew more closely to longstanding economic guidance, and avoid substantially undervaluing future benefits. Each of these advances in the approach to discounting could be incorporated in a revision to the OMB’s Circular A-4, with relevance to both SCC estimation and other contexts. This integrated approach to discounting would harmonize SCC discounting and broader US government guidance on benefit-cost analysis while allowing the US government to update its treatment of the discount rate to reflect the best available science.

*A summary of the research efforts that have informed this article will be published in the* Brookings Papers on Economic Activity *series.*