Natural Resource Damage Assessments Tools and Applications
Environmental justice is a major initiative of the Biden administration and various state governments, with a number of directives and programs to address fairness in the execution of governmental environmental activities. One example of this is the award of $750,000 in grants by the city of Seattle. This grant will go to environmental justice projects impacted by climate change.
Almost all recent attention has been directed to justice across geographies – how current and near-term actions may differentially affect disadvantaged communities. But environmental justice, and climate justice in particular, also has a time dimension. The discount rate used to evaluate the impacts of alternative policies across time has an outsized impact on such evaluations. The issues are especially well reflected in the context of natural resource damage (NRD) cases, where impacts are felt across long time spans, potentially across multiple generations. This article addresses the choice of discount rate in estimating damages, which has implications for numerous other policy arenas as well.
NRDs are a hotly-contested source of potentially large environmental liabilities for unpermitted releases of oil or hazardous substances. One of the most important factors determining damages is the discount rate used to convert effects on natural resources occurring at different points in time to a common “present-value” equivalent. It is fair to say that many practitioners in this area don’t really understand the discount rate and feel that the less they have to think about it the better. The discount rate has been treated as a settled issue in NRDA, not really requiring much thought; that has now changed. Discounting is actively being discussed in cases in ways that lie at the interesting intersection of law, economics, policy, and philosophy.
Historical practice in NRD assessment (NRDA) over the past 25 years has been to use a 3% annual rate as a default value. A default is mostly welcome because you don’t need to think about it. But an incorrect default infuses significant errors into NRD settlements. How significant? Potentially huge, because discount rates act exponentially in time, compounding forward past effects to the present and discounting backward future effects. To illustrate, assume a constant 100% injury to 100 acres of a natural resource starts to accumulate in 1981 (CERCLA was enacted in December, 1980) and lasts until 2021. Assume further that restoration starts in 2022, provides an 100% increase in natural resource services per acre until 2140, and costs $100,000 per acre to construct. In round numbers, at a 1% discount rate NRDs equal $9 million, at 3% they equal $26 million, and at 7% they equal $150 million. Determining which of these is right surely warrants hard thinking, and this range of discount rates is now on the table.
The box in which 3% long resided has been opened: Dunford (2018) and Horsch et al. (2022) (hereafter Horsch) conclude that, if you start with the conceptual and empirical basis for the 3% and then update the estimate using more recent data, you get around 1%. That might appear to be an updated default. However, opening boxes can prove troublesome, and this one is no exception.
Horsch goes on to express concerns about the basis for 3% (now 1%) and propose a novel approach based on tort law and a theory of “coerced loans” (discussed below). Using this approach, you get around 7%. Horsch then opines that, since either of these approaches might be applied, continuing to use 3% provides a pragmatic compromise. But such a large spread of possibilities, with very different justifications, makes it unlikely that 3% will be tucked back into its box.
Since neither paper gets things entirely right, and neither is readily understandable to the non-specialist, this essay attempts to shed light on the issues using rigorous reasoning but a non-technical presentation.
The Discount Rate Demystified
Imagine you find that two of your last three eggs have gone bad when your larder is bare and you were counting on those eggs for a breakfast for three. This is a different event than if you find those two bad eggs when you have another dozen good eggs in the fridge. Suppose an egg fairy appears to compensate you for your two-egg loss by providing a benefit of B more eggs a day hence, after you have been to the store and do have a dozen eggs on hand. Each compensatory egg (received when flush with eggs) generates less incremental wellbeing than the wellbeing loss from those gone bad (inflicted when eggs were scarce). Consequently, the fairy understands that more than two compensatory eggs are needed to make you whole.
Moreover, the fairy saw you unhappily drumming your fingers on the countertop at the prospect of waiting for your compensatory eggs. Due to this impatience, you need more compensatory eggs in the future to replace those lost now because, from today’s perspective, enjoying eggs in the future is just not as highly valued as enjoying them today, irrespective of changes in the baseline number of eggs between the time of loss and when compensation is received.
The discount rate is defined by how much more (or less) you need to be given in the future than what was lost today to ensure no overall net loss in wellbeing. Specifically, if L is the amount lost and C is compensation provided one period hence, the one-period discount rate is defined by the equation L=C÷(1+d). If the fairy divines that three eggs will make you whole for your two-egg loss, then d=(3/2)-1, and your one-period discount rate for eggs is revealed to be 50%.
There are three important lessons here. First, the discount rate doesn’t necessarily have anything to do with financial markets – it is about trades in real stuff. Second, an individual’s discount rate includes (1) impatience to realize wellbeing, plus (2) the effect of changes in the baseline amount of the resource at issue. Third, the discount rate will change depending on what real stuff is being traded through time. If the injury had instead been a decrement in the stock of holiday leftovers after the guests have gone home, the steadily declining resource base implies that future compensatory increments are more highly valued than the incremental loss. In this case, less need be given in the future to compensate for the amount lost and this element of the discount rate is negative. The egg rate is different than the leftovers rate.
The 3% Discount Rate
As recounted by Dunford (2018), the 3% rate was specified in the 1990s using the following reasoning. Suppose that all effects are monetized, as they would be in benefit-cost analysis (BCA). The “thing” that is being traded through time, then, is not eggs or leftovers, nor anything obviously related to natural resources, but an aggregate consumption good, constructed using prices. At a macro level it is (roughly) equivalent to GDP. The rate at which the public is willing to trade consumption good across time – giving up some now for more later and therefore equivalent to the factor d
above – is called the social rate of time preference (SRTP), in this instance, a SRTP for consumption goods.
Now, if a social planner maximizes public wellbeing over time, the SRTP must equal the rate of return to investment in capital. This result is due to Frank Ramsey. A well-functioning market can mirror this outcome, with the SRTP equal to the observed rate of return on investment. The (consumption) SRTP is revealed by the magic of the market place – fairies need not apply.
NRD regulations endorse this measurement strategy. The CERCLA regulations (43 CFR § 11.84(e)(2)) refer to OMB guidance, which calls for a discount rate equal to the average rate of return on investments (which includes risk). Regulations promulgated pursuant to the Oil Pollution Act of 1990 state that a (lower) risk-free SRTP is appropriate for discounting in damage assessment (15 CFR § 990.53(d)(4)) – thereby avoiding double counting risk in both the discount rate and assessment of injury and restoration benefits. A guidance document NOAA (1999) used data on the average risk-free rate of return to investment from 1984 to 1999 to estimate a discount rate of 3%. The default rate was born.
Horsch reiterates this logic, updates the data for the period 1984 to 2021, and computes an average rate of about 1%, consistent with a 2018 recommendation from Dunford. Do we have a new default and can be done thinking about this for a while? Unfortunately, not quite. There are several considerations here, some of which pertain to Horsch’s discomfort with relying on market outcomes; let’s start with a vital one that doesn’t.
SRTPs for NRDA
The “stuff” traded through time in the Ramsey formulation is GDP. However, the preamble to the OPA regulations calls for a “…social rate of time preference for natural resources (i.e. the rate at which society is willing to substitute between current and future consumption of natural resources)….” (61 Federal Register, January 5, 1996, p. 453-454; emphasis added). Does the basis for the 3% rate in Ramsey even apply to NRDA thinking? The answer is both yes and no.
It is routine in NRDA to monetize impacts on recreation. A non-market valuation approach transforms impacts on recreation resources into consumption equivalents, exactly as in BCA. So yes, Ramsey thinking applies to recreation and the associated discount rate is Horsch’s 1%.
As to the “no” portion of the response, the monetization just described for recreation is not undertaken for ecological effects on individual resources (e.g. populations of fish or birds) or on the services provided by habitats (e.g. sediments or wetlands). For these resources, most NRDAs use service equivalency methods in which effects are denominated in bio-physical units, not aggregate consumption. The basis for the market-revealed 1% rate does not apply, even in principle. How to proceed?
Mr. Ramsey and the eggs and leftovers examples point the way. Setting aside impatience, if the resource is growing over time at baseline (the eggs case) the discount rate is positive, while if it is shrinking (the leftovers case), it is negative. Thus, in a NRDA for birds, the sign of the discount rate mirrors the time trend in the relevant population of birds. For habitats, the rate is positive when habitat quantity and/or quality is improving, or negative when these are degrading, due to climate change or other causes. The discount rate for ecological resources will be case and resource specific and among the many other determinations that need to be made.
Market Data and Public Trust Resources
Horsch expresses concerns about relying on market outcomes. They note the “counter-intuitive” potential for negative discount rates and allude to market imperfections that place a wedge between the Ramsey result and rates revealed by markets. Further, both Dunford and Horsch mention issues surrounding long time periods, which raise related considerations, but neither fully deal with these. Some unpacking is in order.
As we have seen, the negative discount issue is not an issue at all; it is a perfectly sensible result when the baseline amount of what you are trading is falling sufficiently fast. Market imperfections, however, are real, as they can embody various cognitive and behavioral slips, trips, and falls, in addition to problems of market power. Should a natural resource trustee, acting on behalf of the public, make decisions based on such outcomes?
The issue seems arguable when all the persons affected are alive and present to be counted. When recreating, people go where they go for various reasons. Trustees may reasonably respect individuals’ choices, whatever their foibles.
But when the effects in a NRDA span very long time periods, as they typically do, things are less clear. Discounting weights more heavily a unit of impact in the past than the same impact today. Impatience in discounting places a weight on persons according to when they are born. If future persons have more or fewer resources than those alive now due to a changing baseline, this is a reasonable basis for differentiating them when choosing compensation. But to additionally weight them according to birth date seems a form of discrimination that should not apply in public decisions. On this basis, impatience might well be dropped discounting in long-lived NRDAs, with only the baseline effect considered. Moreover, some state NRDAs go back much earlier than 1981; when affected persons are no longer alive, do they need to be compensated at all, let alone at a much higher rate?
Horsch’s concerns with the market-based measurements leads them to depart from SRTPs and propose a different approach based on a tort law concept of “coerced loans.” As this leaves behind methods endorsed by the NRD regulations cited above, the begged question is: does the coerced loan approach apply in NRDA?
The Coerced Loan Approach
The new approach is based on keeping the public’s monetary wealth unchanged, not their wellbeing, as in the SRTP. The logic espoused is that, since a monetary payment to address a past harm was not made, there is a “pre-judgement interest” issue interpreted as an involuntary loan made by plaintiffs to defendants. This time delay should be valued, it is argued, at the defendant’s cost of obtaining funds, estimated to be about 7%. Some difficulties arise when applying this idea in NRDA.
First, assume Horsch’s premise that there are past financial losses (the monetized values of the injury), with compensatory payments delayed to today. Pre-judgement interest applies. But symmetrically, future financial gains (monetized restoration benefits) have been pre-paid today. This is a loan from the defendant to the plaintiff, which should be compounded at the plaintiff’s return on savings, not discounted as proposed.
Second, trades in financial wealth are at odds with NRDA in which compensation must be paid in-kind, not in money. While value-based restoration scaling occurs with recreation, once restoration is determined, the device of money disappears and it is the service benefits of projects that provide compensation. The coerced financial loan doesn’t occur.
Third, the public’s loss from a release should not depend on the financial circumstances of the entity responsible for the harm. Under coerced loan theory, if the responsible party has a low cost of capital (the federal government, or a large firm) smaller damages would be identified (all else equal) than if the cost of capital is higher (a small firm, perhaps discriminated against in lending).
Coerced loan theory departs from NRDA’s focus on public wellbeing, assumes monetary payments never made, and has difficult empirical implications. Its relevance for discounting in NRDAs is questionable.
Discussion
Where does this leave us? For recreation assessments, the consumption-based SRTP, newly estimated as 1%, makes sense; it can be placed in a box labeled recreation.
For non-monetized ecological resources, previous empirical analyses are irrelevant. The Ramsey approach continues to hold (similar to the eggs and leftover examples), which aligns with the resource-based SRTP called for in the NRDA regulations. However, site and case specific determinations need to be made.
Finally, in NRDAs with effects spanning multiple generations, “birthist” impatience should be absent, and whether and how to include effects in the distant past needs very careful thought.
Dunford, R.W. 2018. “The Discount Rate for Assessing Intragenerational Natural Resource Damages.” Journal of Natural Resources Policy Research 8: 89–109.
Horsch, E., D. Phaneuf, C. Gihuere, J. Murray, C. Duff, and C. Kroninger. 2022. “ Discounting in Natural Resource Damage Assessment.” Journal of Benefit-Cost Analysis (2022):1-21. doi:10.1017/bca.2022.24
