Ecosystems, Ecological Restoration, and Economics: Does Habitat or Resource Equivalency Analysis Mean Other Economic Valuation Methods Are Not Needed?

Abstract

Coastal and other area resources such as tidal wetlands, seagrasses, coral reefs, wetlands, and other ecosystems are often harmed by environmental damage that might be inflicted by human actions, or could occur from natural hazards such as hurricanes. Society may wish to restore resources to offset the harm, or receive compensation if this is not possible, but faces difficult choices among potential compensation projects. The optimal amount of restoration efforts can be determined by non-market valuation methods, service-to-service, or resource-to-resource approaches such as habitat equivalency analysis (HEA). HEA scales injured resources and lost services on a one-to-one trade-off basis. Here, we present the main differences between the HEA approach and other non-market valuation approaches. Particular focus is on the role of the social discount rate, which appears in the HEA equation and underlies calculations of the present value of future damages. We argue that while HEA involves elements of economic analysis, the assumption of a one-to-one trade-off between lost and restored services sometimes does not hold, and then other non-market economic valuation approaches may help in restoration scaling or in damage determination.

Introduction

The need to address environmental damages with restoration or compensation relates to laws and regulations in several countries around the world. United States federal regulations such as the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) of 1980 create laws that enable trustees of natural resources in the US to hold a party legally responsible for restoring resources that are harmed. Similarly, the European Commission also passed several laws which obligate users of the natural environment to take responsibility for harm and damage (e.g., see the European Commission, Four groups of Directives1 on restoring environmental damages).

This article presents and compares two main environmental restoration or compensation approaches: traditional non-market valuation methods and habitat or resource equivalency analysis—as a service-to-service tool aimed at the determination of resource or habitat replacement. REA is a more general concept than habitat equivalency analysis (HEA), but hereafter in the article, we will use the abbreviation of HEA to mean both REA and HEA unless further clarification is needed. Service-to-service approaches are often applied to determine the amount of restoration or the appropriate compensation for losses caused by human activities (discharges of oil, releases of hazardous substances, or vessel groundings) or natural disasters in coastal areas. Habitats involved may include seagrasses, coral reefs, tidal wetlands, salmon streams, and estuarine soft-bottom sediments (NOAA22000). The primary goal of this manuscript is to consider the role that economics plays in determining whether and how much restoration might be done, but it is mainly intended for non-economists, as an audience.

Background on HEA and Non-market Valuation

There is a potential misconception among some (including economists aware of the methods) that HEA is devoid of economics and that it replaces traditional economic valuation approaches as a method of estimating losses for injuries. This article makes clear that HEA includes many elements that are important in economics, and that each can be potentially used in harmony along with other traditional non-market valuation approaches. Moreover, non-market economic valuation often complements the HEA method. Those intimately involved in the areas of analysis where compensation for losses is common (including natural resource damage assessment or NRDA—National Resource Damage Assessment) will probably not find this to be of any startling news, but this article provides some meaningful insights into the topic and raises at least a few new concerns. The concerns are raised in hopes that the models or assumptions can be further improved.

The importance of many fragile resources is clear, but we use coastal wetlands areas as an example throughout much of this paper, for good reason: coastal areas as of 2003 contained the majority of the United States population (53 %) (Crosett et al. 2004). These areas have been impacted adversely by many different events and activities; by 1980, wetlands in particular had declined about 53 % since the late 1700s (see Dahl 1990). In addition, environmental impacts to coastal areas are still much on the minds of United States Gulf Coast residents in the wake of the Deepwater Horizon offshore oil spill that occurred in the summer of 2010. The newest research suggests that a decline in infrastructure along the Gulf Coast adversely impacts the national ecological health and navigable waterways, leaving the Gulf region vulnerable to sea level rise and more severe weather events (America’s Wetland Foundation February Newsletter 2012). If costal restoration is delayed, the consequences can be quite severe. For example, the Entergy Corporation has estimated that Orleans and Jefferson parishes (Louisiana, USA) have incurred annual losses of USD 878 million caused by impacts from past storms, coastal erosion, subsistence, and land use expansion and development (The Times-Picayune 2012). However, without costal restoration efforts which are in place, the economic losses in this Gulf region would amount to USD 350 thousand million a year by 2030. We are also still reminded of the potential for serious risks of living in coastal areas in the wake of the magnitude 9.0 earthquake and resulting tsunami that hit Japan, in the spring of 2011.

Resources and habitats are also harmed naturally (e.g., coastal areas may be harmed by hurricanes or tropical storms), but still require society to consider the allocation of scarce resources to projects to restore them. Whether the harm is naturally occurring or induced by development or accidents like oil spills, questions arise as to how much restoration should be done to offset losses. A key question today is whether to spend scarce dollars on any one particular restoration project, out of many possible ones. One way to answer these questions is to estimate the losses in social benefits, in monetary terms, so that society knows how much compensation is needed to restore resources to a particular level. In the late 1980s and early 1990s, traditional non-market valuation methods were frequently applied in estimation of these losses in benefits, and this practice continues in some arenas today.

All non-market valuation methods essentially attempt to identify an individual’s maximum willingness to pay (WTP) or minimum willingness to accept (WTA) compensation for resource changes, in monetary terms. Social and total WTP aggregates the individual WTP across all impacted people. To the extent that society’s total WTP is an appropriate benefits measure and can be measured accurately, it tells us how much society values the loss. The presumption is that if this money can be spent to make society whole when harm has occurred, economics has been useful in making this happen, or that society has spent its money wisely to restore resources harmed by nature.

Some non-market valuation methods depend on observed behaviors that reveal preferences (RP will stand for a revealed preference approach), and others on stated preferences (SP). Revealed preference methods depend on uses of resources such as sun-bathing at the beach, fishing, hunting, hiking, swimming, and wildlife viewing. A simple underlying premise in revealed preference is: why would people use these resources unless they were of value to them? This premise is at the foundation for RP methods such as the travel cost method (TCM), where an individual’s cost in travel to and from a recreation destination is used to represent the “price” of a trip there.

SP methods are not dependent on use of the resource and, in theory, values that individuals place on resources are not dependent on use either, so SPs approaches are an important tool to use when passive use or non-use values are sought. Among the SP methods is the contingent valuation method (CVM),3 and, like all traditional market and non-market valuation methods, its use became part of the US federal regulations and has been approved by courts of law (see discussion in Department of the Interior, US 1986; Kopp and Smith 1989). Initially, these non-market valuation methods were the only economic methods used, at least when markets did not exist to provide information about relevant prices associated with the losses.

Today, HEA has largely replaced the sole use of traditional non-market valuation approaches alone (NOAA 2000), especially in the context of NRDA (Zafonte and Hampton 2007). This replacement is in fact tied to controversies that arose in application of non-market valuation methods in damage assessments such as the Exxon Valdez oil spill case; in particular, the oil industry and their lawyers and economists attacked the scientific merit of the CVM with a great deal of effort. Though Peter Diamond won the Nobel Prize in economics much after this for entirely different work, he was among economists working for Exxon, and was highly critical of the CVM (see Hausman 1993; Diamond and Hausman 1994). Even prior to this case, several economists, including those who developed it, had been critical of some features of the CVM, without entirely dismissing it as an important valuation tool (Bishop and Heberlein 1979; Loomis et al. 1990, unpubl.; Kahneman and Knetsch 1992).

Considerably after the Exxon-stimulated debates, NOAA commissioned another expensive economic study of damages that used the CVM in conjunction with estimating damages from DDT4 and PCBs5 coming from the Montrose Chemical Plant (in southern California) and other sources (see Carson et al. 1994), only to have a judge ultimately decide (in 2000) that the economic damage assessment was inadmissible in court. CVM is still widely used today, and thus it is fair to pronounce Diamond and Hausman (1994) wrong in their conclusion that CVM had failed. However, much more attention is paid to the scope test6 and other difficult tests of validity than it was in the past. So the attack on CVM actually led to improvements in the modeling (see Smith and Osborne 1996). Despite these improvements in the CVM, NOAA damage assessment staff and other economists working with them developed HEA (see NOAA 1995), at least in part to avoid the use of the controversial valuation methods.

HEA is designed to assess the amount of restoration efforts sufficient to compensate society for losses resulting from the damage. Note that restoration under CERCLA is usually a separate activity from remedial actions to be performed at a harmed site, such as one designated as a Superfund site. Even if remedial activities restore services at the site to 100 % of baseline levels (i.e., conditions had the impact not occurred), restoration under CERCLA is still required to compensate for the past and interim losses (Allen II et al. 2005).

HEA is also used today in several countries to determine the amount of compensatory mitigation that can make an otherwise unattractive-looking project move forward (see Molianen et al. 2009). Resource managers try to find appropriate offsets in the form of habitat or resources and this is a parallel to the use of non-market valuation methods in NRDA because prior to HEA, a project’s worth may have been decided on the basis of monetary calculation of costs and benefits. Under several Presidential Executive Orders (beginning with Reagan’s 12291 in 1981, then Clinton’s 12866 in 1993), as well as the principles and guidelines for analyzing water resource projects, a project was supposed to have positive, not negative, net benefits (benefits less costs), and one determined if this were true using benefit-cost analysis (BCA).

Today, while several regulatory procedures, including regulatory impact analysis (RIA) still involve BCA, HEA has become a prominent part of these procedures as well. Proponents of a project introduce proposals for compensatory mitigation to offset harm created by the project. The popularity of the HEA/REA approach in ex ante analysis of the future environmental damages is also confirmed by the fact that the US Army Corps of Engineers (USACE)7 uses the HEA/REA approach to identify the scope and scale of environmental mitigation. For example, Ray (2009) describes four proposed projects in which the HEA/REA approach plays a key role. Ray (2009) also emphasizes the importance of incorporating varying levels of uncertainty to assess alternative mitigation strategies [for more details, see discussion of the Craney Island (near Norfolk, Virginia) and Barber’s Point Harbor projects, Ray 2009].

HEA is also used today to scale the size and scope of restoration projects, relative to past injuries. The central idea of the HEA/REA alternatives is to initiate a habitat-based “service-to-service”/“resource-to-resource” approach. For example, if one parcel of wetlands is lost due to a development project, then adequate substitute wetlands might be sought that offer society equal services to the lost wetland’s services. Several pieces of legislation, such as the Clean Water Act (CWA), have been interpreted to mean that society may want to engage directly and indirectly in compensatory mitigation which restores ecological functions and balance of affected habitats. For example, wetland mitigation is consistent with goals of wetland protection that arise in the CWA (see the discussion of wetland mitigation banking by Kaplotwitz et al. 2008). As part of addressing remediation and restoration from mercury and polycyclic aromatic hydrocarbons contamination at its Superfund site, Alcoa transferred 729 acres of land to the US Fishery and Wildlife Service-managed Aransas National Wildlife Refuge (creating 70 acres of inter-tidal salt marsh), and also created 11 acres of new oyster reef habitat in Lavaca Bay.8 And, quite recently, the solar company Bright Source announced that it was mitigating potential impacts of locating solar collectors in the desert on the desert tortoise by purchasing 7000 acres of habitat and employing 40 biologists to manage protection, at a cost of USD 45 million (Hull 2011). Mitigation can be costly. The expenses related to a 3-week-long cleanup after the oil spill which happened near San Juan, Puerto Rico in 1994 were estimated to reach USD 1 million per day (The New York Times 1994; The Telegraph 1994). Figure 1 depicts the process of beach cleanup.

Fig. 1.

The process of beach cleanup after the oil spill which happened near San Juan, Puerto Rico in early 1990s (photo by: W. Douglass Shaw)

A hierarchy for mitigation was legislated and encompasses five crucial steps: (i) avoiding the impact, (ii) minimizing it, (iii) repairing or restoring the affected environment, (iv) reducing the impact over time by preservation or maintenance, and (v) compensating for the impact by replacing or providing the substitute resources or environments (see 40 Code of Federal Regulations 1508.20; Kiesecker et al. 2009). Whereas restoration refers to improving ecological health, compensation is perhaps best thought of as an economic term that refers to human actions to improve resource or service flows. HEA tries to meet the requirements of being a compensation approach (Kiesecker et al. 2009).

In the remainder of this article, we focus more on the details of HEA and how it is similar or different to these traditional economic valuation approaches. We also highlight problems with each approach, and the potential benefits of using them in conjunction with one another. We conclude that traditional valuation methods still need to play a role in restoration of resources such as wetlands in coastal areas, not only via its role in development of the REA/HEA approaches but also to complement them.

Resource/ Habitat Equivalency Analysis

The service-to-service approach dates back at least to King and Adler (1991) who tried to estimate appropriate compensation ratios for wetland mitigation. Unsworth and Bishop (1994) were the first ones we know of to lay out the theoretical economic principles of the HEA method. NOAA also provided an overview and illustration with hypothetical examples (NOAA 1995). The HEA approach essentially is conducted with the aim that the value of habitat services gained with appropriate compensatory restoration equals the value of the lost services prior to resource injury. In fact a simplifying assumption underlying the standard REA/HEA equation is that this equivalence holds. Compensatory restoration actions are then “scaled” to accomplish this balance. To determine the nature of the loss, ecologists try to assess the services that are provided by a resource that is identical to the injured one, in the absence of the injury. The resource equivalency problem can be examined via the solution to an equation. Following Jones and Pease (1997) or Strange et al. (2002), the HEA or REA equation can be written as:

1

where L_t is lost services at time t, R_s is replacement services at time s, t₀ indicates the time that services are first lost, and t_l indicates the last time when lost services cease (the last time when society suffers from lost services); similarly, the s terms indicate the first and last time that replacement services are provided, p is the present time when the claim is presented, and ρ is the discount rate. Figure 2 illustrates the recovery of services provided by the injured habitat (NOAA 1995).

Fig. 2.

Resource service levels at the injury site (NOAA 1995)

The discount rate is a rate that is used to weight future resources (or money or income) relative to present ones. When future resources are discounted, we say we are analyzing their “present value.” Note that the discount rate appears on both sides of Eq. (1).

Penn and Tomasi (2002) and Zafonte and Hampton (2007) each provide similar versions of Eq. (1). There is no indication in the HEA equation above that the lost services or replacement services are to be actually valued monetarily, so obviously one would want to have the type of services lost and replaced be identical to keep the HEA equation simple. Simplification means that the equation is essentially devoid of the economic values that people place on past and future resources. Note that most often, injury occurs in the past, but damages tend to accrue in the near future, while restoration gains accrue solely in the future (often far into future). That gives rise to the pursuit of restoration activities in the first place, while replacement is at best, at present. In practice, restoration activities are more likely to be conducted exclusively in the future because of delays in interested parties coming to terms on what is to be done.

In their example, Strange et al. (2002) consider the losses to a salt marsh from an oil spill, calculating the loss as 500 acre years of services. Strange et al. (2002) calculate the needed number of acres of habitat for compensation in present value terms, but do this for each of several scaling metrics. Changes in these metrics are assumed to reflect proportional changes in ecological functions and quality of habitat and, thus, changes in public value and human welfare (English et al. 2009). For example, first they consider the primary production of above-ground vegetation, then soil development (soil nitrogen), making several different assumptions which lead to quite different numbers of acres of habitat for compensation. Some metrics lead to suggestions of full recovery within a relatively short amount of time, while others do not. The fact that different metrics give rise to different restoration conclusions underscores the difficulty in measuring the impact of environmental change on human well-being in general. That hindrance can be also encountered while using other techniques for scaling environmental restoration.

There exist alternative methods for assessment of a scale, scope, and costs of compensatory mitigation. One of them is Florida’s Uniform Mitigation Assessment Methodology (UMAM), which not only takes into account the magnitude of the restoration effort (in terms of number of acre-years (or hectare-years) of habitat services) but also pays attention to quality of habitat services. UMAM is parallel to economic studies because variation in the quality of habitats should be a key area of economic research. Note that ecologists play a considerable role in the process of quality equivalency determination [see Bardi et al. 2011 or a description of UMAM at the Florida’s Department of Environmental Protection website (FDEP 2011)].

Because the HEA equation is set in present value terms, then more acres of habitat are required when longer lags are inherent in either the recovery process (debit side) or the provision of compensatory gains (credit side). That means it takes time for a restored habitat to get back to ecological balance, thus service flows from it may not begin accruing until the future.

Zafonte and Hampton (2007) directly indicate that “values” underlying the HEA or REA approaches are potentially economic ones, and are still part of an initial formulation of the HEA or REA equation, along with the discount rate. However, the former drop out of the equation altogether as long as the injured and restored values of resources are of the same type and quality. This leaves the only “non-biological” factor as the discount rate itself. As Strange et al. (2002) conclude (p. 297), “…conclusions about equivalency will depend critically on the data and assumptions used to implement scaling methods.” Next, we summarize some concerns about the HEA, highlighting some assumptions underlying this approach.

Some General Concerns About the REA/HEA

Flores and Thacher (2002) raised some general concerns about HEA approaches over 10 years ago, focusing on two major issues: heterogeneous preferences (people are different, and different people like different things) and changing values over time (values for a resource in 20 years may be quite different than today, particularly with increasing scarcity over time). The overarching major concern that economists would have pertains to whether the past and future values cancel each other out in the HEA equation because of similarity between the two. We would note that the modern trend in econometric (statistical) modeling very often finds heterogeneity in preferences across individuals, so Flores and Thacher’s concern has been increasingly supported.

Penn and Tomasi (2002) state that if past and future values do not drop out because of dissimilarities, an “HEA-like model can still be applied as long as the scale of restoration is adjusted to account for the value difference.” (p. 693). However, suppose the differences do stem from different values over time. Knowing how society’s real values for resources change over time is quite difficult. Preferences can change for any number of reasons, including changes in resource scarcity, real income, health, or knowledge of individuals who value the resources, etc. Determining all of these would require that economists have carefully assessed the real change in a set of individuals’ resource values over time for a credible and representative sample of the population, controlling for all possible factors that influence values so that preference heterogeneity over time can be isolated from all other causes that might make it appear that values are different from one date to another. Doing this would likely involve a longitudinal study of the group of people with the same preferences to avoid preference heterogeneity across people and consideration of the same resource over time. Otherwise, differences in people would confound the ability to see if preferences really have changed over time and if so, why.

We know of no carefully done longitudinal study in the literature, although there are many studies of the temporal “reliability” of valuation estimates in the literature (see a summary in McConnell et al. 1998; Loureiro et al. 2010). Most existing studies involve relatively short lapses in the amount of time between data collection: as an example, Carson et al. (1997) consider contingent valuation estimates related to the Exxon Valdez oil spill, where the estimates were obtained 2 years apart, and conclude that there is stability in values over this period. In contrast, Loureiro et al. (2010) find a significant drop in WTP estimates related to the Prestige oil spill in Spain (in 2002) over a 4-year period, though it is difficult to discern what exactly causes this decline.

Dunford et al. (2004) demonstrated that HEA results are quite sensitive to a variety of factors, including price change scenarios, and Zafonte and Hampton (2007) describe three situations where the values on the left- and right-hand sides (i.e., the injured and restored resource values, respectively) do not cancel each other out, thus making elimination of monetary values suspect: (i) the resources are different, (ii) per unit value of injured and restored resources changes over time, and (iii) heterogeneous preferences over natural resources of interest violate the criterion that REA should satisfy, which is that the sum of all individuals’ WTA compensation for injury is zero, including the cost of the correctly scaled restoration. The idea of the last criterion is simply that restoration costs and benefits are balanced for economic efficiency. The central theme is that the gainers compensate the losers (the Hicks or Kaldor criteria for compensating wealth which were discussed by Jones and Pease 1997, among many, many others with a focus on welfare economics).

Zafonte and Hampton (2007) also simulate different outcomes in a REA model to explore how it does when price changes and heterogeneity in preferences are introduced. Doing this requires development of a modified REA model that does include monetary values, as well as several assumptions, including a functional form for a demand function that yields the monetary benefit flows from the restoration project and implies various elasticities (i.e., sensitivity to price or other changes).9 They compare the modified to the basic REA and find four cases where the two models diverge by more than 17 %, which they conclude indicates small differences between the two. However, the support they find for the basic REA approach is tied to the assumption that restored resource’s services are close substitutes for injured ones. They also find that in cases where locals, or those in close proximity to a resource, have much higher values than others, the conventional REA under-compensates the locals as compared to a monetized version of the REA.

Strange et al. (2002) are to be applauded for carefully expressing the framework in a model that is cognizant of the role that discounting plays and, because of their assumptions. Zafonte and Hampton (2007) make this even clearer. The latter do consistently point out that the most relevant economic and biological models connected to the REA framework are dynamic and incorporate the dimension of time, but do not carefully consider any issues related to the discount rate, and therefore, our concerns about it are raised below.

Dynamic Considerations, Uncertainty, and the Discount Rate

In sophisticated economic analysis involving obtaining goods and services and earning income now and in the future, models including discount rates are developed that consider how to maximize utility over the whole life of the individual. Essential in such models is the opportunity cost of consuming something now, in terms of foregone consumption tomorrow and further into the future, or vice versa. As shown in several of the studies of the HEA framework by the above-mentioned economists, this lifetime utility maximization problem is fundamentally connected to what the HEA restoration calculations try to achieve for restoration or at least compensation for losses. Thanks to the inclusion of the discount rate, sophisticated dynamic economic models not only tell us what is consumed, they tell us when to consume, given a host of assumptions about the individuals making decisions, and the resources involved. This timing itself affects the value of goods/services (as does the scarcity and available substitutes). For example, the “optimal” timing of extraction of resources flows from such models. It is worth mentioning that discount rate is an anthropocentric concept and it assumes that as we humans are the decision makers of today, we decide what should be consumed today and what should not. These decisions indirectly attach present and future values to goods and services.

Complex dynamic models often allow for risk (known probabilities) or uncertainty (unknown probabilities) because the future is inherently unknown or risky (e.g., Woodward and Shaw 2008). In economics, the classic model that allows for risk weights an individual’s utility with the probabilities of outcomes, and is called the expected utility model (EUM). When doing a primary valuation exercise, careful thought has to be given to whether the individual is facing risk or uncertainty as she makes her decision regarding WTP. For example, if one is interested in determining WTP to prevent a future oil spill like the Deepwater Horizon from happening again, it is important to consider whether this event will happen with certainty, or with some element of uncertainty. Thus, both valuation and other issues relating economics to HEA, as well as future restoration outcomes, might best be done in the context of the EUM or one of its variants (Shaw and Woodward 2008).

In contrast, the HEA balance in its most simple form (when lost past values and future ones are assumed equal) can be accomplished in any number of time profiles. A simple spreadsheet analysis can be used to show this: replacement acreage or service can come in clumps or chunks in the future, or can be spread out over time, and in either case the equation conditions (left side equals right side) can be met. This approach tells us nothing about the “optimal” timing of replacement from society’s perspective. For example, suppose the present (the p in the HEA equation) analysis is in the year 2010 and we consider an analysis up to the year 2017. If one assumes that losses occur starting in the year 2000 for 3 years, and these losses equal 10, 10, and 5, then a balance can occur by replacing acreage or service with units of 6.24 in each year starting in 2012, presuming a 3 % discount rate. (More is offered on the discount rate, and what it “should” be, below.) The equal chunks of provided service of 6.24 allow society to spread out replacement over these final 6 years in the profile. However, it is quite easy to reallocate future replacement in any number of other ways to accomplish the same balance, more or less. For example, one gets approximately the same balance by offering replacement units of 34.82 all in the year 2012, and none thereafter. Exact calculations are presented in Table 1.

Table 1.

Losses and replacement balance in the HEA analysis using constant discount rate 3 %

Current period 2010
Losses				Replacement units spread over next 6 years
Year	L	(1.03)(p − t)	Discounted value	Year	R	(1.03)(p − s)	Discounted value
2000	10	1.34	13.44	2012	6.24	0.94	5.88
2001	10	1.30	13.05	2013	6.24	0.92	5.71
2002	5	1.27	6.33	2014	6.24	0.89	5.54
				2015	6.24	0.86	5.38
				2016	6.24	0.84	5.23
				2017	6.24	0.81	5.07
		Sum	32.82			Sum	32.82
				Replacement units all in the year 2012
				2012	34.82	0.94	32.82
						Sum	32.82

A specific oil spill example, borrowed from NOAA (1999), is provided below to more easily present the details of discounting. The oil spill occurs in 1997 and injures 50 acres of inter-tidal wetland. Assumptions are that 100 % of wetland services are lost initially, and recovery does not start until 1999, and follows linearly for the next 5 years. Presuming that the loss calculations are done in the year 1998 (i.e., 1998 is the “present” in the calculation), the stream of losses and restoration are in Table 2. The total present discounted value of lost acre years, or the sum of the far right-hand column, is 195.80. This shows several things. First, that even though the initial loss is only 50 acres, the losses over time add up to more than that because each acre offers an annual service flow. Second, the 50 acres actually are compounded10 to be more than 50 acres for the “past” loss, in 1997, from the standpoint of the present, which in the example is 1998. This of course is because the discount factor for the past (1997) is actually greater, not less, than one. Third, the sooner that recovery can be accomplished, the less the discount factor adversely figures into the calculation; by 2002 in the example, discounted acre-years have shrunk to 8.88.

Table 2.

Discounted interim service losses (from NOAA 1999)

Years	% of service losses at start of period	% of service losses at end of period	Raw acre years of service losses	Discount factor	Discounted acre years of service losses
1997	0	100	50	1.03	51.5
1998	100	100	50	1	50
1999	100	80	40	0.97	38.83
2000	80	60	30	0.94	28.28
2001	60	40	20	0.92	18.3
2002	40	20	10	0.89	8.88
2003	20	0	0	0.86	0
2004	0	0	0	0.84	0
2005	0	0	0	0.81	0
2006	0	0	0	0.79	0
2007	0	0	0	0.77	0

Here emerges a key area where other areas of economics can be used along with the HEA approach. A decision on the level of discount rate used in the HEA model is to assess how society (or a representative sample of it) is willing to trade-off present and future consumption of resources not only in its lifetime, but also between generations. As will be seen, studies that estimate individuals’ discount rates often find they differ a great deal from person to person. Therefore, issues connected to setting an appropriate discount rate for use in the analysis are key components.

The Discount Rate

Because of its importance in scaling natural resource services and determining resource restoration scaling, this section is devoted to clarifying the role that the discount rate plays in conventional economic analysis, as well as in the HEA process. To begin, a standard and most common operating procedure in the United States is to use a social discount rate of 3 %, at least for some project evaluations (for recommendations regarding discount rate in the United States see NOAA 1999; Ray 2008). In several European countries, the most prevalent discount rate ranges from 3 to 6 %. High discount rates (10–12 %) are applied to the projects in developing countries, for example, by the World Bank. An ecologist might ask: why is any particular numerical rate used, and what does it mean?

The social discount rate, as mentioned briefly above, represents society’s rate at which it is willing to trade future for present consumption or enjoyment of resources, money, income, or any good or service that can be considered and valued. A positive, non-zero discount rate above zero implies that society values the present more than it does the future. In resource terms, this positive, non-zero rate means that society would rather have something today, such as a resource like a coastal wetland, than it would tomorrow. It does not mean there is no value to society of the wetland “tomorrow,” rather, it merely means that in the calculus of the dynamic utility maximization over lifetime using discount rate, people prefer the present enjoyment more than postponing that until the future. There are host of reasons for the practice of discounting in economics and, in theory, the discount rate can be determined by its components, based on what is called the Ramsey formula (discussed below).

First, some intuition and logic. We humans are impatient. We want our homes now, not in 40 years, after we might have saved enough money to buy them outright. Second, we might not be here tomorrow, i.e., aging and mortality are factors in why we might prefer having something today. Third, we might care a good deal about others in the future when we are gone, but we might also expect growth and technology to make them better off than we are today. Finally, some believe that a fundamental issue is that societies that live one hundred years from now are not here at the present, to represent themselves in decisions, and this should be a factor for how discount rates are chosen.

The usual assumption economists have made is that a discount rate is constant, no matter how long a period of time is evaluated in a tradeoff. This is, however, rarely found to be true in assessing and estimating individual discount rates. A very common pattern is that they decline, the longer the time horizon (see Frederick et al. 2002; Grijalva et al. 2011; and other references to “hyperbolic” discounting found in each of these articles). This suggests that impatience is strongest when obtaining the good is possible most closely to the present, but the impatience fades once we know we cannot have something for quite some time into the future.

Weitzman (2001) conducted a study, asking 2160 economists from 48 countries to state a preferred social discount rate and found a diverse range of answers, but 97.8 % of these economists agreed that a social discount rate should be greater than zero, and it should reflect a positive time preference. However, the main finding of Weitzman’s study (2001) is that even if every individual believes in a constant discount rate, a wide spread of estimates on what it should be makes the effective social discount rate decline significantly over time (so called gamma discounting). Therefore, one may conclude that valuation of resources in restoration projects that mitigate long-term environmental damages might require a discount rate closer to zero, whereas values placed on resources in short-term projects should be calculated with the use of higher discount rates. Naturally, concerns that longer time horizons may involve lower discount rates than shorter time horizons do would carry over to traditional BCA, as well as to HEA.

We keep in mind that it is knowledge of a “social” discount rate that matters, and that begs the question of how to know what this is. In the United States, an early effort to determine the appropriate level of social discount rate was in the year 1981, when the Office of Management and Budget (OMB) issued discounting guidelines for federal agencies. At that time OMB asked that a discount rate of 10 % be used for evaluating federal environmental actions (Kolb and Scheraga 1990). The federal government then used observable transactions involving investment and savings to determine what might best reflect the social discount rate, and this or other rates are recommend for use in analysis of all projects that involve federal government expenditures or funding at some level. This idea relates to the “opportunity cost” of society’s money, and OMB still does this today, although today’s US rate is much lower than in 1981, at about 3 %. Note that in 2011 and early in 2012, many annual rates of return on US Treasury securities are quite close to zero. Other rates of return may be relevant. For example, the current average 30-year mortgage rate provides a clue to society’s average willingness to obtain something now rather than later, and this is about 4 %.

Recommendations of discounting procedures related to restoration are in fact provided under the oil pollution act regulations (see NOAA 1999). The agency recommended that for restoration scaling (involving interim service losses and restoration gains), the discount rate be tied to the consumer rate of time preference. However, for restoration costs, NOAA (1999) recommended that trustees tie discount rates to annual rates of return on US Treasury securities in conjunction with the period of analysis, which could be different than other implied rates of consumer time preference.

As indicated above, the NOAA (1999) memorandum and most mention of discount rates focus exclusively on the rate of time preference as the factor that determines what it should be. However, the original Ramsey formulation of the discount rate includes other factors in addition to the pure rate of time preference (Ramsey 1928). An approximation to the Ramsey formula for the social discount rate can be represented by the following equation:

where r is the social discount rate, δ is the pure time preference rate (also called the utility discount rate, which can be interpreted as an index which measures society’s impatience), η is the elasticity of marginal utility of consumption, which can be interpreted as a curvature of the utility function, and G is the growth rate of consumption per capita.

The second two terms on the right-hand side above are thus the product of the growth rate in consumption and the elasticity of the marginal utility of consumption (see for example, Conceição et al. 2007; or Sterner and Persson 2008 for more details on the standard presentation of the approximate Ramsey formula). Suppose that η is equal to one (this implies a logarithmic relationship between utility and monetary income) and that the growth rate in income is high. This implies a higher discount rate than by ignoring the multiplicative term altogether. The idea is that future generations will be better off than our current one because of their increase in income over our own, so from the present point of view it makes less sense to postpone consumption today. A high elasticity of the marginal utility of money (greater than one) means that society cares less about an additional dollar of income as people get richer. It is generally thought by most economists that rich people value an additional dollar less than poor people, at least at some level of income. Of course, it is not impossible that people will be poorer in the future than they are today, at least in some parts of the globe. But in any case, ignoring the second part of the approximation to the Ramsey formula for determining the discount rate has to be defended, and reasons why it should not be ignored are clearly presented in Sterner and Persson (2008).

Today’s US 3 % rate of discount for project analysis is relatively low, but obviously still has strong implications for the value of something 50 years from now, as compared to a value obtained when a social discount rate of zero is applied. The latter implies that the value of something 50 years from now is the same as it is today: USD 100 valued 50 years from now is still worth USD 100 at a zero discount rate, but only worth about USD 23 at a 3 % discount rate.

In the European Union, the discount rate used in the appraisal of restoration projects ranges from 3 to 6 %. Before the year 2003, discount rates ranging from 3 % in Germany to 8 % in France could be observed (Evans 2006). In 2005, the European Commission published the document (EC 2005) which recommended a 4 % discount rate, but shortly after, a review recommended a 5 % rate (see Rambaud and Torrecillas 2006).11 Other countries in Europe have favored discount rates lower than 5 %: the UK proposes 3.5 % as a social discount rate in The Green Book.12 Spain, in turn, uses 4 % discount rate (Reira 2007). In Sweden, two agencies unanimously suggested that the appropriate discount rate is 4 %, and one of them recommends conducting sensitivity analysis with the 2 % social discount rate (see SIKA 2002; Swedish Environmental Protection Agency 2003).

Still higher discount rates are more common for World Bank projects. According to Lopez (2008), discount rates for projects involving cost-benefit analysis in nine Latin American countries are in the range from 5 to 7 %. Others suggest that the World Bank and other multilateral development banks such as the Asian Development Bank have used a standard real discount rate of 10–12 % to evaluate projects in all sectors and all countries for decades (Belli et al. 1998; Oppermann 2011). This higher rate may reflect the sense of riskiness in the realization of project benefits in developing countries. Risk and its connection to the discount rate are discussed at greater length below.

One gets the sense today that environmentalists, and perhaps some ecologists, think the 3 % discount rate used by the US government agencies to evaluate environmentally related projects is too high, particularly when it comes to debates surrounding climate change. However, one should consider again the fact that both sides of the REA/HEA equation contain the same discount rate. A high discount rate in fact implies that past losses are compounded, or worth more in the present, than they would be with a low discount rate. It is slightly misleading to bring in a discussion of inflation here because discounting and inflation are two different concepts, but the logic that USD 1.00 in 1920 has to be greatly inflated today to provide the same spending power as it had back then illustrates the idea of compounding past losses. Thus, neither a high nor low discount rate is inherently “bad” when it comes to the HEA analysis.

Private Rates of Discount and Heterogeneity

The above discusses the social discount rate as if there were indeed one such rate that everyone should be comfortable with. However, as briefly noted above, different individuals can in fact have different time preferences, and hence, different rates of discount or private discount rates (Frederick et al. 2002). This has been shown in a variety of studies that attempt to estimate private rates of time preference or implied discount rates. These studies accomplish this primarily by relying on tradeoffs that people are asked to make in laboratory experiments, i.e., asking would you rather have USD 1 today or USD 1.50 tomorrow (see the literature review by Frederick et al. 2002). Discount rates that private individuals have might well depend on their preferences and understanding of risks, which also may vary across individuals and change over time. After all, the future is unknown, so it is inherently “risky,” and that may be reflected in one’s discount rate. But individuals often face different real levels of risk because of potential endogeneity13 embedded in that risk, or at least they perceive of risks differently.

Quite recently, Grijalva et al. (2011) found considerable heterogeneity in discount rates as part of their study of environmental tradeoffs and uncertainty. Some people are more or less patient than others. In addition, few studies estimates implied discount rates for any good or service except money, and there is no reason we know of to rule out the possibility that the rates of discount for natural resources might be different than the rates of discount for money or income. The NOAA (1999) memorandum, in fact, contains a footnote (see footnote 12) suggesting that there may be more appropriate rates of discount for use in natural resource management, which could be gleaned using a SP survey.

The concerns above are additional reasons to be considered about application of the HEA analysis when using one simple “social” discount rate, but, of course, this criticism applies to any analysis that would do so, including traditional BCA. Recently, Molianen et al. (2009) greatly expand on the conventional framework for resource compensation, which is strongly related to HEA models, as noted above. They use an “info gap” model that allows for uncertainty, as well as time discounting, where both hopefully contribute toward finding better compensatory mitigation ratios. But economic analysis can be used to find what individuals affected by resource injuries have as discount rates and why they differ, if they do. This knowledge might be used to adjust or scale discount rates used in determining the magnitude of restoration or compensatory mitigation activities. It is important to detect private discount rate heterogeneity because it means that people place different values on lost and restored services. As a result, substitutability between resources in different time periods might be impaired (Zafonte and Hampton 2007).

Risk and Uncertainty in Assessing Recovery Rates Used in HEA

NOAA (1999) and many other agencies have recognized that restoration may involve risk or uncertainty (in the 1999 NOAA memorandum uncertainty and risk are words that are used synonymously, with no differentiation between the two, as noted in their footnote number 8), and that there are at least two ways to introduce uncertainty. The first approach rigorously incorporates uncertainty directly into the calculation of benefits and costs. This is rarely done. The other lets the discount rate be different when facing uncertainty, than when one faces a certain stream of future benefits or costs being evaluated. A standard practice would be to increase the discount rate under conditions of risk, as compared to when there is none, presuming that most agents in society are risk averse. Risk averse agents require a premium to cover their disutility associated with taking a gamble.

Conditions necessary for restoration of habitats and success rates in recovery vary greatly, suggesting the need to incorporate uncertainty in the HEA approach (see Dernie et al. 2003; Lirman and Miller 2003; Wilcox et al. 2006). In the case of reversible environmental damages, it has been suggested that the potential success of ecological restoration “should be demonstrable within 10–50 years” (Jackson et al. 1995). This time frame might be subject to modifications, depending on severity of environmental damages and the overall longevity of the project.

Such uncertainties that accompany the HEA approach are shown by Cole and Kriström (2008) in their case study of a chemical tank release in Helsingborg, Sweden. They conducted a sensitivity analysis in which they analyzed various scenarios, each with different growth rates of the baseline habitat (before damage occurred), adjusting the recovery rate which in turn affects the time horizon over which restoration services should be provided. Similarly, Molowny-Horas et al. (2008) pay close attention to the uncertainties regarding the rate of recovery of a damaged habitat and the rate of growth of a replacement habitat in their study of compensatory mitigation of forest wildfires in Catalonia. They simulate the recruitment of Black pine trees in unburned forests to determine the growth and mortality rates necessary to estimate recovery rates. Next they simulate conditions necessary to specify the probabilities, scale, and location of future forest fires which potentially can increase mortality rates.

Other limitations of ongoing environmental restoration should also be considered as significant sources of uncertainty in restoration success. High costs of suggested restoration projects, lack of social commitment, erroneous judgments and mistakes in valuation, and other ecological circumstances are among the most serious constraints encountered (see Jackson et al. 1995), and these should play an important role in the HEA analysis.

Using Both HEA and Economic Valuation Methods

HEA came into vogue and is probably here to stay. Yet, it would be naive to conclude that either primary or secondary economic valuation now plays no role in determination of compensation for natural injury cases. It does.

Primary environmental valuation methods include all methods which require collection of data essential for the valuation task (e.g., surveys used for the CVM or the TCM). Secondary environmental valuation methods encompass reference to previously completed primary studies (e.g., benefit transfer which is supported by results from studies done in another location or context). Table 3 provides a summary of some additional features of conventional non-market valuation methods. The CVM and TCM are mentioned above. The stated choice experiment modeling (SCM) approach is, like the CVM, an SP approach, but it presents individuals with two or more alternatives that are described by attributes, allowing them to choose between these to show their preference. The hedonic property valuation or pricing method uses the observed value of a property (a home or land) and uses statistical methods to sort out the contribution to that value of an environmental amenity.

Table 3.

Summary of features of non-market valuation methods

Valuation	Value involved	Possible application to HEA
SPs methods
CVM	Use values and non-use values	Can be applied as a method of primary research to determine monetary values
SCM	Use values and non-use values	Especially suitable for evaluation of preference for restoration alternatives
Revealed preference methods
Travel cost	Use values only	Can be applied to estimate a change in frequency of visits or a change in value of visits lost after damage happens
Hedonic pricing method (HPM)	Use values only	Estimates from previous hedonic studies can be transferred to another case study area (benefits transfer)
Market prices	Use-values only	Can be applied as a proxy when other valuation methods cannot be used, but when market prices are available

Source Authors’ adaptation based on Annex 6 to the Toolkit for Performing Resource Equivalency Analysis to Assess and Scale Environmental Damage in the European Union (2008)

HEA/REA and non-market valuation methods are good complements. Whereas non-market valuation methods estimate the monetary value of lost and restored services or resources, the HEA/REA approach deals with service-to-service comparisons. A nice marriage of the two approaches, HEA/REA and primary or secondary economic valuation, emerges in the context of NRDA. As an example, the Rhode Island North Cape Oil spill NRDA case involved both the use of HEA to calculate compensatory mitigation and restoration activities for ecological losses, and a benefits transfer approach to value lost benefits from recreational fishing. Furthermore, the European Union’s four groups of Directives (see Footnote 1) on restoring environmental damages also recommend application of habitat services/resources equivalency methods that hinge on non-market valuation methods. Therefore, one could conclude that regular non-market valuation methods constitute a good augmentation of the HEA/REA approach. In the next section of the article, we present some details on recent applications of the HEA in the European Union.

HEA/REA Application in the European Union

While US HEA and related legislation may focus on situations involving certainty, the EU environmental legal system emphasizes uncertainty and the precautionary principle, and how these relate to benefits and costs of environmental impacts (Christoforou 2004). A recent study, in fact, suggests that the HEA/REA approach is not the most popular way of scaling ecological restoration in the EU (Cox 2007), but it is still recommended as a compensatory mitigation framework [see the EU Birds, Habitats, and Environmental Impact Assessment (EIA) Directives].14 Indeed, there have been several cases in which the HEA/REA approach was applied as a method of complying with the EU Directives.

The chemical spill in Helsingborg, Sweden, was mentioned above and provides a good opportunity to consider some details. This spill occurred on Feb 4, 2005. The Kemira Group’s chemical tank sank and discharged 16300 tons of 96 % sulfuric acid into Kopparverkshamnen, connected to the Baltic Sea. The release of the acid had a detrimental effect on fish, benthic organisms,15 sea plants, and sediments. It was estimated that environmental damage encompassed around 29 acres (12 ha) and reached a depth of 10 m (Cole and Kriström 2008). A restoration effort was undertaken using the HEA approach. According to Cole (2008), as much as 33 discounted hectare-years of habitat services were lost due to the acid spill. Although the restoration project cost €100 000, the equivalency of lost and restored services was not achieved: only 1 discounted hectare-year of sea grass habitat services were provided (Cole 2008).

Another European REA project involved the construction of the Yamal Western Europe gas pipeline and its impact on the habitats around Vistula River in central Poland. Tederko et al. (2008) performed an REA for impacts between 1998 and 2000. They distinguished between primary damages related directly to areas where excavation happened and secondary damages in the terrestrial and aquatic zones which included riparian forest and shallow water habitats, respectively. The REA approach helped researchers in “normalizing damages” and selection of “compensatory remediation projects” (Tederko et al. 2008). It was estimated that 65 discounted service hectare-years were lost in terrestrial habitats and 123 service hectare-years were forfeited in aquatic habitats as a consequence of the gas pipeline expansion. Planned restoration efforts are supposed to restore 40 ha-years for terrestrial habitats and 30 ha-years for aquatic habitats, at a total costs of €170 000–200 000.

The REA approach and the CVM were combined into one compensatory mitigation task in the case study of forest fires in Spain’s Bages-Bergiedá region (Molowny-Horas et al. 2008). In July of 1994, a severe wildfire occurred in the forest, caused by a malfunctioning power line. Approximately 25 000 ha of the European Black pine were burned, creating long-term damage. The habitat was under protection of the EU Habitats Directive. The REA approach was used to estimate the loss of timber and other forest resources, while the CVM was used to estimate the loss in amenity, existence, and recreational site monetary values. Geographical Information System spatial analysis determined the value of resources to be provided by the proposed reforestation projects. Results suggested that to achieve a socially desirable scale of restoration the area of new forest planted needed to be 33 % bigger than the original size of the damaged terrestrial habitat. Findings from the REA analysis also noted that afforestation would render no credit until almost 40 years after planting (Molowny-Horas et al. 2008).

These European examples presented above show how the HEA/REA approach can be applied in the process of complying with remediation requirements of the Habitats Directive and Environmental Impact Assessment Directive (EIA Directive). Apart from the three European cases presented above, there are at least 20 different cases of the HEA/REA applications in countries all around the Europe (e.g., projects in Czech Republic, Germany, UK, Sweden, Poland, and Spain, among others). The HEA/REA approach is increasingly important and is becoming a popular method of ecological restoration, especially in the light of the EU Directives.

Conclusion

The findings presented above provide evidence for the argument that the HEA/REA approach accompanied by other primary and secondary valuation methods can be used to determine whether past lost resources and future restored ones are similar enough to allow the assumption of perfect substitution to be made. As shown in Breffle and Rowe (2002), economic analysis can actually inform ecologists of social desire for certain tradeoffs, supporting or perhaps rejecting the notion of equivalency of resources that may differ. A better understanding of ecological relationships for economists comes as well from study of issues surrounding the HEA/REA. The HEA/REA can also be used as a tool for ex ante analysis (before injury happens: see Roach and Wade 2006; Ray 2009, for example), and may be integrated into standard BCA done in this setting.

In conclusion, we believe that the HEA/REA approach and regular economic valuation methods complement each other and the latter should not be omitted in the scaling of compensatory mitigation and restoration efforts. They each play a significant role in compensatory mitigation and restoration of damaged habitats and lost services. In either case, we first see the need for more careful consideration of the role that the discount rate and discounting play. We also believe that more careful attention needs to be given to dealing with risk and/or uncertainty.

The case studies show that restoration projects are extremely challenging. Restoration results in the above studies were supported with “habitat scalars”—a construct necessary to normalize possible compensatory mitigation alternatives to a single, preferred habitat type. These “habitat scalars” play a key role in scaling of the restoration efforts. Local biologists and ecologists work together on the most appropriate determination of such “habitat scalars.” Their knowledge of the local habitats contributes positively to the restoration process by ensuring that the project provides the highest possible level of habitat productivity and diversity.

The role of ecologists and economists in the HEA/REA approach is crucial. Their active participation in measuring and valuation of habitat services and resources helps in establishing appropriate habitat scalars and conversion rates, perhaps using biomass production modeling (French McCay and Rowe 2003), or direct calculation of the value of environmental services using typical non-market valuation techniques (English et al. 2009). There continues to be a strong need for joint economic and ecological research cooperation.

Biographies

W. Douglass Shaw

is a Professor in the Department of Agricultural Economics and Research Fellow, Hazards Reduction and Recovery Group at Texas A&M University. His research interests include risk and uncertainty, environmental economics, and non-market valuation.

Marta Wlodarz

is a Ph.D. student in the Department of Agricultural Economics at Texas A&M University. Her research interests include non-market valuation and environmental economics.