9. Substitution Methods

From here on, “service” is used to indicate both goods and services, and “environmental services” include ecosystem services.

Before delving into the substitution approaches, it should be noted that other supply-side nonmarket valuation methods exist, most importantly the production factor method. The production factor method computes the value of a nonmarket input in production of items offered for sale. If changes in an environmental condition affect the output levels or costs of firms, producers’ responses to such changes ultimately affect producers’ and potentially also consumers’ surpluses. Good introductions to the production factor method include Freeman (2003, Chapter 9), Point (1994), and Young (2005, Chapter 3). Because that method is conceptually more straightforward than substitution methods and is relatively uncontroversial, this chapter is devoted to the substitution methods.

In the United States, the alternative cost method was in use by the early 1930s and increased in prominence with passage of the Flood Control Act of 1936, which stated that the feasibility of a proposed project is assured only when “the benefits, to whomsoever they may accrue, are in excess of the estimated costs.” This statement essentially provided legislative support for the use of benefit-cost analysis in evaluation of public flood control projects, support that was later extended to other water projects (Griffin 2012). As U.S. government agencies began to implement the Flood Control Act and subsequent guidance in 1940s and 1950s, they searched for ways to measure the benefits of a variety of water resource projects, such as constructing flood control levees, dredging to improve navigation, and damming rivers to provide for water storage. Since then, the use of benefit-cost analysis has become common in the United States, Europe, and elsewhere, is used to evaluate many kinds of public projects, and takes advantage of the full range of valuation methods.

These three conditions were described by Eckstein (1958) and later summarized by Shabman and Batie (1978).

It is sometimes argued that the replacement cost is a lower bound, not an upper bound, on the value of the project or resource at issue. For example, an EPA report (2009, p. 52) states that when Conditions 1 and 3 are met, “It is valid to use the cost of providing the equivalent services via the alternative as a lower-bound estimate of the economic value of the ecosystem service.” This argument essentially says that if Condition 3 is met, we know that society is willing to pay at least the replacement cost to have the service provided, and thus the original ecosystem service or publicly provided resource must also be worth at least that much. While this is surely true, the argument ignores the limitation that the available perfect substitute places on maximum willingness to pay. If one accepts that willingness to pay is limited by the cost of an available perfect substitute, the gross benefit of the publicly provided resource cannot exceed the replacement cost.

A depiction of aggregate demand for the original service—that is, the service in the absence of a perfect substitute—would naturally be shown by a downward sloping demand curve. The horizontal demand curve in Fig. 9.1 depicts the situation faced by a new, or alternative, supplier of the identical service.

de Groot et al. (2012) reported that the replacement cost method (and the avoided cost method) have been the most commonly used methods for valuation of regulating ecosystem services.

A completely unsupported use of RCM occurs where the alternative would be provided by the same entity as that proposing the new public project, and the entity simply offers as the alternative a slightly more expensive project to provide the service, thereby assuring a cost savings if the proposed project were undertaken. This possibility led to the suggestion that the alternative must be provided by a substantially different means from that of the proposed project (Herfindahl and Kneese 1974; Young and Gray 1972). However, the “substantially different” requirement does not assure that Condition 3 is met.

An additional concern is that labor cost does not reflect the true opportunity cost because of lack of full employment of labor. One way to account for this is to estimate the labor cost as its nominal cost times (1 – p), where p is the probability that use of the labor will have no net effect on use of labor elsewhere.

This is not to suggest that the estimate of replacement cost is not socially relevant. Because the residue would be removed from farms to produce cellulosic ethanol to meet a target established by a U.S. federal law, the Energy Independence and Security Act of 2007, the cost of the commercial fertilizer is one part of the full cost of responding to the Act.

The water input in hydropower production is typically available without cost to the hydropower facility. The unit cost savings from construction of the hydropower plant, c ₂ – c ₂, can be attributed to the availability of the water input, and can be considered the value of the water in hydropower production.

In the coastline protection example, use of the RCM is related to use of the defensive behavior method (see Chapter 8). Recall that the defensive behavior method equates the cost of an individual’s defensive measure with the benefit to the individual of a policy that would avoid the need for defensive expenditure, on the assumption that the individual would be willing to pay at least the cost of the defensive measure to avoid the damage. The mangrove example equates the cost of additional dike repair (which avoids damage to crops and other investments) with the benefit of a policy (the mangrove forest) that would avoid the need for the defensive measure, on the assumption that society would be willing to pay at least the cost of the defensive measure to avoid the need for the measure. Defensive measures have been observed in both cases. The situations differ in that with the defensive expenditure method we observe individual defensive measures being taken, whereas in the mangrove case we observe a community-wide defensive measure—maintenance of the dikes.

HEA and REA came about partly in reaction to the cost of economic valuation—especially in situations where the damage is modest, the cost of analysis can be excessive in relation to the importance of the damage—and to court challenges of valuation efforts (Roach and Wade 2006).

The earliest exposition of the REA approach to compensating for resource losses could be King and Adler’s (1991) description of a process for determining compensation for anticipated wetland loss.

For example, Canada’s Fisheries Act contains a no-net-loss provision regarding the productive capacity of fish habitats that could allow for use of REA (Clarke and Bradford 2014; Quigley and Harper 2006).

The European Union’s Environmental Liability Directive was established in 2004 and incorporated into EU member states’ domestic law by 2010. Like comparable U.S. legislation, the ELD gives priority to physical/biological equivalency over value equivalency in the determination of compensatory liability as long as the replacement resources are similar in type, quality, and quantity to the lost resources.

As Randall (1997) explained, the most efficient approach to compensation is to choose the less expensive of two options: reimburse the parties suffering the losses or provide equivalent services.

The National Oceanic and Atmospheric Administration of the U.S. Department of Commerce serves as trustee for coastal and marine resources and determines the damage claims to be filed against parties responsible for certain injuries to natural resources.

The assessment performed following the 1989 Exxon Valdez oil spill is perhaps the best known compensatory analysis using an economic valuation approach (Carson et al. 2003; Goldberg 1994).

Other U.S. federal statutes affecting natural resource damage assessments include the Clean Water Act, the Park System Resource Protection Act, and the National Marine Sanctuaries Act.

Regulations for natural resource damage assessments require discounting: “When scaling a restoration action, … trustees must take account of the value of time in the scaling calculations by discounting to the present the interim lost services or the value of interim lost services due to the injury, as well as the gain in services or service value from the restoration action” (61 FR No. 4, p. 453).

Federal statutes and most state laws require that the baseline recognize the natural change that can occur over time in the absence of the injury, but some states use the pre-injury condition as the baseline. In practice, a lack of data and modeling capability often results in the assumption of a static baseline (Viehman et al. 2009).

An important difference between VEA and HEA or REA is that changes over time in per-unit economic values of the injured and restored resources are generally overlooked in applications of HEA or REA (Dunford et al. 2004; Zafonte and Hampton 2007). Changes over time in economic values can occur as population growth and economic development cause resource values to increase or as the damage causes a reduction in the supply of certain services leading to an increase in marginal value, which is likely to be followed with the recovery of the damaged resource plus the addition of the restoration resource by an increase in total supply and a resulting decrease in marginal value.

This requirement is similar to the first of the three conditions for applying the RCM—that both options provide the same service(s).

See Dunford et al. (2004), Flores and Thacher (2002), Jones and Pease (1997), Mazzotta et al. (1994), Randall (1997), Roach and Wade (2006), Unsworth and Bishop (1994), and Zafonte and Hampton (2007) for more on how HEA or REA compares with an economic valuation (VEA) approach.

Comparability is not only constrained by lack of information about economic values. It can be optimistic to expect that ecological understanding has advanced to the point where ecological “equivalence” can be assessed with much greater efficacy than can economic value. The track record in re-creating wetlands, for example, is apparently not good (Barbier 2011, p. 1841-1843), and the metrics that have been used in HEA (e.g., percent of herbaceous cover) tend to be inadequate to assure re-creation of ecological structure and function.

EA relies on professional judgment at many points, including selection of the metric. Because judgments can differ, involving several professionals, perhaps using a procedure such as Delphi might be advisable. When disagreement persists about a certain estimate, a resolution is sometimes reached by negotiation among the involved stakeholders (Ray 2009).

Use of a 3% discount rate, which is common in REA, results in a discount factor of about 0.23 by Year 50 and 0.05 by Year 100. Because some species, such as slow-growing coral, can take even longer than that to fully recover (Viehman et al. 2009), the discounted service loss—even in the case where the planning horizon is ample—can be rather unreflective of the full extent of the physical loss.

The existence of two separate losses—the loss of services of the salt marsh and also the loss of birds at sea—requires adding a term to the numerator of Eq.(9.11).

For legal reasons it is common in U.S. damage assessment to ignore damages that occurred before passage of the enabling legislation.

The reintroduced Chinook salmon originated from wild salmon donor stocks from an adjacent drainage (rather than being common hatchery salmon).

This summary is based largely on personal communication with Robert Unsworth (June 2013) of Industrial Economics Inc., who prepared an Expert Report for the plaintiff in the Storrie fire case.

The quadratic mean weights larger trees more heavily than would a simple mean, which was understood to more accurately reflect the services that flow from trees of different sizes. Tree diameter is measured at breast height (about 1.5 m from the ground).

The 0.0053 estimate was based on a 102-year fire history during which 27,531 acres burned out of a total of 50,624 acres. It was further assumed that once the area had burned, it would not burn again.