Environmental & Resource Economics (2007) 36:275–293 DOI 10.1007/s10640-006-9028-9
Ó Springer 2006
The Production of Eco-Labels CHRISTOPHER BRUCE1,* and ANDREA LAROIYA2 1 Department of Economics, University of Calgary, 2500 University Dr. N.W., Calgary, AB, T2N 1N4 Canada; 2Alberta Energy Utility Board, Calgary, AB, T2N 1N4 Canada; *Author for correspondence (e-mail:
[email protected])
Accepted 1 August 2006 Abstract. Eco-labeling is a market-based technique for conveying information about consumers’ demands for environmental protection. Most criticisms of eco-labeling have concentrated on demand-side issues – the potential for firms or countries to use eco-labels to manipulate market power – and the difficulty of creating credible labels. In this paper, we argue that fundamental problems also arise on the production side. Specifically, we argue that both increasing returns to scale and complexities in production imply that it will be possible to market only a small percentage of environmental attributes through the use of eco-labels. We present evidence for our hypotheses from a detailed survey of 100 randomly-selected consumer products. Key words: consumer demand, eco-labeling, endangered species, environmental decisionmaking, production function JEL Classifications: P48, Q57
1. Introduction A fundamental problem that is faced by those who are responsible for developing environmental policy is that they have only very approximate measures of consumer preferences. Standard information-gathering techniques, such as elections, lobbying, roundtables, public forums, and public opinion polls, may be able to indicate which issues are of concern to the public; but they are much less able to measure relative strengths of preferences among different groups or among alternative uses of resources. A public forum might be able to identify, for example, that some citizens would like a particular segment of public land to be used for logging, that others would prefer that it be used for hiking and skiing, and yet others would like it to be reserved for the protection of threatened species. The forum will not, however, be able to signal clearly which of those preferences is the strongest.
276
CHRISTOPHER BRUCE AND ANDREA LAROIYA
The result is that policymakers encounter difficulty determining whether the public’s desire to protect some aspect of the environment exceeds its desire to use that aspect in some alternative manner. When issues are welldefined, such as when a decision must be made whether to allow drilling for oil on pasture land, policymakers may be able to rely on insight and personal judgment. But when complex, multidimensional decisions must be made, as is often the case with environmental questions, judgment becomes a much less dependable resource. For this reason, analysts have proposed a number of alternative approaches to measure the strength of the public’s preferences for environmental protection – most importantly: contingent valuation, environmental collaboration, and eco-labeling. The purpose of this paper is to investigate an aspect of the latter approach that we believe has received inadequate attention in the academic literature – the nature of the production function for environmental amenities. Specifically, we will argue that in a large number of cases the techniques required to produce improvements in the environment are such that private firms will encounter difficulty ‘‘tying’’ those improvements to the sale of their consumer products. The paper is divided into four sections, In the first, we briefly review the academic literature concerning eco-labels. In the second and third we discuss two properties of the environmental ‘‘production function’’ that we predict will deter private firms from offering eco-labels. Finally, we present some preliminary empirical findings concerning our hypotheses.
2. The Eco-Labeling Literature An eco-label is a claim by a firm that it has employed environmentallysensitive production or distribution methods. These claims often appear as labels fixed to products; but they can also take the form of advertisements or representations made by salespeople. In each case, however, the ‘‘label’’ is intended to convey information to consumers about the steps the firm has taken to protect the environment. If consumers value these steps, they may be willing to pay a premium for the product to which the eco-label has been attached, thereby creating a ‘‘market’’ for environmental protection. As long as this market operates smoothly, government agencies will not need to estimate consumer preferences as those preferences will be reflected in the prices consumers are observed to pay and in the quantities they purchase. The academic literature discusses three sources of imperfections in the market for eco-labels that might prevent consumer preferences from being revealed in this way. First, because most environmental attributes are public goods, economists anticipated that the free rider problem would
THE PRODUCTION OF ECO-LABELS
277
deter consumers from revealing their preferences in the market place, thereby preventing vigorous markets from developing. Empirical testing has shown, however, that many consumers are willing to pay a premium for eco-labeled products (Gumpper, 2000; Imkamp, 2000; Loureiro et al., 2002; Makatouni, 2002; Moon et al., 2002) and that they do purchase such products (Carlson et al., 1993; Lathrop and Centner, 1998; Teisl et al., 2002). These results have suggested to many economists (Andreoni, 1990, 1995; Cremer and Thisse, 1999) that consumers may be willing to behave altruistically – contrary to standard economic assumptions. No formal theory of altruistic behavior has been developed, however.1 Second, a number of authors have noted that mandatory eco-labeling schemes may introduce imperfections into product markets. Kirchhoff (2000) and Liebi (2002), for example, both develop models that suggest that firms may be induced to over-invest in pollution abatement; whereas Matto and Singh (1994), Swallow and Sedjo (2000), and Dosi and Moretto (2001) all identify circumstances in which eco-labels may induce firms to reduce their investments in environmental protection. Similarly, there is a growing literature that suggests that mandatory eco-labeling might be used as a barrier to trade. (See, especially, Beaulieu and Gaisford, 2002 and the papers contained in Vossenaar et al., 1997.) Third, a large number of researchers (including Erskine and Collins 1997, Lathrop and Centner, 1998, and Thøgersen, 2000) have noted that eco-labels will not provide consumers with the opportunity to reveal their preferences about firms’ environmental policies if consumers find it difficult to check on the veracity of firms’ claims. These researchers commonly call for government imposition of standards for eco-labels, such as are contained in Germany’s Blue Angel program (Kirchhoff, 2000). But, as Church (1994–1995) argues, mandated labeling standards provide consumers with very little true choice. If there is a single government standard, consumers are offered only two choices: buy either the labeled or the non-labeled product. It is only if government intervention takes the form of certification of the firm’s claim (rather than mandating the content of that claim) that eco-labeling can act to convey consumer preferences through the market. In this paper, we argue that even if the three problems identified above could be resolved, there is a fourth factor that could inhibit the development of efficient markets in environmental attributes: the nature of the production function for these goods. Specifically, we argue that small firms may be deterred from providing the socially desirable level of eco-labels if there are economies of scale that they are unable to capture; and that large firms may produce too few eco-labels if there are diseconomies of scale in the provision of information about firms’ environmental practices.
278
CHRISTOPHER BRUCE AND ANDREA LAROIYA
3. Economies of Scale in the Production of Environmental Amenities 3.1. ENVIRONMENTAL AND COMMERCIAL USES OF LAND ARE INCOMPATIBLE
Protection of environmental attributes often requires that commercial activities are either barred from, or severely constrained on, some portion of the available land. Where farm use comes in conflict with the habitat of an endangered species, for example, protection often requires that all such use cease in the area of the affected habitat. (See, for example, Lamberson et al. 1992; Vincent and Binkley, 1993; Swallow and Wear, 1993; and Fahrig and Merriam, 1994.) In these cases, the cost to the commercial operator of altering its activities will be the profit foregone on the output that would have been produced on that land. In an eco-labeling system, the operator will attempt to recoup those profits by raising the price it charges on the crops it produces on its remaining land base. Consider, first, the short-run situation in which the number of hectares of land available to the firm is L, of which LE is to be left fallow as environmental reserve. It uses the remaining portion of its land, (L ) LE), in combination with a fixed amount per hectare, n, of a variable input to produce commercial (non-environmental) output level N.2 If the prices of land and the variable input are pL and pn, respectively, and if the firm operates in a competitive product market with output price pN, its profits from the production of N, on (L ) LE), can be represented as: Y ¼ pN NððL LE ÞÞ pL ðL LE Þ pn nðL LE Þ ð1Þ N Assume, further, that when LE hectares are left fallow, E units of environmental amenities are produced.3 If the firm is able to ‘‘sell’’ each of these units at a price of pE, (by attaching a premium, or eco-label, to each unit of N), its net earnings from production of E will be: Y ¼ PE EðLE Þ pL LE ð2Þ E The firm’s total profits are, therefore Y ¼ pN NððL LE ÞÞ þ pE EðLE Þ pL L pn nðL LE Þ
ð3Þ
and the first order condition for selection of the short-run profit-maximising level of LE is: pN
@N @E pn n ¼ pE @LE @LE
ð4Þ
THE PRODUCTION OF ECO-LABELS
279
As long as the marginal product of environmental reserve, dE/dLE, is decreasing (and there is no corner solution), the firm will be induced to leave some positive amount of land, LE*, fallow. However, there are many situations in which there will be increasing returns to scale in the production of environmental amenities – that is, in which the marginal product of environmental preserve will be increasing in LE (Lamberson et al., 1992, Vincent and Binkley, 1993). Most importantly, endangered species often require large, contiguous tracts of undisturbed land if they are to thrive and reproduce.4 As the amount of such land increases, the probability of survival increases at an increasing rate over some nontrivial range of inputs, as shown in Figure 1. In such cases, if area A is less than area B in Figure 1, it may, once again, be profitable for the firm to set aside an environmental preserve of size LE*. (If A is greater than B, the optimal policy will be to provide no protection for the species in question.) Note, however, that LE* may be greater than the total amount of land available to the firm, L. In such a situation, two outcomes are possible. In the first, L is small enough relative to LE* that even if the firm was to set aside all of its land as preserve, the costs of that action would exceed the benefits. (Such a case is represented by L = L1 in Figure 1.) The socially desirable policy, leaving LE* fallow, will not be taken because individual firms are too small to obtain the necessary economies of scale. In the second, L is large enough that, even though it is less than LE*, the benefits of leaving that land fallow exceed the costs – for example, at L = L2 in Figure 1. Normally, in such a situation, it would be concluded that the firm would select the corner solution: it would maximize profits by leaving all of its land fallow. However, in the case of an eco-labeling scheme, this outcome is not feasible as, in such a scheme, there has to be some output to which the firm can attach a label (and receive compensation for its actions). As long as L is less than or equal to LE* the firm must provide less than the socially optimal amount of environmental protection (in the short run).5
Figure 1. Increasing returns to scale in land.
280
CHRISTOPHER BRUCE AND ANDREA LAROIYA
In the long run some of these problems could be overcome if firms were able to purchase additional land, which they could then convert to environmental reserves. But this solution encounters three difficulties. First, if the socially optimal size of a contiguous reserve is sufficiently large, firms may encounter borrowing constraints when attempting to raise the funds to purchase sufficient additional land. Second, it is not clear that eco-labeling is the optimal method of marketing large tracts of environmental reserves. Once environmental conservation becomes a major focus of the firm’s activities, it may be efficient to market that conservation directly. For example, assume that whereas the average size of a private woodlot is 100 hectares, the optimal size of forest preserve for the protection of an endangered species is 1,000 hectares. Although it is possible that a woodlot owner could purchase 1,000 hectares, and include the cost of that land in the price of the wood from his original 100 hectares, by doing so he would unnecessarily restrict the purchasers of the environmental amenity to the customers of his woodlot. Finally, if the opportunity for eco-labeling encouraged small firms to expand significantly, in order to absorb large environmental reserves, competition in the product market would be inhibited. For example, if owners of 100 hectare woodlots were each to purchase parcels of 10,000 hectares, in order to protect 1,000 hectares each, the total number of woodlots would decline dramatically. 3.2. ENVIRONMENTAL PROTECTION REDUCES OUTPUT OF THE NON-ENVIRONMENTAL GOOD
Additionally, the techniques that are used to increase output of the environmental good may reduce the output of the non-environmental good directly, (that is, without affecting the amount of land available for commercial production). For example, the planting of shade trees on coffee plantations, a change in crop rotation practices, or a move away from clear cutting in the forest industry may each lead to improvements in the environment at the expense of reductions in output of the commercial product. As in the preceding section, the primary cost of environmental protection is a reduction in the output of the commercial good, and the benefit of that protection arises from ‘‘sale’’ of the environmental good through attachment of an eco-label to the commercial product (and consequent increase in retail price). If it is assumed that the producers of eco-labels operate in competitive product markets, the representative firm’s short-run profit function becomes: Y ð5Þ ¼ pN Nðn; e; LÞ þ pE Eðn; e; LÞ pn nL where e represents the effort expended per hectare by the firm to protect the environment (or, alternatively, to produce E). For simplicity, it is
THE PRODUCTION OF ECO-LABELS
281
assumed (i) that the amount of land, L, is fixed; and (ii) that an increase in e has no direct costs: its primary effect is to reduce N. The first order condition with respect to e is, therefore, pN
dN dE ¼ pE de de
ð6Þ
Making the usual assumptions about marginal products, this condition implies that, if the firm is able to sell its environmental output at pE per unit, it will invest some effort level, e*, into the protection of the environment. There is an important circumstance, however, in which the investment level that is chosen by the firm will be less than the socially optimal level, Specifically, in many cases, the marginal impact of the individual firm’s investment in the environment will be an increasing function of the level of investment that has been undertaken by all other firms; that is, there may be increasing returns to scale available to the industry that are not available to the firm (Helfand and Rubin, 1994; Managi, forthcoming). Most importantly, when the environment has been heavily degraded, there may be no appreciable improvement in the environmental amenity, E, until industrywide investment in environmental improvement has reached some minimum level. In such cases, if each firm is small relative to the number that has been causing the degradation, the actions of individual firms may have little or no perceptible effect on E. The marginal costs of protection will exceed the (perceived) marginal benefits, resulting in little (market-based) incentive to provide eco-labels. Yet, the socially optimal level of investment in environmental protection may not be zero. As the number of firms that invested in protection increased, the industry might reach a ‘‘switching point,’’ beyond which the marginal benefits of that protection would rise above the marginal costs. Assuming that this point occurs at a level of protection larger than could be achieved by individual firms, some extra-market incentive would have to be provided to induce firms to invest in this level. This could take the form, for example, of government regulations or of a campaign by environmental groups to induce firms to adopt some standard chosen by them.6 However, although it is likely that the former would achieve its goals, and it is possible that the latter would do so, neither would meet the objective identified in the introduction to this paper: that the market provide a means by which the intensity of consumer preferences for alternative environmental actions could be measured. Rather, in these cases, it would be either the government’s estimate of consumer preferences or the environmental group’s own preferences that would be implemented. Furthermore, in both cases, the agencies (government or environmental groups) could obtain essentially the same results without the use of eco-labels, making the latter superfluous.
282
CHRISTOPHER BRUCE AND ANDREA LAROIYA
The only circumstance in which eco-labels would consistently provide a method for measuring consumer preferences, in the face of industry-wide increasing returns to scale, is that in which the primary sources of environmental harm came from a small number of (relatively large) firms. In that case, each firm might be able to capture the available economies of scale for itself, producing relationships between e and the two marginal values depicted in Figure 2. There, it is assumed that the marginal cost curve has the usual, upward slope while the marginal benefit curve rises at first (due to increasing returns to scale beyond a point like e1) and then falls when diminishing returns to scale begin to set in. In this case, if area B is larger than area A, the firm will maximize profits by investing e* (and attaching ecolabels to its products indicating that it had done so). 3.3. ECONOMIES OF SCALE: SUMMARY
In this section, we have argued that there are many cases in which there are economies of scale in the production of environmental amenities – particularly those in which large tracts of land must be set aside to protect endangered species and those in which improvements in air or water quality have little effect until some ‘‘switching point’’ is reached. In these cases, firms that are able to produce only a small percentage of the desired amenity may find that the costs to them of taking environmental precautions exceed the costs. These firms will fail to offer eco-labels even when the social benefit of an industry-wide precautionary investment exceeds the cost. 4. Diseconomies of scale in the production of eco-labels In Section 3, we implicitly assumed that each firm affected only one aspect of the environment. In such circumstances, production of the eco-label itself is
Figure 2. Increasing returns to environmental protection.
THE PRODUCTION OF ECO-LABELS
283
virtually costless: the firm merely needs to place a sticker on its product, or print a few words on its packaging, to alert the buyer to the environmental action that it has taken. As long as the consumer finds the claim to be credible, both the cost to the firm of producing the label and the cost to the consumer of reading and absorbing the content of the label may be trivial. But most firms affect the environment in a large number of ways. Even an operation as ‘‘simple’’ as a farm can have many different effects on water and air quality and can affect the habitats of numerous threatened and endangered species; and a firm producing a product as complex as an automobile or a television set may affect the environment in hundreds or even thousands of different ways. There is reason to believe that if a firm uses eco-labels to signal to consumers that it has produced environmental amenities, the marginal cost of those labels will increase with the number of labels, (that is, that there are diseconomies of scale in the production of labels). First, if there is limited space on the product’s packaging (such as on a cereal box), as the firm increases the amount of environmental information that it provides, it will have to displace increasingly valuable information about other aspects of the product. Second, as the number of environmental impacts increases, the firm may have to adopt increasingly expensive technologies for displaying information about those impacts. For example, it may have to increase the size of the packaging, add a label or sticker where one did not previously exist, insert a pamphlet, or develop a website. Furthermore, the marginal costs to consumers of obtaining and absorbing the required information can also be expected to increase with the number of environmental impacts being described, Whereas it might be virtually costless for the consumer to recognize one or two labels while perusing the product for other information; as the number of impacts being described rises above that level, the consumer will have to begin making a conscious effort to obtain information about those impacts. At first, this may simply require that the consumer spend time reading the packaging more carefully. But, as the amount of information increases, consumers may be directed to additional, less convenient sources, such as pamphlets or websites. As that will require ever greater amounts of time and effort, the marginal cost of ecolabels will increase with the number of such labels that are associated with each product. When the marginal cost of eco-labels is upward-sloping, the probability that the firm will offer to produce any individual environmental amenity will be a decreasing function of the number of environmental harms caused by the firm. That is, as the net benefit of producing an amenity (or reducing a harm) must exceed the cost of the associated eco-label, if the marginal cost of those labels increases with their number, the firm will be less likely to produce
284
CHRISTOPHER BRUCE AND ANDREA LAROIYA
any one amenity, the greater is the total number of such amenities. Thus, as the number of impacts a firm can have on the environment increases, it may respond by providing eco-labels for a decreasing percentage of those impacts. Alternatively, the firm might reduce the amount of information provided about each amenity by ‘‘packaging’’ a set of environmental amenities into a single label. For example, a firm might advertise that it had used ‘‘environmentally-friendly’’ production techniques or that it had used only ‘‘organic’’ inputs. In the former case, the firm will fail to produce amenities for which the net benefits are positive. In the latter, it will not offer consumers a market-based choice concerning the mix of environmental actions that they would like firms to take. Consumers will only be given the option of accepting or rejecting the package offered by the firm. 5. Empirical evidence The analysis in Sections 3 and 4 suggests two broad hypotheses: (i) firms are less likely to offer eco-labels when there are economies of scale in the production of environmental amenities than when there are constant or decreasing returns to scale; and (ii) firms are more likely to offer an eco-label with respect to an environmental amenity the smaller is the number of such amenities that is affected by the firm. Our ability to test these hypotheses is restricted by two factors. First, it has not been possible to develop a complete theory of eco-labels, primarily because economists (and other social scientists) lack a formal model of altruism. As a result, it is often difficult to distinguish between behaviors that arise from demand-side variables (altruism, for example) and those that arise from supply-side considerations (economies of scale, for example). Second, it is difficult to obtain data concerning many of the elements of our hypotheses. There is very little information available concerning economies of scale in the production of environmental attributes, for example, or with respect to the number of effects that individual firms have on the environment. Nevertheless, we believe that valuable information about the market for eco-labels could be provided if a random sample of all categories of consumer goods was undertaken, in order to determine: (i) whether a representative consumer could, for each such category, find at least one firm that offered an eco-labeled product; and (ii) whether a common pattern could be found among the characteristics of the set of products for which such labels were available. We describe such a survey in subsection 5.1, below. In subsection 5.2, we employ the results of that survey to provide some preliminary evidence concerning the hypotheses developed in Sections 3 and 4 of this paper.
285
THE PRODUCTION OF ECO-LABELS
Table 1. Types of eco-labels Claim made on eco-label
Food Organic Recyclable Organic and recyclable Other dolphin safe (tuna) shade grown (coffee) No claim Total (food) Non-food Energy efficient Phosphate free (soaps)* Sun or wind generated energy Contains recycled components ‘‘All natural ingredients’’ ‘‘Environmentally friendly**’’ No claim Total (non-food)
Number of products on which claim was made
19 2 8 1 1 4 35 3 5 3 4 2 4 44 65
* The five categories of ‘‘soaps’’ were laundry detergent, cleaning and scouring powders, fabric softener, shampoo, and deodorant. In each case, most manufacturers claimed that their products were phosphate free and biodegradable and that they were not tested on animals. ** The four categories were hotels, automobiles, automotive fuels, and computer supplies. 5.1. THE SURVEY
As our goal was to construct a random survey of the universe of consumer products, our first step was to develop a complete list of those products. For this purpose, we began with a comprehensive survey (Statistics Canada, 1992, 1996) of approximately 800 categories of products that are purchased by Canadian consumers. From this list, we deducted all categories representing personal services, interest and mortgage payments, and taxes (on the assumption that these products had little effect on the environment). This left 692 categories, from which 100 were selected at random.7 With respect to each category, three representative8 retail outlets were approached. Every brand sold in those outlets, within that category, was analyzed closely to determine whether an eco-label was offered and, if so, what was contained in that label. Although this procedure may not yield a strictly random sample of products, we argue that (i) it provides a fair indication of the types of ecolabels that are encountered by the typical Canadian consumer; and (ii) it represents the first attempt of which we are aware to obtain such an indication by sampling consumer products rather than by sampling eco-labels (through, for example, a search of web sites and news releases).9
286
CHRISTOPHER BRUCE AND ANDREA LAROIYA
With respect to each of the 100 categories of products that we surveyed, our primary goal was to determine (i) whether an environmentally-concerned consumer could find at least one firm or brand that, through an eco-label, claimed to provide an environmental amenity; and (ii) what types of amenities were offered within those product categories. With respect to food products, We found that within 31 of the 35 categories at least one of the firms or brands surveyed offered some type of eco-label. Those labels took one of five forms. In 19 of the 31 categories, at least one firm claimed that its product was ‘‘organically grown,’’ and no firms made any other environmental claim. In two categories, the only claim that was made (by at least one firm) was that the product was recyclable. In eight categories, there was at least one firm that offered organic products and one whose product was recyclable. (In most cases, at least one firm would claim to be both organic and recyclable.) The remaining two categories were tuna, within which virtually all brands claimed to be ‘‘dolphin safe;’’ and coffee, within which many brands stated that they were ‘‘shade grown’’ (and, hence, supportive of wild bird populations). These results are summarized in the upper section of Table 1. The lower section of Table 1 reports the comparable results for our survey of the 65 non-food categories. The dominant finding here was ‘‘no claim:’’ in 44 categories, our survey found no products that had an eco-label of any kind. Within the remaining categories we found firms that claimed: that their products were energy efficient (3 categories), contained recycled components (4), or had ‘‘all natural’’ ingredients (2); that the energy source for their product was sun- or wind-generated (3); that their product was produced in an ‘‘environmentally-sensitive’’ manner (4); or, in the case of many soap products, that they were biodegradable and used no phosphates or petroleum products (5). In no category did we find that some firms made one of these claims and others in that same category made another claim. The firms who claimed that their products contained recycled materials, for example, all fell within four product categories and no firms in any of those four categories made any other environmental claim. Because of our sampling procedure, we cannot draw any definitive conclusions about the global prevalence of eco-labels. As we do not have a complete list of all manufacturers of the 100 products sampled, we cannot estimate the percentage of firms that offer eco-labels; nor can we obtain an accurate estimate of the percentage of product categories within which ecolabels were offered. Our survey was sufficiently exhaustive, however, that we are confident that our ranking of the relative prevalence of the different types of claims (for example, organic versus energy efficient) provides sufficient information that we can offer a preliminary evaluation of the hypotheses set out in Sections 3 and 4 of this paper.
287
THE PRODUCTION OF ECO-LABELS
Table 2. Type of information provided on product Type of information*
No information General/recyclable Eco-symbol/certification Direct to website Specific Total claims
Food
Non-food
Number
Percentage
Number
Percentage
4 6 9 0 16 35
11.4 17.1 25.7 0 45.7
44 7 2 5 7 65
67.7 10.8 3.1 7.7 10.8
*The five types of information are defined in the text. 5.2. EVIDENCE
First, there are sufficient regularities in the survey data to allow us to draw a number of inferences about the potential market for eco-labels. The observation that eco-labels were available in 52 of the 100 product categories surveyed suggests that the ‘‘free rider/public good’’ problem does not override the willingness of at least some consumers to pay for environmental amenities. Even if it is argued that, with respect to many of the labels, consumers were motivated by self-interest – for example, they may have purchased ‘‘organic’’ foods for the health benefits or energy efficient products for the cost savings – there still remains a substantial number of labels whose values appear to arise primarily from the environmental benefits they provide: for example, recyclable goods, dolphin safe tuna, shade grown coffee, sun and wind generated energy, and phosphate free soaps, The question that has been raised by this paper is whether we can explain why these labels are available for some products and not for others. In this section, we discuss three possible hypotheses. 5.2.1. Cost of information In Section 3 we hypothesized that a firm would be less likely to provide detailed information about the impact that its production had on the environment, the greater was the total number of those impacts. Automobile manufacturers, for example, were predicted to be less likely to provide specific information about the individual effects they had on the environment than were tuna processors. To test this hypothesis, it was necessary to obtain an objective method of determining whether a firm had a ‘‘small’’ or a ‘‘large’’ number of impacts on the environment. As a first approximation, we believe that this division may be obtained by distinguishing between the ‘‘food’’ and ‘‘non-food’’ categories in our survey. Whereas most food products affect the environment in a limited number of ways – perhaps one or
288
CHRISTOPHER BRUCE AND ANDREA LAROIYA
two impacts on each of species habitat and air and water quality – most nonfood products combine a large number of raw and processed inputs in complex manufacturing processes, with each step (potentially) affecting the environment in many ways. Our hypothesis also requires that we distinguish firms that provide ‘‘detailed’’ information from those that provide no or only ‘‘general’’ information about their products. This we did by allocating claims among five broad groups, reported in Table 2. The first, ‘‘no information,’’ refers to categories within which no firm offered any information concerning its environmental practices. The second, ‘‘general/recyclable,’’ refers to a category in which the greatest amount of information provided by any firm was only a very general claim, such as that it was produced in an environmentally responsible manner, that it used recycled components, or that its product was recyclable. ‘‘Eco-symbol/certification’’ refers to a category in which at least one firm claimed that its product met some third-party set of criteria, but in which no information was provided on the product about those criteria. The category ‘‘direct to website’’ refers to products that directed the consumer to a website containing information about the product’s environmental impact. Finally, ‘‘specific’’ refers to a claim that the firm had taken at least one clearly-specified action, such as substituting organic fertilizers for chemical, growing coffee under shade trees, or removing phosphates from the production process. The results of this categorization are reported in Table 2. It is seen there that food products were more than four times as likely as non-food products (45.7% versus 10.8%) to carry specific, detailed information about environmental impacts. A z-statistic indicates that this difference is significant at the 2% level. Even if labels that direct consumers to a website are included in the specific information group, the differences between food (45.8%) and non-food (18.5%) are significant at the 5% level, These findings suggest that most non-food producers find it sufficiently costly to provide full information about their environmental practices that they simply abandon the effort; and when the remaining non-food producers do provide information, it is as likely to be of a generic nature as it is to be specific. 5.2.2. Economies of scale in land use We further anticipated that firms would be unwilling to set aside more than a small percentage of their land base to provide protection for endangered species. On this issue, our survey did not identify even one firm that reported it had set aside land for environmental protection. Indeed, the only related eco-label we found was the claim by some coffee producers that their crops were shade-grown (and, hence, created habitat for tree-dwelling species).10
THE PRODUCTION OF ECO-LABELS
289
These findings are, at least, not inconsistent with our expectations. Furthermore, other explanations are not compelling. One possibility, for example, is that firms with land-intensive production techniques may also affect so many aspects of the environment that they find it too costly to provide information about any one of them (see above). This argument, however, does not conform with the observation that land-intensive production is often associated with primary industries, such as forestry and agriculture, where the number of environmental impacts is relatively small – small enough that most of the food products we investigated claimed to have engaged in more than one environmentally-beneficial practice. These firms appear to have had every opportunity to set aside land for protection but did not (claim to) do so. 5.2.3. Economies of scale in production of amenities We also noted that if economies of scale in the production of environmental amenities could only be obtained at the industry level, individual firms might find the costs of producing eco-labels to be excessive. One method of testing this hypothesis would be to observe the claims that individual farms make with respect to water pollution, as most such producers will have only a negligible effect on the overall quality of water. Our survey identified nineteen food categories in which individual firms competed at the retail level with firms that aggregated the production of numerous farms, through either processing (for example, yoghurt and cheese) or distribution (for example, raisins and beef). If it is assumed that the latter were able to obtain economies of scale that were not available to the former, we would expect to find that the large producers would be more likely to use organic production techniques than would individual farms. In fact, the reverse was true: whereas small firms used organic techniques in seventeen of the nineteen categories, large firms used those techniques in only twelve categories. Not only were large firms not more likely than the small firms to use organic techniques, a z-statistic calculated on the difference between these means suggested that, at the 5% level, small firms were significantly more likely to offer eco-labels than are large firms. These findings suggest that if there are economies of scale in the production of environmental amenities, at least some consumers are willing to contribute to the production of those amenities even though, at low levels of production, the costs exceed the benefits. Perhaps these individuals benefit from the ‘‘warm glow’’ effect that is often observed in contingent valuation studies (Andreoni, 1990, 1995; Cremer and Thisse, 1999): their reward comes from the feeling that they are ‘‘doing something’’ rather than from the actual impact their actions can be expected to have on the environment11.
290
CHRISTOPHER BRUCE AND ANDREA LAROIYA
5.2.4. Summary In this subsection we employed a set of data drawn from a random sample of 100 categories of consumer products to examine a number of hypotheses concerning the production of eco-labels, Our primary finding was, as we expected, that firms were less likely to provide specific information about their environmental practices, the greater was the number of impacts their production processes could have on the environment. Contrary to our expectations, however, firms did not substitute general information for specific; rather, most firms whose production had complex effects on the environment chose to offer no information at all about those effects. The data also offered some support for our expectation that firms would not set aside large portions of their land for environmental protection; but were inconsistent with our expectation that firms would not offer eco-labels when there were industry-wide economies of scale in the production of environmental benefits. 6. Conclusion One of the greatest difficulties facing environmental policymakers is finding a method for weighing competing societal interests against one another. How should they decide, for example, whether a public forest should be used for logging, mining, recreation, or protection of endangered species, or for some combination of those activities? One technique that has been proposed for resolving this issue is to attach eco-labels to products, to provide consumers with information about those products’ environmental attributes. Through their purchases, consumers can then ‘‘vote’’ for those attributes that they prefer. In this paper we have argued that the ability of firms to market environmental benefits through eco-labels will be severely constrained by the manner in which those benefits are produced. Specifically, we argued (i) that fins with complex production processes would affect so many aspects of the environment that it would be extremely difficult for them to provide information about those effects to consumers; and (ii) that there are often significant increasing returns to scale in the production of environmental amenities, making it difficult for industries composed of relatively small firms to provide those amenities. We provided some preliminary information concerning these hypotheses by examining the environmental claims of all firms within a random sample of 100 categories of consumer products. From this examination, we concluded, first, that eco-labeling will have only a limited role to play with respect to the environmental impacts of non-food products, particularly those requiring complex manufacturing processes. Thus, with respect to these
THE PRODUCTION OF ECO-LABELS
291
products, consumer preferences concerning air and water pollution will be only very imperfectly signaled through the use of eco-labels. Second, we argued that it is unlikely that eco-labeling will be able to offer a market-based method for individuals to signal their preferences for environmental attributes that require large expanses of land – for example, protection of the habitats of species such as grizzly bears, wolves, and spotted owls. We have argued both that the primary role of eco-labels in public policymaking is to provide a means by which consumer preferences can be revealed; and that production and cost effects place serious constraints on the ability of labels to transmit that information. If these arguments are accepted, many of the concerns raised in the eco-labeling literature become moot. When labels cannot transmit information, it becomes less important to answer: why governments have been so slow at setting labeling standards, whether labels should report life-cycle effects, or how the credibility of private labels can be improved. Acknowledgment We would like to thank the Donner Canadian Foundation for its generous financial support. We would also like to thank Eugene Beaulieu, Daniel Gordon, and two anonymous referees for their comments. Notes 1. All economic models of altruistic behavior simply assume that individuals behave altruistically. They do not attempt to identify why, or under what circumstances, individuals act this way. 2. By substitution, N is a function only of (L)LE), as the variable input increases in strict proportion to the amount of land. 3. For notational convenience, we represent the firm’s action as being one of producing environmental goods, not of reducing environmental ‘‘bads.’’ 4. For example, it is believed that each pair of Northern Spotted Owls requires 1,000 acres (approximately 400 hectares). (Forsman 1998). 5. Indeed, even if L exceeds LE*, it may only be practical to use eco-labels when the amount to be set aside is a relatively small portion of the total land available. This is because the price on the commercial product must include a premium to compensate the firm for providing the amenity, E. As LE increases relative to L, the amount of commercial product to which this premium can be applied falls and the ratio of the premium to the price of the commercial product rises exponentially. 6. These are what the ISO calls Type I eco-labels; that is, labels whose criteria are established and certified by a third party (as opposed to ISO Type II labels, whose criteria are set by the producer). 7. The selection procedure was such that the probability that a category would be chosen was proportional to the percentage of after-tax income that was spent on it.
292
CHRISTOPHER BRUCE AND ANDREA LAROIYA
8. An attempt was made to survey a ‘‘high-,’’ ‘‘medium-,’’ and ‘‘low-price’’ retailer in each category, Also, in the food item categories, the largest ‘‘alternative’’ food store in the region was included in the set of three retailers. 9. The only other attempt at conducting a random sampling of products of which we are aware is found in Carlson, et al. (1993), They surveyed the advertisements in every issue of eighteen ‘‘representative’’ magazines, in 1989–1990, to identify the types of environmental claims that firms were making. 10. The only other references to endangered species that were identified in our survey were the claim that tuna producers had used dolphin-friendly nets and that one wine producer (Torres Coronas) had acted to protect the Bonelli’s Eagle. However, the former does not require that any fixed factor be left fallow; and the latter appears to have been achieved through selection of appropriate herbicides, pesticides, and fertilizers. 11. We find Cremer and Thisse’s explanation particularly relevant in this situation. They argue (1999, at 576) that: ‘‘individuals consider environmental-friendliness as a particular quality attribute of the products they buy. In other words, consuming a variant with a high environmental quality is in itself gratifying.’’
References Andreoni, J. (1990), ‘Impure Altruism and Donations to Public Goods; A Theory of WarmGlow Giving’, The Economic Journal 100, 464–477. Andreoni, J. (1995), ‘Warm-Glow Versus Cold-Pickle: The Effects of Positive and Negative Framing on Cooperation in Experiments’, The Quarterly Journal of Economics 110(1), 1–21. Beaulieu, E. and J.Gaisford (2002), ‘Labour and Environmental Standards: The ‘Lemons Problem’ in International Trade Policy’, The World Economy 25(1), 59–78. Carlson, L., S.Grove and N.Kangun (1993), ‘A Content Analysis of Environmental Advertising Claims: A Matrix Method Approach’, Journal of Advertising 22(3), 27–39. Church, J. (1994), ‘A Market Solution to Green Marketing: Some Lessons from the Economics of Marketing’, Minnesota Law Review 79, 245–324. Cremer, H. and J. F.Thisse (1999), ‘On the Taxation of Polluting Products in a Differentiated Industry’, European Economic Review 43, 575–594. Dosi, C. and M.Moretto (2001), ‘Is Ecolabelling a Reliable Environmental Policy Measure?’, Environmental and Resource Economics 18, 113–127. Erskine, C. and L.Collins (1997), ‘Eco-labelling: Success or Failure?’, The Environmentalist 17, 125–133. Fahrig, L. and G.Merriam (1994), ‘Conservation of Fragmented Populations’, Conservation Biology 8(1), 50–59. Forsman, E. D. (1998), ÔNorthern Spotted OwlÕ, in M. J.Mac, P. A.Opler, C. E.Haecker and P. D.Doran, eds., Status and Trends of the Nation’s Biological Resources, Vol. 2 (pp. 672– 675). Reston, Va: U.S. Department of the Interior, U.S. Geological Survey. Gumpper, M. (2000), ‘Do Consumers Respond to Ecolabels: Evidence from a Market Experiment Using Contingent Valuation’, Pennsylvania Economic Review 9(1), 14–22. Helfand, G. and J.Rubin (1994), ‘Spreading versus Concentrating Damages: Environmental Policy in the Presence of Nonconvexities’, Journal of Environmental Economics and Management 27, 84–91. Imkamp, H. (2000), ‘The Interest of Consumers in Ecological Product Information is Growing-Evidence from Two German Surveys’, Journal of Consumer Policy 23, 193–202.
THE PRODUCTION OF ECO-LABELS
293
Kirchhoff, S. (2000), ‘Green Business and Blue Angels’, Environmental and Resource Economics 15, 403–420. Lamberson, R., R.McKelvey, B.Noon and C.Voss (1992), ‘A Dynamic Analysis of Northern Spotted Owl Variability in a Fragmented Forest Landscape’, Conservation Biology 6(4), 505–512. Lathrop, K. and T.Centner (1998), ‘Eco-Labeling and ISO 14000: An Analysis of US Regulatory Systems and Issues Concerning Adoption of Type II Standards’, Environmental Management 22(2), 163–172. Liebi T. (2002), ‘Monitoring Eco-Labels; You Can Have Too Much of a Good Thing’, Bern, Switzerland: University of Bern, Diskussionsschraften, dp0207. Loureiro, M., J.McCluskey and R.Mittelhammer (2002), ‘Will Consumers Pay a Premium for Eco-labeled Apples?’, Journal of Consumer Affairs 36(2), 203–219. Makatouni, A. (2002), ‘What Motivates Consumers to Buy Organic Food in the UK?’, British Food Journal 104(3), 345–352. Managi, S. (forthcoming), ‘Are there Increasing Returns to Scale to Pollution Abatement? Empirical Analytics of the Environmental Kuznets Curve in Pesticides’. Ecological Economics (20 pp.). Matoo, A. and H.Singh (1994), ‘Eco-Labelling: Policy Considerations’, Kyklos 47, 53–65. Moon, W., W.Florkowski, B.Bruckner and I.Schonhof (2002), ‘Willingness to Pay for Environmental Practices: Implications for Eco-Labeling’, Land Economics 78(1), 88–102. Statistics Canada 1992. Family Expenditure in Canada, Cat. No. 62–555, Ottawa. Statistics Canada 1996. Family Food Expenditures in Canada, Cat. No. 62–554, Ottawa. Swallow, S. and R.Sedjo (2000), ‘Eco-Labeling Consequences in General Equilibrium, A Graphical Assessment’, Land Economics 76(1), 28–36. Swallow, S. and D.Wear (1993), ‘Spacial Interactions in Multiple-Use Forestry and Substitution and Wealth Effects for a Singe Stand’, Journal of Environmental Economics and Management 25, 103–120. Teisl, M., B.Roe and R.Hicks (2002), ‘Can Eco-Labels Tune a Market? Evidence from Dolphin-Safe Labeling’, Journal of Environmental Economics and Management 43, 339– 359. Thøgersen, J. (2000), ‘Psychological Determinants of Paying Attention to Eco-labels in Purchase Decisions: Model Development and Multinational Validation’, Journal of Consumer Policy 23, 285–313. Vincent, J. and C.Binkely (1993), ‘Efficient Multiple-Use Forestry May Require Land Use Specialization’, Land Economics 69(4), 370–376. Vossenaar, R., S. Zarilli and V. Jha, eds. (1997), Eco-Labeling and International Trade Basingstoke, UK: Macmillan.