UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Wilderness preservation and protection of old-growth forests Watson, Victoria G 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1994-0651.pdf [ 3.05MB ]
Metadata
JSON: 831-1.0087549.json
JSON-LD: 831-1.0087549-ld.json
RDF/XML (Pretty): 831-1.0087549-rdf.xml
RDF/JSON: 831-1.0087549-rdf.json
Turtle: 831-1.0087549-turtle.txt
N-Triples: 831-1.0087549-rdf-ntriples.txt
Original Record: 831-1.0087549-source.json
Full Text
831-1.0087549-fulltext.txt
Citation
831-1.0087549.ris

Full Text

WILDERNESS PRESERVATIONAND PROTECTION OF OLD-GROWTH FORESTSbyVICTORIA G. WATSONB.Sc.(Agr.), The University of British Columbia, 1991A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Department of Agricultural Economics)We accept this thesis as conformingto the required standardTHE UMVERSITY OF BRITISH COLUMBIAJune 1994© Victoria G. Watson, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)___________________Department of (O[V2,fl/M)The University of British ColumbiaVancouver, CanadaDate frDE-6 (2/88)11ABSTRACTThis research on the benefits of increased wilderness preservation has entailed thedevelopment of contingent valuation surveys to elicit consumer’s WTP for the provinceof British Columbia. The study came about after the Protected Areas Strategy (PAS)proposed an additional 6% of B.C.’s land base be set aside for protection. Two surveyswere used, one survey was distributed province-wide, while the other was issued to thirdand fourth year university students in both land use and forestry economics.A dichotomous choice format was chosen as the most appropriate approach due toits simplistic nature and its success in previous studies. Similarly, a logistic model wasapplied to calculate the probability of a person agreeing to pay to a pre-determined offeramount.The results of the province-wide survey indicated that respondents valued additionalwilderness protection in British Columbia at $371.34 per household per year.Aggregating this amount to include all B.C. households yielded a value of $484 millionper year. The results of the classroom survey showed that the respondent’s WTP was$326 per year for a total of $716 million when aggregated for the whole adult populationin B.C. The differnce between WTP values between surveys is partly due to the fact thatthe WTP for the classroom survey is on a per person basis while the estimated WTP forthe provincial survey is on a per household basis. Similarly, the province-wide surveyincluded individuals of all educational backgrounds while the classroom survey includedonly individuals with post secondary education.111Finally, while the Government of B.C. has decided to increase the level ofwilderness protection to 12 percent, the average desired level of protection for bothsurveys used in this paper was 10.75 percent.Abstract.List of TablesList of Figures .AcknowledgementChapter 1: Introduction1.1 Background1.2 Problem Statement1.3 Objectives1.4 Research ProcedureTABLE OF CONTENTSivChapter 2: Review of Biodiversity2.1 The Importance of Biodiversity2.2 Critical Species2.3 Issues in Valuing Biodiversity.2.4 UncertaintyChapter 4: Literature Review4.1 Introduction4.2 Contingent Valuation Method4.2.1 Background4.2.2 CVM Studies4.2.3 Some Critiques of Method4.3 Alternative Methods222222243436Chapter 6: Methodology6.1 Contingent Valuation 49iiviiviiiix13452.5 Biodiversity and B.C.’s Old-Growth Forests68101315Chapter 3: Protected Areas: Wilderness Areas3.1 Overview of B.C.’s Land Base3.2 Protected Areas3.3 Watershed Protection191920Chapter 5: Theoretical Considerations5.1 Background5.2 Theoretical Model39456.2 Closed-ended Approach (Dichotomous Choice).. 49VTable of Contents cont.BiasesProtest ResponsesWTP ModelWelfare MeasuresSurvey Instrument DesignImplementation Procedures6.8.1 The Study Area6.8.2 The DataChapter 7: Results and Implications7.1 Participation Rate7.1.1 Mailout Survey7.1.2 Class Survey7.2 Summary Statistics7.2.1 Mailout Survey7.2.2 Class Survey7.3 Zero Bids and Protest Responses7.3.1 Mailed Survey7.3.2 Classroom Survey7.4 Attitudes and Opinions About Wilderness7.4.1 Province-wide Survey7.4.2 Classroom Survey7.5 Willingness-to-Pay for Increased Protection7.5.1 Provincial Survey7.5.2 The Classroom Results7.6 Biases7.7 Model Specification: Functional Form7.8 Welfare Estimates7.8.17.8.26.36.46.56.66.76.85253535658616161of B.C. ‘s Wilderness Areas6363646464666666676868717474778082848487Integrating Under The C.D.F and Other ApproachesCensored Logistic Regression Approach7.8.2.1 Removal of Bids 917.9 Breakdown of WTP Based on Desired Level of B.C.’s Wilderness Areas 927.10 Aggregate Willingness-to-Pay 93Chapter 8: Summary and Conclusions8.1 Summary 958.2 Conclusions 968.3 Future Recommendations 98viTable of Contents cont.References 100Appendices:Appendix la- Provincial Survey 107Appendix lb - Classroom Survey 113viiLIST OF TABLES7.1 Response by Region 647.2 Statistical Means 657.3 A Description of The Variables Used in The Regression Analysis 757.4 OLS Regression Estimates 767.5 Logit Estimates of The Probability of Answering “Yes” to PreservingX% of B.C.’s Land Base 797.6 Logit Regressions Using Various Functional Forms 837.7 Estimated Mean Willingness-to-Pay Using Cameron’s CensoredLogistic Regression Approach 897.8 Results of Removing The Three Highest Bids 917.9 Average WTP ($) for Each Level of Protection 937.10 Mean ($) WTP Based on Level of Protection 93vii’LIST OF FIGURES5.1 Summary of Attributes of Old-Growth Forests Mentioned byRespondents in Their Working Defmitions 415.2 Percentage of NPS Units With Species Lists For The EntireNPS Unit And Specifically For Old-Growth Forests 425.3 Summary of Natural And Human-Caused Disturbances AffectingOld-Growth Forests 437.1 (a) Provincial Attitudes Concerning Old-Growth Forests in B.C 69(b) Provincial Survey - Breakdown of WTP Values Based on Attributes 707.2 (a) Classroom Attitudes Concerning Old-Growth Forests in B.C 72(b) Class Opinion Questions 737.3 Percentage of Respondents Who Answered “Yes” to the Bid Amount 86ixACKNOWLEDGMENTSFirst I would like to thank my committee chairman, G.C. van Kooten, forintroducing me to the field of resource economics and for his encouragement andsupervision during my graduate career. Chris Gaston deserves special thanks for readingmy thesis, providing me with statistical guidance and correcting statistical errors in mypaper.Many people have provided practical support during the creation of this work. Inparticular, I thank Gwynne Sykes for her invaluable services in the preparation andorganization of the surveys used in my thesis. I also thank Wang Sen for his help in thethe endless task of putting the questionnaires together. I thank Kristy McLeod for readingmy thesis and providing me with valuable insight. Rita Athwal who shared herknowledge and helped with the programming of the model. In addition, I thank LoneSrivastava who was most helpful in giving me ideas for my defense. Specifically, I amin debt to Victor Gaspar who was an immense help in the final preparations for mydefense and for lending a hand whenever it was needed.I would also like to thank all those who participated in the province wide andclassroom surveys. Without their help this thesis could not have been completed. Allthose who allowed themselves to be tested on in the begining are also thanked.1CHAPTER 1INTRODUCTION1.1 BackgroundPreservation of species has been an ongoing concern for many years. Since manyspecies have no market value because they are not traded in the marketplace, thereremains controversy about how to value these “non-market” goods. As E.O. Wilsonwrote:“Today as human populations expand and alter the natural environment, they arereducing biological diversity to its lowest level since the end of the Mesozoic era,65 million years ago. The ultimate consequence of the biological collision arebeyond calculation and certain to be harmful.. .it is a potential source for immenseuntapped material wealth in the form of food, medicine and other commerciallyimportant substances” (Wilson 1989, p. 108).There are several factors that contribute to the loss of biodiversity. Conditionssuch as climatic change and natural selection has resulted in the loss of some speciesthrough extinction. However, the more recent primary cause of loss of species has beenthe destruction of the ecosystems in which these species exist, due to human activity (e.g.,increasing demand for housing and other consumer goods, and thus encroachment ontoagricultural and forest lands).In British Columbia, harvesting of old-growth forests due to their high commercialvalue has in fact raised concern over the loss of certain species. Old growth can be2defmed in both ecological and non ecological terms. However, it is the latter of whicheconomists are able to evaluate in dollar terms. Aside from the many benefits resultingfrom preservation, such as intrinsic and watershed values, old-growth forests alsoconstitute an ecosystem where many species reside. Destruction of these forests forcommercial use inevitably destroys some of the biota of the species.Tn 1991, 81,500 forest workers were employed in B.C.’s forest industry(manufacturing and processing facilities) with an estimated total annual pay and benefitsof $4.5 billion dollars (Government of B.C. Statistics 1992). On average, 66% of B.C.’s94.8 million hectares (ha) is forested and approximately 6.5% (6.2 million ha) of B.C. iscurrently protected as parks or other wilderness areas. However, due to the 1991Protected Areas Strategy, the percentage of preserved wilderness areas is expected toroughly double.Concern over the loss of biodiversity due to harvesting of old-growth timber in thePacific Northwest has shown up in literature for quite some time (e.g., and has recentlycome into focus as a result of the highly-publicized Northern Spotted Owl controversy.The continuing debate between environmentalists and timber companies has raised thequestion of whether “to preserve or harvest” old-growth timber. It is impossible to answerthis question without evaluating the costs and benefits associated with each decision. Thepurely market costs and benefits are relatively easy to determine, primarily since the majormarket benefit, timber, is traded in the marketplace (i.e., costs being the opportunity costsof preservation). However, for goods not traded in the marketplace, or non-market goods,3the estimation of costs and benefits present economists with a challenge.It is a difficult task to determine the economic value of wilderness preservationwhen there does not exist any means to compare these values to commodities traded inthe market. Often public goods have intrinsic value, as well as watershed value, amenityvalue, recreational value, et cetera. Nevertheless, these values are often embedded in anindividual’s utility function and economists try to elicit these values through both indirectand direct measures, and attempt to determine the economic values which under thesegoods and services.1.2 Problem StatementBritish Columbia is Canada’s leading lumber producer and exporter. Only 54% ofB.C.’s land base (94.8 million ha) is designatd productive forestland, and about half of thisis suitable for timber harvesting. The Government of British Columbia has proposed anumber of initiatives that may reduce the amount of timber harvesting activity. Inparticular, the province’s 1991 Protected Areas Strategy (PAS) has increased the currentamount of wilderness protected in: regional and provincial parks; conservation areas;recreation areas; wildlife sanctuaries; and ecological reserves, from 6.5% of the land baseto 12%.Protection of old-growth forests cannot be separated from the larger issue ofwilderness protection. Old-growth can only be protected in contiguous areas thatconstitute fmancially valuable old-growth timber, other financially valuable trees that are4not considered old growth, timber that is uneconomical to harvest, and areas that arebarren (e.g., glaciers and mountain tops). Old growth constitutes 4-12% of an areaprotected under the PAS.The costs of protecting old-growth forests are obvious in economical terms (e.g.,lost jobs, community stability, environmental effects of using non-wood substitutes, andhigher prices to consumbers for forest products) but are more difficult to identify andmeasure for the benefits of preservation. Thus wilderness protection is a controversialissue: preserve wilderness at all costs or to harvest all trees regardless of theirclassification. In doing so, one is risking the chance of losing some of the criticalspecies both known and unknown to man.1.3 ObjectivesThis thesis will consider the problem of the loss of biological diversity as a resultof increased timber harvests of old-growth forests in B.C. In particular, the focus will beto determine the non-timber benefits (value) of preserving old-growth forests and how thisvalue changes at the margin. For example, what are the marginal benefits of preservingan additional unit of forest stands?I will accomplish the aforementioned tasks as follows:1) It is important to define what is meant by old-growth forests. What are theattributes/benefits that make up old-growth forests? How does the Ministry of Forestsdefme old-growth forests? What constitutes the bundle old growth? Does it exist?52) A purpose of the study is to determine the maximum amount society is willing to payto preserve old-growth forests in B.C. given the above attributes and thus find a value forold-growth forests for biodiversity objectives.1.4 Research ProcedureThis paper is structured in the following manner. Chapter 2 provides basicinformation on the importance of biodiversity and its relationship with old-growth forests.A discussion of the newly proposed doubling of B.C.’s protected wilderness areas by theProvince’s Protected Areas Strategy, the basis to which this study is centred, is presentedin Chapter 3. Chapter 4 provides a review of similar studies and previous results withwhich to compare the findings of this paper. Chapter 5 presents the theoreticalbackground for developing the empirical models used in this study. The methodologyand description of the models used in this paper is presented in Chapter 6. Chapter 7provides the results as well as a discussion of the quality results, and finally, summary,conclusions and future recommendations make up Chapter 8.6CHAPTER 2REVWW OF BIODIVERSITY2.1 The Importance of BiodiversityA common definition of biodiversity is as follows:“The full range of genetic diversity (species, subspecies and distinct biologicalpopulations of plants and animals) as well as the full variety of ecosystems inwhich plants and animals occur” (Ledec and Goodland 1988, p. 6).There are six main threats to biodiversity that have been identified by the GlobalBiodiversity Strategy (WRI/1UCN/LJI’.EP, 1992). These threats are as follows:1) habitat loss and fragmentation due to encroachment onto wildiands and increaseduse of natural resources;2) introduced species that result from one species not native to the area originally butwhich now plays a detrimental role in the ecosystem for some species;3) over-exploitation of plant and animal species due to clear-cutting and otherpractices not properly controlled;4) pollution of soil, water, and atmosphere that come about from practices notproperly governed by provincial or federal authorities;5) global climatic change as a consequence of increased greenhouse gases which tendto result in increased global temperatures; and6) industrial agriculture and forestry that has come about as a result of plant and gene7breeding programs which have enabled farmers and foresters to only grow certainplants and trees, or raise only certain livestock due to their enhanced geneticabilities (i.e., higher yielding plants and animal).Not only do we need to preserve species themselves, but, more importantly, weneed to preserve the ecosystems in which they live. For example, in the tropical forestsof the Amazon (where perhaps the greatest diversity of species can be found)deforestation is destroying many ecosystems. The reason is that the Amazon region isquite poor relative to the rest of the world and forests are burned so land can be used forother more productive purposes. Policies prohibiting the destruction of habitats mustincorporate the benefits of preserving species, and the benefits must outweigh the costsif there is to be an incentive to preserve. Unfortunately the benefits of many species arenot yet known or are difficult to calculate.Although the benefits of species do not have commercial value specifically, thisdoes not imply that they are not valuable to humans in other terms. For example, over40% of all prescription drugs in the U.S. contain one or more drugs that originate fromwild species (generating millions of dollars in sales) (McAllister 1992). More recentlyare the discoveries of anti-cancer drugs (e.g., rose-periwinkle, taxol). Obviously, thesediscoveries are essential to the well being of man and the loss of these species could beimmeasurable. Furthermore, species diversity can bring us such satisfaction in theknowledge that such species or a particular state of the environment exists, known as8existence value, or simply that the species will be present for future generations to enjoy,known as bequest value. Furthermore, there is value just in knowing that there is theoption of using the amenity in its given state even if it is never actually used. This iscommonly referred to as option value. Bequest, existence, and option values are allconsidered to be preservation values of a particular amenity. For example, people areoften willing to pay considerable amounts of money just to ensure the Whooping Cranewill be around for future generations to enjoy (bequest value) (i.e., up to $149/yearBowker & Stoll 1989).2.2 Critical SpeciesAlthough it would be desirable to conserve all species, it is impractical andimpossible. Thus, there is an immediate need to identify those species that areendangered and the possible consequences if they are lost forever. It is difficult forecologists to determine which species these are, especially when there is not enoughinformation on every species to identify the critical ones. Therefore, biologists haveclassified some of the critical species into three categories: keystone species, indicatorspecies and flagship species.Keystone Species‘Some animals and plants hold central positions in the meshwork ofinterrelationships that forms a community; if these species are selectively removed, the9community structure begins to collapse” (Malcolm 1988 p. 240). The loss of one of these“key species implies that any loss in future benefits should include losses accruing tothose species that ceased to exist as a result of their extinction. Species on which othersdepend, therefore, have contributory value in addition to their direct uses (Norton 1988).Some species have roles as prey, predator, symbiont or competitor that accentuate theirecological importance beyond what one might predict from their abundance or biomass.In many complex communities, it may be difficult to identify those species that arekeystone species.Indicator SpeciesA particular community is characterized by its most typical members, indicatorspecies that are rarely found in other communities in the vicinity. The term indicatorspecies refers to species that have such ecological tolerance that their presence or absenceis a good indication of environmental conditions (Malcolm 1988). Indicator species havealso been used to warn of environmental dangers. For example, the canary was oftenused to warn miners of methane gas leakages in the mines, or deaths of the brown pelicanprovided information as to the danger of certain pesticides. More importantly, however,indicator species can be used to obtain knowledge of the requirements for specific featureswithin a forest and thus provide a basis for determining what aspects of a forest need tobe preserved for the survival of the species (Malcolm 1988).10Flagship SpeciesUnfortunately, when it comes to endangered species some species attract moreattention than other animals. For example, if the rat were to suddenly become endangeredof extinction, it is unlikely that many would consider this a critical situation (although inSouthern California the rat is used to prevent development). However, the real life threatof extinction of the panda bear or the bald eagle caused quite a stir among the public.It is species like the panda that are considered flagship species—a species that can arouseconsiderable public support and indirectly facilitate the wise conservation of a wholegalaxy of species (Malcolm 1988). Conservationists often use flagship species to promotethe need to preserve the habitat in which these species live. The spotted owl of thePacific Northwest is an excellent example of the controversy surrounding the need toprotect the ecosystem on which these species depend upon, namely, old-growth forests.2.3 Issues in Valuing BiodiversityEconomists in particular have determined numerous ways in which to value publicgoods for which there is no price information available (e.g., preservation value ofbiodiversity). Two methods in particular include implicit pricing methods and contingentvaluation methods. Implicit pricing is defmed as an indirect method in which the unpricedamenity of interest can be purchased as a complement to, or a characteristic of, someordinary unbiased good (Randall 1989). Examples of this type of method are the wellknown travel cost method and the hedonic pricing analysis (defined later in Chapter 4).11A more common and more direct approach is the contingent valuation method which willbe discussed in Chapter 4.Despite the fact that environmentalists and economists both try to achieve the samegoal (i.e., preserve biodiversity), it is the difference in techniques that causes conflict.Environmentalists are frequently opposed to assigning monetary value to biodiversity.The next section deals with this dilemma.Is putting a value on biodiversity ethical?Many environmentalists want to preserve biodiversity, but they are opposed to theidea of putting a “dollar value” on biodiversity. David Ehrenfeld, citing a paper by Clark(1973), in his article “Why Put A Value On Biodiversity?”, suggests that putting aneconomic value on a species may in fact do the opposite of what was intended. Usingthe example of blue whales in Japan, Clark concluded that economically it was morefeasible to kill every blue whale in the ocean as fast as possible and reinvest the profitsin growth industries rather than to wait for the species to recover to the point where itcould sustain an annual catch (Ebrenfeld 1988).Another example of the problems associated with putting an economic value on anon-market good is that by Ludwig and Conrad (1991). Ludwig and Conrad (1991)developed an economic model to determine the optimal amount of old growth forest thatshould be preserved from an economic standpoint, when nontimber benefits such ashabitat, recreation and watershed protection are a function of the area of old growth12remaining in a region. The results of their analysis indicated that, when opportunity costsof preserving old-growth forests are taken into account, it is better to harvest the old-growth timber than to preserve biodiversity. The reason for this is that the majority ofthe opportunity costs were composed of the net revenue from harvesting. These resultsseem to confirm Ehrenfeld’s fear that if a monetary value is placed on a non-market good,it will often be more beneficial to harvest the forest and reinvest the money in a highinterest investment.Ehrenfeld also warned of the chnger of using pharmaceutical value as a means tovalue biocliversity. He claims that reliance on this to promote conservation is only atemporary measure since not far down the road (and currently in progress) drugs whichwere once thought of as depending on the existence of that species from which it wasderived (e.g., the drug taxol) can be reproduced, at a lower cost, “... by computermodelling of the molecular structure, followed by organic synthesis in the laboratory usinga host of new technologies, including genetic engineering” (Ehrenfeld 1988, p. 213).In defence of economists, however, it should be noted that economists are not justconcerned about the existing pharmaceutical value, but rather the “potential”pharmaceutical value of species not yet discovered as cures for diseases. The opportunitycosts of extinction of a species today that may have potential value tomorrow can beextremely high. It is this uncertainty that causes great concern among economists (thisis discussed in more detail later). To summarize, Michael Hanemann (1988) explains theimportance of placing economic value on biodiversity:13“Environmental economists are interested in markets not because they wantto use market prices to multiply something but because they are interestedin measuring the preferences of individuals and ascertaining their trade-offsbetween environmental resources and money or conventional marketcommodities... economists have come to rely quite extensively on simulatedmarkets, or their analogues, in which individuals reveal their preferencesthrough interviews or experimental games involving trade-offs betweenmoney and environmental outcomes. Moreover, when they do analyzeactual markets, economists are not interested in the market prices per se but,rather, in the patterns of selection and the types of preferences that theseimply” (Hanemann 1988, p. 197).Clearly, economists are not trying to put the value of biodiversity into dollar terms perSe, but instead try to illicit its value in terms of how people view it. Thus, the economistuses techniques, such as cost-benefit analysis or implicit pricing, to determine whether ornot to harvest old-growth forests or to preserve species like the Spotted Owl.2.4 UncertaintyWhen arguing the need to conserve a certain species that is on the verge ofextinction, people will often look at the value of the species as it stands at present.Although this is not incorrect or invalid, one should note that it is not thorough. The truevalue of a species should be the value of the species today as well as the “potential” valuein the future. What this implies is that there are species that we know about but whosepotential use is still unknown. If the species is extinct, it is possible that we are losinga valuable resource. This can be further extended to the destruction of ecosystems uponwhich many species depend for survival, If certain ecosystems are destroyed as a resultof deforestation, many species known to man and, perhaps more importantly, those14species not yet discovered may become extinct. The uncertainty as to the possible useof these species is of great concern to economists and ecologists. Once a species is gone,it is gone for good. Uncertainty coupled with irreversibility provide a dilemma for thoseinvolved in policy making. Perhaps one of the species yet to be discovered, whichbecomes extinct, could have become a cure for cancer or another deadly disease. Alsoof importance are the keystone species. As mentioned earlier, keystone species are thosespecies which are critical for the survival of other species. The problem arises when itis not known for certain which species are keystone species. By accidently removing onespecies from an ecosystem, inadvertently several species may also become extinct. Therisks involved are extremely high especially when dealing with irreversibilities.Furthermore, as time passes more and more information may come available:“Because the passing of time brings information about the consequences ofpresent actions, there is a premium on actions that preserve the flexibilityto exploit this information. If a current situation is physically oreconomically irreversible, that flexibility is abandoned. To the extent thatdecision makers disregard the potential value of future information, theywill systematically undervalue policies, such as conservation programs, thatmaintain flexibility and preserve options for future action” (Hanemann 1988,p. 195).Moreover, option values and quasi-option values (defmed below) are directlyrelated to uncertainty. Option value can now be defmed as “the “premium” thatconsumers are willing to pay to avoid the risk of not having something available in thefuture. Munasinghe (1992) further goes on to define quasi-option value:“Quasi-option value is the value of preserving options for future use in theexpectation that knowledge will grow over time. If a development takes15place that causes irreversible environmental damage, the opportunity toexpand knowledge through scientific study of flora and fauna is lost.Uncertainty about the benefits of preservation to be derived through futureknowledge expansion leads to a positive QOV. This suggests that thedevelopment should be postponed until increased knowledge facilitates amore informed decision, If information growth is contingent upon thedevelopment talcing place, which is unlikely in an environmental context,then QOV is positive when the uncertainty regards the benefits ofpreservation, and negative when the uncertainty is about the benefits ofdevelopment” (p. 29).Thus, in the case of irreversibility it is not practical to go ahead with a decision andthen wait to see what information comes available, but rather it is more appropriate towait for information to arise before going ahead with the decision.2.5 Biodiversity and B.C.’s Old-Gniwth ForestsBritish Columbia is particularly at risk in terms of loss of biodiversity. There hasbeen much controversy as to the forest practices concerning old-growth forests in B.C..The most recent and perhaps the most controversial would be that of the ClayoquotSound. The reason that people are so concerned is that the amount of biodiversity thatexists in old-growth forests is particularly high.Old-growth forests contain trees over 200 years of age’; are mainly undisturbed (noharvesting of timber has occurred); contain a variety of large live trees, fallen dead treesand standing dead trees (snags); are often characterized by large canopies (whereby thebranches of the trees overlap to form an umbrella-like effect). The canopy provides‘Old growth is difficult to defme since the age criterion varies depending on species, site, etc. (Focuson Resourses & Our Environment by Government of B.C. 1992)16protection for a large number of species and functions as habitat for others. Both deadand living trees are capable of supporting a wide variety of species, many of which aredependent on old-growth forests for survival. For example, there are approximately 80species of terrestrial vertebrates in B.C. which are strongly identified with or reliant onold growth (Ministry of Forests 1992). Thus, old-growth forests are home to manyspecies, and destruction of these forests could lead to their extinction.British Columbia also has the greatest number of wildlife species in Canada. Inparticular, 70% of breeding birds, 72% of terrestrial mammals, 49% of amphibians, and41% of reptiles in Canada are found only in B.C. In terms of forest dwelling wildlife,77% of the birds are forest dwelling while 81% of mammals live in forests. Furthermore,24% of forest dwelling wildlife are strongly dependent on old-growth forests (GovernmentStatistics 1992 and Bunnell et al. 1991). This is not surprising considering the structureand composition of old-growth forests. Old-growth forests also provide a vast array ofbenefits to users (timber harvest, hunting and other recreation), visitors (wildlife viewing),and even non-users (e.g., those who benefit from medical cures, education and research).In contrast second-growth forests (regenerated forests) have developed after anarea was logged or naturally denuded (e.g., lightning fire). Generally, these forests arecharacterized by younger, more sparse, trees with little or no canopy2. Furthermore, thereis little in the way of dead trees, live or standing, although, second-growth forests can be2Not tme for all regions, in some areas it is difficult to distinguish between old growth and second-growth forests due to the different growing conditions in each region.17managed to include these structures.In B.C., forest structure and composition varies from region to region. Someregions may comprise trees that have a higher market value due to the size andcomposition of the tree. In the Coastal regions, trees often grow to great ages and sizes(1000 years is not uncommon). Individual dead trees (due to old age) often remainstanding in climax forests composed of tolerant species such as western and mountainhemlock, western red cedar and Pacific silver fir. In general, the Coastal region of B.C.has some of the lowest biological risks for growing timber in North America. Overmatureforests are fairly resistant to risks of loss due to insects, disease or fire. This could be inpart due to the fact that Coastal forests are characterized by high rainfall, long growingseasons, and thus a low risk in growing timber.Second-growth forests are also remarkably free from insect and disease problems.Nevertheless, there are large differences in the net returns by site type and by species.Cedar, hemlock and spruce are the higher-valued species and the Coastal region of B.C.is the prime location of high volume sites with low restocking costs.In contrast to the Coastal forests, the Interior forests and boreal forests of B.C. havemuch greater biological risks. Furthermore, the colder climate, temperature and stresseson trees are more extreme, and thus major fires are more common. In the Interior, treestend to be even-aged due to blowdown, insects and catastrophic fires. Second-growthforests also tend to be more susceptible to biological risks. The Interior, in general, hasrelatively little intensive silviculture due to the fact that stand development and yield18prediction of cutover are very uncertain. Since the stands in the Interior do not reach thegreat sizes as on the Coast, they do not command the same market prices, nor do thestands reach the same high yields as on the Coast. Thus, the stands in the coastal regionsof B.C. are much more valuable to producers than those in the Interior forests. However,since the physical characteristics of the Coastal forests are to some, more appealing, theyalso are more valuable in other terms. Therefore, the costs of preserving old-growthforests are higher on the Coast than they are in the Interior, and consequently themotivation for controversy between conservationists and timber companies is established.19CHAPTER 3PROTECTED AREAS: WILDERNESS AREAS3.1 Overview Of B.C.’s Land BaseBritish Columbia is comprised of 94.8 million hectares (ha). Crown provincial landmakes up 86.4 million ha (91%), while 80.7 million ha (85%) are considered Provincialforests (73.8 million ha in timber supply areas and 6.9 million ha in tree farm licences).Roughly 43.3 million ha (45%) of crown provincial forest land is considered productivein timber supply and tree farm licences. Only about 22.6 million ha (23%) of theprovincial forests are suitable for timber harvesting, and only 238 969 ha (.25%) of thisis actually harvested in a given year. Federal land makes up about 1% of B.C.’s total landbase, while the remaining 8% of the land base is held under private ownership.3.2 Protected AreasProtected areas can be defined as land formally designated for conservation orrecreation purposes. Some examples of protected areas include: Provincial and Nationalparks, Recreation and Wilderness areas, Wildlife Sanctuaries and Ecological reserves, etc.Approximately 6.6% (6.3 million ha) of B.C.’s land base is in protected areas. Currentgovernment policy is to increase the amount of the provincial land base protected in parksand ecological reserves to 12% (12.5 million ha) of the land base by the year 2000. Thegovernment has devised a strategy whereby new possible parks and wilderness areas will20be considered for protection. In 1992, 23 new parks or wilderness areas were designatedas protected areas, and additional areas are being assessed. The protected areas programset up by CORE (Commission on Resources and the Environment) focus on provincialparks and recreation areas as well as wilderness areas. Provincial parks and recreationareas protect representative landscapes and special features for conservation andrecreation. Wilderness areas protect wilderness for conservation and recreation whilepermitting compatible, limited resource use.3.3 Watershed PmtectionA watershed is a drainage area with a boundary defmed by a height-of-landupstream and a significant hydrological feature downstream such as a lake or streamconfluence (Ministry of Forests 1992). Watershed benefits include flood control,improved water quality due to a reduction in sediment loads in reservoirs, as well as animpact on fishery. An undeveloped watershed is a watershed in which no more than twoper cent of the area has been developed by human activity (i.e., timber harvesting, roadsand mines) (Ministry of Forests 1992). An unprotected watershed has less than 10 percent of its area protected within the boundaries of a national park, provincial park,ecological reserve, recreation area, or wilderness area. In contrast, a protected watershedhas the entire land area protected within the above areas. An inventory completed by theMinistry of Forests identified a total of 508 undeveloped watersheds in the province (<5000 ha). This was in addition to 58 undeveloped watersheds (< 1000 ha) on Vancouver21Island. In B.C., less than 10% (47) of undeveloped watersheds are totally protected,while another 93 are within parks and wilderness study areas. Approximately, 85% (428)undeveloped watersheds in B.C. are still unprotected with the remainder of watershedspartially protected. On Vancouver Island, 36% (21 of 58) of the undeveloped watershedsare protected with another 14 in park and wilderness study areas (Ministry of Forests1992).22CHAPTER 4LITERATURE REVIEW4.1 IntroductionThis chapter introduces previous studies and literature on topics relevant to the oneat hand. Literature on methods (contingent valuation, conjoint analysis, etc.) used byeconomists and scientists in the past will be reviewed as well as literature on old-growthforests and biodiversity. The purpose is to gather information to be used in this research.A critique section will be included to assess the primary method outlined in this chapter,and to make one aware of some of the problems that need to be worked out to have asuccessful outcome.4.2 Contingent Valuation MethodThere are many methods with which to determine the value of a non-market good.Perhaps the most common form of valuation is the Contingent Valuation Method (CVM).This technique uses surveys to gather information on the respondent’s willingness-to-payor demand for a particular environmental good. The method requires the respondent tostate how much they are willing to pay for increments of decrements for the good inquestion. However, as will be shown later, there are many drawbacks with this approach.4.2.1 BackgroundThe contingent valuation method (CVM) is used to determine an individual’s23willingness-to-pay (WTP) for more or to prevent the loss in the level of an unpriced good.The same approach could be used to calculate an individual’s willingness to acceptcompensation for a decrease/loss in the level of a public good. This approach involvesobtaining information on a consumer’s preferences by using questionnaires. The questionsposed by the surveys are designed to calculate the consumer’s WTP. Often, the subjectis given a certain scenario (contingency or hypothetical situation) regarding the loss orgain of an environmental good. They are then asked to respond to the situation byindicating how much money (usually stated in the survey in the form of bids) they arewilling to pay to prevent the loss or to obtain more of the good. Opponents to thismethod argue that the problems which arise from the design of the questionnaires maybe misleading or may not achieve the desired goals of the researcher. Thus, the valuesmay not give an accurate account of the consumer’s preferences. Some of the reasonsinclude:1) Starting point biases=occurs when the starting value of the bids may be too lowor too high, people anchor their willingness to pay on this amount or use this asan approximation of the true value of the good;2) Strategic bias=the subject may be biased towards the topic and may try to sway theresults in his/her favour;3) Information bias=the respondent may not be familiar with the subject in question;4) The respondent may feel that his/her opinion will have no bearing on the resultsand thus may not give a meaningful opinion;245) The respondent may object to the vehicle of payment, or may feel it is unethicalto place values on environmental goods; and6) The respondent may be trying to please the interviewer by giving answers thathe/she thinks the interviewer wants to hear.7) In general, consumers have not had to price non-market goods and will have noconcept of where a market price may be.8) Values for various similar non-market goods may not be additive.Some of these problems can be overcome through careful design of surveys.Depending on the situation, CVM may be the only available approach to calculating thevalue of unpriced goods.4.2.2 CVM StudiesThere have been many studies aimed at calculating economic values for non-marketgoods using the Contingent Valuation Method. Stevens et al. (1991) examined thevalidity of CVM for estimating the existence value of four wildlife species recentlyintroduced or reintroduced to New England (Atlantic salmon, coyote, bald eagle, and wildturkey). In the study, two separate CVM mail surveys were sent out (one toMassachusetts for the restoration program of Atlantic Salmon and one to New Englandfor valuing bald eagles, wild turkeys and coyotes).The surveys were constructed so as to ask questions that would reveal how each25individual valued the species. The questions were designed to analyze each individual’sdecision-making process and to ascertain whether or not the respondents were consistentin their beliefs.The results of the surveys indicated that people would be willing to contribute acertain amount of money to the Atlantic salmon, bald eagle and wild turkey; however,“80% of survey respondents said that bald eagles, wild turkeys and Atlanticsalmon are important to them, but when confronted with contingentvaluation the majority refused to pay. They were either uncertain abouttheir valuation, believed that wildlife should not be valued in dollar termsor protested the donation payment vehicle. Moreover, most of those whowould pay exhibited behaviour which appears inconsistent with theneoclassical theory underlying the CVM” (Stevens et al., p.399).The results could be explained by the following:1) Most respondents were unfamiliar with the commodity being valued;2) Benefits were primarily viewed as existence values; and3) Many responses were a result of moral or ethical considerations.The results of their study suggest that the CVM may not provide a valid measurefor existence values. Therefore, a benefit-cost approach would not be an appropriatetechnique to value wildlife.Although Stevens et at. concluded that the CVM was not the most appropriatemeans of making decisions about the existence of wildlife, other studies have shown thatthis is not necessarily true. Mohan Munasinghe (1992), in a recent World Bank working26paper, summarized several studies aimed at estimating existence, option and bequestvalues for wildlife and endangered species using CVM. Some of the studies aresummarized below.One study looked at the option price and existence values of grizzly bears andbighorn sheep in Wyoming now threatened by human activity. The study was aimed athunters’ WTP for the right to hunt in a newly designated hunting area, and non-hunters’WTP for the knowledge that the animals would exist in the future. The results (from thehunters’ WTP) indicated that there was a positive relationship between the probability ofan increase in supply of the species and the option price, while the existence value for thegrizzly bear (from non-hunter’s WTP) was quite high compared to the existence value ofbighorn sheep (Brookshire, Eubanks and Randall 1983).Another study by Walsh, Loomis and Gillman (1984) sought to determine theoptimal amount of wilderness area to be protected in Colorado by determining the option,existence and bequest values (i.e., preservation values) for wildlife. Respondents allocatedtheir WTP into a fund for the sole purpose of protecting wildlife in the area. Further, therespondents were asked to categorize their contributions into: recreation use, bequestvalue, option value and existence value. Total preservation value was calculated as theamount left over once recreation use was subtracted from the total WTP for preservingwildlife. The results indicated that recreation use had a positive influence on optionvalue. Moreover, existence values were positively correlated with scenic amenities,ecosystems and biodiversity. Bequest values, which were constant regardless of the27amount of wilderness that prevailed, followed a priori expectations that people gamedsatisfaction from the knowledge that the species will be there for future generations toenjoy.Samples, Dixon and Gowen (1986) tested the hypothesis that, as more informationabout a particular species is disclosed, an individual’s WTP is significantly influenced.The study was divided into two sections. The first part of the study looked at how anindividual’s WTP changed when they were given more information on the species inquestion. The subjects were asked to give their WTP for the preservation of the humpback whale before, and then again after, a film of the whale was viewed. The resultsshowed a significant increase in the value of their bids. The authors concluded that partof this may have been attributed to the fact that the respondents had time to re-evaluatetheir decisions.The second part of the study consisted of respondents who, given a certain amountof money, were asked to allocate that money among different preservation funds. Eachparticular fund had a certain amount of information about a species (e.g., endangeredstatus, physical attributes). The results indicated that the more information that wasavailable about the species, the higher the allocation of monies to that species’preservation fund. The study concluded that people are more likely to contribute higheramounts to a preservation fund as more information about the species is released.All the aforementioned studies (with the exception of Stevens et at.) illustrated howCVM could successfully be used to determine preservation values of wildlife. This is not28to say that CVM is not without its shortcomings, however. It does suggest that, if propermeasures are taken, CVM could give credible results in appropriate situations for theeconomist to analyze (Munasinghe 1992).Contingent valuation studies have been readily used by economists to evaluate aproposed increase (or reduction) in the level of a non-market good such as wildernessareas. Pope and Jones (1990) used CV to examine four proposed increases in wildernessareas in Utah. Using an open-ended framework, telephone surveys were conducted on280 households in the state of Utah and an estimated WTP was obtained for each levelof proposed increase in wilderness areas. The average willingness-to-pay amounts foreach increase of 5%, 10% 15% and 30% were $53, $64, $75, and $92 per year perhousehold. These values were then used to derive an aggregate WTP for the whole stateof Utah by multiplying the average WTP values for each level of wilderness designationby the total number of households in Utah. The aggregate annual amounts ranged from$27 million for a 5% increase to $47 million for a 30% increase. Furthermore, theauthors found that the demand for increased wilderness areas levelled off at about 15%for the state. Thus, the value of setting aside an additional 15% on top of this was noteconomically advisable.In a study similar to the one at hand, Rubin, Helfand and Loomis (1991) used amail-out survey to determine the maximum willingness-to-pay to preserve the NorthernSpotted Owl in the Pacific Northwest. The survey was sent to residents in the state ofWashington. Of the 1,200 surveys distributed randomly throughout Washington, only 25329residents responded to the survey, but, only 216 surveys could be used in the analysis.The results indicated that individuals are willing to pay as much as $49.72 ($34.84 whenadjusted for factors such as education and income levels) to preserve the spotted owl.Although the results were not surprising, the authors did note some problems with thesurvey. The low response rate was an obvious problem, while the unusually higheducation and income level (as compared with society as a whole) may have contributedto survey bias. Generally, people with higher education and income levels often tend tosupport conservation practices.Likewise, a study by Hagen et al. (1991) sought to estimate the non-market benefitsof preserving old-growth forests and protect the spotted owl. The authors used adichotomous choice format for their nationwide mail out questionnaire. The study wentone step further than requesting a “yes” or “no” response. The individual was then askedto reveal his/her maximum and minimum WTP. The estimated values obtained from thesurvey indicate a willingness-to-pay of $86.32 per household per year with an upper limitof $144.28 and lower limit of $47.93. Aggregating these values for the nation gives anestimate of 8.3 billion dollars per year. This is considerably higher than the estimate of$1.5 billion per year obtained by Rubin et al. (1991). More surprising is that Hagen eta!. go even further as to discount the values into the future to arrive at an estimatedannual total of $215 billion over 30 years.A similar study analyzing the value of protecting forest quality (as opposed towilderness areas) was performed by Walsh et al. (1990). Their study involved30interviewing a random sample of households in the Colorado area. An averagewillingness-to-pay value, obtained using an open-ended approach, was calculated to be$47 per household per year. Only one quarter of this value could be attributed torecreation use value, while the remaining three quarters represented preservation values.Similar to the study by Pope and Jones (1990), the authors constructed a demand curvefor forest quality in Colorado. The results indicated diminishing marginal benefits witha maximum of 150 trees/acre and a value of $47 per household. An aggregate value forforest protection was estimated to be $56 million annually (for the state of Colorado).Reliability of the survey is expected to be quite high given the authors provided severalchecks for a variety of bias.A study by Hanley and Ruffell (1992) examined the variation in consumers surplusacross different forest types by placing values on the physical attributes of individualforests. The authors used both the hedonic travel cost method and the contingentvaluation approach. The characteristics obtained were used to help explain total visits toa particular forest. The study included two different CVM studies, one usingphotographs of forests varying in characteristics (e.g., water features, scenic views,recreational facilities, open sky, etc.), and the other using a bid curve analysis whererespondents were asked to bid the amount they would be willing to pay for the option tovisit the forest given the above attributes. In the photo analysis, respondents were askedto choose between different pairs of photographs, each differing by one characteristic.This itself posed problems for the analysis, since not all possible characteristics could be31represented in the photographs. The use of photographs helped to determine whichcharacteristics were important in deciding which forest to visit. However, for surveyswhere photographs were not used, the importance of certain attributes decreased. Theauthors concluded that this is because people visiting the forests had in mind an intendeduse.Many economists have compared different methods of estimating values of non-market goods. In particular, Sellar, Stoll and Chavas (1985) looked at three alternativemethods for valuing recreational boating at the Four Lakes in East Texas, namely, thetravel cost method and two forms of the contingent valuation method (open-ended andclosed-ended). The open-ended survey directly elicited the respondent’s WTP for the lakethey were visiting, whereas the closed-ended questionnaire obtained indirect estimatesbased on 10 alternative bid amounts ranging from $10 to $300. Results of the studyindicated that the open-ended framework gave the lowest WTP values given thatrespondents were probably unfamiliar with valuing a non-market good. The travel costmethod elicited the highest value, but the authors concluded that this was due to the factthat it was for the whole recreational experience and not just the boating experience asin the CV portion of the study. The closed-ended approach seemed to have the closestvalues of consumer surplus to that of the travel cost method. The authors concluded that,for valuing non-market goods, the closed-ended and travel cost methods were theappropriate tools for estimating consumer surplus.There have been numerous studies aimed at comparing different forms of CV32(Sellar, Stoll and Chavas 1985; Johnson, Bregenzer and Shelby 1990 ; Bishop andHeberlein 1979; Boyle and Bishop 1988; Kealy, Dovidio and Rocke 19881; Desvousges,Johnson, Dunford, Boyle, Hudson and Wilson 1992; and Milon 1989). More specifically,Kealy and Turner (1993) compared WTP values across two forms of CV, open-ended andclosed-ended (dichotomous choice) questions. The authors developed a test to analyzethe difference in willingness-to-pay estimates obtained from both open-ended and closed-ended formats for the same sample population. Furthermore, the authors applied aseparate test to examine the difference in estimates using both a public good and a privategood to determine if the attributes of the commodity had any influence on the difference.As in other studies, the results of the study indicate that open-ended questions tend to leadto significantly different results than the closed-ended questions. More precisely, theestimates obtained using the open-ended questions were considerably lower than thoseobtained using the closed-ended format. Regardless of this conclusion, the results did nothold for the private goods. In fact, there was not found to be any significant differencebetween estimates from the two alternative approaches. The reason for this seemed to bethat the private good was a more concrete good where respondents were more familiarwith the value of the good and there was likely to be less strategic behaviour involvedthan in the open-ended, pubic good scenario. The authors also concluded that therespondents may have been unfamiliar with the value of the public good and the fact thatthere may have been strong incentives for strategic behaviour for a public good.Furthermore, individuals may have not been accustomed to the format of the open-ended33questions as opposed to the simplicity of the closed-ended questions.Similarly, there have been many studies which have dealt with the appropriatemethod of estimating WTP values from dichotomous choice questions (Hanemann 1984;Johnson, Bregenzer and Shelby 1990; Ozuna, Jang and Stoll 1993; Duffield and Patterson1991; and Cameron 1988). Bowker and Stoll (1988) using dichotomous choice toestimate the non-market value of Whooping Cranes examined the difference in estimatesof willingness-to-pay across three alternative specifications, two different models ofestimation (logit and probit) for each specification, and compared welfare measuresobtained using both mean WTP and median WTP. Results of the analysis indicate thata logarithmic specification is superior in terms of goodness-of-fit to both a linear andshare specification. Both logit and probit models exhibited similar estimates throughoutthe alternative specifications suggesting that both models perform equally in dichotomouschoice formats. Median and mean estimates of WTP were significantly different amongthe different specifications. Mean WTP was considerably larger than the median WTPacross all specifications, a result that has been found in similar studies. However, themedian estimates were much more sensitive to the functional form used. The authorsconcluded that for their study the logarithmic functional form using the mean WTPprovided the best estimate of non-market values of whooping cranes on the basis ofstatistical fit and other considerations. Thus, caution and judgement should be taken whenchoosing a functional form or a welfare measure.A study by Cooper and Loomis (1992) examined the sensitivity of WTP estimates34to the alternative bid values. Using data obtained from previously completedquestionnaires, the authors tested the sensitively of the models by removing bids (highestbids, lowest bids, and then every other bid) and re-estimating the models. Whenwillingness-to-pay values were restricted to be nonnegative, the model was more sensitiveto the removal of the highest bids than the, unrestricted WTP model. However, theopposite is true when the lowest bids were removed from the data. The use of a restrictedmodel is more suitable when valuing improvements in a public resource, therefore; theresearcher should be more concerned about the effects of the highest bids in the bid spacethan the bids in the lower portion on the results. Their results signify the importance ofchoosing the bid amounts, and thus the sample design, carefully.4.2.3 Some Critiques of MethodKahneman and Knetsch (1992) sought to determine the validity of the contingentvaluation method; in particular, the authors examined the embedding effect ofsurveys/questionnaires used in the analysis.3 A telephone survey was conducted in thegreater Vancouver area and the respondents were asked questions pertaining to increasedavailability of equipment and trained personnel for rescue operations. Three differentgroups were asked variations of the questions. The questions varied from generalquestions on environmental services with the primary question to be examined embedded3The embedding effect is defmed by Knetch as the same good is assigned a lower value if WTP forit is inferred from WTP for a more inclusive good rather than if the particular good is evaluated on itsown.”35in the questionnaire, to questions aimed directly at the question under consideration. Theresults indicated the presence of an embedding effect. Consequently, respondents in eachsample were willing to pay the same amount regardless of the specific good in question.The authors concluded that people were purchasing moral satisfaction rather than actuallyconsidering the good in question, and thus the values obtained from a CVM analysiscannot be considered true economic values.Paul Slovic (1990) examined the role of preference reversals in contingentvaluation surveys. His results indicate that, when an individual prefers one object overanother object, the individual may reverse his/her preference depending on the method ofmeasurement. For example, in one of his studies he found that the respondents provideda higher WTP for an improvement in word processors than an improvement in air quality.However, when the same respondents were asked to value the two options, the majorityof the respondents reversed their preferences. Slovic concluded that, depending onwhether or not value is measured by pricing responses or by choice responses, arespondent’s preference will rely on the method of valuation. Furthermore, Slovicconcluded that, when using contingent valuation as a means of valuing an environmentalgood, one must determine the rationality behind the decision.McKillop (1992) recently illustrated the dangers of using CVM for evaluating howpeople value public goods. In particular, McKillop focussed on the use of CVM innorthern spotted owl studies such as the one by Rubin et al. (1991) (see section 4.2.2).McKillop argued that the respondents were not provided with a complete set of36information (i.e., respondents were told haff truths). For example the respondents weretold that the spotted owl depended heavily on old-growth forests for survival; however,they were not told that there is evidence that the spotted owl also nests in second-growthforests. The respondents were not informed of the major impact on the U.S. timberindustry, globally as well as locally (i.e., extreme job losses), nor were they informed asto the true cost of implementing habitat preservation for the spotted owl. Furthermore,McKillop criticized the Rubin et al. study for suffering many of the biases commonlyfound CVM studies, particularly hypothetical bias. McKillop argued that the amount therespondent is willing to pay is generally much higher than what they would be willing topay in an actual situation, if the respondent does not think they will actually have to paythey are often more generous. Similar results were found in studies by Bishop andHeberlein (1979) and Stevens et al. (1991) (see section 4.2.2).4.3 Alternative MethodsAn alternative approach to examining mulitattribute commodities is the pairwisecomparison approach whereby an individual is given a choice between two attributes andasked to choose one or the other (a modified version extends this approach by allowingthe individual to scale his response). Generally, the individual chooses between attributeA or B by stating which one he prefers. A modffied approach to this method was usedin a study by van Kooten et a!. (1986) to evalUate goal hierarchies among farmers. In thestudy, farmers were given sets of two goal statements and asked to reveal their37preferences by indicating on a line (Goal A at one end, Goal B at the other) the degreeto which they preferred one statement over the other. The results enabled the authors toregress the preferences on a set of farm enterprises and household characteristics.The use of cardinal rating scales to determine how people perceive scenic beautyas a direct result of forest characteristics has been used in several studies (Brown andDaniels 1984; Arthur 1977; Brown et at. 1990; Brunson and Shelby 1992). In thesestudies the authors used photographs to elicit the respondents’s preferences of certaincharacteristics associated with different sites.Conjoint analysis is another approach for estimating the value of a mulitattributegood. This method can use surveys to get respondents to rank attributes of the goodaccording to their preferences. Conjoint analysis is especially useful in estimating thejoint effect of several independent variables on the ordering of the dependent variable.The total utility associated with a combination of attributes can be estimated by summingtogether so called “part-worths”. “A trade-off, representing the amount that one attributemust change to compensate for a change in another attribute (holding utility constant) canbe determined” (Patrick et al. 1983, pp. 4-5).Other methods that attempt to price nonmarket goods include travel cost, hedonictravel cost, hedonic pricing models, etc. The travel cost method uses interviews to obtainan estimate of the demand for a public good based upon the amount of money one spendsto travel to a particular site (e.g., assumes travel cost is a proxy for price, Adamowicz1992). Thus, a demand curve based on the number of trips taken to a particular site as38a function of the amount of money spent on getting to that site can be determined toevaluate the value of the area.Similarly there is the Hedonic Travel Model which assumes that individuals arewilling to spend more on travel costs for sites with higher quality attributes (Adamowicz1992). With this model one is able to obain an estimate of the value of certain sitecharacteristics (e.g., estimate the implicit price of quality attributes).Hedonic price models are generally used to measure the benefits of an improvmentin environmental quality from the implicit effects that change in quality has on marketprices (i.e., property values).39CHAPTER 5THEORETICAL CONSIDERATIONS5.1 BackgroundWhen trying to estimate a value for old-growth forests, it is important to have aclear definition of what is meant by “old growth”. Moreover, what attributes constituteold-growth forests in British Columbia? How do people value these attributes? In theUnited States, each U.S. Forest region has a task force to arrive at some form ofdelineation through surveys and other methods to arrive at 5-6 attributes for old growthin each region. Since the age and composition of species that constitute old growth variesbetween regions, it is impossible to arrive at one specific definition. In the PacificNorthwest, the Old-Growth Defmition Task Group (1986) set out to defme old-growthDouglas-fir and mixed-conifer forests. The Old-Growth Definition Task Group, whichperceived old growth as an ecological concept, determined the minimum standards forold-growth forests applicable to a variety of species found in the Pacffic Northwest andCalifornia.The vast majority of B.C.’s old-growth forests are comprised of softwoods such astrue firs, lodgepole pine, spruce, and, to a lesser degree, Douglas-fir, cedars and Hemlock(Government Statistics 1992). Consequently, the minimum standards for old-growthapplicable in the Pacific Northwest do not necessarily represent the minimum standardsfound in B.C.; however, the standards do provide guidelines for defining old growth in40B.C. The four standard characteristics for old-growth forests are (Old-Growth DefinitionTask Group 1986, p. 4, US Forest Service, and National Park Service):1) Live trees--number and minimum size of species;2) Canopy;3) Snags--minimum number of standing dead trees of specific size; and4) Logs--specific size and minimum tonnage of downed logs.In a study to examine the management and control of old-growth forests on U.S.National Park Service lands, Lucy Tyrrell (1991) sent out questionnaires to rangers inseveral national parks in the major regions of the U.S. (Midwest, Pacific Northwest,Western, etc.) to determine which attributes they found to be linked with old growth. Asshown in Figure 5.1, the attributes defined by the NPS were in much greater detail andincluded such characteristics as species composition (including wildlife) and loggingpractices. Tyrrell’s questionnaire included inquiries into the amount of species associatedwith old-growth forests (see Figure 5.2) and which species were explicitly reliant uponold-growth forests for survival. Moreover, Figure 5.3 shows that when asked what theyfelt was the primary factor affecting old-growth forests, 42% of the respondents indicatedlogging as a primary disturbance of old growth, while wildfire (28%) and windthrow(20%) also have a major effect on old-growth forests (Tyrrell 1991, p. 20).“Towards an Old-Growth Strategy” was a project of the B.C. Ministry of Forests.The purpose of the Ministry’s study was to include the following:Figure5.1SUMMARYOFATTRIBUTESOFOLD-GROWTHFORESTSMENTIONEDBYRESPONDENTSINTHEIRWORKINGDEFINITIONSNotCommerciallyLoggedMinimumAcreageUsePublishedDefinitionStandPresentinCertainYearNoRecent Natural-€CatastropheDon’tUseTermSpeciesCompositionSuccessionalStatusofStand0Virgin‘Pristine“Original’StructuralCharacteristicsMinimumAgeof TreesUncut, Unlogged-%-III05101520253035PercentofNPSUnitsMentioningAttribute404550Source:Tyrrell,LucyE,1991.Old-GrowthForests onNational ParkServiceLands:NPSViews andInformation, p.12.Figure5.2VascularPlants—.Non-Vascular PlantsBirds:Mammals•Reptiles/AmphibiansInvertebratesIButterfliesSoilInvertebrates:]III8090100PERCENTAGEOFNPSUNITSWITHSPECIESLISTSFORTHEENTIRENPSUNITANDSPECIFICALLYFOROLD-GROWTHFORESTS0 z 0 0 C) 0. Cl)Old-GrowthForestEntireNPSUnit010203040506070Percentof NPSUnitsReportingSpeciesListsSource: Tyrrelt,LucyE.,1991.OIrJ-GxowtjiForestsonNational ParkServiceLands:NPSViews andIntomiaton,p.17.Figure5.3SUMMARYOFNATURALANDHUMAN-CAUSEDDISTURBANCESAFFECTINGOLD-GROWTHFORESTSUnknown- -—FireSupressionHuman-causedFireUsesbyNativesEffectsofToxicSpecies-Drainage-______________________FarmingMining__________________________________________________________________________________4).-_____________________________________________________________________LoggingMammalActivity________________-InsectDamage0)DiseasePaludificationLavaFlows-__________________________________________________Wildfire--WindthroworBlowdownsMassWastingGlacialIceandAvalanchesDrought-______________III051015202530354045Percent ofNPSUnitsReportingDisturbanceSource: Tyrreli,LucyE., 1991.OId-GrthForests onNational PakServiceLands:NPSViews andIntomiation,p.2044i) define old-growthii) increase knowledge (amount of inventory of old-growth in B.C., scientific,economic)iii) forest management objectives (social, economic, environmental, scientific)iv) recommendations for actionv) priority issuesvi) criteria for developing old-growth forestsPart of the analysis looked at the different values attributable to old-growth forests;some require the consumption of the forest, while others require preservation. Below isa summary of some of the values associated with old-growth forests obtained from thePublic Review Draft:1) biodiversity 9) option, bequest and existence values2) community recreation 10) science research and education3) community stability 11) spiritual and aesthetic values4) gathering 12) symbolic values5) geoclimate 13) timber and manufactured products6) grazing 14) tourism7) heritage values 15) hydrological features8) wilderness 16) hunting, trapping and fishingMany of these values are not explicitly attributable to old growth and can bemanaged for in second-growth forests.4 Furthermore, it is these values which people aremore familiar with (especially without the aid of photographs) and are able to makeappropriate judgements when valuing environmental amenities. Thus, the above list ofbenefits of old growth forests will be used as the basis for evaluating the preferences ofthe respondents in the survey.4However, many of the ecological characteristics of old-growth forests can not be easily managed forin second-growth forests.455.2 Theoretical ModelThe theoretical model to be used in this paper is based on consumer utility theory.The consumer is considered to have a preference ordering over both private and publicgoods, x and z respectively. The goods traded in markets (x) are considered to havepositive prices (p), whereas the public goods (z) have no prices since they are not tradedin the market. A consumer’s preference can be characterized by the strictly increasing,quasi-concave, ordinal utility function U(x,z), which is maximized subject to theindividual’s household income constraint (Y). The individual’s problem can be bestillustrated as:(5.1) Max U(x ,z; s) s.t. Y = pxThe prices (p) for all other goods (those traded in the market place) as well as thedemographic and socioeconomic factors that help influence the consumer’s ability to payor constrain their behaviour (denoted as the vector s) shall be eliminated for the timebeing since they remain constant throughout the analysis. Thus, the correspondingindirect utility function can be written as follows:(5.2) V(p, z, Y) = max {U(x, z) px Y} = U(x (p ,z, Y), z).The coinciding expenditure function is:(5.3) e(p, z, U) = min px u(x, z) U° = v (p, z, Y) = YThe expenditure function is simply the inverse of the indirect utility function and equalto the consumer’s household income given the optimal utility level U*. The expenditurefunction is the minimum amount of income required to achieve (or minimum cost of46achieving) a given level of utility, whereas the indirect utility function is just themaximum utility obtained from the consumer’s income and the prices of the goods.The indirect utility function is non-decreasing, continuous, twice differentiable inp, y and z, and quasi-concave in z. The expenditure function is continuous, twicedifferentiable in p, y and z, non-increasing and convex in z.There are four different Hicksian welfare measures for contingent valuationsurveys, these are: compensating surplus, compensating variation, equivalent surplus, andequivalent variation. The surpluses differ from the variations in that the former constrainthe quantity of the good being considered, while the latter do not constrain the amountof good being purchased. In cost-benefit analysis, the correct method of valuing a non-market good entails using compensating variation or equivalent variation measures.The Hicksian or compensated demand function, h(p,u), is used to evaluate welfarechanges (increments or decrements) in the level of a public good while utility is keptconstant. The compensating variation function (CV) is used to determine the amount ofmoney that the consumer would be willing to pay (WTP) for an increase in the publicgood from z° to z1, where z0 < z1. The compensating measures assume that the consumeris entitled to the current level of utility or the current endowment of property rights.Similarly, the equivalent variation function (EV) is the minimum amount of money theconsumer would be willing to accept (WTA) as compensation to forgo the increase orimprovement in the public good.In this study, a contingent valuation approach is used to obtain a respondent’s47WTP, therefore, the appropriate choice of measurement is the compensating variation.One way to measure this is to determine the difference between two expenditurefunctions. Alternatively, it is the change in consumer’s income, coupled with the changein the level of the public good that leaves the respondents’ utility unchanged. Thus, CVcan be mathematically stated as follows:(5.4) CV(Z’, Z°, Y) = Y - e(Z’, v(Z°, Y))where z1 is an increase or improvement in the level or availability of the public good andY is the consumer’s income.Compensating variation can also be written as:(5.5) CV = e(p , Z’, Z°, U°) - e(p , Z° , U°),which implies the amount that is needed to make the consumer as well off as before (i.e.,hold U at U°).Taking a Taylor series expansion about Z° and the mean income level, Y, givesthe following expression for CV5:CV = CV(Z°, Z°, + (z1 -z°) + (y-+- z°)2 + - y)2 8CV2 az2 2+ (Z1Z0)(Y+R.(5.6)where R refers to remaining terms. The willingness-to-pay function for a single5As derived by van Kooten (1993).48household for an increase or improvement in a public good is as follows:=+ + c2AZ +c3(y-c4(Y -5Az(y- Y) + R,(5.7)where x= CV(Z°,Z°,Y*)= 0 since no change in a public good in CV would be zero; a1= CV/ aZ; cc2= CV/ aY; CC3 = 1/22 CV! a2 Z; (x4 = 1/2a2CV/Y; CC5 aCV/aZaY.The empirical model would be complete once social factors describing attitudes, age,household makeup and size, etc.49CHAPTER 6METHODOLOGY6.1 Contingent ValuationThere are many different methods for estimating the value of nomnarket goods(briefly defined earlier in Chapter 2), but in this study, contingent valuation was chosen.There are many different ways of eliciting a consumer’s WTP in CV, however, two of themost commonly used methods are the open-ended and closed-ended methods.With an open-ended questionnaire, the interviewer tries to directly estimate therespondent’s willingness-to-pay by asking him or her directly how much they are willingto pay towards the amenity. Although this method directly elicits a WTP value, it toohas many shortcomings (see biases in Chapter 4).6.2 Closed-ended Approach (Dichotomous Choice)The closed-ended approach (also known as referendum or dichotomous choice) usesan indirect approach to estimate a consumer’s WTP. This strategy provides the researcherwith a discrete indicator of WTP rather than an actual WTP.This approach is being used more frequently in analysis of public goods since itdoes not have as many of the drawbacks as the open-ended approach (many of the biasesassociated with the open-ended format are avoided with the closed-ended format). The50dichotomous choice (closed-ended) approach provides the respondent with a given price($X) which he/she must decide whether or not they would be willing to pay for a changein the quantity of the amenity. In other words, the respondent indicates his/her WTP byanswering “yes” or “no” to a single stated price. Often this bid is calculated as theestimated actual cost of preserving the good. Moreover, this design is consistent withutility maximization theory (Chapter 5).One of the benefits that arises from using discrete indicators to elicit a respondent’sWTP is that it gives the respondent some sort of guideline with which to evaluate his/hermaximum willingness-to-pay for the amenity, especially if the good is one that isunfamiliar to them or hard to evaluate in dollar terms. Whereas in the open-endedformat, the individual may have to choose a value without any concept as to the actualvalue of the amenity, as a result the amount may be unrealistically large or too low. Incircumstances where the survey is given absentee of the interviewer or visual aid, thisapproach is often the preferred choice among researchers.Another advantage to this approach is that, unlike the open-ended format where anindividual may overstate his/her WTP, this approach is considered incentive-compatible.Ideally “...it is in the respondent’s strategic interest to say yes if her WTP is greater thanor equal to the price asked, and to say no otherwise” (Mitchell and Carson 1989). Thebenefit of this is the decrease in the probability of strategic bias.However, the dichotomous choice (DC) approach also has its limitations. Oneproblem the analyst may face is that less information is available about the respondent51than with the open-ended questionnaire. For example, an exact value is obtained fromthe respondent with the open-ended format, however only whether the consumer’s trueWTP is greater or less than the amount offered ($X) is obtained in the closed-endedformat. Within the DC approach there is the single-bounded model and the double-bounded model which gives the researcher a more precise hint of the true WTP of therespondent.6 Due to this limitation, a much larger sample size is required.Similarly, the economist must decide what range of bids to offer in the survey.Too small a range may lead to problems in estimating the probability function, or maynot give enough information to the researcher to estimate an accurate demand curve.Furthermore, Mitchell and Carson (1989) have identified other drawbacks to thedichotomous choice approach. The authors suggest that, like the starting bid biasassociated with other methods, the dichotomous choice approach is subject to a nonzerolevel of yea-saying7.Yet another concern is the method by which the mean WTP is obtained. Currentresearchers have found that using either a logistic or probit regression yields a mean WTPdirectly from the parameters of the equation (discussed below). However, using either ofthese functions requires the analyst to make strong assumptions about the mathematical6Recent work has involved the use of double-bounded (and even triple-bounded models), whereby therespondent is then given a second offer depending on the response to the first bid. Generally, a doublingof the first bid if the respondent responds with a “yes” or half of the original bid if the respondentanswers “no” is the usual appmach.7Yea-saying is defmed as when a respondent agrees to the bid amount in an attempt to please theinterviewer.52form of the valuation function. With the open-ended design, the individual’s maximumWTP is calculated directly and gives the researcher a clear indication of how theindividual values the amenity.It is up to the discretion of the researcher which procedure to use, however, theU.S. National Oceanic and Atmospheric Administration (NOAA) panel (Arrow et al.1993) favors the dichotomous choice approach as they find the open-ended approachunreliable.86.3 BiasesApart from the biases mentioned earlier in Chapter 4, another bias that may occurin CV studies is sample selectivity bias. Sample selectivity bias occurs when theresearcher purposely eliminates those respondents for which zero or missing values werefound for the valuation question (including endowment bidding and protest bids).Moreover, if one is to assume that the proportion of the sample remaining (those forwhich a positive value was obtained) is an appropriate representation of the populationas a whole, then one must assume that those who did not respond to the survey(nonrespondents) value the amenity the same way that survey participants do. However,one cannot make this assumption without some sort of proof that sample selectivity biasdoes not exist. One test for this type of bias is the procedure developed by Heckman,8The U.S. National Oceanic and Atmospheric Administration (NOAA) panel was established togive suggestions regarding the use of contingent valuation for valuing environmental goods.53whereby a simple two-stage estimator is used to internalize censored observations (thoserespondents with positive bids) and to test for sample selectivity bias.96.4 Protest ResponsesProtest responses are those responses whereby the respondent, for reasons ofhis/her own, refuses to answer a question (closed-ended surveys) or gives a zero bid (asin the case of the open-ended questions called zero response bids). There are manyreasons why respondents may refuse to answer questions. First, the respondent may, asin the case of public goods, object to putting a dollar amount on a public good. Theymay feel that public goods, such as provincial forests, belong to everyone and that theirvalue cannot be captured by a dollar figure. Second, the respondent may be objecting tothe payment vehicle. For example, if the payment vehicle is increased taxes per year therespondent may not feel that this is the appropriate method of payment, but prefers anapproach where users are charged instead (i.e,. a user fee). Moreover, the respondent mayfeel that the question is not worded properly and may not fully understand what theinterviewer is trying to elicit.6.5 WTP ModelThe first step in calculating the welfare measure is to estimate the parameters ofthe probability function. There are several techniques which can be applied; these are the9For a more detailed explanation of Heckman’s procedure, see Edwards and Anderson 1987.54linear probability model (LPM), the logit model and the probit model. However, the LPMis rarely used as a means of estimating parameters because the error structure isheteroskedastic and non-normal, and probability predictions can be outside the zero to onerange. Conversely, logit and probit regression models are unbounded and linearly relatedto the independent variables, while the probabilities are restricted to 0-1 and related to theindependent variables by a logistic or cumulative normal function. Thus, the probit andlogit models are considered to be more suitable techniques when dealing with binaryresponse models.Often, binary response models (dichotomous choice) use a transformation approach.In these models, an index variable,Z1=XB, representing the utility difference is used. Thelarger the Z1, the greater the probability that the event (respondent says “yes”) will occur.Thus, a monotonic relationship between the probability of the event taking place and theindex variable is established. When this situation occurs, the probability function takeson the characteristic of a cumulative distribution function (c.d.f.). The two mostfrequently used c.d.f.’s are the normal and the logistic functions resulting in the probit andlogit regression models. The logistic function is often preferred to the normal functionbecause of its ease of estimation and that it closely approximates the normal function.It is also for this reason that the logistic function was used in this study. Two methodsthat can be used to estimate the parameters for dichotomous choice models are themaximum likelihood method and generalized least squares (GLS). In this study, amaximum likelihood method was used to estimate the parameters.55The standard logistic function is as follows:1- 1(6.1)+ exp(—2f3) i + exp—(f30I31X+..+f3X+ei)where P is the probability that the respondent will respond with a “yes” (or “no”depending on the analyst’s objective) to the offer. The independent factors that influencethe respondent’s decision are represented by the variables x1...x. The parameters to beestimated are 15 etc. Thus, the probability of the respondent accepting(rejecting) the offer can be written as:(6.2) 11 + exp(XeP)The functional form of the explanatory variables (f(x,b)), depends upon severalfactors, such as economic theory and what functional form best fits the model. Economictheory has been used to help specify the proper functional form by relating it to utilitytheory. Some researchers (Sellar, Chavas and Stoll 1986) have argued that using linearfunctional forms to specify the explanatory variables is incorrect when using data obtainedin dichotomous choice models. The authors claim that the linear form does not satisfythe properties of consumer theory (the Hicksian demand curve is not downward sloping)(see Sellar, Chavas and Stoll 1986p.386-7 for proof). Other researchers concluded that56as long as the functional form meets the minimum requirements for utility theory, thenthe model which is best able to predict observations, best fits the data, and gives the bestgoodness of fit is the more appropriate model (Hanemann 1984; Bishop and Heberlein1979; Stynes and Peterson 1984). Thus, the majority of researchers use either the linearor the log-linear functional form.6.6 Welfare MeasuresAnother important issue to consider when using logit models is whether to use atruncated model or to integrate to infmity when calculating the expected willingness-to-pay. A truncated model uses the maximum bid as the truncation point rather than infmity,which can yield exceptionally high values for the amenity when the estimated functionhas a “fat tail”. When one truncates the model with the highest offer, care must be takento ensure that the probability of a respondent accepting (rejecting) an offer above thisvalue is sufficiently low.Furthermore, the median should be used as the appropriate welfare measure insteadof the expected value, since the median is less affected by the size of the tail than themean. The mean or average of the distribution F(x) implies that the respondent’s WTPis greater than the offered bid. However, the mean may be heavily influenced by theupper tail of the distribution and may represent the values of only a small percentage ofthe population. The median isF1(O.5) and represents the largest amount that at least 50%of the population would be willing to pay (the amount of money the individual would57require to keep them at the point of indifference between paying for the item and doingwithout it). The median is often preferred to the mean because it is likely to be lesssensitive to outlying (unusual) observations that may affect the estimate of thedistribution, although it should be noted that both methods of estimation are consistentwith ordinal utility theory. However, one problem with using the median instead of themean is that the median cannot be aggregated over the entire population even though itis not influenced by the upper tail of the distribution. One solution to some of theaforementioned problems, arising from both the mean and the median, is the truncatedmean which assigns a truncation value T to all WTP values above T before computingthe mean (Duffield and Patterson 1992).Unlike the truncated mean (the difference between two indirect utility functions),the censored logistic regression approach involves specifying a form of the expenditurefunction. This approach enables the researcher to directly calculate the mean WTP byreparameterizing the estimated coefficients obtained through logistic regression, orpreferably, through optimization of the censored logistic regression through maximumlikelihood procedures. Thus, E(WTP) can be calculated by inserting the mean of theexplanatory variables or the values for each respondent into the corresponding censoredlogistic regression equation.Another slightly different welfare measure is the non-parametric approach proposedby Kristrom (1990). Kristrom’s approach is simple: calculate the sequence of proportionof “yes” responses to derive the probability of acceptance of the bid amounts. Graphing58the probabilities of an acceptance with the bid amounts yields an empirical survivorfunction. Linear interpolation of the function allows the researcher to calculate the endsof the function (i.e., where the probability of zero intersects the bid axis). The mean issimply the area under the survival function, and the mean is calculated directly from thegraph at p = 0.5. Kristrom claims that this method is equally comparable to theparametric approaches and is more robust and simpler to calculate.The estimated willingness to pay for an individual respondent can then becalculated by taldng the integral of the above expression (i.e., calculating the area belowthe estimated cumulative density function (c.d.f.)).(6.3) E(WTP) =‘ 1 dKmax Jo 1 + exp—(303X..+f3X)To see how this equation was derived, see Sellar, Stoll and Chavas (1985).6.7 Survey Instrument DesignTwo separate surveys were used in this study. The first survey used a biddinggame format. A total of 1,230 questionnaires were sent to individuals within BritishColumbia. A second survey, using a dichotomous choice format, was distributed among146 students in forestry and land use courses at the university. Both surveys included anintroductory and background page informing the respondents of the issues involved. Themailout surveys included a personalized cover letter to introduce the importance of the59respondent’s participation in the survey and assured them of confidentiality. Opinionquestions were included in the survey to try and achieve some indication of the attitudesof the respondents regarding the subject matter, as well as for consistency of resultsobtained in the valuation questions. The valuation questions, although similar in intent,differed in design between the two surveys. Finally, a section on socioeconomic factors(demographics, income, etc.) was included for statistical purposes. (See Appendix la andlb for the questionnaires).The first section was designed to help assess individual’s opinions of B.C.’sforestlands. The individuals were asked to evaluate the characteristics of old-growthforests (based upon attributes determined by the Old-Growth Definition Task Group) ona scale of importance.The second section of the survey was designed to obtain an estimate of anindividual’s WTP. Using a dichotomous choice format, respondents were giveninformation regarding the current level of preserved forestlands in B.C. For the mailoutsurvey, the question then listed eight levels of possible amounts of land that could bepreserved. A dollar amount was associated with each level of preservation, anapproximation of the cost to put aside the specified amount of land. The respondent wasasked if they would pay at each level. The payment vehicle used in both surveys wasincreased taxes per year. The respondents also had the option to indicate whether theybelieved the current amount of protection was adequate. The in-classroom survey variedslightly in that it allowed the respondents to choose the level of preservation itself.60Furthermore, offers of $75 $1,200 dollars were randomly distributed among thestudents. Students were then asked to respond “yes” or “no” if they would pay the givenamount for the level of preservation specified by the student himself/herself. Furtherquestions would try to reveal the upper or lower limit of the individual’s WTP.Section two of the mailout questionnaire was possibly misleading in its wording.The heading of the section read “Protecting Old-Growth Forests from Timber Harvest”,however, the rest of the details given to the respondent were with respect to B.C.’swilderness areas. It is, therefore, difficult to speculate on whether people were respondingto increased wilderness protection in B.C. per Se, or were responding to preservation ofold-growth forests. A sample of 15 individuals (not included in the mailout survey) weregiven the questionnaire and asked to indicate whether they felt the question was wordedto imply preservation of old-growth forests or increased wilderness protection in B.C. All15 sampled indicated that wilderness preservation was the focus of Section Two.The third section (included in the mail surveys only) was designed to determine ifthe respondents were consistent in their views. They were told to allocate a given budgetfor the protection of forests in B.C. A partial list was provided utilizing several of theattributes from the first section with room to add their own specifications on how thebudget should be distributed.The fourth and fmal section of the questionnaires consisted of a personalinformation page involving questions about age, sex, income, education, # of persons perhousehold as well as information concerning their charitable support for environmental61causes. For the classroom survey, the question regarding income was modified to “Howmuch do you expect to be earning five years after completion of ALL your postsecondary education?” For obvious reasons this section is to be regarded with scepticism.Also included in the classroom simulation, was a section which enabled the interviewerto separate the protest bids from actual zero bids by asking the respondents to indicatetheir reasons for giving a “no” response or not answering the questionnaire.6.8 Implementaion Procedures6.8.1 The Study AreaThe province of British Columbia is divided into six different forest regions; 1)Vancouver; 2) Prince Rupert; 3) Kamloops; 4) Prince George; 5) Nelson; and 6) Cariboo.Vancouver and Prince Rupert regions are coastal regions while the remaining districts areinterior, and different forest characteristics are associated with each classification (coastalvs. interior). For example, in the interior the diameter of old-growth trees is not as largeas in the coastal forests. Furthermore, species type may differ among regions due toenvironmental conditions such as climate, soil quality, temperature and physiographic(topographical elements) considerations. People who live in these regions view oldgrowth differently from those who live in other regions, with some more dependent onthe forest industry for employment (i.e., logging and harvesting) than others. Dependingon the nature of the region, opinions may vary dramatically among the six regions.626.8.2 The DataThe data (name, address, phone number) used in the survey was purchased fromDominion Directories (an independent company who supplies the phone numbers tobusinesses across Canada). The data contained a random list of 5,000 individuals withinthe province of B.C. The number of people per region depended upon the density of thepopulation of that region (i.e., the greatest population was the greater Vancouver region).From the list of 5,000, a random sample of 1,230 names was selected using the sameregional distribution.The data for the classroom surveys came from two third year courses. One coursewas a forest economic course comprising of mainly forestry students while the other wasa land use economic course which students from many disciplines.63CHAPTER 7RESULTS AND IMPLICATIONS7.1 Participation Rate7.1.1 Mailout SurveyResults of the province wide mailout survey are summarized in Tables 7.1,7.2 and7.3. Of the 1,230 surveys mailed out, 35 were undeliverable, and 279 were returned (fora response rate of 24%), but only 246 were usable for summary statistics due to missinginformation. Table 7.1 indicates response rates by region. Prince Rupert exhibited thehighest response rate of 45%, but only 11 surveys were sent to that region. The greatestproportion of surveys were mailed to the Greater Vancouver Area, but only 22% of thesample participated in the survey. The lowest participation rates were in the Nelson andCariboo Forest Regions (8% and 0%, respectively); however, only 1.6% of the 1,230surveys were mailed to these regions compared with the 86% mailed to the VancouverRegion.Although this survey did not have a particularly high response rate, surveys byHelfand et a!. (1991), Adamowicz et al. (1991), and Mannesto and Loomis (1989) havehad similar results. Due to lack of funding it was not possible to send the follow-upletters, pre-letters or make follow-up telephone calls as recommended by DilIman (1978).Similarly, the data provided by Dominion Directories for use in the survey appeared tohave been out of date. Many of the envelopes were returned unopened because the64occupant had passed away or had moved to another location.Table 7.1: Response By RegionMAIL RESPONSEFOREST REGION OUTS RESPONSES RATEVancouver 1055 232 22%Prince Rupert 11 5 45%Kamloops 105 32 33%Prince George 39 9 23%Nelson 13 1 8%Cariboo 7 0 0%7.1.2 Class SurveyThe class survey was distributed among two third year University courses--a landuse course and a forestry economics course. A total of 146 students participated in thesurvey (88 students from the land use class and 66 students from the forest economicsclass).7.2 Summary Statistics7.2.1 Mailout SurveySocioeconomic factors and demographics of survey respondents are summarizedin Table 7.2. The statistical means of the respondents are compared those of all B.C.65residents to inspect for sampling error bias. The results show that the mean householdsize of the respondents was 2.5 persons compared to a provincial average of 2.6 persons(1991 Census). Similarly, the mean household income level of respondents was $40,000 -$50,000 per year, while the mean for B.C. residents was $46,909 per year. The meanage of B.C. residents is 34.3, while it was slightly higher for survey respondents (36-45)(Statistics Canada 1992). Although the average level of post secondary education of B.C.residents was not available for 1991, the Statistics Canada 1986 average education levelwas just over 12 years. Comparatively, the mean level of education for respondents inour survey was 14.3 years (high school plus 1-2 years post secondary education). It isconcluded that survey response bias is not a problem with this survey.Table 7.2: Statistical MeansiTEM MEANSVersion I (II) 138 (141)Sex -Totals: Female (Male) 98 (181)Average Age 36-45Average # People Per Household 2.5Average Education (Total Years) 14.3Average Annual Income” $40,000-$50,000WTP $439/hsehld/yr - -aVersion I arranges WTP questions in ascending order whereas Version II is in descendingorder.bTotal household income before taxes66The mean willingness-to-pay for respondents was $439/year/household. Adjustingfor the respondents who were considered to have given zero values for preservationlowers this mean to $373.15 (i.e., multiplying by the number of those who valued theamenity at some number greater than zero).Endowment bidding, or bidding a high percentage (-p6-10% or more) of one’searnings, accounted for only 5% of the total number of responses.7.2.2 Class SurveyOf the 146 students who participated in the survey, 68% of the students were maleand 32% were female. The majority of students were 25 years or younger. The averageexpected income five years after completion of their post secondary education was$40,000-$50,000 per year. The average WTP was $215 per student per year for anaverage of 10.75% of wilderness protection.7.3 Zero Bids and Protest Responses7.3.1 Mailed SurveyZero bids were considered to be those bids where respondents felt that the currentlevel of protection was adequate. Since the maiout survey failed to include a section forrespondents to indicate their motives for not paying for more protection, it is impossibleto detect protest responses from valid zero bids. Protest responses were, however,considered to be those respondents who failed to answer the valuation questions, or, as67in the case of several of the surveys returned, offered their own reasons for not paying.All protest responses were dropped from the analysis. A test for sample selectivity biasusing the Heckman model indicated that no sample selectivity bias existed (lambda wasrejected at the 0.15% level)’0. Thus, only responses with WTP > 0 were used in theregression analysis and were considered to be representative of protest bidders as well aspositive bidders. Furthermore, the issue of non-response bias was not considered to be asignificant problem due to the wide distribution of responses including zero bids.7.3.2 Classroom SurveyUnlike the provincial survey, a section at the end of the survey provided therespondents with an opportunity to indicate their reasons for not paying the specified bidoffer or completing the questionnaire. This enabled us to identify and separate the protestbidders from the zero bidders. Of the 146 students surveyed, 42 (28%) were identifiedas protest bidders (similar results have been found in other studies). The number ofprotest bidders is exceptionally high given the number of students surveyed. Onereasonable explanation for this situation is that the students (many of which were inforestry) may be biased against the idea of placing a value on public land. A secondpossibility is that those in the discipline of forestry may feel threatened that future jobpossibilities may begin to decrease if the number of forest related jobs diminish due toincreased preservation. Approximately 30% of those students in the forestry class gave‘°Lambda is simply the inverse mill’s ratio.68protest responses compared to the 27% from the land use course which comprised ofstudents from all disciplines including many from forestry. Interestingly, the regressionanalysis indicated that the students in the land use class were less likely to pay to preservemore land than those in the forestry class. The Heckman (1979) procedure was performedhere and the lambda was found to be insignificant as well. Thus sample selectivity biaswas not found to exist.7.4 Attitudes and Opinions About Wilderness7.4.1 Province-wide SurveyThe opinion questions used in the regression did not show any real pattern or meetprior expectations. Four of the 14 opinion questions were found to be significant, and ofthe expected signs. Tests of multicollinearity plus a principal component analysis,indicated that multicollinearity was not a problem.Although the opinion variables did not show any significant influence on WTP,the majority of people who were willing to pay the offered amount had strong views onwilderness protection. Figures 7.1(a) and 7.1(b) show that people who considerrecreation, biodiversity and general ecosystem benefits as being “extremely important” arewilling to pay more than those individuals who indicate wood products and future goodsand services as being “extremely important” to them. Of those individuals who indicateda WTP of $1200, only 5 percent felt that timber and manufactured forest products were“extremely important” to them. Likewise, only 14% felt that provision of goods and69services for the future was “very important”.Similarly, those respondents who presented a zero bid (i.e., felt the current amountof wilderness protection was adequate) valued the benefits of timber products more highlythan habitat protection and recreation. Fifty-five percent of all zero bidders found timberand manufactured products were “extremely important” compared to the 30% and 21%who felt habitat protection and recreational benefits were of “extreme importance”.7.4.2 Classroom SurveyIn the classroom survey, the opinions of zero bidders and those who were willing-to-pay the offered amount are summarized in Figure 7.2 (a) and 7.2 (b). The graphs showthat people who did not want to pay anything indicated that timber and manufacturingproducts were “extremely important” (i.e., 60% of the people who valued wood productshighly were not willing to pay for increased wilderness protection), while only 44% foundwildlife habitat was of significant importance to them. In comparison, of thoserespondents with a WTP> 0, only 34% of the individuals who valued wood productsfavourably were willing to pay anything. Fifty-six percent of the individuals who saidhabitat protection was “extremely important” were also willing to pay some given amount.The differences between the zero bidders and the people who had WTP > 0 aremore noticeable when one examines the latter portion of the graph. Apparentlypreservation values (option, bequest, and existence values) play a more significant rolein determining whether or not a person agrees to the offer than other values, such as theFigure7.1(a)PROVINCIALAlTiTUDESCONCERNINGOLD-GROWTHFORESTSINB.C.3002500) 1% U) I— z w Z1500 0 U) ‘Li0InnotimportantatallHIllsomewhatunimportantElneutralElsomewhatimportantextremelyimportantg...E0E-6Cd•h•‘.(1)C)CI)ECD0.CDC)C.)E•-C.U)p•_.0Cd.2‘.a).0CI)ATTRIBUTESPERCENTL‘)().(fl0)0)CD0000000000IIIIRECRWOODMEDICTOURISMR&DHUNTiiiiiiiiiiiii4JJIJI11IIII!IIIf1111111—1-V0I1-1-u-I-IC-ImCl)w-CD0.0CDt—CD-I0•CCDC,)-nCDC-S-&0•BIODIV!!11!Mit!Lliii1TI1IIIIIIJI[IJI11IIIIII1IwwIwIIIlwIIIIlwwIIIIwIIIIIIIIIIIIlIwwIIIlII!IIIIGEN’LIuiiiJlIi1i1I1iffiiiTIii1i1iJiii1iJi1MiiIIwwiiiIwiIiiiiiiIIiiwwiiiIiiiiiiiiiFUTURESPIRIT’LWILDERnIiii111111111111IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIiIIIIIIIIII‘III{IIIIIIIIIIIIIIIIIIIIIIIIIII111111IIIIIIIIIIIIIIIIIIIIIIIIIIIIIEXIST’NCBEQUESTGATHER/1-72value of timber products. However, since most of the signs of the variables did notconform to a priori expectations and were insignificant when included in the regressionanalysis, this could not be proven.7.5 Willingness-to-Pay For Increased Protection of B.C.’s Wilderness AreasA description of the variables to be used in the regression analysis can be foundin Table 7.3. The variable “Version” was only used for the provincial survey only to testwhether or not the ordering of the questions had any effect on the results. Likewiseregional dummy variables and the opinion variables were included in the province widemaiout survey but not in the classroom surveys. These variables are seif-explanitory andthus not explained in the table.7.5.1 Provincial SurveyThe results from the regression analysis are reported in Table 7.4. The regressionanalysis included not only the socioeconomic factors but the opinion questions as well.The results indicate that men are generally more likely to pay for increased wildernessprotection than are women. As expected, the higher the income class, the morerespondents are willing to pay. This result is often found in CV studies.Education had the correct sign on the coefficient and was statistically significantat the 0.025 level. There is a positive relationship between the level of education and therespondent’s WTP. The coefficient on the variable for age had the correct sign but was•—.0) o0C•) IU) Cl)00-I-ICmU)C)0zC) mzz0I0-l‘10mU) -ICl)za,p#OFRESPONDENTS(146)•_I’3-0)O0I) 0000000recreationIIIIItimbermedicaltourismresearch+It!rM[JllIJ1$ituI!ib1Ihunting1biodiversity03___________CecosystemU)g&sspiritualwildernessexistence-TiCOC-‘CDr’)bequestgathering11FCD()D0DX0CI)00.3C3-+ CDCDCD5• 3.CD-D0••--003- ;-;i--Q)p’0—.-qD70 60 50I z40Lii C.) 30 20 10 0Figure7.2(b)CLASSOPINIONQUESTIONSa:-g 0C) CDCl) D0-CDa, C:6cDD0(!3-u_0.CDC’)ATTRIBUTESII Somewhat ImportantA-----—WTP>0ci) C) C a, C)) x IiiCD 0 a,ci)-C Cu (!3ExtremelyImportant—0-——WTP=075statistically insignificant at the 0.05 level.Table 7.3 A Description Of The Variables Used In The RegressionAnalysisVariable DescriptionWTP Dependent variable used to indicate theindividuals WTPAmount Amount of the offer made to an individual ($)Preservation The amount of land preserved as desired by theindividual (ha)Gender Dummy variable used to identify gender; 1 ifmale, 0 if femaleIncome The household income ($/yr)aMember Dummy variable used to indicated whether therespondent was a member of any environmentalorganizations; 1 if member; 0 otherwiseContribution Dummy variable used to indicated if therespondent contributed regularly to someenvironmental agency or charity; 1 ifcontributed, 0 otherwiseAge The age of the respondentClass Dummy variable to indicate which class thequestionnaire was issued; 1=forest econ; 0=landuseVersion 1=ascending order 0=decending ordera the classroom survey, income was the expected annual income five years after completing all postsecondary education.76Table 7.4: OLS Regression Estimates For the Classroom SurveyVARIABLE ESTIMATED STANDARD T-RATIONAME COEFFICIENT ERROR 230 DFVersion (1=1) -110.30 4554 2422bGender M=1) 83.832 54.28 1.545’Age -29.127 18.18 1.602vFamily -21.382 20.36 -1.05Education 20.453 10.20 2.005cIncome 19.981 12.57Recreation -1.358 31.95 -0.0424Wood Products -52.531 27.28 -l.925’Pharmaceutical 20.493 25.88 0.79 19Tourism -7.452 26.08 -0.2858R&D 37.321 28.79 1.296Biodiversity 59.776 38.06 1570dEcosystem Ben. 19.458 38.94 0.5021Scenic Quality -11.974 30.98 -0.3866Wilderness Val. -3.493 33.78 -0.1034Existence Value 58.464 27.25 2.145’Bequest Value -2.404 37.76 -0.0636Contribution 43.446 65.20 0.664Membership 97.014 79.16 1.225Vancouver 542.74 364.40 1 •734CPrince Rupert 494.90 397.70Kamloops 457.62 371.90 13g8dPrince George 459.67 388.50Constanta j -818.15 426.60 -0.8405‘ihe Nelson 14orest Region is captured in me constant term.b5jgjfjat the 0.01 level; csignificaut at the 0.05 level; dsignificant at the 0.10 level77Of great interest was the sign of the coefficient for the variable describing theversion of the questionnaire. There is a negative correlation with the order in which thebids are asked. Respondents who received a questionnaire with the offer amounts inascending order ($75-s 1200) were not willing to pay as much as those who received theoffer amounts in descending order. In other words, when faced with a higher startingpoint, respondents were willing to pay more than respondents who received a lowerstarting bid ($75 versus $1200). It is possible that respondents receiving the questionnairewhere the first bid offered was also the lowest bid, became tired of answering the iterativequestions and ended the procedure before their maximum bid was obtained. These resultssupport the theory that the order in which the interviewer asks questions can influence therespondent’s decision (see Chapter 4 of this paper). Kahneman and Knetch (1991)suggested that the respondent views the starting bid as a true approximation of the valueof the good and may agree to the offer simply because he/she feels that this is theappropriate thing to do.The four regional dummy variables (for five regions, with the sixth regionexcluded since there were no responses from this region) were all significant at the 5 and10 percent levels. It is likely that the few who participated in the survey viewed theNelson Forest Region as a community which relies on logging practices, and thus, wouldbe averse to preservation of more forest land. Furthermore, respondents may have felt thepercentage of forest land presently preserved there was adequate.787.5.2 The Classroom ResultsThe second questionnaire was distributed to 146 students; its results aresummarized in Table 7.5. The average WTP per respondent was $200 per year topreserve 10.75% of the province’s land base. The results are consistent with a prioriexpectations. The negative sign for the offer amount suggests that, with an increase inthe offer amount, the more likely it is that the respondent will reject the offer. Thepositive sign on the variable for income suggests that as income increases the greater theprobability that the person will accept the offer. This is not unusual since, the moredisposable money they have, the more they may be willing to give to causes suchpreservation of wilderness. However, in this part of the survey the respondent was askedto estimate his/her possible income within the next few years so one should be cautiousabout the significance of the variable. The sign of the constant is negative as expectedimplying that the estimated c.d.f. (cumulative density function) is decreasing. Unlike themaillout survey, the results indicated that the male respondents were less likely to acceptthe offer than the female respondents. Although most studies find the opposite, the resultsare the same as that found by Stevens et al. (1991). One explanation is that more than66 percent of the students surveyed were male students and in the forestry economicsclass whereas the majority of the females surveyed were in the land use class. Similarly,the probability of a “yes” increases as the level of preservation increases. This result wasexpected since the greater the level of protection the more it is worth to the respondent.As expected the variable for membership in an environmental organization was positive79but insignificant at the 10% level. The students in the forestry class were not willing topay as much as those in the land use class. This was not unexpected as those in theforestry class may have had a vested interest in the forestry profession.Table 7.5: Logit Estimates of The Probability of Answering “Yes” to Preserving X% ofB.C.’s Land BaseEstimated CoefficientVariable (t-Statistics) Standard ErrorPreservation 0.3324 0.8 14E-01Amount (4.0820)aAmount -0.16369E-02 0.636E-03(2.5746)aGender -0.84678 0.404(-2.0972)”Income 0.22380 0.144(1 .5569)cMembership 0.68192 0.656(1.0390)Class -0.54141 0.392(1.3796)cConstant -2.8748 1.048(-2.7428)Sample size 146CorrectPredictions (%) 69Maddala R2 0.18pseudo R2 0.16x2(-2log(LOILN)) 28.284Significant at the 1% level; bjgfljfi.j at the 5% level; Csignffict at the 10% level.80As shown, all coefficients used in the final regression exhibited the correct signsindicating a good model. Other tests for goodness of fit were the pseudo R2, Maddala R2,McFadden R2, and the percentage of correct predictions. The pseudo R2 of 0.16 is good(a fit between 0.2 and 0.4 is considered to be a good fit). When more of the explanitoryvariables were included the value increased significantly. The Maddala R2 of 0.18 is alsoquite high, and other studies, such as that the one by Stevens et at. (1991), had the samevalue. The percentage of correct predictions (69%) is quite good. However, when allvariables were included in the regression the amount increased to 78%. Similarly, all thegoodness of fit indicators increased as more of the explanitory variables were added. Alog-linear model was also estimated (not shown) which showed the estimates only variedslightly from the linear and the goodness of fit tests increased slightly.7.6 BiasesInformation bias was not a factor in either survey. The survey was designed toinform people and present all possible sides of the problem. The survey was designed soas no to influence the respondent in one direction or another. Although starting point biascan occur, a regression of the offer amount only on WTP showed that the bid amountsdid not depend on whether people responded favourably or negatively to the bid.Furthermore, the random range of fixed offers varied from $75 to $1200, thus reducingthe likelihood of both starting point bias and strategic bias. However, in the province-wide survey (as previously mentioned), the order in which the questions were asked (high81to low bids versus low to high bids) appeared to have some influence on the level ofWTP. Still, this is not necessarily a case of starting point bias, but perhaps a bias in theordering of questions.Importance bias may also have had some influence in the results of the survey.The survey sought to make the respondent feel that his/her participation was ofimportance in order to persuade him/her to complete the survey, which is compatible withDillman’s (1978) Total Design Method. At no point did the survey mention that theresults would be used to establish the correct level for policy. It was made apparent thatthis was simply part of my Maste?s thesis research in the faculty of graduate studies andpart of an ongoing research by Forestry Canada and the Forest Economics and PolicyAnalysis Research Unit (FEPA).Furthermore, it was hoped that strategic bias would be minimized since respondentswere informed that the results were simply for research for a thesis and not necessarilyfor use in public policy making. In spite of the attempts to minimize these biases, it mustbe noted that those who replied to the survey may have had strong and biased views (i.e.,environmentalists or those in the logging based industries). Many of the returned surveyshad comments made by people who opposed preservation of more forestland and accusedthe survey of being biased in favour of the environmentalists. Conversely, there wereletters returned implying that the survey was aimed towards the forest industry.Nonethelesss, many of these respondents filled in the questionnaire and it was obviousthat they may have overstated or understated (depending on their standpoint) their true82WTP. However, it is difficult to identify all those that may have overstated or understatedtheir WTP outside of protest bidding or endowment bidding. It is unlikely that thesepeople made up a large percent of the respondents.The problem of “yea” or “nay” saying did not seem to be of too much concern inthese surveys as: (1) the survey was a mailout not an in-person survey--people did not feelcompelled to please the interviewer; and (2), since there were two versions of thequestionnaire and the version which began with the highest offer amount yielded thehighest WTP amounts, one can reject the possibility of either “yea” or “nay” saying.Since the classroom survey only offered one amount, there does not seem to be aproblem.7.7 Model Specification: Functional FormTo estimate the sensitivity of the results to the functional form, regressions wereperformed using several common functional forms--linear, share and logarithmic forms.Recent research on the impact of functional form on WTP estimates obtained fromdichotomous choice question formats has shown that, depending on the modelspecification, estimates may vary between models (Bowker and Stoll 1988; Ozuna, Jangand Stoll 1993; and Desvousges et at. 1992). Table 7.6 presents the results of theregressions using the various functional forms.The results show that the estimates obtained from the different functional formsvaried little between models. The logarithmic form (the only model which is not83consistent with Hanemann’s utility theory) gives a slightly higher estimate of mean WTP($559) and has a higher overall goodness of fit.Table 7.6: Logit Regressions Using Various Functional FormsFunctionalFormLinear Logarithmic ShareMean WTP $558 $559 $558Maddala R2 0.18 0.19 0.16Correct 69% 70% 66%PredictionsThe number of correct predictions is 70% for the log-log specification, comparedwith 66% and 69% for the share and linear specifications. Similarly, the Maddala R2,McFadden R2, and the pseudo R2 are also higher than for the other two models. Theshare model exhibited the lowest overall goodness of fit measurements.The variable representing income is expected to cause some imprecise estimatessince students could not possibly predict the actual amount of income they will be earningin the next few years, but rather their expected income. The linear specification was themodel chosen as the correct model for this survey since it conforms to Hanemann’s (1984)utility theory and it has a reasonable goodness of fit.Dropping the socioeconomic variable from the regression equations did not haveany affect on the fmal estimated willingness-to-pay. In all scenarios, the E(WTP) was84$559. Possibly, the upper tail is so “fat” tail.7.8 Welfare EstimatesDue to the concern that the results of dichotomous choice questionaires aresensitive to the functional form used (Desvousges et al. 1992; Bowker and Stoll 1988; andBoyle 1990), several different functional forms were used to test the reliability of theresults found in this paper.The results suggested that the functional form did not affect the estimated meanWTP when using the standard parametric approach of numerically estimating the areaunder the estimated c.d.f. However, when using the censored logistic regression approach(Cameron 1988), the difference in E(WTP) between the log-linear form and the linearform is more apparent (discussed below in Section 7.8.2).7.8.1 Integrating Under The C.D.F. And Other ApproachesBy numerically integrating the area under each estimated willingness-to-payfunction over the range of bid amounts at the mean values of the independent variables(equation 6.3), an annual estimated WTP per person was obtained. The average estimatedWTP from this survey is approximately $558/yr. This amount is considerably higher thanthat obtained other surveys (Void et al. 1994). However, comparing the results from thissurvey with the results of the first survey, one can see that there is only a difference of$185. One possible reason for such a high value is that the bid amounts varied from $7585to $1200. Most other studies do not go as high. The few studies that did employ higherbids also exhibited higher WTP values than the norm. As mentioned earlier, the problemof “yea” saying is quite common in dichotomous choice questionnaires, and even moreso in a controversial situation such as that considered here. However, as noted previously,“yea”-saying is much more prevalent in surveys where an interviewer is present.Similarly, due to difficulties in the truncation procedure for the parametricapproaches, several alternative welfare measures were implemented. The nonparametric(Kristrom 1990), parametric (Bishop and Heberlain 1979, Sellar et al. 1986) and thecensored logistic regression (Cameron 1988) were all implemented using the data obtainedfrom the classroom surveys. The nonparametric approach (described in Section 6.6)became difficult because the sequence of proportion of “yes” responses to the bid amountsdid not form a monotone nonincreasing sequence of proportions. This was due to the factthat once the protest bids were removed from the data set, 100 percent of the remainingrespondents receiving the offer of the highest bid ($1200) agreed to the amount (seeFigure 7.3). Even through modification using the correction factor suggested by Ayer etal. (1955), the results were not considered to be a good estimate of E(WTP) and,therefore, this approach was dismissed.The parametric approach also had its problems since the mean is influenced by theupper tail of the distribution. Due to the problem of a “fat” tail of the estimated c.d.f.(i.e., the probability of accepting the offer was considerably greater than zero at very highbids), integrating to infinity would lead to excessively high WTP values. Consequently,Figure7.37515020025030035045050060065075085095010501200BID($)---PROTESTREMOVED—c--—-PROTESTINCLUDEDPERCENTAGEOFRESPONDENTSWHOANSWERED“YES”TOTHEBIDAMOUNT.-100 90 80 70__60Cl)w >- 040 30 20 10 0I-m..II—II—87the range of integration was truncated at a much lower amount (the maximum bid in thequestionnaire). However, as Boyle et a!. (1988) point out, this is not statistically correct.Therefore, to correct this problem a normalization procedure outlined by Boyle et at. aswell as Duffield and Patterson’s (1991) recommendation to truncate the model at a pointT and replace all WTP values greater than T with T) were used. Both procedures resultedin large variations in results depending on the point of truncation. The values varied from$558 when the maximum bid ($1200) was chosen as the truncation point to $409 when$850 was chosen as the truncation point. The reason this value was chosen as atruncation point was due to the fact that a majority of the respondents whose WTP weregreater than this amount were considered to be endowment bidders (i.e., proportioningmore than 5% of their expected annual income).However, truncating at any point where the probability of anyone accepting a bidhigher than this value is anything but close to zero, can lead to underestimates ofE(WTP). The argument for this is that you would be ignoring those who are actuallywilling to pay greater amounts. This can be minimized by choosing a truncation pointand then setting a maximum bid. Furthermore, pretesting with open-ended questionnairesand using the guidelines suggested by Boyle et a!. (1988) are extremely beneficial inavoiding the problems of “fat” tails.The use of the median as a welfare measure was not really considered in this paper.Although the use of the median ignores outliers in the data, it also tends to neglect allthose who are truly willing to pay the higher offers. Since the results indicate a large88percentage of respondents who are willing to pay the maximum bid, it would beinappropriate to disregard any of those respondents.7.8.2 Censored Logistic Regression AppmachThe results of the censored logistic regression indicate that the linear functionalform yields an overall higher E(WTP) than the log-linear form ($408 versus $326.35).Similarly, when the offer amount was the only independent variable used in theregression, the E(WTP) for the linear model was $360.61, while the E(WTP) for the log-linear model was only $308.96. When income was dropped from the linear model, themean WTP dropped by $5.56 to $402.44. The justification for dropping the variable wasdue to the fact that they were expected annual earnings and thus could not be consideredan ideal indicator of income.”A regression using just the offer amount and preservation amount was run and theresults are shown in Table 7.7. As can be seen, the regression is very sensitive to theaddition of socioeconomic variables (i.e., WTP varied from $360- $408). As in the firstsurvey, the opinion questions were dropped due to some unforeseen problem linking thevariables together. Although the pseudo R2 value and number of right predictionsincreased considerably, most of the opinion questions were either insignificant or of thewrong sign (which in itself is considered an indication of multicollinearity).“Student’s expected earnings varied considerably, but for no apparent reasons other than the student’sarea of study and marks.89It is clear from Table 7.7 that the results obtained through numerical integration(see Table 7.6) differ significantly from those obtained through Cameron’s (1988)censored logistic regression approach. This is unusual, as both approaches should yieldidentical results. One possible explanation is that the upper end of the tail of the c.d.f.is so fat that the numerical integration technique is just too obscure to make any relevantjudgements.Table 7.7: Estimated Mean Willingness-to-Pay Using Cameron’sCensored Logistic Regression ApproachFunctional Variables EstimatedForm Included In Mean WTPRegressionLinear anit $360.61Linear amt, pre $391Linear amt, pre, gen, $402.44cl, co, meLinear amt, pre, in, $408gen, cl, meLog-linear Inamt $308.96Log-linear Inamt, Inpre $319.26Log-linear lnamt, Inpre, $326.35Inin, gen, cl,meThe estimated regression equations using both the linear and log-linearspecifications are as follows:90(7.la) In[Pi/(1-Pi)] = 2.8 14 - 0.49lInamt(1.99) (-2.10)(7.lb) ln[Pi/(1-Pi)] = -3.230 - 0.63SInamt + 2.987lnpre(-1.51) (-2.49) (3.80)(7.lc) In[Pi/(1-Pi)] = -16.324 - 0.837lnamt + 3.486Inpre + 1.30lhiin -(-2.11) (-3.00) (4.09) (1.86)0.487c1 - 0.854gen + 0.784me(-1.26) (-2.10) (1.18)(7.ld) In{Pi/(1-Pi)] = 0.325 - 0.0009amt(1.02) (-1.69)(7.le) In[Pi/(1-Pi)] = -2.55 - 0.OOl2amt + 0.29lpre(-3.12) (-2.08) (3.80)(7.lf) in[Pi/(1-Pi)] = -2.062 - 0.OOl5amt + 0.324pre - 0.404c1 - 0.748gen -(-2.32) (-2.40) (3.97) (-1.07) (-1.89)0.l4lco + 0.688me(-0.22) (0.95)(7.lg) ln{Pi/(1-Pi)J = -2.875 - 0.OOl6amt + 0.332pre + 0.224in - 0.847gen +(-2.74) (-2.57) (4.08) (1.56) (-2.10)0.682me - 0.54 id(1.04) (-1.38)The corresponding censored logistic regressions for the above equations are:(7.2a) E(WTP) = 308.96(1.56)(7.2b) E(WTP) = -5.084 + 4.70lnpre(-1.46) (3.76)(7.2c) E(WTP) = -19.51 + 4.l7lnpre + 1.55mm - 0.58c1 - 1.O2gen + 0.94me(-2.06) (3.96) (1.85) (-1.25) (-2.05) (1.15)91(7.2d) E(WTP) = 360.61(1.01)(7.2e) E(WTP) = -2124.0 + 242.68pre(-3.03) (3.80)(7.2f) E(WTP) = -1382.44 + 216.97pre - 271.19c1 - 501.27gen - 94.62co +(-2.19) (3.79) (-1.02) (-1.72) (-0.23)461 .75me(0.82)(7.2g) E(WTP) = -1756.24 + 203.O6pre + 136.7in - 577.3lgen - 330.75c1 +(-2.03) (4.03) (1.45) (-1.98) (-1.31)416.59me(0.99)7.8.2.1 Removal of BidsThe sensitivity of the model to the removal of bids can be seen when the linearmodel in equation (2f) is re-run with the three highest bids removed (see Table 7.8). Therationale for removing the three highest bids is that the majority of respondents whoseWTP is greater than $850 had allocated a large percentage of their expected earnings.Table 7.8: Results of Removing The Three Highest BidsEstimation Procedure All Bids Included Highest Three Bids(Mean WTP) Removed(Mean WTP)Numerical Integration $559 $404Censored Logistical $408 $367Regression92From the table one can see that the removal of the highest three bids yields a muchlower mean WTP. However, there is a difference of $155 when the estimation procedureof numerical integration is used, and only a $41 difference with the censored logisticregression approach. Thus, the procedure for integration is much more sensitive totruncation procedures than the alternative method. Similar results have been found byCooper and Loomis (1992) and Desvousges et at. (1992).7.9 Breakdown of WTP Based on Desired Level of B.C.’s Wilderness AreasThe mean WTP per respondent that has been broken down based on the level ofdesired wilderness protection can be found in Tables 7.9 and 7.10. For each level ofprotection (6% - 15%), Table 7.9 shows the number of respondents who desired eachamount and the corresponding average WTP for that level. As can be seen from the table,the majority of respondents indicated that 10% was the ideal amount of wildernessprotection. Also shown in Table 7.9 is that the average WTP for protection of 7 percentto 9 percent of the land base was greater than the average WTP for 10 through 12 percentof the land base (i.e., $667-$431 versus $399-$397).The estimated mean WTP from running separate regressions on the different levelsof preservation can be found in Table 7.10. The estimated WTP were calculated usingthe censored regression approach. One can see from Table 7.10 that as the level ofprotection increases the mean WTP almost doubles in comparison.93Table 7.9: Average WTP ($) For Each Level of ProtectionNUMBER OFLEVEL OF RESPONDENTSPROTECTION (WTP> 0) AVERAGE WTP6% 12 (3) $1507% 9 (3) $6678% 14 (1) $5009% 9 (9) $43110% 54 (25) $39911% 4 (2) $32512% 14 (8) $39713% 10 (7) $63614% 5 (4) $68815% 16 (11) $566Table 7.10: Mean ($) WTP Based on Level of ProtectionLEVEL OFPROTECTION WTP6-9% $28610 - 12% $44613-15% 1 $9647.10 Aggregate Willingness-To-PayWith approximately 1.3 million households in the province (Census Canada 1992),and benefits of $373.15 per household per year, the total annual benefits to British94Columbians is roughly $484 million for increased wilderness protection in B.C.Using the estimated WTP from the classroom survey of $326 per respondent peryear, the estimated annual benefits for wilderness protection in B.C. is $716 million basedon a total adult population of 2,196,300 in B.C. This value is based on an average of10.75% of the province’s land base being protected. An important point to make is thatsince the classroom surveys were administered to students in both forestry economics andland use economics the value is not considered to be good representation of the adultpopulation in B.C. as a whole.95CHAPTER 8SUMMARY AND CONCLUSIONS8.1 SummaryThis research on the benefits of increased wilderness preservation entaileddevelopment of contingent valuation surveys to elicit residents’ WTP for the province ofBritish Columbia. The study came about after the Protected Areas Strategy (PAS)proposed an additional 6% of B.C.’s land base be set aside for protection. Two surveyswere used: one survey was distributed province-wide, while the other was issued to thirdand fourth year university students in both land use and forestry economics.An open-ended format was chosen for the mailout survey, while a dichotomouschoice format was chosen in the survey of students. A logistic model was used tocalculate the probability of a person agreeing to pay to a pre-determined offer amount.The results of the province-wide survey indicated that respondents valued additionalwilderness protection in British Columbia at $371.34 per household per year.Aggregating this amount to include all B.C. households yielded a value of $484 millionper year. The results of the classroom survey showed that the respondents’ WTP was$326 per person per year for a total of $716 miffion when aggregated for the whole adultpopulation in B.C. Similarly, the province-wide survey included individuals of alleducational backgrounds while the classroom survey included only individuals with postsecondary education.968.2 ConclusionsProblems with the model included outliers in the set of observations from both theprovincial survey and the classroom survey. In the classroom survey, once protestresponses were removed, all responded who were asked to pay $1200 to increase the levelof protection by X% agreed to do so. This does not meet a priori expectations.Logically, the higher the offer, the greater the probability of a “no” response. However,in the classroom case the findings showed an increase in the probability of a “yes” as theoffer increased in amount. This result made it very difficult to calculate a correct valueof WTP. As a consequence, the tail of the distribution function did not converge. Usingthe median instead of the mean would have helped minimize the problem of outliers inthe data. However, since the results from this survey could not be aggregated over thepopulation, the median was not used. Furthermore, the use of the median would haveeliminated those who were genuinely willing to pay the maximum bid.The results also showed that there were significant differences between the finalWTP estimates depending on the choice of welfare measure employed. The censoredlogistic regression approach seemed to give the best results for this study, while the otherapproaches were heavily influenced by the “fat” tail of the c.d.f. Carefully selecting therange of the offer amounts may reduce the occurence of this problem in future studies.There did not seem to be any variability in WTP values across different functionalforms. Several of the most commonly applied functional forms for discrete response datawere examined in this study, but there did not seem to be any notable disparities in WTP97values between model specifications.The results of the classroom survey can only be taken in a broad sense. Since theamount a person is willing to pay for increased protection of an environmental good issubject to his/her budget constraint, it is important to know the level of income therespondent earns. In the classroom surveys the best the respondent could give was his/herexpected earnings five years after completion of his post-secondary education.It is impossible to know how accurate the student was in estimating their futureearnings as estimates varied between students from $20,000-$25,000 to $80,000 and over.Generally, however, university students tend to. belong to higher income classes and havepotentially higher earnings than the average population.The timing of the survey is expected to have had some impact on the results ofthis research. As mentioned, the province-wide survey was distributed during thecontroversial Clayoquot Sound conflict. Due to the nature of the survey, it is expectedthat many of the respondents were biased in one manner or another. Many of theserespondents bid a large proportion of their annual earnings and were, therefore, eliminatedfrom the data, while others were not as easy to identify. The application of Heckman’s(1979) procedure is expected to have eliminated the possibility of sample selectivity bias.The results of this paper indicate that an average of 10 percent was the ideal levelof protection desired by respondents from both the provincial survey and the classroomsurvey. In both surveys respondents were willing to pay large amounts of money toprotect B.C.’s wilderness areas. Thus, the governments’ decision to double the amount of98wilderness protection was on average comparable to the desired level illustrated in thispaper.8.3 Future RecommendationsThis paper has examined the use of contingent valuation in determining the valueof wilderness in B.C. Although there have been many improvements in the use ofdiscrete choice questionnaires, the use of either the double bounded (systematically givingthe respondent a second offer, higher or lower than the initial bid depending on theresponse to the first one) or even the triple bounded models would allow the researcherto obtain a more accurate estimate of WTP as there is much more information with theseadvanced approaches with respect to the range of the respondent’s WTP.The surveys used in this paper could have been improved upon in many ways. Forexample, the range of offers seemed to cause problems in the fmal estimation of meanWTP. Thus, the approach suggested by Boyle et at. (1988) in the determination of offeramounts (i.e., pretest, bid range, etc.) would have enhanced the findings of this paper.It is highly recommended that, in the future, unless under a tight budget constraint,pre-letters and follow-up phone calls be implemented in order to increase the responserate. If the sample size is not too large, telephone surveys may be an alternative to thoseunder a strict budget However, one should be cautious of the biases that may result frompersonal interviews.It is expected that respondents were knowledgeable in the conflict between99preservationists and timber companies. However, it is not certain as to the familiarity ofthe participants with the benefits (attributes) of old-growth forests (e.g., biodiversity). Nordo we have any means of determining which attributes, if any, people value most inregards to old growth. An alternative measure would be to test whether or not people canjudge the difference between an old-growth and second-growth forest based on theircomposition. Establishing which attributes (ecological and non-ecological) are mostimportant to people, then determining which attributes are solely dependent on old-growthforests and those which can be found in second-growth forests may help the researcherimmensely in analyzing the respondent’s preferences concerning old-growth forests.The use of conjoint analysis could be used to infer preferences for multi-attributeamenities such as old-growth forests. Conjoint analysis can also be used to:1) Determine the utility value of each attribute;2) Determine the importance of old growth as a collection of these attributes;3) Determine the trade-offs between attributes;4) Estimate the effects of a loss in any one of the attributes; andFurthermore, a respondent’s knowledge and familiarity of the subject can beenhanced with the use of photographs and a system of computer simulations developedby the geographic information system (GIS).100ReferencesAyer, Miriam, H.D. Brunk, G.M Ewing, and Edward Silverman, 1955. “An EmpiricalDistribution Function for Sampling with Incomplete Information.” A nnuals ofMathematical Statistics, 26(Mar.) :641-47.B.C. Ministry of Forests, 1991. Outdoor Recreation Survey 1989/90. How BritishColumbians Use and Value their Public Forest Lands for Recreation. RecreationBranch Technical Report 199 1-1. Victoria: Queen’s Printer for British Columbia.B.C. Ministry of Forests, 1992. Old Growth Strategy for British Columbia. Victoria:Queen’s Printer for British Columbia.B.C. Ministry of Forests, 1992. An Inventory of Undeveloped Watersheds in BritishColumbia. Victoria: Queen’s Printer for British Columbia.Bishop, R.C., and TA. Heberlein, 1979. “Measuring Values of Extra-Market Goods: AreIndirect Measures Biased?” A merican Journal of A gricultural Economics,61(Dec.):926-30.Bishop, R.C., Richard Ready, 1991. “Endangered Species and the Safe MinimumStandard”, A merican Journal of A gricultural Economics, 73(2May): 309-3 12.Bowker, J.M., and J.R. Stoll, 1988. “Use of Dichotomous Nonmarket Methods to Valuethe Whooping Crane Resource.” A merican Journal of A gricultural Economics,70(May):372-8 1.Boyle, K.J., M.P. Welsh, and R.C. Bishop, 1988. “Validity of Empirical Measures ofWelfare Change: Comment” Land Economics, 64(Feb.)94-98.Brown, T.C, and T. Daniel, 1984. Modelling Forest Scenic Beauty: Concepts andApplication to Ponderosa Pine, United States Dept. of Agriculture, Research paperRM-256.Brown, T.C., M. Richards, T. Daniel, D. King, 1990. “Scenic Beauty and RecreationValue: Assessing the Relationship, Social Science and Natural ResourceRecreation Management, p.330.Brunson, M., B. Shelby, 1992. “Assessing Recreational and Scenic Quality”, Journal ofForestry, 90(7Jul): 37-41.101Bunnell, F.L., D.K. Daust, W. Kienner, L.L. Kremsater and R.K. McCann, 1991.Managing for B iodiversity in Forested Ecosystems. Victoria: Report to the ForestSector of the Old-Growth Strategy, mimeograph, July 25.Cameron, T.A., 1988. “A New Paradigm for Valuing Non-Market Goods UsingReferendum Data: Maximum Likelihood Estimation by Censored LogisticRegression.” Journal of Environmental Economics and Management, 15(Sept.) :355-79.Cooper, J., and J. Loomis, 1992. “Sensitivity of Willingness-to-Pay Estimates to BidDesign in Dichotomous Choice Contingent Valuation Models.” Land Economics,68(2)(May):21 1-224.Cummings, Ronald G., David S. Brookshire and William D. Schuzle (editors), 1986.Valuing Environmental Goods: An Assessment of the Contingent ValuationMethod. Totowa, H.J.: Rowman and Allanheld.Daly, Herman E. John B. Cobb Jr., 1989. For the Common Good. Boston: Beacon PressDesvousges, W.H., F.R. Johnson, R.W. Dunford, K.J. Boyle, S.P. Hudson, and K.N.Wilson, 1992. “Measuring Natural Resource Damages With Contingent Valuation:Tests of Validity and Reliability.” In Contingent Valuation: A Critical A ssessment.Cambridge Economics, Inc. Washington, D.C., April 2-3, 1992.Diliman, D.A., 1978. Mail and Telephone Surveys: The Total Design Method. NewYork: John Wiley and Sons, Inc.Duffield, J.W., and D.A. Patterson, 1991. “Inference and Optimal Design for a WelfareMeasure in Dichotomous Choice Contingent Valuation”. Land Economics,62(May):225-39.Edwards, S.F., and G.D. Anderson, 1987. “Overlooked Biases in Contingent ValuationSurveys: Some Considerations.” Land Economics, 63(May):168-78.Ebrenfeld, David, 1988. “Why Put a Value On Biodiversity”. In B iodiversity, edited byE.O. Wilson. Washington, D.C.: National Academy Press. pp. 212-216.Ehrenfeld, David, 1970. B iological Conservation. New York: Holt, Rinehart and Winston,Inc.102Focus on Resources and Our Environment: A Report on Recent Initiatives of the BritishColumbia Government. Government of British Columbia Legislative Buildings:Victoria, B.C.. March 1992.Gov. of British Columbia, 1992. “Protected Areas Strategy”. News Release 1992:EL068.N. Vancouver: Ministry of Environment, Land & Parks and Ministry of Forests,May 6.Coy, of British Columbia, 1992. “Khutzeymateen Valley to be Protected Area For GrizzlyBears.” News Release 1992:049. Ministry of Environment, Land & Parks andMinistry of Forests, June 4.Gov. of British Columbia, 1992. “Undeveloped Watershed Inventory Completed”. NewsRelease 1992:104. Ministry of Forests, October 4.Green, P.E., V. Rao, 1971. “Conjoint Measurement for Quantifying Judgmental Data”,Journal of Marketing Research, 8(Aug): 355-63/Green, P.E., F. Carmone, and Y. Wind, 1972. “Subjective Evaluation Models andConjoint Measurement”, Behavioral Science, Vol. 17:288-299.Hagen, D.A., J.W. Vincent, and P.G. Welle, 1992. “Benefits of Preserving Old-GrowthForests and the Spotted Owl.” Contemporary Policy Issues, vol.X(Apr.):13-26.Hanemaun, M.W., 1988. “Economics and the Preservation of Biodiversity”. InBiodiversity, edited by E.O.Wilson with F.M. Peter. Washington, D.C.: NationalAcademy Press. pp. 193-199.Hanemann, M.W., 1984. “Welfare Evaluations in Contingent Valuations Experimentswith Discrete Responses.” American Journal of A gricultural Economics,66(Aug.) :332-41.Hanley, N., Robin Ruffell, 1992. The Valuation of Forest Characteristics. Queen’sDiscussion Paper #849, Institute for Economic Research.Hanley, N., Alistair Munro, 1992. The Effects of Information in Contingent Markets forEnvironmental Goods: A Survey and Some New Evidence. Queen’s DiscussionPaper #848, Institute for Economic Research.Heckman, J.J., 1979. “Sample Selection Bias as a Specification Error.” Econometrica,47:153-61.103Johnson, R.L., N. Stewart Bregenzer, and Bo Shelby, 1990. “Contingent ValuationQuestion Formats: Dichotomous Choice Versus Open-Ended Responses.” InEconomic Valuation of Natural Resources: Issues, Theory, and Applications, editedby Rebecca L Johnson and Gary V. Johnson, Social Behaviour and NaturalResources Series. Boulder, CO: Westview Press.Kahneman, D., Jack L. Knetsch, 1992. “Valuing Public Goods: The Purchase of MoralSatisfaction, Journal of Environmental Economics and Management, 22(lJan): 57-70.Kahneman, D., Jack L. Knetsch, 1992. “Contingent Valuation and The Value of PublicGoods: Reply”, Journal of Environmental Economics and Management, 22(lJan):90-94.Kealy, M.J., J.F. Dovidio, and M.L. Rockel, 1988. “Accuracy in Valuation is a Matterof Degree.” Land Economics, 64(May): 158-70.Kealy, M.J., J.F. Dovidio, and M. Montgomery, 1990. “Reliability and PredictiveValidity of Contingent Values:Does the Nature of the Good Matter?” Journal ofEnvironmental Economics and Managem ent,(Nov.) :244-263.Kenton, R. Miller, 1991. B iodiversity and the Forest Profession. H.R. MacMillan, ForestryLecture Series, UBC. November 7.Kristrom, Bengt, 1990. “A Non-Parametric Approach to the Estimation of WelfareMeasures in Discrete Response Valuation Studies.” Land Economics, 66(May):136-39.Leduc, George, Robert Goodland, 1988. Wildiands: Their Protection and Management inEconomic Development. Washington: World Bank.Leefers, Larry A. (1992). Analyzing Old-Growth Designation Using Ecological andA nalytical Hierarchies. Paper presented at the Workshop on HierarchicalApproaches to Forest Management in Public and Private Organizations held inToronto, ON, May 25-29.Ludwig, D., John Conrad, 1991. Opportunity Cost of Preservation of Old Growth and thePresent Value of Silviculture, FEPA Research Unit Working Paper No. 165,University of B.C., August.104Malcolm L. Hunter, Jr., 1988. Wildlife, Forests, and Forestry. Englewood Cliffs, N.J.:Prentice Hall.McKillop, W., 1992. “Use of Contingent Valuation in Northern Spotted Owl Studies: ACritique”, Journal of Forestry, 90(8Aug): 36-37.McDaniels, T., 1992. A Prescriptive Decision Framework for Preservation ofBiodiversity, Westwater Research Center and School of Community and RegionalPlanning.Muon, J.W., 1989. “Contingent Valuation Experiments for Strategic Behavior.” Journalof Environmental Economics and Management, 17. pp.293-308Milon, J.W., Roger Clemmons, 1991. “Hunters’ Demand for Species Variety”, LandEconomics, 67(4Nov): 401-412.Munasinghe, M., 1992. Environmental Economics and Valuation in DevelopmentDecisionmaking, ENV Working Paper No. 51, World Bank, Washington, D.C.,February 1992.Norgaard, R.B., 1988. “The Rise of the Global Exchange Economy and the Loss ofBiological Diversity”. In Biodiversity, edited by E.O. Wilson with F.M. Peter.Washington, D.C.: National Academy Press. pp. 206-211.Norton, Bryan, 1988. “Commodity, Amenity, and Morality: The Limits of Quantificationin Valuing Biodiversity”. In B iodiversity, edited by E.O. Wilson with F.M Peter.Washington, D.C.: National Academy Press. pp. 200-205.Old-Growth Definition Task Force, 1986. Interim definitions for old-growth Douglas-firand mixed confier forests in the Pacific Northwest and California. USDA ForestService, Pacific Northwest Research Station Research Note PNW-447, Portland,Oregon. ‘7ppPatrick, G.F., E. Loehman, A. Fernandez, 1983. Measuring Trade-Offs: An Applicationof Conjoint A nalysis. American Agricultural Economics Association Meetings,Purdue Univ., July 31-Aug. 3.Patterson, D.A. and J.W. Duffield, 1991. “Comment on Cameron’s Censored LogisticRegression Model for Referendum Data.” Journal of Environmental Economics andManagement, (May):275-283.105Pope ifi, C.A., and J.W. Jones, 1990. “Value of Wilderness Designation in Utah.”Journal of Environmental Economics and Management, 30: 157-174.Randall, Alan, 1988. “What Mainstream Economists Have to Say About the Value ofBiodiversity”. In Biodiversity, edited by E.O. Wilson with F.M. Peter.Washington, D.C.: National Academy Press. pp. 217-223.Randall, Alan, 1991. Thinking About the Value of Biodiversity. Columbus, Ohio: Dept.of Agric. Econ. and Rural Sociology, Ohio State Univ., mimeograph.Rockel, M.L., and M.J. Kealy, 1991. “The Value of Nonconsumptive Wildlife Recreationin the United States.” Land Economics, 67(Nov.):422-34.Roloff, G., J. Haufler (1991). Incorporating Wildlife Objectives into Forest Planning.Department of Fisheries and Wildlife, Michigan State University.Rubin, J., G. Helfand, J. Loomis, 1991. “A Benefit-Cost Analysis of the NorthernSpotted Owl”, Journal of Forestry, 89(l2Dec): 25-30.Schreier, H., W.A. Thompson, G.C. van Kooten , I. Vertinsky, 1992. A DecomposedHierarchical System for Forest Land Use Allocation Decisions With PublicParticipation, Forest Economics and Policy Analysis Research Unit, Univ. of B.C.Sellar, C., J.R. Stoll, and J.P. Chavas, 1986. “Specification of The Logit Model: TheCase of Valuation of Nomnarket Goods.” Journal of Environmental Economics andManagement, 1 3(Dec.):382-90.Sellar, C., J.R. Stoll, and J.P. Chavas, 1985. “Validation of Empirical Measures ofWelfare Change: A Comparison of Nonmarket Techniques.” Land Economics,61(May): 156-75.Shoup, David W. (1981). Using Conjoint Analysis to Forecast Saleable Machine Features.Paper No. 81-1518 presented at the American Society of Agricultural Engineerswinters meetings, Chicago.Sinden, J.A. and A.C. Warrell, 1979. Unpriced Values: Decisions Without MarketPrices. New York: Wiley.Smith, K.V., 1992. “Arbitrary Values, Good Causes, and Premature Verdicts”, Journalof Environmental Economics and Management, 22(lJan): 71-89.106Stevens, T.H., Jaime Echeverria, Ronald J. Glass, Tim Hager, Thomas A. More, 1991.“Measuring the Existence Value of Wildlife: What Do CVM Estimates ReallyShow?”, Land Economics, 67(4Nov): pp. 390-400.Old Growth Strategy Project. Towards an Old Growth Strategy: Public Review Draft,Ministry of Forests, Jan. 1992.Tyrrell, L.E., 1991. Old-Growth Forests on National Park Service Lands: NPS Viewsand Information. Great Lakes Cooperative Park Studies Unit #91-1, Univ. ofWisconsin-Madison.van Kooten, G.C., 1991. Land Resource Economics and Sustainable Development: AnIntroduction to the Resolution of Land Use Conflicts. Vancouver, B.C.: Dept. ofAgric. Econ. and Forest Resources Management, University of British Columbia.van Kooten, G.C., R. Schoney, K. Hayward (1986). “An Alternative Approach to theEvaluation of Goal Hierarchies among Farmers.” Western Journal of A griculturalEconomics, 1 1(lJuly):40-48.Walsh, R.G., R.D. Bjonback, R.A. Aiken, and D.H. Rosenthal, 1990. “Estimating thePublic Benefits of Protecting Forest Quality.” Journal of Environmental Economicsand Management, 30: 175-189.Wilson, E.O., 1989. “Threats to Biodiversity”, Scientific American (September): 108-114.107APPENDIX 1A - PROVINCIAL SURVEYSURVEYPROTECTING B.C.tS FORESTLANDS: DECISIONS FOR THE FUTUREB ackgmundProtection of old-growth forests in British Columbia is a controversial issue: at oneextreme is the view that all old-growth forests should be available for logging; on theother, that all old growth must be preserved. Old growth differs between the Coast andInterior regions of the Province. On the Coast, old growth refers to areas where trees are200 years or older; the forest is undisturbed (no harvesting of timber has occurred); thereis a variety of large living trees, standing dead trees (snags), and fallen trees and debris;and the canopy gives an umbrella-like effect. Interior forests are frequently disturbed byfire, so trees tend to be less magnificent and it is more difficult to distinguish betweenareas that have never been harvested and those that were harvested quite some time ago.Yet Interior old growth provides some unique features that are not available in stands thathave been harvested. The distinction is a direct result of the inevitable disturbance (e.g.,road building) that results from forestry operations. The economic value from harvestingold growth in the Interior is much smaller than on the Coast because trees do not, ingeneral, attain the size they do on the Coast.Since citizens of B.C. are the owners of about 95 per cent of the forest resourcesof the province, changes in harvest and other uses affect them financially. This surveyseeks to determine how you value the preservation of forest resources. The surveyconsists of four sections, followed by a page providing additional information about B.C.forest lands. The information page is located on the back of this panel.This panel can be removed if desired.ANSWERS PROVIDED WILL BE KEPT IN STRICT CONFIDENCE108Provide an environment for recreational activities(such as hiking, canoeing, wildlife viewing, etc.)Provide timber and manufactured forest productsProvide information on medical compoundsProvide tourism incentivesProvide scientific research and education(i.e. how ecosystems work)Provide opportunities for hunting, trappingand fishingProvide a large variety of habitats for a largediversity of speciesProvide general ecosystem benefits, such asregulation of waterflow and absorption ofcarbon dioxide from the airProvide goods and services that you or othersmay want to use in the futureProvide spiritual and aesthetic values(scenic beauty)Provide wilderness value (access to an area notdisturbed for commercial purposes)Provide wilderness areas, even if theseareas are never usedProvide value knowing that this resource will beavailable to future generationsProvide you opportunities for gathering plantproducts for culinary and/or decorative purposes.Section 1: Opinion QuestionsHow important to you are the following benefits from British Columbia’s forest resources.(Please circle the number that best represents your response to the statement indicated).NotExtremely liiipalantImportant Neutral At All5 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 15 4 3 2 1109Section 2: Protecting Old-Growth Forests from Timber HarvestAbout 6.5% or 6 million ha of B.C.’s total land base is currently protected in a variety ofways (national and provincial parks, wildlife refuges, recreation areas, etc.). The amount of landto be protected is going to be increased from about 6 miffion ha to some greater amount. Foreach level of protection, the average person will be required to pay through taxes the amountindicated in perpetuity. This represents our best available, but still rough, estimate of the averagecost to a B.C. household for increasing levels of preservation, but it does not include costs to theenvironment or costs of developing the area for other uses (e.g., developing access roads orcampsites).Please circle one entry in each row indicating whether or not you would be willingthrough taxes for provincial land to be protected.An answer is needed in each row.Land base that is to be protected Actualincreases from 6.0 mil.ha to: Cost to you: Would you pay?14.0 mil.ha.(14.8% of provincial land base) $1,200/year YES NO13.0 mil. ha. (13.7% of provincial land base) $1,050/year YES NO12.0 mil. ha. (12.6% of provincial land base) $ 900/year YES NO11.0 mil. ha. (11.6% of provincial land base) $ 725/year YES NO10.0 mil. ha (10.5% of provincial land base) $ 575/year YES NO9.0 mil. ha (9.5% of provincial land base) $ 400/year YES NO8.0 mil. ha (8.4% of provincial land base) $ 250/year YES NO7.0 mil. ha (7.4% of provincial land base) $ 75/year YES NOCurrent level of protection is adequate YES NO110Section 3: Allocating a Government BudgetThe government has a limited budget for forest protection. You will have in mind some specificbenefits that you value more than others and would like to see these emphasized by the publicforest manager. If you were asked to allocate the provincial government’s forest protectionbudget, how would you allocate it among the items listed.The total should sum to 100 %.BENEFITS: %Preserve species and their habitatsProtect forests for scientific researchand education purposesReserve forests for the following alternative uses:Wilderness recreationHuntingFishing and gatheringOther recreationOther reasons (Please specify):TOTAL 100 %111Section 4: Personal Infonnation1. a) What is your age? (Please check one)25 or under 26-35 36-4546-55 56-65 over 65b) Are you: — Male— Female2. Including yourself, how many individuals are there in your household? —3. What is your level of education? (Please circle)Secondary (Grade): 8 9 10 11 12Post Secondary (Trade School, University, etc.)Years: 1 2 3 4 5 6 7 or more4. What was the approximate gross (before tax) income of your household in 1992? (Checkone)— less than $20,000— $50,000 to less than $60,000— $20,000 to less than $30,000— $50,000 to less than $70,000— $30,000 to less than $40,000— $70,000 to less than $80,000— $40,000 to less than $50,000 $80,000 and over5. a) Do you regularly contribute to any environmental organizations, such as the WorldWildlife Fund? (Please circle)YES NOb) Are you a member of any environmental organizations? (Please circle)YES NOTHANK YOU FOR TAKING THE TIME TO COMPLETETifiS QUESTIONNAIRE112Information PageProtection of Foresfiands in British ColumbiaHow much is protected?Canada’s total land area is 997 million hectares (ha). Of this area, 45% (453 million ha)is designated forest land, with 244 million ha considered productive forest. About 4% of theproductive forest land base or 10 million ha is reserved in national and provincial parks. Largeareas of forested land are also excluded from commercial harvesting simply because it is noteconomically feasible to harvest trees in these areas.B.C. is Canada’s leading lumber producer and exporter. Total provincial land area is 94.8million ha, of which 54% (51.2 miffion ha) is designated productive forestland (96% is publiclyowned, the remainder privately held). The net operable land base that is suitable and designatedfor timber harvest is about half the productive forest area; the Coast accounts for 3.7 million haof operable forest, and the Interior 21.5 million ha. Some 6.5% of the total land base or 6.3million ha is currently reserved in national and provincial parks, regional parks, recreation areas,private conservation areas, wildlife and migratory bird sanctuaries, et cetera. Current governmentpolicy is to increase the amount of provincial land protected in parks and ecological reserves toabout 12.5 million ha, or about 12% of the land base. Some of this increase will come from thenet operable land base that is currently scheduled for timber harvest.What is the cost of protection?A ban on logging has economic costs that will be borne by B.C. residents.(1) An important cost is the net value of the lost timber which, if harvested, would contributeto your overall financial wealth. This loss shows up as lower government revenues, lower levelsof investment and employment in forest product industries, and likely higher prices to consumersfor forest products.(2) There will be costs to governments due to lost jobs, community instability, and so on.These costs are hard to quantify but include such things as unemployment compensation, welfarepayments, retraining allowances, higher government administration costs, and so on.(3) Finally, decreased timber output in B.C. may have national and global environmentaleffects as a result of using non-wood substitutes or increasing harvests in areas (such as thetropics) with more fragile ecosystems.What are some benefits of protection?British Columbia has the greatest number and most diversity of species of any province(e.g., 55% of Canada’s birds breed only in B.C.). Some 80 wildlife species strongly depend onold-growth forests. By harvesting all old growth, some known species could become extinct, asmay species that have not yet been discovered. Uncertainty as to the possible use of these species(e.g. medical compounds, scientific research) is of concern to everyone. It is clear that B.C.’sforest lands are important for maintaining biodiversity, although appropriate management oftimber harvests and forests can prevent extinction of most (but likely not all) species. Protectionof forests also provides benefits from recreational opportunities, hunting, fishing, hiking, viewing,and simply knowing that old-growth forests and wilderness areas exist in B.C. Development offorest policies that take into account biodiversity, recreation, timber and other values will beimportant for future management of B.C.’s forest lands.113APPENDIX lB - CLASSROOM SURVEYSURVEYPROTECTING B.C.’S WILDERNESS AREAS: DECISIONS FOR THE FUTUREB ackgmundCanada’s total land area is 997 million hectares (ha). Of this area, 45% (453 million ha)is designated forestland, with 244 million ha (just over 1/2) considered productive forest.About 4% of the productive forestland base (but a much higher proportion of the totalland area) is reserved in national and provincial parks. Large areas of forested land arealso excluded from commercial harvesting simply because it is not economically feasibleto harvest trees in these areas.British Columbia is Canada’s leading lumber producer and exporter. Total provincialland area is 94.8 million ha, of which 54% (51.2 million ha) is designated productiveforestland (96% is publicly owned, the remainder privately held). The net operable landbase that is suitable and designated for timber harvest is about half the productive forestarea; the Coast accounts for 3.7 million ha of operable forest, and the Interior 21.5 millionha. Some 6.5% of the total land base or 6.3 million ha is currently reserved in nationaland provincial parks, regional parks, recreation areas, private conservation areas, wildlifeand migratory bird sanctuaries, et cetera. Current government policy is to increase theamount of provincial land protected in parks and ecological reserves to about 12 millionha, or about 12% of the land base. Some of this increase will come from the net operableland base that is currently scheduled for timber harvest.One reason for increasing wilderness preservation is to protect old-growth forests. Thisis a controversial issue: on one side, the view is that all old growth should be availablefor logging; on the other, that all old growth must be preserved. Old growth differsbetween the Coast and Interior regions of the Province. On the Coast, old growth refersto areas where trees are 200 years or older; the forest is undisturbed (no harvesting oftimber has occurred); there is a variety of large living trees, standing dead trees (snags),and fallen trees and debris; and, the canopy gives an umbrella-like effect. Interior forestsare frequently disturbed by fire, so trees tend to be less magnificent and it is moredifficult to distinguish between areas that have never been harvested and those that wereharvested quite some time ago. Yet Interior old growth provides some unique featuresthat are not available in stands that have been harvested. The distinction is a direct resultof the inevitable disturbance (e.g., road building) that results from forestry operations.The economic value from harvesting old growth in the Interior is much smaller than onthe Coast because trees do not, in general, attain the size they do on the Coast. Therefore,the financial cost to you from preserving old growth in the Interior is substantially lowerthan on the Coast.114Protection of old-growth forests cannot be separated from the larger issue of wildernessprotection. Old growth can only be protected in contiguous areas that constitutefmancially valuable old-growth timber, other fmancially valuable trees that are notconsidered old growth, timber than is uneconomic to harvest, and areas that are barren(e.g., glaciers and mountain tops). Old growth often constitutes 4-12% of the area thatwould be protected under the province’s Protected Areas Strategy (PAS).What is the cost of protection?A ban on logging has economic costs that will be borne by B.C. residents.(1) An important cost is the net value of the lost timber which, if harvested, wouldcontribute to your overall fmancial wealth. This loss shows up as lower governmentrevenues, lower levels of investment and employment in forest product industries, andlikely higher prices to consumers for forest products.(2) There will be costs to governments due to lost jobs, community instability, and so on.These costs are hard to quantify but include such things as unemployment compensation,welfare payments, retraining allowances, higher government administration costs, and soon. Given government debt and recurring budget deficits, a reduction in timber harvestwill increase the budgetary problems of government, leading to a reduction in otherservices or the need to cut down trees in protected areas at some future date.(3) Finally, decreased timber output in B.C. may have national and global environmentaleffects as a result of using nonwood substitutes or increasing harvests in areas (such asthe tropics) with more fragile ecosystems.What are some benefits of protection?British Columbia has the greatest number and most diversity of species of any province(e.g., 55% of Canada’s birds breed only in B.C.). Some 80 wildlife species stronglydepend on old-growth forests. By harvesting all old growth, some known species couldbecome extinct, as may species that have not yet been discovered. Uncertainty as to thepossible use of these species (e.g. medical compounds, scientific research) is of concernto everyone. It is clear that B.C.’s forest lands are important for maintaining biodiversity,although appropriate management of timber harvests and forests can prevent extinctionof most (but likely not all) species. Protection of forests also provides benefits fromrecreational opportunities, hunting, fishing, hiking, viewing, and simply knowing that oldgrowth forests and wilderness areas exist in B .C. Development of forest policies that takeinto account biodiversity, recreation, timber and other values will be important for futuremanagement of B.C.’s forest lands.ANSWERS PROVIDED WILL BE KEPT IN STRICT CONFIDENCE115Section 1: Opinion QuestionsHow important to you are the following benefits from British Columbia’s forest resources.(Please circle the number that best represents your response to the statement indicated.)NotExtremely ImportantImportant Neutral At AllProvide an environment for recreational activities (such as 1 2 3 4 5hiking, canoeing, wildlife viewing, etc.).Provide timber and manufactured forest productsProvide information on medical compoundsProvide tourism incentivesProvide scientific research and education (i.e., howecosystems work)Provide opportunities for hunting, trapping and fishingProvide a large variety of habitats for a large diversityof speciesProvide general ecosystem benefits, such as regulation ofwaterfiow and absorption of carbon dioxide from the airProvide goods and services that you or others may wantto use in the futureProvide spiritual and aesthetic values (scenic beauty)Provide wilderness value (access to an area notdisturbed for commercial purposes)I am happy just to know these areas exist, even if theyare never usedProvide value knowing that this resource will beavailable to future generationsProvide you opportunities for gathering plant productsfor culinary and/or decorative purposes1 2 3 451 2 3 4 51 2 3 451 2 3 4 51 2 3 4 51 2 3 451 2 3 4 51 2 3 4 51 2 3 451 2 3 4 51 2 3 451 2 3 4 51 2 3 4 5116Section 2: Protecting Old-Growth Forests from Timber HarvestSome 6.5% of B.C.’s total land base is currently protected in a variety of ways (national andprovincial parks, wildlife refuges, recreation areas, etc. The amount of land to be protected isgoing to be increased from about 6 million ha to some greater amount. How much of B.C.’s totalland base do you think ought to be placed in a permanent reserve, so that no harvesting of treeswill be permitted? Please circle the amount you would wish to preserve.6% 7% 8% 9% 10%11% 12% 13% 14% 15%For the level of protection that you indicated above, would you be willing to pay the amountindicated below as an annual surtax on your income in perpetuity. If you currently do not havea source of income, the accrued annual amounts (annual amount indicated below) plusaccumulated interest will be charged at the time that you begin earning an income; since you willbe paying this amount annually once you graduate, the accrued amount will not be due at onetime, but will be spread over the number of years you have spent in post-secondary education.Would you be willing to pay $1,200 per year for the level of wilderness protection indicatedabove? (Circle one)YES NOIf NO, would you be willing to pay the amount indicated for a higher level of wildernessprotection? (Circle)YES NO If YES, for a level of________%of the land base.If YES, what is the minimum level of wilderness protection that you would be willing to acceptfor the dollar amount that you indicated you would be willing to pay?________% of the land base.117Section 3: Personal Information1. a) What is your age? (Please check one.)____under 25____26-35____3 6-4546-55__56-65____over 65b) Are you: — Male — Female2. Realistically, what do you expect to be earning upon completion of ALL your postsecondary education? Please check your approximate, expected gross (before tax) incomein 1993 dollars.less than $20,000____$50,000 to less than $60,000$20,000 to less than $30,000 $60,000 to less than $70,000$30,000 to less than $40,000 $70,000 to less than $80,000$40,000 to less than $50,000 $80,000 and over5. a) Do you regularly contribute to any environmental organizations, such as the WorldWildlife Fund? (Please circle)YES NOb) Are you a member of any environmental organizations? (Please circle)YES NOTHANK YOU FOR YOUR COOPERATION

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0087549/manifest

Comment

Related Items