Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cost and benefits of improving water quality by composting livestock wastes: a contingent valuation approach Athwal, Rita Kaur 1994

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

Item Metadata


831-ubc_1994-0211.pdf [ 3.99MB ]
JSON: 831-1.0094720.json
JSON-LD: 831-1.0094720-ld.json
RDF/XML (Pretty): 831-1.0094720-rdf.xml
RDF/JSON: 831-1.0094720-rdf.json
Turtle: 831-1.0094720-turtle.txt
N-Triples: 831-1.0094720-rdf-ntriples.txt
Original Record: 831-1.0094720-source.json
Full Text

Full Text

COSTS AND BENEFITS OF IMPROVING WATER QUALITY BY COMPOSTING LIVESTOCK WASTES. A CONTINGENT VALUATION APPROACH by RITA KAUR ATHWAL B.Sc.(Agr.)/ The University of British Columbia, 1991 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Agricultural Economics)  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA April 1994 ©Rita Kaur Athwal, 1994  In presenting this thesis  in partial fulfilment  of  the  requirements  for an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or  by  his  or  her  representatives.  It  is  understood  that  copying  or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department of  ^ 4 - l A Ctt-lJICXAJC  The University of British Columbia Vancouver, Canada  Date Of^Ji  DE-6 (2/88)  &*h+.  XCJZ?vt^yi>ufe>  ABSTRACT  Athwal, Rita Kaur, B.Sc.  University of British Columbia,  Vancouver, April 1994. Costs and Benefits of Improving Water Quality by Composting Livestock Wastes. A Contingent Valuation Approach.  Major Professor: G.C. van Kooten.  Water quality tests in the Abbotsford area indicate that approximately 60% of the samples taken from wells in some regions are above the 10 mg/L maximum acceptable concentration of nitrate-nitrogen for drinking water as defined in Health and Welfare Canada's Canadian Drinking Water Quality Guidelines. According to an Environment Canada study, increased nitrate concentrations change the micro-environment in waterways and makes water vulnerable to other types of pollution problems. Livestock wastes have been identified as a major source of ground and surface water pollution.  It is both difficult and expensive  to clean up such wastes.  Scientists have identified composting as a possible solution to the animal waste disposal problem.  Composting yields a net loss  when only private costs are considered.  Linear ordinary least  squares, open-ended; linear tobit; and, log-linear logit, closedended contingent valuation models are used to calculate the amount individuals are willing to pay to reduce the nitratenitrogen concentration in water.  These calculations are  translated as social benefits of composting, and compared to pollution defense expenditures that residents of Abbotsford are ii  making to avoid the pollution problem.  The defense expenditures  are analyzed in linear ordinary least squares and linear logit regressions.  Results from the contingent valuation model indicated an annual willingness to pay of $81.03 to $139.22 per household and annual defense expenditures of $142.94 per household, which is not sufficient to cover the high losses livestock producers would incur by composting all livestock waste.  It would be more  worthwhile to clean-up major incidents of pollution and invest in research for better management practices.  iii  TABLE OF CONTENTS  ABSTRACT  11  TABLE OF CONTENTS  iv  LIST OF TABLES  vi  LIST OF FIGURES  vii  ACKNOWLEDGEMENT  viii  1.0 INTRODUCTION  1  1.1 BACKGROUND  1  1.2 PROBLEM STATEMENT  11  1.3 OBJECTIVES  11  1.4 OVERVIEW OF THESIS  13  2.0 ALTERNATIVES AND COMPOSTING  14  2.1 COMPOSTING METHODS  21  2.2 COMPOSTING COSTS  22  2.3 CONCLUSION ON COMPOSTING  27  3.0 THEORETICAL CONSIDERATIONS  29  3.1 HEDONIC PRICING  29  3.2 CONTINGENT VALUATION METHOD (CVM)  30  3.3 RANDOM UTILITY MAXIMIZATION  35  iv  TABLE OF CONTENTS (CONTINUED)  4.0 METHODOLOGY  41  4.1 CONTINGENT VALUATION  41  4.2 SURVEY DESIGN  43  4.3 WTP MODEL  54  4.4 DEFENSE EXPENDITURES  60  5.0 EMPIRICAL ANALYSIS  63  5.1 SURVEY DESCRIPTIVE RESULTS  63  5.2 WTP MODEL RESULTS  70  5.3 DEFENSE EXPENDITURES RESULTS  78  5.4 COMPARISON OF RESULTS  83  5.5 COST-BENEFIT ANALYSIS  86  6.0 CONCLUSIONS  89  BIBLIOGRAPHY  92  APPENDIX 1  98  v  LIST OF TABLES  Summary of Personal and Background Information for Abbotsford area.  66  Rent or Mortgage Payment per Month for Abbotsford Resident.  67  Abbotsford Residents' Opinion Regarding Off-Farm Pollution.  68  Description of Variables Used in Regressions.  71  Logit Regression Results of Decision Whether to Pay for Nitrate Reduction in Abbotsford Water.  74  Ordinary Least Squares Regression Results for OpenEnded Contingent Valuation Asked to Abbotsford Residents.  77  Tobit Regression Results for Open-Ended Contingent Valuation Questions Asked to Abbotsford Residents.  78  Logit Regression Results of Decision Whether to Take Averting Action for Abbotsford Residents.  80  Ordinary Least Squares Regression Results for Magnitude of Defense Expenditures for Abbotsford Residents.  82  Summary of Measurements from Willingness to Pay and Defense Expenditure Model.  86  Estimated Annual Production of Animal Waste in the Central Fraser Valley Regional District.  87  vi  LIST OF FIGURES 1.  Welfare Measures Related to Changes in Water Quality.  36  2.  The Logit Curve.  58  vii  ACKNOWLEDGEMENT Special thanks is expressed to my parents and brothers, Satvir and Ronnie, for their love and continuous support of my educational goals. I am sincerely grateful to Dr. G.C. van Kooten, my major advisor, for his attentiveness and guidance during the course of this research. Appreciation is extended to Dr. J. Vercammen and Dr. L. Lowe for their constructive advice and for serving on the thesis committee. Deep appreciation is expressed to Dr. A.R. Parikh for his guidance during the tough times. Special thanks to my friend Victoria Watson for her helpful suggestions and encouragement. Similar thanks is expressed to my friends Tina and Shai for their support. Appreciation is extended to R. Paul Dhillon for his time and effort in proofreading the document. Acknowledgement is also made to Robert Flick, Chris Gaston, Richard Porter, Brad Stennes and Gywenne Sykes for their help during the research process.  viii  CHAPTER 1  1.0 INTRODUCTION Farm animal manure has been valued as a source of plant nutrients and organic matter.  Historically, storage and uses of manure  were mainly developed for crop production.  Presently,  agriculture is moving towards more intensification and concentration.  Intensive commercial livestock, swine and poultry  have created potential hazards associated with the storage and disposal of large quantities of manure within a limited land base.  One of the major challenges facing agriculture in South  Coastal British Columbia is the agronomically and environmentally sound use of nitrogen for crop production.  Queries regarding the  safe use of animal manures are valid, despite the fact that animal densities have not reached levels found in some parts of Europe.  Manure disposal problems not only concern farmers and  agriculturalists, but also the public which is exposed to the adverse external effects of manure.  These effects include odour  from nitrogen entering the air as ammonia, and groundwater and surface water contamination by nitrates.  1.1 BACKGROUND There are several particular features of South Coastal B.C. that create unique problems in the sound use of animal manure in the region.  The B.C. Ministry of Agriculture's report (Canadian  Society of Animal Science 1991) on B.C.'s approach to environmentally sound waste management outlines the unique features.  For example, the climate, which is characterized by  wet winters and mild temperatures, is unique in the agricultural 1  areas of the western world.  As a result, comparisons made  directly to Europe for relevant guidance are limited by the greater precipitation in B.C. (Canadian Society of Animal Science 1991).  It is common knowledge that Europe has some serious environmental problems related to agriculture.  B.C.'s climate makes the risk  of contamination from agriculture significantly greater, although current practices in B.C. are less severe.  For example, although  Holland also has a mild, damp climate with temperatures similar to the lower mainland in B.C., Holland receives significantly less precipitation.  The wettest part of Holland, the extreme  southeast, receives a yearly average of more than 34 inches of precipitation (Adam et al. 1989).  The dampest part of B.C., the  western part of Vancouver Island and the Lower Mainland, receives over 100 inches (Bower et al. 1989).  European experience indicates that numbers per hectare can not increase indefinitely without creating problems of overapplication of nutrients.  This is followed by problems of forage  quality and environmental degradation.  Van Kleeck and Kowalenko  (1987) attempted to quantify the amount of nitrogen required and compare it to the amount being applied.  Van Kleeck (1986)  studied areas of high animal densities in Abbotsford and noticed that, in certain areas, crop requirements for nitrogen are significantly less than nitrogen production.  As a result,  environmental contamination occurs due to the over application of nitrogen. 2  Another feature of this region is the rural/urban interface.  The  increasing number of urbanites moving into the rural areas have the rural landscape in mind for recreation, scenic amenities and open space.  This entails some inevitable problems for farming.  In B.C. yield potential is among the highest in the country. This is due to factors such as a reasonable amount of heat, favourable moisture distribution and a long growing season, or high number of growing degree days.  In order to obtain this  potentially high yield other factors such as soil fertility also must be favourable.  In an attempt to achieve high fertility,  great quantities of fertilizer are added to the soil.  Nitrogen  is often one nutrient that is applied more than necessary as a relatively inexpensive assurance of higher yields.  Environmental Issues A sustainable future is what citizens want to be certain of, with the recently growing attention to the environment.  It is clear  from the European experience that a sustainable future includes agriculture.  Pollution is something that people are not willing  to tolerate, whether it be from industry or agriculture. Attention to agricultural practices that contribute to pollution is a recent phenomena.  Previously, this area was given little  attention due to the non-point or dispersed nature of the pollution sources (Anderson 1990).  Non-point pollutants are  difficult to measure since they cover a wide area and vary significantly according to weather conditions such as wind and precipitation. 3  According to the B.C. Ministry of Agriculture and Fisheries (BCMAF) (Canadian Society of Animal Science 1991), related environmental issues of concern include excess nitrates in drinking water, eutrophication of surface water and atmospheric ammonia.  Canadian regulations permit 10 mg/L of nitrates in  drinking water.  "Nitrates are very soluble and mobile in the  soil and hydrological system, and excess nitrate levels (above 10 mg/L) in drinking water have been linked with infant mortality (methaemoglobinaemia)"  (Hall, Schreier and Brown 1991a p. 82).  Nitrogen is an aquatic plant nutrient, thereby reducing biochemical oxygen demand (BOD), and it contributes to contamination of potable groundwater supplies.  Normally, it is  not harmful to adults; however, in some infants nitrate can convert to nitrite, which alters blood hemoglobin so it cannot transport oxygen. baby syndrome.  This results in methaemoglobinaemia or blue  The problem of nitrate contamination is  especially serious in the Lower Fraser Valley, and the major nitrate contributors are leachate from animal wastes and fertilizers (International Reference Group on Great Lakes Pollution From Land Use Activities 1978).  Samples taken by BCMAF  beginning in 1988 from well waters in the Brookswood (South Langley), Hopington and Abbotsford areas have been found to contain nitrate levels above 10 mg/L.  Piezometers or observation  wells were set up by the B.C. Ministry of Environment in March 1988 to determine the severity of the problem and the source of the nitrates.  The piezometers provide indications that the  nitrates are coming from a surface source and that in shallow 4  wells there is an increased amount of the pollutant, which decreases with depth.  Since septic tanks are not considered a  significant source of nitrates, animal manure is the most obvious source. "Groundwater from the Abbotsford area of the Lower Fraser Sub-basin is badly polluted with nitrates largely from animal wastes and agricultural fertilizers. Levels appear to be increasing and there are public health concerns for infants that can develop methaemoglobinaemia" (Hall et al. 1991a, p.71). Phosphorus is less mobile in soil and waterways because it is less soluble than nitrate.  Phosphorous reaches the aquatic  system mostly by soil erosion.  Excessive loads of phosphorus and  other plant nutrients can cause disturbing growths of algae and other aquatic plants, which is known as eutrophication.  Bird and  Rapport (1986) found that concentrations of 0.1 mg/L lead to eutrophication.  Little data on phosphorus is available for the  Lower Fraser Valley, where eutrophication is limited to back water sloughs and slow moving tributaries in the lower section of the Basin.  "Direct run-off from chemical fertilizer and manure,  input via soil erosion and urban run-off are important contributors of phosphorus"  (Hall, Kenneth and Scheier 1991a,  p.83).  Pathogenic bacteria are also contaminants found in waterways from the leaching of fecal matter.  Escheria coli, Pseudomonas  aeruginosa, and Salmonella are bacteria that indicate fecal pollution.  Diseases such as typhoid, cholera, swimmer's itch,  ear infections and leptospirosis are transmitted through fecal contamination of water.  Run-off from animal wastes and direct 5  introduction of such wastes are agricultural inputs of pathogenic bacteria (International Reference Group on Great Lakes Pollution From Land Use Activities 1978).  Another environmental issue is the eutrophication of surface water or the depletion of oxygen in surface water caused by excessive aquatic plant growth.  This excessive growth is a  result of increased quantities of plant nutrients, including phosphorus and nitrogen.  Oxygen depletion from BOD leads to fish  population decreases in streams and several "fish kills" have occurred in the region (Canadian Society of Animal Science 1991).  Atmospheric ammonia is an environmental concern because it causes acid rain.  European estimates indicate that 20% of their acid  rain is caused by ammonia and 80% of this ammonia comes from agriculture (Canadian Society of Animal Science 1991).  In  countries such as Holland, where animal manure is a significantly greater problem, laws have been enacted to control ammonia losses to the atmosphere.  It is known that air around the Abbotsford  area has increased amounts of the ammonia.  BCMAF has identified  poultry as the probable main source (van Kleeck 1992).  Current Practices There are many current practices concerning the management of manure that are of concern.  Over application is common since  animal manure is considered a waste to be disposed of instead of a source of nutrients and organic matter for crop production. Over application is almost always done in areas where animal 6  feeds are purchased instead of grown on the farm.  Bird and  Rapport (1986) determined the average fertilizer application rate for many agricultural areas in Canada to be 125-415 kg/ha/year. The highest rate was found in South-Western B.C. at approximately 625 kg/ha/year. "The Lower Fraser Sub-basin is indeed receiving the highest application of fertilizers. In fact, 52% of all fertilizers used in the Basin are applied in the Lower Fraser Sub-basin. ....[S]ome of these nutrients will find their way into the stream system" (Hall, Schreier and Brown 1991a, p.81).  Furthermore, times.  manure application often occurs at inappropriate  This is due to the lack of storage that makes the  management of manure as a fertilizer very difficult.  Manure is  often applied to lands that are vulnerable to surface run-off and flooding.  Manure from poultry farms is often moved to fields in  the fall and winter months instead of being stored on the farm, resulting in piles of manure exposed to environmental conditions such as wind and rain.  For example, rain saturates them and the  contaminated water leaches into the soil and neighbouring surface water.  The immediate incorporation of surface applied manure is rare and manure injection is not used.  As a result, more nitrogen  fertilizer has to be purchased due to significant losses of ammonia nitrogen.  Important management decisions are given very little attention in the region.  Better management is required to determine the  nutrient content of the manure, ensure homogeneous mixing prior 7  to spreading, and spread the manure uniformly and applying it to match the needs and growth requirements of the crop.  Odour is a common complaint against animal farms.  In the  Abbotsford region, odour is mainly caused by manure spreading but adequate measures are not often taken to reduce this odour.  The intensification of livestock operations has resulted in excess availability of manure in many parts of the world. Legislation restricting animal density according to the land's ability to absorb animal wastes without damaging water resources is now being implemented in countries such as Switzerland, Holland and Denmark.  The nutrient absorption capacity varies  from soil to soil, but 2 animals/ha is assumed to be a reasonable standard for sustainability of water quality (Hall, Kenneth and Schreier 1991a).  This value is exceeded in the Lower Fraser  Valley, where chicken density is 255/ha and pig density is 2.1 animals/ha.  The 625 kg/ha fertilizer application rate adds to  the nutrient disposal problem in the area.  The water pollution problems in the Lower Fraser Basin are the result of other sources as well.  For instance, the area around  the Abbotsford airport represents one of the best aquifers in the Basin.  It supplies drinking and irrigation water to the rural  population and freshwater to a major fish hatchery near Huntington.  The majority of the chicken producers and the most  intensive raspberry operations are located here. "The use of chicken manure and chemical fertilizers and 8  industrial run-off associated with airport activities has contributed to large nitrate loadings, and in 1982-83, 23 of the 24 local wells sampled exceeded the 10 mg/L standard" (Hall, Kenneth and Schreier 1991a, p.82)1 Recent studies show that the problem is getting worse, and there is an additional problem because groundwater flows into the U.S.A. and adjacent municipalities. "Urgent regulation is needed to reduce the nitrate input, to alter cultivation and management by planting grass or cover crops between rows, by reducing chemical fertilizer and manure application, and by restricting the timing of such applications" (Hall, Kenneth and Schreier 1991a, p.81).  Government Intervention The Fraser Valley has been under close examination due to intensive swine, dairy and poultry farms.  Pork producers have  recognized that management of animal wastes requires more attention than the past, since the general public has become more concerned about environmental issues. The Fraser Valley Hog Producers Sustainable Farming Group (FVHPSFG) has been formed as a sub-committee of British Columbia's Hog Marketing Commission (BCHMC) in response to these concerns and to take advantage of funding from the Canada - B.C. Soil and Water Conservation Agreement.  The BCHMC has implemented a $360,000 program, which  runs until March 31, 1994. The objective of this program is to encourage the development, evaluation, transfer, and adoption of soil and manure management practices and technologies that sustain soil and water productivity over the long term (ARDSA 1993). 1  Conservation groups which are not commodity specific have  These authors cite an earlier (1985) study by C. Ott. 9  also been formed on Vancouver Island and in the Okanagan to look at specific problems in those areas.  A new set of environmental guidelines was formed for British Columbia in April 1992. The British Columbia Waste Management Act now includes The Code of Agricultural Practice for Waste Management.  Prior to the inclusion of this Agricultural Code of  Practice under The Waste Management Act, only general conditions existed for acceptable disposal of agricultural wastes. Furthermore, an exemption existed for most farmers from the Ministry of Environment (Castle 1993).  The vagueness of wording  and lack of definition made it difficult for Ministry of Environment officials to deal with complaints and violations of the Waste Management Act.  The new Code establishes recommended practices concerning the storage and spreading of animal wastes, exhaust from building ventilation systems, and the proximity of farm operations to waterways.  Although this new code for waste management will have  some impact on improving the environmental performance of B.C. farmers at very moderate costs to B.C. taxpayers, it will not be sufficient to deal with the serious water quality problems in the Fraser Valley.  The nitrate problem in this area cannot be  addressed through the complaint-response process.  This is due to  the fact that animal wastes leach into waterways via a large number of sources and it would be difficult to effectively deal with the problem by the filing of individual complaints (Castle 1993). 10  1.2 PROBLEM STATEMENT The extent and severity of groundwater degradation was not always well documented, groundwater quality management has become an important public policy issue (Abdalla, Roach and Epp 1992). Farming areas tend to have particular problems with certain contaminants, such as nitrates.  Despite the level of concern  over possible human health effects and legislation relating to groundwater quality, little is known about the economic benefits of groundwater  protection.  Groundwater policies yield a number  of possible benefits, measured in terms of the damages that are avoided.  These include damage to human health (such as increased  mortality or morbidity from exposure to contaminants), ecological damage, and loss of intrinsic values associated with groundwater resources (Abdalla et al. 1992).  Abdalla et al. also indicate that in previous studies of water quality little attention has focused on averting expenditures, or the costs incurred by households, firms or governments to avoid exposure to groundwater contaminants.  In the farm sector, the  economic evaluation, including social benefits, of composting as a means of reducing water pollution has been given very little attention.  How does one measure these effects and reduce the  pollution problem?  Composting is one way to reduce the nitrate  problem, but is it economically feasible?  1.3 OBJECTIVES The economic feasibility of composting needs to be determined in order to assess it as an option to the current manure management 11  practices.  Private costs of composting are readily available.  The objective of this study is to estimate the amount individuals are willing to pay to obtain better water quality in the Abbotsford farming area.  Contingent valuation will be used to  derive individuals' WTP.  In particular, individuals will be  asked how much they are willing to pay to obtain a lower amount of nitrate in their water (estimates of nitrate clean-up costs will be presented).  Since composting animal wastes has been  identified as a way to reduce nitrate in Abbotsford water, this will be used to determine the social benefits of composting. Defense or averting expenditures will be determined and compared to the results obtained from the WTP model.  WTP often results  in overestimates (Cameron 1992), but correction mechanisms are possible.  Pollution defense expenditures which are expenses  incurred by individuals to avoid the pollution problem will be estimated and compared to the values obtained from WTP model. They may be used as a correction factor in the contingent valuation method (CVM), or may be included directly in CVM model (Cameron 1992).  The economic feasibility of composting will be determined from both a private and social point of view.  The few studies that  have focused on this issue look at only the market costs. Opportunity costs from alternatives to composting and government programs encouraging the use of fertilizers (instead of compost) are two costs that will be examined.  The costs and benefits of  composting to farmers will then be estimated.  12  1.4 OVERVIEW OF THESIS There is an externality problem —  farmers don't have an  incentive to change their manure handling practices.  The  government wants to provide the incentive through regulations and subsidies.  One technique has been to encourage farmers to  compost, but subsidies appear to be needed.  The question that  needs to be answered is: Are subsidies worthwhile from a social point of view; i.e., is the benefit cost ratio greater than 1.0.  Chapter 2 presents an overview of composting, as a way to reduce nitrate leaching from animal wastes.  This includes the use of  alternative practices, composting methods and costs. After determining the financial feasibility of composting from a private point of view, a framework will be designed to measure social costs of improving water quality for Abbotsford residents receiving water from the aquifer.  Appropriate social welfare  measures are considered in Chapter 3.  These include hedonic  pricing and the contingent valuation method.  A random utility  model is outlined as the basic underlying theory used in estimating social benefits.  Then the methodology that will be  used to generate results will be outlined is Chapter 4, namely the WTP model and the averting expenditures model.  Contingent  valuation surveys are used to determine willingness to pay for nitrate reduction. used.  Both closed-ended and open-ended models are  Averting expenditures are compared to WTP results. This  analysis, which is presented in Chapter 5, is followed by a conclusion in Chapter 6.  13  CHAPTER 2 2.0 ALTERNATIVES AND COMPOSTING Alternatives to current livestock handling may be useful in reducing the pollution problem or externality imposed on the nonfarm sector.  There are not many economically feasible  alternatives to current practices of handling manure, from a private point of view at least; otherwise, there would be more farmers employing them.  Injections into the surface, better  timing of fertilizer and manure applications, better storage and disposal of animal wastes, and composting are possible remedies for the nitrate externality problem.  Storing poultry manure and transporting it off the aquifer are alternative ways to handling livestock wastes. As Hauser, van Kooten and Cain (1993) point out, total costs range from about $50/tonne to $70/tonne, including storage costs, investment costs, cleaning and handling costs, and transportation.  If  livestock waste can be sold for more than this (some manure is sold for about $90/tonne) (Hauser et al. 1993), then no public intervention is needed.  But these calculations do not take into  account marketing costs, sales of manure to local vegetable and berry producers, and wastes from animals other than chickens. Furthermore, storage and transportation by itself will not solve the nitrogen-nitrate pollution (Hauser et al. 1993).  Injections of manure into the surface is not commonly used.  This  may be due to the lack of available equipment and associated costs.  This method is effective in reducing the odour problem, 14  but may not solve the nitrate problem.  The application is  directly into the ground, which reduces surface leaching. However, in the groundwater there is likely to be nitrate build up contributing to water pollution.  It is difficult to regulate  the timing of fertilizer and manure applications, and farmers would need incentives to do this.  Composting animal wastes is an  alternative that is seriously being considered.  A limited amount of research has focused on the composting of livestock wastes.  Compared to many European countries where  compost has become a significant production input and utilized in residential and land improvement strategies, in B.C. it is still an insignificant production input.  The reliance on fertilizers for plant nutrients has caused the decline in use and value of livestock manures in Canada (Stonehouse 1984).  Non-farm sources of plant nutrients were  more economically attractive until the 1980s because they were relatively inexpensive and abundant. Application of chemical fertilizers was made more convenient and effective by technological innovations regarding their usage on the farm. Along with this, livestock manures have been viewed as an imprecise means of supplying nutrients to plants, and land applications are considered potentially damaging to the environment.  Increased specialization in agricultural  commodities has left Canadian farms without their own source of manure.  This fact, combined with the bulkiness and expensive  transportation of manure, has favoured the use of fertilizers 15  instead of manures (Stonehouse 1984).  As a result, not much  attention was given to composting until recently.  The animal  waste disposal problem and resulting environmental contamination and the rising prices of chemical fertilizers, have sparked more interest in composting.  Composting manure is one way to reduce environmental problems such as leaching and odour.  The B.C. Hog Marketing Commission  supports the possibility of the production of an organic soil conditioner product from composted hog manure.  Composting  represents an environmentally safe alternative to field application of raw manure, which adds harmful leachate into soil and ground water sources.  Furthermore, the transporting and  disposing costs of raw liquid manure  can be significant. An  Agri-food Regional Development Subsidiary Agreement study (ARDSA 1989) found significant potential for production of compost from the Fraser Valley.  Farms in the Fraser Valley contain  approximately 70% of the sows on B.C. farms. According to a representative of the BCMAF (van Kleeck 1991), each of these sows and her young produce 2.2 cubic feet of manure per day on average, 20% of this manure contains solids suitable for composting.  Approximately 63 million litres of manure could be  composted each year if all the farms in the Fraser Valley cooperated in a composting program.  Revenues generated from  selling the finished product could increase the viability of hog farms if composting costs are sufficiently low in comparison.  In general, manure disposal problems not only concern the farmers 16  and agrologists but also the public at large, which is exposed to the external effects of manure, especially people who live near livestock enterprises or use water that could potentially be contaminated.  The externality has never been measured and the  extent of the problem is not known.  Neighbours have a right to a  pollution free environment, and farmers do not take into account the indirect effects of the waste.  These costs are obviously  difficult to measure.  In its 1989 study on the marketing of hog manure, ARDSA (1989) came up with the following conclusions:  1) Significant market potential exists for a composted manure product as an organic soil conditioner in the bulk and retail markets.  2) The largest market for soil conditioners purchased on an annual basis is the retail nursery segment, while the largest bulk  market is the landscape contracting segment.  3) In terms of accessibility, the market with the greatest potential is the B.C. market, especially the Lower Mainland area. The viable market in the prairie provinces, where transportation costs have a significant impact, is the retail market in Alberta. Significant markets also exist in Washington and Oregon.  4) Projected sales were 27.8 million litres by the fifth year for a hog manure composting plant in the Fraser Valley. 17  Some 3.4  million litres of composted hog manure in the first and 11.7 in second year were projected.  5) To penetrate the intended market, the product would have to be priced competitively with comparable products presently used in each market.  If prices equivalent to 6.3 cents per litre and 5.3  cents per litre are charged in the retail and bulk markets, respectively, then the first year revenue is projected to be about $182,000. The projection for the fifth year is $1.5 million in revenues.  6) Marketing costs would be significant.  Transportation and  handling would comprise the largest costs at about 35% of the sales revenue generated per year.  Bagging costs which would  total 30% of sales revenue would also be significant.  Other  marketing costs would include sales force costs and advertising and promotion.  ARDSA (1989) indicated the need for further research in order to promote the development of the industry, especially to determine the economic feasibility of producing composted hog manure in the Fraser Valley.  The ARDSA-sponsored study assumed a very high  value of compost.  Farmers may not be able to sell compost at the  prices indicated because the market may be flooded with compost in the future as more manure in B.C. and other jurasidictions is composted, and as other organic material is composted by municipalities.  The current market price is lower now due to 18  municipal composting programs.  The supply of compost has gone  up, reducing the prevailing price and making the compost investment more economically infeasible for farmers.  A study focusing on swine manure handling and treatment systems developed a basic compost scenario (Fullerton 1990).  Three  different manure treatment systems were analyzed: Standard Treatment System, Partial System, and Full Treatment System.  The standard treatment system does not allow for the production of solids and compost.  Since there are no liquid treatment  facilities, there is no liquid/solid separation for this type of system.  The partial treatment system does contain a liquid/solid  separator; therefore, manure solids and compost are options. Both these systems have the same liquid storage and disposal options.  The full treatment system can produce two commercial products as well: manure solids and compost.  It incorporates a liquid/solid  separator, a settling chamber and a sequencing batch reactor. Liquid storage facilities are not required for this system since liquids are disposed of daily by an electric pump.  Using the compost option for the partial and full treatment systems, income effects of composting solids on 100-400 sow farms were determined by Fullerton (1990).  Compost is assumed to be sold at the nutrient value of 19  $26.00/tonne.2 At this price level, net farm income was negative for all farm sizes, although the 400 sow unit with full treatment only had a small deficit of $3,100/year.  Net farm income  increased significantly, when compost is sold at $39/tonne, 50% above its nutrient value.  Full treatment is at breakeven3 at  this price, except at 100 sow units for which there is a $12,250 loss.  For the partial treatment only the 400 sow unit reaches  breakeven.  At $52/tonne net farm income increases further, resulting in net revenues of $18,000 for full treatment at 200 sow units and $18,000 for partial treatment at 300 sow units.  Based on the  assumptions used, the results indicated the following: "Composting can be a viable alternative for partial and full treatment systems. The feasibility of composting is largely determined by product prices, the size of the farm, production costs and markets. For all farm sizes, there is a direct comparison between compost prices and net farm income. Full treatments reach a breakeven point at 200 sow size, if compost prices are $39.00/tonne (50% increase over nutrient values). At this price, partial systems reach the breakeven point at the 300 sow size." (Fullerton 1990 p.35).  A recent study in Kansas (Berends et al. 1993) assumes compost is sold at $US 6.00 per ton.  This price is based on the actual  price charged to farmers in the area by a commercial compost dealer.  This price is extremely low compared to the prices  2  The current prevailing price was quotes at $22/tonne by the only existing commercial composting operation in B.C., the Answer Garden Farms. 3  Breakeven indicates that net farm income is greater than zero, in other words costs equal at least the gross income. 20  assumed by ARDSA (1989) and Fullerton (1991).  Thus, there likely  to be a downward trend in the B.C. price of compost.  Field  observations indicate that composting is not economically feasible for most farm operations, but it remains to be seen whether it is socially desirable for governments to make up the losses accruing to farmers by composting.  2.1 COMPOSTING METHODS There are several different kinds of composting techniques available for farmers, ranging from simple (Passive) Windrow Composting, which is the method most commonly used by farmers and municipalities to compost solid wastes, to the more capital intensive techniques: Aerated Static Pile Composting and Active/Aerated Windrow Composting (Fullerton 1991).  Active Windrow Composting results in the production of compost in windrow, with mechanical aeration.  A windrow turner, which can  be powered by a farm tractor or self-powered, aerates the compost mix.  Aerated Windrow Composting results in the production of  compost, with both mechanical aeration (windrow turners) and forced air circulation.  Aeration is enhanced by pressurized air  flow when the windrow is placed on top of air ducts. The active composting period will be greater for passive windrow systems, given that operating conditions for both methods are equal.  Some  farmers using the popular Active Windrow System increase costs by using inappropriate farm tractors for windrow turning (Fullerton 1991).  As a result, the easy or inexpensive system is made more  difficult and expensive. 21  Aerated Static Pile Composting yields compost in piles or windrows, with limited or no mechanical aeration.  Aeration  occurs by blowing or drawing air through the composting material, after the windrow or pile is placed on air ducts.  The aeration  system itself can range from a relatively simple technique using electrical motors, fans and ducting, to more sophisticated methods that incorporate various sensors and alarms. Near the end of the active compost period, mechanical aeration may take place in some systems.  Compost in piles or windrows also results from Passive Windrow Composting.  Compost is produced over a long period of time with  minimal or no aeration.  This is a low technology, low labour  approach to composting.  Great improvements in speed and compost  quality can be made by monitoring characteristics such as porosity of the initial mix and particle size.  Similar to  aerated static pile composting, near the end of the active compost period, mechanical aeration may take place in some systems.  2.2 COMPOSTING COSTS One of the earliest studies on composted hog manure was conducted by ARDSA in 1989. The main objective of the study was to assess the market potential for hog compost.  A significant market  potential was found for hog compost in the bulk and retail markets.  The B.C. market was noted as having the greatest potential, while 22  Alberta was found to be a viable market as well.  The largest  cost item was transportation and handling, which was expected to total about 35% of sales revenue generated per annum.  On the  basis of market assessment, there was considerable potential for compost; however, the study did not include technological configurations and financial feasibility of composting.  In order  to determine the economic feasibility one needs to examine the technological and financial aspects.  From the results obtained by ARDSA's marketing study in 1989 (outlined in Section 2.0), the first year of revenues is $182,000.  If price is 6 cents per litre of compost, then we have  approximately 3.1 million litres of compost being produced and sold.  Costs Transportation/handling (§ 30%)  $54,600  Bagging (§ 35%)  63,700  Advertising/marketing (assume @ 20%)  36,400  Production (remainder of costs)  27r300  Total costs  $182,000  Assume normal profits at each stage. When price falls, none of these costs will fall.  Suppose price falls to 5 cents per litre.  Then revenue equals $151,667, which represents an approximate loss of $30,000. As a result, production costs cannot be covered.  23  The economic feasibility was evaluated for the composting systems mentioned earlier for standard operating conditions in B.C. Fullerton (1991) employed an interactive computer model with four different stages, more than 100 operational and financial variables, and six different bulking agents to analyze four different composting systems with regards to inputs, operational efficiency and financial results.  It was found that active  windrow composting, using windrow turners that are either tractor-powered or tractor pulled or self powered, is the most common method utilized for on-farm composting mainly because a significant amount of new machinery does not have to be installed for the procedure since existing farm machinery can be used. This leads to lower annual costs for compost production, which is an important factor to consider since it is the annual costs of production and resulting net benefits that determine the economic feasibility of a composting system.  The investment decision for composting becomes increasingly complex as farm size and capital costs increase, or environmental and/or odour regulations become more restrictive.  A system with  built-in flexibility (e.g., able to efficiently utilize a wide array of bulking agents) has a much better chance of leading to financial success. One obvious benefit would be the ability to alter inputs to arrive at the least cost bulking agent at the time the composting is to be done.  Operational flexibility  allows the investor to adjust specifics in the composting procedure to arrive at the least cost combination.  24  It was found that the bulking agent comprises a large portion of costs.  They can total more than 50% of the cash costs or 40% of  the total operating expenses for an Active Windrow System. Therefore, annual operating and cash expenses are sensitive to bulk prices.  Bulk prices are an important consideration in  determining the profitability of the product.  Competitive  production costs are essential since the compost market itself is competitive (Fullerton 1991).  Other conclusions of the study included the fact that compost quality will vary significantly, but the quality differences may or may not be an important consideration depending on how the compost is to be used.  It was also found that the length of the  active compost period can have a significant impact on both annual expenses and compost production costs especially if building costs are included.  (For a given initial compost mix, a  longer compost period requires larger buildings).  Production  costs for a long composting period (8-12 weeks) could be 30-50% higher than those for a short composting period (4-5 weeks). Since the "curing phase" of composting is not included in these calculations the composting period is longer and the economic costs are be higher when the entire composting process is considered.  Recycling compost as a bulking agent may or may not be economical.  If the compost operation is commercially oriented  (e.g. The Answer Garden Farms), the "opportunity cost" of diverting compost from sales could be significant, particularly 25  if expensive bagged compost is being sold.  The Answer Garden  Farm used sawdust as a bulking agent and, after each compost cycle, the sawdust is separated from the compost and reused.4  A study regarding the economics of off-farm litter disposal options for south coastal B.C. poultry farms was conducted by Stennes (1992).  Poultry manure (litter), except for breeder  manure, contains natural bulk and higher nutrient levels compared to other sources.  The bulky texture of litter creates complications because storage and transportation costs for a bulky product can result in litter with lower economic value.  The costs and returns were derived  for poultry litter shipped from Abbotsford to Delta assuming maximum value for litter. litter.  Costs were greater than the value of  The study found the main benefit of handling litter to  be a social benefit rather than a private one.  This explains why  there are not many farmers composting at present, if any at all. Poultry farmers incur a cost that is partly compensated by the Agricultural Land Development Act (ALDA), which provides loans for storage facilities at rates lower than market rates. Subsidized interest rates were not accounted for in the study.  Stennes (1992) points out the difficulty in valuing litter. end use of the litter determines the value.  The  Factors such as the  NPK ratio, pH, bacterial counts, and organic materials determine  4  Personal communication with Manager, Rick Chase. June 1992. 26  quality, which in turn determines value.  Although the use of compost is technically feasible, compost itself does not provide a large percentage of the nitrogen source in the profit-maximizing combination with commercial fertilizer. Berends et al. (1993) evaluated the economic feasibility of fertilizing irrigated grain sorghum with composted manure in Kansas, using net return budgeting and production function analysis.  This study did not take other nutrients such as  phosphorus, potassium, and micro-nutrients into consideration. Organic matter was considered to be the principal benefit of compost.  The compost costs of $US 6.00 per ton used by Berends  et al. (1993) are based on the actual price charged to farmers in the area by a commercial compost dealer.  This compost price  results in a value of $0.50 per pound nitrogen in compost, whereas fertilizer costs for commercial nitrogen are $0.23 per pound for solid ammonium nitrate and $0.15 per pound for anhydrous ammonia. For purchasing and custom applying compost, $US 8.856 per ton is the lowest possible cost; $12.25 is the highest (Berends et al. 1993).  The results of the study indicate that compost does not appear to be economically feasible, when only the nitrogen value is considered.  2.3 CONCLUSION ON COMPOSTING Most studies indicate a net loss for composting, perhaps the social benefits of composting, via the reduction of nitrate in 27  drinking water, outweigh the losses resulting from it.  Social  benefits are difficult to measure for a non-market good such as improvement in water quality.  Also, the correlation between the  amount of livestock waste composted and resulting reduction in pollution level is not easy to measure.  To evaluate the economic  feasibility of composting or reduction in pollution level, or externality imposed on the non-farm sector, non-market measurement techniques need to be employed.  The next section  examines ways of measuring this externality on the non-farm sector.  28  CHAPTER 3 3.0 THEORETICAL CONSIDERATIONS This chapter looks into methods used to measure non-market goods, in particular externalities similar to water contamination resulting from livestock wastes. An appropriate method is chosen and a basic underlying theory to be used in this research is outlined.  Two applicable methods to this type of research are the hedonic pricing method and the contingent valuation technique.  There  have been few studies done using hedonic pricing and contingent valuation to measure such an externality on the non-farm sector. The choice of method depends on availability of data, along with type of analysis to be done, and other particulars of the study.  3.1 HEDONIC PRICING The Hedonic Pricing (HP) method measures environmental commodities such as air quality, scenic views and water frontage by assuming that households pay a premium to acquire homes that afford access to these amenities.  The demand for such amenities  is inferred from housing prices and the value of the non-market environmental good is obtained as the area under the demand curve.  The HP model may incur identification problems, if not  properly specified.  In judging HP's applicability to this  specific problem, two relationships must be considered.  The  first is whether the link between the environmental good (or bad) and the corresponding environmental attribute firmly established in the minds of the property owners. 29  Second, is the  environmental good likely to be valued by buyers and sellers of housing?  For an environmental good such as air quality, researchers have used pollutant concentration levels in hedonic pricing studies. Where households have good information about air quality, such variables have provided good results when used in HP (Brookshire et al. 1982).  Qualitative differences in air quality are  readily visible to most people and may affect a person's breathing, as well.  For water quality of a lake or river, the  link to an attribute can be established, but some kinds of water quality differentials may not be obvious to households (Willis and Foster 1982).  Part of the problem with water quality is  seasonal fluctuation.  When convenient, most sellers will list  houses when they are most attractive.  HP involves elaborate sets of assumptions and complicated econometrics to arrive at values. But such data is not available for this study. Why not just ask people what values they place on environmental assets?  This is exactly what the contingent  valuation method does.  3.2 CONTINGENT VALUATION METHOD (CVM) The first person to actually implement the contingent valuation method was Davis (1963, 1964).  Davis devised a questionnaire  format, known as the bidding game, that has since been employed in many studies.  Big game hunters in Maine were asked whether  they would continue to hunt if their expenses rose by a specific 30  amount.  If the answer was yes, the amount was increased until a  maximum willingness to pay amount was reached.  Likewise, if the  original level of expenses received a negative response, the amount was lowered until the respondent indicated the maximum level of expenses that was tolerable before participation would cease.  The resulting estimate of maximum willingness to pay of  hunters was then used for comparison with timber values.  Since the initial studies using CVM, a great deal of refinement has taken place.  Basically there are six issues that must be  dealt with in any good CVM study:  (1) What will be the  population base whose values are to estimated? product to be valued be defined? be used?  (3) What payment vehicle will  (4) How will CVM questions be posed?  be analyzed?  (2) How will the  (5) How will data  (6) What supplemental data will be obtained?  Although CVM is a novel (and somewhat controversial) enterprise for economics, it is part of a long tradition of valuation methodologies, spread over many disciplines (Fischhoff and Furby 1988).  CVM is flexible.  Assuming that people's answers are  accurate measures of WTP, a wide variety of resource and environmental issues are amenable to CVM.  Nevertheless, many  within the economics profession have been reluctant to accept CVM (see Kahneman 1984 for more details).  CVM estimates reflect  hypothetical transactions and many economists question whether people are both willing and able to respond in ways that reflect how they would really behave in a well-functioning market. Regarding the willingness to respond accurately, fears have 31  focused on possible "strategic bias" in CVM results(i.e., people wanting to further their own ends).  Regarding the ability of  people to respond, concerns might be lumped under the heading of "hypothetical bias".  Despite widespread a priori fears of biases, empirical research has been at least moderately encouraging about the accuracy of CVM estimates.  Several studies from the public goods' literature  have involved attempts to set up situations where theory predicts that people will behave strategically (e.g. Scherr and Babb 1975; Smith 1977; Marwell and Ames 1979, 1980).  Strategic behaviour  appears to be the exception rather than the rule.  Furthermore, CVM results have been roughly comparable in magnitude to results from studies employing other valuation techniques including Travel Cost models (Smith 1977; Marvel and Ames 1980).  But the Travel Cost Method is not very useful for  calculating the benefits from pollution control (Mitchell and Carson 1989).  A number of laboratory and field experiments have now been completed where CVM results could be compared with value estimates from actual cash transactions.  Statistically  significant differences sometimes emerge between contingent values and actual cash values. However, the contingent values and actual cash values are usually in the same range, at least for willingness to pay.  32  Without doubt, the weight of all this evidence has increased confidence in CVM.  Still doubts remain.  For example, Freeman  (1984) and Kahneman (1984) have voiced concerns about whether results relating to the validity of use of values can be generalized to existence values.  Further research, especially  involving laboratory and field experiments, will help resolve these issues. "In any field, the main intrinsic reason for undertaking a comprehensive perspective is its intellectual challenge. The main intrinsic reason is convincing outsiders of the validity of one's results. CVM studies often have the specific goal of setting policy—by pricing unmarketed goods so that they will be considered in cost-benefit analyses. In such circumstances, one cannot be satisfied with intradisciplinary standards, but must be ready to handle criticism from all comers" (Torgerson 1993, p. 151). In the meantime, it can only be concluded that CVM is becoming more and more firmly established as a major tool for evaluation of non-market goods.  Therefore, this is the method that will be  used for this study.  Survey research is popular.  It permits us to measure the  prevalence of attitudes, beliefs, and behaviours, to study change in them over time, to examine subgroup differences, and to test casual propositions about the sources of attitudes, beliefs and behaviour (Weisberg et al. 1989).  Weisberg et al. (1989) also  state that surveys are a useful tool for social scientific investigations because they have important advantages over other research methods.  For example, experiments involve highly  artificial conditions. Therefore, communications are likely to have different effects in a confined experimental setting than 33  they would in a natural setting.  To monitor attitude and  behaviourial changes over long periods of time, surveys would be a better option than experiments.  There is also a difference in opinion among survey researchers regarding the usefulness of values obtained from CVM to assess environmental damages and, therefore, cost-benefit analysis (Smith 1992).  Randall (1993) and others maintain that CVM is  useful and can be safely used in cost-benefit analysis.  Cameron  (1992) statistically adjusted CVM values by combining them with the travel cost method (TCM).  This may also be done with other  choice based methods (Hauser et al. 1993).  But others concede  that values obtained from this methodology have no meaning and, hence, cannot be employed in cost-benefit analysis (Kahneman and Knetch 1992).  The debate continues and is not explored in this  study.  Data on willingness to pay for improvements in water quality are determined from contingent valuation questionnaires (Musser et al. 1992).  Respondents are basically asked to reveal the maximum  amount they are willing to pay for a hypothetical improvement in water quality.  To minimize bias, it is important that surveys be  appropriately designed and implemented. explored later in the study.  These issues will be  Minimizing bias is particularly  important when questions are of a hypothetical nature.  34  3.3 RANDOM UTILITY MAXIMIZATION The grounding of dichotomous choice welfare measures in utility choice has been demonstrated in Hanemann (1984).  Consumers are  assumed to maximize their utility, which is a function of the amounts of goods that they consume and the quantity of the public good that is available, which, in this case, is the quality of water, denoted Q.  Assuming that consumers' behaviour is  rational, they prefer higher water quality for drinking and other consumptive purposes. As Hauser et al. (1993) demonstrated, the household's economic problem is represented by the following utility model: MaxJJix.Q.z) s.t. m = px,  (1)  where x is the amount of market goods households consume, m is household income, p is a vector of prices, and z is a vector of social and other factors affecting utility (these are dropped from the remaining analysis for convenience).  As indicated in  Figure 1, for a given water quality Q°, the household achieves utility level U0.  35  All Other Goods (X)  -V—  -  EV  WTA  m = px x.  j  WPA --—^__ i  u  j  !  o  Increased Water Quality (Q)  i  0  Q1  Q°  Figure 1. Welfare Measures Related to Changes in Water Quality (From Hauser et al. 1993).  The indirect utility function is:  v(p,Q.m)  - Maxx u{x,Q) \ px-m = 0 = u(x(p,Q,m) ,Q) .  (2)  The associated expenditure function is: e{p,Q, u) = minxpx | u(x,Q) = px(p,Q,ut) = v~x{p, Q,m) .  36  z u°  (3)  The indirect utility function and the expenditure function are assumed continuous and twice differentiable in p, Q and m.  This  function is non-decreasing and quasi-concave in Q; the expenditure function is non-decreasing and convex in Q.  Since  prices remain fixed throughout the analysis, it is dropped as a variable in the remaining analysis.  To evaluate welfare changes from increments or decrements in the availability of a public good, in this case water quality, Hicksian compensated measures are used (Boadway and Bruce 1984). From these welfare measures one can derive the Hicksian compensating variation (CV) (figure 1), which is maximum amount that the household is willing to pay (WTP) for an improvement in water quality from Q° to Q1. right to water quality Q°.  CV assumes that individuals have the The Hicksian equivalent variation  (EV) (figure 1) assumes that they have the property right to Q1. EV is the minimum amount that the household would be willing to accept (WTA) as compensation to forgo the improvement in the water quality.  Mitchell and Carson (1989) discuss the benefits of using WTP over WTA compensation.  "WTA values generally decline over time and,  despite being valid welfare measures, people appear to reject the implicit property right in WTA studies" (Hauser et al. 1993, p.29).  This implies unreliable WTA values.  In practice, WTA and  WTP may vary significantly, despite economic theory suggesting that there should little difference.  37  Knetch (1989) has suggested  that this difference may be due to differences in the allocation of property rights.  The focus is on measurement of compensating variation, since a contingent valuation instrument is used to elicit the respondents' WTPs.  The willingness to pay of the k* household  for the improvement in water quality can be written as: WTPk = a0 + axAQ + a2AQ2 + a2(mk-m) + ai(mk-m)2 + a5AQ(mk-m) + ek,  *'  where: a0 = CV(Q°,Q°,m) = 0, since the CV of no water quality change must be zero; at = the derivative of CV with respect to Q; a2 = the deriv. of CV w.r.t. m; a3 = the deriv. of ax w.r.t. Q; a4 = the deriv. of a2 w.r.t. m; a5 = the deriv. of a3 w.r.t. Q; e = the remaining terms. The addition of social factors describing age, household size, education, and so on, complete the empirical model.  A random sample of respondents are asked if they would be willing to pay some given dollar amount for a hypothetical change in the availability of a particular environmental amenity or attribute. The dollar amounts, called the "bid" amounts (which vary from one person to the next), are recorded along with the yes or no amounts for each respondent.  "Each observation in the sample  reveals whether or not an individual's maximum WTP is greater or less than the fixed dollar bid amount offered" (Kanninen 1993b, p.138).  Usually the sample is segmented and each set is assigned  38  a different bid amount.  This is what makes for a random utility  model, and the question format is known as the dichotomous choice or closed-ended format.  The grounding of dichotomous choice  welfare measures in utility choice has been demonstrated in Hanemann (1984).  There is also a open-ended question format in  which respondents are asked to give an WTP amount on their own. There is an ongoing debate as to which of these question formats are better.  In chapter 5 these two methods are evaluated in the  context of this study.  For practical purposes, an individual's true WTP is unknown to the researcher and can be treated as a random variable.  The mean  of any random variable can be expressed as:  E(WTP) = f[l-F(b)]db  - JF{b)db  0  (5)  —  where the cumulative probability density function (CDF) F(b) is the probability that WTP is less than or equal to b (Cooper and Loomis 1992).  If WTP cannot take on negative values, then E(WTP)  reduces to:  E(WTP) = f[l-F(b)]db  (6>  0  Hanemann's (1984) utility difference model provides one theoretical foundation for deriving the parameters necessary for estimating E(WTP).  The assumption that individuals will always 39  choose the option yielding the highest level of utility is held by a random utility maximization model (Kealy et al. 1988).  The double-bounded bidding procedure is gaining popularity for use in CVM (Kanninen 1993b).  This procedure involves asking  respondents whether they are willing to pay some initial dollar bid amount, and then presenting follow up bids, higher if the response to the initial bid is yes, and lower if it is no.  Hanemann (1984) showed that estimation using a double-bounded procedure is statistically more efficient than results from the single-bounded method.  In addition,  "...many CVM practioners agree that the double-bounded procedure retains a market-type flavour, which improves the reliability of responses. Several CVM studies use the double-bounded procedure, and its use is expected to increase in the future" (Kanninen 1993b, p.138). The optimal experimental design for the double-bounded procedure needs to be derived, and costs of this procedure are not insignificant.  It is difficult to formulate and analyze.  40  4.0 METHODOLOGY The CVM uses surveys to illicit WTP values for a hypothetical change in the availability of a particular environmental amenity such as water quality (Kanninen 1993a).  The WTP responses are  used to estimate welfare benefit measures deemed to be appropriate for use in an economic analysis of that non-market good.  A dichotomous choice is used to determine individual WTP  values.  A random sample of individuals are asked if they would  be willing to pay some given dollar amount for a hypothetical change in the availability of a particular environmental amenity or attribute.  4.1 CONTINGENT VALUATION Any transaction involves the exchange of goods.  For a consumer  good, that change in state commonly involves having more of a particular good and less money.  In the case of a public good  such as water quality, the change is across worlds having different levels of each affected attribute, and of resources required to make that level possible.  The term reference level  will be used to define the current state.  The state obtained if  the transaction is enacted will be known as the target level (Togerson 1988).  Ideally, reference and target levels are specified for every potential attribute affected by a transaction.  In some cases,  specifying even one of these levels for a single attribute may be quite complex.  This is especially due to the variation from area  to area and inaccuracies and difficulties in measurement. 41  A statistical summary  for both target and reference levels is  needed in order to calculate a good's accurate value (Fischhoff and Furby 1988).  One has to keep these values arithmetically  simple enough so that these values are understood by respondents.  Fischhoff and Furby (1988) also indicate that the lower bound for a good's value or it is minimum worth is determined when a respondent agrees to pay the fixed price tag on a proposed transaction.  Only an upper bound is determined if the respondent  refuses the transaction.  A more exact value can be determined  where iterations are possible.  This is done by presenting the  good at iteratively higher or lower prices until the upper and lower bounds converge (Fischhoff and Furby 1988).  The  opportunity to review values may also be provided by this iterative procedure.  But this is done at the risk of being  somewhat tedious and influencing valuations by the choice of starting points and incremental changes (Boyle et al. 1985).  In the open-ended questioning format, which is less time consuming, respondents are asked directly for the maximum they are willing to pay for the good, leaving the price entirely up to the respondent.  This procedure is more demanding, forcing  respondents to imagine various prices and to figure out which is just below the lowest at which they would refuse the transaction (Fischhoff and Furby, 1988).  42  4.2 SURVEY DESIGN A questionnaire was constructed by relying on previous studies of CVM and averting costs on similar public decisions with similar populations (Hauser et al. 1993; Musser et al. 1992; Abdalla et al. 1992).  Survey Administration Technique Survey instruments may be read to the respondent in person or over the telephone, or they may be sent in the mail with a request to fill out and return the questionnaire.  Bradburn  (1983) states that the current agreement among survey researchers as follows:  "Contrary to the common belief favouring face-to-  face interviews, there is no clearly superior method that yields better results for all types of questions" (p.294).  Survey  researchers are recognizing the high costs of in-person surveys and methodological developments in telephone survey technology (Mitchell and Carson 1989).  Major academic survey research  centres successfully experimented with telephone interviews, a methodology which commercial polling houses have used for years (Groves and Kahn 1979).  Another administration method is the  mail survey, which, unlike telephone interviews, permits the use of visual aids. the technique.  Here, too, methodological advances have improved Lower response rate are more prevalent in mail  surveys.  The larger potential cost savings in using telephone and mail surveys has not gone unnoticed by contingent valuation researchers.  Both are commonly used (Mitchell and Carson 1989). 43  One needs to examine other trade-offs between the methods. All methods have advantages and disadvantages (for more details see Mitchell and Carson 1989).  Choice of administration should be  balanced by considerations of cost, speed, response rate, sort of questions asked, length of questionnaire, and quality of data (Weisberg et al. 1989).  The best choice in one setting is not  necessarily the best choice in another setting.  For this study the telephone method of surveying was chosen.  Low  costs, high response rate, speed, and the interviewer being able to answer respondents' questions were considerations in the decision.  However, there exists a possibility of bias in socio-  economic status and interviewer bias.  Measures were taken to  reduce these biases.  Sample and Errors The population to be surveyed was developed from the list of B.C. Tel. Directory for the Abbotsford area by choosing every 10th household.  Population choice bias was greatly reduced since the  random sample represented the population that was receiving water which was polluted by livestock wastes.  Only residents receiving  water from the Abbotsford aquifer area were interviewed, as well as people who did not know the source of their water but were living in the area affected by agricultural wastes in water. There were two boil orders that occurred earlier in the same year.  Surveying individuals after boil orders will affect  results significantly, but measures were taken to reduce this effect.  For example, people were asked to make their decisions 44  about the water ignoring the boil order time periods.  One measure of quality of a survey is sampling procedure. Nonprobability or non-randomized samples can give biased results. Probability samples should be used to get good results (Weisberg et al. 1989) . With these kinds of samples it is possible to calculate the sampling error statistically.  There are other  sources of errors in a sample, but the error that results from sampling is especially important, since it can be estimated statistically.  One of the first questions concerns sample size. A random sample of adults living in the Lower Mainland was available. unclustered proportionate stratified sample is used.  An Stratifying  a sample was dividing the population up into small, manageable chunks and randomly sampling from each chunk (Weisberg et al. 1989).  In this case, a somewhat smaller sample size might  suffice because of the gains in using stratified samples. However, these gains will be small.  It is also important to  obtain a wide range of subclasses, regions, and education levels. A major reason for large samples is that they permit finer divisions of the sample for subclass analysis that can be conducted (Kalton 1983).  The choice of sample size was  determined largely by funding availability, an assessment of the costs of increasing the sample compared with the possible benefits of a more detailed analysis, and an estimate of the return rate.  45  It is important to design probability samples that do not bias the results and that keep costs reasonable.  Weisberg et al.  (1989) point out that, as long as the interviewers cannot select the respondent, the sample size is large, and if there are enough clusters, samples will be highly representative of the population.  Sometimes, this may not be true. A bad sample can  be detected by checking whether the sample approximately matches the percentages for each gender, race, and educational level given by the latest census data.  The more people who are interviewed, the smaller the error. Most national samples employ about 1,500 "observations".  The sampling  error with that size is generally about 3 percent (Weisberg et al. 1989).  To cut a 3 percent sampling error to 1.5 percent, one  would need to use 6,000 observations.  The extra expense can not  be justified.  The sampling fraction, which is the percentage of the population that is being surveyed, also affects the sampling error.  As  Weisberg et al. note, a high sampling fraction of 30 percent results in the likelihood of having the sample being very similar to the general population.  Usually the sampling fraction is very  small. Few samples include more than 1,500 surveys, even when the population is in the millions for a country or hundreds of thousands in major cities. Typically, the sampling fraction is less than 1 percent (assuming a population proportion of 0.5, which yields maximum error), which is too small to matter. Weisberg et al. point out that one needs as large a sample to 46  study the attitudes of 600,000 residents of San Francisco County or 9 million Florida citizens as for the entire United States. A higher sampling error will result if a smaller sample is employed for studying a smaller area.  The response rate must also be taken into consideration when choosing sample size. Non-response can be a problem, though people who refuse to respond usually do not differ much from those who do  respond (other than being less cooperative).  The  higher the refusal rate, the more important it is to determine whether the non-response is coming from a particular segment of society.  As noted above, the demographics of the sample can be  checked by census data to determine how representative it is. Usually, though, non-responses can be ignored (Weisberg et al. 1989).  A high response rate was expected for this survey.  This is  primarily due to the high level of interest in the topic; furthermore, it is a non-controversial issue and the questionnaires were quite short and direct.  A sample size of 115  was used for this study, it seems an adequate size, given the fact that there will be little segmentation done to the data and high quality responses are expected for key questions.  Wording of Questions Surveys must be designed so respondents understand the questions and feel comfortable answering candidly. must be clear to all respondents. 47  The meaning of words  To avoid ambiguity, questions  are short and direct.  Bias must also be avoided in questions.  Biased questions are those that make one response more likely than another, regardless of the respondent's opinion (Weisberg et al. 1989).  Other points such as avoiding double-barrelled  questions (those that asked two questions at the same time) and using unbiased terms were other factors considered in designing questions.  People who lack a real opinion on issues should be considered (Weisberg et al. 1989): "[S]ome respondents develop instant opinions in order to avoid appearing uninformed to the interviewer. The answers they give, though, do not reflect well-informed opinions and sometimes reflect incorrect guesses about what the policy is." (p.71). Two techniques exist to avoid measuring of instant opinions. One method is to explain the issue in neutral terms before asking a question and the second is to first ask whether the person has read or heard much about the topic. were used.  In this study both methods  The first one was usually employed, but individuals  were also asked if they required more information on the topic, which they were provided in neutral terms.  Closed-ended vs. Open-ended In constructing a survey, one needs to decide on the type of question format to use to measure each variable.  Closed-ended  questions offer a series of alternative answers among which the respondent must choose, like a multiple choice examination question.  48  The maximum amount a consumer is willing to pay (WTP) for nonmarket commodities such as improvement in water quality is a measure of the economic value of this resource.  The two most  widely used methods of obtaining WTP are the open-ended and closed-ended questioning methods.  Both methods are used to  estimate the same value and to construct WTP; however, it is not clear whether these two approaches result in same contingent values in practice (Kealy and Turner 1993).  Open-ended questions  ask people to identify their maximum WTP, whereas closed-ended questions specify an amount and ask people whether or not they are willing to pay that amount.  Kealy and Turner (1993) recognized that, if contingent values are sensitive to questioning format, the validity of either one or both of the methods is questionable.  They developed a test to  determine whether the open-ended approach lead to significantly different results.  In their examples it did.  For the public  good, which was preventing additional damage to the Adirondack region aquatic system from acidic deposition, the closed-ended values were 1.4 to 2.5 times as large as the open-ended values, depending on the specification.  The explanations given for a  discrepancy in the errors and the estimates of willingness to pay included perceived differences in incentives for strategic behaviour depending upon the question format and differences in the individual's ability or willingness to ponder the different questions.  It would be useful to establish which questioning method is 49  likely to contain less error.  Previous studies (e.g. Seller et  al. 1985) suggest that, since people are familiar with market situations where a price tag is attached to a good and they must decide whether or not to pay for it, the closed-ended question format should be more reliable (i.e., contain less error) than the open-ended procedure.  It is rare when consumers are faced  with open-ended price situations. Mitchell and Carson (1989) state that respondents find the close-ended procedure easier to answer than the open-ended format. Kealy and Turner (1993) conclude that:  "it is not clear how best to resolve the  inconsistency in welfare estimates based upon question format" (p.329).  They suggest two means of handling the problem: one is  to investigate whether and under what circumstances one method produces consistently better results, and then use that method; the second is to use both methods and employ their estimation method.  If the results are sensitive to question method, one  should show both estimates.  The authors further contend that  open-ended formats generate lower values of WTP than do closed formats.  For the scope of this study, the majority of previous related studies using surveys (for example, Hauser et al. 1993; Hayward 1985; Musser et al. 1992; Sellar et al. 1985, 1986; van Kooten et al. 1986) employed the closed-ended technique.  Since the results  they were seeking were similar to the ones sought in this study, it seems that the closed-ended technique would also be better suited for the current investigation.  Dividing the respondents  up into two categories for the two methods of questioning is not 50  suitable for this case because the focus of this project is not to evaluate the two methods.  The chief advantage of the closed-  ended question format is that it provides the same set of guidelines for all respondents to use in determining their responses (Weisberg et al. 1989). inexpensive to work with.  The resulting data is easy and  The closed-ended techniques will  adequately serve the purpose of this study.  A common format for closed-ended questions is to ask the respondent to agree or disagree with a given statement. However, simple agreement or disagreement with a statement gives no indication as to how strongly her opinion.  the respondent feels about his or  This type of measurement can be made by asking the  respondent whether they strongly agree, agree, neither agree nor disagree, disagree, or strongly disagree.  This is called a  rating scale. Another form of a rating scale is to draw a line between two extremes and then get respondents to indicate their position between the two lines.  In this study, the closed and  open-ended questions are used in symbionce. dichotomous choice question is asked.  First the  This is taken a step  further by asking individuals to give their maximum WTP.  It is  expected that results from the dichotomous choice format will be more accurate.  Pretest Questionnaire construction is really an art, much of which is learned through practice.  In fact, it is so difficult that  without pretesting, researchers rarely use a survey (Mitchell and 51  Carson 1989).  Questions are usually tried out on individuals who  are not going to be a part of the sample. Minimally, it is important to try out the questions on at least a few people (Weisberg et al. 1989).  Problems encountered during the pretest for the questionnaire to be used for this study were rectified in the final version. telephone survey, clarity and brevity were sought.  With  A more  realistic length was needed, since the original one took too long to administer.  The pretest also indicated that individuals  responded more ably to closed-ended questions than open-ended questions.  The open-ended questions had a better response  scenario when the original question was closed and the second part of the question was open.  For example, individuals were  asked if they would be willing to pay $X for a Y amount of reduction in nitrate in water.  If they said yes (or no), then  they were asked to provided the maximum amount the would be willing to pay.  This approach determines an upper and lower  bound on responses. This increases statistical accuracy of the subsequent welfare measures (Kanninen 1993).  Also, the  reluctance of individuals to offer details about their socioeconomic status was noted during the pretest.  The need to  clarify the need for such information and assurance of confidentiality was important.  Survey Outline There were five sections to the survey (see Appendix).  Section 1  comprises four opinion questions on issues such as water quality 52  and environmental concerns relating to agriculture.  The  objective of section 2 is to obtain information on residence, i.e. how long individuals have lived in the area and their residence, whether they owned their home or rented, and so on. This section gave a general idea of how much the respondent can be expected to know about local water problems because it asked individuals how long they have lived in the area.  There may be  residents who do not know much about the concerns facing the region; therefore, this question may be important in identifying why some responses do not align with the answering pattern for the majority of responses.  Section 3 dealt with sewage and water quality more directly. Since most residents in the study region (Abbotsford) are on septic systems, and septic tanks have been suspected as a source of nitrate pollution, a specific question on septic system maintenance was included in the survey.  If respondents indicate  poor septic system maintenance, then education campaigns and mandatory septic cleaning may reduce the environmental concern with little resulting cost.  The questions directly related to water quality included questions on avoiding water pollution using filters and bottled water, i.e. averting expenditures.  Since individuals who are  buying bottled water or installing water filters pay for these services, these figures can be used as comparisons for WTP value. However, because these figures do not include the option value of clean water (i.e. knowledge that clean water exists despite not 53  using it), their avoidance expenditures will underestimate their true WTP.  The WTP question is created by using a demand function derived from the estimated costs of eliminating nitrates in the groundwater by composting manure before application, which are similar to the amounts used by Hauser et al. (1993).  Using this  figure as the maximum cost per household, a demand curve was constructed to determine the quantities to give as options for the WTP.  The reference level used was 12 mg/L nitrate  concentration in water, with no costs to maintain this level. If only one level of environmental quality is given, functional form used for WTP can constrain respondents and may condition the responses.  Therefore, different target levels were provided.  Different prices and water quality levels were presented, which also helps to better differentiate response groups.  To make this  situation realistic, charges for water quality improvements were to be added to the annual water bill or taxes; however, if people do not believe the payment vehicle they may not respond seriously.  This effect can be tested using information from  sections 3 and 5 to determine if the values given are possible considering the persons income and fixed expenses.  4.3 WTP MODEL Both the closed and open-ended models were used to arrive at individuals' E(WTP) amounts.  First individuals were asked if  they were willing to pay $"x" annually, as a part of their water bill, to help reduce the current nitrate level from 12 mg/L to 54  "y" mg/L. next.  Both x and y were varied from one respondent to the  Then they asked to give the maximum amount they would be  willing to pay to reduce the nitrate level to the same "yM amount.  Using this second question defines the respondents'  maximum WTP better than using the dichotomous choice question by itself.  Open-Ended Model The open-ended contingent valuation model was specified as: WTP = f(Q,M,X,R)  (7)  where Q= M= X= R=  the target nitrate level in water; income; initial dollar amount offered in closed-ended question; remaining factor affecting willingness to pay, including age, education, attitudes, household size and so on; and WTP= Hicksian equivalent measure of willingness to pay.  This equation was estimated for the Abbotsford area residents receiving water from the region's aquifer.  Since there was no a  priori assumption about functional form, equation (7) was estimated two ways: linear and linear with a squared term in Q. Sellar et al. 1986 used these functional forms and chose the one most suited to their model.  For open-ended contingent valuation,  the linear OLS and linear tobit are most commonly used (Halstead et al. 1991).  Differentiating the above equation results in an inverse Hicksian demand curve.  Both ordinary least squares (OLS) and Tobit 55  analysis are used and compared.  Generally, Tobit analysis is  considered to be more theoretically correct for this type of analysis.5  Closed-Ended Model Respondents were asked if they were familiar with issues relating to water quality and then given information about the nitrate issue if they requested.  They were then told that 10 mg/L of  nitrate is the government safety standard, and that the current level is approximately 12 mg/L.  Following this they were asked  to give one of two responses, "Yes" or "No," to one of the following randomly selected questions: Are you willing to pay (xlf x2, x3) dollars annually, as an inclusion to your water bill, to bring the nitrate level down to "y" mg/L? where Xj was associated with y=0 and was one of 28, 85 or 200; x2 was associated with y=6 and was one of 57, 85 or 200; and x3 corresponded with y=10 and was chosen from 142, 170 or 200. A binary response model is the proper model to analyze this type of data (Sellar et al. 1985).  In a binary response model the  dependent variable takes one of only two values.  In this case, a  value of "0" for yes responses and "1" for no responses. A cumulative density function (CDF) is then used to calculate the expected willingness to pay, E(WTP).  5  See Halstead et al. 1991 for more details. 56  Cumulative Density Functions Analysis is more difficult for responses to dichotomous choice questions than open-ended ones.  Data on values are in the form  of "yes" or "no" responses to specific, randomly assigned offer amounts (Bishop and Heberlein 1988).  From this data the  inference of maximum WTP is arrived at by estimating logit or probit regression models. These models predict the probability of rejecting the offer as a function of the offer amount and other explanatory variables.  The mathematical expectation of WTP  is then calculated using the probabilities6.  Figure 2 illustrates the calculation of values from dichotomous choice responses (see equations 8 and 9 below for a mathematical representation of P(N) and E(WTP)).  The vertical axis measures  the probability that a randomly selected respondent will answer "no" when asked whether she or he would be willing to pay a given offer amount (x), measured on the horizontal axis.  The  respondent's maximum WTP is less than x if they reject an offer. Thus, F(x), which is the function estimated using logit or probit, is upward sloping to indicate the increasing probability that a randomly selected respondent will reject the offer as the offer amount (x) increases.  "Since a refusal to pay x implies a  refusal to pay all amounts greater than x, F(x) can be interpreted as cumulative density function on the probability of rejecting x" (Bishop and Heberlein 1988, p.89).  Since the  expected value of the random variable equals the area above the  6  At the median, the probability of rejection equals 0.5. 57  CDF and below the line F(x)=l (Hanemann 1984), the estimated expected value of WTP per individual in the subject population equals the shaded area in Figure 2.  Probability of No {P(N)> A 1  v\y" \  \|P{N)  0  Maximum X (M)  X($)  Figure 2: The Logit Curve (From Bishop et al.  1988).  There is very little difference between the logit and probit models.  One should strive to use a functional form that is  consistent with utility maximization and fits the data.  The  probit model has a normal distribution, while the logit has a logistic one.  Since the logistic CDF can closely approximate  that of a normal random variable, there is usually little difference in the empirical results produced by the two models 58  (Sellar et al. 1985).  Since estimates are easier to obtain using  the logistic distribution, logit models are more commonly used, unless there are theoretical reasons preferring the normal to the logistic distribution.  In this study the logit was used due to  its computational simplicity and the fact there were more studies using it for a binary response models than the logit. (e.g., Abdalla et al. 1992; Cooper and Loomis 1992; Seller et al. 1985; Cameron 1992).  The logit model for this study was specified as follows: P(N) = logit (Q.M^X.R) =l/[l+exp(ziB)]  (8>  where zt = estimated coefficients; B = variables used; and P(N) = probability of no (see Figure 2 for illustration).  From the above equation expected willingness to pay is derived as follows (Sellar et al. 1985): M  E(WTP) = M-j'LOGIT(0,M,  X, R) dX  (9)  o  where M is the maximum "bid" or amount presented to respondents; in this case M = 200.  A Log-linear functional form, which is popular for dichotomous choice contingent valuation (Cameron 1988; Sellar et al. 1985, 1986) is the one chosen for this model. 59  Sellar et al. (1986)  point out that for this type of contingent valuation a log-linear form is preferred over a linear one. Theoretically, the following restrictions must be satisfied to ensure that the Hicksian demand functions are positive and downward sloping: fQ = dP(N)/dQ  * 0  U 0 )  and  A linear function form will satisfy only the first condition (equation 10).  It will always be upward sloping, which violates  the theoretical condition in equation 11. This indicates that a linear specification , and therefore the absence of curvature in the utility function are not appropriate for modelling contingent valuations obtained from a dichotomous choice question format (Sellar et al. 1986).  4.4 DEFENSE EXPENDITURES Averting expenditures (or defense expenditures) related to drinking water need to be estimated to obtain a conceptually valid conservative estimate of actual costs or benefits of changes in drinking water quality.  As Abdalla et al. (1992)  indicates, averting measures related to drinking water are easily defined by consumers' purchases of bottled water or home water filtration systems.  To measure averting expenditures made with 60  the intent of avoiding exposure to contaminated water, questions concerning the purchase of bottled water and filter systems are included to the survey.  To address the empirical estimation problem of including expenditures for unrelated reasons, such as change in tastes and preferences, respondents were asked why they purchased bottled water.  Items such as sheer convenience and a fresher, better  taste of the purchased water were separated.  Information was  collected regarding possible factors influencing averting expenditure changes, such as pollution perceptions and demographic factors.  In particular individuals were asked if  their defense expenditures were because of the pollution problem or not.  Household defense expenditures comprise two steps.  First is to  see whether action is being taken to avoid contaminants in the water, in other words the use of bottled water, filters or boiling the water.  The second is to measure the magnitude of  these actions in dollar terms—the amount of money being spent on averting actions.  Logit regression was used to determine factors influencing decisions to undertakes averting actions. The dependent variable was equal to "0" if the household took any sort of averting action to avoid polluted water and was set equal to "one" if no such actions were taken.  Only individuals who made a decision to  avert considered a second decision step—how much to avert by 61  (Abdalla et al. 1992).  Averting behaviour intensity was measured  by how much households spend on defense expenditures.  OLS  regression was used to determine factors contributing to changes in averting expenditures made by households.  62  CHAPTER 5  5.0 EMPIRICAL ANALYSIS This chapter presents the empirical results. The first section presents descriptive results from the survey regarding response rate and socio-economic factors. Then estimates of WTP are provided, followed by an analysis and comparison to averting expenditures obtained from the survey.  5.1 SURVEY DESCRIPTIVE RESULTS It was necessary to mention the topic and purpose of the survey and the fact we were not trying to sell anything, to maximize the number of results and improve quality of responses.  While the  majority of the interviews went smoothly, sometimes a great deal of tact and ingenuity was required to complete the interview. Most interviews took place in the evening, to avoid biasing the sample with an unproportionate number of housewives and retired citizens compared to the residents of the area in general. Interviews generally took no more than 10 minutes to complete. There were a couple of individuals who had a great deal to say about the issue and wanted to talk endlessly.  Care was taken to  remain neutral during the interviews.  Response Rate The response rate was 89%, or 102 completed survey. excellent response rate for analytical purposes.  This is an  This response  rate is significantly higher than similar surveys. A telephone survey conducted in the U.S. by Musser et al. (1992) (for a small town similar to Abbotsford) after a boil order, had a response 63  rate of 61%. A response rate of 27.4% was reached by a mail survey (Hauser et al. 1993), on the same topic, conducted earlier in the same year as this study's survey.  From the non-responses,  two individuals could not speak English; hence, could not answer the survey.  Weisberg et al. (1989) estimates that the average  response rate for telephone surveys is approximately in the 70% range.  The response rate for this survey was increased by  calling back individuals who requested an interview at a later date.  Also, spouses were permitted to answer if the individual  on the list was not home.  Wrong numbers were erased and not  counted in the list.  As expected there was a lot of interest in the topic and people answered most questions well. personal information.  Hesitation arose in giving  But once they were assured of  confidentiality most individuals were cooperative. was not answered well.  One question  It asked individuals to give their  monthly mortgage or rent.  Categories of amounts were offered to  simplify answers, but 10.7% of the respondents refused to answer, although they did reveal whether they rented or owned their residence.  These findings are discussed in the following  section.  Summary of Socio-Economic Factors A summary of the personal and background information obtained from the survey is found in Table 1.  The average age of the  respondents was 41.3 years, while the average education level was one half year of post secondary schooling following the 64  completion of high school. These results confirm the findings obtained by Hauser et al. (1993) for the same region and to the Statistics Canada 1986 average education level of slightly over 12 years.7 As a result, one can assume that there is no educational bias between respondents in the sample and the region of focus for this study.  Average family size was 3.0  individuals, which was the same result Hauser et al. (1993) found.  This is comparable to the 2.9 persons per household  average indicated by Statistics Canada's 1991 Census.  Hauser et al. (1993) found that respondents lived in their homes for an average of 7.0 years and in the neighbourhood for 15.2 years.  This study found results to be fairly close with 5.8  years being the average for current residence and 11.4 years the mean for residence in area.  These findings indicate that  residents should be familiar with water quality issues pertaining to their area.  (In the survey, individuals were also asked  whether they were familiar with such issues. discussed later).  This will be  None of the respondents were farmers.  If they  had been, their views towards agricultural pollution and water quality would be affected by the fact they own farmland.  7  Not all census data are available 1991. 1991 data is used where found; otherwise, 1986 census data is used. 65  Table 1:  Summary of Personal and Background Information for Abbotsford Area. Mean  Coefficient of Variation  Age of respondents  41.3 years  0.444  Female respondents  49.0%  0.985  3.0  0.503  0.50  1.894  80.0%  2.035  Length of time living in area  11.4 years  0.852  Length of time in current home  5.8 years  0.907  Education  12.5 years  0.186  $33,500  0.725  Item  Family size Number of children under five Own their own homes  Household income  The average household income found in the study was $33,500 per year, which is significantly lower than the 1989 census' average of $46,493 and the amount found by Hauser et al. (1993), $44,620. This lower result may be due to the fact that respondents hesitated to give an income figure at first, and then when assured of confidentiality, they finally gave a figure. Weisberg et al. (1992) explains that, although telephone interviews receive a higher response rate to socioeconomic factors than mail out surveys, respondents have a tendency to give biased responses to personal questions. their incomes.  Individuals have a tendency to understate  This may be due to tax evasion purposes.  There  was no significant irrational answers with regards to WTP values, implying that respondents answered rationally to the WTP question, and the amount they gave was feasible, given their income level. 66  About 95% of the respondents had no septic tanks, the remainder did not know whether they had one.  One can eliminate septic  tanks as a probable source of nitrate, since very few have one. Approximately 10.7% of the respondents refused to reveal the amount of rent or mortgage they paid per month. had the lowest response rate.  This question  The majority of households, 80.0%,  owned their own homes and had lived in the residence for a long time.  May have had no mortgage.  Table 2 shows the categorical  responses to the amount of rent or mortgage payment per month. The average monthly rent or mortgage, assuming owners paid some property tax, was $600.  This amount is the same as the one  obtained by Hauser et al. (1993).  Table 2:  Rent or Mortgage Payment per Month for Abbotsford Residents8 % Responses  Number of Respondents  less than $500  59.8  61  $500 - $750  5.8  6  $750 - $1,000  15.7  16  $1,000 - $1,250  8.0  9  more than $1,250  0  0  10.7  10  Rent or Mortgage per Month  no responses  Opinion Questions Four opinion questions were asked in the questionnaire.  This  section descriptively analyzes individuals' responses to these  Assuming owners paid property taxes. 67  four questions.  Table 3 outlines answers received to opinion  questions used in the survey.  The average person was at least  weakly familiar with the issues of nitrates in water and/or the effects of nitrates on health.  Only 13.7% strongly disagreed  with the statement that they were familiar with the aforementioned issues.  It would be safe to assume that most  people were at least somewhat familiar with the issues in question.  Table 3:  Abbotsford Residents' Opinions Regarding Off-Farm Pollution  Statement You are familiar with the issues of nitrates in water and/or the effects of nitrates on health.  % Agree9  Neutral10  % Disagree  58.9%  3.9%  37.2% WD = 23.5% SD = 13.7%  SA = 21.7% WA = 37.2%  Smell and other farm nuisances are a problem.  30.4% SA = 18.6% WA = 11.8% 39.2%  Water quality in your area is adequate.  WD = 45.1% SD = 16.7% 3.9%  SA = 11.7% WA = 27.5% Government needs to impose strong regulations on fertilizer use and the handling of livestock wastes, regardless of cost to farmers.  70.6% SA = 48.1% WA = 22.5%  61.8%  7.8%  56.9% WD = 15.7% SD = 41.2%  11.8%  17.6% WD = 8.8% SD = 8.8%  As far as smell and other farm nuisances are concerned, 61.8% felt that there was no problem with smell and other farm 9  WA = weakly agree; SA = strongly agree; WD = weakly disagree; and SD = strongly disagree. 10  Neutral or no response. 68  nuisances, but only 16.7% felt strongly that there was no problem with this issue.  The answer to this question would depend  somewhat on an individual's proximity to farms.  The closer they  live to farms, the more likely they would feel that there is a problem with farm nuisances.  The need for government to impose strong regulations on fertilizer use and the handling of livestock wastes was echoed in the survey.  About 48% strongly agreed that the handling of  livestock wastes and fertilizer use needs to be regulated by the government, regardless of costs to farmers. A total of 72% at least weakly agreed to this action.  This number indicates  dissatisfaction with current government regulations and current farm practices.  Correspondingly, almost 60% felt that water  quality in their area was not adequate. this conviction strongly.  A total of 41% expressed  The boil order, which was present  immediately prior to this survey, may have contributed to the size of this number, despite efforts to eliminate this effect." Only 11.8% strongly maintained that water quality was adequate. These results show that people did feel that there was a problem with water quality, and they linked it, primarily least partly to the mismanagement of farm wastes.  11  Respondents were told to consider the time period prior to the boil orders, when giving their opinion about water quality. 69  5.2 WTP MODEL RESULTS Results obtained from the WTP models are two fold: open-ended and closed-ended.  For the open-ended procedure tobit and OLS  regression results are discussed, while logit regression results are presented for the closed-ended procedure.  Table 4 outlines variables used in the regressions.  For the most  part, data used for all the models was of good quality.  This  means that there were few gaps and virtually no multicollinearity between variables. multicollinear.  Only two sets of variables were  The variable capturing number of individuals  under the age of five and household size had a correlation coefficient 0.527.  This was to be expected because households  having more children under five will usually have more members in the family as a whole. age group (0.504).  Household size was also collinear with  This simply implies that older respondents  had larger families, which makes sense since younger individuals are usually starting out in life and have few or no children at this point in their life.  Since none of these collinear pairs  were used in unison, modifications were not necessary.  WTPASKED  and Qn were also multicollinear (-0.388), but this was relatively low.  The set up of the question made this happen because  WTPASKED was higher for lower Qn.  This means that the dollar  amount individuals were asked to pay was higher for lower target levels of nitrate.  70  Table 4; Description of Variables Used in Regressions. Variable Familiarity Farm Pollution Opinion about water Gov't role's opinion Years in residence Years in area Rent/Own Qn (Qn)2 Age Gender Household size Individuals under 5 Education Income WTP(N/Y) MAXWTP WTPASKED AVERT(Y/N) AVRTEXP  Description Opinion question 1 on survey, (15 ) : 1 = familiar; 5 = not. Opinion question 2 on survey, (15 ) : 1 = farm pollution is a problem; 5 = not. Opinion question 3 on survey, (15 ) : 1 = adequate; 5 = not. Opinion question 4 on survey, (15 ) : 1 = should regulate; 5 = not. Number of years lived in current residence. Number of years lived in area. Rent - 1; Own = 0. Change in nitrate level, from 12mg/L, Qn = 2, 6, 11. Qn squared. Categorical listing, (1-6): 1 = under 25; 6 = over 65. 1 = Male; 0 = Female. No. of individuals in household. No. of children under the age of 5. Grade completed + number of years post secondary. Categorical listing, (1-9): 1 = under $30,000; 9 = over $100,000. WTP to bring nitrate down to Qn: 1 = no; 0 = yes. Maximum WTP to bring nitrate down to Qn. "Bid" amount offered in WTP closedended question. Existence of any defense expenditures: 1 = yes; 0 = no. Annual amount ($) of defense expenditures.  71  Closed-Ended Model Results The log-linear logit model was specified as in equation (8). WTP(N/Y) was the dependent variable, where no = 1 and yes = 0. Table 5 outlines logit regression results of decision whether to pay for nitrate reduction.  Approximately 71% responded with a  "yes", while the remainder were "no" responses to the dichotomous choice question.  Gender, opinion about water, income and WTPASKED were the parameters that were significant at the 0.025 confidence interval.  An interesting result was that females were more  likely to say yes to payment for such an improvement than males. Women might be more concerned than men about the health of family members.  People who thought water quality was poor in their area  were more likely to pay clean up costs. This result is consistent with logic.  Unexpectedly, individuals with higher  incomes were more likely to say "no" to the bid.  As Table 9  shows, households with larger incomes spend more on bottled water and water filters. As expected, WTPASKED did have a significant effect on WTP(N/Y).  Higher WTPASKEDs resulted in a higher  probability of a "no" response.  The probability of not paying for an improvement in nitrate level went up with age. Maybe the younger generation is more health conscious or more aware of the environmental implications for future generations. However, this was an insignificant result, as were the results for years in area, AVRTEXP, and education. The probability of a "yes" response increases with years lived in 72  area.  Individuals living longer in the area maybe more aware of  the problems with the water, and they maybe more inclined to remain there longer than individuals who have been for a shorter period.  There is a slight increase in the probability of a "yes"  response as AVRTEXP increases.  In other words, respondents who  are buying bottled water of filters, are more likely to say "yes" to WTPASKED.  Results also indicate that people who are familiar  with the issues of water and nitrates in water are more likely to say "yes" to clean-up costs.  Goodness of fit measures indicate that the model used was mediocre.  R-value was 0.22, which is slightly lower than what  other studies obtained for closed-ended models.  For example,  Seller et al. (1985) found R-values to range from 0.40 to 0.31, and this was considered to be a very good fit.  The log-  likelihood of the function was -54.02, which is similar to what Kealy and Turner (1993) found in their linear maximum WTP model. They found this value to be acceptable, although is it quite low. About 76.5% of the observations were explained by the model, which makes the model fairly good.  The functional form served  the purpose for the study, since they were limited observations, one could not do a detailed analysis regardless of how well the model fit the functional form.  73  Table 5: Log-linear Logit Regression Results of Decision Whether to Pay for Nitrate Reduction in Abbotsford water Coefficient  T-Value  Opinion about water  -0.801  (-1.702)*  Familiarity  -0.547  (-1.301)  Years in area  -0.314  (-1.230)  Age (years)  0.611  ( 1.227)  Gender (l=male)  1.021  ( 2.153)*  Individuals under 5  0.404  ( 1.379)  Income  0.883  ( 2.074)*  Variable  AVRTEXP  -0.121  (-1.237)  Qn (mg/L)  -0.376  ( 1.376)  Education (years)  -1.744  (-1.281)  0.921  ( 2.250)*  -0.709  ( 0.853)  R-Value: 0.22  * significant at the 0.025 level  WTPASKED Intercept  102 Observations  76.5% correctly predicted by model R-value is interpreted si milar to R-squared.  Equation 9 was used to calculate E(WTP).  The maximum level was  $194.90 and the minimum was $57.69, which corresponds to the maximum level of 200 and minimum of 57 offered to the respondents.  The mean E(WTP) was $97.73 per year with standard  deviation of 34.59.  (For a comparison of this result to MAXWTP  from the open-ended model and AVRTEXP from the closed ended model see Table 10).  74  Open-Ended Model Results The open-ended question itself generated a maximum WTP amount (MAXWTP).  For a 6 mg/L reduction of nitrate in water, the MAXWTP  was $81.03 annually on average with a standard deviation of 83.081. year.  There was a large range of answers, from 0 to $300 per  This result will be later compared to E(WTP) and AVRTEXP  (Table 10).  An interesting fact was that WTPASKED had very  little correlation with MAXWTP, in other words, the amount respondents were asked to decide on in the closed-ended question, did not significantly influence the dollar amount they were willing to pay in the open-ended question. theoretically correct result.  This is a  As indicated in earlier sections,  usually there is a problematic correlation.  Varying the WTPASKED  amounts and offering a large range may have contributed to reducing this bias.  Using equation (7), "bid" curves were established using data collected from the open-ended question. OLS and tobit regression were used to examine data from the open-ended models. The dependent variable MAXWTP was regressed both in a linear and Qn2 functional from.  OLS results are tabulated in Table 6, with  tobit results in Table 7.  Table 6 indicates that Qn and farm pollution perceptions produced the most significant results. As expected, a lower target nitrate level results in a higher MAXWTP.  People were willing to  pay more for a lower amount of nitrate pollution.  75  Individuals  who considered farms as problematic sources of pollution were inclined to have a higher MAXWTP.  From this one can assume that  the MAXWTP did in fact link to the nitrate problem caused by farming operations.  Another theoretically correct result was that people who considered water quality to be low in their area had a higher MAXWTP.  The variable measuring familiarity with the issue in  question was dropped because it was extremely insignificant in this regression.  However, MAXWTP did increase with education.  Income is an insignificant explanatory variable in this model. Other results were females were inclined to have a higher MAXWTP than males, and the longer one lived in the area the lower the MAXWTP.  However, these results were also quite insignificant.  The regression with Qn2 produced similar results. coefficients changed in value or significance.  Relatively few  The big change  was in Qn, which became more negative but less significant. Since Qn2 was insignificant, the linear regression is considered a better result.  The change in R-squared was virtually non-  existent.  AVRTEXP was very small in size and extremely insignificant; therefore, it was dropped from the open-ended model.  Goodness of fit measures indicate a reasonable fit for the  76  analysis, and given that survey data was used.  For this study a  high degree of fit was neither expected, nor required. OLS regression generated an acceptable R-squared (0.28 and 0.29). Seller et al. (1985) generated an R-squared of 0.25 for a similar model, which was considered sufficient for this type of analysis.  Table 6: Ordinary Least Squares Regression Results for Open-Ended Contingent Valuation Questions Asked to Abbotsford Residents. Variable Farm Pollution Years in area Qn  Regression Coefficient (T-Value)  Qn2 Regression Coefficient (T-Value)  -13.37* (-2.43) 1.32 ( 1-48) 4.75* (-2.56) 14.16* ( 2.78) -11.71 (-0.53) 6.53* ( 2.13) -5.74 (-1.19) 0.15 ( L45) 24.47 ( 0.36)  -13.31* (-2.45) 1.41 ( 1.56) -13.94* (-2.10) 0.35 ( 1.13) 14.45* ( 2.53) -11.41 (-0.45) 7.72* ( 2.07) -5.59 (-1.28) 0.57 ( 1.33) 7.63 ( 0.28)  R-Squared: 0.28  R-Squared: 0.29  (Qn)2 Opinion about water Gender Education Income WTPASKED Intercept 102 Observations * coefficient significant at the 0.025 level.  Tobit regression results (Table 7) were not much different than the OLS results.  Coefficients did not change much in sign,  magnitude and significance. to be insignificant.  As in the OLS model, Qn2 was found  Since one functional form is not clearly  better, the simpler OLS form was chosen for further analysis.  Table 7: Tobit Regression Results for Open-Ended Contingent Valuation Questions Asked to Abbotsford Residents. Variable  Regression Coefficient (T-Value)  Qn2 Regression Coefficient (T-Value)  -14.46* (-1.99) 1.42 ( 1.32) -5.53* ( 2.27)  -12.84 (-1.77) 1.84 ( 1.67) -20.74* (-2.28) 1.60 ( 1.74) 20.39* (-2.27) -26.31 (-1.26) 10.82* ( 2.40) -7.83 (-1.18) 0.17 ( 0.54) -57.47 (-0.82)  Farm Pollution Years in area Qn (Qn)2 Opinion about water  19.12* ( 2.84) -25.79 (-1.22) 9.22* ( 2.08) -7.74 (-1.16) 0.28 ( 0.67) -5.53* ( 2.27)  Gender Education Income WTPASKED Intercept 102 Observations * coefficient significant at the 0.025 level.  Log-1ike1ihood function: -449.36  Log-1ikelihood function: -447.83  5.3 DEFENSE EXPENDITURES When respondents were asked the reason they bought bottled water or filtration systems, 94% indicated the pollution problem as at least one of the reasons. Virtually all respondents boiled water at the time the survey was taken, because there was a boil order in effect.  To get a more accurate estimate on defense  expenditures, individuals were asked their defense expenditure average over the past year.  Households spent an average of  $34.46 per year on filtration systems and $108.49 annually on bottled water.  Table 10 has the value of total averting  expenditures and a comparison to AVRTEXP and MAXWTP. 78  Bottled  water tends to be more expensive, unless the filtration systems are permanently placed into the water outlets.  Small changeable  filters last a long time and are relatively cheap; about 41% opted for the filters, while 59% chose bottled water.  Averting expenditures were defined to be the sum of the results from bottled water and water filters.  For example, if an  individual chooses at least one of these options, they were counted as "averters".  Logit regression results are provided in  Table 8.  Goodness of fit measures indicate that the linear functional form was appropriate.  Musser et al. (1992) and Abdalla et al. (1992)  found the linear logit form to be sufficient for averting expenditure analysis. R-Value was 0.351, which is higher than what Abdalla et al. (1992) (R-Value was 0.30) and Musser et al. (1992) (highest R-Value was 0.17) found in their logit regression for averting expenditures.  29  Table 8:  Logit Regression Results of Decision Whether to Take Averting Actions for Abbotsford Residents  Variable  Coefficient  T-Ratio  Rent/Own  -1.044  -1.244  Farm Pollution  -0.345  -1.381  Years in residence  -0.071  -1.954*  Opinion about water  -1.236  -3.996*  Gov't role's opinion  -0.357  -1.477  Familiarity  -0.411  -1.536  Age  0.483  2.054*  Gender  3.174  3.574*  Children under 5  -0.632  -1.360  3.276  1.572  Intercept  102 Observations  R-Value: 0.351  * significant at the 0.025 level.  83.3% Correctly predicted by model. R-value is interpreted simi lar to R-squared.  Logit regression results indicate that households were more likely to take averting action if they were more familiar with the issues of water quality and nitrates in water.  Education and  income were very insignificant in this model so they were dropped as variables.  House owners were more likely to avert, but this  was statistically insignificant.  Gender and opinion about water quality were two significant results.  As expected, household's opinion about water quality in  their area had much to do in their decision to avert or not. Women were more likely to avert than men. 80  This may be tied in  with the fact that households having children under the age of five were also more likely to avert.  The likelihood of having  defense expenditures increased with age, perhaps individuals are more health conscience with age and know more about the area's water quality.  This may even have to do with the fact that  younger people are still starting out in life as far as career and family are concerned; whereas, older individuals have more financial security and larger families with more children.  The  longer a household had lived in the region the more likely they avert.  People who agree that government should play a more  active role in manure management and fertilizer use were more inclined to avert than those who favoured no government involvement.  The magnitude of averting expenditures has several influencing factors.  OLS results are shown on Table 9.  81  Table 9:  Ordinary Least Squares Regression Results for Magnitude of Defense Expenditures for Abbotsford Residents T-Value  Coefficient  Variable Rent/Own Farm Pollution Years in area  88.196  1.941*  -28.224  -2.256* 1.048  3.531 31.224  2.645*  Age  -36.897  -2.960*  Education  -15.269  -1.923*  31.196  2.856*  335.910  2.420*  Opinion about water  Income Intercept  102 Observations (75 "averted")  R-Squared: 0.27  * Significant at the 0.01 level  A different set of variables are significant in the OLS results for intensity of averting actions.  Intensity of aversion is  greatly affected by whether the individual is renting or owning a home.  Households owning a home averted to a lesser degree than  renters.  Other factors affecting amount spent on defense  expenditures age, education and income.  Older individuals  averted less, and so did more educated ones.  Education was not  significant in the logit regression, but results show that educated and older respondents are more "cautious" about expenditures perhaps, while younger and less educated individuals spend more freely.  As expected, households with a larger income  spent more on bottled water and filtration systems, but the decision to avert or not to had little to do with income. Another predictable result was the fact that attitudes about farm 82  pollution had much influence on the intensity of defense expenditures.  This provides a direct linkage to the perception  of the nitrate problem.  People who felt that farm pollution was  a problem, were inclined to spend more on defense expenditures. Households receiving water from the Abbotsford aquifer spent more on bottled water and/or on filters.  Families with children under  the age of five influenced the decision of aversion, but the extent to which they averted had little to do with the fact they had small children.  R-squared was 0.27, which is much higher than 0.12 obtained by Abdalla et al. (1992) for their OLS regression for intensity of averting actions for Perkasie area households aware of water contamination.  Goodness of fit measures show that the functional  described the model well.  5.4 COMPARISON OF RESULTS Willingness to pay results were obtained from the closed and open-ended models, and are comparable to the amount of defense expenditures individuals are making (Table 10).  Usually results  from contingent valuation are much different than those from defense expenditures (Cameron 1992, Musser et al. 1992). Averting expenditures represent a lower bound to compensating and equivalent variation (Abdalla et al. 1992).  However, some  studies (Musser et al. 1992) did find contingent valuation estimates to be lower than averting expenditures.  In this study,  the amount spent on averting expenditures was greater than both the maximum willingness to pay and the expected willingness to 83  pay.  Normally, expected willingness to pay is assumed to be more  accurate than the maximum willingness to pay, since (as shown in previous chapter) closed-ended questions are answered more accurately.  But in this case, since the open-ended question was  preceded immediately by a dichotomous choice question, the openended is more accurate.  Having the expected willingness to pay  slightly higher than the maximum willingness to pay, indicates that individuals found the amount asked to pay or "bid" to be higher than what they would be willing to pay.  The amount asked  was $131.15 on average, which is much larger than estimates of maximum willingness to pay and expected willingness to pay. Kealy and Turner (1993) also found closed-ended results to be higher than open-ended ones for a public good.  However, Bishop  and Heberlein (1988) found open-ended results to be biased upwards.  The fact that defense expenditures are much higher than results from the willingness to pay models indicates that individuals are willing to pay a lower amount to clean up water, than they are already paying to avoid the pollution problem.  Since normally  expected willingness to pay is expected to be higher, perhaps averting expenditures included some other benefits of defense expenditures.  For instance, some respondents indicated that they  found the taste of filter and / or bottled water to be better than tap water.  Others liked the temperature,  worried about the chlorine problem. to be a problem.  freshness or were  27.3% found excess chlorine  But as mentioned earlier most individuals named  pollution to be at least one of the reasons they purchased 84  bottled water and / or filtration systems. The jointness of benefits of defense expenditures contributed to a larger value. Also, since there was a boil order in effect at the time the survey was taken, averting expenditures were higher, despite attempts to reduce this effect.  Hauser et al. (1993) found  defense expenditures to be $69.59, but with a large standard deviation of $147.47. Compared to Hauser et al. (1993), the responses to the averting expenditures question were more accurate in this survey.  Hauser et al. (1993) found the maximum willingness to pay to be from $68.86 to $90.28 per year, which is similar to the results obtained in this study.  As expected, open-ended results have a  large coefficient of variation (1.025). range of maximum willingness to pay.  85  This allows for a large  Table 10: Summary of Measurements from Willingness to Pay and Defense Expenditure Models. Model  Mean  Coefficient of Variation  Open-Ended (MAXWTP)  81.03  1.025  Open-Ended (at variable means)  108.51  Closed-Ended [E(WTP)]  97.73  0.248  Defense Expenditures (AVRTEXP)  142.94  1.353  5.5 COST-BENEFIT ANALYSIS The fact that composting is not feasible from a private point of view was shown earlier.  But is it economically feasible to  compost from society's point of view? will determine that.  A cost benefit analysis  Given the nature of the data available, one  can only calculate whether it is feasible to compost all animal wastes.  The estimation of the socially optimal amount of wastes  to compost is not possible due to data limitations.  However, the  social feasibility of composting can be determined, and appropriate incentives (and institutions) that encourage livestock producers to compost can be designed.  86  Table 11: Estimated Production of Animal Wastes in the Central Fraser Valley Regional District. Total Waste Animal  Total Number*  (kg/day)b  Cattle  35,666  1,153,201  Pigs  90,069  459,352  1,776  1,954  884  20,332  1,044  2,714  323  65  54,044  140,514  7,221,298  673,988  Sheep Horses Goats Rabbits Mink Poultry Total (tonne/day)  2452  * Statistics Canada, 1991, Census of Agriculture. British Columbia. 1990. Catalogue #95-393. b Manure Production per animal figures from Hauser et al. (1993).  Table 11 provides estimates of livestock waste produced in the Abbotsford region.  Multiplying total annual waste produced by  the cost of converting that waste into manure determines the costs of composting these wastes in any given year.  The amount  of animal waste produced annually is about 890,000 tonnes (Table 11).  The total cost reducing this to compost is between $32.0  and $62.5 million each year, for compost costs of $32-$70/t. Assuming that revenues are $8-$15/t, or $7.1-$13.4 million annually, the shortfall is $18.6-$55.4 million.  The population of Abbotsford was 29,840 in 1991. Hauser et al. (1993) found that about 5,500 households in the Central Valley  87  Regional District are on ground water.  Almost all residents are  on ground water in the District of Abbotsford, which had a population of 18,864 in 1991. As Hauser et al. assumed, an estimate of 3 individuals per household indicates that 6,300 households are on groundwater.  The latter figure is used in  following calculations.  Defense expenditures of $142.92/year per household multiplied by the number of households yields a upper bound estimate of the benefits of improved water quality of $.90 million.  The average  stated expected willingness to pay of $97.73 per household results in $0.62 million in benefits, which is similar to the previous estimate.  The open-ended WTP mean ($81.03 per  household) yields a lower bound of $.51 million per year.  These  figures are similar to the ones obtained by Hauser et al. (1993).  Evidently the social benefits of composting do not exceed the social costs.  Before composting can be considered economically  feasible from a social point of view, private revenues from composting, or any other alternative means of manure disposal, must be very close to private costs. Public subsidies to reduce livestock pollution are not worth implementing if the latter is not true.  88  CHAPTER 6 6.0 CONCLUSIONS The investment decision to compost is not economically feasible for most farmers, with the lower prices of compost resulting from large quantities of supply from rural areas. Municipal programs designed to encourage composting in rural areas have flooded the market with compost, thereby reducing the market price. Transportation and handling costs are also too much, when comparing them to the value of compost.  There is no economic  incentive to compost, but there is an ecological one. Public intervention is not justified since composting is not found to be economically feasible from a social point of view.  This study shows that the pollution of the Abbotsford aquifer is not as serious a problem as originally thought.  The value that  users of groundwater attach to improvements in the quality of their drinking water is not high enough to justify the losses that livestock producers are likely to incur in reducing the pollution problem.  Public subsidies to agricultural producers  will simply encourage composting which costs more than it benefits society.  Taxes or regulation that force farmer to  compost will also be waste of money.  Economic inefficiency will  result from a social point of view and will drive some producers to bankruptcy.  This is the same conclusion reached by Hauser et  al. (1993).  Since these conclusions are based on an "all or nothing", where all or none of the livestock wastes are composted, there may be 89  some benefits from providing subsidies to producers to enable them to clean up the worst cases of pollution.  Perhaps more  research in the areas of better manure application rates timing leading to less nitrates in the aquifer, is a solution.  Low cost  extension and research will result in more efficiency by reducing the nitrate problem than high cost alternatives such as regional manure storage establishment and / or composting facilities. Farm management practices should be improved to allow a lower level of pollution.  The contingent valuation methodology used in the calculation of social benefits is a controversial procedure.  There is an  ongoing debate regarding its usefulness and validity.  Since  there is no other means available to measure such benefits, it was used.  Estimates from pollution defense expenditures were  used as a "check" factor for contingent valuation results. Willingness to pay estimates were fairly close to results obtained from defense expenditures.  Information on the amounts of nitrate in groundwater arriving from farm sources is not known.  Therefore, it is difficult to  estimate the amount of nitrate-nitrogen reduction in water if measures are taken to better management livestock wastes. Furthermore, data about the actual costs of composting techniques is rather limited.  This also holds for alternatives such as  storage and optimal application, surface injection, storage and transportation of animal wastes to other regions, the off-farm costs of manure application (odour problem), using manure as 90  cattle feed, and so on.  It is important to measure the effects of other government programs on the disposal of livestock wastes and use of water. Subsidies and the tax system itself may have an impact on livestock operations.  They may affect location of farming  operations, field crop decisions, preference of fertilizer use over manure, and so on.  Farm-level research is required to  examine the interaction of these factors with production decisions.  91  BIBLIOGRAPHY Abdalla, C.W.; B.A. Roach and D.J. Epp. 1992. "Valuing Environmental Quality Changes Using Averting Expenditures: An Application to Groundwater Contamination", Land Economics. 2:163-169. Adam, R.W.; L.M. Alexander and H.H. Rowen. 1989. "Netherlands: The Climate", The World Book Encyclopedia. 14(N-0):131. Anderson, J.C. 1990. Non-point source surface and groundwater pollution in Canada: Major concerns, and their significance for sustainable development. Ecological Application Division, State of the Environment Reporting Branch, Environment Canada. ARDSA. 1989. "Market Study and Strategic Plan for Composted Hog Manure". BCMAF. September. ARDSA. 1993. "British Columbia Pork Industry Profile". BCMAF. December. Berends, P.T.; P.L. Diebel, J.R. Williams, and J. Schlegel. 1993. "An Economic Comparison of Composted Manure and Commercial Nitrogen with Imperfect Information". Dept. of Agricultural Economics. Kansas State University, Staff Paper 1993-7. Bird and Rapport. 1986. "In" Water in Sustainable Development: Exploring Our Common Future in the Fraser River Basin, ed. Dorcey, A.H.J, and J.R. Griggs. Westwater Research Centre, UBC. Bishop, R.C.; T.A. Heberlein. 1988. "The Contingent Valuation Method". "In" Economic Valuation of Natural Resources, ed. Johnson, R.L. and G.V. Johnson. Westview Press:Boulder, San Francisco, & Oxford. Boadway, R.W. and Neil Bruce. 1984. Welfare Economics. New York: Basil Blackwell. Bower, R.; W.E. Ireland and J.L. Robinson. 1989. "British Columbia: The Climate", World Book Encyclopedia. 2(B):516a. Boyle, K.J.; Bishop, R.C., and M.P. Welsh. 1969. "Starting Point Bias in Contingent Valuation Bidding Games", Land Economics. (1969):188-194. Bradburn, N.M. 1983. "Measurement: Theory and Techniques", "In" P.H. Rossi, J.D. Wright and A.B Anderson, ed. Handbook of Survey Research. New York: Academic Press. Bromley, D.W. ed. 1985. Natural Resource Economics: Policy and Contemporary Analysis. Boston: Kluwer-Nijhoff Publishing.  92  Brookshire, D.S., M.A. Thayer, W.D. Schulze, and R.C. d'Arge. 1983. "Valuing Public Goods: A comparison of survey and hedonic approaches", American Economic Review. 72:165-177. Cameron, T.A. 1988. "A New Paradigm for Valuing Non-Market Goods Using Referendum Data: Maximum Likelihood Estimation by Censored Logistic Regression", Journal of Environmental Economics and Management. 15:355-379. Cameron, T.A. 1992. "Nonuser Resource Values", American Journal of Agricultural Economics. 5:1133. Canadian Society of Animal Science. 1991. Western Branch Meeting Proceedings: "Integration of Animal Management and Crop Production for Positive Environmental Impact". BCMAF. June. Castle, G. 1993. "Agricultural Waste Management in the Great Lakes Region and British Columbia: Comparisons and Recommendations", Sustainable Research Institute. U.B.C. Cichetti, C.J. and L.L. Wilde. 1992. "Uniqueness, Irreversibility, and the Theory of Nonuse Values", American Journal of Agricultural Economics. 74:1122-1125. Cooper, J. and J. Loomis. 1992. "Sensitivity of Willingness-toPay Estimates to Bid Design in Dichotomous Choice Contingent Valuation Models", Land Economics. 68(2):211-224. Coursey, D.L. and W.D. Schulze. 1986. "The Application of Experimental Economics to the Contingent Valuation of Public Goods", Public Choice. 1:47-48. Davis, R.K. 1963. "Recreation planning as an economic problem", Natural Resources Journal 3:239-249. Davis, R.K. 1964. "The value of game hunting in a private forest". Transactions of the Twenty-Ninth North American Wildlife and Natural Resources Conference. 29:393-403. Federal-Provincial Working Group on Drinking Water. 1978. "Guidelines for Canadian Drinking Water Quality". Hull, PQ: Canadian Publishing Centre. Fischoff, B. and L. Furby. 1988. "Measuring Values: A Conceptual Framework for Interpreting Transactions with Special Reference to Contingent Valuation", Journal of Risk and Uncertainty. 2:147-184. Freeman. A.M. III. 1984. "On Assessing the State of the Arts of the Contingent Valuation Method of Evaluating Environmental Changes". "In" Valuing Environmental Goods: A State of the Arts Assessment of the Contingent Valuation Method. R.G. Cummings, D.S. Brookshire, D.L. Coursey, W.D. Schulze. Washington, D.C.: U.S. Environmental Protection Agency. (1984):180-195. 93  Fullerton, D.J. 1990. "Economic Comparison of Swine Manure Handling and Treatment Systems." Dobbin Farms Ltd. A report prepared under contract to the Farm Lab Project, BCMAF. July. Fullerton, D.J. 1991. "Economic Feasibility Composting Systems." Dobbin Farms Ltd. under contract to the BCMAF (and Food) Canada-B.C. Soil Conservation Program.  of Alternative A report prepared as a part of the July.  Gerking, S.; S. De Haan, W. Schulze. 1988. "The Marginal Value of Job Safety: A Contingent Valuation Study", Journal of Risk and Uncertainty. 2:185-199. Groves, R.M. and R. Kahn. 1979. Comparing Telephone and Personal Interview Surveys. New York: Academic Press. Hall, K.J.; H. Schreier and S.J. Brown. 1991a. "Water Quality in the Fraser River Basin". "In" Water in Sustainable Development: Exploring our Common Future in the Fraser River Basin, ed. Dorcey, A.H.J, and J.R. Griggs. Westwater Research Centre. U.B.C. 1991. Hall, K.J.; H. Schreier and S.J. Brown. 1991b. "The Land and Water Interface in the Fraser River Basin". "In" Water in Sustainable Development: Exploring our Common Future in the Fraser River Basin, ed. Dorcey, A.H.J, and J.R. Griggs. Westwater Research Centre. U.B.C. 1991. Halstead J.M.; B.E. Lindsay and C M . Brown. 1991. "Use of the Tobit Model in Contingent Valuation: Experimental Evidence from the Pemigewasset Wilderness Area". Journal of Environmental Management. 33: 79-89. Hanemann, W. 1984. "Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses", American Journal of Agricultural Economics. 66: 332-341. Hauser, A.; G.C. van Kooten, and L. Cain. 1993. "Water Quality and the Abbotsford Aquifer: Overviews and Cost-Benefit Analysis of Livestock Waste Disposal Alternatives using Contingent Valuation Method". Dept. of AgEc. U.B.C. Hayward, K.A. 1985. "An Evaluation of Goal Hierarchies for Agricultural Producers". Dept. of AgEc. MSc Thesis. U.B.C. International Reference Group on Lakes Pollution From Land Use Activities (PLUARG). 1978. Environmental Management Strategy for the Great Lakes System: Final Report to the International Joint Commission. International Joint Commission: Windsor.  94  Johnson, R.L.; N.S. Bregenzer and B. Shelby. 1988. "Contingent Valuation Question Formats: Dichotomous Choice Versus OpenEnded Responses". "In" Economic Valuation of Natural Resources, ed. R.L. Johnson and G.V. Johnson. Westview Press: Boulder, San Francisco & Oxford. Kahneman, D. 1984. "Comments on the Review Panel's Assessment". "In" Valuing Environmental Goods: A State of the Arts Assessment the Contingent Valuation Method. R.G. Cummings, D.S. Brookshire, D.L. Coursey, W.D. Schulze. Washington, D.C.: U.S. Environmental Protection Agency. (1984):226-235. Kalton, G. 1983. Introduction to Survey Sampling. Beverly Hills: Sage Publications. Kanninen, B.J. 1993a. "Optimal Experimental Design for DoubleBounded Dichotomous Choice Valuation", Land Economics. 62(2):138-142. Kanninen, B.J. 1993b. "Dichotomous Choice Contingent Valuation", Land Economics. 69:138-146. Kealy, M.J.; J.F. Dovidio and M.J. Rocket. 1988. "Accuracy in Valuation is a Matter of Degree", Land Economics. 64(2):158171. Kealy, M.J. and R.W. Turner. 1993. "A Test of the Equality of Closed-Ended and Open-Ended Contingent Valuations", American Journal of Agricultural Economics. 75:321-338. Kowalenko, C.G. and R.J. van Kleeck. 1987. "An Evaluation of Nitrogen use in B.C. Agriculture", Dept. of Agriculture Canada. Plant Research Institute. Technical Bulletin 19873E. Larsen, D.M. 1992. "Can Nonuse Value be Measured from Observable Behaviour?", American Journal of Agricultural Economics. 74:1114-1120. Lazo, J.K.; W.D. Schulze, G.H. McClelland and J.K. Doyle. 1992. "Can Contingent Valuation Measure Nonuse Values?", American Journal of Agricultural Economics. 74:1126-1132. Marwell, G. and R.E. Ames. 1975. "Experiments on the provision of public goods I: Resources, interest group size, and the free rider problem", American Journal of Sociology. 84:13351360. Marwell, G. and R.E. Ames. 1980. "Experiments on the provision of public goods II. Provision points, stakes, experience, and the free-rider problem", American Journal of Sociology. 85:926-937.  95  McConnell, H.D. 1991. "Planning for Metropolitan Liquid Waste Management in the GVRD: An Innovative Approach Towards Sustainable Development". "In" Perspective on Sustainable Development in Water Management; Towards Agreement in the Fraser River Basin, ed. Dorcey, A.H.J. Westwater Research Centre. U.B.C. 1991. Mitchell, R.C. and R.T. Carson. 1989. Using Surveys to Value Public Goods: The Contingent Valuation Method. Washington: Resources for the Future. Musser, W.N.; L.M. Musser, A.S. Laughland, and J.S. Shortle. 1992. "Contingent Valuation and Averting Costs Estimates of Benefits for Public Water Decisions in a Small Community." Agricultural Economics and Rural Sociology Dept. College of Agricultural Science. The Pennsylvania State University. Schalger, E. and E. Ostron. 1992. Property Rights Regimes and Natural Resources: A Conceptual Analysis", Land Economics. 3:249-62. Scherr, B.A. and E.M. Babb. 1975. "Pricing Public Goods: An Experiment with two proposed pricing systems", Public Choice. 23:35-48. Scott, A. 1991. "British Columbia's Water Rights: Their impact on the Sustainable Development of the Fraser Basin". "In" Perspective on Sustainable Development in Water Management: Towards Agreement in the Fraser River Basin, ed. Dorcey, A.H.J. Westwater Research Centre. U.B.C. 1991. Sellar, C ; J.R. Stoll and J. Chavas. 1985. "Validation of Empirical Measures of Welfare Changes: A Comparison of Nonmarket Techniques", Land Economics. 61(2): 157-175. Sellar, C ; J. Chavas and J.R. Stoll. 1986. "Specification of the Logit Model: The Case of Valuation of Nonmarket Goods", Journal of Environmental Economics and Management. 13: 382390. Sewall, K. 1990. The Tradeoff Between Cost and Risk in Hazardous Waste Management. New York & London: Garland Publishing, Inc. Smith, V.L. 1977. "The principle of unanimity and voluntary consent in social choice", Journal of Political Economy. 85:1125-1139. Spriggs, J.; G.C. van Kooten. 1983. "An Experience in Defining a Marketing Question", Canadian Journal of Agricultural Economics. 31:79-93. Statistics Canada. Farm Input Price Index. Catalogue no. 62-004. "In" various quarterly issues. 96  Stennes, B. 1992. "The Economics of Off-Farm Litter Disposal Options for South Coastal B.C. Poultry Farms". Stennes Consulting. A report under contract to the BCMAF (and Food) as a part of the Canada-B.C. Soil Conservation Program. Stonehouse, D.P. and Narayanan A.V.S. 1984. "The Contributions of Livestock Manure to Profitability and to Self-sufficiency in Plant Nutrients on Mixed Farms", Canadian Journal of Agricultural Economics. 32:201-211. Teofilo, 0.; K.Y. Jang and J.R. Stoll. 1993. "Testing for Misspecification in the Referendum Contingent Valuation Approach", American Journal of Agricultural Economics. 75:332-338. Torgerson, W.S. 1958. Theory and Methods of Scaling. New York. John Wiley & Sons. van Kleeck, R.J. 1992. B.C. Ministry of Agriculture and Fisheries. Soils and Engineering Branch. Waste Management Engineer. van Kooten, G . C ; R.A. Schoney, and K.A. Hayward. 1986. "An Alternative Approach to the Evaluation of Goal Hierarchies among Farmers", Western Journal of Agricultural Economics. 1:40-49. van Kooten, G.C. 1992. Land Resource Economics and Sustainable Development: An Introduction to the Resolution of Land Use Conflicts. Dept. of Agricultural Economics. U.B.C. Weisberg, H.F.; J.A. Krosnick and B.D. Bowen. 1989. An Introduction to Survey Research and Data Analysis. Second ed. Glenview, Illinois, Boston, and London: Scott, Foresman and Company. Willis, C.E. and J.H. Foster. 1983. "The hedonic approach: No panacea for valuing water quality changes", Journal of the Northeastern Agricultural Economics Council. 12:53-57.  97  APPENDIX 1 SURVEY; Water Quality, and Agricultural Waste Disposal in the Lower Mainland; Decisions for the Future Hello, is Mr. or Mrs. _ = = _ = _ _ in. (If yes: continue on. If no, then check to see if right number has been dialed. Then if number is correct, ask when is a better time to call, and call back at that time or day). Hello my name is . and I'm calling from the University of British Columbia. We are doing a study on water quality for the Abbotsford area, and would like a few minutes of your time. We are not selling anything. Your name was selected at random, and your answers will be strictly confidential. (If person hesitates or says no) : Would you prefer to have us call at another time or speak to someone else in your household? (If person does not want to answer questions, but thinks another person should, then ask if that person is available. If another time would be better—either for the person to whom you are speaking or for the alternative—then find out what time would be best. If person does not want to answer, and does not want you to call back, thank them for their time, and politely say good bye. Record new time and/or alternative person to call.) Once the person has agreed to answer questions: You receive your water from the Abbotsford Aquifer. correct?  Is that  (If no) : From the information we have, it seemed you lived in that area. Since this survey concerns only those people, these questions would not be appropriate for you. But I do want to thank you for your time, anyway. Good bye, now. (Etc.) (If yes, continue with survey).  98  Section 1:  Opinion Questions  Please indicate your agreement or disagreement with respect to each of the following statements. Choices are strongly agree (SA), weakly agree (WA), neutral (N) , strongly disagree (SD), weakly disagree (WD), or no opinion (0). a)  You are familiar with the issues of nitrates in water and the effects of nitrates on health. (If respondents asks about the issue, give then brief unbiased information about the water situation in their area. Then repeat question (a)).  b)  Smell and other farm nuisances are a problem.  c)  Water quality in your area is adequate.  c)  Government needs to impose strong regulations on fertilizer use and handling of livestock wastes, regardless of cost to farmers.  Section 2:  Background Information  (If respondents hesitates or refuses to answer any personal question, once again ensure anonymity and the need for such answers for the study? If still hesitates or says no, gone to the next question) 1.  (a)  How long have you lived in this area?  years  (b)  How long have you lived in your current residence? years  Do you rent or own your current place of residence?. What is your monthly rent or mortgage payment taxes)? _ less than $500 _ $750 to $1,000 $1,250 to $1,500 $1,750 to $2,000 more than $2,250  $500 to $750 $1,000 to $1,250 $1,500 to $1,750 $2,000 to $2,2500  99  (including  Section 3: 1.  Sewage and Water Quality  What type of sewer system do you have? city sewer  (Please check one)  septic tank  If you have a septic tank, when was it last cleaned? one) within last 4 yrs more than 10 yrs ago don't know  don't know (Please check  within 5-10 yrs never  2. (a) Do you use any special filters in your household to improve water quality? (b)  If yes, how much do you spend on this system per week? $  3. (a) Have you purchased any bottled water in the last five years? (b)  If yes, how much do you spend on bottled water pfer week?  (c) If yes to 3(a) or 4(a), why do you purchase bottled water or use a water filter? (For example pollution-free, tastes better, convenient, cold water, looks clearer, etc.)  Scientists have identified composting as a possible solution to the animal waste disposal problem. For various levels of nitrogennitrate concentrations, we have made some rough calculations of the probable costs (after revenues) of cleaning up the pollution via composting. Are you prepared to pay the amounts indicated to clean up farm animal wastes, assuming charges would show up either on your annual water bill or through an increase in rent?  Approximate current nitrate level: 12 mg/L Current government standard: 10 mg/L Future government target: 1 or less mg/L  100  a) Are you willing to pay ($28, $85, $200) n to get to government standard level? If yes, how much more?  $  If no, what is the most you would be willing to pay? $ b) Are you willing to pay ($57, $85, $113) to get to 6 - 4 mg/L, which is one half of current level? If yes, how much more?  $  If no, what is the least you would be willing to pay?$ c) Are you wiling to pay ($142, $170, $200) to get to future target level? If yes, how much more?  $  If no, what is the least you would be willing to pay?$ d) Besides the boil order, do you normally boil your water?  12  Choose one of the following amounts for each respondent when asking questions (a) , (b) or (c) . Try to get the same number of responses for each category. 101  Section 4: 1.  a)  b)  Personal Information  What is your age?  (Please check one)  25 or under  26-35  36-45  46-55  56-65  over 65  Are you:  Male  Female  2. a) Including yourself, how many individuals are there in your household? b) If you have any children under the age of 5 in your household, how many? children under 5 years of age 3.  What is your level of education?  Secondary (Grade):  (Please circle)  8  9  10  Post Secondary (Years): 1  2  3  11  12  5  6  7  5. What was your family's (household's) approximate gross (before tax) income in 1992? (If a farm, income after farm expenses but before personal expenses.) less than $30,000 $30,001 to $40,000 $40,001 to $50,000 $50,001 to $60,000 $60,001 to $70,000 $70,001 to $80,000 $80,001 to $90,000 $90,001 to $100,000 more than $100,000  102  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items