188.8.131.52. Travel Cost
The travel cost method uses observable data on recreation visitation to infer the recreation-related use values of environmental amenities. If it works, changes in environmental amenities can be valued providing the benefit side of a cost-benefit calculation.
The method dates from a 1949 suggestion by Harold Hotelling that travel cost indicates a recreational consumer’s value of visiting a site. From this simple observation, an extensive literature has developed. The studies vary substantially but typically they try to estimate a demand function for site visits that includes as an arguments travel cost and some measure of environmental quality. Area under the demand curve is a dollar denominated measure of consumer surplus. When the environmental quality variable is altered, a different demand curve and surplus result and the difference in the surpluses is deemed the value of the change in the environmental attribute or the willingness to pay for the change.
The EPA-funded literature on this topic includes V. Kerry Smith, Edward R. Morey, Robert D. Rowe, AERE Workshop on Recreation Demand Modeling, June 20, 1985 (EE-0137). The workshop was organized around three themes: the effect of site attributes on demand; modeling demands for recreational sites within a region; and modeling non-user values. Several of the papers employed or reviewed the travel cost approach:
Sharon M. Oster's EPA-funded dissertation, The Incidence of Local Water Pollution Abatement Expenditures: A Case Study of the Merrimack River Basin, June 1974 (EE-0345) contains a contingent valuation survey and discusses the travel cost as an alternative method as a means for estimating benefits but does not attempt any valuation using travel cost.
The 1981 study Measuring the Benefits of Water Pollution Abatement (EE-0178) by Mills and Feenburg includes a discussion justifying the use of economics to investigate benefits. Next, it reviews the three methods typically used to estimate benefits of water pollution abatement: surveys or contingent valuation studies about which the authors expressing some methodological reservations, the travel cost method which is their favored approach and property value studies which they regard as most appropriate for air pollution benefits studies. The report features an extensive econometric investigation and application, one of the earlier travel cost method studies. It also includes some discussion of drinking water and other withdrawal uses. National benefits are also touched upon.
Ivar Strand, K.E. McConnell, Nancy Bockstael, David G. Swartz, Marine Recreational Fishing in the Middle and South Atlantic: A Descriptive Study, August 1991 (EE-0036A), describes data from three surveys on sport fishing that are used in a companion report (EE-0036B) to estimate the benefits of marine sport fishing in the Middle and South Atlantic states and the reduction in that economic value caused by pollution of marine waters. Two of the surveys (one by phone and the other by “intercept” at the site) are part of the Marine Recreational Finfish Statistical Survey sponsored by the National Marine Fisheries Service (NMFS). Those surveys were aimed at estimating total recreational catch. The third survey was conducted by the investigators as part of this project and was designed to generate the information necessary for estimation of economic models.
The companion 1994 report, The Economic Value of Mid and South Atlantic Sportfishing (EE0036B) by McConnell and Strand, uses both contingent valuation and travel cost models to estimate the economic value of marine recreational fishing in the Middle and South Atlantic states, from New York to Florida. There are two values of interest: one is access value, what anglers will pay to have access to the resource; the second is the value of changes in the quality of fishing, what anglers would pay for characteristics of the experience such as catch rate. Pollution and its impact are assessed through changes in the catch rate.
William H. Desvousges, V. Kerry Smith, Matthew P. McGivney, A Comparison of Alternative Approaches for Estimating Recreation and Related Benefits of Water Quality Improvement, March 1, 1983 (EE-0017), is a report on several studies undertaken by the authors to value the benefits of improving water quality. The results of the studies are also compared. The studies use different methods: travel cost, contingent ranking and several varieties of contingent valuation. An effort is made to determine both use value, primarily recreational, and non-use values including option and existence values. The techniques were exercised on data for the Monongahela River in western Pennsylvania. As examples of the results of the studies, the four contingent valuation approaches yielded, for mean annual per household benefits, a range of $4.21 to $30.88 compared with $7.01 for the travel cost method to upgrade water quality from boatable to fishable. As to comparability of results, the authors state, “… the benefit measurement approaches … show consistent results for comparable changes in water quality. Indeed, the range of variation is generally less than that expected in models used to translate the effects of effluents in a water body into the corresponding water quality parameters.”
Cameron has contributed a number of travel cost studies related to marine recreational fishing, including The Effects of Variations in Gamefish Abundance on Texas Recreational Fishing Demand, June 9, 1989 (EE-0094);Trudy Ann Cameron, Combining Contingent Valuation and Travel Cost Data for the Valuation of Non-market Goods, April 21, 1989 (EE-0001); and Using the Basic “Auto-validation” Model to Assess the Effect of Environmental Quality on Texas Recreational Fishing Demand: Welfare Estimates 1988 (EE0065), which uses a subset of the data collected in an earlier study by the author of recreational fishing along the Texas coast (EE0001). For 506 of the observations of the earlier interview of 3366 anglers, respondents provided information on their ex-post subjective assessment of environmental quality at the site. The author found that improving environmental quality from one standard deviation below the mean to one standard deviation above the mean increases the value of the fishery by about $1,400 (about 50%) per angler.
The 1985 report A Methodological Approach to an Economic Analysis of the Beneficial Outcomes of Water Quality Improvements from Sewage Treatment Plants Upgrading and Combined Sewer Overflow Controls (EE0047) by Meta Systems uses a variety of economic techniques, including travel cost, to estimate the benefits of water quality improvements in Boston Harbor that would result form upgrading two sewage treatments plants (Deer Island and Nut Island) and controlling combined sewer overflows (CSOs) in Dorchester Bay, the Neponset and Charles Rivers, Quincy Bay and the Inner Harbor.
Recreational Benefits Transfer Project (EE0151) by Smith and Kaoru (1989) reviews all published and unpublished travel cost recreational demand models done between 1970 and 1986. From this literature a subset of 77 studies are summarized and subject to a meta-analysis. Some of their findings may be highlighted: (1) virtually all estimates of consumer surplus from recreatoin activities are biased because they involve nonlinear transformations of demand parameters; (2) it is possible to summarize into fairly narrow ranges the consumer surplus per unit of use and the price elasticity of demand for recreation sites; and (3) benefits transfer will be facilitated if researchers report a small set of summary statistics in a consistent fashion.
A 1993 report by Jones and Sung, entitled Valuation of Environmental Quality at Michigan Recreational Fishing Sites: Methodological Issues and Policy Applications (EE0289), presents a fishing travel cost study using a random utility model that captures the demand for trips of different durations. The environmental variables included are the catch rate representing the stock of fish and water contamination variables. “First, we calculate the damages to Michigan-licensed recreational anglers from fish kills due to operation of the largest pumped-storage plant in the US. Second, we calculate the benefits of cleaning up PCB contamination in a river in Michigan, which would allow the State to remove dams currently containing contaminated sediments and to open a substantial reach of the river for anadromous runs. The contamination at this site is sufficient to merit designation of the site as an Area of Concern.”
The investigators estimate that closure of the Ludington pumped storage plant would increase anglers’ consumer surplus by increasing the fish stock by $770 thousand annually (1991$). Also, they estimate that cleaning up the PCB contamination of the Kalamazoo River would increase anglers’ consumer surplus by $1.15 million annually.
The NSF/EPA-funded study from 1996 entitled “Updating Prior Methods for Non-Market Valuation: A Bayesian Approach to Combining Disparate Sources of Environmental Values” and conducted by Herriges and Kling will utilize a Bayesian paradigm (1) to integrate contingent valuation and travel cost data and (2) for benefits transfer, i.e., applying the value of a studied resource to an unstudied resource. The integration method will be applied to estimate the value of wetland restoration in the State of Iowa. Assuming success with the approach, that valuation could then be projected to other wetlands using the benefits transfer method.
The NSF/EPA-funded study from 1991, “Valuing Environmental Quality Changes” by Larson is designed to explore some promising new approaches for valuing environmental quality changes from market-related and contingent valuation approaches. An important part of the project is the collection of a set of data on the recreational use of an environmental resource. It will provide a valuable opportunity to investigate the empirical significance and implications of the alternative approaches for valuing quality changes. Several recent theoretical innovations will be implemented: non-parametric testing of whether a single preference function can explain the observed responses, and if not, how many are; combining travel cost information, actual behavior and contingent data on how trips taken may change as quality changes; ascertaining the relationship between minimum expenditure changes with quality; and demonstrating a method of obtaining changes in expenditure with quality changes from market data alone under two-stage budgeting.
Alan Randall argues convincingly in a 1994 article in Land Economics that the travel cost method is seriously limited. He notes that travel cost is not an observed datum but rather is inferred from observation. The method of inference is essentially arbitrary. He notes among other concerns that the opportunity cost of travel time is an 'empirical mystery'; that levels of recreation related durable equipment vary and the allocation of their cost to any particular trip is arbitrary; lodging and subsistence expenditures have a large discretionary component; multi-site and multi-purpose trips have costs allocated arbitrarily; the distance from home to recreation site is exogenously treated but there may be feedback. In sum, travel costs assigned to particular trips depend largely on the investigator’s decisions. Travel costs are ordinal only. Random utility models and household production functions, employed by some researchers, do not fully solve this problem.
Because consumer surplus is dependent on the investigator’s decisions and hence ordinal only, ie, it is only determined up to a monotonic transformation. Inter-personal aggregation is therefore problematic. Different cost allocations and different functional forms would yield different results. A vital link in the chain leading to environmental benefit is missing. Consequently, aggregate benefit estimates are to some extent arbitrary
Randall's criticism suggests that recreation sites are improperly valued with the travel cost method. The valuations for some or all sites may be too high or too low or some may be high and others low because of the way costs are computed and allocated. The estimation of the environmental amenity component is confounded by the general valuation error. If it turns out that all site valuation estimates are inflated more or less proportionately, the difference between relatively clean and dirty sites -- the value of the environmental amenity -- may also be inflated. The reverse might hold if all valuations are too low. If some are high and some low then the direction of the error is not obvious.
It is an area for potential further research to attempt to quantify the problem. That is, how sensitive are estimates to the choices made by investigators? What corrections are possible? What sensitivity analyses must be made by investigators to bound their results?
You can see the full list of reports corresponding to this section in the benefits analysis - valuation - revealed preference - travel cost subview of the subject view of the Environmental Economics Report Manager (EERM) database and the full list of corresponding NSF/EPA projects in the benefits analysis - valuation - revealed preference - travel cost subview of the subject view of the NSF/EPA Funding for Environmental Economics (FfEE) database.
- "The Logit Model and Exact Expected Consumer’s Surplus Measures: Valuing Marine Recreational Fishing" by Morey and Rowe develops a random utility model employing the travel cost method. The authors are able to estimate for a variety of Oregon fishing sites what the option value, the right to fish there, is worth to visitors from various counties in Oregon.
- "The Varying Parameter: In Perspective" by William H. Desvousges begins with a comprehensive review of the travel cost literature. The author introduces the varying parameter version of the travel cost method and then shows its application to a study of 22
- "Valuing Quality Changes In Recreational Resources" by Elizabeth A. Wilman derives, from a general consumer utility maximization model, several resource valuation models including travel cost and hedonic models. The various models are reviewed and considered. A travel cost hedonic model is estimated and applied to the valuation of quality changes in deer hunting sites in the Black Hills National Forest of South Dakota.
- "Modeling the Demand for Outdoor Recreation" by Robert Mendelsohn is a review of three techniques for valuing nonmarketed attributes of recreational assets. According to the author, "Partitioning involves grouping sites into small homogeneous sets and treating each set as a unique good and is best represented by multiple site travel cost models. Hedonics involves disaggregating goods into their component characteristics and modelling the prices and demands for the characteristics and is best represented in the recreation context by the hedonic travel cost method. The index models involve measuring the demand for a standard good and explaining variations in that demand across goods explicitly in terms of observable characteristics. Both generalized travel cost and discrete choice models are members of this last approach."
- "A Model To Estimate the Economic Impacts on Recreational Fishing in the Adirondacks From Current Levels Acidification" by Daniel M. Violette uses a site-characteristics-based travel cost model and a simple participation model to estimate the use values of fishing in the Adirondacks, an area of New York State containing approximately 3,000 lakes. Data came from the New York Anglers’ Survey data set. Site characteristic data came from the Adirondack Lake and Pond Survey. Where possible, the approach was designed to provide upper-bound estimates. Nonetheless, the resulting damage estimates are quite modest; damage due to current levels of acidification could range from $1 to $12 million annually.
- "Modeling Recreational Demand in a Multiple Site Framework" by Bockstael, Hanemann and Kling distinguishes between single and multiple site travel models and concentrates on the latter. They can be subdivided further. The subject paper reviews the substantial to determine how well they account for the benefit changes in a multiple site framework. Several estimation techniques are demonstrated from a common data set.