摘要：The Forest Service controls vast quantities of natural resources including timber, wildlife, watersheds, air sheds, and ecosystems. For many of these resources, recreation is one of the primary uses of the natural asset. Recreation visits taken to National Forests are not "purchased" in the same type of market as other goods (e.g., timber, grazing, or housing). The price of, and ultimately benefit received from, recreation to National Forests cannot be estimated via traditional market prices and quantities. Alternate methods must be employed to estimate the value of recreation access. We use on-site survey data from the Forest Service's National Visitor Use Monitoring database (2000-2003) and stated preference demand estimation methods to model annual recreation trip-taking behavior to National Forests. We then use these models to derive estimates of per-visit net economic benefits across regions and activities. In 2000, the FS began conducting systematic research into recreation visitation levels on National Forest lands under the National Visitor Use Monitoring Project (NVUM). From 2000 to 2003 NVUM has collected data from 120 National Forests providing information on the number of annual visits, primary activity, local area expenditures, satisfaction with facilities, and limited demographic information. These data were collected using an on-site stratified random sampling scheme resulting in over 90,000 completed surveys. Using the NVUM data we estimate the net economic value (NEV) of recreation on National Forest lands. The dataset used to estimate these values contains 73,655 observations. Using a truncated negative binomial estimator, weighted by a composite factor that adjusts for the stratified, on-site nature of the data, we have estimated a series of pooled, multi-site recreation demand models and calculated net economic values for recreational visits to the National Forests for each of fourteen activities and four RPA regions (Pacific, Rocky Mountain, Northern, and Southern) on a per visit per individual value and for a per activity day per individual basis. Our results indicate that for most models and specifications, adjusting for the choice based sampling frame by using a truncated, weighted, stratified, negative binomial estimator, as well as accounting for regional and activity differences, reduces the estimate of the average per day and per activity day values. Forest managers and others involved in managing, planning, and administering resources used for recreation often need an estimate of the economic value of the resource. For many of these resources non-market analysis must be used to generate this information. For forest recreation, many of the values currently available come from secondary sources or from small samples. The values estimated using NVUM represent an improvement over many of the currently available forest recreation values because of the unique nature of the large-scale, stratified random sample.
