Valuation of Forest Amenities: A Macro Approach

A method of estimating forest amenity value based on macroeconomic growth theory is presented. It relies on the assumption that more valuable forest amenities are provided by a forest with a more natural stand structure. We construct a forest naturalness index from stand data that provides a relative measure of the forest amenity provided regionally. This naturalness index is meant to assess the change in forest amenities over time. It is a measure of diversity within the allowable specimens (species/diameter/height) of the natural forest type and observed successional stage. We aggregate the fraction of maximum diversity of the natural forest type for the target successional profiles of the natural forest types of the survey unit to calculate the index for the region. The paper shows a specific form for the path over which aggregate consumption, including forest market goods, and forest amenities, as measured by our naturalness index, evolve. This macroeconomic growth model assumes a trade-off between consumption growth and forest amenities. The present value of future consumption foregone by a marginal increase in the naturalness index is the shadow price of the forest amenity represented by the index increase. The value of any proposed forest policy change is the value of the index change that the new forest policy causes. 1 Authors are respectively, a Research Assistant and Professor in the Department of Forest Ecology and Management, University of Wisconsin , 1630 Linden Drive, Madison, WI 53706. The USDA Forest Service supported the research. Additional support came from the School of Natural Resources, University of Wisconsin, Madison. Environmental Change and Economic Growth Optimal management of the many values supported by a forest resource requires an assessment of a price of each of those values. Whether we are trading off multiple uses of public forests or establishing the optimal level of funding to facilitate some value of the forest, we need prices to analyze the policy choices. This need for prices is also true beyond public lands to the extent that we do establish policies that influence the total forest resources of a region including those on privately held lands. Whether a rational optimization calculation of the best forest policy is formally made or not, we do observe changes in the state of forests which result from policy actions or lack thereof. The overarching concept behind this research is that the state of the forest we observe implies prices for forest values even if they have not been explicitly identified. Prices for many goods provided by the forest are as conceptually straightforward as observing market transactions. Timber prices, grazing rights and many other non-timber commodities are bought and sold so the prices observed in those transactions are the prices that we use in benefit cost analysis. But, many values provided by the forest are not traded in markets. Forests provide environmental services, which are public goods, so due to the free riding problem, market information provides a distorted price. The forest also provides amenities, such as scenic beauty and the value of simply existing that are difficult if not impossible to reduce to a market commodity. Methods have been developed to estimate the value of these non-market goods. With the travel cost method and hedonic pricing method we observe choices of individuals and infer the price individuals are willing to pay for access to amenities. With the contingent valuation method, we experimentally construct markets by polling individuals with questions that either directly or indirectly relate to the value of amenities. One aspect common to each of these valuation methods is that we imply the price by observing the behavior of individuals. Our approach in this project is instead to observe aggregate behavior in a region and infer an aggregate value for forest amenities. In this macro approach there is a trade off between consumption growth and the state of the forest. The present value of consumption goods a region gives up to prevent a marginal degradation in forest naturalness is the price the region effectively puts on that change in forest state. We model this trade-off using an approach based on economic growth theory. This approach has been proposed by Stokey (1998) to explain the Environmental Kuznets Curve (EKC) hypothesis. The EKC hypothesis simply stated is the proposition that for a wide class of environmental amenities, those amenities will at first worsen as per capita national income increases from a low level and then after a peak, the environmental amenity will improve with further increases in income. Forest landcover has been subject to limited study in the EKC empirical framework. Shafik and Bandyopadhyay (1992) found that both net change in forest cover and annual deforestation did not significantly relate to income in an international set of 149 countries between 1961 and 1986. Panayotou (1993) using strictly crosssectional international data reports a turning point in deforestation at a GDP per capita of $1275 in 1985 prices. Cropper and Griffiths (1994) create pooled time series cross-section data for three separate regions of the world. Per capita national income is a significant factor in both Africa and Latin America, but not in Asia. Stern, Common, and Barbier (1996) use the published EKC deforestation data to predict that forest loss stabilizes overall by 2025, but tropical deforestation continues at a constant rate. Finally, Patel, Thomas, and Jaeger (1995) find that the pattern of fuelwood demand in Kenya will eventually cause the decreasing tree cover to increase. Koop and Tole (1999) conclude that the same simple forest cover EKC relationship does not hold without accounting for other international differences. The applicability of these papers to the issue of forest amenities is questionable, since forest cover itself may not be a sufficient measure of EKC type environmental amenities provided by forestland. Measuring Forest Naturalness We will calculate a measure of forest naturalness from FIA survey data using a clearly defined procedure that assesses an overall grade for the forests in a region. In theory, the proposed model can be extended to a multiple dimension measure of the capacity of forests to provide amenities. But, we limit the scope of our current work in this project to a one-dimension naturalness score. Although a consensus has not yet developed on a complete definition of naturalness much less on the relative weightings of different natural aspects of the forest we wish to express as a single score, such a measure is useful for our purposes. If all of the natural amenities provided by the forest increase monotonically with our naturalness index, then a linear approximation is appropriate to compare relatively small differences in naturalness score. In this case, the marginal tradeoffs between the different amenities of interest are relatively constant, so the naturalness measure will faithfully reflect forest amenities averaged together by the naturalness index calculation procedure. We expect that a significant part of the attributes of a forest that make it natural correlate with our measure even though we do not measure many of them directly. To the extent that our measure does not correlate at all with some forest attributes, we have simply defined those attributes as not natural in our sense of naturalness. The objective in defining a naturalness index is not to construct a perfect measure of naturalness, but rather to define one that captures enough natural qualities to proxy for a useful part of the amenities provided by forests. If some single easily defined stand state were the natural condition, then calculation of a score based on the deviation from that stand state would be simple. The most difficult aspect of assessing naturalness is that many states are perfectly consistent with a natural forest in the original-naturalness sense of natural described by Peterken (1996). Great diversity is characteristic of the natural forest including large areas dominated by a few early successional species in some forest types. The natural forest includes recently disturbed sites and areas of forest succeeding that disturbance ranging to a mature forest. This variety in the forest is observed at the stand level and at the landscape level. In our approach we define adherence to a natural disturbance regime and diversity of natural tree types at both the stand and landscape level to be the most natural forest. A rich variety of approaches are taken by past efforts to define naturalness or related qualities of forests. The “health” of a forested landscape is defined by the condition of a multitude of natural attributes of the forest. The Santiago Declaration (Montreal Process, 1995) identifies seven criteria for evaluating the state of conservation and sustainable management of temperate and boreal forests. Multiple indicators are identified for each of the Santiago criteria. For example, criterion 1 is the conservation of biological diversity, which consists of ecosystem diversity, species diversity and genetic diversity. Ecosystem diversity is indicated by a) the area of each forest type, b) the area of forests in age classes and successional stages, c) the area of critical forest types, d) the area of critical forest age classes and successional stages, and e) the fragmentation of forest types. Similarly, The State of the Nation’s Ecosystems (H. John Heinz III Center, 1999) provides a multidimensional assessment of forests. Some efforts to define forest naturalness use somewhat subjectively evaluated scales. Peterken (1996) defines a “scale of naturalness” by describing eight categories from a treeless pasture to virgin forest with increasing degree of wilderness. Schnitzler and Borlea (1998) define a “scale of increasing naturalness” based mostly on stand characteristics plus some landscape structure information. This scale of naturalness is along six dimensions: four at the landscape level and two at the stand level. Numerical criteria are described for placing a forest in a ranked naturalness category for each dimension. We have designed our naturalness index to measure these criteria as directly as we can while restricting ourselves to data collected by FIA surveys. The first component of our naturalness index is stand level diversity of tree types. By tree type, we mean not only species, but also diameter, height, cavity tree status, and crown class. For each plot we assess the diversity weighted by basal area using a Shannon index normalized to a 0 to 1 scale. The tree types must be allowed in the potential natural type and successional stage of the forest to be counted. To the extent that a non-native tree type (e.g. buckthorn is a noted invasive tree species in Southern Wisconsin.) displaces native trees, the naturalness is lowered. This is true for the diversity measure defined below because the partial derivative with respect to the basal area of each allowed tree type is negative when we hold everything else constant except replacing the tree type with an unnatural type. The list of tree types allowed in the forest is a key element in our definition of natural. These tree types are defined specific to both the potential natural forest (PNF) type and the succesional stage we identify for a survey plot. We identify the PNF of a plot from the coordinates and other characteristics of the plot. From the longitude and latitude of the plot, we identify the Omernik Ecoregion potential forest type (USGS, 1990). We further differentiate forest type based upon site physiography, slope and aspect. For the identified PNF, we identify the successional stage from the stand data. For example a site with few trees large enough to be reported serves as a recently disturbed site. A stand dominated by southern pine in some PNFs might serve as an early successional stage and dominance by hardwoods might constitute late successional or a mature forest stage. We calculate of average plot diversity, Sij, for a PNF type i in the survey unit that is in successional stage j: ( ) ( ) ( ) . ln ln 1 ∑ ∑  