Researchers have conducted empirical studies aimed at incorporating forest resources into the national accounts of more than 20 countries since the late 1980s. We obtained and reviewed more than 30 studies. We define a study as a discrete research project. A single "study" might generate several publications, as indeed several have.
Appendix 7 provides details of the studies. In this chapter, we highlight the main characteristics of the studies and what they tell us about the feasibility of improved accounting for the economic benefits of forest resources. We take as points of reference the accounting framework presented in Chapter 2 and the valuation principles presented in Chapter 3. Country and region names in bold-faced type indicate headings in the appendix, to make it easier to look up additional information on the studies.
Our review is more focused on forest resources than some previous ones (e.g., Sheng 1995, Hamilton and Lutz 1996), and more comprehensive than others (e.g., Hartwick 1993, Newson and Gie 1996). Nevertheless, we encourage readers to consult those and other reviews.
General characteristics
The earliest study in our sample is the well-known study of Indonesia by the World Resources Institute (WRI; Repetto et al. 1989). This is the most frequently cited study on natural resources (including forest resources) and the national accounts. A study for the World Bank by Peskin (1989), which commented on the WRI study and contained additional estimates for Tanzania, came next. The number of studies increased more or less steadily during the 1990s, with the greatest number, nearly a dozen, being published or otherwise made available in 1996. We are aware of at least three studies having been published so far in 1997. Several more are in preparation under the auspices of the U.N. Statistical Division.
The greatest number of studies pertains to countries in Asia, and the least to countries in Africa. By region, the countries covered are:
Asia (8 studies): Indonesia, China, Malaysia (2), Nepal, Philippines (2), Thailand
Central and South America (5 studies): Chile, Costa Rica (2), Ecuador, Venezuela
Europe (5 studies): Austria, Finland (2), Sweden (2)
North America (4 studies): Canada (2), Mexico, U.S.
Oceania (4 studies): Australia (2), New Zealand, Papua New Guinea
Africa (2 studies): Tanzania, Zimbabwe
Numbers in parentheses indicate the number of studies in individual countries where more than one study was conducted. We considered as separate studies ones conducted by different research teams, as well as ones conducted by the same team at different points in time but which differed substantially in their approach or coverage.
In addition to the 28 studies in the 21 countries listed above, the sample included one pan-Asian study (Asia) and one global study (Global). These expand the country coverage considerably. The latter, for example, included more than one hundred countries. Not surprisingly, these multi-country studies tend to include less detail than the country-specific studies. They also tend to assume many similarities across countries (e.g., that per-hectare values of forests for nontimber benefits are identical).
Most of the studies covered one or more resources in addition to forests. That is, they did not focus exclusively on forests. They were less forestry accounting studies than broader resource accounting studies. More than half covered subsoil assets (minerals) in addition to forests. Other non-forest resources included agricultural soils, rangeland, fisheries, and, in a few instances, air and water pollution. On the positive side, placing the analysis of forest resources in the context of studies with broader coverage of natural resources creates an opportunity to assess the importance of adjustments for forest resources relative to adjustments for other resources. More negatively, however, the goal of being comprehensive appears to have forced some researchers to sacrifice a certain amount of depth and breadth in analyzing forest resources. The forestry-specific studies tend to cover a broader range of aspects of forest resources (e.g., nontimber aspects as well as timber) and to marshall more data in making adjustments to the accounts.
The forestry-specific studies tend to be the more recent ones. In part because of this, they tend to be the ones covering the longest time periods. The Canada (I) and (II) studies analyzed the longest period, from the early 1960s to the early 1990s. A few studies, generally (but not solely) those that addressed a broad range of resources, analyzed just a single year. Most studies, whether forestry-specific or more comprehensive, covered at least 1 decade.
An implication of these points is that experience with the nuances of forestry accounting is more fresh, and thus more in need of review and discussion, than the total number of studies and their distribution over time (going back nearly a decade) suggests. Unfortunately in this regard, very few reports on the studies were published in academic (refereed) journals. Most were "published" as chapters in conference proceedings, as working papers, as part of organizations’ own report series, or as draft manuscripts. This has three implications. First, identifying and obtaining the reports is difficult. Hence, it is not surprising that later studies do not always appear to draw fully on the results of earlier studies, even when the studies occur in the same country. This also leads us to conclude that Appendix 7 and the "Literature cited" section of this report might be of greater value than we originally expected them to be.
Second, the fact that most material has been published in outlets other than refereed journals means that it has probably been subjected to a less intensive review process. This raises the risk of inappropriate application of accounting procedures and valuation techniques. In a well-established field, this would not be a great concern. In the accounting field, however, there remains considerable controversy about the most appropriate ways of creating economic accounts for natural resources. To be sure, reports of some organizations, for example the World Bank and the World Resources Institute, do undergo internal (and in many cases) external reviews before publication. And the authors of papers presented at conferences and published in proceedings volumes do receive feedback from other conference participants. Nevertheless, in the absence of peer review, one should not be surprised if procedures in some studies are less sound than they could be.
Third, the infrequent publication of studies in the academic literature means that more theoretically-inclined researchers who read and publish predominantly in that literature are less likely to be aware of the issues that applied researchers face in attempting to create economic accounts for forest resources. For example, the long time period between forest regeneration and maturity of the timber crop might appear to theoreticians to be a minor detail that can be glossed over. As we have seen in Chapter 3, however, ignoring the rotation length can lead to severely biased estimates of the net accumulation of timber. Conversely, the fact that minimal dialogue between theorists and practitioners has occurred suggests that practitioners might not be taking full advantage of the implementable results of theoretical research.
The infrequency of publication in academic journals is not a consequence of the studies being conducted primarily by government agencies. In fact, more than half the studies were conducted by researchers with universities, research institutes, or international organizations. Only twelve of the studies appear to have involved government agencies responsible for the preparation of national accounts: Australia (I), Austria, Canada (I) and (II), Chile, Finland (I) and (II), Malaysia (I), Mexico, New Zealand, Philippines (I), and U.S.
Links to national accounts
Virtually all the studies included timber, and nearly half covered one or more nontimber aspects of forest resources. The list includes nonmarket production of fuelwood, berries, mushrooms, game, rattan, and peat; amenity values associated with protected areas and biodiversity; environmental services like watershed protection and soil protection; carbon sequestration; and acid deposition. The Tanzania study was the first to consider a nontimber value, fuelwood. The Sweden (I) study, and two modeled after it, Finland (II) and Sweden (II), covered the greatest range of nontimber values, more than half a dozen in each case.
The studies therefore contain examples of all the types of adjustments discussed in Chapter 2. In many cases, however, they do not indicate clearly the accounting framework guiding the adjustments they make. Most tend to be rather imprecise about adjustments to GDP vs. NDP. Some talk of calculating "green GDP," when the adjustments they make involve net accumulation and thus pertain to NDP. Generally speaking, the more ambitious the study in terms of the number of aspects of forest resources it covered, the more likely it was to be lax in terms of its conceptual framework, although the Sweden (I) study was quite careful in this regard.
These comments point to our principal conclusion: empirical efforts to incorporate forest resources into the national accounts must be guided by economic theory more than they have been. In the absence of a transparent and internally consistent conceptual framework, adjustments to the national accounts become ad hoc and unreliable. The only way to avoid this problem is to develop a framework based on sound economic principles. The framework in Chapter 2 represents our attempt to do this.
Some studies made adjustments to both current and asset accounts, while others made adjustments to just one or the other. Of those that made adjustments to current accounts, about two-thirds made adjustments for net accumulation. In most cases, the net accumulation estimates were for timber, although a few studies included nontimber values in calculations of net accumulation of forestland. Five studies estimated the value of nonmarket production of forest goods, which is an adjustment to GDP (and consequently to NDP, too).
One study, Finland (I), investigated a specific SNA recommendation, that timber growth in "cultivated" forests be valued and added to GDP. New Zealand already implements this recommendation, and has for some time. If this recommendation is implemented, then one must remember to exclude growth from estimates of net accumulation. The China study double-counted the value of timber growth in NDP by both adding it to GDP and using the net depletion method to calculate net accumulation.
About a third of the studies made adjustments related to asset accounts. In most cases, they simply calculated the value of the standing forest, without formally linking the estimates to the accounts. A few did, however, and compared the asset value of forest resources to the value of other assets in the economy.
With two exceptions, the studies all focused on adjustments at the national level. In addition to doing this, the Malaysia (I) and (II) studies also adjusted income accounts at the subnational level. They found that the impacts of net accumulation adjustments varied greatly within the country, with important implications for regional sustainability.
Valuation
Physical accounts
In most studies, making the adjustments was a two-stage process: first, the authors compiled physical information (e.g., area deforested, cubic meters harvested, etc.), and second, they converted physical impacts into monetary terms. The studies varied greatly in terms of the physical information they compiled and used. At one extreme were studies whose physical accounts consisted of just a single number. For example, in valuing the loss of habitat in Australia during 1980-89, Young (1993) assumed that the area of native forest cleared each year was 230,000 ha (Australia (I)). At the other extreme were studies that constructed elaborate GIS databases for analyzing changes in forest cover by forest type (Costa Rica (I)), or used detailed data on forest areas and growing stock by forest type, age class, and ownership category to calculate the forests’ asset value (Australia (II), Canada (II)).
Almost half the studies constructed aggregate growth/drain accounts for estimating the net depletion of timber. The prototype for this approach was the WRI study of Indonesia. That study applied the physical accounting identity,
Closing stock = Opening stock + Additions - Subtractions,
to standing timber volumes. Net depletion equals the difference between subtractions and additions. Additions include growth of existing forests and reforestation, while subtractions include harvest, fire and pest damage, and so on. Studies varied in their implementation of this approach. Some accounted for just timber stocks, while others accounted for forest areas as well.
Some grouped all forests together, while others distinguished between old-growth and second-growth forests or mature and immature timber stocks. A few distinguished forests by dominant species.
Variation in physical accounting approaches is unavoidable. Physical accounts need to be tailored to the particular aspects of forest resources being emphasized, which necessarily vary across countries. The tailoring must be done within the constraints imposed by data availability, which also vary. Hence, we do not advocate any single physical accounting scheme. What we emphasize is that creators of economic accounts for forest resources should have a clear idea of the uses of the accounts and the physical information necessary to create the accounts, and they should take full advantage of whatever data are available.
Net accumulation of timber
We break down the valuation of timber into two components: net accumulation, which is needed for adjusting NDP in the current accounts, and asset valuation, which is needed along with net accumulation to bring timber into the asset accounts. We deal with the former in this section and the latter in the next section.
Most studies used the net depletion method to estimate net accumulation. That is, they multiplied the negative of net depletion of the physical timber stock times net price. The popularity of this method appears to be due to its simplicity — it does not require information on the discount rate or the age of forests — and the example set by WRI’s Indonesia study. One study (Malaysia (I)) used smoothed values of net prices in applying this method, on the assumption that smoothed values, which exclude fluctuations due to short-run market events, provide a closer approximation to long-run user costs. Two studies (Nepal, Philippines (I)) intended to, or actually did, adjust net accumulation for "revaluation" (holding gains/losses) calculated by the residual method outlined in the WRI study: value of closing stock minus value of opening stock minus net accumulation as calculated by the net depletion method. In doing so they followed a suggestion by Peskin (1989). Three others (Finland (II), Sweden (I) and (II)) deducted current silvicultural expenditures (both fixed and variable) from their estimates of net accumulation.
No study applied the correct version of the net price method (the net price variation) presented in Chapter 3. Hence, all the estimates of net accumulation in existing studies based on the net depletion method are biased, with the direction of the bias not clear (because the studies do not provide information on the age class structure of the forests). Most likely, the estimates overstate the decrease in asset value in developing countries, and overstate the increase in developed countries (in view of the discussion about this bias in Chapter 3). Problems with the estimates are compounded by some of the additional adjustments made by certain studies. While holding gains/losses belong in net accumulation, the "revaluation" approach does not yield accurate estimates of holding gains/losses when closing and opening stocks are incorrectly valued by multiplying timber stocks times net price (Chapter 3). This is how the Nepal and Philippines (I) studies valued the stocks. Nor should one deduct silvicultural expenditures from net accumulation (Chapter 2).
A few studies applied other methods. Three (Asia, Malaysia (II), Mexico) applied El Serafy’s method, the first two in its generalized form (i.e., with the marginal cost elasticity not equal to infinity) and the third in the original form (marginal cost elasticity equals infinity). All applied the version for nonrenewable resources, not the correct versions for timber presented in Chapter 3. If timber depletion is primarily a matter of the draw-down of old-growth timber stocks — either because all forestland is converted to other uses following logging, or because second-growth forests have a "normal" age-class structure (i.e., equal areas in each age class up to the rotation age) — then net accumulation of timber is analogous to net accumulation of a nonrenewable resource. Under these circumstances, application of El Serafy’s method to the harvest of old-growth timber should yield accurate estimates of net accumulation, as long as the correct value of the marginal cost elasticity is used. The Malaysia (II) study used this rationale for applying El Serafy’s method, but its assumption that Malaysia’s second-growth forests are "normal" is heroic. The other two studies applied El Serafy’s method to all timber, not just old-growth. For these reasons, net accumulation estimates in studies applying El Serafy’s method are of dubious accuracy.
Two studies (Ecuador, Thailand) applied the replacement cost method, which they somewhat confusing referred to as the user-cost method. In effect, these studies calculated the present value of the hypothetical costs of establishing timber plantations to offset the volume (Thailand) or value (Ecuador) of harvestable timber in natural forests lost due to deforestation. There is no theoretical reason to expect this method to yield estimates of net accumulation bearing any relationship to true values, because the long-run costs of plantation timber could be either greater than or less than the long-run net benefits of timber production in natural forests. For this reason, the replacement cost method is not a sound method for estimating the net accumulation of timber. The Ecuador study attempted to deal with this problem by calculating replacement costs linked to value rather than volume of forgone timber, but it based its value estimates on the net depletion method, which as noted above yields biased estimates.
Two studies (Costa Rica (I), Philippines (I)) calculated net accumulation by applying the basic definition: net accumulation equals change in asset value. They calculated asset value as the present value of future flows of resource rents. The Costa Rica (I) and Philippines (I) studies probably provide the most accurate estimates of net accumulation of timber.
Four studies compared estimates from different methods. The Philippines (I) study compared the net price and present value methods. The former yielded estimates larger (in absolute value) than those of the latter. Assuming that the latter estimates are more accurate, the direction of the bias is what we would expect in a developing country (Chapter 3). The magnitude of the bias was enormous: estimates from the net price method exceeded those from the present value method by a factor of 30 times. The Mexico study compared the net price and El Serafy methods, finding that the former yielded estimates 15 times greater than the latter. Because both methods are problematic when applied to timber, interpreting this discrepancy is difficult. For similar reasons, it is difficult to interpret the comparison of the net price and replacement cost methods in the Ecuador and Thailand studies.
Asset value of timber
More than half the studies calculating the asset value of timber did so by multiplying the standing stock times the net price. As noted in Chapter 3, this does not yield accurate estimates. It understates the value of both mature forests and immature forests. Three studies attempted to avoid the bias for immature forests by multiplying net price times only the stock of mature timber, but, as immature forests do have a positive asset value, this compounds the downward bias.
Six studies (Global, Australia (I), Austria, Canada (II), Costa Rica (I), Philippines (I)) applied the theoretically correct present value method: they set asset value equal to the discounted sum of future rents. Two studies (Global, Canada (II)) assumed that forests were managed on an even-flow basis and that future annual flows of rents would therefore equal current flows. Hence, they simply divided current annual rent by the discount rate to estimate the asset value. (For unsustainably managed forests, the Global study capitalized the flow up to the projected year of complete depletion of timber stocks.) While this simplification is understandable for the Global study, whose objective was to calculate "ballpark" estimates for over one hundred countries, it is less understandable for the Canada (II) study, which apparently had access to detailed data on forests by age class. That study perhaps could have calculated more accurate estimates.
The remaining four studies took forest age into account in calculating asset values. The Austria study applied "age constants," which are implicit discount factors; the other three applied conventional discounting techniques. As in the case of net accumulation, the Costa Rica (I) and Philippines (I) studies, along with the Australia (II) study, applied the most appropriate procedures for estimating timber asset values. For countries that wish to incorporate forest resources in the asset accounts, we advise calculating asset values by using the present value method instead of by multiplying standing timber volume times net price.
Studies employing the present value method used discount rates ranging from 2 to 15 percent. The studies did not always make clear the rationale for the discount rates chosen. The benefit-cost analysis literature contains well-established procedures for selecting the discount rate (Boardman et al. 1996). As the choice of discount rate has a critical impact on estimates from the present value method, studies should make clear the criteria used in selecting the discount rate and should calculate asset values using a set of discount rates believed to span the range of plausible rates for the country in question.
At least three studies (Indonesia, Philippines (II), U.S.), and maybe more (details are not clear), calculated a residual "revaluation" term of the type described earlier, by subtracting opening asset value and net investment (value of change) from closing asset value. We have already given one reason why this procedure is flawed: multiplying standing timber volume times net price yields inaccurate estimates of the asset value of timber stocks, and therefore inaccurate residual estimates of holding gains/losses. A second reason has to do with the fact that gains/losses under this procedure come from changes in net price between the beginning and end of the accounting period. Such short-term fluctuations do not necessarily affect the value of timber when it is actually harvested several years (or decades) hence. Even a large increase in timber price during the current period has no impact on the asset value of immature forests unless it persists until the forests are harvested. The implication is that in calculating timber asset values, analysts should use projections of net prices expected to apply when forests are actually harvested, not current net prices. The simplest, but not necessarily most accurate, way to do this is to determine trend values by either taking a moving average or by applying regression techniques to several recent years of data. The Canada (II) and U.S. studies did the former, while the Malaysia (I) study did the latter.
Deforestation
We noted above that several studies included the volume of standing timber lost during deforestation in their estimates of net accumulation of timber. This procedure is consistent with our recommendation in Chapter 3, as long as the volume refers to merchantable timber (trees of commercial size and species in accessible areas) and is net of defect and normal rates of logging damage. The Costa Rica (I), Ecuador, Malaysia (I) and (II), and Thailand studies stated that they took into account some or all of these considerations. Several other studies instead assumed that all standing timber in deforested areas represented a commercial loss. Those studies overstated the loss in asset value.
A more fundamental issue related to deforestation is how to account for the change in land use from forest to other uses. As noted in Chapter 2, not only should the loss in value associated with the reduction in forest area be deducted in the current and asset accounts, but the gain in value associated with land in its new use should also be added. Only two studies (Mexico, U.S.) included both adjustments, although the Nepal study noted its necessity. This small number partly reflects the fact that many of the studies covered only forestry, but even some of the studies that were more comprehensive ignored the addition for converted land (e.g., Costa Rica (I)). If land markets work perfectly, then as noted earlier the value of converted land should always be greater than or equal to the value of forestland, but land markets are rarely perfectly efficient, especially in the face of nonmarket values and attenuated property rights. Hence, it is important to investigate how the asset value of land in forest use, for the full range of forest-related benefits, compares to the asset value of land in alternative uses, and not simply assume that the latter is at least as large as the former. The Mexico study concluded that the asset value of forestland for timber production exceeded the asset value for agriculture, and therefore deforestation reduced the country’s wealth. The U.S. study, on the other hand, simply assumed that the two were equal.
A comprehensive investigation of the asset value of forestland will always be hindered to a greater or lesser degree by limited data on nonmarket benefits. We discuss nonmarket valuation issues more in the next section. Here, we note the unsurprising consequence, which is that most of the studies focused on a small set of benefits in valuing forestland. Three of the studies (Costa Rica (I), Mexico, Thailand) equated the loss in value associated with deforestation to just the value of forgone timber production. This is not a problem when the only benefit is timber, as the asset value of the forest then simply equals the present value of rents from future timber harvests. That is, when forests provide only timber, asset values for timber and forestland coincide. They cannot be accounted for separately (Chapter 3), unless one introduces some artificial distinction, such as that the latter equals "bare land" timber value (asset value of a forest of age 0) and the former equals the difference between actual asset value and bare land value.
At least one study (U.S.) double-counted by including both the asset value of timber and the asset value of forestland in the asset accounts, without deducting the former from the latter. That study further overestimated the asset value of forestland by setting it equal to the average value of agricultural land, the presumed alternative land use, instead of the marginal value. The margin of competition between forests and agriculture typically occurs on poorer soils and at higher elevations, not in the middle of agricultural land that is already developed and in prime locations.
Because forests offer benefits beyond timber, the total asset value of forestland usually exceeds the timber asset value. Four studies (Australia (I), Costa Rica (II), Malaysia (I), Papua New Guinea) equated the loss in value associated with deforestation to the forgone value of one or more nontimber benefits. They did so by multiplying per-hectare values drawn from the literature, which were assumed to apply to all forestland, times the area deforested. There is nothing wrong with this procedure, as long as the values do indeed apply to all forestland and are asset values (Chapter 3), not simply values of current production. All but the Costa Rica (II) study appear to have used asset values.
Nontimber values
Thirteen of the studies included values related to one or more nontimber aspects of forests. Using the terms in the framework in Chapter 2, we can summarize the coverage as follows:
Nonmarket forest products
Fuelwood: Asia, Nepal, Sweden (I) and (II), Tanzania, Zimbabwe
Fodder: Finland (II), Nepal, Sweden (I) and (II)
Game: Malaysia (I), Finland (II), Sweden (I) and (II)
Other (e.g., mushrooms, berries): Finland (II), Sweden (I) and (II)
Forest amenities
Biodiversity: Australia (I), Global, Malaysia (I), Finland (II), Sweden (I) and (II)
Social and spiritual values: Papua New Guinea
Recreation (inc. tourism): Costa Rica (II), Finland (II)
Environmental services
Watershed protection: Costa Rica (II)
On-site soil erosion: Mexico
Not specified: Papua New Guinea
Airborne pollutants
Acid deposition: Finland (II), Sweden (I) and (II)
Carbon sequestration
Canada (I), Costa Rica (II), Finland (II), Malaysia (I), Sweden (I) and (II)
Aggregate nontimber values
Global
One of the studies (Global) calculated asset values. For forests outside protected areas, it assumed that the annual value of nontimber benefits was constant across forest types, though different in developing and developed countries. It multiplied aggregate per-hectare estimates of the value of those benefits, drawn from the literature, times forest area. It divided the result by ten, on the assumption that one-tenth of total forest area generated appreciable nontimber benefits, and divided again by 4 percent to capitalize the values. For forests within protected areas (mainly biodiversity values), it applied the opportunity cost method: it assumed that the asset value of nontimber benefits was at least as large as the asset value of land in its alternative use, which it assumed to be pasture land. Obviously, the validity of this method depends on the rationality of governments in deciding which areas to protect, and on the opportunity cost indeed being the forgone value of pasture land rather than other alternative uses (e.g., timber production or crops). Even if these conditions are satisfied, one should bear in mind that the method gives lower-bound estimates.
The remaining studies valued flows rather than stocks. For nontimber products, they used two principal valuation methods, both of which were described in Chapter 3. If some portion of production was marketed, several of the studies valued the remaining portion by multiplying times market price. The Asia and Zimbabwe studies, and implicitly the Sweden (I) and (II) studies, used this method for fuelwood. The second method was based on the opportunity cost of labor used in the collection of nontimber products. The Nepal and Tanzania studies used this method for fuelwood. As noted in Chapter 3, this method understates the value of production, except when forest resources are in a state of complete open-access (in which case they generate no rent). Most of the studies used one or the other of these two methods for valuing nonmarket production of nontimber products other than fuelwood.
The studies used a broader range of methods for valuing forest amenities. As noted above, the Global study used the opportunity cost of land to value protected areas. The Finland (II), Sweden (I) and (II), and Papua New Guinea studies also used this method. The first three set the opportunity cost equal to forgone timber rents instead of forgone rents from grazing, while the fourth set the opportunity cost equal to the official compensation rates that tribes are legally entitled to claim when the government appropriates their land. As noted in the previous section, the Australia (I) study set the per-hectare value of habitat protection equal to an estimate drawn from the literature. It did not provide details on the derivation of this estimate. The Costa Rica (II) study set the per-hectare value equal to the estimated net expenditure by international ecotourists. The Finland (II) study used number of recreational visitors instead of forest area as the basis of its estimate, multiplying the former times a literature estimate of the value per visit based on contingent valuation studies. Lastly, the Malaysia (I) study related forest area to number of species, and developed a range of estimates for the value of an individual species, including government expenditure to protected a locally endangered species and government expenditure to reintroduce locally extinct species.
The Papua New Guinea study assumed that compensation rates included unspecified "ecological services." While the asset value of such services is indeed part of the loss in value associated with deforestation, the annual value of the services should not be added to GDP (Chapter 2). As the study only estimated net accumulation, it did not commit this mistake. Likewise, the Costa Rica (II) study was correct in including the value of watershed services in its net accumulation estimates for deforestation. Its estimate of this value is implausible, however, as it set the unit value of water equal to the rate paid by urban consumers, which includes the costs of water treatment and distribution as well as the services provided by forests in providing the raw water. It also multiplied the rate times runoff from all forests, not just municipal watersheds, and it did not capitalize the value.
The Finland (II) and Sweden (I) and (II) studies included values related to airborne pollutants other than CO2. They correctly included losses in forest productivity due to acid deposition in their estimates of net accumulation. The Finland (II) study valued the loss by the productivity change method: forgone timber growth times average stumpage value. The Sweden (I) and (II) studies used instead the replacement cost method: they estimated the cost of limestone and magnesium needed to offset acid deposition. The Mexico study also used this method to value long-run losses in forest productivity due to on-site soil erosion. We discussed the shortcomings of the replacement cost method in the section on Net accumulation of timber, in reference to the Ecuador and Thailand studies.
The Finland (II) study was the only one to value the disposal services provided by forests for air pollutants other than CO2. It did the valuation correctly, multiplying the amount of nitrogen oxides and sulfur dioxide deposited in forests times the estimated abatement costs for the polluting industries. However, it subtracted the resulting estimate from forestry value added instead of adding it (Chapter 2). Moreover, as it was a forestry-specific study, it did not make the required offsetting adjustment by subtracting the estimate from value added in the polluting industries.
The studies that covered carbon sequestration all used the same method. First, they multiplied the net depletion of the timber stock times a parameter to convert from cubic meters of wood to tons of carbon. They then multiplied the resulting quantity times either the local carbon tax (Sweden (I) and (II)) or proposed global carbon taxes (the other studies). This is the method discussed in Chapter 3, although to obtain accurate estimates one should also account for changes in carbon storage in the soil and other parts of the forest ecosystem besides just standing timber. Hoen (1993) provides a useful discussion of this issue, in reference to the Sweden (I) study.
This is not the place for an extended discussion of the pros and cons of the different valuation methods applied to these widely varying nontimber aspects of forests, as many books and articles already provide extended treatments of valuation principles and methods. We highlight just three points of particular relevance for economic accounting. First, most of the estimates are very rough, and rougher than they need to be. While drawing existing estimates from the literature is perfectly acceptable, and often the only option (especially in countries with limited financial and human resources to conduct valuation studies), studies should consider and evaluate a range of estimates, and select the ones that correspond most closely to the situation in the country where the study is being conducted. Studies that calculated original estimates instead of drawing from the literature apparently often did not apply valuation methods correctly (e.g., in setting unit water values equal to municipal water rates). These mistakes were avoidable. Second, some studies apply per-hectare value estimates to the entire f orest, when only part of the forest provides the benefits being valued. This leads to an upward bias in estimates of nonmarket values. The bias could be enormous if the one-tenth assumption in the Global study gives the correct order of magnitude. Third, some studies value benefits that might already be included in the accounts, thus running the risk of double-counting. For example, it is likely that final consumption expenditures in ordinary GDP already include some of the recreational and tourism expenditures estimated by the Costa Rica (II) and Finland (II) studies. Moreover, as emphasized in Chapter 2, overall GDP does not need to be adjusted for the current value of environmental services that forests provide to other industries.
Magnitude of the adjustments
The priority placed on adjusting national accounts for forest resources should depend on how much the adjustments improve macroeconomic aggregates like GDP, NDP, and net investment as indicators of welfare and sustainability. If other prospective adjustments yield greater improvements, then analysts should give them precedence over forest-related adjustments. The studies provide some information on this issue, particularly those studies that cover other resources in addition to forests. We emphasize, however, that this information should be regarded as no more than suggestive, and perhaps even misleading. The sample of studies is small, the countries they cover were not selected randomly, and the accounting approaches and valuation techniques used in them vary considerably, with some suffering from severe biases.
Consider first the magnitude of adjustments to the current accounts. Forest-related net accumulation was large relative to net value added in forestry in some studies, but it was generally smaller than net accumulation of subsoil assets. It was also small relative to NDP and net investment. It tended to be larger in developing countries than in developed countries. These same tendencies were generally true of estimates of final consumption of nonmarket forest products.
In many developed countries, net accumulation of timber was positive: the value of timber resources rose, due to rising timber stocks and capital gains. Even in developing countries, overall net investment tended to be positive after adjusting for changes in the value of forest capital, indicating that investment in human-made capital more than offset disinvestment of forest capital (although whether it outweighed the total depreciation of all forms of natural capital is not always clear, certainly not in the forest-specific studies). Moreover, the actual relative importance of net accumulation and nonmarket production was even less than it seemed in studies that applied the net price method, that included net accumulation of forestland but not net accumulation of converted land, and that assumed the entire forest produced appreciable amounts of nontimber products.
We emphasize that adjustments for net accumulation and nonmarket forest products are theoretically justified and should in principle be made, even if they have a minimal impact on the accounts. Furthermore, small values at the national level do not necessarily imply small values within a country. The Malaysia (II) study found that the loss in value of timber assets was large relative to NDP and net investment in Sabah and Sarawak, though not in Peninsular Malaysia nor at the national level. Given that the abundance of natural resources usually varies considerably within a country, income accounts adjusted for forest resources may be more valuable for provincial policymakers than for national policymakers.
Findings are similar for the asset accounts. Forests account for a much smaller (in many cases, a virtually insignificant) share of total capital than does human-made capital, especially in (not surprisingly) developed countries. They usually account for a smaller share of countries’ natural wealth than do subsoil assets and often cropland. Within the category of forest capital, the asset value of timber resources tends to be larger than the asset value of nontimber resources.
Conclusions for asset accounts are even more tenuous than for product accounts, as considerably fewer studies analyzed the former than the latter. Nevertheless, the findings just noted point in the same direction as those for product accounts: that economic accounting of forest resources is likely to be more important in developing than developed countries, at the subnational than the national level, in mineral-poor than mineral-rich countries, and for timber than for nontimber values.
