Cover Image


View/Hide Left Panel

an example of a global forestry model. Optimization approaches have been widely applied to assess the costs of changes in forest management, e.g., increases in rotation ages or increases in management intensity. Similarly, they have been used to assess the costs of changes in management of agricultural land, e.g., adoption of conservation tillage, or changes in nitrogen fertilizer rates, for example, Murray et al. (2005), and Choi and Sohngen (2009).

There is debate in the literature over which methods are “best” (e.g., Stavins and Richards, 2005; Lubowski et al., 2006), but it is not likely possible to determine which methods are best, or even better, for estimating the costs of carbon sequestration. Each method handles different problems and can be used effectively under a given set of circumstances. For example, bottom-up studies can be very effective tools to get an initial sense for potential costs of different alternatives, particularly when data is limited. Econometric approaches, of course, have the benefit of producing estimates that allow calculation of statistical properties such as confidence intervals. Both econometric and bottom-up estimates, however, typically assume that input and output prices are exogenous, an assumption that may not be tenable under the fairly large land use change programs they seek to evaluate. Recent econometric approaches have developed process-based optimization approaches that utilize the econometric estimates within the context of an optimization model (e.g., Lubowski et al., 2006).

Top down, optimization models, on the other hand, model forest and land management directly, with feedbacks between output, output prices, and the intensity of management. To accomplish this, they are typically constructed to be much more aggregate than the econometric and bottom-up approaches. That is, the typical unit of observation may be a forest type in a specific region of the country (or multiple counties). As computer speeds have increased, modelers have been able to increase their level of disaggregation. The benefit of this approach is that as carbon sequestration policies are modeled, they have impacts on overall land use, which impacts output prices and resource costs. These are modeled explicitly in optimization approaches, allowing direct calculation of the opportunity costs of shifting land from one use to another.

There are relatively few direct comparisons of the approaches. Van Kooten and Sohngen (2007) conducted a meta-analysis of many different studies of carbon sequestration costs and found that methodological differences explained very little of the differences in marginal cost estimates. There is some limited evidence that optimization and econometric estimates are higher cost than bottom up studies, but these results are very dependent on functional form and thus not all that robust. Thus, across a range of 68 currently available studies, their results provide little evidence to support using one method over another to obtain more realistic costs.

Current Cost Estimates

Marginal cost functions for carbon sequestration in forests for three general regions of the world, as derived from IPCC (2007), are shown in Figure C.16. The largest potential exists in tropical countries, due to the carbon benefits (and low costs) of reducing deforestation. The potential in developed countries is fairly large as well, although it is driven by increased forest management. Table C.5 breaks out annual estimates for a number of studies by region and activity, including reduced deforestation, afforestation, and forest management, at a fixed carbon price of $15 per t CO2. These results indicate that around 4.1 billion t CO2 could be sequestered in global forests through various activities over the period 2020-2050 for $15 per t CO2.

At similar carbon prices, $15 per t CO2, national level estimates in the United States of the potential for conservation tillage on cropland to sequester carbon range from 8 t CO2 per year to 168 t CO2 per year (Lewandrowski et al., 2004; Murray et al., 2005). Both of these estimates utilize optimization approaches. The Lewandrowski et al. (2004) study is a single period optimization approach, while the Murray et al., (2005) approach is a multi-period, or dynamic, optimization approach.

A number of regional studies in the United States have examined carbon sequestration through conservation tillage on cropland as well, and they seem to suggest relatively high costs for this activity. Choi and Sohngen (2009) find that in Ohio, Indiana, and Illinois, around 4.1 million t CO2 per year could be sequestered on cropland for $15 per t CO2. These three states account for 26% of the total corn and soybean crop in the United States, so extrapolating these results nationally60 suggests a total potential of only 15.5 million t CO2 per year for $15 per t


The extrapolation is made for expository purposes only, making the very strong assumption that cropland is of similar productivity in other parts of the country and that opportunity costs are similar.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement