Costs of Carbon Sequestration: A Revealed-Preference Approach

Increased attention by policy makers to the threat of global climate change has brought with it considerable attention to the possibility of encouraging the growth of forests as a means of sequestering carbon dioxide. This interest has been due, in part, to: suggestions that sufficient lands are available to use the approach to mitigate a substantial share of annual carbon dioxide (C02) emissions; and claims that growing trees to sequester carbon is a relatively inexpensive means of combating climate change.

This paper demonstrates a methodology whereby reliable estimates of the costs of carbon sequestration can be developed on the basis of revealed-preference evidence from observations of landowners' behavior when confronted with the opportunity costs of alternative land uses. The analytical model takes account of current silvicultural understanding of the intertemporal linkages between deforestation and carbon emissions, on the one hand, and between forestation and carbon sequestration, on the other. Furthermore, the approach employed reflects the actual preferences of landowners, as revealed by landowners' decisions regarding the disposition of their lands in the face of relevant economic signals.

The preliminary results indicate, first, that the marginal costs of carbon sequestration are by no means trivial, despite the claims that have sometimes been made. Second, even putting aside the anticipated responsiveness of land prices to any significant national sequestration program, the heterogeneity of the existing stock of land brings sharply increasing marginal costs of sequestration as higher quality agricultural lands are converted to forested use. Third, there is the somewhat surprising finding that the marginal sequestration cost function shifts upward when periodic harvesting of timber is allowed, despite the fact that opportunity costs for landowners are lowered.

Fourth and finally, there is a striking asymmetry between the marginal costs of carbon sequestration through forestation and those through retarded deforestation. If further research corroborates these findings, this may provide another argument for focusing carbon-sequestration efforts in areas of relatively high rates of deforestation, such as tropical forested regions. In addition to the fact that these areas are more efficient engines of carbon storage than temperate forests and in addition to the lower opportunity costs of land that we would ordinarily anticipate to be associated with such areas, there is the additional reality that in an intertemporal economic context, retarded deforestation provides carbon conservation at much lower marginal costs than does forestation (of the same area).

The major advantage of our approach over the models that have dominated the literature on carbon sequestration is that simulations of marginal costs build directly upon revealed-
preference patterns of how landowners have actually responded to the economic incentives they continually face regarding the alternative uses of their lands. This is in contrast with engineering approaches that build up marginal cost functions by aggregating point estimates of how landowners in a particular region or holding a particular type of land ought to behave, and in contrast with optimization models that often do much the same thing.

As is well known, landowners tend not to behave as they "ought" farmers are notoriously sluggish in responding to the economic signals they face. Stated otherwise, they are strongly affected by non-pecuniary factors, including a desire to stay on the farm for reasons associated more with quality of life than with financial returns. An econometric model that is based upon an underlying optimization model and then allows for "partial adjustment" or other phenomena can capture, albeit in a crude way, such land-use behavior. Hence, the land-use simulations that come from it, along with the respective estimates of carbon-sequestration costs may he better approximations of reality.

Our approach of linking a dynamic simulation model of carbon sequestration with an econometric model of land use has the potential of adding significantly to our understanding of the costs of this frequently discussed approach to addressing the threat of global climate change. There is a small, but growing literature of econometric analyses of forestation and deforestation. Any can serve as the basis for developing revealed-preference analytical models of the respective marginal costs of carbon sequestration.