The additionality problem with offsets: Optimal contracts for carbon sequestration in forests

https://doi.org/10.1016/j.jeem.2013.02.003Get rights and content

Abstract

Carbon offsets are a frequently discussed tool for reducing the costs of an emissions reduction policy. However, offsets have a basic problem stemming from asymmetric information. Sellers of offsets have private information about their opportunity costs, leading to concerns about whether offsets are additional. Non-additional offsets can undermine a cap-and-trade program or, if the government purchases them directly, result in enormous government expenditures. We analyze contracts for carbon sequestration in forests that mitigate the asymmetric information problem. Landowners are offered a menu of two-part contracts that induces them to reveal their type. Under this scheme, the government is able to identify ex post how much additional forest each landowner contributes and minimize ex ante its expenditures on carbon sequestration. To explore the performance of the contracting scheme, we conduct a national-scale simulation using an econometric model of land-use change. The results indicate that for an increase in forest area of 61 million acres, government expenditures are $5.3 billion lower under the contracting approach compared to a uniform subsidy offered to all landowners. This compares to an increase in private opportunity costs of just $110 million dollars under the contracts. Thus, the contracting scheme is preferable from society's perspective.

Introduction

Reducing emissions of greenhouse gases to lessen the impact of future climate change is likely to result in large net benefits globally. However, reducing emissions will impose significant costs. One recent estimate of the costs of achieving optimal abatement of greenhouse gas emissions through control of industrial CO2 emissions was $2.2 trillion [17], which underscores the appeal of any scheme that might lower the costs of carbon reduction. One way to reduce costs that has received a great deal of attention is the use of offsets. The idea is that countries could meet emissions targets by substituting lower-cost offsets for reductions in emissions from energy production.

A particularly promising type of offset involves carbon sequestration in forests. Numerous studies have found that forest sequestration can be used to offset a substantial share of carbon emissions at costs that are similar to or lower than those associated with energy-based mitigation approaches [21], [34], [28]. Other offset categories include carbon storage in agricultural soils and the transfer of clean energy technologies to developing countries not subject to emissions targets. Some types of forest and energy offsets are allowed under the Clean Development Mechanism (CDM) of the Kyoto Protocol, and there is interest in expanding their use under future agreements.2

Despite their potential to reduce costs, offsets have a basic problem stemming from asymmetric information. Sellers of offsets have private information about their opportunity costs of mitigating or abating emissions. This implies that only the seller knows whether she would have undertaken the activity in the absence of a payment for the offset. This leads to the oft-expressed concern about “additionality”: offsets are not true incremental adjustments if they would have happened anyway.3

The additionality of offsets is important in two respects. First, governments will want to avoid paying for non-additional offsets in order to limit their expenditures. A number of studies have argued that government expenditures can reduce net social benefits if there are opportunity costs to raising public funds [1], [5], [6], [8]. The government costs associated with purchases of offsets could be enormous. For example, in the U.S. an average of 1.3 million acres was deforested annually between 1982 and 1997 [32]. While this produced significant carbon emissions that might be avoided at reasonable cost, one must consider that the area of (non-federal) forest in the U.S. is approximately 400 million acres. If the government were to implement a subsidy for avoided deforestation and apply it uniformly across all forested acres, then in the extreme case it would subsidize all forest land when less than 1% of the area would have been deforested. Of course, the government can avoid large expenditures by levying taxes instead of paying subsidies. We regard this option as politically unviable, especially in the context of private landowners in the U.S.

Second, to legitimately use offsets to meet emissions reduction targets, such as those stipulated under international treaties, the government must be able to verify that they are additional. Procedures that fail to clearly identify the increment of sequestration produced have been soundly criticized [20], and, thus, concerns about additionality have been a stumbling block for inclusion of offsets, particularly those from avoided deforestation, in international efforts to address climate change [19]. Even when a private entity such as a regulated emissions source purchases offsets, it faces the same problem as the government does with asymmetric information. As long as it is required to verify the purchase of additional offsets, a private entity will want to avoid paying for non-additional offsets as well as limiting its expenditures on the offsets that are additional. However, sellers have an incentive to exploit the asymmetric information by claiming to have high opportunity costs.

In this paper, we propose a contracting scheme for carbon offsets and investigate its performance empirically with a national-scale simulation. Our paper combines modern contract theory with a careful application of econometric results based on a concrete example. The distinguishing feature of our empirical analysis is that we estimate the distribution over agent types using revealed preference data on individual land-use decisions. As such, we are able to obtain meaningful estimates of the costs, to the principal and agents, associated with employing optimal contracts.4 In our application, we study a national-scale policy that addresses an environmental problem of critical importance. The estimates we obtain allow us to articulate the potential welfare gains from the use of sophisticated contracts that are based on defensible, empirically-based parameter values. Importantly, our results show these potential welfare gains are significant, in a variety of geographical contexts.

Our theoretical model adapts a standard principal-agent framework [12], [22] to the problem at hand. The principal's objective is to maximize expected net societal benefits from afforestation and avoided deforestation (collectively, forestation), where forestation benefits are tied to an exogenously determined carbon price and costs are defined in terms of government expenditures. The problem may be regarded as one of adverse selection: the principal is assumed to know the distribution over landowners' opportunity costs, but not the realization for any particular individual; as a result, the amount of land any particular agent would have placed in forest absent a payment is not observed by the principal.5 In the maximization problem we investigate, the optimality conditions induce a set of optimal contracts, one for each type of agent. Each contract has two ingredients: an amount of land in forest and a payment to the agent. The essential feature of the contract scheme is that it induces agents to truthfully reveal their type (i.e., their opportunity costs). This enables the government to identify ex post how much additional forest is contributed by each landowner. Further, the government is able to minimize ex ante its expenditures on forestation.

We then present a national-scale simulation of the optimal contracts based on individual-level data. The analysis draws on the econometric model of land-use change developed by Lubowski et al. [14]. We use the model to estimate marginal costs distributions for forestation and, with these estimates, compute the optimal contract menus using the theoretical results. We compare costs (both government costs and private opportunity costs) under the contracting approach to the costs of a uniform subsidy offered to all landowners. The results show that government costs are considerably lower under the optimal contracts than under the uniform subsidy. However, because the optimal contract scheme sets different subsidies for different agents, it violates the equi-marginal principle. This social cost inefficiency turns out to be small in relation to the reduction in government outlays associated with the optimal contract scheme. Thus, the contract scheme is socially preferable as well.

The paper proceeds with a description of the theoretical model in Section 2. In Section 3, we present a national-scale simulation of the optimal contracts based on individual-level data. Section 4 presents the results from this simulation analysis; there we show that government costs are considerably lower under the optimal contracts than under the uniform subsidy, and are much closer to first-best costs. Further discussion and conclusions are provided in Section 5.

Section snippets

Theoretical statement of the problem

We suppose there is a governmental agency, which we term the “principal,” that is interested in having land placed in forest. Each unit of land placed in forest generates an amount of sequestration, which yields a benefit Pc to the principal. This induced benefit can naturally be thought of as a value of marginal product, which depends on the price of carbon (which can either be explicit, if a formal carbon market exists, or implicit, as with an emissions trading scheme) and the marginal

Empirical analysis of carbon sequestration contracts

To empirically gauge the ability of our optimal contracting scheme to reduce government costs we conduct a national-level simulation of the carbon sequestration contracts. Two key ingredients for the simulation are the θ distributions and the forest supply functions z(p), which we use to infer the marginal cost associated with a given level of forestation x. These are derived using the econometric model of land use developed by Lubowski et al. [14]. These authors estimate a discrete-choice

Simulation results

By combining Eqs. (5), (6) with the results from Section 2, we are able to simulate the optimal menu of contracts for each land class within each state for a range of VMPs (see Appendix B for details). The simulated contracts can then be used to calculate the expected amount of land in forest, by state and land class, and the associated cost to the principal. We can also use information regarding the distribution over θ and the z(p) function to determine the constant subsidy that would induce

Conclusion

In this paper, we have proposed optimal contracts to encourage carbon offsets from forestation at minimal cost to the government, subject to the landholders having private information about their opportunity costs of placing land in forest. These contracts typically leave some rents in landowners' hands, and so are second-best in nature. But, assuming that the government is concerned with the budgetary effects of the policy, the contracts are a much cheaper approach to inducing the expansion or

References (35)

  • P. Bontems et al.

    Regulating nitrogen pollution with risk averse farmers under hidden information and moral hazard

    American Journal of Agricultural Economics

    (2006)
  • A.L. Bovenberg et al.

    Optimal environmental taxation in the presence of other taxesgeneral-equilibrium analyses

    American Economic Review

    (1996)
  • E.K. Browning

    On the marginal welfare cost of taxation

    American Economic Review

    (1987)
  • J.B. Bushnell, Adverse selection and emissions offsets, Energy Institute at Haas Working Paper 222,...
  • B. Dahlby

    The Marginal Cost of Public Funds

    (2008)
  • P. Gagnepain et al.

    Incentive regulatory policiesthe case of public transit systems in France

    RAND Journal of Economics

    (2002)
  • H. Gorg et al.

    The effect of r&d subsidies on private r&d

    Economica

    (2007)
  • Cited by (0)

    We thank, without implicating, participants at the 2011 Allied Social Science Meetings; 2010 NBER summer institute (Energy and Environmental Economics); the Toulouse conference on Firms, Environment and Natural Resources; the 2010 Heartland Environmental and Resource Economics workshop; the 4th World Congress of Environmental and Resource Economists; and seminar participants at the London School of Economics, University of California—Santa Barbara, University of Paris, University of Puget Sound, University of Siena, and Williamette University. Special thanks are due to Dave Lewis, Brian Murray and Steve Polasky. Two anonymous referees and the Managing Editor provided constructive criticisms that greatly improved the content and presentation of our paper. Funding from the U.S. Department of Energy (grant # DE-FC26-05NT42587) and the U.S Forest Service (grant #PNW 09-JV-11261955-067) is gratefully acknowledged.

    1

    Fax: +1 805 893 7612.

    View full text