مشکل داوطلبانه انتشار تجارت : عدم قطعیت و کژ گزینی در برنامه های اعتباری بخشی

Sectoral crediting has been proposed as a way to scale up project-level carbon offset programs, and provide sector-wide incentives for developing countries to reduce greenhouse gas emissions. However, simulations presented here suggest that information asymmetries and large uncertainties in predicting counterfactual business-as-usual (BAU) emissions are likely to render sectoral crediting an extremely unattractive mechanism in practice, at least for the transportation sector. The regulator faces a tradeoff between efficiency and transfers/environmental damage when setting the crediting baseline in relation to uncertain BAU emissions. A generous baseline promotes efficiency, as more developing countries participate and implement abatement measures. However, a generous baseline also produces large volumes of non-additional offsets, which lead to either increased global emissions, or transfers between developed and developing countries if developed country emission reduction targets are made more stringent in order to leave global emissions unchanged. I show that any crediting baseline that encourages a non-negligible number of countries to participate in a sectoral crediting mechanism results in environmental damage or transfers that are likely to be too high to be politically feasible.

The carbon market is the centerpiece of current efforts to fund low-cost measures to reduce greenhouse gas emissions in developing countries. In particular, the Clean Development Mechanism (CDM), an implementation mechanism of the Kyoto Protocol, allows developed countries to purchase carbon offsets from projects in developing countries as a partial alternative to domestic action. By equalizing marginal abatement costs across sectors and across countries, the CDM can in principle substantially reduce the cost of achieving a given abatement target (Anger et al., 2007). The CDM, however, has come in for substantial criticism in recent years. There is evidence that many of the CDM offsets are not “additional;” i.e., the project would have been undertaken anyway in the absence of the CDM (Wara and Victor, 2008, Haya, 2009, Schneider, 2009, Fujiwara, 2010 and He and Morse, 2010). Other lines of criticism relate to problems with the methodologies used to quantify emission reductions (Millard-Ball and Ortolano, 2010); the lack of broad sustainable development benefits from CDM projects (Sutter and Parreño, 2007); and the inability of the CDM to promote innovation and incentivize long-term transformations in energy systems (Sterk, 2008). Sectoral no-lose targets and other sector-based crediting mechanisms have emerged prominently as a way to address these problems with project-level CDM (Bosi and Ellis, 2005, Figueres, 2006, Center for Clean Air Policy, 2008, Ecofys, 2008, Sterk, 2008, Baron et al., 2009 and IETA, 2010). Developing countries would participate on a voluntary basis, and could generate tradable credits (offsets) by reducing emissions to below a sectoral “crediting baseline” (Fig. 1). Emissions above the crediting baseline would not be penalized (hence the “no lose” designation).There are four fundamental differences between sectoral no-lose targets and the existing CDM. First, the CDM operates at the project level, while sectoral no-lose targets consider aggregate sectoral emissions and do not seek to attribute reductions to any particular project. Second, CDM projects are typically proposed by private investors, while offsets from sectoral no-lose targets would accrue to national governments, who would in turn determine how to pass through incentives to private actors. Third, emission reductions under the CDM are calculated via a two-step process: a binary determination of additionality, followed by an estimate of emission reductions below a counterfactual baseline. In the case of sectoral no-lose targets, both additionality and baseline issues are implicit in determining the crediting baseline. Fourth, the baseline for CDM is typically business-as-usual (BAU).1 In contrast, most discussions of sectoral no-lose targets assume that the crediting baseline would be set below BAU, as implied in Fig. 1, bringing about a net reduction in global emissions. However, the crediting baseline could be set at any level, including at or above BAU. The regulator, such as the UN or other multilateral body, faces a key tradeoff when setting the crediting baseline in the presence of uncertainty over BAU. Set the crediting baseline too stringently, and developing countries may not participate—a rational decision if the costs of reducing emissions to the crediting baseline exceed the revenues from the sale of offsets from further emission reductions. Thus, a stringent baseline risks foregoing low-cost abatement opportunities in countries that do not participate. Set the crediting baseline too generously, and it risks being above counterfactual BAU and enabling developing countries to sell non-additional offsets. These non-additional offsets either represent an environmental cost if global emissions increase, or else a transfer cost from developed to developing countries if targets in developed countries are made more stringent to leave global emissions unchanged. The essential tradeoff faced by the regulator is between efficiency on the one hand, and environmental or transfer costs on the other hand, and the ex-ante optimal baseline depends on their relative importance. One underlying cause of this tradeoff can be overall uncertainty about BAU emissions—i.e., uncertainty that is common to the regulator and the developing country. In this case, the impacts of uncertainty on efficiency, environmental costs, and distributional outcomes will partly depend on how risk is allocated between offset purchasers and the developing country offset suppliers. Another cause can be adverse selection, which arises from information asymmetries between the regulator and individual developing countries. Since a country has more information on its own counterfactual BAU emissions than does the regulator, it can decide to participate if it is granted (by virtue of the regulator's uncertainty) a favorable baseline. Indeed, adverse selection is an issue with any voluntary emissions trading program, including domestic cap-and-trade systems that allow firms to decide whether or not to participate. In the case of the U.S. Acid Rain Program, generating units with a “generous” baseline (one set above their counterfactual BAU emissions) were more likely to participate, resulting in increased SO2 emissions and a net social loss after considering abatement cost savings (Montero, 1999 and Montero, 2000). Adverse selection problems have also been raised in the contexts of crediting rules under project-based CDM (Fischer, 2005) and under opt-in programs for agriculture and forestry (Kerr and Sweet, 2008 and van Benthem and Kerr, 2010). With one main exception (Montero, 1999), however, the impacts of adverse selection in emissions trading have not been estimated empirically. This paper offers three main contributions. First, I numerically simulate the tradeoff between environmental or transfer costs and efficiency. In contrast to Montero (1999), who provides econometric estimates using historical emissions data, my simulations allow me to explore the impacts of a wide range of regulatory decisions over crediting baselines—not just the baseline that happened to have been implemented by the regulator. As discussed in Section 2, analytic results are ambiguous, necessitating the use of simulations. Moreover, I consider the case where uncertainty is common to the regulator and the developing country, as well as the asymmetric information case with adverse selection. Second, I offer the first case study of implementing sectoral crediting in the transportation sector. Transportation is important because of the sheer size of the sector—it accounted for 23% of global energy-related CO2 emissions in 2007 (International Energy Agency, 2009). Moreover, its underrepresentation in the CDM, accounting for less than 1% of emission reductions, suggests that the gains from moving to a sectoral approach may be large (Bradley et al., 2007, Ellermann, 2009, Schneider and Cames, 2009 and Wittneben et al., 2009). Third, I contribute to the policy literature on sectoral crediting and other market-based mechanisms to engage developing countries in greenhouse gas abatement. There is a considerable literature advocating sectoral crediting as a policy solution, but it has focused on conceptual design issues with little detailed analysis of how to set the crediting baseline. This paper is the first to quantify the costs that uncertainty and adverse selection may impose on such a mechanism. The conclusions of this paper stand in contrast to the excitement over sectoral no-lose targets evident in the policy literature, as well as the theoretical attraction of using voluntary market mechanisms to equalize marginal abatement costs across the globe. I show that the tradeoff between efficiency and environmental or transfer costs is likely to be stark and unappealing in practice, at least for the transportation sector. Any crediting baseline that encourages a non-negligible number of countries to participate is too generous from the standpoint of additionality—more than 75% of offsets under most scenarios are non-additional, leading to major increases in global emissions or transfers that can exceed $10 billion per year. At root, this result is due to the imprecision with which the regulator can predict counterfactual BAU emissions in developing countries that are rapidly growing. Thus, while the efficient outcome2 is attainable, the necessary transfers from developed to developing countries are likely to be too high to be politically feasible. These transfers are assumed to be made through adopting more stringent emissions targets in developed countries, in order to maintain global emissions at the optimum level. If targets are not adjusted, then there may be no efficiency gain at all due to increased environmental damage. The paper proceeds as follows. In Section 2, I present a theoretical model that specifies participation and abatement decisions by developing countries, and the baseline-setting decision by the regulator. Section 3 describes the empirical approach to estimating abatement cost functions and business-as-usual emissions. In Section 4, I present the results of the simulations. Section 5 concludes with implications for the design of policy instruments to fund abatement in developing countries.

In principle, sectoral no-lose targets are a compelling mechanism to provide incentives for emission reductions in developing countries. However, their feasibility is conditional on the ability of both individual developing countries and an international regulator to make reasonably accurate predictions of business-as-usual emissions. The results presented in this paper suggest that, at least for the transportation sector, the uncertainties in predicting BAU are extremely large relative to expected abatement. This is the case even when, as here, contemporaneous GDP and oil prices are used to make the predictions (i.e., the baseline is dynamic). As a result of the uncertainties, a large fraction of offsets are non-additional, rendering sectoral no-lose targets an unattractive option. The efficient solution requires setting an extremely generous baseline—generous enough to compensate for the regulator's prediction error—to encourage as many countries as possible to participate. However, in this case, almost all of the resultant offsets will be non-additional. If Annex I countries were not to tighten their own emission caps in response, which is perhaps the most likely outcome, global emissions would be higher on the order of 500 Mt CO2 per year. If Annex I caps are tightened, then environmental impacts are avoided but large transfers (payment for non-additional offsets) that can exceed $10 billion per year are required. For comparison, the total mitigation assistance pledged under the Copenhagen Accord was just $30 billion. Large transfer payments may be justifiable from an ethical or equity point of view, in that they will tend to flow from some of the largest emitters to countries that bear little historical responsibility for CO2 emissions. Politically, however, monetary transfers of this magnitude are almost certainly unacceptable. Moreover, as BAU cannot be calculated ex post, neither can additionality or the amount of transfer to a particular country; thus, transfers cannot be made in lieu of direct overseas development assistance for mitigation or adaptation. An alternative regulatory approach would be to focus not on efficiency, but on environmental goals or minimizing transfer payments. The regulator might seek to maximize global emission reductions or the percentage of additional offsets. However, such an approach will leave sectoral no-lose targets largely irrelevant, as the baseline would be set so stringently that few countries participate. At low carbon prices, moreover, even such stringent baselines are insufficient to ensure that most offsets are additional. The results here assume that governments can pass on the carbon price signal to firms and consumers, or enact regulations to achieve the same goal. They ignore the potential for national governments to manipulate emissions data. They also assume that both the regulator and individual countries have perfect information on abatement cost curves and carbon prices. To the extent that these assumptions do not hold, the potential for sectoral crediting mechanisms may be even bleaker than suggested here. The inability to make precise predictions about transportation emissions, particularly over a five- to ten-year time horizon, is hardly surprising. Even in the much more static and data-rich environment of the U.S., predictions of regional travel demand models can err by 6% (Rodier, 2004; see also Flyvbjerg et al., 2005 and Transportation Research Board, 2007). Despite a sophisticated energy modeling system, aggregate five-year U.S. transportation energy forecasts were off by an average of 6.6% during the 1980s and 1990s (Winebrake and Sakva, 2006; see also Fischer et al., 2009). Nor is the problem of predictive performance limited to transportation, which suggests that similar analyses might reveal problems of uncertainty and adverse selection in other sectors. Even in those considered more “straightforward,” such as the electricity generation sector with its uniform product, there are large uncertainties in estimating baselines (Zhang et al., 2006). In the U.S., the 6.6% average error for transportation compares to 8.0% for industrial production, 5.3% for commercial, and 2.8% for residential (Winebrake and Sakva, 2006). Similar, the International Energy Agency's predictions for industrial energy demand in individual countries are no better than those for transportation energy demand (Linderoth, 2002). Fischer et al. (2009) also suggest that the commercial and industrial sectors can be more difficult to predict than transportation. Thus, while this paper analyses only the case of transportation, it would be wise to be cautious about the feasibility of similar crediting mechanisms in other sectors. To the extent that policymakers wish to pursue sectoral no-lose targets, they might be advised to focus on sectors and countries where the prediction error is likely to be small in relation to expected abatement. This implies that an “open to all” system might not be the most attractive option. Instead, sectoral no-lose targets might be implemented on an invitation-only basis to specific countries for specific sectors where emissions have historically been relatively easy to predict. Furthermore, sectoral crediting might only be implemented once the carbon price reaches a given threshold, as another way to increase the volume of expected abatement relative to prediction errors. An invitation-only system might also bring to bear non-financial pressures on countries to participate. The analysis in this paper assumes that participation decisions are made solely on financial grounds. However, political pressure on countries might promote greater participation and improve performance on all three metrics—offset quality, participation and global emissions. Critiques of offsets and other tradable credit-based approaches to reducing emissions in developing countries have already identified a wide range of challenges, such as inattention to sustainable development co-benefits; the focus on shorter-term, measurable projects; and payment of the market clearing price rather than incremental cost for emission reductions, which reduces the abatement that can be secured for a given sum of money. This paper provides further evidence that the more we study offsets and similar crediting mechanisms, the more problems we uncover. Both policy design and estimates of abatement potential in developing countries need to take into account the impacts of uncertainty, information asymmetries and other barriers to realizing the full potential. Meanwhile, researchers and policymakers might usefully compare offsets against other potential climate policy instruments such as results-based agreements, grants and technology transfer (Kerr and Millard-Ball, 2012). While tradable credits offer many attractions in principle, not least the ability to equalize marginal abatement costs across sectors and countries, other instruments may offer more robust ways to fund mitigation in developing countries in practice.