IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group III: Mitigation of Climate Change Equilibrium modelling of carbon leakage from the Kyoto Protocol

Paltsev (2001) uses a static global-equilibrium model GTAP-EG to analyse the effects of the Kyoto Protocol. He reports a leakage rate of 10.5%, with an uncertainty range of 5–15% covering different assumptions about aggregation, trade elasticities and capital mobility, but his main purpose is to trace back non-Annex B increases in CO2 to their sources in the regions and sectors of Annex B. The chemicals and iron and steel sectors make the highest contributions (20% and 16% respectively), with the EU being the largest regional source (41% of total leakage). The highest bilateral leakage is from the EU to China (over 10% of the total). Kuik and Gerlagh (2003) use a similar GTAP-E model and conclude that, for Annex I Kyoto-style action, the main reason for leakage is the reduction in world energy prices, rather than substitution within Annex I. They find that the central estimate of 11% leakage is sensitive to assumptions about trade-substitution elasticities and fossil-fuel supply elasticities and to lower import tariffs under the Uruguay Round, and they state a range of 6% to 17% leakage.

In a more recent study, Babiker (2005), using a model with different assumptions about production and competition in the energy-intensive sector, reports a range of global leakage rates between 25% and 130%, depending on the assumptions adopted. The main reasons for the higher estimates are the inclusion of increasing returns to scale, strategic behaviour in the energy-intensive industry and the assumption of homogeneous products. Rates above 100% would imply that mitigation action in one region leads to more global GHG emissions rather than less.

However, other studies point to real world conditions that make these outcomes unlikely. Significant carbon leakage arises when internationally tradeable energy-intensive production moves abroad to non-abating regions. This is frequently referred to as a competitiveness concern. In industrialized countries, these sectors account for 15–20% of CO2 emissions (IEA, 2004). Results with high leakage therefore reflect conditions in which countries implement policies that lead to most emission savings being obtained by industrial relocation (to areas of lower-cost, and in some cases less efficient, production), rather than in the less mobile sectors (such as power generation, domestic, services etc). In practice, most countries have tended to adjust policies to avoid any such outcome (for example through derogation, exemption or protection for such sectors).

Sijm et al. (2004) provide a literature review and an assessment of the potential effects of Annex I mitigation associated with the EU emissions trading scheme (ETS) for carbon leakage, and especially in developing countries. Technological spillovers discussed in this paper are considered in section 11.7.6 below. In the empirical analysis of effects in energy-intensive industries, the modelling studies reporting high leakage rates look at many other factors in addition to price competitiveness. They conclude that, in practice, carbon leakage is unlikely to be substantial because transport costs, local market conditions, product variety and incomplete information all favour local production. They argue that a simple indicator of carbon leakage is insufficient for policymaking. Szabo et al. (2006) report production leakage estimates of 29% by 2010 for cement given an EU ETS allowance price of about 50 US$/tCO2 and a detailed model of the global industry. Leakage rates rise with the allowance price. More generally, Reinaud (2005) surveys estimates of leakage for energy-intensive industries (steel, cement, newsprint and aluminium) assuming the EU ETS. She comes to a similar conclusion to Sijm et al. (2004) and finds that, with the free allocation of CO2 allowances, ‘any leakage would be considerably lower than previously projected, at least in the near term.’ (p. 10). However, ‘the ambiguous results of the empirical studies in both positive and negative spillovers warrant further research in this field.’ (p.179).