19.4.2.4 Cost–benefit analysis

Cost–benefit analyses (CBAs) of climate change in general are reviewed in Fisher et al., 2007 Section 3.5.3.3. The discussion here focuses on the suitability of CBA for avoiding key vulnerabilities and DAI. Most early cost–benefit analyses of climate change have assumed that climate change will be a gradual and smooth process. This assumption has prevented these analyses from determining a robust optimal policy solution (Hall and Behl, 2006), as it neglects important key vulnerabilities. Recognising the restrictions of this assumption, an extensive literature has developed extending cost–benefit analyses and related decision-making (e.g., Jones, 2003) in the context of Article 2, with a particular emphasis on abrupt change at global and regional scales (Schneider and Azar, 2001; Higgins et al., 2002; Azar and Lindgren, 2003; Baranzini et al., 2003; Wright and Erickson, 2003).

Several papers have focused on incorporating damages from large-scale climate instabilities identified as key vulnerabilities, such as climate-change-induced slowing or shutdown of the MOC (Keller et al., 2000, 2004; Mastrandrea and Schneider, 2001; Link and Tol, 2004). For example, quantifying market-based damages associated with MOC changes is a difficult task, and current analyses should be interpreted as order-of-magnitude estimates, with none carrying high confidence. These preliminary analyses suggest that significant reductions in anthropogenic greenhouse gas emissions are economically efficient even if the damages associated with a MOC slowing or collapse are less than 1% of gross world product. However, model results are very dependent on assumptions about climate sensitivity, the damage functions for smooth and abrupt climate change and time discounting, and are thus designed primarily to demonstrate frameworks for analysis and order-of-magnitude outcomes rather than high-confidence quantitative projections.

Several researchers have implemented probabilistic treatments of uncertainty in cost–benefit analyses; recent examples include Mastrandrea and Schneider (2004) and Hope (2006). These probabilistic analyses consistently suggest more aggressive mitigation policies compared with deterministic analyses, since probabilistic analyses allow the co-occurrence of high climate sensitivities (see Key caveat in Section 19.4.2.2 on low confidence for specific quantitatitive results) with high climate-damage functions.