An often overlooked source of uncertainty in assessments of impacts and vulnerability is the wide difference in assumptions (often not even stated) in the initial conditions and trends of environmental systems and socioeconomic conditions. These assumptions include information on population and related variables (e.g., population density), economic trends (e.g., income levels, sectoral composition of GDP, or levels of trade), other social indicators (e.g., education levels, private- and public-sector institutions), culture, land cover and use, and availability and use of other resources such as water. They are important not only for determining the forces driving global changes but also for understanding the general capabilities available to societies for adaptation. Projections of these factors for time periods such as the middle of the 21st century are at least as uncertain as projections of future climate; hence, it is probably most advisable to use such information as scenarios of change, or conditioning assumptions (IPCC, 1998). Moreover, culture exerts important influences on socioeconomic processes, problemsolving methods, and the like. The formation of coalitions, social movements, and educational programs directed toward changing institutional norms that might influence people’s behavior concerning climatic change is culturally determined, like other complex social and psychological processes. Cultural processes and economic behavior, for example, can be modeled to capture some of the complexity of the social processes, structures, and cognitive behavior involving culture (e.g., Rotmans and van Asselt, 1996; Koizumi and Lundstedt, 1998). Thus, it is simply impossible to predict with high confidence how societies and economies will develop in the future—hence the extent to which they will have the capacity needed for adaptation. The use of scenarios to assess driving forces and adaptive capacity is one way to explicitly acknowledge these kinds of structural uncertainties (e.g., IPCC, 2000). Socioeconomic scenarios, as already noted, are not predictions of future states of the world but consistent and plausible sets of assumptions about issues such as population growth, economic development, values, and institutions.

Although the emphasis on adaptation to reduce vulnerability and take advantage of emerging opportunities is increasing in impact assessment, many uncertainties remain regarding the effectiveness of different options, the relationship between adaptation to short-term climate fluctuations and long-term climate change, and constraints and opportunities that will be imposed by factors such as existing institutional structures, economic and financial limitations, and cultural resistance (IPCC, 1998).

Uncertainties are pervasive throughout climate change impact assessment. For some sectors, such as agriculture, uncertainty is large enough to prevent a highly confident assessment of even the sign of the impacts. Until a few years ago, uncertainties in assessments were so great that few researchers were willing to carry their analysis through to numerical estimates of monetary impacts. Even today, as the applicability of subjective probabilities is becoming more accepted, impact estimates with explicit confidence intervals are the exception rather than the rule (a few exceptions are Peck and Teisberg, 1992; Hope et al., 1993; Nordhaus, 1994a; Manne and Richels, 1995; Morgan and Dowlatabadi, 1996; Titus and Narayanan, 1996; Roughgarden and Schneider, 1999). Figure 2-2 (Moss and Schneider, 2000) graphically depicts how uncertainties in emissions scenarios feed into uncertainties in carbon cycle response, climate sensitivity, regional climate responses, and ranges of impacts in an “explosion” or “cascade” of widening uncertainty bounds. However, despite this daunting expansion of uncertainty, methods to classify and formally treat such uncertainties via subjective probability distributions are available in the literature (see Box 1-2 and Section 2.6) and can help to clarify which subcomponents of the overall human-environment system are most critical to integrated assessments of the costs and benefits of climatic changes or climate policies.

Efforts to deal with low-probability, potentially catastrophic events in integrated assessments of climate change are not well-represented in the literature. One possibility would be to treat these risks like any hazard and use methods from risk analysis: The value of the risk is the probability of occurrence multiplied by the consequences of the event. For rare and catastrophic possibilities, there is very little frequency data; thus, probabilities assessed are based largely on subjective methods (e.g., Nordhaus, 1994b; Roughgarden and Schneider, 1999). Equally important, under these conditions the expected cost estimate would be very sensitive to the analyst’s (subjective) assumptions about the costs of catastrophic events. Subjective probabilities can vary widely from analyst to analyst under such conditions. This partly explains why most analysts have been reluctant to include low-probability but potentially catastrophic events in integrated assessments (for a recent exception, see Mastrandrea and Schneider, 2001). However, absence of analysis does not necessarily imply absence of risk, and many risk management decisions in the private and public sectors are based on strategic hedging against low-probability but highly costly possibilities, such as insurance and deterrence (see Chapter 8). However, the expected cost approach would imply a risk neutrality—an uncomfortable position for those holding risk averse values in the face of possibilities such as collapse of the “conveyor belt” circulation in the North Atlantic Ocean (e.g., Broecker, 1997; Rahmstorf, 1999; Chapter 19) or melting of the West Antarctic Ice Sheet (e.g., Oppenheimer, 1998). Risk-averse individuals often worry about the possibility that a forecast for a high-consequence event is either accurate or an underestimate—the “type 2 error.” Such individuals have argued that a better way to treat the possibility of catastrophe is to ensure that all possible efforts are taken to avoid it—the “precautionary principle” (see, e.g., Wiener, 1995). However, spending valuable, limited resources to hedge against possible catastrophic outcomes with a low probability of occurring is infeasible in practice; scarce resources could have been used more productively elsewhere, including dealing with more probable climatic threats. People who are concerned about “squandering” resources on what they perceive to be unlikely threats or even an erroneous forecast—the “type 1 error”—often are engaged in contentious debates with those more concerned with type 2 errors—a situation that is well-known in risk management disciplines. Thus, it is difficult to apply the precautionary principle unambiguously to justify a hedging strategy against a potential catastrophic climatic event without also applying it to the possibility of negative outcomes from the hedging strategy itself, then weighing the relative risks of type 1 versus type 2 errors (Wiener, 1995).

Although much progress on valuation techniques is being made, as noted in Box 1-2, uncertainties are still large, and many impact estimates are “highly speculative” (Nordhaus and Boyer, 2000). Impacts can be divided into market and nonmarket impacts.

Market impacts occur in sectors or activities such as agriculture, forestry, provision of water, insurance against extreme events, transportation, tourism, and activities that use low-lying coastal land. Where these activities produce marketed goods, a monetary estimate of impacts (in units of dollars per °C, for example) sometimes can be made with fairly straightforward techniques, at least under present-day conditions; this has been the most common approach in impact studies to date (e.g., Mendelsohn et al., 2000). Market prices, adjusted to correct for market distortions (e.g., externalities), are the appropriate measure for unit impacts. Although the techniques are well established, the numbers obtained still are approximate as a result of all the uncertainties that surround impact assessments. Working out how the impacts will unfold in the distant future is much less straightforward. Impacts could increase as the intensity and scale of the activity increases (e.g., loss of coastal property) or decrease as more modern and robust systems replace existing ones (e.g., new crop strains are introduced with more climatic adaptability). Also, as noted in Box 1-2, impacts expressed in economic terms embed the values people attribute to the impacts across several numeraires, as well as the values of future generations (see Section 2.5.6 for further elaborations).

For example, the use of highly aggregated decision analysis frameworks (see Box 1-2 and Chapter 2) can be controversial because aggregation of positive and negative costs of even a limited number of market category sectors involves the arithmetic sum of many subelements that contain large uncertainties and are related to different regions. Furthermore, important market costs could be incurred by political instability (e.g., Kennedy et al., 1998), migration of displaced persons (e.g., Myers, 1993), diminished capacity of damaged ecosystems to provide accustomed services (e.g., Daily, 1997), or loss of heritage sites from sea-level rise (e.g., Schneider et al., 2000b). Moreover, losses in nonmonetary categories (i.e., other numeraires such as biodiversity lost, lives lost, quality of life degraded, or inequity generated—all per °C) are very controversial (e.g., Goulder and Kennedy, 1997, discuss attempts to estimate the intrinsic value of species). Any aggregation over such numeraires into a common metric—usually the dollar—cannot be accomplished transparently unless a variety of assumptions are explicitly given for the valuation of each of these numeraires before aggregation hides the underlying assumptions of how valuation was accomplished.

Aggregation of various damages into a single estimate sometimes is appropriate to provide policymakers with information about the magnitude of damages that can be expected on a global scale. However, as noted in Box 1-2, Section 1.5.5, and Section 2.6.4, there also is the risk that such aggregation conceals rather than highlights some of the critical issues and value-laden assumptions that are at stake.

As a hypothetical but concrete example, assume that climatic change would cause destruction of lives, ecosystems, and property in Bangladesh, corresponding to a loss of 80% of its GDP. This loss to Bangladesh would amount to roughly 0.1% of global GDP. If the global economy grows at 2% yr^-1, this assumed impact on Bangladesh would correspond to a delay in global income growth of less than 3 weeks. It is debatable whether adding, say, the possible benefits for temperate agriculture to the losses of lives resulting from sea-level rise in Bangladesh helps to assess the severity of climate change impacts because the “winner” does not compensate the “loser” (i.e., benefits for temperate agriculture offer little relief to those who have been affected by sea-level rise in other regions). Authors in the literature have expressed concern about trading the costs of emission reduction in some countries (e.g., more efficient end-use energy technologies) with large-scale losses of lives and human health in others (e.g., Munasinghe, 2000). Still, this is implicitly done in most conventional cost-benefit analyses of climate change available in the literature. As noted above and in Section 2.6.4, this points to the necessity of using appropriately disaggregated cost and benefit data to make the analysis more transparent. Possible ways of incorporating equity concerns include use of distributional weights in cost-benefit analysis (e.g., Azar and Sterner, 1996; Fankhauser et al., 1997; Azar, 1999).

Owing to the complexities of valuation and aggregation analyses described above and in the preceding subsection, the TAR authors are cautious about the applicability of single “optimal” answers. Instead, they attempt to examine ranges of outcomes calculated under a variety of assumptions available in the literature, for which alternative valuation methods can be applied to different categories across various numeraires.

IPCC Homepage