Climate Change 2001:
Working Group II: Impacts, Adaptation and Vulnerability
Other reports in this collection

2.5.5. Equity and Distribution

Assessments of the impacts of alternative climate change scenarios require assessments of their impacts on different groups, societies, nations, and even species. Indeed, this report reveals that many sectors and/or regions are at greater risk to climate change than others. This section addresses this need. Interpersonal Comparisons

First principles of economic theory offer two approaches for comparing situations in which different people are affected differently. In the first—the utilitarian approach attributed to Bentham (1822) and expanded by Mills (1861)—a situation in which the sum of all individual utilities is larger is preferred. Because Bentham's view of utility reflected "pleasure" and "pain," this approach embraces the "greatest happiness principle." Many objections have been raised against it, however, primarily because the whole notion of interpersonal comparisons of utility is problematic. Indeed, Arrow (1951 and 1963) objected strenuously in arguing that "interpersonal comparisons in the measurement of utilities has no meaning and, in fact, there is no meaning relevant to welfare comparisons in the measurability of individual utility." For example, it is impossible to compare the pleasure that a person receives from listening to a concert with what another gets from watching a dance. Second, maximizing the sum total of utility, if it were possible, would require that the marginal utilities of all individuals be equal. But this would say nothing about the level of utility for each individual. They could be quite different, so the utilitarian rule is insensitive to distributional issues except in the special case in which all individuals have identical utility functions.

These difficulties led to the development of a second approach—the welfarist approach, in which a social welfare function of individual utilities is postulated. Utilitarianism is thus a special case in which the social welfare function is simply the sum of individual utilities. There are other options, of course. The Gandhian principle, for example, can support a function that judges every possible action on the basis of its impact on the poorest of the poor.

It also is possible to compare two situations without defining an explicit social welfare function and without making interpersonal comparisons of individual utilities. The Pareto principle offers one method, by which one judges any situation better than another if at least one person is better off and no one else is worse off. A partial social ordering with which unambiguous comparisons can be made in some (but not all) cases can be constructed from the Pareto principle if cardinal utilities can be added across individuals, if society accepts the principle of anonymity (i.e., only the distribution matters, not which particular person is in a particular place), and if there is an aversion to regressive transfers (i.e., transfers from the poor to the rich). To see how, consider two situations, X and Y. Assume that there are n individuals ordered from poorest to richest. Let them have incomes (or utilities) {X1, ..., Xn} and {Y1, ..., Yn} in X and Y, respectively. X can be deemed preferable to Y if X1 Y1, [X1 + X2] [Y1 + Y2], and so on through [X1 +...+ Xn] [Y1 +...+ Yn], with at least one strict inequality holding. Note that showing that X is not preferred to Y is not sufficient to show that Y is preferred to X

Rothschild and Stiglitz (1973) took these notions further by showing three alternative but equivalent ways of comparing distributions X and Y. They concluded that X would be preferred to Y if all of the following obtain:

Note, in passing, that Lorenz curves simply plot the percentage of income received by various percentiles of populations when they are ordered from least to greatest. Rothschild and Stiglitz (1973) also point out, however, that these measures apply only to a one-good economy. This requirement is equivalent to assuming that income is desired by all individuals and there are no externalities; the implications of more than one good are "substantial."

None of these measures speaks to estimating the cost of inequity when comparisons can be made. But just as insurance can be used as a utility-based measure of the cost of uncertainty, similarly constructed estimates that are based on social welfare functions that display aversion to inequality can be constructed. Insurance premiums computed in these cases simply represent a measure of what society would willingly pay to eliminate inequality. Such an approach assumes the possibility of defining an international social welfare function. Let us now look at the difficulties involved in defining it.

Other reports in this collection