Beyond Cost-Benefit Analysis

Governmental policies often have important distributional consequences. Those who benefit from a policy may be different (in their welfare-relevant characteristics) from those who are made worse off. For example, the beneficiaries of a given policy may tend to have higher or lower income, to be in better or worse health, to have higher or lower life expectancy, or to live in more or less environmentally degraded surroundings, than the “losers.”

How should the policymaker take account of these sorts of distributional impacts? In choosing between the status quo and various regulatory alternatives, how should she decide which option is “fairest”? And if it turns out that fairness and efficiency point in different directions—that one option does best with respect to fair distribution, while a second maximizes overall well-being—how should the policymaker balance the two factors?

As I argue in a new book, Well-Being and Fair Distribution: Beyond Cost-Benefit Analysis, the “social welfare function” (SWF) provides a very plausible answer to these questions. The SWF framework is a systematic, normatively appealing methodology for incorporating distributional considerations into policy analysis.

Cost-benefit analysis (CBA), now the dominant policy-evaluation tool, is insensitive to distributive impacts—at least in its standard form.  Standard CBA ranks policies by summing monetary equivalents. If an individual benefits from a policy, as compared to the status quo, the monetary measure of her benefit is the amount she is willing to pay (WTP) in return for it; if she is harmed, the measure of her setback is the amount she is willing to accept (WTA) as compensation.   The sum of WTP amounts, minus the sum of WTA amounts, is then taken as an overall indicator of the attractiveness of the policy.

Note that an individual who is well-off with respect to income, health, or some other dimension of well-being could have a larger WTP or WTA amount for a given policy impact (e.g., a change in environmental quality) than an individual who is worse off. Note also that a pure redistribution of income from higher- to lower-income individuals (with no administrative costs, and no labor-disincentive effects) is seen by CBA as a neutral event—and yet many view a more equal distribution of income as fairer, at least if the income inequality thus corrected was not attributable to harder work and higher effort by those earning higher incomes.

By contrast, the SWF framework is sensitive to fairness considerations. This framework originates in theoretical welfare economics, and is currently used in various scholarly subfields within economics: optimal tax theory, growth theory, and some parts of environmental economics. While CBA quantifies welfare effects in dollars, the SWF framework measures well-being in utilities. A given policy corresponds to a particular pattern of utilities; policies are then ranked as a function of these.

Formally, the SWF framework maps a given outcome x onto a list of N utility numbers—denote this list as u(x). It then uses a function w(.) to rank these lists.   In other words, the SWF framework says: outcome x is at least as good as outcome y if and only if w(u(x)) ≥ w(u(y)).

To be sure, elaborating and implementing the SWF framework raises a host of difficult theoretical and practical issues—issues that require engagement with a wide range of scholarly literatures, including theoretical economics, applied economics, decision theory, moral philosophy, and law.

First, what is the basis for assigning “utility” numbers to different individuals based upon their various characteristics? The SWF framework requires not merely the assignment of such numbers but, indeed, that the utilities be interpersonally comparable. If Sally’s utility is larger than Lou’s, this means that Sally is better off than Lou.   Economists who use the SWF approach are comfortable with interpersonal utility comparisons, but the intellectual tradition that gave rise to CBA was skeptical of interpersonal comparability—and many economists retain such skepticism.

As I show in Well-Being and Fair Distribution, the concept of “extended preferences” (originating in the work of John Harsanyi) provides a firm basis for constructing an interpersonally comparable utility function.  Moreover, under normal conditions, we can estimate individuals’ “extended preferences” by looking to the ordinary preference data that are the bread and butter of applied economics. We can use behavior or surveys to infer individuals’ preferences over bundles of attributes such as health, income, leisure, or environmental goods—and thereby also to make inferences about their “extended preferences.”

Second, what particular social welfare function, w(.), should be used to compare different arrangements of individual utility? It is here that questions of fairness come to the fore. One possible w(.) is utilitarian. This SWF simply adds up individual utilities. But as I explain in Well-Being and Fair Distribution, utilitarianism is insufficiently sensitive to distributive impacts. A more attractive SWF is a “prioritarian” SWF: one that gives greater weight to utility changes affecting individuals at lower utility levels, as compared to individuals at higher utility levels. “Prioritarianism” is an ethical position that derives from the work of Derek Parfit, and has been much discussed in recent philosophical writing on equality. (Formally, a “prioritarian” SWF—sometimes referred to by economists as a “generalized utilitarian” SWF—is the sum of a concave transformation of individual utility.)

The contrast between a prioritarian SWF and traditional CBA is stark. Consider a pure income transfer (without administrative costs or disincentive effects), from higher- to lower-income individuals. Prioritarianism recommends such a transfer on two grounds: first, because the poorer individuals have higher marginal utility of money; and, second, because the poorer individuals are at a lower level of utility.

Many additional issues must be confronted in developing the SWF approach. How shall we calibrate the degree of priority given to those at lower utility levels? How does time figure into the analysis? &
nbsp; Does the fair distribution of well-being focus on individuals’ lifetime utility or their utility during some time-slice (for example, annual utility)?  How should the prioritarian SWF be applied under conditions of uncertainty (where the policymaker does not know for sure which outcome will result from a given policy, and so sees each as a probability distribution across outcomes)? Should the policymaker be concerned with the fair distribution of individuals’ expected utilities, or instead with the distribution of realized utility in each possible outcome?

These questions are undoubtedly difficult—but we can make progress addressing them, through careful, open-minded analysis that uses both the technical tools of economics, and the normative vocabulary of moral philosophy.

While this brief presentation has stressed the differences between the SWF approach and CBA, an important connection should be noted.   CBA with distributional weights sums individual WTP/WTA amounts that are adjusted by weighting factors.  This is a much less widespread version of CBA, but one that has been discussed by economists and employed by some governments.  It turns out that virtually any SWF—and certainly the prioritarian SWF—can be implemented via CBA with an appropriate set of weighting factors. Distributionally weighted CBA is controversial—in part because no clear basis has yet been furnished for the calculation of weights. The prioritarian SWF gives us the intellectual foundation for distributionally weighted CBA.