Oxford University Press's
Academic Insights for the Thinking World

Why cost-benefit analysis is flawed and how to improve it

Cost-benefit analysis is a key component of the US regulatory state. How it works and the function it plays in policymaking is not widely understood, however. Even the most substantive media outlets rarely discuss it. But cost-benefit analysis is a linchpin of the regulatory process. Its structure and role—and its flaws—should therefore be grist for an informed public conversation.

Any given proposed regulation would have various impacts on people’s  well-being. Consider, for example, a proposed rule that will reduce air pollution—the sort of rule that the Environmental Protection Agency issues. This rule will have at least three different types of effects. Breathing polluted air causes various diseases. These diseases, first, may increase the risk of premature death. Thus the anti-pollution rule, if enacted, will reduce people’s fatality risks. The diseases, second, will be associated with impaired health quality (e.g., pain, reduced mobility) on top of the increased death risk they may produce. Thus the anti-pollution rule will improve health quality. Finally, implementing the rule will require material resources—for example, firms may need to install costly anti-pollution devices—and these resource costs will ultimately show up in reduced income, for firms’ shareholders, employees, and/or consumers.

Cost-benefit analysis measures all of the impacts of a proposed regulation on a monetary scale. We convert impacts into monetary equivalents, by asking what people are willing to pay (if they are made better off) or willing to accept (if made worse off). Consider the anti-pollution rule. The monetary equivalent for a reduction in someone’s fatality risk is what she is willing to pay for that reduction, as estimated by various sources of evidence (market prices of safety devices, surveys). The monetary equivalent for a health improvement is, again, what the person is willing to pay for that improvement. Finally, converting income losses into dollars is automatic. The monetary equivalent for a $100 loss of income is just −$100.

Cost-benefit analysis now says that a proposed regulation is worth enacting if the regulation’s sum total of monetary equivalents is positive. In the case of the anti-pollution rule, this means that the rule’s total benefits, as measured on a dollar scale, are larger than its total costs in loss income.

For nearly the last forty years, all federal agencies in the Executive Branch have been required, by executive order, to employ cost-benefit analysis in considering proposed regulations, and to submit these regulations and the accompanying cost-benefit documents for review by a powerful oversight body within the Office of Management and Budget.

Cost-benefit analysis has a wider role than even this. In the US government, cost-benefit analysis is principally used as a tool for evaluating regulation (legal rules). But its methodology is applicable to any type of governmental policy. For example, we can assess a proposed infrastructure project by comparing the monetized benefits of the project (e.g., better transportation) to the monetized costs. Cost-benefit analysis is important to the policymaking process in a number of countries, not just the U.S. And it has given rise to a vast body of academic literature. But cost-benefit is flawed. We can improve on it.

The problem is that money is an imperfect metric of well-being. Money has diminishing marginal well-being impact. The greater someone’s income, the smaller the effect of a given monetary change on that person’s well-being. For example, a $1,000 increase in the income of someone earning $10,000 a year makes a much bigger difference to her well-being than a $1,000 increase in the income of someone earning $100,000—in turn a bigger difference than a $1,000 increase in the income of someone earning $1 million.

How does this infect cost-benefit analysis? If Casey and Dalia experience the very same well-being impact as the result of some policy, but Casey has more income than Dalia, Casey’s monetary equivalent for the policy will be larger than Dalia’s. Conversely, if Casey experiences a smaller well-being impact than Dalia, Casey’s monetary equivalent may still be larger than or equal to Dalia’s.

Government tries to circumvent this feature of cost-benefit analysis by using population-average monetary equivalents. Thus, for, example, the Environmental Protection Agency converts fatality risk reduction into dollars by ascertaining what people, on average, are willing to pay for risk reduction. But this averaging can have perverse consequences. Consider the case in which a poor person not only receives various non-income benefits from a policy (risk reduction, health improvement, etc.), but also has to pay for those benefits in reduced income—so that, on balance, he is worse off from  the policy. Cost-benefit analysis using population-average monetary equivalents might indicate the opposite.

It’s possible to improve upon cost-benefit analysis by using social welfare functions to evaluate governmental policy. The social-welfare-function framework is already used in some areas of economic scholarship, although not (yet) in governmental practice.

This methodology measures well-being with a utility scale rather than a dollar scale. A utility function is a mathematical device that reflects someone’s preferences. If Xavier prefers one bundle of goods to a second, and therefore is better off with the first bundle, his utility function will assign it a larger number.

Utility generally increases as income increases, but at a decreasing rate. Assume that Alice and Bob have the same preferences, and thus the same utility function. However, Alice is poorer than Bob. Increasing Alice’s income would translate into a larger utility change than growing Bob’s income by the same amount. Further, the utility approach to measuring well-being is better because it is based on people’s actual preferences.

The social-welfare-function framework is quite flexible in how it aggregates utilities across persons. The framework can simply sum up individual utilities. This is a “utilitarian” approach; Jeremy Bentham’s famous idea of utilitarianism is implemented, in the social-welfare-function methodology, via a straight summation of utility numbers. However, it is also possible to employ a “prioritarian” social welfare function, which accords greater weight to the utility of people at a lower utility level.

In a recent analysis, I compared cost-benefit analysis to the social-welfare-function approach using a model of risk regulation based upon actual US data. Individuals of different ages are modelled as facing “lotteries” over life-histories. A given life-history is some lifespan (for example, living to the age of 75), with a certain amount of income each year alive. The effect of a risk-regulation policy is to reduce individuals’ current fatality risk, and thereby increase their chances of life-histories with longer lifespans; but also to reduce the amount of income they earn if alive. For purposes of the social-welfare-function methodology, income is converted into “utility” with a logarithmic utility function—a very standard utility function in economic scholarship.

My analysis finds that the utilitarian social welfare function is somewhat biased towards the rich. Reducing a rich person’s fatality risk is accorded greater social value than reducing a poor person’s fatality risk. However, cost-benefit analysis is shown to be much more biased toward the rich than utilitarianism. Moreover, the utilitarian bias can be avoided by shifting to a prioritarian social welfare function. Here, I find a preference for risk reduction among the poor. Cost-benefit analysis with population-average monetary equivalents avoids the problem of being biased towards the rich, but has a different problem: it accords the same value to risk reduction independent of age. For example, risk reduction for a 20-year-old is assigned exactly the same value as risk reduction for a 60-year-old. By contrast, both utilitarian and prioritarian social welfare functions prefer to allocate risk reduction to the young. This seems more appropriate, ethically; someone who dies at age 20 can be expected to lose more years of life than someone who dies at age 60.

In short, we can improve on cost-benefit analysis, and the social-welfare-function framework shows how.

Featured Image Credits: by Helloquence on Unsplash

Recent Comments

  1. Oliv

    Thinking about the utility functions of the HS2 railway project…whom will it benefit the most? *I wonder*.

Leave a Comment

Your email address will not be published. Required fields are marked *