1. The virtues of flawed cost-benefit analyses

In most general terms, cost-benefit analysis is a reasoning strategy that

permits us to estimate the desirability of various tradeoffs available to us.

The most familiar form of cost-benefit analysis places a dollar value on the

costs and benefits of each available option. This procedure is supposed to

allow any person or institutional body to identify the option that promises

the greatest total benefit. For example, in deciding whether to expand a

successful company’s production capacity, cost-benefit analysis would

have us compare the expected profits to be gained by expansion with current

expected profits. The objection to this standard approach to costbenefit

analysis is as obvious as it is serious: It relies on money as its only

measure of value. A deeper and more interesting objection is that there is no ‘‘neutral’’ measure of value for comparing very distinct sorts of goods.

Values are incommensurable. We cannot reduce all value to money, and

there is no realistic way to assign commensurable units of value to freedom,

happiness, personal security, or a (relatively) pristine Grand Canyon.

These objections show that it is not possible to measure the net benefits

(or costs) of different options against each other. We are sympathetic to

many of these objections to cost-benefit analysis (Anderson 1993, Sen

2000). Still, we would argue that flawed cost-benefit analyses can be very

useful and important.

A cost-benefit approach to a problem can be very useful even if it is

plagued by theoretical problems. It permits us to estimate the desirability

of various tradeoffs available to us, no matter what kind of value we place

on them. In doing so, it can be useful for helping us to avoid making decisions

that do not reflect our own values. Explicitly engaging in costbenefit

analysis, even a flawed analysis, allows us to slow down, cool off,

compare the value we assign to certain outcomes, and determine what

strategies we ought to adopt to achieve them. The primary virtue of costbenefit

analyses, even when flawed, is that they can help us to set and adjust

our priorities in ways that better reflect our values.

We can distinguish between two kinds of flawed but useful costbenefit

approaches to problems, incomplete cost-benefit analyses and

unreduced cost-benefit analyses. An incomplete cost-benefit analysis focuses

on only a subset of the values at stake in a decision. For example,

cost-benefit analyses that focus only on money as a unit of value are typically

incomplete. An unreduced cost-benefit analysis focuses on a realistic

set of values that are at stake in making a decision, but it does not attempt

to reduce those values to standard units of costs and benefits. To make our

case that these sorts of flawed cost-benefit analyses can be useful in helping

us to set and reset our priorities, let’s consider some examples.

Many companies offer their employees (sometimes quite generous)

retirement packages; but a surprising percentage of people don’t participate

in them. Many of these people would benefit from an unreduced

cost-benefit analysis of the choice of whether to invest in such a retirement

account. Such an analysis would not even attempt to reduce all the costs

and benefits of the options to a single value (like current dollars). Consider

the benefits of current spending (on, say, a larger mortgage on a larger

home) and the potential future costs of inadequate retirement funds

(lower standard of living, dependence on kin or children). While these

costs and benefits cannot be reduced to a single unit of value, we suspect

that if people thought clearly and coolly about the various options, they might be surprised to find that they would explicitly reject the values

implicit in their actual decisions.

Let’s consider an example of some radically incomplete cost-benefit

analyses that as a historical matter affected public-policy decision making

for the better: attempts to put a monetary value on human lives. In late

seventeenth-century Britain, William Petty ‘‘argued in favor of measures

to ward off or mitigate the effects of the plague on the grounds that by

spending a few thousand pounds the king might protect an investment

many times greater in the lives of his subjects’’ (Porter 1994, 214–15). By

the early part of the twentieth century, professional actuaries working for

insurance companies had taken over the business of placing monetary

values on human lives. In The Money Value of a Man (1930), Louis Dublin

and Alfred Lotka attached a dollar figure to human lives, basing their calculations

on the expected net present value of future earnings. Their analyses

explicitly recognized that they could not put a monetary figure on the

‘‘intangibles of life’’ but instead viewed humans (and in particular, men)

as just so much equipment:

Man has much in common with the industrial aids, machines, manufacturing

plants, and so forth, of which he makes use to conduct the business of

life. Like them he has a ‘‘cost of installation,’’ . . . running expenses, interest

on capital invested, and the loss of a certain proportion of children that do

not live to attain adult age, just as in manufacturing processes allowance

must be made for losses by ‘‘spoilage’’ of material that never reaches the

‘‘finished’’ stage. (1930, 44, quoted in Porter 1994, 216)

This kind of cost-benefit analysis inevitably strikes us as appallingly crass.

But besides figuring out how much life insurance to sell people, these

analyses were also typically put to political use:

None of these calculations . . . was part of a formal assessment of costs and

benefits. They were intended not to guide policy in detail, but to soften

slightly those famously hardheaded administrators and businessmen who

were reluctant to invest dollars, pounds, and francs in what would come to

be known as human capital. The sentimental and self-regarding emotions

could be left out because the numbers were already large enough to draw

attention, and that was their only real purpose . . . [W]hile these formulas

may not have been very accurate, they were surprisingly precise. . . . This was

almost indispensable if calculation was to figure significantly in decisions

vulnerable to public criticism. (Porter, 217)

The cost-benefit analyses suggested that the preventable suffering and

death of poor laborers was bad for business. We might view it as wearily predictable that it was profit that spurred captains of industry to support

the public works projects that prevented the deaths of many of our

ancestors. And one would be right to object that the cost-benefit approach

is no guarantor of a positive outcome. The morally right policies were

implemented because of contingent facts about the profit structure of local

industrial economies rather than because of the intrinsic moral worth of

human laborers. But while we may be appalled that ethical considerations

did not move the captains of industry quicker, we can be glad that costbenefit

considerations kept them from moving slower still.

The cost-benefit analyses that convinced ‘‘hardheaded administrators

and businessmen’’ that they were underestimating the monetary value of

their ‘‘human capital’’ was radically incomplete. It focused only on economic

considerations. But the right course of action was so clearly right

that any number of different cost-benefit analyses—even those informed

by values we might find crass or repugnant—delivered the correct result.

When the world is obliging in this way, incomplete cost-benefit analyses

can lead us to effectively revise our priorities and actions.

Flawed cost-benefit analyses are useful because they dramatize the

opportunity costs of a favored course of action. For example, after airplane

crashes, there are often voices pressing for greater air safety regulations.

While such regulations might make air travel safer, they might also have the

effect of making travel more dangerous. That’s because the increased regulations

may increase the price of flying and so lead more people to drive—

which is a more dangerous form of transportation (Sunstein 2000, 1073).

(Further, given its relative safety, making air travel a tiny bit safer may well

be less cost effective than, say, repaving the Pennsylvania Turnpike.) Now it

may be that we are willing to accept a greater number of total traffic injuries

and fatalities in exchange for slightly safer (and more expensive) air travel.

But by engaging in cost-benefit analysis—even one that only considers the

rather crass value of money spent per death prevented—we can recognize

that the decision to make air travel slightly safer has this effect. Even a

flawed cost-benefit analysis can help us to expose the aims and values

implicit in our decisions; and this exposure can lead us to change, challenge,

or clarify our values and priorities. Perhaps we so dread the terrifying

nature of airplane deaths that we are willing to accept an increased probability

of injury or death for a decreased probability of a terror-filled death.

But only by exposing this value underlying our decision can we even begin

to question it. Even a flawed cost-benefit analysis can help us set priorities

and effectively marshal our scarce resources so as to improve our decision

making (Sunstein 2000).

As citizens of a democracy, we have an interest in how members of

governmental safety boards (e.g., OSHA, EPA, etc.) decide how to allocate

scarce resources to improve occupational and other federal safety standards.

In making such decisions, it is useful to know that (according to the

Consumer Product Safety Commission) banning unvented space heaters

is among the cheapest ways to save lives, costing $100,000 per prevented

death. Limiting exposure to asbestos in the workplace is considerably

more expensive, at 8.3 million dollars per prevented death (OSHA). And

the cost of listing wood-preserving chemicals as hazardous waste is estimated

by the EPA to be 5.7 trillion dollars per prevented death (Sunstein,

2000). By being clear about what measures are likely to save the greatest

number of people (or minimize the number of people getting sick), we can

better make decisions that accurately reflect our values.

In most general terms, cost-benefit analysis is a reasoning strategy that

permits us to estimate the desirability of various tradeoffs available to us.

The most familiar form of cost-benefit analysis places a dollar value on the

costs and benefits of each available option. This procedure is supposed to

allow any person or institutional body to identify the option that promises

the greatest total benefit. For example, in deciding whether to expand a

successful company’s production capacity, cost-benefit analysis would

have us compare the expected profits to be gained by expansion with current

expected profits. The objection to this standard approach to costbenefit

analysis is as obvious as it is serious: It relies on money as its only

measure of value. A deeper and more interesting objection is that there is no ‘‘neutral’’ measure of value for comparing very distinct sorts of goods.

Values are incommensurable. We cannot reduce all value to money, and

there is no realistic way to assign commensurable units of value to freedom,

happiness, personal security, or a (relatively) pristine Grand Canyon.

These objections show that it is not possible to measure the net benefits

(or costs) of different options against each other. We are sympathetic to

many of these objections to cost-benefit analysis (Anderson 1993, Sen

2000). Still, we would argue that flawed cost-benefit analyses can be very

useful and important.

A cost-benefit approach to a problem can be very useful even if it is

plagued by theoretical problems. It permits us to estimate the desirability

of various tradeoffs available to us, no matter what kind of value we place

on them. In doing so, it can be useful for helping us to avoid making decisions

that do not reflect our own values. Explicitly engaging in costbenefit

analysis, even a flawed analysis, allows us to slow down, cool off,

compare the value we assign to certain outcomes, and determine what

strategies we ought to adopt to achieve them. The primary virtue of costbenefit

analyses, even when flawed, is that they can help us to set and adjust

our priorities in ways that better reflect our values.

We can distinguish between two kinds of flawed but useful costbenefit

approaches to problems, incomplete cost-benefit analyses and

unreduced cost-benefit analyses. An incomplete cost-benefit analysis focuses

on only a subset of the values at stake in a decision. For example,

cost-benefit analyses that focus only on money as a unit of value are typically

incomplete. An unreduced cost-benefit analysis focuses on a realistic

set of values that are at stake in making a decision, but it does not attempt

to reduce those values to standard units of costs and benefits. To make our

case that these sorts of flawed cost-benefit analyses can be useful in helping

us to set and reset our priorities, let’s consider some examples.

Many companies offer their employees (sometimes quite generous)

retirement packages; but a surprising percentage of people don’t participate

in them. Many of these people would benefit from an unreduced

cost-benefit analysis of the choice of whether to invest in such a retirement

account. Such an analysis would not even attempt to reduce all the costs

and benefits of the options to a single value (like current dollars). Consider

the benefits of current spending (on, say, a larger mortgage on a larger

home) and the potential future costs of inadequate retirement funds

(lower standard of living, dependence on kin or children). While these

costs and benefits cannot be reduced to a single unit of value, we suspect

that if people thought clearly and coolly about the various options, they might be surprised to find that they would explicitly reject the values

implicit in their actual decisions.

Let’s consider an example of some radically incomplete cost-benefit

analyses that as a historical matter affected public-policy decision making

for the better: attempts to put a monetary value on human lives. In late

seventeenth-century Britain, William Petty ‘‘argued in favor of measures

to ward off or mitigate the effects of the plague on the grounds that by

spending a few thousand pounds the king might protect an investment

many times greater in the lives of his subjects’’ (Porter 1994, 214–15). By

the early part of the twentieth century, professional actuaries working for

insurance companies had taken over the business of placing monetary

values on human lives. In The Money Value of a Man (1930), Louis Dublin

and Alfred Lotka attached a dollar figure to human lives, basing their calculations

on the expected net present value of future earnings. Their analyses

explicitly recognized that they could not put a monetary figure on the

‘‘intangibles of life’’ but instead viewed humans (and in particular, men)

as just so much equipment:

Man has much in common with the industrial aids, machines, manufacturing

plants, and so forth, of which he makes use to conduct the business of

life. Like them he has a ‘‘cost of installation,’’ . . . running expenses, interest

on capital invested, and the loss of a certain proportion of children that do

not live to attain adult age, just as in manufacturing processes allowance

must be made for losses by ‘‘spoilage’’ of material that never reaches the

‘‘finished’’ stage. (1930, 44, quoted in Porter 1994, 216)

This kind of cost-benefit analysis inevitably strikes us as appallingly crass.

But besides figuring out how much life insurance to sell people, these

analyses were also typically put to political use:

None of these calculations . . . was part of a formal assessment of costs and

benefits. They were intended not to guide policy in detail, but to soften

slightly those famously hardheaded administrators and businessmen who

were reluctant to invest dollars, pounds, and francs in what would come to

be known as human capital. The sentimental and self-regarding emotions

could be left out because the numbers were already large enough to draw

attention, and that was their only real purpose . . . [W]hile these formulas

may not have been very accurate, they were surprisingly precise. . . . This was

almost indispensable if calculation was to figure significantly in decisions

vulnerable to public criticism. (Porter, 217)

The cost-benefit analyses suggested that the preventable suffering and

death of poor laborers was bad for business. We might view it as wearily predictable that it was profit that spurred captains of industry to support

the public works projects that prevented the deaths of many of our

ancestors. And one would be right to object that the cost-benefit approach

is no guarantor of a positive outcome. The morally right policies were

implemented because of contingent facts about the profit structure of local

industrial economies rather than because of the intrinsic moral worth of

human laborers. But while we may be appalled that ethical considerations

did not move the captains of industry quicker, we can be glad that costbenefit

considerations kept them from moving slower still.

The cost-benefit analyses that convinced ‘‘hardheaded administrators

and businessmen’’ that they were underestimating the monetary value of

their ‘‘human capital’’ was radically incomplete. It focused only on economic

considerations. But the right course of action was so clearly right

that any number of different cost-benefit analyses—even those informed

by values we might find crass or repugnant—delivered the correct result.

When the world is obliging in this way, incomplete cost-benefit analyses

can lead us to effectively revise our priorities and actions.

Flawed cost-benefit analyses are useful because they dramatize the

opportunity costs of a favored course of action. For example, after airplane

crashes, there are often voices pressing for greater air safety regulations.

While such regulations might make air travel safer, they might also have the

effect of making travel more dangerous. That’s because the increased regulations

may increase the price of flying and so lead more people to drive—

which is a more dangerous form of transportation (Sunstein 2000, 1073).

(Further, given its relative safety, making air travel a tiny bit safer may well

be less cost effective than, say, repaving the Pennsylvania Turnpike.) Now it

may be that we are willing to accept a greater number of total traffic injuries

and fatalities in exchange for slightly safer (and more expensive) air travel.

But by engaging in cost-benefit analysis—even one that only considers the

rather crass value of money spent per death prevented—we can recognize

that the decision to make air travel slightly safer has this effect. Even a

flawed cost-benefit analysis can help us to expose the aims and values

implicit in our decisions; and this exposure can lead us to change, challenge,

or clarify our values and priorities. Perhaps we so dread the terrifying

nature of airplane deaths that we are willing to accept an increased probability

of injury or death for a decreased probability of a terror-filled death.

But only by exposing this value underlying our decision can we even begin

to question it. Even a flawed cost-benefit analysis can help us set priorities

and effectively marshal our scarce resources so as to improve our decision

making (Sunstein 2000).

As citizens of a democracy, we have an interest in how members of

governmental safety boards (e.g., OSHA, EPA, etc.) decide how to allocate

scarce resources to improve occupational and other federal safety standards.

In making such decisions, it is useful to know that (according to the

Consumer Product Safety Commission) banning unvented space heaters

is among the cheapest ways to save lives, costing $100,000 per prevented

death. Limiting exposure to asbestos in the workplace is considerably

more expensive, at 8.3 million dollars per prevented death (OSHA). And

the cost of listing wood-preserving chemicals as hazardous waste is estimated

by the EPA to be 5.7 trillion dollars per prevented death (Sunstein,

2000). By being clear about what measures are likely to save the greatest

number of people (or minimize the number of people getting sick), we can

better make decisions that accurately reflect our values.