I hope we can all agree on the following basic, general assessment of the merits and limitations of cost-benefit analysis (CBA).  I’m not that well informed about CBA issues, but this all makes sense to me.

Limitations of CBA

1. CBA is not scientifically objective.  Its conclusions are not “knowledge,” though it can provide us with knowledge, such as by revealing certain trade-offs or valuations that were formerly implicit in a regulatory scheme.  CBA depends on value judgments, because values determine what counts as a cost or a benefit, as well as how costs and benefits should be quantified.  So CBA is not “self-justifying,” though I think its justifications are extremely powerful.

2. CBA should not necessarily be used to determine how a law or regulation should be enforced.  I’m not sure how exactly we should go about deciding when CBA is acceptable or required, but certain laws and regulations are clearly not based on consequentialist reasoning and therefore should not be enforced at the direction of a consequentialist tool.  The law is not a thoroughly consequentialist enterprise.  For example, the Americans With Disabilities Act contains certain “reasonable accommodations” requirements.  It may be the case (I’m not sure about this) that the reasonableness of requiring an accommodation is clearly not meant to be determined based on CBA, but rather should be based on a common-sense, reasonable-person-type analysis, even if this results in the provision of inefficient accommodations that are theoretically suboptimal.  (For the avoidance of doubt, this is not to suggest that the ADA is a bad law: it may be the case that it was the only politically viable option and is preferable to having done nothing or to having waited for more favorable political circumstances.  I wouldn’t know.)

In addition, there are probably situations in which CBA is not worth the candle, in that the cost of engaging in CBA is higher than the marginal expected value it would provide compared to a readily available alternative, such as a common-sense, bright-line rule.

3. CBA isn’t easy.  It is difficult to quantify costs and benefits in a way that is both consistent and in accordance with our values.  For example, when should people’s willingness to pay to avoid a cost be used as a basis for quantifying that cost?  If some sort of average willingness-to-pay is used, how should it be calculated?  (Which people should count towards the average, etc.?)  When should cost-of-mitigation be used instead?  Each of these measures is an imperfect proxy, and each is imperfect in different ways.

4. Relatedly, CBA is not the be-all, end-all of regulatory decisions.  It only tells us what the overall costs and benefits of a proposed policy are – though the quantification of costs and benefits may be, and perhaps should be, influenced by distributive and positional concerns.  It doesn’t tell us how exactly to implement the policy.  And perhaps the potential means of implementation should in some cases influence how the costs and benefits are calculated – for example, if the policy decision in question is where to place a garbage-processing facility, perhaps separate CBA calculations should be done with respect to the differently situated residents of the particular neighborhoods under consideration.

5. CBA may be more vulnerable to political abuse than “command-and-control” alternatives.  I am highly skeptical of this view, but the argument is that courts have much more discretion to strike down CBA-based regulations than regulations that are specified in the governing statute itself, and courts are highly politicized; therefore, CBA opens the door to effective regulation by corrupt quasi-dictators.  I am skeptical of this, because (i) courts’ corruption is decently mitigated by various factors, including the need to publicly justify most decisions, whereas legislative corruption is out of control, tends to happen behind closed doors, and often comes into play when regulated companies and industries lobby for better treatment, and (ii) judicial review of CBA-based regulations is a fairly formalistic exercise that is not designed to facilitate a court’s substitution of its own CBA judgments for those of the regulator, but rather is designed to ensure that the regulator properly engaged in the process of CBA, so it isn’t that easy for courts to deviously reach politically motivated conclusions (cf. Scalia doesn’t always vote for the result he wants, because his political capital is limited, and I believe he is known for deviating from his politics more so when it comes to technocratic cases, where its harder to handwave and the political stakes are not particularly high, at least compared to hot-button cases such as those involving gay rights and universal health care); moreover, the very structure of judicial review of administrative decisions limits courts’ ability to usurp regulators’ authority, because courts can’t affirmatively make regulations, but rather can only strike them down and direct the agency to try again in accordance with certain procedural requirements.  I’m not saying courts are great; I’m saying Congress sucks more and has more regulatory power.

Also, keep in mind the level at which this political abuse argument operates: it says that a given regulation is more abusable by courts if it is the result of administrative CBA as opposed to legislative specification; this ignores the fact that CBA-based regulations tend to be much better than command-and-control-based ones, so even if courts are more corrupt than Congress, if they can’t always reach the results they want, we're likely better off under a CBA regime.

Merits of CBA

Cass Sunstein has made a career of setting forth the merits of CBA, and I’m not that familiar with his work, but the merits of CBA that are readily apparent to me seem like more than enough to recommend its widespread use.  I’ll keep this section simple.

1. On balance, CBA is unambiguously superior to any other means of consequentialist decision-making.  This point smacks of tautology – consequentialism is inherently concerned with comparing costs and benefits, broadly speaking – but it is important to keep in mind.  Many, if not most, policy decisions are based on consequentialist reasoning, albeit of a form that is not explicit, systematic, and transparent about the inevitable trade-offs at stake.  CBA minimizes these defects to great avail.  In short, by increasing decision-makers' awareness and understanding of trade-offs, CBA enables decision-makers to optimize more effectively.  Actual CBA is preferable to handwavy quasi-CBA.  As I put it in another forum:
I think the obvious defense of explicit, aboveboard weighing, however speculative, is that there are some valuations that people would reject, and the only way to ensure that we do not implicitly use such valuations is to be clear about our calculations.  Also, I think there is something to be said for crude quantifications and modeling in some cases.  For example, if everyone agrees that several items being compared are valuable essentially because they each possess the same four qualities, it might be a good idea to quantify the degree to which they do so and/or to rank the importance of these qualities in one’s value calculus, in order to avoid acting inconsistently with one’s actual preferences due to cognitive shortcomings.


Unrigorous explicit valuations may irritate you, but the alternative irritates me – it reminds me of this time in my environmental policy seminar when a student earnestly argued that certain types of environmental values (beauty, biodiversity, and the like) are incommensurable with the other values that one should factor into environmental policy decisions.  The class quickly came around to what I regard as the obviously sensible view that any decision implicitly places a value on these “incommensurables,” and that it would be better to crudely quantify them, thereby crudely maximizing people’s preferences, rather than affording them some indeterminate, unrankable value.


I certainly think something should change if a regulation in one industry implicitly valued a human life at $1 million and a regulation in another industry did so at $10 million.  Similarly, however we value human life, I think we should do so consistently in different regulatory contexts, so I think it’s worth examining where we can get the most bang for our buck.


Modeling isn’t inherently that costly (for an extreme example, imagine a model that simply ordinally ranked various locales in terms of biodiversity and beauty and weighted these factors 2 to 1), and there are clear benefits to using quantities in cost-benefit calculations.  So I think it should be a rare case where we reject quantitative modeling entirely.
2. CBA is superior to non-consequentialist decision-making in the overwhelming majority of regulatory cases.  This claim obviously implicates a sprawling, age-old debate that I have no desire to get into here.  But I will make the point that I suspect “the will of the people” with respect to the vast majority of regulatory decisions is something vaguely utilitarian.  Regulation tends to be about issues such as safety, which do not, at least not in a straightforward way, implicate deontological concerns.  People may say things like “each human life is infinitely valuable,” or “every sperm is sacred,” but they don’t actually believe these things, and it would obviously be suicidal, if not impossible, to make policies that are in accordance with such statements.  What people actually want is a certain amount of safety that varies with the relevant costs and benefits (which is not to suggest that people’s purported safety desires or revealed preferences should always be dispositive of safety policy; for example, people can trade off cost and safety to some extent in deciding which car to buy, but they can’t meaningfully do so in deciding which flight to take, and this may very well be a good thing).  So I think CBA is clearly preferable to non-consequentialist regulatory decision-making, which ignores the utilitarian trade-offs that people care about.  I think examples of non-consequentialist regulatory laws, such as my ADA discussion above, are exceptional and unrepresentative of the state's regulatory apparatus.


CBA isn’t perfect.  It isn’t scientifically objective.  It isn’t free of vexing value-judgment calls.  But it’s the best tool for the job.  And it doesn’t inherently purport to be anything more!  It doesn’t inherently purport to “generate objective assessments.”  It isn’t inherently “basically a set of policy preferences dressed up as science.”  It is only such things according to its most overzealous defenders.

I recognize that it’s dangerous and irritating to “fetishize” flawed methodologies such as CBA.  CBA should not be placed on a pedestal with the scientific method (which I do think belongs on a pedestal even if most of our attempts at science fail to do it justice).  But it should indeed be regarded as the best system we have for making regulatory decisions.


  1. Point 1 under "merits" seems unsupported. What if I'm a hardcore wealth-maximizer? Then won't I prefer a system that maximizes wealth, rather than whatever CBA maximizes? What if I don't weigh all people's utility equally (which CBA purports to do)? (For instance, if I could administer a drug to prisoners that would make the remainder of their sentences feel like pure bliss, CBA would say to do it, but common sense would say not to.) What if I'm pro-life and CBA doesn't put any value on fetal life? Or what if I'm pro-choice and CBA attributes the same value to fetal life as to all other human life?

    If you can prove this point, then I think you win, but I don't think you can prove this point.

  2. (Needless to say point 2 under "merits" is similarly unsupported and suffers from the same defect.)

  3. In fact, imagine that a pro-life country like Ireland implements a modification of CBA (let's call it "ICBA") in which fetal life is given the same weight as all other human life. And let's say ICBA comes to different conclusions on a wide variety of issues, some of them seemingly unrelated to fetal health (air pollution, seatbelts, that kind of thing). Let's call CBA as practiced in the US "USCBA."

    Now, we know from Alan that "CBA is unambiguously superior to any other means of consequentialist decision-making." And so we know that either ICBA or USCBA isn't really CBA, since at least one of them is suboptimal. But which one is it? Do we have a way of identifying the "true" CBA and ferreting out the impostors? If not, how do we know when we have achieved true CBA? What if USCBA is so far from "true" CBA that we would be better off flipping a coin?

  4. I think we have a misunderstanding about the substantive commitments inherent in CBA. My understanding is they are minimal. I conceive of CBA as a process that involves identifying all relevant costs and benefits, quantifying them in monetary terms, applying a discount rate in order to translate all quantities into net present values, and then comparing the total costs and total benefits. Accordingly, CBA can properly be used by decision-makers with very different substantive values to reach very different conclusions: it can be used by wealth-maximizers to determine whether a proposed policy would be wealth-maximizing; it can be used by those who value fetal life as human life; it can be used by those who value the cost of noise pollution based on people's willingness to pay to avoid it; etc. CBA does not inherently purport to weigh all people's utility equally, to place a certain value on fetal life, or to make any other such substantive value commitments. Thus, my point is simply that whatever conception of utility a consequentialist policy-maker has, he is generally better off using CBA to determine whether a proposed policy would be utility maximizing than to engage in some other less explicit, less transparent, and less systematic consequentialist reasoning process.

    To illustrate, take your ICBA vs. USCBA critique. Right off the bat, your misunderstanding is apparent: it does not make sense to refer to ICBA as a "modification" of CBA, because CBA is not inherently committed to a position on the question of how to value fetal life. So no, I don't have to come up with a way of identifying "true" CBA and ferreting out impostors based on substantive values. Both ICBA and USCBA -- assuming they are carried out in accordance with the procedural commitments inherent in CBA that I note above -- are forms of "true" CBA and are meritorious because they facilitate utility maximization according to the decision-maker's conception of utility. "True" CBA is simply that which properly engages in the process of CBA. I don't have to prove that CBA always yields the objective right answer on every policy question (assuming such answers even exist) in order to prove that it is better than its alternatives; I just have to prove that it's the best way for a consequentialist decision-maker to maximize her conception of utility.

    In other words, your critique is as flawed as maintaining that CBA can't possibly be the best means of consequentialist decision-making, because it can yield an infinitude of different answers to a policy question based on an infinitude of different inputs. Right. But my point is that all of these CBAs are better than non-CBA.

    Of course, given a particular set of values or a particular country, it is true that certain means of valuing costs and benefits are deemed wrong for substantive reasons. Consequently, a CBA that yielded a certain conclusion based on a "wrong" valuation will be deemed wrong by the relevant audience (e.g., an Irish CBA that afforded de minimis value to fetal life). But this does not mean that CBA itself is flawed, or that CBA is necessarily committed to any position on the valuations in question.

    This is why I feel that these "merits" points are simple and uncontroversial, and why I feel comfortable describing CBA as "unambiguously" superior to its alternatives. CBA basically says: if you want to figure out whether the consequences of a proposed policy are on balance good or bad, be systematic and explicit about the considerations you are weighing. The alternative is basically to wave your arms around. Now, I like to wave my arms around, but I don't think we should make regulatory policy that way.

  5. But so in your view, CBA is really just instrumental rationality with strong (and largely undefended) assumptions about the value of spending a bunch of money on data-gathering and -processing. By that definition, pretty much anything is CBA, so long as it involves a bunch of technocratic math. It needn't consider costs, it needn't consider benefits, it needn't maximize anything in particular.

    I don't think that's what this debate is about.

  6. Another point to make here is that even if we accept that CBA consists solely of instrumental rationality, no particular instance of CBA can ever be regarded as "the best regulatory policy." It would only be "the best regulatory policy" for the last chef, that is, for the person whose values it embeds. Everyone else should regard the policy as sub-optimal. And even the person whose values are perfectly reflected may not think that CBA yields the best policy if CBA is unduly expensive or time-consuming.

    But in any case I don't think you can characterize CBA as merely instrumental rationality + a jobs act for quantitative bureaucrats.