David R. Henderson  

Can One Conduct Cost-Benefit Analysis of a Policy?

Illegal Means Illegal... The bizarre way economists cal...

The answer to the question I asked in the title seems as if it should be "Yes." And not just "yes," but "Obviously yes."

Yet economist John Whitehead says that one can't conduct cost-benefit analysis of a policy. Specifically, he writes:

The costs of cap and trade do not outweigh the benefits. It might be the case that the costs of a climate policy, any climate policy, outweigh the benefits. But cap and trade is a policy instrument, not something for which you conduct a benefit-cost analysis. The economics says that if the government decided to undertake climate policy, cap and trade would be one of the most cost-effective ways of doing it.

Notice his third sentence. Cap and trade is a policy instrument. But, contrary to Whitehead, you can conduct cost/benefit analyses of policy instruments. You would compute the costs and compute the benefits. It might be difficult to get a handle on those things, especially on the benefits. But in principle it's doable. So the answer to the question I raised in the title is "Yes."

Whitehead seems to be saying that one can do only cost-effectiveness analyses of policy instruments. That's true only if you can't compute benefits.

By the way, in the same post Whitehead also addresses the issue of carbon taxes with offsetting reductions in other taxes. Although he doesn't address the issue directly, I'll bet dollars to doughnuts that he is not aware of this article and the important tax-interaction issue it addresses.

HT to Mark Thoma.

Comments and Sharing

CATEGORIES: Cost-benefit Analysis

COMMENTS (17 to date)
Thomas writes:

Digging deeper, I would say that one can conduct a cost-benefit analysis of government policy, but that such an analysis will be practically meaningless in most cases. Why practically meaningless? Because government policy, in the end, confers benefits on and extracts costs from individuals. And there is seldom complete overlap between the group of individuals that bears the costs and the group of individuals that reaps the benefits. In fact, I suspect that the degree of overlap is usually on the low side. If you accept my premises, then you should agree that cost-benefit analysis of government policy is practically meaningless because it compares costs and benefits that are mostly incommensurate. My favorite homely example goes like this: Adam may derive a lot of pleasure by punching Bob in the mouth, but Adam's pleasure by no means cancels Bob's pain (assuming reasonably that Bob isn't a masochist). Thus it is with most (possibly all) government policies and programs.

Pierre Lemieux writes:

I haven't read Whitehead's article, so I might be off the track. I think I can see part of his argument, though: costs and benefits depend on your starting point -- in this case, no climate policy or a climate objective. But the problem of cost-benefit analysis is deeper than this: see my article on this topic at http://www.independent.org/pdf/tir/tir_11_01_02_lemieux.pdf. Anthony de Jasay says it in two sentences: “it is the intuition of the person making the comparison which decides, or there is no comparison. . . . In an analogous manner, the
two statements ‘the state found that increasing group P’s utility and decreasing that of group R would result in a net increase of utility’ and ‘the state chose to favor group P over group R ’ are *descriptions of the same reality*.” (emphasis in original).

Tom West writes:

I think where climate change is concerned, there are enough degrees of freedom in the problem that it's nigh impossible to come up with concrete benefits:

- Save the planet?
- Provide subtle pressure to persuade other countries to take CO2 cutbacks seriously?
- Feel better that we've actually done something to make a difference?

etc., etc.

It's like trying to apply cost-benefit analysis to deciding to have children. It can be mildly useful for enumerating one's thoughts, but the economics are a vanishingly small part of the decision and the actual calculation of the benefit is often near meaningless.

Cost analysis is useful, however.

Charlie writes:

"I'll bet dollars to doughnuts that he is not aware of this article and the important tax-interaction issue it addresses."

Should I make the same bet that Murphy isn't aware of the importance of the interaction, between carbon/gasoline and leisure in the choice of an optimal gas tax? More interestingly, this evidence comes from work by his primary source Lawerence Goulder (and a former professor of mine Roberton Williams III). In a later article, they find that even in the absence of any externality, gasoline should be taxed, because it is a shadow tax on leisure and inhibits the disincentive effect of other taxes.

Optimal taxation and cross-price effects on labor supply: Estimates of the optimal gas tax


This study estimates parameters necessary to calculate the optimal second-best gasoline tax, most notably the cross-price elasticity between gasoline and leisure. Prior theoretical work indicates the importance of this elasticity, but despite this, almost none of the prior studies of commodity taxation (and none of the studies on second-best environmental regulation) actually estimate it. Using household data, we find that gasoline is a relative complement to leisure, and thus that the optimal gasoline tax is significantly higher than marginal damages-the opposite of the result suggested by the bulk of the prior literature. Indeed, even if there were no externalities at all associated with gasoline, the optimal tax rate would still be almost equal to the average gas tax rate in the U.S. Following this approach to estimate cross-price elasticities with leisure could strongly influence estimates of optimal rates for other important commodity or pollution taxes.

An Igyt writes:

Whitehead is no ignoramus. I suspect he meant, one cannot do CBA without the detaills -- i.e., without specifying how stringent the carbon cap would be. What Ryan said could be true, if even the highest net benefit among all possible cap levels were negative, but that is not what economists have calculated, not even close. For example, the new coal plant rules have positive net benefits according to EPA (and still positive, but small, if done more traditionally, see Stavins). Nordhaus's models imply a large range of positive net benefits... and many more.

JLV writes:

You could of course do cost-benefit analysis of a policy of reducing CO2 emissions by X tons using a cap and trade regime, but this is not the same as doing a cost benefit analysis of cap and trade.

You can only compute costs and benefits for cap and trade conditional on a specific policy (but then you are just computing costs and benefits of the policy, not cap and trade.)

JLV writes:

And to add, I think Whitehead's cost-effectiveness point is that if we have decided on a policy, and the benefits are independent of instrument, cost-effectiveness analysis is all we need to do. The choice is the best way to do the policy, not whether to do the policy.

Hazel Meade writes:

Perhaps what he was grasping at is that nobody has any idea what the benefits are or has any means of valuing them.

For some people, carbon reduction is just a binary goal. If you're going to do it, you don't really care what it costs, except in the sense of picking the mechanism that costs least. But you don't ever decide NOT to do it because the cost is too high.

This fits hand in glove with the believe that climate change is a catastrophe whose costs will be incalculably high. If you believe it, there , yes, it's impossible, or rather, pointless, to do cost-benefit analysis, because the "benefit" is infinity.

Alex Godofsky writes:

I think you're overreacting. I think he's saying that you can't usefully perform a cost/benefit analysis of an instrument, you can only perform one of an instrument setting, i.e. a particular cap.

David R. Henderson writes:

And to add, I think Whitehead's cost-effectiveness point is that if we have decided on a policy, and the benefits are independent of instrument, cost-effectiveness analysis is all we need to do. The choice is the best way to do the policy, not whether to do the policy.
That doesn’t make sense. Remember that he’s criticizing Paul Ryan’s claim, and Ryan is challenging the policy. So there’s no “we have decided on a policy."

David R. Henderson writes:

Thanks for this cite. I don’t know if Bob Murphy was aware of it, but I was not and, I would bet, over 90% of the readers of this blog were not.

J Scheppers writes:

Dr. Henderson and Charlie:

I am befuddle, just because gasoline is a relative compliment to leisure does not justify a higher gas tax. The information in the citation seems to jump to the conclusion that leisure should be taxed, and even more than that, leisure should be taxed via the gas tax. Seems quite a leap to me. Please help me out if I not understanding the mechanism that justifies a increased tax on relative compliments.

I do think I agree with Charlie on the concept that renaming a revenue does not address the issue. The purpose of the existing gas tax is a user fee to pay for infrastructure. This transaction of the gas tax for roadways is trade between parties. Nominally changing the gas tax to carbon tax and then still using the revenue for roadways has not instituted a price on carbon, it has only green washed the gas tax.

David R. Henderson writes:

@J Scheppers,
I am befuddle[d], just because gasoline is a relative compliment [sic] to leisure does not justify a higher gas tax. The information in the citation seems to jump to the conclusion that leisure should be taxed, and even more than that, leisure should be taxed via the gas tax. Seems quite a leap to me.
I haven’t read the article yet and so I can’t comment on whether it jumps to a conclusion. But here would be my guess about the reasoning. We know that high marginal tax rates cause people to take more leisure. You (possibly) and I (certainly) would like to handle that problem by reducing marginal tax rates. But another policy, the one that the article suggests, is to tax leisure. Taxing leisure is difficult: it’s hard to measure and you can easily imagine some pretty intrusive ways of taxing it. What’s another way of handling the issue? Find complements to leisure that are relatively easy to tax. One such complement, according to the authors, is gasoline. So tax gasoline.

Do I know that this is the authors’ argument? No. But the abstract seems to suggest that.

Dallas writes:


You misunderstand John.

He's making the point that Paul Ryan isn't making any sense when he says "the costs of cap and trade outweigh the benefits". Cap and trade is just one policy instrument for reducing GHG emissions. You can't actually conduct a benefit-cost analysis UNTIL you propose a specific cap.

At the same time, he is also saying that regardless of the cap you choose, the cap-and-trade *policy instrument* will always achieve the objective at lowest cost (i.e. be cost effective).

I hope this eliminates the confusion.

Charlie writes:

@JScheppers and @DH

DH nailed the reasoning, which shows he has a pretty solid intuition for Public Finance. RWIII used this as an example in the very first public finance course I ever took, and I was quite befuddled as well. But it is reasoning right out of public finance 101. The reasoning DH espouses holds as long as there is any positive income tax and otherwise untaxable leisure (some real world considerations like collection/enforcement costs not withstanding). Taxing labor disincentivizes work, so why not offset it by disincentivize leisure as well?

An unresolved issue, though, that is brought up in the Murphy piece is the effect on the capital stock. It sounds like Goulder's earlier work addresses carbon taxation in a more dynamic setting than the RWIII+Goulder paper. I don't know if the question has ever been addressed empirically in the subsequent literature. I know as of 2007 or so that it hadn't been.

It'd be interesting to know Goulder's views today, given that he's wrote about the issues from several different perspectives.

Daublin writes:

Dallas, while you can't know the exact costs and the exact benefits without knowing a particular policy, you can most certainly figure out their relative magnitudes. That is especially so if you have a way to estimate the probable magnitudes that would ever be reasonable.

In the case of CO2 controls, most policies being pushed by proponents don't make much effect on CO2. Thus the benefit can't be significant, but the costs substantially non-zero.

In fact I would quite like to see a serious proposal that will predictably move the level of CO2 in the world, or at least the US, to such a degree that global temperatures will be plausibly affected. I believe that such a policy would be obviously unacceptible to the bulk of Americans. The proposals we are seeing are only getting anywhere because they are tempered down to almost nothing.

J Scheppers writes:

@Dr. Henderson & @ Charlie,

Sorry for the long delay, but if you ever check here is my reply:

I was able to find the Goulder & Williams reference and it is as advertised stating that leisure should be taxed to induce people to work more to reach the optimal work level.

I disagree that a gas tax as a substitute for a leisure tax is a good idea.

I spend plenty of money when I and my family take leisure time, and those expenditures during my leisure are significantly taxed along with significant taxes applied to hotel clerks and rental car agents that assist me in my family's leisure. Even if I were to stay home and do yoga, I still have to pay my mortgage and utilities with all the associated taxes. It is also true that I pay a gas tax for the portion of my leisure that involves driving. These examples should provide plenty of costs to show leisure is not close to free.

It is my view that these taxes and costs are aimed at payment of user fees based on the cost (including profits) of providing the associated overhead to enable the leisure activities that I participate in. There is not a perfect correlation between tax and overhead, but nothing ever is perfect.

I find it very paternalistic to tax extra for gas for the specific purpose to induce me to work more. Ask me to pay my share for the government overhead to support my activities, and I would say definitely within reason. However this gas/leisure tax goes further to attempt to change my choices for the good of society.

There are significant distortions to leisure markets if the gas tax was large enough to materially change work behavior, with more people having stay-cations instead of vacations. It would be extraordinarily difficult to tax leisure gas and not also the gas used to get to work or other "productive" activities.

From my libertarian view, I am opposed to increasing the tax when I choose to work less and smile more. I also reject the concept that since everything is more distorted than the gas tax, we should modify it to make it equally distorted to fix the distortion.

Comments for this entry have been closed
Return to top