The Tyranny of Cost-Benefit Analysis

Posted by on . 4 comments.

Ezra has a post on the curious privilege of economists to weigh in on policy issues seemingly well outside their field. He’s right; dismal scientists basically get a free pass to talk about any policy question, and they’re often looked to as experts on issues not necessarily related to their immediate areas of research.

I don’t think this is that difficult to understand. Economists essentially have a monopoly on the methodology of public decision making. The cost-benefit analysis is the beginning and end of many policy questions, and it’s highly economic in nature. You assume certain parameters — growth rates and such — tot up costs and benefits, choose a discount rate, and determine how much current people should be willing to pay or invest to achieve certain future goal. Future goal can be nearly anything: completed highway project, x number of jobs created, y percentage of the population insured, z billions of dollars in climate change costs averted.

Policymakers want to know what the benefits of any particular project are worth. Economists can give them a simple, easy to understand answer. They can occupy an op-ed slot and write things like, “Such-and-such project is a bad investment. It will deliver far less in benefits than it will cost in spending. Here are the numbers to prove it.” That’s very difficult to argue with.

You can see this at work in an interesting back and forth between Robert Pindyck and Martin Weitzman on climate change policy, described here. Pindyck takes the down-the-line approach to the problem, estimating the future costs of warming and the benefits of policy action and determining that we should spend at most 2.5% of world income to slow or halt climate change. Weitzman, on the other hand, says that this is nuts. By tweaking just one or two variables, he says, you get wildly different outcomes. Simply by concluding that it matters whether the world ends in 400 years or not (irrelevant in Pindyck’s approach), one can then compute that up to 99% of income could reasonably be spent to avert warming. Cost-benefit analysis just doesn’t make sense in the context of climate change, Weitzman declares.

Well, this just about makes Pindyck’s head explode. He begins warning that abandoning cost-benefit analysis will “cede the debate to radical environmentalists.” This seems absurd, but it does get at a difficult question — if we can’t use cost-benefit, how do we make these decisions? How in hell do we figure out which trade-offs are sound ones and which are damaging to society on net? To the economist, and indeed to many policymakers, eliminating cost-benefit analysis is like depriving them of language — how can you discuss the problem without it?

In fact, we really have no other way of wrestling with these issues. One can argue by moral imperative — that we don’t have the right to impose serious costs on others — but we still must determine how much compensation or preventative action those others are owed, and from where the resources to pay or take action should be drawn. These questions involve trade-offs, which must be weighed in some fashion, and so again we find ourselves turning to economists.

What this says to me is that we must rely on economists and on cost-benefit analysis, but we must not rely exclusively on those things. When an economist declares that we should spend no more than 2.5% of income to slow or halt warming, that’s a data point worth considering, but it’s not the only data point worth considering. Other things — moral responsibilities and acknowledged uncertainties, for instance — should be taken into account. This will surely drive many economists nuts. But if you were lost in a forest with an imperfect map, you wouldn’t guide yourself by looking at that map and nothing else. You’d keep your eyes on the terrain around you, to make sure you weren’t blundering into any obviously dangerous situation. The map is a guide only, and it should be understood as such.

Comments

  1. Doug says:
    June 3, 2009 at 1:29 pm

    2.5% of world income would be a huge increase, wouldn’t it? I think the fear that makes some of us cling to economic analysis is just the one Pindck points out. That without a perspective based on estimated value, we’ll quickly end up with the goal of a world without carbon. There’s no evidence for this, but it’s a fear nonetheless.

  2. Karl Smith says:
    June 3, 2009 at 1:32 pm

    Economists get a big say on policy choices because this is “uh, like the science of choice, dude”

    More seriously the question “should I do this or not” is the essence of economics. The grand value of economics is that in answering questions about how society should decide it helps to look at how people actually decide.

    For example, you might say that perhaps we should consider it important whether or not the world exists 400 years from now.

    The first question for an economist, however, is “do people behave as if they care whether or not the world exists 400 years from now?” If the answer is consistently and overwhelmingly, “no” then I question the utility of that perspective.

    In particular I suspect that what people really want is to “feel as if they care what happens in 400 years” but in reality they do not care. This is exemplified by them consistently not counting 400+ year costs when making their own choices but then publically expounding on the need to do so. In short, by them being hypocrites.

    Now maybe there is a need for a greater, deeper public morality. I am open to that. However, that doesn’t invalidate cost-benefit. Cost-benefit doesn’t have to be utilitarian. It can be Rawlsian, or with fancy new computer models any sort of social contract theory that you want.

    On a much much more practical level I think appeals to abandoning Cost-Benefit often come because Cost-Benefit reveals the actual complexity of the problem.

    The fact that Cost-Benefit clearly yields radically different answers based on different assumptions is a feature not a bug. It is telling you that your assumptions, explicit or implicit are incredibly important to this question.

    The fact that other arguments might bury their assumptions makes those assumptions no less crucial to the conclusion.

    In the particular case of climate change I think the discount rate debate is extremely important, BECAUSE it yields such different results.

    I have a lot more to say about this but let me be brief. Very small uncertainties about the future path of our society do make big differences in the relative costs of global warming.

    All endogenous growth models show some connection between discount rates, interest rates and long term growth rates. In many models all three must be identical. In simple terms it makes sense that a society with a higher discount rate would invest more and therefore grow faster.

    A society in which labor productivity is growing a 1% will have over 50 times the resources per person to deal with the consquences of global warming. That’s huge. So huge it can be hard to wrap our mind around it.

    For the US it would be like going from $30K a year to $1.5M a year. Obviously the choices available to you at $1.5M are vastly different. Moving to a different part of the country is no big deal. Food is a trivial part of your budget, etc.

    However, with labor productivity growing at 2.5% as they did in the 1990s the differences are far far more massive. $30K a year becomes $600M.

    At 5% a year, more like rates in some developing countries income increases by a factor of 300M. Even people living on less than a dollar a day today would become fabulously wealthy.

    I am not suggesting that sustained increases in productivity of 5% are possible but simply that the wild difference in answers mirrors the wild differences in states of the world we could be facing.

    A would in which the poorest among us have what would today be $300M in resources available to them is unimaginably different. If your moral calculus is based on assuming that the world 400 years from now is even quasi-recognizable to the world we have today then small differences in growth rates could mean that you are grossly grossly mistaken.

    Its also important to note that while productivity growth rates in the West are stabilizing, world productivity growth is accelerating. The next 400 years are likely to see much more change even in percentage terms than the last 400.

    So long story short, the trade-offs here are not intuitive they are highly unintuitive. The advantage of cost-benefit is that it reveals that.

  3. dWj says:
    June 3, 2009 at 10:28 pm

    Ultimately, it does come down to cost-benefit analysis, but the inputs into the analysis are not fundamentally economic. Just as climate scientists have no special moral ground to tell us what tradeoffs to prefer, neither do economists; climate scientists can tell us what is how likely to result in what, and economists can tell us the same sort of thing about a different set of questions. Cost-benefit analysis can force us to keep our preferences honest and coherent, but it can’t answer the ultimate normative questions beyond that.

  4. jeff says:
    June 4, 2009 at 5:58 pm

    Karl notes that the importance of assumptions is a feature of cost-benefit, not a bug. Sure, but it’s a feature that allows the analysis be skewed however the analyst wants to skew it. Like most long-term analyses (DCF, demographic trends, etc.), any good analyst can make the numbers say whatever he/she wants. So its incumbent on the analyst to make clear those assumptions, and the biases that led to them.

    But more to the point of the original post, some outcomes seem fairly unquantifiable. What number do you put on the end of the earth, whether it’s in 400 years or 4,000? Moral absolutes can create lead to odd results (“spend everything on global warming, right now!”), but so can efforts to quantify existential consequences. This leads back to Ryan’s eventual conclusion: cost-benefit is useful, but it can’t be our only guide.