16 January 2012

How NOT to do benefit-cost analysis

A political economist* has written a study on the effects of piracy on local incomes in Somalia [BBC version], concluding that "the positive economic impacts of piracy are widely spread, so a military strategy to eradicate piracy could seriously undermine local development."

Let's draw a few parallels by analogy, e.g.,
  • The positive economic impacts of slavery are widely spread, so the eradication of slavery could seriously undermine local development.
  • The positive economic impacts of the war on drugs are widely spread, so an end to the war on drugs could seriously undermine local development.
  • The positive economic impacts of attacking foreign nations are widely spread, so an end to war could seriously undermine local development.
What we are seeing here is a classic example of failing to see the forest for the trees. Sure, there's some benefit to pirates who attack, seize and ransom ships, but the negative impacts of these actions -- from direct impacts to shippers, passengers and crews to indirect impacts to consumers and taxpayers -- are far far greater.

Ask yourself this question a different way: is piracy so good for development that we should have it in as many places as possible? I hope you're able to see why that answer is no.

Bottom Line: There are costs and benefits to every activity. It's not just important to make sure that benefits exceed costs, it's important to make sure that those who get the benefits also bear the costs!

* I've been telling people I am a political economist for awhile now, but I just updated the blog banner from "the economics of water (and other stuff)" to "the political-economy of water (and other diversions)." Better?

12 comments:

  1. I mostly agree, any sort of benefit that happens from illegal activity should not be counted as a benefit in BCA. But I disagree with your bottom line since there are good policies where the benefits and costs are not borne by the same people.

    ReplyDelete
  2. @JW -- they may be "good" from a social perspective, but they are often unfair when one group bears costs and another the benefits. I can say benefits to special interests paid for by the population are far more destructive than benefits paid to students (young people) that are paid for by older people. In fact, we can have a good debate on this, since I am not so sure about ANY policy that divides costs and benefits by a significant amount...

    ReplyDelete
  3. But, equity isn't what BCA is made for. A BCA can be used to identify potential Pareto improvements (B>C) and that is about all. A well done analysis should describe when the benefits and costs are borne by different groups. It is then the decision maker's role to determine if the policy doesn't pass an equity test.

    So, we agree, except about the extent to which economists can use BCA (and I assume economists aren't decision makers).

    ReplyDelete
  4. @John -- I'll go with that, except that it's OFTEN used by decisionmakers without due consideration (or bias consideration) of equity -- so I think we need to add that instead of allowing BCA to be abused...

    ReplyDelete
  5. F. Bastiat's Broken Window?

    ReplyDelete
  6. Can't it simultaneously be true that piracy fails a global CB analysis but would pass a local one for pirate communities? JW's dead wrong that we have to ignore benefits of illegal activities - read David Friedman on this one.

    ReplyDelete
  7. Firstly, it's worth noting that the study doesn't claim to be a benefit-cost analysis, but instead looks at one hitherto overlooked category of benefits. It seems reasonable and useful to investigate these. The study also doesn't suggest that we should maximise piracy because of its local economic benefits. I’m not sure where you got that from. In fact it suggests a few ways in which piracy could be addressed.

    Secondly, it's perfectly possible that "the positive economic impacts of piracy are widely spread, so a military strategy to eradicate piracy could seriously undermine local development" with piracy at the same time failing a broader cost-benefit analysis. In any case, if we wanted to do a BCA of piracy globally it would be important that we quantified the local development impacts, as well as the costs (ransoms, lost goods and trade, buying security for ships, etc.).

    From a broader policy perspective, not recognising the benefits piracy can bring to some communities would be incredibly short-sighted as it is precisely this lack of local development that is behind much of the piracy around the Horn of Africa (lack of alternative economic opportunities and poor governance being closely though complicatedly related, both in general and in the Somali case).

    It's clear that this study gives useful information for policy generally, and is the sort of information that would be useful to a narrower BCA of the issue. Sure, it's not a BCA, but so what? Even if we think an BCA is the end-goal, that would need other studies to feed into it by scientists and other social scientists of the subject matter itself (we need to know the impacts before we can value them!).

    ReplyDelete
  8. My apologies for double-posting, but to quote directly from the conclusion of the study in question:

    "The conclusion that a large group of people can be expected to benefit from piracy should not discourage the international community from seeking a land-based solution. The total cost of piracy off the Horn of Africa (including the counter-piracy measures) was estimated to be in the region of US$7–12 billion for 2010, while ransoms were said to be in the region of US$250 million. Even if Somali communities received all of the ransom money, replacing this source of income (for example with a combination of a foreign-funded security forces and development aid) would be considerably cheaper than continuing with the status quo."

    ...

    ReplyDelete
  9. @EEP -- the local benefits that you highlight come with considerable costs. They are then, de facto, not worth discussing. It's the same -- as Scott mentions -- as the "benefits" that accrue to glaziers from broken windows. In other words, there is NEVER a case in favor of piracy (or other theft) -- no matter how accurate your quantification. My point -- if you're still wondering -- is that this academic exercise is a total waste of time.

    ReplyDelete
  10. Eric Crampton:

    Could you provide a link?

    My point is that criminals have no standing in benefit-cost analysis and therefore their benefits and costs should not be included.

    When considering this point Zerbe and Bellas (A Primer on Benefit-Cost Analysis, p. 45) begin with this "Yet it does not seem morally defensible to include utility from bad acts in a normative analysis." And conclude: "This may be taken to deny standing to the value of goods ot the thief or the utility that a murderer gains from murder."

    In Boardman et al's chapter 16 (Cost-Benefit Analysis: concepts and practice) on shadow prices they provide estimates of the cost of crime (p. 417) and the cost of crime control programs but none for the benefits of criminal activity. Implicitly, I take this to suggest that a BCA should compare the cost of reducing crime with the benefits (the avoided cost of crime).

    Finally, in the end isn't the benefit of crime just a transfer? Transfers net out in BCA.

    Note that this is different from trying to understand criminal behavior with rational choice theory.

    ReplyDelete
  11. David Zetland: I see what you are saying, but if you read the actual research paper it doesn't say anywhere in it that there is a case in favour of piracy (in fact it says the opposite - although I know what you are saying re. the benefits not being real benefits), and it isn't doing a benefit-cost analysis.

    That doesn't matter, though, as there is more to useful research than cost benefit analysis, e.g. if you want to come up with ideas of how to stop piracy, knowing about how its proceeds are distributed tells you about the incentives involved, which is exactly what you'd need to know to design effective policy. Hence the paper could be very useful to policy, and certainly doesn't fall into the heffalump traps you say it does.

    Policy isn't just a question of whether or not we should favour piracy, but about how we can effectively deal with it and to weigh up different options. It's not all about NPV; it's a question of how to deal with a problem, and you can do a BCA for that option only after that. Unless we are going down the route of saying that BCA is the only policy-useful research, which is demonstrably false.

    On a sidenote about criminality and BCA/ethics, it only makes sense to exclude all illegal acts if we say that illegal = immoral (which seems dubious), and only makes sense if we see moral quality as a criterion for inclusion of something in BCA -- this is, to my mind, clearly beyond the remit of an economist performing a BCA. That's a question of ethics for the policymaker/public/other stakeholders to address.

    The whole implicit ethical framework on which BCA says that people are best placed to know their own interests, and act in a way that maximizes this self interest, as shown via what they choose to do/buy -- hence why it would be desirable to maximise utility as defined by people's preferences, as in BCA in the narrowest sense. If you're adopting that ethical framework (and we do every time we do BCA) it's quite hard to, at the same time, introduce finer ethical distinctions (which an economist shouldn't probably make of their own initiative, as they're the domain of other stakeholders int the policy process). War on drugs is a good example. If illegal = immoral = excluded from BCA, and drugs are illegal and drug users criminals, we should not include the massive benefits drug users would gain from legality in any BCA of drugs policy. Or decriminalising homosexuality - the benefits to gay people would not be counted in a BCA of the policy because they would be counted as criminals pre-decriminalisation. Both of which are obviously absurd. I think this might stem from distinction (or lack of) between BCA sensu stricto and broader policy evaluation of which what I sometimes call narrow BCA is a component (where equity/distributional consequences, etc., are also accounted for).

    ReplyDelete
  12. @EEP -- I agree with most of your comment, so I think we may be talking with (or past) each other. My original post (and point) was that the paper was *presented* as an argument for considering the positive impacts of piracy on the local community -- as if (and this is perhaps implied) there was a reason that it should continue. How to end piracy, as you discuss, is a different matter that got lost in that headline.

    ReplyDelete

Spammers, don't bother. I delete spam.