When I was doing microeconomics at university, cost-benefit analysis (CBA) was considered boring but quite easy - calculating net present values, and so forth. Doing them in real life I found the task more challenging. For what CBA is really all about is seeking to quantify, in a quite rigorous way, choices over government policy.
It is challenging because there are often key gaps in the statistics and evidence base. Moreover, the task is almost always incomplete, because you can seldom quantify everything (quantifying less direct macro benefits or estimating dynamic behavioural effects can be particularly tricky).
All this is a background to why I found the Economic Focus column in this week's Economist of interest. The regulators' best friend? argues the case for CBA, against those critics who apparently consider it fatally flawed and intrinsically anti-regulatory:
The Centre for Progressive Regulation, a think-tank that shelters many sceptics, thinks so. It objects to two features in particular: the “translation of lives, health, and the natural environment into monetary terms” and “the discounting of harms to human health and the environment that are expected to occur in the future”.
Those who question cost-benefit analysis doubt that a price tag can ever be put on life. How could one seriously count the cost of death and injury caused by road accidents, for example? But, as Robert Frank, an economist now at Cornell University, has pointed out, even the fiercest critics do not get their brakes checked every morning. They have more pressing uses of their time.
Indeed. Done well, CBA can add much to the policy decison-making process - by challenging policy colleagues' to rigourous questioning, by more easily being able to compare policy options, and by assessing whether a policy's benefits outweigh the costs. Done badly, it is just dressing up a bad policy in economic drag. Either way, the inherent limitations of the methodology need to be acknowledged. As I see it there are three key ones:
1. The law of unexpected consequences - You can't really know what the impact of a policy will be before it is introduced, particularly indirect effects (hence a need for proper monitoring and evaluation).
2. Quantification problems - You can't quantify everything, and often it is easier to quantify costs than benefits, as they're more direct (e.g. implementation costs). That, along with a tendency to make fairly conservative assumptions, may lead to an under-estimate of net benefits.
3. Ignorance and dodgy data - The evidence base for specific policy areas can often be surprisingly sparse, particularly in new or emerging policy areas. In such cases pilots or random controlled trials are often a better way forward than full-scale implementation straight off the bat.
None of this means CBA is pointlesss, or worthless. It isn't. But economists also need to be honest about its limitations, and recognise the need - once all the facts and estimates are in - for judgement. That's why we have politicians, I guess.
UPDATE: For those interested in the UK government's approach to cost-benefit analysis, see Treasury's 'Green Book', aka Appraisal and evaluation in central government, and related guidance.