In the fire example, the scope of “return” is loss prevention. The NPV of a given fire department is the present value of expected loss prevention minus the present value of the costs to have a fire department. ROI would state it almost the same way.

The goal of ESL is to cause foreign language speakers to acquire English faster and more easily than they would do in normal classes. The scope of “return” might be some measure of the economic benefit of more rapid English acquisition. In this case, that return is negative, because ESL students learn English more slowly than non-ESL. The NPV is then the present value of expected economic benefit (which is negative) minus the present value of the cost of the program. If you get rid of ESL, the NPV calculation is zero minus zero, which is improved, with high confidence, hence you should get rid of ESL. (If I saw data to suggest that ESL students actually learn English much faster, then there would be basis to change my mind.)

So there is nothing inherently business-oriented about this. It is just a way to make decisions.

What is absolutely true, though, is that NPV is a minefield, highly sensitive to tiny estimation errors in certain factors. So you need a huge margin of safety to be able to say anything with confidence.

Opportunity cost is the crucible for all decision making. ROI (maybe “expected NPV” is more precise) is shorthand for the attempt to measure opportunity cost. As I see it, business and finance are merely narrow applications of this general view of decision-making, not the other way around.

To compare alternatives, you would subtract one expected return from another, and see which is better. But I’m not arguing for precise calculations. In business and elsewhere, hairsplitting precision is a path to error. I’m sure you remember Feynman’s comment that it’s far more important to be clear than precise. Instead, you want to pick obvious cases where a result is so far outside the margin of uncertainty that there is no question what to do.

This is what I meant earlier by saying that even if two potential investments have a return of minus 100% — outcome is no different than if you had simply burned the money — you can choose between them, because the one with the smaller outlay results in a smaller total loss.

The examples in the post are intended to be something like that. Each of them has specific, stated aims. Everyone agrees that not 1% of those aims have been achieved in decades of trying. Thus, on the face of it, it is uncontroversial to say these expenditures resulted in outcomes little different from simply burning the money. In some cases it’s worse: spending a dollar on ESL is equivalent to burning more than a dollar, because the rate of language acquisition becomes slower than if the student had joined a regular class.

In the case of the dominoes, this was known to the leadership by the mid-1960s at the very latest. That’s why the Pentagon Papers were such a bombshell to the public — LBJ was admitting that containing communism had little to do with our reasons for remaining in Vietnam.

The level of fire protection is a much harder question, because fire departments actually put out fires. But I will bet anything that somewhere on the county payroll is a statistician/economist that runs exactly this calculation to determine how many fire trucks we need.

Counterfactual problem: I’m repeating myself, but yes, big policy questions commonly suffer this, for lack of sample size. The way out, if any, would seem to be deductive rather than inductive, but that places great demands on controlling one’s bias (hence the invocation of Kahneman).

Making this up as I go (as usual), but for example, could communism realistically be spread, by force, by countries with no naval power (Russia and China), across oceans, to the four most prosperous countries in east Asia (in 1960 these were AU, NZ, JP and PH)? If it couldn’t be spread there by force, then did a military commitment make sense? It doesn’t wash. Or shouldn’t have.

http://www.astrid-online.it/Cartella-p/Orientamen/FOREIGN-POLICY_Why-Hawks-Win.pdf

Rob

The post did not assert the “opposite” policy would have a better result.

It really is true that big policy questions have the counterfactual problem. History rhymes but doesn’t repeat. As a result, it’s hard to argue inductively, especially when trying something new, because we lack applicable, statistically significant data. Does Keynesianism “work?” We still don’t know. Same problem.

The examples I chose are so painfully bad that it becomes easy to argue deductively rather than inductively. It also becomes easy to show that doing nothing would be better. Not the opposite, but simply nothing. If ROI is negative 100%, then investing less gives a better outcome, automatically.

The Amsterdam comment implies some cognitive capture (maybe jokingly), i.e. if we aren’t carpet bombing Colombia, then we must be offering free needles and subsidizing opium dens. In past posts, I have argued for a narrowly defined middle ground: suppress demand, not supply.

http://nostradoofus.com/2005/09/26/war-on-drugs-misunderstands-microeconomics/

That argument is deductive rather than inductive, and is straightforward.

Since the current policy is totally ineffective, any cheaper policy (such as education) is automatically superior, even if still totally ineffective, because the loss is reduced.