15 January 2009

The mismeasure of money

One of my pet irritations is that money is an accounting tool that's more or less taken over the world, and the number of places where the map is not merely not the territory, but some place lamentably unrelated to the territory is increasing.

The simplest example of this I can think of is error rate.

Anybody, doing anything, will make mistakes. You can, and it is very often done, relate these mistakes to a cost; if you have a robot on an assembly line tightening bolts, and it snaps one bolt in 400, rather than successfully inserting and tightening that bolt, you can figure out how much that costs to fix (somebody has to drill and tap the snapped bolt to get an extractor in it, and then replace it with a bolt at the correct torque, and they want to be paid...) and then build that cost in as representing the error rate.

Only, it, well, doesn't. It represents the pricing of the error under some set of accounting assumptions, and assuming the prices come out the same, there's no difference in that model between the bolt example and lighting every 10,000th thing that's having the bolt inserted on fire.

Which is a bad model. If there is no difference in the model between "defective robot applies over-torque" and "defective bolts snap sometimes", you can't tell those two cases apart. Since "I wonder how we fix the robot, to keep it from mangling these good bolts" and "I wonder how many of those defective bolts are almost at the point of snapping, allowing the wheels to fall of cars at speed" are extremely different questions when it comes to fixing the problem, you want a model that allows you to distinguish these cases.

Similarly, prosperity is difficult to measure with money, since prosperity is a measure of predictability (in the future, will I be better off? worse? about the same?) that's at least somewhat relative to the people around you (do you want to be highly prosperous by the standards of Early Modern London?) but which can be measured in fairly absolute ways across populations (literacy, age at death, etc.) but this does not stop wide swathes of media and government from using median income as a nominal measure of prosperity and collapsing huge swathes of differing access to choice into an inappropriate proxy measurement.

The worst assumption is that money is a good proxy for the things available for you to choose to do; that anybody with 10 dollars, or ten thousand, can do the same things with them.

Since that's obviously not a factual statement, and since the ways in which that isn't factual are in many respects proper targets of public policy, dropping the monetary proxy and starting to count the actual access to choice would be the better thing. (The more challenging thing, too, but we have really good computers these days.)


Zeborah said...

This reminds me of a conference presentation Marilyn Waring gave pointing out that money isn't a solution to problems. Money might enable you to solve problems, but just throwing money at the problem does nothing.

She said that for example she didn't think New Zealand should immediately increase its GDP percentage for international aid to the desired (what is it?) 0.7% or something - because the aid agencies that would receive the money don't have the infrastructure to deal with such a rapid increase. Best case, they'd throw the money at idiotic projects.

She gave an example of an island where some foreign aid group wanted to build a road to support motor vehicle travel. But everyone on the island walked or used bicycles; there were no motor vehicles to support.

She was encouraging us to be careful about what questions we asked, to avoid distorting the problem by quantifying it with eg money because that can too often lead to meaningless answers and useless solutions.

Graydon said...


You can (sorta) proxy access to resources with money, but you can't proxy organization, and what you get out of those resources is mostly a function of organization.

Ms. Waring is decidedly correct that you have to know what the problem actually is before attempting to solve it, too.

Though you can readily spend aid money usefully without feeding it through in-country agencies. You give it to individuals as educational scholarships, so you get a couple hundred engineers and doctors trained, send them back home, wait six months, and then you ask them what to do with the rest of the development aid you have available.