I thought it was us that invented AIDS but it seems we didn't: the economists got there first. An Almost Ideal Demand System lies at the heart of a recent paper on what would happen if we put a 20% tax on fizzy drinks.

Unfortunately, the economists' AIDS is utterly incomprehensible. It's full of notions like this:

"Provided (10), (1 1), and (12) hold, equation (8) represents a system of demand functions which add up to total expenditure (Ewi = 1), are homogeneous of degree zero in prices and total expenditure taken together, and which satisfy Slutsky symmetry."

Science builds on what went before, so if you go back in time methods get simpler. The inventors of AIDS open their 1980 paper with a tribute to a 1954 paper by a Richard Stone. Google scholar takes us to a 1945 paper by the same author called 'The analysis of market demand', which I can just about understand. Stone used 19 years' worth of data on household expenditure to see what happened to spend on, say, beer, as prices went up and down: this is a straightforward regression analysis with various predictors (including price) of an outcome (national beer consumption). So I assume that AIDS does the same, but with a rather better mathematical model - they didn't have computers in 1945. When I say 'better' I mean that the AIDS model satisfies Slutsky symmetry etc etc.

Back to the present day. The authors of the paper on sugar sweetened drinks don't have 19 years worth of data so they can't be looking at what happens as price goes up and down. They have data from one year, but at household level. Some of the houses are rich and others are poor, and they spend different amounts on fizzy drinks. So I guess (can't be sure) that they are using this variation between households to estimate what will happen if fizzy drinks become more expensive relative to other drinks.

But if you don't understand it fully, you shouldn't use it (as bankers found out with collateralised debt swaps). So I won't be using this paper because I really don't understand what lies at the heart of it. And that annoys me. If this had been published in an economics journal you could expect readers to understand (I suppose) AIDS. But it's published in a medical journal. The authors owe us more than a bald reference to an incomprehensible paper.

And that's not all! Another key part of the paper is working out the link between calorie intake and obesity. This was done using the PRIME model. Following my own advice (see last week's rant) I thought I'd better find out more about how PRIME works. But the reference cited is a webpage which as far as I can tell (and I used in the in-page search function) says nothing about PRIME. I emailed the lead author last week to ask if this citation was erroneous but he hasn't bothered to reply.

Oh dear. That's two rants already this month. Am I becoming a grumpy old man?

Meanwhile here is a paper I do understand: it sets out rather nicely some practical problems with fat taxes.