20 June 2007

Good health is worth $631,000 a year?

Nattavudh Powdthavee claims that improving your health from "very poor" to "excellent" makes you as much happier as $631,000 extra per year would.

Does this seem reasonable to you? It doesn't to me. From what I understand, studies such as this are done by asking people to assume they are in excellent health, and then asking how much money they would accept to have (say) a 1% chance of being in very poor health. On average they say $6,310 a year. (In the case of health, it might work by looking at how much people are willing to pay for health insurance; this seems like the sort of thing the folks at Freakonomics might do, although I don't know if they've done it.) Divide by .01 and you get this $631,000 figure.

It gets even weirder when you look at it in reverse. The article claims that "Widowhood packs a psychic punch of $421,000 a year in losses". Taken literally, this means that the average widow would pay $421,000 to have her husband back for a year -- despite the fact that this is almost certainly a large multiple of her entire annual income. Maybe more than her house costs. She'd become homeless in order to have her husband back? For a year?

Or: Increasing face time with friends and relatives from "once or twice a month" to "on most days" feels like getting a $179,000 raise. That's about $500 a day. Are you saying you'd go out today to see your friends if I offered you five hundred bucks to stay in?

I suspect the problem here is that it's hard to express happiness in terms of money. Dollars are not the natural unit of happiness. If you gave me an extra $20,000 a year, I'd be happy -- that would be about a doubling of my income. If you gave the punks who hang out on the street in my neighborhood that money who have no income, they'd be even happier -- now they'd have a roof over their head! But if you gave Bill Gates that money, he wouldn't care at all. For him it's a rounding error. Some people have claimed that, say, getting a 10% raise feels the same to everybody, which seems a lot more reasonable. So what people are trying to maximize is actually the logarithm of the amount of money they have.

Note that if you choose to redistribute all the money in the world so that the sum of the logarithms of everybody's net worth (or income) is maximized, then everybody should have the same amount of money. Intuitively, if I have $30,000 and you have $10,000, then the total amount of happiness is less than if we both have $20,000; if you apply this to all pairs of people then you get this flat distribution. I leave critiquing this argument as an exercise for the reader.

1 comment:

Elizabeth said...

I think you're misunderstanding how the study was conducted. Actually surveying people about how much money they'd pay for things is well-known to be very inaccurate because people have no idea how much they'd actually pay for anything.

It looks like what they actually did is survey a whole bunch of people about how happy they were as well as how much money they make and how much they do other things. I guess that, for example, they found that somebody with "very poor" health had to be making $631,000 more than someone in "excellent" health to report the same level of happiness.

It still seems like the study is *extremely* subjective (especially rating one's happiness on a 7-point scale), and the amounts of money per year are so large that they're bound to be inaccurate (because few people actually make $631,000/year more than other people), but the methodology isn't quite as flawed as simply asking people how much they'd pay.

Also, dollars don't seem like a natural measure of happiness, but often to make a decision you need to know which one makes you more happy, and dollars (or any form of compensation) are probably the most natural and familiar unit by which to compare the relative "happiness level" of two options. It's a typical criticism of economics that dollars aren't a good measure for happiness, but assuming you need a numerical measure, I can't think of a better one... (except in theoretical economics, where they tend to use a fake unit called 'utils')