Bullshit from economists, Vol. MMCXIV
According to some economists, apparently the monetary value of a statistical live year (VSLY) is somewhere in the six to seven figure range (see the appendix to this paper). A closely related statistic to the value of a statistical life (VSL), both are a near inevitable part of bureaucratic cost-benefit analysis. For example, there was a big hooplah a couple of years ago when the EPA changed its VSL from $7.8m to $6.9m – effectively decreasing the environmental regulatory burden on business. For example, if a regulation came with an annual cost of $500m to the economy, the old system suggests it would have to save at least 65 lives a year to be worth implementing, compared to 75 now*.
I would like to submit the thesis that these figures are bullshit. This may not be a controversial position, but I hope to quickly demonstrate that these statistics are pretty much inconsistent with the very premises that economists use to generate them (or at least have wildly implausible implications). The idea is this: people are rational, and their actual decisions ‘reveal’ their underlying preferences. Suppose I am at a grocery store that sells apples and oranges at the same price, if I buy an orange then that ‘reveals’ my preference for oranges to apples. In the case of the VSL and VSLY, economists look at people’s decisions concerning how much we are willing to pay for mortality risk mitigation. Now, if these dollar values of these ‘revealed’ preferences are to mean anything of interest, then they should imply that if I have value x at $100, then I should accept $110 in exchange for x. The key assumption is that people are rational, and by looking at the actual decisions people make, we can figure out how much they implicitly value things in a common currency.
For the sake of argument, let’s assume the VSLY is $300,000. The number doesn’t matter very much, but rather the order of magnitude. For my example, let us suppose smoking decreases the number of those statistical life years a smoker would live by 10 (again, order of magnitude is what matters). Now, whilst smoking may not be the activity would one want to point to as a paradigmatic example of rational economic thinking, but it strikes me the decision to try and quit smoking (and how hard to try) is as good a candidate as any. It’s certainly as good as the kinds of decisions economists put into VSL(Y) models. Now, since lots of people today don’t either think of quitting or trying that hard to quit smoking despite being quite aware of all the risks, if they are rational economic actors then the revealed value to them of smoking/not trying hard to quit must at least be something in the region of $3m ($300,000 VSLY times 10, for the years they are likely to lose). That is to say, if they are rational then if the value of smoking less than the value of the life-years they lose due to smoking, then they will at least try and quit (even if this proves impossible). Given that there are people who don’t try, then the value of not trying must be at least equal to the value of the lost life-years.
Now, there was a time when people didn’t know the health risks of smoking, and indeed even believed it to be healthy. I am going to make the assumption that at that time we have no reason to think the value of smoking/not trying to quit would be less than it is today. I actually suspect we have reason to think it would be higher, given the social stigma currently attached to smoking compared to its former glorification. The significance of this is that revealed preference theory predicts that if you paid some young fellow (whose body could recover) back then the equivalent of $1m on the condition he successfully stopped smoking, he wouldn’t necessarily even try.
Think about it. If he is rational and and is one of those who value smoking somewhere to the tune of $3m – well, that number is a lot bigger than $1m. It’s a no-brainer for Homo economicus! I submit that this result is overwhelmingly implausible. It is, however, an implication of revealed preference theory with the additional premise that the implicit value of not having to try and quit cigarettes is about the same sixty years ago as it is today. It is at least extremely plausible for it to be in the same order of magnitude.
So, it seems you have to either believe that your average Joe Schmo wouldn’t immediately leap at such a chance for $1m, believe the value of smoking has sky-rocketed since the time it was cool and trendy, or reject revealed preference theory. At this point, it might be complained that since decisions as to whether a costly regulation is worthwhile have to be made, we cannot but utilize these kinds of statistics in bureaucratic decision-making. This may or may not be true, but that doesn’t mean we should pretend these numbers are remotely scientific, or that they really reflect what it is we actually value.
*This is, strictly speaking, only correct in highly specific circumstances coupled with some even more highly dubious moral theory. I, however, cannot attest to the EPA’s actual decision making process and the way in which these figures are in fact used.