Tuesday, February 23, 2010

"The Economics of Snake Oil"

Steven Novella has an excellent post about this. That snake-oil products and services exist shouldn't be too surprising; there's a poor critical thinker born every minute, and five more who want to part that person from his or her dollars. What's strange is not that snake oil products exist, but that they persist - and in some types of products more than others.

[Added later: also offering insight on this question is Razib Khan at Gene Expression: "The history of medicine is, unfortunately, rather similar to the history of astrology. In fact for much of history doctors are likely to have increased, rather than decreased, mortality, thanks to an ignorance of germ theory and false paradigms such as Humorism. The demand-side pressures for cures & prevention seems to still exert a powerful push toward the rise & fall of fads (see google trends for "low carb" for example). A difference between pre-modern and contemporary fads though is that they're not all capricious today."]

Why is the persistence of medical nonsense strange? Because we can frequently count on exchanges of money to cleanse people of irrational beliefs. By that I mean: go find somebody making an irrational claim. Easy, right? But talk is cheap - so now, try to get that person to make a financial decision (bet, investment, etc.) based on their claimed belief. I try this all the time (with 2012ers, Scientologists, young-Earth creationists, you name it) and I've yet to find a single taker. It's a great acid test for beliefs. Plus, it usually shuts them up when they're so obviously not ready to put their money where their mouth is. People believe all kinds of strange things but as soon as immediate material self-interest comes into play, they can find a way to rationalize making the same decision that a person without their weird beliefs would make.* Bullshit evaporates quickly in the harsh glare of financial reality or personal safety.

Or, it usually does. Novella cites the work of American economist Werner Troesken who looked at the patent medicine industry of his time (late nineteenth/early twentieth century), wondering why people kept buying it when it couldn't have been more obvious that the stuff wasn't working - the "inelasticity of demand with respect to product failures". I'd never heard of this Troesken guy before and this work is fascinating for several reasons. Without getting too much into the economics, John Stuart Mill actually used the drug industry of the nineteenth century as an example where a manufacturer-direct-to-end-consumer model would spell disaster, for much the same reason that Troesken said it wouldn't work: consumers themselves with no formal training tried to assess whether the drugs were doing anything for them - we can all tell whether it made us feel better or not, right? In fact, no, we can't. We can tell whether we feel better now than we did an hour or a day ago, but whether it's because of Dr. Wadsworth's Special Elixir is another matter. Medicine is ripe for all kinds of fun heuristics and fallacies like confirmation bias and post hoc propter hoc. And as long as "Dr." Wadsworth was selling direct to consumer, these fallacies ran wild (which is why things like prescription control exist). Another way of looking at this problem is that medicine consumers couldn't get good information about their products or the marketplace as a whole, which is why the market didn't work. Paul Krugman's early work was on the same problem, but he was looking at used cars.

But there's more to it than ignoring product failures. The first objection is that humans are rationally self-interested only to a first approximation, and of course our behavior in markets reflects this. Otherwise, advertising wouldn't work and there would be no such thing as brand loyalty or viral marketing (memes, anyone?) As an economist, Troesken was explaining observations that clearly didn't accord with the rational optimizer model of economics - he wondered why people kept buying this junk when they knew damn well it didn't work. There are two solutions that make the behavior seem less demanding of an explanation. The first solution erodes the rational optimizer model - people are muddle-headed, and they actually don't know most of the time (in any clear sense) that the junk didn't work. The second reason is that even if the people did know, they make buying decisions for all kinds of reasons that don't have to do with product efficacy, but might still be rationally self-interested in the long run (display status, social networking, etc.)

What's more interesting is that among areas of commerce, healthcare seems to suffer more from "inelasticity of demand with respect to product failures" than others. Even after a century of amazing progress in the life sciences, we're still fighting homeopathy and anti-vaxxers. Compare this to the auto industry - yes, there might be families or individual with attachments to Ford or Chevy but there's no equivalent to faith healing and homeopathy in the auto industry. (I once had a Ford I considered taking to an exorcist but that's another story.) So what's special about healthcare that makes quackery uniquely persistent? Some possibilities:

- Healthcare by its nature has unclear cause-effect relationships that facilitate misinterpretation via cognitive shortcomings (as described above)

- Healthcare involves health of self or family members; ironically, as with other central moral values, this makes it more difficult for us to think objectively and critically

- Healthcare is purchased under duress. Suffering people (and their frightened loved ones) make poor decisions, and when you're in pain you'll latch onto anything

- Healthcare is seen as subject to group norms, especially where care is given to dependents, and thus subject to cognitively-distorting social pressures not applicable in other areas of commerce ("What do you mean you won't give echinacea to your daughter? You monster, I'm telling the building association about you!")

- In Troesken's time, there was far less medicine that actually did work. People didn't have ibuprofen as a comparator for Wadsworth's Elixir; they didn't know what it was like when a drug really did make premenstrual cramps go away. (As a result, I would retrodict that we should expect there to have been more snake oil in the nineteenth century than there is today, and that snake oil today is more likely to focus on painful ailments for which our treatments are still not fully effective.)

In the end, the most important part of the Enlightenment project we're extending is the impact on morality and the human condition. Consequently, for those of us who aspire to be more rational, the morality at the intersection of skepticism and commerce is an important area of further inquiry.

*When we talk about rationality, it's worth keeping in mind that there are on one hand people who can keep the lights on and who are theoretically subject to rational argument, and then on the other hand there are people whose irrationality is quite outside the bounds of their free will, like Alzheimers and schizophrenia patients. The practical upshot is this. You could morally collect bet money from a crestfallen 2012er on 1 January 2013; you could not morally sell tinfoil hats to the man under the freeway who thinks the CIA has implanted mind control chips in his brain.


Philippe said...

I remember learning about a psych experiment where people were persuaded to pay varying amounts for some "entertaining" event. The event was designed by experimenters to be boring and tedious. People who paid little money were more likely to say the event sucked. People who paid more money were more likely to say it was a good event. The greater sacrifice made, the more likely people are to justify it to others. I'm sure this must apply to alternate medicine. And restaurants.

Michael Caton said...

Yep. The psychologists' verification of a coping mechanism for buyer's remorse. A common one is "I live in neighborhood X and I pay 50% more rent than you do, therefore it MUST be a better neighborhood." I think my favorite price-affects-perception experiment was the Stanford wine club fMRI work. Give them two samples of the same wine, tell them one is $12/bottle, the other is $75. They reported, and legitimately perceived (according to brain activation) greater pleasure drinking the "$75" wine. And these were people who thought they knew what they were doing around a pinot noir.