Friday 15 October 2010

Mont Pelerin

I yesterday spoke to the Mont Pelerin Society about paternalism as a threat to liberty. The presentation is embedded below; it looks best in full screen. Hit the "more" button at the bottom to get the full screen version. The link is here should the embedded version fail.

I argued that dodgy cost studies have pushed out the demand curve for paternalistic policies by making them seem more necessary; at the same time, 'nudge' policies have served as a supply shock reducing the perceived cost to liberty of paternalistic policies. The policies are far less beneficial and are more costly than voters expect.

Being at Mont Pelerin has been awesome; I'm very grateful for having been invited.

I'll put up the conference paper sometime down the track after I've incorporated the excellent comments I there received.


  1. I will wait for the paper, which sounds very interesting. The prezi thingy is right up there with flying bullet points in the annoyance scale.

  2. Your timing is fantastic! Do browse across to your choice of local news site and read about the latest 'research' from Otago Uni. This time evil capitalism has gone too far! Alcohol cheaper than water ?!

    Now I can justify stockpiling Tui for the next earthquake. Sweet!


  3. I'm really digging that Prezi thing, Luis!

  4. I'm guessing the slides don't do your argument justice, but I'm really interested in your arguments against nudges. You start out saying that the the cost of nudge policies are underestimated, but your section on nudges seem to be about motivation and effectiveness. Should I wait until you post the paper?

  5. 1. The planner can't see into your utility function, so can't know whether a nudge makes you better or worse off. That's the socialist calculation debate argument taken to current policy. Consequently, the paternalist will need to substitute something else for your utility function, and that's what we see in the healthist cost literature: a substitution of an epidemiological standard to stand in for that which could be "rationally" consumed.

    2. Rizzo and Whitman's "slippery slopes" argument: soft paternalism progresses to hard paternalism as soft paternalism proves less effective than desired.

    3. In principle, if used to roll back existing hard paternalism, nudges could be pretty cool. But there are few campaigns to do that; instead, nudges are being used to expand the range of paternalism.

    4. It's dangerous to let the modeler stand outside of the model; how can we assume that the bureaucrats in charge of nudges are less subject to the biases that plague the subject of their attentions? Moreover, if voters ultimately monitor, their behavioural problems are larger when acting as voters than when acting in their private lives.

    5. Behavioural economic theory remains fairly contested: it's unclear that it moves well from lab to field. Even hyperbolic discounting has been argued nonexistent (see Anderson, Lau et al 2010).

    6. Decent evidence that voters suffer more anomalies than shoppers. Why not introduce nudges at the ballot box?

    7. There's risk that pervasive nudges infantilize us - we stop using our moral muscles and judgmental skills.

  6. The whole point of a nudge is that it doesn't *make* you do anything, right?

    Let's take smoking. If smokers were rational agents, then they are buying cigarettes because they get net utility from them after taking the health costs into account.

    So, a nudge policy like displays ban would have no impact. That same rational agent would get the same utility from cigarettes, and face the same cost (maybe +5 sec on their purchase time), so they would make the same decision.

    So, doesn't the assumption that nudges *can* make people worse off (i.e. Cause people to make irrational choices) undermine the entire rational agent assumption?

  7. @Keith: Do note how the Aussie productivity commission's noting of nudge benefits to govt subsidization of precommitment mechanisms turned to Gillard's proposal of mandatory precommitment.

    More to your point, though, adding transaction costs and finding changes in behaviour hardly shows that the behaviour was rational in one case and not in the other. Imagine the limit case: no displays of anything at the store. You walk to the counter, tell them what you want, and you're given it. I'd bet the content of your cart changes. But I'd not call you irrational in either state.

  8. Yeah, I take your point about the Gillard decision.

    But for the second point, given that people walk to the corner dairy and spend $15(?) on a packet of cigarettes, it's a pretty big call to say that the additional transaction time of a few seconds is a significant - or even noticeable - increase in costs.

    But let's back away from the specifics for a moment. Your limit case is interesting.

    A few years back, I did some work for a multinational ad agency, writing PR stuff for their in-store research outfit. Their new thing at the time was using eye-tracking cameras to test how placements, colour combinations, etc. affected attention and purchasing decisions. It was all quite scientific - and quite effective too, supposedly. (Who knows - maybe they were lying their asses off about its effectiveness.)

    Of course, from an outsider's perspective, it was creepy as hell. They can influence purchasing decisions without changing the products or the price, and they've boiled it down to a science.

    This is the central debate around nudges, isn't it? Whether people actually make rational choices based on sound understand of cost-benefit, or whether they just buy shit because it's at eye-level and of a pretty colour. (Not a rhetorical question. I actually want to know whether this is what you're disputing.)

    If you're interested, I can show you some of the store display guidelines that tobacco companies send out to dairies. They have really specific instructions on how to construct "Power Walls".

    Retail display advertising in general and tobacco display advertising in particular are, basically proponents of nudge, and have been for decades. Their models of consumer behaviour are, clearly, much more successful than economists'.

    The motivations and competence of the government is a different question, but once you pull down the rational agent theory, then the debate changes completely...

  9. I would never dispute that things at eye level are more likely to sell than things that are harder to see; I'd have to reckon every store's slotting fees to be hopelessly inefficient if that were the case. And of course tobacco displays, like every other product's display, will be designed to maximize their chances of selling products.

    But I can't see why any of that means that the rational agent model is wrong. Customers are optimizing while they're scanning the shelves, and things at eye level are easiest to spot. And I'm not sure that we wind up at any kind of inefficiency in which products are placed where so long as the store is profit maximizing and charging slotting fees properly. They'd want their highest margin products at eye-level, unless some supplier has paid them enough to put a lower margin product at that spot. Either way we wind up with a bit of price discrimination where shoppers willing to spend extra time and effort to scan the shelves for bargains do a bit better than the ones that don't. Absolutely none of that sounds like irrationality to me.

    Why can't "I like shit that's of a pretty colour" be part of a rational optimizer's utility function?