Slippery slopes are dangerous places

Apparently New York is banning large soft drink cup sizes to reduce sugar consumption and obesity. Eric has used this as a launchpad for a slippery slope argument attacking libertarian paternalism (LP). The original idea of Sunstein and Thaler was that you could design choices to minimise the likelihood of bad decisions without restricting choice. Every time people make a choice the person providing the choice gets to frame it. The premise is that, if they can do that in a way that helps people to make a good decision, then they should.

A ban on large soft drink cups would clearly not be a ‘nudge’ in the sense that Thaler and Sunstein mean it. It would restrict the choice set of diners so it’s purely paternalistic. That’s probably why Eric has pivoted to attacking the idea of nudges as being too easy for politicians to misinterpret:

For all the protests that “nudge” was supposed to have strong opt-out provisions, it was awfully predictable that it wouldn’t turn out that way in practice. I don’t know how much time Thaler spent working to ensure choice was preserved in his proposed choice-preserving architecture, but he did spend a bit of time telling libertarians that this sort of thing couldn’t happen.

This argument against LP is silly for a number of reasons.

  1. First, as Eric acknowledges, the proposed regulation doesn’t fall within the set of mechanisms described by Sunstein and Thaler. The ban isn’t within the ambit of LP.
  2. Secondly, products have been banned for health reasons since long before LP existed. Even if the ban can be described as LP, it would have happened anyway.
  3. Thirdly, misuse of a concept by others is a cost that needs to be balanced against the benefits of the concept when used appropriately. Even if LP is responsible for the ban, it may still be a helpful idea for motivating regulation

Thaler and Sunstein have pointed to many instances in which better choice architecture would result in welfare increases. Even if they can be held responsible for bans such as this — which I find a stretch — the concept could still be beneficial in sum. I’m reminded of the way many economic concepts are misapplied to justify investment in stadiums, limits on alcohol sales, or restrictions in foreign investment. I don’t hear many complaints from economists that the whole idea of economics is a slippery slope to terrible policy and I don’t see how LP is different: it’s a useful idea that can help shape great policy. Unfortunately, there will always be a few people who misuse it to further their own ends. That doesn’t make it a bad idea.

  • Did you catch Justin Wolfers arguing that it was a nudge rather than a shove because people could still buy two cups?

    • Yes, I saw that but I didn’t want to make the post any longer that it already is! I think Matt replied on Twitter saying that it’s still a reduction in the choice set, even if you can achieve the same level of consumption by other means. I’d call a change in the default cup size a nudge, but it’s hard to see how anything classed as a ban is maintaining the choice set.

      • Hazy definitions around what gets to count as choice preserving always meant that LP was going to be bad in practice. Worse, because of the sunk cost fallacy in politics, any time a policy is advocated on LP grounds, harder paternalism in support of the goal is more likely to be supported subsequent to any failure of the LP policy to achieve the goal. All you need for this is the sunk cost fallacy to apply among voters. “We’ve already tried LP in pursuit of X which we decided MUST be worth the cost because we initiated policy LP. Now isn’t the time to back down! We need P!”

        And I’ll bite the bullet and say that the way that externalities and Pigou is typically taught at principles level does more harm than good given the ease with which complete rubbish gets passed off in public debate as measuring social cost. Actual set of externalities to which Pigou or regulation is appropriate is very narrow relative to the full domain of externalities; the way we teach it is a slippery slope to bad policy.

        • I think your starting point is wrong: people intuitively want to regulate things that they view as ‘bad’ and their default is to ban them or something close. If they’re aware of LP then they might consider more liberty-preserving measures than they otherwise would. I think LP is unlikely to result in more regulation than would otherwise have existed, although it might moderate the mechanisms that are used.

          Similarly, I doubt that the way externalities are taught makes much difference to the things that are regulated. Poor economics may provide a useful veneer of authority to regulations, but the desire for them is based on people’s pre-existing value judgments. If you look at the example of regulatory CBA (RIA in NZ), the required analyses have no discernable impact on the policies enacted, and that result holds across the Western world.

        • Bad social cost studies make folks think more things are bad for them through their pocketbook than is really the case; libertarian paternalism makes paternalism seem cheaper than it really is. Both work to increase the effective demand for paternalistic policy. 

          Agreed that LP can be used to scale back from hard paternalism to weak paternalism. Any examples of it ever happening anywhere? Can think of a fair bit of new paternalism couched as LP: precommitment cards for gamblers in Oz; all the soda nonsense; default opt-in Kiwisaver; some even push fat taxes as being nudges. Can’t think of any existing bit of paternalism that has been replaced by soft paternalism. 

        • Bad social cost studies make folks think more things are bad for them through their pocketbook than is really the case

          That’s possible, but I’d like to see evidence of it before being sure. I don’t know how much weight anyone really gives to the numbers that economists tout. Most research into policy development processes finds that the numbers are used to validate policies that would hvae been enacted regardless.

          Can think of a fair bit of new paternalism couched as LP: precommitment cards for gamblers in Oz; all the soda nonsense; default opt-in Kiwisaver

          Are precommitment cards and opt-in Kiwisaver not LP? I’m not familiar with the details so perhaps they’re not. I’m not really aware of any policy analyses that pay attention to the paternalistic nature of the proposed intervention. If they use LP terminology then I doubt it’s intentional except insofar as they try to minimise the perceived intrusion on people’s lives to gain acceptance of their proposal. I’d be really surprised if Thaler and Sunstein’s article had a measurable impact on policy development, as opposed to policy commentary!


        • Why would anybody bother commissioning dodgy social cost studies unless they helped sway public opinion to justify policy?

          Agree that opt-out Kiwisaver is actually LP. Mandatory precommitment cards aren’t – everybody has to go through the hassle of getting one, no opt-out [see the Australian policy debates].

          Look up the UK Nudge Unit sometime. Thaler was advisor there…. 

        • Why would anybody bother commissioning dodgy social cost studies unless they helped sway public opinion to justify policy?

          Because of the move towards evidence-based policy in the public sector, as demonstrated by the RIA requirements. If you check out the literature on CBA for policy development (Robert Hahn and Claudio Radaelli do a lot of work in the field) they struggle to find any measurable effect on policy. There’s still a strong belief that it’s because the processes and expertise aren’t in place, rather than because it’s inherently unworkable, though.

      • That’s not a crazy hypothesis. But the MoH commissioned stuff on alcohol didn’t feed directly into any RIS – it hit the LC’s review, and maybe they’d intended that it be around to justify stuff that would come post LC, but it’s awfully hard to read that paper and reckon that anybody at Treasury looking at the adequacy of an RIS would find it fit for purpose. The main point of it seemed instead to be the 5 second soundbite on Radio NZ on how horrible drinkers are in costing everybody else money, with the subsequent benefit of misleading Sir Geoffrey Palmer. 

        My hypothesis, that I still have to figure out a way of testing, is that these reports push the median voter, on dimensions where the median voter has weak preferences, to support greater paternalism because the issue framing flips from “other people doing silly things” to “other people doing silly things that cost me money”. 

  • Joe Connell

    Would it be LP if only smaller cups were sold, but there were free refills?

    • My feeling is that intervention to change the elements of the choice set steps outside the bounds of a nudge. It may be that many people who favour LP would like interventions that are more heavy-handed, but I think it’s hard to then call it libertarian in any sense.

      Changing defaults seems fairly uncontroversial, but the difficult bit arises if you consider any intervention that overcomes a behavioural bias to fall within the ambit of LP. That’s when smoking taxes to correct internalities get brought up, which obviously have large costs from heterogeneity of consumers in addition to any benefits from precommitment. Since it’s unlikely to be optimal from any single person’s point of view it’s harder to justify. Eric may well argue that all such schemes are advanced under the same umbrella but that conceals a lot of variation in the classes of regulation that are proposed.

  • Joe Connell

    And I’d say default opt-in kiwisaver, which I think most people would call out-out kiwisaver is LP.

  • Pingback: The importance of economists » TVHE()