Friday 1 June 2012

Libertarian paternalism is an oxymoron

Remember those Sci-Fi movies from the '50s where the mad scientist, who only had the good of the world at heart and proceeded with his experiments despite much warning, winds up on his knees crying out to the sky "It wasn't supposed to be like this! This isn't what I wanted!"

Here's Richard Thaler on Mayor Bloomberg's ban on soda:
To which Justin Wolfers replied:
There isn't much distance from nudge to shove, especially when preference heterogeneity keeps nudgers from recognizing when they're shoving. Justin doesn't see it as too costly, likely because he isn't a soda-fan either. So how could it be costly? A religious nudger could similarly see it as pretty costless to encourage at-home parenting by banning having a nanny employed more than 3 days a week; a parent could always opt-out by hiring a second nanny.

Should we blame Thaler? Here's what he had to say two years ago when pressed on whether his libertarian paternalism wouldn't lead to bans and harder paternalism:
In short, the risk of the slippery slope appears to be a figment of Professor Whitman’s imagination, and clear evidence of his bathmophobia. To be fair to him, this phobia is hardly unique to him and Professor Rizzo. Slope-mongering is a well-worn political tool used by all sides in the political debate to debunk any idea they oppose.
Instead of slope-mongering we should evaluate proposals on their merits. (We devote a chapter of Nudge to an evaluation of the choice architecture used in Sweden’s social security experience.) Helping people make better choices, as judged by themselves, is really not a controversial goal, is it?
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.

I still like my old review of Nudge, in which I proposed some nudges at the ballot-box:
While Sunstein worries about our decisions over investment plans or our weakness of will at the buffet table, I worry about our decisions at the voting booth. We vote infrequently, there’s no feedback from our personal voting decision to any policy outcome (unless you happen to hit Lotto by breaking a tie), the voting decision is complex and we may have little grasp of the issues at stake let alone our own positions on those issues. In my own research, I’ve found that only about half of voters in 2005 could place National, United Future, and Labour correctly on a left-right spectrum, for example, and that individuals’ political knowledge independently affects their policy and party preferences even after controlling for income, education, race, employment, gender, and other demographic characteristics. And so I think we (by which I mean you) need a nudge. Under my libertarian paternalistic voting system, your electoral enrolment would be linked to your census details. You’d then answer a brief questionnaire when entering a computerized voting booth, and I’d tell you, through the computer’s algorithms, for whom you should vote. Trust me: I’d be choosing the option that really would be best for you, if you only understood all of the policies supported by each of the parties and had a PhD economist’s understanding of the likely effects of these policies. You’d still be free to pick some other candidate or party, but you’d have to first reject the default choice I’d pick for you. The remaining options would then be presented in an order designed to maximize the chances of your choosing the next best option.

I trust that you find this kind of scheme repugnant. I’d find it great, so long as I got to be the choice architect. But opinions surely would vary, and I’d surely oppose the scheme if anyone other than me got to be the architect. The problem is that most of the arguments against my scheme cut similarly against Sunstein’s. More worrying, Sunstein seems pretty happy to blur the line between nudges and shoves: increasing cigarette taxes to discourage smoking is surely paternalistic, but is a bit stronger than a nudge. And, honestly, even the choice preserving nudges, like cars that nag you about the petrol you could save by easing up on the pedal, sound thoroughly unpleasant: I’d be nudged into learning enough automotive electronics to cut the right wires.

Update: It's occurred to me that I can't assume that everybody in the world has read the brilliant piece that started all this: Sunstein and Thaler's "Libertarian Paternalism Is Not An Oxymoron". Restricting it to choice of defaults, it's pretty tough to fault their argument. Except it never gets restricted to choice of defaults, now, does it?


  1. I'm reminded of the Maori Party's proposal that Maori voters be automatically enrolled on the Maori roll with the option of opting out. That's quite close to your reducto ad absurdum - they're not quite opting you in to voting for the Maori Party, but they are opting you in to creating more seats for the Maori Party to win. (Though it's possible that this will be negated by adding new voters who are probably less inclined to support the Maori Party).

    1. Oh wow. I had entirely missed that one. Very nice play on their part; I salute them. Lovely gerrymandering.

  2. You might be interested in my discussion of the subject, including two examples of a real world slippery slope, at:

    1. Very nice! Thanks for the reminder. Your closing paragraph there was great.

  3. Came across a relevant passage from an article in the European Economic Review (Haavio and Kotakorpi, 2011, political economy of sin taxes):

    "Many economists have been concerned that uniform sin taxes as well as other paternalistic policies aimed at helping irrational individuals are often detrimental for the welfare of rational individuals. This has resulted in an emphasis on a search for policies that help irrational individuals while having only a small or no impact on those who are rational. However, it seems natural that economists should not restrict themselves to study minimal interventions, but we should also engage in analysing optimal policies."

  4. Paper's cute.

    But it applies to a world where the regulator wants to help people with self-control problems as those with self-control problems judge things.

    Run that system instead in a world where a bunch of the regulators see the sin tax as something aiming at abolishing sin ("Sin Free by 2025!"), and you can see why I might not like the system. It'd morph almost immediately in policy into a system where each person gets a maximum allocation based on what the Ministry of Health thinks is consistent with health rather than an allocation based on what you want to set for self-binding. And then the individualised data generated by the system would turn into personalised nudges and shoves.

    The second this starts emerging as any kind of viable policy option, I'm running to buy a still. They're legal here. But wouldn't be for long under this setup.

  5. Ha - I was checking a working paper version that differed substantially from the one you'd cited - the titles were close, but they're clearly different papers. The working paper I'd read and here noted was looking at a system where people buy control cards at the start of the year saying what their drinking quota was to be; the card then didn't let them get more than that level of consumption. Rather different paper.