Asymmetric paternalism
Policies that help people who make mistakes while imposing little or no cost on those who are fully rational.
What it means
Asymmetric paternalism is a design principle for regulation: an intervention qualifies if it creates large benefits for people prone to error or self-control problems while imposing only small costs (or none) on people who already act in their own best interest. The asymmetry is the point — it sidesteps the standard objection to paternalism by protecting the fallible without coercing the competent, making it attractive across the political spectrum. Cooling-off periods, clear default options, and mandated disclosures are classic examples: they rescue the impulsive or uninformed while a savvy person can opt out at negligible cost. It is closely related to libertarian paternalism but framed explicitly around a cost-benefit test weighted by who bears the cost. Its critique is that judging who is 'irrational' and pricing the costs is contestable. It matters as the ethical and economic justification many nudges actually rest on.
The test behind the label
Camerer, Issacharoff, Loewenstein, O'Donoghue and Rabin did not simply argue that helping the fallible is decent. They set out arithmetic. Split the affected population into people who err and people who do not. The intervention's expected value is the benefit to the first group, weighted by its size, minus the cost borne by the second, weighted by its size, minus what the policy costs to run and enforce. Asymmetry is a claim about the shape of that sum, not about the designer's intentions. The title, Regulation for Conservatives, advertises the pitch: if the competent lose almost nothing, the liberty objection loses its bite. The same framing sorts interventions by how much they charge the competent, from information, through framing and defaults, to cooling-off periods and bans. Loewenstein, Brennan and Volpp later carried the framework into health behaviour, where the same test governs clinical and public-health design.
What the evidence shows
In practice the benefit side of the test gets measured constantly; the cost side almost never does. DellaVigna and Linos assembled 126 trials covering 23 million people across two US nudge units and found an average take-up effect of 1.4 percentage points — an 8% lift on a 17.4% baseline — against 8.7 points, a 33% lift, in comparable academic publications. The large benefits half of the test is routinely overstated by a mix of publication bias and the fact that trials run at scale are mostly low-touch letters, so an intervention that looked asymmetric on paper can shrink to a fraction of its advertised size once it ships. The cost side is not zero by assumption either. Bernheim, Fradkin and Popov modelled why 401(k) defaults stick at all, separating opt-out costs, procrastination, inattention and anchoring, and found the welfare verdict flips with the model chosen, supporting a default at the employer match rate or, against received wisdom, at zero. The asymmetry is an empirical claim that is rarely tested.
Where the asymmetry breaks down
Two premises carry the argument, and both are contestable. First, someone has to identify who is erring, yet the same behaviour is often error in one person and preference in another. Sunstein's account of nudges that fail turns on exactly this: strong prior preferences, or a counternudge from a firm with the opposite interest, absorb the intervention. Rizzo and Whitman press harder, arguing that pricing another person's error requires knowing their true preferences, and that rational-choice axioms get used as the normative yardstick rather than defended as one. Second, opting out is not free. If defaults work partly by anchoring rather than by inertia alone, the escape hatch does not leave the competent untouched; it moves them too, and the claimed asymmetry quietly disappears.
Using it in practice
Treat the label as a test to be passed, not a badge to be claimed. Before building, write down four numbers: how many people are erring, what the error costs them, what the escape hatch costs everyone else in time and attention, and what the policy costs to run. If you cannot fill in the third, you have not shown asymmetry, you have assumed it. Design the exit so it takes one action and demands no justification; a cancellation that requires a phone call is sludge wearing the label. Then read the opt-out rate as evidence rather than leakage. A high rate among the people you classed as competent is the test failing in public, and the cheapest signal you will get that the diagnosis was wrong.
Examples
A mandatory cooling-off period on payday loans helps impulsive borrowers reconsider while barely inconveniencing a borrower who genuinely needs the cash and will return.
Automatically enrolling staff into the workplace pension rescues everyone who never got round to signing up, while someone who genuinely needs the cash today opts out with a single click.
Requiring a plain reminder email before a free trial starts charging saves the people who forgot, and costs the people who meant to subscribe ten seconds and a delete.
A pharmacy app that defaults a repeat prescription to the generic catches the patients who never think to ask for it, and costs a patient with a real reason to want the brand one tap to switch back.
A betting site that makes every new user set a deposit limit before their first bet forces an active choice rather than a default: the cost to the competent is one screen, not an opt-out, while people who would never think to set a limit end up with one.
First described in Camerer, Issacharoff, Loewenstein, O'Donoghue & Rabin (2003).
Key references
- DellaVigna, S., & Linos, E. (2022). RCTs to scale: Comprehensive evidence from two nudge units. Econometrica, 90(1), 81-116. doi.org/10.3982/ECTA18709
- Rizzo, M. J., & Whitman, G. (2020). Escaping paternalism: Rationality, behavioral economics, and public policy. Cambridge University Press. doi.org/10.1017/9781139061810
- Sunstein, C. R. (2017). Nudges that fail. Behavioural Public Policy, 1(1), 4-25. doi.org/10.1017/bpp.2016.3
- Bernheim, B. D., Fradkin, A., & Popov, I. (2015). The welfare economics of default options in 401(k) plans. American Economic Review, 105(9), 2798-2837. doi.org/10.1257/aer.20130907
- Loewenstein, G., Brennan, T., & Volpp, K. G. (2007). Asymmetric paternalism to improve health behaviors. JAMA, 298(20), 2415-2417. doi.org/10.1001/jama.298.20.2415
- Camerer, C., Issacharoff, S., Loewenstein, G., O'Donoghue, T., & Rabin, M. (2003). Regulation for conservatives: Behavioral economics and the case for "asymmetric paternalism". University of Pennsylvania Law Review, 151(3), 1211-1254. doi.org/10.2307/3312889