
Behavioral Economics — Choice Architecture and Nudges
Behavioral Economics — There Is No Neutral Way to Present Choices
For any choice, someone arranges the options. A cafeteria has to put food somewhere, and a pension enrolment form has to pre-fill or leave blank every box. Thaler and Sunstein called this arrangement choice architecture, and called a design that changes behaviour predictably without removing options or significantly changing monetary incentives a nudge.
1. The power of defaults
Madrian and Shea (2001) compared outcomes before and after a US firm switched its retirement plan (401(k)) to automatic enrolment. Among employees hired when they had to apply to join, the participation rate was about 37%; among those hired when they were enrolled automatically unless they opted out, it was about 86%. The options were identical; the only thing that changed was the outcome of doing nothing.
The same study also showed a downside of defaults. Most people stayed at the default contribution rate (3%) and default investment of automatic enrolment. Some employees who would otherwise have chosen higher contribution rates were stuck at the default and actually saved less. Defaults raise participation, but the default level itself must be chosen carefully.
Defaults are powerful because several concepts from this course work at once.
| Channel | Related chapters |
|---|---|
| Putting off the effort of changing (present bias) | Chapter 4 |
| Feeling a move away from the status quo as a loss | Chapters 2 and 3 |
| Taking the default as the designer's recommendation | Chapter 5 (information as a signal) |
2. Social norms and comparative information
The US utility company Opower sent households reports comparing their electricity use with the average of their neighbours. Using data from randomized experiments on hundreds of thousands of households, Allcott (2011) estimated that the reports cut electricity use by about 2% on average. That may look small, but it is large for a letter that costs almost nothing, equivalent to the reduction that would require raising electricity prices by several per cent.
The UK tax authority raised payment rates by adding to reminder letters for late payers the sentence “most people in your area pay their tax on time.” It is an administrative application of the conditional cooperation of Chapter 7.
Comparative information carries the risk of a boomerang effect: households that used less than average may learn that “I use less than average” and increase their use. Opower reduced this effect by adding a mark of praise for low users.
3. Making information understandable
Chapter 5 showed that presenting frequencies instead of probabilities greatly reduces base-rate neglect. The same principle applies to disclosure design. Choices improve when loans show total repayment rather than only the annual interest rate, or when mobile phone plans show expected bills based on past usage. The key is not to give more information but to give it in the form the decision needs.
4. How effective are nudges?
| Source | Average effect |
|---|---|
| Nudge experiments published in academic journals | About 8.7 percentage points |
| All experiments run by two US nudge units with government | About 1.4 percentage points |
DellaVigna and Linos (2022) collected all 126 randomized experiments run by two US nudge units (a federal government unit and the North American office of the Behavioural Insights Team) and compared them with published nudge studies. The average effect of the government experiments was about one-sixth of the journal average. Most of the difference was explained by publication bias (only studies with large effects get published) and differences in experiment size. Nudges work, but often less than hoped, and the average of published studies should be discounted when estimating policy effects.
5. Libertarian paternalism and its critics
Thaler and Sunstein justified nudges as libertarian paternalism: steering people towards outcomes they would choose on reflection (more saving, healthier diets) while leaving freedom of choice intact. Since choice architecture cannot be avoided, there is no “neutral design,” and if we have to design anyway, we should design in people’s interest.
The criticisms run in three directions.
- Who decides what is “in people’s interest”? The designer’s judgement may differ from individual preferences. This is the problem of behavioral welfare economics in Chapter 10.
- The risk of manipulation: the same techniques can be used to harm people — subscription designs that make cancelling hard, or designs that set an unfavourable option as the default. Sunstein called this sludge, the opposite of a nudge.
- Sidestepping structural problems: if saving is low because incomes are low, adjusting defaults is not a fundamental solution. Nudges should be seen as complements to taxes, subsidies and regulation, not substitutes.
Check your understanding
In a firm with 1,000 new employees, if the participation rate is 37% under opt-in and 86% under automatic enrolment, how many more participants does automatic enrolment bring? 490. But if half of the 370 opt-in participants originally contributed 6% and after automatic enrolment stay at the 3% default, the contributions of those 185 people are halved. The number of participants and total saving are different indicators, and default design must look at both.
References
- Brigitte Madrian and Dennis Shea, “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior,” Quarterly Journal of Economics (2001)
- Hunt Allcott, “Social Norms and Energy Conservation,” Journal of Public Economics (2011)
- Stefano DellaVigna and Elizabeth Linos, “RCTs to Scale: Comprehensive Evidence from Two Nudge Units,” Econometrica (2022)
- Richard Thaler and Cass Sunstein, Nudge: The Final Edition (2021)
OIYO Editorial
Editorial DeskThe OIYO editorial desk researches money, law, lifestyle, and self-understanding topics against primary sources and public statistics. Every piece carries source notes and is reviewed on a regular cycle for accuracy and usefulness.