
Behavioral Economics — Replicability, Behavioral Welfare Economics and the Limits of Policy
Behavioral Economics — What to Believe, and What Standard to Design Policy By
Chapters 1–9 looked at where the assumptions of the standard model go wrong. The final chapter closes the course with two questions. How far can those findings be trusted? And if people’s choices may not be good for them, what standard should policy use?
1. Replicability: what survived?
In the 2010s, a wave of studies re-tested famous findings in psychology and the social sciences on a large scale. When the Open Science Collaboration (2015) replicated 100 psychology studies, about 36% produced statistically significant results in the same direction as the originals. When Camerer et al. (2016) replicated 18 laboratory experiments from leading economics journals, 11 (about 61%) replicated.
| Finding | Strength of evidence | Basis |
|---|---|---|
| Direction of loss aversion and the endowment effect | Robust | Repeated across methods and field data (size depends on context) |
| Present bias | Robust | Field behaviour (gyms, saving) and demand for commitment |
| Default effects | Robust | Large natural experiments in administrative data |
| Overall size of nudge effects | Overestimated | The average across all nudge-unit experiments is about one-sixth of the journal average (Chapter 9) |
| Ego depletion | Failed to replicate | Effect close to 0 in a preregistered replication across 23 laboratories |
| Priming (irrelevant words change behaviour) | Mostly failed to replicate | Multiple replication studies |
Ego depletion is the hypothesis that willpower is a resource that runs down with use. Early studies reported that after a self-control task, persistence on the next task fell. In the preregistered replication by Hagger et al. (2016) across 23 laboratories (about 2,100 participants), the effect was almost zero. The widely cited study claiming that “judges grant fewer paroles just before meals” was also reanalysed, showing that the order of cases was not random — for instance, cases without lawyers were bunched at the end of sessions. This does not deny any relationship between fatigue and decision-making, but the “willpower fuel” model and the dramatic effect sizes have lost their support.
The lesson of the replication problem is methodological: give priority to results backed by preregistration, large samples, joint replication across laboratories and cross-checks against field data.
2. When choices do not reveal welfare
Standard welfare economics relies on revealed preference: if someone chose A, A is better. Behavioral economics cuts this link. A person with the present bias of Chapter 4 puts off exercise today, but yesterday the same person wanted to exercise today. Which of the two choices reveals welfare?
| Approach | Standard | Difficulty |
|---|---|---|
| Long-run self | The preferences of a planner using only δ from the β-δ model | Why is the long-run self the 'true' one? |
| Choice-consistent domains | Use for welfare judgements only choices that do not contradict each other | The more contradictions, the less can be judged |
| Experienced utility | Satisfaction actually felt (survey and happiness data) | Measurement error, adaptation |
Whatever standard is chosen shapes the conclusion, so policy analysis must state which welfare standard it uses.
3. Internalities and corrective taxes
The externalities of Public Finance Chapter 1 were costs imposed on others. Costs that present bias leads people to impose on their own future selves are called internalities. Part of the cost of smoking and excessive sugar consumption is borne by one’s future self. If the long-run self standard is accepted, then, as O’Donoghue and Rabin showed, corrective taxes on internalities also raise welfare.
If the future health cost of a can of soft drink is 200 won and consumers take only half of it into account, the marginal internality is 100 won. The problems are distribution and heterogeneity. A sugar tax falls more heavily on low-income households (it is regressive) and is an unnecessary tax for consumers with little bias. The case strengthens when the revenue is returned to low-income households, or when there is evidence that more strongly biased groups respond more to the tax.
4. Do markets correct biases or exploit them?
The intuition that competition eliminates biases does not always hold. Firms have an incentive to design contracts that exploit biases.
- Gym fees (Chapter 4): monthly memberships are highly profitable with naive present-biased customers.
- Hidden add-on costs: strategies that shroud add-on costs, like printers and ink or bank fees, can survive competition (Gabaix and Laibson). A firm that discloses add-on costs loses naive customers, while sophisticated customers already avoid the add-ons, so disclosure gains nothing.
- Auto-renewing subscriptions: designs that make cancelling a chore (the sludge of Chapter 9) exploit present bias and status quo bias.
Here is the role of regulation. Mandatory disclosure of add-on costs, simplified cancellation procedures and cooling-off periods (the right to withdraw) are institutions that reduce the profitability of contracts that exploit biases.
5. Course summary
| Chapter | Key model or formula | Signature evidence |
|---|---|---|
| 2 | v(x): reference point, λ ≈ 2.25 | Reflection effect, the Asian disease problem |
| 3 | w(p): inverse-S weighting | Allais paradox, mug experiment |
| 4 | β-δ discounting | Gym memberships, SMarT |
| 5 | Departures from Bayes' rule | Base-rate neglect, anchoring |
| 6 | Acquisition utility + transaction utility | Season tickets, beer on the beach |
| 7 | Fehr-Schmidt utility | Ultimatum and public goods games |
| 8 | Limits to arbitrage | Disposition effect, excessive trading, equity premium |
| 9 | Defaults and social norms | 401(k) automatic enrolment, energy reports |
Check your understanding
A tax of 250 won is imposed on a good with a marginal external cost of 150 won and a future health cost of 100 won that consumers ignore. Is this rate right for an unbiased consumer (one who takes all future costs into account)? For that consumer the internality is 0, so the corrective tax should be 150 won. The extra 100 won only creates excess burden for them. When biases differ across people, a single rate can only target the average internality, and this excess burden is the price.
References
- Open Science Collaboration, “Estimating the Reproducibility of Psychological Science,” Science (2015)
- Colin Camerer et al., “Evaluating Replicability of Laboratory Experiments in Economics,” Science (2016)
- Martin Hagger et al., “A Multilab Preregistered Replication of the Ego-Depletion Effect,” Perspectives on Psychological Science (2016)
- B. Douglas Bernheim and Antonio Rangel, “Beyond Revealed Preference: Choice-Theoretic Foundations for Behavioral Welfare Economics,” Quarterly Journal of Economics (2009)
- Ted O’Donoghue and Matthew Rabin, “Optimal Sin Taxes,” Journal of Public Economics (2006)
- Xavier Gabaix and David Laibson, “Shrouded Attributes, Consumer Myopia, and Information Suppression in Competitive Markets,” Quarterly Journal of Economics (2006)
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.