Learning to Unlearn: Education as a Remedy for Misspecified Beliefs
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866909875009748992 |
|---|---|
| author | Fedyaeva, Daria Lukyanov, Georgy Tollié, Hannah |
| author_facet | Fedyaeva, Daria Lukyanov, Georgy Tollié, Hannah |
| contents | We study education as a remedy for misspecified beliefs in a canonical sequential social-learning model. Uneducated agents misinterpret action histories - treating actions as if they were independent signals and, potentially, overstating signal precision - while educated agents use the correct likelihoods (and may also enjoy higher private precision). We define a misspecified-belief PBE and show existence with a simple structure: education is a cutoff in the realized cost and actions are threshold rules in a single log-likelihood index. A closed-form value-of-education statistic compares the accuracy of the educated versus uneducated decision at any history; this yields transparent conditions for self-education. When a misspecified process sustains an incorrect cascade, uniformly positive private value and a positive flip probability imply that education breaks the cascade almost surely in finite time, with an explicit bound on expected break time. We quantify welfare gains from making education available and show how small per-education subsidies sharply raise de-cascading probabilities and improve discounted welfare. Extensions cover imperfect observability of education choices and a planner who deploys history-dependent subsidies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_24735 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Learning to Unlearn: Education as a Remedy for Misspecified Beliefs Fedyaeva, Daria Lukyanov, Georgy Tollié, Hannah Theoretical Economics We study education as a remedy for misspecified beliefs in a canonical sequential social-learning model. Uneducated agents misinterpret action histories - treating actions as if they were independent signals and, potentially, overstating signal precision - while educated agents use the correct likelihoods (and may also enjoy higher private precision). We define a misspecified-belief PBE and show existence with a simple structure: education is a cutoff in the realized cost and actions are threshold rules in a single log-likelihood index. A closed-form value-of-education statistic compares the accuracy of the educated versus uneducated decision at any history; this yields transparent conditions for self-education. When a misspecified process sustains an incorrect cascade, uniformly positive private value and a positive flip probability imply that education breaks the cascade almost surely in finite time, with an explicit bound on expected break time. We quantify welfare gains from making education available and show how small per-education subsidies sharply raise de-cascading probabilities and improve discounted welfare. Extensions cover imperfect observability of education choices and a planner who deploys history-dependent subsidies. |
| title | Learning to Unlearn: Education as a Remedy for Misspecified Beliefs |
| topic | Theoretical Economics |
| url | https://arxiv.org/abs/2510.24735 |