Metraipolis › archive › dc_ca458952399e

Can Large Language Models Anticipate Behavioral Responses to Social Policies? A Case of Pension Enrollment Prediction among China's Flexible Workers — arXiv:2609.05189

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.05189
License (as archived): CC0-1.0
Archived: 2026-10-16T22:35:33.945Z
SHA-256 of the archived text: 352c40d8bae5af85f373bee6b00f00230317a81a211f2f44f1f839bba4758263
Stable URL of this page: https://metraipolis.com/documento/dc_ca458952399e

How to cite

Can Large Language Models Anticipate Behavioral Responses to Social Policies? A Case of Pension Enrollment Prediction among China's Flexible Workers — arXiv:2609.05189 — arXiv (en), https://arxiv.org/abs/2609.05189 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ca458952399e (Metraipolis archive dc_ca458952399e, sha256 352c40d8bae5af85f373bee6b00f00230317a81a211f2f44f1f839bba4758263)

BibTeX

@misc{dcca458952399e,
  title = {Can Large Language Models Anticipate Behavioral Responses to Social Policies? A Case of Pension Enrollment Prediction among China's Flexible Workers — arXiv:2609.05189},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.05189},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ca458952399e; sha256 352c40d8bae5af85f373bee6b00f00230317a81a211f2f44f1f839bba4758263},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.05189. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Assessing the impacts of social policy changes is a widely acknowledged challenge for policymakers. Econometric methods can be unreliable when extrapolating to hypothetical scenarios, while field pilot programs are highly costly. In this paper, we propose using large language models (LLMs) as policy-assessment tools adapted from general-purpose models. We present FlexPension-LLM, the first domain-specialized large language model for a hierarchical pension-enrollment prediction task among flexible workers in China, and introduce DKI-RDistill, which injects policy-grounded cues into the prompt, including Probit-derived marginal effects and hukou-province pension rules. The method then uses LoRA/SFT to distill rationale-augmented supervision into an open-weight MoE student, with teacher errors corrected by regenerating those cases under ground-truth labels. On a CHFS 2019 blind split, FlexPension-LLM achieves 0.9316 Composite F1, surpassing its Claude Sonnet 4.5 teacher and 15 of 17 baselines, and is statistically indistinguishable from Claude Opus 4.6. Across four external surveys, it averages 0.7549 Composite F1 and shows the narrowest performance range among the strongest systems. Component analysis shows that gains come mainly from policy-grounded cue injection and error-filtered supervision, while rationales provide decision traces that can be checked against policy rules.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.05189 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T22:21:16.299Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).