Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Simulating Tenant Responses to Energy Policy Interventions with Transaction-Cost-Aware LLM Age

Source

arXiv

arxiv.org/abs/2607.24341v1

Read the source

Source

Plain-Language Brief

Researchers developed a method to simulate how tenants respond to energy-efficient renovation policies using large language models (LLMs). They introduce a friction-aware persona modeling approach that takes into account perceived transaction costs (PTCs) such as information burden, administrative effort, coordination demands, and perceived uncertainty. The study compares various LLMs and evaluates two methods: supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO). Results indicate that the PTC-based persona design improves model performance across both prompt-only and fine-tuned settings.

What the Document Says

  • The research uses LLMs to simulate tenant responses to energy-efficient renovation policies.
  • They introduce a friction-aware persona modeling approach that considers PTCs.
  • PTCs include information burden, administrative effort, coordination demands, and perceived uncertainty.
  • Tenants are represented by their demographics as well as their perceptions of costs, benefits, barriers, and uncertainties associated with proposed renovation plans.
  • The study collects survey data from 1,068 Dutch citizens to compare prompt-only and fine-tuned settings across GPT-3.5-turbo, Ministral-8B-Instruct, and Llama-3.1-8B-Instruct.
  • They evaluate SFT and GRPO for local open-weight models.
  • Results show that incorporating PTC-based personas and reasoning improves model performance across both prompt-only and fine-tuned settings.

Key Concepts Mentioned

  • Large language models (LLMs)
  • Energy-efficient renovation
  • Perceived transaction cost (PTC)
  • Persona modeling approach
  • Survey data
  • GPT-3.5-turbo
  • Ministral-8B-Instruct
  • Llama-3.1-8B-Instruct
  • Supervised fine-tuning (SFT)
  • Group Relative Policy Optimization (GRPO)

Claims Made by the Document

  • Incorporating PTC-based personas and reasoning improves model performance.

Evidence Used

  • Results from the study show improved model performance across both prompt-only and fine-tuned settings.

Methods / Approach

  • LLM-based simulation
  • friction-aware persona modeling approach
  • survey data collection
  • comparison of LLMs
  • supervised fine-tuning (SFT)
  • Group Relative Policy Optimization (GRPO)

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

The study may not account for all potential limitations, as the text does not provide explicit details on any limitations.

Open Questions

  • What are the specific benefits and costs that tenants perceive associated with proposed renovation plans?
  • How does the incorporation of PTC-based personas impact the accuracy of the simulations?

Notes for Future Q&A

May include interpretation. Labeled where it does.

  • The study compares multiple LLMs but does not provide details on how they differ.
  • It's unclear what specific PTC factors are considered in the persona modeling approach.

Raw Abstract / Excerpt

Recent studies use Large language models (LLMs) to simulate human opinions and decisions by prompting models with demographic, attitudinal, or persona-based descriptions. Yet such simulations rarely model the practical, cognitive, or social frictions that shape how people respond to policy interventions. Perceived transaction cost (PTC) provides a useful lens for modeling the practical frictions that shape policy responses, such as information burden, administrative effort, coordination demands, and perceived uncertainty. We use this lens to develop a friction-aware persona modeling approach for LLM-based simulation. In the context of energy-efficient renovation (EER), tenants are represented not only by who they are demographically, but by how they perceive the costs, benefits, barriers, and uncertainties associated with proposed renovation plans. Using survey data collected from 1,068 citizens in the Netherlands, comprising approximately 40,548 survey question and answer pairs, we compare prompt-only and fine-tuned settings across GPT-3.5-turbo, Ministral-8B-Instruct, and Llama-3.1-8B-Instruct, and evaluate supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO) for local open-weight models. Results show that incorporating PTC-based personas and reasoning consistently improves model performance across both prompt-only and fine-tuned settings, suggesting that PTC-based persona design provides a useful bridge between institutional policy theory and interpretable LLM-based policy simulation. Code is available at https://github.com/xiaweijie1996/socialagent.