An LLM Approach to Study Expectation Management Frictions under China's Dual-Track Regulation and Multi-Objective Constraints

Zeqin Liu, Zongwu Cai and Ying Fang
Additional contact information
Zeqin Liu: School of Statistics, Shanxi University of Finance and Economics, Taiyuan, Shanxi 030006, China
Zongwu Cai: Department of Economics, The University of Kansas, Lawrence, KS 66045, USA
Ying Fang: The Wang Yanan Institute for Studies in Economics, Xiamen University, Xiamen, Fujian 361005, China and Department of Statistics & Data Science, School of Economics, Xiamen University, Xiamen, Fujian 361005, China

No 202613, WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS from University of Kansas, Department of Economics

Abstract: Expectation management is a critical yet challenging task for central banks, particularly within the Chinese context of dual-track regulation and multi-objective constraints. This paper investigates the transmission efficacy of the People's Bank of China's (PBoC) forward guidance by proposing an LLM-powered analytical framework. First, we employ adaptive semantic segmentation to partition communication texts based on genuine meaning shifts. Second, we apply tense-by-topic tagging to precisely separate forward-looking signals from retrospective ones and isolate monetary policy stances from macroeconomic assessments and other auxiliary themes. Third, we further decompose these forward-looking policy stance units into granular signals regarding overall tone, quantity tools, price tools, and targeted objectives, and subsequently classify these signals as accommodative, neutral, or tight. Based on this granular data, we construct three specialized indices: forward-looking net policy intention (NPSf), quantity-price signal divergence (QPSD), and multi-objective communication dispersion (MDI). Controlling for actual policy operations and macroeconomic variables, we employ local projections to identify the dynamic effects and friction mechanisms of expectation management. Empirical results reveal that forward-looking monetary policy communication is the cornerstone of expectation management, whereas retrospective statements have been fully priced in by the market. Specifically, the credit channel functions effectively; forward-looking intentions drive substantive adjustments in credit growth, real financing costs, and risk premiums. Conversely, transmission through interest rate expectations and asset price channels remains limited. Further analysis demonstrates that price-quantity signal divergence systematically weakens transmission across all channels. Moreover, multi-objective communication triggers an overshooting response in both short-term money market benchmarks and credit risk premiums, while objectively dampening the pricing sensitivity of equity assets. We suggest that central bank expectation management should prioritize strengthening forward-looking path guidance, supported by highly coordinated price-quantity signals and clearly defined dominant objectives, to enhance policy efficacy.

Keywords: Expectation Management; Large Language Models; Forward Guidance; Dual-Track Monetary Policy; Multiple Policy Objectives; Local Projections (search for similar items in EconPapers)
JEL-codes: C55 E52 G12 (search for similar items in EconPapers)
Date: 2026-05, Revised 2026-05
New Economics Papers: this item is included in nep-inv and nep-mon
References: Add references at CitEc
Citations:

Downloads: (external link)
https://kuwpaper.ku.edu/2026Papers/202613.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:kan:wpaper:202613

Access Statistics for this paper

More papers in WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS from University of Kansas, Department of Economics Contact information at EDIRC.
Bibliographic data for series maintained by Professor Zongwu Cai ().