We provide a novel explanation for the geographic dispersion of borrower-lender relationships based on information frictions rather than competition. Firms may strategically select distant banks to increase lenders’ information production costs, securing larger loans under information-insensitive contracts. Our model predicts that higher-quality firms prefer distant lenders for information-insensitive contracts, while lower-quality firms use local lenders with information-sensitive terms. Using transaction-level data from a major Chinese bank, we find strong empirical support: higher-rated firms exhibit greater propensity for distant borrowing; local loans show stronger negative correlation between amounts and interest rates; and distant loan pricing demonstrates weaker sensitivity to defaults.

This paper investigates how transparency in lawsuit information affects capital allocation and aggregate industrial production. Greater transparency enhances the availability of information about firms' fundamentals, which can influence resource distribution. We exploit regional variations in courts' compliance with mandated judicial document disclosures in China, implemented since 2014, as a natural experiment. For firms with initially high marginal revenue products of capital (MRPK), a 10-percentage-point increase in legal transparency results in a 4.4% increase in physical capital and a 7.9% reduction in MRPK, relative to firms with lower MRPK. Additionally, regions with higher transparency experience a rise in aggregate output. Further analysis differentiating firms by ownership type, public listing status, and industry-level contract intensity enhances the robustness of our findings.

We examine whether large language models (LLMs) can extract contextualized representation of Chinese public news articles to predict stock returns. Based on representativeness and influences, we consider seven LLMs: BERT, RoBERTa, FinBERT, Baichuan, ChatGLM, InternLM, and their ensemble model. We show that news tones and return forecasts extracted by LLMs from Chinese news significantly predict future returns. The value-weighted long-minus-short portfolios yield annualized returns between 35% and 67%, depending on the model. Building on the return predictive power of LLM signals, we further investigate its implications for information efficiency. The LLM signals contain firm fundamental information, and it takes two days for LLM signals to be incorporated into stock prices. The predictive power of the LLM signals is stronger for firms with more information frictions, more retail holdings and for more complex news. Interestingly, many investors trade in opposite directions of LLM signals upon news releases, and can benefit from the LLM signals. These findings suggest LLMs can be helpful in processing public news, and thus contribute to overall market efficiency.

How does labor risk affect corporate’s bond financing? Using the unique institutional feature of government regulations in China, we provide robust evidence that firms with a larger employment size have significantly better access to bond credit. This effect is more pronounced in times of local labor market deterioration or economic slowdown, for low-skill intensive industries, or in places with career-driven government officials. Our results are not driven by differential financial constraints or information frictions. We further show that the employment bias allocates bond credit towards under-performing large employers and the performance-enhancing benefits from bond issuance diminishes with employment size.