1. Attention to Long-Horizon Information

Abstract: I study how sell-side analysts allocate attention to information about the distant future. Using approximately 5.1 million equity research reports from 1998 to 2024, I extract discussions of outcomes at least five years ahead and term them as "long-horizon attention." Such attention is more prevalent when distant information is more relevant for valuation and when analysts face lower processing costs. Long-horizon attention is associated with more accurate earnings and target price forecasts. Recommendation revisions by analysts with greater historical long-horizon attention elicit stronger market reactions. The findings highlight selective attention to distant fundamentals in analysts' belief formation. 

2. Macro-Equity Logic Chains in the Stock Market

• Coauthors: Yukun Liu and Pengfei Sui

Abstract: We extract and evaluate "macro-equity logic chains"-explicit causal reasoning pathways linking macro shocks to firm-level outcomes—from five million analyst reports using large language models. We document substantial cross-sectional heterogeneity and unconditional sparsity in their adoption. Consistent with a framework of bounded rationality, analysts strategically adopt concise, firm-relevant logic chains, reflecting a fundamental trade-off between the predictive value of macroeconomic insight and the cognitive costs of sequential reasoning. We show that lower baseline cognitive costs, proxied by formal economic education, and accumulated domain-specific experience increase the propensity to employ these structures. Furthermore, the integration of explicit macro-equity reasoning improves earnings forecast accuracy, particularly when the logic chains align with the firm's empirical exposure to the underlying macroeconomic factor. Beyond forecast accuracy, this reasoning path translates into observable asset pricing impacts. 

3. Expected Earnings Smoothness

• Coauthors: Zhi Da, Rui Shen and Pengfei Sui

Get in touch at zhileliu@link.cuhk.edu.cn / LinkedIn .