Do differing tones in Chinese and English public discourse influence the A–H premium amid the U.S.–China trade friction

This dataset covers the daily A–H share premium for 85 firms that are dual-listed in Mainland China (A shares) and Hong Kong (H shares) from 2 January 2025 to 30 May 2025, together with a Chinese–English media/public-discourse tone-gap measure (Tone_gap) constructed from the GDELT Global Database. The dataset is organized as a firm-by-date daily panel with 149 calendar days and 85 firms (12,665 observations). To align stock-market variables with the daily tone series, a balanced calendar panel is created for each firm and missing premiums on non-trading days (or suspensions) are filled by linear interpolation; is_trading_day flags original trading-day observations. The dataset also includes macro/market controls—exchange-rate futures volatility (future_vol_filled), an A–H return-gap control (ret_ah_filled), and the IPO-count gap between the A-share and H-share markets (ipo_gap)—as well as firm classifications (WIND sector: sector; GICS sector: gics). Recommended use: winsorize continuous variables at the 1st/99th percentiles and standardize prior to estimation, then estimate panel models with firm fixed effects.