Files
Abstract
Does the recent surge in Chinese imports affect the media slant against China in the United States? Using a dataset of 157 U.S. local newspapers from 1998 to 2017, I construct a new measure of media bias through machine-learning-based sentiment analysis. Implementing the shift-share empirical strategy on the newspaper level, this paper finds that newspapers whose circulation states face greater exposure to Chinese import competitions report more negative news about China. The source of negative description more stems from non-trade-related topics and editorials. Though the findings in Lu et al. (2018) remain unchanged, this paper argues an overestimation tendency in the previous literature by matching keywords and tries to provide less unbiased point estimates.