The China Backlash: Quantifying the Narrative Discourse on China
Using text-as-data methods, we quantify the U.S. discourse on China, place it in historical perspective, and compare its evolution across different sources of public narrative. Our analysis is based on a large corpus of materials drawn from public records (presidential and congressional records, newspaper articles, social media, think tank reports). We develop two measures of narrative discourse: an indicator of the frequency with which China is covered and a proxy for how positively or negatively China is presented. We document a sustained decline in net sentiment towards China in presidential sources since the early 2000s, predating Donald Trump’s first term in office. In contrast to presidential documents, congressional, news media, social media, and think tank sources exhibit persistent negative sentiment towards China even in decades prior to the 2000s. Instead of a deterioration in sentiment across the board, what we find is a convergence in presidential sentiment to other sources of narrative discourse. We find no evidence that presidential leadership has played a significant role in setting the narrative tone for the nation as a whole. Historical and shorter-term evidence both point to a bottom-up process of diffusion of narratives, rather than top-down diffusion.
-
-
Copy CitationMert Geyiktepe and Dani Rodrik, "The China Backlash: Quantifying the Narrative Discourse on China," NBER Working Paper 35539 (2026), https://doi.org/10.3386/w35539.Download Citation