Twitter (X) and inflation dynamics: Investigating the role of Twitter sentiment in Turkiye's hybrid new Keynesian Phillips curve

This study explores the role of Twitter-based sentiment in shaping inflation expectations within Türkiye's Hybrid New Keynesian Phillips Curve (HNKPC) in a closed-economy framework. By incorporating a novel Twitter-based inflation expectations variable (TWT), derived using Natural Language Processing (NLP) with the Bidirectional Encoder Representations from Transformers (BERT) model, as a forward-looking component, this research offers an alternative to traditional methods like rational expectations and survey data. The findings reveal that while backward-looking inflation (past inflation) remains the dominant driver of expectations, Twitter sentiment, representing forward-looking expectations, influences actual inflation, though with a relatively weaker effect. This suggests that while past inflation plays a crucial role, TWT offers useful, up-to-date observations of the public's views on future inflation. The study also shows that in this closed model, the output gap is not significant, reinforcing the notion that inflation persistence remains central to Türkiye's inflation dynamics. The results suggest that policymakers should consider integrating Twitter sentiment into their strategies to enable more effective and dynamic communication with the public. By better understanding and responding to inflation expectations across traditional and social media, especially Twitter, authorities can enhance the credibility of inflation-targeting frameworks and potentially reduce inflation more efficiently.