Twitter-Based Political Sentiment and Macro-Financial Indicators: A Case Study of Indonesia’s 2024 Presidential Transition

Periods of presidential transition in emerging markets may coincide with heightened political and economic uncertainty. This study examines whether Twitter-based political sentiment during Indonesia’s 2024 presidential transition is associated with short-term USD/IDR movements and year-on-year headline inflation. IndoBERT and IndoELECTRA were adapted to Indonesian political discourse using approximately 150,000 unlabeled tweets and evaluated on 10,000 manually labeled tweets using stratified five-fold cross-validation. Weekly USD/IDR log returns were analyzed using OLS regressions with lagged political sentiment, Brent crude oil prices, the Effective Federal Funds Rate, and Bank Indonesia’s BI-Rate, with Newey–West inference and additional diagnostic and sensitivity checks. Monthly Pearson correlations were used separately to examine contemporaneous associations between political-discussion measures, the monthly average USD/IDR exchange rate, and inflation. Both sentiment models achieved accuracy of approximately 0.84. Lagged sentiment was not statistically significantly associated with weekly USD/IDR returns, and the monthly exchange-rate correlations were also weak and statistically imprecise. Monthly inflation correlations were positive but uncertain, resembled the association observed for total tweet activity, and weakened when the partial October observation was excluded. Overall, Twitter-based political sentiment appears more useful as a contextual measure of political discussion than as a stand-alone signal of short-term macro-financial movements.