Analysing the sentiment-based variation in finance ministers’ communication
摘要
This study underscores the importance of contextually customizing the lexicon framework to improve the accuracy of sentiment scoring in textual data. We adapt the commonly used Loughran and McDonald lexicon to reclassify, remove, and incorporate words to capture their nuanced meanings within the public finance language accurately. The revised methodology shifts the sentiment score from negative to positive in 43% of budget speeches delivered by Indian finance ministers since 1980. Our analysis indicates a progressive increase in positive sentiment within budget speeches since 2016, except for the COVID-19 years, along with a shortening of speech length. The regression analysis suggests that budget speeches delivered by finance ministers during the single-party governments formed by the political party, the Indian National Congress, were the least positive, controlling for the macroeconomic environment. In contrast, those during the Bhartiya Janata Party (BJP) majority government were the most positive. The budget speeches during the coalition governments fall between these extremes, but the speeches by the BJP finance ministers within a coalition have a higher positive sentiment. Finance ministers’ speech sentiment is negatively associated with lagged inflation rates but not with economic growth and crisis periods.