A study of Implicit disagreement in MPC meetings for selected Eastern European Economies using the Taylor Rule

Abstract

This paper in an empirical attempt to study implicit disagreement among MPC members of central banks contained in the minutes of the MPC meetings for three Eastern European economies, viz., Poland, Czech Republic and Hungary. NLP sentiment analysis techniques, specifically a BERT model trained and fine-tuned on the corpus of the text of the minutes of the MPC meetings of the three countries, are applied to extract sentence level sentiment scores in terms of hawkish, dovish or neutral sentiment which are then used to calculate meeting level Disagreement Index(DI) variable by taking the Standard Deviation of the sentence level sentiment scores. We augment the standard Taylor rule formula for setting monetary policy rates with this DI variable and see that it impacts the interest rate negatively, suggesting that higher implicit disagreement might restrain the MPC members from being too hawkish which might be influenced by higher uncertainty about economic conditions, especially fears of a looming recession. So, we find that disagreement has a significant impact, even implicit disagreement, on the policy rate decisions of the MPC members. This offers insights into why central banks might hesitate to raise rates immediately in face of higher inflation or output gap pressures.