20231003-IMF-Mining_the_Gap_Extracting_Firms_Inflation_Expectations_From_Earnings_Calls_46页_5mb

Mining the Gap: Extracting Firms' Inflation Expectations From Earnings Calls

This paper introduces a novel approach to extract firms' inflation expectations using natural language processing (NLP) on earnings call transcripts. The methodology, called the Earnings Calls-based Firm’s Inflation Expectations (ECFIE) index, is built using advanced NLP algorithms to analyze the intensity of discussions about inflation in these transcripts.

The index is validated through cross-country comparisons with survey-based measures, such as the Livingston Survey, and demonstrates strong correlations. For instance, in the U.S., it shows a correlation of 0.84 with inflation expectations and 0.8 with consumer price inflation. The study also constructs an attention index to the central bank, revealing that firms paying more attention enhance the effectiveness of monetary policy.

Key stylized facts include violations of rational expectations, with firms extrapolating their own costs and sentiment influencing expectations. Attention to the central bank amplifies monetary policy effects by approximately 10%, but this diminishes during periods of high disagreement, suggesting that communication strategies could improve policy outcomes.

The findings underscore the importance of firms' expectations in inflation dynamics and monetary policy transmission, providing a tool for policymakers to better understand and influence these factors.