Template-Type: ReDIF-Paper 1.0
Author-Name: Douglas Kiarelly Godoy de Araujo
Author-X-Name-First: Douglas Kiarelly
Author-X-Name-Last: Godoy de Araujo
Author-Name: Nikola Bokan
Author-X-Name-First: Nikola
Author-X-Name-Last: Bokan
Author-Name: Fabio Alberto Comazzi
Author-X-Name-First: Fabio Alberto
Author-X-Name-Last: Comazzi
Author-Name: Michele Lenza
Author-X-Name-First: Michele
Author-X-Name-Last: Lenza
Title: Word2Prices: embedding central bank communications for inflation prediction
Abstract: Word embeddings are vectors of real numbers associated with words, designed to capture semantic and syntactic similarity between the words in a corpus of text. We estimate the word embeddings of the European Central Bank's introductory statements at monetary policy press conferences by using a simple natural language processing model (Word2Vec), only based on the information and model parameters available as of each press conference. We show that a measure based on such embeddings contributes to improve core inflation forecasts multiple quarters ahead. Other common textual analysis techniques, such as dictionary-based metrics or sentiment metrics do not obtain the same results. The information contained in the embeddings remains valuable for out-of-sample forecasting even after controlling for the central bank inflation forecasts, which are an important input for the introductory statements.
Creation-Date: 2025-03
File-URL: https://www.bis.org/publ/work1253.pdf
File-Format: Application/pdf
File-Function: Full PDF document
File-URL: https://www.bis.org/publ/work1253.htm
File-Format: text/html
Number: 1253
Keywords: embeddings, inflation, forecasting, central bank texts
Classification-JEL: E31, E37, E58
Handle: RePEc:bis:biswps:1253