国际清算银行-Word2Prices：嵌入央行通信以预测通胀（英）

BIS Working PapersNo 1253Word2Prices: embedding central bank communications for inflation prediction by Douglas K G Araujo, Nikola Bokan, Fabio Alberto Comazzi and Michele Lenza Monetary and Economic Department March 2025 JEL classification: E31, E37, E58. Keywords: embeddings, inflation, forecasting, central bank texts. BIS Working Papers are written by members of the Monetary and Economic Department of the Bank for International Settlements, and from time to time by other economists, and are published by the Bank. The papers are on subjects of topical interest and are technical in character. The views expressed in them are those of their authors and not necessarily the views of the BIS. This publication is available on the BIS website (www.bis.org). © Bank for International Settlements 2025. All rights reserved. Brief excerpts may be reproduced or translated provided the source is stated. ISSN 1020-0959 (print) ISSN 1682-7678 (online) Word2Prices: Embedding central bank communications forinflation prediction∗Douglas Araujo (BIS), Nikola Bokan (ECB)Fabio Alberto Comazzi (ESM), Michele Lenza (ECB and CEPR)AbstractWord embeddings are vectors of real numbers associated with words, designed to capturesemantic and syntactic similarity between the words in a corpus of text. We estimate theword embeddings of the European Central Bank’s introductory statements at monetary policypress conferences by using a simple natural language processing model (Word2Vec), only basedon the information and model parameters available as of each press conference.We showthat a measure based on such embeddings contributes to improve core inflation forecastsmultiple quarters ahead. Other common textual analysis techniques, such as dictionary-basedmetrics or sentiment metrics do not obtain the same results. The information contained inthe embeddings remains valuable for out-of-sample forecasting even after controlling for thecentral bank inflation forecasts, which are an important input for the introductory statements.JEL classification: E31, E37, E58Keywords: Embeddings, inflation, forecasting, central bank texts∗The authors thank Ben Cohen, Michael Ehrmann, Marek Jarocinski, Sujit Kapadia, Hanno Kase, Joan Paredes,Fernando P´erez-Cruz and several seminar participants for comments. Johannes Damp provided excellent researchassistance in the first phase of this project. The opinions in this paper are those of the authors and do not necessarilyreflect the views of the European Central Bank, the Eurosystem, the CEPR, the BIS or the ESM. Contact author:Michele Lenza (michele.lenza@ecb.europa.eu).11IntroductionThe wide and ever increasing availability of text in digital format, coupled with the advances inartificial intelligence to decode it, opens new ground for the use of text as data (Gentzkow et al.,2019; Dell, 2024; Ash and Hansen, 2023). Perhaps the most popular methodology to draw insightfor the purpose of macroeconomic analysis, including nowcasting and forecasting (see,