Abstract:
This paper evaluates the influence of central bank's projections and narrative signals provided in the summaries of its Inflation Report on the expectations of professional forecasters for inflation and GDP growth in the case of Mexico. We use the Latent Dirichlet Allocation model, a textmining technique, to identify narrative signals. We show that both quantitative and qualitative information have an influence on inflation and GDP growth expectations. We also find that narrative signals related to monetary policy, observed inflation, aggregate demand, and inflation and employment projections stand out as the most relevant in accounting for changes in analysts' expectations. If the period of the COVID-19 pandemic is excluded, we still find that forecasters consider both types of information for their inflation expectations.