Forthcoming in Journal of Econometrics:  Central Bank Mandates and Monetary Policy Stances  

“Central Bank Mandates and Monetary Policy Stances: through the Lens of Federal Reserve Speeches” with Isaiah Hull, Robin Lumsdaine, and Xin Zhang, Journal of Econometrics, forthcoming.

  • Abstract: The Federal Reserve System has an institutional mandate to pursue price stability and maximum sustainable employment; however, it remains unclear whether it can also pursue secondary objectives. The academic literature has largely argued that it should not. We characterize the Fed’s interpretation of its mandate using state-of-the-art methods from natural language processing, including a collection of large language models (LLMs) that we modify for enhanced performance on central bank texts. We apply these methods and models to a comprehensive corpus of Fed speeches delivered between 1960 and 2022. We find that the Fed perceives financial stability to be the most important policy concern that is not directly enumerated in its mandate, especially in times when the debt-to-GDP ratio is high, but does not generally treat it as a separate policy objective. In its policy discourse, it has frequently discussed the use of monetary policy to achieve financial stability, which we demonstrate generates movements in asset prices, even after rigorously controlling for macroeconomic and financial variables. (C55, E42, E5, E61, G28)
A word cloud of terms and groups of terms that were identified by extractive question answering as the speaker’s main concern in Federal Reserve speech paragraphs with low dual mandate content scores over the period 1984-2022.

Keywords: Large Language Models, Machine Learning, Central Bank Communication, Financial Stability. An earlier version is available as Riksbank Working Paper No. 417

Riksbank WP No. 432 on International Central Bank Communication

“Four Facts about International Central Bank Communication”

  • Abstract: This paper introduces a novel database of text features extracted from the speeches of 53 central banks from 1996 to 2023 using state-of-the-art NLP methods. We establish four facts: (1) central banks with floating and pegged exchange rates communicate differently, and these differences are particularly pronounced in discussions about exchange rates and the dollar, (2) communication spillovers from the Federal Reserve are prominent in exchange rate and dollar-related topics for dollar peggers and in hawkish sentiment for others, (3) central banks engage in FX intervention guidance, and (4) more transparent institutions are less responsive to political pressure in their communication. (C55, E42, E5, F31, F42)
The figure visualizes the output of the t-stochastic nearest neighbors (t-SNE) algorithm applied to the exchange rate, U.S. dollar and international trade text features (see Section 2.1) for 21 central banks. We construct three categories for the visualization: “base” currencies in blue (U.S. dollar/Federal Reserve System, Euro/ECB), floating currencies in orange, and pegged currencies in green, which includes the euro NCBs.
  • Keywords: Exchange Rates, Natural Language Processing (NLP), International Spillovers, Monetary Policy.