Does Real‐Time Macroeconomic Information Help to Predict Interest Rates?

Abstract

We analyze the predictive ability of real‐time macroeconomic information for the yield curve of interest rates. We specify a mixed‐frequency macro‐yields model in real time that incorporates interest rate surveys and treats macroeconomic factors as unobservable components. Results indicate that real‐time macroeconomic information is helpful to predict interest rates, and that data revisions drive a superior predictive ability of revised macro data over real‐time macro data. We also find that interest rate surveys can have significant predictive power over and above real‐time macroeconomic variables.

Venturing into Risk Management: Examining Credit Risk and Performance in Pakistani Commercial Banks through Panel VAR Analysis from a Business Perspective

“…For instance, high inflation reduces the value of loans, causing the risk of default to increase; GDP growth generally improves credit quality and reduces default rates. Fluctuations in interest rates impact borrowing costs and the likelihood of default (Caruso & Coroneo, 2023).…”

Exchange rates and macroeconomic fundamentals: Evidence of instabilities from time‐varying factor loadings

“…Thereby, rather than simply using the first release of each observation, we take into account that investors have knowledge of revisions to published data prior to the current month, which could potentially impact their decisions. Caruso and Coroneo (2022) show that this improves forecasts of financial variables relative to just relying on first releases.…”

“…1 Therefore, rather than just relying on first releases, we include revisions made to past data with each new release. That is, we accurately replicate the information set available at each time step, which has been shown to improve forecasts of financial variables (Caruso & Coroneo, 2022). The information inherent in these series is extracted via principal components that serve as factor estimates.…”

Slow Expectation-Maximization Convergence in Low-Noise Dynamic Factor Models

“…Yet, it has been shown that the EM algorithm becomes inefficient in such a low-noise environment (Bermond and Cardoso, 1999;Petersen et al, 2005), causing extremely slow convergence, especially for the factor loading estimates. Unfortunately, this issue seems to have been overlooked by Bańbura and Modugno (2014) and subsequent applications of their approach (see, among others, Coroneo et al, 2016;Scotti, 2016;Alvarez et al, 2016;Bok et al, 2018;Barigozzi and Luciani, 2019;Cascaldi-Garcia et al, 2021;Caruso and Coroneo, 2023). Moreover, these low-noise issues could also arise more naturally whenever the series exhibit a strong factor structure with a high signal-to-noise ratio.…”