Macroeconomic factors in the term structure of interest rates when agents learn

Finance models of the term structure of interest rates have for a long time relied on unobserved factors as explanatory variables. In a seminal paper, Ang and Piazzesi (2003) have examined the potential role of macroeconomic variables in explaining the term structure. They, and subsequent studies, have found that macroeconomic variables such as inflation and real activity have explanatory power for bond yields, but have also found that latent variables still explain a great deal of the variation in bond yields. A limitation of these studies is that they assume that relationships between macroeconomic variables and interest rates are constant over time. Recent research has highlighted the presence of structural breaks among key U.S. macroeconomic variables over the past four decades. Time-invariant macro models may therefore provide poor representations of market participants' expectations of future interest rates. Given the crucial role of expectations for the term structure, this paper focuses on modeling expectations. We use a flexible framework to allow for time variation in agents' forecasting models and pay close attention to the data they were most likely to observe and forecast. We show that the explanatory power of macro factors is dramatically improved in this way. Time variation is shown to be an important feature of the estimates of both intercepts and slope coefficients. These results emphasize the importance of the expectations channel, especially of perceptions about monetary policy, in the transmission of monetary policy