AbstractLittle has been done to estimate the magnitude of rational inattention and fear for model misspecification especially by using actual data on macro-finance variables such as yields. Our goal in this paper is two-fold. First, we develop a class of affine term structure models that allow for the role of bounded rationality by incorporating either information-processing constraint or fear for mis-specification into an otherwise canonical representation of the three-factor arbitrage-free affine model. For each of these models, the presence of bounded rationality creates a new additional factor that is not spanned by conventional factors such as level, slope, and curvature factors. Second, we explore the possibility that macro-finance affine models with such a bounded rationality factor help estimate the magnitude of bounded rationality. Our empirical results indicate that substantial amounts of information capacity constraint and robustness preference for model misspecification are needed to explain the observed behavior of yields when we use affine term-structure models. In addition, out-of-sample forecasts of our models indicate that the incorporation of bounded rationality can enhance an affine term-structure model's capability to forecast yields. We also assess the empirical performance of our approach in terms of the role of bounded rationality factors to help explain excess returns of bond holdings. Specifically, the rational inattention constraint for current states generates a shock to the information set of agents that does not move the current time-t yield curve but affect the equilibrium dynamics of yields and thus expected excess returns of holding bonds.