This article develops and estimates a dynamic arbitrage-free model of the current forward curve as the sum of (i) an unconditional component, (ii) a maturity-specific component and (iii) a date-specific component. The model combines features of the Preferred Habitat model, the Expectations Hypothesis (ET) and affine yield curve models; it permits a class of low-parameter, multiple state variable dynamic models for the forward curve. We show how to construct alternative parametric examples of the three components from a sum of exponential functions, verify that the resulting forward curves satisfy the Heath-Jarrow-Morton (HJM) conditions, and derive the risk-neutral dynamics for the purpose of pricing interest rate derivatives. We select a model from alternative affine examples that are fitted to the Fama-Bliss Treasury data over an initial training period and use it to generate out-of-sample forecasts for forward rates and yields. For forecast horizons of 6 months or longer, the forecasts of this model significantly outperform those from common benchmark models.