The term structure of interest rates is modeled as a random field with conditional volatility. Random field models allow consistency with the current shape of the term structure without the need for recalibration. However, most such models are Gaussian, with no conditional volatility. State-dependent volatility is introduced while a key property of Gaussian random field models is retained. Each forward rate is part of a low-dimensional diffusion process, simplifying estimation and derivatives pricing. The modeling approach also implies that, in general, the set of zero coupon bonds does not complete the market, and term structure derivatives cannot always be priced by arbitrage.

We develop and implement a technique for closed-form maximum likelihood estimation (MLE) of multifactor affine yield models. We derive closed-form approximations to likelihoods for nine Dai and Singleton (2000) affine models. Simulations show our technique very accurately approximates true (but infeasible) MLE. Using US Treasury data, we estimate nine affine yield models with different market price of risk specifications. MLE allows non-nested model comparison using likelihood ratio tests; the preferred model depends on the market price of risk. Estimation with simulated and real data suggests our technique is much closer to true MLE than Euler and quasi-maximum likelihood (QML) methods.

We extend the standard specification of the market price of risk for affine yield models, and apply it to U.S. Treasury data. Our specification often provides better fit, sometimes with very high statistical significance. The improved fit comes from the time-series rather than cross-sectional features of the yield curve. We derive conditions under which our specification does not admit arbitrage opportunities. The extension has extremely strong statistical significance for affine yield models with multiple square-root type variables. Although we focus on affine yield models, our specification can be used with other asset pricing models as well.