This dissertaion is comprised of two essays on econometric evaluation of models of commodity futures prices. The first essay develops a frequency- domain volatility bound approach that can be used to evaluate possibly mis-specified models. The proposed method allows us to detect model failures at specific frequencies, for example, the seasonal frequencies, the business cycle frequencies, etc. This is particularly useful when the data exhibit significant cyclical behavior. As an application of the proposed method, the consumption based capital asset pricing models for commodity futures are evaluated using crude oil and corn futures price data. The equilibrium price conditions are derived. Empirical results overwhelmingly reject the consumption based capital asset pricing models across all frequencies. The second essay proposes an econometric method that can be used to determine the appropriate number of factors in affine term structure models. The proposed method estimates the affine model by solving a nonlinear least squares problem, and the number of factors is determined by minimization of a proposed model selection criteria. Simulation study shows that the proposed method can identify the right number of factors under general conditions. Finally, the empirical issue of how many number of factors are needed for crude oil and corn futures prices is examined using the proposed method.
