Schwartz mean-reverting commodity models
Models spot and futures dynamics with mean reversion and alternative sets of risk factors.
How it works
Schwartz compares commodity models with different combinations of factors, including spot, convenience yield and interest rates. Mean reversion can make temporary supply disturbances fade rather than move all futures maturities equally. The forward curve and its dynamics carry information about which factors a model needs.
Worked example
In a one-factor log-price illustration, the state lies 0.20 above its long-run mean and reverts at speed 1.2 per year. Its expected deviation after a year is 0.20 exp(−1.2), about 0.0602. The expectation refers to log price. Converting it into expected spot requires accounting for variance.
Validation and limitations
Estimate using multiple maturities and separate seasonality, rolling contracts and delivery conventions. Stress the persistence of a price spike and inspect long-maturity futures errors. Compare with the two-factor Schwartz–Smith model.