Equity-Imposed Tilts in Affine Term Structure Models: Evidence from Option-Implied Asymmetries

Affine term structure models, such as the arbitrage-free Nelson–Siegel (AFNS) framework, assume symmetric Gaussian dynamics under the risk-neutral measure, masking the directional information embedded in option markets. This article develops a tilted AFNS specification that incorporates option-implied asymmetries (i.e., an equity-imposed tilt) into the factor dynamics of the yield curve. Using S&P 500 Mini (XSP) option data from 2019 to 2020, we recover risk-neutral probabilities that deviate systematically from the 50/50 benchmark, with pronounced shifts during the Federal Reserve’s dovish stance in late 2019 and the onset of the COVID-19 crisis. Embedding these asymmetries into the AFNS structure enhances yield-curve fit, reduces out-of-sample forecast errors by roughly 40% to 50%, and produces term premiums that respond more realistically to market stress. Cross-sectional Fama–MacBeth regressions show that the tilted model generates significantly priced factors, while symmetric benchmarks contribute little once asymmetry is accounted for. These results indicate that option-implied asymmetries constitute a priced state variable for fixed-income markets and that incorporating them within affine models improves both statistical performance and economic interpretation.