Expected-Shortfall Envelopes and Hansen-Jagannathan Kernels
Uses regularized conditional quantiles to compare downside-risk envelopes and admissible Hansen-Jagannathan kernels across asset universes.
This project develops a regularized conditional-quantile workflow that maps portfolio tail information into tail-state measures and, when feasible, admissible empirical stochastic discount factors. The core distinction is between an expected-shortfall tail-probability envelope and a no-arbitrage pricing kernel that must satisfy normalization, pricing, and nonnegativity restrictions.
Together with Joachim Grammig and Johannes Schief, I estimate a shared portfolio and conditional-quantile surface with elastic-net penalties. Residual ranks define an expected-shortfall envelope; only after normalization and projection can this object be interpreted as a nonnegative stochastic discount factor. The benchmark is a state-dependent Hansen-Jagannathan minimum-second-moment kernel estimated under the same data conventions.
The empirical application uses monthly Fama-French portfolios and U.S. macro-financial states. The current evidence points to a clearer volatility relation than labor-market relation, and treats failures of admissibility as diagnostics rather than hidden estimator details.
Current manuscript: non-peer-reviewed working paper.