Back
2025ECO007.pdf
Doctoral thesis

Three essays in asset pricing

  • 2025

PhD: Università della Svizzera italiana

Asset pricing
Return predictability
Market return predictors
Online learning
Statistical learning methods
Variance swaps
Sharpe ratio
Stochastic discount factor
Consumption–wealth return
Long-run risk models
Hansen–Jagannathan bound
English This dissertation develops optimization-based approaches to core challenges in financial economics. The first chapter, From surveys and theory to statistical methods: Which predictors can we trust? And when?, examines the predictive performance of more than twenty market return predictors, including survey-based measures, economic indicators, and predictors constructed using modern statistical learning techniques. Using an online learning framework that adapts continuously as new data arrive, the study evaluates predictors over 2001–2018 and shows that predictive power is highly context dependent. While no predictor proved reliable during the Global Financial Crisis, certain survey-based measures and statistical methods displayed informativeness in more stable periods. The second chapter, Optimal Variance Swaps (with Emanuele Luzzi, Paul Schneider, and Rohan Sen), develops a novel nonparametric recovery of the Sharpe-ratio–optimal convex contract used to replicate variance swaps with static option positions. The method is supported by both asymptotic and finite-sample guarantees. Monte Carlo experiments demonstrate that it recovers the theoretical stochastic discount factor, and empirical evidence from SPX option data reveals that the proposed optimal variance swap systematically outperforms the standard VIX-implied contract, providing sharper insights into market-priced uncertainty. The third chapter, What is the consumption–wealth return? (with Paul Schneider and Paul Whelan), addresses the definition of this key state variable in long-run risk models. Whereas existing approaches often rely on tractable log-linearizations or polynomial approximations, the chapter introduces a nonparametric minimum-variance specification obtained by minimizing the Hansen–Jagannathan bound. The results contrast sharply with those produced by conventional methods, overturning several qualitative predictions previously attributed to long-run risk models and underscoring the necessity of grounding such specifications in economic principles.
Collections
Language
  • English
Classification
Economics
License
License undefined
Open access status
green
Identifiers
Persistent URL
https://n2t.net/ark:/12658/srd1333212
Statistics
Document views: 4 File downloads:
  • 2025ECO007.pdf: 1