Browse by subject:
Applied and bridge studies study record
Methods & measurementA Disciplined Cross-Section: Bayesian Model Averaging over the Factor Zoo Under the Joint-Hypothesis Constraint
STATUS · Manuscript in progressSSRN · Not yet posted
MECHANISM
Identify the incentive structure and the condition that would falsify the claim.
RULE CHANGE
Read the intervention only after the paper shows how the current payoff space fails to support system welfare.
READER USE
Use the summary to see where private gain creates system exposure, then check the study record.
Contribution — what this adds to the conversation
Resolves a foundational inference problem in empirical asset pricing; nests all major competitors (FF3, FF5, HLZ, KMZ) as special cases; provides a unified Bayesian framework that bridges Bayesian and frequentist traditions.
WHAT'S NEW · First Bayesian model-averaging framework for the full factor zoo (K≥200) under the joint-hypothesis constraint; novel likelihood decomposition separating pricing-model failure from market-efficiency failure; derived posterior-odds stopping rule.
The paper formalizes Fama's joint-hypothesis constraint as a Bayesian likelihood decomposition and delivers a posterior-odds stopping rule for factor inclusion in asset pricing. Applied to 207 factors from the Chen-Zimmermann library, the framework identifies 7-9 robust factors, compressing the factor zoo by 96% while preserving economic interpretability. Out-of-sample R² of 0.74 is competitive with dense machine-learning approaches. Welfare implications are quantified at $500 billion annually through the Missing System Theory.