Notes

Market Participants vs. Dealer Belief

(Beliefs Identification) I jointly observe primary dealers’ (PD) and market participants’ (MP) belief operators under the same information set from 2014 to 2024.
(Which States Matter) Belief heterogeneity arises because market participants underweight long-horizon downside and stress states, while dealers price these states—leading to first-order effects in CDS premia and credit spreads.
(Generative AI) I use LLMs to apply an identical tail-state classification to PD and MP survey texts collected separately, enabling apples-to-apples comparison across agents.

(Key Findings) Under a common information set, "dealers" internalize downside tail risk while "market participants" underweight it. I show that this belief gap is embedded in credit spreads.

Example of NY Fed Survey Questions (SPD & SMP)

Word Cloud of Responses: Dealers

Google Sites

Report abuse