TL;DR
This paper investigates how the unique dual-market trading structure of agency mortgage-backed securities (MBS) – Specified Pool (SP) and To-Be-Announced (TBA) – influences their pricing and trading. The authors develop a model and find that MBS heterogeneity significantly impacts pricing through "venue selection" and "venue backup" channels, affecting yields and trading volumes in ways distinct from prepayment risk.
What Problem Does This Paper Address?
The agency MBS market is one of the largest fixed-income markets in the U.S., playing a crucial role in monetary policy implementation. Its unique structure involves two parallel trading mechanisms: Specified Pool (SP) trading, where individual MBSs are traded with specific contracts, and To-Be-Announced (TBA) trading, where standardized contracts allow for delivery flexibility. While the liquidity of this market is well-recognized, the specific impact of this dual-trading environment on asset pricing has received less attention. Existing research often focuses on prepayment risks. This paper aims to fill this gap by exploring how the parallel trading structure, particularly MBS heterogeneity within cohorts, affects MBS pricing and trading beyond prepayment considerations.
Methods
The authors develop a stylized model that abstracts from prepayment risk to focus on the economic effects of the parallel TBA/SP trading environment. The model considers MBSs traded in two rounds, with liquidity shocks in the second round. It highlights a "venue selection channel" where sellers choose between TBA and SP markets based on expected CTD discounts and transaction costs, and a "venue backup channel" where buyers consider the TBA market as a future resale option for SP MBSs. Empirically, they measure MBS heterogeneity using the dispersion in prepayment characteristics (specifically, weighted-average original FICO scores, WAOCS) within coupon cohorts. They then test hypotheses derived from the model using panel regressions on MBS yields (Option-Adjusted Spreads - OAS) and trading activity data.
Datasets
The study utilizes several datasets:
- Fannie Mae 30-year MBS data: This forms the basis for computing MBS heterogeneity measures, covering coupon cohorts from 2.5% to 7% from June 2003 to December 2018.
- Freddie Mac Primary Mortgage Market Survey (PMMS): Used to construct monthly series of mortgage rates.
- TRACE dataset from FINRA: Provides MBS transaction data (trades, volume, etc.) from May 2011 onwards, used to measure trading activities.
- Wall Street MBS dealer data: Used to obtain OAS series for SP MBSs and TBA contracts, as well as hedged returns and prepayment forecasts.
- IO/PO strip data: Used to differentiate MBS heterogeneity effects from prepayment risk.
Key Insights and Ideas
- Dual Markets Create Unique Pricing Channels: The existence of both TBA and SP markets creates distinct "venue selection" and "venue backup" channels that influence MBS pricing. This is a novel contribution beyond traditional prepayment risk models.
- Heterogeneity Matters Beyond Prepayment: MBS heterogeneity, measured by the dispersion in loan characteristics within a cohort, has a significant impact on pricing. This effect is distinct from and additive to prepayment risk.
- TBA Market's "Lemon's Problem": The TBA market's structure, where sellers can deliver the cheapest eligible MBS (CTD), creates a discount that is absent in the individually-priced SP market. This CTD discount is a key driver of the observed pricing effects.
- Liquidity Shocks Amplify Heterogeneity Effects: The impact of MBS heterogeneity on SP yields is stronger when investors anticipate heavier selling pressure or liquidity shocks. This highlights the importance of market liquidity in asset pricing.
- Trading Activity Responds to Heterogeneity: Increased MBS heterogeneity leads to dampened trading activity in both TBA and SP markets, but also shifts a larger proportion of trades towards the SP market.
Key Results
- MBS Heterogeneity Increases SP Yields: A greater MBS heterogeneity leads to higher yields on SP MBSs, particularly for those more likely to be delivered into TBA contracts. This effect is weaker for more valuable MBSs.
- Selling Pressure Amplifies Effects: When investors expect heavier selling pressure, the dependence of SP yields on MBS heterogeneity becomes stronger.
- Trading Volume Decreases, SP Share Increases: Higher MBS heterogeneity is associated with a decrease in both TBA and SP trading volume, but a significant increase in the ratio of SP to TBA trading volume.
- Distinguishing from Prepayment Risk: The study provides evidence that the effects of MBS heterogeneity are distinct from prepayment risk, even when considering premium/discount markets and IO/PO strips.
Limitations and Future Work
- Model Simplifications: The model abstracts from prepayment risk and assumes exogenous transaction costs for simplicity. While the authors argue their main results would hold even with endogenous liquidity, this is a simplification.
- Data Limitations: The study relies on specific data sources and time periods. The availability of certain data (e.g., OAS from alternative dealers) was limited.
- Future Work: The authors suggest that endogenizing TBA market liquidity could offer further insights into how MBS heterogeneity affects TBA liquidity itself. They also highlight the potential for applying TBA-like trading mechanisms to other OTC fixed-income markets.
Summary updated