Jinzhao Wang
Head of Client Solutions
The annual congregation for securitized products is much more optimistic, innovative, and synergetic than it was fifteen years ago as portrayed in the Big Short. Infima’s team including our Founder Kay Giesecke, CEO Hendrik Bartel, Head of Sales Derk Osenberg, and Head of Client Solutions Jinzhao Wang engaged in various panels, meetings, and conversations with players in the securitized products industry and distilled the following key takeaways from SFVegas 2023.
1. Preference for Mortgages in a Stormier Economy
Market participants in agency mortgage-backed securities (MBS) at the conference believe that the demand for housing will remain high, especially given the current strong employment. That translates into stable home prices in the near term, during which regional differences will prevail. Furthermore, market participants agree that MBS has historically received preferential treatment against corporates due to the lower volatility of MBS performance. For that reason, MBS also tends to have a better Sharpe ratio than corporates. Especially when the economy falters, the historically high yields and spreads on agency MBS make it an attractive option with nearly no credit risk compared with other asset classes in fixed income. Right now, banks are the main buyer in MBS, and they have preferred Ginnie Mae collateral to Treasuries for the higher yield and to Conventional (GSE) collateral for better capital treatment.
2. Dialogue between Government Agencies and Private Sector
Dynamic dialogue took place among government agencies, legal firms, and market participants at each panel of the conference. Government agencies want to ensure that their programs keep running to fulfill their mission and want to hear from the private sector about their investment needs and innovative solutions. A case in point is the question of when a functioning specified pool market actually erodes the liquidity of the TBA market. Since specified pools tend to trade well in an environment with higher rates (as higher rates lower dollar rolls), more buying in specified pools tends to decrease the trading volume of TBAs. And because TBAs are the primary instrument in which government agencies such as the GSEs and Ginnie Mae provide liquidity to mortgage borrowers, it is in their interest to balance between a healthy TBA market and a functioning Specified Pool market.
3. Relative Value within MBS
MBS basis could be rich at around 120 bps, cheap at around 150 bps, and will likely remain range-bound in the near term. For the discount coupons which compose nearly 95% of the market, three factors may most contribute to prepayment speeds: turnover, cash-out, and curtailment. On top of the coupon stack, new production pools last year have steeper S-curves due to both technological advancement and the declining capacity of originators. Some recent homeowners took out mortgages already with the thought of refinancing, given several programs from originators to waive refinancing fees. Tier 2 servicers could provide a convexity advantage for these newly originated pools. Additionally, high LTV stories in these higher coupons could provide further call protection, especially given that home price appreciation has slowed and the pay-ups on these pools are still trading cheap.
4. True Opportunities for AI in Fixed Income
Panelists at the conference see tremendous opportunity for cutting-edge AI technology to support core fixed-income investment workflows. In MBS, for example, the name of the game is understanding and forecasting future borrower behavior such as prepayment, which drives MBS returns. AI, and specifically Deep Learning, is significantly better at this than legacy approaches because Deep Learning can harness all available loan-level data (billions of data points) and uncover hidden patterns in individual borrower behavior that legacy models miss. Currently, portfolio managers make security selection decisions based on decades-old legacy analytics systems whose output has been performing poorly. The pandemic and 2022 turned out to be especially challenging. Infima has shown that AI can be a game-changer for these portfolio managers, delivering a 10x boost in prepayment forecast accuracy and reliable risk/return analytics and enabling them to make more confident high-stakes security selection, portfolio construction, and risk management decisions across market regimes.
5. Explainable AI Helps Adoption
Panels also see Explainable AI as both a challenge and an opportunity to expand market understanding with model insights. State-of-the-art artificial intelligence ranges from a complete black box to a rule-based model with utter clarity. There has been a fair amount of research over the recent years that contributes to the explainability of AI models, making them as transparent as legacy models. It would now be easier to explain the repeatability and interpretability of model results to regulators. That is essential to ensure fairness, trust, and reliability around the model output. Furthermore, it would help market participants validate and understand where the model’s recommendations come from. Infima, for instance, has solutions that supply narratives around security selection. For example, it has developed tools that highlight the drivers of speeds and provide prepay narratives around specific bonds or cohorts. In this way, rich AI systems yield important new insights that were not previously available. These solutions encourage market participants to leverage powerful machines to acquire a sustainable edge in the market.
Nothing can compete with in-person interactions that facilitate the exchange of ideas, especially across sectors, including market participants, legislators, and innovators. Infima’s attendance at the conference is a tailwind to not only collaboration with other players in structured finance but also the advancement of interdisciplinary knowledge combining AI with fixed income.
