This piece originally appeared in the December 2024 edition of MortgagePoint magazine, online now.
If you were a research oncologist looking to extract insight from massive amounts of patient data, or an engineer evaluating climatological patterns to identify renewable energy sites, or an e-commerce strategist looking for new ways to use customer data to enhance online shopping experiences, you wouldn’t think twice about using artificial intelligence (AI) to advance your work. For many industries, AI is already firmly entrenched in the technology toolbox. But while the common promises of faster, more accurate and transparent data processing are highly applicable to the fixed-income world, adoption continues to be cautious.
Let’s be clear. AI is not simply the next, natural iteration of machine learning (ML). It’s multidimensional in its abilities, combining human-like logic with something humans will never be able to do—crunch and extrapolate information from massive lakes of both structured and unstructured data from hundreds of different sources—at scale, and then, as a logic-driven, decision-making tool, distill all that into recommendations, reports, and business-critical insight.
Mapping AI to Asset Management Objectives
At the ground level, fixed-income asset managers are focused on four core functions for optimizing returns:
- Liquidity management
- Portfolio construction and diversification
- Yield enhancement and risk management functions
- Performance and investor reporting
Myriad data-driven decisions and actions steer the strategies for these functions, as well as their successful execution. AI, with its power to drive out data management inefficiencies while driving in data veracity and transparency, enables asset managers to perform these functions not only faster, but with more confidence in their conclusions and recommendations. In doing so, AI is effectively bringing the power of logic-based automation to the trifecta of fixed-income management: Best Asset, Best Management, and Best Execution.
Creating the Best Asset
Improving asset quality and tradability improves asset value. In the mortgage world, a significant portion of an asset’s value depends on its history. The more accurate and complete the history, the less risk associated with the asset and, therefore, the more it is worth. This sounds straightforward enough until you factor in all the different partners and players that “touch” an asset’s data throughout its lifecycle, including multiple asset owners, servicers, law firms, and custodians. Every party has its own data collection and management process. Different departments within a servicer, (loss mitigation, for example) may collect different borrower data points, and typically have different methods for gathering, reviewing, and cataloging key back-up documentation, such as requisite legal and title reports.
Residing across such disparate sources, none of the data, nor corresponding documentation, is easily accessible, much less validated, vetted for errors or reconciled. And the longer the seasoning of the asset, the more potential for inconsistent, incomplete, or inaccurate data. In other words, the less it’s worth.
Questionable loan-level data severely compromises the validity and applicability of analytics and insights the data informs. The ensuing trust deficit has a domino effect, impacting asset management strategy and execution, continuing to drag on asset value and ROI.
How AI Helps Achieve Best Management
Lack of centralized, validated loan data casts a long shadow, impacting asset management effectiveness and decisioning at both the operational and strategic levels and eventually dragging on investor yield.
The weightiest asset management challenges lie primarily in secondary market operations where verification, manual validation, and extensive documentation requirements cause friction, delays, incremental costs, and potential disputes.
Compounding these challenges are third party diligence providers tasked with unearthing risk associated with ownership transfers and other asset conditions that may impede asset management efficiency and subsequent execution of the disposition strategy.
Meanwhile, AI has the power to make light work of even the most complex, labor-intensive diligence tasks; using sophisticated querying to help determine critical decision loan attributes such as:
- Lawful and rightful ownership and the seniority/subordination of the asset
- The right to foreclose or take ownership of the underlying collateral in the event of borrower default
- Compliance at the origination of the asset
- Value and condition of the underlying collateral
- Borrower payment history and likelihood of paying a loan to maturity, paying off early, or default [the borrower credit profile]
- Determination that proper servicing practices are in place, including loss mitigation and borrower contact
In essence, AI can be leveraged to identify, evaluate, validate, reconcile, tabulate, rank, verify, search, append, compose, analyze … just about any data-enabled logic task associated with managing an asset or portfolio of assets through their lifecycles.
These agents also enhance curative functions, not only pinpointing clouded assets and documentation issues but automating instructions for remediation and improving asset value. Even more impressive is AI’s ability to perform complex asset management tasks and optimization across more esoteric asset classes such as reverse mortgages, HELOCs, and EBOs.
Using AI to de-risk and improve value, enhances liquidity and tradability and, in the case of the more complex asset classes, provides transparency and confidence to help increase participation on both the buy and sell side of the transaction equation.
AI’s Power to Reshape Best Execution
The true transformative power of AI for fixed-income lies in its potential to create a strong-form efficient market where every asset is a known quantity and portfolio pricing and trading are driven more by the future potential of an asset than its risky past.
When coupled with blockchain for ensuring full transparency and data immutability, AI can automate a more intricate pricing structure that takes more data points into account. In addition to due diligence findings, such as borrower credit and payment history, predictive algorithms can factor into valuations, probability algorithms on early repayment, possible default, legal timelines and expenses, and ongoing servicing costs. Because the data feeding these predictions is accurate and verifiable, the resulting valuation, pricing, financing, and hedging are better informed.
Layering in additional data points regarding property condition and home improvements history also bolsters valuation and pricing while helping to minimize investor risk by ensuring the asset value covers the mortgage balance.
If you’re thinking “this is already being done,” the answer is yes, but not at scale … and not with the level of veracity necessary to protect asset valuation and pricing and optimize returns.
Additionally, the capacity of AI to automate the generation of investor and servicing reports brings an unprecedented level of speed and efficiency to the marketing and trading of portfolios creating greater options for liquidity.
Inputting custom queries that “ask” AI specific counterparty and investor questions provides more reliable, data-informed answers in a fraction of the time it would take a team of humans.
How AI Goes Beyond Speed and Cost-Efficiency
If you believe in the potential of a strong form efficient market, you believe in the potential of AI. We simply will not get there without its (and blockchain’s) ability to enable the free-flowing exchange of fixed-income assets whereby investors have trust in the assets and the integrity of the processes and systems for buying and selling them.
Furthermore, AI and blockchain technologies are fueling innovation and ideation in an industry that is certainly not known for either. Fractional ownership of assets through tokenization can open up market participation exponentially, eventually leading to direct access by retail investors.
What does not take much imagination is this: we are only at the very beginning of understanding how AI will eventually reshape the fixed-income industry and market.
