Month: April 2024

Artificial intelligence is bringing about a radical transformation in the financial services sector and nowhere is it more apparent than in mortgage underwriting. AI’s ability to ingest and process huge volumes of multivariate information makes it far more powerful than even the most experienced and knowledgeable underwriter.

While human judgment remains key, AI has introduced an unprecedented degree of automation and augmentation. It’s no surprise, therefore, that 65% of lenders are acutely familiar with AI, and the benefits it can unlock for their business. However, one important challenge remains – driving AI adoption.

Despite industry-wide awareness of the growing need for AI transformation, only 7% have deployed it within their current mortgage processes including both underwriting and mortgage loan servicing. This means that businesses are losing out on potential gains, even as financial services become increasingly more technology-driven.

Read more: The Rise of Generative AI: Transforming the Tech Landscape

How Does Predictive Analytics in Mortgage Underwriting Work?

Predictive analytics, also known as propensity modeling, is the science of leveraging historical data to identify and measure the possibilities of future events.

To take a simple example, based on previous information and customer behavior patterns, one might reasonably predict that when someone in their 30s gets married, they tend to purchase a house within three years. Therefore, analytics can help anticipate first-time home buyer trends and allow lenders to prepare their mortgage underwriting processes accordingly.

Interestingly, predictive analytics is not a new technology and has actually been around for a long time. Modern predictive modeling dates back to the 1940s, when data was first used to forecast future events and changes in customer behavior, financial risks, and market conditions.

As hardware and software technology evolved over the years, machines became better equipped to process data without relying on manual calculations. Then, since the 1990s and the 2000s, explosive digital adoption led to high volumes of data generation from prospective home buyers – all of which would eventually feed into mortgage underwriting processes and analytics models.

Indeed, modern data generation has far surpassed what traditional analytics systems could consume, necessitating the use of artificial intelligence or AI.

Read more: Gen AI Mortgage Metamorphosis: What’s in Store for Tomorrow?

Unpacking AI-Powered Predictive Analytics in Mortgage

Artificial intelligence (AI) is a leapfrog advancement in data analysis and automation that allows machines to understand and act on data that was previously comprehensible only to human beings. For example, AI allows data engines to understand natural languages, images, and objects. This has far-reaching impacts on the mortgage sector – say, for document management.

AI has two major advantages when compared to traditional models. One, it can process raw or big data.  From screenshots and photographs to social media and file metadata, AI can convert all types of information into a machine-readable format. Two, it can process data at scale. Instead of a few spreadsheets or SQL tables, you can feed gigabytes or even terabytes of information into modern AI engines. For instance, ChatGPT was trained on approximately 570 GB of data sets.

When it comes to predictive analytics for mortgage underwriting, AI opens up infinite possibilities. It can predict the likelihood of borrower acceptance, credit risk levels, and market fluctuations with significantly more accuracy than conventional statistical tools.

Further, thanks to machine learning, AI-driven predictive analytics only gets better with time. It can learn from the decisions and outcomes resulting from previous prediction cycles, to become iteratively more accurate with repeated use. This also significantly reduces the pressure of model maintenance. Rather than having to update statistical algorithms every time there’s a new trend or event (e.g., the rise of Generation Z), AI can self-teach new correlations to maintain the accuracy of mortgage underwriting processes.

Read more: How will Individual Artificial Intelligence Impact the Mortgage Industry?

AI Predictive Analytics Use Cases in Mortgage Underwriting

By employing AI-powered predictive analytics in mortgage underwriting processes, lenders can unlock a plethora of benefits:

• Pre-approved loan offers: AI can predict, with incredible accuracy, which of your existing customers may be pre-approved for a mortgage. It can automatically reach out to potential borrowers, engage with them using chatbots, and complete loan disbursal with zero intervention from human executives.
• Lending to economically disadvantaged groups: Traditional models may draw broad conclusions about mortgage eligibility, which could exclude economically disadvantaged groups. AI predictive analytics can look at parameters other than, let’s say formal college education when making underwriting
• Risk assessments for small and new businesses: Commercial real estate (CRE) mortgages can be difficult to acquire for small or new businesses with limited credit histories. AI predictive analytics could level the playing field for such entities, also unlocking new business for mortgage lenders.
• Income verification from unconventional sources: Home buyers today may present a variety of income sources during the mortgage underwriting process, including unconventional ones like capital gains from cryptocurrency or fluctuating income from freelancing. AI predictive analytics is well-suited for data analysis in such unusual cases, while also minimizing risk.
• Selecting fixed rate vs. adjustable mortgages: Choosing between different mortgage rates can be difficult and the wrong decision can negatively impact both the lender and the home buyer. AI predictive analytics could factor in larger, more diverse datasets to make a recommendation with greater accuracy.

Read more: What Is Decision Intelligence and How Can It Impact the Mortgage Industry

Navigating the Ethical and Technical Complexities of AI

Artificial intelligence, whether in mortgage loan servicing or underwriting, signals a paradigmatic change in your operations. Not only do you require a strong foundation of data capabilities and interconnected systems, but you also need to be prepared for a culture shift that results from the reduction of reliance on human work and judgment. At Nexval, we help today’s leading lenders navigate accelerating digital transformation and come out on the winning side.

Speak with our experts to learn more.

Generative artificial intelligence or gen AI had its breakout year in 2023. While overall AI adoption holds steady at 55%, over 2 in 3 companies are planning to pilot gen AI technology. This is because of its incredible versatility, putting it at par with once the internet and more recently the cloud in terms of its business potential. Mortgage processing too is a prime candidate for gen AI metamorphosis given its data-intensive nature, providing plenty of material for gen AI models to train and learn.

How Does Gen AI in Mortgage Processing Work?

At its core, gen AI works by analyzing past information and using statistical algorithms to learn how to mimic it and create new information. These algorithms can be of various types like variational autoencoders (VAEs), generative adversarial networks (GANs), and autoregressive models. They allow the AI system to learn from content over time (typically several gigabytes of data) and get incrementally smarter.

Gen AI companies blackbox this technology into software that you can implement and interact with using simple GUI or conversational interfaces. The technical barriers to gen AI adoption are significantly lower than other forms of artificial intelligence, which can be harder to build, integrate, and maintain.

In a recent EY survey of financial service leaders, all respondents say that gen AI is part of their roadmap; only 1 in 5 said that they are “nervous or skeptical” about its impact.

In mortgage processing, AI already plays a critical role in tasks like credit risk assessments, underwriting, document management, and customer communications. Gen AI would take this a step further by adding a natural language wrapper around the entire process, making it widely accessible to even non-technical mortgage processing executives.

Read More: How Mortgage Businesses Can Benefit from Generative AI

What is the Future of Gen AI in Mortgage Processing? Unraveling the Metamorphosis

This new form of artificial intelligence goes beyond simple automation of manual tasks. It signals a paradigmatic shift in how we approach mortgage processes, interpret data, and make decisions. Here are the potential impact areas in the not-so-far-away future:

1. The mortgage approval process shrinks dramatically

Now: Today, the average mortgage processing timeline from the first application to closure takes anywhere between 30 and 60 days. A lot of this is due to protracted approvals, where mortgage applicants must jump through several hoops to complete verifications and obtain funds. Not only is this detrimental to the customer experience, but it also lowers mortgage processing capacity and, consequently, profitability.

Tomorrow: Sophisticated Gen AI platforms can crunch applicant data — both structured and big data — to generate approval estimates in an easy-to-understand natural language. Eventually, it could facilitate instantaneous approvals, accessible 24/7.

2. Servicing workflows need minimal documentation

Now: While technology has progressed, servicing workflows have, paradoxically, become less efficient. This is due to a variety of factors like new regulations, economic pressures, and the pandemic. Since 2019, the cost of servicing alone has increased by over 9%, with admin expenses contributing to 20% of costs. At the same time, the average number of loans serviced by one mortgage executive dropped from 900 to 723.

Tomorrow: Gen AI could significantly reduce the documentation burden involved in mortgage servicing by automatically extracting data from multiple disparate sources. It could also help build contextualized memory, which means that each customer communication is automatically contextualized with previous information without the need for repeated documentation exchange.

3. Personalized mortgage products help navigate financial crises

Now: Prospective home buyers have encountered numerous challenges in the last few years, from COVID-related restrictions to rising interest rates, from inflation cutting into savings to limited real estate inventory. Buying a home remains out of reach for most Gen Z home buyers, even millennials, due to unprecedented changes in the country’s socio-economic front.

Tomorrow: While much of this is systemic, gen AI could offer innovative workarounds and product/service packages tailored to individual buyers’ financial conditions. What’s more, the technology would be able to complete this complex mortgage processing task in a fraction of the time it’d take us humans, enabling personalization without increasing headcount or costs.

4. Mortgage decision-makers are better equipped to adapt to a fast-changing future

Now: The mortgage industry has witnessed rapid changes in the last few years, with several trends bucking our expectations. For example, despite high-interest rates and inflation, there’s still huge demand for new properties. Decision makers need to be able to quickly analyze information, draw inferences, and take the right action for their mortgage processing businesses. Unfortunately, siloed data, lack of data science skills, and complicated dashboards often get in the way of this.

Tomorrow: Gen AI would be able to generate data snapshots from vast amounts of information, including market intelligence, regulatory documents, and mortgage processing analytics. These natural language snapshots will include recommendations with quantifiable accuracy estimates and weighted evaluations.

Read More: What Is Decision Intelligence and How Can It Impact the Mortgage Industry

Preparing for the Gen AI Metamorphosis

While it may take a few years to reach maturity, the era of gen AI in mortgage processing is already here. Organizations need to start preparing by shoring up their data infrastructure, cloud-based systems, and, most importantly, their operational culture.

Gen AI is poised to upend your current working model — and Nexval is here to help you navigate this change. Our team of 1000+ SMEs works with top American mortgage companies to pave the way for digital transformation now and in the future.

Speak with our experts to know more.